Discussion with comments

Why is it so important to formulate your brief for a data presentation? Discuss some ways you would implement to formulate an effective brief. What are some advantages to your methods? What are some disadvantages?

According to Kirk (2016), The essence of “Formulating Your Brief” is to “identify the context in which your work will be undertaken and then define its aims: it is the who, what, where, when and how.” It could be formal or informal as any project you think you must make it. This phase is where you create a vision for your work.

Reference

Kirk, A. (2016). Data Visualisation: A Handbook for Data Driven Design. Thousand Oaks, CA: Sage Publications, Ltd.

Data Visualisation
2

3

Data Visualisation
A Handbook for Data Driven Design
2nd Edition
Andy Kirk
Los Angeles
London
New Delhi
Singapore
Washington DC
Melbourne
4

SAGE Publications Ltd
1 Oliver’s Yard
55 City Road
London EC1Y 1SP
SAGE Publications Inc.
2455 Teller Road
Thousand Oaks, California 91320
SAGE Publications India Pvt Ltd
B 1/I 1 Mohan Cooperative Industrial Area
Mathura Road
New Delhi 110 044
SAGE Publications Asia-Pacific Pte Ltd
3 Church Street
#10-04 Samsung Hub
Singapore 049483
© Andy Kirk 2019
First edition published 2016. Reprinted four times in 2016, twice in 2017, three times in 2018,
and three times in 2019.
Apart from any fair dealing for the purposes of research or private study, or criticism or review, as
permitted under the Copyright, Designs and Patents Act, 1988, this publication may be
reproduced, stored or transmitted in any form, or by any means, only with the prior permission
in writing of the publishers, or in the case of reprographic reproduction, in accordance with the
terms of licences issued by the Copyright Licensing Agency. Enquiries concerning reproduction
outside those terms should be sent to the publishers.
Library of Congress Control Number: 2018964578
British Library Cataloguing in Publication data
A catalogue record for this book is available from the British Library
ISBN 978-1-5264-6893-2
ISBN 978-1-5264-6892-5 (pbk)
Editor: Aly Owen
Editorial assistant: Lauren Jacobs
Production editor: Ian Antcliff
5

Copyeditor: Neville Hankins
Proofreader: Christine Bitten
Indexer: David Rudeforth
Marketing manager: Susheel Gokarakonda
Cover design: Shaun Mercier
Typeset by: C&M Digitals (P) Ltd, Chennai, India
Printed in the UK
At SAGE we take sustainability seriously. Most of our products are printed in the UK using responsibly sourced papers and
boards. When we print overseas we ensure sustainable papers are used as measured by the PREPS grading system. We undertake
an annual audit to monitor our sustainability.
6

Contents
Acknowledgements
About the Author
Discover Your Textbook’s Online Resources
Introduction
PART A FOUNDATIONS
1 Defining Data Visualisation
2 The Visualisation Design Process
PART B THE HIDDEN THINKING
3 Formulating Your Brief
4 Working With Data
5 Establishing Your Editorial Thinking
PART C DEVELOPING YOUR DESIGN SOLUTION
6 Data Representation
7 Interactivity
8 Annotation
9 Colour
10 Composition
Epilogue
References
Index
7

Acknowledgements
I could not have written this book without the unwavering support of my wonderful wife, Ellie,
and my family. The book is dedicated to my inspirational Dad who sadly passed away before its
publication. I want to acknowledge the contributions of the thousands of data visualisation
practitioners who have created such a wealth of exceptional design work and smart writing. I
have been devouring this for over a decade now and I am constantly inspired by the talents and
minds behind it all. I also want to express my gratitude to the people and organisations who have
granted me permission to reference and showcase their visualisation work in this book. Sincere
thanks to the many people at Sage who have played a role in making this book grow from the
first proposal and now to a second edition. Finally, to you the readers, I am hugely thankful that
you chose to invest in this book. I hope it helps you in your journey to learning about this super
subject.
8

About the Author
Andy Kirk
is a freelance data visualisation specialist based in Yorkshire, UK. He is a visualisation design
consultant, training provider, teacher, author, speaker, researcher and editor of the award-
winning website visualisingdata.com.
After graduating from Lancaster University in 1999 with a BSc (hons) in Operational
Research, Andy’s working life began with a variety of business analysis and information
management roles at organisations including CIS Insurance, West Yorkshire Police and the
University of Leeds.
He discovered data visualisation in early 2007, when it was lurking somewhat on the fringes
of the Web. Fortunately, the timing of this discovery coincided with his shaping of his
Master’s (MA) degree research proposal, a self-directed research programme that gave him
the opportunity to unlock and secure his passion for the subject.
He launched visualisingdata.com to continue the process of discovery and to chart the
course of the increasing popularity of the subject. Over time, this award-winning site has
grown to become a popular reference for followers of the field, offering contemporary
discourse, design techniques and vast collections of visualisation examples and resources.
Andy became a freelance professional in 2011. Since then he has been fortunate to work
with a diverse range of clients across the world, including organisations such as Google,
CERN, Electronic Arts, the EU Council, Hershey and McKinsey. At the time of
publication, he will have delivered over 270 public and private training events in 25
different countries, reaching more than 6000 delegates. Alongside his busy training
schedule, Andy also provides design consultancy, his primary client being the Arsenal FC
Performance Team, since 2015.
In addition to his commercial activities, he maintains regular engagements in academia.
Between 2014 and 2015 he was an external consultant on a research project called ‘Seeing
Data’, funded by the Arts & Humanities Research Council and hosted by the University of
Sheffield. This study explored the issues of data visualisation literacy among the general
public and, inter alia, helped to shape an understanding of the human factors that affect
visualisation literacy and the effectiveness of design.
Andy joined the highly respected Maryland Institute College of Art (MICA) as a visiting
lecturer in 2013 teaching a module on the Information Visualisation Master’s Programme
through to 2017. From January 2016, he taught a data visualisation module as part of the
MSc in Business Analytics at the Imperial College Business School in London through to
2018. As of May 2019, Andy has started teaching at University College London (UCL).
9

Discover Your Textbook’s Online Resources
Want more support around understanding and creating data visualisations? Andy Kirk is here to
help, offline and on!
Hosted by the author and with resources organized by chapter, the supporting website for this
book has everything you need to explore, practice, and hone your data visualisation skills.
Explore the field: expand your knowledge and reinforce your learning about working with
data through libraries of further reading, references, and tutorials.
Try this yourself: revise, reflect, and refine your skill and understanding about the
challenges of working with data through practical exercises.
See data visualisation in action: get to grips with the nuances and intricacies of working
with data in the real world by navigating instalments of the narrative case study and seeing
an additional extended example of data visualisation in practice. Follow along with Andy’s
video diary of the process and get direct insight into his thought processes, challenges,
mistakes, and decisions along the way.
Chartmaker directory: access crowd-sourced guidance that aims to answer the crucial
question ‘which tools make which charts?’ with this growing directory of examples and
technical solutions for chart building.
Ready to learn more? Go beyond the book and dive deeper into data visualisation via the rest of
Andy’s website (www.visualisingdata.com), which contains data visualisation tools and
software, links to additional influential further reading, and a blog with monthly collections of
the best data visualisation examples and resources each month.
10

http://www.visualisingdata.com

Introduction
The primary challenge one faces when writing a book about data visualisation is to determine
what to leave in and what to leave out. Data visualisation is a big subject. There is no single book
to rule it all because there is no one book that can truly cover it all. Each and every one of the
topics covered by the chapters in this book could (and, in several cases, do) exist as books in their
own right.
The secondary challenge when writing a book about data visualisation is to decide how to weave
the content together. Data visualisation is not rocket science; it is not an especially complicated
discipline, though it can be when working on sophisticated topics and with advanced
applications. It is, however, a complex subject. There are lots of things to think about, many
things to do and, of course, things that will need making. Creative and journalistic sensibilities
need to blend harmoniously with analytical and scientific judgement. In one moment, you might
be checking the statistical rigour of an intricate calculation, in the next deciding which shade of
orange most strikingly contrasts with a vibrant blue. The complexity of data visualisation
manifests in how the myriad small ingredients interact, influence and intersect to form a whole.
The decisions I have made when formulating this book’s content have been shaped by my own
process of learning. I have been researching, writing about and practising data visualisation for
over a decade. I believe you only truly learn about your own knowledge of a subject when you
have to explain it and teach it to others. To this extent I have been fortunate to have had
extensive experience designing and delivering commercial training as well as academic teaching.
I believe this book offers an effective and proven pedagogy that successfully translates the
complexities of this subject in a form that is fundamentally useful. I feel well placed to bridge the
gap between the everyday practitioners, who might identify themselves as beginners, and the
superstar talents expanding the potential of data visualisation. I am not going to claim to belong
to the latter cohort, but I have certainly been a novice, taking tentative early steps into this world.
Most of my working hours are spent helping others start their journey. I know what I would have
valued when I started out in this field and this helps inform how I now pass this on to others in
the same position I was several years ago.
There is a large and growing library of fantastic books offering different theoretical and practical
viewpoints on this subject. My aim is to add value to this existing collection by approaching the
subject through the perspective of process. I believe the path to mastering data visualisation is
achieved by making better decisions: namely, effective choices, efficiently made. I will help you
understand what decisions need to be made and give you the confidence to make the right
choices. Before moving on to discuss the book’s intended audience, here are its key aims:
To challenge your existing approaches to creating and consuming visualisations. I will
challenge your beliefs about what you consider to be effective or ineffective visualisation. I
will encourage you to eliminate arbitrary choices from your thinking, rely less on taste and
instinct, and become more reasoned in your judgements.
To enlighten you I will increase your awareness of the possible approaches to visualising
11

data. This book will broaden your visual vocabulary, giving you a wider and more
sophisticated understanding of the contemporary techniques used to express your data
visually.
To equip is to provide you with robust tactics for managing your way through the myriad
options that exist in data visualisation. To help you overcome the burden of choice, an
adaptable framework is offered to help you think for yourself, rather than relying on
inflexible rules and narrow instruction.
To inspire is to open the door to a subject that will stimulate you to elevate your ambition
and broaden your confidence. Developing competency in data visualisation will take time
and will need more than just reading this book. It will require a commitment to embrace
the obstacles that each new data visualisation opportunity poses through practice. It will
require persistence to learn, apply, reflect and improve.
12

Who Is This Book Aimed At?
Anyone who has reason to use quantitative and qualitative methods in their professional or
academic duties will need to grasp the demands of data visualisation. Whether this is a large part
of your duties or just a small part, this book will support your needs.
The primary intended audiences are undergraduates, postgraduates and early-career researchers.
Although aimed at those in the social sciences, the content will be relevant to readers from across
the spectrum of arts and humanities right through to the natural sciences.
This book is intended to offer an accessible route for novices to start their data visualisation
learning journey and, for those already familiar with the basics, the content will hopefully
contribute to refining their capabilities. It is not aimed at experienced or established visualisation
practitioners, though there may be some new perspectives to enrich their thinking: some content
will reinforce existing knowledge, other content might challenge their convictions.
The people who are active in this field come from all backgrounds. Outside academia, data
visualisation has reached the mainstream consciousness in professional and commercial contexts.
An increasing number of professionals and organisations, across all industry types and sizes, are
embracing the importance of getting more value from their data and doing more with it, for both
internal and external benefit. You might be a market researcher, a librarian or a data analyst
looking to enhance your data capabilities. Perhaps you are a skilled graphic designer or web
developer looking to take your portfolio of work into a more data-driven direction. Maybe you
are in a managerial position and though not directly involved in the creation of visualisation
work, you might wish to improve the sophistication of the language you coordinate or
commission others who are. Everyone needs the lens and vocabulary to evaluate work effectively.
Data visualisation is a genuinely multidisciplinary discipline. Nobody arrives fully formed with
all constituent capabilities. The pre-existing knowledge, skills or experiences which, I think,
reflect the traits needed to get the most out of this book would include:
Strong numeracy is necessary as well as a familiarity with basic statistics.
While it is reasonable to assume limited prior knowledge of data visualisation, there should
be a strong desire to want to learn it. The demands of learning a craft like this take time and
effort; the capabilities will need nurturing through ongoing learning and practice. They are
not going to be achieved overnight or acquired alone from reading this book. Any book that
claims to be able magically to inject mastery through just reading it cover to cover is over-
promising and likely to under-deliver.
The best data visualisers possess inherent curiosity. You should be the type of person who is
naturally disposed to question the world around them. Your instinct for discovering and
sharing answers will be at the heart of this activity.
There are no expectations of your having any prior familiarity with design principles, but an
appetite to embrace some of the creative aspects presented in this book will heighten the
impact of your work. Time to unleash that suppressed imagination!
If you are somebody fortunate to possess already a strong creative flair, this book will guide
you through when and crucially when not to tap into this sensibility. You should be willing
13

to increase the rigour of your analytical decision making and be prepared to have your
creative thinking informed more fundamentally by data rather than just instinct.
No particular technical skills are required to get value from this book, as I will explain
shortly. But you will ideally have some basic knowledge of spreadsheets and experience of
working with data irrespective of which particular tool.
This is a portable practice involving techniques that are subject-matter agnostic. Throughout this
book you will see a broad array of examples from different industries covering many different
topics. Do not be deterred by any example being about a subject different to your own area of
interest. Look beyond the subject and you will see analytical and design choices that are just as
applicable to you and your work: a line chart showing political forecasts involves the same
thought process as would a line chart showing stock prices changing or average global
temperatures rising. A line chart is a line chart, regardless of the subject matter.
The type of data you are working with is the only legitimate restriction to the design methods
you might employ, not your subject and certainly not traditions in your subject. ‘Waterfall charts
are only for people in finance’, ‘maps are only for cartographers’, ‘Sankey diagrams are only for
engineers’. Enter this subject with an open mind, forget what you believe or have been told is the
normal approach, and your capabilities will be expanded.
Data visualisation is an entirely global community, not the preserve of any geographic region.
Although the English language dominates written discourse, the interest in the subject and work
created from studios through to graphics teams originates everywhere. There are cultural
influences and different flavours in design sensibility around the world which enrich the field
but, otherwise, it is a practice common and accessible to all.
14

Finding the Balance
15

Handbook vs Manual
The description of this book as a ‘handbook’ positions it as distinct from a tutorial-based
manual. It aims to offer conceptual and practical guidance, rather than technical instruction.
Think of it more as a guidebook for a tourist visiting a city than an instruction manual for how
to fix a washing machine.
Apart from a small proportion of visualisation work that is created manually, the reliance on
technology to create visualisation work is an inseparable necessity. For many beginners in
visualisation there is an understandable appetite for step-by-step tutorials that help them
immediately to implement their newly acquired techniques.
However, writing about data visualisation through the lens of selected tools is hard, given the
diversity of technical options that exist in the context of such varied skills, access and needs. The
visualisation technology space is characterised by flux. New tools are constantly emerging to
supplement the many that already exist. Some are proprietary, others are open source; some are
easier to learn but do not offer much functionality; others do offer rich potential but require a
great deal of foundation understanding before you even accomplish your first bar chart. Some
tools evolve to keep up with current techniques; they are well supported by vendors and have
thriving user communities, others less so. Some will exist as long-term options whereas others
depreciate. Many have briefly burnt brightly but quickly become obsolete or have been
swallowed up by others higher up the food chain. Tools come and go but the craft remains.
There is a role for all book types and a need for more than one to acquire true competency in a
subject. Different people want different sources of insight at different stages in their
development. If you are seeking a text that provides instructive tutorials, you will learn from this
how to accomplish technical developments in a given technology. However, if you only read
tutorial-based books, you will likely fall short in the fundamental critical thinking that will be
needed to harness data visualisation as a skill.
I believe a practical, rather than technical, text focusing on the underlying craft of data
visualisation through a tool-agnostic approach offers the most effective guide to help people learn
this subject.
The content of this book will be relevant to readers regardless of their technical knowledge and
experience. The focus will be to take your critical thinking towards a detailed, fully reasoned
design specification – a declaration of intent of what you want to develop. Think of the
distinction as similar to that between architecture (design specification) and engineering (design
execution).
There is a section in Chapter 3 that describes the influence technology has on your work and the
places it will shape your ambitions. Furthermore, among the digital resources offered online are
further profiles of applications, tools and libraries in common use in the field today and a vast
directory of resources offering instructive tutorials. These will help you to apply technically the
critical capabilities you acquire throughout this book.
16

17

Useful vs Beautiful
Another important distinction to make is that this book is not intended to be seen as a beauty
pageant. I love flicking through glossy ‘coffee table’ books as they offer great inspiration, but
often lack substance beyond the evident beauty. This book serves a different purpose to that. I
believe, for a beginner or relative beginner, the most valuable inspiration comes more from
understanding the thinking behind some of the amazing works encountered today, learning
about the decisions that led to their conceptual development.
My desire is to make this the most useful text available, a reference that will spend more time on
your desk than on your bookshelf. To be useful is to be used. I want the pages to be dog-eared. I
want to see scribbles and annotated notes made across its pages and key passages underlined. I
want to see sticky labels peering out above identified pages of note. I want to see creases where
pages have been folded back or a double-page spread that has been weighed down to keep it
open. It will be an elegantly presented and packaged book, but it should not be something that
invites you to look but not touch.
18

Pragmatic vs Theoretical
The content of this book has been formed through years of absorbing knowledge from as many
books as my shelves can hold, generations of academic work, endless web articles, hundreds of
conference talks, personal interactions with the great and the good of the field, and lots and lots
of practice. More accurately, lots and lots of mistakes. What I present here is a pragmatic
distillation of what I have learned and feel others will benefit from learning too.
It is not a deeply academic or theoretical book. Experienced or especially curious practitioners
may have a desire for deeper theoretical discourse, but that is beyond the intent of this particular
text. You have to draw a line somewhere to determine the depth you can reasonably explore
about a given topic. Take the science of visual perception, for example, arguably the subject’s
foundation. There is no value in replicating or attempting to better what has already been
covered by other books in greater quality than I could achieve.
An important reason for giving greater weight to pragmatism is because of the inherent
imperfections of this subject. Although there is so much important empirical thinking in this
subject, the practical application can sometimes fail to translate beyond the somewhat artificial
context of a research study. Real-world circumstances and the strong influence of human factors
can easily distort the significance of otherwise robust concepts.
Critical thinking will be the watchword, equipping you with the independence of thought to
decide rationally for yourself which solutions best fit your context, your data, your message and
your audience. To accomplish this, you will need to develop an appreciation of all the options
available to you (the different things you could do) and a reliable approach for critically
determining what choices you should make (the things you will do and why).
19

Contemporary vs Historical
I have huge respect for the ancestors of this field, the dominant names who, despite primitive
means, pioneered new concepts in the visual display of statistics to shape the foundations of the
field being practised today. The field’s lineage is decorated by pioneers such as William Playfair,
W. E. B. Du Bois, Florence Nightingale and John Snow, to name but a few. To many beginners
in the field, the historical context of this subject is of huge interest. However, this kind of
content has already been covered by plenty of other book and article authors.
I do not want to bloat this book with the unnecessary reprising of topics that have been covered
at length elsewhere. I am not going to spend time attempting to enlighten you about how we live
in the age of ‘Big Data’ and how occupations related to data are or will be the ‘sexiest jobs’ of our
time. The former is no longer news, the latter claim emerged from a single source. There is more
valuable and useful content I want you to focus your time on.
The subject matter, the ideas and the practices presented here will hopefully not date a great deal.
Of course, many of the graphic examples included in the book will be surpassed by newer work
demonstrating similar concepts as the field continues to develop. However, their worth as
exhibits of a particular perspective covered in the text should prove timeless. As time passes there
will be new techniques, new concepts and new, empirically evidenced rules. There will be new
thought-leaders, new sources of reference and new visualisers to draw insight from. Things that
prove a manual burden now may become seamlessly automated in the near future. That is the
nature of a fast-growing field.
20

Analysis vs Communication
A further distinction to make concerns the subtle but critical difference between visualisation
used for analysing data and visualisation used for communicating data.
Before a visualiser can confidently decide what to communicate to others, he or she needs to have
developed an intimate understanding of the qualities and potential of the data. In certain
contexts, this might only be achieved through exploratory data analysis. Here, the visualiser and
the viewer are the same person. Through visual exploration, interrogations of the data can be
conducted to learn about its qualities and to unearth confirmatory or enlightening discoveries
about what insights exist.
Visualisation for analysis is part of the journey towards creating visualisation for communication,
but the techniques used for visual analysis do not have to be visually polished or necessarily
appealing. They are only serving the purpose of helping you truly to learn about your data.
When a data visualisation is being created to communicate to others, many careful considerations
come into play about the requirements and interests of the intended audience. This influences
many design decisions that do not exist alone with visual analysis.
For the scope of this book the content is weighted more towards methods and concerns about
communicating data visually to others. If your role is concerned more with techniques for
exploratory analysis rather than visual communication, you will likely require a deeper treatment
of the topic than this book can reasonably offer.
Another matter to touch on here concerns the coverage of statistics, or lack thereof. For many
people, statistics can be a difficult topic to grasp. Even for those who are relatively numerate and
comfortable working with simple statistical methods, it is quite easy to become rusty without
frequent practice. The fear of making errors with intricate statistical calculations depresses
confidence and a vicious circle begins.
You cannot avoid the need to use some statistical techniques if you are going to work with data. I
will describe some of the most relevant statistical techniques in Chapter 4, at the point in your
thinking where they are most applicable. However, I do believe the range and level of statistical
techniques most people will need to employ on most of their visualisation tasks can be overstated.
I know there will be exceptions, and a significant minority will be exposed to requiring advanced
statistical thinking in their work.
It all depends, of course. In my experience, however, the majority of data visualisation challenges
will generally involve relatively straightforward univariate and bivariate statistical techniques to
describe data. Univariate techniques help you to understand the shape, size and range of a single
variable of data, such as determining the minimum, maximum and average height of a group of
people. Bivariate techniques are used to observe possible relationships between two different
variables. For example, you might look at the relationship between gross domestic product and
medal success for countries competing at the Olympics. You may also encounter visualisation
challenges that require a basic understanding of probabilities to assist with forecasting risk or
modelling uncertainty.
21

The more advanced applications of statistics will be required when working with larger
complicated datasets, where multivariate techniques are employed simultaneously to model the
significance of relationships between multiple variables. Above and beyond that, you are moving
towards advanced statistical modelling and algorithm design.
Though it may seem unsatisfactory to offer little coverage of this topic, there is no value in
reinventing the wheel. There are hundreds of existing books better placed to offer the depth you
might need. That statistics is such a prolific and vast field in itself further demonstrates how
deeply multidisciplinary a field visualisation truly is.
22

Chapter Contents
The book is organised into three main parts (A, B and C) comprising ten chapters and an
Epilogue. Each chapter opens with a preview of the content to be covered and closes with a
summary of the most salient learning points to emerge. There are collections of further resources
available online to substantiate the learning from each chapter.
For most readers, especially beginners, it is recommended that you start from the beginning and
proceed through each chapter as presented. For those setting out to begin working on their own
visualisation, you might jump straight into Chapters 2–5 to ensure you are fully prepared for
some of the important preparatory activities you need to accomplish before moving on to look at
developing your design solution. For those with more experience and/or prior exposure to this
subject, who are perhaps looking to fine-tune specific aspects of their design skills, most of your
interest will lie in Part C, comprising Chapters 6–10. For readers who just want to dip in and out
of specific topic areas, although each chapter builds sequentially from the preceding ones, they
can all be read in isolation. Follow any sequence that satisfies your needs. The coloured tabs on
the outer edge will provide quick visual navigation through the distinct parts and chapters
within.
23

Part A: Foundations
Part A introduces some important foundational understanding about data visualisation as a
subject area and as an activity. The contents of the first two chapters give shape to the coverage
across the rest of the book.
Chapter 1 ‘Defining Data Visualisation’ will be the logical starting point for those who are new
to the field, providing a definition for the subject and exploring some of the tensions that enrich
this subject. The second section explains some of the distinctions and overlaps with other related
disciplines. If you already know what data visualisation is about, you might choose to pass on
this; it does, though, help frame many of the discussions elsewhere.
Chapter 2 ‘The Visualisation Design Process’ introduces the value of following a design process,
the sequence of activities around which the book’s contents in Parts B and C are organised. It
explains what is involved and offers some useful tips to help you seamlessly adopt this approach.
Where the process offers organisation and efficiency, design principles ensure effectiveness. The
second section will describe what separates the good from the bad in visualisation design,
building up your convictions to help with your upcoming decision making.
24

Part B: The Hidden Thinking
Part B profiles the first three stages of the data visualisation design process. These are the hidden
preparatory stages that will significantly influence the path you take towards an eventual solution.
Chapter 3 ‘Formulating Your Brief’ covers the opening tasks involved in initiating, defining and
planning the requirements of your work. The first section looks at issues around context,
specifically about the importance of defining curiosity and identifying the circumstances that will
shape your project. The second section considers the vision of your work, looking at what
purpose it intends to serve and how you might creatively define the type of work you will need to
pursue. Finally, a short section looks at the value of harnessing initial ideas.
Chapter 4 ‘Working With Data’ commences your practical involvement with your data,
stepping through the four distinct steps that acquaint you with the potential of your critical raw
material. Data acquisition outlines the different origins of and methods for obtaining your data.
Data examination profiles the different characteristics that define the type, extent and condition
of your data. Data transformation builds on your examination work to find ways of modifying
and enhancing your data to prepare it for use. Finally, data exploration discusses methods for
discovering more about the qualities and insights hidden away in your data.
Chapter 5 ‘Establishing Your Editorial Thinking’ reflects on the possibilities offered by your data
and explains the importance of committing to an editorial path. The chapter opens with a
definition about the influence of editorial thinking, using two case studies to explain how
editorial definitions influence design choices later in the process.
25

Part C: Developing Your Design Solution
Part C represents the main part of this book and covers the five distinct layers of the data
visualisation anatomy. They are presented in separate chapters to help organise your thinking and
to avoid being overwhelmed by the detailed options that exist. However, they are ultimately
interrelated matters and the chapter sequencing across this part is carefully arranged to support
this. Each chapter follows a similar structure, opening with an array of different possible design
options and supplemented by guidance on the factors that will influence your choices. Initially,
you will need to make decisions about what elements to include around data representation
(charts), interactivity and annotation. You will then complete your thinking about the
appearance of these elements, through colour and composition.
Chapter 6 ‘Data Representation’ introduces the act of visual encoding and then expands on this
to provide a detailed profile of 49 distinct chart types to help broaden your visual vocabulary.
The chapter closes with a run through the key factors that will influence the suitability of your
data representation choices.
Chapter 7 ‘Interactivity’ introduces the potential value of incorporating interactive features in
your work, profiling a wide range of options – such as filtering, highlighting and animating –
that will enable users to interrogate and control a visualisation. The chapter closes with the main
considerations that will influence your selection of interactive features.
Chapter 8 ‘Annotation’ describes the importance of providing useful assistance to your viewers,
including headings, chart apparatus, and labels. The chapter closes with a look at which factors
will inform the choices you make.
Chapter 9 ‘Colour’ commences with an overview of different colour models. This provides the
basis for understanding the different ways of applying colour to facilitate data legibility and
deliver functional decoration. Once again, having introduced the options, we will look at how
you arrive at appropriate choices.
Chapter 10 ‘Composition’ explores the final element of developing your design solution
concerning how you organise the placement and sizing of all your visual elements within the
space you have to work. Looking at matters of layout, arrangement and chart sizing, we will then
wrap up this topic with a discussion about how to make your decisions.
Epilogue: To close the book, the epilogue will summarise the development cycle of activities you
will need to undertake as you move your detailed design specification to a fully executed
solution.
26

Digital Resources
The opportunity to supplement the print version of this book with further digital companion
resources helps to offer readers a range of additional learning materials:
a written and video-based case-study of a visualisation project that demonstrates the design
process in action;
an extensive and up-to-date catalogue of over 350 data visualisation tools;
a large collection of tutorials and resources to help develop your technical capabilities in
making a wide range of different charts;
useful exercises designed to help embed the learning covered in each chapter;
a digital gallery of all the artwork included in this book and many further examples of the
concepts presented across all chapters;
refreshed reading resources to support ongoing learning about the subjects covered in each
chapter.
27

Glossary
Consistency in the meaning of language and terms used in data visualisation is important.
Though data visualisation is no different to many fields that get bogged down by superfluous
semantic noise, it can only help to establish clarity about its usage in this book at least.
28

Roles
Visualiser: This is the role I am assigning to you – the person making the visualisation.
Sometimes people prefer to use terms like researcher, analyst, developer, storyteller or even
‘visualist’. Designer would also be particularly appropriate, but I want to broaden the scope of
the role beyond just design to cover all activities involved in this discipline.
Viewer: This is the role assigned to the recipient, who is viewing or using your visualisation
product. It offers a broader and better fit than alternatives such as consumer, reader, user or
customer. However, ‘user’ will be temporarily adopted during the more active chapter about
interactivity.
Audience: This concerns the collective group of viewers for whom your work is intended.
Within an audience there will be cohorts of different viewer types that you might characterise
through distinct personas to help your thinking about serving their varied needs.
Consuming: This will be the general act of the viewer, to consume. I will use more active
descriptions like ‘reading’ and ‘using’ when consuming becomes too passive or vague, and when
distinctions are needed between reading a chart and using interactive features.
29

Data
Raw data: For the purpose of this book, raw data will be the initial state of data you have
collected, received or downloaded that has not yet been subjected to any statistical or
transforming treatment. Some people take issue with the implied ‘rawness’ this label implies,
given that data will have already lost its raw state having been recorded by some instrument,
stored, retrieved and maybe cleaned already. I appreciate this viewpoint but think it is the most
pragmatic label relevant to most people’s understanding.
Data source: This is the term used to describe the origin(s) of the raw data used in a
visualisation.
Dataset: A table of data is an array of values visually arranged into rows and columns, usually
existing in a spreadsheet or database. The rows are the records – instances or items – and the
columns are the variables – details about the items. Datasets are visualised in order to ‘see’ the
size, patterns and relationships that are otherwise hard to observe. A dataset may comprise one or
a collection of several tables.
Tabulation: For the purpose of this book, I distinguish between types of datasets that are
‘normalised’ and others that are ‘cross-tabulated’. This distinction will be explained in context in
Chapter 4.
Data types: The variables (columns) in a table that hold details about items (records) will have
different scales of measurement or data types. At the most general level, distinctions in
quantitative (e.g. salary) and categorical (e.g. gender) data are important in how you will
statistically and visually handle them. A detailed distinction between data types, with examples,
will again be offered in Chapter 4.
Series: A series of values is essentially a sequence of related values in a table. An example of a
series would be the highest recorded temperatures in a city for each day over a month. Though
individual daily values will be stored as distinct moment-in-time measurements, the activity of
temperature never stops ‘happening’ and therefore the collected values have a legitimate
continuous relationship through the series.
30

Visualisation
Project: For the purpose of this book, we will consider the development of a data visualisation as
being a project. Even though you might consider something a quick, small task, it will still need
to involve the thinking consistent with the stages of the process covered in this book.
Chart type: Charts are visual representations of data. There are many ways of represent- ing your
data, using different combinations of marks, attributes, layouts and apparatus. Their
combinations form archetypes of charts more commonly named chart types, such as the bar chart,
dendrogram or treemap.
Graphs, plots, diagrams and maps: Traditionally the term graph has been used to describe
visualisations that display network relationships, while chart would be commonly used to label
common devices like the bar or pie chart. Plots and diagrams are more specifically attached to
special types of displays but with no pattern of consistency in their usage. All these terms are so
interchangeable that any energy expended in explaining meaningful difference is redundant. For
the purpose of this book, I will generally stick to the term chart to act as the main label to cover
all representation types. In places, this ‘umbrella’ term will incorporate thematic maps, for the
sake of convenience, even though they clearly have a visual structure that is quite different to
standard charts.
Graphic: The term graphic will be used when referring to visuals more focused on information-
led displays such as explanation or process diagrams as distinct from charts that are concerned
with data-driven visuals. It might also be used to refer more broadly to a visualisation that
incorporates charts, text and images.
Format: This concerns the difference in output form between printed work, digital work and
physical visualisation work.
Functionality: This concerns the difference in whether a visualisation is static or interactive.
Interactive visualisations allow you to manipulate and interrogate a computer-based display of
data. They are published on the Web, exist within apps, or are on larger digital displays, as in
galleries. In contrast, a static visualisation displays a non-changeable, still display of data that
could be published in print but also digitally. Just because something is published digitally does
not automatically make it interactive.
Axes: Many common chart types have axis lines that provide a reference for measuring
quantitative values or positioning categorical values. The horizontal axis is known as the x-axis
and the vertical axis is known as the y-axis.
Scales: Scales exist in two forms, typically. Firstly, as a set of marks along an axis that indicate
positions for the range of values included in a chart. Scales are normally presented in regular
intervals (10, 20, 30, etc.) representing units of measurement, such as prices, distances, years or
percentages. A scale may also be presented in a key to explain associations between, for example,
different sizes of areas or classifications of different colour attributes.
31

Legend: Charts that employ visual attributes, such as colours, shapes or sizes to represent values
of data, will often be accompanied by a legend to house visual explanations of classifications,
known as keys.
Outliers: Outliers are points of data that are outside the normal range of values. They are the
unusually large or small or simply different values that stand out and generally draw a viewer’s
attention.
Correlation: This is a measure of the presence and extent of a mutual relationship between two
or more variables of data. For example, you would expect to see a correlation between the height
and weight of people or age and salary of workers. Devices like scatter plots, in particular, help
visually to portray possible correlations between two quantitative values.
32

Part A Foundations
33

1 Defining Data Visualisation
This opening chapter will introduce data visualisation through the prism of a proposed
definition. Each component that forms this definition will be explored in depth to illustrate some
of the main characteristics and complexities of this subject.
The second part of the chapter will position data visualisation in the context of other related
disciplines or fields, explaining where overlaps or clear distinctions exist. Overall, this chapter will
seek to forge a shared understanding that will help set the tone and reasoning for the structure of
this book.
34

1.1 What Is Data Visualisation?
It is useful to commence this book with a definition of data visualisation (Figure 1.1). It helps to
ensure we (you the reader, me the writer) have a mutual understanding, from the outset, about
what is meant by data visualisation in the context of this text. The components of this definition
carve the subject into distinct perspectives around which the contents of this book are organised.
Figure 1.1 A Definition for Data Visualisation
Let me delve into this and describe the roles of and relationships between each component
expressed. I will also explain where and how these topics will be covered. Firstly, let’s look at
data.
Data is names and amounts. It is groupings, descriptions and measurements. It is dates and
locations. It will be helpful for discussions in this book to think of data as being typically
structured in table form, with rows of records and columns of variables. Most data we commonly
encounter will exist in textual, numeric or a combined form, but it is also worth noting the
opportunities that increasingly exist for working with data assets in media forms of images, audio
and video.
In Chapter 4 you will learn about the importance of developing an intimate understanding of
your data to acquaint yourself fully with its properties, its condition and its qualities.
You will see that data is the fundamental element driving the decisions across this design process.
Without data there is no material to feed nor necessitate a visualisation. Conversely, without
visualisation the value of data can be unfulfilled. This is not to say we should always visualise
data, absolutely not, but in most circumstances, to harness the maximum value of data, there are
missed opportunities if we do not.
To explain, here is a simple illustration. When data is presented in a table, it is a straightforward
task for a viewer to scan the rows and columns to seek out values of relevance or to discover
particular data points that trigger interest. For instance, by viewing the table in Figure 1.2 it
should prove quite simple to find out what the percentage share of online sales for a Company X
was during April 2016. Now look for the percentage share of store sales during December 2011.
35

Figure 1.2 Proportion of Sales % by Channel Over Time
As a viewer your task is simply to find the relevant row and column intersection: look at the value
display and read it. The percentage share of online sales for Company X during April 2016 is 84,
and for store sales during December 2011 it is 71.
To find which sales channel had the second largest percentage share of sales during August 2014,
again just find the relevant row, compare the three quantitative values along that row, and then
determine which channel column contains the second-ranked amount. For this month, the
online channel, at 44, had the second largest percentage share of sales.
The limitations of reading data when it is presented in this form emerge when we want to answer
broader questions: that is, enquiries that transcend the scope of an answer originating from a
single or small number of adjacent data points. From the same table, how easy do you find it to
identify the headline trends across each sales channel over the period of time displayed?
You can probably ascertain that the percentage share of sales for stores starts quite high then
drops to nothing, the percentage share of online sales starts quite low and then reaches the 100%
maximum, and the percentage share of sales via telephone is consistently tiny.
Though it takes a while to study the values under each sales channel column in order to form this
summary observation, it is still possible. But what if your observations need to be formed more
quickly? What if you needed to know more about the localised patterns of ups and downs within
those global trends? What if you wanted to identify the first occasion when the percentage share
of online sales exceeded the percentage share of store sales? When was the last occasion the
percentage share of store sales exceeded that of online sales? During which periods did the
different sales channels experience the most accelerated upward or downward changes?
36

These are harder questions to answer efficiently and accurately from the data alone. This is
because synthesising observations from multiple values across different rows and columns to
perceive broader relationships fails to exploit fully the capabilities of our visual system – how our
eyes and mind work together to make sense of objects and patterns. To read values in isolation,
store them in our short-term memory and compare them in our head with other isolated values is
mentally challenging. It is not impossible, since we can still accomplish this with just a table of
data, but it will take an excessive amount of time and effort.
This workload will also only increase as the data grows in volume and complexity. For instance,
what if this table were 1000 rows deep and there were 20, 50 or 100 different columns to work
through? Or, what if the quantities had similar value sizes and more modest variation? How easy
would it then be to notice significant patterns?
The crux of all this is that we can look at data, but we cannot really see it. To see data, we need to
represent it in a different, visual form.
Returning to the definition, the term visual representation is arguably the quintessential activity
of data visualisation. Representation involves making decisions about how you are going to
portray your data visually so that the subject understanding it offers can be made accessible to
your audience. In simple terms, this is all about charts and the act of selecting the right chart to
show the features of your data that you think are most relevant.
The building blocks of any chart are marks and attributes. Marks can be points, lines or shapes
and they are used to represent items of data. An example of an item of data from the table in
Figure 1.2 would be the ‘percentage share of sales from stores during June 2014’. Not the value
itself, more the thing the value is about.
Attributes, sometimes described as channels, are visual variations of marks to represent the values
associated with each. These include properties such as different scales of size, colour or position.
If the item of data is ‘percentage share of sales from stores during June 2014’, an attribute would
be used to represent the associated value, in this case 72. If marks and attributes are the
ingredients, the different combinations used create different chart types – the recipes.
Figure 1.3 shows a chart of the data shown in the table from Figure 1.2. Here the data is
represented using a line chart, a common chart type used to show how quantitative values change
over time. In this case the items of data are represented by point marks, positioned at the
intersection of the relevant x and y positions for each reporting month and channel. The
attributes used here are, firstly, the connected lines that join the continuous series of values for
each channel and, secondly, the distinct colours applied to distinguish each line path and
associate them with their respective sales channel category.
Figure 1.3 Proportion of Sales Percentage by Channel over Time
37

As a viewer, you scan this chart to form observations about the three sales channels individually
and then compare them with each other. The comparisons made between separate channels are
especially relevant for this data as the quantities shown are representative of parts of a 100%
whole. This means that at any given point along the timeline, the change in value for one
channel will have an effect on the values across the two others.
Consuming this data in chart form, as opposed to reading a table, enables a viewer to process
clusters of multiple data points simultaneously to identify the slopes and flats, the peaks and
troughs, as well as gaps and cross-overs between lines. Though the precision of determining an
individual data point (e.g. from the chart, what was the percentage share of online sales during
April 2016?) is slightly diminished compared with the ease of performing the same task with data
in table form, observations about the collective patterns and relationships, in turn, become more
precise. The story of the rise in dominance of online sales and the related decline of store sales is
immediately apparent, but what is striking here is an intense pattern of ebb and flow during the
time period of mid-2014 to mid-2015, out of which the significant respective changes in
trajectory of online and store sales materialised and continues.
The chart has the same data as the table, but it is represented differently. Whether this chart view
is better than a table view depends on the purpose of your communication and the needs of your
audience. You do not chart data because you can, you do it because it provides a window for
seeing different features of data. We will explore the judgements you need to make about what
you want to show your audience in Chapter 5 and, in particular, in Chapter 6 where you will
learn about the wide range of established chart types that are commonly used by the visualisers of
today. These charts vary in complexity and composition. Each is capable of accommodating
different types of data and portraying different types of analysis. This chapter will broaden your
visual vocabulary, giving you an appreciation of more ways to express your data. It will also
increase the sophistication of how you go about making effective choices.
The next component of the definition is presentation, which concerns all the other design
decisions that make up the full anatomy of any visualisation. As this text is focused on creating
visualisation as a means for communicating to others, presentation concerns how we choose to
‘package up’ a visualisation work to impart it to an audience, irrespective of the medium or
disseminating method.
38

Visual presentation includes design choices such as the possible application of interactivity,
features of annotation, all matters around colour usage, and the composition of the work.
Considering the line chart in Figure 1.3, if this was intended for the Web, you could envisage
interactivity being useful to offer tooltip details of value labels as you hover over parts of each
line. You could offer controls to modify the x-axis time range or filtering to hide or show
different lines of interest. There are some features of annotation already on display with this
chart, such as the title, the colour legend, and the x- and y-axis scales. You could also add
captions to provide explanations about some of the most noticeable patterns in the data. As
mentioned, colour is already used as an attribute, to distinguish the lines for each sales channel
category, but the application of colour extends across every visible element including the
background shading, gridline colours, and colouring of any text or labels. Finally, composition
relates to the size and placement of all design elements, like the dimensions of the chart area, the
alignment and size of the title, and the placement of the axis labels.
The thinking that goes into designing the full anatomy of a visualisation – combining visual
representation and presentation – is inevitably interconnected. The selection of a chart type
inherently triggers a need to think about the space and place it will occupy on your screen or
page; a clickable interactive feature that reveals annotated captions requires careful thought about
how to style the text and what colours to use.
There are lots of seemingly small design decisions to make in visualisation, little things that add
up to having a big impact. During the early stages of learning this subject it is helpful to partition
your presentation thinking and tackle these design concerns as separate layers. Chapters 7–10 will
explore each of these design matters separately, but in sufficient depth, profiling the options
available and the factors that influence your decisions. As you gain experience and assurance, the
interrelated nature of the choices you make will become more seamless and you will be
stimulated by the depth of thinking demanded of you.
The final component of the definition expresses that data visualisation aims to facilitate
understanding. Everything in this book essentially boils down to helping you accomplish this
objective. We will deal with the term facilitate shortly, but let’s focus for now on the word
understanding.
The notion of understanding is quite broad. To best explain its relevance to data visualisation
requires us, again, to turn to the perspective of a viewer.
When consuming a visualisation, a viewer will go through a process of understanding involving
three phases: perceiving, interpreting and comprehending (Figure 1.4). These are not just synonyms
for the same word, rather they convey distinctions in cognitive focus.
Figure 1.4 The Three Phases of Understanding
39

For the benefit of this illustration we will consider them to occur in a linear sequence, with
successive phases being dependent on the preceding phase having been accomplished. To a
viewer, consciously trying both to understand a visualisation and to extract understanding from a
visualisation, these different phases will feel rather indiscernible. They might appear to occur in
parallel. Viewers are human – there are occasions when rapid interpretations of a chart’s headline
features are made before the whole content has had a chance to be perceived first.
Let’s look at the characteristics of and differences between these phases referring, initially, to an
example chart (Figure 1.5) that presents some headline statistics about footballer Lionel Messi’s
career with FC Barcelona.
The first phase is perceiving, and this concerns the act of reading a chart: ‘what do I see?’. A
viewer decodes how the data is represented to form initial observations about the main features of
the displayed data:
What chart is being used?
What items of data do the marks represent? What value associations do the attributes
represent?
What range of values are displayed?
Are the data and its representation trustworthy?
Figure 1.5 Lionel Messi: Games and Goals for FC Barcelona
40

Source: Data from transfermarkt.com
In the example we see a clustered bar chart showing quantitative values of pairs of categories over
time. This is a chart type I am familiar with and so I feel instantly at ease with the prospect of
consuming it.
I see time is plotted on the x-axis in years – or, more specifically, football seasons – and a shared
quantitative measure is on the y-axis. There are two distinct categories of bars for each season,
with the colour association explained by an explanation key integrated into the title. The
burgundy bars show the games played in a season and the blue bars the number of goals scored.
This title also helps establish clarity about what the data is showing. As the representation
method is understood, initial observations begin to form about the main characteristics of the
display:
What features – shapes, patterns, differences or connections – are observable?
Where are the largest, mid-sized and smallest values? (known as ‘stepped magnitude’
judgements).
Where are the most and the least? Where is the average or normal? (‘global comparison’
judgements).
When scanning the chart, my eyes are drawn to the dominant bars in the middle and towards the
right of the display. I am particularly interested in the highest pair of bars in 2011/12. With
assistance offered by the horizontal gridlines and axis labels I can perceive with reasonable
confidence that the highest number of goals scored was 73 and the most games played was 60. I
can see that the burgundy bars – showing games played – are relatively stable in size since around
2008/09, but the blue bars are more erratic. The bar heights for both categories are much smaller
the further left the time series goes. Looking between the categories, there is no consistency in the
relationship as the burgundy bars are sometimes larger than their blue neighbours, sometimes
smaller.
Interpreting, the second phase of understanding, translates these observations into quantitative
and/or qualitative meaning. Interpreting involves assimilating what you have observed against
what you know about the subject. What does what you have seen mean, given the subject?
What features – shapes, patterns, differences or connections – are interesting?
What features are expected or unexpected?
What features are important given the subject?
The task of drawing interpretations from the observations I made on the chart is helped
considerably by my interest in and knowledge of football. I know that if a player is scoring more
than 25 goals in a season this is very good, and to score over 35 is exceptional. To achieve 50, 60
or indeed 70 goals in a season is frankly preposterous, especially at the highest level of the game. I
know it is rare for a player to be scoring at a ratio of greater than one goal per game played, so the
seasons where a blue bar exceeds the height of the burgundy bars represent a quite remarkable
statistic. I could elaborate on some of the features I expected to (and do) see in this chart based
on knowing the periods when different managers were in charge of Barcelona, which other
players were in the team, and how the team performed from one season to the next. I know what
41

http://transfermarkt.com

to expect in terms of the classic shape of a footballer’s career arc and can map that onto Messi’s,
anticipating that at some point – but not yet – the classic rise, peak and plateau will inevitably be
followed by steady decline.
As this commentary demonstrates, a viewer’s ability to perform rational interpretation will be
significantly determined by factors external to the visualisation itself. The degree of knowledge
viewers possess about the portrayed subject and their capacity to close a knowledge gap is
fundamental. To fulfil the perceiving of a chart, viewers need the context of scale; to fulfil the
interpreting of a chart, viewers need the context of subject. Furthermore, there is the matter of
willingness. At the time of consuming a visualisation, not everyone has the inclination to engage
with it, especially if they have no interest in a subject or if it has no immediate relevance to their
needs.
My connection with the subject of football helped me understand more about the meaning of
the features of data compared with other viewers who might possess no knowledge of the sport.
Switching the subject from football to a completely made-up topic, but using the same chart
with the same data, reinforces this. In Figure 1.6 we see a chart displaying data about the
sightings of Winglets and Spungles.
Figure 1.6 Total Sightings of Winglets and Spungles
I can still perceive the chart, observing the same features as I did when it was portraying Messi’s
quantities of games and goals, but as I have no knowledge of this subject I cannot interpret it. I
have no idea what Winglets and Spungles are, so I cannot form any reasonable sense of what is
interesting, surprising or important about the features of this display. My process of
understanding stops after the perceiving phase.
As this illustrates, any deficit in a viewer’s connection to a subject will fundamentally impede
progress towards performing interpretation. Additionally, this may heighten the risk of the viewer
drawing spurious or unsupported interpretations from a visual display.
In situations where a potential viewer might not possess sufficient knowledge of a subject, it will
require the visualiser to assist in bridging the gap between observation and meaning. This can be
achieved through simple design elements like the provision of captions, inclusion of headlines
42

and astute use of colour to create emphasis, for example. The viewer must then take
responsibility to learn from the assistance provided. As the purple colouring of the middle phase
circle shown in Figure 1.4 denotes, forming useful and reasonable interpretations is a shared
responsibility.
The final phase of understanding is comprehending, which is the consequence or reflective legacy
of the communication experience. The viewers now consider what the interpretations mean to
themselves. What can be inferred as being important to you about the interpretations you have
made?
What has been learnt? Has it reinforced or challenged existing knowledge? Has it been
enlightened with new knowledge?
What feelings have been stirred? Has the experience had an impact emotionally?
What does one do with this understanding? Is it just knowledge acquired or something to
inspire action, such as making a decision or motivating a change in behaviour?
In my case, the outcome of the understanding achieved from the Messi chart is nothing too
dramatic or emotional. There is no direct action linked to it, rather I simply reflect on gaining a
heightened impression, formed out of this data, about how sensational a footballer he has been
and continues to be. For Barcelona fanatics who watch him play every week, they will have
already formed this understanding. This information would only reaffirm what they already
knew. To others less familiar with the subject, it might be more enlightening, but only if they
had any requisite interest.
One person’s ‘wow’ is another person’s ‘I knew that’ is another person’s ‘I don’t care’. Even if
you have just two people in your target audience group, you have potentially two different viewer
profiles. We cannot always anticipate what they do not know, what they want to know and what
is the relevance to them of knowing something.
Visualising data is just an agent of communication and not a guarantor for what a viewer does
with the opportunity for understanding that is presented. There are different flavours of
comprehension, different consequences of understanding formed through this final phase. Many
visualisations will be created with the ambition simply to inform, like the Messi graphic achieved
for me, perhaps to add just an extra grain to the pile of knowledge held about a subject. Not
every visualisation exists to lead a viewer towards some Hollywood-esque moment of grand
discovery, surprising insight or life-changing decision. That is OK, though, as long as the
outcome fits with the intended purpose, something we will discuss in more depth in Chapter 3.
Once again, the association a viewer has with the subject portrayed will greatly influence this
comprehending phase. Returning to the data shown earlier about the percentage sales by channel
over time for Company X, let’s suppose this was a chart produced to assess the effectiveness of a
corporate strategy to consolidate operations towards an online-only sales model.
The outcome of the interpretations formed from this chart might be to draw the conclusion that
whatever actions were taken, they have succeeded. Depending on when the 100% online sales
target was expected, it may be that this chart demonstrates complete success. It might also reveal
belated success. Maybe the company was hoping for 100% online sales far sooner than when
they were achieved. Conversely, the analysis shown might reveal unexpected patterns of sales.
43

Online channels are clearly dominating, but what if the company is still maintaining the expense
of running stores, with staff costs and stock tied up in what appears to be an expired model?
There might be substantial costs assigned to telephone operators waiting for the phone to ring in
order to make a potential sale. But nobody is phoning, so maybe the company should look at
restructuring.
All these are reasonable avenues that comprehending this data could lead to. But, at this point, I
should reveal that the real context of this data actually had nothing to do with sales. That subject
was picked for illustration purposes, but the data was actually about something else.
Specifically, this was data about the ever-shifting forecasts during the night of the 2016 US
Election. The data values came from FiveThirtyEight, a respected website noted for its use of
statistical techniques to analyse and tell stories about elections and several other data-rich
subjects. The quantities relate to the ‘chances of winning the presidency’ forecasts for the two
main party candidates, as well as a residual ‘other result’ percentages to make up the 100%
aggregate. The temporal dimension concerned the times during the night of the election (8
November 2016), when key results were declared, influencing the changes shown in the
forecasted outcome at each point.
Figure 1.7 shows the same chart as before, with the same quantities plotted and with the same
design, but now reflecting the true context of the subject matter, as indicated by the updated
title, colour key and x-axis scales.
Figure 1.7 Forecasted % Chance of Winning Presidency (US Election, 8 November 2016)
Irrespective of where you sit politically, the revised context of the data portrayed in this chart will
unquestionably change how you feel about what you now see. It is no longer just a routine sales
chart restricted in relevance to a small group of people at Company X. It is now a visualisation
about a momentous event in modern history, the outcome of which most people on the planet
have some connection with or awareness of.
There are consequences of emotion to consuming this data. Some will relive the wild jubilation
of their candidate’s unexpected victory, others will recoil in horror at the memory of their
candidate’s unexpected defeat.
44

There are consequences of enlightenment. Some will be seeing these compelling patterns of ebb
and flow for the first time, others will at least recollect this roller-coaster story playing out via TV
or web coverage during the night itself.
There are also rational reactions. Consuming this chart now, many months or years later, offers
the opportunity for more considered analysis, in contrast to the original setting of this data being
consumed live across the USA and the rest of the world via a dynamically – and dramatically –
changing forecast tracker. In the cold light of day questions can be asked (and have been) about
the rigour of polling methods as well as the calculations used to create such forecasts. ‘How could
they be so wrong?’ some have asked, while others have countered with ‘How could they be
expected to be more right, it’s a complicated electoral system!?’
From your perspective as the visualiser, this final phase of understanding is something you will
have limited control over. Everything depends. It can be frustrating for people who are learning
visualisation and who just want the answer: ‘How do I deliver understanding to my audience?!’
In my experience, the factors that most influence the success of a visualisation are not technical,
they are contextual and, furthermore, human. Viewers are people. People are different, and
people are complex. They can be irrational and unpredictable, or impassive and disengaged. You
can lead a horse to water, but you cannot make it drink: you cannot force viewers to be interested
in reading your work, nor to understand the meaning of what you present, nor control how they
react to that experience. Even if your visualisation clearly shows action needs to be taken, you
cannot guarantee the viewers will recognise there is a need to act, will be in a position to act, and
indeed will know how to act.
It is at this point that we must recognise the ambitions and – more importantly – the limitations
of what data visualisation can deliver. Returning to the definition for a final time, the
illustrations we have gone through in this chapter support why the term facilitating is
realistically the most a visualiser can do. It might feel like a rather tepid duty, something of a cop-
out that abdicates responsibility for the outcome – why not aspire to achieve something more
concrete than ‘facilitate’?
I use facilitate because it gets to the heart of the tensions that visualisers face. There are times
when the onus is on us, and other times when the onus is on the viewer. Visualisation design
cannot change the world, it can only make it run a little smoother. Visualisers can control the
output but not the outcome; at best we can expect to have only some influence on it. The rest of
this book concerns how we optimise this influence.
45

1.2 Distinctions
Having delved into the proposed definition for data visualisation, it is now worth acknowledging
some other associated terms and disciplines that you may be familiar with or aware of.
The subtleties and semantics of defining fields are recurring concerns as new technologies
develop and creative techniques evolve. As participation has grown over the past decade, data
visualisation has been cross-pollinated with creative and analytical sensibilities arriving from
different origins. The traditional boundaries begin to blur and the practical value of preserving
dogmatic distinctions reduces accordingly. Ultimately, when one is tasked with creating a visual
portrayal of data, does it really matter if the creation is labelled and filed under ‘data visualisation’
or ‘infographic’ as long as it achieves the aim of helping the audience to achieve some form of
understanding?
However, subject distinctions do need to be understood. It is important for people to identify
with a particular discipline in which they have recognised expertise. It is therefore worth
clarifying some proposed distinctions, so, once again, we are on the same page of understanding.
Infographics: The classic distinction between infographics and data visualisation concerns the
format and the content. Infographics were traditionally created for print consumption, in
newspapers or magazines, for example. The best infographics explain things graphically –
systems, events, stories – and can often be generalised as explanation graphics. Infographics
contain charts (visualisation elements) but may also include illustrations, photo-imagery,
diagrams and text. These days, the art of infographic design continues to be produced for static
output – as opposed to interactive – irrespective of how and where the work is published.
Earlier this decade there was an explosion in different forms of infographics. From a purist
perspective, this wave of work was generally viewed as being an inferior form of infographic
design. These pieces were primarily driven by marketing desire for ‘clicks’, above any real desire
to facilitate understanding. If your motive is ‘bums on seats’ then I feel this is a different
endeavour to pure infographics and I would question the legitimacy of attaching the term
infographic to these designs; perhaps instead info-posters or tower graphics (they commonly
existed with a fixed-width dimension and huge length in order to be embedded into websites and
onto social media platforms) could be used. It is important not to dismiss entirely the evident – if
superficial – value of this type of work, as demonstrated by the occasional viral success story. But
I sense the popular interest in these forms has now waned and the authentic superior-quality
infographic has managed to rise back out of this noise.
Information visualisation: Smarter people than me use labels of data visualisation and
information visualisation interchangeably, without a great deal of thought for the relevant
differences. The general distinction tends to be shaped by one’s emphasis in focus towards either
the input material (data) or the nature of the output form (information). It is common for
information visualisation to be used as the term to define work that is primarily concerned with
visualising abstract data structures such as trees or graphs (networks) as well as other qualitative
data (therefore focusing more on relationships rather than quantities).
46

Information design: Information design is a design practice concerned with the presentation of
information. It is often associated with the activities of data visualisation; indeed sometimes it is
presented as the major field in which data visualisation belongs. Unquestionably, both share an
underlying motive to facilitate understanding. However, in my view, information design has a
much broader application concerned with the design of many different forms of visual
communication, particularly those with an instructional or functional slant, such as way-finding
devices like hospital building maps or in the design of utility bills.
Data journalism: Also known as data-driven journalism (DDJ), this concerns the increasingly
recognised importance of having numerical, data and computer skills in the journalism field. In a
sense it is an adaption of data visualisation but with unquestionably deeper roots in the
responsibilities of the reporter/journalist.
Visual analytics: Some people use this term to relate to analytical-style visualisation work, such
as dashboards, that serve the role of operational decision support systems or provide instruments
of business intelligence. The term is also used to describe the analytical reasoning and exploration
of data facilitated by interactive visual tools. This aligns with the role of exploratory data analysis
that I will be discussing in Chapter 4.
Data science: As a field, data science is hard to define, so it is easier to consider it through the
lens of a data scientist’s duties. Data scientists are somewhat unicorn-like in that they possess – or
are expected to possess – an almost preposterous repertoire of capabilities covering the gamut of
demands involved with gathering, handling, analysing and presenting data. Typically, the data
scientist works with data of large size and complexity. Data scientists have strong mathematical,
statistical and computer science skills, not to mention astute business experience, and are also
expected to possess so-called ‘softer’ abilities like problem solving, communication and
presentation.
Scientific visualisation: This is another form of a term used by many people for different
applications. Some label exploratory data analysis as scientific visualisation (drawing out the
scientific methods for analysing and reasoning about data). Others relate it to the use of
visualisation for conceiving highly complex and multivariate datasets specifically concerning
matters with a scientific bent (such as the modelling functions of the brain or molecular
structures).
Data art: Apart from the disputes over the merits of certain infographic work, data art is arguably
the other discipline related to visualisation that has historically stirred up the most debate. Again,
maybe it is reasonable to suggest the noise is quieter these days, but its sheer existence still
manages to wind up certain sections of the data visualisation illuminati. Data artists work with a
similar raw material in the form of data, but their goal is not driven by facilitating the kind of
understanding that a data visualisation would seek. Data art is more about pursuing a form of
self-expression or aesthetic exhibition using data as the paint and algorithms as the brush. As a
viewer, the meaning you draw from displays of data art are entirely down to the personal
interpretation it invites.
Dashboard: These are popular methods for displaying multiple visualisations and statistical
information. Dashboards often take the form of some organisational instrument that offers both
47

at-a-glance and detailed views of many different analytical and information dimensions.
Dashboards are not a unique chart type themselves, but rather should be considered
compositions that comprise multiple chart types.
Storytelling: This is an increasingly common term that is often misused and misunderstood,
which is quite understandable. Stories are usually constructed upon some notion of movement,
change or narrative. Charts showing trends or activities over a temporal plane or maps portraying
spatial relationships offer displays that are most consistent with the idea of a story. A bar chart
alone does not represent a story, in most people’s sense of the term, but if you show a pair of bar
charts to represent a before-and-after comparison, you have created a change dynamic.
Similarly, if you incorporate charts into some temporal presentation like a slideshow or video, the
chart becomes a prop and a narrator may draw out the story verbally. In this case it is the setting
and delivery that are consistent with the notion of storytelling, not the chart itself.
A further distinction to make is between stories that are explicitly communicated and stories that
form through interpretation. The famous six-word story For sale: baby shoes, never worn by Ernest
Hemingway is not presented as a story, rather the story is triggered in our mind when we read
this passage and start to infer meaning, implication and context. A story is being presented only if
it is accompanied by some explanation of the meaning of the data. Otherwise, any story derived
is what the viewers form themselves.
48

Summary: Defining Data Visualisation
In this chapter you have been introduced to the subject of data visualisation, learning a definition
that will shape much of the structure and content of this book:
The visual representation and presentation of data to facilitate understanding.
The different components that form this definition have been explained, with particular focus on
the nuances around facilitating understanding. The three distinct phases of understanding were
described:
Perceiving: what do I see?
Interpreting: what does it mean, given the subject?
Comprehending: what does it mean to me?
The second section explained some of the distinctions and overlaps with other related disciplines,
supplementing the glossary provided in the Introduction.
What now? Visit book.visualisingdata.com
EXPLORE THE FIELD Expand your knowledge and reinforce your learning about working with data through this
chapter’s library of further reading, references, and tutorials.
TRY THIS YOURSELF Revise, reflect, and refine your skill and understanding about the challenges of working
with data through these practical exercises.
SEE DATA VISUALISATION IN ACTION Get to grips with the nuances and intricacies of working with data in
the real world by working through this next instalment in the narrative case study and see an additional extended
example of data visualisation in practice. Follow along with Andy’s video diary of the process and get direct insight
into his thought processes, challenges, mistakes, and decisions along the way.
49

http://book.visualisingdata.com

2 The Visualisation Design Process
In this second chapter I will outline the data visualisation design process around which the
book’s chapters are arranged. You will learn why using a process approach is important to
organise and optimise your thinking – taking you from the initial spark of curiosity, through
wrangling with data, to juggling the myriad options that shape a design solution.
The process organises the activities into a sequence of manageable chunks so that the right things
are tackled in the right order. You cannot expect just to land on a great solution by chance if
your working practices are chaotic and confused. You will be aided by some additional practical
tips and good habits to employ across the whole process.
The quality of your decision making is the main difference between a visualisation that succeeds
and one that fails. To maximise the effectiveness of facilitating understanding for your audience,
the sectional parts of the chapter will introduce the three principles of good visualisation design.
50

