u JC. ..

wk?»

.w. .-

“we”... ..r .. Hm“
4w.“ in: .V... . , «va

I4 n}
h...

. 7
a

r. u .
HM;

1430..
.1:

. «13%.
$ka

.2‘

..
7....
1;“...4
.1. .4‘

r r r r
ﬁnk». :

3E a
in g

cm x 5..
2 » .
2.11an
. . 1
g
i

L 5:,

. 43”»
a} .

V 2.

W 79”.“!

V r .4.
c.

I.
amﬁ.
vi

 

T5153} "8“-

ft

5
£603

5% 77%?

 

LIBRARY
Michigan State This is to certify that the
U niversity d'ssem ”W“

 

 

 

The Effects of a Family-Supportive Work Environment on
Work-to-Family Conﬂict, Family-to-Work Conﬂict, and
Emotional Exhaustion - Does Income Level Matter?

presented by
Cynthia Ozeki

has been accepted towards fulﬁllment
of the requirements for the

Doctoral degree in Labor and Industrial Relations

Major Professor's Signature
ay 20 l 200 3

Date

 

MSU is an Afﬁrmative Action/Equal Opportunity Institution

__._ W .~‘.. . 4i "'
ﬂ
‘ ﬂ —. g. #—

‘.——. a

_. _‘ _' “——

_ T . -.." ‘4 4 ‘4 '- ‘W- 1 '

PLACE IN RETURN BOX to remove this checkout from your record.

TO AVOID FINES return on or before date due.
MAY BE RECALLED with earlier due date if requested.

 

DATE DUE

DATE DUE

DATE DUE

 

with w»

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

6/01 o'JCIRC/DateDuepes-sz

THE EFFECTS OF A FAMILY-SUPPORTIVE WORK ENVIRONMENT ON
WORK-TO-FAMILY CONFLICT, FAlyﬂLY-TO-WORK CONFLICT,
AND EMOTIONAL EXHAUSTION - DOES INCOME LEVEL MATTER?

By

Cynthia Ozeki

A DISSERTATION

Submitted to
Michigan State University
In partial fulﬁllment of the requirements for the degree of

DOCTOR OF PHILOSOPHY

School of Labor and Industrial Relations

2003

ABSTRACT

THE EFFECTS OF A FAMILY-SUPPORTIVE WORK ENVIRONMENT ON WORK-
TO-FAMILY CONFLICT, FAMILY-TO-WORK CONFLICT,
AND EMOTIONAL EXHAUSTION —— DOES INCOME LEVEL MATTER?
By

Cynthia Ozeki

Drawing on the conservation of resources model, this study investigates the impact of a
family-supportive work environment on work-to—family conﬂict, family-to-work conﬂict, and
emotional exhaustion in the context of other important personal and work-related demands and
resources in a national sample of 2,877 employees. The study ﬁnds that working in
organizations with environments that are more supportive is associated with less work-to-family
and family-to-work conﬂict, as well as lower levels of emotional exhaustion. Part of the impact
of a supportive work environment on emotional exhaustion was mediated by conﬂict between
work and family. The relationship between the outcomes and working in a supportive
enviromnent appeared particularly strong for workers from lower-income households, who have
fewer alternative resources to draw on. Access to more dependent care beneﬁts was associated
most closely with less work-to-family conﬂict among lower income workers. Supportive
supervisors and cultural norms that don’t penalize workers for putting family ﬁrst were related to
lower levels of family-tO-work conﬂict, with the effects being stronger for lower income workers
than those with more ﬁnancial power. For all workers, informal support was more closely

associated with reduced conﬂict and exhaustion than the availability of formal work/life beneﬁts.

This work is dedicated to my family.

I v
commute;
Richard B
dlSSCIIaIIL‘

I

COUIIIDUI‘. I

 

ACKNOWLEDGEMENTS

I would like to acknowledge the invaluable support and advice of my dissertation
committee chair, Ellen Ernst Kossek, as well as the wise assistance I received from
Richard Block, my coursework committee chair, and from the other members of my
dissertation committee, Ann Marie Ryan, Clifford Broman, and Peter Berg. Their

contributions helped to improve this paper immensely.

iv

LIST OF

CHAPTE
LNTROD
RI

 

TABLE OF CONTENTS

LIST OF TABLES AND FIGURES... viii
CHAPTER 1:
INTRODUCTION
Research Needed on Links Between Work, Family, and Burnout... .. .. 1
Gaps in Research on the Effects of Work/Family Policies. .. .... 1
Conﬂict Between Work and Family An Important Outcome... ... 2
Burnout a Critical Outcome for Study. 3
Family-Supportive Environments: Formal and Informal Elements. 5
Purposeofthe Study... 7
Main Research Questions... .. .... .. .... . 7
Contributions. 8
Underlying Assumptions... 8
Organization of the Dissertation 10
CHAPTER 2:
THEORETICAL UNDERPINNINGS AND PREVIOUS RESEARCH
Construct Deﬁnition: A F amily-Supportive Work Environment... .. 11
Previous Research on Family-Supportive Organizational Environments. 12
Deﬁning Work-to-Family and Family-to-Work Conﬂict... .. 14
Research on Work/Family Conﬂict... l4
ConsequencesofConﬂict... 15
Antecedents of Conﬂict... . .. .... 16
Research on Family-Supportive WOrk Envirbnments and Conﬂict. ......... 17
Burnout: DeﬁningtheConstruct... l9
PreviousResearchonBumout... 20
ConsequencesofBurnout . 20
Antecedents ofEmotional Exhaustion 20
Research on Family-Supportive Environments and Burnout ................ 21
Major Theories Underlying the Study... . . 22
Work Environment Studies: Theories of Seeial Exbhange and Symbolism" 23
Underlying Research on Work and Family: Role Theory .................... 24
Helping Explain Emotional Exhaustion: Conservation of Resources. . .. 25
CHAPTER 3:
LITERATURE REVIEW AND HYPOTHESES
Research Overview... 28

Pt

 

 

 

Da

CWTEI

Sal

Personal Demands and Resources... .
Gender as an Indicator of Demands

Caring for Elders and Disabled Adults Can Ee a ”Burden. I i 7

Children: A Key Source of Demands...
Partners Can Provide Support...

Household Income: A Key Resource .
Age Means More Life Experiences, An ImpOrtant Resource
Work Demands and Resources...

Long Work Hours and An Irregular Schedule: SOurces of Stress:

Autonomy Allows Workers to Take Control of Problems...
Tenure Means Valuable Skills...
A Family-Supportive Work Environment

Flexibility in Work Time Reduces Conﬂlcts and Stress.

Flexibility in Work Location May Be Helpful...

Dependent Care Assistance Contributes to a Sense Of support

Supervisor Support Can Promote More Positive Outcomes...

Family-Supportive Cultural Norms May Reduce Concerns...
Potential Moderators: Income Level and Single Parent Status...

Work/familyConﬂictasaMediator........ .. ...

CHAPTER 4:
RESEARCH METHODS

SampleBackground.................
DataCollectionProcedures................................

Measures... .
Personal Demands and Resources ..
Work-Related Demands andResources...

Dependent Variables.........
Data Analysis.....

CHAPTER 5:
RESULTS
Outline of the Chapter...

Sample Characteristics and Descriptive Statistics fer MajOr Vanables ...
116

117
118
.. 119

Demographics... .

Personal Demands and Resources

Work Demands and Resources... . .
Organizational Work/Family Environment

vi

34
35
38
39
41
43

49
50
51
53
55
56
57
61
65
68
71
73
78

80
80
8 1
86
9O
93
94

100
116

St.

cram
DISCUSS
ct

 

 

 

 

APPERD
APPEND.
APPEND

REFEREE

Results of Regression Analyses and Evaluation of Hypotheses. .. ..
Control Vanables ..
Personal Demands and Resources
Work Demands and Resources... .

F amﬂy-Supporhve Organizational Env1ronment

Interaction Effects... .

Mediation Effects for Full Sample

Overall Variance Explained...

Why Income Matters...

Regression Results By InCOme Level . ...

Higher Income Associated with Smaller Environment Impact ......
Summary of Results...

CHAPTER 6:
DISCUSSION

Contributionsofthe Study
153
156
158

..162

Implications for Future Research...

Implications for Practice. .

Study Limitations and Suggestions for Future Research
Conclusions...

APPENDIX A: SURVEY ITEMS AND SCALES...”

APPENDIX B: SUPPLEMENTARY TABLES...... ..

APPENDIXC:BACKGROUNDONTHENSCWF...........

vii

120

120
121
.. 127
129
131
..... 134
135
139
. 142

146

. 146

148

164

171

190

196

Table II

Table 1'. I.

Worms

 

   

Table 3. l
Family-It"

Table 5:
Table 81;
Table BI.
Table B3_
Table 84
FISH“: I:
Figure 2;

FIgure 3_
Family C

FlElite <1
work-IO

Flgure S.
F 31111134

LIST OF TABLES AND FIGURES

Table 1: Means, Standard Deviations and Intercorrelations...

Table 2: Results of Multiple Regression Analysis Predicting
Work-to-Family Conﬂict... .. . . .. . .

Table 3: Results of Multiple Regression Analysis Predicting
Family-to-Work Conﬂict. .. .... .. .... .

Table 4: Results of Multiple Regression Analysis Predicting
EmotionalExhaustion...

Table5:SummaryonypothesesandResults..........................................
Table Bl: Results ofPrincipal Components Factor Analyses...
TableBZ:Intercorrelations forLowerIncome Group
TableB3zlntercorrelations forMiddleIncomeGroup................................
TableB4:Intercorre1ations forUpperIncomeGroup...........................
Figure1:ResearchModel—DirectEffects................ .

Figure 2: Research Model—IndirectEﬂ‘ects

Figure 3: Interaction of Income and Total Environment Work-to-

Family Conﬂict

Figure 4: Interaction of Income and Dependent Care Beneﬁt Index

Work-to-Family Conﬂict...

Figure 5: Interaction of Income and Total Environment

Family-to-Work Conﬂ1ct

Figure 6: Interaction of Income and Supervisor Support
Family-to-Work Conﬂict. . .

Figure 7: Interaction of Income and Cultural Norms

Family-to-Work Conﬂ1ct

viii

101

. 105

. 109

. 114

137

172

175

179

183

29

30

187

187

188

188

189

 

Research
A:
strikingl}

has broug

 

while husT
and doing
common ;

family has

 

1998), W1
bOUndaries
°0nnectj0r
and Policy
in The relax
MS and 12
Gaps in R.

