Advertising

Article

Save Time On Research and Writing
Hire a Pro to Write You a 100% Plagiarism-Free Paper.
Get My Paper

A Meta-Analysis of When and How
Advertising Creativity Works

Sara Rosengren, Martin Eisend, Scott Koslow, and Micael Dahlen

Abstract
Although creativity is often considered a key success factor in advertising, the marketing literature lacks a systematic empirical
account of when and how advertising creativity works. The authors use a meta-analysis to synthesize the literature on advertising
creativity and test different theoretical explanations for its effects. The analysis covers 93 data sets taken from 67 papers that
provide 878 effect sizes. The results show robust positive effects but also highlight the importance of considering both originality
and appropriateness when investing in advertising creativity. Moderation analyses show that the effects of advertising creativity
are stronger for high- (vs. low-) involvement products, and that the effects on ad (but not brand) reactions are marginally stronger
for unfamiliar brands. An empirical test of theoretical mechanisms shows that affect transfer, processing, and signaling jointly
explain these effects, and that originality mainly leads to affect transfer, whereas appropriateness leads to signaling. The authors
also call for further research connecting advertising creativity with sales and studying its effects in digital contexts.

Keywords
advertising, affect transfer, creativity, meta-analysis, processing, signaling

Save Time On Research and Writing
Hire a Pro to Write You a 100% Plagiarism-Free Paper.
Get My Paper

Online supplement: https://doi.org/10.1177/0022242920929288

Creativity is important in marketing and is often considered to

be at the heart of the advertising industry. The importance of

creativity is highlighted, for example, in the popularity of

industry competitions, such as the Cannes Lions International

Festival of Creativity, and the growing academic literature on

its effects (e.g., Reinartz and Saffert 2013; West, Koslow, and

Kilgour 2019). However, the value of creativity is also subject

to longstanding debate (Baack et al. 2015; Levitt 1963), and

recent reports highlight that marketers are increasingly

growing skeptical of advertising creativity (Parsons 2019;

Premutico 2019) and decreasing their investments in it

(Forrester 2019).

When and how is advertising creativity most valuable? Mar-

keters wanting to answer these questions will find little gui-

dance in the academic literature. Although the link between

advertising expenditure and advertising effects has been con-

sistently supported (Joshi and Hanssens 2010; Sridhar et al.

2016), to date, there is no comprehensive account of advertis-

ing creativity and its influence on consumer response. Even

Vakratsas and Ambler (1999) failed to account for creativity

as a factor in their insightful and influential review of how

advertising works.

Several factors seem to hold back scholarship in advertising

creativity: (1) contrasting empirical results on its effects in

terms of ad and brand outcomes (e.g., Lee and Mason 1999;

Smith, Chen, and Yang 2008; Till and Baack 2005), (2) dis-

agreements over what creativity is and how it should be

assessed (e.g., Modig and Dahlen 2019; Smith, Chen, and Yang

2008), (3) limited understanding of moderators of its effect

(e.g., Yang and Smith 2009), and (4) ambiguity about the kind

of theories that best explain how creativity works (e.g., West,

Koslow, and Kilgour 2019; Yang and Smith 2009). Given the

apparent confusion about what advertising creativity is and

when it might benefit a brand, it is not surprising that marketers

often make the wrong decisions when investing in advertising

creativity (Reinartz and Saffert 2013).

In this article, we synthesize the fragmented literature on

consumer response to advertising creativity. Based on a litera-

ture review, we develop a conceptual framework linking adver-

tising creativity to consumer outcome responses in terms of ad,

brand, and sales. Through a meta-analysis, we then integrate

878 effect sizes in the first quantitative empirical overview of

Sara Rosengren is Professor, Center for Retailing, Stockholm School of

Economics, Sweden (email: sara.rosengren@hhs.se). Martin Eisend is

Professor, Marketing Department, European University Viadrina, Germany

(email: eisend@europa-uni.de). Scott Koslow is Professor, Department of

Marketing, Macquarie University, Australia (email: scott.koslow@mq.edu.au).

Micael Dahlen is Professor, Center for Consumer Marketing, Stockholm

School of Economics, Sweden (email: micael.dahlen@hhs.se).

Journal of Marketing
2020, Vol. 84(6) 39-56
ª The Author(s) 2020

Article reuse guidelines:
sagepub.com/journals-permissions
DOI: 10.1177/0022242920929288

journals.sagepub.com/home/jmx

https://doi.org/10.1177/0022242920929288

mailto:sara.rosengren@hhs.se

mailto:eisend@europa-uni.de

mailto:scott.koslow@mq.edu.au

mailto:micael.dahlen@hhs.se

https://sagepub.com/journals-permissions

https://doi.org/10.1177/0022242920929288

http://journals.sagepub.com/home/jmx

the literature. Thus, we capture the impact of advertising crea-

tivity on 19 different consumer responses taken from 93 data

sets in 67 papers. We thereby contribute a comprehensive and

empirically grounded account of how and when advertising

creativity works, providing researchers with generalized find-

ings that can serve as benchmarks and a common foundation

for future studies of this important topic.

First, we provide an empirically validated account of how

advertising creativity works. The results show robust positive

effects of advertising creativity on consumer responses but also

inform researchers about the relative importance of various

consumer responses to advertising creativity. Overall, effects

are stronger for ad rather than brand responses (r ¼ .491 vs.
.317) and for attitudinal rather than memory outcomes (all

below .140). This suggests that the main advantages of adver-

tising creativity are not grabbing attention and making ads

memorable but rather the ability to foster positive ad and brand

attitudes.

Second, we highlight that advertising creativity is different

from originality. Effects on consumer response are greater

when creativity is measured as a bipartite construct comprising

of originality and appropriateness. Effects of originality only

on ad and brand attitudes are comparatively small (.362 and

.164), suggesting that marketers who view creativity as synon-

ymous with originality will not reap the full benefits of invest-

ments in advertising creativity.

Third, we show that the different theoretical accounts used

in the literature to explain how advertising creativity works are

complementary. Although previously considered separately,

affect transfer, processing, and signaling provide the best

account when considered jointly. The results further show that

processing is a key mediator of the effects, whereas originality

fosters affect transfer and appropriateness signaling. When

marketers invest in bipartite creativity, affect transfer and sig-

naling occur in parallel to processing, which can explain the

stronger effects of creativity compared with originality only.

Fourth, we find that when the three theoretical accounts are

considered jointly, the effects of advertising creativity on brand

response are not dependent on ad responses. This is in line with

the signaling account of advertising creativity and suggests that

to fully understand how advertising creativity works, marketers

should assess consumer responses in terms of brand rather than

ad outcomes.

For managers, the results provide guidance on how, when,

and why to invest in advertising creativity. For example, adver-

tising creativity leads to greater ad responses in high (vs. low)

involvement communication contexts (.653 vs. .340) and (mar-

ginally) for unfamiliar (vs. familiar) brands (.577 vs. .435). The

literature review also highlights the need for more studies on

the relationship between advertising creativity and sales, as

well as its effects in digital media contexts, two areas that are

especially important given the ongoing debate about the value

of advertising creativity in contemporary marketing practice.

Conceptual Framework

Figure 1 presents the conceptual framework guiding the meta-

analysis. We developed this framework on the basis of a review

of the literature on consumer responses to advertising creativ-

ity. The framework focuses on consumer responses that have

been empirically studied and distinguishes between immediate

and outcome responses. The categorization of immediate

responses is based on the three main theoretical accounts of

how advertising creativity works found in the literature: affect

transfer, processing, and signaling (i.e., consumer response in

terms of affect, processing, and perceived signals at the time of

exposure to creative advertising). The categorization of out-

come variables is based on ad and brand responses and char-

acterized in terms of attitudinal or memory responses (i.e.,

longer-lasting responses related to the ad and brand, such as

attitudes, memory effects, and sales). The framework also high-

lights two moderators of these responses: definitions and

Advertising creativity Consumer response

Creativity
moderator

Originality and
appropriateness

Originality only

Contextual
moderator

Involvement

Familiarity

Immediate
response

Affect

Processing

Signals

Outcome
response

Ad
Attitude/memory

Brand
Attitude/memory

Salesa

Figure 1. How and when advertising creativity works: conceptual framework.
aNot tested empirically due to lack of data.

40 Journal of Marketing 84(6)

assessments of advertising creativity and properties of the com-

munication context (involvement and familiarity).

Whereas research on advertising creativity generally has

found positive effects on immediate and outcome responses

(for reviews, see Sasser and Koslow [2008] and Smith, Chen,

and Yang [2008]), empirical findings suggest that the effects

vary between different types of consumer responses. Findings

generally show that advertising creativity has benefits in terms

of immediate responses, such as attention (Pieters, Warlop, and

Wedel 2002; Smith et al. 2007), positive affect (Haberland and

Dacin 1992; Yang and Smith 2009), and signals, such as per-

ceived sender effort (Dahlen, Rosengren, and Törn 2008;

Lange, Rosengren, and Blom 2016), but results are inconsistent

regarding when and how this might also lead to outcome

responses, such as attitudes and intentions (Modig and Roseng-

ren 2013; Smith, Chen, and Yang 2008) or memory effects

(Pieters, Warlop, and Wedel 2002; Till and Baack 2005).

Results also varied regarding whether attitude and memory

outcomes are affected (Baack, Wilson, and Till 2008). In line

with the overall literature, we hypothesize that advertising

creativity has positive effects on immediate and outcome

responses:

H1: Advertising creativity has positive effects on (a) imme-

diate responses and (b) outcome responses.

However, from a managerial perspective, understanding

whether investments in advertising creativity mainly affect ad

rather than brand response, and whether advertising creativity

is better at stimulating attitude or memory outcomes, is also

important. Given the inconsistencies in the literature, we qua-

lify this hypothesis by asking what type of consumer responses

are most affected:

RQ1: Is consumer response stronger for ad versus brand

outcomes?

RQ2: Is consumer response stronger for memory versus

attitude outcomes?

Defining and Measuring Advertising
Creativity

As indicated in the introduction, a key challenge in the litera-

ture is the different views on advertising creativity. Creativity

is a general construct that has been widely researched in fields

such as psychology and organizational behavior, as well as in

marketing (Im and Workman 2004; Sasser and Koslow 2008).

Creativity can be used to describe individuals (e.g., an art

director at an advertising agency), processes (e.g., design think-

ing methods used to brainstorm advertising campaigns), or out-

puts (e.g., the actual ad executions used in a marketing

campaign). In this article, we adopt the output perspective.

Drawing on the creativity literature (Amabile 1996; Hen-

nessey and Amabile 2010; Runco and Jaeger 2012), we define

advertising creativity as advertising execution(s) that is (are)

original and appropriate. This bipartite definition of creativity

has been widely adapted in the marketing literature, in which

the definition has been applied in advertising (Chen, Yang, and

Smith 2016; Kilgour, Sasser, and Koslow 2013), new product

development (Burroughs et al. 2011; Im and Workman 2004),

and consumer behavior (Burroughs and Mick 2004; Moreau

and Engeset 2016). As argued by Amabile (1996), a bipartite

definition of creativity is required, because outputs that are

original or unique but carry no use or meaning are perceived

as weird or bizarre. However, any judgments of creativity are

subjective and likely to vary across context and time. For

example, judgments about originality and appropriateness in

a work of art differ from the same judgments in an advertising

context (even if the actual object being judged is the same).