2.1 Design Process: Organising Your Decision Making
For those new to the field, one of the first things to grasp is the idea that any notion of perfect in
data visualisation does not exist. It can prove simultaneously frustrating and liberating to learn
that there are good and bad solutions, but there are no perfect ones. To have perfect you need
immaculate conditions that are free of pressure, constraint or flaw. That is how things operate
now in real life. There will always be demands pushing and pulling you in different directions.
There will be shortcomings in the data that frustrate you or limitations in technical ability that
impede you. As described in Chapter 1, people, as recipients, introduce a diversity of need that
realistically cannot always be fulfilled. Recognising that perfect is unobtainable helps unburden
us from a nagging sense that somehow we might have missed finding the perfect solution. There
will never be just one single possible solution to a problem.
The central premise in this book is that decision making is the key competency in data
visualisation: namely, effective decisions, efficiently made. To accomplish this you need to follow
a design process that organises your thinking and is underpinned by robust principles to optimise
your thinking.
We will discuss principles shortly, but firstly let’s look briefly at the design process overall (Figure
2.1).
Figure 2.1 The Four Stages of the Data Visualisation Design Process
Across the four stages that make up this process there are two main phases. The first three stages,
presented in Part B of this book through Chapters 3 to 5, involve activities that I describe as
concerning the ‘hidden thinking’ of data visualisation. These stages cover the preparatory work
that informs what you are visualising, for whom and, crucially, why:
1. Formulating your brief: planning, defining and initiating your project.
2. Working with data: gathering, handling and preparing your data.
3. Establishing your editorial thinking: defining what you will show your audience.
The second main phase of the process sits entirely with stage 4 and this involves developing your
design solution, the visual manifestation of the preparatory work you have conducted. This stage
is concerned with the how.
The five distinct design layers that make up the anatomy of any visualisation solution – data
representation, interactivity, annotation, colour and composition – are covered in Part C of this
book, in Chapters 6 to 10 respectively. As explained earlier, a detailed treatment of technical
activities is beyond the scope of this text.
I am not going to describe these process stages in more depth here – the next eight chapters exist
51

to do that. Instead, here are some observations about why it is important to follow a design
process.
Reducing the randomness of your approach: The value of this design process is that it shapes
your entry and closing points. How do you start a process? How do you know when you have
finished? As I have mentioned, the sheer extent of things you will have to think about, even with
simple projects, can be quite an overwhelming prospect. This approach breaks down key stages
into a connected system of thinking that will help progress your work and preserve cohesion
between your activities. It incrementally leads you towards developing a solution, with each stage
building on the previous one and informing the next.
‘I tend to keep referring back to the original brief (even if it’s a brief I’ve made myself) to keep checking that the
concepts I’m creating tick all the right boxes. Or sometimes I get excited about an idea but if I talk about it to friends
and it’s hard to describe effectively then I know that the concept isn’t clear enough. Sometimes just sleeping on it is all
it takes to separate the good from the bad! Having an established workflow is important to me, as it helps me cover all
the bases of a project and feel confident that my concept has a sound logic.’ Stefanie Posavec, Information Designer
Every project is different: Every visualisation presents new challenges. Even if you are just re-
producing the same report every month, no two instances of that report will involve the exact
same context. Just by having one extra month of data, for example, may expose you to larger
values, smaller values, new values and expired values. Whether you have simple data, or vast
amounts of complex data, two hours or two months, the process you follow will always be the
same. You should follow the same sequence of thinking regardless of the size, speed and
complexity of your challenge. The main dif-ference is that any extremes in the circum- stances
you face will amplify the stresses at each stage of the process and place greater demands on the
need for thorough, effective and timely decision making.
Adaptability: The term process contrasts considerably with procedure. The process outlined in this
book provides a framework for thinking, rather than instructions to learn and follow. A good
process should offer adaptability and remove the inflexibility of a defined procedure. In any
visualisation project, you will need to respond to revised requirements, additional data that
emerges, or a shift in creative direction. A good process safeguards adaptability and cushions the
impact of changing circumstances like these. Although the activities presented in this book are in
a linear arrangement, there will always need to be room for iteration. There will be plenty of
occasions when you have to revisit decisions or redo activities in a different way, especially if you
make mistakes. What is more important in these situations is how gracefully you fail and how
quickly you recover.
Protect experimentation: The process approach I am advocating is not overly systematic and
does not compromise on allowing space for experimentation. When there are pressures on time,
the need to focus and avoid distraction is understandable. Aspiring to reduce wasted effort and
improve efficiency is entirely reasonable, but one must still seek out opportunities – in the right
circumstances – for imagination to blossom. In reality, few projects will offer too much scope for
far-reaching creative exploration, but when an opportunity presents itself for you to work on a
subject that befits creativity, you should embrace it. And do not forget to enjoy it!
‘I truly feel that experimentation (even for the sake of experimentation) is important, and I would strongly encourage
it. There are infinite possibilities in diagramming and visual communication, so we have much to explore yet. I think
52

a good rule of thumb is to never allow your design or implementation to obscure the reader understanding the central
point of your piece. However, I’d even be willing to forsake this, at times, to allow for innovation and
experimentation. It ends up moving us all forward, in some way or another.’ Kennedy Elliott, Graphics Editor,
National Geographic
The first occasion, not the last: Each activity you commence across the distinct stages in the
process will likely represent the first occasion you pay attention to these matters, but not the final
occasion. Think of the sequencing as being akin to a trickle-down effect. Take, for instance, the
recurring concern about thinking about your audience. You will first encounter the need to
define a profile of your anticipated audience’s characteristics during the first stage of the process,
‘Formulating your brief’. However, the concern about what they know, what they need to know,
and how interested they will be will reoccur right through to the end. Concerns like these should
never drop off your radar. The list of concerns will only build, but the intention is that the
process gives you the best chance of keeping all the necessary plates spinning for as long as they
need to be.
Across the book there are frequent vignettes of advice and useful tips for you to adopt to get the
most out of working through this process. These are informed by interviews with people working
in the field, as well as from my own practical experiences, and are provided with each topic in the
book. There are some recommended habits that are applicable to all stages in this process,
relevant to novices or experienced visualisers alike, as follows.
Time management: Any creative work quickly swallows up all the available time. You get
tempted to try things, to explore different ideas, to attempt one final pass at seeking out
interesting features of your data. It is easy to be consumed by the stretching demands of the
activities across this process. As you then reach a deadline you either sink or swim: for some the
pressure of the clock ticking is crippling, especially impacting their creative thinking; others
thrive on the adrenaline it stirs, sharpening their focus as a result. Regardless of how you respond
to looming deadlines, good planning is vital.
Time management is the essence of good planning. It keeps a process cohesive and on track.
From experience working on different projects your ability to anticipate how much time to
allocate to different activities will improve. That said, each project introduces its own profile of
demands, so always find time before you set off to estimate where your likely commitments will
be most required. Do not forget to factor-in time for easily neglected responsibilities, such as
supervisor meetings, Skype calls, research and file management.
Mindsets: Irrespective of the type of visualisation you are working on, your process will involve a
mixture of conceptual and practical activities. Sometimes these will be allocated across a team,
exploiting the range of talents at different times through the process. On other occasions you will
be working alone, and the diversity of these activities will stretch your mind considerably.
Sometimes you are thinking, sometimes you are creating; sometimes you need to be creative,
sometimes you need to have an eye for detail.
‘You need a design eye to design, and a non- designer eye to feel what you designed. As Paul Klee said, “See with one
eye, feel with the other.”’ Oliver Reichenstein, Founder of Information Architects (iA)
Thinking: The duties here will be conceptual in nature, requiring imagination and
53

judgement, such as formulating your curiosity, defining your audience’s needs, reasoning
your editorial perspectives, and making decisions about viable design choices.
Doing: These are active duties that engage the brain through more practical undertakings,
such as sketching ideas, conducting research, holding discussions with a client, or checking
data.
Making: These are more hands-on constructive duties characterised by using tools for
activities like handling data, creating charts, and designing presentation features.
For the scope of this book, the focus is largely on thinking. I find the notion of brain ‘states’
relevant here, especially the ‘alpha’ state. This is the state our mind is in, most commonly, when
we feel especially relaxed. Occupying this state helps heighten your imagination and thought
process. I find I do some of my most astute thinking in the shower or just before going to sleep at
night. These are the occasions when I am most likely drifting into a relaxed state. I find the same
conditions when undertaking long train journeys or flights. I use it to help contemplate the
progress I am making on a task. It lets me escape the noise present when doing more practical
tasks.
Documenting: It is mawkish to claim the humble pen and paper are the most important tools
for visualisers. After all, unless you are producing artisan hand-drawn work, technical
applications will be more applicable for most of your process. However, pen and paper will prove
to be a real ally to help you document thoughts and capture sketches. Do not rely on your
memory; if you have a great idea, sketch it down. You do not need great artistry, you just need to
get things out of your head and onto paper, particularly if you are collaborating with others. If
you are fortunate to be fluent with a tool and find it more natural to use that for ‘sketching’ ideas
than pen and paper, then this is absolutely fine, as long as it is the quickest medium to do so.
Whether using pen and paper, or a tool like Word or Google Docs, note-taking is a useful habit
to develop. It helps you document important details such as:
task lists with details of deadlines and precedents;
information about the sources of data you are using;
details of complicated calculations or manipulations you have applied to your data;
a log of any assumptions you have made;
terminology, abbreviations, acronyms – technical properties of your data that are crucial to
its understanding;
questions and answers you have received or are yet to;
issues or problems you have experienced or can foresee;
wish lists of features or ideas you would like to explore;
sources of inspiration, like websites or magazines you discover;
ideas you have had or rejected.
Note-taking is more easily preached about than done. I am the least competent of note-takers,
but I have found a way to make it a forced habit and it does prove valuable.
Communication: Communication is a two-way activity. It is about listening to stakeholders and
to your audience: what do they want, what do they expect, what ideas do they have? In
particular, what knowledge do they have about your subject? Communication is about speaking
54

to others: presenting ideas, updat- ing on progress, seeking feedback, sharing your thoughts
about possible solutions, and promoting and selling your work (regardless of the setting, you will
need to do this). You cannot avoid the demands of communicating, so do not hide behind your
laptop – get out there and interact with people who can help or whom you can help.
‘Because I speak the language of data, I can talk pretty efficiently with the experts who made it. It doesn’t take them
long, even if the subject is new to me, for them to tell me any important caveats or trends. I also think that’s because I
approach that conversation as a journalist, where I’m mostly there to listen. I find if you listen, people talk. (It sounds
so obvious, but it is so important.) I find if you ask an insightful question, something that makes them say “oh, that’s
a good point,” the whole conversation opens up. Now you’re both on the same side, trying to get this great data to the
public in an understandable way.’ Katie Peek, Visualisation Designer and Science Journalist
Associated with the need for good communication skills is the importance of research. You
cannot know everything about your subject, about the meaning of your data, about the relevant
and irrelevant qualities it possesses. As you will see later, data itself can only tell us so much; often
it just tells us where interesting things might exist, not what actually explains why they are
interesting. Talk to smart people who know a subject better than you or people who do not
know the subject but are just smart.
‘Research is key. Data, without interpretation, is just a jumble of words and numbers – out of context and devoid of
meaning. If done well, research not only provides a solid foundation upon which to build your graphic/visualisation,
but also acts as a source of inspiration and a guidebook for creativity. A good researcher must be a team player with
the ability to think critically, analytically, and creatively. They should be a proactive problem solver, identifying
potential pitfalls and providing various roadmaps for overcoming them. In short, their inclusion should amplify, not
restrain, the talents of others.’ Amanda Hobbs, Researcher and Visual Content Editor
Attention to detail: The process you follow embodies the concept of the ‘aggregation of
marginal gains’. You need to sweat the small stuff. Even if many of your decisions seem tiny and
inconsequential, they deserve your full attention. Like note-taking, the importance of checking
every detail may not be a natural trait for some. However, errors found in your work can be
damaging and will certainly undermine your audience’s trust, as you will learn about shortly. I
know through experience how one mistake can undermine the integrity of an entire project, even
if this feels unfair and disproportionate considering everything that was correct. Start every
project with a commitment to eliminate mistakes and learn from the pain when you fail. It is not
easy: I have no doubt I will leave at least one mistake in this book and it will haunt me. It can
help, if you are so immersed in your own work and become blind to it, to seek others to help
you.
‘Kill your darlings’: A recurring consequence of facing so many decisions in visualisation is the
need to demonstrate the discipline of not doing something. It is easy to applaud oneself over
brilliant ideas, but occasionally these ideas you have invested in deeply just will not work out.
Even though you have invested heavily in time and emotional energy, do not be stubborn. When
something is not working, learn to kill it. Otherwise, such preciousness will impede the quality of
your work. Being blind to things that are not working, or ignoring constructive feedback from
others, will prove destructive.
Learn: Reflective learning is about looking back over your work, examining the output and
evaluating your approach. What did you do well? What would you do differently? How well did
you manage your time? Did you make the best decisions you could, given the constraints that
55

existed? Learn from others. Read how other people undertake their visualisation challenges.
Maybe share your own? You will find you truly learn about something when you find the space
to write about it and explain it to others. Write up your projects, present your work to others
and, in doing so, this will force you to think ‘why did I do what I did?’
Also use reflective learning to find your process. What is presented in this book is proposed, not
imposed. If you cannot get this approach to fit your personality, your project’s purpose, or the
rhythm of how you need to work with others, modify it. We are all different. Take this as a
recommended framework but then bend it, stretch it and make it work for you. As you become
more experienced (and confident through having been exposed to different challenges) the
activities involved in data visualisation design will become second nature. You will probably
become blissfully unaware of even observing a process.
56

2.2 Design Principles: Optimising Your Decision Making
If the goal of data visualisation, as defined in the first chapter, is to facilitate understanding, all
judgements made through the design process have to contribute to accomplishing this.
Most choices are relatively clear cut and basing your judgement on common sense, informed by
the first three preparatory stages, will be entirely reasonable. However, for more nuanced
situations, when there might be several complex options presenting themselves, you will face a
dilemma. Making a choice will need more than just common sense. This is when it helps to
consult a framework of design principles.
‘I say begin by learning about data visualisation’s “black and whites”, the rules, then start looking for the greys. It
really then becomes quite a personal journey of developing your conviction.’ Jorge Camoes, Data Visualisation
Consultant
In data visualisation there are relatively few universal rules to follow. There are evidence-based,
useful suggestions that nudge you towards ‘always do this’ and ‘never do that’, but even they are
exposed to legitimate breaking point. This is because each decision that needs to be made is
accompanied by many contextual dependencies.
The principles that inform my own visualisation design convictions originated from beyond the
boundaries of this subject. Dieter Rams was a German industrial and product designer most
famously associated with the Braun company. Around the late 1970s and early 1980s, he was
becoming concerned about the state and direction of design thinking and, given his prominent
role in the industry, felt a responsibility to challenge himself, his own work and his own
thinking. He posed the simple question: ‘Is my design good design?’ By dissecting his work in
response to this question he conceived ten principles that symbolised the important aspects of
what he considered to be good design (Figure 2.2).
Figure 2.2 Dieter Rams’ ‘Ten Principles of Good Design’
In ‘De architectura’, a thesis on architecture written around 15 BC by Marcus Vitruvius Pollio, a
Roman architect, the author declares that the essence of quality in architecture is framed by the
social relevance of the work, not the eventual form or workmanship towards that form. He states
that good architecture can only be measured according to the value it brings to the people who
use it. In a 1624 translation of the work, Sir Henry Wotton offers a paraphrased version of one
of Vitruvius’ most enduring notions that a ‘well building hath three conditions: firmness,
commodity, and delight’. An updated translation of this would read as ‘sturdy, useful and
beautiful’.
57

Collectively, these separate sources have informed the three key principles shown in Figure 2.3
that I believe apply to any judgement of effectiveness in data visualisation.
Figure 2.3 The Three Principles of Good Visualisation Design
As you go through the process, these principles will guide the choices you make. Let’s look at
them in more detail, framing them in relation to Rams’ ten principles of good design as well as
Vitruvius’ three.
58

Principle 1: Good Visualisation Design Is Trustworthy
This first principle is to make your visualisation trustworthy, which maps directly onto one of
Dieter Rams’ ten principles (Figure 2.4) and embodies Vitruvius’ desire for sturdy. This is about
reliability and is achieved by securing and sustaining the trust of your audience.
Figure 2.4 Mapping ‘Trustworthiness’ onto Rams’ Ten Principles
This is presented as the first of the three principles because it is about the fundamental legitimacy
of a data visualisation. Without trust, the opportunity to facilitate understanding vanishes. You
can disregard any value from achieving ‘accessible’ and ‘elegant’ design; if you lose trust, you lose
your audience. Game over.
‘Good design is honest. It does not make a product appear more innovative, powerful or valuable than it really is. It
does not attempt to manipulate the consumer with promises that cannot be kept.’ Dieter Rams
There is an important distinction to make about the relationship between trust and truth.
Achieving trust is an aim, presenting truth is an obligation. There should be no compromise
here. You should never create work you know to be misleading, through either its content or its
representation. You should never claim something presents the truth if it cannot be reasonably
supported. The difference between a truth and an untruth should be beyond dispute. The fact
that it is not, these days, is a sad indictment of modern society. Nevertheless, the imperative for
truthfulness must be clear.
The difficulty for a visualiser comes when there are potentially multiple different, but legitimate,
versions of a ‘truth’ within the same data or subject context. This muddies things somewhat. A
glass that is half full is also half empty. Both viewpoints are objectively truthful, but the one you
choose to focus on is subjective. In data visualisation there is rarely a single view of the truth. You
make the call about what is most relevant in the context of your work. This is something we will
explore in Chapter 5 about editorial thinking.
Even though the choice made might be impartial, it is the act of choosing that creates an
unavoidable form of bias. When you choose to do one thing you are usually choosing not to do
something else. Deciding to show the forecasts for a political party winning an election changing
over time using a line chart might also incorporate a decision not to show how the forecast looks
geographically, assuming the data was available. Any visualisation will be an aggregation of
decisions among which many are shaped by reasonable subjectivity. No visualisation is purely
objective, even if it seems to portray this quality.
Rather than get consumed by the inevitability of biases rippling through your work, and perhaps
seeing this as good reason not to undertake it, your focus is better directed towards ensuring your
59

chosen path is trustworthy. In the absence of a single objective truth, what can you do to secure
trust in your subjectively selected truth?
Trust must be earned, but it is hard to secure and easy to lose. As the translation of a Dutch
proverb states, ‘trust arrives on foot and leaves on horseback’. Trust is something a visualiser
must try to nurture through accuracy and transparency, eliminating doubts or legitimate dispute
from a viewer. Easier said than done, though, as visualisers have only a certain amount of control
over this because our audience is people. Yes, them again. A visualisation can be truthful but
viewed, unreasonably, as being untrustworthy. Conversely, a visualisation that might not be
truthful might still be trusted (perhaps a more dangerous outcome that opens a separate
discussion, but one for other books to tackle). Neither of these are satisfactory experiences: the
latter we can and should avoid; the former is something we strive to overturn.
Let’s consider an example that illustrates the fragility of trust. In Figure 2.5, the chart shown
plots the number of murders committed using firearms in Florida over a given period of time.
The data is framed around the enactment of the ‘Stand Your Ground’ law in 2005. The chart
uses an inverted vertical y-axis with a red-filled area occupying the space beneath the x-axis
baseline, growing downwards as the number of deaths increases. Some of the peak values are at
1990 and 2007.
Figure 2.5 Gun Deaths in Florida (Reuters Graphics)
When this piece was published, many commentators hastily cried foul, remarking on how the
inversion of the y-axis had deceived them. They had mistaken the red area as the background and
60

saw the data as formed by the ‘white mountain’ emerging in the foreground. In misreading the
chart, they were instead seeing peak values as being those for 1999 and 2005, the highest points
of this apparent white mountain. This illusion is caused by an effect known as figure–ground
perception whereby a background form (the white area) can become inadvertently recognised as
the foreground form, and vice versa with the red area seen as the background. Despite eventually
reading the chart and being able to discover the correct view of the chart, for many viewers, any
trust had been lost: they felt they had been tricked. An accusation exacerbated, no doubt, by the
emotive nature of the subject: any chart about gun crimes will stir passions regardless of its form.
Figure 2.6 Iraq’s Bloody Toll, by Simon Scarr (South China Morning Post)
Although the approach to inverting the y-axis may not be entirely conventional, it was a
legitimate approach. The problem was arguably caused by the visually prominent gridline for
1000 which inadvertently framed the ‘white mountain’ but, creatively speaking, attempting to
convey the effect of dribbling blood was a plausible metaphor to pursue. It was emulating a
prominent and celebrated visualisation work, from several years ago, showing the death toll
during the Iraq conflict (Figure 2.6). Being inspired and influenced by the techniques
demonstrated by other visualisers is something to be encouraged as an important way of
furthering our skills.
The key point is that there was no intention to mislead. The lack of trust expressed by some was
61

the consequence of a well-intended set of design decisions. It demonstrates how vulnerable trust
is, especially in the pressured environment of a newsroom where the output of work is relentless
and there is often only a single opportunity to publish a given piece to a huge, widespread
audience. Even in this era of largely digital media platforms, it is hard to intercept, withdraw and
revise work. ‘You don’t get a second chance to make a first impression’, as the saying goes.
Is the Handling of the Data Reasonable and Faithful to the
Subject?
Trustworthiness is a cause that should guide all your decisions, not just those that emerge in the
design-focused final stage of the process. The principles are framed as design principles because it
is through your design work that all your decisions will visually materialise. Earning trust is
something that reaches right the way back to the earliest preparatory task.
‘Data and data sets are not objective; they are creations of human design. Hidden biases in both the collection and
analysis stages present considerable risks [in terms of inference].’ Kate Crawford, Principal Researcher at Microsoft
Research NYC
It is during the first stage that the initial seeds are sewn for how you might creatively handle your
subject matter. As you have just witnessed, if you are working on potentially emotive topics, this
will only heighten the potential exposure to prejudgement and opinion. As you will learn in
Chapter 3, when considering your subject matter and establishing your ideas about the purpose
of your work, there will be some contexts that lend themselves to exploiting the emotive qualities
of your subject but others that will not. Trust will be jeopardised if you have misjudged the tone
of voice.
Working with data, the second stage of the process, is arguably where trust is most at stake. You
are the custodian with a responsibility for being faithful to the data you have and the subject it
embodies. You need to be careful with your handling of the data and transparent with what you
decide to do with it. There are critical questions you may need to answer to ensure your
approach to handling the data is reasonable, such as the following:
How was the data collected: from where and using what method?
What calculations or modifications have you applied?
Have you made significant assumptions or applied any specific counting rules?
What criteria were used for the data values you decided to include and exclude from the
display?
Does the Representation and Presentation Design Have Integrity?
A fundamental tenet of data visualisation is never to deceive the receiver. Avoiding possible
misunderstandings, inaccuracies, confusions and distortions is of primary concern when thinking
about the integrity of how your work is both represented and presented. There are many possible
features of visualisation design that can lead to varying degrees of mistrust, whether intended or
not, as exhibited in the previous example; I will explore many of these further later in the book:
62

Sometimes charts are used in ways that effectively corrupt their usage. An example would be
using pie charts to display percentage parts that exceed 100%.
Unreliable functional experiences with interactive projects. Does the solution work and,
specifically, does it work in the way it promises or is expected to do (such as the speed of
loading)?
Missing annotations like clear titles, introductions, axis titles, labels, footnotes and data
sources. All these features help a viewer understand what he or she is consuming; when they
are missing it can lead to confusion and suspicion.
Mistakes made with any statistics or captions presented will tarnish the perceived accuracy
of the entire work.
The quantitative axis of a bar chart should not be ‘truncated’. That is, the baseline origin
value should be zero, otherwise the resulting display will distort the perceived bar size
judgements.
The size of geometric areas, such as circle sizes, can sometimes be miscalculated by using
diameter as the basis of size variation rather than shape area. This results in the quantitative
values being disproportionately sized.
When a chart based on two dimensions of data is presented in 3D form but consumed in
2D format (such as a static display on a screen or in print), this decorative design choice
distorts the perceived value sizes – you cannot adjust your viewpoint to accommodate
perspective, process distance, or see obstructed features. Thus 3D should only be considered
when there are dynamic means for a viewer to change his or her viewpoint in order to
navigate around a 3D form and see it from bespoke 2D perspectives.
The aspect ratio (height vs width) of a line chart’s display can affect the perceived steepness
of connecting lines which reveal the trends of a continuous series of values over time. If the
chart area is too narrow the steepness will be embellished; too wide and the steepness is
dampened.
When portraying spatial analysis through a thematic map, there are different mapping
projections which translate a spherical globe into a flattened 2D map. The mathematical
treatment applied to this translation can significantly alter the perceived size or shape of
regions, potentially distorting their perception. More on this in Chapter 10.
The size or sequencing in the layout of a work might raise suspicions if seemingly important
contents are diminished in the visual hierarchy, such as pushed to the bottom or shrunk in
size.
The examples in Figures 2.7 and 2.8 display two charts showing the same analysis but presented
differently. Take a moment to reflect on how much trust you feel is achieved by these respective
pieces. For context, both were extracted from articles discussing issues about home ownership, so
they would normally be presented alongside written analysis in their original published form.
Figure 2.7 Housing and Home Ownership in the UK (ONS Digital Content Team)
63

Figure 2.8 Falling Number of Young Homeowners (Daily Mail)
As I have said, both charts use the same data and use the same chart type to represent the
analysis; they even arrive at the same summary finding. However, in my view, the first chart,
produced by the UK Office for National Statistics (ONS), commands greater credibility and
authority, and therefore far more of my trust, than the second visualisation, produced by the
Daily Mail.
The first reason for this opinion concerns how the ONS piece is more informative and
transparent about the data and subject. Whereas the Daily Mail piece refers to the ONS as the
source of the data, it fails to include any more details about the data source, information which is
included on the ONS graphic alongside other features like the subtitle, an explanation about the
yearly periods. colour choices for the bars. The option to see and download the associated data is
unique to the ONS chart as it was published on the Web, so it is unfair to contrast the absence of
this feature on the Daily Mail graphic.
64

The second reason is more instinctive and influenced by my personal taste. The colours used in
the ONS graphic are reserved but aesthetically engaging and convey a certain assuredness. By
contrast, the Daily Mail’s colour palette feels needy. It seems to be craving attention with sickly
sweet coloured sticks. The house key image in the background is visually harmless but feels lazy
and derivative. The typeface, font size and colouring of the text feel cheap. The ONS text feels
polite and conveys authoritativeness. These presentational features are stylistic and therefore
more weighted towards being matters of pursuing ‘elegance’ in your design, as we will describe
shortly, but the three principles are unquestionably interconnected in places.
Overall, my opinion about the level of trust I hold for these pieces is partially reasonable and
partially irrational. Crucially, it is also greatly influenced by the prejudices I bring to the
encounter. I do not trust the Daily Mail as a source of any information, whereas I do trust the
ONS. It is hard to undo those feelings and biases that we bring to the viewing process. The
platform and location in which your work is published (e.g. website or source location) will prove
influential: visualisations encountered in already-distrusted media will create obstacles that are
hard to overcome.
65

Principle 2: Good Visualisation Design Is Accessible
This second principle is to make your visualisation accessible, which maps onto three of Dieter
Rams’ ten principles of good design (Figure 2.9) as well as Vitruvius’ desire for useful.
Figure 2.9 Mapping ‘Accessibility’ onto Rams’ Ten Principles
Accessibility in visualisation design is concerned with giving your audience access to useful
understanding. It must be relevant to the subject and relevant to their needs. This needs to be
achieved in a way that does not require undue effort to perceive, interpret and comprehend.
Is the Portrayal of the Data and the Subject Relevant?
The first aspect of accessibility concerns the relevance of the visualisation you are portraying to
your audience. Judging relevance is a subjective and contextually driven matter relating to the
potential usefulness of your visualisation: am I providing my audience with access to the most
useful understanding about this subject? Relevance is a somewhat shifting concept that is, in part,
based on qualities such as interestingness and pertinence.
It is also, fundamentally, shaped by what you actually have available to present to your audience.
You might create the most beautifully designed visualisation, but if nobody finds the specific
analysis you choose to portray about a subject relevant, any motivation your audience had to
engage with your work may be undermined.
‘The key difference I think in producing data visualisation/infographics in the service of journalism versus other
contexts (like art) is that there is always an underlying, ultimate goal: to be useful. Not just beautiful or efficient –
although something can (and should!) be all of those things. But journalism presents a certain set of constraints. A
journalist has to always ask the question: How can I make this more useful? How can what I am creating help
someone, teach someone, show someone something new?’ Lena Groeger, Science Journalist, Designer and
Developer at ProPublica
Imagine you are visiting a city for the first time and you ask a passer-by for help with directions
to the main railway station. Unfortunately, they cannot guide you to the station, but they do
know how to get to the main library. In these circumstances, what was useful for you to learn was
not matched by what was possible for the passer-by to impart. Directions to the library do not
give you access to the understanding you actually need and is therefore irrelevant. However,
suppose you later discover that the railway station is across the road from the library; the
information available about how to get to the library is now instantly promoted to being relevant.
It is just a shame it is too late.
Any judgement of relevance will also be determined by the level of content sophistication. In the
66

Introduction’s glossary of terms, you will have seen the distinction between terms like complex
and simple. As a visualiser, you might misjudge the level of sophistication required by your
audience and oversimplify a complex subject. Maybe there were lots of interesting perspectives
about the subject that you could and should have included, but instead you just presented one
simplified chart and that might disguise many of the key nuances about the subject. Equally,
maybe you made a great effort to include multiple related chart views that provide a wide
coverage of different aspects of your subject, when what your audience need is just a simple chart
and some quick headline insights. In each case you have failed to provide suitable access to the
relevant content.
This is evidently a hard thing to judge, especially when you have a varied audience with diverse
interests. You can only do so much, and you certainly should not expect to get it right for every
potential viewer. But it is a crucial matter to care about. Most of the topics and activities covered
in Chapters 3, 4 and 5 are concerned with guiding you towards a reasonable judgement of
relevance.
Is the Representation and Presentation Design Suitably
Understandable?
In contrast to relevance, the suitability of design looks at usefulness from a different perspective.
This is less about the consequence of a visualisation’s use and more about understanding how to
use it.
Accessibility in design is fulfilled by removing any design- and content-related obstructions faced
by your viewers. Expressed another way, thinking of the opposite of accessible (confusing), can
you ensure a viewer avoids a confusing experience with your work? Confusion is the friction
between the act of understanding (effort) and the achieving of understanding (reward).
What constitutes minimum friction is shaped, inevitably, by context and the characteristics of the
audience, which makes the notion of accessibility a somewhat variable concept. It is not always
possible to eliminate friction, which is why a judgement of ‘suitable’ fric-tion is a pragmatic
perspective to take. The efforts need to feel proportional to the rewards on offer. Not every
process of under-standing can or should be quick and simple, but to achieve accessibility means
to elimi-nate unnecessary delays to this process.
Demonstrating empathy for your audience, appreciating the setting in which they encounter
your work and how they need to use it are at the heart of accessible design thinking. Here are
some of the most crucial factors.
‘We should pay as much attention to understanding the project’s goal in relation to its audience. This involves
understanding principles of perception and cognition in addition to other relevant factors, such as culture and
education levels, for example. More importantly, it means carefully matching the tasks in the representation to our
audience’s needs, expectations, expertise, etc. Visualizations are human-centred projects, in that they are not universal
and will not be effective for all humans uniformly. As producers of visualizations, whether devised for data exploration
or communication of information, we need to take into careful consideration those on the other side of the equation,
and who will face the challenges of decoding our representations.’ Isabel Meirelles, Professor, OCAD University
(Toronto)
67

Understanding a subject: What your audience know and do not know about a subject will have
a significant bearing on the degree to which they consider a visualisation to be accessible. What is
considered inaccessible to one audience group could be fully accessible to another. Reading a
street sign written in Japanese is entirely inaccessible to non-Japanese speakers, yet it would be
fully accessible to someone who knows the language. If that street sign is encountered in Tokyo it
is contextually appropriate, but in Newcastle upon Tyne it would not be.
As I demonstrated in the first chapter when interpreting the charts about football and ‘Winglets
and Spungles’, if you do not understand a subject, this instantly raises the chances of confusion
through ignorance. Interpretation is prevented.
Though existing knowledge is important, the principal property of subject matter that most
influences accessibility is the intellectual level it embodies. In other words, is it a complicated,
complex or simple subject for anyone to grasp? This leads us again into a discussion about the
semantics of language, but these differences are crucial in how we make judgements about the
suitability of accessibility:
Complicated relates to subject knowledge or a skill that is typically intricately technical,
probably unique and difficult to understand. It requires a certain level of intellect or
inherent talent to do so. The mathematics that underpinned the Moon landings is
complicated. The inner workings of a boiler are complicated. Making Baked Alaska
(successfully) is complicated. The knowledge or skill in question is acquirable and the
learning involved surmountable, if steep, but only achieved through lots of time, hard work
and, usually, with assistance from external expertise.
Complex is associated with systems or contexts that have no perfect conclusion or even no
end state. Managing relationships is complex. Same with parenting: there are books and
people offering advice but there is no rulebook for how to do it well all the time – no
definitive way of accomplishing it. The elements of parenting might not be necessarily
complicated – such as remembering to cut the crusts off Sergio’s sandwiches to avoid a
tantrum – but the interrelated natures of events and pressures are shaping and colliding to
make it feel very hard to master.
Simple, for the purpose of this book, concerns a matter that is inherently easy to understand.
It may be small in dimension and scope, meaning there is not a lot of knowledge to acquire
and it is unlikely to require lots of practice to sustain, irrespective of prior experience. It is
also quite isolated in that it does not have other interconnections affecting its state.
When working with a complex or complicated subject, your instinct might be to seek to simplify
it. Simplifying is a reductive process that translates a complex or complicated state into a
simplified form, usually by eliminating details or nuance. There are situations that will warrant
making the process of understanding quicker and easier, though this is not a universal goal.
‘Strive for clarity, not simplicity. It’s easy to “dumb something down,” but extremely difficult to provide clarity while
maintaining complexity. I hate the word “simplify.” In many ways, as a researcher, it is the bane of my existence. I
much prefer “explain,” “clarify,” or “synthesize.” If you take the complexity out of a topic, you degrade its existence
and malign its importance. Words are not your enemy. Complex thoughts are not your enemy. Confusion is. Don’t
confuse your audience. Don’t talk down to them, don’t mislead them, and certainly don’t lie to them.’ Amanda
Hobbs, Researcher and Visual Content Editor
68

Not everything can or should be simple. The process of simplification might risk the subject
being oversimplified to the point of obscurity. In removing important subtleties and
technicalities this can be just as detrimental to the perceived accessibility as leaving a complex or
complicated subject too intellectually demanding. What if your audience are sufficiently
sophisticated with the capability and motivation to handle the learning process required in
grasping a hard topic? By simplifying things, they would be denied that learning opportunity and
denied access to relevant understanding. An audience in this case may justifiably feel patronised
when faced with an oversimplified portrayal.
When considering the level of your subject matter and the nature of your analysis, if you do not
think your audience will understand what you are presenting, you have a choice: to simplify or
clarify.
Simplify when your audience do not have the knowledge or capacity to handle a
complicated subject and do not need to acquire deep understanding about it.
Clarify when your audience do not have the knowledge but do have the capacity to handle a
complicated subject, with assistance. Provide features of annotations to explain what the
subject is about, how to read it, what features are significant, and what it all means. Do not
underestimate the capacity of your audience to be willing and able to grasp topics they have
no prior understanding about. Often, this is what they want out of a visualisation
experience.
A further risk to creating confusion is through carelessness and complacency: do not assume
domain expertise among all your audience with the use of ambiguous acronyms, abbreviations or
technical language. Explain what needs to be explained. Include annotations for titles, scales and
units, explaining data sources and colour associations. These are all features that contribute a
great deal towards eliminating confusion.
Representation design: As well as subject complexity, another issue to consider is representation
complexity. That is, the perceived complexity of uncommon or unfamiliar chart types. Not every
chart we encounter is familiar. And when something is not familiar, it is understandable how this
exposes a risk of confusion.
Sometimes, the lack of familiarity is a trigger to blame the visualiser: ‘Why did they use a chart
I’ve never seen before?’ This deficit in knowing how to read a new or unfamiliar chart type is not
a failing on the part of the viewer, it is simply the viewer’s lack of prior exposure to these
different methods. But what if there was good reason to use that chart? It might be the most
astute way to portray the most relevant analysis about a subject. Sure, it might be visually
complicated, but this might be the only way to show it.
Everything is new once. This is why we learn and why we are capable of learning. To overcome
chart types that are unfamiliar a viewer must be provided with the means to learn how to
perceive and interpret a chart. This prospect can naturally frustrate some people who see it as a
critical obstacle they are unwilling to entertain. Even with the best intent and the provision of
helpful guidance, if a viewer is simply unwilling to make the effort, you have little further
influence in overcoming this blockage.
Time: This concerns the characteristics of the encounter and the duration of time or attention
69

viewers might have available to process understanding. You need to consider if, at the point of
consuming a visualisation, the viewers are in a pressured situation. Are they in a rush? Do they
need quick insights or is there scope for a more prolonged engagement? Maybe they do not have
time to read about the background to a subject or learn how to read a given chart because there
are direct operational decisions at stake. If so, only through the immediacy of understanding will
this visualisation be considered effective. Furthermore, if you create an interactive solution with
an excessive number of features, the richness in functionality may undermine its usage. The
audience may lack the necessary desire to make an effort to interrogate and manipulate the
display. More clicks can equate to more obstacles towards understanding.
Attitude and emotion: As viewers, on occasion we are not in the right mood. We might be tired
and lazy, or we have had a particularly bad day. At these times the prospect of engaging with even
the most compelling and well-designed visualisation might feel too much. I spend all my days
looking at visualisations and can recognise how indifferent I feel towards work when fatigue kicks
in.
An extension of mood is confidence. Sometimes the audience may not feel sufficiently equipped
to embark on a visualisation if it is about an unknown subject, even if assistance is available. It
might be somewhat irrational, but they do not wish to engage, especially if it takes them beyond
their comfort zone in terms of the demands asked of them to interpret and comprehend.
Presentation design: In what format will your viewers need to consume your work? Are they
going to need work created for a print output or a digital platform? Does this need to be
compatible with a small display as on a smartphone or a tablet? Will any viewers have visual
impairments that need accommodating?
If what you create is consumed away from its intended native format, like viewing a large
infographic with small text on a mobile phone, that will impede the experience for the viewer.
How and where your work are consumed will often be beyond your control and you cannot
mitigate for every eventuality.
70

Principle 3: Good Visualisation Design Is Elegant
The third principle is to make your visualisation elegant, which maps again onto three more of
Dieter Rams’ ten principles of good design (Figure 2.10) and Vitruvius’ desire for beautiful.
Elegance is concerned with creating an aesthetic that will appeal to your audience and endure,
sustaining positive sentiment throughout the experience, far beyond just the initial moments of
engagement.
Figure 2.10 Mapping ‘Elegance’ onto Rams’ Ten Principles
Elegant design is presented as the third principle for good reason. Any choices you make towards
achieving ‘elegance’ must not undermine the accomplishment of trustworthiness and accessibility
in your design. Indeed, the pursuit of the other principles often already leads to a certain elegance
as a by-product.
‘When working on a problem, I never think about beauty. I think only how to solve the problem. But when I have
finished, if the solution is not beautiful, I know it is wrong.’ Richard Buckminster Fuller, Celebrated Inventor and
Visionary
Is the Representation and Presentation Design Appealing?
The pursuit of elegance is as elusive as a practical definition. What gives something an elegant
quality? The adjectives that surface in my mind are stylish, dignified and graceful. Elegance has a
timelessness that transcends more fleeting notions like fancy or cool.
Elegance is most conspicuous when it is missing. This is when a visualisation’s design lacks
cohesion and inspiration, especially across the colour and composition elements that so inform its
appearance. By contrast, as expressed by Rams’ principle ‘Good design is as little design as
possible’, elegant design accelerates you to the content and to understanding.
In his book The Shape of Design, designer Frank Chimero references a Shaker proverb: ‘Do not
make something unless it is both necessary and useful; but if it is both, do not hesitate to make it
beautiful.’ In serving the principles of trustworthy and accessible design, you will hopefully have
covered both the necessary and useful. As Chimero suggests, if we have served the mind, our heart
is telling us that now is the time to think about beauty. There are several components of design
thinking I believe contribute to achieving elegance in design.
Eliminate the arbitrary: As with any creative work, good editing is a hugely valuable skill. Every
single design decision you make – every dot, every pixel – should be justifiable. Nothing that
remains in your work should be considered arbitrary, based on random tastes, nor redundant,
offering superfluous value. These will distract and, worse, may distort the process of
71

understanding. Even if your choices are not based on empirical reasoning, you should still be able
to offer justification for every feature that is included as well as any significant feature excluded.
‘“Everything must have a reason.” A principle that I learned as a graphic designer that still applies to data visualisation.
In essence, everything needs to be rationalised and have a logic to why it’s in the design/visualisation, or it’s out.’
Stefanie Posavec, Information Designer
Eliminating the arbitrary should not be confused with the pursuit of minimalism, which is a
brutal approach that strips away the arbitrary and then cuts deeper. In the context of
visualisation, minimalism can be an unnecessarily savage and austere act that may be
inappropriate with the style of work needed.
Thoroughness: A dedicated visualiser should be prepared to agonise over the smallest details and
want to resolve even the smallest pixel-width inaccuracies. The desire to treat your work with this
level of attention demonstrates respect for your audience: you want them to be able to experience
quality, so pride yourself on precision. Not all decisions share the same significance, but we need
to attend to every single decision equally, caring about the small details. Do not neglect to check
things and do not cut corners by not testing. It will be worth it. We are, though, only human
and not every single issue can always be detected and eradicated.
‘You don’t get there [beauty] with cosmetics, you get there by taking care of the details, by polishing and refining
what you have. This is ultimately a matter of trained taste, or what German speakers call fingerspitzengefühl (“finger-
tip-feeling”)’. Oliver Reichenstein, Founder of Information Architects (iA)
Style: This is another hard concept to pin down, especially as the word holds different meaning
to different people. It has been somewhat tarnished by the age-old complaints around something
demonstrating style over substance. When it feels like style over substance has been at the heart
of decision making, the consequence will usually prove to be an obstructed or distorted
experience. Developing a style is a manifestation of elegant design. The decisions around colour
selection, typography and composition are all matters that determine your style. So too does
experimentation in the deployment of different representation techniques or interactive features.
The development of a style creates a degree of consistency and reliability in your strongest design
values that can be repeatedly deployed. It is something that needs time to develop as you find
your design voice.
‘I suppose one could say our work has a certain signature. Style, to me, has a negative connotation of “slapped on” to
prettify something without much meaning. We don’t make it our goal to have a recognisable (visual) signature,
instead to create work that truly matters and is unique. Pretty much all our projects are bespoke and have a different
end result. That is one of the reasons why we are more concerned with working according to values and principles
that transcend individual projects and I believe that is what makes our work recognisable.’ Thomas Clever, Co-
founder of CLEVER°FRANKE, a Data-Driven Experiences Studio
Many news and media organisations actively seek to devise their own style guides to help
visualisers, graphics editors and developers navigate through the choppy waters of design
thinking. This is a conscious attempt to foster consistency in approach as well as create efficiency.
In these organisations, the pressure of tight timescales from the perpetual demands of the news
cycle means that creating efficiency is of enormous value. In removing the burden of having
always to think from scratch about their design choices, the visualisers are left to spend more time
72

on the fundamental challenge of what to show and not get bogged down by how to show it. The
most effective styles stand out as instantly recognisable: there is a reason why you can instantly
pick out the work of the New York Times, the Guardian or the Financial Times.
Decoration should be additive, not negative: The decorative arts are historically considered to
be an intersection of that which is useful and beauty. The term decoration when applied to data
carries a different connotation. It is often used in criticism of the perceived dressing up of data
using superfluous visual flourishes to provoke attention, usually when content is uninteresting or
wafer thin in its substance.
In moderation, visual embellishments can offer effective means for securing and sustaining the
appeal of an audience. We will consider the role of ideas in Chapter 3 where the headline advice
is always to be primarily led by data and your audience, not your ideas. However, there are
occasions in the design process when you should embrace creative flair, novelty and fun. People
like nice things. Sometimes viewers crave something that stirs a more upbeat and upfront
emotional engagement. A singularity of style is a dull existence for all of us.
In certain circumstances you may need to consider employing aesthetic seduction to create an
appeal form that attracts viewers and encourages them to engage with a subject they might not
otherwise have found relevant. This could involve the use of novel visual or functional devices
that attract and perform a useful role.
This is especially the case when your ideas are congruous with the subject matter or key message.
The works featured in Figures 2.11 and 2.12 show appealing enhancements to the background
presentation and representation styling. The choices in each case are harmonious with the
respective subjects of cocoa prices (‘The KitKat’ bar chart) and online razor sales (bar charts
created by scraping away lengths of shaving foam). In each case these design choices offer quite a
charming design solution that does not undermine the message. They supplement the
information presented without obstruction or distraction.
Figure 2.11 Asia Loses its Sweet Tooth for Chocolate, by Graphics Department (Wall Street
Journal)
73

Figure 2.12 Razor Sales Move Online, Away from Gillette, by Graphics Department (Wall Street
Journal)
74

Some may argue that viewers will be encouraged to engage with a visualisation if it is relevant to
them and they should not need to be seduced by novel appearance choices. Indeed, if they need
to be convinced to look at something, maybe they should not be considered to be the intended
audience? Perhaps in a business or operational setting, the needs of individuals, roles and groups
are much more clear cut. Elsewhere, in the real world, there are usually more nuanced
considerations about how best to connect to a potentially diverse audience demographic. Indeed,
as a viewer, your interest in a subject may only materialise as a consequence of experiencing a
visualisation. It might not have existed beforehand as a motivating prerequisite, and without
some initial sense of attraction or intrigue felt towards the prospect of a visualisation, a viewer
may miss the chance to discover this connection.
‘I love the idea of Edward Tufte’s assertion that “Graphical excellence is that which gives to the viewer the greatest
number of ideas in the shortest time with the least ink in the smallest space.” But I found that when I developed
magazine graphics according to that philosophy, they were most often met with a yawn. The reality is that Scientific
American isn’t required reading. We need to engage readers, as well as inform them. I try to do that in an elegant, and
refined, and smart manner. To that end, I avoid illustrative details that distort the core concept. But I’m happy to
include them if the topic could benefit from a welcoming gesture.’ Jen Christiansen, Graphics Editor at Scientific
American
To conclude this discussion about principles, let me explain why I feel three of Rams’ original ten
do not quite fit as universal ideals for data visualisation (Figure 2.13).
Figure 2.13 Considering Rams’ Non-universal Principles
75

Good design is innovative: Most visualisations use tried and tested charting methods that have
been in play for years. Unlike in product design, for example, it is not necessary for visualisers to
conceive constantly new forms of representation or techniques for presentation. You might have
a personal desire to be innovative, aligned to personal goals about the development of your skills,
perhaps through rethinking how to tackle previous projects. It is not that data visualisation is
never about innovation, just that it is not a universal principle.
That said, there are of course circumstances when innovation is important. In the context of
limitations or imposed constraints, in order to overcome a particular challenge, innovation
materialises. You also need it when established solutions fail to resolve new problems.
I sometimes try to look at restrictions in a positive light because of this. Consider the
circumstances faced by Director Steven Spielberg while filming Jaws. The early attempts to create
a convincing-looking shark model proved to be so flawed that for much of the film’s scheduled
production Spielberg was left without a visible shark to work with. Such were the diminishing
time resources that he could not afford to wait for a solution to film the action sequences, so he
had to work with a combination of props and visual devices. Objects being disrupted, like
floating barrels or buoys and, famously, a mock shark fin piercing the surface, were just some of
the tactics he used to create the suggestion of a shark rather than actually show it. Eventually, a
viable shark model was developed to serve the latter scenes but, as we all now know, in not being
able to show the shark for most of the film, the suspense was immeasurably heightened. This
made it one of the most enduring films of its generation. The necessary innovation that emerged
from the limited resources and increasing pressure led to a solution that surely transcended any
other outcome that would have emerged had there been freedom from restrictions. Embrace
circumstances that heighten your need to be innovative; just do not feel it is a mandatory pursuit
for all visualisation contexts.
‘I’m always the fool looking at the sky who falls off the cliff. In other words, I tend to seize on ideas because I’m
excited about them without thinking through the consequences of the amount of work they will entail. I find tight
deadlines energizing. Answering the question of “what is the graphic trying to do?” is always helpful. At minimum the
work I create needs to speak to this. Innovation doesn’t have to be a wholesale out-of-the box approach. Iterating on a
previous idea, moving it forward, is innovation.’ Sarah Slobin, Visual Journalist
Good design is long-lasting: The translation of this principle to the context of data visualisation
can be taken in different ways. ‘Long-lasting’ could be related to the desire to preserve the
ongoing functionality of a digital project, for example. It is quite demoralising how often browser
bookmarks pointing to old visualisations have now elapsed or how often digital works expire due
to a lack of sustained support. The evolution of technology also risks rendering older
functionality obsolete. Thankfully, the long history of print visualisation and infographic work in
particular has a legacy that is simpler to preserve, in many respects.
Another way to interpret ‘long-lasting’ is in the durability of the technique. Bar charts or line
76

charts, for example, are always useful, always being used, always there when you need them.
‘Long-lasting’ can also relate to the rejection of current fashions, preserving a timeless approach
to design thinking that chimes with the discussion about elegance.
I feel ‘long-lasting’ most closely applies to the subject matter and the data portrayed in a
visualisation. Expiry in the accuracy of data about an activity that has since changed undermines
a project’s long-lasting potential. This is particularly the case with subjects concerning current
affairs. Analysis about the loss of life during the Second World War is timeless because nothing is
now going to change the nature or extent of the underlying data (unless new discoveries emerge).
Analysis of the highest grossing movies today will change as soon as new big movies are released
and time elapses. So, again, the idea of long-lasting is context specific, rather than being a
universal goal for data visualisation.
Good design is environmentally friendly: This is of course a noble aim, but the relevance of this
principle has to be positioned again at the contextual level, based on the specific circumstances of
a given project. If your work is to be printed, the resources used will undermine that project’s
environmental friendliness. Developing a rich interactive that is being constantly hammered by
users around the world places a burden on the hosting server and the associated energy supply.
Specific judgements about the scope of environmental impact of visualisation work realistically
resides with the protagonists and stakeholders involved.
Finally, a comment about the need for a visualisation to be memorable. This is often proposed as
a universal aim in data visualisation, but I disagree. If the seamless accessibility of a visualisation
leads to its also being memorable, then wonderful. If the elegance of your design thinking,
possibly including certain memorable visual flourishes, leave such a legacy in the mind of the
viewer, then this will be a terrific by-product of your work. As an objective in itself, achieving
memorability has to be considered, again, at the contextual level based on the specific goals of a
given piece of work and taking into consideration the capacity of the intended audience.
77

Summary: The Visualisation Design Process
78

Design Process
In this chapter you were introduced to the design process, the sequence of activities around
which the book’s contents are organised:
1. Formulating your brief: planning, defining and initiating your project.
2. Working with data: gathering, handling and preparing your data.
3. Establishing your editorial thinking: defining what you will show your audience.
4. Developing your design solution: making design choices about how you represent and
present what it is you want to show your audience.
It explained why a process is important to follow:
It reduces the randomness of your approach.
It offers adaptability to accommodate changing requirements and circumstances.
It protects the value of experimentation.
Each stage reached represents the first occasion you will start to undertake that activity, not
the last.
79

Design Principles
Where the process offers efficiency, design principles ensure effectiveness. The second section
introduced three key principles to help build the clarity of your convictions around the difference
between effective and ineffective visualisation design:
1. Good data visualisation is trustworthy: Is it reliable? Is the portrayal of the data and the
subject faithful? Do the representation and presentation design have integrity?
2. Good data visualisation is accessible: Is it usable? Is the portrayal of the data and the subject
relevant? Is the representation and presentation design suitably understandable?
3. Good data visualisation is elegant: Is it aesthetic? Is the representation and presentation
design appealing?
80

General Tips and Tactics
You were also presented with some general tips ahead of putting the process into practice:
The importance of good time management.
The need to occupy different mindsets at different times, switching seamlessly between
thinking, doing, and making.
Documenting your thought process, capturing sketches and keeping notes.
Communication is a two-way relationship: it is about speaking and listening.
Attention to detail is an obligation: the integrity of your work is paramount.
Do not be precious, have the discipline to not do things, to kill ideas, to avoid scope creep.
Use reflective learning to improve your capabilities and to make the process work for you.
What now? Visit book.visualisingdata.com
EXPLORE THE FIELD Expand your knowledge and reinforce your learning about working with data through this
chapter’s library of further reading, references, and tutorials.
TRY THIS YOURSELF Revise, reflect, and refine your skill and understanding about the challenges of working
with data through these practical exercises.
SEE DATA VISUALISATION IN ACTION Get to grips with the nuances and intricacies of working with data in
the real world by working through this next instalment in the narrative case study and see an additional extended
example of data visualisation in practice. Follow along with Andy’s video diary of the process and get direct insight
into his thought processes, challenges, mistakes, and decisions along the way.
81

http://book.visualisingdata.com

Part B The Hidden Thinking
82

3 Formulating Your Brief
In Chapter 2 we learnt about the importance of adopting a process to tackle data visualisation
challenges through an organised sequence of activities. Supplemented by the guiding benefit of
design principles, this offers a framework to help you make good decisions.
This third chapter initiates the design process commencing with stage 1. This stage encompasses
activities concerned with ‘formulating your brief’ to forge initial clarity about the context and
vision of your work.
The manifestation of a brief can be as informal or as formal as your situation requires. For
example, it can be useful when working with other stakeholders to document information about
the requirements and conditions of a project. It can then be shared, agreed upon and referred
back to. It will be in the interests of all parties to have such a source of mutual understanding,
especially for matters to do with the expected deliverables. For more personal projects you might
need to make basic notes to capture your thoughts.
The primary task of this stage is to establish why you are producing this data visualisation. What
is its raison d’être? This involves identifying the origin curiosity that will drive your work. This is
an articulation of the appetite for understanding you are addressing through your visualisation.
No visualisation project is ever undertaken free of constraint, so you will also spend time defining
the influential contextual matters around the who, the where and the when. These requirements
and factors will shape the conditions of the project you are about to undertake and need to be
recognised early.
You will then switch focus, looking ahead to consider the vision of your work. Thinking about
this vision represents early conceptual thinking about what it is you might be developing,
providing early clues about the best-fit tone, functional experience and style your visualisation
may need to demonstrate. The design-centric specifics of how you will fulfil this will be kept on
ice until we reach stage 4 later in the process.
In defining your project’s purpose, you will consider more deeply what it is for: what are you
trying to accomplish? The type of understanding you are facilitating is important. For example,
are you imparting key messages to your audience or enabling them to make their own
discoveries? Are you placing an emphasis on the precision of readability or amplifying the feeling
of a subject? We will also look at harnessing instinctive ideas that form in our minds, concerning
the keywords, imagery, metaphors and external inspiration that might be relevant to the subject.
83

3.1 Defining Your Project’s Context
84

What Is the Motivating Curiosity?
Answering the question ‘Where does a data visualisation process start?’ might seem
straightforward. Instinctively, one might suggest it starts with a request. Someone asks someone
else to do a visualisation. They share some background information about the requirements,
maybe provide access to some data, and this sets the process in motion.
This type of scenario is clearly commonplace. However, though this might be how a process
starts, it is not quite representative of where things truly start. You see, before a request is issued,
before any data is shared and certainly before any design work is commenced, a curiosity has
formed: some origin interest held by someone about a subject.
‘Be curious. Everyone claims she or he is curious, nobody wants to say “no, I am completely ‘uncurious’, I don’t want
to know about the world”. What I mean is that, if you want to work in data visualisation, you need to be relentlessly
and systematically curious. You should try to get interested in anything and everything that comes your way. Also,
you need to understand that curiosity is not just about your interests being triggered. Curiosity also involves pursuing
those interests like a hound. Being truly curious involves a lot of hard work, devoting time and effort to learn as much
as possible about various topics, and to make connections between them. Curiosity is not something that just comes
naturally. It can be taught, and it can be learned.’ Professor Alberto Cairo, Knight Chair in Visual Journalism,
University of Miami, and Visualisation Specialist
The dictionary definition for curiosity is ‘possessing a desire to know or to learn something’. If
visualisation is about facilitating understanding, these are the two ends that meet. Curiosity
therefore represents the why of your process: the instigating, driving motive for a visualisation
project to be developed.
You do not create a visualisation because you happen to have data. You create a visualisation
because there is a definable appetite for the understanding it offers, whether this appetite is held
by you or somebody else (that you essentially inherit). Any visualisation work undertaken in the
absence of a definable curiosity will lead to an uncertain and aimless decision-making process.
By identifying your project’s origin curiosity, it gives shape to your subsequent decisions,
especially those concerned with the content side of your work. This enables you to keep checking
that your choices help to contribute towards facilitating understanding about the most relevant
matters.
A key attribute of any curiosity is to recognise from whom it originates. Figure 3.1 shows an
example of a visualisation produced in response to my own recognised curiosity. This type of
work is often characterised as being a ‘pet’ or ‘passion’ project that is entirely self-initiated, with
no other stakeholder involved. You have freedom to follow your own enquiry, shaped only by the
limitations of your imagination and interests. The ‘Filmographics’ project was entirely motivated
by a curiosity I had about the movie industry: ‘What are the patterns of success or failure in the
movie careers of a range of notable actors/directors?’
Figure 3.1 Filmographics, by Andy Kirk and Matt Knott
85

Articulating my curiosity in this way helped focus my decisions about what data to gather, what
analysis to conduct and, thereafter, what features of representation and presentation to employ.
There were many things I could have explored about the movie industry, but this was the
particular slice of content that intrigued me the most. I wanted to know about the story of
Steven Spielberg’s career. I wanted to know whether Meryl Streep’s critical successes had been
matched by financial success. When was De Niro last consistently making good movies? Why has
Adam Sandler been allowed to make any?
Although the Filmographics project was initiated by me and served my appetite for
understanding, it was published publicly in anticipation of its also being relevant to certain other
audiences who perhaps share an interest in the subject matter. I was not explicitly serving their
expressed curiosity, rather expecting that some would share my curiosity.
Sometimes the curiosity you are pursuing does not originate from you. Stakeholders are the
people involved in a visualisation project, other than yourself, who may influence what curiosity
your work should pursue. Stakeholders exist as managers, academic supervisors, clients or
colleagues, and it might be them tasking you to undertake a visualisation project based on the
curiosity they express to you. In these situations, you are inheriting their interest and you have to
own it from then on.
In certain situations, stakeholders may have a dual role of initiator and intended recipient (e.g.
‘can you show me trends of sickness absence among staff in my department this year?’), in others
they might be expressing to you what they expect a separate audience would find interesting (e.g.
‘can you produce an analysis showing trends of sickness absence among staff this year to share
with all heads of departments?’).
An important point from the previous chapter needs to be reinforced here: this is the first but not
the last occasion when you will have the opportunity, and reason, to refine the definition of your
ultimate curiosity. Depending on the particular situation of your work, it may that your initially
expressed curiosity is not fixed, not specific and not even singular.
The need for specificity in your curiosity will vary from one situation to the next. ‘How much did
86

the public engage with the previous Australian election?’ is a far broader curiosity than ‘What was
the percentage turnout across each electoral region of Australia compared with the previous
election?’ If you are a runner with a fitness-tracking device or application, you might finish a run
and wonder ‘how good was that run?’ This is a broad enquiry. To form an answer requires the
synthesis of several distinct pieces of information (‘How far? What time? What route? What
achievements? What previous times?’) that collectively provide a notion of how good the run was.
By contrast, if you just want to know ‘in what time did I complete the run?’, this is a specific
curiosity that can be effectively answered by a single piece of information.
On many occasions I have embarked on a visualisation project with an initial curiosity in mind
but then, having become better acquainted with the subject through its data, other legitimate
enquiries subsequently have emerged as simply being more relevant. It is easier to justify shifting
your focus when working on your own projects. When you are being tasked by another
stakeholder, you might find there is less room for manoeuvre beyond the pursuit you have been
tasked with. That said, you should still seek constant dialogue with your stakeholder if you
strongly feel another route might be more interesting. After all, it serves nobody’s purpose if you
remain anchored to an enquiry that no longer reflects the most relevant aspects of a subject. The
important thing to challenge is whether any shift in focus should be embraced or curtailed. No
matter how relevant or interesting your new possibilities are, you might simply be drifting
beyond the scope of the work.
Sometimes your process is being driven by the need to serve the known or anticipated interests of
your intended audience. If you know your audience well enough and are able to predict their
potential needs, your origin curiosity will form around what it is you think they want answering.
Of course, there will be situations where you are in position to consult your audience directly to
define explicitly what appetite they have about the topic in question. Otherwise, you will need to
make reasonable judgements to anticipate what most people will likely find most interesting.
These are the situations in which I find most indecision in a visualisation process, mainly because
there is always so much choice and so much temptation to mark up everything as being of equal
interest. Additionally, at the outset of a project, it might not be reasonable to expect you to be
aware of all the potential features of interest about your subject or in your data. This is
something that will develop as you work further through stage 2 (‘Working with data’) and stage
3 (‘Editorial thinking’).
It is often the case that committing to just a single avenue of curiosity is not feasible. Embarking
on multiple distinct curiosities, bound by a shared connection to the same subject matter, might
give you more work to do but might be necessary, especially with data or subjects that are
unfamiliar to you at the outset.
Suppose you are a student studying the history of music and encounter some data about the
structure of popular songs. You suspect there are many potentially fascinating things to discover
in this data, but you do not yet know what the single most interesting curiosity will be. A key
activity of the visualisation process will be to explore the data, unlock its key qualities, test out
the multiple different enquiries, and thereafter determine which perspectives offer the most
relevant or interesting findings. You are possibly pursuing multiple speculative curiosities in order
to see which ones emerge as being most relevant. You might still end up with several, only one,
or indeed no legitimate curiosities.
87

In my experience, forming a question tends to be the most useful and comfortable articulation of
a curiosity. In doing this you are positioning your visualisation as a means of providing some
notion of an answer. I find this a natural way to keep my mind focused on pursuing this answer.
An alternative way to approach this, especially when you are anticipating what others might find
interesting, is to switch your viewpoint away from the question form and think more in terms of
what it is you are aiming to present to your audience. You might extend the wording of the data
visualisation definition to describe what the facilitated understanding will be about, as illustrated
in Figure 3.2.
Figure 3.2 Sample Statements of Curiosity
Regardless of your curiosity’s origin, specificity, permanence and form, you just need somewhere
to start from. Seek to express the most useful overriding curiosity that best encapsulates, at this
stage, what you are setting out to pursue for you and/or your audience. Define now, refine later.
88

Identifying Project Circumstances
The second aspect of contextual thinking concerns identifying a project’s circumstances. These
are the frictions and freedoms that are imposed on you or determined by you. Whether you are a
full-time visualisation professional, a student, a researcher, working in a business or doing
visualisation as a pastime, there are common influencing factors that characterise the conditions
of the projects you are undertaking. They will determine the boundaries of your creative
ambitions.
‘Context is key. You’ll hear that the most important quality of a visualisation is graphical honesty, or storytelling
value, or facilitation of “insights”. The truth is, all of these things (and others) are the most important quality, but in
different times and places. There is no singular function of visualisation; what’s important shifts with the constraints
of your audience, goals, tools, expertise, and data and time available.’ Scott Murray, Principal Learning Scientist,
O’Reilly Media
When commencing a project, probably not all of the circumstances that may potentially
influence your work will be definable. Things change. That is why we need to be prepared to
accommodate elegantly the impact of new factors at any point in the process. Of course, the
more things you can define, the more things become fixed and this reduces uncertainties. Ideally,
we want to eliminate as much of the unknown as we can. Useful definition also exists through
identifying the absence of restriction or requirement. Knowing you have freedom to determine
choices yourself is of clear value. It gives you control. Sometimes you might see merit in
imposing restrictions on yourself, where none exist, to aid your focus. Constraints are not always
a bad thing; indeed, they can often help us innovate.
People
Stakeholders: In project situations where you have been requested to develop a visualisation by
somebody else, it is important to establish an understanding of who is involved. ‘Who is the
ultimate customer?’ is a key question to answer. The customer will always be particularly invested
in what you develop. This may or may not be the person(s) who has directly commissioned you,
but they will usually determine whether the work is on the right track, in their view. They will
also critically determine when the work is ultimately of sufficient quality to be considered
finished. In my world as a freelance design consultant, my customer determines when (and if) I
will get paid.
Other stakeholders might exist as subject-matter experts, available to offer advice about domain-
specific queries. They might be able to guide you on what the most salient issues are in a subject.
They might be points of contact to raise issues around technology requirements or advise on
some of the nuances you encounter around values held in datasets. On rare occasions, some
stakeholders can hinder progress by unduly influencing design decisions beyond their remit and
capability. They can become interferers, making conditions harder for you.
Identifying this cast and crew of people who have a stake in your work will help you anticipate
the interactions and relationships you might need to manage: what personalities exist, what help
you might exploit, and what obstacles might need navigating. Diplomacy will be required.
89