W1
Work and 1
Understanc
on the Job
Mines Ih;
limited hm

wOrk’farml

CHAPTER 1: INTRODUCTION

Research Needed on the Links Between Work, Family, and Burnout

As we enter the 2 1 st century, the faces that make up the American workforce are
strikingly different from half a century ago. A dramatic shift over the past few decades
has brought more women, including mothers of young children, into all types of jobs,
while husbands and fathers have become more involved in caring for family members
and doing household work. Today households headed by single working parents are a
common phenomenon, nearly half of all workers are women, and the modal American
family has two working parents (Barnett and Hyde, 2001; Bond, Galinsky and Swanberg,
1998). With more people combining heavy family responsibilities with work, the
boundaries between work and family have become more permeable. Understanding the
connections between work and home is important for employers, individuals, researchers,
and policy makers. Changing workforce demographics have led to an increased interest
in the relationship between people’s work and private lives, how this may affect their
jobs and families, and what can be done promote positive experiences.
Gaps in Research on the Eﬂects of Organizational Work/Family Policies

While changes in workforce demographics have inspired a stream of research on
work and family, key gaps in our knowledge remain. Particularly important to
understand are the effects of organizational efforts to support workers balancing demands
on the job and at home. Yet quality research on the effects of human resource (HR)
policies that aim to help workers manage their lives inside and outside the workplace is
limited both in availability and scope. Recent meta-analyses on the impacts of

work/family policies (Kossek and Ozeki, 1998) and ﬂextime and compressed workweeks

(Baltes a

 

statistical

 
  
   
  

job satisl‘.
organizar:
611011 by '
1995),an1
Smith anc
(Allen, 20
Boudreau.

the degree

 

ainount of
022111, 199
organizatir
another an
reported e'
Only a is“
it dlﬁicu] t

C""ﬂt'cr 5

(Baltes er al., 1999) each located fewer than 30 published journal articles that reported
statistical estimates of the effects of these policies on common work outcomes, such as
job satisfaction and absences. The studies that have looked at such indications of
organizational support have linked them to individual job outcomes like greater extra-role
effort by employees (Lambert, 2000), organizational commitment (Grover and Crooker,
1995), and loyalty (Roehling et al., 2001), as well as organizational performance (Perry-
Smith and Blum, 2000). However, only a handful of studies involving unique samples
(Allen, 2001; Thompson, Beauvais and Lyness, 1999; Thomas and Ganster, 1995; Judge,
Boudreau, and Bretz, 1994; Goﬂ‘, Mount, and Jamison, 1990), have actually examined
the degree to which organizational efforts to be family-supportive actually reduce the
amount of conﬂict between work and family experienced by employees (Kossek and
Ozeki, 1998). Also important to understand is how the work/family interface —— including
organizational supports — is related to burnout (Maslach and Jackson, 1985), and this is
another area where limited work has been done. Recent meta-analytic reviews have
reported evidence of strong links from conﬂict between work and family to burnout, but
only a few studies that examined such relationships, all with unique samples which make
it difﬁcult to generalize their results (Kossek and Ozeki, 1998; Allen et al., 2000).
Conﬂict Between Work and Family An [ml-taut Outcome

While researchers interested in human resource policies and practices have tended
to focus on familiar employment-related outcomes like organizational commitment and
performance, a large stream of research on conﬂict between work and family has shown
the importance of this construct, both as an unpleasant outcome for individuals and as

precursor of outcomes unfavorable for employers. Conﬂict between work and family has

been lini.

 

2000 ), p1
physical
depressir‘l
dwbm
related to
commitm
Performa:
01., 1988.
have, in g:
StIessors a

OUIcomes

 

Barnum a
As
i"dictates j
mmmm
Widrome,
emmlorml c
Leiler and I
dlained, blu
the moming

and the Olga

been linked to lower satisfaction with marriage, family and life in general (Allen et al.,
2000), psychological strain (Grandey and Cropanzano, 1999; Burke, 1988), poorer
physical health (Googins and Burden, 1987; F rone, Russell and Barnes, 1996), and
depression (Frone, Russell, and C00per, 1992). On the employment side, studies have
also found that conﬂicts associated with combining work and family responsibilities are
related to reduced job satisfaction (Kossek and Ozeki, 1998), lower organizational
commitment (Good, Sisler and Gentry, 1988; Lyness and Thompson, 1997), poorer job
performance (Frone, Yardley, and Markel, 1997), and greater interest in quitting (Good et
al., 1988; Ayree, 1992). While considerable research has been done in this area, studies
have, in general, focused on the relationships between gender, job- and family-related
stressors and negative spillover or conﬂict, and on the relationship of conﬂict to
outcomes such as those described above. The effects of speciﬁc work/life policies have
not generally been included (Kossek and Ozeki, 1998).
Burnout a Critical Outcome for Study

As mentioned previously, burnout is another negative outcome that research
indicates is linked to conﬂict between work and family, and it is an outcome that is
important both for employees and employers. The most critical part of the burnout
syndrome, validated in a large stream of previous research, is a work-related state of
emotional exhaustion (Cordes and Dougherty, 1993; Lee and Ashforth, 1993; 1996;
Leiter and Durup, 1996). Emotional exhaustion refers to a sense of feeling completely
drained, burned out, tired when you think about facing another day — even first thing in
the morning Such deep-seated fatigue obviously has negative implications for workers

and the organizations who employ them. Research has shown that workers with higher

iC\€lS or:
their org;
their org;|
has also 1'
satisfactilI
c |
than orig:
mainly \\
female-dr
The conCe
use Of "p;-
that W'orkt
meme”
Wes ofjg
and What 1.
majority 01

[he Slﬁdrm

 

levels of emotional exhaustion are less satisfied with their jobs, are not as committed to
their organizations, have greater intention to quit and are more likely to actually leave
their organizations (Lee and Ashforth, 1996; Wright and Cropanzano, 1998). Burnout
has also been linked to negative outcomes on the home front, including reduced marital
satisfaction (Leiter and Durup, 1996).

Current research on burnout indicates that this syndrome may affect more workers
than originally thought. Until quite recently, burnout was conceptualized as affecting
mainly workers whose jobs involved intense interaction with others, particularly those in
female-dominated human service professions like nursing, teaching, and social services.
The concept was gradually expanded to cover other occupations that required extensive
use of “people skills”, like sales and management. However, recent research indicates
that workers in a wide range of professions - even those that involve little contact with
peOple - may also suffer from burnout (Demerouti et al., 2001). If workers in most or all
types of jobs are susceptible to burnout, understanding what causes emotional exhaustion
and what helps to alleviate it is extremely important. To date, with few exceptions, the
majority of burnout research has focused on job-related stressors and buffers and on how
the syndrome unfolds, relying on homogenous samples of service workers.

The handful of studies that have examined burnout in the context of work and
family responsibilities have found strong correlations with conﬂict between work and
family, particularly for women (e. g. Greenglass and Burke, 1989; Leiter and Durup,
1996; Bacharach et al., 1991; Etzion, 1988; Netemeyer er al., 1996). These relationships
imply that successful efforts to reduce conﬂict between work and family have the

potential to reduce emotional exhaustion as well. Organizational efforts to create a

 

family-s;
both con
family».l

R

 

 

combina:
tried to re
worklife
work em

H1

 

the job an
members,
telecomm
Challenges
“’Orkers n

A]
limited, tl‘.
companiej
telecOmm
9111mm (F1
Potential S
orgam- 2:11;.
in Starling

home 0' o:

farnily-supportive work environment represent a key resource that may help decrease
both conﬂict and bumout.
Family-.S'upportive Environments: Formal and Informal Elements

Recognizing that today’s workers are increasing challenged by a complex
combination of often-conﬂicting responsibilities at work and at home, employers have
tried to respond to the needs of their changing workforce by introducing a wide range of
work/life policies and initiatives that can contribute to the creation of a family-supportive
work environment.

How can employers support workers in their efforts to combine responsibilities on
the job and at home? Formal beneﬁts like time off to care for new babies and sick family
members, help locating and paying for dependent care assistance, ﬂextime and
telecommuting programs represent important signals that employers recognize the
challenges involved in combining work and family and are making efforts to help
workers meet them.

Although research on the effects of such indicators of workplace support is
limited, their use is fairly widespread. During the heady economic boom of the 19905,
companies competing for quality personnel developed programs introducing
telecommuting and ﬂexible hours, and deliberately worked to create a family-friendly
culture (Feldman and Gainey, 1997; Osterman 1995), something many ﬁrms viewed as a
potential source of competitive advantage (Allen, 2000). In a recent survey of 1,057
organizations with 100 or more employees, 68 percent reported offering some ﬂexibility
in starting and quitting times and 33 percent said they allowed employees to work at

home or off-site on a regular basis. While only 9 percent provided child care at or near

the work

for child

 

 

 

child car.
emploxejl
lrl
other pen
family-so
supervise
1999; Alll
Work fan:
al., 1999;
Th

 

has been d

 

0f family-5
described .
illeImmC 1
envimnmeI
exhaustion

can 2111311211;

the work site, nearly half offered dependent care assistance plans that help employees pay
for child care with pretax dollars and 36 percent provided access to information on local
child care providers. Twenty-three percent said they provided similar referrals for
employees seeking assistance with eldercare issues (Galinsky and Bond, 1998).

In addition to formal programs and policies, research indicates that the attitudes of
other people in the workplace can make an important contribution to the creation of a
family-supportive work environment. Particularly important are how supportive
supervisors are when workers attempt to resolve work/family conﬂicts (Thompson et al.,
1999; Allen, 2000) and cultural norms that don’t penalize workers’ careers for utilizing
work/family beneﬁts or putting family needs ﬁrst when they are important (Thompson et
al.1999)

The effects of supervisor support with work issues on conﬂict between work and
family and on burnout have been studied fairly extensively; however, far less research
has been done on the effects of supervisor support with work/family issues. The effects
of family-supportive cultural norms, and well as formal policies and programs like those
described above, represent important areas where research is needed. Critical gaps in the
literature include information on whether and how a supportive organizational
environment affects conﬂict that emanates from family to affect work and emotional
exhaustion. In addition, our knowledge of the degree to which a supportive environment

can alleviate conﬂict that emanates from work to affect family is sketchy at best.

Purpose

 

I
work eml
context 0
Previous
be inﬂ uei
beneﬁu. LI
material 1
and a ma} .
examininzl
related de
M the “or

heleul.