Similarly, judgments of the creativity of the same object can

vary over time. In the art domain, for example, there are several

artists who are now considered creative but whose art was

controversial or even rejected in their time (e.g., Monet,

Picasso, Dalı́). Such works were initially seen as “weird” or

“bizarre,” mainly because the type of appropriateness expected

of them at the time of creation was a literal representation of

reality. Thus, these artists were redefined as creative later,

when judgment of appropriateness changed.

In the advertising context, the best documented dimension

of creativity is originality. This dimension has also been

referred to as novelty, divergence, unexpectedness, and new-

ness (Kim, Han, and Yoon 2010; Koslow, Sasser, and Riordan

2003; Sheinin, Varki, and Ashley 2012; Smith et al. 2007).

Originality has positive effects on consumer responses to

advertising, as originality makes advertising more likely to

be attended and processed (Pieters, Warlop, and Wedel 2002;

Smith, Chen, and Yang 2008). Originality also has a positive

effect on consumer response, as people have a predisposition to

appreciate divergent stimuli and deem them intrinsically inter-

esting (Yang and Smith 2009). Advertising practitioners typi-

cally view originality as the most defining aspect of advertising

creativity (Koslow, Sasser, and Riordan 2003; Modig and Dah-

len 2019), especially when it comes to advertising creativity

awards (Choi et al. 2018; Smith et al. 2007). Thus, it is not

surprising that many scholars focus primarily or exclusively on

originality when assessing advertising creativity (Krishen and

Homer 2012; Pieters, Warlop, and Wedel 2002).

When it comes to appropriateness, this dimension comple-

ments originality by connecting the advertisement with brand

strategy and consumer problem-solving abilities and goals

(Ang, Lee, and Leong 2007; El-Murad and West 2004; Modig

and Dahlen 2019; Smith et al. 2007). In the advertising litera-

ture, appropriateness is also referred to as relevance and use-

fulness (and when assessed by practitioners as “on strategy”;

cf. Kilgour, Sasser, and Koslow 2013; Sasser and Koslow

2008). Appropriateness as such has received much research

attention (often using the term “relevance”; e.g., Hayes et al.

2020). However, in contrast with originality, scholars rarely

consider appropriateness to be an indicator of creativity in and

of itself. Instead, researchers typically view appropriateness as

a prerequisite for advertising to be interesting to its intended

audience regardless of its level of creativity.

Rosengren et al. 41

Theoretically, it is clear that creativity is both originality and

appropriateness. Some scholars also argue that additional

dimensions could be needed to fully understand advertising

creativity (Ang, Lee, and Leong 2007; Haberland and Dacin

1992). They argue in favor of including a third advertising-

specific dimension of creativity—namely, the quality of the

ad execution, also referred to as artistry or production (Modig

and Dahlen 2019; Smith et al. 2007). In the literature, we

distinguish four approaches to empirically assess advertising

creativity. First, some studies measure advertising creativity as

a holistic perception of the creativity of an ad, typically by

using a single item “creative” or multiple creativity items that

do not refer specifically to different dimensions (e.g., Roseng-

ren, Dahlen, and Modig 2013). Second, other studies rely on

only one dimension of advertising creativity (typically origin-

ality; e.g., Maniu and Zaharie 2014). Third, acknowledging the

bipartite definition of creativity, some researchers use the inter-

action between originality and appropriateness as a creativity

measure (e.g., Smith et al. 2007). Fourth, acknowledging the

multidimensionality of the bipartite definition, some studies

rely on measures of both originality and appropriateness (Kim,

Han, and Yoon 2010), sometimes combined with one or more

additional dimensions related to the production quality or artis-

tic value (Modig and Dahlen 2019; Reinartz and Saffert 2013).

We argue that researchers who focus on originality only

(e.g., Maniu and Zaharie 2014; Pieters, Warlop, and Wedel

2002) are likely to get different results from those who study

creativity in terms of a bipartite (e.g., Modig and Dahlen 2019;

Smith et al. 2007). Although we cannot test the validity of

different assessments, we propose that the best measure of

advertising creativity should explain more variance in outcome

response variables than alternative measures of creativity, lead-

ing to stronger effect sizes (for a similar argument, see Eisend

2015). The approach that has the highest explanatory value

should also be the most managerially relevant. Given that crea-

tivity is defined as a combination of originality and appropri-

ateness, we propose that the effect sizes should be stronger

when both dimensions are considered and weaker when only

originality is used. Thus,

H2: The effect of advertising creativity on (a) ad response

and (b) brand response is stronger when creativity is

assessed as a bipartite versus as originality only.

When Advertising Creativity Works:
Contextual Moderators

Although we expect advertising creativity to generally have

positive effects on consumer response (H1), we also expect

properties of the communication context to moderate these

effects. In the literature review, it was apparent that little atten-

tion has been paid to such contextual moderators in the existing

literature (Yang and Smith 2009). However, we identified two

theoretically relevant moderators: involvement and familiarity.

Both variables have been found to affect consumer response to

advertising in general, but of interest here is how they affect

consumer responses to advertising creativity. Specifically, the

literature suggests that advertising creativity (i.e., a combina-

tion of originality and appropriateness) has benefits regardless

of the type of processing (peripheral vs. central; e.g., Cacioppo

and Petty 1984) depending on these moderators.

Involvement

Consumers’ involvement with advertising reflects their level of

interest in brand evaluation in any given context and has been

found to moderate the effects of advertising processing and

response (e.g., MacInnis and Jaworski 1989; Meyers-Levy and

Malaviya 1999). Specifically, consumer responses to advertis-

ing differ depending on how much effort goes into processing

it. For example, high involvement with a product category

motivates consumers to pay attention to and actively process

advertising. When involvement is low, attention is typically

allocated to other things, and consequently, ad processing is

limited and utilizes few processing resources and peripheral

cues (e.g., Cacioppo and Petty 1984; MacInnis and Jaworski

1989).

Although advertising creativity has typically been thought

of as an attention-grabbing device (e.g., Pieters, Warlop, and

Wedel 2002), implying that it would work best in low-

involvement contexts (where it can foster situational involve-

ment; Cacioppo and Petty 1984), creativity has been found to

have additional processing advantages when it comes to high-

involvement contexts (Smith and Yang 2004; Yang and Smith

2009). In a low-involvement context, any additional processing

stimulated by a creative ad is likely to be shallow (MacInnis

and Jaworski 1989; Yang and Smith 2009). In a high-

involvement context, however, additional processing makes

consumers more likely to actively assess the claims. In this

context, the combination of originality and appropriateness

fosters more open-minded and less defensive processing of

claims made (“willingness to delay closure”; Kardes et al.

2004; Yang and Smith 2009). This means that consumers will

be more open to new information about the brand and less

likely to use defensive mechanisms when processing advertis-

ing messages that are communicated creatively. Whereas

advertising creativity in low-involvement contexts stimulates

more affective processing, in high-involvement contexts crea-

tivity influences affective and cognitive processing (Yang and

Smith 2009). In both contexts, advertising creativity should

have a positive impact on consumer response, but in a high-

involvement context, in-depth processing, coupled with the

willingness to delay closure, makes the effects stronger. Thus,

H3: The effect of advertising creativity on (a) ad response

and (b) brand response is stronger for high-involvement

versus low-involvement products.

Familiarity

Familiarity reflects the extent of consumers’ direct and indirect

experience with a stimulus, such as a product or a brand (Alba

42 Journal of Marketing 84(6)

and Hutchinson 1987; Campbell and Keller 2003). Consumer

response to advertising has been found to vary with familiarity

(Machleit, Allen, and Madden 1993; Sethuraman, Tellis, and

Briesch 2011). Specifically, the effects of advertising are gen-

erally stronger for unfamiliar than familiar brands. This effect

is due to consumers not being able to draw from previous

experiences (neither their own nor the experiences of others)

when evaluating unfamiliar brands, which makes advertising

more important for these brands. However, advertising for

unfamiliar brands wears out faster (Campbell and Keller

2003). For marketers of unfamiliar brands, this poses a chal-

lenge, as they need advertising to build familiarity but also

must be careful about how they advertise to avoid negative

reactions caused by (too much) repetition.

Familiarity has also been found to moderate the effects of

advertising creativity in terms of familiarity with the specific

ad (Chen, Yang, and Smith 2016; Pieters, Warlop, and Wedel

2002). Chen, Yang, and Smith (2016) found that advertising

creativity has two main benefits when it comes to repetition:

(1) it generates more positive effects upon initial exposure,

and (2) it resists wear-out over multiple exposures. The latter

finding is in line with results showing that advertising crea-

tivity (in terms of originality) helps draw attention to familiar

ads that might otherwise be overlooked due to tedium (Pieters,

Warlop, and Wedel 2002). For unfamiliar brands, these

advantages are more important (Campbell and Keller 2003).

Overall, this suggests that advertising creativity should be

beneficial for unfamiliar and familiar brands, but given its

attention-grabbing character (Pieters, Warlop, and Wedel

2002), the immediate wear-in effect that it can generate

(Chen, Yang, and Smith 2016), and the higher impact of

advertising in general (Machleit, Allen, and Madden 1993;

Sethuraman, Tellis, and Briesch 2011), the effects should be

stronger for unfamiliar brands. Thus,

H4: The effect of advertising creativity on (a) ad response

and (b) brand response is stronger for unfamiliar versus

familiar brands.

How Advertising Creativity Works:
Mediators

In the literature, scholars have used three main theories to

explain the effects of advertising creativity on consumer

responses. These accounts focus on different immediate

responses as key mediators of the effects of advertising crea-

tivity on outcome responses. The affect transfer model

focuses on the potential of creativity to evoke positive feel-

ings that spill over into consumer responses to the ad and

brand (i.e., regarding “positive affect” as a key mediator;

Yang and Smith 2009). The processing model focuses primar-

ily on the ability of creativity to get consumers interested in

the ad and brand (i.e., “ad processing” is the key mediator;

Smith et al. 2007). The signaling model proposes that crea-

tivity works as a marketing signal, directly influencing per-

ceptions about the sender and thus, consumer responses to the

brand (i.e., “perceived sender effort” is the key mediator;

Dahlen, Rosengren, and Törn 2008).
1

Affect Transfer Model

A common explanation for the effects of advertising creativity

is based on affect transfer (De Houwer, Thomas, and Bayens

2001; also referred to as affect infusion; Forgas 1995). This

explanation focuses on the ability of affectively loaded infor-

mation to transfer into other, more or less unrelated, targets. In

the advertising creativity context, the affect transfer model

builds on the fact that consumers generally like creative ads

(Rosengren, Dahlen, and Modig 2013; Smith, Chen, and Yang

2008). Processing creative ads is seen as intrinsically motivat-

ing and pleasurable (Rosengren, Dahlen, and Modig 2013;

Yang and Smith 2009), which means that consumers are likely

to experience positive affect while exposed to such advertising.