If there are no stakeholders, and the project is effectively a solo pursuit, there will be much more
autonomy, which can be liberating, but with this comes more responsibility on you to direct all
matters yourself.
Audience: There are several characteristics of your target or expected audience that will need
careful consideration. Your audience will bring irrationalities and inconsistencies. Identifying
their varied traits and accommodating their influence into your decision making is a permanent
concern throughout the design process and one that requires clear judgement:
What is your audience’s relationship with the subject matter? What knowledge do they have
or, conversely, lack about a subject? What assistance might they need to interpret the
meaning of the subject? Do they have the capacity to comprehend what it means to them?
What is their motivation for acquiring the understanding you intend to provide for them?
Do they have a direct, expressed need or are they more passive and indifferent? Might you
need to find a way to persuade them or even seduce them to engage?
What are their visualisation literacy capabilities? Might they require assistance perceiving the
chart(s) produced? Are they sufficiently comfortable with operating features of interactivity?
Do they have any visual accessibility issues, such as red–green colour blindness, that will
need to be factored into your design thinking?
As you will discover, there are usually lots of textual elements included in any visualisation.
What regions of the world do your audience come from and, therefore, what language
considerations must you take into account? Might you need to create multiple translated
versions of the eventual solution?
Sometimes, you have direct knowledge of your audience and can easily characterise their needs
and mould your choices accordingly. For example, if you have a fixed group of viewers who you
know will understand the technical context of the data you are presenting, you probably will not
need to include the detailed explanations that would be necessary for a less knowledgeable
audience. On other occasions, your audience’s characteristics may be more ambiguous and so
distant from you that you can only rely on reasonable imagination. You might form in your
mind estimated personas of the types of people you could expect to be the main beneficiaries. If
you have an especially wide-ranging and diverse profile of audience characteristics, you are
unlikely to be able to satisfy the needs of each variation; you might need to commit to
prioritising some audiences over others. One size does not fit all.
Visualiser(s): Data visualisation design is truly a multidisciplinary endeavour. It is this variety
that fuels the richness of the subject and makes it a particularly compelling challenge. To master
it requires a repertoire of skills, knowledge and different attitudes that dominate different stages
of this process. Inspired by Edward de Bono’s Six Thinking Hats (1985), the seven hats of data
visualisation in Figure 3.3 represent my attempt to deconstruct the specification of an imagined
‘perfect’ visualiser. The attributes listed under each of these hats can be viewed as a wish list of
personal or team capabilities, depending on the context of your data visualisation work.
‘There is not one project I have been involved in that I would execute exactly the same way second time around. I
could conceivably pick any of them – and probably the thing they could all benefit most from? More inter-
disciplinary expertise.’ Alan Smith OBE, Data Visualisation Editor, Financial Times
90

Figure 3.3 The Attributes that Comprise the ‘Seven Hats of Visualisation Design’
Across these capability groups, which attributes do you possess, or can you demonstrate? Where
are your weaknesses, in terms of both gaps and potentially overly dominant traits? I am painfully
aware of the things I am simply not good enough at (programming), the things where I rely on
instinct more than skills gained from training (graphic design) and also the things I do not enjoy
(finishing, proof-reading, note-taking). If certain skills are not available to you compromises may
be required and ambitions may need to be lowered.
If collaboration is possible, there are clear advantages in pooling diverse capabilities into a shared
challenge. The best functioning visualisation teams will offer a balanced blend of skills across all
these hats. Success will be hard to achieve if a team comprises a dominance skewing the diversity
of abilities, so what is the best way to allocate or occupy different duties to optimise your design
process with a team?
Constraints
Timescales: The primary constraint is usually how much time there is to develop your solution.
Most projects have a deadline attached to them, whether this is imposed by other stakeholders,
mutually agreed or set by yourself. Even if you do not need necessarily to adhere to a deadline,
let’s say for personal projects, it can still be useful to define a target date to help sharpen your
progress. At the opposite end of the timeline, there is the start date. This may not be now. You
may have to wait for certain conditions to be in place before you can even commence your work.
91

If you are conducting an analysis of some survey results, you will not have a complete, final
dataset of responses to work with until the survey is closed.
During your project there may be certain milestones to factor in as well. These might be
occasions when you need to show work in progress or critical points when you switch to working
with real data rather than sample data that may be used to draft early ideas.
The most crucial aspect of time is task duration. There are clearly going to be large differences in
the ambitions of a project to be completed in two days compared with another in two months.
But if the two-day deadline concerns a small-scale task that will take only a few hours, that is
going to be deliverable. Though a two-month deadline sounds great, if you are facing three
months’ work it will be a struggle to accomplish it in time.
Estimating project duration to any reliable degree is a difficult thing to judge. You usually do not
know how long a project will take until it is completed, which is often too late to be useful. Even
with experience from working on a diverse range of projects, seemingly similar projects can end
up with very different task durations. I would always recommend noting down the duration of
each major task across your design process, so you can more surgically evaluate how you have
spent your time and be better placed to estimate accurately commitments on future projects.
‘What is the least this can be? What is the minimum result that will 1) be factually accurate, 2) present the core
concepts of this story in a way that a general audience will understand, and 3) be readable on a variety of screen sizes
(desktop, mobile, etc.)? And then I judge what else can be done based on the time I have. Certainly, when we’re down
to the wire it’s no time to introduce complex new features that require lots of testing and could potentially break
other, working features.’ Alyson Hurt, News Graphics Editor, NPR, on dealing with timescale pressures
Pressures: Depending on the context of your project, certain cost factors may exist. Going back
to the matter of time, how much can you afford to spend on a project? Some projects may have a
budget allocated and so the associated staff activity costs need to be managed sensibly. You might
need to outsource to external parties with specialist expertise, for example transcription services,
third-party data sources, illustration work, but can you afford the costs involved? What costs will
be incurred in paying for hardware, software, licences to use photography or audio?
Further pressures may emerge from the politics surrounding your subject, the data, or the
messages coming from this data. I have been involved in several projects where the charts
produced showing data about cities or countries had to be sorted alphabetically, and not by any
other ranking measure, in order to preserve a certain diplomatic neutrality. You may receive
guidance from your stakeholders that certain messages need to be downplayed or amplified.
These can be difficult matters to handle: you want to respect any requirements received, but also
you do not want to undermine the integrity of what you are representing.
There may be cultural sensitivities to consider if you are creating work for audiences from
different regions. Issues around the use of imagery, colour connotations, or symbology of certain
forms may need to be carefully handled. There may also be environmental considerations,
particularly concerning the output of your work, that need to be observed.
Design: Restrictions around certain design choices are common, often informed by style
guidelines that must be adhered to through the use of specific colours, typeface and fonts. Where
92

possible, I always attempt to challenge these rules somewhat because they can be unnecessarily
restrictive, but permission is not always granted. You may need to include logos, which can take
up valuable space and unbalance your composition, but rules are rules and therefore we need to
know about these things at the outset, not later.
Layout or size restrictions may also exist, dictating the space in which you have to work. For
example, when producing graphics for journals or for digital outputs that need to work on a
tablet or smartphone, you might have quite a small amount of space to utilise. Conversely, your
output might need to be very large, which can introduce different challenges with legibility and
resolution quality.
Further creative pressure might materialise in the form of what I describe as market influences.
The visualisation you develop may have to compete for attention alongside other work. For
example, if you are creating a visualisation for a charitable organisation, how do you get a
message across louder and more prominently than others competing for the same eyeballs? If you
are working on an academic research project, how do you get your findings heard among all the
other studies battling to create an impact? Creative influences can emerge internally, through the
unique dynamics of an organisation, and externally, through broader competition across the
entire marketplace and industries. Although it is not the most important factor, a desire to
emulate the best or differentiate from the rest can prove to be a strong motive in your design
thinking.
Technological: As I have mentioned in the Introduction, there are myriad tools, applications
and programming libraries in data visualisation, offering a varied landscape of capabilities. The
technology you have access to will affect how digitally ambitious your work can be and/or how
efficiently you will be able to make it. You can only achieve what your tools enable you to
achieve. This influence will shape several stages of your process:
Working with data: Technologies to help with acquiring, examining, transforming and
exploring data. How much data can your tools handle? How quickly do they perform
actions, especially with large data? Do they enable automation, maybe through scripting?
What range of statistical techniques is available? How effective are the methods for
modifying data? Do they enable you to explore your data visually?
Data representation: Technologies to help with making charts. What range of different
charts do they enable you to make? Is the process of constructing charts automatic or
manual? Do the tools facilitate workarounds or means for potentially expanding their
standard capabilities?
Interactivity: Technologies to develop features for exploration and control. What range of
different interactive features do they offer? Is the process of developing these features
automatic or manual? Do the tools facilitate workarounds or means for expanding their
potential capabilities?
Data presentation: Technologies to manage the inclusion of annotations, the use of colour
and the composition of your work. What range of annotated features can you include? Are
you able to control fully the appearance of these features? To what extent can you manage
the colour applied to every visual element? Likewise, what degree of control do you have
over the size and placement of all elements?
Publishing: Technologies to disseminate your work. Do you present through a slide deck?
93

Do you compile a printed report? Do you publish your work through a website? Do you
upload it to be accessed/downloaded from the Web? Do you send files via email to others
with the same applications? Do you compile a video? Do you publish as a gif on social
media?
Deliverables
Setting: This concerns the characteristics of the setting in which your work is going to be
encountered and consumed by your audience. Firstly, is it going to be consumed remotely –
away from you – or presented in person? If you are not personally present to offer verbal
explanations of key features and findings, descriptions of the data gathering process, assumptions
or calculations, you may need to include these as annotated properties. Secondly, is the nature of
the engagement one that needs to facilitate especially rapid understanding, or does it lend itself to
a more extended, prolonged engagement? I usually think of four broadly typical settings:
‘I like to imagine that I have a person sitting in front of me, and I need to explain something interesting or important
about this data to them, and I’ve only got about 10 seconds to do it. What can I say, or show them, that will keep
them from standing up and walking away?’ Bill Rapp, Data Visualisation Designer, discussing an audience
scenario setting he conceives in his mind’s eye
The boardroom: A setting characterised by limited time, limited patience and limited
attention. Immediate insights and key messages need to be imparted at a glance. There will
likely be reduced appetite for engaging with anything that requires effort, such as an
unfamiliar chart type or rich interactivity, unless someone is there to present it.
The coffee shop: A more relaxed setting that might be compatible with a piece of work that
involves more effort and requires more time to learn about the subject. Unfamiliar
representations might be reasonable to use as long as sufficient assistance is provided about
how to read the displays. Interactive features to enable interrogations may not pose the same
obstacles to understanding in the way they would in other situations.
The cockpit: A situation characterised by operational need, whereby the visualisation is
offered as a tool or instrument to provide immediate signals that stand out at a glance.
Sufficient breadth and depth of additional content will also be required to serve the
multitude of different potential scenarios that might arise. Think of a wide-ranging
organisational dashboard or a reference map that serves multiple potential levels of enquiry
that aid the operational needs of navigation, from high-level orientation to in-depth
localised detail.
The prop: Here a visualisation plays the role of a supporting visual device to accompany a
presenter’s verbal facilitation of the key understandings (perhaps in the form of a talk or
video) or an author’s write-up of key findings that refer to an accompanying chart or figure
(report or article).
Medium: You will need a clear understanding about the specific format of the deliverables
required. Is your intended output to be produced for print, for digital or maybe even physical?
Will it be static, interactive or animated? An important thing to reinforce again: just because
something is published on the Web does not mean it is interactive.
Maybe you will have to produce something that will published across multiple media. For
94

example, newspapers typically publish graphics in their printed version, on the web version, on
their mobile apps, and often also share them via social media. While it may be the same graphic
produced four times over, there may still be subtle alterations required to optimise the
presentation for each respective platform, especially when conside-ring the impact of the size
restrictions that will exist. This increases your workload.
‘I love, love, love print. I feel there is something so special about having the texture and weight of paper be the canvas
of the visualisation. It’s a privilege to be able to design for print these days, so take advantage of the strengths that
paper offers – mainly, resolution and texture. Print has a lot more real estate than screen, allowing for very dense,
information packed visualisations. I love to take this opportunity to build in multiple story strands, and let the reader
explore on their own. The texture of paper can also play a role in enhancing the visualisation; consider how a design
and colour choices might be different on a glossy magazine page versus the rougher surface of a newspaper.’ Jane
Pong, Data Visualisation journalist at the Financial Times
Having spoken about technology, maybe you do not need any? In the right context, it may be
possible to embrace more analogue or artisanal approaches, as demonstrated in the example
shown in Figure 3.4, which is a visualisation (or possibly better termed a ‘data physicalisation’)
created using Play-Doh. This analysis presents data about the various plug types used around the
world. The imprints of the configuration of pins for each plug type are stamped into coloured
chunks of Play-Doh, and then white lengths are measured out to represent the populations of
people from the countries whose power systems use each type.
Quantity: As well as the medium, it is also important to get a sense of the project’s deliverables
in terms of expected quantities. How many things are you making? How much, what type, what
shape and what size? For example, are you producing 12 different graphics for a month-by-
month slide deck or contributing to a large report that will need two charts developed for each of
the 50 questions asked in a survey? Perhaps it’s a web-based project with four distinct sections,
each requiring at least one interactively adjustable view, or it could be just a single chart to be
emailed to your manager. It is not always possible to determine output quantities this early in the
process, but you should certainly maintain awareness of how realistic the expected deliverables
are going to be, given the project’s resources (timescales, skills and budget, where relevant).
Frequency: The issue of frequency concerns how often a particular project will need to be
reproduced and what its lifespan will be. It might be a regular (e.g. monthly report) product, in
which case the efficiency and reproducibility of your design solution will be paramount. If it is a
one-off or irregular piece (e.g. election polling graphic updated after each new release), you will
have more justification to create a bespoke solution so long as the cost–benefits involved remain
positive. Maybe it is a one-off project in development terms but will be constantly updated and
republished on a frequent basis thereafter, such as a monthly report. There may be more upfront
work to develop a functionally robust template, though the task of generating each subsequent
monthly report may only involve a limited amount of work. You may need to consider if there
will be any future benefits from reusing some of the techniques you have employed, so there is
some recyclable value. Can you justify investing time, for example, in programmatically
automating certain parts of the construction process if they can be reused to save time in the
future?
Figure 3.4 Popularity of International Outlets, by Amy Cesal
95

96

3.2 Establishing Your Project’s Vision
97

Defining Your Project’s Purpose
Identifying the curiosity that motivates your work establishes the project’s origin. The
circumstances you have just considered define the conditions you will experience and need to
accommodate through your project. To supplement this contextual thinking, the vision you have
for your work needs some early consideration.
The definition of vision is ‘the ability to think about or plan the future with imagination or
wisdom’. What are you hoping to accomplish with your visualisation: what is its purpose?
Articulating your project’s purpose represents a statement of intent. It offers clarity about what
you see as your destination. As before with the origin curiosity, purpose might evolve as you
progress through the process, but the sooner you can establish at least some degree of focus, the
better, especially in being able to eliminate potential creative avenues that will have no relevance
to your aims.
We have established that the overriding goal of a visualisation is to facilitate understanding,
though the nature of understanding can vary from one project to the next. Some visualisations
aim to be quite impactive, attempting to shock an audience into changing behaviour or perhaps
inspiring viewers to take significant action. For example, you may seek to change attitudes among
parents about the effect of sugary drinks in contributing to the rise of obesity. To accomplish this
might require an emotive style that amplifies the feelings about the subject, attracting the
attention of impassive viewers and then striking home with a powerful message that, hopefully,
resonates deeply with anyone in a position to act. Affecting people to this extent can be
ambitious, but it might be the purpose of the work to be this ambitious.
In another context, a visualisation about obesity may hold more modest ambitions of just seeking
to inform viewers about a subject. Let’s suppose you are offering health professionals an interface
that lets them explore obesity trends in their local area. They probably will not need convincing
about the significance of this topic or any aesthetic seduction to encourage them to participate.
As health professionals they are likely to have an operational need to know this and a certain
responsibility to educate citizens themselves. The most suitable style of visualisation in this case
might therefore be more low key, perhaps imparting an authoritative tone with an emphasis on
technical precision and clear functionality. If all it achieves is to reinforce existing understanding
or just add an extra small grain of understanding, this will possibly represent success. Purpose
achieved.
There are different types of visualisations that demonstrate different design characteristics, the
suitability of which will be largely determined by what you are trying to accomplish. In Chapter
1, I described how viewers go through three phases of understanding: perceiving, interpreting and
comprehending. I explained how, as visualisers, we have limited control over the final phase,
comprehending, which is largely determined by a viewer’s attitude and connection to the subject
matter. We do, though, have control over how our viewers perceive and interpret our
visualisations. They are particularly influenced by the choices surrounding two significant design
characteristics: tone and experience.
98

99

Judging the Tone of Your Visualisation
The tone conveyed by a visualisation has an influence on the perceiving phase of understanding.
In judging the most suitable tone for your project, you are deciding whether to place more
emphasis on the viewer being able to read data or feel data (Figure 3.5).
Figure 3.5 The Spectrum of ‘Tone’
Reading tone: A visualisation that conveys a reading tone places emphasis on optimising the
precision and efficiency of perceiving the represented data. The visual quality that embodies a
reading tone of voice might be described by adjectives like pragmatic, authoritative, analytical,
conservative, utilitarian and (necessarily) boring.
A reading tone might be suitable in circumstances where there is no need to employ any form of
visual stimulation to impart a message more potently, nor to seduce an audience through
aesthetic appeal. There will be good reason to portray the underlying subject in a statistical style
and there is no desire or relevance in amplifying any emotional devices.
The design choices employed will seek to make it easier for a viewer to determine the magnitude
of and the relationships between values. Representation methods like bar charts, as shown in
Figure 3.6, embody this tone of voice. By representing the size of a quantitative value using the
proportional size of a line, bar charts facilitate both general sense-making and precise point-
reading, thus heightening the perceptual accuracy for the viewer. In this example you can quickly
ascertain that the USA value is about three times the size of the next largest, the UK one.
Although similar in magnitude, you can see that the value for Italy is slightly larger than the one
for The Netherlands which is slightly larger than those for Canada and Switzerland. You can
probably estimate France’s absolute value to be around 55 and Germany’s around 90. For
exactness in reading the values you would offer direct value labelling, though the degree of
accuracy in judgements here is already quite high and probably sufficient.
Figure 3.6 Nobel Laureates Awarded (1901–2017), by Country of Birth
Bar charts are so ubiquitous and so necessary because they make it easier to answer the quite
reasonable question: ‘What is the size of that value?’ Most of the visualisations you will ever
100

produce will likely lean towards offering this kind of reading tone.
Indeed, you might reasonably ask why would you ever not seek to optimise the accuracy and
efficiency of value judgements? Surely anything that compromises on this is undermining the
accessibility of your design and maybe even jeopardising its trustworthiness? Well, this is why the
definition of your purpose is so significant. There are other considerations, as typified by the
feeling end of this continuum.
Feeling tone: In contrast to reading values, sometimes we might justifiably place more emphasis
on feeling data. The visual quality that embodies a feeling tone of voice might be described using
adjectives like emotive, seductive, figurative, big-picture, fun and dramatic.
Sometimes, the perceptual judgements that are most important for your viewer may align more
with the notion of ‘getting the gist’. This means the viewer can quickly and easily form headline
observations of the hierarchy of large, medium and small properties of your data. The viewer
might gain a general sense of major patterns that reveal things going up and going down, the
major clusters of connections and the major components of a whole. A representation of data
that facilitates ‘at-a-glance’ viewing is sometimes the most suitable way to portray a subject’s
values. The consequence of this is that perceiving precise readings is diminished.
As described earlier, the benefit of visually representing data is that it offers something different,
and often something better than a table of data, by helping a viewer see quantitative and
qualitative relationships in a subject. There are occasions when we want to do more than just let
a viewer see the subject through its data. Sometimes you will be working with subject matter that
has the potential to stir strong emotions or relates to inherently imprecise or abstract concepts.
The projects displayed in Figures 3.7 and 3.8 exist at the very intersection of these notions,
portraying the imprecision of emotion as conveyed through the use of language. The work in
Figure 3.7 is an excerpt from a project looking at the emotional arcs of the past ten US
presidential inaugural addresses using the Microsoft Emotion API to analyse facial expressions
and match them to common emotion classifications. Each ‘feather’ form represents a full
inaugural address and each barb of the feather is a moment during the speech where the
president displayed an emotion: positive emotions are drawn above the quill, negative emotions
below. The length represents the intensity of the emotion.
Figure 3.8 analyses the emotions found in Taylor Swift’s song lyrics. In this project the data was
processed using IBM Watson to derive emotions of happy, sad, mad and scared. The emotional
mix of each track is then represented like a mix of gooey liquid with different measures of yellow,
blue, red and purple. The size of each blob represents a measure of the confidence in how the
process has identified explicit emotions; the smaller songs indicate Watson could not absolutely
detect outright emotion.
Figure 3.7 One Angry Bird, by Periscopic
101

The data in both these works is based on a good degree of automated subjectivity, but the
respective portrayals perfectly embody the feeling of the subject: you cannot read them with
precision and you should not seek to read them with precision, because the data represented in
them does not convey precision.
As I mentioned, sometimes we do not need to read values precisely because it is more important
just to get the gist. In the project illustrated in Figure 3.9, you see visuals from an analysis of the
families who have most financial clout when it comes to funding presidential candidates. The
data quantities are portrayed using stacks of Monopoly house pieces piled up outside on the
White House lawn. The red houses represent the small number of families who have contributed
nearly half of the initial campaign funding, the green pieces are representative of the total
households in the USA. You cannot count the number of pieces piled up. You cannot get
remotely close to estimating their quantity, but you can get a sense of their relative proportion
from the juxtaposition of many compared to few. It offers a visual approximation of the
remarkably disproportionate balance and power of wealth. That is the only level of readability
offered and intended.
Data is more than just a bunch of numbers and text values. Thinking about tone is to recognise
semantically what your subject is about: what activity, instance or phenomenon does it represent?
Is it about people, places, products? Is it about similarities or differences, change or growth?
Figure 3.8 Taylor Swift is Mostly Happy, Quite Often Sad, Sometimes Mad, and Occasionally
Really Scared, by Shirley Wu
102

Figure 3.9 Buying Power: The Families Funding the 2016 Presidential Election, by Wilson
Andrews, Amanda Cox, Alicia DeSantis, Evan Grothjan, Yuliya Parshina-Kottas, Graham
Roberts, Derek Watkins and Karen Yourish (New York Times)
103

‘There’s a strand of the data viz world that argues everything could be a bar chart. That’s possibly true but also
possibly a world without joy.’ Amanda Cox, Editor, The Upshot
Learning about the underlying phenomena of your data helps you feel its spirit more clearly than
just looking at values in isolation. This prepares you for the level of responsibility and potential
sensitivity you will face in curating a visual representation of this subject matter. As we saw in
Chapter 2, with the case of the ‘Florida Gun Deaths’ graphic (Figure 2.5), some subjects are
inherently more emotive than others. You might choose to amplify or suppress the emotion of
the subject, and you need a clear conviction in deciding how to find the most suitable tone of
voice.
For subjects that carry the weight of strong emotion, there might be good reason to exploit the
inherent feelings. Encapsulating emotional sensations like fear, disgust, fun and humanity
through your design choices might accelerate the meaning of the subject and potentially affect
the most elusive phase of understanding, comprehending and how viewers feel.
This approach could be seen as somewhat manipulative. To a certain degree it probably is and
there are risks associated with misjudging the employment of emotional attributes. A playful
approach to portraying data about a serious topic will demonstrate insensitivity and possibly
undermine the trustworthiness of your work, even if you have created an elegant solution. As
long as you are faithful to the underlying data and the subject’s visual embodiment is not
superficial, artificial or deceptive, I believe it is an entirely appropriate motive when the
circumstances suit.
‘Find loveliness in the unlovely. That is my guiding principle. Often, topics are disturbing or difficult; inherently
ugly. But if they are illustrated elegantly there is a special sort of beauty in the truthful communication of something.
Secondly, Kirk Goldsberry stresses that data visualization should ultimately be true to a phenomenon, rather than a
technique or the format of data. This has had a huge impact on how I think about the creative process and its results.’
John Nelson, Cartographer
It is important to note that any visualisation work that leans more towards ‘feeling’ is typically
the exception and will be relevant to a minority of situations. A skilled visualiser needs an
adaptive view and the ability to judge the appropriate occasions whereby the purpose of
visualisation will support such an exceptional approach.
When it comes to defining the best-fit choice of tone, it is often possible to think of a blend of
options in combination. There will be projects you work on that involve multiple chart assets,
104

multiple interactions, different pages and deeper layers. The mantra proposed by Ben
Shneiderman (1996), one of the most esteemed academics in this field, namely ‘Overview first,
details on demand’, informs the idea of thinking about different layers of readability and depth in
your visualisation work accessed through interactivity. Some of the chart types that you will meet
in Chapter 6 can only ever hope to deliver a gist of the general magnitude of values (the big, the
small and the medium) and not their precise details. A treemap, for example, is never going to
facilitate the detailed perceiving of values. In the example shown in Figure 3.10, showing S&P
500 stock, the area of each rectangular shape represents the size of market capital for each
company included. The colours indicate the change over the past 24 hours.
Figure 3.10 Finviz: Standard & Poor’s 500 Index Stocks (www.finviz.com)
Our perceptual system is quite poor at estimating scales of areas, as you will learn later. If you
wish to compare the size of one stock (e.g. Google, top left) with another (e.g. Amazon, top
middle) it will not be easy to make accurately such a judgement. However, you do get a sense
that they are both relatively large and a chart like this usually seeks only to give a general sense of
the hierarchy of values (big, medium, small) as well as prominent observations of colour (vivid
red vs vivid green). The clue is perhaps in the name – treemap – in that some charts often
provide multiple layers of detail, navigating from a broad understanding of how complex or
dense a system of content towards more detailed specific enquiries thereafter. In this case, as you
can see, features of interactivity exist allowing the user to hover over a given shape to reveal a
tooltip containing precise details as value labels.
In the Better Life Index, shown in Figure 3.11, the initial view is based around a series of charts
that look like flowers. This is attractive, intriguing and offers a nice single-page summary at a
glance. The task of reading the petal sizes with any degree of precision is hard but that is not the
intent of this first layer. The purpose is to achieve a balance between a form that attracts the user
and a function that offers a general sense of where the big, medium and small values sit within
the data. For those who want to read the values with greater accuracy, once again, they just need
to hover over the flower shapes to view an alternative representation of the same data in the form
of a bar chart.
Figure 3.11 OECD Better Life Index, by Moritz Stefaner and Dominikus Baur, Raureif GmbH
105

http://www.finviz.com

In both these examples the viewer’s task of perceiving the chart has been adapted from a general
feeling of the data towards a more precise reading of the values. Sometimes this initial ‘gateway’
layer is required as a primary view, to seduce your audience and/or to provide a big picture at a
glance (feeling), and then the audience move towards more perceptually precise displays of data
(reading). This is usually achieved through interaction or by any means of sequencing, perhaps
by navigating through the pages of a report or advancing through a slide deck.
106

Judging the Experience Offered by Your Visualisation
The experience offered by a visualisation influences the interpreting phase of understanding.
Whereas tone embodies a continuum, the judgement of the most suitable experience is more
distinct and concerns different methods of enabling interpretation: explanatory, exhibitory or
exploratory (Figure 3.12).
Figure 3.12 The Classifications of ‘Experience’
Explanatory visualisations offer an experience characterised by the visualiser taking responsibility
to present important observations and interpretations to help the viewer more quickly assimilate
the meaning of what is presented. I find quotation marks are emblematic of explanatory
visualisations, as they are associated here with a visualiser saying something.
The simplest, perhaps mildest method of creating an explanatory experience is through the
inclusion of simple devices that direct the eye’s attention towards key features of a display.
Visually emphasising values of most interest through the use of contrasting colour properties can
offer cues that establish the hierarchy of importance. Annotation properties like value labels,
captions or summaries can provide explicit textual commentary about key findings to accelerate
the interpretation.
An example of this kind of explanatory experience is shown in Figure 3.13, which was published
in an article reporting on protests across US schools (in November 2015) regarding the under-
representation of black students.
Here you see a scatter plot comparing the share of enrolled black students for different public
research universities (along the vertical y-axis) with the share of the college-age black populations
in the respective states (along the horizontal x-axis). With the protests beginning at the
University of Missouri, the chart uses red to highlight this data point as the primary item of
interest. Other notable colleges, as mentioned in the article, are emphasised using darker dots and
labels to illustrate useful comparisons. With additional visual overlays like the trend line and
dotted line indicating proportional representation, the viewer’s attention is drawn to the
implication of what it means to be positioned in different regions of this chart – is it good or bad,
typical or atypical?
A useful way to consider the role of an explanatory visualisation is to think how you would
verbally present key insights from any chart in person. What features would you point out as
being the most interesting? Which values would you mention, and which would you ignore? The
traits of a good explanatory visualisation are that it effectively does the job of communicating the
main features you would remark on if you were there. It can stand alone without the need for in-
person explanation, yet still beckons the viewer towards important interpretations.
Figure 3.13 Mizzou’s Racial Gap is Typical on College Campuses, by FiveThirtyEight
107

A more intensive example of an explanatory visualisation would be characterised by work that
enlightens through narrative sequences in the form of sophisticated articles, animations or
presentations. Some might describe this as ‘narrative’ visualisation. This is where the most
tangible demonstration of storytelling is relevant. One example that typifies this classification is
seen in a powerful video illustrated in Figure 3.14 through a selection of still images. The video
employs an animated graphic sequence to weave together a data-driven narrative describing issues
of wealth inequality in the USA. There is an affecting voiceover that verbally presents the main
insights of the subject at hand, delivered via a linear story that unfolds. As a viewer, you sit back,
listen and process what you are being told.
Common to any explanatory visualisations is a need for the visualiser to possess sufficient
knowledge – or have the skill and capacity to acquire sufficient knowledge – about the topic
being displayed. The visualiser needs to be able to identify the most relevant and interesting
insights to present to the viewer. Creating explanatory visualisations forces a visualiser to
challenge how well he or she actually knows a subject. If you cannot explain or articulate what is
insightful, and why, to others, then this probably means you do not know the reasons yourself.
Figure 3.14 Excerpts from Wealth Inequality in America, by YouTube user ‘Politizane’
108

Explanatory projects will mainly be for audiences who do not have the knowledge, capacity or
time to form for themselves the meaning of a visualisation. Furthermore, if you have something to
say, indeed if you have to say something, say it with an explanatory visualisation.
Exploratory visualisations differ from explanatory in that they are focused more on helping the
viewers or – more specifically in this case – the users discover and form their own interpretations.
Almost universally, these types of works will be digital and interactive in nature. I find the
question mark is emblematic of exploratory visualisations, as they are associated with a visualiser
helping a user answer a question.
The most basic level of exploratory visualisation provides simple interrogation and manipulation
of the data. You might offer your user the ability to filter a display to show only certain categories
of interest or switch the view to different data parameters.
An example of this type of visualisation is shown in an interactive project in Figure 3.15. It was
developed to allow users to explore different measures concerning the dimensional changes of
different wood species, over time, across selected cities of the world. There are no captions or
conclusions. There are no indications of what is significant or insignificant. There is no assistance
from the visualiser to help the user interpret the meaning of this data – what is ‘good’ or ‘bad’?
This project exists simply to provide a visual window into the subject through this data to enable
users to interact with the different indicators and selections offered to let them find features that
resonate and form their own interpretations.
The responsibility for then translating ‘what it means’, the essence of interpretation, is passed to
them. This kind of experience will only be suitable if the audience have the requisite knowledge
and motivation to form such interpretations themselves. Indeed, the assumption would be that
the users will be better equipped to do this than the creators.
Figure 3.15 Dimensional Changes in Wood, by Luis Carli (luiscarli.com)
109

A deeper exploratory experience goes beyond just offering means to interact and more towards
what might be described as offering a participatory or contributory experience. The prospect of
greater control, a deeper array of features and the possibility of contributing one’s own data to a
visualisation can be very seductive. Users are naturally drawn to challenges like quizzes and
projects that allow them to make sense of their place in the world (e.g. how does my salary
compare with others; how well do I know the area where I live?). Figure 3.16 shows the New
York Times’ so-called ‘Dialect quiz map’. This is just one contemporary example employing this
participatory approach to great effect.
In this case users are invited to complete 25 questions about their use of language terms in
different scenarios. Based upon their responses and the others gathered in the associated (and
ever-growing) study, the similarity or otherwise of their apparent dialect compared across the
USA is revealed graphically. This is a custom map display shaped by the contributions of the
participating users. It shows them who they are. You might think this outcome is more
characteristic of an explanatory experience, but the end state is only reached as a result of the
user’s participation. And even then, it is down to the user to interpret the meaning of the results
shown. There was not any one thing the visualiser wanted to say, rather there were many
thousands of things – the only way to impart this was by handing over control to the users to let
them discover for themselves.
Figure 3.16 How Y’all, Youse and You Guys Talk, by Josh Katz (New York Times)
110

The biggest obstacle to the success of an exploratory visualisation’s impact is the ‘so what?’ factor.
‘What do you want me to do with this project? Why is it relevant? What am I supposed to get
out of this?’ If these are the reactions you are getting from users, then there is a clear disconnect
between the intentions of your project and the experience (or maybe expectations) of those using
it.
Increasingly there is a trend for visualisation projects to blend different types of experiences into
the same overall project – the term ‘explorable explanations’ has been coined to describe them. A
project like ‘Losing Ground’ by ProPublica (Figure 3.17) is an example of this as it moves
between telling a story about the disappearing coastline of Louisiana and enabling users to
interrogate and adjust their view of the data at various milestone stages in the sequence.
Exhibitory visualisations are characterised by being neither explicitly explanatory nor
functionally exploratory. With exhibitory visualisations the viewers have to do the work to
interpret meaning, relying on their own capacity to perceive and translate the features of a
visualisation. I generally describe these visualisations as simply being visual displays of data and
find the ellipsis is emblematic of exhibitory visualisations, as it represents the idea of a visualiser
leaving the viewer to finish the task of gaining understanding.
Think of this type of experience in relation to exhibiting an artwork: it takes the interpretative
capacity of the viewer to be able to understand the content of a display as well as the context of a
display. In contrast to exploratory visualisation, for exhibitory pieces this is conducted just by
looking and thinking. But like exploratory experiences, exhibitory projects rely entirely on the
audience having the motivation and capacity to interpret.
You might wonder what the value is of an exhibitory visualisation. Sometimes the circumstances
of the audience encountering a visualisation do not require technical exploration or direct
explanation. If you have a very specific audience whom you know to be sufficiently
knowledgeable about the domain and the analysis you have provided, it might not be necessary
to emphasise any of the key insights as you would with an explanatory visualisation.
Furthermore, the extent of the analysis might be so narrow that there is no value in enhancing
the experience with interactivity. Indeed, it might not even be technically feasible to do so.
Figure 3.17 Losing Ground, by Bob Marshall, The Lens, Brian Jacobs and Al Shaw (ProPublica)
111

In Figure 3.18, there is an analysis of the top 100 highest earning athletes. This is an exhibitory
piece because it leaves you to form your own interpretation about what you see presented. In this
case, it is quite clearly about the lack of representation of any female athlete among these top 100
earners. It does not need to be spelt out any more explicitly with captions or comments. It speaks
for itself.
Sometimes a visualisation cannot speak for itself, but we can. I described earlier the imagined
scenario of presenting a chart to an audience to get you to imagine what features you would
point out and comment on. For explanatory visualisations you try to recreate this layer of insight
directly within the work itself. However, if this was a real scenario you might use an exhibitory
visualisation as your main prop and accompany this with your observations and gestures to
provide an overall explanatory experience. I would define this as an exhibitory visualisation but
with the understanding facilitated through an explanatory setting.
Furthermore, perhaps you are presenting a graphic as a figure within a written article or report.
In and of itself the visualisation does not explain things in a stand-alone sense, but instead exists
as a visual figure to reference from within and supplement the writing. The text therefore
provides the explanatory narrative.
Figure 3.18 Forbes: The World’s 100 Highest-paid Athletes, by Andy Kirk
112

Another common context for using an exhibitory visualisation might exist in the situation of
producing a visual for stakeholders who have directly requested you to create something for
them. They might not need to see anything other than the basic chart of data. They know what
they are looking for and how to find it. The problem is that many visualisation projects
mistakenly fall into the void of being exhibitory visualisations when they really needed to be
more supportively explanatory or functionally exploratory.
113

Harnessing Ideas
The second aspect of establishing your project’s vision offers an opportunity to harness
imagination by capturing your initial, instinctive ideas. These are the earliest seeds of any
thoughts you may have for what the eventual solution you are working towards might look like.
In Thinking Fast and Slow, author Daniel Kahneman describes two models of thought that
control our thinking activities. He calls these System 1 and System 2 thinking: the former is
responsible for our instinctive, intuitive and metaphorical thoughts; the latter is much more
ponderous, by contrast, much slower, and requiring of more mental effort when being called
upon. System 1 thinking is what you want to harness at this part of the first stage: what are the
mental impressions that form quickly and automatically in your mind when you first think about
the challenge you are facing?
You cannot switch off System 1 thoughts. Mental visualisations are what we instinctively ‘see’ in
our mind’s eye when we consider the subject or nature of a task. You will not be able to stop
them happening when thinking about a problem. Rather than stifling your natural mental habits,
this stage of the process presents the best possible opportunity to allow yourself space to begin
imagining.
What colours do you see? Sometimes instinctive ideas are reflections of our culture or society,
especially the connotations of colour usage. What shapes and patterns strike you as being
semantically aligned with the subject? This can be useful not just to inspire but also possibly to
obtain a glimpse into the similarly impulsive way the minds of your audience might connect with
a subject when consuming the solution.
Think back to the example shown in Figure 3.9 about political ‘buying power’. As a commonly
recognisable metaphor of wealth, using Monopoly pieces was an entirely reasonable way to
represent the data. Presenting this huge, imaginary pile on the lawn of the White House was
symbolically congruent with the subject involved. The visualisation in Figure 3.19 concerns the
wine industry, showing the top grape varieties grown. In the upper part of the graphic, the size of
production for each grape variety is shown using a bubble chart, which creates a metaphorical
representation of a bunch of grapes.
You can clearly see how this design might have been conceived from early ideas formed before
the data was even collected and analysed. Not only is the representation consistent with the
subject, but it also offers an immediately recognisable metaphor. Any viewer will make a seamless
connection between subject and form.
To help unlock your imagination it is useful to be influenced and inspired by the world around
you. Exposing your senses to different sources of influence can only help to broaden the range of
solutions you might be able to conceive. Research the techniques that are being used across the
visualisation field, look through books and see how others might have tackled similar subjects
portraying similar types of data. Outside of the visualisation sphere, consider other forms of
design or imagery: colours, patterns, shapes and metaphors from everyday life whose aesthetic
qualities you just like. Start a scrapbook or project mood board that compiles the sources of
114

inspiration you come across and helps you form ideas about the style, tone or essence of your
project. They might not have immediate value for your current project but may materialise as
useful for future work.
‘I focus on structural exploration on one side and on the reality and the landscape of opportunities in the other … I
try not to impose any early ideas of what the result will look like because that will emerge from the process. In a
nutshell I first activate data curiosity, client curiosity, and then visual imagination in parallel with experimentation.’
Santiago Ortiz, Founder and Chief Data Officer at DrumWave, discussing the role – and timing – of forming
ideas and mental concepts
It is important to acknowledge the boundaries of this activity. Influence and inspiration are
healthy: the desire to emulate what others have done is understandable. Plagiarism, copying and
stealing uncredited ideas are wrong. There are ambiguities in any creative discipline about the
boundaries between influence and plagiarism, and the worlds of visualisation and infographic
design are not spared that challenge.
Figure 3.19 Grape Expectations, by S. Scarr, C. Chan and F. Foo (Reuters Graphics)
115

Being influenced by the research you do and the great work you see around the field is not
stealing, but if you do incorporate in your work explicit ideas influenced by others, at the very
least you should do the noble thing and credit the authors, or, even better, seek them out and ask
them to grant you their approval. You do not have to credit William Playfair every time you use
the bar chart, but there are certain unique visual devices that will unquestionably be deserving of
attribution.
‘Look at how other designers solve visual problems (but don’t copy the look of their solutions). Look at art to see how
great painters use space, and organise the elements of their pictures. Look back at the history of infographics. It’s all
been done before, and usually by hand! Draw something with a pencil (or pen … but NOT a computer!). Sketch
often: The cat asleep. The view from the bus. The bus. Personally, I listen to music – mostly jazz – a lot.’ Nigel
Holmes, Explanation Graphic Designer, on inspirations that feed his approach
Sketching is also an important habit to develop. It does not require strong artistic talent, but it
can prove to be a useful method for extracting ideas out of your mind and quickly capturing
116

them in visual form. Figure 3.20 shows a montage of various sketches and sources of inspiration
that influenced the design concept of a project visualising the spread of the #MeToo movement.
‘It is easy to immerse yourself in a certain idea, but I think it is important to step back regularly and recognise that
other people have different ways of interpreting things. I am very fortunate to work with people whom I greatly
admire and who also see things from a different perspective. Their feedback is invaluable in the process.’ Jane Pong,
Data Visualisation Designer
For some people, the most fluent and efficient way to ‘sketch’ is through their software
application of choice rather than on paper. Regardless of the medium you use, sketching is useful
when you are working with collaborators or for stake-holders as a means of discussing ideas,
getting input and other’s thoughts on the brief. I find it particularly helpful when trying to
conceive innovative solutions to unusual or particularly complex challenges. It may be that my
eventual solution looks nothing like my rudimentary sketches, but it gives me a way to cycle
rapidly through iterations of concepts that may be worth exploring later on.
‘I draw to freely explore possibilities. I draw to visually understand what I am thinking. I draw to evaluate my ideas
and intuitions by seeing them coming to life on paper. I draw to help my mind think without limitations, without
boundaries. The act of drawing, and the very fact we choose to stop and draw, demands focus and attention. I use
drawing as my primary expression, as a sort of functional tool for capturing and exploring thoughts.’ Giorgia Lupi,
Co-founder and Design Director at Accurat
There are limits to the value of ideas and the role they should be allowed to play. After all, data is
your raw material, your ideas are not. It may be that your ideas are ultimately incompatible with
the properties of the data you are working with, in which case you should just let go and move
on.
Early sparks of inspiration in your thinking should be embraced, but do not be precious or
stubborn. Always maintain an open mind and recognise that ideas have a limited role. This is
why harnessing is the appropriate term used to describe this activity.
Figure 3.20 MeTooMentum, by Valentina D’Efilippo (design) and Lucia Kocincova
(development)
117

Finally, it is worth noting the diplomatic prospect of taking on board other people’s ideas. One
of the greatest anxieties I face in my client work comes from working with stakeholders who are
unequivocally and emphatically clear about what they think a solution should look like – from
the very start.
Often your involvement in a project may arrive after these ideas have already been formed,
during which time they have shaped the brief issued to you by the stakeholders (‘Can you make
this, please?’). This is where your tactful but assured ‘communicator’ hat comes to the fore. The
ideas presented to you may be reasonable and well intended, but it is your responsibility to lead
the creation process. You can welcome input in the form of proposed concepts but, as with the
limitations of your own ideas, there will be other factors with a greater influence: the nature of
the data, the type of curiosities you are pursuing, the essence of the subject matter, and the
nature of the audience, among many other things. These will be the factors that ultimately
dictate whether any early vision of potential ideas ends up being of value.
118

Summary: Formulating Your Brief
This chapter commenced the opening stage of the design process concerned with initiating,
defining and planning the requirements of your work.
119

Context
The first section looked at issues around context, specifically about the importance of defining
the motivating curiosity and identifying all the circumstances that will shape your project. These
included factors such as follows.
People
Stakeholders: Who is the ultimate customer? Who are the influencers, interferers, subject
matter experts (SMEs)?
Audience: What is their knowledge (informed or ‘layperson’)? Receptive or indifferent?
Visualiser(s): What skills/knowledge are possessed? Individual or team?
Constraints
Timescales: When is it due? When can you start? Milestones? Available duration?
Pressures: Financial? Political? Cultural? Environmental?
Design: Style restrictions (colour, type, logo), size?
Technological: What software, hardware, infrastructure exist? Platform compatibility?
Deliverables
Setting: Rapid or prolonged? Consumed remotely or live?
Medium: What is the intended output format?
Quantity: How many outputs are being produced?
Frequency: One-off project or a regular/repeated task?
120

Vision
The second section considered the vision of your work, firstly looking at its core purpose. What
is it for? What are you trying to accomplish? Depending on your defined purpose, you will need
to pursue the right balance in the tone and experience through which understanding will be
facilitated:
Tone: The distinction between ‘reading’ and ‘feeling’ data.
Experience: The difference between ‘explanatory’, ‘exhibitory’ and ‘exploratory’
visualisations.
Finally, you learnt about the value and limitations of harnessing ideas. What mental images,
shapes, forms and keywords instinctively come to mind when thinking about the subject matter
of this challenge? What influence and inspiration can you source from elsewhere that might start
to shape your thinking?
121

General tips and tactics
Not all circumstantial factors can be defined, nor will they be stable throughout. Certain
things may change in definition, some things will emerge, some defined things
will need to be reconsidered, some things are just always constraint-free.
Notes are so important to keep about any thoughts you have had that express the nature of
your curiosity, articulation of purpose, any assumptions, things you know and do not know,
where you might need to get data from, who the experts are, questions, things to do,
issues/problems, wish lists, etc.
Keep a ‘scrapbook’ (digital bookmarks, print clippings) of anything and everything that
inspire and influence you – not just data visualisations. Log your ideas and inspire yourself.
This stage is about ambition management and it will be to your benefit if you treat it with
the thoroughness it needs. The impact of any corners being cut here will be amplified later
on.
What now? Visit book.visualisingdata.com
EXPLORE THE FIELD Expand your knowledge and reinforce your learning about working with data through this
chapter’s library of further reading, references, and tutorials.
TRY THIS YOURSELF Revise, reflect, and refine your skill and understanding about the challenges of working
with data through these practical exercises.
SEE DATA VISUALISATION IN ACTION Get to grips with the nuances and intricacies of working with data in
the real world by working through this next instalment in the narrative case study and see an additional extended
example of data visualisation in practice. Follow along with Andy’s video diary of the process and get direct insight
into his thought processes, challenges, mistakes, and decisions along the way.
122

http://book.visualisingdata.com

4 Working With Data
In Chapter 3, the design process was initiated, with early attention paid to defining matters of
context and vision. The discussion about context looked at identifying the source curiosity and
the circumstances that would influence the conditions of your work. Vision was a more forward-
facing glance towards the purpose of your work, thinking about how you might accomplish the
type of understanding you are seeking to facilitate for your audience. We closed the chapter by
looking at the value of harnessing ideas, through sketching and research.
In this second chapter, your thinking will switch to the practical mechanics of working with data.
In this chapter you will be respecting its role as the critical raw material of this process, learning
how to nurture its potential but equally being prepared for it to frustrate you, on occasion. You
will work through four distinct activities to develop a close acquaintance with it, as follows:
Data acquisition: Sourcing and gathering the raw material.
Data examination: Familiarising yourself with the key physical properties and condition of
your data.
Data transformation: Refining your data through modification and consolidation.
Data exploration: Using exploratory analysis and research techniques to discover insights.
I often encounter people who declare their love for data. Data does indeed have the capacity to
earn and merit love, but I personally do not love it all that much. Data always demands so much
attention yet consistently seems to conspire against you. You do not need to love data but,
equally, you should not fear data – you should just respect it.
Some readers might feel confident working with data but might not have much direct experience
when working with it in the context of a visualisation challenge. For those readers I want to
provide you with a strong appreciation of the influence data has on your editorial and design
thinking and dispel any sense that this is especially complicated. For those with more experience
and confidence with this topic, this chapter might help to reinforce some of the ways of thinking
about the impact of your data on your visualisation project.
123

4.1 Step 1: Data Acquisition
The first step when working with data in a visualisation project is to get the data. There are
several distinct origins and methods involved in acquiring data. Some are characterised by your
having to do most of the work yourself, others involve people making it available for you to
access in different ways. In each of these cases you need to be assured about the reliability of the
data you are gathering, whether it is you or others who have curated it. As discussed in Chapter 2
when describing the importance of ‘trustworthiness’, there may be collection issues creating
inaccuracies and biases that can affect the quality of your data at source. You need to be
discerning in the degree of trust you place in it, at least to begin with, until you have a chance to
examine it more closely.
Supplied: The simplest method for acquiring data involves getting it from somebody else. In
projects where you have been commissioned by a stakeholder (manager, client or colleague), you
will often be issued with the data you need. The extent of your efforts may therefore just involve
saving an attachment from an email. You should, however, still undertake as much background
research as possible about the original source and collection method used to form the data you
have been given.
System download: When working in organisations, there will inevitably be many occasions
where the data will come from internal reporting tools or exports from corporate systems. There
are many organisations offering publicly accessible data to download through the Web,
sometimes through interfaces that let interested users construct detailed queries and download
structured data customised to their need. There is an increasing marketplace for data.
Web scraping: This involves using special programs to extract structured and unstructured items
of data published on web pages. For example, you may wish to extract information from a hotel
chain, product details from the IKEA website or data about the history of the Winter Olympics
on Wikipedia. Depending on the tools used, you can often set routines in motion to extract data
across multiple pages of a site based on the connected links that exist within it. This is known as
web crawling. An important consideration to bear in mind with any web scraping or crawling
activity concerns rules of access and the legalities of extracting the data held on certain sites.
Always check – and respect – the terms of use before undertaking this.
APIs: Certain specialist websites or services offer an API (Application Programming Interface) to
enable people to access streams of data. A popular example would be the access provided to real-
time data on topics like air quality, traffic disruptions and London Underground passenger levels
provided by Transport for London (TfL), which encourages people to develop bespoke software
applications. Many commercial services now offer extensive sources of curated and customised
data that would otherwise be very complex to gather or difficult to obtain. An example might
include large, customised extracts from social media platforms like Twitter based on specific
keyword criteria.
Primary collection: If the data you need does not exist in digital form, you might need to
consider gathering primary data. This is where you collect observations or capture measurements
about bespoke phenomena specific to your needs. These might include:
124

Transcribing a political speech from unstructured video or audio recordings to explore
patterns of sentiment and/or rhetoric.
Designing a participant questionnaire to collect relevant data about a research study.
Using measurement devices to track fitness activity or health information over a period of
time.
This type of data gathering activity can be expensive in time and cost. The benefit is that you will
be able to control carefully the collection of the data to ensure its value is optimised for your
needs.
‘Don’t underestimate the importance of domain expertise. At the Office for National Statistics (ONS), I was lucky in
that I was very often working with the people who created the data – obviously, not everyone will have that luxury.
But most credible data producers will now produce something to accompany the data they publish and help users
interpret it – make sure you read it, as it will often include key findings as well as notes on reliability and limitations
of the data.’ Alan Smith OBE, Data Visualisation Editor, Financial Times
Data foraging: If the data you need does not exist in a convenient single form or location, you
may need to forage for it. This usually involves manually sourcing relatively small amounts of
disparate or dispersed data values. For example, suppose you wanted to compare the lifetime
costs associated with a range of different mobile phones to help you decide which model or tariff
to go with. You would find the information on the Web, locate the specific data items and values
you need, collect them in a spreadsheet, and then repeat for each model or tariff to build up your
table of comparable data. Sometimes data foraging involves extracting data from documents,
such as pdf files. There are tools that assist in accelerating this task, enabling you seamlessly to
extract tables of data and convert them into more usable Excel or CSV (Comma-Separated
Values) formats.
125

4.2 Step 2: Data Examination
Once you have acquired your data – whether this is all of it or just a starting point – the second
step is to examine it thoroughly. Before you choose what meal to cook, you need to know what
ingredients you have, how much and in what condition. The same applies to data. Before you
can contemplate any design thinking you first need to familiarise yourself fully with the physical
characteristics and state of your data. Examining your data specifically involves learning about the
types of data you have, the size and range of values held, and its condition.
‘Data inspires me. I always open the data in its native format and look at the raw data just to get the lay of the land.
It’s much like looking at a map to begin a journey.’ Kim Rees, Head of Data Experience Design at Capital One
126

Data Types
Before we look specifically at a classification for different types of data, we first need to establish
the difference between types of tabulation. A normalised form of tabulated data offers the most
detailed, granular form of data, organised by variables (columns or fields) and items (rows or
records). The table in Figure 4.1 is a simple, small-scale example of a normalised dataset. Each
column in the table represents a different variable describing movies in each film series.
Figure 4.1 Example of a Normalised Dataset
In contrast, cross-tabulated datasets present a summarised form of normalised data, displaying
values that are the result of statistical operations such as grouped totals, maximums and
minimums. If normalised data is sometimes colloquially described as being ‘tall and thin’, cross-
tabulated data is ‘short and fat’. In Figure 4.2, you will see a cross-tabulated form of the data
shown in the normalised table in Figure 4.1.
Figure 4.2 Example of a Cross-tabulated Dataset
If you are working with data that exists in cross-tabulated form and you do not have access to the
underlying normalised data, the avenues of potential analysis will be reduced in scope. As you
can see in Figure 4.2, you have a far reduced set of data items and values to work with. There is
no granularity, no sense of the distribution and variety of values that lie beneath these statistical
aggregates.
127

The type of data tabulation is also influential when conducting analysis and generating charts.
Certain tools need data to be shaped in specific ways. For example, charting in Excel is usually
performed by linking a selected chart template to a range of cells that are usually organised in
cross-tabulated form. Working with a tool like Tableau involves connecting to normalised data
and constructing a chart from the ground up. I find it always preferable to work with normalised
data, where you have far more detail that you can then choose to aggregate should you wish. The
key is that you have the choice.
For the purpose of this chapter, we will principally look at working with data in normalised
form.
Next you need to develop a thorough understanding of your data types. Also commonly referred
to as levels of data or scales of measurement, data types define the nature of the values held under
each variable and about each item in your dataset. The different types of data you might have
will have a major influence over several key aspects of the design process, such as:
determining the type of statistical analysis methods you can use;
shaping the editorial perspectives you will pursue;
filtering the specific chart types you can or cannot use;
informing the appropriateness of your colour associations; and
guiding composition decisions on size, placement and layout.
In simple terms, data types are distinguished by being either qualitative or quantitative in nature.
Below this high-level view there are more nuanced but crucial distinctions that need to be
understood.
The most useful taxonomy I have found to distinguish types of data, in the context of developing
a visualisation, comes from psychologist researcher Stanley Stevens. He devised the NOIR
classification which represents: Nominal, Ordinal, Interval and Ratio. The order of these distinct
types is deliberate, as each subsequent level of measurement embodies a certain increase in
precision. This is a particularly relative approach to working with data in the context of social
research. I find it useful to extend the acronym, adding a leading ‘T’ – for Textual – to reflect the
contemporary experiences of working with a greater variety of qualitative data.
‘Absorb the data. Read it, re-read it, read it backwards and understand the lyrical and human-centred contribution.’
Kate McLean, Smellscape Mapper and Senior Lecturer Graphic Design
Textual data is qualitative and characteristically ‘human’, usually existing in unstructured form
like passages of text or sections of a report. Examples of textual data might include:
Responses to ‘Any other comments?’ in a questionnaire.
A newspaper write-up about a football match.
The abstract for an academic research article.
The description of a product on Amazon.
The transcript of a speech given by a politician.
When working with textual data you will typically need to transform it to extract certain
properties and relational characteristics in some way, such as counting the frequency of certain
128

keywords or using natural language processing techniques to derive sentiment classifications.
Nominal data is another form of qualitative data and the second distinct data type. Nominal
data exists as categories, which offer means of separating different values and grouping similar
values together. Examples of nominal data might include:
The gender of a survey participant.
The meals available on a restaurant menu.
The name of your country of birth.
The genre of a movie.
The sport events in the Olympics.
Nominal data does not exclusively mean text-based data; nominal values can be numeric. For
example, a student ID number is a categorical device used to uniquely identify all student
records. The shirt number of a footballer is a way of helping teammates, spectators and officials
recognise each unique player. You might find measurements of gender captured as 1 (male), 2
(female) and 3 (other), but these numeric values should not be considered quantitative values –
adding ‘1’ to ‘2’ does not equal ‘3’, for gender.
Another characteristic of nominal data is the potential for a hierarchical relationship to exist
between two or more major and sub-categorical variables. For example, a major category holding
details of ‘Country’ and a sub-category holding ‘Airport’; or a major category holding details of
‘Industry’ and a sub-category holding details of ‘Company Names’. Recognising this type of
relationship will become important when deciding how to portray your data using certain chart
types.
Ordinal data is the third qualitative data type. Unlike nominal data, ordinal data is characterised
by their being some notion of order in the relationship between different categorical values.
Examples of nominal data might include:
The options available to answer a survey question based on the extent to which you agree or
disagree with a statement.
General temperature observations from very hot to very cold.
The size of T-shirts from XS to XXL.
The rank of a police officer.
The gold, silver and bronze medal categories at the Olympics.
Recognising a categorical variable as being ordinal rather than nominal in nature will be
particularly relevant when you make decisions about classifying values using different colour
scales.
Interval data is a quantitative measurement defined by difference on a scale but not by relative
scale. Examples of interval data might include:
The body mass index for measuring obesity.
The forecasted temperature in °C.
The latitude and longitude coordinates of a given location.
129

The principal characteristic of interval data is that the absolute difference between two values is
meaningful, but any arithmetic operation, such as multiplication, is not. For example, the
absolute difference between 15°C and 20°C is the same difference as between 5°C and 10°C.
However, the relative difference between 5°C and 10°C is not the same as the difference between
10°C and 20°C (where in both cases you multiply the lower value by two or increase by 100%).
This is because the zero state of an interval scale is not a true zero value, it is just an established
scale position. A temperature reading of 0°C does not mean there is no temperature; it is a
quantitative scale for measuring relative temperature.
Ratio data is the second quantitative type of data and the one you are most likely to encounter in
most visualisation project situations. Examples of ratio data might include:
The age of a survey participant in years.
The forecasted amount of rainfall in millimetres.
The estimated budget for a research grant proposal in GBP (£).
The number of sales of a book on Amazon.
The distance of the winning long jump at the 2016 Olympics in metres.
Ratio data values are numeric measurements with significant properties of both difference and
scale. The absolute difference in age between a 10- and 20-year-old is the same as the difference
between a 40- and 50-year-old. The relative difference between a 10- and a 20-year-old is the
same as the difference between a 40- and an 80-year-old (‘twice as old’). Unlike interval data, a
zero value for a ratio variable is a true zero, meaning there is no amount.
There are other important data-type distinctions. One key distinction is between values that are
discrete or continuous. This distinction is particularly influential in how you might visually
represent the relationship between such values. Discrete data is associated with all classifying
measurements that have no ‘in-between’ state. This applies to all qualitative data types and any
quantitative values for which only a whole is possible. Examples include the heads or tails
outcome of a coin toss, the days of the week and the number of seats in a cinema.
In contrast, continuous measurements can hold the value of an in-between state and indeed any
value between the natural upper and lower limits, if such fine degrees of measurement detail are
possible. Think of them as moment-in-time measurements, with examples including height,
weight and temperature.
There are some data-type classifications that are hard to define on the TNOIR scale, due to
special or varied characteristics that are not universally compatible with this taxonomy. One such
example would be time-based data, which can shift across the TNOIR classification depending
on the format and purpose of using this data (e.g. for grouping, for labelling or for calculations of
duration).
Whereas most quantitative measurements you will deal with are based on a linear scale, there are
exceptions. Variables about the strength of sound (decibels) and magnitude of earthquakes
(Richter) are actually based on a logarithmic scale. An earthquake with a magnitude of 4.0 on the
Richter scale is 100 times bigger and 1000 times stronger (based on the amount of energy
released) than an earthquake of magnitude 2.0. Logarithmic values, as well as other
mathematically derived types of data, are often still considered as ratio variables but are
130

distinguished as being non-linear scaled variables.
131

Data Size: Amount and Range
Once you have established an understanding of the different types of data, you can switch your
examination towards the shape and size of this data, looking at the quantitative attributes across
all variables and for all items. The main questions you will ask of your data include:
For quantitative variables (interval or ratio), what is the lowest and the highest value in each
case?
In what format are the numeric values presented (i.e. how many decimal places or comma-
formatted)?
For a categorical variable (nominal or ordinal), how many different values are held?
If you have textual data, what is the maximum and minimum character length or word
count?
Statistical methods will assist in describing further physical characteristics. Here are some analyses
you might find useful at this stage. These are not the only methods you will ever need to use, but
it is likely they will be the most common.
Frequency distribution: Applied to quantitative values to learn about the shape of the
distribution of values.
Measurements of central tendency: These describe the summary attributes of a group of
quantitative values, including: the mean (the average value); the median (the middle value if
all quantities are arranged from smallest to largest); the mode (the most common value).
Frequency counts: Applied to categorical values to understand the frequency of different
instances.
Measurements of spread: These are used to describe the dispersion of values above and below
the mean:
Maximum, minimum and range: the highest and lowest and magnitude of spread of
values.
Percentiles: the value below which x% of values fall (e.g. the 20th percentile is the
value below which 20% of all quantitative values fall).
Standard deviation: a calculated measure used to determine how spread out a series of
quantitative values are.
As mentioned at the end of the previous chapter, it is worth repeating that your ideas may
stimulate certain design thinking, but the shape and size of your data will drive it. The
quantitative characteristics of your data will have a strong bearing on what may or may not
qualify as a suitable design solution. For example, look at the shape of data in the ‘Better Life
Index’ project that you saw earlier. As you can see in Figure 4.3, the analysis of the quality of life
covers 38 OECD member states and uses a flower structure for each country comprising 11
petals. Each petal represents a different quality of life indicator.
Figure 4.3 OECD Better Life Index, by Moritz Stefaner and Dominikus Baur, Raureif GmbH
132