 

Th

Tammy and

Purpose of the Study

The main purpose of this study is to examine the effects of a family-supportive
work environment on emotional exhaustion and conﬂict between work and family in the
context of other relevant demands and resources in both domains, work and home.
Previous research has helped to identify these constructs as important outcomes likely to
be inﬂuenced by such a organizational support, and work in all three areas - work-family
beneﬁts, burnout, and work-family conﬂict — provides rich theoretical and empirical
material to draw on. However, research that draws on all three perspectives is limited,
and a major goal of this paper is to link the thinking and ﬁndings of these literatures. By
examining the effects of a supportive environment relative to other work- and family-
related demands and resources, the study should contribute to a clearer conceptualization
of the work/family interface. It should help to identify what poses the strongest
challenges for US workers in combining work and family today, and what appears to be
helpful.

The study will examine these issues in a large sample of workers from all walks
of life and all areas of the US, in contrast to the vast majority of research on work and
family and burnout as well, which has generally relied on small convenience samples
comprised of workers in the same ﬁeld and/or organization. This should enable the study
to make another contribution, by examining how much previous ﬁndings can be
generalized to the population of US working adults.

Main Research Questions
The study will attempt to shed light on these questions: How do work and family

demands and resources combine to create conﬂict between work and family and

emotion

 

efforts It
‘ I
envrronn
and farm
cmironn
lower-inc
CODITIbL'
Ti'
family-fr;
how it al‘.‘
examini né
50W econr

“QM”

 

Walks of 11
Personal 211
Supportive
Pleture of 1
eXhausu'On
laden-ﬁn:

A n
mmwm
“hammer,

emotional exhaustion? In particular, how are these outcomes affected by organizational
efforts to create a more family-supportive environment? Does having such an
environment impact emotional exhaustion through its effect on conﬂict between work
and family? Do more positive outcomes result from working in a family-supportive
environment for individuals with lower levels of personal resources, such as those from
lower-income households and single parents?
Contributions

The study will contribute to the literature by being the ﬁrst to look at how a
family-friendly organizational environment may affect burnout, the ﬁrst to investigate
how it affects family-to-work conﬂict in a diverse sample, and one of the most broad
examining its relationship to work-to-family conﬂict. It will is also the ﬁrst to consider
how economic status (an important resource) may moderate these relationships. The
study will maximize variance in all measures by utilizing a sample with workers from all
walks of life, making it easier to see relations and generalize results. By including both
personal and work-related resources and demands in addition to measures of how family-
supportive the work environment is, the study should help provide a more complete
picture of the interface between work and family and how it may be related to emotional
exhaustion.
Underlying Assumptions

A number of important assumptions underlie the design of the study. First, it is
assumed that, consistent with recent research and reconceptualizations, emotional
exhaustion is a construct that has applicability for workers in all occupations. While

most work in this area has previously focused on workers in jobs that involved

1
I

evidence

Demeror

 

consider

makes it
and occu

would be

them.

and contl
meaning?
alld fam11~

hCIC \Ve m

 

the Conﬂic
Again, pr
iImportant
19964 Net:
moderately
meaningfu
“summer
as Selma:

A ll
mean‘ineﬁil

considerable human interaction, the few studies that do involve broader samples ﬁnd
evidence of burnout, particularly emotional exhaustion, in all types of workers (e.g.
Demerouti et al., 2001; Ayree, 1993; Schutte er al., 1999). The design of the study also
makes it possible to evaluate the possibility that this assumption does not hold. Industry
and occupation are entered as control variables; large effect sizes for those variables
would be a strong indicator that there are, indeed, important differences associated with
them.

A second assumption is that emotional exhaustion, conﬂict from work to family,
and conﬂict from family to work represent three distinct constructs that can be
meaningfully distinguished from one another. The difference between work-to—family
and farnily-to-work conﬂict will be discussed in greater detail in the following chapter",
here we note that both are outcomes of interest in this study, based on the assumption that
the conﬂict between work and family can be differentiated by the direction of effects.
Again, previous research provides strong indications that these represent separate and
important ideas. In studies that include combinations of them (e. g. Leiter and Durup,
1996; Netemeyer et al., 1996, Frone et al., 1992), correlations between them are
moderately strong, but generally not above a 6.0, which might indicate that they might be
meaningfully integrated. The design of the study also allows for examination of this
assumption; factor analyses and intercorrelations are examined to verify whether these
are separate constructs.

A third assumption is that research questions described above can be
meaningfully examined in a cross-sectional analysis. Cross-sectional research cannot

prove the existence of effects that unfold over time. It can, however, reveal whether

 

as follow
related to
this secti

Chélpter 3
literature 1
sample. sc
Smdl'. incl
CODCIUdeS
research :11

and main (

constructs and variables are related as would be predicted, thus shedding light on whether
causal effects may exist without necessarily proving that they do. The question becomes,
then, whether such cross-sectional research is useful. Like much of the work in this area
and the social sciences in general, this study adopts the perspective that it is, provided
that the limitations of cross—sectional research are considered when drawing conclusions.
Organization of the Dissertation

With these assumptions in mind, the organization of this dissertation will proceed
as follows. The major constructs and theories underlying the study as well as research
related to them will be explained in greater detail in Chapter 2, which directly follows
this section. The model this research draws on will be described at the beginning of
Chapter 3, followed by a description of the hypotheses to be tested and a review of
literature relevant to them. The research methods used, including a discussion of the
sample, scales, and analytical approach, will be covered in Chapter 4. The results of the
study, including tables summarizing the ﬁndings, are included in Chapter 5. Chapter 6
concludes the dissertation with a discussion of major ﬁndings, and their implications for
research and practice. Information on study limitations, suggestions for future research,

and main contributions of the study conclude this chapter.

10

research

outcome
emironr

discussio

emironm
organza:
consider I
outcomes
TCSponseg
from each
construct

As
increasing?
their Work
“COT ill I] g I

among mar,

 

CHAPTER 2: THEORETICAL UNDERPINNINGS
AND PREVIOUS RESEARCH

The constructs emphasized in this study and the theories that have inﬂuenced
research about them are described in this chapter, which begins with deﬁnitions of the
outcome variables and a discussion of the elements of a family-supportive work
environment. Previous research in each area is brieﬂy summarized, followed by a
discussion of the four main theories that inform the current study.

The main independent variable of interest in this study, a family-supportive work
environment, is a multi-faceted construct. Given the high level of resources that
organizations have committed to becoming more “family-friendly,” it is important to
consider the ways that this can be accomplished and how effective they are. The
outcomes chosen for study share important characteristics: they represent individual-level
responses to stresses. However, as will be seen from the discussion to follow, they differ
from each other in key respects.

Construct Deﬁnition: A Family-Supportive Work Environment

As noted in the introduction, changing demographics have lead US employers to
increasingly adopt formal policies and programs aimed at helping employees balance
their work and nonwork responsibilities (Galinsky and Bond, 1998). Equally important,
according to recent research, are efforts to bolster informal support for struggling workers
among managers (Allen, 2000) and to create a culture that does not penalize workers’
careers when they make decisions that show a high value placed on family (Thompson et

al., 1997). Workers consider the existence of formal policies and beneﬁts, combined

11

111th ass
how far
f

deﬁned a

demands
consider:

and locat

—_- _

with fam:
decisions
Previous

Vl‘
been enm-
hal'e tend,
“ﬁh P0511;
Allen, 209*
Tho“IDS-on
al., 1989, II
mama} Stat
conﬁrm am

 

The
both 1for}; a‘
1

film} these 5

 

with assessments of informal support, and use them to develop overall perceptions of
how family-supportive their organizations are (Allen, 2000).

For the purpose of this study, a family-supportive organizational environment is
deﬁned as one where workers are supported in their efforts to balance work and family
demands. Five indicators of support identiﬁed as important in previous research are
considered as contributing to such an environment. They are: ﬂexibility in work timing
and location, the extensiveness of dependent care beneﬁts provided, supervisor support
with family issues, and cultural norms that do not penalize workers’ careers for making
decisions that reﬂect a strong emphasis on family.

Previous Research on Family-Supportive Organizational Environments

While some recent work in the area of organizational work/family supports has
been enriched by drawing heavily on the work/family role conﬂict literature, most studies
have tended to look at whether the existence or use of such such beneﬁts is associated
with positive work-related outcomes, such as greater job satisfaction (Rothausen, 1994;
Allen, 2001) and increased loyalty or commitment (Roehling et al., 2001; Allen, 2001;
Thompson et al., 1999; Grover and Crooker, 1995). In some studies (e.g. Greenberger et
al., 1989) the sample is limited to working parents. In other cases, researchers have used
marital status, gender or the presence of children as a type of proxy for work/family
conﬂict and have looked to see whether more positive effects are found for people more
likely to be dealing with work/family issues (e.g. Grover and Crooker, 1995).

The notion that providing resources for employees struggling with demands from
both work and home leads to positive work attitudes has received considerable support

from these studies. However, even non-parents and other workers who appear to have

12

less nee
organiz
employ
effects 0

(eg Gl’C

 

have pro
to worke

fairly tre '

 

less need for a more family-friendly work environment have more positive attitudes in
organizations that offer such support. Flexibility, in particular, seems to be valued by
employees of all types. Dependent care beneﬁts have also been found to have positive
effects on work attitudes among employees milikely to use them in some cases as well

(e. g. Grover and Crooker, 1995, Roehling er al., 2001). One explanation: researchers
have proposed that a family-ﬁiendly environment is a powerful symbol which indicates
to workers that the organization cares about them and their concerns, and that they will be
fairly treated (Grover and Crooker, 1995; Allen, 2001).

Whatever the reasons, research on how a supportive work environment affects
work attitudes has shown some signiﬁcant positive relationships, a particularly important
outcome for employers. However, just as important to understand is how the
organization’s environment may be associated with the level of stress actually
experienced by workers. While some work has addressed the effects of a supportive
environment on work-to-family conﬂict, little is known about how it may be related to
family-to—work conﬂict or emotional exhaustion. This study will contribute to
understanding of the effects of a family-supportive workplace environment by looking at
how indicators of support are related to both work-to-family and family-to-work conﬂict
as well as emotional exhaustion. The large national sample to be used means results
should be quite generalizeable. To a greater extent than is common in most previous
research in this area, personal and job-related factors that past studies indicate may have
an impact will be included in the model to improve the accuracy of estimates about the

effects of workplace support.

