Home TECH INEQUITIES

TECH INEQUITIES

Weekly Discussion

Technology & Work Related Inequities

Reflect on this week’s resources and the role of technology in easing/exacerbating work related issues. Integrate the findings of two scholarly articles to connect your experience within broader context of work related inequities. Compare and contrast your experience with your findings. Be sure to address the widening digital divide when you analyze the role of technology in easing/exacerbating work related inequities.

Resources

Required References

Click url to play videos

Charoensukmongkol, P. (2014, July 01). Effects of support and job demands on social media use and work outcomes. Computers in Human Behavior, 36(71), 340-349. Retrieved from http://www.looooker.com/wp-content/uploads/2014/05/Effects-of-support-and-job-demands-on-social-media-use-and-work-outcomes (Links to an external site.)

Hargittai, E. (2003). The digital divide and what to do about it. In D. C. Jones (Eds.), New Economy Handbook (822-841). San Diego, CA: Academic Press. 2003.

Offer, S. (2014). The Costs of Thinking About Work and Family: Mental Labor, Work-Family Spillover, and Gender Inequality Among Parents in Dual-Earner Families. Sociological Forum, 29(4), 916-936. doi:10.1111/socf.12126 Available in Academic Search Premier

Noonan, M. C., & Glass, J. L. (June 01, 2012). The hard truth about telecommuting. Monthly Labor Review, 135(6), 38-45. Retrieved from http://www.bls.gov/opub/mlr/2012/06/art3full

TEDx Talks. (2010, May 25). TEDxDesMoines – Christian Renaud – Future of the workplace and workforce [Video file]. Retrieved from

TEDxDesMoines – Christian Renaud – Future of the Workplace and Workforce

Computers in Human Behavior 36 (2014) 340–349

Contents lists available at ScienceDirect

Computers in Human Behavior

j o u r n a l h o m e p a g e : w w w . e l s e v i e r . c o m / l o c a t e / c o m p h u m b e h

Effects of support and job demands on social media use and work
outcomes

http://dx.doi.org/10.1016/j.chb.2014.03.061
0747-5632/

⇑ Tel.: +66 (662) 727 3003.
E-mail address: peerayuth@outlook.com

Peerayuth Charoensukmongkol ⇑
International College, National Institute of Development Administration, 118 Moo3, Sereethai Road, Klong-Chan, Bangkapi, Bangkok 10240, Thailand

a r t i c l e i n f o a b s t r a c t

Article history:

Keywords:
Social media
Social exchange
Social support
Social capital
Media synchronicity theory
Job performance

Studies related to the use of social media in the workplace are still somewhat scarce despite their increas-
ing popularity in social media research. This paper aims to investigate how employee perceptions of a
workplace related to coworker support, supervisor support, and job-related demands can determine
the degree of attachment some employees feel to social media use at work. The study also explores some
consequences of social media use at work by analyzing its associations with job satisfaction, job perfor-
mance, and cognitive absorption. The data was collected through the snowball sampling technique of 170
employees in Thailand and analyzed using partial least squares regression. For the factors predicted to
influence social media use at work, the analysis found that coworker support and job demands are
positively associated with social media use intensity, while supervisor support is negatively associated
with it. The analysis also found a positive association between job satisfaction and job performance
and social media use at work. An indirect relationship between social media use and cognitive absorption
was also found through the mediating effect of job satisfaction. Overall, the evidence suggests that social
media use at work may not necessarily lead to negative job-related outcomes.

1. Introduction

Leading online social media sites such as Facebook, Twitter,
MySpace, Pinterest, Instagram, and so forth have become an inte-
gral part of the daily activities of many people around the world.
In the academic arena, various aspects related to social media
use have received the preponderance of attention from scholars.
One of the main focuses in research is the impact of social media
use intensity on psychological and societal outcomes (Oh,
Ozkaya, & LaRose, 2014; Reinecke & Trepte, 2014). However, most
of this research was mainly conducted in educational institutions
and used students as subjects (Chang & Heo, 2014; Kalpidou,
Costin, & Morris, 2011; Kirschner & Karpinski, 2010). So far, less
is known about the effect of social media use in organizations.
Studies conducted in this context are important as some organiza-
tions have become concerned about employees’ access to social
media sites during work hours. Some argued that social media
can interrupt work and affect employee performance. Currently,
a lack of empirical evidence exists concerning this impact to guide
organizational policies regarding the use of social media in the
workplace.

The objective of this study, which attempts to fill this research
gap, is twofold. First, some factors that can influence the degree to
which employees believe that social media is important for them
at work are explored, with specific focus on the role of social sup-
port within an organization. Two aspects of social support that are
focused on are coworker support and supervisor support; these
two factors were selected since coworkers and supervisors are per-
sons who not only closely interact with employees in a workplace
but also influence their behavior (Bakker & Bal, 2010; Schreurs,
Hetty van Emmerik, Günter, & Germeys, 2012). Moreover, as the
use of social media is mainly driven by the degree of social connec-
tion that people have with others (Sacks & Graves, 2012), the first
research question is whether the quality of the relationships that
employees develop with people in an organization can influence
how they perceive the importance of social media in the work-
place. In addition, this research also focuses on the impact of some
job characteristics on social media use at work. Specifically, the
perceived importance of social media in terms of favorable or unfa-
vorable work conditions with respect to job demands is explored.

The second objective of this research is to investigate the rela-
tionship between social media use intensity at work and job-related
outcomes of employees. Following a study by Moqbel, Nevo, and
Kock (2013), two job-related outcomes that this research empha-
sizes are job satisfaction and perceived job performance. These
two factors are selected because research has shown that they are

http://crossmark.crossref.org/dialog/?doi=10.1016/j.chb.2014.03.061&domain=pdf

http://dx.doi.org/10.1016/j.chb.2014.03.061

mailto:peerayuth@outlook.com

http://dx.doi.org/10.1016/j.chb.2014.03.061

http://www.sciencedirect.com/science/journal/07475632

http://www.elsevier.com/locate/comphumbeh

P. Charoensukmongkol / Computers in Human Behavior 36 (2014) 340–349 341

considered key indicators that can determine the success of an orga-
nization. In order to extend the study of Moqbel et al. (2013), one
additional job-related outcome in this research is the cognitive
absorption that employees have toward their work. This outcome
factor is selected as it reflects ‘‘the intensity of focus and immersion
experienced by the employees when working’’ (Ho, Sze-Sze, & Chay
Hoon, 2011: p. 26). Specifically, the author aims to explore if the
degree of social media use at work can affect the degree to which
employees can focus on work activities. Moreover, the study was
conducted using samples from Thailand. As previous studies on
social media were mainly conducted in the United States, it is neces-
sary to determine whether these results would be replicated in dif-
ferent cultures. Specifically, Thailand is a country where social
media is very popular among people of all ages. The number of social
media users in Thailand has grown significantly, as recent statistics
showed that in 2013, there were about 18 million social media users
in Thailand, about 27 percent of the total population (Millward,
2013). This increase in social media usage in Thailand makes the
country appropriate for social media research.

The samples used for this research (employees working in ser-
vice and manufacturing industries) were obtained through the
snowball sampling method. Even though using this nonprobability
sampling technique may raise some concerns regarding the repro-
ducibility of the findings in other contexts, results from this study
will add to the existing literature, in which there is a paucity of evi-
dence regarding some factors behind the use of social media in the
workplace and related job outcomes. In addition, the results will
also provide implications to managers about policies toward the
use of social media during work.

2. Background and hypotheses

2.1. Social media

Social media is defined by Kaplan and Haenlein (2010: p. 61) as
‘‘a group of Internet-based applications that build on the ideologi-
cal and technological foundations of Web 2.0 and that allow the
creation and exchange of User Generated Content.’’ Social media
has gained significant popularity worldwide because it not only
allows users to maintain personal relationships with family,
friends, and colleagues but also provides them with opportunities
to make new social connections (Raacke & Bonds-Raacke, 2008).
In addition to the use of social media for personal networking, peo-
ple can access social media for other purposes, such as information
seeking and entertainment (Park, Kee, & Valenzuela, 2009). More
recently, social media has also been applied for business and mar-
keting purposes to advertise products and services online (Kaplan
& Haenlein, 2010).

Despite the benefits that social media provides, concern has
been increasing regarding its potentially negative impact on users.
In particular, concerns related to social media addiction have
recently been raised (Andreassen, Torsheim, Brunborg, &
Pallesen, 2012; Griffiths, 2012). A clinical report showed that some
people are prone to developing social media addiction disorder
(defined as being unable to refrain from checking one’s social
media) (Karaiskos, Tzavellas, Balta, & Paparrigopoulos, 2010); this
behavior dramatically interferes with their daily lives since they
cannot focus on their jobs and/or other responsibilities. In addition,
empirical research on social media has reported various outcomes.
For example, Kalpidou et al. (2011) conducted a study on the
impact of Facebook-use intensity on self-esteem using a sample
of seventy undergraduate students. They found that increased time
spent on Facebook correlated with lower self-esteem. A study by
Kirschner and Karpinski (2010) on the effects of Facebook-use
intensity on academic performance found that students who spent

more time on Facebook not only spent less time studying but also
had poorer academic performance compared with students who
spent less time on Facebook. On the other hand, Kim and Lee
(2011) conducted a study about the relationship between the num-
ber of Facebook friends and the subjective well-being of university
students and found a positive link between the two factors. A study
by Reinecke and Trepte (2014) found that social media tended to
enhance the psychological well-being of subjects who used it as
a form for authentic self-presentation.

A recent study by Moqbel et al. (2013) is considered to be the
pioneering work on the impact of social media use in an organiza-
tion on job-related outcomes, including job satisfaction, job perfor-
mance, and organizational commitment. By using a sample of 193
employees in the United States, they found that social media use
intensity was positively associated with job performance and orga-
nizational commitment through the mediating effect of job satis-
faction. Generally, their results provided support regarding the
positive effect of social media use in an organization. Despite these
new findings, it is still necessary to obtain additional evidence in a
different context. Moreover, to date, some of the workplace factors
that lead workers to believe that social media use is important for
them remain .

2.2. Social support and social media use intensity at work

Social support is widely conceptualized in research as ‘‘the
functions performed for the individual by significant others, such
as family members, friends, and colleagues’’ (Schreurs et al.,
2012: p. 263). Although this conceptualization encompasses sup-
port from various sources, this research focuses on support from
coworkers and supervisors as they are the individuals who have
the most influence on employees’ behaviors within an organiza-
tion. Specifically, the author argues that the level of interpersonal
support employees perceive from their coworkers and from their
supervisors can determine the degree to which employees feel that
social media is important for them at work. However, these two
aspects of social support at work may have different effects on
the intensity of social media use in an organization.

First, with respect to the role of coworker support, this study
hypothesizes that coworker support can increase the intensity of
social media use at work. Coworker support was defined in litera-
ture as ‘‘the extent to which one’s coworkers are helpful, can be
relied upon in times of need, and are receptive to work-related
problems’’ (Menguc & Boichuk, 2012: p. 1360). Accordingly, the
main reason why perceived high levels of support from coworkers
can increase social media use at work is because good relationships
among colleagues make employees feel connected to one another,
which in turn motivates them to communicate and interact more
often (Fay & Kline, 2011). Social media can serve as an online plat-
form that makes it more convenient for employees to connect with
their colleagues (Raacke & Bonds-Raacke, 2008). Instead of face-to-
face interaction, employees may use social media to discuss work
and nonwork issues with each other regardless of where they are
located within or outside an organization (Skeels & Grudin,
2009). The positive association of coworker support and social
media use intensity is also consistent with a study by Oh et al.
(2014) that found a connection between perceived social support
and social media use. Therefore, our hypothesis is as follows:

H1. Coworker support will positively associate with social media
use intensity at work.

While coworker support was hypothesized to increase social
network use at work, the study predicts that the relationship
between supervisor support and social media use intensity at work
will be negative. Supervisor support reflects the degree to which

342 P. Charoensukmongkol / Computers in Human Behavior 36 (2014) 340–349

employees perceive that their supervisor cares about their well-
being. In particular, supervisors are considered to have control over
rewards and punishments of employees. Thus, the degree to which
an employee perceives that the supervisor cares about his or her
well-being can determine the employee’s behavior in the work-
place (Ertureten, Cemalcilar, & Aycan, 2013). In particular, the con-
tribution of supervisor support to employee behaviors can be
explained by the social exchange theory, suggesting that ‘‘employ-
ees who perceive their organizational environment as supportive
will feel obligated to reciprocate with behaviors that are beneficial
to the organization’’ (Zhang & Jia, 2010: p. 747). Moreover, studies
related to the leader-member exchange theory have showed that
employees who received good treatment from their supervisor
tended to avoid counterproductive behaviors and care more about
how their work and work-related behaviors benefit the organiza-
tion (Tse, Huang, & Lam, 2013). Accordingly, when employees
receive favorable treatment from their supervisor, they are less
likely to spend time during work hours accessing their social media
accounts because they may consider this behavior inappropriate
and unfair to the supervisor. Therefore:

H2. Supervisor support will negatively associate with social media
use intensity at work.

2.3. Job demands and social media use intensity at work

In addition to the role of social support in the workplace on the
perceived importance of social media, the characteristics of a given
job can also determine the level of social media use intensity at
work. In particular, this work proposes that strenuous job demands
are one of the main determining factors behind social media use. Job
demands refer to any physical, psychological, social, or organiza-
tional aspects of a job that create psychological strains for employ-
ees (Demerouti, Bakker, Nachreiner, & Schaufeli, 2001). Some
examples of job demands include high-volume workloads, role con-
flict, and other unfavorable working conditions. Research has shown
that job demands are critical factors that make employees experi-
ence work-related stress and burnout (Demerouti, Bakker, & Fried,
2012). For this reason, it can be possible that employees who expe-
rience high job demands may perceive that social media is impor-
tant for them during work. This argument can be supported by the
job demands–resources model (Hausser, Mojzisch, Niesel, &
Schulz-Hardt, 2010), which suggests that employees who experi-
ence a high degree of stress due to job demands tend to need some
type of support to help them deal effectively with external stressors.
Accessing social media can serve as one solution to help employees
relax and take breaks from their stressful work surroundings.
Accordingly, employees who are involved with highly demanding
jobs are proposed to be more likely to access social media compared
to those in less stressful positions. Therefore:

H3. Job demands will positively associate with social media use
intensity at work.

2.4. Social media use intensity during work and job satisfaction

Locke (1976: p. 1304) defined job satisfaction as ‘‘a pleasurable
or positive emotional state resulting from the appraisal of one’s job
or job experiences.’’ Since employees are considered one of the
most critical assets of an organization, firms tend to be concerned
about the level of satisfaction that employees have toward their
jobs. In particular, studies have shown that employees who were
unsatisfied with their jobs not only demonstrated counterproduc-
tive behaviors in a workplace but also had a high tendency to leave
the organization (Dong, Mitchell, Lee, Holtom, & Hinkin, 2012).

Conversely, research has shown that employees who develop
favorable attitudes toward their jobs are more likely to work pro-
ductively and contribute more to their workplace (Bouckenooghe,
Raja, & Butt, 2013; Yunxia & Jianmin, 2010). In this regard, job sat-
isfaction can be considered one major factor that affects employee
work performance. Therefore:

H4. Job satisfaction will positively associate with job performance.

The chance to access social media at work can affect employee
satisfaction. On one positive side, occasionally using social media
can help employees feel relaxed during work. In addition, access
to social media in a workplace can allow employees to have more
opportunity to connect to their family and friends both inside and
outside their organization (König & Caner de la Guardia, 2014).
This social interaction is crucial since it can help employees allevi-
ate work-related stress (Schreurs et al., 2012). In particular, this
positive contribution is also in line with literature related to social
support theory, which shows the positive influence of employee
attitudes toward their jobs and good relationships at the workplace
(Gaan, 2008; Singh, Suar, & Leiter, 2012). This positive contribution
of social media is also consistent with a study by Moqbel et al.
(2013), which argued that social media can promote job satisfac-
tion of employees as it helps them achieve work-life balance and
reduce work-family conflict (Michel, Kotrba, Mitchelson, Clark, &
Baltes, 2011). Therefore:

H5. Social media use intensity at work will positively associate
with job satisfaction.

2.5. Social media use intensity at work and cognitive absorption

According to Schaufeli, Bakker, and Salanova (2006: p. 702),
engagement refers to ‘‘a more persistent and pervasive affective–
cognitive state that is not focused on any particular object, event,
individual, or behavior.’’ It reflects the degree of vigor, dedication,
and absorption that employees have about performing their jobs.
Vigor refers to the degree of effort that people invest in their jobs;
dedication refers to the level of involvement that people have
toward their job; absorption refers to the degree to which people
are fully concentrated and happily engrossed in their work
(Schaufeli, Salanova, González-romá, & Bakker, 2002; Schaufeli
et al., 2006). Specifically, the current study focuses on the absorp-
tion aspect of work engagement, which is considered a cognitive
aspect of engagement that reflects the psychological state in which
employees are ‘‘absorbed in a job so much that everything else is
forgotten’’ (Babcock-Roberson & Strickland, 2010: p. 316). The cog-
nitive aspect of work engagement is crucial since it has a strong
influence on the level of dedication and involvement that people
put to their work activities (Ho et al., 2011). Generally, people
who are highly absorbed in a job tend to feel psychologically
attached to it and tend to perceive that time passes by quickly
while they are doing the task (Schaufeli et al., 2006). For this rea-
son, employees who are highly absorbed in their jobs tend to be
more dedicated to work and are more likely to devote themselves
to the performance of their tasks (Babcock-Roberson & Strickland,
2010; Lin, 2010). Existing studies support the concept that employ-
ees who are highly engaged in the performance of their jobs tend
to report higher work commitment and performance (Bakker &
Bal, 2010; Bakker, Schaufeli, Leiter, & Taris, 2008). Therefore:

H6. The level of cognitive absorption that employees demonstrate
during work will positively associate with job performance.

To date, there is no empirical evidence regarding the link
between social media use intensity and the degree to which people
feel psychologically engaged in their jobs. On one hand, some

P. Charoensukmongkol / Computers in Human Behavior 36 (2014) 340–349 343

research has shown that greater social media intensity can inhibit
the ability of people to focus their attention on work activities
(Kirschner & Karpinski, 2010). From this reasoning, it can be less
likely that the use of social media at work would have a positive
contribution to the level of cognitive absorption that employees
display toward their work. However, it is possible that using social
media at work can help employees in some way develop this cog-
nitive absorption. This paper suggests that instead of a direct asso-
ciation between cognitive absorption and social media intensity,
there rather is a relationship that is mediated through job satisfac-
tion. A number of empirical studies have shown that when
employees are satisfied with their jobs, they are more likely to feel
attached to their work (Mohr & Zoghi, 2008; Zatzick & Iverson,
2011). The job satisfaction that employees achieve through the
chance to access social media during work to balance their work-
life activities can cause employees to develop favorable attitudes
toward their job. As employees are happy with their jobs, they
are more likely to enjoy working and can have a high tendency
to feel attached to their work. Therefore:

H7. There is no significant direct relationship between social
media use intensity at work and the level of cognitive absorption
that employees display toward their work.

H8. Job satisfaction will positively associate with the level of cog-
nitive absorption that employees have toward their work.

H9. Job satisfaction will mediate the positive relationship between
social media use intensity at work and cognitive absorption during
work.

2.6. Social media use intensity during work and job performance

Following Moqbel et al. (2013), this research also predicts a
positive relationship between social media use intensity and job
performance. Specifically, Moqbel et al. (2013) argued from the
vantage of work-life balance that employees’ satisfaction that is
rooted in social media access in a workplace is a major reducer
of employee stress and absenteeism, which will subsequently help
them achieve higher levels of performance. In addition, social net-
work research posits that ties people develop with others through
networking can be a good source of critical information, advice,
and assistance that individuals can use to perform their jobs
(Sparrowe, Liden, Wayne, & Kraimer, 2001). Chow and Chan
(2008) also argued that the quality of social capital as reflected
by connections that people build with others can facilitate knowl-
edge sharing within a network of people. Based on the social cap-
ital theory, using social media at work can have a positive impact
on work performance because social media makes it become more
convenient and easier for employees to obtain advice from friends
or colleagues who are in their social media network. This benefit of
social media also coincides with the media synchronicity theory,
which posits that communication performance will be enhanced
when a variety of media are used (Dennis, Fuller, & Valacich,
2008). According to Cao, Vogel, Guo, Liu, and Gu (2012: p. 3940),
‘‘social media are exactly a combination of different media, provid-
ing the ideal combination of media capabilities for knowledge
transfer.’’ They also argued that social media can enhance work
performance because it serves as a communication channel where
explicit and implicit knowledge can be effectively transferred
among employees (Cao et al., 2012). Therefore:

H10. Social media use intensity at work will positively associate
with job performance.

However, the mere use of social media during work may not be a
single factor that directly affects the job performance of employees.
It is more likely that the positive association between these two fac-
tors could be mediated through the level of job satisfaction that
employees have from using social media during work. Employees
who are satisfied with their jobs are more willing to work effec-
tively, generally leading to higher job performance (Smayling &
Miller, 2012). This indirect contribution of social media and social
networking is consistent with the study by Moqbel et al. (2013) that
found that job satisfaction positively mediated the relationship
between social media use intensity and job performance among
employees in the United States. Therefore:

H11. Job satisfaction will mediate the positive relationship
between social media use intensity at work and job performance.

3. Methodology

3.1. Measures

Following the study by Moqbel et al. (2013), the main depen-
dent variable, social media use intensity during work, was measured
using the modified version of the scale originally developed by
Ellison, Steinfield, and Lampe (2007). The original scale was
designed specifically to measure perceptions people have regard-
ing their personal levels of attachment to Facebook. Therefore, to
be consistent with the previous research, the authors performed
wording modifications by replacing ‘‘Facebook’’ with ‘‘social net-
working’’ and included Facebook and MySpace as examples to clar-
ify the public type of social media. Sample items include ‘‘At work,
social networking sites have become part of my daily routine’’ and ‘‘At
work, I feel out of touch when I haven’t logged onto social networking
sites for a while’’.

Supervisor support was measured using a scale developed by
Eisenberger, Huntington, Hutchinson, and Sowa (1986). It consists
of five items. Sample items include ‘‘My work supervisor really cares
about my wellbeing’’ and ‘‘My supervisor cares about my opinions’’.

Coworker support was measured using a scale developed by
Tang (1998) that consists of six items. Sample items include ‘‘My
coworkers are very helpful when I encounter difficulties with my
work’’ and ‘‘When I encounter a problem I usually seek help from
my coworkers’’.

Job demands was measured using five items from the Job Con-
tent Questionnaire survey developed by Karasek et al. (1998). This
construct was measured in terms of quantitative workload (e.g.,
work hard, work fast).

Job satisfaction was measured using a scale developed by
Cammann, Fichman, Jenkins, and Klesh (1983). The scale consists
of three items. Sample items include ‘‘All in all, I am satisfied with
my job’’ and ‘‘I like working for my current organization’’.

Job performance was measured in terms of subjective perfor-
mance. The scale for job performance was adopted from Rehman
(2011). It consists of three questions. Sample items include ‘‘I am
very satisfied with my performance in my current job’’ and ‘‘My per-
formance in my current job is excellent’’.

Cognitive absorption was measured using six items from Utrecht
Work Engagement Scale developed by Schaufeli et al. (2006). Sam-
ple items include ‘‘Time flies when I am working’’ and ‘‘I am
immersed in my work’’.

All question items for social media use intensity at work, cow-
orker support, supervisor support, job demands, job satisfaction,
job performance, and cognitive absorption were scored on a five-
point frequency scale, ranging from 1 (strongly agree) to 5
(strongly disagree). These constructs were measured as reflective
latent variables.

