Professional Documents
Culture Documents
Research Collection Lee Kong Chian School Of Lee Kong Chian School of Business
Business
2009
Kelly S. Wilson
Michigan State University
Citation
Ilies, Remus; Wilson, Kelly S.; and WAGNER, David Turley. The Spillover of Daily Job Satisfaction onto
Employees’ Family Lives: The Facilitating Role of Work-Family Integration. (2009). Academy of
Management Journal. 52, (1), 87-102.
Available at: https://ink.library.smu.edu.sg/lkcsb_research/1744
This Journal Article is brought to you for free and open access by the Lee Kong Chian School of Business at
Institutional Knowledge at Singapore Management University. It has been accepted for inclusion in Research
Collection Lee Kong Chian School Of Business by an authorized administrator of Institutional Knowledge at
Singapore Management University. For more information, please email cherylds@smu.edu.sg.
The Spillover of Daily Job Satisfaction onto Employees' Family Lives: The Facilitating Role
of Work-Family Integration
Author(s): Remus Ilies, Kelly Schwind Wilson and David T. Wagner
Source: The Academy of Management Journal, Vol. 52, No. 1 (Feb., 2009), pp. 87-102
Published by: Academy of Management
Stable URL: http://www.jstor.org/stable/40390277
Accessed: 11-06-2018 05:53 UTC
JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide
range of content in a trusted digital archive. We use information technology and tools to increase productivity and
facilitate new forms of scholarship. For more information about JSTOR, please contact support@jstor.org.
Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at
http://about.jstor.org/terms
Academy of Management is collaborating with JSTOR to digitize, preserve and extend access
to The Academy of Management Journal
This content downloaded from 202.161.43.77 on Mon, 11 Jun 2018 05:53:39 UTC
All use subject to http://about.jstor.org/terms
° Academy of Management Journal
2009, Vol. 52, No. 1, 87-102.
REMUS ILIES
KELLY SCHWIND WILSON
DAVID T. WAGNER
Michigan State University
The longitudinal, multisource, multimethod study presented herein examines the role
of employees' work-family integration in the spillover of daily job satisfaction onto
daily marital satisfaction and affective states experienced by employees at home. The
spillover linkages are modeled at the within-individual level, and results support the
main effects of daily job satisfaction on daily marital satisfaction and affect at home,
as well as the moderating effect of work-family integration on the strength of the
within-individual spillover effects on home affect. That is, employees with highly
integrated work and family roles exhibited stronger intraindividual spillover effects on
positive and negative affect at home.
Copyright of the Academy of Management, all rights reserved. Contents may not be copied, emailed, posted to a listserv, or otherwise transmitted without the copyright holder's express
written permission. Users may print, download or email articles for individual use only.
This content downloaded from 202.161.43.77 on Mon, 11 Jun 2018 05:53:39 UTC
All use subject to http://about.jstor.org/terms
88 Academy of Management Journal February
This content downloaded from 202.161.43.77 on Mon, 11 Jun 2018 05:53:39 UTC
All use subject to http://about.jstor.org/terms
2009 lhes, Wilson, and Wagner 89
This content downloaded from 202.161.43.77 on Mon, 11 Jun 2018 05:53:39 UTC
All use subject to http://about.jstor.org/terms
90 Academy of Management Journal February
This content downloaded from 202.161.43.77 on Mon, 11 Jun 2018 05:53:39 UTC
All use subject to http://about.jstor.org/terms
2009 Hies, Wilson, and Wagner 91
Hypothesis 6. The
because they
tendency
did not complete the final measurefor of em
higher (rather than
work-family rolelower) daily
integration. Finally, after match- jo
ing the work and home
to experience higher data from employees
daily and
marital
at home (as predicted by
spouses or significant others, we removed data for 9 Hyp
stronger for employees
of the remaining participants because whothey did not are
than less work-family
have at least four matchedrole-integra
sets of responses.