13

Deﬁning Work-to-Family and Family-to-Work Conﬂict

According to the most common deﬁnition, work-family conﬂict occurs when the
demands of work and family roles are incompatible in some respect so that participation
in one is more difﬁcult because of participation in the other (Greenhaus and Beutell,
1985). For example, in her role as a manager, a woman may be expected to attend a
meeting with an important client, but at the same time, in her role as a parent, she may be
expected to care for a sick child.

Considerable research supports the notion that conﬂict between work and family
has two important dimensions, deﬁned by the direction of the effects (Netemeyer et al.,
1996; Carlson et al., 2000). For example, our manager may have to put in many hours of
overtime one week, reducing the amount of time and energy she has to help her son with
a science project. Because the effects emanate from work and impact family, this is
termed work-to-family conﬂict. If our manager were to miss an important deadline for
reporting results because she was helping an elderly parent obtain emergency medical
treatment, the effects would emanate from family to impact work; this would be an
example of family-to-work conﬂict. For purposes of simplicity, where both types of
conﬂict are discussed, the term work/family conﬂict will be used throughout the
remained of this paper.

Research on Work/Family Conﬂict

In studies that have distinguished between ways that work has a negative effect
upon family and ways that family has a negative effect upon work (e.g. Gutek, Searle,

and Klepa, 1991; Netemeyer et al., 1996; Wiley, 1987; Leiter and Durup, 1996; and

Adams et al., 1996), workers have generally reported more work-to-family conﬂict than

14

family-’
to ensur

helps to

spurred
consider
of work
Consequ
VI
function :
depressn
conﬂict 1.
Cllect 0n
have beer
10b 8&1le ;

l

”ﬁller an

1996), de

Somewhat
CXcem lot

“’Ork (K05

 

family-to-work conﬂict, perhaps because when choices must be made people are careful
to ensure that their personal lives don’t negatively affect their performance at a job that
helps to support their lifestyles.

Since the 19705, when the increasing number of women entering the workforce
spurred interest in the intersection of work and family (Barnett and Hyde, 2001),
considerable research has been done on the consequences and antecedents of both types
of work/family conﬂict.

Consequences of Conﬂict

Work-to-family conﬂict has, by deﬁnition, strong implications for family
functioning, and it has been associated with reduced family and marital satisfaction,
depression, and even alcohol abuse (Allen et al., 2000). Extensive family-to-work
conﬂict is likely to have a negative impact on an employee’s career, and, through its
effect on individual performance, can be harmful to organizations. Both types of conﬂict
have been associated with negative work outcomes in previous research, including lower
job satisfaction (Kossek and Ozeki, 1998), more interest in quitting (Netemeyer et al.,
1996), decreased commitment (Wiley, 1987; Netemeyer et al., 1996), greater burnout
(Leiter and Durup, 1996; Netemeyer et al., 1996), and poorer performance (N etemeyer et
al., 1996). A recent meta-analytic review indicates that work-to-family conﬂict may be
somewhat more closely associated with these outcomes than family-to-work conﬂict,
except for performance, which suffered more when family demands spilled over to affect

work (Kossek and Ozeki, 1999).

15

Antecedents of Conﬂict

Although they have frequently been included in research designs as controls rather than
major emphases of research, demographic and family characteristics have been shown to
affect the experience of work/family conﬂict. Much of the research in this area has relied
on homogenous samples of workers, for example, female health care workers (Bacharach
er al., 1991; Garland, Oyabu and Gipson), and has limited participants to those with
children and or spouses (e.g. Leiter and Durup, 1996; Ayree, 1993). Despite a tendency
for variance to be reduced through such sample selection techniques, there is support for
the practical notion that people who tend to have greater family demands — women,
partners in dual-career relationships, and single parents — are more susceptible to conﬂict
between work and family (e.g. Kossek, 1990; Pleck, Staines and Lang, 1980). Other
inﬂuences from the family side include instrumental and social support from family
members, as well as strain within family roles, such as ambiguity or conﬂict (Frone,
Yardley, and Markel, 1997; Parasuraman et al., 1992). The results of studies in these
areas are not completely consistent, implying that research that increases our
understanding would be useful.

One ﬁnding does seem to be consistent: family-related variables appear to have
the strongest impact on family-to-work conﬂict, while work-related inﬂuences are more
closely associated with work-to-family conﬂict (F rone, Russell and C00per, 1992; Frone
Yardley and Markel, 1997; Ayree, Fields and Luk, 1999).

Among the most important job-related inﬂuences is how demanding one’s work
is, whether measured in terms of hours worked (F rone, Yardley and Markel, 1997), a

feeling of being overloaded (Greenhaus et al., 1989), or how hard and fast employees

16

must work (Pleck, Staines and Lang, 1980). Another job characteristic that has been
found to affect work/family conﬂict is type of work schedule (Pleck, Staines, and Lang,
1980). Among the resources identiﬁed in previous research are autonomy, tenure
(Greenhaus et al., 1989), and social support from supervisors and co-workers
(Parasuraman et al., 1992).
Research on Family-Supportive Work Environments and Conﬂict

A relatively small body of work deals with the inﬂuence of an important work-
related resource, organizational efforts to create a family-supportive environment. Only a
handful of studies have looked at the effects of family-related policies and beneﬁts on
work-to-family conﬂict, and family-to-work conﬂict has received even less consideration
in this literature, which will be reviewed more extensively in Chapter 3. In evaluating the
effects of organizational efforts to create a supportive work environment for those
combining work and family roles, it is important to consider both work-to—family and
family-to—work conﬂict. Presumably, companies introducing ﬂextime, building on-site
daycare centers, or offering workers information on where to turn for eldercare assistance
are seeking to reduce both negative effects on families from work and family intrusions
that may limit a worker’s effectiveness. This study will contribute to research in this area
by examining the effects of various personal and organizational factors on both work-to-
family and family-to work conﬂict.

Previous research on the effects of organizational efforts to support workers
balancing job and family demands contains some contradictory ﬁndings, but overall
provides grounds for cautious optimism. It is important to note that only a few studies

published in peer-reviewed academic journals have looked at how organizational efforts

17

10 C1681

i
one of 1
it difTicl
awomeI
graduat.l
I995. m|
ludgel
emploie
midwest
inclusior
\Tl'iuallt
work'fan'
1996‘). 3
family ﬁr
not in 3]]
bmeﬁts 1.
Section.

Sll
homogenc
le g Bon d
POIenii‘ai n
“demand

Olmnﬂiq

 

to create a family-supportive environment are related to work/family conﬂict, and all but
one of these (Galinsky, Bond, and Friedman, 1996) have used unique samples that make
it difﬁcult to generalize (Allen, 2001 , workers in communications ﬁrms and members of
a women’s business association; Thompson, Beauvais and Lyness, 1999, alumni of
graduate programs in business and labor and industrial relations; Thomas and Ganster,
1995, mainly female health care professionals; Ray and Miller, 1994, hospital workers;
Judge, Boudreau, and Bretz, 1994, male executives; Goff, Mount, and Jamison, 1990,
employees at a communications ﬁrm; Kossek and Nichol, 1992 employees of a
midwestem hospital; Greenberger et al., 1989 parents of preschoolers). Aside from its
inclusion in the Judge et al. (1994) study of executives, family-to-work conﬂict has been
virtually ignored or mixed in with work-to-family conﬂict to create a single bi-directional
work/family conﬂict measure (e.g. Goff er al., 1990; Galinsky, Bond and Friedman,
1996). Support from supervisors and coworkers and a culture that allows workers to put
family ﬁrst have been fairly consistently associated with lower work/family conﬂict, but
not in all cases (see Ray and Miller, 1994). Evidence for ﬂexibility and dependent care
beneﬁts is mixed (Kossek and Ozeki, 1999), as will be illustrated in the literature review
section.

Study ﬁndings, then, are not consistent, perhaps inﬂuenced by the reliance on
homogenous samples. Even in cases where large, representative samples have been used
(e. g. Bond, Galinsky and Swanberg, 1998; Galinsky, Bond, and Friedman, 1996)
potential moderators have not been examined. This study will contribute to our
understanding by examining the effects of a supportive work environment on both types

of conﬂict in a large, highly diverse national sample. It will also be the ﬁrst to look at

18

whether people with fewer persona] resources experience a greater reduction in work-
family conﬂict from working in a family-supportive environment than workers who may
have a lighter load at home or more options for help, thus ﬁlling critical gaps in the
literature.

Burnout: Deﬁning the Construct

Strongly inﬂuenced by work by Maslach and colleagues (e.g. Maslach and
Jackson, 1981a,b; Maslach, 1982) burnout has generally been viewed as a syndrome that
has three dimensions: emotional exhaustion, a state in which one feels drained of energy
and enthusiasm for work; depersonalization, a response to excessive demands in which
workers take a cynical view of the people and they work with and the work they do; and
diminished personal accomplishment, a feeling that one is not making a meaningful
contribution through work.

Considerable work in this area has focused on how the dimensions related to each
other, as well as to relevant sources of stress and support. Studies have endorsed
conceptions of burnout as a syndrome involving from two (e.g. Demerouti et al., 2001) to
eight (Golembiewski et al., 1986) phases or dimensions. Several models have been
proposed regarding the order in which the dimensions occur, but there does appear to be
considerable consensus on one point: emotional exhaustion plays a key role in the
process (Lee and Ashforth, 1993; Cordes and Dougherty, 1993; Wright and Cropanzano,
1998). The focus of this research will therefore be on emotional exhaustion, or a sense of

overwhelming fatigue associated with work.