This positive affect spills over to the ad and brand, leading

them to be evaluated more favorably (Haberland and Dacin

1992; Yang and Smith 2009). According to this explanation,

the positive effects are driven by creative ads being more

enjoyable and liked, and the positive feelings that this stimu-

lates. Although theoretically exceptions might occur, as may be

the case for executions of fear appeal advertising that combines

originality and appropriateness, the reviewed literature on

advertising creativity unanimously provides empirical support

for positive reactions to advertising creativity. Thus,

H5: The effect of advertising creativity on (a) ad response

and (b) brand response is mediated by positive affect.

Processing Model

Explanations of how advertising works often draw on informa-

tion processing models (e.g., MacInnis and Jaworski 1989;

Meyers-Levy and Malaviya 1999). Specifically, they explain

consumer responses to advertising based on different levels of

ad processing. This is also a common explanation for the

effects of advertising creativity. Creativity is said to stand out,

thus making creative ads more likely to be attended to and

processed (Smith et al. 2007; Yang and Smith 2009). This

means that more creative advertising stimulates more ad pro-

cessing, resulting in longer exposure and greater attention

(Haberland and Dacin 1992; Pieters, Warlop, and Wedel

2002), which has positive effects on consumer outcome

response. According to this explanation, the positive effects

of creativity are driven by creativity being more interesting and

therefore, processed more. Thus,

1
The models we use in this article focus on key variables that are discussed in

the extant literature of the three theoretical accounts. We could not include

additional variables presented in Table 1 because of data constraints, which we

explain in detail in the “Methods” section. It should also be noted that all three

accounts have primarily been developed using experimental approaches.

Rosengren et al. 43

H6: The effect of advertising creativity on (a) ad response

and (b) brand response is mediated by ad processing.

Signaling Model

A third explanation for the effects of advertising creativity

focuses on creativity as a signal of brand and company

ability (e.g., Dahlen, Rosengren, and Törn 2008; Lange,

Rosengren, and Blom 2016). This model builds on research

on marketing signals (Kirmani 1997; Kirmani and Rao

2000), showing that certain behaviors on behalf of a firm

(e.g., offering long-lasting warranties) can be used to sig-

nal unobservable quality to consumers. For example,

advertising spending (i.e., monetary investments) has been

found to work as a signal of brand quality (Kirmani 1990;

Kirmani and Wright 1989). Similarly, advertising creativ-

ity has been found to be perceived by consumers as a

signal that the sender has invested effort (in terms of time

and money) in their brand (Dahlen, Rosengren, and Kars-

berg 2018; Dahlen, Rosengren, and Törn 2008). Through

creative advertising, a brand conveys that it is committed

to its advertising and its products, which is interpreted as a

signal that positively affects how the brand is perceived

and evaluated. According to this explanation, the positive

effects of creativity are driven by creativity signaling

effort on behalf of the sender, thus affecting the ad and

brand positively. In contrast to the other two models, this

account considers process reactions to creative advertising

to be about the immediate perceptions of the brand rather

than the ad. Thus,

H7: The effect of advertising creativity on (a) ad response

and (b) brand response is mediated by perceived sender

effort.

Full Model

Although the three theoretical accounts typically have been

used in isolation (for an exception, see Yang and Smith

[2009]), combining the three models should provide a more

comprehensive account of how advertising creativity works.

Furthermore, the omission of any one of the intermediary

effects might lead to the overestimation of the other (Vakratsas

and Ambler 1999). Therefore,

H8: A model of advertising creativity that considers (a)

positive affect, (b) ad processing, and (c) perceived sender

effort jointly better explains the effects of advertising crea-

tivity on ad and brand responses than any of the three mod-

els used separately.

To test this hypothesis, we propose a full model, which incor-

porates all three theoretical mechanisms (see Figure 2). Given

that the initial models focus on different variables, we propose

several additional relationships in the full model. First, ad pro-

cessing is likely to spur stronger perceptions of sender effort.

This is because processing facilitates a more careful under-

standing of the ad (Smith, Chen, and Yang 2008), which

should, in turn, enhance perceptions of sender effort stimulated

by advertising creativity. Second, affective responses can influ-

ence not only the ad and brand but also perceptions of sender

Perceived
sender
effort

Advertising
creativity

Ad
processing

Positive
affect

Attitude
toward

advertisement

Attitude
toward
brand

.293***

.337***

.341***

.373***

.388***

.051***

.351***

.234***

.462***
.304***

.078†

.206*** .085**

.088**

Figure 2. How advertising creativity works: Full model including estimation results (standardized path coefficients).
yp < .10.

*p < .05. **p < .01. ***p < .001.

Notes: Dotted lines indicate paths that were added to the combined model to form the full model.

44 Journal of Marketing 84(6)

effort. The underlying logic is, again, affect transfer (Yang and

Smith 2009). Furthermore, affect and processing should be

positively related, because feelings ease processing (mood the-

ory), and easy processing is experienced as a good feeling

(processing fluency theory).

Method

Data Set

For this meta-analysis, we selected papers that provide estimates

of the effects of advertising creativity on various consumer

responses. According to our bipartite definition of advertising

creativity, advertising creativity comprises originality and appro-

priateness. To be able to assess the relevance of different assess-

ments of creativity, we included all studies that identify as “ad*

creativity” studies independent of their definition and operationa-

lization of advertising creativity. This means that we also included

all studies that relied on advertising stimuli judged to be creative

(even if they did not use the bipartite definition), as well as studies

that investigated the two main dimensions of creativity, even if

they did not use the term “creativity” (Lee and Mason 1999).

To identify relevant papers, we first referred to review arti-

cles that provide an overview of previous research on advertis-

ing creativity (e.g., Sasser and Koslow 2008). We applied an

ancestry tree search by searching all papers that refer to the

review papers that were available in the Web of Science data-

base. Second, we performed a keyword search of electronic

databases (e.g., ABI/INFORM, Emerald, Elsevier, EBSCO,

and ProQuest Dissertation Publishing) using “advertising

creativity,” “ad creativity,” “advertisement creativity,” and

“advertising creative,” “ad creative,” and “advertisement

creative” as key words, followed by a search with key words

that relate to the two main dimensions of advertising creativity

(“original*,” “novel*,” “newn*,” “unexpected*,” “divergen*,”

“innovati*,” “incongru*,” “relevan*,” “appropriate*,”

“useful*,” and “meaningful*” combined with “advertis*”). The

database search was complemented by a search on Google

Scholar. Third, we performed a manual search of the journal

outlets that turned out to be major sources for articles on adver-

tising creativity. Fourth, once we identified a paper, we exam-

ined the references in a search for additional studies. The

search period covered all papers (published and unpublished)

that were available by December 2018. The retrieval approach

was consistent with recommendations in the literature (Hunter

and Schmidt 2004) and closely followed the steps taken in

recent meta-analyses published in marketing (Roschk and Hos-

seinpour 2020; Zlatevska, Dubelaar, and Holden 2014).

After identifying manuscripts for potential inclusion in the

data set, we applied inclusion and exclusion criteria to deter-

mine which manuscripts to retain. We included all empirical

studies that measured or manipulated advertising creativity (as

described previously) and provided estimates on its effects on

consumer responses. We excluded any manuscripts outside this

scope. For instance, we excluded studies that investigated non-

consumer response to creative ads (e.g., advertisers; Wang

et al. 2013), or studies on creative media choice, but not crea-

tive ads (Dahlen, Friberg, and Nilsson 2009). We also excluded

studies that failed to provide sufficient data for the meta-

analysis and for which necessary data could not be retrieved

from the authors.

To avoid duplications in the data set, a document with orig-

inal analyses and findings by the same authors (e.g., journal

article, working paper, conference paper) is called a “paper.”

In some papers, the authors analyzed more than one distinct data

set (e.g., a paper with several experiments), while some data sets

were analyzed in more than one paper (e.g., a study that was

published as a conference paper and a journal paper). The anal-

ysis is based on data sets. Each data set can provide single or

multiple effect sizes that refer to the effect of advertising crea-

tivity on any consumer response variable. The search resulted in

67 usable papers covering 93 data sets (see Web Appendix

Table 1). The sample includes journal articles, book chapters,

working papers, unpublished theses, and conference proceed-

ings, thus reducing the risk of a biased representation of the state

of research because of the source of publication. The variation of

sources is similar, and the number of papers and data sets is

higher than in other major meta-analyses in marketing (Chang

and Taylor 2015; You, Vaddkkepatt, and Joshi 2015).

Coding

We categorized the consumer response variables measured in

the studies based on the conceptual framework (see Figure 1).

Specifically, we classified consumer responses in terms of

immediate responses (affect, processing, and signals) and last-

ing outcomes related to the ad and brand (none of the data sets

provided data for sales). The outcome responses were further

divided based on attitude and memory responses. In addition,

we identified a few consumer responses that did not fit in either

category (e.g., actual creativity, brand familiarity, willingness

to pay). Because these consumer response variables appeared

either in only one or two data sets or in only one paper, we

eliminated them from further analysis.
2

We did this to ensure a

minimum degree of generalizability, because a meta-analysis

should provide a high degree of generalization and thus,

requires more information than a single manuscript or a

single-study manuscript followed by a replication study. This

left 878 effect sizes. For an overview of the consumer response

variables and categorization scheme, see Table 1.

In terms of creativity moderators, we coded the variables at

the effect size level, meaning that multiple effect sizes from

one data set can be assigned different codes. Specifically, we

coded whether creativity was assessed as originality only, as

appropriateness only, as an interaction effect between origin-

ality and appropriateness, as a multidimensional concept

2
We excluded the following variables from further analysis (mean correlations

in parentheses): negative thoughts (�.105, p < .01), other thoughts (.047, n.s.), negative feelings (.083, n.s.), actual creativity (.189, p < .01), brand familiarity (.192, p < .10), presumed influence (.309, p < .01), and willingness to pay (.429, p < .10).

Rosengren et al. 45

(including originality, appropriateness, and potentially more

dimensions), or as a holistic concept (measured with a single

item “creative” or corresponding multiple items or manipulated

as such; this is the base alternative in the model). As an illustra-

tion, Yang and Smith (2009) presented results based on origin-

ality and appropriateness separately, as well as for the interaction

between the two allowing us to code three types of measure-

ments for each of the variables studied. Although our main

interest is in comparing a bipartite view of advertising creativity

with a view of advertising creativity as originality only, this

coding process allows a more complete understanding of how

different assessments of creativity affect consumer response.

In terms of communication context moderators, we dummy

coded the variables on the data-set level. Specifically, we

coded the data sets 1 if the advertised category was a high-

involvement product and if the advertised brand was familiar.

In addition, we added three control variables that captured

substantial differences between studies and that could be

related to the context variables (medium, year, and award).

Two authors independently assigned variables in the primary

studies to consumer responses and coded the moderators and

control variables based on the information available in each

study. The agreement rate was above 98% (Krippendorff’s
alpha ¼ .932), and inconsistencies were resolved by discussion.