In sum, the final sample consisted of 101 employ-
ees and their spouses or significant others (67 percent
METHODS of the initial sample). We obtained job titles for all of
the employee participants in the final sample. An
Participants
analysis of these job titles revealed that participants
The sample of employees included in the current held jobs dealing with secretarial or administrative
study participated in a broader study on work and support work (34%), communications and coordina-
family by Hies, Schwind, Wagner, Johnson, DeRue, tion (15%), research activities (13%), information
and Ilgen (2007). However, with the exception of technology (13%), and other areas (25%).
the scores reflecting employees' positive and nega- The employees had an average age of 42.7 years
tive affect at work (assessed at the end of the work- and an average tenure of 12.7 years. Seventy-eight
day), none of the variables used in that study were percent were female; 48 percent had no children at
used in the hypothesis tests reported here. We re- home; and 43 percent had one or two children
cruited participants from a pool of university em- living at home. In the final sample, all but one of
ployees including administrative professionals, su- the focal participants were married, and the one
pervisors, and clerical-technical employees. In the exception lived in a domestic partnership. No de-
recruitment e-mail, we briefly described the study mographic measures were obtained from spouses
and emphasized that, to enroll in it, employees had or significant others. We performed independent
to be married or cohabiting with a significant other, sample t-tests to compare the demographic data
and their spouses/significant others needed to agree from the employees included in the data analysis
to participate. Interested participants were directed to with the demographic data of the employees who
an online registration page where they created a submitted the latter but did not complete the entire
unique user name and password. The first 150 indi- study. We found no significant differences between
viduals to register were allowed to participate in the the two groups of employees on any of the demo-
study and comprised our initial sample. graphic variables mentioned above.
Participation in the study was voluntary, and
employees and their spouses/significant others Procedures
were compensated for their participation (employ-
ees received up to $75, and spouses/significant oth- Surveys were used to assess perceptions and/or
ers up to $40, depending on how many surveys feelings from employees and their spouses/signifi-
they completed). Of the university employees who cant others. These surveys were given via the inter-
signed up to participate in the study, 147 com- net to employees at work and provided to their
pleted an initial survey requesting demographic spouses/significant others on paper. A paper sur-
information. Of this sample, 7 individuals indi- vey was also given to the employees in this study to
cated they were not married or did not live with a use when reporting their daily marital satisfaction in
significant other willing to participate. Twenty ad- the evening of each workday. The paper surveys con-
ditional participants did not respond to any of the tained an instruction to immediately seal them in
daily surveys at work. Although we do not know accompanying envelopes addressed to the research-
the reasons that prevented each of these employees ers and to mail these the next morning. Telephone
from participating, we received e-mails from sev- interviews with employees* spouses or significant
eral individuals who indicated that they could not others also occurred between 7 and 9 p.m. of each
continue participating because their spouses or sig- workday of the study (beginning in phase 2).
nificant others could no longer participate; they In phase 1 of the study, the employees' Internet-
had received new work assignments; they had to based survey assessed demographic variables and
travel during the study period and could not fulfill various individual difference variables such as
the study requirements; or health or family issues "trait positive affect" and "trait negative affect."
prevented continuing. Four of the remaining par- Employees' spouses or significant others also com-
ticipants did not complete any home measures, and pleted a paper survey in which they rated their
thus their data were not usable. Six additional par- spouses'/significant others' general levels of job
ticipants were eliminated from the final analyses satisfaction and their own global marital satisfac-
This content downloaded from 202.161.43.77 on Mon, 11 Jun 2018 05:53:39 UTC
All use subject to http://about.jstor.org/terms
92 Academy of Management Journal February
This content downloaded from 202.161.43.77 on Mon, 11 Jun 2018 05:53:39 UTC
All use subject to http://about.jstor.org/terms
2009 Hies, Wilson, and Wagner 93
Employees9 positive
homeand negativ
variables, w
fect at work. We assessed employ
measuring the ti
negative affect at work during
workday job satis the
noon positive affectee'sat home work survey
and af
affect at work) of tion, each and workday (2) the o t
the 20 adjectival descriptors
significant of other moo
tive and Negative We controlled
Affect for these amounts of time in all the
Schedule (
within-individual analyses.