19

Previous Research on Burnout

In earlier research, this syndrome was considered to apply only to workers whose
jobs involved extensive interactions with people, such as teachers, nurses, and social
workers. However, recent studies have found that the concept of burnout and its
dimensions are applicable to a broad range of jobs, even those which involve little
contact with people (Demerouti et al., 2001).
Consequences of Burnout

Research has shown that burnout has negative effects for both employees and
employers. A recent meta-analysis found that burned out workers are less satisﬁed with
their families, their marriages, and their lives. They have lower levels of job satisfaction,
and organizational commitment, and are more likely to both consider quitting and to
actually do so (Lee and Ashforth, 1996). Given these negative consequences, it is
important to develop a complete understanding of what causes burnout and how it can be
prevented.
Antecedents of Emotional Exhaustion

Previous research has identiﬁed a number of resources and demands associated
with emotional exhaustion, mainly in the work domain. A recent meta-analysis of the
burnout literature found that key resources linked to lower work-related burnout include
social support and autonomy. The strongest relationships, however, appear to be for
demands, particularly workload. Also closely related were conﬂict and ambiguity within
the work domain, although there are concerns that similarity between items commonly
used to measure emotional exhaustion, within-role conﬂict, and ambiguity may be

worded in ways that lead to overestimates the strength of their relationships (Lee and

20

Ashforth, 1996). As noted above, the same work-related resources and demands have
also been linked to conﬂict between work and family roles (Frone, Russell and Cooper,
1992)

Work/family conﬂict represents another factor that may contribute to increased
emotional exhaustion. The handful of studies that have examined burnout in the context
of work and family responsibilities have found strong relationships with work-to-family
conﬂict, particularly for women (e.g. Greenglass and Burke, 1989 (but no relationship
found for men); Leiter and Durup, 1996; Bacharach et al., 1991; Etzion, 1988;
Netemeyer et al., 1996). Although the relationship is weaker, family-to-work conﬂict
also appears to be correlated with emotional exhaustion (Netemeyer er al., 1996; Leiter
and Durup, 1996).

Research on Family-Supportive Work Environments and Burnout

Given the links that have been found between work/family conﬂict and burnout, it
appears important to examine whether or not efforts to support workers in balancing job
and home responsibilities also have important implications for burnout. If working in a
family-supportive environment is associated with less burnout, organizations would have
another important reason to work towards creating one. However, research in this area is
limited.

Previous studies indicate that supervisors who are supportive when work
problems arise contribute to lower levels of emotional exhaustion (Lee and Ashforth,
1996), and similar results have been found for supervisor support with family-related
issues (e.g. Leiter and Durup, 1996; Ray and Miller, 1994). However, only a single study

was located that looked at the relationship between schedule ﬂexibility and burnout. In a

21

study of dual-career parents of preschoolers in Singapore, Ayree (1993) found that
greater difficulty in adjusting work days and times was associated with more emotional
exhaustion among women, but not men. He posited that the differences reﬂected
Singaporean expectations regarding the greater involvement of women in their childrens’
education, underscoring the need for further study in a contrasting sample of workers.
No studies were located that dealt with the relationship between dependent care
assistance, or organizational cultural norms and any of the burnout dimensions.

This study will extend the limited research that has looked at the effects of family
responsibilities on burnout by exploring whether organizational efforts to provide a
family-supportive environment are associated with lower levels of emotional exhaustion,
particularly for workers who have high personal demands or who are less able to afford
alternate support. Another contribution will be looking at the degree to which a
supportive enviromnent affects emotional exhaustion through its impact on work/family
conﬂict, and the degree to which it reduces it directly or in other ways.

Major Theories Underlying the Study

Three theoretical frameworks help to identify the ways in which family-
supportive work environments, work-to-family conﬂict, family-to-work conﬂict, and
emotional exhaustion are linked. While theories of social exchange and symbolism have
exerted the greatest inﬂuence on previous research related to family-related beneﬁts,
incorporating the insights of a major inﬂuence underlying work/family research, role
theory, can enrich and extend thinking in this area. The model proposed by the

conservation of resources theory, which has informed much of the more recent research

22

on 13111
influei

Work

is som
given 1
positix
this the
cases \\
useful.

be foun

 

on burnout, can be used to develop a logical perspective and may help identify important
inﬂuences.
Work Environment Studies: Theories of Social Exchange and Symbolism

Underlying the majority of studies in the area of work/family beneﬁts and policies
is some form of social exchange or balance theory; researchers expected that employees
given more beneﬁts would repay their organizations for supporting them with more
positive work attitudes (Grover and Crooker, 1995; Roehling er al., 2001). Implied by
this theoretical perspective is the notion that reciprocation from workers is most likely in
cases where a supportive work-family environment is noticed by employees and found
useful. That is, stronger effects for organizational work/family support are most likely to
be found among workers who ﬁnd them the most useful.

As discussed earlier, research based on these theories has found that workers in
supportive environments have more positive attitudes, but the notion that workers most
likely to use beneﬁts would have the most positive attitudes received less support (Grover
and Crooker, 1995; Roehling et al., 2000). For example, parents of young children or
people who were anticipating having a child in the near future were not more inﬂuenced
by dependent care support and maternity/paternity leave policies (Grover and Crooker,
1995). One explanation researchers have proposed is that a family-ﬁiendly environment
may act as a powerful symbol which indicates to workers that the organization cares
about them and their concerns, and that they will be fairly treated (Grover and Crooker,
1995; Allen, 2001).

The theoretical framework underlying previous research in this area has three

important implications for the cunent study. First, while emotional exhaustion is

23

conceptualized more as a state than a job attitude, it may be that workers who receive
greater support will reciprocate by dedicating a larger amount of their energy to their
work. Second, working in a family-supportive environment may have the most beneﬁt
for workers who have the most need for assistance in this area, either due to a higher
level of demands in the two domains or a lack of alternative resources for coping with
them. Third, even if workers do not derive clear beneﬁts from working in a supportive
environment (for example, because they have few family responsibilities), the existence
of supportive policies and norms may serve as a symbol of overall concern and support
on the part of the organization. Perceptions that support is available — even if it is never
used — have been associated with less burnout (Lee and Ashforth, 1993).
Underlying Research on Work and Family: Role Theory

For over twenty years, role theory has provided an important framework for
research on the relationships between work and family (Voydanoff, 1988). Both on the
job and in the home individuals face behavioral expectations often termed “roles”
(Jackson and Schuler, 1995; Katz and Kahn, 1978). For example, in her role as a parent,
3 mother may be expected to assist her children with homework and see that they are
appropriately dressed and fed. In her role as a manager, the same woman may be
expected to meet with clients, assign responsibility for different projects among her
subordinates, and evaluate their work. It is clear that the very deﬁnition of work/family
conﬂict is based on the belief that people occupy multiple roles with expectations that can
conﬂict. As can be seen from the this example, not all role expectations are necessarily

associated with conﬂict, but in the case of our manager, time and energy her family

24

expects her to use for helping with homework may sometimes be consumed in meetings
with clients and subordinates.

Role theory as it has been applied to work/family research highlights the
importance of considering all of the roles that are relevant in both domains, as well as the
most demanding expectations associated with them, because these are the most likely to
be associated with conﬂicts. Research in the area of work/family conﬂict supports this;
major stressors associated with the parental, spouse, and employment roles all have been
shown to have important effects, particularly eldercare (Martire and Stephens, 2003),
children, work hours, and work scheduling (Frone et al., 1992; 1997).

Also drawing from role theory is the notion each role one occupies may offer
enriching experiences and resources (Marks, 1977; Barnett and Hyde, 2000), again
highlighting the importance of considering the various roles that are applicable both at
home and at work and the beneﬁts they may provide. Social support from family
members, supervisors and co-workers has been identiﬁed as a beneﬁt that may help
buffer the effects of stressors in the two domains, as have greater life and work
experience and autonomy (Frone et al., 1992; 1997; Ayree and Luk, 1999).

The importance of including these constructs in any effort to examine inﬂuences
on work-to-family and family-to-work conﬂict represents a key implication role theory
has for the present study.

Helping Explain Emotional Exhaustion: Conservation of Resources Theory

An emphasis on resources is the distinguishing feature of a theoretical framework

that has informed much recent research on burnout over the past few years (e.g.

Demerouti er al., 2001; Wright and Cropanzano, 1998; Lee and Ashforth, 1996) as well

25

as two recent studies involving work-farnily conﬂict (Allen, 2000; Grandey and
Cropanzano, 1998). The conservation of resources (COR) model proposed by Hobfall
(1989) holds that when the resources people rely on are suddenly reduced, threatened, or
are insufﬁcient to help them meet the demands they face, stress and other negative
personal consequences result. Perceptions of resource shortfalls that continue for
relatively long periods of time may lead to burnout, particularly emotional exhaustion. In
response, people look for ways to conserve their resources, often through withdrawal
behaviors like quitting a demanding job (Hobfall, 1989; Lee and Ashforth, 1996).

The COR model is broad enough to subsume much of the research and theory on
stress, including not only burnout, but also conﬂict between work and family. The
breadth and ﬂexibility of this theoretical framework allow it to incorporate the most
important contributions of the other theoretical ﬁameworks described in this section.
Therefore, it has been used as the primary theory in developing hypotheses for the current
study.

The COR model implies that heavy expectations and enhancing features of work
and family roles identiﬁed by role theory-based research as important can be viewed as
demands and resources. Within this framework, work/family conﬂict can be considered
as a stress-related outcome (Grandey and Cropanzano, 1998) and a family supportive
work environment can be seen as potentially important resource (Allen, 2000) that may
supplement those that previous research has indicated are important at work and at home.
Conﬂict between work and family may also be seen as another demand that likely to

increase the experience of burnout.

26

The implications of the theories and research described in this chapter are further

deveIOped in Chapter 3, which introduces the hypotheses to be examined by this study.

27

CHAPTER 3: LITERATURE REVIEW AND HYPOTHESES

The conservation of resources theory implies that work-to—family conﬂict, family-
to—work conﬂict, and emotional exhaustion may all be caused by having insufﬁcient
resources to meet demands in the two realms, work and home. Workers arrive on the job
with personal demands and resources arising from their personal and family situations.
Employers have somewhat more control over work-related demands and resources,
which may vary signiﬁcantly based on the characteristics of the job and the worker. At
the organizational level, they can work to create a more family-supportive environment.
An overview of the main model guiding this research is provided in Figures 1 and 2,
which appear on the following pages. As can be seen by examining the model, the three
outcome variables, work-to-family conﬂict and family-to-work conﬂict are presumed to
be inﬂuenced by three sets of variables: personal demands and resources, work demands
and resources, and in addition to job-level factors, the degree of support for balancing
work and family evidenced by the environment of the employing organization. Personal
demands and resources, work-related demands and resources, and the organization’s
work/family environment are also hypothesized to affect emotional exhaustion, both
directly (see model 1) and indirectly (see model 2). Indeed, the effects of a supportive
work environment are expected to be largely mediated by work-to-family and family-to-
work conﬂict. Moderating effects are also predicted for the variables of greatest interest
in the study, those related to a supportive organizational environment. Illustrated in
ﬁgure 2, these will be discussed at the end of this section. As noted in model 1, race,

occupation, and industry are included as control variables in all analyses.