What if there were 25 indicators? Or just 3? What if the analysis was expanded to cover 150
countries? Would the idea of using the flowers as a metaphor for quality of life still work for such
different dimensions of data? Arguably not.
‘My design approach requires that I immerse myself deeply in the problem domain and available data very early in the
project, to get a feel for the unique characteristics of the data, its “texture” and the affordances it brings. It is very
important that the results from these explorations, which I also discuss in detail with my clients, can influence the
basic concept and main direction of the project. To put it in Hans Rosling’s words, you need to “let the data set
change your mind set”.’ Moritz Stefaner, Truth & Beauty Operator
Another relevant example of the impact of the shape of data can be demonstrated in the
‘Spotlight on profitability’ graphic in Figure 4.4. Although there are several movies that slightly
exceed the maximum scale value of $1000 million, there is one movie, Avatar, that emphatically
bursts through the spatial ceiling of the chart. As an isolated but extreme outlier, in this case, it
was accommodated using a bespoke approach. As you can see, the towering shape of this movie
trespasses above into the space afforded by two empty rows. This solution emphasises the value’s
exceptional quality.
Figure 4.4 Spotlight on Profitability, by Krisztina Szücs
How do you elegantly handle quantitative measures that have hugely varied value ranges?
Accommodating the full range of your data values into a single chart scale can have a distorting
impact on the space it occupies. Normally, you would not have the luxury of being able to apply
a customised approach to handling such diverse data values. At this stage, we are not so
concerned about working out a design solution, rather the general point is to become aware of
133

the existence of the ‘Avatars’ in your data that will eventually have an impact on your design
thinking.
134

Data Condition: Quality and Representativeness
Undiscovered and unresolved issues around the quality of your data will undermine the trust in
and the accuracy of your work. You will need to discover and address these issues during this
stage of the process. Features to look out for may include:
Missing values: Are empty cells assumed to be of no value (zero/nothing) or no
measurement (n/a, null)? This is a subtle but important difference.
Erroneous values: Typos and any values that clearly look out of place (such as a gender value
in an ‘age’ column).
Inconsistencies: Capitalisation, units of measurement, value formatting.
Duplicate records.
Expired values: Values that might have elapsed in their current relevance or accuracy, like
someone’s age or any statistic that would be expected to have subsequently changed.
Uncommon system characters or line breaks.
Leading or trailing spaces: A subtle but particularly evil issue!
Date issues around format (dd/mm/yy or mm/dd/yy) and basis (systems like Excel’s base
dates on daily counts since 1 January 1900, but not all do that).
An extension of examining the condition of your data is to consider how representative it is. This
is, in part, about appreciating what is missing, not just in value terms, but more in relation to the
items of data you have and do not have about your subject matter. You need to be healthily
sceptical about your data, seeking constant reassurance of its quality and condition, so you can be
confident that what you are presenting is legitimate. Inaccuracies in judging what your data truly
represents can have an even greater impact on trust than the damage caused by individual
elements of missing or inaccurate data. The questions you need to ask of your data are:
Does it represent genuine observations about a given phenomenon or is it influenced by the
limitations of a collection method?
Does your data reflect the entirety of a particular phenomenon, a recognised sample, or
maybe even an obstructed view caused by hidden limitations in the availability of data
about that phenomenon?
It is simply not reasonable to expect always to have access to the entirety of data about your
subject matter. Most projects you will work with will resemble a sample of a population. This is
not an obstacle to progressing with a visualisation, it is about caution rather than cessation. You
must be clear about the basis on which your sample is formed and how you might faithfully
represent and communicate this to your audience. You might even exploit the gap that exists
between the data you have and all the data that could exist about the phenomena to shine a light
on what is missing. Make that your key focus.
135

4.3 Step 3: Data Transformation
Once you complete your examination of your data you will have a good idea about what actions
may be needed to transform your data. This is to prepare it for the analysis and charting steps
you will soon move on to. What do you need to do to get it into shape and fit for purpose?
It is worth noting that transforming your data is a prime example of a step in this process that
may start now but will likely continue right the way through to the latter stages of your design
thinking. As you reach that stage you will often encounter the need to tweak further the shape
and size of your data.
In accordance with the desire for trustworthy design, any modifications or enhancements you
apply to your data need to be noted and potentially explained to your audience. You must be in a
position to share the thinking behind any significant assumptions, calculations and conversions
you have made.
There are three different types of potential activity involved in transforming your data: cleaning,
creating and consolidating.
‘When I first started learning about visualisation, I naively assumed that datasets arrived at your doorstep ready to roll.
Begrudgingly I accepted that before you can plot or graph anything, you have to find the data, understand it, evaluate
it, clean it, and perhaps restructure it.’ Marcia Gray, Graphic Designer
Cleaning: I have already discussed the importance of data quality. There is no need to revisit the
list of potential issues to look out for, but the point is that now is the time to begin to address
these.
There is no single approach for how best to conduct data cleaning. Some issues can be resolved
through a straightforward ‘find and replace’ (or remove) operation. Other tasks might be much
more intricate, requiring manual intervention, often in combination with inspection features like
sorting or filtering, to find, isolate and modify any problem values.
A further part of cleaning your data involves eliminating what you do not need. Any variable or
items of data that serve no ongoing value will take up space and attention. My tactic is usually to
gather as much data as I can initially and then remove or at least archive it later to help reduce
the clutter.
Remember, also, to keep backups. Before you undertake any transformation, make a copy of
your dataset. After each major iteration, save further copies. It can be useful to preserve your
original unaltered data so you can easily return to that state should you ever need to.
Creating: The most substantial transformation work often comes in the form of creating new
data from existing values. This task is something I refer to as the hidden cleverness, where you
expand your data to form new calculations and derive new groupings or any other mathematical
treatments. In doing so you broaden the range of analytical options open for you to explore.
There are unlimited different approaches to doing this depending on the data you have and what
you need from it, though they might at least include:
136

Creating percentage calculations based on existing quantities.
Creating a calculation to establish a rolling 12 monthly total.
Using ‘start date’ and ‘end date’ values to calculate the duration in days.
Converting absolute quantities associated with different geographic locations into ‘per
capita’ values based on population numbers in each.
Using logic-based formulae to create new categorical values out of quantities, such as
checking if an age value is under 18, in which case the ‘Age group’ value would be ‘Child’,
otherwise ‘Adult’.
I mentioned earlier in this chapter how sometimes your data does not exist in a tabulated form,
but instead in an unstructured, qualitative document. In these cases you may choose to derive
reasonable categorical or quantitative values from the original form. For example, when
performing categorical transformations from textual data, you might seek to:
Identify keywords or summary themes from text and convert these into categorical
classifications.
Identify and flag up instances of certain cases existing or otherwise (e.g. X is mentioned in
this passage).
Identify and flag up the existence of certain relationships (e.g. A and B were both
mentioned in the same passage, C was always mentioned before D).
Use natural language processing techniques to determine sentiments, to identify specific
word types (nouns, verbs, adjectives) or sentence structures (around clauses and punctuation
marks).
With URLs, isolate and extract the different components of website address and sub-folder
locations.
Similarly, for quantitative transformations of textual data, here are some common approaches:
Calculate the frequency of certain words being used.
Analyse the attributes of text, such as total word count, physical length, potential reading
duration.
Count the number of sentences or paragraphs, derived from the frequency of different
punctuation marks.
Position the temporal location of certain words/phrases in relation to other words/phrases
or compared with the whole (e.g. X was mentioned at 1m 51s).
Position the spatial location of certain words/phrases in relation to other words/phrases or
compared with the whole.
Figure 4.5 presents a graphic showing an analysis of the profanities used by CEOs from a review
of recorded conference calls over a period of time. This work demonstrates two ways of utilising
textual data in visualisation. Firstly, the visualiser has extracted categorical classifications and
quantitative measurements to show the trends in usage over time and to compare the frequency
of different, certain swear words being used. Secondly, we see annotated captions lower down the
page which preserve the original qualitative form of the textual data, without any transformation
applied. This provides the viewer with some examples of the context of a swear word shown
within a sample passage of the original transcript.
137

Figure 4.5 Graphic Language: The Curse of the CEO, by David Ingold and Keith Collins
(Bloomberg Visual Data) and Jeff Green (Bloomberg News)
Handling textual data will always create more work and you will need to judge the reward vs
effort of such activity: how much effort will I need to expend in order to extract usable, valuable
content from the text? Some of the approaches you might use will be quite straightforward to
undertake, but others are more complicated and require sophisticated tools to assist with more
algorithmic techniques.
‘Although all our projects are very much data driven, visualisation is only part of the products and solutions we create.
This day and age provides us with amazing opportunities to combine video, animation, visualisation, sound and
interactivity. Why not make full use of this? Judging whether to include something or not is all about editing: asking
“is it really necessary?”. There is always an aspect of gut feel or instinct mixed with continuous doubt that drives me in
these cases.’ Thomas Clever, Co-founder CLEVER°FRANKE, a data-driven experiences studio
A further transformation task involves converting the layout and format of your data. Formatting
data for its appearance in printing is not usually compatible with how we need it arranging when
analysing or visualising it. For example, you might need to go through a spreadsheet and
‘unmerge’ any cell values that are formatted across several table columns. Sometimes you might
138

encounter visual formatting like background shading or the colouring of a font to represent a key
status. This might be useful when reading the table, but few tools will be able to ‘see’ these
attributes – they will need to exist as actual values.
Consolidating: There will be occasions where you may seek to source and introduce additional
data to expand (more variables) or append (more items) your data further in order to enhance its
analytical potential:
Expand: This is where you want to broaden the values of data you have to work with. For
example, if you have location data at the level of detail of country, you might want to group
and aggregate your analysis to a higher level. You could therefore source continent
groupings, so you can then create this hierarchical relationship, giving you the option to
analyse at both levels.
Append: This might occur if your original dataset is no longer representative of the most up-
to-date state and newer data items are available for you to access. If you are doing an
analysis about movies, as soon as another week elapses new movies will be released, and your
existing data will no longer have the most current items.
You may also use this moment in the process to start thinking about sourcing other assets that
might enhance your data representation options later on. Perhaps the elegance of your work will
be improved through possible access to photo-imagery, written anecdotes, video clips or physical
media?
Even though it will be a while until we reach the design thinking stage, it is useful to start
thinking about this as early as possible in case the collection of these additional assets requires
significant time and effort. It might also reveal to you any particular obstacles involved in
obtaining permissions for use or blockages to sourcing high-quality media. If you know you are
going to have to do this asset gathering, do not leave it too late – reduce the possibility of such
stresses by acting early.
139

4.4 Step 4: Data Exploration
140

Widening the Viewpoint
The examination step was about forming a deep acquaintance with the physical properties of
your data. In doing this you will have found reasons and ways to enhance the data by
transforming it. Next, you ideally need to build in some time to interrogate your data further to
give yourself every opportunity of discovering the potential insights and qualities of
understanding your data may provide. This is where we embark on visual exploration, an activity
that is especially important when you are working with big datasets and/or datasets that are
unfamiliar to you.
Undertaking data exploration involves the use of both statistical and visual techniques to help
you go beyond just looking at data and to begin to start seeing it. What answers to your overriding
curiosity can you find? What other enlightening features of your data can you unearth?
Sometimes, it might present new discoveries that will motivate you to pursue different avenues of
enquiry.
‘After the data exploration phase you may come to the conclusion that the data does not support the goal of the
project. The thing is: data is leading in a data visualisation project – you cannot make up some data just to comply
with your initial ideas. So, you need to have some kind of an open mind and “listen to what the data has to say” and
learn what its potential is for a visualisation. Sometimes this means that a project has to stop if there is too much of a
mismatch between the goal of the project and the available data. In other cases this may mean that the goal needs to
be adjusted and the project can continue.’ Jan Willem Tulp, Data Experience Designer
Overall, you are trying to widen your viewpoint and be truly acquainted with the full potential of
what your data is offering you. As I have emphasised about the benefit of this whole design
process, to make the best decisions you first need to be aware of all the options. Data exploration
is about broadening your awareness of the potentially interesting things you could show your
audience about your subject. Making choices about ultimately which ones you will pursue comes
next.
To frame this discussion I find it useful to refer to the transcript of a news briefing given by the
then US Secretary of Defense, Donald Rumsfeld, in February 2002. This was his infamous
‘known knowns’ statement:
Reports that say that something hasn’t happened are always interesting to me, because as we
know, there are known knowns; there are things we know we know. We also know there are
known unknowns; that is to say we know there are some things we do not know. But there
are also unknown unknowns – the ones we don’t know we don’t know. And if one looks
throughout the history of our country and other free countries, it is the latter category that
tend to be the difficult ones.
There has been much said about the apparent lack of elegance in the language used and criticism
of the muddled meaning in this passage, but I disagree. I think it is probably the most concise
way he could have articulated this, at least in written or verbal form. The essence of this
statement, as presented in visual form in Figure 4.6, was to distinguish between an awareness of
141

what is knowable about a subject from the status of acquiring this knowledge. When thinking
about the role of data exploration, there is much we can gain from this concept.
The known knowns are aspects of understanding present in your data about your subject that you
are aware of: you know about these things. When working with familiar data it can be tempting
to consider the known knowns as being the only relevant perspectives to build your analysis
around. Indeed, your knowledge of the subject may have influenced the nature of the curiosity
you are pursuing.
Figure 4.6 Making Sense of the Known Knowns
However, what you know to be interesting about a subject may only represent a quite narrow
viewpoint. It is therefore important to be willing to seek other interesting qualities of your data,
especially if it or the subject is unfamiliar. Indeed, your known knowns may be entirely absent if
you know nothing about a subject and it can be helpful, on occasion, not to be burdened by
existing knowledge.
Although there may contexts where some unknown knowns exist – things you did not realise you
knew about a subject – the most important matter to address is the known unknowns and the
even more elusive unknown unknowns. Analytical tactics are needed to help plug these gaps as far,
as deep and as wide as possible. You need the capacity to convert unknowns into knowns. In
doing so it will optimise the viewpoint of your subject and better support your judgement about
whether to continue with, to refine or to redefine your origin curiosity.
142

Exploratory Data Analysis
In visualisation, the task of addressing the unknowns you may have, as well as substantiating the
knowns that already exist, involves the use of exploratory data analysis (EDA). This integrates
statistical methods with visual analysis to offer ways of extracting wider understanding about
what qualities are hidden in your data. We need statistical analysis to describe what is in our data;
we need visual analysis to show us what is in our data and, crucially, show us what is not there.
‘When the data has been explored sufficiently, it is time to sit down and reflect – what were the most interesting
insights? What surprised me? What were the recurring themes and facts throughout all views on the data? In the end,
what do we find most important and most interesting? These are the things that will govern which angles and
perspectives we want to emphasise in the subsequent project phases.’ Moritz Stefaner, Truth & Beauty Operator
The chart shown in Figure 4.7 is a useful illustration of the value of supplementing stats with
visuals. This analysis shows a histogram distribution of the finish times of 9 million marathon
runs. The features of this chart follow the classic bell-shape curve that is found in plots about
many natural phenomena, such as the height measurements of a large group of people. If we
described this data using just statistical analysis we would have found common observations such
as the average, maximum, minimums and variance.
However, when you visualise the data, and closely scrutinise some of the patterns, you discover
interesting gaps emerging just after the key target finish times on or just before the three-, four-
and five-hour marks. You see the shape peaks just before these thresholds and then noticeably
drops. This reveals the significance of runners setting themselves and practising for target finish
times, often rounded to hourly or half-hourly milestones. This is a quality found in this data that
is only revealed by visualising it and only observed through the shape of the results.
Figure 4.7 What Good Marathons and Bad Investments Have in Common, by Justin Wolfers
(New York Times)
This example demonstrates the value of EDA, but it does not disclose the secret of how you find
such discoveries. There is no instruction manual for this. As John Tukey, the father of EDA,
described: ‘Exploratory data analysis is an attitude, a flexibility, and a reliance on display, not a
bundle of techniques’ (1980: 23). There is no single path to undertaking this activity effectively
or efficiently; it requires a repertoire of technical, practical and conceptual capabilities, as follows:
143

Instinct of the analyst: This is the primary matter. The attitude and flexibility that Tukey
describes are about recognising the importance of the analyst’s traits. Effective EDA is not about
the tool. There are many vendors out there pitching their applications as the magic ‘point and
click’ solution to make deep discoveries, and technology inevitably plays a key role in facilitating
this endeavour. However, the value of a good analyst cannot be underestimated. It is arguably
more influential than the differentiating characteristics between one tool and the next. In the
absence of a defined procedure for conducting EDA, an analyst needs to possess the capacity to
recognise and pursue the scent of enquiry. A good analyst will have that special blend of natural
inquisitiveness and the sense to know what approaches (statistical or visual) to employ and when.
Furthermore, when these traits combine with a strong subject knowledge, clearer judgements can
be made in distinguishing the significant from the insignificant.
Reasoning: Efficiency is a particularly important aspect of this exploration stage. The act of
interrogating data, waiting for it to volunteer its secrets, can take a lot of time and energy. Even
with smaller datasets you can find yourself tempted into trying out myriad combinations of
analyses, driven by the desire to find an as-yet-undiscovered nugget of golden insight.
‘At the beginning, there’s a process of “interviewing” the data – first evaluating their source and means of
collection/aggregation/computation, and then trying to get a sense of what they say – and how well they say it via
quick sketches in Excel with pivot tables and charts. Do the data, in various slices, say anything interesting? If I’m
coming into this with certain assumptions, do the data confirm them, or refute them?’ Alyson Hurt, News Graphics
Editor, NPR
There are so many statistical methods and, as you will see, so many visual means for seeing views
of data that you simply cannot expect to have the capacity to unleash the full exploratory
artillery. In most project circumstances you cannot afford to spend time trying out everything.
EDA is about being smart, recognising that you need to be discerning with your tactics.
Reasoning can help reduce the size of this prospect. In academia there are two distinctions in
approaches to reasoning – deductive and inductive – that I feel are usefully applied in this
discussion. Ideally, you will accommodate both approaches to help you confirm your knowns and
address those elusive unknowns:
Deductive reasoning is targeted: You follow a specific curiosity or hypothesis, framed by
subject knowledge, and interrogate the data in order to determine whether there is any
evidence of relevance or interest in the concluding finding.
Inductive reasoning is much more open in nature: You ‘play around’ with the data, based on
your sense or instinct about what might be of interest, and wait and see what emerges. In
some ways this is like prospecting, hoping for that moment of serendipity when you unearth
gold. It is important to give yourself room to embark on these somewhat less structured
exploratory journeys.
An analogy I often think is useful to help describe EDA concerns a ‘Where’s Wally?’ visual
puzzle. The process of finding Wally is unscientific. You often start out by unleashing your eyes
around the scene quite randomly. After this initial burst, perhaps subconsciously, you then adopt
a more considered process of visual analysis, eliminating different parts of the scene as ‘Wally-
free’ zones. This aids your focus and informs your strategy for where to look next. As you then
move across each mini-scene your eyes are pattern matching, looking out for the giveaway
144

characteristics of the boy wearing glasses, a red-and-white-striped hat and jumper, and blue
trousers.
The objective of this task is clear and singular in definition: you know what you are looking for.
Unfortunately, the challenge of EDA is rarely that clean, even if you have a source curiosity to
shape your pursuit, and you manage to find evidence of your ‘Wally’ somewhere in the data. In
EDA, unlike the ‘Where’s Wally?’ challenge, you have scope also to find other features in your
data that might change the definition of what qualifies as an interesting insight. In unearthing
other discoveries you might determine that you no longer care about Wally – finding him no
longer represents the main enquiry.
‘I kick it over into a rough picture as soon as possible. When I can see something then I am able to ask better
questions of it – then the what-about-this iterations begin. I try to look at the same data in as many different
dimensions as possible. For example, if I have a spreadsheet of bird sighting locations and times, first I like to see
where they happen, previewing it in some mapping software. I’ll also look for patterns in the timing of the
phenomenon, usually using a pivot table in a spreadsheet. The real magic happens when a pattern reveals itself only
when seen in both dimensions at the same time.’ John Nelson, Cartographer, on the value of visually exploring his
data
Research: It is important to learn as much as possible about the domain and the data you are
working with. Interpreting – the second phase of understanding where you establish meaning – is
only possible with domain knowledge. Without this, or having access to resources to help, you
will not know if what you are seeing is meaningful. Often, the consequence of EDA is that you
simply become more acquainted with the questions you need to ask, rather than any answers.
How to go about addressing this is really just common sense. You need to explore the places
(books, websites) and consult the people (experts, colleagues) to give you the best chance of
getting accurate answers to the questions you have. Good communication skills are vital. It is not
just about talking to others, it is about listening and learning. If you are in a dialogue with
subject-matter experts you will have to find an approach that allows you to understand
potentially complicated matters and also cut through to the most salient matters of interest.
Nothings: What if you have found nothing? You might reach a dead end, discovering no
significant relationships and finding nothing interesting about the shape or distribution of your
data. What do you do then? In these situations you need to change your mindset: nothing usually
still means something. Reaching a dead end or going down blind alleys can be helpful because
they help you eliminate dimensions of possible analysis. Traits of nothingness in your data or
analysis – gaps, nulls, zeros and no insights – can prove to be the main insight, as described
earlier.
‘My main advice is not to be disheartened. Sometimes the data don’t show what you thought they would, or they
aren’t available in a usable or comparable form. But [in my world] sometimes that research still turns up threads a
reporter could pursue and turn into a really interesting story – there just might not be a viz in it. Or maybe there’s no
story at all. And that’s all okay. At minimum, you’ve still hopefully learned something new in the process about a
topic, or a data source (person or database), or a “gotcha” in a particular dataset – lessons that can be applied to
another project down the line.’ Alyson Hurt, News Graphics Editor, NPR
There is always something interesting in your data. If a value has not changed over time, maybe it
was supposed to? That is an insight. If everything is the same size, maybe that is the story? If
145

there is no significance in the quantities, categories or spatial relationships, make the absence of
significance your main insight.
Not always needed: It is important to close this discussion about exploration with some
pragmatic reality. Not all visualisation challenges will involve much EDA and not all visualisation
projects will give you space to do much EDA. Your data might be immediately understandable,
and you might have a sufficiently strong knowledge of your subject (lots of known knowns already
in place). If you are working with small datasets they might not warrant broad visual
investigation. You need to be ready and equipped with the capacity to undertake this type of
exploration activity when it is needed, but the key point is to judge when it is needed.
146

Summary: Working with Data
147

The Four Steps
This chapter commenced your practical involvement with your data, taking you through the four
distinct steps that comprehensively acquaint you with the potential of your critical raw material.
Data acquisition looked at the different origins of and methods for accessing your data,
including data that is supplied to you, accessed via system download or through web scraping,
obtained using an API, gathered through foraging, or involves methods of primary collection.
Data examination profiled the different characteristics that define the type, size and condition of
your data. To usefully distinguish different types of data, the ‘TNOIR’ mnemonic was proposed:
Textual (e.g. responses to ‘Any other comments?’ in a questionnaire).
Nominal (e.g. the gender of a survey participant).
Ordinal (e.g. the rank of a police officer).
Interval (e.g. the forecasted temperature in °C).
Ratio (e.g. the number of sales of a book on Amazon).
Data transformation built on your examination work, identifying ways of modifying and
enhancing your data to prepare it for use, including:
Cleaning: resolve any data condition issues.
Creating: consider developing new calculations and value conversions.
Consolidating: think about introducing further data to expand or append to what you
already have.
Data exploration discussed the value of using visualisation techniques to supplement statistical
approaches as a way to discover more about the qualities and insights hidden away in your data.
148

General Tips and Tactics
Perfect data (complete, accurate, up to date, truly representative) is an almost impossible
standard to reach, especially given the typical presence of time constraints. You will often
need to make a call about when good enough is good enough. You might recognise a point
after which your ongoing efforts to refine may result in diminishing returns.
Do not underestimate the demands on your time; working with data will always be
consuming of your attention and effort. Ensure you have built plenty of time into your
handling of this data stage; be disciplined by keeping your focus and not getting sidetracked
into exploring every possible interesting avenue; be patient and persevere.
Clerical tasks like file management are important: maintain backups of each major iteration
of data, employ good file organisation of your data and other assets, and maintain logical
naming conventions.
Data management practices around data security and privacy will become important when
you are working with data that involves more sensitive/confidential subjects.
Keep notes about where you have sourced data, what you have done with it, any
assumptions or counting rules you have applied, ideas you might have for transforming or
consolidating, issues/problems, things you do not understand.
Anticipate and have contingency plans for the worst-case scenarios for data, such as the
scarcity of data availability, null values, odd distributions, erroneous values, long values, bad
formatting, data loss.
Ask questions. If you do not know something about your data, do not assume or stay
ignorant. And then listen: always pay attention to key information offered by subject-matter
experts.
Attention to detail is of paramount importance at this stage, so get into good habits early
and do not cut corners.
Maintain an open mind and do not get frustrated. You can only work with what you have.
If it is not showing what you expected or hoped for, you cannot force it to say something
that is simply not there.
The visuals produced during your data exploration work do not need to be elegantly
designed. Do not waste time making your analysis ‘pretty’, it only needs to inform you.
What now? Visit book.visualisingdata.com
EXPLORE THE FIELD Expand your knowledge and reinforce your learning about working with data through this
chapter’s library of further reading, references, and tutorials.
TRY THIS YOURSELF Revise, reflect, and refine your skill and understanding about the challenges of working
with data through these practical exercises.
SEE DATA VISUALISATION IN ACTION Get to grips with the nuances and intricacies of working with data in
the real world by working through this next instalment in the narrative case study and see an additional extended
example of data visualisation in practice. Follow along with Andy’s video diary of the process and get direct insight
into his thought processes, challenges, mistakes, and decisions along the way.
149

http://book.visualisingdata.com

150

5 Establishing Your Editorial Thinking
In Chapter 3, you initiated the design process by expressing the curiosity your work will attempt
to satisfy – for you, a stakeholder or your audience. In Chapter 4, you sequentially built up a
more intimate understanding of your subject through its data. This may have helped to confirm,
revised or expanded the scope of your curiosity.
In the present chapter we move forward to what is, arguably, the least technical stage of the
process: establishing your editorial thinking. In the context of a visualisation project, editorial
thinking is about determining what analysis you are going to portray visually to your audience.
The matter of how follows next: this stage is the critical bridge between your curiosity definition,
your data work and the design steps that follow.
In this chapter you will learn about the importance of editorial thinking in visualisation, what it
involves, and how it influences so much of the design thinking that will follow after this stage.
151

5.1 What is Editorial Thinking?
Editorial thinking is concerned with making informed judgements about the content you intend
to include in your visualisation. In my view this is one of the most defining activities that
separates the best visualisers from the rest, possibly even more so than any technical talent or
design flair. Before we move on to making design choices, you need to consider: given all the
things you could show, what will you show?
There are two words often used to define the essence of editorial thinking. One is editing; the
other is opinion. In the context of a data visualisation project, these are both relevant.
Editing is about making selections: choosing what clips you leave in a movie, what contents you
leave in a book, how you arrange music into a coherent whole. In visualisation we need to make
selections about what analysis we are going to portray to our audience in order to satisfy our
articulated curiosity. Regardless of whether origin curiosity still represents our core focus, or if it
has evolved following the ‘working with data’ stage, we will need to decide what analysis will
contribute towards facilitating the most relevant understanding about this subject. You can rarely
show everything – you rarely should show everything – so what are you going to show?
The term opinion can imply being impulsive or irrational but, in this context, it means you
making discerning, informed judgements. The emphasis is clear. It is you who are ultimately
responsible for the editing process. Even if it is influenced by guidance you have sought from
your project stakeholders or through dialogue with representatives of your audience, every
visualisation you ever produced is the consequence of your subjective choices. There is no
rulebook for this, no set procedure to lean on, no notion of perfect – it is down to you to make
the most reasonable call.
You need to decide what you are going to do now because you are about to move towards the
design phase that will involve picking chart types, deciding on which colours to use, what layout
to construct, and more besides. Before you decide how to design your content, you have to
determine what content to include.
‘A photo is never an objective reflection, but always an interpretation of reality. I see data visualisation as sort of a new
photojournalism – a highly editorial activity.’ Moritz Stefaner, Truth & Beauty Operator
When explaining what it means to establish editorial thinking in practice I find it helpful to
consider the parallels that exist between data visualisation and photography (or perhaps more
specifically, photojournalism, when there is a more explicit aim to communicate and report).
Think of a chart as a photograph of your data. By considering some of the decisions involved in
taking a photograph, you will find useful perspectives to shape your editorial thinking in
visualisation. There are three particular perspectives to consider: angle, framing and focus.
Angle: The first aspect of editorial thinking is concerned with choosing the angle of analysis –
the view of your data – that you think will best support the understanding required for this
subject. What content will answer or contribute towards answering the overriding curiosity?
152

In photography the angle is the position from which you take a shot and the view of your subject
this position gives you. Just as with any photograph, a visualisation is limited in that it cannot
provide a panoramic 360° view of all your data simultaneously. There is only so much a single
chart can show.
So, what are you going to show? Will it be how values have changed over time, or how they look
spatially, on a map? Is it more important to show a categorical breakdown or portray important
relationships between different variables?
Each different chart you will meet in the next chapter provides a different view of a subject
through its portrayal of your data. Before you choose to use a chart, you need to nail down what
angle of analysis you want to provide. Furthermore, will one angle be sufficient, or might you
need several different views?
It is easy to find yourself reluctant to commit to just a singular choice of angle. Even in a small
dataset, there are typically multiple possible angles of analysis you could conduct. It is often hard
to ignore the temptation of wanting to include multiple angles to serve more people’s interests. I
often find it far too easy to see everything as being potentially interesting to my audience, the
curse of the analyst.
It is important, though, not to fall into the trap of lazy thinking that if you throw together
multiple angles of analysis into your work, eventually one will serve the interests of your
audience. Just because you take 100 photographs of your holiday does not mean you should
show them all to somebody. You need to demonstrate discipline.
‘I think this is something I’ve learned from experience rather than advice that was passed on. Less can often be more.
In other words, don’t get carried away and try to tell the reader everything there is to know on a subject. Know what
it is that you want to show the reader and don’t stray from that. I often find myself asking others “do we need to show
this?” or “is this really necessary?” Let’s take it out.’ Simon Scarr, Deputy Head of Graphics, ThomsonReuters
Let’s suppose you are becoming increasingly active as a runner and are curious about how well
you are doing, drawing from run data you have been collecting using a tracking device or smart
watch which you can download and analyse yourself.
Your core curiosity may be expressed as ‘How well am I running?’ This is a rather open-ended
enquiry and is not going to be answerable by a single statistic or one single view of data. Even if,
after a specific run, there is only one thing dominating your attention at that specific moment
(such as ‘what distance did I go today?’), after each subsequent run there might be different
perspectives of interest as your focus shifts. To answer this ongoing curiosity, you will need access
to several distinct angles of analysis that, when synthesised, will collectively provide the
understanding you need, as listed in Figure 5.1.
Figure 5.1 The Questions That Might Help Answer the Query, ‘How Well Am I Running?’
153

As explained in the first chapter when discussing articulating your curiosity, I find forming data
questions helpful at this point. It keeps me focused on what I am trying to answer, albeit now at
a degree of increased specificity. All the questions listed in Figure 5.1 reflect reasonable
contributors towards collectively answering the curiosity. Some of these questions will be
answered by individual statistics, some will require charts to show an answer.
Framing: After defining which angle or angles of analysis you might need to include, framing is
the second editorial perspective concerned with refining the contents to be included in your
analysis. In photographic parlance this relates to choices about the field of view: what will be
included inside the frame of the photograph and what will be left out? Just like a photographer, a
visualiser must demonstrate careful judgement about what data items to include, what data items
to exclude, and why, for each statistical or chart display. A balance must be struck to find the
most representative view of your content.
They say the camera never lies, but photographs can certainly be distorting. With visualisations,
if you filter off too much of the content, it might disguise important context required for
interpreting the significance and meaning of values. Conversely, if you avoid filtering your
content you may fail to make visible the most salient discoveries. One of the key motives of
framing is to remove unnecessary clutter – there is only so much that can be accommodated in a
single view and only so much your audience will be able to process. Do not give them a puzzle if
you can give them the answer.
The criteria for your framing judgements will be influenced by the amount of data available to
show and the complexity of what you want to portray. Sometimes, showing lots of things –
indeed, all the things – creates visual complexity and that actually supports the point you are
making (‘look how crazily complex this system is!’). Further considerations about the setting,
such as the need for rapid insights or scope for deeper, more prolonged engagements, and the
dimensions and medium of the output, will also have a bearing on this matter.
Returning to the scenario of running data, the editorial framing decisions about this may cover
the following:
Do I include only my latest run, other recent runs or all my runs?
Do I include only those runs where I covered a certain minimum distance?
Do I include only runs where I have used specific routes or all routes?
Do I include only my data in this analysis or are there other people whose data I can
compare (e.g. a partner or running mate)?
154

Focus: This is the third editorial perspective and concerns which items of your data you might
choose to emphasise, thus generating some level of focus for the attention of your viewer. This
decision is not a function of filtering – that is the concern of your framing decisions. Focus is
about subjectively choosing to contrast visually features of your display that you deem to be more
important than others.
The best photographs are able to balance light and colour, not just to set the mood of a subject,
but to help illuminate key elements and convey visual depth. In addition to the depiction of the
relative sizing and arrangement of different contents, this provides a sense of visual hierarchy to
direct the eye of the viewer.
Whereas framing judgements were about balancing the clutter of your content, this is about
balancing the volume. If everything is shouting, nothing is heard; if everything is in the
foreground, nothing stands out; if everything is bold, nothing dominates.
The relevance of editorial focus is primarily associated with explanatory visualisations, whereby
elevating key insights to the surface of a display is a key attribute of the experience they provide.
So, what features of your data need to be brought into the foreground, left in the mid-ground, or
relegated into the background? What needs to be bigger and more prominent and what needs to
be less so?
For your running data, you might use colour highlighting to emphasise your above-average runs
and add labels to point out the fastest and the furthest. It will depend on what features you are
focusing on.
155

5.2 The Influence of Editorial Thinking
It is useful to ground this discussion practically by explaining how these editorial perspectives will
affect your design thinking. This is not just going to influence the best choice of chart to
represent your angle of analysis, but may influence the way you employ interactivity, the
annotations you include, the colours you apply, and the composition of your content. Let’s look
at two examples that illustrate this connection of influence between editorial and design
thinking.
156

‘The Fall and Rise of US Inequality’
The first example (Figure 5.2) is a chart taken from an article published on the ‘Planet Money:
The Economy Explained’ section of the US-based National Public Radio (NPR) website. The
article is titled ‘The Fall and Rise of US Inequality in 2 Graphs’. As the title suggests, the full
article includes two charts, but I just want to focus on the second of these for the purpose of this
illustration.
Let’s assess the editorial perspectives of angle, framing and focus as demonstrated by this work.
Angle: The main angle of analysis can be expressed as: ‘What is the relationship between two
quantitative measures (average income for the bottom 90% and for the top 1% of earners) and
how has this changed over time (year)?’ This angle would be considered relevant because the
relationship between the haves and the have-nots is a key indicator of wealth distribution. It is a
topical and suitable choice of analysis to include with any discussion about inequality in the
USA.
Framing: The parameters that define the inclusion and exclusion of data in the displayed analysis
involve filters for time period (1917 to 2012) and country (just for the USA). The starting point
of the data commencing from 1917 may just be an arbitrary cut-off point or could be a more
significant milestone in the narrative. More likely, it probably represents the earliest available
data. The up-to-date-ness of any chart can expire immediately after publication, but although
this data only reaches as far forward in time as 2012 (despite publication in 2015), the analysis is
of such historical depth that it should be considered suitably representative of the subject matter.
To focus only on the USA is entirely understandable given the scope of the piece.
Figure 5.2 The Fall and Rise of US Inequality, in 2 Graphs
157

Source: World Top Incomes Database; Design credit: Quoctrung Bui (NPR)
Focus: The visualisation includes an interactive ‘time slider’ control that allows users to move the
focus incrementally through each year, colouring each consecutive yearly marker for emphasis.
The colours are organised into three classifications to draw particular attention to two main
periods of noticeably different relationships between the two quantitative measures.
Now let’s switch our view and have a look at how the five layers of design are influenced by this
editorial thinking.
Data representation: The angle is what fundamentally shapes the data representation approach.
In simple terms, it determines which chart type is used. In this example, the desired angle of
analysis is to view the relationship between two quantitative measures over time (average income
for bottom 90% vs top 1% of earners). A suitable chart type to portray this visually is the scatter
plot, as selected. You will learn in the next chapter that the scatter plot belongs to the ‘relational’
family of chart types in that it primarily displays the relationships that might exist or otherwise
between different variables. Given there was also a dimension of time expressed in the intended
angle of analysis, a chart type from the ‘temporal’ family of charts could have been used.
However, with the main emphasis being to show the relationships, the scatter plot was the better
choice. The framing perspective defines what data will be included in the chosen chart: as
mentioned, only data for the USA and the time period 1917–2012 is displayed.
158

Interactivity: As you will discover in Chapter 7, the role of interactivity is to enable adjustments
for what data is displayed and how it is displayed. The sole feature of interactivity in this project
is offered through the ‘time slider’ control, as mentioned, which sequences the unveiling of the
data points year by year in either a manual or automated fashion. The inclusion of such
interactivity can be influenced by the editorial focus, in this case to unveil the yearly values in
sequence emphasising the position of each consecutive value over time.
Annotation: The primary chart annotations used here are the two arrows and associated
captions, drawing attention to the two prominent patterns that characterise a fall and a rise in
inequality. The inclusion of these captions would be a consequence of editorial focus to determine
that these patterns in the data should be surfaced for the viewer.
Colour: As you will learn in Chapter 9, one of the key applications of colour is to create ordinal
emphasis that brings important content to the surface and draws the eye’s attention. This would
influence the decision to deploy four colour states within the chart: a neutral colour to show all
points at the start of the animation and then three different emerging colours used to separate the
three clustered groups visually. The absolute colour choices of red, green and yellow tones are not
directly informed by editorial thinking, rather it is the identified need to have four different
colours to draw out the key patterns.
Composition: This element of design concerns decisions about all of the positioning and sizing
decisions. In this example, editorial thinking will have had limited influence over the
composition choices in this chart.
159

‘Why Peyton Manning’s Record Will Be Hard to Beat’
In this second example, published in ‘TheUpshot’ section of the New York Times website, there
are three charts presented in an article titled ‘Why Peyton Manning’s Record Will Be Hard to
Beat’. Here I will look at all three charts.
Let’s once again start by assessing the editorial perspectives of angle, framing and focus as demonstrated
by this work, one chart at a time.
Figure 5.3 Why Peyton Manning’s Record Will Be Hard to Beat, by Gregor Aisch and Kevin
Quealy (New York Times)
Angle: The first chart (Figure 5.3) portrays an angle of analysis that can be expressed as: ‘How
have quantitative values (cumulative NFL touchdown passes) changed over time (year) for
multiple categories (quarterbacks)?’ This analysis was relevant at the time due to the significance
of Peyton Manning setting a new record for NFL quarterback touchdown passes, an historic
moment, and, according to the article, ‘evidence of how much the passing game has advanced
through the history of the game’. Inspired by this achievement, the question posed by this article
overall is whether the record will ever be bettered – which would have likely been the origin
curiosity that drove the visualisation project in the first place. On its own, this analysis would be
deemed insufficient to support the overarching enquiry, as evidenced by the inclusion of two
further charts.
Framing: The criteria for data inclusion are shaped by the time period (1930 to 19 October
2014) and qualifying quantitative threshold (minimum of 30 touchdown passes). They are
representative of the truth at the moment of publication (i.e. up to 19 October 2014), though
clearly the data would no longer be up to date as soon as the next round of games took place. At
the time of publishing this book, Manning’s record is likely to be surpassed by either Tom Brady
or Drew Brees, and possibly both. This reinforces the idea of a chart being a frame in time. The
inclusion of players with at least 30 touchdown passes would be informed either by knowledge of
the sport (and if 30 touchdowns were seen as a common threshold) or possibly from discovering
that this was a logical cut-off value having visually explored the shape of the data for every
quarterback.
Focus: There is editorial emphasis applied to highlight the record holder as well as distinguishing
a selection of other current players. This helps to see which other contemporary players (at the
time) could have a chance of pursuing this record. Knowing what we know now, it was not
160

unreasonable to expect Brady and Brees to be among the main candidates to pursue Manning’s
record. There is also further emphasis applied to contrast selected all-time NFL touchdown pass
leaders with every other qualifying player.
Angle/Framing: In the second chart (Figure 5.4), the same editorial perspectives apply for angle
and framing, but the focus has changed. Even though the charts now comprise several small,
repeated views, each one focusing on the career trajectories of a selected previous record holder, it
is still fundamentally the same underlying view of data and includes the same data.
Focus: Colour is used to emphasise the previous record-holding players’ career line in each chart
panel, with background colour banding used to illuminate the duration of their record standing.
Value labels reveal the number of touchdowns achieved by each.
Figure 5.4 Why Peyton Manning’s Record Will Be Hard to Beat, by Gregor Aisch and Kevin
Quealy (New York Times)
The third and final chart (Figure 5.5) has many similarities with the first (Figure 5.3). Once
again it maintains the same consistent definition for framing and it has the same focus as the first
chart, but now there is a subtle difference in angle.
Figure 5.5 Why Peyton Manning’s Record Will Be Hard to Beat, by Gregor Aisch and Kevin
Quealy (New York Times)
Angle: The third and final chart (Figure 5.5) has many similarities with the first (Figure 5.3), but
the view of data now shown would be expressed as: ‘How have quantitative values (cumulative
161

NFL touchdown passes) changed over time (age) for multiple categories (quarterbacks)?’ The
difference is the temporal measure plotted along the x-axis and is now about the players’ ages at
the time of each touchdown pass, rather than when it occurred. This is a small but relevant
difference as it changes the nature of the analysis. It is included in support of the enquiry posed
about whether ‘the quarterback who will surpass Manning’s record is playing today?’
Incidentally, the article concludes it is going to be a very difficult record to beat.
Framing/Focus: This chart maintains the same defined perspectives for framing and focus as the
first chart.
Let’s now switch our view and have a look at how the five layers of design are influenced by this
editorial thinking.
Data representation: As I have stated, the angle and framing dimensions are hugely influential in
the reasoning of chart-type requirements. In each of the charts used we are shown different
perspectives around the central theme of how touchdown passes have changed over time for each
qualifying quarterback. A line chart was an entirely appropriate way to show the trends of
cumulative touchdown values for all the players included. Not surprisingly, the line chart belongs
to the ‘temporal’ family of chart types, as you will see in Chapter 6. Alternative angles of analysis
may have been possible to pursue, such as exploring the relationship between the age of players
when they reached their highest total and the absolute total touchdown passes. For this analysis, a
scatterplot would have been ideal to show this. However, showing the stories of each player’s
trajectory towards their cumulative touchdown total made for a more compelling display.
Interactivity: The only feature of interaction is included in the first and third charts, offering
mouseover events to reveal annotations of the names and total passes of any of the players
presented as grey lines. This serves the appetite of any viewer curious about the names of those
players without a label, but also preserves a certain elegance by not over-cluttering the main chart
display: detail is only available on demand, one player at a time.
Annotation: The interactive-enabled labelling is effectively a joint matter concerned also with
annotation. The decision to include permanent annotated value labels in each chart provides
editorial emphasis (in the first and third charts) for Peyton Manning, selected current
quarterbacks and selected all-time NFL touchdown pass leaders. The second chart only provides
single-value labels in each chart panel for the respective record holder on display.
Colour: The approach to creating focus is further amplified using colour. In the main chart,
emphasis is again drawn to Peyton Manning’s line, as the record holder (thick blue line), other
current players (highlighted with a blue line) as well as selected all-time NFL touchdown pass
leaders (dark-grey line). For the second chart the light-blue-coloured banding draws out the
period of the records held by selected players down the years. This helps the viewer to perceive
the duration of their records.
Composition: The sequencing of the charts in the article is a function of editorial focus – what
should go first, second and last, and why? Given the limitations of screen space to consume this
article, the ordering of the charts in this way will be to support the main narrative and essentially
answer the curiosity of the piece, as expressed in the title.
162

163

Determining Relevance
So, you can see how editorial definitions influence the design thinking that follows. But how do
you arrive at these definitions? What determines if you have astutely identified the right editorial
perspectives?
In Chapter 2, we discovered that one of the key principles for good visualisation design is that it
should be accessible. A characteristic of fulfilling accessibility in design was relevance, specifically
a concern about whether you are providing your audience with access to the most useful
understanding about this subject.
A lack of relevance is a curse that strikes a lot of visualisation work. Turning data into a visual
just because you happen to have it available is an aimless exercise. That is why we need to
instigate from the origin of a curiosity. This should provide a reasonably informed view about
what could be most useful to your audience, at least initially. Now that you have spent more time
deliberating over your audience’s needs, maybe even asking them what they need to know, it is
time to determine what is truly useful to them. This involves a blend of considerations:
Timeliness: Is the understanding beneficial at the moment of encounter?
Interestingness: Is the topic stimulated by new understanding (‘man bites dog!’) or
representative of helping to reinforce existing understanding (‘dog bites man!’)?
Pertinence: Does the viewer have an established association with the topic?
Sufficiency: Is the level of detail provided appropriate to the viewer’s needs at the moment
of encounter?
‘It requires the discipline to do your homework, the ability to quiet down your brain and be honest about what is
interesting.’ Sarah Slobin, Visual Journalist
Usefulness as perceived by the audience is not the only factor that shapes relevance, though. It is
also informed by what you actually have available to show them (Figure 5.6): what data do you
have and what analysis is available for you to present? If you do not have it in your data, you
cannot show it.
Figure 5.6 An Illustration for Determining ‘Relevance’
164

Having gone through the process of examining and exploring your data, in particular, you will
now have a far more informed view about what you could show your audience in order to meet
all their needs of usefulness. Indeed, sometimes your audience will not really be best placed to
know what is useful to them, in which case you may lead on what you want your audience to
know. Depending on the context, and your proximity to the subject and its data, you might have
the autonomy to dictate what you want to say, more so than what you think the audience want
to see.
It is vital not to fall into the trap of going through the motions. Just because you have spatial data
does not mean that the most useful angle of analysis will concern the ‘where’ of your data. If the
interesting insights about that subject are not significantly influenced by the spatial dimension, a
map may not provide the most relevant window on that data. You might actually find the
location information is more useful as a categorical device to group or separate analysis and so
other forms of analysis may be more insightful to pursue.
Although presented as consecutive stages, ‘working with data’ and ‘editorial thinking’ are quite
iterative: working with data influences your editorial perspectives; and your editorial perspectives
in turn may influence activities around working with data. In practice there will be much toing
and froing between the two (in contrast to the linear way I have to write and present this book).
The data transformation activity, in particular, is a key link. Editorial definitions may trigger the
need for more data to be gathered about the specific subject matter or further consolidation to
support the desired angles of analysis or the framing dimensions. Editorial definitions might also
influence the need for further calculations, groupings or general modifications to refine the
preparedness of your data for displaying the analysis.
165

Summary: Establishing Your Editorial Thinking
166

Editorial Perspectives
In this chapter you reflected on the possibilities offered by your data and learned about the
importance of committing to an editorial path. You defined three key editorial perspectives that
should be relevant to your audience in support of the overriding curiosity you are pursuing:
Angle: What view(s) of your data is most relevant? In language terms, what question should
your eventually chosen charts answer?
Framing: What data items and values will you include and exclude? What is most
representative of your subject?
Focus: Are there any features of your data you would wish to emphasise? This is especially
relevant to explanatory visualisations: if you have something to say, say it.
167

General Tips and Tactics
If your data is riddled with data condition issues, such as gaps or errors, perhaps consider
making this the story: invert the editorial view to be about the data, rather than about the
subject through its data.
There is always something interesting in your data: you just might not be equipped with
sufficient domain knowledge to know this or it may not be currently relevant. Get to know
the difference between relevant and irrelevant by researching and communicating to learn
more about your subject.
To help with your editorial angle, think about what title you would attach to this work.
What would be the headline? What would be the question posed that the visualisation
might answer?
What now? Visit book.visualisingdata.com
EXPLORE THE FIELD Expand your knowledge and reinforce your learning about working with data through this
chapter’s library of further reading, references, and tutorials.
TRY THIS YOURSELF Revise, reflect, and refine your skill and understanding about the challenges of working
with data through these practical exercises.
SEE DATA VISUALISATION IN ACTION Get to grips with the nuances and intricacies of working with data in
the real world by working through this next instalment in the narrative case study and see an additional extended
example of data visualisation in practice. Follow along with Andy’s video diary of the process and get direct insight
into his thought processes, challenges, mistakes, and decisions along the way.
168

http://book.visualisingdata.com

Part C Developing Your Design Solution
169

6 Data Representation
In Chapters 3, 4 and 5 you have been working through activities that embody what I consider to
be the hidden thinking of a visualisation project. These preparatory stages have helped you define
the requirements and aims of your work, given you steps to become acquainted with your data,
and, most recently, provided a structure for defining your editorial intent.
This chapter commences the fourth stage of the design process and represents a shift in focus
towards design thinking. ‘Developing your design solution’ begins with arguably the most
significant element of the visualisation design anatomy, namely data representation. How will
you visually portray your data?
We start the discussion by looking at the fundamentals of visual encoding, exploring the building
blocks that underpin all data representation thinking. From this bottom-up viewpoint we will
switch to the more pragmatic perspective of selecting chart types. To close the chapter, you will
learn about the influencing factors that will inform the choices you make.
170

6.1 Visual Encoding and Charts
Representing your data visually involves the act of visual encoding. As visualisers, we encode our
data using two main visual properties, marks and attributes. Marks are visual placeholders
representing data items, such as distinct records or discrete groupings, depending on the form of
your tabulation. These are the four main types of marks, as shown in Figure 6.1.
Attributes are variations in the visual appearance of marks to represent the values associated with
each data item. The main attributes you will encounter include those given in Figure 6.2.
The creative scope of some projects may use variation in attributes around the auditory (sound),
haptic (touch), gustatory (taste) and olfactory (smell) senses, otherwise these visual attributes are
the most commonly used options.
Figure 6.1 A Classification of Different Types of ‘Mark’ Encodings
Figure 6.2 A Classification of Different Types of ‘Attribute’ Encodings
171

It is worth noting that sometimes you do not need to encode data. Displaying values in their
original numeric or textual form may suffice, perhaps as presented in a table or through callout
statistic headlines.
172

Understanding visual encoding is of fundamental importance and is of particular relevance when
representing data using tools that adopt a bottom-up approach. However, for most people’s
needs, it can often be more pragmatic to think about data representation techniques through
selecting chart types.
Figure 6.3 The ‘CHRTS’ Families of Chart Types
If marks and attributes are the ingredients, chart types are the recipes. Different charts offer
different established ways of representing data, each one comprising combinations of marks and
attributes. As the field has matured, and practitioners have developed new recipes of marks and
attributes, the range of established chart-type options has grown.
To acquaint you with a broader repertoire of charting options, over the coming pages I will
present a collection of some of the common and useful chart types being used across the field
today. This gallery aims to provide you with a valuable reference that will help you to decide how
best to show what it is you want to say. I have organised each chart into five main families
(Figure 6.3) based on the primary editorial relationship you are trying to understand. The five-
letter mnemonic CHRTS provides a useful taxonomy for organising your thinking about which
chart(s) to use for your data representation needs.
Each chart-type profile is presented with supporting details that will help you fully understand
the role and characteristics of each option, including:
The primary name used to label each chart type as well as some further alternative names
that are often used.
An indication of which CHRTS family each chart belongs to, based on their specific
primary role, as well as a sub-family definition for further classification.
A description of the chart’s representation method, detailing what it shows and what each
mark and attribute encoding it deploys.
An applied example of the chart type in use with a description of what it specifically shows.
Presentation tips about the potential interactivity, annotation, colour or composition design
choices you might consider.
‘Variations and alternatives’ that describe further derivatives to understand other uses and
different purposes.
This gallery of charts is by no means an exhaustive list and I have excluded some options because
they were not different enough from other charts that have been profiled. I have mentioned some
charts that are legitimate derivatives or alternative applications of other similar charts, but have
assigned a whole page to profile these separately. For example, the Voronoi treemap is really just a
variation on the treemap that is profiled. It uses a different algorithm to arrange its constituent
pieces within different spatial layouts, like circles. The appearance and method of making this
173

might be slightly different, its usage is not.
I have wrestled with the value of including some of the charts presented, often due to limitations
and shortcomings in aspects of their usage. Some charts have merit for specific contexts, but can
be quite narrow in scope. Therefore, by including certain partially flawed charts I am attempting
to signpost relevant shortcomings, so you know how to use them sparingly. A word cloud, for
example, is a chart with absolutely quite limited value, but nonetheless it does have a role, as does
the often-derided pie chart. All chart types offer value for different situations; you just need to
use discretion to select them only under specific circumstances.
Although I have excluded several charts on grounds of demonstrating only a slight variation on
profiled charts, there are some types included that do exhibit small derivations from other charts
(such as the bar chart and the clustered bar, or the scatter plot and the bubble plot). In these
cases I felt there was sufficient difference in their practical application, and they were in common
usage, to merit their separate inclusion, despite the similarities.
Another point to make is that certain charts do not just fit into a single family. All charts that
belong to the hierarchical, relational, temporal and spatial families can include features of
categorical breakdown. Using a line chart to show how quantitative values have changed over
time for different categories could warrant being classified in either the temporal or the
categorical families. However, the change over time dimension is the primary dimension of
analysis and enables comparison between categories as a secondary perspective, so it is assigned to
the temporal family. I have therefore concentrated the taxonomy around the angle of analysis
each chart primarily conveys.
Finally, the spatial family of charts often relates to thematic maps that would not normally be
considered charts in purist terms. For convenience, though, I am badging them all as charts. It is
worth noting too that not all spatial analysis is geographic. Any of the spatial methods presented
could be used for non-geographic contexts, such as the anatomy of the body, the layout of a
building, the seat plan of an airliner.
Bar Chart
ALSO KNOWN AS Column chart, histogram (wrongly), lollipop chart
174

Representation Description
A bar chart displays quantitative values for different category items. The chart comprises line marks (bars) with the
size attribute (length or height) used to represent the quantitative value for each item.
Figure 6.4 The Countries with the Most Land Neighbours
175

Presentation Tips
ANNOTATION: The inclusion of chart apparatus devices like tick marks and gridlines can help increase the
precision of judging the quantitative values. If you include axis-scale labels you should not need to label
directly each bar value, as this will lead to label overload.
COMPOSITION: The bars should be proportionally sized according to the associated quantitative value –
nothing more, nothing less – otherwise the perception of the bar sizes will be distorted. Most commonly, this
means setting the quantitative value scale to an origin of zero. There is no significant difference in perception
between vertically or horizontally arranged bar charts; it will depend on which layout makes it easier to
accommodate the range of values and to read the item labels associated with each bar. Including a small gap
between each bar will help to preserve a clear distinction between each category item. Aim to make the sorting
of values in the chart as meaningful as possible.
176

Variations & Alternatives
A variation in the application of a bar chart would be to show quantitative values over time. This would be an option
to consider over the line chart when you have quantities for discrete periods (such as totals over a monthly period)
rather than a purely continuous series of point-in-time measurements. ‘Spark bars’ are mini bar charts that aim to
occupy only a word’s length amount of space. They are often seen in dashboards where space is at a premium and
there is a desire to optimise the density of the display. If you want to include further categorical subdivisions, an
alternative might be the ‘clustered bar chart’, to compare two or more or adjacent values, or the ‘stacked bar chart’, if
there is a part-to-whole relationship. ‘Dot plots’ offer a useful alternative for situations where you have large
quantitative values with a narrow range of difference and this difference is important to make visible. For contexts
where you have diverse value sizes and many categorical items, the ‘proportional symbol chart’ is an option to
consider.
Clustered Bar Chart
ALSO KNOWN AS Clustered column chart, paired bar chart, grouped bar chart
177

Representation Description
A clustered bar chart displays quantitative values for different primary category items with a secondary categorical
breakdown enabling local comparisons. The chart comprises line marks (bars) with the size attribute (length or
height) used to represent the quantitative value for each item. An attribute of colour is also used to distinguish further
the secondary categorical groupings.
Figure 6.5 The Ten Actors Who Have Received the Most Oscar Nominations for Acting
178

Presentation Tips
ANNOTATION: The inclusion of chart apparatus devices like tick marks and gridlines can help increase the
precision of judging the quantitative values. If you include axis-scale labels you should not need to label
directly each bar value, as this will lead to label overload. Any colours used must be explained through the
inclusion of a legend.
COMPOSITION: The bars should be proportionally sized according to the associated quantitative value –
nothing more, nothing less – otherwise the perception of the bar sizes will be distorted. Most commonly, this
means setting the quantitative value scale to an origin of zero. There is no significant difference in perception
between vertically or horizontally arranged clustered bar charts; it will depend on which layout makes it easier
to accommodate the range of values and to read the item labels associated with each cluster. Including a
noticeable gap between each cluster of bars will help to preserve a clear distinction between each primary
category item. Sometimes one bar might be slightly hidden behind the other if the display concerns a before
and after relationship. Aim to make the sorting of values in the chart as meaningful as possible.
179

Variations & Alternatives
Like the bar chart, clustered bar charts can also be used to show how values have changed over time. Alternatives
would include the ‘connected dot plot’, particularly to compare the quantitative size of two categories across a
number of major category items. If your clusters comprise many distinct categories, the display might become too
busy. You therefore might consider creating separate bar charts for each category item or using a ‘matrix chart’
structure to show the quantitative values at the intersection of two categorical dimensions.
Bullet Chart
ALSO KNOWN AS {No other names}
180

Representation Description
A bullet chart is effectively a bar chart displaying quantitative values for different categories, but incorporating
additional bandings to assist with interpreting the bars. The chart comprises line marks (bars) with the size attribute
(length or height) used to represent a quantitative value for each item. An attribute of colour (usually the lightness
property) is commonly used to distinguish contextual bandings behind each bar to aid interpretation.
Figure 6.6 The Top 20 Ranked Batters in Men’s Test Cricket (October 2018)
181

Presentation Tips
ANNOTATION: The inclusion of chart apparatus devices like tick marks and gridlines can help increase the
precision of judging the quantitative values. If you include axis-scale labels you should not need to label each
bar value directly, as this will lead to label overload. Any colours used to indicate meaningful bandings or
markers should be explained through the inclusion of a legend.
COMPOSITION: The bars should be proportionally sized according to the associated quantitative value –
nothing more, nothing less – otherwise the perception of the bar sizes will be distorted. Most commonly, this
means setting the quantitative value scale to an origin of zero. There is no significant difference in perception
between vertically or horizontally arranged bullet charts; it will depend on which layout makes it easier to
accommodate the range of values and to read the item labels associated with each bar. Aim to make the sorting
of values in the chart as meaningful as possible.
182

Variations & Alternatives
Like the bar chart, bullet charts can also be used to show how values have changed over time. Further point markers
(usually small circles or thin lines) can be included in the bullet chart to offer further useful comparisons and to
optimise the interpretation.
Waterfall Chart
ALSO KNOWN AS Cascade chart
183

Representation Description
A waterfall chart provides details of how a total or net quantitative value has been formed through an ordered
sequence of bars representing quantitative values for discrete categorical components. It is essentially a visual
calculation showing different components of positive and negative values, represented by size and direction, to
establish a running total. A common application of a waterfall chart would be to break down the calculation of profit
as formed by different categories of income and expenditure.
Figure 6.7 Nearly Half of New Zealand’s Annual Migration Gain is From Asia, by Kat Greenbrook
184

Presentation Tips
ANNOTATION: Direct value labelling is usually applied to each step describing what each relates to as well
as the quantitative amount. A dotted line is sometimes added to make more discernible what the running total
is at each stage. Any colours used must be explained through the inclusion of a legend.
COLOUR: Attributes of colour are often established to classify visually each categorical stage or to distinguish
further the positive and negative direction of quantitative values.
COMPOSITION: Most commonly a waterfall will be presented in landscape form with a left-to-right
sequence arriving at a final total or net amount at the final right-side position.
185

Variations & Alternatives
One alternative would be to consider the stacked bar chart, as long as there were no negative quantitative values and
all components are included that comprise a total, which represents a meaningful whole. The clustered bar chart may
also be used to split the categorical parts of the final total in close proximity, but with all bars sized from a common
baseline.
Radar Chart
ALSO KNOWN AS Filled radar chart, star chart, spider diagram, parallel coordinates
186

Representation Description
A radar chart plots values across multiple quantitative variables for one or several categorical items to enable general
pattern forming. It uses a radial (circular) layout comprising several axes emerging from the centre-like spokes on a
wheel, one for each variable. The quantitative values are then plotted along each scale using the attribute of position
and then joined by connecting lines to form a unique geometric shape. Sometimes the lines or the shape fill is
coloured for emphasis or for categorical differentiation when more than one item is plotted.
Figure 6.8 Global Competitiveness Report 2017–2018, by the World Economic Forum
187

Presentation Tips
ANNOTATION: The inclusion of chart apparatus devices like tick marks and gridlines can help increase the
precision of judging the quantitative values. Gridlines are only relevant if there are common scales across each
quantitative variable. If so, the gridlines must be presented as straight lines, not concentric arcs, because the
connecting lines joining up the values are themselves straight lines. If your quantitative values are on different
scales, do not forget to display the values ranges on each. Any colours used must be explained through the
inclusion of a legend.
COLOUR: When radar shapes are filled with a colour, sometimes a degree of transparency is applied to allow
the chart apparatus to be still partially visible.
COMPOSITION: The cyclical ordering of the quantitative variables should be as meaningful as possible and
consistent, as the shape formed will change for any ordering permutation. This will have a major impact on
the readability and meaning of the resulting chart shape. A radar chart works best when the neighbouring
pairings have some significant comparable value (such as values being plotted around the face of a clock or
compass).
188

Variations & Alternatives
If you have common scales across the quantitative variables, a ‘polar chart’ is an alternative, should the radial layout
be important to preserve. Otherwise, a ‘bar chart’ or ‘dot plot’ would be better options. While not strictly a variation,
‘parallel coordinates’ display a similar technique for plotting several independent quantitative measures in the same
chart. The main difference is that parallel coordinates use a linear layout. If you have multiple category items, rather
than plot them all on the same radar chart, consider using small multiples formed of distinct radars for each
individual item instead.
Polar Chart
ALSO KNOWN AS Coxcomb plot, polar area plot, circular barplot
189

Representation Description
A radar chart plots values across multiple quantitative variables for one categorical item to enable general pattern
forming. It uses a radial (circular) layout comprising several equal-angled circular sectors – like slices of a pizza, one
for each variable. In contrast to the radar chart (which uses position along a scale), the polar chart uses variation in the
size of the sector areas to represent values for each quantitative variable. It is, in essence, a radially arranged bar chart.
Colour is an optional attribute, sometimes used to differentiate between different quantitative variables.
Figure 6.9
190

Presentation Tips
ANNOTATION: The inclusion of chart apparatus devices like tick marks and gridlines can help increase the
precision of judging the quantitative values. Gridlines are only relevant if there are common scales across each
quantitative variable. If so, the gridlines must be presented as arcs reflecting the outer shape of each sector.
Connecting lines joining up the values are themselves straight lines. Each sector typically uses the same
quantitative scale for each quantitative measure but, on the occasions when this is not the case, do not forget to
display the values ranges on each. Any colours used must be explained through the inclusion of a legend.
COMPOSITION: The cyclical ordering of the quantitative variables should be as meaningful as possible and
consistent, as the shape formed will change for any ordering permutation. This will have a major impact on
the readability and meaning of the resulting chart shape. A polar chart works best when the neighbouring
pairings have some significant comparable value (such as values being plotted around the face of a clock or
compass). The sizing of the sectors needs to be carefully calculated. Each sector should have a proportionally
consistent angle of the whole and, to encode the quantitative values, the area of the sector, not the radius
length, should be used.
191