Clark, 1994). Specifically, the ten a
tors from the positive scale
For the analyses were
predicting positive "int
and negative
siastic," "excited," "strong," "pro
affect at home, to account for possible autocorrela-
tion in the home
spired," "attentive," affect scores, we controlled
"active" and for the "
ten adjectives from the
positive and negative negative
affect scores employees had
set," "irritable,"
reported in the morning from work. These control "g
"distressed,"
"ashamed," "nervous," "jittery,"
variables were assessed every morning, via the in- "a
tile." This survey was administer
ternet, using the same scales and response options
used to assess were
internet. Employees afternoon positive and negative af-to
asked
fect at work (see
tent to which they experienced eac above).
described by the Finally,
PANAS following the suggestion
adjectivesof an anony-
were completingmous the survey
reviewer, we created two job type (1 = "s
classifica-
all," to 5, "very tions;
much").we defined the first type as a supporting role
Employees9 home at work, which positive
included such time-bound and
jobs as
To assess the home positive
secretary, administrative and
assistant, and other simi- n
lar positions;in
the focal employees we defined
this the secondstudy,
type as more
significant othersautonomouswere roles, which included jobs such
asked by as t
human resourcescale,
cate, via a five-point analyst, communications
the manager,
exte
believed that the focal
and editor. employee
The primary distinction we sought to
various positive and
highlight concerned negative
the relatively rigid temporalaf
and spatial requirements of
emotions that evening. the supporting
The roles,
telep
presented the 20 versus theadjective
more flexible boundaries of the autono-descr
PANAS (see above) mous roles, in and oura implicit
randomexpectation was that o
ing positive and employees
negativewith more autonomousaffect jobs would have
more work-family role-integrated
telephone interviewer instructed roles. We in-
cluded the job type
cant others to select one classification
of as a the
control in thefo
"slightly or not analyses
at examining
all," the moderating
"a influence
little of
"quite a bit," orwork-family
"very role integration
much," on the strength of the an
were coded 1, 2, job3, satisfaction
4, spillover.
and 5, respec
Employees9 work-family role in
extent to which each participant in
Analytic Strategy
grated work and family roles was
slightly modified To test the intraindividual spillover
version of the effects of t
daily job satisfaction and the cross-level moderat-
Family Role Integration-Blurring S
ing hypotheses,
et al., 2005). At the end we used aof hierarchical
the linear mod-
stu
responded to the eling (HLM) framework. For each items:
following criterion, we
cult to tell wheretestedmy a model that regressed daily criterion
work life scores
end
on the dailyto
life begins," "I tend job satisfaction
integrate scores at the first level m
ily duties when I (within
work participants) ofat analysis. Importantly, the
home," a
there is a clear daily
boundary betwee
job satisfaction scores were centered relative to
each respondent's
my role as a parent or average family
score to form deviations m
scored). The anchor
from respondents'points characteristic means. As for
many au- t
thors have explained
measure work-family (e.g., Hies, Scott,
role & Judge, 2006;
integrat
Sonnentag, 2003), this centering
1, "strongly disagree," to approach5, eliminates
"stro
therefore, higher all the between-individual
scores variance
are in the predictor
asso
integrated scores and therefore the roles.
work-family estimates represent strictly
Control within-individual
variables. associations.