Effect Size Computation

The effect size metric selected for the meta-analysis was the

correlation coefficient; higher absolute values of the coefficient

indicate a stronger influence of advertising creativity on con-

sumer responses. For papers that reported other measures (e.g.,

Student’s t, mean differences), we converted those measures

following guidelines for meta-analysis (Lipsey and Wilson

2001; Peterson and Brown 2005).
3

We adjusted all correlations

for unreliability. When a paper did not report the reliability, or

when the paper used a single-item measure, we used the mean

reliability for that construct across all studies, following the

procedure in previous meta-analyses in marketing (e.g., Kirca,

Jayachandran, and Bearden 2005).

We dealt with integrating dependencies between effect sizes

using the following approach. When a data set provided

Table 1. Variables Used in the Meta-Analysis.

Variable Description Coding Scheme (Reliability)

Immediate Responses Consumer responses at the time of
exposure in terms of…

Affect Emotions Positive affect, humor
Processing Processing Attention, interest in ad, ad processing, complexity of ad/difficult to

comprehend, positive thoughts
Signals Perceptions Perceived sender effort, perceived brand value/quality, perceived trust,

perceived credibility
Outcome Responses Lasting consumer responses in terms of…

Ad response Attitude Aad
Memory Ad recall, ad recognition

Brand response Attitude Abrand, purchase/behavioral intention
Memory Brand recall, brand recognition, brand memory (mix recall/recognition)

Sales response Brand/product sales N.A.
Moderators
Creativitya Creativity assessed in terms of…

Originality only 0 ¼ other, 1 ¼ originality only
Appropriateness only 0 ¼ other, 1 ¼ appropriateness only
Originality � appropriateness 0 ¼ other, 1 ¼ interaction only
Multidimensional measure 0 ¼ other, 1 ¼ multidimensional

Familiarity Degree of brand familiarity 0 ¼ unfamiliar/fictitious/mixed, 1 ¼ familiar (AR ¼ 97.3%,
a ¼ .938)

Involvement Degree of product involvement 0 ¼ low involvement/mixed, 1 ¼ high involvement (AR ¼ 94.5%, a ¼
.786)

Control Variables
Medium Type of medium used to convey ad 0 ¼ print/outdoor, 1 ¼ TV/movies
Year Year of publication Continuous
Award Whether the studied ad has won a

creative award
0 ¼ other, 1 ¼ award winning (AR ¼ 94.5%, a ¼ .888)

aThe moderator variable is measured at the effect-size level, while all other moderators are measured at the data-set level.
Notes: Intercoder reliability is provided for all high-inference coding with AR ¼ agreement rate and a ¼ Krippendorff’s alpha.

3
Of 878 effect sizes, we converted 21 from coefficients in multivariate

regressions via the formula suggested by Peterson and Brown (2005). These

parameters were partial correlations, and therefore, we checked whether they

had an influence on the meta-regression results by including a dummy variable

that distinguishes between partial correlations and correlations. Because partial

correlations did not appear in the set of correlations referring to Abrand, the

dummy was included in the Aad model only.

46 Journal of Marketing 84(6)

findings for different consumer response variables, we treated

the findings as independent, because we integrated and analyzed

the estimates for each consumer response variable separately.

Some data sets reported multiple relevant tests for the same

consumer response variable. We accounted for the dependencies

of the effect sizes and the nested structure of the meta-analytic

data by using a mixed-effects multilevel model (Raudenbush

and Bryk 2002). We estimated the following model:

r ij ¼ g 00 þ m 0j þ e ij; ð1Þ

where i ¼ 1, 2, 3 . . . I effect sizes, j ¼ 1, 2, 3 . . . J data sets. This
formula estimates the average effect size g00, the deviation of
the average effect size in a data set from g00 (m0j), and the
deviation of each effect size in the kth data set from g00 (eij).
The latter two terms have variances that follow a normal dis-

tribution and are uncorrelated.

To address publication bias, we computed fail-safe Ns

(Rosenthal 1979), which represents the number of additional

effects with null results needed to render the results for an inte-

grated effect size not statistically significant at p ¼ .05. The fail-
safe Ns were calculated for all statistically significant integrated

effect sizes (p < .05) using the effect size estimates that were adjusted for measurement error. Furthermore, we provided a

homogeneity test as an aid in deciding whether the observed

effect sizes were more variable than would be expected from

sampling error alone. If they are, there is a strong basis for

including moderators. The homogeneity test involves the Q sta-

tistic, in which the distribution is similar to a chi-square with

K � 1 degrees of freedom (Hedges and Olkin 1985).

Moderator Analysis

If the homogeneity test indicated heterogeneity, we proceeded

with a moderator analysis. We added the moderators specified

by the hypotheses and the control variables simultaneously to

Equation 1 and ran multilevel meta-regression models in hier-

archical linear modeling separately for the major outcome vari-

ables. The model was a mixed-effects model, because fixed

effects for the moderators were considered in addition to ran-

dom components. We specified the following model:

r ij ¼ g 00 þ g 01 � involvement j
� �

þ g 02 � familiarity j
� �

þg 04 � medium j
� �

þ g 05 � year j
� �

þ g 06

� award j
� �

þ g 10 � originality ij
� �

þ g 20

� appropriateness ij
� �

þ g 30 � interaction ij
� �

þg 40 � multidimensionality ij
� �

þ g 50

� partial correlation ij
� �

þ u 0j þ e ij;

ð2Þ
where rij is the ith effect size describing the relationship

between advertising creativity and the respective consequence

variable reported within the jth data set.

Assuring the robustness of the model required a sufficient

sample size. The major restriction is often the higher-level

sample size, and the literature recommends a sample of around

50 to avoid biased estimates of the second-level standard errors

(Maas and Hox 2005). Thus, we applied the model only to the

outcome variables in the data set with a sufficiently large sam-

ple of data sets: Aad and Abrand (43 and 44 data sets,

respectively).

Structural Model Estimation

To investigate the different processes that explain how adver-

tising creativity works, we developed a correlation matrix

including integrated effect sizes of the consumer responses

to advertising creativity and added integrated effect sizes for

the interrelationships between the consumer response vari-

ables. We followed recommendations in the literature about

collecting meta-analytic data for the correlation matrix,

deciding about sample size, analytical decisions, and report-

ing (Bergh et al. 2016). We searched the papers in the meta-

analysis for correlations for the interrelationships between

consumer response variables. For a construct to be included

in such analysis, multiple study effects must relate it to every

other construct in the model. Therefore, no additional vari-

ables shown in Table 1 could be considered. For example,

because we did not find correlations between sender effort

and recall or memory measures, the latter could not be

included in the model. We found at least three correlations

for each relationship, which equals or exceeds the require-

ments of other meta-analytic correlation matrices found in the

literature (Geyskens, Steenkamp, and Kumar 1999). We inte-

grated and adjusted the correlations in the same way as the

correlations between advertising creativity and consumer

response variables. That is, we first adjusted all correlations

for unreliability. We accounted for the dependencies of effect

sizes and the nested structure of meta-analytic data by using a

mixed-effects multilevel model as described previously (Rau-

denbush and Bryk 2002).

We then used this correlation matrix (see Web Appendix

Table 2) as input in a structural equation modeling (SEM)

analysis using the maximum likelihood method. The matrix

was based on 449 correlations, and the harmonic mean of the

cumulative sample size for each cell equaled 1,293. Each con-

struct was measured with a single indicator in the structural

model. We fixed the error variances for these indicators to zero

because we had already considered measurement errors when

we integrated the effect sizes. We used the harmonic mean of

the cumulative sample size underlying each integrated effect

size (i.e., effect size cells comprising each entry in the correla-

tion matrix) as the sample size for the analysis.

Results

Table 2 reports the integration of the reliability-corrected cor-

relations between advertising creativity and all consumer

response variables.

Rosengren et al. 47

Looking at immediate responses, we found statistically sig-

nificant effects on affect in terms of positive affect and per-

ceived humor. Interestingly, although positive affect has been

studied more, the effects of humor were significantly stronger

as indicated by nonoverlapping confidence intervals (95% CI
for positive affect [.198, .388] vs. humor [.428, .832]). We also

found significant positive effects on processing in terms of

attention, interest in the ad, and ad processing, but only a mar-

ginal effect on complexity and positive thoughts. The effects on

attention, interest in the ad, and ad processing were comparable

in size (95% CI for attention [.218, .592], interest in ad [.215,
.615], and ad processing [.015, .659]). Furthermore, advertising

creativity had statistically significant positive effects on per-

ceived signals: sender effort, brand value/quality, brand trust,

and brand credibility. These effects were comparable in terms

of size (95% CI for perceived sender effort [.282, .510], value/
quality [.171, .407], brand trust [.171, .603], and brand cred-

ibility [166, .628]).

Turning to outcome responses, advertising creativity had a

statistically significant effect on all ad responses: Aad, ad recall,

and ad recognition. The strongest and most widely studied

effect was that on Aad. In terms of brand responses, the effects

followed a similar pattern: Abrand was the most widely studied

variable and statistically significantly affected. We also found a

statistically significant positive effect on purchase/behavioral

intention and brand memory, but not on brand recall or brand

recognition. Overall, the pattern of results support H1 by high-

lighting that advertising creativity has positive effects on con-

sumer reactions in terms of ad and brand. Answering RQ1, we

found that the effect on Aad was statistically significantly larger

than that on Abrand and purchase intentions (95% CI for Aad
[.407, .575] vs. Abrand [.235, .399]) and purchase intention

[.225, .387]), and that the effects on ad recall ([.214, .408])

were significantly larger than the effects on brand memory

([.072, .208]). This suggests that ad responses are more affected

than brand responses. Related to RQ2, the pattern of results

Table 2. Influence of Advertising Creativity on Consumer Responses (H1).

# Papers
# Data

Sets
# Effect

Sizes
Total

Sample Size Average r
Homogeneity

Test Q Fail-Safe N

Immediate responses Affect
Positive affecta 6 10 32 2,610 .293*** 158.184*** 595
Perceived humor 4 4 10 1,208 .630*** 142.097*** 1,860

Processing
Attention 12 13 30 4,365 .405*** 2,853.202*** 20,410
Interest in ad 7 11 40 1,829 .415** 15,766.431*** 267,621
Ad processinga 3 4 6 1,037 .337* 822.686*** 429
Complexity of ad/difficult to

comprehend
4 6 15 2,357 �.217y 417.683*** —

Positive thoughts 2 3 57 743 .177y 23.194*** —
Signals

Perceived sender efforta 6 7 38 6,310 .396*** 1,042.326*** 1,333
Perceived brand value/quality 8 10 27 2,623 .289*** 724.790*** 9,735
Perceived brand trust 3 4 6 626 .387*** 81.835*** 220
Perceived credibility 5 7 8 2,138 .397*** 848.119*** 2,434

Outcome responses Ad Attitude
Aad

a
37 44 192 19,729 .491*** 23,134.086*** 1,446,838

Ad Memory
Ad recall 18 24 91 2,712 .311*** 3,575.699*** 30,142
Ad recognition 11 15 32 3,334 .252** 2.485.365*** 7,133

Brand Attitude
Abrand

a 35 43 138 11,434 .317*** 12,438.120*** 232,128
Purchase/behavioral intention 29 34 83 28,950 .306*** 3,214.985*** 122,938

Brand Memory
Brand recall 11 14 36 3,825 .129 793.672*** —
Brand recognition 8 10 21 2,148 .052 2,752.091*** —
Brand memory 4 5 16 742 .140* 8.860y 173

yp < .10. *p < .05. **p < .01. ***p < .001.

a
These variables are used to test H2–H7.