Variations & Alternatives
If you have inconsistent scales across the quantitative variables, a ‘radar chart’ is an alternative should the radial layout
be important to preserve. Otherwise, a ‘bar chart’ or ‘dot plot’ would be better options.
Connected Dot Plot
ALSO KNOWN AS Dot plot, dumbbell chart, range chart, dot chart, arrow chart
192

Representation Description
A connected dot plot displays quantitative values for different primary category items with a secondary categorical
breakdown enabling local comparisons. The plot is typically formed of two point marks plotting the quantitative
value positions for each secondary categorical grouping. Joining the two points together is a connecting line which
effectively represents the ‘delta’ (difference) between the two values through its size. Attributes of colour or variation
in symbol are commonly used to distinguish the secondary categorical groupings.
Figure 6.10 Gender Pay Gap UK, by David McCandless, Miriam Quick (Research) and Philippa Thomas (Design)
193

Presentation Tips
ANNOTATION: The inclusion of chart apparatus devices like tick marks and gridlines can help increase the
precision of judging the quantitative values. If you include axis-scale labels you should not need to label each
value directly, as this will lead to label overload. Any colours used must be explained through the inclusion of a
legend.
COLOUR: Colour may be used to help emphasise the directional basis of the connecting lines.
COMPOSITION: As the representation of the quantitative values is encoded through position along a scale
and not size, the quantitative axis does not need to have a zero origin. However, a zero origin may be helpful
to establish the scale of the differences depending on the subject matter being portrayed. If you do not
commence from an origin of zero, this will need to be clearly annotated. Aim to make the sorting of values in
the chart as meaningful as possible.
194

Variations & Alternatives
A variation in the application of the ‘connected dot plot’ would be to plot and compare values representative of two
different points in time for the same measure. An alternative would be to use a variation of the ‘Gantt chart’, and
rather than a single line starting from a minimum date and extending to a maximum date, you would just use this
line to show the position and difference between quantitative values. An ‘arrow chart’ is an extension of this whereby
the arrowhead is used to emphasise the directional basis of the line. Similarly, the ‘carrot chart’ uses line width
tapering to indicate direction. If the number of secondary categories grows in number, the ‘dot plot’ would be useful
to show the distribution of values rather than attempting to compare differences between just two values. A ‘clustered
bar chart’ offers a further alternative for showing comparisons between secondary categorical dimensions.
Pictogram
ALSO KNOWN AS Isotype chart, pictorial bar chart, symbol chart
195

Representation Description
A pictogram displays quantitative values for different primary category items with the option for secondary categorical
breakdown. The basis of the pictogram is the repetition in use of point marks, in the form of symbols or pictures, to
represent an associated quantitative count. Each point mark may be representative of one or many quantitative units
(e.g. a single symbol may represent 100 people). Secondary categorical dimensions can be incorporated through
differentiation in the attribute of colour or symbol.
Figure 6.11 Forbes: The World’s 100 Highest-paid Athletes, by Andy Kirk
196

Presentation Tips
ANNOTATION: The choice of symbols should be as recognisably intuitive as possible. If not, any legends
should be presented close to the display to enable quick reference for determining the categorical and
quantitative association of each symbol variation used.
COMPOSITION: If the quantities of markers exceed a single row, try to make the number of units per row
logically ‘countable’, such as displaying in groups of 5, 10 or 100. To aid readability, make sure there is a
sufficiently noticeable gap between clusters of grouped units. Aim to make the sorting of values in the chart as
meaningful as possible.
197

Variations & Alternatives
When showing a part-to-whole relationship, the ‘waffle chart’ is similarly formed using point marks and symbol or
colour attributes to differentiate the constituent parts of a whole.
Proportional Symbol Chart
ALSO KNOWN AS Proportional shape chart, graduated symbol plot, bubble chart, circle packing diagram
198

Representation Description
A proportional symbol chart displays quantitative values for different category items. The chart comprises shape
marks with the size attribute (area) used to represent the quantitative value for each item. An attribute of colour may
be used to accentuate the quantitative scale or organise marks by the distinct categories. Estimating and comparing
the size of areas with accuracy is not as easy, so this chart type works best when you have a diverse range of
quantitative value sizes.
Figure 6.12 For These 55 Marijuana Companies, Every Day is 4/20, by Alex Tribou and Adam Pearce (Bloomberg
Visual Data)
199

Presentation Tips
INTERACTIVITY: Proportional symbol charts may be accompanied by interactive features that let users
select or mouseover individual shapes to reveal annotated values of the quantity and category.
ANNOTATION: If interactivity is not achievable, a quantitative size key should be included, or direct
labelling incorporated. Though labelling can make a display cluttered (and be hard to fit when working with
small-sized shapes) it will help overcome some of the limitations of judging area size. Any colours used must be
explained through the inclusion of a legend.
COMPOSITION: The geometric accuracy of the shape mark size calculation is paramount: it is the area you
are modifying, not the diameter/radius. Typically, the layout is quite free-form with no baseline or central
gravity binding the display together. Otherwise, you might employ clustering or containers to help organise
the categorical distinctions, though the colouring of each shape may already achieve this. Aim to make the
sorting of the shapes in the chart as meaningful as possible.
200

Variations & Alternatives
Often, the data shown represents many parts of a whole. A ‘circle packing diagram’ uses circular shapes and packs the
contents into a neat circular layout representing a whole. The ‘bubble plot’ also uses differently sized shapes (usually
circles) but the position is meaningful across two quantitative variable dimensions. By removing the size attribute
(and effectively replacing the shape mark with a point mark) you could use the quantity of points clustered together
for different categorical totals to create a variation of the ‘pictogram’.
Word Cloud
ALSO KNOWN AS Tag cloud, proportional symbol chart
201

Representation Description
A word cloud shows the frequency of individual word items within a passage of textual data. Each item is represented
by words and then the font size of each is scaled according to the frequency of its usage. Words already have varied
lengths, so it is important to remember that it is effectively the area of the word, not its length, that encodes its
quantitative measure.
Figure 6.13 Comparing the Frequency of Words Used in Chapter 1 in the First Edition of this Book
202

Presentation Tips
INTERACTIVITY: Interactivity that lets users interrogate, filter and scrutinise the words in more depth,
perhaps presenting examples of their usage in a passage, can be quite useful features to enhance the value of a
word cloud.
ANNOTATION: Word clouds are most useful when you are trying to form a quick sense of some of the
dominant keywords used in the text. Relative comparisons can be aided by including a key to explain how the
font size scales equate to word frequency. Any colours used must be explained through the inclusion of a
legend.
COMPOSITION: The arrangement of the words within a word cloud is typically based on a layout process
that calculates the best placement of each word to occupy the optimum space.
203

Variations & Alternatives
Variations may include colours being used as a second form of quantitative encoding to accentuate the larger
frequencies further or to organise useful groupings categorically. You might also consider using containers to separate
out different clusters. Any alternative method from this categorical family of charts would more usefully display the
counts of text, such as a bar chart or a proportional shape chart where the word label sits inside a sized shape mark.
Heat Map
ALSO KNOWN AS Matrix chart, mosaic plot, table chart, XY heatmap, 2D density plot
204

Representation Description
A heat map displays quantitative values across the intersections of two categorical and/or discrete quantitative
dimensions. The chart comprises two categorical axes with each distinct value presented across the row and column
headers of a tabular layout. The corresponding cells effectively house a point mark with the attribute of colour
(usually, colour lightness) used to represent the associated quantitative value.
Figure 6.14 How Popular is Your Birthday?, by ONS Digital Content team
205

Presentation Tips
ANNOTATION: Direct value labelling is possible but normally a clear legend to indicate colour associations
will suffice. It is not easy for the eye to determine the exact quantitative values represented by the colours, even
if there is a colour scale provided; heat maps mainly facilitate more a gist of the order of magnitude.
COLOUR: Decisions need to be made about whether to use a smooth colour gradient or employ discrete
classifications for different value intervals. Different approaches will affect the patterns that emerge. There is
no single right answer – you will arrive at it largely through trial and error/experimentation – but it is
important to consider, especially when you have a diverse distribution of values.
COMPOSITION: Logical sorting and/or even grouping of the categorical values along each axis will aid
readability and may help to surface key relationships.
206

Variations & Alternatives
A ‘radial heat map’ offers a structure variation whereby the table may be portrayed using a circular layout. As with any
radial display, this is really of value only if the cyclical ordering means something for the subject matter. A variation
would see the colour lightness replaced by a categorical colouring approach if the values plotted were not quantitative
in nature. An alternative chart approach would be the ‘matrix chart’ using the size of a shape or the frequency of
clustered point marks to indicate a quantitative value.
Matrix Chart
ALSO KNOWN AS Table chart, correlogram
207

Representation Description
A matrix chart displays quantitative values across the intersections of two categorical and/or discrete quantitative
dimensions. The chart comprises two categorical axes with each distinct value presented across the row and column
headers of a tabular layout. The corresponding cells effectively house a geometric shape with scaled area size or
clusters of point marks repeated in quantity to represent the associated quantitative value. Attributes of colour are
often used visually to distinguish further categorical detail.
Figure 6.15
208

Presentation Tips
INTERACTIVITY: When using point mark clusters, interactive features can be useful to enable users to
discover the labels of each item through tooltips.
ANNOTATION: When shape marks are used, direct value labelling is possible but normally a clear key to
indicate the size associations will suffice. Any colours used must be explained through the inclusion of a
legend.
COLOUR: Colours may not be necessary because the tabular layout already establishes separation across the
two categorical dimensions. However, employing an additional attribute of colour can help to distinguish
further the horizontal or vertical categorical values.
COMPOSITION: If there are diverse value sizes with some especially large outliers, it may be necessary for
the size of the shape marks or the quantity of point clusters to outgrow the space of the relevant cell. This
might help to emphasise editorially the outlier status. Controlling this may not be possible, in which case the
largest quantitative value will usually fill no more than the maximum space available. Logical sorting (and
maybe even sub-grouping) of the categorical values along each axis will aid readability and may help surface
key relationships. The geometric accuracy of the shape mark size calculation is paramount: it is the area you are
modifying, not the diameter/radius.
209

Variations & Alternatives
A variation may involve the intersecting cells being representative of categorical values (nominal or ordinal), and
therefore you might substitute quantitative attributes of size or quantity with variation in symbols and/or colour
attributes. An alternative chart type might be the ‘heat map’, which similarly indicates quantitative values at the
intersections of two categorical and/or discrete quantitative dimensions.
Dot Plot
ALSO KNOWN AS Univariate scatter plot, 1D scatter plot, instance chart, strip plot, barcode chart
210

Representation Description
A dot plot displays the distribution of quantitative values for data items, sometimes broken down by a categorical
dimension, to show the range and shape of quantities. The plot is typically formed of point marks positioned along a
quantitative scale. The point marks may be small circles or thin lines (‘strips’). If categorical differentiation is
necessary, attributes of colour or variation in symbol may be employed within a single plot, otherwise several separate
plot views will be created for each discrete category grouping.
Figure 6.16 Bloomberg Billionaires, by Bloomberg Visual Data (Design and Development), Lina Chen and Anita
Rundles (Illustration)
211

Presentation Tips
ANNOTATION: The inclusion of chart apparatus devices like tick marks and gridlines can help increase the
precision of judging the quantitative values. If you include axis-scale labels you should not need to label each
value directly, as this will lead to label overload. Direct labelling will normally be restricted to noteworthy
points only. Any colours used must be explained through the inclusion of a legend.
COLOUR: Colour may be used to establish the focus of certain points and/or distinction between different
sub-category groups to assist with interpretation. To overcome occlusion caused by plotting several marks at
the same value position, you might use unfilled or semi-transparent filled circles to convey value frequency.
COMPOSITION: As the representation of the quantitative values is encoded through position along a scale,
the quantitative axis does not need to have a zero origin, unless this is meaningful to the subject. If you do not
commence from an origin of zero, this will need to be clearly annotated.
212

Variations & Alternatives
A variation in the encoding of the dot plot may see the point marks replaced by shape marks (usually circles) in order
to represent a second quantitative measure through size variation. This might be a useful method to represent the
frequency of observations when several items share a similar value. The variation in the role of the dot plot would be
through the ‘instance chart’, which plots events over a temporal axis rather than a quantitative scale. An alternative
chart type would be the ‘beeswarm plot’, especially when you have a non-uniform distribution of values that cluster
around similar quantities. You could also use a ‘scatter plot’ with its second axis offering the scope to plot two data
quantitative variables with the items spread across the associated coordinate positions.
Beeswarm Plot
ALSO KNOWN AS Jitter plot
213

Representation Description
A beeswarm plot displays the distribution of quantitative values for data items to show the range and shape of
quantities. The plot is typically formed of point marks, usually small circles, positioned along a quantitative scale. The
points are then evenly distributed using a second dimension of space above and below the quantitative axis baseline,
not to represent any quantitative measure, but to accommodate closely packed points that have similar value
positions. If categorical differentiation is necessary, attributes of colour or variation in symbol may be employed
within a single plot, otherwise several separate plot views will be created for each discrete category grouping.
Figure 6.17
214

Presentation Tips
INTERACTIVITY: Interactive features can be useful to enable users to discover the value labels of each item
through tooltips.
ANNOTATION: The inclusion of chart apparatus devices like tick marks and gridlines can help increase the
precision of judging the quantitative values. If you include axis-scale labels you should not need to label each
value directly, as this will lead to label overload. Direct labelling will normally be restricted to noteworthy
points only. Any colours used must be explained through the inclusion of a legend.
COLOUR: Colour may be used to establish the focus of certain points and/or distinction between different
sub-category groups to assist with interpretation.
COMPOSITION: As the representation of the quantitative values is encoded through position along a scale,
the quantitative axis does not need to have a zero origin, unless this is meaningful to the subject. If you do not
commence from an origin of zero, this will need to be clearly annotated.
215

Variations & Alternatives
A variation in the encoding of the beeswarm plot may see the point marks replaced by shape marks (usually circles) in
order to represent a second quantitative measure through size variation. An alternative chart type would be the ‘dot
plot’, which removes the second dimension spread of values and overlays similar values. You could also use a
‘histogram’ to show the frequency and distribution of values in discrete quantitative groupings.
Histogram
ALSO KNOWN AS Bar chart (wrongly), population pyramid
216

Representation Description
A histogram displays the frequency and distribution of quantitative measurements across grouped values for data
items. Whereas bar charts compare quantities for discrete nominal categories, a histogram uses discrete quantitative
‘bins’ to form ordinal value groupings. The representation is formed using variation of line size (if the value groupings
have equal intervals) or of shape area (if the value groupings have unequal value intervals) to represent the frequency
of measurements.
Figure 6.18 Beauty Brawl: How Inclusive are Beauty Brands Around the World?, by Amber Thomas, Jason Li and
Divya Manian for ‘The Pudding’
217

Presentation Tips
ANNOTATION: The inclusion of chart apparatus devices like tick marks and gridlines can help increase the
precision of judging the quantitative values. If you include axis-scale labels you should not need to label each
value directly, as this will lead to label overload.
COMPOSITION: Unlike the bar chart there should be no (or, at most, a very thin) gap between bars to help
the collective shape of the frequencies emerge. The sorting of the quantitative value bins must be presented in
ascending order so that the reading of the overall shape preserves its meaning. The number of value bins and
the range of values covered by each have a prominent influence over the appearance of the histogram and the
usefulness of what it might reveal: too few bins may disguise interesting nuances, patterns and outliers; too
many bins and the most interesting shapes may be abstracted by noise above signal. There is no singular best
approach, the right choice simply arrives through experimentation and iteration.
218

Variations & Alternatives
For an analysis that looks at the distribution of values across two dimensions, such as populations by age group across
binary gender categories, you might consider a ‘back-to-back histogram’ also commonly known as a ‘population
pyramid’. A ‘box-and-whisker plot’ is an alternative approach that reduces the display of the distribution of values to
just five key statistical measures. To reveal more granular detail, the ‘dot plot’ and ‘beeswarm plot’ display all items
individually across a quantitative scale.
Density Plot
ALSO KNOWN AS Ridgeline plot
219

Representation Description
Density plots display the frequency and distribution of quantitative values for data items. Whereas histograms
compare quantities using discrete quantitative ‘bins’ to form ordinal value groupings, a density plot can be considered
a smoothed histogram. The plot is typically formed of a quantitative scale along which a line connects measurements
of the frequency of each quantitative value. The line gets higher as the frequency gets higher. The connected line is
then smoothed using various statistical techniques (that will depend on the subject context) and the area below is
filled with colour to help visibility of the resulting shape. This creates the appearance of an ‘area chart’. Often the
density plot comprises multiple rows to separate observations across discrete category groupings.
Figure 6.19
220

Presentation Tips
ANNOTATION: The inclusion of chart apparatus devices like tick marks and gridlines is not usually
necessary with density plots as they are more about getting a sense of the shape and patterns.
COMPOSITION: Depending on the nature of the quantitative measurements, and in particular the presence
of outlier shapes in the distribution of values, the density plot is often presented in a way whereby high-value
area ‘spikes’ intrude into and over the row space occupied by categories above. The arrangement of discrete
categories is important to avoid too much occlusion and/or wasted empty space.
221

Variations & Alternatives
A variation in the design of a density plot is the ‘violin plot’ whereby the shape of distribution is plotted
symmetrically creating a two-sided violin-like, rather than the one-sided shape of the density plot. An alternative role
for the density plot would be in the form of an ‘area chart’, which plots quantitative trends over a temporal axis rather
than a quantitative scale. An alternative chart type would be the ‘beeswarm plot’ to show the quantitative values of
individual data items or a ‘histogram’ to show the frequency and distribution of values in discrete quantitative
groupings.
Box-And-Whisker Plot
ALSO KNOWN AS Box plot, candlestick chart, OHLC chart
222

Representation Description
A box-and-whisker plot displays the distribution and shape of a series of quantitative values for different categories.
The display is formed by a combination of lines and point markers to indicate (through position and length),
typically, five different statistical measures. Three of the statistical values are common to all plots: the first quartile
(25th percentile), the second quartile (or median) and the third quartile (75th percentile) values. These are displayed
with a box (effectively a wide bar) positioned and sized according to the first and third quartile values with a marker
indicating the median. The remaining two statistical values vary in definition: usually the minimum and maximum
values or the 10th and 90th percentiles. These statistical values are represented by extending a line beyond the bottom
and top of the main box to join with a point marker indicating the appropriate position. These are the whiskers. A
single plot will be produced for each relevant, discrete category grouping.
Figure 6.20 This Chart Shows How Much More Ivy League Grads Make Than You, by Christopher Ingraham
(Washington Post)
223

Presentation Tips
ANNOTATION: The inclusion of chart apparatus devices like tick marks and gridlines can help increase the
precision of judging the quantitative values. If you include axis-scale labels you should not need to label each
value directly, as this will lead to label overload. Direct labelling will normally be restricted to noteworthy
points only.
COMPOSITION: The quantitative value axis does not need to commence from zero, unless it means
something significant to the interpretation, as the ranges of values themselves do not necessarily start from zero
and the focus is more on the statistical properties between the outer values. There is no significant difference in
perception between vertically or horizontally arranged box-and-whisker plots; it will depend on which layout
makes it easier to accommodate the range of values and to read the item labels associated with each bar. When
you have several plots in the same chart, where possible try to make the categorical sorting meaningful, perhaps
by organising values in ascending or descending order based on the median value.
224

Variations & Alternatives
Variations mainly concern changing the number of statistical measures included in the display. Sometimes you might
remove the ‘whiskers’ to show just the 25th and 75th percentiles through the lower and upper parts of the ‘box’. The
‘candlestick chart’ (or ‘OHLC chart’ used in stock market analysis to track the opening, highest, lowest and closing
prices of stocks) uses a similar method and is often used to show the distribution and milestone quantitative values for
events that encounter constant change, such as stock market analysis over a given time frame based on showing the
opening, highest, lowest and closing prices.
Pie Chart
ALSO KNOWN AS Pizza chart, donut chart (wrongly)
225

Representation Description
A pie chart shows how proportions of quantities for different constituent categories make up a whole. It uses a
circular display divided into sectors for each category, with the angle representing the percentage proportions and
attributes of colour to separate the discrete categories. The resulting size of the sector (in area terms) is a spatial by-
product of the angle and so offers an additional means for judging values. The total of all sector values must be 100%
and the constituent parts must be exclusive and representative of a meaningful ‘whole’, otherwise the chart will be
corrupted.
Figure 6.21 Breakdown of Michael Schumacher’s F1 Career Over 308 Races
226

Presentation Tips
ANNOTATION: Directly labelling each category and associated value can enhance readability but may create
inelegant clutter depending on the shape of the data and the size of the label values. Any colours used must be
explained through the inclusion of a legend.
COLOUR: Colour is used to classify the categorical associations of each sector, so aim to vary the hue
property of each colour to maximise the visible difference. When you have multiple sectors, you might choose
to emphasise only two or three parts through editorial selection.
COMPOSITION: Positioning the first slice at the vertical 12 o’clock position gives a useful baseline to help
judge the first sector angle value. The ordering of sectors using descending values or ordinal characteristics
helps with the overall readability and allocation of effort.
227

Variations & Alternatives
The principal variation of the pie chart would be the ‘donut chart’. Its function is exactly the same, but the donut has
the centre removed, often to accommodate a labelling property. This removes the possibility of judging the sector
angles at the circle origin, so the reading is formed by the arc lengths. The role of a pie chart is primarily about being
able to compare a ‘part to a whole’ than being able to compare one part to another part. If you want to display and
compare multiple parts, the ‘bar chart’ will offer a better option. For showing many parts, especially if they are
organised into hierarchical groupings, the ‘treemap’ is a good option. Depending on the allocated space, a ‘stacked bar
chart’ may provide an alternative layout to the pie chart, especially if your categorical values have an ordinal
relationship. A ‘nested shape chart’, typically based on square or circle marks, enables comparisons across a series of
one-part-to-whole relationships showing absolute values (through size) and proportions (through relative size).
Waffle Chart
ALSO KNOWN AS Square pie, unit chart, grid plot
228

Representation Description
A waffle chart shows how proportions of quantities for different constituent categories make up a whole. It uses a
square display divided typically into 100 points arranged in a grid layout. Each constituent proportion is displayed
through colour coding the relevant number of points. The role of the waffle chart is to simplify the counting of
proportions in contrast to the angle judgements of the pie chart, though the display is limited to only showing integer
values. The total of all sector values must be 100% and the constituent parts must be exclusive and representative of a
meaningful ‘whole’, otherwise the chart will be corrupted.
Figure 6.22 Percentage of Hours During 2017 the Sun was Above the Horizon in Nuorgam, Finland, by Hanna
Kumpula (@kumpulahanna)
229

Presentation Tips
ANNOTATION: Chart apparatus is rarely applied to a waffle chart, though direct labelling may be included,
perhaps using a nearby caption to indicate a category and quantitative label. Any colours used must be
explained through the inclusion of a legend.
COLOUR: Adding outlines to each point mark (grid cell or circle) can be useful to help discern individual
units.
COMPOSITION: A waffle chart is quicker to read when clusters of units, such as groups of five or ten, can
be easily recognised. You may therefore seek to arrange the cells in groups to facilitate this. When you have
several parts in the same waffle chart, where possible try to make the categorical sorting meaningful.
230

Variations & Alternatives
Rather than using colour, sometimes variations in symbols will be used to classify different categories or groupings.
For example, you might see figures or gender icons used to show the makeup of a given sample population. A
variation in the role of a waffle chart is to show quantitative counts rather than proportions of a whole, and this
approach somewhat overlaps with applications of the ‘pictogram’. A ‘nested shape chart’ using sized rectangular
shapes may provide an alternative way of showing a part-to-whole relationship while also occupying a squarified
layout.
Stacked Bar Chart
ALSO KNOWN AS Stacked chart, packed bars
231

Representation Description
A stacked bar chart shows how quantitative values for different constituent categories make up a whole across major
category items. The proportion of each constituent categorical ‘part’ is represented by separate bars that are sized
according to their quantitative proportion and then stacked to create the whole. Sometimes the whole is standardised
to represent 100%, otherwise it will be representative of an absolute total. Colour attributes are used to classify the
discrete categorical parts. Stacked bar charts often work best when the categories are ordinal in nature, and it is the
overall pattern of spread across the whole that is important. If the parts are representative of nominal categories,
judging and comparing the size of individual stacked parts become quite hard without a common baseline, so you
might seek to reduce the number of discrete values.
Figure 6.23 In a Nation of Cynics, We’re Flocking to the Fringe, by ABC
232

Presentation Tips
ANNOTATION: Direct value labelling can become very cluttered when there are many parts, so you may
choose to focus only on labelling noteworthy values. Axis scales using logical intervals will be helpful, as will
the inclusion of gridlines, especially highlighting key features such as the 50% position when your data is
displaying a 100% stacked total. Any colours used must be explained through the inclusion of a legend.
COLOUR: If you are representing categorical ordinal data, colour can be astutely deployed to give a sense of
the general balance of values within the whole, but this will only work if their sorting arrangement within the
stack is logically applied. For categorical nominal data, ensure the stacked parts have sufficiently different
colour hues so their distinct bar lengths can be easily detected.
COMPOSITION: The bars should be proportionally sized according to the associated quantitative value –
nothing more, nothing less – otherwise the perception of the bar sizes will be distorted. Most commonly, this
means setting the quantitative value scale to an origin of zero. There is no significant difference in perception
between vertically or horizontally arranged stacked bar charts; it will depend on which layout makes it easier to
accommodate the range of values and to read the item labels associated with each cluster. Including a
noticeable gap between each stack of bars will help to preserve a clear distinction between the primary category
items. Aim to make the sorting of values in the chart as meaningful as possible.
233

Variations & Alternatives
The main alternative would be to create multiples of bar charts each showing the quantitative values for just a single
constituent part for each major category item. The ‘waterfall chart’ splits out the individual constituent parts to create
a step-by-step breakdown of a single stacked whole. Like their unstacked siblings, stacked bar charts can also be used
to show how value proportions have changed over time.
Diverging Bar Chart
ALSO KNOWN AS Back-to-back bar chart, paired bar chart, spine chart
234

Representation Description
A diverging bar chart shows how quantitative values for different constituent categories make up a whole across major
category items. The proportion of each constituent categorical ‘part’ is represented by separate bars that are sized
according to their quantitative proportion and then stacked to create the whole. Sometimes the whole is standardised
to represent 100%, otherwise it will be representative of an absolute total. In contrast to the stacked bar chart, the
diverging bar chart arranges constituent categorical parts either side of a common baseline depending on the discrete
nominal or ordinal relationships that benefit from such separation. Colour attributes are commonly used to classify
the discrete categorical parts.
Figure 6.24 Political Polarization in the American Public, Pew Research Center, Washington, DC (February, 2015)
(http://www.people-press.org/2014/06/12/political-polarization-in-the-american-public/)
235

http://www.people-press.org/2014/06/12/political-polarization-in-the-american-public/

Presentation Tips
ANNOTATION: Direct value labelling can become cluttered when there are many constituent parts, so you
may choose to focus only on labelling noteworthy values. Axis scales using logical intervals will be helpful, as
will the inclusion of gridlines. Any colours used must be explained through the inclusion of a legend.
COLOUR: If you are representing categorical ordinal data, colour can be astutely deployed to give a sense of
the general balance of values within the whole, but this will only work if their sorting arrangement within the
stack is logically applied. For categorical nominal data, ensure the stacked parts have sufficiently different
colour hues so their distinct bar lengths can be easily detected.
COMPOSITION: There is no significant difference in perception between vertically or horizontally arranged
diverging bar charts; it will depend on which layout makes it easier to accommodate the range of values and to
read the item labels associated with each cluster. Including a noticeable gap between each stack of bars will
help to preserve a clear distinction between the primary category item. Aim to make the sorting of values in the
chart as meaningful as possible.
236

Variations & Alternatives
A variation would be a ‘diverging histogram’ whereby the major categories have ordinal qualities, like increasing age
groups, and the resulting shape of the chart has meaning about the distribution of values. If you want to facilitate
direct comparison, a ‘clustered bar chart’ showing adjacent bars may offer a better alternative solution.
Marimekko Chart
ALSO KNOWN AS Mekko chart, mosaic plot, proportional stacked bar
237

Representation Description
A Marimekko chart is effectively a two-dimensional stacked bar chart with variation in size for both height and width
to display parts of a whole simultaneously across two dimensions. It is often used to contextualise percentage part-to-
whole comparisons of major categories with a second dimension of absolute numbers that make up a total. Attributes
of colour are commonly used to provide categorical classifications.
Figure 6.25 The Growth in Participants and Female Participation at the Summer Olympics
238

Presentation Tips
ANNOTATION: With two quantitative axes and two dimensions of categorical division, labelling
Marimekko charts can become quite cluttered. At the very least, the two axes should be clearly titled, and some
size scales provided, either through axis interval labelling or direct labelling of noteworthy items. Any colours
used must be explained through the inclusion of a legend.
COLOUR: It will usually be possible to distinguish classifications visually across only one of the categorical
dimensions.
239

Variations & Alternatives
An alternative to the Marimekko chart would be the treemap which shows part-to-whole relationships when there are
many category parts and there is some hierarchical organisation of those categories.
Treemap
ALSO KNOWN AS Heat map (wrongly)
240

Representation Description
A treemap is an enclosure diagram providing a hierarchical display that shows how quantitative values for different
constituent categorical parts make up a whole. It uses a contained rectangular layout, often termed squarified,
representing the 100% total. This is divided into proportionally sized shape marks (rectangular tiles) for the
quantitative values associated with each categorical part. The organisation of each shape is based on a tiling algorithm
to optimise the overall space usage and to cluster related categories into larger rectangle-grouped containers.
Attributes of colour are often used to represent further quantitative measures or categorical associations. Treemaps are
most commonly used, and of most value, when there are many parts to the whole. The constituent parts must be
exclusive and the total representative of a meaningful ‘whole’, otherwise the chart will be corrupted.
Figure 6.26 Finviz: Standard and Poor’s 500 Index Stocks (www.finviz.com)
241

http://www.finviz.com

Presentation Tips
INTERACTIVITY: Interactive features that enable selection events to trigger annotated tooltips can be
useful, providing direct value labels and details. There may also be scope for modifications to temporal
dimensions, changing the sizes and colouring accordingly, or zooming techniques to get a closer view of small
constituent parts.
ANNOTATION: Group or container labels can be hard to allocate space to efficiently, so borders are usually
applied to indicate the relevant enclosure areas. Effective direct value labelling becomes difficult as the
rectangles get smaller, so only the most prominent values may be annotated, especially if interactivity is not
available. Any colours used must be explained through the inclusion of a legend.
COMPOSITION: As the tiling algorithm used by any given tool to create a treemap will be focused on
optimising the dimensions and arrangement of the rectangular shapes, treemaps may not always be able to
facilitate meaningful sorting of high to low values within each enclosure. However, you will generally find
larger areas appear in the top left and work outwards towards the smaller constituent parts.
242

Variations & Alternatives
A variation of the treemap sees the overall rectangular layout replaced by a circular one and the tiles represented by
polygonal shapes. These are known as ‘Voronoi treemaps’ as the tiling algorithm is informed by a Voronoi
tessellation. The ‘circle packing diagram’, a variation of the ‘bubble chart’, similarly shows many parts to a whole but
uses non-tessellating circular shapes. The ‘Marimekko chart’ is similar in appearance to a treemap but, in contrast to
the treemap’s hierarchical display, presents a breakdown of quantitative percentages and/or absolute values across two
categorical dimensions.
Sunburst Chart
ALSO KNOWN AS Icicle chart, radial treemap, ring bracket
243

Representation Description
A sunburst chart displays hierarchical and part-to-whole relationships across multiple tiers of categorical dimensions.
In contrast to the dendrogram, the sunburst uses layers of concentric rings, one layer for each tier. Starting from the
centre ‘parent’ tier, the outward adjacency of the constituent parts of each tier represents the ‘parent-and-child’
hierarchical composition. Each ring layer is divided into proportional quantitative parts for each constituent category
across that tier. The size of the quantitative parts is represented by the size of a circular arc section (in length; width is
constant). Colours are often used to achieve further categorical distinction.
Figure 6.27 Which Fossil Fuel Companies are Most Responsible for Climate Change?, by Duncan Clark and Robin
Houston (Kiln) published in the Guardian, drawing on work by Mike Bostock and Jason Davies
244

Presentation Tips
INTERACTIVITY: Interactive features that enable selection events to trigger annotated tooltips can be
useful, providing direct value labels and details. There may also be scope for modifications to temporal
dimensions, changing the sizes and colouring accordingly, or zooming techniques to get a closer view of small
constituent parts.
ANNOTATION: Effective direct value labelling becomes difficult as the constituent parts get smaller, so only
the most prominent values may be annotated, especially if interactivity is not available. Any colours used must
be explained through the inclusion of a legend.
COMPOSITION: Sometimes, the hierarchical tiers do not necessarily have a parent–child relationship, so
their ordering can be legitimately switched around. Therefore, careful decisions are needed about the most
logical hierarchical sequencing given the subject matter and enquiry. There is also scope for arranging the
sequencing of constituent parts within each tier in a meaningful way.
245

Variations & Alternatives
Whereas the sunburst chart uses a radial layout, the ‘icicle chart’ uses a vertical, linear layout starting from the top and
moving downwards. The choice of a linear or radial tree structure will be informed largely by the space you have to
work in, as well as by the legitimacy of the cyclical nature of the content in your data. A variation of the sunburst
chart would be the ‘ring bracket’. This might show a hierarchical sequence of data related to subjects like sporting
competitions showing a knock-out sequence.
Dendrogram
ALSO KNOWN AS Radial tree, layout tree, cluster tree, tree hierarchy
246

Representation Description
A dendrogram is a node–link diagram that displays hierarchical relationships across multiple tiers of categorical
dimensions. It displays a hierarchy based on multi-generational ‘parent-and-child’ relationships. Starting from a
singular origin root node (or ‘parent’) each subsequent set of constituent ‘child’ nodes, a tier below and represented
by points, is connected by lines (curved or straight) to indicate the existence of a relationship. Each constituent node
may then have further constituencies represented in the same way, continuing through to the lowest tier of detail.
Figure 6.28 Making Sense of Skills: A UK Skills Taxonomy, by Dr Cath Sleeman
247

Presentation Tips
INTERACTIVITY: A useful interactive feature would be to enable filtering or highlighting of branches of
interest and selection options for revealing tooltips if labelling is too difficult to accommodate elegantly.
ANNOTATION: If labelling is required, depending on the number of tiers and nodes, the size of the text will
need to be carefully considered to ensure readability and minimise the effect of clutter.
COLOUR: Colour would be an optional attribute for accentuating certain nodes or applying further detail of
categorisation. The colour of the connecting lines is usually based on a neutral option like black or grey.
COMPOSITION: The layout can be based on either a linear tree (typically left to right, top to bottom) or
radial tree (outwards from the centre) structure. Sometimes, the hierarchical tiers do not necessarily have a
parent–child relationship, so their ordering can be legitimately switched around. Therefore, careful decisions
are needed about the most logical hierarchical sequencing given the subject matter and enquiry. There is also
scope for arranging the sequencing of constituent parts within each tier in a meaningful way.
248

Variations & Alternatives
Variations of the dendrogram involve incorporating some additional quantitative representation such as using the
length or width of connecting lines and, on replacing the point marks for each node, varying the size of node shapes.
An alternative approach would be to consider the ‘sunburst chart’ which would show a part-to-whole relationship
across the constituent categories in each hierarchical tier.
Venn Diagram
ALSO KNOWN AS Set diagram, Euler diagram (wrongly)
249

Representation Description
A Venn diagram shows collections of and relationships between multiple sets. It typically uses circular containers to
represent all independent and intersecting permutations with value labels or point marks used to place all category
items in the appropriate container. The size of the contained area is (typically) not important; what is important is in
which containing region an item resides. Variations in the attributes of colour or symbol may be commonly used to
represent further unique distinction among the items displayed.
Figure 6.29 This Venn Diagram Shows Where You Can Both Smoke Weed and Get a Same-sex Marriage, by Phillip
Bump (Washington Post)
250

Presentation Tips
ANNOTATION: The main annotation feature required will be to make clear which containers relate to
which set or membership grouping. When the permutations increase in number (e.g. three- or four-way
Venns) it can be hard to accommodate reasonable labels in each possible container.
COLOUR: Colour is often used to create more immediate distinction between the intersections and
independent parts or members of each container, especially when multi-way Venns are being used.
COMPOSITION: As the attributes of size and shape of the containers are of no significance there is more
flexibility to manipulate the display to modify the layout to accommodate the number or size of items in each
container group. While it is theoretically possible to exceed five-way Venn diagrams, the ability of readers to
make sense of such displays diminishes significantly.
251

Variations & Alternatives
A common variation or alternative to the Venn (but often mistakenly called a Venn) is the ‘Euler diagram’. The
difference is that an Euler diagram does not need to present every possible intersection and independency from all
categorical sets. A different approach to visualising sets (especially larger numbers) can be achieved using the ‘UpSet’
technique, which uses a matrix layout to present all possible set combinations and then a second, aligned method like
a bar chart to reveal a quantitative count for each set.
Scatter Plot
Also known as Scatter graph, scatter chart
252

Representation Description
A scatter plot displays the relationship between two quantitative variables for different category items. The display is
formed by point marks for each item, plotted positionally along each quantitative axis. Sometimes attributes of colour
hue are used to distinguish categorical dimensions across all items.
Figure 6.30 How Long Will We Live – And How Well?, by Bonnie Berkowitz, Emily Chow and Todd Lindeman
(Washington Post)
253

Presentation Tips
ANNOTATION: The inclusion of chart apparatus devices like tick marks and gridlines can help increase the
precision of judging the quantitative values. Reference lines, such as a trend line of best fit, might also aid
interpretation. If you include axis-scale labels you should not need to label each value directly, as this will lead
to label overload. Direct labelling will normally be restricted to noteworthy points only. Any colours used must
be explained through the inclusion of a legend.
COLOUR: To overcome occlusion caused by plotting several marks at the same value position, you could use
unfilled or semi-transparent filled circles to help convey value frequency.
COMPOSITION: As the representation of the quantitative values is encoded through position along a scale,
the quantitative axis does not need to have a zero origin, unless this is meaningful to the subject. If you do not
commence from an origin of zero, this will need to be clearly annotated. Ideally a scatter plot will have a
squared aspect ratio (equally tall as it is wide) to help patterns surface more evidently. If one quantitative
variable (e.g. weight) is likely to be affected by the other variable (e.g. height), it is general practice to place the
former on the y-axis and the latter on the x-axis. If you have to use a logarithmic quantitative scale on either or
both axes, you need to make this clear to viewers.
254

Variations & Alternatives
A ‘ternary plot’ is a variation of the scatter plot through the inclusion of a third quantitative variable axis. The ‘bubble
plot’ also incorporates a third quantitative variable, this time through encoding the size of a geometric shape
(replacing the point marker). A ‘scatter plot matrix’ involves a single view of multiple scatter plots presenting different
combinations of plotted quantitative variables, used to explore possible relationships among larger multivariate
datasets. A ‘connected scatter plot’ compares the shifting state of two quantitative measures over time.
Bubble Plot
ALSO KNOWN AS Bubble chart
255

Representation Description
A bubble plot displays the relationship between three quantitative variables for different category items. In contrast to
the scatter plot, the bubble plot uses shape marks (usually circles) for each category item, plotted positionally along
each quantitative axis with the variation in size of each mark representing a third quantitative measure. Sometimes
attributes of colour hue are used to distinguish categorical dimensions across all items.
Figure 6.31 Debtors or Creditors of the World? Looking at Countries’ Ecological Footprint Versus Biocapacity, by
Lisa Rost and Edith Maulandi
256

Presentation Tips
INTERACTIVITY: A useful interactive feature would be to enable filtering or highlighting of certain
categorical items, especially if there are several distinct categories and lots of items to make sense of.
Furthermore, selection options for revealing tooltips can be helpful if direct labelling is too difficult to
accommodate elegantly.
ANNOTATION: The inclusion of chart apparatus devices like tick marks and gridlines can help increase the
precision of judging the quantitative values. Reference lines, such as a trend line of best fit, might also aid
interpretation. If you include axis-scale labels you should not need to label each value directly, as this will lead
to label overload. Direct labelling will normally be restricted to noteworthy points only. Any colours used must
be explained through the inclusion of a legend.
COLOUR: If colours are being used to distinguish the different categories, ensure these are as visibly different
as possible. When data values are especially diverse in range, the size of shapes may vary from very small to
quite large. The largest shapes may overlap, in spatial terms, with other values or even hide them completely.
The use of outline borders and semi-transparent colours can help avoid the effect of total occlusion.
COMPOSITION: As the representation of the quantitative values is encoded through position along a scale,
the quantitative axis does not need to have a zero origin, unless this is meaningful to the subject. If you do not
commence from an origin of zero, this will need to be clearly annotated. Ideally a bubble plot will have a
squared aspect ratio (equally tall as it is wide) to help patterns surface more evidently. If one quantitative
variable (e.g. weight) is likely to be affected by the other variable (e.g. height), it is general practice to place the
former on the y-axis and the latter on the x-axis. If you have to use a logarithmic quantitative scale on either or
both axes, you need to make this clear to viewers. The geometric accuracy of the shape mark size calculation is
paramount: it is the area you are modifying, not the diameter/radius.
257

Variations & Alternatives
If the third quantitative variable is removed, the chart type would revert to a ‘scatter plot’. Variations on the bubble
plot might see the use of different geometric shapes as the markers, maybe introducing extra meaning through the
shape and dimensions used.
Network Diagram
ALSO KNOWN AS Node–link diagram, graph, hairball graph, social network
258

Representation Description
Node–link diagrams display relationships through the connections between data items. The common version of this
type of diagram displays items as nodes, represented by point marks, with links or edges (represented by lines)
depicting the existence of a connection. These connecting lines will sometimes encode an attribute of direction to
indicate the influencer relationship. Replacing point marks with a geometric shape and using attributes of size is a
further variation. In some versions a further quantitative weighting is applied to show the relationship strength,
maybe through increased line width. Attributes of colour may be used to indicate a quantitative value (e.g. number of
connections) and/or some categorical classification.
Figure 6.32 The Rise of Partisanship and Super-cooperators in the U.S. House of Representatives, visualisation by
Mauro Martino, authored by Clio Andris, David Lee, Marcus J. Hamilton, Mauro Martino, Christian E. Gunning,
and John Armistead Selde
259

Presentation Tips
INTERACTIVITY: Node–link diagrams are particularly useful when offered with interactive features,
enabling the user to interrogate and manipulate the display to facilitate visual exploration. The option to apply
filters to reduce the busyness of the display, and maybe enable isolation of individual node connections, can
help viewers to focus on specific parts of the network rather than face the whole system at once.
ANNOTATION: The complexity revealed by these diagrams is often a reflection of the underlying subject
complexity, so it can be helpful to use annotation to surface key observations about significant clusters or label
those items with the most connections.
COLOUR: Aside from the possible categorical colouring of each node, decisions need to be made about the
colour of the connecting lines, especially with regard to how prominent these links will be in contrast to the
nodes.
COMPOSITION: Composition decisions will be so varied for any network diagram depending on the
complexity and volume of the data and the output constraints around space and consumption format. There
are several common algorithmic treatments used to compute custom arrangements to optimise network
displays, such as force-directed layouts using the physics of repulsion and springs to amplify relationships.
There are also simplifying techniques such as edge bundling to aggregate or summarise multiple similar links.
260

Variations & Alternatives
There are many derivatives of the node–link diagram, as explained, based on variations in the use of different
attributes. ‘Hive plots’ and ‘BioFabric’ offer alternative approaches based on replacing nodes with vertices.
Sankey Diagram
ALSO KNOWN AS Alluvial diagram
261

Representation Description
Sankey diagrams display categorical composition and flows of quantitative relationships between different major
ordinal dimensions. The original application of the Sankey diagram displayed flow relationships over many discrete
ordinal stages, but it would be reasonable to say most common forms involve a two-sided parallel display. The two
sides represent different states of a paired, ordinal relationship, such as input vs output, or time A vs time B. On each
side there is effectively a stacked bar chart displaying proportionally sized and differently coloured (or spaced apart)
constituent parts of each whole. These might show categorical breakdowns of income vs categories of expenditure or
the categorical composition of some whole in a before and after state. Curved bands join each side of the display to
represent the connecting categories (origin and destination) with proportionally sized thickness representing the
quantitative flow of this relationship.
Figure 6.33 A Month in an Animal Shelter, by Sarah Campbell
262

Presentation Tips
INTERACTIVITY: Sankey diagrams are particularly useful when offered with interactive features, enabling
the user to interrogate and manipulate the display to facilitate visual exploration. The option to apply filters to
reduce the chaos of the visual and enable isolation of individual or groups of flows helps users to focus on
specific relationships of interest. Interactively enabled labelling can also be beneficial as direct labelling is
difficult to incorporate elegantly.
ANNOTATION: Direct labelling will normally be restricted to noteworthy points only. Any colours used
must be explained through the inclusion of a legend.
COLOUR: Colouring is often used visually to indicate the categories of the connecting bands, though this
can become complicated by an origin categorical colour joining to a different destination colour. A sense of
this change can be conveyed by blending the origin and destination colours half-way across.
COMPOSITION: The main arrangement decisions concern sorting. Firstly, by deriving as much logical
meaning from the categorical values within each stack and, secondly, by deciding on the sorting of the
connecting lines in the z-dimension – if many lines are crossing, there is a need to think about which will be
on top and which will be below. There is no significant difference between a landscape or portrait layout; it
will depend on the subject-matter ‘fit’ and the space within which you have to work. Sometimes the stacks on
each side are curved and appear more like stacked arcs.
263

Variations & Alternatives
The Sankey is closely related to the ‘alluvial diagram’, which tends to show changes in composition and flow over
time and often across multiple stages (rather than just the common paired structure of the Sankey). These days, the
labels are often applied interchangeably. A ‘chord diagram’ is a variation that uses a radial display to enable certain
origins and destinations to be one and the same. Showing how component parts have changed over time could just be
displayed using a ‘stacked area chart’. Plotting composition and flow can also be applied to a spatial display to create a
variation of the geographical ‘flow map’.
Chord Diagram
ALSO KNOWN AS Radial Sankey diagram, radial network, arc diagram
264

Representation Description
A chord diagram displays relationships through connections between and within category items. The diagram is
formed around a radial display with categories located around the edge: either shown as individual nodes or as arc
segments proportionally sized around the circumference to represent a part-to-whole breakdown. Emerging inwards
from each origin position are curved lines that join with related categorical destinations around the edge. The
connecting lines are proportionally sized according to a quantitative measure and a directional or influencing
relationship is often indicated. Attributes of colour hue are commonly used to distinguish different category
groupings visually.
Figure 6.34
265

Presentation Tips
INTERACTIVITY: Chord diagrams are particularly useful when offered with interactive features, enabling
the user to interrogate and manipulate the display to facilitate visual exploration. The option to apply filters to
reduce the chaos of the visual and enable isolation of individual or groups of flows helps users to focus on
specific relationships of interest. Interactively enabled labelling can also be beneficial as direct labelling is
difficulty to incorporate elegantly.
ANNOTATION: Direct labelling will normally be restricted to noteworthy points only. Any colours used
must be explained through the inclusion of a legend.
COLOUR: Aside from the categorical colouring of each node, decisions need to be made about the colour of
the connecting lines, especially on deciding how prominent these links will be in contrast to the nodes.
Sometimes the connections will match the origin or destination colours, or they will combine the two (with a
start and end colour blend to match the relationship).
COMPOSITION: The main arrangement decisions concern sorting. Firstly, by deriving as much logical
meaning from the categorical values within each stack and, secondly, by deciding on the sorting of the
connecting lines in the z-dimension – if many lines are crossing, there is a need to think about which will be
on top and which will be below. Showing the direction of connections can be achieved using arrowheads or
colour lightness variation. One common, subtle solution is to pull the destination join away from the edge of
the destination arc to contrast with connecting lines that emerge directly from an origin.
266

Variations & Alternatives
Variations of the chord diagram would be to consider using a single baseline axis, placing all category items along it
and forming connections between using semi-circular arcs. Additionally, a ‘Sankey diagram’ would be relevant if there
are distinct origins and destination relationships to reveal.
Line Chart
ALSO KNOWN AS Fever chart, stock chart
267

Representation Description
A line chart shows how quantitative values have changed over time for different categorical items. Line charts are
typically structured around a continuous temporal x-axis and quantitative y-axis with values plotted using point
marks at relevant coordinates. Connecting lines join up adjacent and related categorical items to form slopes which
are then extended along the full timescale to display a complete change over time. Multiple categories can be
displayed in the same view, each represented by a discrete line often with categorical or editorial colour attributes.
The connecting lines are typically straight, but sometimes curved line ‘interpolation’ may be applied to help
emphasise a general trend above precise point reading.
Figure 6.35 Cricketer Alastair Cook Plays His 161st and Final Test Match, by Financial Times / John Burn-
Murdoch
268

Presentation Tips
INTERACTIVITY: Interactivity may be especially helpful if you have many discrete categorical lines and
wish to enable the user to isolate a certain category of interest, either through filtering to exclude the others or
using a contrasting colour to emphasise its shape among the rest.
ANNOTATION: Sometimes the point mark is quite pronounced, to aid value judgements and possibly to
provide space for a value label, but on most occasions only the position of the connecting lines is displayed.
Ranking labels can be derived from the vertical position along the scale so direct labelling is usually
unnecessary. The inclusion of chart apparatus devices like tick marks and gridlines can help increase the
precision of judging the quantitative values. You might choose to annotate specific values of interest (highest,
lowest, specific milestones). Any colours used must be explained through the inclusion of a legend. If the shape
of the data presents an opportunity, you might consider directly labelling each or specific category lines, at the
first or last point mark position.
COLOUR: When many categories are shown, rather than colouring each line with a distinct categorical
classification, it may only be viable to emphasise lines of interest using colour hue or saturation.
COMPOSITION: The chart’s dimensions will need to be carefully considered, specifically the aspect ratio
formed by its height and width. The upward and downward slopes can seem more significant if the chart
width is narrow and less significant if it is more stretched out. There is no single practical rule to obey other
than using common sense to ensure you do not overly amplify or underplay features of your data. The
sequencing of values tends to follow a chronological left-to-right direction for the time-based x-axis and low
values rising up to high values on the y-axis; you will need a good (and clearly annotated) reason to break this
convention. Line charts do not always require the quantitative axis origin to start from zero, as the size of a
value is represented through position along a scale rather than the size of a line or shape. If zero has
significance for the interpretation of the trends portrayed, given the subject matter, then you should start the
baseline at zero.
269

Variations & Alternatives
Variations of the line chart may include the ‘bump chart’, to show rankings over time, and the ‘slope graph’, to
compare trends over two points in time. ‘Spark lines’ are mini line charts that aim to occupy only a word’s length
amount of space. They are often seen in dashboards where space is at a premium and there is a desire to optimise the
density of the display. An alternative would to use the ‘bar chart’ when you have quantities for discrete periods (such
as totals over a monthly period) rather than a purely continuous series of point-in-time measurements.
Bump Chart
ALSO KNOWN AS Bumps chart, rank chart
270

Representation Description
A bump chart shows how quantitative values have changed over time for different categorical items, where the
quantitative values are ranking measurements. These charts are typically structured around a continuous temporal x-
axis and quantitative y-axis with values plotted using point marks at relevant coordinates. Connecting lines join up
adjacent and related items to form slopes which are then extended along the full timescale to display a complete
change over time. A common extension of the bump charts uses variation in the size (width) of each line to represent
a quantitative measure, usually the absolute value that informs the ranking measurement. Multiple categories are
commonly displayed in the same view, each represented by a discrete line often with categorical or editorial colour
attributes. The connecting lines are typically straight, but sometimes curved line ‘interpolation’ may be applied to
help emphasise a general trend above precise point reading.
Figure 6.36 These are the 15 Most Important Political Problems in Germany, [Translated] by Berliner Morgenpost
271

Presentation Tips
INTERACTIVITY: Interactivity may be especially helpful if you have many discrete categorical lines and
wish to enable the user to isolate a certain category of interest, either through filtering to exclude the others or
using a contrasting colour to emphasise its shape among the rest.
ANNOTATION: Sometimes the point mark is quite pronounced, to aid value judgements and possibly to
provide space for a value label, but on most occasions only the position of the connecting lines is displayed.
Ranking labels can be derived from the vertical position along the scale so direct labelling is usually
unnecessary. You might choose to annotate specific values of interest (highest, lowest, specific milestones). Any
colours used must be explained through the inclusion of a legend. If the shape of the data presents an
opportunity, you might consider directly labelling each or specific category lines, at the first or last point mark
position, or even both.
COLOUR: When many categories are shown, rather than colouring each line with a distinct categorical
classification, it may only be viable to emphasise lines of interest using colour hue or saturation.
COMPOSITION: The sequencing of values tends to follow a chronological left-to-right direction for the
time-based x-axis and highest ranking values dropping to lowest ranking values on the y-axis; you will need a
good (and clearly annotated) reason to break this convention.
272

Variations & Alternatives
Consider alternatives like ‘line charts’ and ‘area charts’ if the ranking measurement is of secondary interest to plotting
absolute quantitative values.
Slope Graph
ALSO KNOWN AS Slope chart, parallel coordinates
273

Representation Description
A slope graph shows how quantitative values have changed over two points in time for different category items. The
display is based on (typically) two parallel quantitative axes with a common value range. A line is plotted for each
category connecting the two axes together with the vertical position on each axis representing the respective
quantitative values. These connecting lines form slopes that indicate the upward, downward or stable trend between
the two temporal axes. Attributes of colours are often used to distinguish different categorical lines or to surface major
trends among the items plotted.
Figure 6.37
274

Presentation Tips
INTERACTIVITY: Depending on the number of category values being presented, slope graphs can become
quite busy, especially if there are bunches of similar quantitative values with slope transitions. This also causes
a problem with accommodating multiple labels on the same value. On these occasions you might find
interactive features useful to enable filtering of certain items, to exclude others or to highlight a selection.
Discovering value labels of each item through interactive tooltips can also be beneficial.
ANNOTATION: Labelling of each category item on both sides will often be necessary, though this can be
challenging composition-wise when there are several items positioned in close proximity. You might therefore
choose to annotate only specific values of interest (highest, lowest, of editorial interest). The parallel axes will
need clear labels to explain the respective points in time. Any colours used must be explained through the
inclusion of a legend.
275

Variations & Alternatives
Rather than comparing two points in time, some variations in the application of a slope graph are used to show the
relationship between discrete quantitative variables for related category items. In this case the connecting line is not
indicative of a trend, rather a join to connected related items. This approach can also lead to the slope graph being
extended beyond just two parallel axes and thus evolving into the technique known as ‘parallel coordinates’. An
alternative chart type would be to consider the ‘connected dot plot’ which can also show comparisons of quantities for
two points in time across multiple category items.
Connected Scatter Plot
ALSO KNOWN AS Trail chart, comet chart
276

Representation Description
A connected scatter plot displays the relationship between two quantitative measures over time. The display is formed
of two quantitative x- and y-axes and with the values represented by point marks at the respective coordinates, one for
each measurement over time. The individual points are then connected (think of a dot-to-dot drawing puzzle) using
lines joining each consecutive point in time to form a sequence of change.
Figure 6.38 Holdouts Find Cheapest Super Bowl Tickets Late in the Game, by Alex Tribou, David Ingold and
Jeremy Diamond (Bloomberg Visual Data)
277

Presentation Tips
INTERACTIVITY: The biggest challenge is making the connections and the sequence as visible as possible.
This becomes much harder when values change very little and/or they loop back to previous positions, crossing
back over themselves. It is especially hard to label the sequential time values elegantly. One option to overcome
this is through animated sequences which might build up the display, connecting one line at a time and
unveiling the date labels as time progresses. It is often the case that only one series will be plotted. However,
interactive options may allow the user to overlay one or more for comparison, switching them on and off as
required.
ANNOTATION: The inclusion of chart apparatus devices like tick marks and gridlines can help increase the
precision of judging the quantitative values. If you can elegantly include direct labels to each point value,
indicating the time period it relates to, then this can be helpful. Connected scatter plots are unfamiliar to many
audiences and it can be quite demanding to learn how to read them. It may be necessary to provide a ‘how to
read’ guide illustrating what the axis values represent and what it means when connecting lines are moving in
different directions. Also consider labelling parts of the chart region that carry particular meaning, so if a
connecting line moves into that region, the interpretation of what this means can be accelerated.
COLOUR: Colour might be used to explain the temporal status of the connecting lines, for instance using a
faded colour for the past and a more vivid colour for the present. Otherwise, you might use attributes of colour
to accentuate certain sections of a sequence that might warrant particular attention.
COMPOSITION: As the representation of the quantitative values is encoded through position along a scale,
the quantitative axis does not need to have a zero origin, unless this is meaningful to the subject. If you do not
commence from an origin of zero, this will need to be clearly annotated. Ideally a scatter plot will have a
squared aspect ratio (equally tall as it is wide) to help patterns surface more evidently. If one quantitative
variable (e.g. weight) is likely to be affected by the other variable (e.g. height), it is general practice to place the
former on the y-axis and the latter on the x-axis.
278

Variations & Alternatives
The ‘comet chart’ is to the connected scatter plot what the ‘slope graph’ is to the ‘line chart’ – a summarised view of
the relationship between two quantitative measures over two points in time. The comet aspect is demonstrated
through the cone shape of the connecting line, with the more recent period of time generally having a thicker width.
A variation of the connected scatter plot is simply the ‘scatter plot’ where there is no time dimension or elements of
connectedness between points.
Area Chart
ALSO KNOWN AS Density plot
279

Representation Description
An area chart shows how quantitative values have changed over time for a single categorical item. The charts are
typically structured around a continuous temporal x-axis and quantitative y-axis with values plotted using point
marks at relevant coordinates. Connecting lines join up adjacent and related items to form slopes which are then
extended along the full timescale to display a complete change over time. The connecting lines are typically straight,
but sometimes curved line ‘interpolation’ may be applied to help emphasise a general trend above precise point
reading. To accentuate the shape of the trends, the area beneath the line is filled with colour, which means the height
of the area at any given point also reveals its quantity.
Figure 6.39 Crude Oil Prices for Brent (Dollars per Barrel) 2008–2018
280

Presentation Tips
ANNOTATION: Sometimes the point mark is quite pronounced, to aid value judgements and possibly to
provide space for a value label, but on most occasions only the position of the connecting lines is displayed.
The inclusion of chart apparatus devices like tick marks and gridlines can help increase the precision of judging
the quantitative values. You might choose to annotate specific values of interest (highest, lowest, specific
milestones).
COMPOSITION: The chart’s dimensions will need to be carefully considered, specifically the aspect ratio
formed by its height and width. The upward and downward slopes can seem more significant if the chart
width is narrow and less significant if it is more stretched out. There is no single practical rule to obey other
than using common sense to ensure you do not overly amplify or underplay features of your data. The
sequencing of values tends to follow a chronological left-to-right direction for the time-based x-axis and low
values rising up to high values on the y-axis; you will need a good (and clearly annotated) reason to break this
convention. Unlike the line chart, the quantitative axis for area charts must have an origin of zero as it is the
height of the coloured area under the trend line that is used to perceive the quantitative values.
281

Variations & Alternatives
A variation of the area chart is the ‘stacked area chart’, which can be used to show how multiple categories form a
whole and how this composition changes over time. The stacks may amount to an absolute total or form a 100%
proportion view. A ‘density plot’ appears like an area chart but is used to show the distribution of values across a
quantitative axis, rather than a time axis. Another variation is the ‘horizon chart’, which is based on an area chart but
for space-limited contexts. Values that exceed an imposed fixed maximum y-axis range are coloured to indicate
different bands of magnitudes, with the extremes usually darker. Like slicing layers off a mountain, each distinct band
of values above the maximum y-axis range is chopped off and dropped down to the baseline in front of its foundation
base. The final effect shows overlapping layers of increasingly darker colour-shaded areas occupying the same vertical
space. An alternative may be simply to consider the ‘line chart’, especially if you want to compare against several
discrete categorical items.
Stacked Area Chart
ALSO KNOWN AS Area chart, horizon chart
282

Representation Description
A stacked area chart shows how quantitative values have changed over time for multiple categorical items. These
charts are typically structured around a continuous temporal x-axis and quantitative y-axis with values plotted using
point marks at relevant coordinates. Connecting lines join up adjacent and related items to form slopes which are
then extended along the full timescale to display a complete change over time. The connecting lines are typically
straight, but sometimes curved line ‘interpolation’ may be applied to help emphasise a general trend above precise
point reading. To accentuate the shape of the trends, the area beneath the line is filled with colour, which means the
height of the area at any given point also reveals its quantity. When there are multiple discrete categories, separate
stacked areas, sized in height proportionally to their shifting values, are distinguished through distinct stacked regions
often coloured to establish their categorical association. The resulting display reveals how a part-to-whole relationship
changes over time.
Figure 6.40 Daily Indego Bike Share Station Usage, by Randy Olson (@randal_olson)(www.randalolson.com)
283

http://www.randalolson.com

Presentation Tips
INTERACTIVITY: Interactivity may be especially helpful if you have many discrete categorical stacks and
wish to enable the user to isolate a certain category of interest, either through filtering to exclude the others or
using a contrasting colour to emphasise its shape among the rest. Revealing the quantitative value, time and
category label at any point on the chart through a selectable tooltip can also be useful.
ANNOTATION: The inclusion of chart apparatus devices like tick marks and gridlines can help increase the
precision of judging the quantitative values. You might choose to annotate specific values of interest (highest,
lowest, specific milestones). Directly labelling the discrete category stacks can be helpful, otherwise use a clear
colour legend to explain associations.
COMPOSITION: The chart’s dimensions will need to be carefully considered, specifically the aspect ratio
formed by its height and width. The upward and downward slopes can seem more significant if the chart
width is narrow and less significant if it is more stretched out. There is no single practical rule to obey other
than using common sense to ensure you do not overly amplify or underplay features of your data. The
sequencing of values tends to follow a chronological left-to-right direction for the time-based x-axis and low
values rising up to high values on the y-axis; you will need a good (and clearly annotated) reason to break this
convention. Unlike the line chart, the quantitative axis for stacked area charts must have an origin of zero as it
is the height of the coloured areas used to perceive the quantitative values. Try to make the sorting of the
categorical stacks as meaningful as possible, perhaps placing the most important values on the bottom stack to
give it a consistent baseline.
284

Variations & Alternatives
The main variation in stacked area charts will be based on the quantitative values plotted and whether they are
representative of an absolute total or a proportional total forming a 100% whole. Rather than stacking categories you
might consider using small multiples of single-category area charts, especially as this will display each from a common
baseline and therefore make judgements of shape and size a little easier. An alternative may be simply to consider the
‘line chart’ formed of multiple lines for discrete categorical items.
Stream Graph
ALSO KNOWN AS Theme river
285