Because work
over effects are To test our hypothesis
likely to that employees
depend will expe- o
time elapsed rience higher positivethe
between affect at home on days when
assessm
This content downloaded from 202.161.43.77 on Mon, 11 Jun 2018 05:53:39 UTC
All use subject to http://about.jstor.org/terms
94 Academy of Management Journal February
This content downloaded from 202.161.43.77 on Mon, 11 Jun 2018 05:53:39 UTC
All use subject to http://about.jstor.org/terms
2009 lhes, Wilson, and Wagner 95
TABLE 1
1. Afternoon positive affect at work 2.93 0.79 (.94) -.15** .39** -.01 .09+ -.14*
(self-rated)
2. Afternoon negative affect at work 1.22 0.33 -.08 (.86) -.35** .03 -.07* .18**
(self-rated)
3. Daily job satisfaction (self-rated, at 3.66 0.59 .54** -.24* (.86) .08* .10** -.18*
work, in the afternoon)
4. Daily marital satisfaction (self- 4.11 0.59 .15 -.02 .13 (.93) .10* -.16**
rated, at home, in the evening)
5. Home positive affect (spouse-rated, 2.89 0.69 .12 -.06 -.03 .06 (.88) -.11*
at home, in the evening)
6. Home negative affect (spouse-rated, 1.24 0.29 -.26* .09 -.24* -.05 .09 (.91)
at home, in the evening)
7. General job satisfaction (spouse- 3.56 0.72 .17 -.02 .38** .09 .18 -.22+ (.83)
rated; one-time rating)
8. Spouse general marital satisfaction 4.34 0.71 -.11 .14 -.01 .46** .19 -.12 .17 (.91)
(spouse-rated; one-time rating)
9. Work-family integration (self-rated; 2.34 0.73 -.02 .10 .07 -.11 -.01 .14 .17 .19+ (.72)
one-time rating)
10. Job type (coded from job titles) 0.34 0.48 .05 -.25* .03 -.02 .14 -.11 .06 -.07 -.19+
a The correlations below the diagonal represent between-individual associations (for variables 1 through 8, w
individual correlations using individuals' aggregated scores; n = 86-101, pairwise). The correlations above
within-individual associations (over time) and were estimated from fixed-effects HLM models with single level
2 predictors [n - 621-867, pairwise). Reliabilities are reported on the diagonal; for the within-individual variabl
averaged over measurements.
bJob type was coded as follows: support role (e.g., secretarial) = 1, other = 0. The morning affect control
employees at work) are not included in this table.
+ p< .10
* p < .05
**p< .01
***p < .001
Two-tailed tests.
affect (r = .24, p < .001) and negative affect (r = .34, which shows a positive and significant effect of
p < .001), which supports the validity of the spou- daily job satisfaction on positive affect at home
sal reports of affect. (ß = 0.15, p < .01) in our test of model 1 to predict
Before explicitly testing the hypotheses, we positive affect at home.
examined whether there was sufficient intraindi- Hypothesis 2 predicted that on days when em-
vidual variance in the daily scores to warrant mod- ployees experience high daily job satisfaction, they
eling intraindividual relationships by computing will experience lower negative affect at home than
the proportion of the total variance in the daily they will on days when they experience low daily
scores that was associated with within-individual job satisfaction. Support for this is seen in Table 2,
variation (from null HLM models). Indeed, a sub- which shows a negative and significant effect of
stantial proportion of the total variance (25-55%) daily job satisfaction on negative affect at home
was caused by day-to-day fluctuations in each mea- (ß = -0.17, p < .01) in our test of model 1 to predict
sured score, which is consistent with previous negative affect at home.
work examining within-individual fluctuations in Hypothesis 3 predicted that on days when em-
affect and satisfaction (e.g., Hies & Judge, 2002). ployees experience high daily job satisfaction, they
will report higher daily marital satisfaction than
they will on days when they experience low daily
Hypothesis Tests
job satisfaction. This hypothesis was also sup-
Hypothesis 1 predicted that on days when em- ported, as seen in Table 2: The effect of daily job
ployees experience high daily job satisfaction, they satisfaction on daily marital satisfaction was posi-
will experience higher positive affect at home than tive and significant (ß = 0.10, p < .05) in our test of
they will on days when they experience low daily model 1 to predict daily marital satisfaction.