28

Figure 1: Research Model - Direct Effects

 

 

Personal Demands (+)
Sex
Elder/disabled care time
# kids 0-12
# kids 13-18

 

 

 

 

Personal Resources (-)
Partner
Familial child care
Age
Household income

 

 

 

 

 

 

Work Demands (+)
Weekly work hours
Irregular schedule

 

Work-to-
Family
Conﬂict

 

 

 

 

 

 

 

Work Resources (-)
Autonomy
Tenure

 

 

Family-to-
Work Conﬂict

 

 

 

 

 

 

 

F amily-Su pportive

Organizational Environment (-)

Flexibility in work time
Flexibility in work location
Dependent care beneﬁts
Supervisor support
Cultural norms

 

 

29

 

Emotional
Exhaustion

 

 

 

Controls
Occupation
Industry
Races

 

 

 

 

 

Figure 2: Research Model - Indirect Effects

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Household
Income

11‘.
Family-Supportive :
Organizational
Environment *
Flexibility in work time H13 Work/Family H13 Emotional
Flexibility in workplace ’ Conﬂict . Exhaustion
Dependent care beneﬁts (Both Directions)
Supervisor support
Cultural norms E

N

 

 

 

 

 

 

Controls
Occupation
Industry
Race

Single Parent

 

 

 

 

 

 

Personal demands indicators include sex, weekly hours spent caring for elderly
and disabled adults, and the number of children 0-12 and 13-18. While the other
indicators clearly represent sources of demands, gender is included because women often
shoulder a heavier load at home than men (Voydanoff, 2002). Higher demands are
expected to increase conﬂict and burnout, so positive relationships are anticipated
between the personal demand indicators and work-to-family conﬂict, family-to-work
conﬂict, and emotional exhaustion. Personal resources considered here include living
with a partner, having a family member provide some or all needed child care, a higher
household income level, and the wisdom we associate with being older. Higher levels of

resources are expected to reduce conﬂict and burnout, so negative relationships are

30

resources are expected to reduce conﬂict and burnout, so negative relationships are
anticipated between the personal resource indicators and work-to-family conﬂict, family-
to—work conﬂict, and emotional exhaustion. Personal demands and resources are
expected to exert the strongest inﬂuence on family-to-work conﬂict, since by its nature
such conﬂict springs from the nonwork domain.

Work demands are represented by longer work hours and an irregular schedule;
they are expected to have positive relationships with the three outcome variables,
reﬂecting COR theory’s implications that greater demands are associated with negative
consequences. Work resources are represented by autonomy and work experience, and
are expected to have negative relationships with all three outcome variables. However,
as work-to-family conﬂict and burnout both involve work-associated depletion of
resources, work-related factors are expected to be more important than personal ones in
explaining them. Therefore, work demands and resources are expected to exert the
strongest inﬂuence on these two outcomes.

Hours, schedules, autonomy and tenure affect workers at the individual level, and
vary based on the job or person. At the organizational level employers have recently
shown and interest in supplementing the personal and work resources employees have for
dealing with complex combinations of demands on the job and home front. A family-
supportive environment represents a potentially important resource for today’s workers,
and one that may help reduce work-to-family and family-to-work conﬂict, as well as
emotional exhaustion. Five indicators of the employing organization’s work/family
environment have been selected, including the degree of ﬂexibility in work timing and

location, the number of dependent care beneﬁts offered, the degree to which supervisors

31

are supportive when work/family issues arise, and cultural norms about how choosing
family over work can affect employees’ careers. A more family-supportive
organizational environment is expected to be associated with less conﬂict and emotional
exhaustion; negative relationships between these indicators and the outcome variables are
expected.

The major focus of this study is looking at the effects of a family-supportive work
environment on work-to—family conﬂict, family-to—work conﬂict, and emotional
exhaustion in the context of other relevant demands and resources in both the work and
nonwork domains. Therefore, more complex hypotheses involving mediating and
interaction effects will be tested for this set of variables, illustrated in model 2, which
shows hypotheses related to indirect effects.

Several previous studies have identiﬁed work/family role conﬂict as a correlate of
burnout, and in general it has been viewed as a stressor that may contribute to increased
emotional exhaustion (e. g. Ayree, 1993; Greenglass and Burke, 1988). As family-
supportive programs, policies, and initiatives are designed to help employees more
successfully manage the combined demands of work and nonwork roles, the clear
irnplicatiou is that they will reduce conﬂict between work and family in both directions.
It is expected that, by reducing such conﬂict, a family-supportive work environment can
also help to alleviate emotional exhaustion. In other words, work-to-family and family-
to-work conﬂict are expected to largely mediate the effects of a supportive work
environment on emotional exhaustion. It is also possible that a supportive work

environment may have a direct effect on emotional exhaustion by creating a sense of

32

support, but the policies studied here are not designed to do so. Therefore, because of the
nature of the programs, fairly strong mediation effects are expected

Illustrated in model 2 is the notion, derived from COR theory, that the
combination of demands and resources is important in inﬂuencing outcomes, implying
that workers with fewer alternate resources will beneﬁt most from working in a
supportive organizational environment. While a large number of combinations could be
examined, two will be the focus of this study because workers in these groups are
relatively easy to identify and their circumstances appear to make them particularly
strong candidates for needing additional assistance to deal with the work/home interface.
Lower income workers are less able to purchase assistance with household tasks and
child care, and thus may be expected to beneﬁt more from working in a supportive
organizational environment. Support may also be more useful to single parents, who
have higher demands and fewer resources on the home front, than other workers.

It is important to note that the model presented here, although it appears complex,
is actually a simpliﬁcation of the work/family interface. Several relationships between
the study variables that have been identiﬁed in previous research but are not a focus of
this study are not shown. For example, although they appear to be separate constructs,
research has consistently found signiﬁcant relationships between work-to-family and
family-to-work conﬂict (e.g. F rone et al., 1998; Netemeyer et al., 1996). Previous
research also implies that the relationship between burnout and work/family conﬂict may
be reciprocal (Leiter and Durup, 1996), but this idea will not be examined here, as it is
best studied using longitudinal data The goal, as stated earlier, is not to test a

comprehensive model of the work/family interface, but to look at the effects of a

33

supportive organizational environment in the conjunction with demands and resources
that previous research has identiﬁed as likely to inﬂuence the outcomes selected for
focus: work-to-family conﬂict, family-to-work conﬂict, and emotional exhaustion.

In the remainder of the chapter, hypotheses are outlined for the way that personal
and work-related demands and resources, as well as the organization’s environment, are
related to the outcome variables.

Personal Demands and Resources

As noted in the previous chapter, the COR approach to thinking about stress-
related outcomes focuses on the demands faced by and individual and the resources he or
she has available to deal with them. In examining work/family conﬂict and burnout,
demands and resources related to a worker’s personal life are important to consider.
Personal demands are those that employees bring to the job with them; for example, a
family that includes young children, a disabled adult or elderly parent who need to be
cared for. While not necessarily a demand in itself, gender can represent a strong
indicator of demands, as women still appear to shoulder a larger portion of the burden
associated with caring for family members and doing housework (Rothbard, 1999).
Employees, of course, also have personal resources to help them cope with such demands
on the home front. Potential resources include a live-in partner, a family member who
provides most or all needed child care outside school hours, and a higher household
income, which can be used to purchase assistance. Age may also be considered a
resource, as the greater life experience associated with being older often exposes people
to a wider variety of coping techniques and can help them develop patience and a more

balanced perspective (Lee and Ashforth, 1993). COR theory implies that negative stress-

34

related outcomes like work/family conﬂict and burnout will be more likely for workers
with high demands, and may be somewhat alleviated by having more resources. Previous
research and hypotheses regarding the personal demands and resources outlined here will
be described in greater detail in the next section.

Gender as an Indicator of Demands

As noted in the introduction, evidence suggests that work/family values in the US
seem to be changing somewhat, or at least patterns of behavior are. More women are
entering the workforce and remaining there while mothering even yormg children, while
men are contributing more at home (Bond, Galinsky and Swanberg, 1998). However,
even research from the last decade indicates that traditional views of gender roles
continue have an inﬂuence (Gutek, Searle, and Klepa, 1991) and women in general still
take more responsibility for housework and child care than their male partners in most
ethnic groups (Bond, Galinsky and Swanberg, 1998; Goff et al., 1990; Broman, 1988;
Ayree, 1993). While being female, then, is not in itself a “personal demand,” it may be
indicative of a higher level of demands at home.

Gender is the most consistently studied social category in research on work and
family (Voydanoff, 2002). While it may seem obvious that women, because they
generally take more responsibility for children and housework, would experience more
conﬂict between work and family in both directions, research has not always supported
that idea. One recent review (Voydanoff, 2002) concluded that although, particularly
where children are involved, reduced work participation and family interference with
work appear more common for women, there is no strong evidence that gender is related

to work-to-family conﬂict. Considerable published research supports the notion that

35

Women are more prone to family-to-work conﬂict (e.g. Gutek et al., 1991; Kirchmeyer,
1992; Williams and Alliger, 1994; Carlson et al., 2000;Grzywacz et al., 2002). Research
has also shown that women are more likely than their partners to take time off when a
child is sick (Kossek, 1990; Crouter, 1984). However, there are some contradictions.
One study found evidence of family involvement enriching - but not depleting -
resources women had available for work (Rothbard, 2001).

The evidence of a link between gender and work-to-family conﬂict is not
consistent, with no relationship found in several studies (e. g. Allen, 2001; Thompson et
al., 1999) but there is some support. Being female was associated with greater conﬂict,
both from work-to-family and from family-to-work, in a recent study of university
professors (Grandey and Cropanzano, 1999). In a nationally representative study of mid-
life adults, women experienced more conﬂict in both directions (Grzywacz et al., 2002).
A study of working adults in various occupations also found that women experienced
more time-, strain- and behavior-based family-to-work conﬂict along with more strain-
based work-to-family conﬂict (Carlson et al., 2000). Women experienced a depletion of
resources available for family when strongly involved in work, while men did not, in a
study of university employees (Rothbard, 2001). And using a unique design that
involved asked study participants to report on their moods and attitudes at various
intervals Williams and Alliger (1994) found that women experienced stronger spillover
from work to home as well as from home to work.