344 P. Charoensukmongkol / Computers in Human Behavior 36 (2014) 340–349

In addition to the main independent variables, this study con-
trolled for organizational factors and demographic factors that
can influence the endogenous variables. Control variables for
social media use intensity at work include age, gender,
educational level, working class, job position, job tenure, salary,
organizational policy about the use of social media at work, and
the degree to which the use of social media is related to work.
Control variables for job satisfaction, job performance, and cogni-
tive involvement include age, gender, educational level, working
class, job position, job tenure, salary, coworker support, supervi-
sor support, and job demands.

Age was measured in years. Gender was measured as a
dummy variable (females = 0; males 1). Educational level was
measured using an ordinal scale (1 = less than a bachelor’s
degree; 2 = bachelor’s degree; 3 = master’s degree; 4 = doctoral
degree). Working class was measured as a dummy variable
(white-collar workers = 0; blue-collar worker = 1). Job tenure
was measured as the number of years that respondents had
worked for their organizations. Job position was measured using
an ordinal scale (1 = junior staff; 2 = senior staff; 3 = junior man-
ager; 4 = middle-level manager; 5 = senior-level manager). Salary
was measured as monthly payment that employees received.
Organizational policy about the use of social media at work
was measured as a dummy variable (allow = 1; do not allow = 0).
The degree to which the use of social media is related to work
was measured using an ordinal scale ranging from 1 (not related
at all) to 5 (highly related).

Table 1
Descriptive statistics of respondents.

Age (in years) Mean: 28.69
Standard deviation: 5.72

Gender Male: 64 (38%)
Female: 106 (62%)

Education Below bachelor’s degree:
19 (11%)
Bachelor’s degree: 109
(63%)
Master’s degree: 40 (23%)
Doctoral degree: 2 (1%)

Type of organization Manufacturing: 73 (43%)
Service: 97 (57%)

Working class White collar: 159 (94%)
Blue collar: 11 (6%)

Job position Junior staff workers: 111
(65%)
Senior staff workers: 19
(11%)
Lower-level managers: 19
(11%)
Middle-level managers:
17 (10%)
Senior-level managers: 4
(3%)

Salary (in Thai Baht) Mean: 26,647
Standard deviation:
18,398

Job tenure (in years) Mean: 3.35
Standard deviation: 2.8

Policy towards the use of social media at work Allowed: 118 (69%)
Not allow: 52 (31%)

The degree to which the use of social media is
related to work

Not related: 50 (29%)

A little bit related: 40
(24%)
Moderately related: 57
(34%)
Pretty much related: 17
(10%)
Highly related: 6 (3%)

3.2. Samples and data collection

A self-administered questionnaire was used to collect data
from respondents. In order to obtain a diverse set of samples
from various organizations, a snowball sampling technique was
used (Oh et al., 2014). In this research, graduate students who
took the research methodology class were asked to distribute
questionnaires to their colleagues at their workplace. Question-
naire distribution was voluntarily undertaken by students
without any compensation or extra class credit. A total of 250
questionnaires were taken and distributed by students; 170 com-
pletely filled surveys were returned, yielding a 68 percent
response rate.

The final sample is composed of 64 males (38 percent) and 106
females (62 percent). Mean age is 28.69 years old (standard devia-
tion = 5.72). For educational level, 19 had not received their bach-
elor’s degrees (11 percent), 109 held bachelor’s degrees (63
percent), 40 held master’s degrees (23 percent), and 2 held doctoral
degrees (1 percent). For type of organization, 97 respondents
reported that they worked in the service sector (57 percent) while
73 reported that they worked in the manufacturing sector (43 per-
cent). For working class, 159 were white-collar workers (94 per-
cent) and 11 were blue-collar workers (6 percent). For job
position, 111 were junior staff workers (65 percent), 19 were
senior staff workers (11 percent), 19 were lower-level managers
(11 percent), 17 were middle-level manager (10 percent), and 4
were senior-level managers (3 percent). Mean salary is 26,647 Thai
Baht (standard deviation = 18,398). The mean job tenure was
3.35 years (standard deviation = 2.8). For policy towards the use
of social media at work, 118 reported that their firm allowed social
media access (69 percent), while 52 reported that their firm did not
allow (31 percent). For the degree to which the use of social media
is related to work, 50 reported that it was not related (29 percent),
40 reported that it was a little bit related (24 percent), 57 reported
moderately related (34 percent), 17 reported pretty much related
(10 percent), and 6 reported highly related (3 percent). Descriptive
statistics for the respondents are reported in Table 1.

3.3. Analysis strategy

This study uses partial least-squares (PLS) regression for the
analysis. PLS was selected for the analysis because it offers more
flexibility in comparison with covariance-based standard error of
the mean (SEM) techniques. Specifically, PLS does not require data
to be normally distributed (Fornell and Bookstein, 1982). An addi-
tional advantage of PLS is that it allows smaller sample sizes com-
pared to other SEM techniques (Chin and Newsted, 1999). PLS
analysis was performed using WarpPLS version 3.0 (Kock, 2012).

4. Results

Prior to PLS model estimation, it is important to perform a series
of analyses. Firstly, construct reliabilities were evaluated using
Cronbach’s alpha (a) and composite reliability coefficients. The
results, as shown in Table 2, indicated that all coefficients exceeded
0.7 as recommended by Fornell and Larcker (1981). Secondly, the
convergent validity of latent variables was evaluated using factor
loadings. The results indicated that all factor loadings were greater
than 0.5, which is satisfactory as suggested by Hair, Black, Babin,
and Anderson (2009). Next, the test for discriminant validity was
performed using average variance extracted (AVE). As recom-
mended by Fornell and Larcker (1981), the square root of the
AVE of each construct must be greater than other correlations

Table 2
Construct reliability indicators.

SN CWS SPS JDM JSAT JPFM CA

Composite reliability .909 .947 .9 .868 .852 .879 .851
Cronbach’s alpha .879 .936 .848 .81 .737 .793 .789

Note: SN = social media use intensity at work, CWS = coworker support,
SPS = supervisor support, JDM = job demands, JSAT = job satisfaction, JPFM = job
performance, CA = cognitive absorption.

P. Charoensukmongkol / Computers in Human Behavior 36 (2014) 340–349 345

involving that construct in order for discriminant validity to exist.
The results were also satisfactory. Table 3 reports Spearman corre-
lations among variables as well as reliability and discriminant
validity indicators of latent variables.

Finally, the test for the possible presence of multicollinearity
among the indicators was performed using full Variance Inflation
Factor (VIF) statistics. The full VIFs in the model ranged from
1.128 to 2.203, which is considerably below the critical value of
3.3 as suggested by Petter, Straub, and Rai (2007). In addition,
Kock and Lynn (2012) argued that the full VIF test can serve as a
technique that captures the possibility of common method vari-
ance (Lindell & Whitney, 2001) in the PLS model. According to
Kock and Lynn (2012), the full VIF test may be seen as a vari-
ance-based SEM similar to the common method bias test used in
covariance-based SEM. They suggested that common method bias
can be a serious issue if the full VIF value is higher than 3.3. In this
study, the test results suggested that all of the full VIF values were
considerably lower than the critical value.

Results from the PLS analysis that included all control variables
are presented in Fig. 1. The standardized coefficient and t-values
were calculated using a jackknifing resampling procedure. This
resampling method was performed to help minimizing problems
associated with the presence of outliers due to errors in data col-
lection (Chiquoine & Hjalmarsson, 2009). The author also used
ranked data in the estimation in order to deal with an outlier issue.
In particular, WarpPLS 3.0 allows users to select an option whereby
all the data is automatically ranked prior to the SEM analysis in
order to conduct the analyses with only ranked data. When data
is ranked, typically the value distances that typify outliers are sig-
nificantly reduced, effectively eliminating outliers without
decreasing the sample size (Kock, 2012).

Table 3
Correlations among latent variable and discriminant validity indicators.

SMUI CWS SPS JDM JSAT JPFM CA

SMUI (.791)
CWS .05 (.831)
SPS �.08 .434** (.833)
JDM .09 �.048 �.008 (.754)
JSAT .133 .401** .399** �.165* (.812)
JPFM .195* .223** .166* .197* .368** (.841)
CA .144 .205** .285** .041 .459** .335** (.7)
AGE �.228** �.071 .028 .095 .036 .003 .188*
GEN �.069 �.144 �.137 .033 �.151* �.03 �.138
EDU �.048 .019 .002 .111 �.127 .053 �.03
SMRW .559** �.087 .033 .008 .08 .133 .221*
WC �.004 �.013 �.018 �.031 .026 .119 .018
JPOS �.132 �.117 �.042 .192* �.09 �.007 .079
JTN �.087 �.032 �.038 .131 .059 .041 .142
OPSN .236** .076 .127 �.074 .221** .061 .113
SAL �.148 .02 �.004 .223** �.129 .11 �.124

Spearman correlation coefficients are reported; square roots of average variance abstrac
SN = social media use intensity at work, CWS = coworker support, SPS = supervisor supp
CA = cognitive absorption, AGE = age, GEN = gender (male = 1), EDU = education, SMRW
WC = working class (blue-collar worker = 1), JPOS = job position, JTN = job tenure, OPSN =
** Notes: Significant level at 1% respectively.
* Significant level at 5% respectively.

Given the statistical methods outlined above, the nine hypoth-
eses either met or did not meet the criteria for statistical signifi-
cance as described below. To summarize quickly, hypotheses 1–6
and hypotheses 8–10 showed significant relationships between
social media use and the other factor(s) measured; hypotheses 7
and 11 did not reveal significant relationships. Other than hypoth-
esis 11 (predicting an indirect relationship between social media
use intensity during work and job performance, as mediated by
job satisfaction), all hypotheses were confirmed by the data
obtained. The details are as follows:

– Hypothesis 1 predicted a positive link between coworker sup-
port and social media use intensity at work. The result revealed
a positive and significant relationship between them (b = .124;
p < .05). Thus, hypothesis 1 is supported.

– Hypothesis 2 predicted a negative link between supervisor sup-
port and social media use intensity at work. The result indicated
a negative and significant relationship between them
(b = �.172; p < .05). Thus, hypothesis 2 is supported.

– Hypothesis 3 predicted a positive link between job demands
and social media use intensity at work. The analysis also
showed that they were positively and significantly associated
(b = .13; p < .05). Thus, hypothesis 3 is supported.

– Hypothesis 4 predicted a positive association between job satis-
faction and job performance. The analysis showed that their
relationship is positive and significant (b = .204; p < .05). Thus, hypothesis 4 is supported.

– Hypothesis 5 predicted a positive relationship between social
media use intensity at work and job satisfaction. The results
indicate a positive and significant relationship between the
two variables (b = .144; p < .05). Therefore, hypothesis 5 is supported.

– Hypothesis 6 predicted a positive link between cognitive
absorption and job performance. The result is positive and
strongly significant (b = .372; p < .001). Thus, hypothesis 6 is supported.

– Hypothesis 7 predicted that there is no direct relationship
between social media use intensity at work and cognitive
absorption. The result indicated that even though these two
variables are positively related, the association is not statisti-
cally significant (b = .073; p = .238). Therefore, hypothesis 7 is
supported.

AGE GEN EDU SMRW WC JPOS JTN OPSN

(1)
.03 (1)
.114 .019 (1)

* �.084 �.08 �.14 (1)
.144 .151* �.138 .149 (1)
.208** .156* .134 �.062 .076 (1)
.425** .02 .062 .052 .116 .23** (1)
�.039 �.061 �.046 .279** .112 �.1 �.029 (1)
.229** .151* .46** �.302** �.052 .554** .255** �.15

ted are in parentheses.
ort, JDM = job demands, JSAT = job satisfaction, JPFM = job performance.
= social media use related to work.
organizational policy about the use of social media at work (allow = 1), SAL = salary.

Fig. 1. PLS results. Notes: ��, ��, � indicate significant level at 0.1 percent, 1 percent, and 5 percent, respectively; solid lines represent significant results. SN = social media use
intensity at work, CWS = coworker support, SPS = supervisor support, JDM = job demands, JSAT = job satisfaction, JPFM = job performance, CA = cognitive absorption; control
variables are pointed to all endogenous variables.

346 P. Charoensukmongkol / Computers in Human Behavior 36 (2014) 340–349

– Hypothesis 8 predicted a positive link between job satisfaction
and cognitive absorption. The result showed that the associa-
tion is positive and strongly significant (b = .372; p < .001). Thus, hypothesis 8 is supported.

– Hypothesis 9 predicted an indirect relationship between social
media use intensity during work and cognitive absorption as
mediated by the variable of job satisfaction. In order to test
the mediating effect, the method suggested by Preacher and
Hayes (2004) was performed in WarpPLS 3.0. The analysis
shows a positive and significant result (b = .054; p < .05). This finding confirms the mediating effect of job satisfaction on the positive link between social media use intensity at work and job performance. Therefore, hypothesis 9 is supported.

– Hypothesis 10 predicted a positive link between social media
use intensity at work and job performance. The result shows
that they are positively and significantly related (b = .126;
p < .05). Thus, hypothesis 10 is supported.

– Finally, hypothesis 11 predicted an indirect relationship
between social media use intensity during work and job perfor-
mance, as mediated by job satisfaction. Although the result
from the analysis is positive, it is not statistically significant
(b = .042; p = .069). Therefore, hypothesis 11 is not supported.

Lastly, the significance of the relationships between control
variables and key dependent variables can be described as follows:

– Social media use intensity at work positively associated with
the degree to which the use of social media is related to work
(b = .537; p < .001) and organizational policy that allows social media access at work (b = .119; p < .05); the result also indi- cated that social network use intensity at work is higher in females than in males (b = �.17; p < .01).

– Job satisfaction positively associated with coworker support
(b = .308; p < .001), supervisor support (b = .226; p < .05), but negatively associated with job demands (b = �.158; p < .05).

– Job performance positively associated with salary (b = .233;
p < .05) and job demands (b = .26; p < .01); moreover, blue-col- lar workers reported higher job performance than white-collar workers (b = .1; p < .05).

5. Discussion

This research investigated some factors that may explain why
employees perceive that social media is important for them at
work, as well as to explore some job-related outcomes associated
with social media use during work. For the factors that lead to
social media use at work, the analysis found evidence that social
support in a workplace tended to associate strongly with the
degree to which employees attached to social media during work.
Specifically, employees who perceived that their coworkers were
supportive tended to develop a greater attachment to social media
at work. On the other hand, employees who perceived that their
supervisor was supportive tended to have less attachment to social
media. In addition, the study also found evidence of a positive link-
age between the level of job demands and social media use inten-
sity. In particular, employees who experienced unfavorable
working conditions tended to be those who perceived that social
media was important for them at the workplace. Regarding the
job-related outcomes of social media use intensity at work, the
study found supporting evidence about its positive association
with job satisfaction, job performance, and cognitive absorption.
However, while the analysis confirmed its direct association with
job satisfaction and job performance, the association with cogni-
tive absorption was found to be mediated through job satisfaction.

In particular, the results regarding the job-related outcomes of
social media use intensity in a workplace are consistent with the
results from the work by Moqbel et al. (2013) that used samples
from the U.S. Here, the author found similar and consistent results
in samples of the Thai workplace. However, instead of the indirect
effect that social media use intensity had on job performance
through job satisfaction, the present study found strong support
for a direct relationship between them, but not through the medi-
ating effect of job satisfaction. Importantly, the positive associa-
tions between social media use intensity on the key dependent
variables were still significant even though the model controlled
for other factors that strongly influenced job-related outcomes.

Moreover, this study contributes to the previous research by
adding the consideration of the possible impact of social media
use at work on the ability of employees to focus on their tasks.

P. Charoensukmongkol / Computers in Human Behavior 36 (2014) 340–349 347

Interestingly, the evidence supported the positive link between the
two factors, contradicting the argument that the use of social
media can hamper the ability of people to concentrate on their
work (Kirschner & Karpinski, 2010). As predicted, this positive link
appeared to be mediated through satisfaction that employees had
with their job instead of a direct association. This result suggests
that just using social media may not be a sufficient condition for
people to focus better on their task; rather, it could create favor-
able conditions for improving focus on job-related tasks. For the
scope of this research, the satisfaction that employees have from
the opportunity to better balance their work lives and their per-
sonal lives may be among the various conditions that allowed
them to concentrate with more involvement on their jobs.

In addition to the evidence of the job-related outcomes associ-
ated with the use of social media at work, this research also makes
an additional contribution to literature by identifying some factors
that motivate employees to feel attached to social media in an
organization. This contribution is crucial, as research concerning
this is currently lacking. Although several studies have identified
some factors that explain why people tend to use social media
more or less often (Chang & Heo, 2014), none has explored factors
in a workplace that determine social media use intensity at work.

From a theoretical perspective, the overall findings about the
associations between social media use at work, job satisfaction
and performance, and various other work-related factors are in line
with the job demand-resource model, which emphasizes the role
of resources that employees can rely on to help them effectively
deal with stressors at work. In this regard, employees may consider
social media as a means by which they can maintain personal con-
nections with family, friends, and colleagues. The quality of the
relationships that employees maintain with these individuals in
their social network is regarded as a source of support for them
in terms of advice, caring, and empathy. In collectivist cultures
such as that of Thailand, where people tend to value social connec-
tions, employees may be more likely to perceive social media’s
importance as a way of obtaining support from coworkers
(Nadkarni & Hofmann, 2012). These cultural factors can also
explain why access to social media in a workplace tended to asso-
ciate positively with job satisfaction and job performance. As
human life is driven by social needs, the ability to maintain per-
sonal connections with friends while at work can be a crucial factor
that allows employees to balance their work/life activities more
effectively. Employees who are fully satisfied in their personal lives
are more able to concentrate on their work. The findings also coin-
cide with research related to a social capital theory and a theory of
media synchronicity that posited that online social media is an
effective communication tool that allows people to share informa-
tion and benefit from knowledge transfers among friends in their
network (Cao et al., 2012). This is another primary reason why
access to social media can enhance performance of employees.
These findings taken collectively explain the positive contribution
that use of social media during work has on job-related outcomes.
Results regarding these positive contributions are also consistent
with research related to job characteristics that suggests that orga-
nizations should provide favorable working environments for
employees to help them develop positive attitudes towards their
work and organization (Biswas, 2011; Zatzick & Iverson, 2011).

Moreover, the negative linkage between supervisor support and
the intensity of social media use at work is consistent with social
exchange theory that emphasizes the role of treatment that
employees receive from managers on employee behaviors in a
workplace. Specifically, employees who perceived that their super-
visor was supportive and cared about their well-being were found
to have less attachment to social media while at work. This finding,
therefore, suggest a crucial implication. If managers believe that
using social media at work is counterproductive and want to

reduce this behavior, it is important for them to provide adequate
support to employees. Perceived support from supervisors can per-
suade employees to refrain from using social media at work as they
perceive that doing so is not fair to the organization. On the con-
trary, employees that have a negative perception of their supervi-
sors may develop a cynical attitude, and subsequently feel
motivated to spend more time on social media at work to express
their negative job-related attitudes or complain about their super-
visors on their social media site.

5.1. Limitation and future research directions

Firstly, the data used in this research is cross-sectional. Using
cross sectional data in the analysis makes it difficult to justify
the direction of causality between the constructs (Maxwell &
Cole, 2007). For example, while social media use intensity was
hypothesized to affect job performance, it is possible that employ-
ees with poor job performance have less opportunity to spend time
accessing social media at work since they may need to spend more
time improving their performance. This is a common limitation
experienced in previous social media research (Chang and Heo,
2014). Secondly, it is possible that using a self-evaluation question-
naire to collect the data may cause some bias or inaccuracy in the
measurements. Thirdly, the use of a snowball sampling technique
to obtain respondents may cause bias in sample selection. While
this sampling technique is used when researchers have no way
to obtain a representative sample, it is susceptible to issues regard-
ing the degree to which the results can be replicated in further
studies. Future works may employ qualitative analysis and include
triangulating the sources of data to strengthen the findings. More-
over, a study based on small-scale data collection from only two
industries may limit the generalizability of the results. Future stud-
ies may need to expand the sampling frame in order to cover more
samples by using respondents from different industries.

The current research also offers some direction for future studies.
Although this research provides new evidence that extend our
understanding about social media use at work, more research ave-
nues need to be explored. Future studies may investigate the impact
of social media access at work on other job-related outcomes.
Researchers may focus on the behavioral and emotional aspects of
job engagement, in addition to the psychological aspect of engage-
ment explored in the present study. Equally important, it is also
important to explore other antecedents that cause employees to feel
that social media is necessary for them in a workplace. Finally, since
to date research on social media has tended to be limited to a single
country (that is, each study tends to focus on a single country, even if
multiple countries are covered by the breadth of studies), it is
important for future studies to conduct a cross-cultural analysis to
compare the effects of social media between samples from different
countries in order to determine if culture can influence the results. A
single study that examines social media use across cultures will use
the same methodology and study design across the cultures, allow-
ing for better comparison. For example, researchers may investigate
if the role of social support can influence social media use intensity
of employees in individualist cultures.

6. Conclusion

In conclusion, this research expands our understanding about
the motivations social media is used in the workplace, and the con-
sequences of this use. Despite research limitations, the study pro-
vided some evidence that organizational factors including
coworker support, supervisor support, and job demands were
among the key reasons that explained why employees felt a
greater or lesser attachment to social media at work. However,

348 P. Charoensukmongkol / Computers in Human Behavior 36 (2014) 340–349

the study also provided support that using social media in a work-
place does not necessarily result in negative outcomes at work.
Overall, these findings suggested that organizations may allow
employees to access social media, as it may help employees allevi-
ate work-related stress, accordingly helping them perform their
work more effectively.

Acknowledments

This research received financial support from the International
College of National Institute of Development Administration. The
author is very grateful to two anonymous reviewers for valuable
comments and suggestions. All remaining errors and omissions
are the author’s responsibility.

References

Andreassen, C. S., Torsheim, T., Brunborg, G. S., & Pallesen, S. (2012). Development of
a Facebook addiction scale. Psychological Reports, 110, 501–517.

Babcock-Roberson, M. E., & Strickland, O. J. (2010). The relationship between
charismatic leadership, work engagement, and organizational citizenship
behaviors. Journal of Psychology, 144, 313–326.

Bakker, A. B., & Bal, P. M. (2010). Weekly work engagement and performance: A
study among starting teachers. Journal of Occupational and Organizational
Psychology, 83, 189–206.

Bakker, A. B., Schaufeli, W. B., Leiter, M. P., & Taris, T. W. (2008). Work engagement:
An emerging concept in occupational health psychology. Work and Stress, 22,
187–200.

Biswas, S. (2011). Psychological climate as an antecedent of job satisfaction and job
involvement. Indian Journal of Industrial Relations, 46, 465–477.

Bouckenooghe, D., Raja, U., & Butt, A. N. (2013). Combined effects of positive and
negative affectivity and job satisfaction on job performance and turnover
intentions. Journal of Psychology, 147, 105–123.

Cammann, C., Fichman, M., Jenkins, G. D., & Klesh, J. R. (1983). Assessing the
attitudes and perceptions of organizational members. In S. E. Seashore, E. E.
Lawler, P. H. Mirvis, & C. Cammann (Eds.), Assessing organizational change: A
guide to methods, measures, and practices (pp. 71–138). New York, NY: Wiley.

Cao, X., Vogel, D. R., Guo, X., Liu, H., & Gu, J. (2012). Understanding the influence of
social media in the workplace: An integration of media synchronicity and social
capital theories. Paper presented at the Hawaii international conference on system
science (HICSS). Maui, HI.