Representation Description
A stream graph shows how quantitative values have changed over time for multiple categorical items. The graphs are
generally used when you have many concurrent, constituent categories at any given point in time and these categories
may start and stop at different points in time rather than continue throughout the presented time frame. As befitting
the name, the appearance of the graphs is characterised by a flowing, organic display of meandering layers. They are
structured around a temporal x-axis with quantitative values plotted to quantify height above a local baseline, which is
not a stable zero baseline but rather a shifting shape formed out of other category layers. Connecting lines join up
adjacent and related items to form slopes which are then extended along the relevant time frame to create a unique
categorical layer. The area occupied by this layer is filled with an attribute of colour to represent a further quantitative
value scale or to associate with categorical classifications. The stacking arrangement of the multiple categorical layers
can shift up and down the implied y-axis dimension, in order to optimise the layout, but not to indicate any notion
of positive or negative values.
Figure 6.41 The Ebb and Flow of Movies: Box Office Receipts 1986–2008, by Mathew Bloch, Lee Byron, Shan
Carter and Amanda Cox (New York Times)
286

Presentation Tips
INTERACTIVITY: Interactivity may be especially helpful if you have many discrete categorical layers and
wish to enable the user to isolate a certain category of interest, either through filtering to exclude the others or
using a contrasting colour to emphasise its shape among the rest. Revealing the quantitative value, time and
category label at any point on the chart through a selectable tooltip can also be useful.
ANNOTATION: Chart apparatus devices are generally of limited use in a stream graph with the priority
being more on offering a general sense of pattern above precision of value reading. Direct labelling of the
discrete categorical layers may be possible, depending on the shape of the data, otherwise use a clear colour
legend to explain associations.
287

Variations & Alternatives
If you have relatively few discrete categorical items, you might consider using an alternative chart like the ‘stacked
area chart’ or small multiples of individual ‘area charts’. A ‘stacked bar chart’ would be a consideration, again if there
are relatively few categories to include and the quantitative measurements are based on discrete periods (such as totals
over a monthly period) rather than a purely continuous series of point-in-time measurements.
Gantt Chart
ALSO KNOWN AS Range chart, floating bar chart, Priestley timeline
288

Representation Description
A Gantt chart displays time-based intervals for different categorical items. The charts are typically structured around a
continuous temporal x-axis with a separate row allocated to each major categorical item. Intervals are formed by line
marks positioned according to a starting point and sized through length according to a closing point in time. Point
marks at each end of this line are sometimes included and presented with discrete symbols or attributes of colour to
highlight their distinction. The line may also display an attribute of colour to relate to some categorical status.
Figure 6.42 Establishment of the U.S. National Parks, by Nicholas Rougeux (www.c82net)
289

http://www.c82net

Presentation Tips
ANNOTATION: The inclusion of chart apparatus devices like tick marks and gridlines can help increase the
precision of judging the date values and durations. If you include axis-scale labels you should not need to label
each bar value directly, as this will lead to label overload.
COMPOSITION: The bars should be proportionally sized according to the associated duration length –
nothing more, nothing less – otherwise the perception of the bar sizes will be distorted. There is no significant
difference in perception between vertically or horizontally arranged Gantt charts; it will depend on which
layout makes it easier to accommodate the range of values and to read the item labels associated with each
category. Landscape layouts with time chronologically sequenced from left to right would be the most
common arrangement. Where possible, try to sequence the categorical items in a way that makes for the most
logical reading, organised by either the start/finish dates or maybe the durations (depending on which has most
relevance).
290

Variations & Alternatives
Gantt charts share many characteristics with the ‘connected dot plot’, but the measurement dimension here is of time
duration rather than quantitative difference. If duration between points in time is less important than individual
milestones or events, the ‘instance chart’ would be worth considering. Sometimes interval lines join up with other
adjacent categories, rather than being bound by discrete rows. This might be representative of the merging of
activities or the absence of ‘discreteness’ between activities, and the technique may therefore evolve more towards
being a ‘connected timeline’.
Instance Chart
ALSO KNOWN AS Dot plot, barcode plot, strip plot
291

Representation Description
An instance chart displays time-based events for different categorical items. It is typically structured around a
continuous temporal x-axis with a separate row allocated to each major categorical item. Events are represented by
point markers, plotted along the timeline, using combinations of symbols and colours to represent different types.
Figure 6.43 How the ‘Avengers’ Line-up Has Changed Over the Years, by Jon Keegan (Wall Street Journal)
292

Presentation Tips
ANNOTATION: The inclusion of chart apparatus devices like tick marks and gridlines can help increase the
precision of judging the date values. If you include axis-scale labels you should not need to label each bar value
directly, as this will lead to label overload. Any colours, symbols or size attributes used must be explained
through the inclusion of a legend.
COMPOSITION: There is no significant difference in perception between vertically or horizontally arranged
instance charts; it will depend on which layout makes it easier to accommodate the range of values and to read
the item labels associated with each category. Landscape layouts with time chronologically sequenced from left
to right would be the most common arrangement. Where possible, try to sequence the categorical items in a
way that makes for the most logical reading, organised by either the earliest or latest points in time or maybe
some measure of quantity (such as which category has the most recorded events).
293

Variations & Alternatives
Instance charts share many characteristics with the ‘dot plot’ but the measurement dimension here is of time rather
than quantitative value. Variations may see the point mark replaced by a geometric shape sized to represent a
quantitative measure associated with each event. If the data is more about durations and intervals between events, the
‘Gantt chart’ will be the best-fit option.
Choropleth Map
ALSO KNOWN AS Heat map
294

Representation Description
A choropleth map displays quantitative values for distinct, definable spatial regions. Each region is represented by a
defined polygonal shape, with each distinct shape collectively arranged to form the entire landscape. An attribute of
colour is used to represent a quantitative measurement. Choropleth maps should only be used when the quantitative
measure is directly associated with and continuously relevant across the spatial region on which it will be displayed. If
the quantitative measure is related to the consequence of more people living in an area, consider transforming your
data by standardising it as per capita or per acre (or other spatial denominator) accordingly.
Figure 6.44 How Voters in the 11,000 Municipalities Voted, [Translated] by Berliner Morgenpost
295

Presentation Tips
INTERACTIVITY: Interactivity may be especially helpful to offer selectable tooltips to view quantitative
values and category or location labels for any region on the display.
ANNOTATION: Depending on the shapes of the regions displayed, direct labelling may be limited to just a
number of noteworthy values. Any colours used must be explained through the inclusion of a legend. If you
choose to include a detailed map image in the background, do not include any unnecessary geographic details
that add no value to the spatial orientation or interpretation (e.g. roads, building structures).
COLOUR: The outline colour and stroke width for each spatial area should be distinguishable enough to
define the shape but not so prominent as to dominate. Usually, a light-grey or white-coloured stroke will
suffice. Sometimes variation in pattern may be included, as well as colour, to represent values that may be
uncertain or incomplete. When background map images are included, consider making them semi-transparent
or light in colour to avoid competition for attention with the more important data layer.
COMPOSITION: There are many different mapping projections for spatially representing the regions of the
world on a plane surface. Be aware that the transformation adjustments made by some of these projections can
distort the size of regions of the world, inflating their size relative to other regions, so you will need to pick a
projection that is appropriate to the spatial view you are providing.
296

Variations & Alternatives
Some choropleth maps may be used to indicate categorical association rather than quantitative measurements.
Alternative thematic mapping approaches to representing quantitative values might include the ‘proportional symbol
map’, using sized shapes over locations, and the ‘dot density map’, which plots a representative quantity of dots
equally (but randomly) across and within a defined spatial region. ‘Dasymetric mapping’ is similar in approach to
choropleth mapping but breaks the constituent regional areas into much smaller, more specific sub-regions better to
represent the realities of the distribution of human and physical phenomena within a given spatial boundary. This
might include details of individual buildings, for example.
Isarithmic Map
ALSO KNOWN AS Contour map, isopleth map, isochrone map
297

Representation Description
An isarithmic map displays distinct spatial surfaces on a map that shares the same quantitative classification. The
spatial definition here is not framed by geopolitical boundaries, rather it is organic regions that share a certain
quantitative value or interval scale. The regions are formed by interpolated ‘isolines’ connecting points of similar
measurement to form distinct surface areas. Each area is then colour coded to represent the relevant quantitative
value.
Figure 6.45 How Y’all, Youse and You Guys Talk, by Josh Katz (New York Times)
298

Presentation Tips
INTERACTIVITY: Interactivity may be especially helpful to offer selectable tooltips to view quantitative
values and category or location labels for any region on the display.
ANNOTATION: Depending on the shapes of the regions displayed, direct labelling may be limited to just a
number of noteworthy values. Any colours used must be explained through the inclusion of a legend. If you
choose to include a detailed map image in the background, do not include any unnecessary geographic details
that add no value to the spatial orientation or interpretation (e.g. roads, building structures).
COLOUR: The outline colour and stroke width for each spatial area should be distinguishable enough to
define the shape but not so prominent as to dominate. Usually, a light-grey or white-coloured stroke will
suffice. Sometimes variation in pattern may be included, as well as colour, to represent values that may be
uncertain or incomplete. When background map images are included, consider making them semi-transparent
or light in colour to avoid competition for attention with the more important data layer.
COMPOSITION: There are many different mapping projections for spatially representing the regions of the
world on a plane surface. Be aware that the transformation adjustments made by some of these projections can
distort the size of regions of the world, inflating their size relative to other regions, so you will need to pick a
projection that is appropriate to the spatial view you are providing.
299

Variations & Alternatives
There are specific applications of isarithmic maps used for showing elevation (‘contour maps’), atmospheric pressure
(‘isopleth maps’) or travel-time distances (‘isochrone maps’). Sometimes you might use isarithmic maps to show a
categorical status (perhaps a binary state) instead of a quantitative scale. ‘Choropleth maps’ will be the method used if
your data is organised by bound regions.
Proportional Symbol Map
ALSO KNOWN AS Graduated symbol map
300

Representation Description
A proportional symbol map displays quantitative values for locations on a map. The values are represented via
proportionally sized shapes (usually circles), which are positioned with the centre mid-point over a given location
coordinate. Colour is sometimes used to introduce further categorical distinction.
Figure 6.46 Here’s Exactly Where the Candidates’ Cash Came From, by Zach Mider, Christopher Cannon, and
Adam Pearce (Bloomberg Visual Data)
301

Presentation Tips
INTERACTIVITY: Interactivity may be especially helpful to offer selectable tooltips to view quantitative
values and category or location labels for any region on the display.
ANNOTATION: Depending on the size and overlapping of shapes displayed, direct labelling may be limited
to just a number of noteworthy values. Any size scales and colours used must be explained through the
inclusion of a legend. If you choose to include a detailed map image in the background, do not include any
unnecessary geographic details that add no value to the spatial orientation or interpretation (e.g. roads,
building structures).
COLOUR: The outline colour and stroke width for each spatial area should be distinguishable enough to
define the shape but not so prominent as to dominate. Usually, a light-grey or white-coloured stroke will
suffice. The largest shapes may overlap, in spatial terms, with other nearby locations and sometimes even hide
them completely. The use of semi-transparent colours can help avoid the effect of total occlusion. When
background map images are included, consider making them semi-transparent or light in colour to avoid
competition for attention with the more important data layer.
COMPOSITION: The geometric accuracy of the shape mark size calculation is paramount: it is the area you
are modifying, not the diameter/radius. There are many different mapping projections for spatially
representing the regions of the world on a plane surface. Be aware that the transformation adjustments made
by some of these projections can distort the size of regions of the world, inflating their size relative to other
regions, so you will need to pick a projection that is appropriate to the spatial view you are providing.
302

Variations & Alternatives
The main variations usually involve different geometric shapes being used. Alternatives include the ‘choropleth map’,
which colour codes regions, or the ‘dot map’, which uses dots to represent all items across a spatial region.
Prism Map
ALSO KNOWN AS Isometric map, spike map, datascape
303

Representation Description
A prism map displays quantitative values for locations on a map. The values are represented via proportionally sized
lines, appearing as 3D bars, that typically cover a fixed surface area of space and are then sized through height to
proportionally represent the quantitative value at each location. Attributes of colour are sometimes used to emphasise
large values in particular.
Figure 6.47 Trillions of trees, by Jan Willem Tulp
304

Presentation Tips
INTERACTIVITY: Ideally prism maps would be accompanied with interactive features that allow panning
around the map region to offer different viewing angles that overcome the perceptual difficulties of judging the
3D presentations of data in a 2D view. Otherwise, smaller values can find themselves hidden behind larger
forms, just as small buildings are hidden by skyscrapers in a city.
ANNOTATION: Direct labelling is usually impractical, so the most important feature of annotation is to
indicate the size scales used in the map display. If you choose to include a detailed map image in the
background, do not include any unnecessary geographic details that add no value to the spatial orientation or
interpretation (e.g. roads, building structures).
COLOUR: When background map images are included, consider making them semi-transparent or light in
colour to avoid competition for attention with the more important data layer.
305

Variations & Alternatives
Alternatives to the prism map, especially to avoid the 3D form, include the ‘proportional symbol map’, which uses
proportionally sized geometric shapes, and the ‘choropleth map’, which colour codes regional shapes.
Dot Map
ALSO KNOWN AS Dot distribution map, pointillist map, location map, dot density map
306

Representation Description
A dot map displays the distribution of phenomena on a map. It uses point marks to plot data items at specific
geographic coordinates. Items might be representative of instances of people, notable sites or incidences. The point
marks are usually small circles with attributes of colour used to distinguish categorical classifications. Sometimes a dot
represents a one-to-one phenomenon (i.e. a single record at that location) or one-to-many phenomena (i.e. for an
aggregated statistic whereby the location represents a logical mid-point), usually depending on the potential relevance
and/or sensitivity of directly plotting phenomena at precise locations.
Figure 6.48 The Racial Dot Map: Image Copyright, 2013, Weldon Cooper Center for Public Service, Rector and
Visitors of the University of Virginia (Dustin A. Cable, creator)
307

Presentation Tips
INTERACTIVITY: One method for dealing with viewing high quantities of observations is to provide
interactive semantic zoom features, whereby each time a user zooms in by one level of focus, the unit quantity
represented by each dot decreases, from a one-to-many towards a one-to-one relationship. Filtering options to
exclude or highlight certain selections may also aid the process of understanding.
ANNOTATION: Direct labelling is rarely applied. Clear legends explaining the dot unit scale and any colour
associations should ideally be placed as close to the map display as possible. If you choose to include a detailed
map image in the background, do not include any unnecessary geographic details that add no value to the
spatial orientation or interpretation (e.g. roads, building structures).
COLOUR: If colours are being used to distinguish the different categories, ensure these are as visibly different
as possible. When background map images are included, consider making them semi-transparent or light in
colour to avoid competition for attention with the more important data layer.
COMPOSITION: Dot maps should be displayed using an equal-area projection, as the precision of the
plotted locations is usually paramount. From a readability perspective, try to find a balance between making
the size of the dots small enough to preserve their individuality but not too tiny as to be indecipherable.
308

Variations & Alternatives
A ‘dot density map’ is a variation that involves plotting a representative quantity of dots equally (but randomly) across
and within a defined spatial region. The position of individual dots is therefore not to be read as indicative of precise
locations but used to form a measure of quantitative density. This offers a useful alternative to the choropleth map,
especially when categorical separation of the dots through colour is of value.
Flow Map
ALSO KNOWN AS Connection map, route map, stream map, particle flow map
309

Representation Description
A flow map shows the characteristics of movement or connections between phenomena across spatial regions. There
is no fixed recipe for a flow map, but it generally displays characteristics of origin and destination (positions on a
map), route (using organic or vector paths), direction (using arrow or tapered line width), categorical classification
(colour) and quantitative measurement (line weight or, if animated, motion speed).
Figure 6.49 Arteries of the City, by Simon Scarr (South China Morning Post)
310

Presentation Tips
INTERACTIVITY: Animated sequences may provide a useful presentation method when the phenomena are
characteristic of some notion of movement.
ANNOTATION: Annotation needs will be unique to each approach and the inherent complexity or
otherwise of the display. Often the general patterns may offer the sufficient level of readability without the
need for imposing amounts of value labels, but clear legends explaining the associations with any attributes
used will be important to include. If you choose to include a detailed map image in the background, only
include any relevant geographic details that offer spatial orientation or interpretation to the nature of flow
being represented (e.g. roads, rivers, oceans).
COLOUR: If colours are being used to distinguish the different categories, ensure these are as visibly different
as possible. When background map images are included, consider making them semi-transparent or light in
colour to avoid competition for attention with the more important data layer.
COMPOSITION: Some degree of geographic distortion or smoothing of flow routes may be required.
Decisions about the degree of interpolation applied to line smoothing or the merging of relatively similar
pathways may be entirely legitimate, but ensure that this is made clear to the viewer. There are many different
mapping projections for spatially representing the regions of the world on a plane surface. Be aware that the
transformation adjustments made by some of these projections can distort the size of regions of the world,
inflating their size relative to other regions, so you will need to pick a projection that is appropriate to the
spatial view you are providing.
311

Variations & Alternatives
There are several variations for how you might label different applications of displaying flow. It generally depends on
whether you are showing point A to point B journeys (‘connection maps’), more intricate pathways (‘route maps’) or
organic phenomena (‘particle flow maps’).
Area Cartogram
ALSO KNOWN AS Contiguous cartogram, density-equalising map
312

Representation Description
An area cartogram displays the quantitative values associated with distinct, definable spatial regions on a map. Each
geographic region is represented by a polygonal area based on its outline shape with the collective regional shapes
forming the entire landscape. Quantitative values are represented by proportionately distorting (inflating or deflating)
the relative size of and, to some degree, shape of the respective regional areas. Traditionally, area cartograms strictly
aim to preserve the neighbourhood relationships between different regions. Attributes of colour are often used to
represent the quantitative measurements and/or to associate the region with a categorical classification. Area
cartograms require the reader to be relatively familiar with the original size and shape of regions in order to be able to
establish the degree of relative change in their proportions.
Figure 6.50 The Carbon Map, by Duncan Clark and Robin Houston (Kiln)
313

Presentation Tips
INTERACTIVITY: Animated sequences enabled through interactive controls can help to better identify
instances and degrees of change, but usually only over a small set of regions and only if the change is relatively
smooth and sustained. Manual animation will help to provide more control over the experience. Selectable
tooltips to view quantitative values and category or location labels for any region on the display may also prove
useful.
ANNOTATION: Directly labelling the regional areas with geographic details and the value they hold is likely
to lead to too much clutter. As it is difficult to assess the degree of distortion and, indeed, often to identify the
regions themselves, it can be useful to present a thumbnail view of the undistorted original geographic layout
to help readers orient themselves with the changes. Additionally, a limited number of regional labels might be
included to provide direct spatial context and orientation. Any colours used must be explained through the
inclusion of a legend.
COLOUR: The outline colour and stroke width for each spatial area should be distinguishable enough to
define the shape but not so prominent as to dominate. Usually, a light-grey or white-coloured stroke will
suffice.
314

Variations & Alternatives
Unlike contiguous cartograms, non-contiguous cartograms tend to preserve the shape of the individual polygons but
modify the size and the neighbouring connectivity to other adjacent regional polygon areas. The best alternative ways
of showing similar data would be to consider using the ‘choropleth map’ or ‘Dorling cartogram’.
Dorling Cartogram
ALSO KNOWN AS Demers cartogram
315

Representation Description
A Dorling cartogram displays the quantitative values associated with distinct, definable spatial regions on a map. Each
geographic region is represented by a circular mark which is proportionally sized to represent a quantitative value.
The placement of each circle loosely resembles the region’s geographic location with general preservation of
neighbourhood relationships between adjacent shapes. Attributes of colour hue are often used to associate each spatial
region with a categorical classification.
Figure 6.51 Share of Individuals Using the Internet, 2015, by Lisa Rost
316

Presentation Tips
INTERACTIVITY: Interactivity may be helpful to offer selectable tooltips to view quantitative values and
category or location labels for any region on the display.
ANNOTATION: Directly labelling the shapes with geographic details and the values they hold is common,
though you might restrict this to only circles that are of sufficient size to hold such annotations. Any colours
used must be explained through the inclusion of a legend.
COMPOSITION: Preserving the layout adjacency with neighbouring regions is important. Dorling
cartograms tend not to allow circles to overlap or occlude, so some accommodation of large values might result
in some location distortion.
317

Variations & Alternatives
A variation on the approach, called the ‘Demers cartogram’, involves the use of rectangular marks instead of circles.
This offers an alternative way of connecting adjacent shapes. Other alternative chart types to consider would be the
‘area cartogram’ or the ‘choropleth map’.
Grid Map
ALSO KNOWN AS Cartogram, bin map, equal-area cartogram, hexagon bin map
318

Representation Description
A grid map displays the quantitative values associated with distinct, definable spatial regions on a map. Each
geographic region (or a statistically consistent interval of space, known as a ‘bin’) is represented by a fixed-size
uniform shape, sometimes termed a tile. The marks used tend to be squares or hexagons, though any tessellating
shape might help to arrange all regional tiles into a collective shape that roughly fits the real-world geographic
adjacency. Attributes of colour are applied to each regional tile either to represent a quantitative measurement or to
associate the region with a categorical classification.
Figure 6.52 Share of People Voting to Leave and Remain During the EU Referendum in 2016, by Ben Flanagan
319

Presentation Tips
INTERACTIVITY: Interactivity may be helpful to offer selectable tooltips to view quantitative values and
category or location labels for any region on the display.
ANNOTATION: Directly labelling the shapes with geographic details is usually impractical due to the small
size of each point mark, unless short abbreviated values can suitably represent the location label. Legends
explaining the colour associations must be included.
COMPOSITION: The main composition challenge is to determine the right geographic level for each
constituent tile to be representative of, and to optimise, the best-fit collective layout that preserves as many
neighbouring relationships as possible.
320

Variations & Alternatives
‘Hexagon bin maps’ are specific deployments of the grid map that offer a layout formed by a high resolution of
smaller hexagons to preserve localised details.
321

6.2 Influencing Factors and Considerations
You have now been through the gallery of chart-type options learning more about their specific
roles and what design features may enhance their particular deployment. Even if you have a fairly
clear idea about which chart(s) you might choose, there are other factors that may influence your
final decision of how to represent your data. There is a blend of considerations to draw from
your progress through the first three preparatory stages of the design process, supplemented by
the enduring need to satisfy the three principles of good visualisation design, as presented in
Chapter 2.
Technological: What charts you can actually make and how easily you can personally create
them is a big factor. Data visualisation technologies offer different chart-making capabilities and
it can be hard navigating through the options that exist. To assist with this, you might consider
consulting the ‘Chartmaker Directory’ (chartmaker.visualisingdata.com/). This digital resource
organises a huge catalogue of useful references that will offer an answer to the most common of
questions: ‘Which tool do you need to make that chart?’
Figure 6.53 Screenshot of the ‘Chartmaker Directory’
The directory’s content is presented through a tabular layout (Figure 6.53). Across the top of the
table are a selection of around 40 chart-making tools. A comprehensive list of different chart
types is presented down the side matching the gallery you have just explored. Inside the
intersecting cells, you will find unfilled and filled circular markers representing a reference in the
directory:
An unfilled mark represents a link to an example, providing evidence that a given chart can
be made in a given tool. Read it as ‘here’s a link to a bar chart made using Excel’, for
example.
A filled mark represents a link to a solution, which provides guidance on how to create a
given chart with a given tool. Solutions might exist as ‘how-to’ tutorials with step-by-step
instructions, video demonstrations, downloadable workbooks/templates or reusable code.
The directory is constantly growing as more chart-making solutions and examples are discovered
for each tool. In particular, many valuable references present smart workarounds or ‘out of the
box’ thinking that employ unconventional techniques to create a chart in a tool that normally
322

would not seem possible.
‘The capability to cope with the technological dimension is a key attribute of successful students: coding – more as a
logic and a mindset than a technical task – is becoming a very important asset for designers who want to work in Data
Visualisation. It doesn’t necessarily mean that you need to be able to code to find a job, but it helps a lot in the design
process. The profile in the (near) future will be a hybrid one, mixing competences, skills and approaches currently
separated into disciplinary silos.’ Paolo Ciuccarelli, discussing students on his Communication Design Master
Programme at Politecnico di Milano
Purpose: Having the technical ability to create a broad repertoire of chart types is the vocabulary
of this discipline; judging when to use them is the literacy. The first question to ask yourself is if
you even need to represent your data in chart form. Will this enable new qualities and
relationships in your data to be seen? Do not rule out the value of a table if providing a means for
your viewer to look up and reference values. It might be a more suitable solution option.
This brings us back to the discussion about the importance of defining the tone of your project.
Are you aiming to facilitate the reading of data or the feeling of data? Is it more important to
offer precise value judgements or should more emphasis be placed on general sense-making about
the big, medium and small values? Were there emotional qualities you wanted to emphasise or
suppress?
In his book Semiology Graphique, published in 1967, Jacques Bertin proposed the idea that
different ways of encoding data might offer varying degrees of accuracy in the perception of data
values. In 1984, William Cleveland and Robert McGill published a seminal paper, ‘Graphical
Perception: Theory, Experimentation, and Application to the Development of Graphical
Methods’. This offered more empirical evidence of Bertin’s thoughts. From this study they
developed a general ranking that explained which attributes used to encode quantitative values
would facilitate the highest degree of perceptual accuracy. In 1986, Jock Mackinlay’s paper,
‘Automating the Design of Graphical Presentations of Relational Information’, further extended
this to include proposed rankings for encoding categorical (nominal and ordinal) data, as well as
quantitative values. The table shown in Figure 6.54 presents the ‘Ranking of Perceptual Tasks’.
Figure 6.54 The Ranking of Perceptual Tasks, adapted from Mackinlay (1986)
323

What this ancestry of studies reveals is that the use of certain attributes to encode certain types of
data may make it quicker, easier and more accurate to judge the values portrayed. Two classic
illustrations of this notion are shown below. Looking at Figure 6.55, if A is 10, how big is B?
In both cases the answer is B equals 5. Although B in the bar chart being of size 5 feels about
right, the idea that circle B is also 5 feels less so. Our visual system is superior in its accuracy
when performing relative judgements for a line, in comparison with a shape. This is explained by
the fact that judging the variation in size of lines involves detecting change in a linear dimension
(length), whereas the variation in size of a geometric shape like a circle happens across a quadratic
dimension (area). If you look at the rankings in Figure 6.54 in the ‘Quantitative’ column, you
will see the encoding attribute of Length is ranked higher than the attribute of Area.
Figure 6.55 Comparison of Judging Line Size vs Area Size
Now let’s consider a demonstration of perceptual accuracy when using different dimensions of
colour variation to represent nominal values. In the charts shown in Figure 6.56 you can see that
different attributes are used to represent the categorical groupings in the two scatter plots. On the
left you see variation in the attribute of colour hue (blue, orange and green) to classify the
distinct categories; on the right you see the attribute of symbol (diamond, circle and square)
applied similarly.
What you will find is a more immediate, effortless and accurate experience in identifying the
groupings of the coloured category markers compared with the symbol-based equivalents. It is
easier to observe classifications through variation in colour than it is using variation in symbol, as
supported by colour hue being ranked higher than shape for nominal data types as shown in the
table in Figure 6.54.
Figure 6.56 Comparison of Judging Categorical Associations Using Variation in Hue vs
Variation in Shape
You can see from these simple demonstrations that there are clearly ways of encoding data that
will make it easier to read values accurately and efficiently. However, as Cleveland and McGill
stress in their paper, this should only be taken as guidance, commenting that the ranking of
324

attributes ‘does not result in a precise prescription for displaying data but rather is a framework
within which to work’.
This is important to acknowledge because you have to weigh up whether precise perceiving is
actually what you wish to offer your viewers. As stated in Chapter 3, sometimes getting the ‘gist’
of data values is sufficient. You might therefore determine that selecting a chart that uses the
attribute of size through variation in area, which is lower down the quantitative attribute
rankings, offers a suitable balance. Judging the hierarchy of large, medium and small features may
be sufficient for your needs. It depends on your purpose.
Sometimes, you will have scope in your encoding choices to incorporate a certain amount of
visual immediacy in accordance with your topic. I warned earlier about the need to be driven by
your data and not by your ideas, but sometimes there is scope to squeeze out extra stylistic
associations between the visual and the content. The flowers of the Better Life Index feel
consistent in metaphor with the idea of better life: the more in bloom the flowers, the more
colourful and prouder each petal appears and the better the quality of life in that country.
Figure 6.57 Simulated Dendrochronology of U.S. Immigration, by Pedro Cruz, John Wihbey,
Avni Ghael and Felipe Shibuya
325

What you are trying to represent may not be possible using a conventional chart. Another
example that draws from nature is shown in Figure 6.57. This piece uses the notion of
dendrochronology – the study of tree rings – to create a compelling portrayal of the history of
immigration into the USA. Each ring is a decade container, working outwards through
chronological decades. Within each container ring are dots corresponding to 100 immigrants.
The colours indicate the origin continents or major regions. The outcome is a stunning concept
that perfectly aligns subject matter and visual encoding.
‘I’ve come to believe that pure beautiful visual works are somehow relevant in everyday life, because they can become
a trigger to get people curious to explore the contents these visuals convey. I like the idea of making people say “oh
that’s beautiful! I want to know what this is about!” I think that probably (or, at least, lots of people pointed that out
to us) being Italians plays its role on this idea of “making things not only functional but beautiful”.’ Giorgia Lupi,
Co-founder and Design Director at Accurat
Data type and shape: The types of data and range of values you are trying to display will have a
bearing on which charts you can use, and, of those, which will best portray what you want to say.
Any chart type will only accommodate certain types of data. For example, if you want to use a
line chart, you will need one or more continuous series of quantitative values that have a
326

dimension of temporal data. Additionally, the viability of any chart choice will be determined by
how well it accommodates the range of values you wish to include. As ever, this depends on what
it is you want to say.
Let’s suppose you are producing some simple analysis about the market share of browsers. The
first chart you consider is the pie chart. To use this you will need quantitative values, in the form
of percentages, for different categories that aggregate to a true ‘whole’ (nothing more, nothing
less, than 100%). The data shows there is a market share breakdown across 30 discrete browsers.
As you can see in Figure 6.58, there are a few issues with the pie chart (A). If it is important for a
viewer to judge values for each of the 30 browsers with a certain degree of accuracy, the pie chart
will not be fit for purpose. It gets harder to perceive the size of each slice after the first three or
four. Furthermore, the colour associations as shown in the legend are indiscernible. We need a
plan B. In this case you might switch to a bar chart. Even though this chart belongs to a different
‘family’ you can still use it to represent parts-of-a-whole percentages for each browser item. This
will offer an improved option to make it more readable, both in judging the values and through
the proximity of the category labels to each bar. It does, though, result in a lot of empty space
due to the skewed shape of the data values.
If you are really seeking to enable the readability of each value, you may try to convey just how
dominant Chrome is as one part of this whole. You might therefore revert to using a pie chart
(C) to include all the discrete browser categories, but label only the Chrome part and summarise
the rest as a single ‘All others’ value. You only need the Chrome value to be seen as ‘biggest’
compared with the many other competitors battling for but losing out on the dominant market
share. If the visibility of the many other parts is not important, group them into a single ‘All
others’ value so now you have a simple two-slice pie (D) or a donut chart (E) if you wish to
exploit the empty centre to accommodate the summary value labels.
Figure 6.58 Iterations of Different Chart Options to Show the Same Data
327

This illustration demonstrates how you only know if a chart will serve your purpose once you try
it out with real data. After that, consider variations in chart and/or transformations of your data
to find the best way to show what you really want to say.
Data exploration: One consistently useful pointer to how you might visually communicate your
data is to consider which techniques helped you to unearth key insights when you were visually
exploring the data. What chart types have you already tried out and maybe found to reveal
interesting patterns? Exploratory data analysis, in many ways, offers this bridge to visual
communication: the charts you use to see data for yourself often represent prototype thinking
about how you might communicate real data to others. The way you style the chart may differ,
but if a method is already working, why not utilise the same approach again?
Editorial angle: When defining your editorial angle(s) you are expressing what specific aspect of
understanding you are attempting to portray to your viewers. This helps you to determine which
chart type might be most relevant or at least which family across the CHRTS taxonomy will
provide the best option to pick from. Always give yourself time to spend on the editorial stage,
carefully articulating what you want to say before you get too carried away with picking how.
Trustworthy design: In the discussion about tone I explained how you might sacrifice precision
in the perception of values to suit the purpose of your work. Precision in perception is one thing,
328

but precision in design is a different matter and one for which there should be no compromise.
Being accurate in your portrayal of data is a fundamental obligation. There are many ways in
which viewers can be deceived through incorrect and inappropriate encoding choices, whether
they are intended or not.
Geometric miscalculations are a common mistake. When using the area of shapes to represent
different quantitative values, the underlying geometry needs to be calculated accurately. For
example, using circular shapes to show a quantitative value of 20 compared with another of 10,
you would just half the diameter of the second, right? Wrong.
The illustration in Figure 6.59 shows the incorrect and correct ways of encoding two quantitative
values through circle size, where value A is twice the size of B. The orange circle for B has half the
diameter of A, the green circle for B has half the area of A. Using variation in diameter distorts
the perceived size of circle B as being far smaller than the value actually is. Viewers base estimates
of quantitative size through the area of a circle, not its diameter. Therefore, the green circles
demonstrate the correct way to encode these values.
Figure 6.59 The Correct and Incorrect Way to Encode Variation in Shape Size
Another representation accuracy issue causing problems for size judgements concerns truncated
axis scales. When quantitative values are encoded through the height or length of size (e.g. for bar
charts), truncating the value axis (not starting the range of quantitative values from the origin of
zero) distorts the size judgements. I will revisit this issue in Chapter 10 because it is ultimately a
consideration about the sizing of chart-scale ranges, which I deem to be a matter of composition.
Another design issue that can distort data is 3D decoration. In the majority of cases, the use of
3D charts is, at best, unnecessary and, at worst, hugely distorting. Though I concede that there
can be a certain appeal to the physical appearance of 3D charts, it is not an effective choice for
trustworthy practices. It is often seen applied to a chart when the visualiser is motivated by a
desire to demonstrate technical competence with a tool or encouraged by stakeholders who want
to see charts made to look ‘fancy’ or ‘cool’.
Using psuedo-3D decoration, when you have only two dimensions of data, is gratuitous and will
distort the viewer’s ability to judge values with any degree of acceptable accuracy. As illustrated in
Figure 6.60, when forming value estimates of the angles and sectors in the respective pie charts,
the 3D version makes it much harder to form accurate judgements. The tilting of the isometric
plane amplifies the front part of the chart and diminishes the back. It also introduces a raised
‘step’ which is purely decorative, thus embellishing the judgement of the sector sizes.
Figure 6.60 Illustrating the Distortions Created by 3D Decoration
329

For charts genuinely based on three dimensions of data, a 3D representation should only be
considered reasonable if the viewer is provided with the means to adjust the field of view. This
will help to overcome the distortion of distance and perspective, creating multiple potential 2D
viewing angles. ‘All the Buildings in Manhattan’, Figure 6.61, offers a slick interactive experience
that lets users navigate around a 3D view of New York City to observe the size of the buildings
around Manhattan. This means you can change your field of view to determine properly the
height of the modelled building shapes and make comparisons across the city.
Another legitimate application of 3D visualisation is through the potential of physical displays,
perhaps using 3D printing techniques, as demonstrated by the piece shown in Figure 6.62. This
portrays trajectories for every home run scored by Kris Bryant of the Chicago Cubs during 2017,
including the height, distance and landing position of each shot.
Figure 6.61 All the Buildings in Manhattan, by Taylor Baldwin (tbaldwin.net, @taylorbaldwin)
Figure 6.62 Representing Three Dimensions of Data (Baseball Home Run Trajectories) in a 3D
Space
330

The final matter related to trustworthiness concerns thematic mapping, specifically the often
contentious matter of choosing a map projection. The Earth is not flat. Although advances in
technology are enabling interaction with 3D portrayals of the Earth within a 2D space, the
dominant form through which maps are presented portrays the Earth as a flat surface. Features
such as size, shape and distance can be measured accurately on Earth, but when projected onto a
flat surface a compromise has to occur. Only some of these qualities can be preserved and
represented accurately. Although there are exceptionally complicated calculations attached to
each spatial projection, the main features most of us need to know about are that:
every type of map projection has some sort of distortion;
the larger the area of the Earth portrayed as a flat map, the greater the distortion;
there is no single right answer – it is often about choosing the least-worst case.
Thematic mapping (as opposed to mapping spatially for navigation or reference purposes) is
generally best carried out using mapping projections based on ‘equal-area’ calculations (so the
sacrifice is more on the shape, not the size). This ensures that the phenomena per unit – the
values you are typically plotting – are correctly represented by proportion of regional area. For
choosing the best specific projection, in the absence of perfect, the decision is usually based on
which one will distort the spatial truth the least given the level of mapping required. There are
many variables in play, however, based on the scope of view (world, continent or country/sub-
region), the potential distance from the equator of your region of focus and whether you are
focusing on land, sea or sky (atmosphere), to name but a few. As with many other topics in this
field, a discussion of mapping projections requires a dedicated text, but let me at least offer a
brief outline of five different projections (Figure 6.63).
Figure 6.63 A Selection of Commonly Deployed Mapping Projections. Images from Wikimedia
Commons published under the Creative Commons Attribution-Share Alike 3.0 Unported
Licence
331

332

Summary: Data Representation
333

Visual Encoding and Charts
This chapter introduced the act of visual encoding, the fundamentals of how you represent data
visually. All charts are based on a combination of marks and attributes:
Marks: Visual placeholders representing data items, such as distinct records or discrete
groupings.
Attributes: Variations in the visual appearance of marks to represent the values associated
with each data item.
Expanding on this introduction, you were then introduced to a wide gallery of chart types,
including profiles of 49 distinct approaches, to give you a sense of the common options that
exist. The charts were organised into five family groupings, based on what each type is primarily
used to show:
Categorical: Comparing categories and distributions of quantitative values.
Hierarchical: Revealing part-to-whole relationships and hierarchies.
Relational: Exploring correlations and connections.
Temporal: Plotting trends and intervals over time.
Spatial: Mapping spatial patterns through overlays and distortions.
334

Influencing Factors and Considerations
If these were the options, how did you make your choices? The influencing factors included:
Technological: What charts can you make and how efficiently?
Purpose: What is the intended ‘tone’ of voice your representation should convey? Where is
the emphasis between reading and feeling data?
Data type and shape: The types of data and range of values you are trying to display will
have a bearing on which charts you can use.
Data exploration: What charting methods did you use to explore your data and did any of
those represent possible means for communicating to your audience?
Editorial angle: What is the specific angle of enquiry that you want to portray visually? Is it
relevant and representative of the most interesting analysis of your data?
Trustworthy design: Avoid deception through mistaken geometric calculations, 3D
decoration, truncated axis scales, corrupt charts.
335

General Tips and Tactics
Do not arrive at this stage with fixed, preconceived ideas about wanting to use certain chart
types: be driven by your data and by your editorial thinking.
Do not be precious: acknowledge when you have made a wrong call or gone down a dead
end.
What now? Visit book.visualisingdata.com
EXPLORE THE FIELD Expand your knowledge and reinforce your learning about working with data through this
chapter’s library of further reading, references, and tutorials.
TRY THIS YOURSELF Revise, reflect, and refine your skill and understanding about the challenges of working
with data through these practical exercises.
SEE DATA VISUALISATION IN ACTION Get to grips with the nuances and intricacies of working with data in
the real world by working through this next instalment in the narrative case study and see an additional extended
example of data visualisation in practice. Follow along with Andy’s video diary of the process and get direct insight
into his thought processes, challenges, mistakes, and decisions along the way.
336

http://book.visualisingdata.com

7 Interactivity
In the previous chapter we explored a wide range of different options for representing data
visually and learnt about the factors informing your choices. In this chapter we move on to the
second element of design thinking concerning the potential features of interactivity.
It is not much more than a generation ago that most visualisations would have been created
exclusively for print. The advancement of technology has now entirely altered the nature of how
visualisations are produced, shared and consumed. The capabilities of modern devices and
proliferation of high-speed web access have created a particularly fertile landscape for talented
developers to produce engaging interactive experiences.
Not everything can, will or should be interactive. The careful judgements that characterise this
entire visualisation design process are especially important when handling this layer of anatomy.
Features of interactivity must be fundamentally justifiable. They must enhance and not obstruct
the facilitation of understanding.
In this chapter we will temporarily switch labels from ‘viewer’ to ‘user’ as this implies a more
active role for discussing interactivity. For a user to become active, there needs to be sufficient
reward for making the effort. Moreover, the visualiser must ensure that offering interactivity does
not reflect an abdication of responsibility. Do not pass on to the user the task of discovering
insights if a context necessitates the provision of an explanatory experience. This often betrays a
certain lack of commitment to editorial clarity.
However, when the circumstances are appropriate, incorporating features of interactivity into
your visualisation can offer several advantages:
It expands the physical limits of what can be consumed in a given space.
It broadens the variety of analysis to serve different curiosities within a single project.
It facilitates manipulations of data to accommodate varied interrogations.
It amplifies the overall control and potential customisation of an experience.
It increases the range of techniques for engaging users with dynamic displays.
Before deciding what features of interactivity you will use, we will first consider what you could
use. We will explore some of the common methods employed to equip you with the means for
interrogating, manipulating and navigating through rich digital experiences. As we encounter
each option, it will be useful for you to understand the distinctions between an event, the control
and the function:
The event is the user action, such as a single click of a mouse.
The control is the feature to which the event action is applied, such as a dropdown menu.
The function is the operation that is performed, such as selecting a category.
337

338

7.1 Features of Interactivity
339

Filtering
The first set of techniques (Table 7.1) enables users to specify what data they wish to include or
exclude from a chart display. This action effectively modifies the editorial ‘framing’ perspective
for the current view of data.
Table 7.1
The first example in Figure 7.1 shows an excerpt from ‘The Pursuit of Faster’, a visualisation
project looking at the evolution of result times throughout the history of the Summer Olympics.
By using the categorical menus at the top of the screen, users can modify the selection of which
sport and event results are currently displayed in the chart. Further categorical filters can then be
applied to show or hide results for different genders and for individual medal series.
‘How Nations Fare in PhDs by Sex’, shown in Figure 7.2, explores the under-representation of
women among the workforce of science and engineering organisations. This gender disparity is
revealed by looking at the PhDs awarded to men and to women across different countries and
subject areas. The interactive options provided in the dropdown menu let users select and apply
different subject filters to help reveal different patterns of disparity.
Figure 7.1 The Pursuit of Faster, by Andy Kirk and Andrew Witherley
340

Shown in Figure 7.3 is a census of the prevalence of species of trees found around the boroughs
of New York City. This initial big-picture view creates a beautiful tapestry made up of a wide
range of different tree populations across the region.
Figure 7.2 How Nations Fare in PhDs by Sex, Interactive by Periscopic; Research by Amanda
Hobbs; Published in Scientific American
341

Figure 7.3 NYC Street Trees by Species, by Jill Hubley
To observe patterns for individual tree types is hard: with 52 different tree species there are
simply too many classifications to be able to allocate sufficiently unique colours to each, as you
will learn about in Chapter 9. To overcome this functionally, the project features a useful pop-up
filter list which allows users to adjust the data on view based on revealing only the species of
selected interest. Incidentally, notice the big void where JFK Airport is located.
When browsing through the chart gallery you will have already seen the ‘treemap’ used by the
‘FinViz’ stock market analysis site (see Figure 6.26). In Figure 7.4 you can see the companion
bubble plot used to show this data from a different angle. Here, you can apply a quantitative
filter by modifying the positions of the pair of parameter handles along the x- and y-value axes.
Doing so will apply changes to the maximum and minimum scales, which means only the values
that fall into the new range will be displayed. Further options exist to change the variables plotted
on each axis, using the dropdown menu provided.
Figure 7.4 Finviz: Standard & Poor’s 500 Index Stocks (www.finviz.com)
342

http://www.finviz.com

343

Highlighting
The second group of interactive features (Table 7.2) offer visual emphasis to highlight data items
or values of interest. Whereas the filtering options we have just looked at modified what data
items would be included and excluded, these features do not eliminate from a display but create
visual or positional contrast. This may be achieved through temporarily modifying attributes of
colour or by reordering the arrangement of data items. In contrast to the filtering options, these
functions modify the editorial ‘focus’ perspective.
Table 7.2
Figure 7.5 Nobel Laureates, by Matthew Weber (Reuters Graphics)
The example in Figure 7.5 demonstrates the use of radio buttons which let you pick different
cohorts of all Nobel Laureates to emphasise the matching items across the chart display. As you
can see, the selections include focusing on women, shared winners and those who are still living
at the time. The selected Laureates are not coloured differently, rather it is the residual values that
344

are significantly lightened to create emphasis through contrast.
In Figure 7.6 we see a bump chart, produced by the Office for National Statistics (ONS), that
plots rankings for the 100 most popular names given to baby boys and girls over the past century
and beyond. As we learnt in the previous chapter, bump charts can quickly become visually
complex when there are multiple items included and the passage of their lines is chaotically up
and down. You cannot reasonably colour code 100+ discrete lines for all name categories, so, in
this case, users are able to select or enter up to six different names to see how their individual
ranking trends have formed over the period.
Figure 7.6 Baby Names in England and Wales: 2017, by ONS Digital Content team
The next example (Figure 7.7) portrays the increase or decrease in workers’ compensation
benefits by US state. This project demonstrates an example of the technique known as data
‘linking’, where hovering over a mark item in one chart display will highlight an associated item
in another chart, thus draw attention to the shared relationship. In this case, hovering over a US
state circle, in any of the grid maps, will highlight the same state in the adjacent maps to draw
your eye to their respective statuses.
Figure 7.7 Workers’ Compensation Reforms by State, by Yue Qiu and Michael Grabell
(ProPublica)
345

Linking and brushing are common approaches used in exploratory data analysis tools where you
might have several chart panels and wish to see how selected items compare across each display.
Figure 7.8 How Big Will the UK Population Be in 25 Years’ Time?, by ONS Digital Content
team
The example in Figure 7.8, once again from the ONS, looks at the UK Census estimates for
2011 and demonstrates both techniques. In this case, you use the cursor to ‘brush’ a selected
range of items from within one of the population histograms to inform calculated statistics about
the age groups you have chosen, as shown in the panels above the charts. This example also
employs ‘linking’ by highlighting the associated regions of the chart in both panels.
In ‘US Gun Deaths’ (Figure 7.9), we see a different approach taken to highlight visually a
selection of values of interest. In this piece you can use pop-up check-box lists at the bottom of
the page to select different categorical groups. The chosen data items are plotted in the chart view
above the baseline separate from the rest, and details of the selection criteria are displayed on the
left. Usefully, the ‘remove filters’ option is available in the control panel to reset the display
quickly back to the original settings. Note how the transparency of the filter menu allows the
data displayed behind it to still be seen. Though it does partially occlude the chart, it is not
entirely intruding.
Figure 7.9 US Gun Deaths, by Periscopic
346

Sorting is another way of highlighting patterns in your data. In Figure 7.10, featuring work by
the Thomson Reuters graphics team on ECB bank test results, you can see a tabular display with
interactive features in the column headers that allow you to reorder specified columns of data.
Columns of categorical data values will be ordered alphabetically; quantitative data values will be
reconfigured into ascending or descending order. You can also hand-pick individual records from
anywhere in the table to drag them to another position in the table, perhaps to promote them
towards the top of the display to facilitate easier comparisons with adjacent records.
Figure 7.10 ECB Bank Test Results, by Monica Ulmanu, Laura Noonan and Vincent Flasseur
(Reuters Graphics)
347

Participating
The techniques presented so far have modified what data you are viewing or how you are viewing
it. The next group of features (Table 7.3) involves users taking a more active role by contributing
data to help customise a participatory experience.
Table 7.3
In the project ‘Who Old Are You?’ (Figure 7.11), users are invited to enter their date of birth
using the input box and, based on the calculated age, they are taken to a customised view that
compares their age with the ages of a range of famous or celebrated people at the time of major
milestone achievements in their lives.
Figure 7.11 Who Old Are You?, by David McCandless and Tom Evans
Figure 7.12 How Well Do You Know Your Area?, by ONS Digital Content team
348

The next example (Figure 7.12), titled ‘How well do you know your area?’, employs a simple
quiz engine to challenge or confirm the knowledge users have about the demographics of their
local area. Having entered a UK postcode to establish the neighbourhood, users are asked seven
questions, such as ‘For every 100 people, how many are aged 65 or over?’ To respond, the
position of the handle can be modified along the slider to indicate a guess, which will be
illustrated by the companion waffle chart. When this estimate is submitted, a correct answer is
revealed and an indication of how close or otherwise the guess was, compared with this actual
value, is displayed.
Figure 7.13 Sugar Quiz: How Much Sugar Is in Our Food?, by Claudine Ryan, Ben Spraggon
and Colin Gourlay
349

Asking people ‘what do you think?’ and providing immediate feedback to their response is a
compelling way to challenge or reinforce people’s perceived understanding about a subject. In
this next work by ABC in Australia (Figure 7.13), a similar approach is taken to ask people ‘how
much sugar is in our food?’ using a series of 10 questions to test participants’ knowledge of the
sugar content in some of the most popular groceries. Each question is framed the same way,
asking how much of a given item can be consumed until six teaspoons of sugar have been
reached. Like the ONS quiz, users enter their estimates using a slider, with a nice additional
feature being the modified imagery to offer a visual that matches your estimate.
The next example is a project that records and reuses the data it collects. Figure 7.14 shows
screens from ‘Do you remember where Germany was divided?’, where users are invited to draw a
line representing their estimate of the route of the former border between East and West
Germany. There is no assistance provided in terms of town or city markers, so, using the pencil
cursor, you draw a line to reflect your best recollection of its shape. When you have completed
your drawing, a more detailed map view is automatically loaded up to place your suggested route
in the context of the actual route. Additionally, having collected and saved the drawings of other
participants, it shows and calculates a comparison of how accurately your drawing was compared
to other people.
Figure 7.14 Do You Remember Where Germany Was Divided? [Translated], by Berliner
Morgenpost/Funke Interaktiv
350

351

Annotating
As mentioned numerous times, some ways of representing data do not place an emphasis on
users being able to judge values easily to any degree of precision. This might be entirely
consistent with the intended tone of a project. When interactivity is available, it is possible to
address the quite reasonable appetite some users may have to see more details about the data they
are seeing.
Providing a temporary display of data details ‘on demand’ can be achieved in different ways
(Table 7.4). One issue to be aware of when creating pop-up tooltips is to ensure the place in
which they appear does not risk obstructing the view of other adjacent marks in the region you
are currently interrogating. This can be an intricate thing to handle, especially when you have a
lot of annotated detail to share.
Table 7.4
The example shown in Figure 7.15 uses a heatmap to show the relationships between different
‘Avengers’ characters. Specifically, it plots how often the main characters have appeared together
in the same comic book titles over time. The colour coding applied to a heatmap lets users form
a general sense of the main patterns of frequent and infrequent shared appearances. For those
who want more detailed values, however, they can simply hover over the chart and click on the
intersecting cell of interest. The space available to the right of the chart is then occupied with a
detailed annotation presenting images of the pair of chosen characters as well as the statistic for
how often they have appeared together. Notice also how background shading and bold font
labels are added to the heatmap to help orientate which row and column has been selected.
Figure 7.15 How the ‘Avengers’ Line-up Has Changed over the Years, by Jon Keegan (Wall
Street Journal)
Not all project layouts provide sufficient empty space to accommodate annotation in this way
352

and so pop-up displays offer a solution to overcome spatial constraints. The example in Figure
7.16 analyses the rhetoric used by US Presidents through history in the annual ‘State of the
Union’ address. Circle marks are proportionally sized to indicate the standardised frequency of
different words being used in the speeches given by each president over time. This display
facilitates a sense of the main patterns, but, to learn about the exact values, users can hover over
each circle to bring up an annotated tooltip displaying details of which year, which president and
how frequently the given word or phrase of interest was used.
Figure 7.16 History Through the President’s Words, by Kennedy Elliott, Ted Mellnik and
Richard Johnson (Washington Post)
As you will learn in the next chapter, annotation is about providing useful assistance to your
users. One key potential feature of assistance can be a ‘how to read’ guide, helping users to
understand how to read chart types.
Connected scatter plots are unfamiliar chart types to many audiences. Recognising that users may
not necessarily understand how to read them, Bloomberg’s visual data team provide a pop-up
‘How to read this graphic’ guide when they visit the project shown in Figure 7.17. This guide
can be closed but remains available for those who may need to refer to it again. The connected
scatter plot was the right choice for this analysis, showing the relationship between two
quantitative variables over time. Rather than use a different and possible inferior representation
approach, it is to the authors’ credit that they respected the capacity of their users to learn how to
read this graphical form.
Figure 7.17 Holdouts Find Cheapest Super Bowl Tickets Late in the Game, by Alex Tribou,
David Ingold and Jeremy Diamond (Bloomberg Visual Data)
353

354

Animating
Data that has a temporal dimension can present opportunities for being displayed using some
form of animated sequencing. Seeing values transition over time can sometimes expose
interesting patterns that may otherwise be hidden or imperceptible through comparing static
views in isolation.
Table 7.5
In some respects, using the label of interactivity to describe the functions (Table 7.5) of an
animated visualisation can be misplaced. For some animated pieces, one could argue they are
more a matter related to composition thinking and indeed, oftentimes, they will not actually be
controllable by any means. The increasingly popular animated gif is such an example, whereby
you effectively open it and it runs automatically.
In many cases there will be at least some control for starting and stopping an animated sequence,
like this first example in Figure 7.18, titled ‘Breathing Earth’. This work simulates the health of
vegetation around the planet between states of lush and arid, pulsing in different ways in
different places through the seasons of a year. The more a region is covered in a darker colour
shade indicates a greater measure of ‘greenness’, the shorthand term used to represent the
scientific vegetation index. The main purpose of this piece is to witness the data presented in this
dynamic fashion and to experience repeatedly this animated loop in order to find new seasonal
and spatial observations. As expressed by Nadieh Bremer in the description that accompanies her
work, ‘the more often you watch the year go by, the more the small details will start to stand
out’.
Figure 7.18 Breathing Earth, by Nadieh Bremer (Visual Cinnamon)
355

The next example (Figure 7.19) plots the distribution of height and weight of NFL footballers
over time and presents the data using an animated heatmap to help reveal this shifting pattern.
When you land on the page, the animation automatically initiates, but you can then assume
control by using the play button to start and stop the animation when you wish. Additionally,
you can grab and move the handle along the time slider manually to reposition the time frame
view.
Figure 7.19 NFL Players: Height and Weight over Time, by Noah Veltman (noahveltman.com)
356

http://noahveltman.com

Navigating
One of the main benefits of interactivity is to overcome the limitations of space. You might have
lots of detail or contents to share but not enough room to make it reasonably and simultaneously
accessible. The next group of dynamic features (Table 7.6) enable users to access multiple views
or explore greater levels of detail.
Table 7.6
The first example is simply titled ‘Earth’ (Figure 7.20) and offers a powerful, elegant and widely
used tool to explore live patterns of wind, weather and ocean conditions anywhere on the planet.
It offers a multitude of different interactive features, including another demonstration of data
being animated – wind, weather and the oceans are clearly phenomena that lend themselves to
dynamic representation. However, for the scope of this section of features, it is the enabling of
users to navigate and view the map across any position around the globe – known as ‘panning’ –
and to adjust the scale of their view – known as ‘zooming’.
Figure 7.20 Earth, by Cameron Beccario (earth.nullschool.net)
The dot map in Figure 7.21 displays an incredibly detailed representation of population density
across part of the USA. There are over 300 million dots plotted, one for each person residing in
the USA, colour coded by race and ethnicity, based on data from the 2010 Census. Like the
Earth wind map, users can pan around different locations on the map and then use a scrollable
zoom or scaled zoom buttons incrementally to change the level of detail displayed. This act of
zooming to increase the magnification of the view is known as a geometric zoom, effectively re-
framing the included and excluded data through the window at each scale level.
Figure 7.21 The Racial Dot Map: Image Copyright, 2013, Weldon Cooper Center for Public
357

Service, Rector and Visitors of the University of Virginia (Dustin A. Cable, creator)
Both these recent works demonstrate methods for navigating vast landscapes in detail, enabling
users to dictate which views and levels of scale to explore the data through. An alternative
approach to navigating involves the offer of a more linear, explanatory experience, building up a
narrative about a subject through a series of discrete sequences. It is arguably the quintessential
example of storytelling with data and is often presented using a technique known as
‘scrollytelling’, whereby users scroll to move vertically up and down through the steps of a story.
Figure 7.22 Killing the Colorado: Explore the Robot River, by Abrahm Lustgarten, Al Shaw, Jeff
Larson, Amanda Zamora and Lauren Kirchner (ProPublica) and John Grimwade
The project featured in Figure 7.22 is a prime exhibit of this kind of dynamic interface. It offers a
step-by-step journey down the length of the Colorado River to investigate the impact of some of
the major infrastructure projects that have caused the gradual draining of this vital source of
water for millions of Americans. To break out of the linear navigation, users are also able to jump
ahead or back to different chapters of interest using the left-hand menu. This can be a
particularly useful feature after you have been through the full sequence once. Another helpful
device is the inclusion of a thumbnail image to help orientate the location of current focus within
the context of the overall journey down the river.
Sequentially building up a story can prove to be a powerful way of facilitating understanding. In
358

Figure 7.23, the project ‘What’s Really Warming the World’ presents a sequence of possible
causes for climate change. As you scroll down the page (or, alternatively, click through the page
steppers or directional arrow) it takes you through the different hypotheses, overlaying data about
each onto a chart plotting the observed changes in temperature. Eventually, you reach the
conclusion and the big reveal: it is due to greenhouse gases.
Figure 7.23 What’s Really Warming the World?, by Eric Roston and Blacki Migliozzi
(Bloomberg Visual Data)
Figure 7.24 100 Years of Tax Brackets, in One Chart, by Alvin Chang
Figure 7.25 OECD Better Life Index, by Moritz Stefaner and Dominikus Baur, Raureif GmbH
359

The work shown in Figure 7.24 looks at 100 years of tax brackets in the USA. This project
employs a similar sequencing approach to unfold content, but rather than stepping through a
series of different charts or discrete views of a subject, the sequence here gradually builds up the
user’s understanding about the subject matter. It steps through information about why tax
brackets are a relevant topic, what they are, how they affect people, and what are some of the
main historical patterns to have emerged from this analysis. When you have technical topics like
this it might trigger indifference among an audience through their lack of domain knowledge.
So, rather than drop them straight into the deep end, a skilfully executed and carefully
considered step-by-step presentation like this, coaching them rather than exposing them, can lead
to increased engagement.
A final demonstration of a navigating device is characteristic of a ‘drill-down’ feature, giving users
access to data which might exist at a lower granularity or hierarchical level. The flower
representations used in the Better Life Index project (Figure 7.25) effectively co-exist as menu
items. These offer ways of choosing a country to navigate to a separate report that provides far
more detailed analysis and commentary, supplementing the summarising data as displayed in the
initial flower view.
360

7.2 Influencing Factors and Considerations
You should now have established a good sense of the wide range of possibilities for incorporating
interactive features into your work. So let’s turn our attention to consider the factors that will
have most influence on which of these techniques you might need to or choose to apply.
Constraints: The main factor that will shape the scope for employing interactivity is
unquestionably the technical skills you possess and the capabilities of the technology you have
access to. If you are technically limited in being able to develop any of the features profiled, it
immediately rules them out of your thinking. In the online resources that accompany this book, I
include a guide through some of the contemporary applications, tools and programming libraries
that enable you to develop visualisations with interactive features. So, looking beyond technology
for now, another major constraint will exist through the pressure of time. If you have a limited
time frame in which to complete your work, you are going to find it a challenge to pursue
particularly ambitious or bespoke interactive solutions, even if you possess extensive technical
competence.
Deliverables: Understanding the expected deliverables of your work is vital. Just because a
visualisation might be created and published digitally, the output may still be non-interactive:
digital does not necessarily or automatically mean interactive. If your work is for print only,
interactivity is not needed, and this entire chapter of thinking will be outside your radar of
concern.
The main question to ask is whether the characteristics of the setting in which your audience will
consume a visualisation are compatible with the prospect of needing them to interact. Will your
audience have the time, the patience and indeed the know-how to exploit such features?
Additionally, what are the varied device specifications on which your solution will need to
function? To what extent will you be seeking to emulate the same experience across mobile,
tablet and desktop devices? How adaptable might your solution need to be, and might there be a
need for compromises?
Where once we were limited to the mouse or the trackpad as the common peripheral, over the
past decade we have seen the emergence of touch-screens, through smartphones and tablets. This
has introduced a whole new challenge for developers to find ways of creating consistent solutions
that are compatible across different platforms.
For simplicity and consistency, in this chapter we have focused on events associated with using a
mouse or trackpad, but here is a translation of the equivalent touch events (Table 7.7). The
primary difference between the two concerns the inability to enact the equivalent of a hover (or
‘mouseover’) event with touch-screens.
361

Table 7.7
Purpose: Though all visualisations that offer exploratory experiences will need to be interactive,
not all interactives are exclusively for offering exploratory experiences. Some of features we have
profiled, such as those that navigate through sequences, are classic demonstrations of using
interactivity to fulfil an explanatory experience.
Your definition about what experience to offer will inform the features of interactivity you may
seek to incorporate. As mentioned in Chapter 3, there may also often be scope for a blended
approach. For instance, you might open a visualisation with an explanatory experience, based
around showing some main findings and telling your audience something. Through interactivity,
you may then transition the users towards more of an exploratory interface that invites them to
interrogate the data to pursue their own particular curiosities about the subject.
Data representation: Some charts are inherently visually complex, and this can create obstacles
for the user trying to understand them. The bump chart, chord diagram and Sankey diagram are
just a few of the charts that commonly have multiple lines and bandings crossing over in the
same space. Offering interactive features that enable filtering and/or highlighting of certain data
items can help them become a more palatable prospect for the user to engage with. The Sankey
diagram shown in Figure 7.26 looks at a month of data about animals entering a shelter. The left
side of the diagram displays the origin stories which are connected with the proportional bands to
the associated outcomes on the right side. With so many crossing paths, the ability to choose a
single category on the origin or outcome side of this display helps reveal some of the important
stories more discernibly. The chart becomes more readable and, by extension, more usable.
Figure 7.26 A Month in an Animal Shelter, by Sarah Campbell
Trustworthy design: The reliability, consistency and functional performance of a visualisation is
362

something that influences the perceived ‘trustworthiness’. Does it do what it promises, and can
the user trust the functions that it performs?
Inevitably, issues around data privacy and intended usage of data collected will be important
matters to handle with integrity and transparency, otherwise the trustworthiness of your work
may be critically undermined. In most projects, any data contributed by a user is collected only
for a temporary period of participation (i.e. not held beyond the moment of usage). However, if
you intend to collect and save this data, perhaps to append to an original dataset and use this to
improve the content, you should make your intentions very clear to the user.
‘Confusing widgets, complex dialog boxes, hidden operations, incomprehensible displays, or slow response times …
may curtail thorough deliberation and introduce errors.’ Jeff Heer and Ben Shneiderman, taken from ‘Interactive
Dynamics for Visual Analysis’ (2012)
Is it a one-off piece of work or something that will run on regularly updated data? In which case
how robust is the design going to be to accommodate new data? Will it cope with new categories
and larger or smaller value ranges? Who will keep running it and who will support it thereafter to
ensure it continues to offer a quality experience that preserves the trust of its users?
Accessible design: Seek to minimise the friction between the act of using an interactive feature
and the understanding it facilitates. Do not create unnecessary obstacles that stifle curiosity.
Indeed, resort to interactivity only when you have exhausted the possibilities of an appropriate
and effective static solution.
For example, let’s consider features of animating. If your data is not changing much, an animated
sequence may be of no merit. If it is changing a lot, maybe it will be too chaotic and the
movements too sudden to be observable. Your intention may have been to exhibit this chaos, but
the main value of animated sequences should be to help reveal dynamic patterns of change rather
than random variation.
It really depends on what it is you want to show: the dynamics of a ‘system’ that changes over
time or a comparison between different states over time? With animated sequences, there is a
reliance on memory to conduct a constant comparison of change. Yet, our recall ability is fleeting
at best and weakens the further apart (in time) the basis of the comparison has occurred.
The speed of an animation is also a delicate matter to judge as you seek to avoid the
phenomenon of change blindness. Rapid sequences will cause the stimulus of change to be
missed; a tedious pace will dampen the stimulus of change and key observations may be lost.
Access to the available understanding becomes diminished.
If you wish to facilitate direct comparisons you ideally need to juxtapose individual frames within
the same view. The common technique used to achieve this in visualisation is through small
multiples, where you repeat the same representation for each moment in time of interest and
present them collectively in an adjacent view, often through a grid layout. This enables far more
incisive comparisons. The famous ‘Horse in Motion’ work by Eadweard Muybridge (Figure
7.27) was carried out to learn about the galloping form of a horse by seeing each stage of the
motion through individually framed moments.
363