job satisfaction. Support for this is seen in Table 2, Hypothesis 4 predicted that the tendency for
This content downloaded from 202.161.43.77 on Mon, 11 Jun 2018 05:53:39 UTC
All use subject to http://about.jstor.org/terms
96 Academy of Management Journal February
TABLE 2
Variables ß t ß t ß t
Model 1
Intercept 2.91 38.15** 1.25 35.11** 4.12 67.38**
Time lag (coded) 0.11 1.93* 0.04 0.65 -0.09 -1.52
Daily job satisfaction (self-rated, in the afternoon) 0.15 3.28** -0.17 -2.55** 0.10 2.13*
Model 2
Intercept 2.91 36.15** 1.25 34.02** 4.12 67.34**
Afternoon positive affect at work (self-rated) 0.04 0.47 -0.01 -0.31
Afternoon negative affect at work (self-rated) 0.11 1.74* 0.07 1.41
Time lag (coded) 0.10 1.88* 0.05 1.10 -0.10 -1.79*
Daily job satisfaction (self-rated, in the afternoon) 0.16 2.73** -0.13 -2.03* 0.10 2.24*
a Daily job satisfaction, measured with reports provided by employees from work each afternoon before leavin
predictor in these analyses. In the model 2 analyses, we included the affect scores reported by employees at the end
positive and negative affect) as predictors to account for work-to-home mood spillover. When predicting positive a
home, models 1 and 2 also controlled for morning positive or negative affect at work, respectively, to account for
in the data. The regression coefficients for these controls were close to 0 (ranging from -.03 to .01) and nonsignific
1 predictor scores at the individuals' means to eliminate between-individual variance. The main effects of job satisf
standardized.
* p < .05
**p< .01
One-tailed directional tests.
employees with higher (rather than lower) daily isfaction and work-family role integration when pre-
job satisfaction to experience higher positive af- dicting home negative affect (see Table 3). This
fect at home (as predicted by Hypothesis 1) will interactive effect is illustrated in Figure 2, which
be stronger for employees who are more rather shows that individuals with highly integrated work
than less work-family role-integrated. As shown and family roles strongly feel the effects of a dissatis-
in Table 3, which presents the estimates for fying day on the job, resulting in high levels of nega-
model 3, the data supported this hypothesis intive affect on days when they report low daily job
that the interactive effect of daily job satisfaction
satisfaction, and low levels of negative affect on days
and work-family role integration was positive when they experience high daily job satisfaction. In
and significant when predicting home positive
contrast to the employees rating high on work-family
affect. Figure 1 depicts this interactive effect
role integration, those employees with segmented
graphically, showing that individuals with
roles experienced a positive (albeit weak) relation-
highly integrated work and family roles experi-
ship between daily job satisfaction and negative af-
ence a strong, positive relationship between self-
fect, and there were relatively large differences in
ratings of daily job satisfaction at work and spou-
sal ratings of their positive affect at home, home negative affect on days of dissatisfying work
whereas the relationship was weak (actually neg- between employees with highly integrated, versus
ative) for those with segmented roles. highly segmented, roles.