Chang, C.-W., & Heo, J. (2014). Visiting theories that predict college students’ self-
disclosure on Facebook. Computers in Human Behavior, 30, 79–86.

Chin, W. W., & Newsted, P. R. (1999). Structural equation modeling analysis with
small samples using partial least squares. In R. Hoyle (Ed.), Statistical Strategies
for Small Sample Research (pp. 307–341). Thousand Oaks, CA: Sage.

Chiquoine, B., & Hjalmarsson, E. (2009). Jackknifing stock return predictions. Journal
of Empirical Finance, 16, 793–803.

Chow, W. S., & Chan, L. S. (2008). Social network, social trust and shared goals in
organizational knowledge sharing. Information and Management, 45, 458–465.

Demerouti, E., Bakker, A. B., & Fried, Y. (2012). Work orientations in the job
demands-resources model. Journal of Managerial Psychology, 27, 557–575.

Demerouti, E., Bakker, A. B., Nachreiner, F., & Schaufeli, W. B. (2001). The job
demands-resources model of burnout. Journal of Applied Psychology, 86,
499–512.

Dennis, A. R., Fuller, R. M., & Valacich, J. S. (2008). Media, tasks, and communication
processes: A theory of media synchronicity. MIS Quarterly, 32, 575–600.

Dong, L. I. U., Mitchell, T. R., Lee, T. W., Holtom, B. C., & Hinkin, T. R. (2012). When
employees are out of step with coworkers: How job satisfaction trajectory and
dispersion influence individual- and unit-level voluntary turnover. Academy of
Management Journal, 55, 1360–1380.

Eisenberger, R., Huntington, R., Hutchinson, S., & Sowa, D. (1986). Perceived
organizational support. Journal of Applied Psychology, 71:3(71), 500–507.

Ellison, N. B., Steinfield, C., & Lampe, C. (2007). The benefits of Facebook ‘‘friends’’:
Social capital and college students’ use of online social network sites. Journal of
Computer – Mediated Communication, 12, 1143–1168.

Ertureten, A., Cemalcilar, Z., & Aycan, Z. (2013). The relationship of downward
mobbing with leadership style and organizational attitudes. Journal of Business
Ethics, 116, 205–216.

Fay, M. J., & Kline, S. L. (2011). Coworker relationships and informal communication
in high-intensity telecommuting. Journal of Applied Communication Research, 39,
144–163.

Fornell, C., & Bookstein, F. L. (1982). Two structural equation models: LISREL and PLS
applied to consumer exit-voice theory. Journal of Marketing Research, 19,
440–452.

Fornell, C., & Larcker, D. (1981). Evaluating structural equation models with
unobservable variables and measurement error. Journal of Marketing Research,
18, 39–50.

Gaan, N. (2008). Stress, social support, job attitudes and job outcome across gender.
ICFAI Journal of Organizational Behavior, 7, 34–44.

Griffiths, M. D. (2012). Facebook addiction: Concerns, criticism, and
recommendations – A response to Andreassen and colleagues. Psychological
Reports, 110, 518–520.

Hair, J. F., Black, W. C., Babin, B. J., & Anderson, R. E. (2009). Multivariate data analysis
(7th ed.). Upper Saddle River, NJ: Prentice Hall.

Hausser, J. A., Mojzisch, A., Niesel, M., & Schulz-Hardt, S. (2010). Ten years on: A
review of recent research on the Job Demand-Control (-Support) model and
psychological well-being. Work and Stress, 24, 1–35.

Ho, V. T., Sze-Sze, W., & Chay Hoon, L. (2011). A tale of passion: Linking job passion
and cognitive engagement to employee work performance. Journal of
Management Studies, 48, 26–47.

Kalpidou, M., Costin, D., & Morris, J. (2011). The relationship between Facebook and
the well-being of undergraduate college students. Cyberpsychology, Behavior,
and Social Networking, 14, 183–189.

Kaplan, A. M., & Haenlein, M. (2010). Users of the world, unite! The challenges and
opportunities of Social Media. Business Horizons, 53, 59–68.

Karaiskos, D., Tzavellas, E., Balta, G., & Paparrigopoulos, T. (2010). P02–232 – Social
network addiction: A new clinical disorder? European Psychiatry, 25, 855.

Karasek, R., Brisson, C., Kawakami, N., Houtman, I., Bongers, P., & Amick, B. (1998).
The job content questionnaire (JCQ): An instrument for internationally
comparative assessments of psychosocial job characteristics. Journal of
Occupational Health Psychology, 3, 322–355.

Kim, J., & Lee, J.-E. R. (2011). The Facebook paths to happiness: Effects of the number
of Facebook friends and self-presentation on subjective well-being.
CyberPsychology, Behavior and Social Networking, 14, 359–364.

Kirschner, P. A., & Karpinski, A. C. (2010). Facebook� and academic performance.
Computers in Human Behavior, 26, 1237–1245.

Kock, N. (2012). WarpPLS 3.0 User Manual. Laredo, TX: ScriptWarp Systems.
Kock, N., & Lynn, G. S. (2012). Lateral collinearity and misleading results in variance-

based SEM: An illustration and recommendations. Journal of the Association for
Information Systems, 13, 546–580.

König, C. J., & Caner de la Guardia, M. E. (2014). Exploring the positive side of
personal internet use at work: Does it help in managing the border between
work and nonwork? Computers in Human Behavior, 30, 355–360.

Lin, C.-P. (2010). Modeling corporate citizenship, organizational trust, and work
engagement based on attachment theory. Journal of Business Ethics, 94,
517–531.

Lindell, M., & Whitney, D. (2001). Accounting for common method variance in
cross-sectional research designs. Journal of Applied Psychology, 86, 114–121.

Locke, E. A. (1976). The nature and causes of job satisfaction. In M. D. Dunnette (Ed.),
Handbook of Industrial and Organizational Psychology (pp. 1297–1349). Chicago,
IL: Rand McNally.

Maxwell, S. E., & Cole, D. A. (2007). Bias in cross-sectional analyses of longitudinal
mediation. Psychological Methods, 12, 23–44.

Menguc, B., & Boichuk, J. P. (2012). Customer orientation dissimilarity, sales unit
identification, and customer-directed extra-role behaviors: Understanding the
contingency role of coworker support. Journal of Business Research, 65,
1357–1363.

Michel, J. S., Kotrba, L. M., Mitchelson, J. K., Clark, M. A., & Baltes, B. B. (2011).
Antecedents of work-family conflict: A meta-analytic review. Journal of
Organizational Behavior, 32, 689–725.

Millward, S. (2013). Thailand now has 18 million social media users. Tech in Asia.

(Retrieved 2.01.14).

Mohr, R. D., & Zoghi, C. (2008). High-involvement work design and job satisfaction.
Industrial and Labor Relations Review, 61, 275–296.

Moqbel, M. A., Nevo, S., & Kock, N. (2013). Organizational members’ use of social
networking sites and job performance: An exploratory study. Information
Technology and People, 26, 240–264.

Nadkarni, A., & Hofmann, S. G. (2012). Why do people use Facebook? Personality and
Individual Differences, 52, 243–249.

Oh, H. J., Ozkaya, E., & LaRose, R. (2014). How does online social networking
enhance life satisfaction? The relationships among online supportive
interaction, affect, perceived social support, sense of community, and life
satisfaction. Computers in Human Behavior, 30, 69–78.

Park, N., Kee, K. F., & Valenzuela, S. (2009). Being immersed in social networking
environment: Facebook groups, uses and gratifications, and social outcomes.
CyberPsychology and Behavior, 12, 729–733.

Petter, S., Straub, D., & Rai, A. (2007). Specifying formative constructs in information
systems research. MIS Quarterly, 31, 623–656.

Preacher, K. J., & Hayes, A. F. (2004). SPSS and SAS procedures for estimating indirect
effects in simple mediation models. Behavior Research Methods, Instruments, and
Computers, 36, 717–731.

Raacke, J., & Bonds-Raacke, J. (2008). MySpace and Facebook: Applying the uses and
gratifications theory to exploring friend-networking sites. CyberPsychology and
Behavior, 11, 169–174.

Rehman, M. S. (2011). Exploring the impact of human resources management on
organizational performance: A study of public sector organizations. Journal of
Business Studies Quarterly, 2, 1–16.

Reinecke, L., & Trepte, S. (2014). Authenticity and well-being on social network
sites: A two-wave longitudinal study on the effects of online authenticity and
the positivity bias in SNS communication. Computers in Human Behavior, 30,
95–102.

Sacks, M. A., & Graves, N. (2012). How many ‘‘Friends’’ do you need? Teaching
students how to network using social media. Business Communication Quarterly,
75, 80–88.

http://refhub.elsevier.com/S0747-5632(14)00189-7/h0005

http://refhub.elsevier.com/S0747-5632(14)00189-7/h0010

http://refhub.elsevier.com/S0747-5632(14)00189-7/h0015

http://refhub.elsevier.com/S0747-5632(14)00189-7/h0020

http://refhub.elsevier.com/S0747-5632(14)00189-7/h0025

http://refhub.elsevier.com/S0747-5632(14)00189-7/h0030

http://refhub.elsevier.com/S0747-5632(14)00189-7/h0035

http://refhub.elsevier.com/S0747-5632(14)00189-7/h0045

http://refhub.elsevier.com/S0747-5632(14)00189-7/h9010

http://refhub.elsevier.com/S0747-5632(14)00189-7/h0050

http://refhub.elsevier.com/S0747-5632(14)00189-7/h0055

http://refhub.elsevier.com/S0747-5632(14)00189-7/h0060

http://refhub.elsevier.com/S0747-5632(14)00189-7/h0065

http://refhub.elsevier.com/S0747-5632(14)00189-7/h0070

http://refhub.elsevier.com/S0747-5632(14)00189-7/h0075

http://refhub.elsevier.com/S0747-5632(14)00189-7/h0080

http://refhub.elsevier.com/S0747-5632(14)00189-7/h0085

http://refhub.elsevier.com/S0747-5632(14)00189-7/h0090

http://refhub.elsevier.com/S0747-5632(14)00189-7/h0095

http://refhub.elsevier.com/S0747-5632(14)00189-7/h9015

http://refhub.elsevier.com/S0747-5632(14)00189-7/h0100

http://refhub.elsevier.com/S0747-5632(14)00189-7/h0105

http://refhub.elsevier.com/S0747-5632(14)00189-7/h0110

http://refhub.elsevier.com/S0747-5632(14)00189-7/h0115

http://refhub.elsevier.com/S0747-5632(14)00189-7/h0120

http://refhub.elsevier.com/S0747-5632(14)00189-7/h0125

http://refhub.elsevier.com/S0747-5632(14)00189-7/h0130

http://refhub.elsevier.com/S0747-5632(14)00189-7/h0135

http://refhub.elsevier.com/S0747-5632(14)00189-7/h0140

http://refhub.elsevier.com/S0747-5632(14)00189-7/h0145

http://refhub.elsevier.com/S0747-5632(14)00189-7/h0150

http://refhub.elsevier.com/S0747-5632(14)00189-7/h0155

http://refhub.elsevier.com/S0747-5632(14)00189-7/h0165

http://refhub.elsevier.com/S0747-5632(14)00189-7/h0170

http://refhub.elsevier.com/S0747-5632(14)00189-7/h0175

http://refhub.elsevier.com/S0747-5632(14)00189-7/h0180

http://refhub.elsevier.com/S0747-5632(14)00189-7/h0185

http://refhub.elsevier.com/S0747-5632(14)00189-7/h0190

http://refhub.elsevier.com/S0747-5632(14)00189-7/h0195

http://refhub.elsevier.com/S0747-5632(14)00189-7/h0200

http://www.techinasia.com/thailand-18-million-social-media-users-in-2013/

http://refhub.elsevier.com/S0747-5632(14)00189-7/h0210

http://refhub.elsevier.com/S0747-5632(14)00189-7/h0215

http://refhub.elsevier.com/S0747-5632(14)00189-7/h0220

http://refhub.elsevier.com/S0747-5632(14)00189-7/h0225

http://refhub.elsevier.com/S0747-5632(14)00189-7/h0230

http://refhub.elsevier.com/S0747-5632(14)00189-7/h0235

http://refhub.elsevier.com/S0747-5632(14)00189-7/h0240

http://refhub.elsevier.com/S0747-5632(14)00189-7/h0245

http://refhub.elsevier.com/S0747-5632(14)00189-7/h0250

http://refhub.elsevier.com/S0747-5632(14)00189-7/h0255

http://refhub.elsevier.com/S0747-5632(14)00189-7/h0260

P. Charoensukmongkol / Computers in Human Behavior 36 (2014) 340–349 349

Schaufeli, W. B., Bakker, A. B., & Salanova, M. (2006). The measurement of work
engagement with a short questionnaire: A cross-national study. Educational and
Psychological Measurement, 66, 701–716.

Schaufeli, W., Salanova, M., González-romá, V., & Bakker, A. (2002). The
measurement of engagement and burnout: A two sample confirmatory factor
analytic approach. Journal of Happiness Studies, 3, 71–92.

Schreurs, B. H. J., Hetty van Emmerik, I. J., Günter, H., & Germeys, F. (2012). A weekly
diary study on the buffering role of social support in the relationship between
job insecurity and employee performance. Human Resource Management, 51,
259–279.

Singh, P., Suar, D., & Leiter, M. P. (2012). Antecedents, work-related consequences,
and buffers of job burnout among Indian software developers. Journal of
Leadership and Organizational Studies, 19, 83–104.

Skeels, M. M., & Grudin, J. (2009). When social networks cross boundaries: A case
study of workplace use of Facebook and LinkedIn. Paper presented at the ACM
2009 International conference on supporting group work. New York, NY.

Smayling, M., & Miller, H. (2012). Job satisfaction and job performance at the
internship level. Journal of Leadership, Accountability and Ethics, 9, 27–33.

Sparrowe, R. T., Liden, R. C., Wayne, S. J., & Kraimer, M. L. (2001). Social networks
and the performance of individuals and groups. Academy of Management Journal,
44, 316–325.

Tang, H. K. (1998). An inventory of organizational innovativeness. Technovation, 19,
41–51.

Tse, H. H. M., Huang, X., & Lam, W. (2013). Why does transformational leadership
matter for employee turnover? A multi-foci social exchange perspective.
Leadership Quarterly, 24, 763–776.

Yunxia, Z., & Jianmin, F. (2010). Does the relationship between job satisfaction and
job performance depend on culture? Academy of Management Perspectives, 24,
86–87.

Zatzick, C. D., & Iverson, R. D. (2011). Putting employee involvement in context: A
cross-level model examining job satisfaction and absenteeism in high-
involvement work systems. International Journal of Human Resource
Management, 22, 3462–3476.

Zhang, Z., & Jia, M. (2010). Using social exchange theory to predict the effects of
high-performance human resource practices on corporate entrepreneurship:
Evidence from China. Human Resource Management, 49, 743–765.

http://refhub.elsevier.com/S0747-5632(14)00189-7/h0265

http://refhub.elsevier.com/S0747-5632(14)00189-7/h0270

http://refhub.elsevier.com/S0747-5632(14)00189-7/h0275

http://refhub.elsevier.com/S0747-5632(14)00189-7/h0280

http://refhub.elsevier.com/S0747-5632(14)00189-7/h0290

http://refhub.elsevier.com/S0747-5632(14)00189-7/h0295

http://refhub.elsevier.com/S0747-5632(14)00189-7/h0300

http://refhub.elsevier.com/S0747-5632(14)00189-7/h0305

http://refhub.elsevier.com/S0747-5632(14)00189-7/h0310

http://refhub.elsevier.com/S0747-5632(14)00189-7/h0315

http://refhub.elsevier.com/S0747-5632(14)00189-7/h0320

Effects of support and job demands on social media use and work outcomes

1 Introduction
2 Background and hypotheses
2.1 Social media
2.2 Social support and social media use intensity at work
2.3 Job demands and social media use intensity at work
2.4 Social media use intensity during work and job satisfaction
2.5 Social media use intensity at work and cognitive absorption
2.6 Social media use intensity during work and job performance
3 Methodology
3.1 Measures
3.2 Samples and data collection
3.3 Analysis strategy
4 Results
5 Discussion
5.1 Limitation and future research directions
6 Conclusion
Acknowledments
References

The Digital Divide and What To Do About It

Eszter Hargittai

papers-at-eszter-dot-com
Sociology Department
Princeton University

This is a pre-print version of the book chapter to appear in the “New
Economy Handbook” edited by Derek C. Jones. San Diego, CA: Academic
Press. 2003.

Please do not post this document on any Web sites or distribute it on any
mailing lists. You can point people to its online location here:
http://www.eszter.com/papers/c04-digitaldivide.html

Thank you.

Abstract

In a society where knowledge-intensive activities are an increasingly important
component of the economy, the distribution of knowledge across the population is
increasingly linked to stratification. Much attention among both academic
researchers and in policy circles has been paid to what segments of the population
have access to the Internet or are Internet users. Although the medium has seen
high rates of diffusion, its spread has been unequal both within and across nations.
In this chapter, I look at (a) individual-level inequality in Internet access and use in
the United States, (b) cross-national variation in connectedness, and (c) inequality
from the side of content producers in gaining audiences for their material online.

Outline

I. Introduction

II. Defining the “Digital Divide”

III. From Digital Divide to Digital Inequality

IV. Global Digital Inequality

V. Inequality in Content Production and Distribution

VI. Conclusion

Acknowledgements

Some of the material in this chapter draws on work with Paul DiMaggio to whom I
am grateful for many discussions on the topic. I also thank the National Science
Foundation (grant IIS0086143) and the Dan David Foundation for supporting this
work.

Eszter Hargittai The Digital Divide :: 2

Glossary

Autonomy of Use: The freedom to use technologies when, where and how one
wishes

Digital Divide: The gap between those who have access to digital technologies and
those who do not; or the gap between those who use digital technologies and those
who do not understood in binary terms distinguishing the “haves” from the “have-
nots”

Digital Inequality: A refined understanding of the “digital divide” that emphasizes
a spectrum of inequality across segments of the population depending on differences
along several dimensions of technology access and use

Online Skill: The ability to use the Internet effectively and efficiently

Portal: a Web site that primarily presents itself as a one-stop point-of-entry site to
the content of the Web

Universal Service: Policy to ensure that everyone has affordable access to the
telecommunications network

Eszter Hargittai The Digital Divide :: 3

I. Introduction

In a society where knowledge-intensive activities are an increasingly

important component of the economy, the distribution of knowledge across the

population is increasingly linked to stratification. The mass diffusion of the Internet

across the population has led many to speculate about the potential effects of the

new medium on society at large. Enthusiast have heralded the potential benefits of

the technology suggesting that it will reduce inequality by lowering the barriers to

information allowing people of all backgrounds to improve their human capital,

expand their social networks, search for and find jobs, have better access to health

information and otherwise improve their opportunities and enhance their life

chances. In contrast, others caution that the differential spread of the Internet

across the population will lead to increasing inequalities improving the prospects of

those who are already in privileged positions while denying opportunities for

advancement to the underprivileged.

Much attention among both academic researchers and in policy circles has

been paid to what segments of the population have access to the Internet or are

Internet users. Access is usually defined as having a network-connected machine in

one’s home or workplace. Use more specifically refers to people’s actual use of the

medium beyond merely having access to it. The “digital divide” is most often

conceptualized in binary terms: someone either has access to the medium or does

not, someone either uses the Internet or does not. In this chapter, I offer a refined

understanding of the “digital divide” to include a discussion of different dimensions

of the divide focusing on such details as quality of equipment, autonomy of use, the

presence of social support networks, experience and online skill. In addition to

discussing inequalities at the national level, I also look at the unequal diffusion of the

Internet across countries. Furthermore, I consider the divide that exists at the level

of content production and distribution. Finally, I discuss what type of policy

approach may help in avoiding possible new inequalities emerging from differential

access to and use of the Internet.

Eszter Hargittai The Digital Divide :: 4

II. Defining the “Digital Divide”

Although the Internet has been around for several decades, it saw wide

diffusion only in the second part of the 1990s. Its growth has been especially large

since the emergence of graphical browser software for the Web in 1993. The

number of Americans online grew from 25 million in 1995 when only three percent

of Americans had ever used the Internet to 83 million in 1999, with 55 million

Americans going online on a typical day in mid-2000. In 1994, just 11 percent of

U.S. households had online access. By the end of 1998 this figure had grown to 26.2

percent. Less than two years later it stood at 41.5 percent, and well over 50 percent

of individuals between the ages of 9 and 49 reported going online at home, work, or

some other location. By 2001, over half of the American population was using the

Internet on a regular basis (see Figure 1 for basic Internet user statistics in the United

States over time).

Figure 1. The percentage of the adult US population online, 1994-2001

With the rise of the Internet’s importance in all spheres of life there has been

an increasing concern regarding the patterns of its diffusion across the population.

Reports have documented the presence of an Internet “digital divide”, i.e.

inequalities in access to and use of the medium, with lower levels of connectivity

among women, racial and ethnic minorities, people with lower incomes, rural

Eszter Hargittai The Digital Divide :: 5

residents and less educated people. (See Figures 2-7 for information about the

percentage of various population groups online.)

Figure 2. The percentage of racial groups online among the adult US population, 1994-
2001

Figure 3. The percentage of racial groups online among the adult US population, 1994-
2001

Eszter Hargittai The Digital Divide :: 6

Figure 4. The percentage of non -Hispanic and Hispanic groups online among the adult
US population, 1994-2001

Figure 5. The percentage of groups with different income online among the adult US
population, 1994-2001

Eszter Hargittai The Digital Divide :: 7

Figure 6. The percentage of urban and non-urban groups online among the adult US
population, 1994-2001

Figure 7. The percentage of groups with different educat ional attainment online among
the adult US population, 1994-2001

Eszter Hargittai The Digital Divide :: 8

While most reports identify differences among various segments of the

population, over time studies emphasize the increasing diffusion of the medium

among the population at large. There is considerable disagreement about whether

inequalities in access and use are increasing or decreasing across different

demographic categories. Some argue that with time the majority of the population

will be online and no policy intervention is necessary to achieve equal distribution of

the medium across the population (Compaine 2001). Others emphasize the

increasing differences among various segments of the population at large (Dickard

2002).

These approaches are in stark contrast despite the fact that most of these

reports often rely on the same source of data: the Computer and Internet Use

Supplement of the Current Population Survey administered by the U.S. Census

Bureau. The positions differ because there are different ways in which one can

interpret the data. Let us consider, for example, the Internet use statistics for

Hispanics and non-Hispanics (see Figure 4). It is certainly the case that use has

dramatically increased in both segments of the population: the percentage of

Hispanics online has grown from 5.6 percent in 1994 to 31.4 percent in 2001 while

the percentage of non-Hispanics online increased from 13.3 to 56.9 percent. From

this perspective, Internet use is clearly on the rise in both groups. Moreover,

whereas the percentage of non-Hispanics online increased just over four times, the

growth among Hispanics was over five-fold. Such interpretation suggests optimism

at curbing inequality between groups. However, if we look at Figure 4 we see that

the gap between the two lines has increased from 7.7 percentage points in 1994 to

24.5 percentage points by 2001 suggesting that the overall difference in the

percentage of users is increasing, potentially leading to more inequality among these

two segments of the population. How we interpret the figures has much to do with

what type of divide – if any – we see. Comparing penetration rates across

population groups is more informative than considering numbers about any one

population segment in isolation. Comparison across groups suggests that certain

divides persist and in some cases are growing with respect to Internet diffusion.