‘Generations of masterpieces portray the legs of galloping horses incorrectly. Before stop-gap photography, the
complex interaction of horses’ legs simply happened too fast to be accurately apprehended … but in order to see the
complex interaction of moving parts, you need the motion.’ [Paraphrasing] Barbara Tversky and Julie Bauer
Morrison, taken from Animation: Can it Facilitate?
Elegant design: The self-discipline required to avoid the temptation of feature creep is
indisputable at this stage of the design process. For those people with a natural technical flair,
there can be a strong temptation to incorporate interactivity when it is neither required nor
helpful: just because you can does not mean to say you should.
Judging the degree of flexibility is something of a balancing act: you do not want to overwhelm
the users with more adjustments than they need, nor do you want to narrow the scope of their
likely interrogations. If the characteristics of your audience are varied, you may be
understandably inclined to try including more features than are necessary. For a one-off project
you have to rely on your own best judgement; for projects that will be repeatedly used you will
have more potential to seek and accommodate feedback to inform refinements.
Figure 7.27 The Horse in Motion, by Eadweard Muybridge. Source: United States Library of
Congress’s Prints and Photographs division, digital ID cph.3a45870
Elegance in designing interactive features extends to their appearance and the seamlessness with
which they can be accessed and used. With visualisation you are aiming to make invisible insights
visible. Conversely, the features of visible design should be as inconspicuous and intuitively
packaged as possible.
The captivating quality of a well-conceived interactive visualisation is how it can introduce new
means of engaging with data that simply could not have been delivered without the incorporated
features. It is important to acknowledge that there can be pleasure created by thoughtfully
conceived interactivity. Even if there are features that offer only ornamental benefit, a sense of
fun and playability can be appealing to any audience type, so long as the circumstances are right
and such features do not obstruct access to understanding.
364

365

Summary: Interactivity
366

Features of Interactivity
This chapter introduced the potential value of incorporating interactive features into your work,
profiling a wide range of options that will enable users to interrogate and control a visualisation.
These included:
Filtering: Enabling users to specify what data they wish to include or exclude from a chart
display.
Highlighting: Features that apply visual emphasis to highlight data items or values of
interest.
Participating: Inviting users to contribute data to help customise a participatory experience.
Annotating: Offering users ways to see more details about the data they are seeing.
Animating: Displaying data with a temporal dimension using animated sequencing.
Navigating: Features that enable users to access multiple views or explore greater levels of
detail.
367

Influencing Factors and Considerations
If they were the options, how did you make your choices? The influencing factors included:
Constraints: The technology and skills possessed, as well as the timescales, will shape
ambitions.
Purpose: What experience are you facilitating and how might interactive options help
achieve this?
Data representation: Certain chart choices may require interactivity to enhance readability.
Trustworthy design: Functional performance and reliability will substantiate the perception
of trust from your users.
Accessible design: Interactive features should be useful and unobtrusive – minimise the
clicks.
Elegant design: Beware of feature creep but embrace the potential of fun and playability.
368

General Tips and Tactics
Good project management is critical when considering the development of an interactive
solution.
Do not be distracted by working on interaction features that seem ‘cool’ or ‘fancy’, but do
not add enough value to warrant precious resources being allocated (time, effort, people).
Keep focusing on what is important and relevant. A technical achievement may be great for
you and your CV, but is it needed for the project?
What now? Visit book.visualisingdata.com
EXPLORE THE FIELD Expand your knowledge and reinforce your learning about working with data through this
chapter’s library of further reading, references, and tutorials.
TRY THIS YOURSELF Revise, reflect, and refine your skill and understanding about the challenges of working
with data through these practical exercises.
SEE DATA VISUALISATION IN ACTION Get to grips with the nuances and intricacies of working with data in
the real world by working through this next instalment in the narrative case study and see an additional extended
example of data visualisation in practice. Follow along with Andy’s video diary of the process and get direct insight
into his thought processes, challenges, mistakes, and decisions along the way.
369

http://book.visualisingdata.com

8 Annotation
The third element of the visualisation design anatomy is annotation. This concerns judging the
level of assistance an audience may require in order to understand the background, function and
purpose of a project, as well as what guidance needs to be provided to help viewers perceive and
interpret the data representations.
In contrast to the more theoretical and technical concerns around data representation, colour and
interactivity, judgements about what annotated features to offer your viewers can be more heavily
informed by common sense. This is an influential but often neglected layer of thinking that
really exposes the amount of care a visualiser shows towards the audience.
The sequence of suggestions roughly follows the typical organisation of the layout of your work,
beginning with the features that might typically exist at the start of an experience, working
through to those usually found towards the end. Towards the end of the chapter we will look at
the factors that will influence your choices, but first let’s profile some of the key features of
annotated design you might consider including in your visualisation.
370

8.1 Features of Annotation
371

Headings and Introductions
The primary aim of a heading is to inform your viewers efficiently about the content they are
about to encounter and to orientate themselves within the hierarchy of this content. Main
headings typically occupy prominent places in your project’s layout, perhaps as a title to
introduce a report or at the top of a page or screen.
The suitability of your heading comes down to the language used: what are you going to use this
key feature to say? There is no universal practice for what constitutes good use of headings; it will
vary considerably between subject areas, project contexts and audience settings. However, I find
there are generally four approaches to constructing and using them, as follows.
Statement: These are short headline forms of titles that may highlight a key observation or
finding that emerges from a visualisation work. The statement title (Figure 8.1) might be most
commonly used with visualisations that offer an explanatory experience, based on the mantra ‘if
you have something to say, say it’.
Figure 8.1 Examples of ‘Statement’ Titles
Question: A title presented as a question (Figure 8.2) can offer a compelling way to align your
audience’s minds with the essence of the curiosity that has driven the project. It prepares them to
inherit an appetite to find some notion of an answer to the question posed, which the
visualisation should serve. These titles work well for exploratory visualisation experiences.
Figure 8.2 Examples of ‘Question’ Titles
Descriptive: These types of titles generally articulate what is represented on a chart. They are
more functional and less editorial in style, but perhaps more informative about what the viewer is
about to encounter. Characteristically, descriptive headings (Figure 8.3) tend to be aligned with
exhibitory visualisations. This approach would also usually be applied to lower level headings,
such as section dividers and localised chart titles, offering details about what each element is
about.
372

Figure 8.3 Examples of ‘Descriptive’ Titles
Artistic: This approach tends towards the use of short, succinct and rather enigmatic phrases to
convey roughly the nature of the topic, but mainly to pique the curiosity of the audience. The
titles are consistent in style with titles typically used for creative endeavours like movies or
artworks (Figure 8.4). Due to their punchier size they might be more easily remembered than
other approaches. However, they are often necessarily supplemented by more descriptive sub-
headings that expands on the detail.
Figure 8.4 Examples of ‘Artistic’ Titles
Introductions are commonly provided in close proximity to a heading in the form of short
paragraphs that concisely explain in further detail what a project is about, why it exists and what
it is for. The content of this introduction might usefully explain matters such as:
details of the reason for the project, perhaps articulating the origin story of the curiosity;
an explanation of the relevance of this analysis;
a description of the analysis that is presented;
a few comments about the main messages or findings the work is about to reveal.
The extracted introduction shown in Figure 8.5 accompanies a main heading to explain the
background of the subject, why the analysis has been undertaken, and information about the
experts providing their headline observations.
Figure 8.5 Excerpt from History Through the President’s Words, by Kennedy Elliott, Ted
Mellnik and Richard Johnson (Washington Post)
373

Introductions are, naturally, logically offered near the top or start of a project. Sometimes,
through interactivity they are made available on demand through a separate window or pop-up
to provide the necessary details. This would be appropriate if they were quite detailed in nature
and would otherwise occupy too much precious space. Furthermore, the viewer might encounter
the work on a repeated basis, but will only need to read an introduction on the first occasion.
For some projects, the introduction may be used to provide a more extensive description of the
data, where data has come from, how it has been transformed, and comments about any
assumptions or potential shortcomings.
374

User Guides
Projects that include features of interactivity may need to offer some level of instruction in the
form of prompts or more in-depth user guides. Though the features may not necessarily be overly
technical – and easy to learn how to use – instructions can help to enhance the accessibility of
your project. In Figure 8.6, there is an input box inviting users to enter a job title. The short
instructive sentence explains what to do, and the example text provides clues about the format
and phrases you might attempt to search for.
Figure 8.6 Excerpt from Making Sense of Skills: A UK Skills Taxonomy, by Dr Cath Sleeman
It can be a mistake to assume that every user will be sufficiently sophisticated to understand
immediately the workings of the functionality you are offering. It can also be a mistake to assume
that every user will find all the functions you are offering. A more dedicated user guide that
introduces and explains the full repertoire of features might be necessary. Projects like ‘Kindred
Britain’ (Figure 8.7) provide a vast array of means for exploring aspects of history about the
British royal family and aristocracy. Without offering a detailed user guide, many users may miss
out on some of the skilfully crafted opportunities to interrogate the data. It is therefore in
everyone’s interest to provide this type of assistance. Moreover, it is in everyone’s interest to
provide this assistance using an elegantly presented accessible form, as this example certainly
exhibits.
Figure 8.7 Kindred Britain, version 1.0 © 2013 Nicholas Jenkins – designed by Scott Murray,
powered by SUL-CIDR
375

376

Reader Guides and Legends
Reader guides offer a different type of assistance to user guides in that they focus on helping
viewers to understand how to read a chart. If you have used an unfamiliar and/or particularly
complicated chart type, with many attributes that need to be decoded and synthesised, you might
need to provide instructions to assist viewers who need it.
The first reader guide example, shown in Figure 8.8, comes from work designed by Accurat, a
studio renowned for innovative and expressive representation techniques. Given the relative
complexity of the encodings used in this piece, it is necessary to equip the viewer with guidance
about the layout of each chart panel, what the shading portions and arced lines represent, what
the dots and musical notes stand for, and what to imply from the length of the bars.
Figure 8.8 The Life Cycle of Ideas, by Accurat
In Figure 8.9, you can see another guide taken from the Gantt chart we saw in the chart gallery,
looking at timeline histories of the current and former US national parks. This offers a
description about the arrangement of the items, the associations of several elements of symbol
and colour usage, as well as definitions of the acronyms used.
Figure 8.9 Establishment of the US National Parks, by Nicholas Rougeux (www.c82net)
There are many similarities between reader guides and legends. A legend is usually positioned
adjacent to a chart and contains one or several keys used to explain associations between
categorical attributes or classifications of quantitative scales. Figure 8.10 presents a range of
377

http://www.c82net

different examples from work you will encounter across this book. The main difference between
a legend and a reader guide is that a legend offers far less written explanation, whereas a reader
guide more actively coaches a viewer through the reading task.
Figure 8.10 Selection of Example Legends
There are different ways of being creative in how you portray a legend. In Figure 8.11 you will
see how colour associations are integrated into the introductory text, with certain words
highlighted through shading or in the colour of their font. Rather than having a separate colour
key elsewhere on the page, this saves space and provides all the setup information and reading
instructions in the same place.
Figure 8.11 Boom and Bust: The Shape of a Roller-coaster Season, by Andy Kirk
In Figure 8.12, you can see a smart way of squeezing more information out of a legend in a
visualisation that analyses the language of tweets posted around New York City. The legend takes
the form of a bar chart that acts as both a colour key, explaining the association with the different
languages, and a method of showing a quantitative summary of the total number of tweets for
each language.
Figure 8.12 Twitter NYC: A Multilingual Social City, by James Cheshire, Ed Manley, John
Barratt and Oliver O’Brien
378

379

Chart Apparatus and References
Chart apparatus relates to the range of structural components found in different chart types, such
as axis lines, gridlines or tick marks. I consider these elements of annotation because their role is
to help orient viewers in making judgements about size and/or position.
Not all chart types have the same structural apparatus. For example, a pie chart does not have
axis lines, a Sankey diagram will not have gridlines. The scatter plot shown in Figure 8.13 is just
one example of a work we have seen earlier that includes the full array of chart apparatus. There
is no fixed recommended approach on whether to include or exclude these features, but the
default treatment applied by most chart applications would generally become quite heavy with
the apparatus. Your choices will generally be informed by the degree of emphasis you place on
viewers efficiently judging values with precision and also be influenced by your desired
presentation style. Mentioning these features in this section is as much about prompting you not
just to go through the motions by accepting default thinking.
Beyond the structural elements of chart apparatus, you may find value in incorporating
additional markings on your charts to help viewers with the task of interpretation. Chart
references can be usefully included as visual overlays to provide context of scale, to clarify the
expected and unexpected, and to separate the normal from the exceptional. In some ways these
features might be considered an extension of data representation, as they will be formed through
values of data, but I see them more as annotations focused on providing assistance. There are
several different types of references that may be useful to include:
Bandings: These are typically shaded areas that provide some frame of contextual judgement
for data values, such as providing a range of historic or expected values. As we saw in the
bullet chart in the chart gallery, there are various shaded regions that might help to indicate
whether the bar’s value should be considered great, good or just average. They might also be
used to indicate confidence intervals to convey probabilities or to represent a contextual
measurement of margin of error to explain degrees of uncertainty.
Markers: Adding points or symbol markers to a chart might be useful to help indicate
statistical features such as comparison against a target, forecast or a previous value of note.
Reference lines: These are useful in chart displays that use position or size along an axis as an
attribute for a quantitative value. Line charts or scatter plots are particularly enhanced by
the inclusion of reference lines, helping to direct the eye towards calculated trends,
constants or averages and, with scatter plots specifically, the lines of best fit or correlation.
The example in Figure 8.14 uses a heat map display to show the relative incidence of
measles per 100,000 population across each US state over time. The patterns were already
indicating a fascinating trend but, by adding the reference line to indicate when the vaccine
was introduced in 1963, the compelling story of cause and effect jumps off the page.
Figure 8.13 Mizzou’s Racial Gap Is Typical on College Campuses, by FiveThirtyEight
380

Figure 8.14 Battling Infectious Diseases in the 20th Century: The Impact of Vaccines, by
Graphics Department (Wall Street Journal)
381

Chart Labelling and Captions
There are three main features of labelling that you will need to consider adding to your charts,
depending on the type of chart you are using. As demonstrated, again, by the chart in Figure
8.13, these include axis titles, axis scales and value labels:
Axis titles describe what values are plotted along each axis. This might be a single word or a
short sentence depending on what best fits the needs of your viewers. Often the role of an
axis is already explained (or implied) by headings or introductions, but do not always
assume this will be automatically clear to your viewers.
Axis scales provide references along each axis to identify the categorical items, quantitative
value intervals or the dates with a time frame. For categorical axes (as seen in bar charts and
heat maps) one of the main judgements relates to the readability of the label and how well it
fits into the space you have. For non-categorical data the main judgement will be what
quantitative intervals to use. This will be shaped by considering the most relevant interval
for the subject matter and the required precision in readability. It can also come down to
what offers the most elegant rhythm in your display – does it feel too fussy or too sparse?
Depending on your editorial framing definitions, you might also expand your quantitative
range outside the observed value range in order to use empty chart space to support a point
of narrative.
Value labels will appear in proximity to specific mark encodings inside a chart. Typically,
these labels will be used to reveal a quantity, such as showing the percentage sizes of the
sectors in a pie chart or the size of categorical bars. Having the option to reveal values
through interactive annotations can be a nice option, as it reduces clutter from a display.
Unless you are highlighting important values for key items, if you have clear axis labels you
should not need to double up your value reading assistance. Choose one or the other.
Captions are often included in explanatory projects to offer more localised comments that
transcend the general role of an introduction, user or reader guide. The excerpt shown in Figure
8.15 comes from an interactive project that offers users scrollable navigation through discrete
sequences of analysis about the history and growth of the #MeToo movement. At each milestone
stage, captions are displayed to offer historical narrative and context to where we are in the story.
Furthermore, more detailed value labels emerge, in the form of significant tweets, showcasing
some of the most influential actions, actors and conversations in this campaign.
Figure 8.15 MeTooMentum, by Valentina D’Efilippo (design) and Lucia Kocincova
(development)
382

In ‘US Gun Deaths’ (Figure 8.16), there is a clever feature that combines annotated captions
with interactive data adjustments. Below the main chart there is a ‘What This Data Reveals’
section which presents some of the main findings. The captions double up as clickable shortcuts
that, when selected, quickly apply the relevant filters to the main display so users can see in the
chart the relevant data that supports each commentary.
Figure 8.16 US Gun Deaths, by Periscopic
Figure 8.17 Excerpt from Wealth Inequality in America, by YouTube user ‘Politizane’
383

As creative tools become more ubiquitous, there are new opportunities for incorporating audio as
a means for verbally narrating a subject and explaining key messages. One of the standout
projects using this approach in recent years was the video ‘Wealth Inequality in America’ (Figure
8.17), as mentioned in Chapter 3. In this videographic, the voiceover provides a compelling and
cohesive narrative about the subject with the visuals supporting what is being described.
384

Footnotes and Methods
Often the final visible feature of your display, footnotes provide a convenient place to share useful
information that further substantiates the transparency of your work:
Data sources from where your raw material was acquired should always be provided, ideally
in close proximity to the relevant charts if several are to be included.
Credits will list the authors and main contributors of the work, often including details about
methods of contact for further information or feedback.
Attribution is important if you wish to recognise the influence of other people’s work in
shaping your ideas or, for instance, to acknowledge your use of open source applications or
typeface.
Usage information might explain the circumstances in which the work can be viewed or
reused, whether there are any confidentialities or copyrights involved.
Time/date stamps are useful to include so they indicate the moment of production/
publication and from which it will be possible to determine the work’s ongoing accuracy
and relevance.
Annotation is one of the most important aids to ensure you secure and sustain trust from your
viewers by demonstrating integrity and openness. If you have to undertake a lot of work to
transform your data for use in your analysis, it may be necessary to provide more detail using a
methods statement, an example of which is shown in Figure 8.18. In the spirit of ‘show your
workings’, these sections will typically extend beyond comments about the sources and methods
of collecting data to explain any assumptions that have been made, details of calculations applied,
the criteria for editorially framing the work (inclusions/ exclusions), and maybe what
imperfections existed in the data and how you have handled them.
Figure 8.18 The Pursuit of Faster, by Andy Kirk and Andrew Witherley
385

8.2 Influencing Factors and Considerations
You have now been introduced to the roles of different annotation options, so how do you
decide what types of annotations to incorporate and what level of assistance to offer your
audience? Let’s look at some of the main factors and considerations that will influence your
decisions.
Audience: Given that most annotations serve the purpose of viewer assistance, your decisions will
inevitably be influenced by the characteristics of your intended audience. Having an appreciation
of and empathy towards the knowledge and capabilities of the different cohorts of viewers is of
principal concern. It can be hard to find a solution that suits all, especially if your viewers are
diverse, but here are some of the main issues to consider:
Subject: How well acquainted are they with your project’s subject matter? Will they
understand what the data is about? Will they understand language, such as specific
terminology, acronyms or abbreviations?
Perceiving: How familiar are they with the chart type(s) you have used? If they are
unfamiliar, is it easily learnable or will they need some explanations?
Interpreting: Will they know how to interpret the meaning of what is presented? Do they
need help in determining what features are good or bad, significant or insignificant?
Interactive functions: How confident will they be in understanding how to use different
features of interactivity?
‘Think of the reader – a specific reader, like a friend who’s curious but a novice to the subject and to data-viz – when
designing the graphic. That helps. And I rely pretty heavily on that introductory text that runs with each graphic –
about 100 words, usually, that should give the new-to-the-subject reader enough background to understand why this
graphic is worth engaging with and sets them up to understand and contextualize the takeaway. And annotate the
graphic itself. If there’s a particular point you want the reader to understand, make it! Explicitly!’ Katie Peek,
Visualisation Designer and Science Journalist, on making complex and/or complicated subject matter accessible
and interesting to her audience
Setting: Providing a sufficient amount of assistance is about balance: too much assistance makes
the annotations included feel overburdening; too little and there is too much scope for
misconceptions and misinterpretation to prosper.
A setting that is consistent with the need to deliver immediate insights will need annotations to
help fulfil this rapid exchange of understanding. There will be no time for long introductions or
patience with explanations about how to read charts. Conversely, a visualisation about subject
matter that is inherently complex may warrant more assistance and invite the viewer to embark
on a process of learning. If there is time for a viewer to engage with a user or reader guide, it is
entirely reasonable to include such features since the rewards should outweigh the efforts
involved.
Purpose: Your definitions for the intended tone and experience of your work will influence the
type and extent of annotation features required. If you are working on a solution that leans more
towards the ‘reading’ tone, you are placing an emphasis on the perceptibility of the data values. It
therefore makes sense that you should aim to provide as much assistance as possible (especially
386

through extensive chart annotations) to maximise the efficiency and precision of this reading
process. In combination with your editorial ‘focus’ thinking, you might choose to emphasise
specific items over others, in which case: What are the criteria and reasoning for this? Will the
selective labelling be fixed, or should it be driven by interactive selection?
If you are providing an ‘explanatory’ experience it would be logical to employ as many devices as
possible that will help inform your viewers about how to read the charts (assisting with the
‘perceiving’ phase of understanding) and also bring some of the key insights to the surface,
making clear the meaning of the quantities and relationships displayed (thus assisting with the
‘interpreting’ phase). The use of captions and reference markings will be particularly helpful in
enabling this.
‘Exploratory’ experiences are likely to need instructive guides, ensuring that viewers (or
specifically, in this case, users) have as much understanding as possible about the functional
controls available. Features like reader guides, chart apparatus and labelling will still be relevant,
irrespective of the intended experience. Characteristically, ‘exhibitory’ work includes only chart-
level annotation, as it is more about providing a visual display of the data rather than offering
explanatory insights or instructions for exploratory interrogation. The assumptions with
exhibitory experiences are that the audience do not require extensive assistance.
Accessible design: Although equally connected with concerns about elegance, decisions on
typography will have an influence on the accessibility of your work. As you will have observed,
many features of annotation utilise text. This means your decisions will be concerned not just
with what text to include, but also with how it will look. Typography is another element of data
visualisation design thinking that exists as a significant subject in its own right, but here is a short
guide to inform your thinking:
A typeface is the styled glyphs representing individual letters, numbers and other symbols.
Tahoma and Century Gothic are different typefaces. A typeface can have one or many
different fonts in its family.
Fonts are variations in the dimension of your typeface, such as weight, size, condensation
and italicisation. This font and this font both belong to the Georgia typeface family but
display variations in size, weight and italicisation.
Serif typefaces are characterised by extra little flourishes in the form of a small line at the
end of the stroke in a letter or symbol. Garamond is an example of a serif font. Serif
typefaces are generally considered to be easier to read for long sequences of text (such as the
full body text) and are especially used in print displays.
Sans-serif typefaces have no extra line extending the stroke for each character. Verdana is an
example of a sans-serif typeface. These typefaces are commonly used for shorter sections of
text, such as axis or value labels or titles, and for screen displays.
Your typeface and font choices should be based on optimising the legibility and meaning of text
elements across your display:
In terms of legibility, viewers need to be able to read the words and numbers on display
without difficulty. Quite obvious, really. Some typefaces (and specifically fonts) are more
easily read than others. Some are better applied to help make numbers clearly readable,
387

others work better for words and passages or sentences.
Just as variation in colour implies meaning, so does variation in typeface and font. If you
make some text capitalised, large and bold-weight, this will suggest it carries greater
significance and portrays a higher prominence across the object hierarchy than text
presented in lower case, with a smaller size and thinner weight. In general, you should seek
to limit the variation in font where possible and only vary it when the property it is applied
to needs to be distinguishable from other properties.
Deciding on the most suitable choice of typeface and variety of font is something that will
ultimately come down to experience and being influenced by other creative work you encounter.
We all have our own preferences but, in practice, I find most typographic decisions I make rely
on experimentation.
Elegant design: A final judgement about annotations concerns avoiding the potential clutter and
obstruction caused by them. Any annotation feature included in your work will add more
content. These features have to be located somewhere. Too much and the display becomes
cluttered, overwhelming and potentially undermines the intention of being helpful; too little and
viewers may be faced with the demands of working things out themselves, when assistance would
make that prospect far easier.
388

Summary: Annotation
389

Features of Annotation
This chapter described the importance of providing useful assistance to your viewers, introducing
some of the many helpful features to consider, including:
Headings and introductions: Titles, subtitles and section headings, often combined with
longer passages to describe the background and aims of a project.
User guides: Advice or instructions on how to use interactive features.
Reader guides and legends: Detailed instructions advising viewers how to perceive and
interpret the chart, describing the associations between data values and attribute
classifications.
Chart apparatus and references: Structural components found in different chart types, such
as axis lines, gridlines or tick marks, as well as markings that assist with interpretation.
Chart labelling and captions: Axis titles, axis labels, value labels and commentaries.
Footnotes and methods: Include data sources, credits, and time/date stamps. May be
expanded to provide more detailed description of data handling processes, assumptions and
shortcomings.
390

Influencing Factors and Considerations
If these were the options, how did you make your choices? The influencing factors included:
Audience: Considering the characteristics and needs of the audience to determine what
assistance they might need.
Setting: Will the audience have the scope to engage with annotations if the encounter is
characterised by time pressures?
Purpose: The tone and experience offered will influence the type of annotations required.
Accessible design: Many annotations are based on text displays and so you need to consider
the legibility of the typeface you choose and the logic behind the font–size hierarchy you
display.
Elegant design: Minimise the clutter.
391

General Tips and Tactics
Attention to detail is imperative. All introductory information, project instructions,
captions and value labels need to be accurate. Always spell-check digitally and manually and
ask others to proofread if you are too ‘close’ to the work to see it rationally.
Testing with sample members of an audience may save you a lot of pain by intercepting any
shortcomings or excesses in the annotations you plan to offer.
What now? Visit book.visualisingdata.com
EXPLORE THE FIELD Expand your knowledge and reinforce your learning about working with data through this
chapter’s library of further reading, references, and tutorials.
TRY THIS YOURSELF Revise, reflect, and refine your skill and understanding about the challenges of working
with data through these practical exercises.
SEE DATA VISUALISATION IN ACTION Get to grips with the nuances and intricacies of working with data in
the real world by working through this next instalment in the narrative case study and see an additional extended
example of data visualisation in practice. Follow along with Andy’s video diary of the process and get direct insight
into his thought processes, challenges, mistakes, and decisions along the way.
392

http://book.visualisingdata.com

9 Colour
Having now established the charts you intend to use, the interactive features that might be
required and the elements of annotation that will be useful, you have effectively selected all the
visible contents of your visualisation. The remaining two layers of design thinking are concerned
with the appearance of these contents. In the final chapter after this we will consider decisions
about composition, but before that we look at the critical issue of making choices about colour.
Colour is a most potent visual stimulus. The choices we make will have an immediate impact on
the eye of the viewer, offering sensory cues about the meaning and organisation of a display.
Variation in colour implies significance. When a colour looks like it conveys meaning, the viewer
will think about that and spend time establishing what the meaning is. Many of the chart types
employ an attribute of colour to represent data values. Whether it is used to classify quantitative
scales or to associate with discrete categorical values, there is a lot riding on your colour choices
being astutely judged.
The chapter opens with an overview of colour models, offering a foundation for your
understanding about this topic. After that you will learn about the different ways and places in
which colour is to be used, starting from inside a chart and then working outwards across the rest
of the visualisation anatomy. Your objective is to establish meaning first and worry about
decoration last. As before, you will then reflect on the main factors that will ultimately shape
your choices.
393

9.1 Overview of Colour Models
Colour is a vast theoretical subject rooted in the science of optics – the branch of physics
concerned with the behaviour and properties of light – as well as colorimetry – the science and
technology used to quantify and describe human colour perception. The challenge of writing
about colour in the context of visualisation thinking is to establish a pragmatic cut-off point that
avoids sinking too deeply into these sciences, but still provides a rigorous basis for the
recommendations that follow.
The most relevant starting point for this overview is to recognise that, when dealing with issues
of colour in data visualisation, you will almost always be creating work using a computer. There
are exceptions of course, as we have seen in work created using Play-Doh and colouring pencils,
but mostly you will be using software viewed through an electronic display.
A discussion about colour theory needs to be framed around the RGB (Red, Blue, Green) colour
model. This is used to define the combination of light that forms the colours you see on a screen,
conceptually laid out in a cubic space based on variations across these three attributes. Even if
you create work with tools that use hexadecimal codes to specify your colour choices, these
specifications are still based on a mix of red, green and blue light. The ‘hex’ values take the form
of #RRGGBB using two-digit codes for each component ranging from 00 to FF.
The output format of your work will vary between screen display and print display. If you are
creating something for print you will shift your colour output settings to CMYK (Cyan,
Magenta, Yellow and Black). This is the model used to define the proportions of inks that make
up a printed colour. This is known as a subtractive model, which means that combining all four
inks produces black. RGB is an additive model as the three screen colours combine to produce
white.
CMYK communicates from your software to a printer, telling it what colours to print as an
output. RGB does the same but communicates the colour messages to a screen display. Neither
of these, though, really offer a logical model for us to think about the choices we might make on
which colours to use in a visualisation design. We require an alternative model of colour
thinking.
One of the most accessible colour models for considering the application of colour in data
visualisation is known as HSL (Hue, Saturation, Lightness), and was devised by Albert Munsell
at the start of the 20th century. These three dimensions (Figure 9.1) combine to make up what is
known as a cylindrical-coordinate colour representation of the RGB colour model.
Figure 9.1 HSL Colour Cylinder: Image from Wikimedia Commons published under the
Creative Commons Attribution-Share Alike 3.0 Unported license
394

Hue is considered the true colour. When you are describing or labelling colours you are most
commonly referring to their hue: think of colours of the rainbow ranging through mixtures of
red, orange, yellow, green, blue, indigo and violet (Figure 9.2). Hue is considered a qualitative
colour attribute because it is defined by difference and not by scale. With hue there are no shades
(adding black), tints (adding whites) or tones (adding grey).
Figure 9.2 The Colour ‘Hue’ Spectrum
Saturation defines the purity or colourfulness of a hue. This does convey a scale ranging from
intense, pure colour (high saturation) through increasing tones to what is technically the ‘no-
colour’ state of grey (low saturation) (Figure 9.3). In language terms think vivid through to
muted.
Figure 9.3 The Colour ‘Saturation’ Spectrum
Lightness defines the contrast of a single hue from dark to light (Figure 9.4). It is not a measure
of brightness – there are other models that define that – rather a scale of light tints (adding
white) through to dark shades (adding black). In language terms I actually tend to think of
lightness more in terms of degrees of darkness, but this is just a personal mindset. Also note that,
technically speaking, black, white and grey are not considered colours, but for the sake of this
chapter we will continue to think of them as being so.
Figure 9.4 The Colour ‘Lightness’ Spectrum
Alternative models exist offering variations on a similar theme, such as HSV (Hue, Saturation,
Value), HSI (Hue, Saturation, Intensity), HSB (Hue, Saturation, Brightness) and HCL (Hue,
Chroma, Luminance). These are representations of the RGB model space but involve different
mathematical translations into and from RGB. They also bring differences in the meaning of the
same terms (definitions of hue and saturation vary local to each model). The biggest difference,
395

though, concerns whether the models specify colour from the perspective of its quality (as in how
a colour is intended to appear) or as it is perceived (as in how a colour is ultimately experienced).
Pantone is another colour space that you might recognise. It offers a proprietary colour-
matching, identifying and communicating service for print, essentially giving ‘names’ to colours
based on the CMYK process.
There are arguments against defining colour thinking using the HSL model. While it is fine for
colour setting (i.e. an intuitive way to think about and specify the colours you want to set in your
visualisation work), the resulting colours will not be uniformly perceived the same, from one
device to the next. This is because there are many variables that affect the projection of light
when displaying colour, which means the same perceptual experience will not be guaranteed.
Some make a case for other models, such as CIELAB, as being more rigorous in the way they
offer an absolute, rather than relative, definition of colour for both input and output. Though I
understand the rationale, models like this can become too detached from the ideal pragmatism of
conceiving appropriate colour choices for visualisation design thinking. For the purpose of this
chapter, I will therefore draw suggestions from the HSL model.
396

9.2 Features of Colour
397

Data Legibility
Data legibility concerns the astute use of colour attributes to represent data values. The term
legibility places an emphasis on making sure the differences between and associations of any
colours used are readable and meaningful. There are different ways of optimally using colour to
represent values, depending on whether they are showing quantitative or categorical data.
Figure 9.5 What are the Current Electricity Prices in Switzerland? [Translated], by Interactive
things for NZZ
Colouring quantitative scales: When using colour to classify quantitative values the primary aim
is to create a sufficiently intuitive scale that facilitates an understanding of the hierarchy of values.
Variation in the lightness of a hue is typically the approach used for differentiating quantities.
The viewer should be able to discern, at least, whether a particular colour represents a larger or
smaller quantitative value than another quantity. Assessing the relative contrast between two
colours is generally how we construct a quantitative hierarchy. Absolute judgements can be
harder, even with a colour key provided for reference, and especially if you employ a continuous
gradient scale. The visual system is not always capable of reliably matching exact differences in
colour.
To maximise the efficiency of judging absolute values, quantitative colour scales are often divided
up into discrete classes, with each increasing in shade (towards dark) or tints (towards white).
This helps viewers detect local variations of colour. In the choropleth map in Figure 9.5, showing
the variation in electricity prices across Switzerland, the darker shades of blue indicate the higher
values, the lighter tints the lower prices.
Similarly, there are fascinating patterns that emerge in the next map (Figure 9.6), comparing
increases in the percentage of people gaining health insurance in the USA during 2013–14. The
398

data is broken down to county-level detail with a colour scale showing a darker red for the higher
percentage increases.
Figure 9.6 Obama’s Health Law: Who Was Helped Most, by Kevin Quealy and Margot Sanger-
Katz (New York Times)
Aside from the big-picture observations of the darker shades in the west and the noticeably
lighter tints to the east and parts of the mid-west, take a closer look at some of the interesting
differences at a more local level. Notice the stark contrast across state lines between the dark
regions of southern Kentucky (to the left of the annotated caption) and the light regions in the
neighbouring counties of northern Tennessee. Despite their spatial proximity, there are clearly
strong differences in enrolment on the programme among residents of these regions.
These two examples both employ a converging colour scale, moving through discrete variations in
the lightness of a single hue to represent small through to large quantities. Sometimes the shape
and range of your data may warrant a diverging colour scale. This is when you want to show how
quantities are changing in two directions either side of a specified breakpoint. This breakpoint is
commonly set to separate values visually above or below zero or those either side of a meaningful
threshold, such as a target, an average or a middle value.
The map in Figure 9.7 demonstrates this approach, through plotting data about the
demographics of the neighbourhoods around Berlin with a specific focus on the proportions of
inhabitants who are new or native Berliners. There is a diverging colour scheme used to indicate
whether there is a dominance of new Berliners (light orange to dark) or native Berliners (light
blue to dark). As this work is also interactive, the readability of the colour scales is supplemented
by annotated tooltips presenting the actual values in each location.
Figure 9.7 Native and New Berliners – How the S-Bahn Ring Divides the City, by Julius
Tröger, André Pätzold, David Wendler (Berliner Morgenpost) and Moritz Klack (webkid.io)
399

Although entirely continuous colour scales are not uncommon, usually there is value in dividing
up your converging or diverging scales into discrete classes. This needs careful thought to ensure
you get the right balance between aiding judgements of the relative order of magnitude as well as
the absolute magnitudes. There are two key factors to consider when judging your scales:
Are you plotting observed data or observable data? You might have data based on responses to
a survey that measures levels of satisfaction. The values in your dataset range from 42% to
82%. This is the observed data. However, it was possible for these responses to have ranged
from 0% to 100%, so will your colour classifications be based on the observed range or on
the potentially observable data range?
What is the distribution of your data? Does it make sense to create equal intervals for your
colour classifications or are there more meaningful divisions that better reflect the shape of
your data? Sometimes, you will have legitimate outliers that, if included, will stretch your
colour scales far beyond the meaningful concentration of where most values reside.
For diverging scales, the respective colour classification increments either side of a
breakpoint need to imply the same quantitative increment in both directions. For example,
if you use a shade of colour to represent +10% on one side of the breakpoint, the respective
colour shade for −10% on the other side of the breakpoint should have the same shade
intensity but for a different hue.
Additionally, the darkest shades of hues at the extreme ends of a diverging scale must still be
discernible. Sometimes darkest shades will be too close to black and viewers will no longer
be able to distinguish differences in the underlying hue.
One of the common mistakes in using colour to represent quantitative data is in the use of the
much-derided rainbow scale. The map in Figure 9.8 shows some particular alarming high
temperatures across Australia. Consider the colour key to the right of the map. Is this a
sufficiently intuitive scale to identify quantitative classifications? If there were no key provided,
would you still be able to perceive the order of magnitude relationship between the colours on
the map? If you saw a purple colour next to a blue colour, which would you expect to mean
hotter and which colder?
Figure 9.8 Highest Max Temperatures in Australia, produced by the Australian Government
Bureau of Meteorology
400

While the general implication of blue = ‘colder’ to red = ‘hotter’ is represented in parts of this
temperature scale, the presence of other hues obstructs the accessibility and creates inconsistency
in logic. For instance, do the colours used to show 24°C (light blue) jumping to 26°C (dark
green) make sense? How about 18°C (grey) to 20°C (dark blue), or the choice of the mid-brown
used for 46°C which interrupts the sequence of red shades? If you saw on the map a region with
the pink tone, as used for 16°C, would you be confident that you could easily distinguish this
from the lighter pink used to represent 38°C? Unless there are meaningful thresholds within your
quantitative data – justifiable breakpoints – you should only vary your colour scales through the
lightness dimension, not the hue dimension.
Colouring categorical classifications: When using colour to classify nominal categories the
primary aim is to create a clear, visible distinction between each unique categorical association.
The viewer should be able to discern different values as efficiently and accurately as possible. You
are not seeking to imply any notion of order or magnitude. You just want to help differentiate
each category from the others in a way that preserves a sense of equity among the colours
deployed.
Figure 9.9 Executive Pay by the Numbers, by Karl Russell (New York Times)
401

Variation in hue is typically the colour dimension to consider using for differentiating categories.
From a stylistic perspective, you might choose to vary the saturation across all hues, but you
should not consider using variations in the lightness dimension. As you can see demonstrated in
Figure 9.9, the lightness variation of a blue tone makes it quite hard to connect the colour
associations presented in the key at the top with the colours displayed in the stacked bars
underneath. With the shading in the column header and the 2011 grey bar also contributing
similar tones, the viewer’s visual processing system has to work much harder to determine the
associations than it should need to do.
Often the quantity of distinct categories you will need to differentiate between using colour will
be relatively few in number. In Figure 9.10, two colours are used to separate and associate the
panels of analysis in the charts showing margins of victories across all Olympic 100m events for
women and men respectively.
Figure 9.10 The Pursuit of Faster, by Andy Kirk and Andrew Witherley
402

In Figure 9.11, colours are used to represent key moments from the transcript of the Senate
testimony of then Supreme Court nominee Brett Kavanaugh and the woman accusing him of
sexual assault, Christine Blasey Ford. During the course of the hearing, you can see moments
when each was asked a question by the senators and prosecutor, with the colour indicating
whether those questions were actually answered or otherwise.
Figure 9.11 Every Time Ford and Kavanaugh Dodged a Question, by Alvin Chang
Figure 9.12 How Long Will We Live – And How Well?, by Bonnie Berkowitz, Emily Chow and
Todd Lindeman (Washington Post)
403

As the range of different categories grows, you need to expand the range of noticeably different
colours. In the scatter plot shown in Figure 9.12, six different hues of colour are used to classify
visually all point marks based on which countries are from the different continents of the world.
The ability to preserve clear differentiation becomes harder as the unique colours available
diminish. A useful guide to follow is once you exceed 12 categories, there are no longer
sufficiently different hues available to assign to categories 13+. There are variations of hues, of
course, but they are not different enough to preserve sufficient legibility. Just because variations
are available does not make them useful. You will be increasing the viewer’s cognitive burden
significantly, trying to learn, recall and recognise each association. This delays the process of
understanding and undermines the accessibility. There are three ways of handling excessive
numbers of distinct categories:
If interactivity is an option, consider offering filters to modify which category or categories
are displayed at any given point in a visualisation, as demonstrated in Chapter 7 in the
visualisation of tree species across New York City (Figure 7.3). Alternatively, use a
highlighting feature, like the ‘Baby names’ project example (Figure 7.6) to emphasise some
selected values, leaving the remainder presented in grey.
You might need to loop back to do some further data transformation, by considering how
to exclude or combine categories in order to reduce the number of distinct classifying
colours needed.
You may also consider supplementing the use of colour with texture or pattern to create
further visible distinctions. In Figure 9.13 you can see two patterns being used occasionally
as additive properties to show the structure of tracks on the Beatles’ album.
Figure 9.13 Charting the Beatles: Song Structure, by Michael Deal
404

Sometimes, your categorical data is actually about categories of colour. There can be an
immediate explicit relationship between the colours you use and their associated data values. In
Figure 9.14, Vienna is reduced to an illustrative 100m2 apartment whereby the floor plan
presents the proportional composition of the different types of space and land in the city. The
colours and textures of each component explicitly embody the visual characteristics of the
associated categorical value. This is another example of acceptable gratuitousness: the colour and
appearance demonstrate an additive quality, not a distracting one, creating topic immediacy that
accelerates the value recognition.
Another treemap, as shown in Figure 9.15, shows the wide range of different colours used in
official rapid transit diagrams of every system in the world. The colours associated with every line
on every worldwide metro or subway system are grouped by colour family with the box sizes
based on the number of stations served by that line. So, for example, the yellow Circle line is one
of 14 different lines on the London Underground system and serves 36 stations along its route.
Ordinal categories are handled a bit differently to nominal categories because they introduce
properties of order. When using colour to classify different ordinal categories you are striving not
only to create visible distinction between each distinct category, but also to portray the
hierarchical relationship that exists between them. The colour approaches used to achieve this
tend to align more with how you would represent quantitative values.
Figure 9.14 If Vienna Would Be an Apartment, by NZZ
405

Figure 9.15 Colors of the Rails, by Nicholas Rougeux (www.c82net)
A typical example of a diverging ordinal scale might be seen in a stacked bar chart. The example
shown on the left in Figure 9.16 applies ordinal colour classifications to reveal the responses to a
range of survey questions. The categories are representative of sentiment and strength of feeling,
based on a scale from strongly agree, agree, no opinion, disagree to strongly disagree. By colouring
the agreement categories in green, the disagreement categories in pink and ‘no opinion’ in grey, a
viewer can quickly perceive the general balance of feelings being expressed. Darker shades
emphasise the strongest feelings at each end of the stacked bar rows.
Figure 9.16 Contrasting Approaches to Colouring Stacked Bar Charts Displaying Ordinal Data
406

http://www.c82net

Although the nominal colouring applied to the same chart on the right-hand side still enables
you to learn, look up and see the distinct categories, it fails to make the collective ordinal patterns
as discernible as the approach on the left.
Another example of ordinal data might be to represent the notion of recency. In Figure 9.17 you
can see a display plotting the 2013 Yosemite National Park fire. Colour is used to display the
recorded day-by-day progress of the fire’s spread. The colour scale is based on a temporal
spectrum with darker shades being more recent, lighter tints being more in the past. It applies the
metaphor of the past having somewhat faded away.
There is an extension in the potential application of ordinal colouring which becomes relevant
when you might wish to apply the notion of hierarchical emphasis to draw out significant
categorical features of your data that would otherwise merit nominal colouring practices.
This is about drawing contrast between important features that should appear prominently in the
foreground for the viewer against other features of less importance that should be more subdued
in their appearance. As introduced in Chapter 3, bringing key insights to the surface of your
charts contributes towards facilitating an explanatory experience. It bears repeating: if you have
something important to say, say it. In this case, say it with colour.
It is here that grey will prove to be a strong ally helping you to convey a sense of depth in your
work. You will recall from the opening section that grey is the unsaturated form of a hue. When
an unsaturated colour is juxtaposed with a saturated colour, contrast is achieved. In the bar chart
shown in Figure 9.18, the analysis presents a summary of the most prevalent men’s names that
feature among the CEOs of the S&P 1500 companies. As you can see, there are more guys
named ‘John’ or ‘David’ than the percentage of all the women CEOs combined. With the
emphasis of the analysis on this startling statement of inequality, the bar for ‘All women’ is
emphasised in a strong burgundy-coloured hue, contrasting with the grey bars of all the men’s
names. Notice also that the respective axis and bar value labels are both presented using a bold
font for the ‘All women’ bar, which adds further emphasis.
Figure 9.17 ‘Rim Fire’ – The Extent of Fire in the Sierra Nevada Range and Yosemite National
Park, 2013: NASA Earth Observatory images by Robert Simmon
407

Figure 9.18 Fewer Women Run Big Companies Than Men Named John, by Justin Wolfers
(New York Times)
Sometimes, only modest emphasis is required. There is no need to shout in order to establish
contrast. The chart in Figure 9.19 creates more subtle distinction between the slightly darker
shade of green (and emboldened text), emphasising New York’s figures, compared with the other
listed departments that appear in lighter green. The object of our attention aligns with the topic
of interest. In this case, it concerns a drive to recruit more NYPD officers. This does not need to
be any more contrasting than it appears; it is sufficiently noticeable and sometimes that is the
right level of volume to apply.
Figure 9.19 NYPD, Council Spar over More Officers, by Graphics Department (Wall Street
Journal)
408

The chart shown in Figure 9.20 shows an example of when you want to use colour to establish
discrete associations between categorical values, but then also apply an ordinal separation to
create contrast between the values you want viewers to read and those included only to provide
scale and context. The analysis in this chart looks at international cricketers (specifically batsman)
and their cumulative run scoring across their Test match careers based on their age at the time
they scored their runs. Seven current or recent celebrated batsmen are elevated to the foreground
and categorised using distinct colours for each series line. The rest of the players included in the
chart are given a grey shade and thus relegated to the background. We want to see the shapes of
their careers, but in this analysis we do not care about finding out who they are.
Figure 9.20 Cricketer Alastair Cook Plays His 161st and Final Test Match, by John Burn-
Murdoch for Financial Times
409

Functional Decoration
After making colour choices for optimising data legibility in your charts, you must turn to
consider functional decoration. This is concerned with colouring every other element of your
visualisation display: your interactive features, your chart apparatus and any annotations need to
be coloured in order to be visible, but in a way that is harmonious with the colour schemes you
have used to represent your data.
Functional decoration may sound like an oxymoron, but it captures the delicate balancing act
you face. You have room to experiment in how you might colour your annotations and
interactive elements, but only to the extent that those choices do not compromise their
functional purpose, which is to support the legibility of data.
There is no single pathway towards achieving this. The Wind Map project (Figure 9.21) conveys
a highly aesthetic quality yet uses only a monochromatic palette. There is no colouring of the sea,
no topographic detail, no emphasising of any extreme wind speed thresholds being reached. It
exhibits artistic and functional beauty.
Figure 9.21 Wind Map, by Fernanda Viégas and Martin Wattenberg
I am not advocating a need to pursue minimalism. Creating something that is pleasing to the eye
and equally fit for purpose functionally is a hard balance to achieve. Though you can create
elegant work through a limited palette of colours, justifying the use of colours is not the same as
unnecessarily restricting the use of colour. Sometimes you will just find a role for many more
colours, to help capture the right look and feel for your subject matter. It is why this phase of
design thinking is characteristically iterative and often relies on a degree of trial and error in your
approach.
Some of the most influential colour practices in data visualisation come from the field of
cartography (as do many of the most passionate colour purists). Just think about the amount of
visual detail shown in a reference map that relies on colour to help differentiate types of land,
indicate the depth of water or the altitude of high ground, mark out routes of road and rail
networks, etc. The best maps pack an incredible amount of detail into a single display and yet,
somehow, they are still legible and functional.
410

As with reference maps, every design feature you incorporate into your visualisation display will
have a property of colour, otherwise they would be invisible. And all these colour choices are
connected. Even though you will often make decisions about colouring features in isolation,
there will always be a consequence of that choice elsewhere.
The colour choices for chart annotations, including apparatus like gridlines, axis lines and value
labels, are particularly sensitive given their proximity to colours assigned already to the data
values. If you choose to classify a category in your data using a shade of grey, using the same grey
for your gridlines may create confusion as the eye may lose track of which line relates to which
feature.
Additionally, once you commit a colour to mean something you should not use the same colour
to mean something different, at least not in the same view or page. Exclusivity in a colour’s
association is important to preserve for as long as possible so the viewer does not have to relearn
its meaning. The graphic on ‘Ring-necked Parakeets’, featured in Figure 9.22, establishes a
quantitative association between a scale of green and pink tints. Once you have learnt this
association, you can rely on the same colour associations being continued right across the whole
graphic. The viewer can relax into scanning without wondering if the meaning has evolved from
one section to another. This significantly amplifies the accessibility of the work and also enhances
the elegance through the limited but meaningful colour palette.
Figure 9.22 Ring-necked Parakeets, by Sophie Sparkes and Jonni Walker
Figure 9.23 Art in the Age of Mechanical Reproduction: Walter Benjamin, by Stefanie Posavec
411

Figure 9.24 Lunge Feeding, by Jonathan Corum (New York Times); Whale Illustration by
Nicholas D. Pyenson
412

If you must use the same colours for different associations, at least try to maximise the ‘gap’
between each instance of a different association, such as physical gaps (different pages, interactive
views), time spans (the duration between reading displays with different associations) or editorial
breaks (new subject matter, new angle of analysis). This space will help effectively to cleanse the
palate (yes, pun intended) in the mind of the viewer. At each new assignment of a colour, clear
explanations are of course mandatory.
‘When something is not harmonious, it’s either boring or chaotic. At one extreme is a visual experience that is so
bland that the viewer is not engaged. The human brain will reject under-stimulating information. At the other
extreme is a visual experience that is so overdone, so chaotic, that the viewer can’t stand to look at it. The human
brain rejects what it cannot organise, what it cannot understand.’ Jill Morton, Colour Expert and Researcher
The quality of harmony across all your colour choices is a hard thing to achieve. It shares the
same enigmatic quality as ‘elegance’, in that you notice it more when it is missing. It is apparent
in the colours used by Stefanie Posavec in her visualisation of the structure of Walter Benjamin’s
essay ‘Art in the Age of Mechanical Reproduction’ (Figure 9.23). There is an almost effortless
cohesion between the colours used across the entire design anatomy of this work: the petals,
413

branches, labels, titles, legend and background.
The ‘Lunge Feeding’ graphic, Figure 9.24, similarly demonstrates the importance of functional
decoration. The blue-shaded panel, getting darker as the sea depth increases down the page,
provides a notion of scale for the journey taken by a whale when feeding. This draws contrast
from the rest of the layout, establishing the panel as the centrepiece to which all other elements
are anchored. The thin grey-shaded columns emerging from the bottom of this panel indicate the
occasions of a lunge action, which ties in with the same grey bandings used in the small charts
that assist the sequence of illustrations of a whale’s feeding action on the left. The style of these
illustrations is coherent with the overall tone of the work. Rather than being jarringly different,
they feel seamlessly integrated and decorate the work in a functional way, helping the viewer to
see the act that is being described.
There are no universal rules for the benefits of any particular colour for shading the background
of projects or charts. Your choice will depend on the circumstances and conditions in which your
viewers are consuming the work, the inherent association with your subject matter, and the style
you are trying to convey. It is not uncommon to see background colours being drawn from the
set of neutral options presented in Figure 9.25.
Figure 9.25 Examples of Common Background Colour Tones
Above all else, the colours you have selected to establish data legibility will be key. In general, a
white background gives viewers the best chance of being able to perceive accurately the different
colour attributes used in your data encoding and especially scales that use degrees of lightness. In
Figure 9.24, the influence relationship was inverted: in using blue to colour the background of
the sea, the hue of orange offered the most contrasting option to ensure the path of the whale’s
dive was most visible. This work also demonstrates the value of emptiness or white space to
establish layout. Think of it as visual punctuation, offering moments in your work where the
viewer can pause, reflect and then move on to the next discrete element.
With thematic maps, there is often merit in including some kind of reference map in the
background to assist with orientation. The dot map in Figure 9.26 looks at the language of
tweets posted over a period of time across the New York City area. Given the density and
number of discrete data points, permanently and simultaneously including the detailed features
shown in the mapping layer becomes too visually cluttered. The developers employ a smart
414

interactive solution to overcome this, offering an adjustable slider that allows users to modify the
transparency of the network of roads to reveal the apparatus of the mapping layer.
Figure 9.26 Twitter NYC: A Multilingual Social City, by James Cheshire, Ed Manley, John
Barratt and Oliver O’Brien
For interactive projects, every control needs colour in order to be visible, but it must also
demonstrate ‘affordance’: properties that indicate what functional events are possible and, where
relevant, what events have been activated. The example shown in Figure 9.27 examines the
connected stories of casualties and fatalities from the Iraqi and Afghan conflicts. You will see
several interactive controls, all of which are astutely coloured in a way that feels consistent with
the overall tone of the project, but also makes it functionally evident what each control’s selected
setting is. This is achieved through subtle but effective combinations of dark and light greys that
help to indicate what has been selected or highlighted, such as the ‘Afghanistan’ and ‘Iraq’ tabs at
the top. Filter controls at the bottom use brighter greys to show the selected range of values, but
also preserve visibly what other currently unselected values are available to include.
Figure 9.27 Casualties, by Stamen, published by CNN
415

416

9.3 Influencing Factors and Considerations
We have already touched on the strong influence of using colour when displaying different types
of data, but there are many other factors that will influence your decisions about how colour
should be used in your work.
Medium: If your printed work will need to be compatible for both colour and black and white
output, before finalising your decisions check that the legibility and intended meaning of your
colour choices are being maintained across both. It might seem obvious but there is a significant
difference in how colours appear when published in colour and how they appear when published
in greyscale because different hues have different levels of brightness. The purest blue is darker
than the purest yellow, and so if printed using black and white settings, blue would appear a
darker shade of grey and imply it is representative of a higher order of value.
For digital displays, the conditions in which the work will be consumed will have some influence
over the choice between, for example, light and dark backgrounds. It can be hard to mitigate for
all the subtleties of variation in light present at the time of consumption, but if your work is
generally intended for consumption in a light environment, lighter backgrounds with saturated
foreground colours tend to be more fitting; likewise, darker backgrounds will work best for
consuming in darker settings.
Colour rules: In many organisations, publications and websites, there are style guidelines that
require the strict observation of colour rules. These are often established with good intent, driven
by a desire to create conformity and consistency in style and appearance. Developing a
recognisable ‘brand’ and not having to think from scratch about what colours to use every time
you face a new project are things that can be very helpful, especially across a team environment.
However, in my experience, the contents of such colour guides rarely offer optimal application
for allocating colours to fulfil a variety of different roles in a data visualisation work.
Compromising on the rules would be the ideal scenario, but the main point is to discover the
constraints that exist early in your process so you do not arrive at this stage ignorant of the
restrictions you will be facing.
Purpose: Colour choices will strongly influence the visible tone of your work. Does it need to be
modest or stimulating? Can you select from a vivid and varied palette or should you be striving
for more muted and distinguished options? If you are pursuing an explanatory experience, you
may have determined that you will be seeking to say something with colour, using it to draw out
significant features of your data.
Accessible design: Approximately 5% of the population have visual impairments that
compromise their ability to discern particular colours and colour combinations. Deuteranopia is
the most common form, often known as red–green colour blindness, and is a particular genetic
issue associated with men. The traffic light scheme of green = ‘good’, red = ‘bad’ is a common
approach for using colour as an indicator. It is a convenient metaphor, especially in the corporate
world, and the reasons for its use are entirely understandable. However, as demonstrated in the
treemap of Figure 9.28, which has been rendered to simulate deuteranopia, the meaning of reds
and greens will not be at all distinguishable and it will be inaccessible to those affected.
417

Figure 9.28 Finviz: Standard & Poor’s 500 Index Stocks (www.finviz.com)
If nobody in your audience has such visual impairments, it is not necessary to avoid the use of
red or green, but if your audience are large and you may need to consider colour-
blind-friendly alternatives. Some options are presented in Figure 9.29. The first three options
show variations of green tones alongside different secondary pairings that might be considered
instead of the standard default red. The fourth option switches the metaphor to red = hot =
good, blue = cold = bad. The final option uses secondary encoding through symbols to convey
the association if the colours cannot be perceived.
Figure 9.29 Colour-blind-friendly Alternatives to the Standard Green and Red Tones
Deuteranopia is not the only visual impairment to be concerned about. For those with limited
eyesight, features that offer magnified views and voice assistance may be worth considering,
should they be viable.
An extension of accessible design thinking is to consider the impact of potentially exploiting
established colour associations with your subject matter. In politics, sport, brands and in nature,
there are many subjects that already have immediate associations that offer the viewer a shortcut
to accelerate recognition. These might be applied to encode your data or functionally decorate
your work.
However, while some colours can offer useful and positive associations, in some cases there can
be negative connotations that should be handled sensitively or even avoided. You would not use
bright, happy colours if you were portraying matters of death or disease. To use a blue colour in a
project about depression would be insensitive. Using any notion of skin colour to represent
ethnic groups is something that would be understandably considered offensive unless there were
very good reasons for skin colour being intrinsic to the data.
Occasionally, established colour associations are out of sync with contemporary culture or
society. For example, when you think about colour and the matter of gender, because it has been
418

http://www.finviz.com

so endlessly adopted down the years, it is almost impossible not to think instinctively blue for
boys, pink for girls. My personal view is that this association should be avoided. I agree with
many commentators who say the association of pink to signify the female gender, in particular, is
clichéd, outdated and no longer fit for purpose. This is not a universal view, and I have
encountered many who disagree with it. However, I do not think it is too much to expect viewers
to learn the association of two different colours for representing gender.
Cultural sensitivities and inconsistencies are also important to consider. In China, for example,
red is a lucky colour and so the use of red in their stock market displays, for example, indicates
rising values. In Western society red is often the signal for a warning or danger.
419

Summary: Colour
420

Features of Colour
This chapter introduced you to colour theory and presented different ways of applying colour in
a visualisation to facilitate data legibility and deliver functional decoration.
Data legibility: Using colours to represent different types of data, with distinctions in
approach for representing classifications for quantitative data and associations with
categorical (nominal) data. A further distinction was made for using colour to emphasise the
relationships between ordinal categories.
Functional decoration: Concerning decisions about applying colour to every other visual
element in your work, including interactive features and annotations.
421

Influencing Factors and Considerations
If these were the options, how did you make your choices? The influencing factors included:
Medium: The intended output format of your work will affect both colour choices and how
they are perceived.
Colour rules: The need to observe potentially restrictive colour guidelines.
Purpose: What tone of voice are you trying to convey and how might colour choices shape
this?
Accessible design: Pay attention to potential visual impairments across your audience. Be
aware of the sensitivities and positive or negative colour connotations.
422

General Tips and Tactics
Use the squint test. Shrink things down and/or half close your eyes to see what coloured
properties are most prominent and visible. Are these the right features of your display that
should be emphasised?
Experimentation: Trial and error is often necessary in colour selection, especially for
functional decoration.
Print quality and consistency is a factor. Graphics editors who create work for print
newspapers or magazines will often consider using colours as close in tone as possible to
pure CMYK, especially if their work is quite intricate in detail. This is because the colour
plates used in printing presses will not always be 100% aligned and thus mixtures of colours
may be slightly compromised.
Developing a personal style guide for colour usage saves you having to think from scratch
every time and will help your work become more immediately identifiable (which may or
may not be an important factor).
Make life easier by ensuring your preferred (or imposed) colour palettes are loaded up into
any tool you are using, even if it is just the tool you are using for analysis rather than for the
final presentation of your work.
If you are creating for print, make sure you do test print runs of the draft work to see how
your colours are looking – do not wait for the first print when you (think you) have finished
your process. What looks like a perfect colour palette on screen may not ultimately look the
same when printed.
What now? Visit book.visualisingdata.com
EXPLORE THE FIELD Expand your knowledge and reinforce your learning about working with data through this
chapter’s library of further reading, references, and tutorials.
TRY THIS YOURSELF Revise, reflect, and refine your skill and understanding about the challenges of working
with data through these practical exercises.
SEE DATA VISUALISATION IN ACTION Get to grips with the nuances and intricacies of working with data in
the real world by working through this next instalment in the narrative case study and see an additional extended
example of data visualisation in practice. Follow along with Andy’s video diary of the process and get direct insight
into his thought processes, challenges, mistakes, and decisions along the way.
423

http://book.visualisingdata.com

10 Composition
Composition is the final part of your design anatomy. It concerns the management of space. By
definition, composition can be seen as both the act of and result of arranging a mixture of visual
ingredients together to form a final whole.
You should not infer that discussing this topic in the final chapter means it is the least important
part of developing your design solution. Far from it. It is just that only after having considered
annotation can you reasonably move past thinking about what will be included and then
defining how it will appear.
Charts, interactive controls and features of annotations all occupy space. The decisions you make
in the final step cover the physical attributes of, and relationships between, every design element
that is to be included in your final work. By extension, interactivity can also affect how you use
your space, how you overcome the limitations of your space, and how you navigate to other
space.
424

10.1 Features of Composition
425

Layout
Judging layout is the essence of composition thinking. Well-considered layouts optimise the
readability and meaning of the collected content. They are a function of the relative positioning
and sizing of all your design elements in the space you are working with. Just as variation in
colour implies meaning, so too does variation in position and size. A chart that is larger than
another chart will imply it carries greater importance. Charts of equal size but located in different
places will lead to extra attention being commanded by the one positioned at the top of a screen
or presented first in a sequence.
The purpose of your layout is to establish a hierarchy of meaning and importance, offering the
viewers clues about the journey they should take through the content.
Figure 10.1 City of Anarchy, by Simon Scarr (South China Morning Post)
426