Notes: Only relationships for which effects were available in more than one paper and/or more than two independent data sets are shown. The corrected average
correlation coefficients (r) are the sample size-weighted, reliability-corrected estimates of the population correlation coefficients. The fail-safe N indicates the
number of nonsignificant, unpublished (or missing) effects that would need to be added to a meta-analysis to reduce an overall statistically significant (p < .05) observed result to nonsignificance.

48 Journal of Marketing 84(6)

suggests that effects of advertising creativity are stronger for

attitudes than for memory. Aad was statistically significantly

different from ad recognition (95% CI for Aad [.407, .575] vs.
ad recognition [.107, .307]), and marginally different from ad

recall ([.214, .408]). Similarly, the effect on Abrand was signif-

icantly stronger than the effect on brand memory (95% CI for
Abrand [.235, .399] vs. brand memory [.072, .208]).

All homogeneity tests (except for brand memory) were sta-

tistically significant at p < .05 and showed that the variation in effect sizes cannot be explained by sampling error alone. The

fail-safe N indicates that the statistically significant integrated

correlations do not suffer from publication bias according to

Rosenthal’s (1979) rule of thumb (fail-safe N should be at least

5 times the number of effects plus 10).

Table 3 presents the results for the multilevel moderator

regression model for the relationship between advertising crea-

tivity and Aad and Abrand. To investigate whether the positive

effects of advertising creativity depend on the type of assess-

ment used, we examined the moderating effect of creativity

assessments. The analysis showed that relying only on origin-

ality led to lower effect sizes for Aad and Abrand; thus, H2 was

supported. We found a similar pattern for assessments relying

on appropriateness only and for interaction effects, although

the negative effect was only marginally significant for the latter

when it came to Abrand. The findings also showed that multi-

dimensional measures of advertising creativity led to stronger

effect sizes for Abrand, but not for Aad. Overall, this pattern of

results suggests that assessing advertising creativity only in

terms of (1) originality, (2) appropriateness, or (3) an interac-

tion effect between the two will lead to an underestimation of

the effects. From a managerial perspective, the result also sug-

gests that a multidimensional view of advertising creativity is

the most relevant, as brand responses are more important than

ad responses.

We then turned to the moderating effect of the communica-

tion context. The results showed stronger effects on Aad and

Abrand for high-involvement products; thus, H3 was

supported.

Furthermore, the effects on Aad were marginally stronger for

unfamiliar products, but there was no statistically significant

difference in terms of Abrand. Thus, H4 was only partially sup-

ported. The control variables showed that using a partial cor-

relation coefficient led to smaller effects on Aad. None of the

remaining control variables affected Aad. However, the effects

on Abrand were higher for audiovisual media (TV/movies) and

marginally lower for award-winning ads. We did not find any

statistically significant differences in terms of year of study.
4

To better understand why advertising creativity has positive

effects on consumer responses, we performed a SEM analysis

of the different models using the meta-analytic correlation

matrix (cf. Web Appendix Table 2). Table 4 presents the results

of the SEMs (standardized coefficients and model fit statistics).

Table 3. Influence of Moderator Variables on Effect Sizes: Multivariate Meta-Regression Analysis Results (H2–H4).

Moderator (Hypothesis) Moderator Values

Aad Abrand

b (SE) Predicted b (SE) Predicted

Intercept .625 (.090)*** .308 (.082)***
Creativity (H2) Other vs. originality only �.202 (.052)** .564 vs. .362 �.170 (.074)* .334 vs. .164

Other vs. appropriateness only �.228 (.074)** .549 vs. .320 �.149 (.048)** .304 vs. .154
Other vs. interaction only �.270 (.089)** .510 vs. .240 –.120 (.068)y .275 vs. .156
Other vs. multidimensional .105 (.102) .231 (.013)*** .260 vs. .492

Involvement (H3) Low vs. high involvement .259 (.078)** .340 vs. .653 .223 (.093)* .196 vs. .420
Familiarity (H4) Unfamiliar vs. familiar �.142 (.080)þ .577 vs. .435 �.064 (.085)
Medium (Ctrl) Print/outdoors vs. TV/movies �.049 (.078) .182 (.081)* .206 vs. .388
Year (Ctrl) Continuous �.004 (.005) �.003 (.003)
Award (Ctrl) Others vs. award winning �.093 (.091) �.214 (.110)y .327 vs. .113
Partial correlation (Ctrl) Other vs. effect converted from

multivariate regression coefficient
�.523 (.052)*** .519 vs. �.004 —

y
p < .10.

*p < .05. **p < .01. ***p < .001.

4
To further explore the role of originality and appropriateness in explaining

these effects, we also tested several plausible interactions between creativity

measurements and moderators (similar to the procedure in Sethuraman,

Tellis, and Briesch 2011). For Aad, we tested interactions between

measurements (using dummy variables for originality, appropriateness, and

multidimensions, the cell sizes for interactions with “interaction only” were

too small to provide a robust analysis) and the hypothesized moderating

variables (involvement, familiarity). The analysis showed a significant

interaction effect for appropriateness � familiarity (b ¼ �.239, SE ¼ .062,
t ¼ 3.830, p < .001), suggesting that appropriateness is more important for unfamiliar brands. There was also a significant interaction effect for

appropriateness � involvement (b ¼ .271, SE ¼ .058, t ¼ 4.643, p < .001), suggesting that appropriateness is more important in high-involvement

contexts. We conducted the same analysis for Abrand, where we were able to

test interactions between measurements focusing on originality and

appropriateness (the cell sizes for “interaction only” and “multidimensional”

were too small to provide a robust analysis). However, we did not find any

statistically significant interactions.

Rosengren et al. 49

As we suggested alternative models implying that the relation-

ship between advertising creativity and Aad is mediated by

more than one mediating variable, we added a path between

advertising creativity and Aad that captured alternative pro-

cesses to each model. All three individual models showed a

very good model fit, and all paths were statistically significant

and in line with the suggested effects; thus, H5, H6, and H7 were

supported.

The model that combines the three individual models

showed a comparatively weak fit but was significantly

improved by adding the proposed relationships between pro-

cessing and perceived sender effort and positive affect

and perceived sender effort suggested by the full model

(Dw2/d.f. ¼ 96.527/5, p < .001; see Figure 2). To determine whether the full model provided a better explanation than the

three parsimonious models that were nested within it, we com-

pared the fit of the full model that was restricted to any of the

nested models with the fit of the full model with unrestricted

paths. The model fit worsened significantly when it was

restricted to the affect transfer model (Dw2/d.f. ¼ 1,629.935/
8, p < .001), the processing model (Dw2/d.f. ¼ 1,733.093/8, p < .001), or the signaling model (Dw2/d.f. ¼ 1,528.916/8, p < .001). Thus, the full model provides an explanation that goes

beyond the explanatory power of each nested model; H8 was

empirically supported. Interestingly, in the full model, the med-

iating effect of Aad on Abrand dropped from around .5 in the

individual models to a marginally significant effect of .078

(Dw2/d.f. ¼ 96.512/1, p < .001). This suggests that the effect of advertising creativity on brand response is only weakly

mediated by ad response, which adds additional insight into

RQ1 about the effects of creativity on ad versus brand response.

We performed two additional analyses to further explore

how well the three models explain the effects of creativity on

consumer response. First, we compared how much of the var-

iance in Aad was explained directly by advertising creativity

and indirectly by either process suggested by the three individ-

ual models (we could not apply this comparison to Abrand, as

there was no direct effect of creativity on Abrand in the model).

We computed the proportion of mediation as the ratio of indi-

rect to total effect; that is, the indirect path(s) was/were divided

by the sum of the direct path and indirect path(s) (Iacobucci,

Saldanha, and Deng 2007). The proportion of mediation via

positive affect was 26.8%, via ad processing was 28.3%, and
via perceived sender effort was 33.9%. When we tested the
mediation paths in the full model against each other by restrict-

ing two corresponding paths at a time (see Web Appendix

Table 3), we found no differences between the paths from

advertising creativity to any of the three mediators (positive

affect, ad processing, and sender effort). However, the effect of

sender effort on Aad was significantly different and stronger than

the effect of either positive affect or ad processing on Aad. The

findings indicate that signaling explains more variance in Aad
than the two other models, thus providing the strongest explana-

tion for the effect of creativity on Aad of the three models.

Table 4. Coefficients and Fit Indices of the Meta-Analytic SEMs (H5–H8).

Affect Transfer
Model (H5)

Processing
Model (H6)

Signaling
Model (H7)

Combined
Model Full Model (H8)

Creativity ! Positive affect .293*** .293*** .293***
Creativity ! Ad processing .337*** .337*** .337***
Creativity ! Perceived sender effort .396*** .396*** .341***
Creativity ! Attitude toward the ad .315*** .301*** .250*** .051** .051***
Positive affect ! Perceived sender effort .085**
Positive affect ! Attitude toward the ad .515*** .388*** .388***
Positive affect ! Attitude toward the brand .266*** .373*** .373***
Positive affect +! Ad processing .206***
Ad processing ! Attitude toward the ad .490*** .351*** .351***
Ad processing ! Attitude toward the brand .131*** .234*** .234***
Ad processing ! Perceived sender effort .088**
Perceived sender effort ! Attitude toward the ad .546*** .462*** .462***
Perceived sender effort ! Attitude toward the brand .128*** .304*** .304***
Attitude toward the ad ! Attitude toward the brand .477*** .562*** .556*** .078y .078y
Model Statistics
w2/d.f. 1.840/1 1.067/1 .655/1 97.903/4*** 1.376/1
Goodness-of-fit index .999 1.000 1.000 .975 1.000
Adjusted goodness-of-fit index .993 .996 .997 .867 .993
Comparative fit index 1.000 1.000 1.000 .973 1.000
Root mean square residual .008 .006 .005 .085 .004
Root mean square error of approximation .025 .007 .000 .134 .017

yp < .10. *p < .05. **p < .01. ***p < .001.

50 Journal of Marketing 84(6)

Second, we compared the theoretical explanation offered by

the full model between the two dimensions of creativity by using

correlation matrices that considered the variable relationships

with either originality or appropriateness instead of creativity

(see Web Appendix Table 4). The results showed that the pos-

itive effects on ad processing are equally strong for both dimen-

sions. However, affect transfer mainly explains the effects of

originality as indicated by the fact that the path from creativity

to positive affect was statistically significant for originality, but

not for appropriateness. When it comes to signaling, however,

appropriateness seems more important, as indicated by the sig-

nificantly stronger link between creativity and sender effort.

Discussion

Summary of Findings

In this article, we offer a comprehensive synthesis of the effects

of advertising creativity on consumer responses. The study

highlights the importance of advertising creativity by showing

robust positive effects on a wide range of immediate and out-

come responses. The effects are stronger for ad responses com-

pared with brand responses and for attitudinal compared with

memory outcomes. Moderation analyses show that the effects

of advertising creativity are weaker when creativity is assessed

as originality only, compared with a bipartite comprising ori-

ginality and appropriateness. This suggests that the effects of

advertising creativity go beyond those of originality alone. The

results further show that advertising creativity has stronger

effects in high-involvement contexts, and that effects on ad

response are (marginally) stronger for unfamiliar brands.