Eszter Hargittai The Digital Divide :: 9

III. From Digital Divide to Digital Inequality

Figures 2-7 show that Internet use is spreading at varying rates across

different segments of the population. Some have cautioned that the differential

spread of the Internet will lead to increasing inequalities benefiting those who are

already in advantageous positions and denying access to better resources to the

underprivileged. Robert Merton (1973) called this the “Matthew Effect” according

to which “unto every one who hath shall be given” whereby initial advantages

translate into increasing returns over time.

Research on information technologies has found support for this latter

expectation. Mass media seem to reinforce knowledge gaps across the population.

Past studies have found evidence for this in the realm of general foreign affairs

information (Robinson 1967), political knowledge and participation (Eveland and

Scheufele 2000), diffusion of daily TV news information (Robinson and Levy 1986)

and in a broad range of other information contexts (Gaziano 1983). With respect to

the Web, the Matthew effect predicts that those having more experience with

technologies and more exposure to various communication media will benefit more

from the Web by using it in a more sophisticated manner and for more types of

information retrieval. Evidence has already been presented regarding the connection

between the use of traditional news and entertainment media, and computers and the

Internet (Robinson, Barth and Kohut 1997; Robinson, Levin and Hak 1998). Such

findings suggest that use of the Internet leads to greater information gaps.

As more people start using the Web for communication and information

retrieval, it becomes less useful to merely look at binary classifications of who is

online when discussing questions of inequality in relation to the Internet. Rather, we

need to start looking at differences in how those who are online access and use the

medium. Such a refined understanding of the “digital divide” implies the need for a

more comprehensive term for understanding inequalities in the digital age; DiMaggio

and Hargittai (2001) suggest that the term “digital inequality” better encompasses the
various dimensions along which differences will exist even after access to the

medium is nearly universal.

Some scholars have suggested ways in which we need to distinguish between

different types of Internet use. One such approach (Norris 2001) suggests

Eszter Hargittai The Digital Divide :: 10

distinguishing between divides at three levels: the global divide which encompasses

differences among industrialized and lesser developed nations; the social divide

which points to inequalities among the population within one nation; and a

democratic divide which refers to the differences among those who do and do not

use digital technologies to engage and participate in public life. Wilson (2000) took

this classification a step further by identifying four components of full social access:

i) financial access which indicates whether users (individuals or whole communities)

can afford connectivity; ii) cognitive access which considers whether people are

trained to use the medium, and find and evaluate the type of information for which

they are looking; iii) production of content access which looks at whether there is

enough material available that suits users’ needs; and iv) political access which takes

into account whether users have access to the institutions that regulate the

technologies they are using. Warschauer (2002) has also offered an alternative

approach suggesting that in addition to the physical sides of access, other factors

such as content, language, literacy, education and institutional structures must also be

taken into consideration when assessing the level of information and communication

technology use in a community. These researchers all call for a more holistic

approach to the study of digital inequality.

As the refined approaches above illustrate, there are factors beyond mere

connectivity that need to be considered when discussing the potential implications of

the Internet for inequality. In addition to relying on basic measures of access to a

medium, we need to consider the following more nuanced measures of use:

1. technical means (quality of the equipment)

2. autonomy of use (location of access, freedom to use the medium for one’s

preferred activities)

3. social support networks (availability of others one can turn to for assistance

with use, size of networks to encourage use)

4. experience (number of years using the technology, types of use patterns)

These four factors together contribute to one’s level of skill. Skill is defined as the

ability to efficiently and effectively use the new technology. Here, I consider these five

components which should guide our analyses of digital inequality at the individual

user level.

Eszter Hargittai The Digital Divide :: 11

Technical means

For Internet use several dimensions of equipment quality are relevant to

questions of equal access. People who have access to top quality computers with

good and reliable Internet connections at home or at work are much more likely to

exhibit high levels of sophistication than those without access to such technical

resources. Better hardware, better software and faster connection are the

infrastructural basis of having access to all that the Web has to offer. When using

outdated equipment, more time may be necessary to reach online resources resulting

in fewer opportunities for users to acquaint themselves with and explore varied

corners of the Web. Users may become frustrated by long download times and the

inability to access certain sites potentially leading to less enthusiasm toward the

medium and less time spent exploring its features.

Autonomy of use

Although theoretically many Americans have access to the Internet at a

public library, access remains easiest for those who are connected through home or

work computers. There are differences in how easily people can reach libraries

quickly (e.g. do they live close enough not to require substantial time and monetary

commitments to go there), and whether they are free at times when these resources

are available (e.g. do their work or family responsibilities make it difficult to

capitalize on such resources?). Regarding on-the-job access, those with restrictions

on their work computer use will not have the freedom to enhance their online skills

due to the limitations placed on them by their employment environments. These

differences in autonomy of use are likely to influence people’s level of Web use

sophistication. Those who have easier access to resources and more freedom to use

them are likely to extract more from the medium.

Social support network

The literature on the diffusion of innovations emphasizes the importance of

social support networks in the spread of new technologies. Those with exposure to

innovations in their surroundings are more likely to adopt new technologies such as

personal computers. The availability of friends and family who are also Internet

users provides support for problems encountered while using the medium and is also

a source of new knowledge via advice and recommendations. It is also a source of

Eszter Hargittai The Digital Divide :: 12

encouragement to go online as there are more people with whom to communicate

and share.

For online skills in particular, this implies that people who are able to draw

on their social contacts for information on how to use the medium will learn more

quickly and will be exposed to a broader repertoire of online services than those who

have few people to whom they can turn for advice with their Web use. A study of

home computer diffusion found that people were more likely to give up using the

technology when they had no neighbors or friends to call on for support (Murdock,

Hartmann and Gray 1992). By contrast, people whose social circles include users

knowledgeable about the Web can draw on their networks for site recommendations

and suggestions when they run into problems.

Experience

Experience is a relevant dimension to consider because it tells us whether

people are investing time in a technology to become familiar enough with it for

convenient and efficient use. The amount of prior experience people have with the

Internet is likely to affect their online actions. People who require use of a computer

and online resources for their job or school will have invested time in acquiring

higher level skills in this activity as the acquired knowledge is necessary to perform

their work. People who spend more time online – whether at work or any other

location – will likely acquire more knowledge about the Web and thus will have

better online skills. Finally, people who have been Internet users for longer are

expected to be better at finding information online as they have more experiences to

draw on. Moreover, these are people who were early adopters and thus tend to be

more innovative suggesting more willingness to explore the new medium and

familiarize themselves with it.

Skill

A look at the evolution of how literacy has been defined and refined over

time is a helpful comparison to show that the focus on and necessity of basic access

to a medium is gradually replaced by more refined understandings of what it means

to have efficient access to a communication medium (Kaestle 1991). Whereas

initially literacy simply meant the ability to sign one’s name, someone possessing

solely those writing skills today would not be deemed literate. Such baseline writing

Eszter Hargittai The Digital Divide :: 13

skills today cannot be equated with efficient access to information whether in the

form of government documents or job application forms. Similarly, when

considering the potential implications of the Internet for social inequality, we cannot

rely on a binary classification of who is a user and who is not. Rather, we must also

focus on people’s ability to use the technology effectively and efficiently.

But how is it possible that skill is a relevant factor when it comes to Internet

use given that material posted online – all billions of pages worth – is equally

available to all users via the correct Web address? Beyond the hurdle of gaining

access to a network-connected machine, the zeros and ones that transfer the

multitude of information on the network to the user do not discriminate among

people. (It is important to note here that the plans for the next generation Internet

protocol (IPv6) would allow routers to discriminate among packets which would lead

to increasing inequalities especially with respect to issues discussed in Section V

below.) Once the correct Web address is entered, the data are accessed and the

information is readily available. But how does a user find the particular Web site?

Consider the following scenario. A user is looking for information about

political candidates, in particular, she is interested in comparing the views of two

presidential candidates about a controversial issue, say abortion. There are

thousands of Web sites that describe, critique, and compare political actors.

However, a simple search on the candidate’s name or using the word abortion will

not yield any obvious results, rather, it will present the user with hundreds if not

thousands of possible links to pages with only one of the two topics.

In this particular case, a user who understands how search queries can be

refined through the use of quotation marks (to signal proximity of terms), the use of

Boolean operators (to suggest whether terms should all be included in a search or

whether some terms should be explicitly excluded) and through the use of multiple

terms in a query will likely turn up helpful results almost regardless of the search

engine used. A knowledgeable user may type the following into a search box: bush

gore abortion and quickly find relevant results. Nonetheless, even the use of such

refined search queries requires additional know-how on the part of the user. Many

sites come cluttered with images and text – often in an attempt to make a

commercial venture viable – and it sometimes becomes quite challenging to find

Eszter Hargittai The Digital Divide :: 14

specific information on a page. Among the one hundred participants in a study that

surveyed a random sample of Internet users’ online skills (Hargittai 2003), only one

ever used the Find function (available in all browsers and on all platforms) to search

for a term on a Web page. In the case of this task, looking for the word “abortion”

through use of the Find function would have aided many participants. This action

can significantly reduce the effort it takes to find specific content on a page yet

almost no one uses it. The findings from this study suggest that users differ

significantly in their online skills.

As the above examples illustrate, in addition to demographic characteristics

the five dimensions of user attributes – technical means, autonomy of use, social

support networks, experience and skill – are all important for understanding how

exactly technologies are being adopted by users and to what extent their uses are

similar across different segments of society. Had such nuanced information been

collected on other communication media in their early years, we would have a much

better understanding of their true diffusion across the population and how they may

have contributed to new social inequalities. The above dimensions of user attributes

must all be considered in our discussions of digital inequality but are only starting to

become part of researchers’ agendas in the field. (For a discussion of information

technology skills and the labor market, see the chapter on “The New Economy and

the Organization of Work”. To learn more about how use of the Internet differs

amongst segments of the population for job searches, see the chapter on “The

Internet and Matching in Labor Markets”. )

IV. Global Digital Inequality

Similarly to rapid Internet diffusion within the United States, the number of

users has also grown drastically worldwide from approximately 20 million users in

1995 to 520 million in 2001 (see Figure 8 for details). Although at first glance the

figures suggest that Internet access is becoming a reality for vast segments of the

global population it is important to note that even in 2001 less than ten percent of

the world’s inhabitants had ever used the Internet. Moreover, the medium is

diffusing at considerably different rates across countries. Figure 9 shows that

disproportionate numbers of users are from the North American and European

continents whereas other world regions are vastly underrepresented. Most work on

Eszter Hargittai The Digital Divide :: 15

international Internet diffusion has tried to uncover the reasons for such differential

rates in spread.

Figure 8. Number of Internet users worldwide, 1995-2001 (Data source: Nua Internet
Surveys)

Figure 9. Proportion of Internet users from different geographic regions as compared to
proportion of world population in these regions, 2001 (Data source: Nua Internet Surveys)

Eszter Hargittai The Digital Divide :: 16

Most initial reports focused on bivariate analyses showing high correlation

between economic indicators and diffusion rates. Education has also been

considered an important predictor of Internet use cross-nationally. Recently, some

more refined studies have also considered the effects of institutional factors.

Hargittai (1999) found that among OECD countries, in addition to national wealth,

competition in the telecommunications sector was an important predictor of

connectivity. Along similar lines, Kiirski and Pohjola (2002) found that access price

was an important determinant of connectivity in OECD countries, a factor likely

influenced by the telecommunications policy variable. Guillen and Suarez (2002)

also found similar effects of regulatory environment when looking at diffusion rates

across over one hundred nations. Research on the diffusion of mobile telephony has

also found that competition has a positive effect on the spread of the technology

(Gruber and Verboven 2001; Koski and Kretschmer 2002).

Although in its initial years of mass diffusion the Internet was widely

heralded as a potential equalizing tool across nations, the largely unequal patterns of

its diffusion globally suggest that it may end up contributing more to rising

inequalities rather than leveling the playing field across nations. (See the chapter on

“The Adoption and Diffusion of ICT Across Countries: Patterns and Determinants”

for more on global diffusion patterns.)

V. Inequality in Access to Content Production and Distribution

In addition to looking at individual level variables to see how new media are

adopted by users we must also consider institutional factors that shape new

technologies. The rapid increase in the number of Internet users was complemented

by exponential growth in the amount of information available on the Web. In 1995,

there were fewer than 20,000 Web sites. That number grew to over 38 million by

2002 representing billions of Web pages with as many as two million pages added

daily.

A large portion of these billions of Web pages is available on the Web for

public use. Any individual or organization with the know-how to create a site can

contribute content to the public Web. The technicalities of making such content as

available to users as the most popular Web sites are more or less the same.

However, information abundance still leaves the problem of attention scarcity.

Eszter Hargittai The Digital Divide :: 17

Attention scarcity leads individual creators of content to rely on online gatekeepers

to channel their material toward users and leads users to rely on such services to find

their way to content on the Web. Web services that categorize online information

can be considered gatekeepers on the World Wide Web.

The term ‘gatekeeper’ refers to points that function as gates blocking the

flow of some material while allowing other information to pass through. Although

there may be numerous high quality sites on the Web, there is no guarantee that

anyone will find their way to them. The central concern is no longer what is

produced, but what consumers hear and know about. Accordingly, gatekeeping

activity still occurs, but now takes place at the level of information exposure. Its

location has shifted from the decision about what should be produced to control of

what materials get to consumers and what they become aware of. Users with more

advanced Web use skills will be less dependent on such gatekeepers and can more

easily sidestep them to find information of interest to them.

In order to understand the implications of gatekeeping for the reachability of

online content – whether commercial or not-for-profit content, individual or

governmental materials – it is important to distinguish between content that is

merely present on the Web in contrast to content that users are readily exposed to.

To make this distinction, I use the word ‘available’ to refer to material that exists

online and use ‘accessible’ to denote content that is easily within the reach of Web

users. Whereas ‘availability’ means mere existence, ‘accessibility’ implies relative ease

of reachability.

As the amount of Web content grew exponentially, search engines became

increasingly important in sifting through online material. According to one survey,

85 percent of users have ever used a search engine (Pew 2002). Although seemingly

neutral, search engines systematically exclude certain sites in favor of others either by

design or by accident. Search engines index no more than a small portion of all Web

pages and even collectively the largest engines only a ccount for a combined coverage

of just a fraction of all information online. This suggests that there is great

discrepancy between what is physically available on the Web and what information is

realistically accessible to users.

Eszter Hargittai The Digital Divide :: 18

Undoubtedly, the entry of the private sector into the Internet world

encouraged its wide spread and the growth in online content. Search engines and

portal sites assist millions of users every day in finding information online. So why is

it a problem that commercial interests sometimes guide the content selection on

popular sites? The concern is that search engines that are guided by profit motives

will point people away from the most relevant and best quality sites in favor of those

that have paid the highest bids for placement on the results page regardless of their

quality and specific relevance to the search query.

Analyses of large-scale search engine usage data suggest that users mainly rely

on the first page of results to a search query. A study analyzing almost one billion

queries on the AltaVista search engine showed that in 85 percent of the cases users

only viewed the first screen of results (Silverstein et al. 1999). Web users’ habits have

not changed much over the years. Another study (Spink et al. 2002) compared data

on the use of the Excite search engine from 1997, 1999, and 2001 and found that the

mean number of results pages users looked at had decreased over time. The data in

this study also showed that the majority of users rely on simple queries without the

use of advanced search features mentioned earlier.

These findings suggest that users heavily rely on sites for presenting them

with information rather than using sophisticated search strategies to fine-tune their

queries. This implies that information prominently displayed on portal sites –

whether selected because of high content value or for commercial reasons – has a

good chance of being the destination of visitors. If users do not possess advanced

know-how about how content is organized and presented to them online then they

are especially at the mercy of what content sites decide to feature prominently and

make easily accessible to them.

Sites spend significant resources on optimizing their content to show up as

results. In fact, an entire industry has sprung up around “search engine

optimization” offering advice on how companies and others can best assure that

their Web sites climb to the top of search engine results. In contrast, the sites with

the most relevant content may be posted by a non-profit or an individual on his or

her own initiative and only appear far down the results list because the owners of

such sites do not have the resources to optimize for search engine positioning. So

Eszter Hargittai The Digital Divide :: 19

the overall concern due to the prominence of commercial interests on the Web is not

that users will unknowingly be roped into purchasing information they could

otherwise obtain for free – although this may happen as well – but that they may not

find what they are looking for or may miss the best available information because

those resources are crowded out by the profit-seeking ventures. Accordingly,

inequality exists at the level of content production and distribution in the digital

world.

VI. Conclusion

The prevailing approach to the “digital divide” focuses on a binary

classification of Internet use merely distinguishing those who are connected from

those who do not have access to the medium. Related policy discussions also limit

their focus to targeting connectedness without expanding the issue to questions of

skill which can only be achieved by also paying serious attention to training. The

binary classification is due to historical precedent. U.S. telecommunications policy,

for years, has been concerned with “universal service” whereby all citizens should

have access to affordable telephone service (Schement 1996).

Following this approach, discussions about Internet use have focused on

access only at the expense of considering details about use. In the case of the

telephone it makes sense to target access only as there are only a limited number of

ways in which one may use that medium. In contrast, effective access to the Internet

means much more than simply having a network connected machine. Rather, it

includes the ability to use the medium effectively and efficiently enabling users to

benefit from the medium. These necessary online skills can only be achieved

universally by focusing policy not only on improving access but also investing in

training. For example, Bolt and Crawford (2000) found that although there has been

a rapid increase in the number of public schools offering Internet access, support for

the necessary training and staffing has lagged behind.

Instead of drawing parallels to policy debates about telephone access when

considering Internet access policy, a better analogy is to reflect on the varied

dimensions of literacy. We do not think about literacy in binary terms. Children are

not simply given a book in first grade and expected to read. Nor are they given

excerpts from Shakespeare on their first day of class. Instead, we invest in teaching

Eszter Hargittai The Digital Divide :: 20

students how to read gradually. The history of literacy shows that our understanding

of functional literacy has evolved considerably over time requiring flexibility in

education policy to keep up with the changing landscape. Similarly, it is too

simplistic to assume that merely providing an Internet connection to people will

obliterate all potential access differences among users. Rather, a more refined

approach to the “digital divide”, a more comprehensive understanding of digital

inequality is necessary if we are to avoid increasing inequalities among different

segments of the population due to disparities in effective access to all that the

Internet has to offer.

Eszter Hargittai The Digital Divide :: 21

Further Reading

Bolt, D., and R. Crawford. 2000. Digital Divide: Computers and Our Children’s Future. New
York: TV Books.

Compaine, B (Ed.). 2001. The Digital Divide: Facing a Crisis or Creating a Myth? Cambridge,
MA: MIT Press.

Dickard, N. 2002. “Federal Retrenchment on the Digital Divide: Potential National
Impact.” Washington, DC: Benton Foundation.

DiMaggio, Paul, and Eszter Hargittai. 2001. “From the ‘digital divide’ to ‘digital
inequality’: Studying Internet use as penetration increases.” Princeton: Center for
Arts and Cultural Policy Studies, Woodrow Wilson School, Princeton University.

Eveland, W.P., and D.A. Scheufele. 2000. “Connecting News Media Use with Gaps in
Knowledge and Participation.” Political Communication 17(3):215-237.

Gaziano, C. 1983. “The Knowledge Gap – An Analytical Review of Media Effects.”
Communication Research 10(4):447-486.

Gruber, H, and F Verboven. 2001. “The Evolution of Markets Under Entry and
Standards Regulation – The Case of Global Mobile Telecommunications.”
International Journal of Industrial Organization 19(1189-1212.

Guillen, M, and S Suarez. 2002. “The Political Economy of Internet Development: A
Cross-National, Time-Series Analysis.” Philadelphia: The Wharton School,
University of Pennsylvania.

Hargittai, E. 1999. “Weaving the Western Web: Explaining difference in Internet
connectivity among OECD countries.” Telecommunications Policy 23(701-718.

—. 2003. “How Wide a Web? Social Inequality in the Digital Age.” in Sociology Department.
Princeton, NJ: Princeton University.

Kaestle, C.F. 1991. Literacy in the United States. New Haven: Yale University Press.
Kiiski, S., and M. Pohjola. 2002. “Cross-Country Diffusion of the Internet.” Information

Economics and Policy 14(2):297-310.
Koski, H., and T Kretschmer. 2002. “The Global Wireless Telecommunications Markets –

What Will Shape Their Future?” London: London School of Economics.
McDonald, Sharon, and Linda Spencer. 2000. “Gender Differences in Web Navigation:

Strategies, Efficiency, and Confidence.” Pp. 174-181 in Women, Work, and
Computerization: Charting a Course to the Future, edited by Ellen Balka and Richard.
K. Smith. Boston: Kluwer Academic Publishers.

Merton, R.K. 1973. The Sociology of Science: Theoretical and Empirical Investigations. Chicago:
University of Chicago Press.

Murdock, G., P. Hartmann, and P. Gray. 1992. “Contextualizing Home Computing:
Resources and Practices.” Pp. 146-160 in Consuming Technologies, edited by E.
Hirsch. New York, NY: Routledge.

Norris, P. 2001. Digital Divide: Civic Engagement, Information Poverty and the Internet in
Democratic Societies. New York: Cambridge University Press.

Pew Internet and American Life Project. 2002. “Search Engines:
A Pew Internet Project Data Memo.”
Robinson, J.P. 1967. “World Affairs Information and Mass Media Exposure.” Journalism

Quarterly 44(1):23-31.
Robinson, J.P., K. Barth, and A. Kohut. 1997. “Social Impact Research – Personal

Computers, Mass Media, and Use of Time.” Social Science Computer Review 15(1):65-
82.

Eszter Hargittai The Digital Divide :: 22

Robinson, J.P., S. Levin, and B. Hak. 1998. “Computer Time.” American Demographics:18-
23.

Schement, J. 1996. “Beyond Universal Service: Characteristics of Americans without
Telephones, 1980-1993.” in Communications Policy Working Paper #1. Washington,
D.C.: Benton Foundation.

Silverstein, C., M. Henzinger, H. Marais, and M. Moricz. 1999. “Analysis of a Very Large
Web Search Engine Query Log.” SIGIR Forum 33(1):6-12.

Spink, A., B.J. Jansen, D. Wolfram, and T. Saracevic. 2002. “From E-Sex to E-
Commerce: Web Search Changes.” IEEE Computer 35(3):107-109.

Warschauer, M. 2002. “Reconceptualizing the Digital Divide.” First Monday 7(7).

The Costs of Thinking About Work and Family: Mental Labor,

Work–Family Spillover, and Gender Inequality Among Parents
in Dual-Earner Families

Shira Offer
2

One of the aspects unaccounted for in previous assessments of employed parents ‘distribution of time

is the mental dimension of tasks and demands. This aspect, referred to as mental labor, is conceptual-

ized as the planning, organization, and management of everyday activities. Using the experience sam-

pling method, a unique form of time diary, and survey data from the 500 Family Study (N = 402
mothers with 16,451 signals and 291 fathers with 11,322 signals), this study examined the prevalence,

context, and emotional correlates of mental labor among parents in dual-earner families. Results show

that fathers reported thinking more frequently about job-related matters than mothers but these

concerns did not spill over into unpaid work. By contrast, mothers’ job-related thoughts tended to spill

over into unpaid work and free-time activities. When engaging in mental labor, mothers and fathers

were equally likely to think about family matters, but these thoughts were only detrimental to emo-

tional well-being in mothers. Among both mothers and fathers, paid work was relatively insulated from

thoughts about family matters. Overall, findings highlight mothers’ double burden and suggest that

mental labor may contribute to mothers’ emotional stress and gender inequality among dual-earner

families.

KEY WORDS: dual-earner families; experience sampling method; mental labor; time distribution; well-
being; work-family interface.