Research in the area of burnout also indicates that gender may inﬂuence burnout
(Lee and Ashforth, 1993), with studies fairly consistent in showing that men more are

inclined to depersonalization (e. g. Maslach and Jackson, 1985; Greenglass and Burke,

36

1988). While ﬁndings are not consistent, some research also supports the notion that
women are more prone to emotional exhaustion than men (e.g. Maslach and Jackson,
1985; Etzion, 1988; Ayree, 1993). Cordes and Dougherty (1993) report women
experiencing signiﬁcantly greater levels of emotional exhaustion in four studies, but not
in six others. In only one case did men report a statistically signiﬁcant higher degree of
emotional exhaustion, and this was only true for a subgroup made up of managers (Pretty
et al. (1992). All of these studies involved workers in a single occupational area and/or
organization, which is likely to have reduced variance and the likelihood of ﬁnding
effects.

Given that women take on a higher share of domestic work, they may have higher
overall demands than men, and should thus logically be more prone to emotional
exhaustion (Greenglass and Burke, 1988). Women are also more likely to be employed
in the human service professions that burnout was originally conceived about because of
the intensive nature of the interactions involved (Ray and Miller, 1994). At home,
women are often expected to be nurturing, empathetic, and sensitive to others, the same
expectations that are placed on human service workers and considered in much research
to contribute to burnout (Lee and Ashforth, 1996). When forced to deal with demands
both at home and at work, women may ﬁnd that they give to everyone but themselves,
making them prime candidates for burnout (Ray and Miller, 1994; Maslach, 1982).

H l a, b, 0: Being female will be associated with higher levels of work-to-family conﬂicf’,

family-to-work conﬂicf’, and emotional exhaustion“.

37

Caring for Elders and Disabled Adults Can Be a Burden

As Americans have begun living longer, one issue that has attracted increasing
attention is the care of elderly relatives. Parents and other adult family members in ill
health can be a powerful and unpredictable source of personal demands, and one that can
increase negative stress-related outcomes in the COR theory view.

More and more workers are ﬁnding themselves with eldercare responsibilities;
over the course of a year, an estimated one-fourth of US employees provide care for an
elder (Bond, Galinsky and Swanberg, 1998). Eldercare responsibilities are a logical
source of work-to-family conﬂict and a precursor to burnout, and this has generally been
supported by previous research (e.g. Martire and Stephens, 2003). Similar demands and
reactions are likely to be experienced by workers providing care for disabled adults not
yet in their sixties. Because caring for a disabled adult can also present similar
challenges, rewards, and experiences, both have been combined to create a single
measure of special caregiving demands here.

Previous research has shown that eldercare givers are more likely to report work-
to-family conﬂict (Sharlach and Boyd, 1989; Tennestedt and Gonyea, 1994; Martire and
Stephens, 2003). Caregiving has also been associated with increased family-to-work
conﬂict (Martire and Stephens, 2003) as well as absence and tardiness, both indicative of
such family-to—work conﬂict (Anastas et al., 1990; Sharlach and Boyd, 1989; Tennestedt
and Gonyea, 1994). Caregivers have also reported more frequent stress on the job and
high levels of fatigue (Tennestedt and Gonyea, 1994), and can be considered more likely

to experience emotional exhaustion.

38

H2a, b, c: Spending more time caring for elders and disabled adults will be associated
with higher levels of work—to-family conflict", family-to—work conﬂictb, and emotional
exhaustion”.
Children: A Key Source of Demands

On the home front, among the ﬁrst personal demands that come to mind are those
associated with raising children. Infants require almost round-the-clock supervision and
care, and while older children may be more self-sufﬁcient, parents still must take care of
basic needs like food and clothing, support them in their efforts to learn, ensure that they
have appropriate supervision and activities, and provide emotional and ﬁnancial support.
The time demands are considerable; a study of working parents found that they spent, on
average, 3.2 hours on workdays and 8.2 hours on nonworkdays caring for their children.
In addition, while non-parents reported that household tasks required about 2 hours on
workdays and 4.4 hours on nonworkdays, parents spent 2.5 hours on work days and 5.2
hours on nonworkdays on chores (Galinsky, Bond and Friedman, 1996).

Research has shown that having children, especially young children, is associated
with more family-to-work conﬂict (e.g. Grzywacz et al., 2002; Judge et al., 1994).
Findings are less consistent regarding the relationship of age or number of children to
work-to-family conﬂict, with some studies ﬁnding no signiﬁcant relationship (e. g. Allen,
2001; Thompson et al., 1999; Goff et al., 1990) and others ﬁnding a moderate link (e.g.
Judge et al., 1994; Ray and Miller, 1994). This is consistent with a large body of research
that shows that family-related variables are more strongly related to family-to-work
conﬂict, while work-related variables are more closely related to work-to-family conﬂict

(e. g. Frone, Yardley and Markel, 1997; Frone, Russell and Cooper, 1992; Ayree, Fields,

39

and Luk, 1999). Another contributing factor is likely that most studies in this ﬁeld only
include workers with spouses or children, or both, reducing variance. While having more
children means more people who require attention and thus is likely to contribute to
work-to-family conﬂict, the effects are likely to be much weaker than for family-to-work
conﬂict.

But in spite of — or perhaps because of - the great amount of attention that
children require, some previous research has found that parents tend to experience lower
levels of burnout, rather than higher. Maslach and Jackson compared mean levels of
emotional exhaustion for parents and nonparents employed in public contact positions at
a government agency, and found that nonparents were more exhausted. However, they
do not report controlling for age in this analysis, and indeed posit that workers with
children may be older and more mature, which could help them to cope better with job
demands. Other proposed reasons included the notion that children provide needed
distractions and a non-work source of positive feedback and support or that parents are
less likely to see work as their main source of personal fulﬁllment and may not get as
deeply involved in it as nonparents. In the only other study located that actually reported
on the relationship between having children and burnout, Ayree (1993) found a small
positive correlation between the number of children under six and burnout among dual-
career couples in Singapore. The relationship, estimated at r= .1 l for men and .14 for
women, was not statistically signiﬁcant, likely because of the relatively small sample size
and the fact that all participants had at least one young child. Other studies in this area

have also tended to focus on workers with families, reducing variance, and have

40

generally used the number or existence of children as control variables without reporting
associated effects estimates.

Therefore, while studies show that children are likely to increase work/family
conﬂict, previous research has found both positive and negative effects on burnout.
Because this study utilizes the number of children employees are responsible for rather
than parental status, we expect results consistent with a demand perspective —- workers
with more kids are expected to experience greater emotional exhaustion.

H3a, b, c: Having more children, particularly young children, will be associated with
higher levels of work-to-family conflict" and family-to-work conflict” and emotional
exhaustion".

Partners Can Provide Support

A spouse or live-in partner can be a strong resource for workers in dealing with
the multiple demands of their lives, and this is especially true for people with children. A
partner can provide practical assistance by handling household chores, preparing meals,
or taking care of children. A partner can also be a strong source of emotional support, by
listening, offering advice, and providing comfort when problems are encountered.
Support from partners and families has been associated with lower levels of emotional
burnout (Leiter and Durup, 1996; Ray and Miller, 1994; Lee and Ashforth, 1996) and
both work-to-family and family-to-work conﬂict (Leiter and Durup, 1996; F rone,
Yardley and Markel, 1997).

Results from studies have not always been consistent, but marital status and
partner support have generally been associated with lower family-to-work conﬂict

(Frone, Yardley and Markel, 1997). However, helpful as a spouse may be, just having

41

someone who expects to spend time with you can be a source of demands that increases
work-to-family conﬂict. While some studies have reported a reduction in such conﬂict
due to spouse support (e. g. Parasuraman et al., 1992, but for women only, Grandey and
Cropanzano, 1999, but due to a small sample the effect was not statistically signiﬁcant)
many seem to ﬁnd no relationship between work/family conﬂict and marital status (e. g.
Thompson et al., 1999; Judge et al., 1994; Ray and Miller, 1994) or that having a spouse
is associated with more work-to-family conﬂict (e.g. Burke, 1988; Staines and O’Conner,
1980; Ray and Miller, 1994). Although a partner can have both positive and negative
inﬂuences, here it is hypothesized that being in a relationship will be associated with less
conﬂict in both directions.

Research generally indicates that family support and resources are associated with
less emotional exhaustion (Lee and Ashforth, 1996), but results are not completely
consistent. A study of mental health workers found that people who indicated that their
families used external resources such as extended family or neighbors when problems
crop up were less likely to suffer increasing levels of emotional exhaustion over a six-
month period (Leiter, 1990). While Greenglass and Burke (1989) found that marital
satisfaction was a signiﬁcant predictor of burnout for female educators, the same did not
hold true for their male colleagues. Leiter and Durup (1996) found that higher levels of
family support were associated with less emotional exhaustion among nurses, but in
another study of health care professionals Ray and Miller (1994) found higher levels of
both family and co-worker support associated with more emotional exhaustion. This was
the only study to ﬁnd that support from a spouse had a positive association with burnout.

While not all studies included in Cordes and Dougherty’s (1993) review found that being

42

fm
res
red
cor
Wh:
H4r
can
Far.
“in;
con:

Strox

married had a statistically signiﬁcant association with less emotional exhaustion, none
found that single people tended to be less emotionally exhausted. Overall, then, most
research does seem to indicate that having a spouse, particularly a supportive one, may
reduce feelings of emotional exhaustion (Lee and Ashforth, 1996). While most studies
consider actual marital status, a live-in life partner should be able to play a similar role,
whether or not a wedding has taken place.

H4a, b, c: Living with a partner will be associated with lower levels of work-to-family”
conﬂict, family-to-work conﬂicd and emotional exhaustion“.

Family Child Care Reduces Burdens

While simply having a spouse has not consistently been associated with less work/family
conﬂict, having one who shoulders the main burden of child care demands can be a
strong resource, as shown by research on dual- and single-eamer families. Other family
members — grandparents, aunts and uncles, and older siblings — can also provide workers
with a greater sense of security and more reliable assistance when they take responsibility
for a some or all child care. For example, Kossek (1990) found that workers who relied
completely or partially on family members to care for their youngster reported more
positive attitudes about managing work and child care responsibilities than those who had
to depend on nonfamilial assistance. Kossek and Nichol (1992) also found that family
help with child care was associated with more positive attitudes towards managing work
and child care responsibilities, fewer problems with care, and fewer child-care related
absences as estimated by supervisors. Thomas and Ganster (1995), too, found that having
a spouse who handled child care was associated with less work-to-family conﬂict and

better health-related outcomes. However, comparing means, Goff, Mount and Jamison

43

(1‘

ch

for

con
aSSC
bCei
men
H5a
(1.8M)
emor

Hou,

nOnvs

(1990) did not ﬁnd signiﬁcant differences between those with spouses who cared for
children and those who didn’t in terms of work/family conﬂict or absences, possibly
because those whose spouses cared for their children generally had more young children.
Despite this negative ﬁnding and the fact that childless respondents would have no need
for such assistance, it has strong implications for the experiences of working parents.