Furthermore, we find empirical support for all three theoretical

accounts (affect transfer, processing, and signaling) used in the

literature, but also that a full model (where the three accounts

are considered jointly) best explains the effects of advertising

creativity on consumer outcome response. In the full model, the

effect of the three advertising creativity mediators (positive

affect, ad processing, and perceived sender effort) on brand

response is only marginally mediated by ad response, suggest-

ing that although ad responses are generally more affected than

brand responses, they are not needed for advertising creativity

affect brand response. Additional analyses show that affect

transfer mainly explains the effects caused by originality and

that signaling provides the strongest account of advertising

creativity in terms of ad response.

Theoretical Implications

Although marketing researchers and practitioners tend to agree

that advertising creativity is important, there are contrasting

views on what advertising creativity is, and how and when it

can lead to positive outcomes. Through this meta-analysis, we

provide a synthesis of the growing, but dispersed, literature on

advertising creativity, thus building a common foundation for

future studies of this important topic. The results inform about

important outcome variables and moderators of advertising

creativity effects. The meta-analytic findings can serve as

benchmarks for future advertising creativity studies, as well

as for studies dealing with other ad execution elements. Future

findings can be compared against the meta-analytic results in

terms of explained variance as a measure of advertising effec-

tiveness. The results also have several implications for future

studies of advertising creativity.

First, we offer an empirically validated understanding of

how advertising creativity works. The pattern of results sug-

gests that advertising creativity has a role to play in stimulating

positive consumer responses that goes beyond being a source of

attention. If the attention-grabbing nature of advertising crea-

tivity were the key benefit, its effects should be greater for

memory rather than attitudinal responses, and in communica-

tion contexts where consumers are less likely to attend to and

process ads, such as for low-involvement products and for

unfamiliar brands (Dahlen, Rosengren, and Törn 2008; Pieters,

Warlop, and Wedel 2002), which is not in line with the empiri-

cal results. Although claims that advertising creativity enables

advertising to “cut through clutter” and make advertising more

memorable (Pieters, Warlop, and Wedel 2002) are true, they

risk directing focus away from attitudinal consumer responses,

which are more affected. The fact that advertising creativity

has stronger effects in high-involvement contexts suggests that

processing is important for the effects to occur. It also raises the

question of what to expect from advertising creativity in con-

texts where consumers are unlikely to pay attention to and

process ads, such as digital and mobile media. The meta-

analysis did not include any such studies, but the results suggest

that effects should be weaker in media such as smartphones

where focus is very directed at other focal tasks (Melumad and

Meyer 2020). At the same time, effects should be stronger for

advertising content in own channels and in media where con-

sumers voluntarily seek out advertising (Rosengren and Dahlen

2015). However, future research is needed to explicitly study

the role of advertising creativity in these contexts.

Second, we contribute insights into how to define and assess

advertising creativity. In line with the creativity literature

(Amabile 1996; Runco and Jaeger 2012), the results indicate

that creativity is not just about originality. A bipartite definition

and multidimensional assessments of creativity offer better

explanations of the effects (for a similar argument, see Ang,

Lee, and Leong 2007 and Modig and Dahlen 2019). This sug-

gests that researchers should be mindful when using the term

advertising creativity and restrict it to studies of original and

appropriate ads. When studying original advertising only, the

term creativity should be avoided. It also suggests that the

reliance on advertising awards as an operationalization of

advertising creativity is not valid, as such awards tend to focus

on originality (Choi et al. 2018; Kilgour, Sasser, and Koslow

2013). The fact that empirical studies have found positive

effects of original and award-winning ads, however, is reassur-

ing, as the results suggest that, if anything, those studies under-

estimate the effects.

Third, we contribute to the theoretical understanding of how

advertising creativity works. The findings show that the different

Rosengren et al. 51

theoretical accounts of advertising creativity available in the

literature are complementary, but also that they have different

relationships with creativity dimensions. Our meta-analytic path

analysis show that originality primarily stimulates affect trans-

fer, whereas appropriateness is more important for signaling. We

also find that signaling has the highest explanatory value. Again,

this reinforces the notion that a bipartite view of advertising

creativity is most relevant, as ads that combine originality with

appropriateness allow these mechanisms to work simultane-

ously, whereas original ads do not. It also suggests that future

studies of advertising creativity should include more comprehen-

sive theoretical frameworks than what has previously been the

case. Together, these insights offer the basic building blocks for

a more complete processing model of advertising creativity

called for by West, Koslow, and Kilgour (2019).

Fourth, the finding that the three theoretical mediators of

advertising creativity have direct effects on brand response

(Abrand) that are only weakly mediated through ad response

(Aad) adds further to our understanding of how advertising

creativity works.
5

It shows that although creativity has stronger

effects on ad responses than brand responses, these effects are

not necessarily dependent on ad response. Again, this pattern

can be understood in terms of the combination of (high) origin-

ality and (high) appropriateness in creative ads. In line with

Smith et al.’s (2007) finding that originality has advantages in

terms of attention and that appropriateness stimulates down-

stream effects and brand response, advertising creativity allows

the two to work in parallel, which also has more direct brand

outcomes. This finding is in line with the signaling account of

advertising creativity that suggests a more direct effect on the

brand. For researchers, it suggests that when studying the

effects of advertising creativity, brand (and sales) responses

must be included through direct measures rather than relying

on Aad or other ad responses as proxies of such effects.

Overall, the empirical results provide convincing evidence

of the positive effects of advertising creativity on consumer

responses and thus highlight the need for marketing scholars

to consider not only media investments (ad spend; Joshi and

Hanssens 2010; Sridhar et al. 2016) but also creativity invest-

ments in models of how advertising work.

Managerial Contributions

For marketers, we contribute a systematic account and empiri-

cal evidence of the value of advertising creativity. Specifically,

we offer important insights into how, when, and why to invest

in advertising creativity. Given the ongoing debate about the

value of creativity in advertising (Forrester 2019; Premutico

2019), this contribution is timely and useful. It also shows no

evidence of advertising creativity becoming less (or more)

effective over time.

When it comes to how to invest, Reinartz and Saffert (2013)

found that many marketers make suboptimal decisions regard-

ing investments in advertising creativity. We suggest that a

tendency to focus on originality might be the root of this prob-

lem. Creativity is more than originality, and by incorporating

appropriateness consumer response will be more positive. To

achieve this, marketers must find ways to assess advertising

creativity. This is easier said than done, given that creativity

judgments are subjective and vary across context and time. We

find that award-winning ads lead to marginally weaker brand

response, suggesting that consumer rather than professional

judgments should be used. This supports Ang, Lee, and

Leong’s (2007) argument that marketers should involve con-

sumers more in advertising development. Whereas there is a

growing literature focusing on consumers as cocreators of

advertising (Dahlen and Rosengren 2016; Thompson and

Malaviya 2013), consumers could also be engaged as prejudges

of advertising. A post hoc analysis of the role of ad judges

provided additional support for this notion. Specifically, we

coded a variable that distinguished between ads that were

judged to be creative by either consumers, by experts, or

selected from award shows. As some studies did not provide

details on ad judges, we first ran analysis of variance models

for a combination of all three outcome responses (Aad, Abrand,

and intentions) to ensure sufficiently large sample sizes. We

found significant effects (F(2, 351) ¼ 4.931, p ¼ .008) on
outcome response. The effects were stronger when consumers

judged advertising creativity (.373) compared with experts

(.300) or award shows (.193). When we analyzed the three

responses separately, the effect held for Abrand and intentions,

but not for Aad. As brand outcomes are more valuable for

marketers, this reinforces the potential in allowing consumers

to (pre)judge advertising creativity.

When it comes to when to invest, the results suggest that

advertising creativity has positive effects in general but also

that the effects are stronger for attitudinal rather than memory

response and marginally stronger in audiovisual media (TV/

movies vs. print/outdoor). Furthermore, the effects are stronger

for high-involvement contexts. For marketers, this challenges

the established view of advertising as a tool for gaining atten-

tion and suggests that creativity is especially valuable in con-

texts where consumers are likely to process advertising.

Although we studied product involvement, this logic should

also hold for media context involvement, meaning that creativ-

ity is more likely to work in situations where more focused ad

processing occurs. Thus, advertising creativity should be more

important for media contexts in which consumers voluntarily

direct their attention to, or are forced to focus directly on,

advertising than in in media contexts that rely on incidental

and divided attention (see also Dahlen and Rosengren 2016;

Rosengren and Dahlen 2015).

We also find that advertising creativity has marginally

stronger effects for unfamiliar compared with familiar brands.

However, this effect is related only to ad rather than brand

response. As suggested by Campbell and Keller (2003), ad

response is a strong indicator of brand response for unfamiliar

5
This suppression effect is especially interesting because the Aad and Abrand

relationship is typically very strong, as evidenced in previous meta-analyses

(e.g., Brown and Stayman [1992]: r ¼ .600 [# effect sizes ¼ 33], Eisend [2007]:
r ¼ .581 [4], Eisend and Küster [2011]: r ¼ .624 [11]).

52 Journal of Marketing 84(6)

brands (as consumers have little other information on which to

base evaluations), suggesting that this finding is still manage-

rially important. By investing in advertising creativity, such

brands can increase the value of their advertising to consumers

(“advertising equity”; Rosengren and Dahlen 2015). Taken

together, this suggests that advertising creativity is especially

valuable when establishing a new brand in the market.

When it comes to why advertising creativity works, the

mechanisms underlying its positive effects are more profound

than many marketers might think. An in-depth understanding

of how affect transfer, processing, and signaling jointly con-

tribute to brand response help make investments in advertising

creativity less risky (West, Koslow, and Kilgour 2019).

Although marketers who focus on originality can expect posi-

tive effects due to affect transfer, they will miss out on the

potential effects of signaling and appropriateness. By investing

in bipartite advertising creativity, marketers can increase the

chances of their ads being liked, processed, and interpreted as

signals of what the brand has to offer. It also means that there is

little risk that positive effects will be for ad response only.

From a managerial perspective, the effects of signaling are

especially important to consider, as they offer the strongest

explanation for the effects on ad response and because appro-

priateness is especially important in high-involvement and

low-familiarity contexts, where advertising creativity also has

the strongest effects. It suggests that advertisements can pro-

duce effects by way of the signals they send rather than the

specific messages they convey. Signals are especially impor-

tant in situations where there is information asymmetry

between marketers and their customers (Chase and Murtha

2019; Kirmani and Rao 2000). This is arguably the case for

unfamiliar brands and high-involvement products as well as in

other situations where the decision-making process is complex,

such as in business-to-business, business-to-government, and

recruitment contexts (Chase and Murtha 2019; Dahlen,

Rosengren, and Karsberg 2018). In fact, recent research sug-

gests that the effect of advertising signals extends beyond con-

sumers to other stakeholders, such as employees and investors

(Dahlen, Rosengren, and Karsberg 2018), though this is beyond

the scope of the present study.