INTRODUCTION

As job expectations and parenting standards have become more demanding,
many parents today experience severe time squeezes and are constantly in need of
attending to the demands of both family and work (Bianchi, Robinson, and Milkie
2006; Christensen and Schneider 2010; Gerson 2009; Jacobs and Gerson 2004).
Studies suggest that these pressures have led parents to frequently engage in mental
labor as part of their juggling act (Daly 2002; Darrah, Freeman, and English-Lueck
2007; Hessing 1994). Mental labor is conceptualized as the planning, organization,
coordination, and management of everyday tasks and duties, and it reflects parents’
concerns about their ability to get through the day in an efficient and timely manner
(Hessing 1994; Mederer 1993; Shaw 2001; Walzer 1996). It can be viewed as a

1
I thank Karen Cerulo and the three anonymous Sociological Forum reviewers for their invaluable com-
ments on previous drafts of this article. The findings of this study were presented at the annual meeting
of the American Sociological Association in August 2014 in New York City.

2
Department of Sociology and Anthropology, Bar-Ilan University, Ramat Gan, 52900 Israel; e-mail:
shira.offer@biu.ac.il.

Sociological Forum, Vol. 29, No. 4, December 2014

DOI: 10.1111/socf.12126

916

“mental spreadsheet” that helps parents accomplish daily activities that require
mental effort, such as preparing for meetings at work, arranging car pools for their
children, planning birthday parties, scheduling doctor’s appointments, and meeting
job deadlines.

Paradoxically, although parents engage in mental labor in their attempt to
balance multiple responsibilities at home and work, mental labor can be a major
source of burden. For example, in their ethnographic study of the organization
of time among dual-earner families, Darrah and his associates (2007:93) noted
that “These internal mental activities became something more to do, thus increas-
ing their [parents’] sense of crowding in everyday life.” Yet despite its implica-
tions for parents’ well-being, mental labor has received relatively limited
empirical attention. Most of the literature on family and work has focused on the
visible or physical dimension of tasks and demands (Nippert-Eng 1996). Follow-
ing Shaw’s (2008) claim that the work of thinking typically goes unnoticed by
outside observers, including researchers, this study seeks to fill this gap in the
literature by examining how frequently employed parents engage in mental labor,
what they think about, and how they feel when they do so. Special attention is
dedicated to thoughts that involve domain crossing, namely mental spillover
between work and family—for example, how often parents think about job-
related matters in nonwork domains (e.g., when they engage in unpaid work or
during free-time activities) and how often they think about their family when they
engage in paid work.

The mental dimension of the work–family interface is examined in this article
using data from the 500 Family Study on the work and family life of parents in
dual-earner middle-class families. Indisputably, some parental concerns are univer-
sal (Nelson 2011), but others are likely to depend on families’ specific socioeco-
nomic conditions. For example, all parents want their children to be safe, but this
is a greater concern for low-income families who more often live in unsafe neigh-
borhoods (Furstenberg et al. 1999). Low-income parents also face a more precari-
ous labor market and are thus more likely to be concerned about economic issues
such as securing jobs and making ends meet (e.g., Edin and Lein 1997). Neverthe-
less, even if these data are not representative of all families in the United States,
the focus on middle-class parents covers a segment of the population that has
experienced particularly severe time pressures and is therefore of much interest to
scholars and policymakers (Christensen and Schneider 2010; Jacobs and Gerson
2004).

Participants in the 500 Family Study were surveyed and filled in the experience
sampling method (ESM), a unique time diary method that provides high-quality
information about respondents’ activities, thoughts, and emotional states as they
occur in natural settings (Hektner, Schmidt, and Csikszentmihalyi 2007). A major
advantage of the ESM is that it also provides information about what respondents
think about when they engage in different types of activities. Therefore, the ESM is
particularly suited for studying mental labor and makes it possible to estimate the
extent of mental spillover between work and family and examine its emotional
repercussions.

Costs of Thinking About Work and Family 917

THE CONCEPTUALIZATION OF MENTAL LABOR

In the qualitative literature, mental labor is viewed as another dimension of
family work for which women bear the major responsibility. Family work is often
mental in nature because it involves not merely the accomplishment of physical
tasks, such as cleaning, cooking, and chauffeuring children, but also the planning,
organization, and management of these tasks, as well as the thinking and feelings
that accompany them (Daly 2002; DeVault 1991, 1999; Hessing 1994; Mederer
1993; Shaw 2001; Walzer 1996). Mental labor is generally considered part of
women’s role as household managers. In most dual-earner families, mothers are the
ones who assure that routine activities are maintained and daily needs are met.
Mothers are also expected to be constantly on call and prepare backup plans for
last-minute needs and unexpected events (Arendell 2001; Bianchi et al. 2006;
Coltrane 1996, 2000; Doucet 2006; Hochschild 1989).

Recent scholarship, however, has underscored fathers’ increasing involvement
at home (Bianchi et al. 2006; Cabrera et al. 2000; Sayer, Bianchi, and Robinson
2004). This trend is related, among other things, to the shift toward more egalitarian
relationships and changing norms of fathering, which pressure fathers to take an
active role in child rearing. As a result, a growing number of men include hands-on
care and emotional connectivity, not just financial provision, in their definition of
“good” fathering (Coltrane 1996; Doucet 2006; Pleck 2010). The wish to be more
involved with their family is likely to increase fathers’ sense of stress and conflict,
especially in light of current workplace cultures and policies that provide little
support for fathers’ nurturance efforts and continue to portray the ideal worker as a
man whose family responsibilities do not intrude into his work life (Gerson 2010;
Ranson 2012; Williams 2012).

It is thus not surprising that gender inequality in the domestic sphere persists
and that family care is still viewed as the major responsibility of women (Coltrane
2000; Doucet 2006; Townsend 2002). The qualitative literature suggests that moth-
ers’ role as the household manager and primary, or “default,” parent (i.e., the par-
ent who is the most accessible and attentive to the child’s needs) is also reflected at
the mental level. Studies show that mothers more frequently worry about their chil-
dren (Doucet 2006; Kurz 2002; Walzer 1996), advocate for their children’s needs
(DeVault 1999), coordinate family members’ schedules (Arendell 2001; Coltrane
1996; Daly 2002; Hessing 1994), and organize family activities and events (Arendell
2001; Daly 1996; Shaw 2001, 2008).

Mothers’ concerns about family matters may also affect their engagement in
other domains. It has been argued, for example, that mothers’ job effort and pro-
ductivity may be impaired because they spend more time worrying for their children
and taking care of family and household needs while at work (Anderson, Binder,
and Krause 2002–2003; Budig and England 2001). A similar argument has been
raised as regard free-time activities. Studies have shown that mothers’ experience of
free time was more often fragmented and “contaminated” by activities that revolve
around the care and concern of children, making leisure activities less enjoyable for
mothers than for fathers (Arendell 2001; Bittman and Wajcman 2000; Mattingly
and Bianchi 2003).

918 Offer

Overall, these trends suggest that because concerns about family matters
require the expenditure of considerable time and energy they often take a heavy toll
on parents’, especially mothers’, emotional well-being. Yet this type of mental labor
remains highly invisible and is usually taken for granted, even to those involved
(Darrah et al. 2007; Hessing 1994). For this reason, mental labor plays an impor-
tant role in the reproduction of gender inequality among dual-earner families
(Hochschild 1989; Mederer 1993; Shaw 2008; Walzer 1996).

Most qualitative research has focused on mental labor in the family domain,
but employed parents may also be preoccupied with issues pertaining to their job.
Current workplace cultures require workers, particularly those in high-status occu-
pations, to be available around the clock to deal with job issues (Blair-Loy 2003;
Milliken and Dunn-Jensen 2005; Schieman, Milkie, and Glavin 2009). These expec-
tations have been facilitated by the development of communication technologies,
which allow people to work “anytime and anywhere” (Chesley, 2005; Olson-Bucha-
nan and Boswell 2006). Consistent with these trends, studies show that workers fre-
quently think about their job outside of the workplace and that this type of
preoccupation can impinge upon their family life and the performance of family
roles (Ashforth, Kreiner, and Fugate 2000; Bakker and Demerouti 2007; Nippert-
Eng 1996; Voydanoff 2004).

Research further suggests that these types of concerns, which revolve around
job-related issues, may be especially acute among men because men are still com-
pelled to be the main breadwinner of their family. In contemporary culture, paid
work is central to men’s masculine identity, sense of self-worth, and social status.
Therefore, despite increasing participation at home, being a provider continues to
dominate men’s understandings of what it means to be a “good” father and a “suc-
cessful” man (Christiansen and Palkovitz 2001; Townsend 2002; Williams 2012)
and most men still prioritize paid work over family care (Gerson 2010; Ranson
2012).

In this study I adopt a broad conceptualization of mental labor, which includes
general concerns about tasks to be done and time constraints that arise at both
home and work, referred to as cross-domain mental labor, as well as domain-
specific thoughts, namely thoughts about family matters (i.e., family-specific men-
tal labor) and thoughts about job-related matters (i.e., job-specific mental labor).
I then examine the context of these three types of mental labor and the extent to
which they spill over into the domains of paid work, unpaid work, and free-time
activities.

THE MENTAL DIMENSION OF WORK–FAMILY SPILLOVER

Research on work–family spillover has examined the ways in which overflows
between the domains of work and family manifest themselves, including how feel-
ings, attitudes, and behaviors that develop in one domain are carried over into the
other domain (Bellavia and Frone 2005; Greenhaus and Beutell 1985; Keene and
Reynolds 2005; Roehling, Moen, and Batt 2003; Stevens et al. 2007; Voydanoff
2004). Most attention in this field has been dedicated to the experience of negative

Costs of Thinking About Work and Family 919

spillovers, which result from the existence of multiple demands that compete for
individuals’ finite resources, including time, energy, and attention, and make it diffi-
cult for them to fulfill the requirements of both domains (e.g., Bakker and Geurts
2004; Greenhaus and Beutell 1985; Grzywacz, Almeida, and McDonald 2002;
Mennino and Brayfield 2005).

The literature further acknowledges the directionality of the interference by
distinguishing between work-to-family and family-to-work spillover (Byron 2005;
Grzywacz et al. 2002; Jacobs and Gerson 2004; Roehling et al. 2003; Voydanoff
2004). Studies show that negative work-to-family spillover (e.g., when people find it
difficult to do things at home because their job makes them feel too tired) are more
prevalent than family-to-work spillover (e.g., when family matters keeps people
from spending adequate time on their job), thus testifying to the greater importance
attached to paid work over family (Jacobs and Gerson 2004; Mennino and Bray-
field 2005; Roehling et al. 2003). These findings resonate with Nippert-Eng’s (1996)
observation regarding cross-realm talk. Nippert-Eng notes that talking about work
issues at home is more common than talking about home and personal matters at
work because, among other things, workplace cultures expect employers to dedicate
their full attention and energy to work while they are at the workplace. No equiva-
lent norms exist for the home.

In this study I focus on the mental dimension of the work–family interface by
examining the extent to which thoughts associated with one domain intrude into
other domains. This is a relatively neglected issue in the quantitative literature on
work and family. Most measures of work–family spillover used in survey-based
studies ask how “family life,” “family demands,” or “personal life” in general affects
the individual’s paid work or vice versa. Certain indices do probe mental spillover.
Respondents in the Work, Stress, and Health survey, for example, were asked how
often they think about things going on at work when they are not working (Schi-
eman et al. 2009). In the National Survey of Midlife Development in the United
States, respondents were asked how often personal or family worries and problems
distract them when they are at work (Grzywacz and Marks 2000). These items,
however, were not treated separately but rather were averaged with the other items
in the scale to create a global work–family spillover score. Hence the present study
expands over previous research by specifically focusing on mental work–family
spillover.

A focus on mental spillover is important also because it is indicative of the
degree of permeability or integration between roles. According to border/bound-
ary theory, the extent to which elements from one domain enter the other
domain determines the permeability of the boundaries of the domains (Ashforth
et al. 2000; Bellavia and Frone 2005; Clark 2000; Nippert-Eng 1996). A high
degree of permeability, it is argued, facilitates integration between roles and
between the different aspects of the self and makes the physical and mental

3
By focusing on the detrimental aspects of the work–family interface, the concept of negative spillover
is analogous to the concept of work–family conflict or interference (Roehling et al. 2003; Stevens
et al. 2007). As a form of interrole conflict (Greenhaus and Beutell 1985), both concepts examine the
transfer of emotions and behaviors from home to work and from work to home (Roehling et al.
2003).

920 Offer

transitions between home and work easier.
4
In contemporary society both work

and family have “boundary-spanning demands” that require parents to perform
work-related activities at home or address family-related issues at work and can
lead to role-blurring (Voydanoff 2004, 2005). Mental labor can be an important
source of role-blurring because it allows people to be “physically located in the
role’s domain but psychologically and/or behaviorally involved in another role”
(Ashforth et al. 2000:474). A major goal of the present study is to examine how
frequently these mental intrusions occur and how respondents feel when they
occur.

WORK–FAMILY SPILLOVER AND WELL-BEING

Research has shown that work-to-family and family-to-work spillovers were
both associated with outcomes such as increased stress, depression, and psychologi-
cal distress (see reviews in Allen et al. 2000; Bellavia and Frone 2005). Additionally,
studies have found that work demands leading to role-blurring, such as bringing
work home or receiving phone calls from work at home, were detrimental to well-
being (Ashforth et al. 2000; Glavin et al. 2011; Voydanoff 2005). These findings
raise the question of whether the experience of mental work–family spillover is simi-
larly detrimental to employed parents’ well-being. While qualitative studies have
documented the feelings of stress and emotional burden that often accompany
engagement in mental labor among employed parents, particularly mothers (Hes-
sing 1994; Hochschild 1989; Walzer 1996), this important issue has been relatively
overlooked in quantitative research.

Prior studies, however, suggest that mental work–family spillover may have
negative implications for well-being. Mental spillover requires a quick transition
from a family to a work mind-set or vice versa. These transitions, it is argued, can
make it difficult to concentrate on tasks (Ashforth et al. 2000; Glavin et al. 2011;
Nippert-Eng 1996), create strain-based or time-based conflict (Ashforth et al. 2000;
Greenhaus and Beutell 1985; Olson-Buchanan and Boswell 2006), and induce a neg-
ative self-appraisal that may make people feel that they fail meeting prescribed role
expectations (Voydanoff 2005).

GENDER CONTINGENCIES AND STUDY HYPOTHESES

The experience of mental spillover, as well as its emotional repercussions, may
differ by gender. Despite ideological changes favoring egalitarianism, the salience
and meaning of family and work roles continue to be shaped along gender lines. As
Doucet (2006:191) noted, “men’s and women’s lives as caregivers and earners are

4
While both spillover theory and border, or more generally boundary, theory deal with how the domains
of work and family influence each other, they relate to distinct concepts (see reviews in Bellavia and
Frone 2005; Glavin, Schieman, and Reid 2011; Voydanoff 2005). Research on negative spillover, as a
form of interrole conflict, focuses on how demands from multiple domains are incompatible such that
participation in one role makes it difficult to participate in the other role. By contrast, border/boundary
theory emphasizes the extent to which different role domains overlap, such as when individuals receive
phone calls from work when they are at home.

Costs of Thinking About Work and Family 921

affected by deeply felt moral and social scripts about what they should do within
and outside household life.” Reflecting the view that women should be responsible
for unpaid domestic work, research has shown that women experience greater nega-
tive family-to-work spillover than men (Byron 2005; Chesley 2005; Dilworth 2004;
Keene and Reynolds 2005; Martinengo, Jacob, and Hill 2010). These findings also
suggest that it may be more socially acceptable for family matters to intrude into
mothers’ than into fathers’ paid work and other activities (Butler and Skattebo,
2004; Korabik, McElwain, and Chappell 2008).

By contrast, the continued centrality of breadwinning for men’s paternal
identity (Christiansen and Palkovitz 2001; Townsend 2002) may lead fathers to
think more frequently than mothers about work-related matters in unpaid work
contexts and consider the intrusion of job demands into the family domain as
legitimate. Glavin et al. (2011), for example, showed that men were more often
contacted by coworkers, supervisors, managers, or clients about work matters
outside normal work hours. Simon (1995) argued that because the roles of work
and home are more likely to be interrelated among men, working for pay does
not contradict prescribed role expectations, as is often the case among women
(see also Garey 1999). Much to the contrary, financial provision is often regarded
as fulfilling men’s family obligations. I therefore expect that mothers will more
frequently engage in cross-domain and family-specific mental labor (Hypothesis
1) whereas fathers will more frequently engage in job-specific mental labor
(Hypothesis 2). I also hypothesize that cross-domain and family-specific mental
labor will more frequently spill over into mothers’ than fathers’ paid work and
free-time activities (Hypothesis 3) whereas job-specific mental labor will more fre-
quently spill over into fathers’ than mothers’ unpaid work and free-time activities
(Hypothesis 4).

Furthermore, women are the ones typically held accountable for how their
children fare and how their households are run (Arendell 2001; Doucet 2006;
Walzer 1996), suggesting that mental labor may be a more stressful and negative
experience for mothers than for fathers. Mothers are also more likely than
fathers to be criticized for being involved too much in paid work (Deutsch and
Saxon 1998), which when compounded by the expectation that mothers should
be fully devoted to their job makes the juggling act especially challenging for
mothers (Blair-Loy 2003). Glavin and colleagues (2011) showed that having to
deal with job-related matters outside of normal work hours was related to a
higher level of guilt among mothers only. Furthermore, they found that guilt par-
tially mediated the association between work contact and psychological distress
only among mothers. These findings resonate with Simon’s (1995) study showing
that employed mothers had more guilt feelings and experienced greater role con-
flict because they were more likely to feel that their employment prevented them
from fulfilling their family obligations. Hence I expect that the intrusion of
thoughts about family matters into paid work and free-time activities (Hypothesis
5), and the intrusion of thoughts about job matters into paid work and free-time
activities (Hypothesis 6), will be negatively associated with emotional well-being
but that these associations will be more pronounced among mothers than
fathers.

922 Offer

DATA AND METHODS

The Sample

The data for this study come from the 500 Family Study, a large interdisciplin-
ary multimethod research project that explored the work and family experiences of
parents in middle-class dual-earner families.

5
Respondents were recruited through

posts at local schools and newspapers in eight urban and suburban communities
across the United States in 1999–2000. The study included predominantly non-
Hispanic white families with highly educated parents whose average earnings were
above the national average for married parents in the United States (for more infor-
mation about the 500 Family Study, its design, and sample characteristics, see
Schneider and Waite 2005).

Participants in the 500 Family Study completed a survey and filled in the
experience sampling method (ESM), a unique form of time diary that collects infor-
mation about the content and context of individuals’ daily experiences in the course
of a typical week (Csikszentmihalyi and Larson 1987). In this study respondents
carried a paging device that was preprogrammed to randomly emit eight signals
during their waking hours for 7 consecutive days. When signaled, they were asked
to report and evaluate their activities, thoughts, and affective states on a self-report
questionnaire. The ESM is considered a highly valid and reliable instrument that
provides accurate information about individuals’ time uses and subjective experi-
ences as they occur in their natural setting (Hektner et al. 2007). The ESM response
rate in the 500 Family Study was 78% and 73% for mothers and fathers,
respectively.

The analyses conducted in this study are based on a subsample of parents
who provided both survey and ESM data (402 mothers with 16,451 signals and
291 fathers with 11,322 signals). Consistent with previous studies that used the
ESM (Schneider 2006), participants who responded to less than a fourth of the sig-
nals (15 mothers and 11 fathers) were excluded from the sample. A multiple impu-
tation technique (MI) with AMELIA II software (Honaker, King, and Blackwell
2009) was used to impute for missing data in the ESM emotional outcomes and
survey measures.

6
Imputing missing data did not yield significantly different

results.

Measures

Mental Labor. In the ESM, respondents were asked the open-ended question
“What was on your mind?” when signaled. I used responses to this question to

5
This research project was directed by Barbara Schneider and Linda Waite and funded by the Alfred P.
Sloan Center on Parents, Children, and Work at the University of Chicago. The 500 Family Study data
are available through the Inter-University Consortium for Political and Social Research (ICPSR) and
can be downloaded at http://www.icpsr.umich.edu/icpsrweb/ICPSR/studies/4549 or http://dx.doi.org/
10.3886/ICPSR04549. I

6
In this study, 695 signals could not be imputed because the respondent did not answer any of the emo-
tional well-being items. The advantage of the ESM, however, is that if respondents have a sufficient
number of signals overall, they can fail to respond to one or several signals without being excluded
from the analyses.

Costs of Thinking About Work and Family 923

construct three measures of mental labor: (1) Cross-domain mental labor (CD-ML)
refers to general concerns about daily tasks and one’s ability to accomplish them in
a timely manner and it includes items such as thinking about things to do, being
late, and the day’s or next day’s schedule; (2) Family-specific mental labor (F-ML)
includes thoughts about family, children, and spouse; (3) Job-specific mental labor
(J-ML) includes thoughts about work, clients, and supervisor. All three measures of
mental labor are dummy variables indicating whether the respondent engaged in
this type of mental labor when signaled (yes = 1; no = 0). Responses such as thinking
about food, life, God, society, and sex, which accounted for 22% of all signals in
which thoughts were reported, were excluded from this measure because they did
not fit the conceptualization of mental labor.

Main Activity. To examine the context of mental labor and assess the extent of
mental spillover the three mental labor measures were considered by the type of
activity parents were engaged in when they reported mental labor. In the ESM
participants were asked to report what they were doing when signaled.
Responses to this item were originally coded by trained coders into more than
400 activity codes (all the items were double coded, with interrater reliabilities
ranging from j = .79–.95). I grouped these codes into the three following cate-
gories: (1) paid work refers to employment-related activities conducted at work,
home, and other settings, including commuting to and from work; (2) unpaid
work includes family care and domestic activities, such as household chores,
child care, and shopping; and (3) free time refers to personal care and discre-
tionary activities, including eating, recreation, leisure, and participation in reli-
gious and social events. As shown in Table I, mothers reported spending
significantly less time in paid work and more time in unpaid work and free-time
activities than fathers.

Emotional Well-Being. In the ESM, respondents were asked to report on their feel-
ings when signaled. I used these items to construct two composite emotional well-
being measures that were used as dependent variables in the multilevel analyses.
These include the means of positive affect—feeling cheerful, happy, relaxed, good
about oneself, and succeeding in the activity when signaled (a =.85)—and negative
affect—feeling nervous, strained, worried, and stressed (a = .88). The response cate-
gories for all the ESM items ranged from 0 (not at all) to 3 (very much). As Table I
indicates, mothers and fathers in this study reported similar levels of positive and
negative affect.