In Figure 10.1 we see the infographic ‘City of anarchy’. This demonstrates a clear visual
hierarchy across its layout. There is a prominent heading that introduces the work and manages,
temporarily, to pull your eye away from the temptation of focusing too prematurely on the
beautifully illustrated ‘cutaway’ diagram, which is the centrepiece of the work. Annotated
captions orbit around the graphic making interesting observations available when your eye
reaches that part of the display. The thumbnail references at the top of the page offer spatial
context, providing a localised map view and a world map to explain where this building is
located. Those references are there to assist when you want them. At the bottom of the page there
are small illustrations to provide supplementary analysis about the history of the city’s growth
over time. It is clear through their placement at the bottom of the page and their diminutive
stature that this analysis will only be encountered much later in the reading process.
427

When starting to think about your potential layout you cannot usefully isolate your thoughts
about position from matters of size. When you arrange furniture in a room the decisions you
make about where to put things are informed by how big those things are. But if the absolute size
of the furniture is not yet defined then the permutations of different arrangements increase in
number substantially.
To break this impasse, there are two approaches to help start shaping your composition ideas:
wireframing and storyboarding. Wireframing involves creating low-fidelity sketches of the
potential layout of all your design elements within a single page, like an infographic or an
interactive where all functions apply within the same page or view. Figure 10.2 shows the
iteration of wireframe designs leading towards the final composition for the ‘Filmographics’
project. This shows the evolution of ideas for the placement of the charts, the annotations and
the interactive controls.
Figure 10.2 Filmographics, by Andy Kirk and Matt Knott
Storyboarding is used to establish the overall structure of your work when it will entail multiple
distinct views (e.g. a report or presentation, wide-ranging interactive). It organises your thinking
about the sequencing of and navigation between each distinct view of content. The composition
within each of these views then goes through more detailed wireframing.
Regardless of whether this activity is carried out using pen and paper, basic tools or more
sophisticated technologies, always start with rough ideas and from there the precision will emerge
through iteration and experimentation.
It stands to reason that charts will and should be the centrepiece of any visualisation work.
Anchoring your layout around where your viewers should encounter your charts can be a useful
starting point. Thereafter, the placement of interactive controls and features of annotation should
be supporting the experience of understanding, not dominating it.
428

Interactive controls will ideally be located as close to where the functions will be performed so
the eye and hand have far less distance to travel between the two. For annotations, the order in
which I profiled the different potential features in Chapter 8 was based on the arrangement of
where those elements are typically located within a visualisation work. Starting from headings
and introductions, moving through reader and/or user guides, on to chart-related apparatus and
labelling, and then finishing with footnotes and methods. You might need the setup of
introductions and guides to be seen before any charts are seen in order to enhance the reading
process. Although there are certain conventions you might follow for the efficiency of your
thought process, you have the freedom to determine whatever is the best layout suitable for your
project’s purpose.
429

Arranging
Arranging concerns the ordering and direction of your data content as it is displayed within a
chart. This is an important consideration to help viewers perceive your representations in the
most relevant formation. There are several different approaches to sorting data.
Alphabetical sorting is a cataloguing approach that facilitates efficient lookup and reference of
textual or categorical values. You would use this arrangement when you need to offer your
viewers an efficient way to look up specific categorical values when there are many items
included. In Figure 10.3, investigating different measures of waiting times in emergency rooms
across the USA, the bar charts are presented using alphabetical sorting of each state name. This is
the default setting, but users can also choose to reorder the bars in other ways across the other
columns.
Alphabetical sorting is a common approach but one that often reflects a default choice rather
than a useful one. It can be the least interesting way to arrange your data values. However, it
might be seen as a suitably diplomatic option should you need to avoid politically displaying
your data using any form of ranking, in particular. Additionally, it is absolutely sensible to
employ alphabetical ordering for values listed in interactive controls like dropdown menus.
Figure 10.3 ER Wait Watcher: Which Emergency Room Will See You the Fastest?, by Lena
Groeger, Mike Tigas and Sisi Wei (ProPublica)
Chronological sorting is used when the data has a temporal dimension and you need your display
to expose patterns over time. In Figure 10.4, you can see a snapshot of a graphic that portrays the
rain patterns in Hong Kong since 1990. Each row of data represents a full year of daily readings
running from left to right, with the years arranged vertically from the past to the present.
Figure 10.4 Extract from Rain Patterns, by Jane Pong (South China Morning Post)
430

Locational sorting involves sequencing content according to a spatial dimension, particularly
when you are not using a mapping technique to portray your data and it is more about relative,
rather than fixed, location relationships. This could involve sorting values based on geographic
relationships (such as presenting data for all the stations along a train route) or a non-geographic
relationship (like a sequence of values based on the positions of parts of the body, from head to
toe). Ordering data by location will only be relevant if you believe there is interest in or
significance between comparing adjacent locations. An example of this approach is exhibited by
the project ‘On Broadway’ (Figure 10.5), which is an interactive installation that stitched
together a bunch of different data measurement and media items relating to intervals of life along
Broadway. This collective work offers compelling views of the fluctuating characteristics of the
different communities and neighbourhoods as you journey down the spine of New York City,
stretching 13 miles (21km) across and beyond the length of Manhattan.
Figure 10.5 On Broadway, by Daniel Goddemeyer, Moritz Stefaner, Dominikus Baur and Lev
Manovich
Ordinal sorting can be usefully applied to arrange categorical data that has characteristics of
ordered and, potentially, hierarchical relationships. In Figure 10.6, you can see a dendrogram
that looks at the consequences of two major beer brands merging, namely SAB Miller and
ABInBev. The diagram shows all the individual beer brands that were previously owned by each
discrete brand. The hierarchical layout organises this display around a radial structure, starting
from the inside tier of a continent, moving out to countries, and then finally to the outer nodes
431

detailing each product.
Finally, ranked sorting may be the most common and useful way to arrange data values based on
ascending or descending quantitative rankings. In Figure 10.7, the bar chart shows the artists and
songs that have held the number one position in the UK charts for the greatest number of weeks.
The values are arranged in descending order from the longest duration to the shortest.
Figure 10.6 The 200+ Beer Brands of SAB and AB InBev, by Maarten Lambrechts for Mediafin
Figure 10.7 The Songs That Were #1 in the UK Charts for the Greatest Number of Weeks
There are further arrangement decisions to make with certain chart types that have overlapping
432

lines, like line charts or bump charts, or criss-crossing connection bandings, like Sankey diagrams
and chord diagrams. These charts introduce the need to contemplate value sorting in a z-
dimension: that is, which of these features should be displayed on top and which underneath,
and why?
The orientation of your chart is another arrangement consideration which might help squeeze
out an extra degree of readability and meaning from your display.
The primary thought about chart orientation regards the readability of axis value labels. A
vertically arranged bar chart, with multiple categories along the x-axis, will potentially cause
viewers to tilt their neck in order to read the labels. You could try adjusting the orientation of the
labels to 45° or 90°, but my preference is to transpose the chart so the labels are on the y-axis, the
bar sizes are directed along the x-axis, and the category labels are presented in a more readable
fashion.
Figure 10.8 Kasich Could Be the GOP’s Moderate Backstop, by FiveThirtyEight
The example in Figure 10.8 rotates a scatter plot by 45° to help guide the viewer’s interpretation
of what it means for a point mark to be found in each quadrant region. It is also used to
emphasise the distinction between being in the top half and bottom half of the chart, which
defines the degree of popularity, as this is the principal angle of interest.
433

Sizing
Do not be afraid to shrink your charts. The eye can detect, with great efficiency even at small
resolution, variation in size, position, colour and pattern. The technique of ‘small multiples’ is
commonly used to replicate distinct chart displays for multiple categories or points in time and
arrange them usually in a grid layout. This enables the eye to compare and contrast features
across all charts in a simultaneous view. Otherwise, you might have to browse through multiple
pages or navigate through different selections using interactivity and remember each chart view in
order to compare against it. The main obstacle to shrinking chart displays is the impact on text
size. The eye will not cope too well with small fonts for axis and value labels, so there has to be a
trade-off.
‘Using our eyes to switch between different views that are visible simultaneously has much lower cognitive load than
consulting our memory to compare a current view with what was seen before.’ Professor Tamara Munzner,
Department of Computer Science, University of British Columbia
The project featured in Figure 10.9 demons-trates a beautiful example of small multiples. This
work is called ‘Coral Cities’ and looks at how easy it is for people to move within and out of
cities. The organic forms displayed represent the distance and routes that can be reached within
30 minutes by car when leaving each city centre. The 40 cities selected are based on the Mercer
‘Quality of Living City’ rankings. Although created for a large print output, even when looking
at a smaller scaled version, any viewer can investigate the shape of patterns for each city but also
seamlessly turn their attention to look at all the patterns collectively to explore and find
commonalities and exceptions.
Decisions about chart sizing extend to defining axis scales and value intervals. Although the clues
about the most meaningful range of values to include will be shaped by your work at the data
examination stage of the process, there are certain conventions that also need to be observed.
When a chart encodes quantitative values using size, the viewer needs to see the full,
representative size of the mark, otherwise it will be a distortion of the truth. For example, with
bar charts that show data from a common baseline, viewers need to see the full size of a bar based
on the value it represents, nothing more and nothing less. To do this you must set the origin of
the quantitative value axis to zero. If you start this baseline position from any other value, the
effect will be to truncate the axis range and the perceived size of the bars. This creates a
distortion: the viewer is only presented with part of a bar’s true size. The charts in Figure 10.10
show the tallest buildings and structures around the world that are at least 250m in height. The
scale used in the chart with the red bars uses a quantitative axis with an origin of 250. This
distorts the lengths compared with the true sizing as shown in the first bar chart with the blue
bars, which has an origin of 0.
Figure 10.9 Coral Cities, by Craig Taylor, Data Visualisation Design Manager at Ito World
434

Figure 10.10 Illustrating the Effect of Truncating Quantitative Axis Scales for Bar Charts
Not all charts that use bars necessarily need to start from a zero origin. Variations in the use of
the waterfall chart, for example, might be used to show quantitative changes or differences
between absolute values (the ‘delta’). In this case the base of a bar may be positioned to start from
any quantitative position and not necessarily from zero. So long as you still encode its full
435

representative size, that is fine.
In contrast to the bar chart, a line chart does not necessarily need to have a value axis origin
always set to zero. It encodes quantitative values through point marks positioned along a value
scale, not through size. Truncating the quantitative value axis may be relevant when the notion
of a quantity of zero might represent an impossible measurement. It should be made clear to the
viewer through clear axis labelling that the baseline position does not represent zero.
In the chart shown in Figure 10.11, we see a line chart plotting the history of 100m record times.
Although results have quickened, there is a physical limit to what humans can achieve: running
the 100m in a time that is anywhere near zero seconds is impossible. So, starting the value axis at
9 seconds through to a maximum of 11 seconds provides a reasonable axis range in which to plot
the observed measurements.
Figure 10.11 Doping under the Microscope, by S. Scarr and W. Foo (Reuters Graphics)
436

10.2 Influencing Factors and Considerations
Decision making about composition is greatly shaped by common sense but equally burdened by
the unsatisfactory shrug of ‘it depends’. Here are some of the main considerations.
Medium: Naturally, as composition is about spatial arrangement, the nature and dimensions of
the canvas you have to work with will have a fundamental bearing on the decisions you make.
There are two concerns here: what will be the shape and size of the primary format; and how
transferable will your solution be across the different platforms on which it might be used or
consumed?
Decisions about layout will vary depending on whether your work is to be published on a single-
page or screen view, such as an infographic or interactive visualisation, or across multiple linear
pages (like a report, presentation) or multiple different views (interactively driven navigation
controlled by the user). The best solutions composition-wise will vary considerably for each.
The varied characteristics of modern devices present visualisers (or perhaps more appropriately, at
this stage, developers) with real challenges. Getting a visualisation to work consistently, flexibly
and portably across device types, browsers and screen dimensions (smartphone, tablet, desktop) is
hard.
Although many organisations, especially the media, are focused on a mobile-first strategy, the
reduced canvas size and the intricacies of requiring users to interact precisely with diminutive
controls create difficulties. Given the choice, most visualisers would probably see the desktop as
their preferred canvas. Solutions designed for mobile and, to a certain extent, tablet will aim to
preserve as much continuity in the core experience as possible but may require certain
compromises.
For ProPublica’s work on ‘Losing Ground’ (Figure 10.12), the approach taken to determine
what degree of cross-platform compatibility should be preserved was informed by using the
heuristic ‘smallify or simplify’. Features that worked on ProPublica’s primary platform of the
desktop would either be simplified to function practically on mobile or just be reduced in size.
You will see in the pair of contrasting images how the map display is both shrunk and cropped,
and the introductory text is stripped back to include only the most essential information.
Figure 10.12 Losing Ground, by Bob Marshall, The Lens, Brian Jacobs and Al Shaw
(ProPublica)
437

Another consideration about medium will relate to whether your work will be published as an
interactive for the Web and also as a static piece for print. The features that make up an effective
interactive project may not necessarily translate directly into static form. You might need to
pursue two parallel solutions to suit the respective characteristics of each output format.
Quantitative value range: When discussing the physical properties of data in Chapter 4, I
described the influence of the shape of your data on your chart composition choices. If you have
30 distinct categorical values in your data, and they all need to be shown, you will need to
allocate space for 30 categorical items in your chart layout. The lengths of the words of each item
will also need to fit as labels in or adjacent to the chart.
We have already discussed the conventions of setting axis scales, but there are particular
composition challenges when you have a wide range of quantitative values. Legitimate outliers
will potentially distort your ideal scale choices but will need to be accommodated somehow in
the space you are working with.
One solution for dealing with this is to use a non-linear logarithmic (often just known as a ‘log’)
scale. Essentially, each major interval along a log scale increases the value at that marked position
by a factor of 10 (or by one order of magnitude) rather than by equal increments.
In Figure 10.13, looking at ratings for thousands of different board games, the x-axis is presented
on a log scale in order to accommodate the wide range of values for the ‘Number of ratings’
measure. This also helps to fit the chart into a neat square layout, which can sometimes be a
requirement to enable graphics to be optimally sized for publishing on social media platforms.
Had this x-axis remained as a linear scale, in order to preserve this square layout the values below
1000 would have had to be squashed into such a tightly packed space that you would hardly see
the patterns. A wide, rectangular chart would have been necessary but impractical, given the
limitations of the space this chart would occupy.
Figure 10.13 The Worst Board Games Ever Invented, by FiveThirtyEight
438

Editorial thinking: Your editorial thoughts will probably have extended to establishing a sense of
hierarchy that might inform which analysis should be displayed more prominently and which less
so.
If you have decided to include a number of different angles of analysis in your work, this will
amplify the challenge of composition. The more analysis you include naturally increases the
demands on space. You might need to compromise by reducing the size of your chart elements or
by offering a non-simultaneous arrangement, through multi-page layouts or sequences reached
through interactive navigation.
Data representation: Different charts bring different spatial consequences. A treemap generally
occupies far more space than a pie chart simply because there are usually many more ‘parts’ being
shown. A polar chart is circular in shape, whereas a waffle chart is square. Each chart you include
introduces a uniquely shaped element that will need to be arranged into your layout.
The table in Figure 10.14 summarises the main chart structures and the typical shapes they
occupy, based on the chart types profiled in Chapter 6.
Figure 10.14 List of Different Chart Shapes and Structures
Elegant design: Like colour, composition decisions are always relative: an object’s place and the
439

space it occupies within a display create a relationship with everything else in the display. Unity
in composition provides a similar sense of harmony and balance between all objects. The flow of
content should feel logical and meaningful.
‘I’m obsessed with alignments. Sloppy label placement on final files causes my confidence in the designer to flag.
What other details haven’t been given full attention? Has the data been handled sloppily as well? … On the flip side,
clean, layered, and logically built final files are a thing of beauty and my confidence in the designer, and their
attention to detail, soars.’ Jen Christiansen, Graphics Editor at Scientific American
The enduring idea that elegance in design is most appreciated when it is absent is just as relevant
with composition. Design solutions that felt effortless to navigate through visually will lead to a
superior experience compared with those that felt punctured, chaotic and confusing.
Thoroughness in the precision and consistency of your layout is important because any
shortcomings will be immediately noticeable and will undermine the elegance. Pay attention to
the smallest things: care about every last dot or pixel.
440

Summary: Composition
441

Features of Composition
This chapter explored the final element of developing your design solution concerning how you
will organise the placement and sizing of all your visual elements within the space you have to
work.
Layout: What is the visual hierarchy of your project? Making decisions about the relative
size and placement of all your visual elements, including charts, interactive controls and
annotations.
Arranging: Concerning the ordering and direction of your data content as it is displayed
within a chart with different options including alphabetical, chronological, locational,
ordinal and ranked sorting.
Sizing: Ways of astutely using the technique of small multiples and correct approach to
sizing charts through axis-scale ranges for different chart types.
442

Influencing factors and considerations
If these were the options, how did you make your choices? The influencing factors included:
Medium: What space have you got to work within?
Quantitative value range: What is the minimum and maximum value range and are there
legitimate outliers (large or small) that will skew the distribution of values and create
challenges for accommodating them?
Editorial thinking: How many different angles (charts) might you need to include? Is there
any specific hierarchy of importance or sequence that needs to be conveyed?
Data representation: All charts have a spatial consequence and have varied structures and
sizing requirements that will need to be accommodated.
Elegant design: The unity of your layout, offering a seamless visual journey to the viewer, is
another contributing factor that will create elegance in your work.
443

General Tips and Tactics
Empty space is like punctuation in visual language – use it to break up content when it
needs that momentary pause, just as a comma or full stop is needed in a sentence. Do not be
afraid to use empty space more extensively across larger regions as a device to create impact.
Like the notes not played in jazz, effective composition can be achieved through the
distinction between something and nothing.
At the deepest level of composition thinking you will become evermore consumed by the
tiniest of precision judgements. The task of nudging things by fractions of a pixel and
constantly resizing and realigning features will dominate your progress.
As your energy starts to diminish, and your deadlines start to emerge, you will need to
maintain a commitment to thoroughness and a pride in precision right through to the end!
What now? Visit book.visualisingdata.com
EXPLORE THE FIELD Expand your knowledge and reinforce your learning about working with data through this
chapter’s library of further reading, references, and tutorials.
TRY THIS YOURSELF Revise, reflect, and refine your skill and understanding about the challenges of working
with data through these practical exercises.
SEE DATA VISUALISATION IN ACTION Get to grips with the nuances and intricacies of working with data in
the real world by working through this next instalment in the narrative case study and see an additional extended
example of data visualisation in practice. Follow along with Andy’s video diary of the process and get direct insight
into his thought processes, challenges, mistakes, and decisions along the way.
444

http://book.visualisingdata.com

Epilogue The Development Cycle
The previous five chapters have been leading you through design decision making, broadening
your awareness of what options exist and then informing you about the things that will shape
your choices. The consequence of this will be a fully reasoned design specification. This
represents your conceptual thinking – a detailed plan of what you intend to develop.
In this closing section of this book, I want to leave you with an understanding of what happens
next as you continue the process of developing your design (Figure E.1), translating your concept
idea into a technically executed solution.
Figure E.1 From Concept to Solution: A Wireframe Sketch and Final Design for Nobels, No
degrees, by Accurat
The development cycle (Figure E.2) is characterised by loops of iteration at the intersection of
each successive step as you gradually nudge your ideas forward through successive rounds of
enhancements. The degree of iteration required will depend on the size of the gap that exists
between the idealism of your plan and the reality of what is technically achievable. Often, you
will not really know what is feasible until you try it. Thoughtful planning can help restrict the
size of this gap, but even then, the process of construction will likely incur impromptu revision or
unforeseen compromises. Trade-offs between ambition and pragmatism are inevitable. At times,
when things are just not progressing as you wish, you might need to go back to the drawing
board, returning to the start of your design thinking to conceive new choices and carve out a
different path.
Figure E.2 The Stages of the Development Cycle
The relevance of and rigour required across each of these steps is for you to determine. If the
solution you are creating is relatively simple in nature it will not likely involve the same demands
445

of development as would be necessary with creating a complex interactive visualisation. You
should see this cycle as an indicative outline; adapt it and use it as you see fit.
Let’s learn more about the specific components of the development cycle. As mentioned, the
consequence of the previous five chapters of this book led to completion of the first step,
‘developing a concept design specification’, which would have entailed creating initial
storyboards and/or wireframes to capture what you intend to create. So what follows this?
Create a mock-up and/or prototype: Whereas wireframing and storyboarding are characterised
by the creation of low-fi sketches, the development of mock-ups or prototypes advances the detail
of your draft ideas towards a technical solution (assuming you are not producing your output by
hand). This effectively leads to the creation of a first version that should closely demonstrate
some of the main features of the eventual solution. With the next step in the cycle concerned
with testing, this will understandably need to be focused on obtaining the most helpful feedback.
The respective terms tend to be used interchangeably but I feel mock-up is more applicable for
developing static work, whereas prototype is more relevant to interactive work.
Conduct testing: To move forward from a mock-up or prototype version requires testing. The
first round of testing is done automatically and constantly by you and/or your collaborators to
iron out any obvious immediate problems. In software development parlance, this would
generally be consistent with alpha testing. Naturally, beta follows alpha and next you will need to
seek others to ‘use’ it, evaluate it and give feedback on it. Even in small projects it is worth
considering this, even if it involves offering the eventual viewer a first look. Testing happens
regardless of the output format; it does not need to be a digital, interactive project to merit being
tested. There will be different aspects of testing to conduct, and it is worth revisiting the three
design principles to organise these:
Trustworthy design testing concerns assessing the reliability of the work, in terms of the
integrity of its content and performance. Are there any inaccuracies, mistakes or even
deceptions? Are there any design choices that could lead to misunderstandings? Any aspects
in how the data has been calculated or counted that could undermine trust? If it is a digital
solution, what is the speed of loading and are there any technical bugs or errors? Is it
suitably responsive and adaptable in its use across different platforms? Try out various user
scenarios: multiple and concurrent users, real-time data, all data vs sample data, etc. Ask the
people testing your solution to try to break it so you can find and resolve any problems
now.
Accessible design testing relates to how intuitive or sufficiently well explained the work is.
Do viewers understand how to read it and what all the encodings mean? Is the viewer
provided with a sufficient level of assistance that would be required in accordance with the
defined characteristics of the intended audience? Can testers find answers to the questions
you intended them to find and do this suitably quickly? Can they find answers to the
questions they think are most relevant?
Elegant design testing relates to questions such as: Is the solution suitably appealing in its
design? Are there any features which are redundant or superfluous design choices that are
impeding your engagement? Do you feel the appearance sustains any positive initial
sentiment?
446

Whoever you invite to test your work will vary considerably in each project context. Generally,
you may consider different cohorts of testers:
Stakeholders: The ultimate customers/clients/colleagues who have commissioned the work
may need to be included in this stage, if not for full testing then at least to engage them in
receiving initial concept feedback.
Audience: You might choose a small sample of your target audience and invite those viewers
to take part in the beta testing.
Critical friends: Peers/team/colleagues with suitable knowledge and appreciation about your
working process may offer a more sophisticated means of expressing feedback.
You: Sometimes (often) it may ultimately be down to you alone to undertake all testing,
through either lack of access to other people or due simply to the lack of time. To
accomplish this effectively you have to find a way to detach yourself from the mindset of the
visualiser and try to occupy that of the viewer. You need to see the wood and the trees.
‘We can kid ourselves that we are successful in what we “want” to achieve, but ultimately an external and critical
audience is essential. Feedback comes in many forms; I seek it, listen to it, sniff it, touch it, taste it and respond.’ Kate
McLean, Smellscape Mapper and Senior Lecturer Graphic Design
The timing of when to seek feedback through testing is another matter to consider. Sometimes
the pressure from stakeholders requesting to see progress will determine this. Otherwise, you will
need to judge carefully the right moment to do so. You do not want to get feedback when it is
too late or change is expensive. Similarly, it can be risky showing nominated testers your
undercooked concepts, perhaps just sharing early wireframes, when they might not have the
capacity to imagine how this will materialise into a polished final solution.
Refine and complete: Based on the outcome of your testing process, this will likely trigger a
need to address any issues that have been flagged or embrace new opportunities that emerge.
Troubleshooting is one characteristic of this stage, as too is editing, which is more aligned with
fine-tuning than problem solving. Regardless of the label, some of the features you are looking to
cover here will include:
correcting identified errors or issues;
stripping away superfluous content;
checking and enhancing the remaining content;
squeezing out final degrees of sophistication from every layer of your design;
improving the consistency and cohesion of your choices;
double-checking the accuracy of every component;
revisiting initial requirements and agreed definitions.
In any creative process a visualiser is faced with having to declare work complete. Judging this can
be quite a tough call to make. While the looming presence of a deadline (and, at times, agitated
stakeholders) will sharpen the focus, often it comes down to a fingertip sense of when you feel
you are entering the period of diminishing returns – when the refinements you make no longer
add sufficient value for the amount of effort you invest in making them. Eventually, you will
reach a judgement that your work is good enough. Completion is perhaps never truly reached in a
creative process lacking the single perfect solution – you just need to finish.
447

‘You know you’ve achieved perfection in design, not when you have nothing more to add, but when you have
nothing more to take away.’ Antoine de Saint-Exupéry, Writer, Poet, Aristocrat, Journalist and Pioneering Aviator
Launch and evaluate: The nature of launching your work will vary significantly, based, as
always, on the context of your challenge. It might simply be emailing a chart to a colleague or
presenting your work to an audience. In other cases, it could be a graphic going to print for a
newspaper or involve an anxious go-live moment with the launch of a digital project. Whatever
the context of your launch, there are a few characteristic matters to bear in mind. These will not
be relevant to all project scenarios but, over time, you might encounter them in different
situations:
‘Admit that nothing you create on deadline will be perfect. However, it should never be wrong. I try to work by a
motto my editor likes to say: No Heroics. Your code may not be beautiful, but if it works, it’s good enough. A
visualisation may not have every feature you could possibly want, but if it gets the message across and is useful to
people, it’s good enough. Being “good enough” is not an insult in journalism – it’s a necessity.’ Lena Groeger,
Science Journalist, Designer and Developer at ProPublica
Are you ready? Regardless of the scope of your work, as soon as you declare work completed
and published you are at the mercy of your decisions. You are no longer in control of how
people will interpret your work and in what way they will truly use it. If you have a
particularly large, diverse and potentially emotive subject matter, you will need to be ready
for the questions and scrutiny that might head in your direction.
Communicating your work is a big deal. The need to publicise and sell its benefits is of
particular relevance if you have a public-facing project. You might promote it loudly or
leave it as a ‘slow-burner’ to spread through word of mouth. For more modest or intimate
audience types you might need to consider directly presenting your work to these groups,
coaching them through what it offers. This is particularly necessary on those occasions when
you may be using an unfamiliar representation approach.
What ongoing commitment exists to support the work? This uniquely relates to digital
projects. Do you have to maintain a live data feed? Will it need to sustain operations with
variable concurrent visitors? What happens if it goes viral – have you got the necessary
infrastructure? Have you got ongoing access to the people/skills required to keep the project
alive and thriving?
Will you need to revise, update and rerelease the project? Will you need to replicate this
work on a repeated basis? What can you do to make the reproduction as seamless as
possible?
What is the work’s likely shelf life? Does it have a point of expiry after which it could be
archived or even killed off? How might you digitally preserve it beyond its useful lifespan?
There are two components in evaluating the outcome of a visualisation solution that will help to
refine your capabilities: what was the outcome of the work; and how do you reflect on your
performance?
Measuring the effectiveness of a data visualisation from an outcome perspective remains an
elusive task. This is largely because it can only be determined according to contextual measures of
success. This is why defining ‘purpose’ is necessary early on.
448

Sometimes effectiveness is tangible, but most times it is entirely intangible. If the purpose of the
work is to further the debate about a subject, to establish reputation or voice of authority, these
are hard things to measure in terms of positive outcome. One option may be to invert the
measurement to seek evidence of tangible ineffectiveness. For example, there may be significant
reputation-based impacts should decisions be made on inaccurate, misleading or inaccessible
visual information.
There are, of course, relatively free quantitative measures available for digital projects, including
web-based metrics such as visitor counts and social media engagement (e.g. likes, retweets,
mentions). These, at least, provide a surface indicator of success in terms of the project’s apparent
appeal and spread. Ideally, however, you should aim to supplement this by collecting more
reliable qualitative feedback, even if this can, at times, be rather expensive to secure. Some
options include:
capturing anecdotal evidence from comments submitted on a site, opinions attributed to
tweets or other social media descriptors, feedback shared in emails or in person;
informal feedback through polls or short surveys;
formal case studies which might offer more structured interviews and observations about
documented effects;
experiments with controlled tasks/conditions and tracked performance measures.
A personal reflection of your contribution to a project is important for your own development.
The best way to learn is by considering the things you enjoyed and/or did well (and do more of
those things) and by identifying the things you did not enjoy/do well (and do less of those things
or do them better!). Look back over your project experience and consider the following:
Were you satisfied with your solution? If yes, why; if no, why and what would you do
differently?
In a different context, what other design solutions might you have considered?
Were there any skill or knowledge shortcomings that restricted your process and/or
solution?
Are there aspects of this project that you might seek to recycle or reproduce in other
projects? For instance, ideas that did not make the final cut but could be given new life in
other challenges?
How well did you use your time? Were there any activities on which you feel you spent too
much time?
Developing effectiveness and efficiency in your data visualisation work will take time and will
require your ongoing efforts to learn, apply, reflect and repeat again. I am still learning new
things every day. It is a journey that never stops because data visualisation is a subject that has no
ending.
‘All of us who do creative work, we get into it because we have good taste … [but] there is this gap and for the first
couple of years that you’re making stuff, what you’re making is just not that good … It’s trying to be good, it has
potential, but it’s not. But your taste, the thing that got you into the game, is still killer. And your taste is why your
work disappoints you. A lot of people never get past this phase, they quit. Most people I know who do interesting,
creative work went through years of this. We know our work doesn’t have this special thing that we want it to have.
We all go through this. And if you are just starting out or you are still in this phase, you gotta know it’s normal and
449

the most important thing you can do is do a lot of work. Put yourself on a deadline so that every week you will finish
one story. It is only by going through a volume of work that you will close that gap, and your work will be as good as
your ambitions. And I took longer to figure out how to do this than anyone I’ve ever met. It’s gonna take a while. It’s
normal to take a while. You’ve just gotta fight your way through.’ Ira Glass, Host and Producer of ‘This American
Life’.
450

These references relate to content mentioned in the body text and/or attributed quotes that do
not come from individual interviews with the author. Extensive further reading lists to support
each chapter’s content are provided in the companion digital resources.
Bertin, Jacques (2011) Semiology of Graphics: Diagrams, Networks, Maps. Redlands, CA: ESRI
Press.
Chimero, Frank (2012) The Shape of Design. http://shapeofdesignbook.com/
Cleveland, William S. and McGill, Robert M. (1984) ‘Graphical Perception: Theory,
Experimentation, and Application to the Development of Graphical Methods’. Journal of the
American Statistical Association, vol. 79, no. 387, pp. 531–54.
Cox, Amanda (2013) Harvard Business Review. https://hbr.org/2013/03/power-of-
visualizations-aha-moment/
Crawford, Kate (2013) ‘The Hidden Biases in Big Data’. Harvard Business Review.
http://blogs.hbr.org/cs/2013/04/the_hidden_biases_in_big_data.html
de Bono, Edward (1985) Six Thinking Hats. New York: Little, Brown.
Glass, Ira (2009) Open Culture.
www.openculture.com/2009/10/ira_glass_on_the_art_of_story_telling.html
Heer, Jeffrey and Shneiderman, Ben (2012) ‘Interactive Dynamics for Visual Analysis’. ACM
Queue, vol. 10, no. 2, p. 30.
Ive, Jonny, Kemp, Klaus and Lovell, Sophie (2011) Dieter Rams: As Little Design As Possible.
London: Phaidon Press.
Kahneman, Daniel (2011) Thinking Fast and Slow. New York: Farrar, Straus & Giroux.
Mackinlay, Jock (1986) ‘Automating the Design of Graphical Presentations of Relational
Information’. ACM Transactions on Graphics (TOG), vol. 5, no. 2, pp. 110–41.
Morton, Jill (2015) Color Matters. www.colormatters.com/color-and-design/basic-color-theory
Munzner , Tamara (2014) Visualization Analysis and Design. Boca Raton, FL: CRC Press.
451

http://shapeofdesignbook.com/

https://hbr.org/2013/03/power-of-visualizations-aha-moment/

http://blogs.hbr.org/cs/2013/04/the_hidden_biases_in_big_data.html

Ira Glass on the Art of Story Telling

http://www.colormatters.com/color-and-design/basic-color-theory

Reichenstein, Oliver (2013) Information Architects. https://ia.net/know-how/learning-to-see
Rumsfeld, Donald (2002) US DoD News Briefing.
https://en.wikipedia.org/wiki/There_are_known_knowns
Shneiderman, Ben (1996) ‘The Eyes Have It: A Task by Data Type Taxonomy for Information
Visualizations’. Proceedings of the IEEE Symposium on Visual Languages. Washington, DC:
IEEE Computer Society Press, pp. 336–43.
Stefaner, Moritz (2014) Well-formed Data. http://well-formed-data.net/archives/1027/worlds-
not-stories
Stevens, Stanley Smith (1946) ‘On the Theory of Scales of Measurement’. Science, vol. 103, no.
2684, pp. 677–680.
Tukey, John W. (1980) ‘We Need Both Exploratory and Confirmatory’. American Statistician,
vol. 34, no. 1, pp. 23–5.
Tversky, Barbara and Bauer Morrison, Julie (2002) ‘Animation: Can it facilitate?’. International
Journal of Human-Computer Studies, Special issue: Interactive graphical communication, vol.
57, no. 4, pp. 247–62.
Vitruvius Pollio, Marcus (15 BC) ‘De architectura’.
Wotton, Sir Henry (1624) The Elements of Architecture. London: Longmans, Green.
452

https://ia.net/know-how/learning-to-see

https://en.wikipedia.org/wiki/There_are_known_knowns

Worlds, not stories

Index
In this index, titles or caption descriptions of visualisations are printed in italics. The numbers of
the pages where they appear are in bold print. Numerals at the beginning of an entry are treated
as words, e.g. ‘3D decoration’ is filed after ‘thoroughness’.
accessibility 45–50, 127, 243–244
testing 293
Accurat 194, 236, 295
acquisition of data 71, 96–97, 115
adaptability 33
aesthetic seduction 53
aims of book 1–2, 5
Aisch, Gregor 126, 127
All the Buildings in Manhattan 197, 198
alphabetical sorting 280–281
analysis and communication 6–7
analysts, qualities and traits of 112
Andrews, Wilson 79
angle of analysis 119–121, 123, 125, 126, 128
animating 217–219, 218, 227
annotating (interactivity) 215–217
annotations 19, 82, 125, 128, 231–247
audience 244
audio 242–243
captions 240–241
chart apparatus 238, 239
chart references 238–239
elegant design 246
footnotes 243
headings 231–233
labels 240–241, 290
and layout 280
legends 237–238
methods statements 243
missing annotations 43
and purpose 244–245
reader guides 235–236
and setting 244
for specific chart types see charts, types, gallery of charts
and text size 285
user guides 233, 234–235
Annual Staff Perception Survey 261, 262
API (Application Programming Interface) 96
453

appending data 109
areas of shapes 196
arranging 280–285
Art in the Age of Mechanical Reproduction: Walter Benjamin 267–270, 268
Asia Loses its Sweet Tooth for Chocolate 53
asking questions 116
‘at-a-glance’ viewing 76
attention span 49–50
attention to detail 36, 116, 247, 290, 291
attitudes 50
attributes (channels) 17, 18, 135, 137
audiences:
and accessibility 45–50, 129
attention span 49–50
attitudes and emotions 50
beta testing 295
definition 10
diversity of 31, 67
empathy for 47
and interactivity 227–228
interests of 64
knowledge of subject 67, 244
language 67
listening to 35
motivation 67, 86
needs 127
and relevance 46, 64
respect for 51–52
thinking about in design process 33–34, 129
and use of annotations 244
and use of colour 273–274
visualisation literacy 67, 244
visualiser’s knowledge of 67
see also complexity/simplicity; understanding
audio 240–241
‘Avengers’ characters’ appearance over time 179
Average Weekly Brent Crude Oil Prices, 2008–2018 175
axes 12
axis scales 240–241, 285–287
axis titles 240
Baby Names in England and Wales 2017 209, 259
backups 116
Baldwin, Taylor 198
bandings 239
bar charts 21, 43, 75, 76, 140, 194
454

axis scales 285, 286, 287
truncation 286, 287
Barratt, John 238, 271
Baseball Home Run Trajectories 198
Battling Infectious Diseases in the 20th Century: The Impact of Vaccines 240
Baur, Dominik 81, 103, 282
Beccario, Cameron 221
Bees, Drew 126
Berkowitz, Bonnie 259
Berliner Morgenpost 215, 254
Bertin, Jacques: Semiology Graphique 190
bias 39
Black Students Are Underrepresented On Campus 82, 83, 239
Bloomburg Billionaires 152
Bloomberg News 108
Bloomberg Visual Data 108
boardrooms 72
Boom and Bust: The Shape of a Roller-coaster Season 237
Brady, Tom 126
brain ‘states’ 34–35
Breakdown of Michael Schumacher’s F1 Career Over 308 Races 157
Breathing Earth 218, 219
Bremer, Nadieh 219
brief: formulating 32, 61
brushing 210
Bryant, Kris 197, 198
bubble plots 167, 207
budget 70
Bui, Quoctrung 124
bump charts 172, 209, 284
Burn-Murdoch, John 265
Buying Power: The Families Funding the 2016 Presidential Election 77, 79, 89
Cable, Dustin A. 221
Cairo, Alberto 62
Camoes, Jorge 37
Campbell, Sarah 227
captions 240–241
Carli, Luis 85
Carbon Map 186
cartesian charts 291
Casualties 272
Cesal, Amy 73
Chan, C. 90
Chang, Alvin 224, 258
channels see attributes (channels)
455

chart apparatus 238
chart references 238–239
Charting the Beatles: Song Structure 260
Chartmaker Directory 189–90
charts:
and graphs, plots and diagrams 12
lines and attributes 17–18
and tables 19
technological constraints 189–190
3D form 43
types 11–12, 18
‘CHRTS’ family of types 138
gallery of charts 138–188
area cartogram 186
area chart 175
bar chart 140
beeswarm plot 153
box-and-whisker plot 156
bubble plot 167
bullet chart 142
bump chart 172
chord diagram 170
choropleth map 180
clustered bar chart 141
connected dot plot 146
connected scatter plot 174
dendogram 164
density plot 155
diverging bar chart 160
Dorling cartogram 187
dot map 184
dot plot 152
flow map 185
Gantt chart 178
grid map 188
heat map 150
histogram 154
instance chart 179
isarithmic map 181
line chart 171
Marimekko chart 161
matrix chart 151
network diagram 168
pictogram 147
pie chart 157
polar chart 145
456

prism map 183
proportional symbol chart 148
proportional symbol map 182
radar chart 144
Sankey diagram 169
scatter plot 166
slope graph 173
stacked area chart 176
stacked bar chart 159
stream graph 177
sunburst chart 163
treemap 162
Venn diagram 165
Voronoi treemap 138
waffle chart 158
waterfall chart 143
word cloud 149
and range of values 194
and types of data 194, 195
unfamiliar types 49
usage 43
Cheshire, James 238, 271
Chimero, Frank, The Shape of Design 51
choices:
of chart types 80
and restrictions 66
what to include/exclude 122
see also decision making
chord diagrams 170, 284
Chow, Emily 259
Christiansen, Jen 54, 290
Chrome Dominates a Cluttered Browser Market 195
chronological sorting 281
City of Anarchy 277–279, 278
Ciuccarelli, Paolo 190
clarification 49
cleaning data 106
Cleveland, William and McGill, Robert: ‘Graphical Perception: Theory, Experimentation,
and Application to the Development of Graphical Methods’ 190, 192
Clever, Thomas 52, 108
CNN 271
cockpit setting 72
coffee shops 72
collaboration 69
Collins, Keith 108
Colors of the Rails 260, 261
457

colour 247–274
applications 19, 125, 126, 127
background neutral tones 270
categorical classifications 256–265
categories of colour 260
with large numbers of categories 259–260
CIELAB model 252
CMYK (Cyan, Magenta, Yellow, Black) model 250
contrasting 82, 253
converging/diverging colour scales 254, 255
functional decoration 265–272
harmony 267
HSL (Hue, Saturation, Lightness) model 250–252
impact of 45
judging variation 192
legibility 252
maps 266
medium 272
print quality 275
and purpose 273
quantitative scales 253–256, 254, 255
rainbow scale 255–256
RGB (Red, Blue, Green) model 250
rules and guidelines 272–273
and visual impairments 273–274
colour-blind-friendly altenative tones 274
commitment 297
communication 35–36, 114
comparing circles 196
Comparing the Degree of Trust by Australians for Different Institutions 159
Comparing Relative Values and Daily Changes of Market Capital for Stocks 160
completing 296
complexity/simplicity 46, 48–49
in representation 49
complicated subjects 48
composition 19, 129, 277–292
arranging 280–285
and data representation 290
and data types 99
and editorial thinking 125, 289–290
and elegance 51, 290
features 291
layout 277–279, 288
and medium 288
quantitative value range 288, 289
sizing 285–287
458

for specific chart types see charts, types, gallery of charts
see also style
comprehending (phase of understanding) 24–5, 74
confusion 47, 48, 49
connected scatter plots 174, 217
consolidating data 108–109
constraints 66, 69–71
consuming: definition 11
content 86
context:
circumstances 65–66
constraints 69–71
and content 86
deliverables 71–73
motivating curiosity 62–65
people 66–69
Coral Cities 285, 286
correlation 12
Corum, Jonathan 269
Countries With The Most Land Neighbours 140
Cox, Amanda 79
Crawford, Kate 42
creative influences 70
creativity 33, 36, 53
Cricketer Alastair Cook Plays His 161st Final Test Match 264, 265
critical thinking 5
Cruz, Peter 193
Cultural Politics: Marijuana Use and Same-sex Marriage in USA 165
cultural sensitivities 70
curiosity 62–65, 74
multiple curiosities 64–65
and question forming 65, 119
sample statements 65
specificity 64
customers see stakeholders
Daily Indego Bike Share Usage in Philadelphia 176
Daily Mail 44, 45
Daily Price and Availability of Super Bowl Tickets 174
dashboards 28
data:
creating 106–107
perfect data 115
quality/condition 105, 131
security/privacy 116
technologies for presentation 71
459

types 11, 16, 97–102
working with 32, 96–115
acquisition 71, 96–97, 115
examination 71, 97–105, 115
exploration 109–115
sorting 280–285
transformation 71, 106–109, 115
data art 28
data journalism (data-driven journalism (DDJ) 28
data representation 133–199
chart types 11–12, 18
gallery of charts 136–186
and composition 288
influenced by angle and framing 124–125, 128
interactivity 226
technological constraints 71, 189–190
trustworthy design 196–200
visual encoding 135–137, 136
data science 28
data source: definition 11
data visualisation: definition 15
datasets 11
cross-tabulated 98–99
expanding/appending 108–109
normalised 98
de Bono, Edward, Six Thinking Hats 67
De Niro, Robert 63
deadlines 69
Deal, Michael 260
decision making 20, 31, 39, 118
optimising 37–38
decoration 52–55
3D decoration 197
see also functional decoration
deductive reasoning 113
D’Efilippo, Valentina 92, 241
deliverables 71–73
interactivity 225
DeSantis, Alicia 79
design characteristics 74
design principles 37–56, 38, 57
accessibility 45–50, 245–246
elegance 50–56, 290
trustworthiness 38–45, 196–200, 226, 227
design process 31–37, 57
process and procedure 33
460

reasons for following 32–37
stages 32
see also design principles
design restrictions 70
development cycle 291–296
stages 292
diagrams: definition 12
‘Dialect quiz map’ 85, 86
differences in projects 32–33
digital resources 10, 225
Dimensional Changes in Wood 85
discrete/continuous data 102
Do You Remember Where Germany Was Divided? 215
documenting 35
doing (practical undertakings) 34
and not doing 36
domain expertise 97
donut charts 194, 195
Doping under the Microscope 288
Du Bois, W.E.B. 6
duration of task 69–70
Earth 217, 219
Ebb and Flow of Movies: Box Office Receipts 1986–2008 177
ECB Bank Test Results 211, 211
Ecological Footprint and Biocapacity 167
editing 51, 119
editorial angle 196
editorial thinking 119–131
and composition 125, 289–291
establishing 32
examples:
Fall and Rise of US Inequality 123–125, 124
Why Peyton Manning’s Record Will Be Hard to Beat 125–129, 126, 127
effectiveness 297, 298
elegance 50–56
in annotations 246
and composition 51, 290
interactivity 226–227
testing 293
eliminating the arbitrary 51
Elliot, Kennedy 33, 217, 234
emotions 26, 50, 76–80, 190
manipulation of 79
emphases 122
empty space 292
461

enclosure charts 291
enlightenment 26
environmentally friendly design 56
Equal Earth projection 199
ER Wait Watcher: Which Emergency Room Will See You the Fastest 281
erroneous values 105
evaluation 297–298
Evans, Tom 213
Every Time Ford and Kavanaugh Dodged a Question 257–8
examination of data 71, 97–105, 115
Excel 99
Executive Pay by the Numbers 256–7
exhibitory visualisations 86–88, 245
expanding data 108–109
experiences offered by visualisation 82–88
experimentation 33
explanatory visualisations 82–84, 87, 245
explorable explanations 86
exploration of data 109–115, 195–196
exploratory data analysis (EDA) 111–115
exploratory visualisations 84–86, 245
facilitating 26, 37
Fall and Rise of US Inequality, The 123–125, 124
Falling Number of Young Homeowners 44
familiarity 49
feedback 295–296
from friends 295
feeling tone 76–81, 190
figure-ground perception 40
Filmographics 62–3, 279
filtering 202–205
Financial Times 67, 97, 265
FinViz: Standard & Poor’s 500 Index Stocks 80, 207, 274
FiveThirtyEight 83, 239, 284, 290
Flasseur, Vincent 212
focus 122–123, 124, 126, 127, 128
fonts 245, 246
Foo, F. 90
footnotes 243
foraging for data 97
Forbes: The World’s 100 Highest-paid Athletes 87, 88, 147
Ford, Christine Blasey 257
Forecast % Chance of Winning Presidency (US Election, 8th November 2016) 25–6
formats 12
formatting data 108
462

Four Teams in Group F of 2018 World Cup 145
framing 122, 123, 126, 128
frequency 73
frequency counts 103
frequency distribution 103
Frequency of Words Used in Ch 1 of First Edition of This Book 149
Fuller, Richard Buckminster 51
functional decoration 266–272
functionality 12
Funds Raised Across USA for Election Candidate Hillary Clinton 182
Funke Interaktiv 215
Gender Pay Gap US/UK 146
geometric miscalculations 196
German General Election Results Showing Winning Party for Each Location 180
Ghael, Avni 193
Glass, Ira 298
Global Flow of People 170
goal of visualisation 74
see also purpose
Goddemeyer, Daniel 282
Goldsberry, Kirk 79
Gourlay, Colin 214
Grabell, Michael 210
Grape Expectations 90
Graphic Language: The Curse of the CEO 107–8
graphics 12
graphs 12
see also charts
Gray, Marcia 106
Green, Jeff 108
Grimwade, John 222
Groeger, Lena 46, 281, 296
Grothjan, Evan 79
Growth in Participants and Female Participation at the Summer Olympics 161
Gun Deaths in Florida 40, 42, 79
Guys Named John, and Gender Inequality 263
harmony 265
harnessing ideas 88–93
headings/titles:
artistic 232–233
descriptive 232
and introductions 233, 234
as questions 232
statements 231–233
and storyboarding 280
463

Heer, Jeff and Shneiderman, Ben: ‘Interactive Dynamics for Visual Analysis’ 226
Hemingway, Ernest 29
hidden thinking 32
Highest Max Temperatures in Australia 255–6
History Through the President’s Words 216, 217, 233
Hobbs, Amanda 36, 48
Holdouts Find Cheapest Super Bowl Tickets Late in the Game 216, 218
Holmes, Nigel 91
honesty 38, 39
Horse in Motion 229
Household Incomes for Simulated Population of Chicago Residents 153
Housing and Home Ownership in the UK 44, 45
How the ‘Avengers’ Line-up Has Changed over the Years 216
How Big Will the UK Population Be in 25 Years’ Time? 210
How Each State Generates Electric Power (2004–2014) 173
How Good is ‘Good’? 155
How Inclusive are Beauty Brands Around the World? 154
How Long Will We Live – And How Well? 166, 259
How Nations Fare in PhDs by Sex 204, 206
How Popular is Your Birthday? 150
How Well Am I Running? 121, 122
How Well Do You Know Your Area? 212, 213
How Y’all, Youse and You Guys Talk 86, 181
Hubley, Jill 206
hue 250–252
Hurt, Alyson 69, 113, 114
If Vienna Would Be an Apartment 261, 261
inductive reasoning 113
info-posters 27
infographics 27
information design 27–28
information visualisation 27
Ingold, David 108
innovative design 55
integrity 43–45, 295
interactivity 125, 128, 203–230
accessibility 227–228
animating 217–219, 218
annotating 215–217
with colour for categorical classification 259
data representation 226
elegance 228–229
event, control, function 204
in explorative visualisations 84, 85
filtering 204–207
464

highlighting 207–211, 208
influencing factors 225–229, 226
and layout 280
and medium 288
navigating 219–224, 220
participating 211–215, 212
technologies for 71, 225
user guides 233, 234–235
interestingness 129
internet as source of data 96
interpreting 22–23
interval data 101
introductions 233, 234
Iraq’s Bloody Toll 41, 42
iteration 33, 130
Ito World 286
Jacobs, Brian 87, 289
Jenkins, Nicholas 235
Johnson, Richard 217, 234
judging comparative size 191
Kahneman, Daniel, Thinking Fast and Slow 88–89
Kasich Could Be the GOP’s Moderate Backstop 284–5
Katz, Josh 86
Kavanaugh, Brett 257
Keegan, Jon 214
Killing the Colorado: Explore the Robot River 222
Kindred Britain 235
Kirchner, Lauren 222
Kirk, Andy 63, 237, 243, 257, 279
Klack, Moritz 254
Klee, Paul 34
Knott, Matt 63, 279
Known Knowns 111, 113
Kocinova, Lucia 92, 241
labels 240–241, 290
Lambert Arimuthel Equal-area projection 199
Lambrechts, Maarten 283
Larson, Jeff 222
launching 296–297
layout 277–280
and annotations 280
and interactivity 280
and size 279
learning 114, 116
465

legends 12
legibility 246
colour 252
levels of data 99
Li, Jason 154
Liberals Most Likely to Favor No Restrictions on Abortion 160
Life Cycle of Ideas 236
lightness (colour) 251, 253
Lindemann, Todd 259
line charts 18, 239, 284
aspect ratio 43
axis values 276, 287
linking 208
Lionel Messi: Games and Goals for FC Barcelona 20–22, 21
locational sorting 281–282
log scales 289, 290
long-lasting data 56
long-lasting design 55–56
Losing Ground 86, 87, 288, 289
Lunge Feeding 269–71
Lupi, Giorgia 91, 194
Lustgarten, Abraham 222
McCandles, David 213
McGill, Robert 190
Mackinlay, Jock: ‘Automating the Design of Graphical Presentations of Relational
Information’ 190, 191
McLean, Kate 100, 295
making 34
Making Sense of Skills: A UK Skills Taxonomy 234
Manian, Divya 154
Manley, Ed 238, 271
Manning, Peyton 125, 126, 127, 128
Manovich, Lev 282
maps 12, 291
and colour 266
projection 197, 198–200
thematic 43, 197, 198–200, 199
see also charts
markers 239
Market Capitalisation of Companies 148
market influences 70
market share browsers 194, 195
marks 17–18, 135, 136
Marshall, Bob 87, 289
measurement of central tendency 103
466

measurements of spread 103
media (formats) 72–73
and colour 272
and composition 288
and interactivity 288
Mediafin 283
Meirelles, Isabel 47
Mellnik, Ted 217, 234
memorability 56
Mercator projection 199
Mercer ‘Quality of Living City’ 285
MeTooMentum 92, 241
Migliozzi, Blacki 223
minimalism 51
minimum friction 47
missing values 105
mistakes 36, 43
and geometric miscalculations 196
Mizzou’s Racial Gap is Typical on College Campuses 82, 83
mock-ups 294
Mollweide projection 199
Month in an Animal Shelter 169, 226, 227
Morrison, Julie Baur 228
Morton, Jill 267
multiple media production 72
Munzer, Tamara 285
Murray, Scott 66, 235
Muybridge, Eadweard 228
narrative visualisations 83–84
National Geographic 33
Native and New Berliners – How the S-Bahn Ring Divides the City 254
navigating 219–224, 220
Nearly Half of New Zealand’s Migration Gain is From Asia 143
Nelson, John 79, 116
New York City Street Trees by Species 204, 205, 206, 259
New York Times 79, 85, 86, 112, 126, 127, 253, 256, 263, 269
news and media organisation: style 52
NFL Players: Height and Weight over Time 219, 220
Nightingale, Florence 6
Nobel Laureates 207, 208
Nobel Laureates Awarded (1901-2017) by Country of Birth 75–6
Nobel Laureates by Category and Country of Birth 151
Nobels, No Degrees 295
NOIR (TNOIR) classification 99–100, 115
nominal data 100
467

Noonan, Laura 212
note-taking 35, 94, 116
nothings (in data) 114
Number of Vehicles Using Hong Kong’s Network of Roads 2011 185
NYPD Staffing Compared With Other Cities 264
NZZ 252, 261
Obama’s Health Law: Who Was Helped Most? 253, 254
objectivity 39
O’Brien, Oliver 238, 271
observation-meaning gap 23
OECD Better Life Index 81, 103, 192
Ireland 224
Office for National Statistics (ONS) 44, 45, 97, 209, 210, 213
On Broadway 282
One Angry Bird 76, 77
100 Years of Tax Brackets, in One Chart 223, 224
opinion 120
ordinal data 101
ordinal sorting 282, 283
orientation of charts 284–285
Ortiz, Santiago 88
outliers 12
Parshina-Kottas, Yuliya 79
participating 211–215, 212
Peek, Katie 35, 244
perceiving (phase of understanding) 21–22
Percentage of Hours During 2017 the Sun was Above the Horizon in Nuorgam, Finland 158
perceptual accuracy 191–193
perfection, impossibility of 31
Periscopic 77, 206, 211, 242
pertinence 129
photography 120
physical displays 197, 198
pie charts 157, 194, 195
plagiarism 89, 91
Playfair, William 6
plots 12
see also charts
political pressures 70
Politically Important Topics for Germans Between 1998 and 2017 172
Politizane 84, 242
Pong, Jane 72, 91, 277
Popularity of International Outlets 72, 73
Posavec, Stephanie 33, 51, 268
presentation 19, 50
468

pressures 70
primary data 97
print as medium 72, 288
project circumstances 65–66
project management 230
projects: definition 11
Proportion of Sales Percentage by Channel over Time 16–17, 18, 24–25
ProPublica 87, 210, 222, 281, 288, 289
prototypes 294
publicising work 297
publishing 71
purpose 61, 74, 76
and annotations 244–245
and choice of charts 190–194
and colour 273
and interactivity 225, 226
Pursuit of the Faster 257
Pursuit of the Faster (footnotes) 205, 243
Pyensen, Nicholas D. 269
Qiu, Yui 210
quantity 73
Quealy, Kevin 126, 127, 253
question forming 65
Racial Dot Map, The 222
radial structures 291
radio buttons 207
Rain Patterns 281
Rams, Dieter: Principles of Good Design 37, 38, 45, 50, 55
ranked sorting 282
Ranking the Ivies 156
Ranking of Perceptual Tasks 191
Rapp, Bill 71
ratio data 101
Raureif GmbH 81, 103
raw data 11
Razor Sales Move Online, Away from Gillette 53, 54
reading tone 75–76, 190, 245
reasoning 112–113
recurrence of concern 33–34
reducing randomness of approach 32
reference lines 239
refining 296
reflective learning 36–37
Reichenstein, Oliver 34, 52
relevance 46–47, 129–130, 230
469

representation complexity 49
research 36, 114
Reuters Graphics 90, 210, 287
Rim Fire – The Extent of Fire in the Sierra Nevada Range and Yosemite National Park 260,
261
Ring-Necked Parakeets 266–7
Roberts, Graham 79
Roston, Eric 223
Rougeux, Nicholas 236, 261
Rumsfeld, Donald 110
Runs Scored in Test Matches by English Batsmen 171
Russell, Karl 256
Ryan, Claudine 214
Saint-Expuéry, Antoine de 296
samples 105
Sandler, Adam 63
Sanger-Katz, Margot 253
Sankey diagrams 167, 284
sans-serif typefaces 245
saturation 251
scales 12
scales of measurement 99
Scarr, Simon 41, 90, 121, 278
scatter plots 82, 83, 125, 239
Scientific American 54, 290
scientific visualisation 28
scrapbooks 89, 94
‘scrollytelling’ 222
series 11
serif typefaces 245
settings 71–72
‘Seven Hats of Visualisation Design’ 68
shape of data 103–104
Share of Individuals Using the Internet 2015 187
Share of People Voting to Leave and Remain During the EU referendum 188
Shaw, Al 87, 222, 289
Shibuya, Felipe 193
shifting focus 64
Shneiderman, Ben 80, 226
Simmon, Robert 263
simplifying 48–49
size 277, 283–285
and quantitative value range 288, 289
restrictions 70
shrinking 285
470

size of data 102
sketching 35, 91, 92
skills needed for this book 3
skills of visualisers 8, 67–69
Sleeman, Cath 234
Slobin, Sarah 56, 127
small mulitples 285, 286
Smith, Alan 67, 97
Snow, John 6
Songs That Were #1 in the UK Charts for the Greatest Number of Weeks 282, 283
sophistication of content 46
sorting data 209, 280–285
sources of data see data, acquisition
South China Morning Post 41, 278, 281
Sparkes, Sophie 267
spatial analysis 291
Spielberg, Steven 63
Jaws 55
Spotlight on Profitability 104
Spraggon, Ben 214
squint test 273
stakeholders 63–64, 66, 295
Stamen 271
‘Stand Your Ground’ law (2005) 40
statistics:
knowledge of 7
supplemented with visuals 111–112
Stefaner, Moritz 81, 103, 104, 111, 120, 282
Stevens, Stanley 99
storyboarding 279, 280, 294
storytelling 28–29, 222–223
Streep, Meryl 63
style 52, 53, 70
for colour usage 272–273, 275
subject distinctions 27–29
subject-matter experts 66
subjectivity 39
sufficiency 129
Sugar Quiz: How Much Sugar Is in Our Food? 214
SUL-CDR 235
supplied data 96
Swift, Taylor 76, 78
System 1 and System 2 thinking 88–89
system download 96
Szücs, Krisztina 104
471

tabular designs 291
tabulation of data 11, 16–17, 24–25, 98–99
NOIR (TNOIR) classification 99–100
Tabuleau 99
Tallest Buildings Around the World (Effect of Truncation on Bar Charts) 286, 287
Taylor, Craig 286
Taylor Swift is Mostly Happy, Quite Often Sad, Sometimes Mad, and Occasionally Really
Scared 76, 77, 78
technological constraints 71, 189–190
interactivity 223
Ten Actors Who Have Received the Most Oscar Nominations for Acting 141
testing annotations 247
testing ideas 294–296
text size 285
textual data 100, 107–108
The Lens 87, 289
thinking 34
Thomas, Amber 154
thoroughness 51–52
3D decoration 197
Tigas, Mike 281
time 49–50, 118
time management 34
time-based data 102
timeliness 129
timescales 69–70
tone 74, 75–81, 190
choice of 75, 80
tones (colour) see saturation
Top 20 Ranked Batters in Men’s Test Cricket 142
Total Sightings of Winglets and Spungles 23
transformation of data 71, 106–109, 115
Transport for London (TfL) 96
Tree for US Immigration, A 193, 194
treemaps 80
Trillions of Trees 183
Tröger, Julius 254
troubleshooting 296
trust and truth 39–40
trustworthiness 38–45
data acquisition 96
in design 196–200, 226, 227
integrity 43–45
testing 295
Tufte, Edward 54
Tukey, John 112
472

Tulp, Jan Willem 109
Tversky, Barbara and Morrison, Julie Baur: Animation: Can it Facilitate 228
Twitter NYC: A Multilingual Social City 238, 270–1
200+ Beer Brands of SAB and AB InBev 282, 283
typefaces 245, 246
typology 245–246
Ulmanu, Monica 210
understanding 47–50, 67, 74
and confusion 47, 48, 49
content and context 86
delivering 26
facilitating 37, 74
phases 20–25, 74
United Kingdom: Global Competitiveness 144
UK Skills Taxonomy 164
univariate/bivariate/multivariate techniques 7
University of Missouri 82
unknowns 111, 113
updating projects 297
Upshot, The 79
US Guns Deaths 211, 241, 242
US National Parks 178, 236
US residents Based on Location in 2010 Census 184
usefulness 129
value intervals 285
value labels 241
variables 98
Veltman, Noah 220
Viégas, Fernanda 266
viewers:
definition 10
diversity of 26
knowledge 22, 23, 24
vision: definition 74
visual analytics 28
Visual Cinnamon 219
visual encoding 135–137
attributes (channels) 17, 18, 135, 137
marks 17–18, 135, 136
visual immediacy 192, 194
visualisation literacy 67, 190
visualisation as prop 72
visualisers 67
definition 10
as leader 92–93
473

‘Seven Hats’ (skills) 67–69, 68
visualising data 111
Vitruvius Pollio, Marcus, ‘De architectura’ 38, 45, 50
Voting Patterns for Democrats and Republicans Across Members of US House of Representatives
168
Walker, Jonni 265
Wall Street Journal 53, 54, 216, 240, 264
Washington Post 217, 234
waterfall charts 143, 286
Watkins, Derek 79
Wattenberg, Martin 266
Wealth Inequality in America 84, 242, 243
web crawling 96
web scraping 96
Weber, Matthew 208
Wei, Sisi 281
Wendler, David 254
What are the Current Electricity Prices in Switzerland? 252, 253
What Good Marathons and Bad Investments Have in Common 111, 112
What’s Really Warming the World? 223
‘Where’s Wally?’ 113
Which Companies Caused Global Warming? 163
Who Old Are You? 212, 213, 214
Why Peyton Manning’s Record Will Be Hard to Beat 125–129, 126, 127
Wihbey, John 193
Wind Map 265, 266
wine industry 89, 90
Winkel-Tripel projection 199
wireframing 279–280, 294
Witherley, Andrew 243, 257
Wolfers, Justin 112, 263
Workers’ Compensation Reforms by State 209, 210
World Top Incomes Database 124
Worst Games Ever Made, The 289, 290
Wotton, Sir Henry 38
Wu, Shirley 78
x-axis 12, 284, 289
y-axis 12, 42, 284
Yourish, Karen 79
Zamora, Amanda 222
zooming 221
474

475

Half Title
Publisher Note
Title Page
Copyright Page
Contents
Acknowledgements
About the Author
Discover Your Textbook’s Online Resources
Introduction
Part A Foundations
1 Defining Data Visualisation
2 The Visualisation Design Process
Part B The Hidden Thinking
3 Formulating Your Brief
4 Working With Data
5 Establishing Your Editorial Thinking
Part C Developing Your Design Solution
6 Data Representation
7 Interactivity
8 Annotation
9 Colour
10 Composition
Epilogue The Development Cycle
References
Index