Hypothesis 5 predicted that the tendency for em- Finally, Hypothesis 6 predicted that the ten-
ployees with higher (rather than lower) daily job sat- dency for employees with higher (rather than
isfaction to experience lower negative affect at home lower) daily job satisfaction to experience more
(as predicted by Hypothesis 2) will be stronger for daily marital satisfaction at home (as predicted by
employees who are more rather than less work-family Hypothesis 3) will be stronger for employees who
role-integrated. This hypothesis was also supported are more rather than less work-family role-inte-
by the data, as evidenced by the negative and signif- grated. This hypothesis did not receive support, as
icant effect for the interaction between daily job sat- the interactive effect of daily job satisfaction and
This content downloaded from 202.161.43.77 on Mon, 11 Jun 2018 05:53:39 UTC
All use subject to http://about.jstor.org/terms
2009 lhes, Wilson, and Wagner 97
TABLE 3
Results of Cross-Level HLM Analyses Testing the Moderating Effect of Work-Family Integration
Daily Marital
Home Positive Affect Home Negative Affect Satisfaction (Self-
(Spouse-Rated, in the (Spouse-Rated, in the Rated, in the
Evening) Evening) Evening)
Variables ß t ß t ß t
Model 3
Intercept 2.90 38.95** 1.25 35.32** 4.12 68.96**
Time lag (coded) 0.10 1.75* 0.03 0.73 -0.09 -1.39
Job satisfaction (self-rated, in the afternoon) 0.17 3.88** -0.22 -3.12** 0.09 1.95*
Work-family integration (self-rated, one- -0.02 -0.26 0.04 0.65 -0.12 -1.53
time rating)
Work-family integration X job satisfaction 0.12 2.02* -0.12 -2.33* 0.02 0.38
a The models predicting positive and negative affect at home controlled for morning positive or negat
account for possible autocorrelation in the data (the regression coefficients for these controls are were .0
negative affect, respectively, both nonsignificant). In estimating the moderating effects of work-fami
type at level 2 (the effects of job type are discussed in the section presenting supplementary results). Le
at the individuals' means to eliminate between-individual variance. The intraindividual effects o
standardized; the main effect of work-family integration is an interindividual effect; the level 2 estimat
the increase in the magnitude of the intraindividual effect of job satisfaction, in standardized poi
deviation increase in the work-family integration score.
* p < .05
**p< .01
One-tailed directional tests.
FIGUREI
Interactive Effect of Work-Family Integration and Job Satisfaction on Home Posit
3.30-1
3.10- ^r
Positive Hom
Spouse-Reported 2*90" ^^ ^"^ -^
^r "** *£
2.70- ^^
2.50-1 - -
a Low job satisfaction and low work-family integration represent scores one standard deviation below the grand means on the
respective measures; high scores are one standard deviation above grand means.
work-family role integration was not significant over is controlled, we did several supplemental
when predicting daily marital satisfaction. tests. First, we tested whether daily job satisfaction
predicts home affect when afternoon affect at work
is controlled. Introducing the positive and negative
Supplemental Results affect scores reported by employees at the end of
To test whether the effects specified in Hypoth- their workdays (on the same survey on which they
eses 1, 2, and 3 are maintained when mood spill- reported job satisfaction) into the regressions pre-
This content downloaded from 202.161.43.77 on Mon, 11 Jun 2018 05:53:39 UTC
All use subject to http://about.jstor.org/terms
98 Academy of Management Journal February
FIGURE 2
Interactive Effect of Work Family Integration and Job Satisfaction on Home Negative Affect
1.70-1
1.50- ^V.
0.90-1
Low Job Satisfaction High Job Satisfaction
This content downloaded from 202.161.43.77 on Mon, 11 Jun 2018 05:53:39 UTC
All use subject to http://about.jstor.org/terms
2009 Hies, Wilson, and Wagner 99
This content downloaded from 202.161.43.77 on Mon, 11 Jun 2018 05:53:39 UTC
All use subject to http://about.jstor.org/terms
100 Academy of Management Journal February
This content downloaded from 202.161.43.77 on Mon, 11 Jun 2018 05:53:39 UTC
All use subject to http://about.jstor.org/terms
2009 Hies, Wilson, and Wagner 101
This content downloaded from 202.161.43.77 on Mon, 11 Jun 2018 05:53:39 UTC
All use subject to http://about.jstor.org/terms
102 Academy of Management Journal February
This content downloaded from 202.161.43.77 on Mon, 11 Jun 2018 05:53:39 UTC
All use subject to http://about.jstor.org/terms