Control Variable. The multilevel models controlled for several familial, demo-
graphic, and occupational characteristics that have been shown in previous
research to be associated with work–family spillover (e.g., Bellavia and Frone
2005; Byron 2005; Dilworth 2004; Grzywacz et al. 2002; Keene and Reynolds
2005; Martinengo et al. 2010; Stevens et al. 2007). Number of children is a dummy
variable indicating whether the respondent has one or two children (three or more
children is the reference category). Age of youngest child includes two dummies:
whether the respondent’s youngest child is below age 2 or between the ages of 2

924 Offer

and 6 (youngest child above age 6 is the reference category). Age is measured in
years. Graduate degree refers to whether the respondent has a graduate or profes-
sional degree (yes = 1; no = 0). The variable work hours was originally measured
with seven categories referring to the number of hours the respondent spends
working in a typical week (“1–15,” “16–25,” “26–37,” “38–45,” “46–50,” “51–60,”
and “more than 60”). This question was recoded into two dummy variables:
whether the respondent works 38–50, or 51 or more hours per week (37 or fewer
hours per week is the reference category). Regular work schedule refers to whether
the respondent works on a regular daytime (anytime 6 a.m. to 6 p.m.) schedule
(yes = 1; no = 0). Job autonomy is the mean of three items asking respondents, on
a scale ranging from 1 (“not true at all”) to 4 (“very true”), whether they have a
lot of opportunities at work to make decisions on their own, and say over what
happens at their job, and whether they can design or plan most of their daily
work (a = .88). Job stress is assessed with a single item asking, on a scale ranging
from 0 (“never”) to 4 (“always”), how often the respondent finds his or her work
stressful. As Table I shows, more than half of the parents in this study had one to
two children, about 6% of them had a child under the age of 2, and parents’ aver-
age age was approximately 45. More than 80% of mothers and nearly 90% of
fathers had a graduate or professional degree. A significantly higher percentage of
fathers reported working 38–50 hours per week (61%, compared to 48% among
mothers) and 51 or more hours per week (26%, compared to 6% among
mothers). The percentage of respondents who reported a regular daytime schedule
was also significantly higher among fathers than mothers (83% and 70%,

Table I. Means and standard deviations for emotional well-being, main activity, and control variables
by gender

Mothers Fathers

Emotional well-being (ESM)
a

Positive affect 1.92 (.35) 1.92 (.32)
Negative affect .42 (.30) .40 (.31)

Primary activity (ESM)
a

Paid work .33 (.16) .45*** (.15)
Unpaid work .28 (.13) .18*** (.10)
Free time .39 (.12) .37 (.13)

Controls (survey)
One or two children .61 (.49) .55 (.03)
Youngest child is under 2 years .06 (.24) .07 (.25)
Youngest child is 2–6 years .23 (.42) .22 (.42)
Age 44.96 (6.21) 45.85 (6.42)
Graduate degree .81 (.39) .90* (.31)
Work hours (38–50) .48 (.50) .61*** (.49)
Work hours (51+) .06 (.25) .26*** (.44)
Regular work schedule .70 (.46) .83*** (.38)
Job autonomy (0–4) 3.18 (.85) 3.26 (.80)
Job stress (1–4) 2.08 (.77) 2.15 (.77)

Note: N = 402 mothers with 16,451 signals; 291 fathers with 11,322 signals.
a
The estimate is calculated at the aggregated person-level.
*p < .05; ** p < .01; *** p < .001 (two-tailed tests).

Costs of Thinking About Work and Family 925

respectively). No significant gender differences were found for job autonomy or
job stress.

Analytic Strategy

The first part of the study provides descriptive information about the preva-
lence and content of mental labor and extent of mental spillover by type of activity.
The second part of the study takes advantage of the nested structure of the data and
employs hierarchical linear modeling to estimate the association between mental
labor and the two ESM emotional well-being outcomes (i.e., positive affect and neg-
ative affect), while controlling for familial, demographic, and occupational charac-
teristics. The aggregated measure of time spent in unpaid work is also used as a
control variable in these models. The major advantage of hierarchical linear model-
ing is that it accounts for the nonindependence of observations within individuals
and allows estimating the signal (within-individual) level and the person (between-
individual) level simultaneously (Raudenbush and Bryk 2002). The hierarchical lin-
ear models can be expressed with two sets of equations. The within-individual level
equation models emotional well-being as a function of the three mental labor mea-
sures:

Positive affectij ¼ b0j þ b1j(CD-ML)ij þ b2j(F-ML)ij þ b3j(J-ML)ij þ eij ð1Þ

The dependent variable is the score on emotional outcome (positive affect in
this illustration) on signal i for person j. This equation estimates positive affect
as a linear function of mental labor, where b0j is the intercept, b1j, b2j, and b3j
respectively denote the association between cross-domain, family-specific, and
job-specific mental labor on positive affect, and eij is the within-person residual.
In this equation the intercept is allowed to vary randomly across individuals.
This variation is then explained using the between-individual equation, which
estimates variation across respondents in b0j as a function of covariates Z:

b0j ¼ c00 þ c01Z1j þ c02Z2j þ . . . þ c0kZkj þ m0j ð2Þ

In this equation c00 is the average score on positive affect, c01 to c0k are the regres-
sion coefficients of the estimated effects of the covariates Z (i.e., familial, demo-
graphic, and occupational controls) on the adjusted positive affect score, and m0j is
the person-level error term assumed to be normally distributed with mean zero and
unknown variance.

In the next stage, to test whether the association between emotional well-being
and mental labor varies by context, or type of activity (i.e., paid work, unpaid work,
and free time), interaction terms between the mental labor and activity measures
were added to the model. Free time is used as the omitted category. These models
made it possible to examine how parents feel when they experience mental
spillovers.

926 Offer

RESULTS

The Prevalence, Content, and Context of Mental Labor

The descriptive results presented in Table II indicate that overall mothers were
more likely than fathers to engage in mental labor. Mothers reported engaging in
mental labor in 26.14% of the signals, the equivalent of slightly more than one-
fourth of their waking time, whereas fathers reported doing so in about one-fifth of
their waking time (21.46% of signals). An examination by type of mental labor
showed that compared to fathers, mothers were more likely to engage in cross-
domain and family-specific mental labor and less likely to engage in job-specific
mental labor. It should be noted, however, that although significant, these differ-
ences were small in size (3% or lower).

Table II further shows the within-mental labor distribution of signals. These
results indicate that the share of cross-domain mental labor out of all mental labor
signals was slightly higher among mothers than fathers (53.8% and 48.5%, respec-
tively). The share of family-specific mental labor signals out of all mental labor sig-
nals, however, was similar for mothers and fathers. Stated differently, when
engaging in mental labor both mothers and fathers spent about 30% of their time
thinking about family matters. This result, together with the finding that the gender
difference in time spent in family-specific mental labor throughout the day was
extremely small (less than 3%) fails to support Hypothesis 1. By contrast, fathers
spent a significantly larger proportion of their mental labor time thinking about
job-related matters (23.1% as compared to 14.6% among mothers). These findings
lend some support to Hypothesis 2.

Next, to examine whether mental labor varies by context and estimate the
extent of mental spillover between domains, I cross-tabulated the three mental labor
measures by the type of activity. As Table III shows, mothers engaged in family-
specific mental labor while doing unpaid work slightly more often than fathers
(13.7% of all mental labor signals among mothers compared to 9.3% among

Table II. Gender differences in the distribution of mental labor signals: percentages

Mothers Fathers

Mental labor (out of all signals) 26.14 (12.64) 21.46 (12.79)***
Type of mental labor (out of all signals)
Cross-domain mental labor 13.77 (9.19) 10.24*** (8.07)
Family-specific mental labor 8.36 (7.07) 5.92*** (5.89)
Job-specific mental labor 4.00 (5.26) 5.29** (7.07)

Total 26.13 21.45
Type of mental labor (out of all mental labor signals)
Cross-domain mental labor 53.8 48.5**
Family-specific mental labor 31.6 28.4
Job-specific mental labor 14.6 23.1***

Total 100 100

Note: N = 402 mothers with 16,451 signals; 291 fathers with 11,322 signals. All the estimates are calcu-
lated at the aggregated person-level.
*p < .05; **p < .01; ***p < .001 (two-tailed tests).

Costs of Thinking About Work and Family 927

fathers), but mothers and fathers were equally likely to engage in family-specific
mental labor while doing paid work (5.7%) and during free-time activities (about
13%). Furthermore, no gender differences were found for cross-domain mental
labor in the three contexts. Hypothesis 3, which posited that cross-domain and fam-
ily-specific mental labor would spill over more frequently into mothers’ than into
fathers’ paid work and free time was thus not supported.

The results did not support Hypothesis 4 either. Table III indicates that moth-
ers engaged more frequently in job-specific mental labor while doing unpaid work
than fathers (16.1% and 9.4% among mothers and fathers, respectively) and during
free-time activities (18.1% and 15.6% among mothers and fathers, respectively).
These results suggest that mothers rather than fathers, as expected in Hypothesis 4,
experience more mental work-to-family spillover.

Mental Work–Family Spillover and Emotional Well-Being

Table IV presents the results of the hierarchical linear models predicting the
two ESM emotional well-being outcomes. As Models 1 and 5, respectively, show
cross-domain mental labor (CD-ML) was associated with lower positive affect
among both mothers (b = –0.04, p < .01) and fathers (b = –0.04, p < .01). This associ- ation did not vary by context for mothers as none of the coefficients of CD-ML in the model with interaction effects (Model 2) was significant. For fathers, however, Model 6 suggests that this association was derived from paid work (b = 0.03 – 0.11 = –.08, p < .001). Stated differently, fathers reported lower positive affect when they engaged in cross-domain mental labor in the context of paid work only. Thinking about tasks to be done and time constraints during paid work appeared to be a negative experience for both mothers and fathers. As Models 4 and 8 indicate both mothers and fathers reported higher negative affect when they engaged in

Table III. Gender differences in the distribution of mental labor signals by type of mental labor and
main activity: percentages (out of all mental labor signals)

Mothers Fathers

Cross-domain mental labor
Paid work 10.0 16.6
Unpaid work 2.2 1.9
Free-time activities 3.1 5.8

Family-specific mental labor
Paid work 5.7 5.7
Unpaid work 13.7 9.3***
Free-time activities 13.4 13.1

Job-specific mental labor
Paid work 17.8 22.7***
Unpaid work 16.1 9.4***
Free-time activities 18.1 15.6***

Total 100 100

Note: N = 402 mothers with 16,451 signals; 291 fathers with 11,322 signals.
*p < .05; **p < .01; ***p < .001 (two-tailed tests).

928 Offer

T
a
b
le

IV
.

H
ie
ra
rc
h
ic
a
l
li
n
ea
r
m
o
d
el
in
g
re
su
lt
s
p
re
d
ic
ti
n
g
p
a
re
n
ts
’
em

o
ti
o
n
a
l
w
el
l-
b
ei
n
g
w
it
h
m
en
ta
l
la
b
o
r,
m
a
in

a
ct
iv
it
y
,
a
n
d
co
n
tr
o
ls

M
o
th
er
s

F
a
th
er
s

P
o
si
ti
v
e
a
ff
ec
t

N
eg
a
ti
v
e
a
ff
ec
t

P
o
si
ti
v
e
a
ff
ec
t
N
eg
a
ti
v
e
a
ff
ec
t

M
o
d
el
1

M
o
d
el
2

M
o
d
el
3

M
o
d
el
4

M
o
d
el
5

M
o
d
el
6

M
o
d
el
7

M
o
d
el
8

b
(
S
E
)

b
(
S
E
)
b
(
S
E
)
b
(
S
E
)
b
(
S
E
)
b
(
S
E
)
b
(
S
E
)
b
(
S
E
)

S
ig
n
a
l-
le
v
el
v
a
ri
a
b
le
s

C
D
-M

L
–

.0
4
*
*
(.
0
1
)

–

.0
4
(.
0
2
)

.0
4
*
*
(.
0
1
)

.0
0
(.
0
2
)

–
.0
4
*
*
(.
0
1
)

.0
3
(.
0
3
)

.0
2
(.
0
1
)

–

.0
1
(.
0
2
)

F
-M

L
–
.0
4
*
(.
0
2
)

–

.0
3
(.
0
2
)

.0
5
*
*
*
(.
0
2
)

.0
4
(.
0
2
)
.0
1
(.
0
2
)

–

.0
1
(.
0
3
)

–
.0
3
(.
0
2
)

–
.0
5
(.
0
3
)

J-
M
L

–
.0
4
(.
0
2
)

–
.0
1
(.
0
5
)

.0
2
(.
0
2
)

.0
2
(.
0
4
)

–
.0
2
(.
0
2
)

.0
3
(.
0
4
)

.0
3
(.
0
2
)

–

.0
7

(.
0
5
)

C
D
-M

L
9

p
a
id

w
o
rk

–

.0
2
(.
0
3
)

.0
7
*
(.
0
3
)

–
.1
1
*
*
*
(.
0
3
)

.0
7
*
*
(.
0
3
)

C
D
-M
L
9

u
n
p
a
id

w
o
rk
.0
1
(.
0
3
)

.0
4
(.
0
3
)

–

.0
4
(.
0
4
)

–
.0
1
(.
0
3
)

F
-M
L
9
p
a
id
w
o
rk

–
.0
2
(.
0
5
)

.0
6
(.
0
4
)

–

.0
1
(.
0
6
)

.1
0

(.
0
6
)

F
-M
L
9
u
n
p
a
id
w
o
rk

.0
0
(.
0
3
)

.0
2
(.
0
3
)

.0
5
(.
0
5
)

.0
0
(.
0
4
)

J-
M
L
9

p
a
id
w
o
rk

–

.0
3
(.
0
5
)

–
.0
6
(.
0
5
)

.1
3
*
*
(.
0
5
)

J-
M
L
9
u
n
p
a
id
w
o
rk
.0
1
(.
0
6
)

–
.0
8
(.
0
6
)

–

.0
9

(.
0
7
)

.1
0
(.
0
9
)

In
te
rc
ep
t

1
.8
9
*
*
*
(.
0
6
)

.5
5
*
*
*
(.
0
5
)

2
.0
4
*
*
*
(.
0
8
)

.2
9
*
*
*
(.
0
7
)

P
er
so
n
-l
ev
el
v
a
ri
a
b
le
s

E
ff
ec
ts
o
n
in
te
rc
ep
t

O
n
e–
tw

o
ch
il
d
re
n

.0
1
(.
0
3
)
.0
1
(.
0
3
)
.0
1
(.
0
3
)
.0
1
(.
0
3
)

.0
1
(.
0
4
)

–
.0
1
2
(.
0
3
5
)

–
.0
5
(.
0
3
)

Y
o
u
n
g
es
t
ch
il
d
< 2

.0
4
(.
0
6
)

–
.1
3
*
(.
0
5
)

–
.1
3
(.
0
8
)

.0
4
(.
0
8
)

Y
o
u
n
g
es
t
ch
il
d
2
–
6

.0
8
(.
0
5
)

–
.0
8
*
(.
0
4
)

–
.0
8
*
(.
0
4
)
–
.0
7
(.
0
5
)
–
.0
7
(.
0
5
)
.0
2
(.
0
4
)
.0
3
(.
0
4
)

A
g
e

.0
0
(.
0
0
)

–
.0
0
(.
0
0
)

–
.0
0
(.
0
0
)
–
.0
0
(.
0
0
)
–
.0
0
(.
0
0
)

–
.0
1
(.
0
0
)

–
.0
0
(.
0
0
)

G
ra
d
u
a
te

d
eg
re
e

–
.0
1
(.
0
5
)
–
.0
1
(.
0
5
)

–
.0
7
(.
0
4
)

-0
6
(.
0
6
)

–
.0
6
(.
0
6
)

.1
2
*
*
(.
0
5
)

T
im

e
in

u
n
p
a
id
w
o
rk

–
.1
0
(.
1
4
)

–
.1
1
(.
1
4
)

–
.1
6
(.
1
2
)

.1
6
(.
1
8
)

.1
5
(.
1
8
)

–
.5
6
*
*
*
(.
1
7
)

–
.5
5
*
*
(.
1
7
)

W
o
rk

h
o
u
rs

(3
8
–
5
0
)

–
.0
1
(.
0
4
)

–
.0
1
(.
0
4
)
–
.0
1
(.
0
3
)
–
.0
1
(.
0
3
)

–
.0
4
(.
0
6
)

–
.0
4
(.
0
6
)
–
.0
1
(.
0
5
)
–
.0
1
(.
0
5
)
W
o
rk
h
o
u
rs

(5
1
+
)

–
.0
2
(.
0
8
)

–
.0
4
(.
0
5
)

–
.0
5
(.
0
5
)

–
.0
1
(.
0
7
)

–
.0
3
(.
0
6
)

R
eg
u
la
r
w
o
rk

sc
h
ed
u
le

.0
4
(.
0
4
)
.0
4
(.
0
4
)

–
.0
7
*
(.
0
3
)

–
.0
7
*
(.
0
3
)
–
.0
1
(.
0
5
)
–
.0
1
(.
0
5
)
.0
3
(.
0
4
)
.0
3
(.
0
4
)

Jo
b
a
u
to
n
o
m
y

.0
5
*
(.
0
2
)

–
.0
0
(.
0
2
)

.0
6
*
*
(.
0
2
)

.0
6
*
*
(.
0
2
)
–
.0
0
(.
0
2
)
–
.0
0
(.
0
2
)

Jo
b
st
re
ss

–
.1
4
*
*
*
(.
0
2
)

.1
2
*
*
*
(.
0
2
)

–
.0
9
*
*
*
(.
0
2
)

.1
4
*
*
*
(.
0
3
)

.1
5
*
*
*
(.
0
3
)

V
a
ri
a
n
ce

co
m
p
o
n
en
ts

S
ig
n
a
l-
le
v
el
(l
ev
el
1
)

.3
0

.2
4

.2
6

.2
1

P
er
so
n
-l
ev
el
(l
ev
el
2
)

.1
0
.0
7
.0
9
.0
7

N
o
te
:
C
D
-M

L
=
cr
o
ss
-d
o
m
a
in

m
en
ta
l
la
b
o
r,
F
-M

L
=
fa
m
il
y
-s
p
ec
ifi
c
m
en
ta
l
la
b
o
r,
J-
M
L
=
jo
b
-s
p
ec
ifi
c
m
en
ta
l
la
b
o
r.
N

=
4
0
2
m
o
th
er
s
w
it
h
1
6
,4
5
1
si
g
n
a
ls
;
2
9
1
fa
th
er
s
w
it
h

1
1
,3
2
2
si
g
n
a
ls
.

*
p
< .0 5 ; * * p < .0 1 ; * * * p < .0 0 1 (t w o -t a il ed

te
st
s)
.

Costs of Thinking About Work and Family 929

cross-domain mental labor during paid work (b = 0.00 + 0.07 = 0.07, p < .05 among mothers; and b = –0.01 + 0.07 = 0.06, p < .01 among fathers). By contrast, cross- domain mental labor during unpaid work or free-time activities was not signifi- cantly associated with either positive affect or negative affect.

Hypothesis 5 posited that engaging in family-specific mental labor during paid
work or free-time activities would be negatively associated with emotional well-
being and that these associations would be more pronounced among mothers than
fathers. The results presented in Table IV provide partial support to this hypothe-
sis. Family-specific mental labor (F-ML) was overall a negative experience for
mothers, regardless of context. It was associated with lower positive affect in Model
1 (b = –0.04, p < .05) and higher negative affect in Model 3 (b = 0.05, p < .001). Mod- els 2 and 4, however, indicate that these associations did not depend on the type of activity because none of the interaction terms was significant. By contrast, family- specific mental labor was not significantly related to either positive affect or negative affect in fathers (see Models 6 and 8).

According to Hypothesis 6, the spillover of job-related thoughts into unpaid
work and free-time activities was expected to be detrimental to emotional well-
being. The results, however, indicate that no significant associations were found
between job-specific mental labor (J-ML) in the contexts of unpaid work or free
time and the two emotional well-being outcomes in either mothers or fathers.
Thinking about job-related matters during paid work was associated with higher
negative affect in fathers only (b = –0.07 + 0.13 = 0.06, p < .01 in Model 8).

DISCUSSION AND CONCLUSION

Previous research has shown that although men and women in dual-earner
families have relatively similar workloads, that is, they spend approximately the
same amount of time on paid and unpaid labor combined (Bianchi et al. 2006;
Bittman and Wajcman 2000), employed mothers experience substantially greater
burden and emotional stress than employed fathers (Arendell 2001; Hochschild
1989; Mattingly and Sayer 2006). In their attempt to explain this contradiction,
scholars have argued that previous assessments of parents’ distribution of time have
not adequately measured tasks and demands (Coltrane 2000; Nippert-Eng 1996;
Twigges, McQuillan, and Ferree 1999; Walzer 1996). One of the aspects unac-
counted for in these studies is the mental dimension of tasks and demands, which
was referred to here as mental labor. Mental labor is difficult to tap in quantitative
research because it does not necessarily translate into physical activities that can be
estimated using conventional time diary methods. The major contribution of the
present study was that by using the ESM, mental labor could be tabulated and char-
acterized. Hence the findings shed light on a relatively underexplored aspect of the
gendered division of labor among employed families.

Results showed that mothers were more likely than fathers to engage in mental
labor, but the gender gap in mental labor was overall small. Mothers engaged in
mental labor in about one-fourth, and fathers in one-fifth, of their waking time.
Nevertheless, several important differences in the content, context, and emotional

930 Offer

implications of mental labor by gender were revealed. Consistent with the idea that
breadwinning is an important issue for men (Christiansen and Palkovitz 2001;
Gerson 2010; Townsend 2002), results showed that when they engaged in mental
labor, fathers spent more time thinking about job-related matters than mothers.
However, and contrary to expectations, these thoughts were less likely to spill over
into fathers’ than into mothers’ nonwork domains. It may be that fathers are more
adept than mothers at leaving their job-related concerns behind when they leave
work. But interestingly, these results diverge from previous research showing that
fathers were disturbed by work matters while at home more often than mothers
(Glavin et al. 2011). An alternative explanation relates to the specific characteristics
of the sample used in this study. The percentage of fathers working long hours in
this sample was very high; nearly 60% reported working 46 or more hours per week.
Thus the especially long hours that these fathers spend at work may leave them with
little time to think about job-related matters outside of work or prompt them to
reserve the little time they have left to think about other, nonwork-related things.

By contrast, job-specific mental labor among mothers was distributed almost
evenly across the three contexts of paid work, unpaid work, and free-time activities,
suggesting that their thoughts about paid work easily spilled over into other
domains. These findings could reflect what Mennino and Brayfield (2005:110), who
also found women to experience work-to-home spillover more often, refer to as
women’s greater awareness “of the ways in which employment encroaches on family
life.” Women may be more aware of and affected by these influences because, unlike
men, the obligations entailed by their work and family roles are often at odds with
each other (Garey 1999; Simon 1995). Working mothers, especially among the mid-
dle class, are expected to be fully committed to their job and at the same time inten-
sively engaged with their children (Bianchi et al. 2006; Blair-Loy 2003). These
“competing devotions” require mothers to develop strategies for reconciling work
and family demands. Because mothers are the ones who usually alter and adapt
their work schedule to meet family demands, such as leaving work early to pick up
children from day care or staying at home with a sick child (Gerson 2010; Jacobs
and Gerson 2004; Keene and Reynolds 2005), they may feel that they “have to catch
up” on work and as a result are easily preoccupied with job-related matters outside
of the workplace.

Another gender difference pertains not to the quantitative but qualitative expe-
rience of family-specific mental labor. Contrary to expectations, the results showed
that the gender difference in the amount of time parents spent thinking about family
matters throughout the day was negligible. Moreover, when engaging in mental
labor, mothers and fathers were equally likely to think about family matters and the
extent of spillover of thoughts about family matters into the other domains was sim-
ilar for mothers and fathers. Echoing the cultural image of “new fathers” and con-
sistent with previous studies, these findings suggest that fathers in dual-earner
middle-class families are involved in those aspects of child care that pertain to
parental responsibility and are closely linked to the concept of mental labor, namely
participation in decision making about and planning for their children (Doucet
2006; Pleck 2010). However, the emotional implications of family-specific mental
labor were substantially different for mothers and fathers. Only among mothers was

Costs of Thinking About Work and Family 931

this type of mental labor detrimental to emotional well-being. Regardless of con-
text, thinking about family matters was significantly associated with lower positive
affect and higher negative affect in mothers. By contrast, thinking about family mat-
ters was not related to fathers’ emotional well-being. It appears that even if fathers
are as preoccupied with family matters as mothers are, this type of mental labor is a
source of emotional burden and stress for mothers alone.