Limitations and Further Research

Given the nature of a meta-analysis, we could study only

consumer responses that previous researchers had investi-

gated. This means, for example, that we could not consider

potential negative effects of creativity on, for example, con-

fusion, negative affect, and fear appeals. However, we found a

marginally significant negative effect of complexity, suggest-

ing that the potential downsides of creativity warrant further

investigation.

Similarly, the literature review revealed a lack of studies on

the effects of advertising creativity on sales (for an exception,

see Reinartz and Saffert [2013]) and the effects of advertising

creativity in digital contexts, such as the effects of advertising

creativity on social media influencer engagement (Hughes,

Swaminathan, and Brooks 2019). Future studies are needed

to explore how advertising creativity works in those contexts.

Studies linking the effects of advertising creativity to beha-

vioral measures, such as brand choice or sales, seem especially

important. This could be done by combining quantitative

(advertising spend) and qualitative (advertising creativity)

assessments of advertising investments with behavioral

outcomes, for example, adding advertising creativity in

marketing-mix models or adding sales as a dependent variable

in experimental studies. In such efforts, additional moderators,

such as clutter (Pieters, Warlop, and Wedel 2002) and repeti-

tion (Chen, Yang, and Smith 2016), should also be considered.

As another limitation, the present study focused on con-

sumer responses to advertising creativity only. There are sev-

eral related issues in the literature that could contribute to our

understanding of advertising creativity. For example, there is a

vast literature on creative processes in agencies that foster

creativity in advertising (Goldenberg, Mazursky, and Solomon

1999; Kilgour and Koslow 2009), and synthesizing this litera-

ture should bring additional insights to marketers. Relatedly,

there should be room to further integrate the literature on adver-

tising creativity with creativity research focusing on other mar-

keting contexts (Burroughs et al. 2011; Dean, Griffith, and

Calantone 2016) to allow for a more complete understanding

of how creativity works in marketing more broadly. It is our

hope that this article can contribute to this development.

Acknowledgments

The authors dedicate this article in loving memory of Dr. Sheila Sas-

ser, a brilliant creativity scholar and dear friend. The authors are very

grateful to the JM review team for their insightful and constructive

comments that greatly benefited the article.

Associate Editor

Wayne Hoyer

Declaration of Conflicting Interests

The author(s) declared no potential conflicts of interest with respect to

the research, author

ship, and/or publication of this article.

Funding

The author(s) received no financial support for the research, author-

ship, and/or publication of this article.

References

Alba, Joseph W. and J. Wesley Hutchinson (1987), “Dimensions of

Consumer Expertise,” Journal of Consumer Research, 13 (4),

411–54.

Amabile, Teresa M. (1996), Creativity in Context: Update to the

Social Psychology of Creativity. London: Routledge.

Ang, Swee Hoon, Yih Hwai Lee, and Siew Leng Leong (2007), “The

Ad Creativity Cube: Conceptualization and Initial Validation,”

Journal of the Academy of Marketing Science, 35, 220–32.

Baack, Daniel W., Rick T. Wilson, and Brian D. Till (2008),

“Creativity and Memory Effects: Recall, Recognition, and an

Rosengren et al. 53

Exploration of Nontraditional Media,” Journal of Advertising, 37

(4), 85–94.

Baack, Daniel W., Rick T. Wilson, Maria M. van Dessel, and Charles

H. Patti (2015), “Advertising to Businesses: Does Creativity Mat-

ter?” Industrial Marketing Management, 55, 169–77.

Bergh, Donald D., Herman Aguinis, Ciaran Heavey, David J. Ketchen,

Brian K. Boyd, Peiran Su, et al. (2016), “Using Meta-Analytic

Structural Equation Modeling to Advance Strategic Management

Research: Guidelines and an Empirical Illustration via the Strate-

gic Leadership-Performance Relationship,” Strategic Management

Journal, 37 (3), 477–97.

Brown, Steven P. and Douglas M. Stayman (1992), “Antecedents and

Consequences of Attitude Toward the Ad: A Meta-Analysis,”

Journal of Consumer Research 19(1) 34–51.

Burroughs, James E., Darren W. Dahl, Page C. Moreau, Amitava

Chattopadhyay, and Gerald J. Gorn (2011), “Facilitating and

Rewarding Creativity During New Product Development,” Jour-

nal of Marketing, 75 (4), 53–67.

Burroughs, James E. and David Glen Mick (2004), “Exploring Ante-

cedents and Consequences of Consumer Creativity in a Problem-

Solving Context,” Journal of Consumer Research, 31 (2), 402–11.

Cacioppo, John T. and Richard E. Petty (1984), “The Elaboration

Likelihood Model of Persuasion,” in Advances in Consumer

Research, Vol. 11,Thomas C. Kinnear, ed. Provo, UT: Association

for Consumer Research, 673–75.

Campbell, Margaret C. and Kevin Lane Keller (2003), “Brand Famil-

iarity and Advertising Repetition Effects,” Journal of Consumer

Research, 30 (2), 292–304.

Chang, Woojung and Steven A. Taylor (2015), “The Effectiveness of

Customer Participation in New Product Development: A Meta-

Analysis,” Journal of Marketing, 80 (1), 47–64.

Chase, Kevin S. and Brian Murtha (2019), “Selling to Barricaded

Buyers,” Journal of Marketing, 86 (6), 2–20.

Chen, Jiemiao, Xiaojing Yang, and Robert E. Smith (2016), “The

Effects of Creativity on Advertising Wear-In and Wear-Out,”

Journal of the Academy of Marketing Science, 44 (3), 334–49.

Choi, Hojoon, Larry Kelley, Leonard N. Reid, Jan Uhrick, and Kevin

Kou (2018), “Judgments of Highly Creative Advertising: Presence

of Functional Matching and the FCB Planning Model in Clio-

Winning Advertisement,” Creativity Research Journal, 30 (2),

152–63.

Dahlen, Micael, Lars Friberg, and Erik Nilsson (2009), “Long Live

Creative Media Choice,” Journal of Advertising, 38 (2), 121–29.

Dahlen, Micael and Sara Rosengren (2016), “If Advertising Won’t

Die, What Will It Be? Towards a Working Definition of Adver-

tising,” Journal of Advertising, 45 (3), 334–45.

Dahlen, Micael, Sara Rosengren, and John Karsberg (2018), “The

Effects of Signaling Monetary and Creative Effort in Ads. Adver-

tising Effort Can Go a Long Way Influencing B2B Clients,

Employees, and Investors,” Journal of Advertising Research, 58

(4), 433–42.

Dahlen, Micael, Sara Rosengren, and Fredrik Törn (2008),

“Advertising Creativity Matters,” Journal of Advertising Research,

48 (3), 392–403.

De Houwer, Jan, and Frank Bayens (2001), “Association Learning of

Likes and Dislikes: A Review of 25 Years of Research on Human

Affective Conditioning,” Psychological Bulletin, 127 (6), 853–69.

Dean, Tereza, David A. Griffith, and Roger J. Calantone (2016), “New

Product Creativity: Understanding Contract Specificity in New

Product Introductions,” Journal of Marketing, 80 (2), 39–58.

Eisend, Martin (2007), “Understanding Two-Sided Persuasion: An

Empirical Assessment of Theoretical Approaches,” Psychology

& Marketing, 24 (6), 615–40.

Eisend, Martin (2015), “Have We Progressed Marketing Knowledge?

A Meta-Meta-Analysis of Effect Sizes in Marketing Research,”

Journal of Marketing, 79 (3), 23–40.

Eisend, Martin and Franziska Küster (2011), “The Effectiveness of

Publicity Versus Advertising: A Meta-Analytic Investigation of Its

Moderators,” Journal of the Academy of Marketing Science 39(6)

906–21.

El-Murad, Jaafar and Douglas C. West (2004), “The Definition and

Measurement of Creativity: What Do We Know?” Journal of

Advertising Research, 44 (2), 188–201.

Forgas, Joseph P. (1995), “Mood and Judgment: The Affect Infusion

Model (AIM),” Psychological Bulletin, 117 (1), 39–66.

Forrester (2019), “The Cost of Losing Creativity,” research report

(June 17), https://go.forrester.com/agency-creativity/.

Geyskens, Inge, Jan-Benedict E.M. Steenkamp, and Nirmalya Kumar

(1999), “Generalizations About Satisfaction in Marketing Channel

Relationships Using Meta-Analysis,” Journal of Marketing

Research, 36 (2), 223–38.

Goldenberg, Jacob, David Mazursky, and Sorin Solomon (1999), “The

Fundamental Templates of Quality Ads,” Marketing Science, 18

(3), 333–52.

Haberland, Gabriele S. and Peter A. Dacin (1992), “The Development

of a Measure to Assess Viewers’ Judgments of the Creativity of an

Advertisement: A Preliminary Study,” in Advances in Consumer

Research, Vol. 19,John F. Sherry Jr., and Brian Sternthal, eds.

Provo, UT: Association for Consumer Research, 817–25.

Hayes, Jameson L., Guy Golan, Brian Britt, and Janelle Applequist

(2020), “How Advertising Relevance and Consumer–Brand Rela-

tionship Strength Limit Disclosure Effects of Native Ads on

Twitter,” International Journal of Advertising, 39 (1), 131–65.

Hedges, Larry V. and Ingram Olkin (1985), Statistical Methods for

Meta-Analysis. Orlando, FL: Academic Press.

Hennessey, Beth A. and Teresa M. Amabile (2010), “Creativity,”

Annual Review of Psychology, 61, 569–98.

Hughes, Christian, Vanitha Swaminathan, and Gillian Brooks (2019),

“Driving Brand Engagement Through Online Social Influencers:

An Empirical Investigation of Sponsored Blogging Campaigns,”

Journal of Marketing, 83 (5), 78–96.

Hunter, John E. and Frank L. Schmidt (2004), Methods of Meta-

Analysis. Correcting Error and Bias in Research Findings, 2nd

ed.

Thousand Oaks, CA: SAGE Publications.

Iacobucci, Dawn, Neela Saldanha, and Xiaoyan Deng (2007), “A

Meditation on Mediation: Evidence That Structural Equations

Models Perform Better Than Regressions,” Journal of Consumer

Psychology, 17 (2), 139–52.

54 Journal of Marketing 84(6)

https://go.forrester.com/agency-creativity/

Im, Subin and John P. Workman (2004), “Market Orientation, Crea-

tivity, and New Product Performance in High-Technology Firms,”

Journal of Marketing, 68 (2), 114–32.

Joshi, Amit and Dominique M. Hanssens (2010), “The Direct and

Indirect Effects of Advertising Spending on Firm Value,” Journal

of Marketing, 74 (1), 20–33.

Kardes, Frank R., Maria L. Cronley, James J. Kellaris, and Steven S.

Posavac (2004), “The Role of Selective Information Processing in

Price-Quality Inference,” Journal of Consumer Research, 31 (2),

368–74.

Kilgour, Mark and Scott Koslow (2009), “Why and How Do Creative

Thinking Techniques Work? Trading Off Originality and Appro-

priateness to Make More Creative Advertising,” Journal of the

Academy of Marketing Science, 37, 298–309.