In light of previous studies it could be expected that these fathers, who work
particularly long hours, would have guilt feelings about not spending enough time
with their children and not meeting the cultural expectations of involved fatherhood
(Daly 1996), but the results of this study do not seem to support this idea. Instead,
what they suggest is that despite the increase in fathers’ involvement at home, moth-
ers continue to be cast as the bearers of the moral responsibility for the quality of
family life and the well-being of children (Arendell 2001; Coltrane 1996; Hochschild
1989; Townsend 2002). Mothers are viewed and judged on the basis of a specific
moral standpoint that defines what women ought to be doing and thinking as pri-
mary caregivers (Doucet 2006). Worrying, it is argued, allows mothers to express
moral responsibility and gain social acceptability. Hence, mothers are more likely
than fathers to fear that they will be judged adversely by others in their family and
community if they do not behave in ways that conform to the ideal of intensive par-
enting (Doucet 2006; Ruddick 1995; Walzer 1996).

Notwithstanding these differences, several gender similarities are noteworthy.
First, thinking about things to be done and time constraints during paid work
appeared to be detrimental to the emotional well-being of both mothers and fathers.
It should be noted, however, that the level of resolution of the ESM was not high
enough to know what exactly parents had in mind when they reported cross-domain
mental labor. This is an important limitation of the present study because the mea-
sure of general mental labor did not allow distinguishing between the family and
paid work domains. It was therefore impossible to determine, for example, whether
parents were concerned about family (e.g., scheduling a doctor’s appointment for
the child) or job-related matters (e.g., returning e-mails to clients or colleagues)
when they reported about “things to do.” Nevertheless, despite being a coarse mea-
sure, the high prevalence of this type of mental labor among both mothers and
fathers provides yet another important indication of the time-squeezed and rushed
pace of life of employed parents in contemporary dual-earner families (Bianchi
et al. 2006; Jacobs and Gerson 2004; Mattingly and Sayer 2006; Roxburgh 2012).

Second, among both mothers and fathers paid work was found to be overall
relatively insulated from thoughts about family matters. Both mothers and fathers
reported thinking about family matters during paid labor in less than 6% of all
mental labor signals.

The low level of mental family-to-work spillover found in this study is indica-
tive of a high degree of role segmentation (Ashforth et al. 2000; Olson-Buchanan
and Boswell 2006) in the context of paid work, where both mothers’ and fathers’
job rarely tends to be interrupted by concerns associated with their role as a parent
or spouse. These findings resonate with Nippert-Eng’s (1996:39) observation that
our current culture is dominated by a segmentist approach to home and work,
which requires us to “reserve more of our home-related thoughts, activities, people,

932 Offer

and objects for more private space and time.” It appears that both the mothers and
fathers in this study managed to create clear boundaries around paid work and pre-
clude its infringement by nonwork-related concerns and thus conformed to the
norm of the ideal worker as someone who can dedicate all his/her attention to work
while working.

While this is not surprising for fathers, the finding that paid work was highly
insulated from family-related thoughts among mothers is intriguing because previ-
ous research has shown that mothers were substantially more likely than fathers to
have their paid work disturbed by family matters (Mennino and Brayfield 2005;
Nippert-Eng 1996; Roehling et al. 2003). It is also striking that mental family-to-
work spillover was not significantly associated with well-being for either mothers or
fathers. These results could be explained by the particular characteristics of the sam-
ple. Recall that the vast majority of the mothers included in this study were highly
educated and employed in demanding professional and managerial positions. These
educational and occupational characteristics may have led to the attenuation of
gender differences reported elsewhere.

Third, the scope and emotional experience of mental spillovers into free-time
activities were overall similar by gender. Findings showed that mothers and fathers
were equally likely to think about family matters during free-time activities and that
this type of mental intrusion was not associated with either mothers’ or fathers’
emotional well-being. These findings, which paint a slightly different picture than
the one portrayed in previous research showing that women’s free-time activities
were more often contaminated by nonleisure, mainly family-related, activities and
demands than fathers’ free-time activities (Bittman and Wajcman 2000; Mattingly
and Bianchi 2003), call for further investigation of this issue in future research.

Altogether, the present study reveals several important gender similarities with
respect to mental labor, most notably, that fathers and mothers are equally likely to
think about family matters. However, it also highlights the challenge that mothers
in dual-earner middle-class families experience as they struggle to juggle multiple
responsibilities at home and work. The findings suggest that mental labor may con-
tribute to gender inequality among these families mainly because mothers’ concerns
about their job tend to intrude into their other domains and because mothers
endure the emotional repercussions of engaging in mental labor that revolves
around family matters.

REFERENCES

Allen, Tammy D., David E. L. Herst, Carly S. Bruck, and Martha Sutton. 2000. “Consequences Associ-
ated With Work-to-Family Conflict: A Review and Agenda for Future Research.” Journal of Occupa-
tional Health Psychology 5: 2: 278–308.

Anderson, Deborah J., Melissa Binder, and Kate Krause. 2002–2003. “The Motherhood Wage Penalty
Revisited: Experience, Heterogeneity, Work Effort, and Work-Schedule Flexibility.” Industrial and
Labor Relations Review 56: 2: 273–294.

Arendell, Teresa. 2001. “The New Care Work of Middle Class Mothers: Managing Childrearing,
Employment, and Time.” In Kerry J. Daly (ed.), Minding the Time in Family Experience: Emerging
Perspectives and Issues, Volume 3: pp. 163–204. Oxford, UK: JAI.

Ashforth, Blake E., Glen E. Kreiner, and Mel Fugate. 2000. “All in a Day’s Work: Boundaries and
Micro Role Transitions.” Academy of Management Review 25: 3: 472–491.

Costs of Thinking About Work and Family 933

Bakker, Arnold B. and Evangelia Demerouti. 2007. “The Job-Demands-Resources Model: State of the
Art.” Journal of Managerial Psychology 22: 3: 309–328.

Bakker, Arnold B. and Sabine A. E. Geurts. 2004. “Toward a Dual-Process Model of Work-Home Inter-
ference.” Work and Occupations 31: 3: 345–366.

Becker, Penny E. and Phyllis Moen. 1999. “Scaling Back: Dual-Earner Couples’ Work-Family Strate-
gies.” Journal of Marriage and Family 61: 4: 995–1007.

Bellavia, Gina M. and Michael Frone. 2005. “Work-Family Conflict.” In Julian Barling, Kevin E.
Kelloway and Michael R. Frone (eds.), Handbook of Work Stress: pp. 113–148. Thousand Oaks, CA:
Sage.

Bianchi, Suzanne M., John P. Robinson, and Melissa A. Milkie. 2006. The Changing Rhythm of American
Family Life. New York: Russell Sage Foundation.

Bittman, Michael and Judy Wajcman. 2000. “The Rush Hour: The Character of Leisure Time and Gen-
der Equity.” Social Forces 79: 1: 165–189.

Blair-Loy, Mary. 2003. Competing Devotions. Cambridge, MA: Harvard University Press.
Budig, Michelle J. and Paula England. 2001. “The Wage Penalty for Motherhood.” American Sociologi-

cal Review 66: 2: 204–225.
Butler, Adam B. and Amie Skattebo. 2004. “What Is Acceptable for Women May Not Be for Men: The

Effect of Family Conflicts With Work on Job-Performance Ratings.” Journal of Occupational and
Organizational Psychology 77: 4: 553–564.

Byron, Kristin. 2005. “A Meta-Analytic Review of Work-Family Conflict and Its Antecedents.” Journal
of Vocational Behavior 67: 169–198.

Cabrera, Natasha J., Catherine Tamis-LeMonda, Robert H. Bradley, Sandra Hofferth, and Michael E.
Lamb. 2000. “Fatherhood in the Twenty-First Century.” Child Development 71: 1: 127–136.

Chesley, Noelle. 2005. “Blurring Boundaries? Linking Technology Use, Spillover, Individual Distress,
and Family Satisfaction.” Journal of Marriage and Family 67: 5: 1237–1248.

Christensen, Kathleen and Barbara Schneider. 2010. “Introduction: Evidence of the Worker and Work-
place Mismatch.” In Kathleen Christensen and Barbara Schneider (eds.), Workplace Flexibility:
Realigning 20th-Century Jobs for a 21st-Century Workforce: pp. 1–14. Ithaca, NY: Cornell University
Press.

Christiansen, Shawn L. and Rob Palkovitz. 2001. “Why the ‘Good Provider’ Role Still Matters: Provid-
ing as a Form of Paternal Involvement.” Journal of Family Issues 22: 1: 84–106.

Clark, Sue. 2000. “Work/Family Border Theory: A New Theory of Work/Family Balance.” Human Rela-
tions 53: 6: 747–770.

Coltrane, Scott. 1996. Family Man: Fatherhood, Housework, and Gender Equity. New York: Oxford Uni-
versity Press.

Coltrane, Scott. 2000. “Research on Household Labor: Modeling and Measuring the Social Embedded-
ness of Routine Family Work.” Journal of Marriage and Family 62: 4: 1208–1233.

Csikszentmihalyi, Mihaly and Reed Larson. 1987. “Validity and Reliability of the Experience Sampling
Method.” The Journal of Nervous and Mental Disease 175: 9: 526–536.

Daly, J. Kerry. 1996. “Spending Time With the Kids: Meanings of Family Time for Fathers.” Family
Relations 45: 4: 466–476.

Daly, J. Kerry. 2002. “Time, Gender, and the Negotiation of Family Schedules.” Symbolic Interaction 25:
3: 323–342.

Darrah, Charles N., James M. Freeman, and J. A. English-Lueck. 2007. Busier Than Ever: Why American
Families Can’t Slow Down. Stanford, CA: Stanford University Press.

Deutsch, Francine M. and Susan E. Saxon. 1998. “The Double Standard of Praise and Criticism for
Mothers and Fathers.” Psychology of Women Quarterly 22: 4: 665–683.

DeVault, Marjorie L. 1991. Feeding the Family: The Social Organization of Caring as Gendered Work.
Chicago: University of Chicago Press.

DeVault, Marjorie L. 1999. “Comfort and Struggle: Emotion Work in Family Life.” The Annals of the
American Academy of Political and Social Science 561: 52–63.

Dilworth, Jennie E. Long. 2004. “Predictors of Negative Spillover From Family to Work.” Journal of
Family Issues 25: 2: 241–261.

Doucet, Andrea. 2006. Do Men Mother? Fathering, Care, and Domestic Responsibility. Toronto: Univer-
sity of Toronto Press.

Edin, Kathryn and Laura Lein. 1997. Making Ends Meet: How Single Mothers Survive Welfare and Low-
Wage Work. New York: Russell Sage Foundation.

Furstenberg, Frank F. Jr, Thomas D. Cook, Jacquelynne EcclesGlen H. Elder Jr, and Arnold Sameroff.
1999. Managing to Make It: Urban Families and Adolescent Success. Chicago: University of Chicago
Press.

Garey, Anita I. 1999. Weaving Work and Motherhood. Philadelphia: Temple University Press.

934 Offer

Gerson, Kathleen. 2009. “Changing Lives, Resistant Institutions: A New Generation Negotiates Gender,
Work, and Family Change.” Sociological Forum 24: 4: 735–753.

Gerson, Kathleen. 2010. The Unfinished Revolution: How a New Generation Is Reshaping Family, Work,
and Gender in America. New York: Oxford University Press.

Glavin, Paul, Scott Schieman, and Sarah Reid. 2011. “Boundary-Spanning Work Demands and Their
Consequences for Guilt and Psychological Distress.” Journal of Health and Social Behavior 52: 1: 43–
57.

Greenhaus, Jeffrey H. and Nicholas J. Beutell. 1985. “Source of Conflict Between Work and Family
Roles.” Academy of Management Journal 10: 1: 76–88.

Grzywacz, Joseph G., David M. Almeida, and Daniel A. McDonald. 2002. “Work-Family Spillover and
Daily Reports of Work and Family Stress in the Adult Labor Force.” Family Relations 51: 1: 28–36.

Grzywacz, Joseph G. and Nadine F. Marks. 2000. “Reconceptualizing the Work-Family Interface: An
Ecological Perspective on the Correlates of Positive and Negative Spillover Between Work and Fam-
ily.” Journal of Occupational Health Psychology 5: 1: 111–126.

Hektner, Joel M., Jennifer A. Schmidt, and Mihaly Csikszentmihalyi. 2007. Experience Sampling
Method: Measuring the Quality of Everyday Life. Thousand Oaks, CA: Sage.

Hessing, Melody. 1994. “More Than Clockwork: Women’s Time Management in Their Combined
Workloads.” Sociological Perspectives 37: 4: 611–633.

Hochschild, Arlie R. 1989. The Second Shift. New York: Viking Penguin.
Honaker, James, Gary King, and Matthew Blackwell. 2009. Amelia II: A Program for Missing Data

(Version 1.2-6). [Computer software]. Cambridge, MA: Harvard University.
Jacobs, Jerry A. and Kathleen Gerson. 2004. The Time Divide: Work, Family, and Gender Inequality.

Cambridge, MA: Harvard University Press.
Keene, Jennifer R. and John R. Reynolds. 2005. “The Job Costs of Family Demands.” Journal of Family

Issues 26: 3: 275–299.
Korabik, Karen, Allyson McElwain, and Dara B. Chappell. 2008. “Integrating Gender-Related Issues

Into Research on Work and Family.” In Karen Korabik, Lero, Donna S. and Whitehead, Denise L.
(eds.), Handbook of Work-Family Integration: pp. 215–232. London: Elsevier.

Kurz, Demie. 2002. “Caring for Teenage Children.” Journal of Family Issues 23: 6: 748–767.
Martinengo, Giuseppe, Jenet I. Jacob, and Jeffrey Hill. 2010. “Gender and the Work-Family Interface:

Exploring Differences Across the Family Life Course.” Journal of Family Issues 31: 10: 1363–1390.
Mattingly, Marybeth J. and Suzanne M. Bianchi. 2003. “Gender Differences in the Quantity and Quality

of Free Time: The U.S. Experience.” Social Forces 81: 3: 999–1030.
Mattingly, Marybeth J. and Liana C. Sayer. 2006. “Under Pressure: Gender Differences in the Relation-

ship Between Free Time and Feeling Rushed.” Journal of Marriage and Family 68: 1: 205–221.
Mederer, Helen J. 1993. “Division of Labor in Two-Earner Homes: Task Accomplishment Versus

Household Management as Critical Variables in Perceptions About Family Work.” Journal of Mar-
riage and the Family 55: 1: 133–145.

Mennino, Sue F. and April Brayfield. 2005. “Job-Family Trade-Offs.” Work and Occupations 29: 2: 226–
256.

Milliken, Frances J. and Linda M. Dunn-Jensen. 2005. “The Changing Time Demands of Managerial
and Professional Work: Implications for Managing the Work-Life Boundary.” In Ellen Ernst Kossek
and Susan J. Lambert (eds.), Work and Life Integration: Organizational, Cultural, and Individual Per-
spectives: pp. 43–59. Mahwah, NJ: Erlbaum.

Nelson, Margaret K. 2011. Parenting Out of Control: Anxious Parents in Uncertain Times. New York:
New York University Press.

Nippert-Eng, Christina E. 1996. Home and Work: Negotiating Boundaries Through Everyday Life. Chi-
cago: University of Chicago Press.

Olson-Buchanan, Julie B. and Wendy R. Boswell. 2006. “Blurring Boundaries: Correlates of Integration
and Segmentation Between Work and Nonwork.” Journal of Vocational Behavior 68: 432–445.

Pleck, Joseph A. 2010. “Paternal Involvement: Revised Conceptualization and Theoretical Linkages
With Child Outcomes.” In M. E. Lamb (ed.), The Role of the Father in Child Development: pp. 67–107.
Hoboken, NJ: Wiley.

Ranson, Gillian. 2012. “Men, Paid Employment and Family Responsibilities: Conceptualizing the
‘Working Father’.” Gender, Work and Organization 19: 6: 741–761.

Raudenbush, Stephen W. and Anthony S. Bryk. 2002. Hierarchical Linear Models. Thousand Oaks, CA:
Sage.

Roehling, Patricia, Phyllis Moen, and Rosemary Batt. 2003. “Spillover.” In Phyllis Moen (ed.), It’s About
Time: Couples and Careers: pp. 101–121. Ithaca, NY: Cornell University Press.

Roxburgh, Susan. 2012. “Parental Time Pressures and Depression Among Married Dual-Earner Par-
ents.” Journal of Family Issues 33: 8: 1027–1050.

Costs of Thinking About Work and Family 935

Ruddick, Sara. 1995. Maternal Thinking: Towards a Politics of Peace. Boston: Beacon.
Sayer, Liana C., Suzanne M. Bianchi, and John P. Robinson. 2004. “Are Parents Investing Less in Chil-

dren? Trends in Mothers’ and Fathers’ Time With Children.” American Journal of Sociology 110: 1: 1–
43.

Schieman, Scott, Melissa A. Milkie, and Paul Glavin. 2009. “When Work Interferes With Life: Work-
Nonwork Interference and the Influence of Work-Related Demands and Resources.” American Socio-
logical Review 74: 6: 966–988.

Schneider, Barbara. 2006. “In the Moment: The Benefits of the Experience Sampling Method.” In M. Pitt
Catsouphes, Ellen Ernst Kossek and Stephen Sweet (eds.), The Work and Family Handbook: pp. 469–
488. Mahwah, NJ: Erlbaum.

Schneider, Barbara and Linda J. Waite. 2005. Being Together, Working Apart: Dual-Career Families and
Their Work-Life Balance. Cambridge, UK: Cambridge University Press.

Shaw, Susan M. 2001. “The Family Leisure Dilemma: Insights From Research With Canadian Families.”
World Leisure 4: 53–62.

Shaw, Susan M. 2008. “Family Leisure and Changing Ideologies of Parenthood.” Sociology Compass 2:
2: 688–703.

Simon, Robin W. 1995. “Gender, Multiple Roles, Role Meaning, and Mental Health.” Journal of Health
and Social Behavior 36: 2: 182–194.

Stevens, Daphne Pederson, Krista Lynn Minnotte, Susan E. Mannon, and Gary Kiger. 2007. “Examin-
ing the Neglected Side of the Work-Family Interface.” Journal of Family Issues 28: 2: 242–262.

Townsend, Nicholas. 2002. The Packaged Deal: Marriage, Work and Fatherhood in Men’s Lives. Philadel-
phia: Temple University Press.

Twigges, Joan E., Julia McQuillan, and Myra Marx Ferree. 1999. “Meaning and Measurement: Recon-
ceptualizing Measures of the Division of Household Labor.” Journal of Marriage and Family 61:
August: 712–724.

Voydanoff, Patricia. 2004. “The Effects of Work Demands and Resources on Work-to-Family Conflict
and Facilitation.” Journal of Marriage and Family 66: 2: 398–412.

Voydanoff, Patricia. 2005. “Consequences of Boundary-Spanning Demands and Resources for Work-to-
Family Conflict and Perceived Stress.” Journal of Occupational Health Psychology 10: 4: 491–503.

Walzer, Susan. 1996. “Thinking About the Baby: Gender and Divisions of Infant Care.” Social Problems
43: 2: 219–234.

Williams, Joan C. 2012. Reshaping the Work-Family Debate: Why Men and Class Matter. Cambridge,
MA: Harvard University Press.

936 Offer

Copyright of Sociological Forum is the property of Wiley-Blackwell 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.

38 Monthly Labor Review • June 2012

Telecommuting

The hard truth about telecommuting

Telecommuting has not permeated the American workplace, and
where it has become commonly used, it is not helpful in reducing
work-family conflicts; telecommuting appears, instead, to have
become instrumental in the general expansion of work hours,
facilitating workers’ needs for additional worktime beyond the
standard workweek and/or the ability of employers to increase or
intensify work demands among their salaried employees

Mary C. Noonan
and
Jennifer L. Glass

Mary C. Noonan is an Associate
Professor at the Department of
Sociology, The University of Iowa;
Jennifer L. Glass is the Barbara
Bush Regents Professor of Liberal
Arts at the Department of Sociol-
ogy and Population Research
Center, University of Texas at
Austin. Email: mary-noonan-1@
uiowa.edu or jennifer-glass@
austin.utexas.edu.

Telecommuting, defined here as work tasks regularly performed at home, has achieved enough
traction in the American workplace to
merit intensive scrutiny, with 24 percent
of employed Americans reporting in recent
surveys that they work at least some hours
at home each week.1 The definitions of
telecommuting are quite diverse. In this ar-
ticle, we define telecommuters as employ-
ees who work regularly, but not exclusively,
at home. In our definition, at-home work
activities do not need to be technologically
mediated nor do telecommuters need a
formal arrangement with their employer to
work at home.

Telecommuting is popular with policy
makers and activists, with proponents
pointing out the multiple ways in which
telecommuting can cut commuting time
and costs,2 reduce energy consumption
and traffic congestion, and contribute to
worklife balance for those with caregiving
responsibilities.3 Changes in the structure
of jobs that enable mothers to more effec-
tively compete in the workplace, such as
telecommuting, may be needed to finally
eliminate the gender gap in earnings and
direct more earned income to children,
both important public policy goals.4

Evidence also reveals that an increasing num-
ber of jobs in the American economy could be
performed at home if employers were willing
to allow employees to do so.5 Often, employees
can perform jobs at home without supervision
in the “high-tech” sector, in the financial sector,
and many in the communication sector that are
technology dependent. The obstacles or barriers
to telecommuting seem to be more organiza-
tional, stemming from the managers’ reluctance
to give up direct supervisory control of workers
and from their fears of shirking among workers
who telecommute.6

Where the impact of telecommuting has
been empirically evaluated, it seems to boost
productivity, decrease absenteeism, and increase
retention.7 But can telecommuting live up to its
promise as an effective work-family policy that
helps employees meet their nonwork responsi-
bilities? To do so, telecommuting needs to be
both (1) widely used by workers who most need
it and (2) instrumental in substituting hours at
home for hours onsite.8 Popular perceptions of
telecommuting conjure images of workers re-
placing hours worked onsite with hours more
comfortably worked at home, for mothers and
other care workers, especially. Yet, we know little
about how telecommuting in practice has be-
come institutionalized in American workplaces.

Which workers telecommute? Is telecom-

mailto:mary-noonan-1%40uiowa.edu?subject=

mailto:ennifer-glass%40austin.utexas.edu?subject=

Monthly Labor Review • June 2012 39

muting an effective strategy that lowers employees’
average hours worked onsite, or is telecommuting as-
sociated with longer average weekly work hours? To
preview our results here, we find that telecommuting
has not extensively permeated the American work-
place, and where it has become commonly used, it is
not unequivocally helpful in reducing work-family
conflicts. Instead, telecommuting appears to have be-
come instrumental in the general expansion of work
hours, facilitating workers’ needs for additional work-
time beyond the standard workweek and/or the ability
of employers to increase or intensify work demands
among their salaried employees.

We use two nationally representative data sources,
the National Longitudinal Survey of Youth (NLSY)
1979 panel (hereafter, noted as the NLSY) and special
supplements from the U.S. Census Current Population
Survey (CPS), to ascertain (1) trends over time in the
use of telecommuting among employees in the civilian
labor force, (2) who telecommutes across the population
of employees, and (3) the relationship between telecom-
muting and longer work hours among employees. These
two data sources provide information on telecommut-
ing from the mid- to late 1990s through the mid-2000s,
a period in which interest and capacity for telecom-
muting dramatically increased among U.S. businesses.
(Note that we did not use more recent data because the
Work Schedules and Work at Home May CPS supple-
ment was not fielded after 2004.)

Together, these two datasets allow us to ascertain
any general changes over time in the proportion of
employees who telecommute and the time intensity of
their telecommuting at their main job. We further disag-
gregate telecommuting hours into those hours that are
encapsulated within the 40-hour workweek (such that,
regardless of the day or time worked, these telecommut-
ing hours do not raise total work hours per week above
the statutory 40-hour threshold) and those hours that
extended the total number of hours worked per week
beyond 40. By dividing telecommuting hours into these
two categories, we are able to determine whether tele-
commuting either replaces hours that otherwise would
have been worked onsite during a standard 40-hour
workweek or expands the workweek beyond the 40 or
more hours already worked onsite.

In the following sections, we briefly describe our data
sources and measures, provide results from our analysis
of the data, and summarize the lessons learned from
investigating the implementation of telecommuting in
American workplaces.

Methods

The NLSY is a national probability sample of 12,686 women
and men living in the United States and born between 1957
and 1964. The sample was interviewed annually from 1979 to
1994 and biennially thereafter. In 1989, the NLSY began ask-
ing questions about the amount of time respondents worked at
home. To most closely match the years of the CPS supplements
(described in the next paragraph), we pool 3 years from the
NLSY for our analysis: 1998, 2002, and 2004.

The CPS is a monthly survey of about 50,000 households
representing the nation’s civilian noninstitutional population
16 years of age and over. We use data from the special Work
Schedules and Work at Home supplement to the May 1997,
2001, and 2004 CPS, which asks workers whether they worked
at home as part of their job. The advantage of the CPS data is
that, unlike the NLSY, it covers a broader age range of workers
so that we can compare a cohort similar in age with the NLSY,
as well as a younger cohort of workers who might be more
technologically sophisticated and more amenable to telecom-
muting. As such, we restrict the CPS sample to workers 22 to
40 years of age in 1997, 26 to 44 in 2001, and 29 to 47 in 2004.

We further restrict our sample to individuals who worked
at least 20 hours per week in nonagricultural jobs and who
provided valid data on all the key variables. Workers who
were self-employed or worked exclusively at home are also
excluded from the sample. The final sample sizes are 16,298
for the NLSY and 50,452 for the CPS.

Our two main variables of interest are total hours worked
per week for the main job and total hours worked per week at
home for the main job.9 We use these two measures to cre-
ate two dummy variables indicating respondents who worked
overtime (i.e., more than 40, 50, and 60 hours per week) and
who telecommuted (i.e., worked at least 1 hour at home per
week), respectively. Finally, for those respondents who tele-
commuted, we disaggregate telecommuting hours into regu-
lar telecommuting hours and overtime telecommuting hours. We
create these two variables by first creating a variable equal
to hours worked per week onsite for the main job. If total on-
site work hours are less than 40, we categorize telecommut-
ing hours that do not raise total work hours above 40 hours
as regular telecommuting hours. If total onsite work hours
equal 40 or more, we categorize all telecommuting hours as
overtime telecommuting hours. We do not know the day or
time that onsite and/or telecommuting hours were worked;
instead, in our categorization, we assume that onsite hours
are “worked first” and telecommuting hours come second.
Note that some workers reported both types of telecommut-
ing hours. For example, a worker reporting 45 total hours of
work per week, of which 10 are worked exclusively at home,

Telecommuting

40 Monthly Labor Review • June 2012

would yield 5 hours of regular telecommuting and 5 hours
of overtime telecommuting by our definitions.

Control variables include occupation (measured with
three categories: managerial/professional, sales, and
other), education (measured with four categories: less than
high school, high school diploma, some college, and col-
lege degree or higher), gender, race/ethnicity (measured
with three categories: other [White, Asian, etc.], Black,
and Hispanic), marital status (measured with three catego-
ries: never married, married, and separated/divorced/wid-
owed), parental status (dummy variable indicating whether
a child 0 to 18 years old lives in household), and age.

Finally, we create synthetic age cohorts for the CPS data
based on the age range of the NLSY sample (32 to 40 years
old in 1997). We define the older cohort for the CPS as 32 to
40 years old in 1997, 36 to 44 years old in 2001, and 39 to
47 years old in 2004. The younger cohort from the CPS, by
contrast, incorporated workers who were 22 to 29 years old
in 1997, maturing to 26 to 33 in 2001 and 29 to 36 in 2004.
These two cohorts effectively cover the career stages in which
most earnings growth occurs, from the mid-20s to late 40s.

To begin our analysis, we present trends over time in the
use of telecommuting for each sample as a whole and then
for various demographic groups. Next, we present descriptive
statistics on all variables by telecommuting status for the CPS
sample and the NLSY sample. For each sample, we perform
statistical tests to determine if differences exist between tele-
commuters and nontelecommuters. We pay special attention
to the average hours of telecommuting among telecommuters
and discuss how much telecommuting replaces onsite hours
within the first 40 hours worked and how much telecom-
muting extends the workweek beyond 40 hours. Finally, we
estimate logistic regression models to predict the likelihood
of working overtime based on telecommuting status, includ-
ing the control variables just described. Important to note is
that neither the CPS nor the NLSY provides information on
whether or not the employee has an option to telecommute.
Our regression model assumes that all workers are able to
telecommute and that “telecommuting status” is exogenous
to work hours. In reality, the ability to telecommute is likely
a function of one’s occupational type and, within occupation,
one’s performance. Both occupation and employee perfor-
mance are likely correlated with hours worked. We deal with
this endogeneity problem by controlling for occupation in
our models; data on employee performance are not available.

Results

To begin, we examine trends over time in telecommuting
for all workers and then for various demographic groups.

According to our NLSY and CPS estimates, approximately
10 percent of workers telecommuted in the mid-1990s
(chart 1). The rate of telecommuting increased slightly to
17 percent in the early 2000s and then remained constant
to the mid-2000s.10 Our results suggest that telecommut-
ing rates are not significantly different between younger
and older cohorts of workers. Furthermore, no evidence
suggests that, among telecommuters, the number of hours
spent telecommuting has increased over time (results not
shown). For the remainder of our analysis, we use a single
CPS sample, not differentiated by age (i.e., the younger
and older cohorts are pooled together).

Next, we examine how telecommuting varies by educa-
tional attainment, occupation, and parental status (chart 2).
Here, we present data from the CPS only; results from the
NLSY are similar to the CPS results. CPS results show that
parents are no more likely than the population as a whole
to telecommute, and mothers do not telecommute more
than fathers (about 17 percent for each group, results not
shown). However, college-educated workers and those in
managerial and professional occupations are significantly
more likely to telecommute than the population as a whole.

Table 1 presents descriptive statistics on our key vari-
ables by telecommuting status for both datasets, NLSY
(1998, 2002, 2004) and CPS (1997, 2001, 2004). Most no-
tably, telecommuters worked between 5 and 7 total hours
more per week than nontelecommuters. Telecommuters
were significantly less likely to work a regular schedule
(i.e., between 20 and 40 hours per week) and were more
likely to work overtime, regardless of how overtime is de-
fined (i.e., as working more than 40, 50, or 60 hours per
week).

Among telecommuters, the average number of hours
spent telecommuting each week is relatively modest, ap-
proximately 6 hours per week in both the CPS and NLSY
samples. But fully 67 percent (i.e., 4.17/6.20) of telecom-
muting hours in the NLSY and almost 50 percent (i.e.,
3.21/6.75) in the CPS occur in the overtime portion of
the weekly hours distribution (see table 1, “Hours worked
by location”). This finding suggests that telecommuting is
not being predominately used as a substitute for working
onsite during the first 40 hours worked per week.

Telecommuters are significantly more likely to have
a college degree and to work in managerial/professional
occupations compared with those who do not work at
home. Interestingly, parents are only slightly more pre-
dominant among telecommuters than nontelecommut-
ers. Telecommuters are less likely to be Black or Hispanic
and less likely to be married compared with those not
telecommuting.

Monthly Labor Review • June 2012 41

Chart 1.

Percent Percent

Percentage of workers telecommuting over time, by cohort

1997/1998 2001/2002 2004

NOTES: Younger cohort is 22–29 years old in 1997. Older cohort is 32–40 years old in 1997.
SOURCES: National Longitudinal Survey of Youth (NLSY) 1979 panel and special supplement from the U.S. Census Current Population

Survey (CPS).

40
30
20
10
0

Chart 2.

Percent Percent
50
40
30
20
10
0

Percentage of workers telecommuting over time, by education, occupation, and parental status

1997 2001 2004

50
40
30
20
10
0

SOURCE: Special supplement from the U.S. Census Current Population Survey (CPS).

NLSY, older cohort

CPS, younger cohort

CPS, older cohort

All

Managerial/professional
College educated

Parent

Telecommuting

42 Monthly Labor Review • June 2012

Table 2 presents the results of our logistic regression
models predicting the likelihood of working overtime
as a function of telecommuting status. We model three
versions of overtime: working more than 40 total hours
per week, more than 50 total hours per week, and more

than 60 total hours per week. In each model, we control
for occupation, education, gender, race/ethnicity, marital
status, and age. Since the 2001 CPS did not collect data
on parental status, we do not include this variable in the
models. Because logistic regression coefficients do not

Variable
NLSY (1998, 2002, 2004)

Telecommuting status Statistical
test

CPS

(1997, 2001, 2004)

Telecommuting status Statistical

test
No Yes No Yes

Total hours worked per week 41.11 47.81 (1) 40.79 45.45 (1)

20–40 73 22 (1) 72 47 (1)
41 or more 27 78 (1) 28 53 (1)
51 or more 7 30 (1) 9 22
61 or more 2 7 (1) 2 6 (1)

Hours worked by location
Onsite 41.11 41.61 (2) 40.79 38.70 (1)
At home — 6.20 — — 6.75 —

Regular — 2.03 — — 3.54 —
Overtime — 4.17 — — 3.21 —

Occupation (percent)
Managerial/professional 26 70 (1) 27 71 (1)
Sales 7 12 (1) 10 14 (1)
Other 67 18 (1) 63 15 (1)

Education (percent)
Less than high school 8 2 (1) 11 1 (1)
High school diploma 47 17 (1) 35 11 (1)
Some college 25 20 (1) 30 20 (1)
College degree or higher 21 60 (1) 24 68 (1)

Gender (percent)
Male 51 54 (3) 55 53 (3)
Female 49 46 (3) 45 47 (3)

Race/ethnicity (percent)
Other (White, Asian, etc.) 78 88 (1) 73 88 (1)
Black 16 8 (1) 12 6 (1)
Hispanic 7 5 (1) 14 6 (1)

Marital status (percent)
Never married 15 11 (1) 26 20 (1)
Married 63 75 (1) 60 69 (1)

Separated/divorced/widowed 22 14 (1) 14 11 (1)

Parental status (percent)

Parent (1 = yes) 65 74 (1) 75 77 (1)

Age 40.24 40.54 (3) 34.88 36.30 (1)

Number 14,100 2,198 — 43,188 7,264 —

1 p < .001. 2 p < .10. 3 p < .01.

NOTES: All statistics are weighted. Sample includes respondents who
were not self-employed, worked at least 20 hours per week, and

worked at least 1 hour onsite. In Current Population Survey (CPS) data,
the parental status question is only asked in 2001 and 2004; statistics
for this variable represent only these years.

SOURCES: National Longitudinal Survey of Youth (NLSY) 1979 panel
and special supplement from the U.S. Census CPS.

Table 1. Descriptive statistics by telecommuting status

Hours worked per week (percent)

Monthly Labor Review • June 2012 43

Variable
NLSY hours worked per week (1998, 2002, 2004) CPS hours worked per week (1997, 2001, 2004)

41 or more 51 or more 61 or more 41 or more 51 or more 61 or more

Telecommute status (1 = yes) 2.17 1.79 1.70 0.89 0.95 0.85
1(.07) 1(.08) 1(.16) 1(.03) 1(.04) 1(.08)

Occupation

Managerial/professional .39 .18 –.43 .36 .23 .11
1(.06) 2(.09) 3(.19) 1(.03) 1(.05) (.09)

Sales .42 .07 –.46 .46 .39 .17
1(.09) (.13) 2(.26) 1(.04) 1(.06) (.10)

Other (4) (4) (4) (4) (4) (4)

Education

Less than high school –.17 .36 .31 –.28 –.17 –.04
2(.10) 3(.15) (.29) 1(.05) 3(.08) (.15)

High school diploma –.10 .30 .32 –.06 –.02 .05

(.06) 5(.10) 2(.20) (.03) (.05) (.09)

Some college –.19 .17 .17 –.06 –.05 .00
5(.07) 2(.10) (.21) 2(.03) (.05) (.09)

College degree or higher (4) (4) (4) (4) (4) (4)

Gender

Female –1.32 –1.33 –1.40 –1.07 –1.20 –1.18
1(.05) 1(.07) 1(.15) 1(.02) 1(.04) 1(.07)

Male (4) (4) (4) (4) (4) (4)

Race/ethnicity

White (4) (4) (4) (4) (4) (4)

Black –.26 .01 .38 –.51 –.32 –.12
1(.05) (.07) 1(.13) 1(.04) 1(.07) (.12)

Hispanic –.21 –.02 –.01 –.46 –.42 –.42
1(.06) (.08) (.15) 1(.04) 1(.07) 1(.13)

Marital status

Never married –.18 –.04 .05 –.13 –.16 –.07
5(.07) (.10) (.17) 1(.03) 1(.05) (.08)

Married (4) (4) (4) (4) (4) (4)

Separated/divorced/widowed .12 .20 .29 .06 –.04 .17
3(.06) 3(.08) 2(.16) (.03) (.05) 2(.09)

Age –.01 –.01 .02 .00 .00 –.01

(.01) (.01) (.02) (.01) (.01) (.01)

Constant –.14 –2.04 –4.51 –.44 –1.95 –3.04

(.27) 5(.40) 5(.79) 1(.08) 1(.12) 1(.20)

Number 16,298 16,298 16,298 50,452 50,452 50,452

1 p < .001. 2 p < .10. 3 p < .05. 4 Omitted category. 5 p < .01.

NOTES: All statistics are weighted. Sample includes respondents who
were not self-employed, worked at least 20 hours per week, and worked
at least 1 hour onsite. Parental status not included in regression models
because it is not available in the 1997 Current Population Survey (CPS).

SOURCES: National Longitudinal Survey of Youth (NLSY) 1979 panel and
special supplement from the U.S. Census CPS.

Logistic regression coefficients predicting working overtimeTable 2.

Telecommuting

44 Monthly Labor Review • June 2012

show how much the probability of an event changes when
the predictors change, we translate the coefficients into
predicted probabilities for four “ideal types” (cases) in table
3. For each case, we calculate the probability of working
overtime, assuming the individual is not a telecommuter
and again assuming the individual is a telecommuter. In
both datasets and in all models, the probability of work-
ing overtime is higher for telecommuters compared with
nontelecommuters. The difference in the probability of
working overtime between the two groups is largest when
we define overtime as 41 hours or more, and smaller, but
still significant, when overtime is defined as working 61
hours or more.

OUR ANALYSIS OF TELECOMMUTING has yielded
several surprising findings. Though more and more employ-
ers claim to be offering flexible work options, the propor-
tion of workers who telecommute has been essentially flat
over the mid-1990s to mid-2000s and is no larger among
younger cohorts of workers than older cohorts. Moreover,
the average number of hours spent telecommuting each

week is relatively modest, around 6 hours per week in both
the CPS and NLSY samples. No evidence suggests that the
number of hours spent telecommuting is increasing over
time.

Our descriptive results suggest that labor demand for
work-family accommodation does not seem to propel
the distribution of telecommuting hours. None of the
expected relationships under such a scenario are present
in the data—parents of dependent children, for example,
are no more likely to telecommute than the population
as a whole. Meanwhile, indicators that suggest a supply-
side explanation—such as occupational sector and work
hours—are more strongly related to telecommuting
hours. As others have noted, the ability to work at home
appears to be systematically related to authority and sta-
tus in the workplace. Managerial and professional work-
ers are more likely than others to have the type of tasks
and autonomous control of their work schedule necessary
to perform work at home. While telecommuting may in
theory be a solution to the dilemmas of combining work
and family, telecommuting in practice does not unequiv-

Table 3. Predicted probability of working overtime as a function of telecommuting status and other variables

Case
NLSY hours worked per week

(1998, 2002, 2004)
CPS hours worked per week

(1997, 2001, 2004)
41 or more 51 or more 61 or more 41 or more 51 or more 61 or more

Case 1:
Man, college degree, managerial/professional

No telecommuting 49 10 1 47 16 4

Yes telecommuting 90 40 8 68 33 8
Difference 40 30 6 21 17 5

Case 2:
Man, high school diploma, other occupation

No telecommuting 37 11 3 37 13 4
Yes telecommuting 84 43 15 59 27 8

Difference 47 32 12 22 15 4
Case 3:

Woman, college degree, managerial/professional
No telecommuting 21 3 0 23 5 1
Yes telecommuting 70 15 2 42 13 3
Difference 49 12 2 19 7 2

Case 4:
Woman, high school diploma, other occupation

No telecommuting 14 3 1 17 4 1
Yes telecommuting 58 17 4 33 10 3
Difference 45 13 3 16 6 1

NOTES: In all predictions, the worker is White, married, and 40 years
old. Predictions based on estimated coefficients from table 2.

SOURCES: National Longitudinal Survey of Youth (NLSY) 1979 panel and
special supplement from the U.S. Census Current Population Survey (CPS).

[In percent]

Monthly Labor Review • June 2012 45

ocally meet the needs of workers with significant caregiv-
ing responsibilities.

The most telling problem with telecommuting as a
worklife solution is its strong relationship to long work
hours and the “work devotion schema.”11 Fully 67 percent
of telecommuting hours in the NLSY and almost 50 per-
cent in the CPS push respondents’ work hours above 40 per
week and essentially occur as overtime work. This dynamic
suggests that telecommuting in practice expands to meet
workers’ needs for additional worktime beyond the stan-
dard workweek. As a strategy of resistance to longer work
hours at the office, telecommuting appears to be somewhat
successful in relocating those hours but not eliminating
them. A less sanguine interpretation is that the ability of
employees to work at home may actually allow employers
to raise expectations for work availability during evenings
and weekends and foster longer workdays and workweeks.

Future research employing longitudinal data should explore
whether employees increase their work hours after initia-
tion of telecommuting.

Since telecommuting is intrinsically linked to infor-
mation technologies that facilitate 24/7 communication
between clients, coworkers, and supervisors, telecommut-
ing can potentially increase the penetration of work tasks
into home time. Bolstering this interpretation, the 2008
Pew Networked Workers survey reports that the majority
of wired workers report telecommuting technology has
increased their overall work hours and that workers use
technology, especially email, to perform work tasks even
when sick or on vacation.12 Careful monitoring of this
blurred boundary between work and home time and the
erosion of “normal working hours” in many professions
can help us understand the expansion of work hours over-
all among salaried workers.

NOTES

1 See American Time Use Survey—2010 Results, USDL-11-0919
(U.S. Bureau of Labor Statistics, June 22, 2011).

2 See Aleksandra Todorova, “Company Programs Help Employees
Save on Gas,” Smart Money, May 29, 2008, http://www.smartmoney.
com/spend/family-money/company-programs-help-employees-
save-on-gas-23179.

3 For review, see Ravi S. Gajendran and David A. Harrison, “The
Good, the Bad, and the Unknown about Telecommuting: Meta-Anal-
ysis of Psychological Mediators and Individual Consequences,” Journal
of Applied Psychology 92, no. 6 (2007), pp. 1,524–1,541.

4 See Nancy Folbre, Who Pays for the Kids? Gender and the Structure
of Constraint (New York: Routledge, 1995), and see Joan Williams, Un-
bending Gender: Why Family and Work Conflict and What to Do about It
(New York: Oxford University Press, 2000).

5 See Gartner, Inc., “Dataquest Insight: Teleworking, The Quiet
Revolution (2007 Update),” Gartner (May 14, 2007).

6 See Mary Blair-Loy, Competing Devotions: Career and Family
among Women Executives (Cambridge, MA: Harvard University Press,
2003). See Arlie Hochschild, The Time Bind (New York: Metropolitan
Books, 1995); See Pamela Stone, Opting Out? Why Women Really Quit
Careers and Head Home (Berkeley, CA: University of California Press,
2007).

7 See Gajendran and Harrison, “The Good, the Bad, and the Un-
known about Telecommuting,” pp. 1,524–1,541.

8 We use the term “onsite” to mean the location where workers labor
under the direction of their employer—an office, store, or other work-
site. In the datasets we use for the analysis, we have measures of total
hours worked and total hours worked at home. For simplicity, we refer
to the “hours worked not at home” as hours worked “onsite.” We use
the terms “work at home” and “telecommuting” interchangeably.

9 Differences exist in questionnaire wording both (1) over time in
the CPS and (2) between the CPS and NLSY that limit comparability of
work hour estimates across time periods and surveys. With all three
CPS surveys (1997, 2001, and 2004), we measure total work hours with
a question referring to actual hours of work (pehract1). “Last week,

how many hours did you actually work at your job?” To measure tele-
commuting, all three May CPS questionnaires have a lead-in question
asking, “As part of this job, do you do any of your work at home?” The
follow-up question varies slightly depending on which year of the CPS
survey is being used. The May 1997 CPS questionnaire asks, “Last week,
of the ___ actual hours of work you did, approximately how many of
them did you do at home for this job?” The May 2001/2004 CPS ques-
tionnaire, on the other hand, asks, “When you work at home, how
many hours per week do you work at home for this job?” Furthermore,
the questionnaire wording in the NLSY is slightly different than the
CPS. The NLSY question on hours worked (both at home and not at
home) measures usual hours, not actual hours: “How many hours per
week do you usually work at this job?” and then, “How many hours per
week do you usually work at this job at home?” Studies comparing the
two measures of hours worked (actual versus usual) find that estimates
of actual hours worked are generally lower than estimates of usual
hours worked (See Richard D. Williams, “Investigating Hours Worked
Measurements,” 2004, Labor Market Trends 112, no. 2 (2004), pp. 71–
79. Our results suggest a similar pattern. Finally, “it varies” is a valid
response option in the May 2001/2004 CPS question asking workers
for the number of hours worked at home. Approximately one-third
of the telecommuters in each year selected “it varies” as their response.
We imputed the mean telecommuting hours for those who replied “it
varies” (6.40 for 2001 and 6.74 for 2004) and created a dummy variable
to indicate that the respondent’s value for telecommuting hours was
imputed. This indicator was included in the logistic regression models
predicting overtime; the substantive results from these models are not
sensitive to the inclusion of the indicator variable.

10 Our telecommuting estimates from 2004 are lower than the
American Time Use Survey (ATUS) estimates for 2010: 17 percent ver-
sus 24 percent. The most likely explanation for the difference is sample
composition. We exclude workers who are self-employed and/or who
work exclusively at home; the ATUS does not.

11 Outlined by Blair-Loy, Competing Devotions.
12 See Mary Madden and Sydney Jones, Networked Workers (Pew

Research Center, September 24, 2008), http://pewinternet.org/Re-
ports/2008/Networked-Workers.aspx.

http://www.smartmoney.com/spend/family-money/company-programs-help-employees-save-on-gas-23179

http://pewinternet.org/Reports/2008/Networked-Workers.aspx.

Turn in your highest-quality paper
Get a qualified writer to help you with

“ TECH INEQUITIES ”

Get high-quality paper

NEW! AI matching with writer

Hire a Writer

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.

New to Your Trusted Assignment Help Service? Sign up & Save

Calculate the price of your order

Type of paper needed:

Pages:

You will get a personal manager and a discount.

Academic level:

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 Now Order Now