Kilgour, Mark, Sheila Sasser, and Scott Koslow (2013), “Creativity

Awards: Great Expectations?” Creativity Research Journal, 25 (2),

163–71.

Kim, Byoung Hee, Sangpil Han, and Sukki Yoon (2010), “Advertising

Creativity in Korea. Scale Development and Validation,” Journal

of Advertising, 39 (2), 95–108.

Kirca, Ahmet H., Satish Jayachandran, and William O. Bearden

(2005), “Market Orientation: A Meta-Analytic Review and Assess-

ment of Its Antecedents and Impact on Performance,” Journal of

Marketing, 69 (2), 24–41.

Kirmani, Amna (1997), “Advertising Repetition as a Signal of Qual-

ity: If It’s Advertised So Much, Something Must Be Wrong,”

Journal of Advertising, 26 (3), 77–86.

Kirmani, Amna (1990), “The Effect of Perceived Advertising Costs on

Brand Perceptions,” Journal of Consumer Research, 17 (1), 160–71.

Kirmani, Amna and Akshay R. Rao (2000), “No Pain, No Gain: A

Critical Review of the Literature on Signaling Unobservable Prod-

uct Quality,” Journal of Marketing, 64 (2), 66–79.

Kirmani, Amna and Peter Wright (1989), “Money Talks: Perceived

Advertising Expense and Expected Product Quality,” Journal of

Consumer Research, 16 (3), 344–53.

Koslow, Scott, Sheila L. Sasser, and Edward A. Riordan (2003),

“What Is Creative to Whom and Why? Perceptions in Advertising

Agencies,” Journal of Advertising Research, 43 (1), 96–110.

Krishen, Anjala S. and Pamela M. Homer (2012), “Do Opposites

Attract? Understanding Opposition in Promotion,” Journal of

Business Research, 65 (8), 1144–51.

Lange, Fredrik, Sara Rosengren, and Angelica Blom (2016), “Store-

Window Creativity’s Impact on Shopper Behavior,” Journal of

Business Research, 69 (3), 1014–21.

Lee, Yih Hwai and Charlotte Mason (1999), “Responses to Information

Incongruency in Advertising: The Role of Expectancy, Relevancy,

and Humor,” Journal of Consumer Research, 26 (2), 156–69.

Levitt, Theodore (1963), “Creativity Is Not Enough,” Harvard Busi-

ness Review, 41 (3), 72–83.

Lipsey, Mark W. and David T. Wilson (2001), Practical Meta-Anal-

ysis. Thousand Oaks, CA: SAGE Publications.

Maas, Cora J.M. and Joop J. Hox (2005), “Sufficient Sample Sizes for

Multilevel Modeling,” Methodology, 1 (3), 86–92.

Machleit, Karen A., Chris T. Allen, and Thomas J. Madden (1993), “The

Mature Brand and Brand Interest: An Alternative Consequence of

Ad-Evoked Affect,” Journal of Marketing, 57 (4), 72–82.

MacInnis, Deborah J. and Bernard Jaworski (1989), “Information Pro-

cessing from Advertisements: Toward an Integrative Framework,”

Journal of Marketing, 53 (4), 1–23.

Maniu, Andreea-Ioana and Monica-Maria Zaharie (2014),

“Advertising Creativity—The Right Balance Between Surprise,

Medium and Message Relevance,” Procedia Economics and

Finance, 15, 1165–72.

Melumad, Shiri and Robert Meyer (2020), “Full Disclosure: How

Smartphones Enhance Consumer Self-Disclosure,” Journal of

Marketing, 84 (3), 28–45.

Meyers-Levy, Joan and Prashant Malaviya (1999), “Consumers’ Pro-

cessing of Persuasive Advertisements: An Integrative Framework

of Persuasion Theories,” Journal of Marketing, 63 (4), 45–60.

Modig, Erik and Micael Dahlen (2019), “Quantifying the Advertising-

Creativity Assessments of Consumers Versus Advertising

Professionals,” Journal of Advertising Research, 60 (1), 1–14.

Modig, Erik and Sara Rosengren (2013), “More Than Price? Explor-

ing the Effects of Creativity and Price in Advertising,” in Advances

in Consumer Research, Vol. 41,Simona Botti and Aparna Labroo,

eds. Duluth, MN: Association for Consumer Research, 373–77.

Moreau, C. Page and Marit Gundersen Engeset (2016), “The Down-

stream Consequences of Problem-Solving Mindsets: How Playing

with LEGO Influences Creativity,” Journal of Marketing

Research, 53 (1), 18–30.

Parsons, Russell (2019), “Proving the Effectiveness of Creativity Is

Key to Profitable Marketing,” Marketing Week (June 10), https://

www.marketingweek.com/prove-effectiveness-creativity/.

Peterson, Robert A. and Steven P. Brown (2005), “On the Use of Beta

Coefficients in Meta-Analysis,” Journal of Applied Psychology, 90

(1), 175–81.

Pieters, Rik, Luk Warlop, and Michel Wedel (2002), “Breaking

Through the Clutter: Benefits of Advertisement Originality and

Familiarity for Brand Attention and Memory,” Management Sci-

ence, 48 (6), 765–81.

Premutico, Leo (2019), “Only the Brave Will Reap the Benefits of

Creativity,” Ad Age (September 26), https://adage.com/article/opin

ion/only-brave-will-reap-benefits-creativity/2198811.

Raudenbush, Stephen W. and Anthony S. Bryk (2002), Hierarchical

Linear Models. Application and Data Analysis Methods, 2nd ed.

Thousand Oaks, CA: SAGE Publications.

Reinartz, Werner and Peter Saffert (2013), “Creativity in Advertising:

When It Works and When It Doesn’t,” Harvard Business Review,

106–12.

Roschk, Holger and Masoumeh Hosseinpour (2020), “Pleasant Ambi-

ent Scents: A Meta-Analysis of Customer Responses and Situa-

tional Contingencies,” Journal of Marketing, 84 (1), 125–45.

Rosengren, Sara and Micael Dahlen (2015), “Exploring Advertising

Equity: How a Brand’s Past Advertising May Affect Consumer

Willingness to Approach Its Future Ads,” Journal of Advertising,

44 (1), 1–13.

Rosengren, Sara, Micael Dahlen, and Erik Modig (2013), “Think

Outside the Ad: Can Advertising Creativity Benefit More Than

the Advertiser?” Journal of Advertising, 42 (4), 320–30.

Rosenthal, Robert (1979), “The ‘File Drawer Problem’ and Tolerance

for Null Results,” Psychological Bulletin, 86 (3), 638–41.

Rosengren et al. 55

Proving the effectiveness of creativity is key to profitable marketing

Proving the effectiveness of creativity is key to profitable marketing

https://adage.com/article/opinion/only-brave-will-reap-benefits-creativity/2198811

https://adage.com/article/opinion/only-brave-will-reap-benefits-creativity/2198811

Runco, Mark A. and Garrett J. Jaeger (2012), “The Standard Defini-

tion of Creativity,” Creativity Research Journal, 24 (1), 92–96.

Sasser, Sheila L. and Scott Koslow (2008), “Desperately Seeking

Advertising Creativity: Engaging an Imaginative ‘3Ps’ Research

Agenda,” Journal of Advertising, 37 (4), 5–19.

Sethuraman, Raj, Gerard J. Tellis, and Richard Briesch (2011), “How

Well Does Advertising Work? Generalizations from Meta-

Analysis of Brand Advertising Elasticities,” Journal of Marketing

Research, 48 (3), 457–71.

Sheinin, Daniel A., Sajeev Varki, and Christy Ashley (2012), “The

Differential Effect of Ad Novelty and Message Usefulness on

Brand Judgments,” Journal of Advertising, 40 (3), 5–17.

Smith, Robert E., Jiemiao Chen, and Xiaojing Yang (2008), “The

Impact of Advertising Creativity on the Hierarchy of Effects,”

Journal of Advertising, 37 (4), 47–61.

Smith, Robert E., Scott B. MacKenzie, Xiaojing Yang, Laura M.

Buchholz, and William K. Darley (2007), “Modeling the Determi-

nants and Effects of Creativity in Advertising,” Marketing Science,

26 (6), 819–33.

Smith, Robert E. and Xiaojing Yang (2004), “Toward a General The-

ory of Creativity in Advertising: Examining the Role of

Divergence,” Marketing Theory 4(1) 31–58.

Sridhar, Shrihari, Frank Germann, Charles Kang, and Rajdeep Grewal

(2016), “Relating Online, Regional, and National Advertising to

Firm Value,” Journal of Marketing, 80 (4), 39–55.

Thompson, Debora V. and Prashant Malaviya (2013), “Consumer-

Generated Ads: Does Awareness of Advertising Co-Creation Help

or Hurt Persuasion?” Journal of Marketing, 77 (3), 33–47.

Till, Brian D. and Daniel W. Baack (2005), “Recall and Persuasion:

Does Creative Advertising Matter?” Journal of Advertising, 34 (3),

47–57.

Vakratsas, Demetrios and Tim Ambler (1999), “How Advertising

Works: What Do We Really Know?” Journal of Marketing, 63 (1),

26–43.

Wang, Guangping, Wenyu Dou, Hairong Li, and Nan Zhou (2013),

“Advertiser Risk Taking, Campaign Originality, and Campaign

Performance,” Journal of Advertising, 42 (1), 42–53.

West, Douglas, Scott Koslow, and Mark Kilgour (2019), “Future

Directions for Advertising Creativity Research,” Journal of Adver-

tising, 48 (1), 102–14.

Yang, Xiaojing and Robert E. Smith (2009), “Beyond Attention

Effects: Modeling the Persuasive and Emotional Effects of Adver-

tising Creativity,” Marketing Science, 28 (5), 935–49.

You, Ya, Gautham G. Vaddkkepatt, and Amit M. Joshi (2015),

“A Meta-Analysis of Electronic Word-of-Mouth Elasticity,” Jour-

nal of Marketing, 79 (2), 19–39.

Zlatevska, Natalina, Chris Dubelaar, and Stephen S. Holden

(2014), “Sizing Up the Effect of Portion Size on Consump-

tion: A Meta-Analytic Review,” Journal of Marketing, 78 (3),

140–54.

56 Journal of Marketing 84(6)

Copyright of Journal of Marketing is the property of American Marketing Association and its
content may not be copied or emailed to multiple sites or posted to a listserv without the
copyright holder’s express written permission. However, users may print, download, or email
articles for individual use.

Calculate your order
Pages (275 words)
Standard price: $0.00
Client Reviews
4.9
Sitejabber
4.6
Trustpilot
4.8
Our Guarantees
100% Confidentiality
Information about customers is confidential and never disclosed to third parties.
Original Writing
We complete all papers from scratch. You can get a plagiarism report.
Timely Delivery
No missed deadlines – 97% of assignments are completed in time.
Money Back
If you're confident that a writer didn't follow your order details, ask for a refund.

Calculate the price of your order

You will get a personal manager and a discount.
We'll send you the first draft for approval by at
Total price:
$0.00
Power up Your Academic Success with the
Team of Professionals. We’ve Got Your Back.
Power up Your Study Success with Experts We’ve Got Your Back.

Order your essay today and save 30% with the discount code ESSAYHELP