No studies of bumout were located in which the degree of family child care was
considered. However, this is a form of instrumental support that should be highly
associated with the family and spouse support measures which, as noted above, have
been correlated with emotional exhaustion in previous research. A spouse or other family
member who handles child care represents a valuable resource for working parents.

H5a, b, c: Having a most or all needed child care provided by a family member will be
associated with lower levels of work-to-family conﬂict“, family-to-work conflict" , and
emotional exhaustion”.

Household Income — A Key Resource

One of the most important personal resources a worker has in dealing with
nonwork demands is household income. Whether the money comes from the employee’s
salary, a partner’s earnings, investments, inheritance, or other sources, a higher income
makes it possible to purchase assistance with child care and household tasks (Thompson,
Beauvais, and Lyness, 1999). Financial resources can be used to make mealtimes less
demanding and more pleasurable by paying for take-out food, restaurant service, or at the
high end, a personal chef. While low-income families may sometimes struggle just to
afford basic goods and services, middle income families may take the cost of child care in

stride, and higher-income families can easily pay for assistance with yard and housework.

Alu
chﬂc
onr
to m
{ami
paw
cﬂder

ofen

 

 

A higher income also may make it possible to pay for more or hi gher—quality day care for
children, or for care options that may provide working parents with more ﬂexibility and a
stronger sense of security, like a live-in nanny. Income may also affect a worker’s ability
to take advantage of the beneﬁts an organization offers. One study found that use of
family-related beneﬁts that required workers to cover a portion of the costs (such as
paying for on-site daycare or covering the cost of company-sponsored summer camps for
older children) were more likely to be used by higher-income workers, except in the case
of emergency care for sick family members (Lambert, 1995).

Most research in the area of work and family has involved fairly homogenous
groups of workers with the same occupation, often from middle- to upper-class
backgrounds (Kossek and Ozeki, 1998). With a few notable exceptions (e. g. Lambert
1995), where family income has been considered, it is usually treated as a control
variable and its effects are often not reported (e.g. Frone, Yardley and Markel, 1997;
Greenglass and Burke, 1989). In general, researchers have not deeply examined the
effects of ﬁnances or social class and have not provided a clear picture of how the
work/family interface may differ for those with higher or lower incomes (Voydanoff,
2002). Similar criticisms can be applied to the burnout literature, which until very
recently focused only on workers in human service ﬁelds (Demerouti et al., 2001).
Research on burnout is comprised mainly of studies that involve only teachers, just
nurses, managers, or other homogenous groups likely to be similar in salary, meaning that
where it is considered at all the variance in family income is reduced.

Because of the focus on work/family conﬂict and burnout, household income,

rather than salary, is the appropriate measure for this study. While salary levels may

45

contribute to job satisfaction or turnover intentions, the level of total household income
provides a better picture of the ﬁnancial resources workers can draw on to help them
manage conﬂicting responsibilities.

Previous research does indicate that having a high-salary job is associated with
more work/family conﬂict (e. g. Allen, 2001; Parasauraman et al., 1992; Judge et al.,
1994), although this may be more true for women than for men (Parasauraman et al.,
1992). Since having more money would not logically make it more difﬁcult to balance
work and family, the effect likely captures the increased demands associated with jobs
that involve higher pay, especially for women who already have high demands at home.
Judge et al. (1994), noted, for example, that the mean levels of work-to-family and
family-to-work conﬂict reported by the male executives in their study were higher than
those reported by Gutek et a1. (1991) for a study involving groups of psychologists and
managers as well as the mean levels reported by Frone et al. (1992) for a heterogeneous
cross-section of workers. Pointing out that the male executives they studied also worked
an average of 56 hours a week, considerably more than participants in other studies,
Judge et al. suggested that the major difference may be because their jobs were more
demanding. With occupation and industry also included in the analysis, the effects of
job differences should be controlled and the positive effects of having a higher income
should be more clear.

In the one study located that relied on a broad national sample of workers in
which household income was considered a major explanatory variable, hours worked and
a six-category industry variable were the only job characteristics included in the analysis,

which didn’t ﬁnd a signiﬁcant relationship between being in the lowest income quartile

C(
DC
var

H784

cont
fami.
singlc
r'ncom

assis

and greater work/family conﬂict in either direction (Grzywacz et al., 2002). Another
possible reason is that some earners in the second quartile may also have had relatively
low incomes, but this could not be determined because the study provided no information
on quartile cutoffs or how they compared with Census Bureau estimates. In a study of
working parents of LA. preschoolers, Greenberger et al. (1989) found that among
married and single mothers a higher household income was actually related to more role
strain, which they conceptualized as combining work-to-family and family-to-work
conﬂict, as well as a sense of overload and conﬂict within each role. For men, there was
no relationship. Again, the results likely reﬂect the inﬂuence of job demands on both
variables, especially since strain within the work role was included in the role strain
measure.

Another issue possibly affecting results is that the effects of income are likely not
continuous. As noted by Voydanoff (1988), a minimum level of income is necessary for
family stability. Below a certain threshold, the cost of daycare assistance for even a
single child represents an almost overwhelming proportion of income. At moderate
income levels, such costs can be more easily absorbed, even if more expensive forms of
assistance are less likely to be considered. There should be little difference between the
merely wealthy and the super-rich in their ability to afford nannies, house cleaners, yard
service and other assistance. Logically, the lower, middle, and upper income classes
commonly used in social and economic research appear the most appropriate way the
examine the effects of income level, although it is difﬁcult to know where to draw the

lines dividing the groups. For the purpose of this study, US Census bureau statistics on

47

household incomes were used as a starting place. More information on income
classiﬁcations used is included in the methods section.

Previous research on burnout has generally involved homogenous samples of
human service workers, so income has not been a major focus of study, although
contingent rewards have been shown to serve as a resource associated with reduced
emotional exhaustion (Lee and Ashforth, 1996). Even among the handful of studies
located that combined a focus on burnout and work/family conﬂict, none reported
information on salary or household income. As noted earlier, however, they do offer
considerable support for the notion that minimizing conﬂict between work and family
roles may lead to less burnout. Studies on the relationship between work-to—family
conﬂict and emotional exhaustion have consistently reported strong connections; in a
recent review, Allen and colleagues (2000) reported that the weighted mean correlation in
10 identiﬁed samples was .42. While the estimated effect size was smaller, Leiter and
Durup (1996) also reported a signiﬁcant relationship between family-to-work conﬂict and
emotional exhaustion both at the time of the survey and three months later. Therefore,
personal resources that are useful in reducing such conﬂict should be associated with
lower levels of emotional exhaustion.

Aside from reducing conﬂict, money that can allow workers to hire paid help with
household tasks should enable them, if they choose, to free up energy that might
otherwise be used on chores at home and use it in other domains, like work. It may also
be used for equipment that can make tasks at home or at work easier and less time-
consuming (like electronic personal planners, lap top computers, cell phones,

dishwashers, and power lawn mowers) and for recreatiorml activities that can serve as a

48

way to rest and recharge. Having a higher income should also be associated with less
worry about having enough money to pay for necessary expenses, a potential source of
stress at home that may contribute to emotional exhaustion.

H6a, b, c: When occupation and industry are controlled, a higher household income level
will be associated with lower levels of work-to-family conﬂict", family-to-work conﬂictb,
and emotional exhaustion”.

Age Means More Life Experiences, An Important Resource

Being older means having more experience in a wide variety of situation, which
may help workers both at home and on the job, as well as in balancing the demands of
both realms. As the common adage implies, with age comes wisdom, a powerful
personal resource.

Previous research has shown that, even with gender and the presence of young
children controlled for, being older is associated with less conﬂict, both from work to
family and from family to work (Grzywacz, Almeida, and McDonald, 2002; Grandey and
Cropanzano, 1999). Being older is also associated with reduced burnout (Cordes and
Dougherty, 1993; Maslach, 1982), likely because experience helps people to develop
better coping skills (Lee and Ashforth, 1993), and older workers are likely to have more
different types of life experiences and coping skills to draw on. Age has generally been
included in research on burnout and work and family as well, often as a control variable.
However, the COR model highlights the importance of age as a valuable potential
resource rather tlmn a miscellaneous source of variance in the outcome measures. Thus,
while the inclusion of age as a predictor of work/family conﬂict and burnout is not new,

here we draw attention to its role as a personal resource.

49

H 7a, b, c: Being older will be associated with lower levels of work-to-family conflict”,
family-to—work conflict”, and emotional exhaustion”.
Work Demands and Resources

No study on emotional exhaustion or work/family conﬂict could provide an
accurate estimate of the effects of other inﬂuences without considering work-related
demands and resources. Job characteristics have been signiﬁcant predictors of these
stress-related outcomes in previous research, as COR theory, with its emphasis on
demands and resources, would lead one to anticipate.

In their very conceptualizations, work-to—family conﬂict and emotional
exhaustion both reﬂect the view that work can drain people of valuable resources,
depleting their stores of time and energy and negatively affecting their ability to perform
family roles (in the case of work-to-family conﬂict) and the level of emotional energy
they bring to work (in the case of burnout). Family-to-work conﬂict may also be more of
an issue when the expectations associated with a job are very high, leaving fewer
resources for dealing with non-work roles. Some jobs clearly involve more demands than
others. Two types of work demands previous research shows may be important have been
included here: weekly work hours and having an irregular work schedule. Note that these
are likely to be characteristics of the job, rather than the employing organization as a
whole — managers are more likely to work long hours and shift work is more common at
lower levels.

Among the resources workers can use to deal with these demands, autonomy and
experience in the organization appear to be particularly useful. Jobs which are designed

to provide workers with greater autonomy give them a sense of control and the ability to

50

make sure their work is done in a way that also takes into consideration nonwork
demands. Longer tenure in an organization in general may mean stronger personal
networks and a better understanding of the most effective ways to do one’s job, as well as
ways to acquire organizational resources and manage job-related demands.

In this section, research on how these work demands and resources may affect
work/family conﬂict and emotional exhaustion is reviewed and related hypotheses are
presented.

Long Work Hours and An Irregular Schedule: Sources of Stress

Sheer volume of work represents perhaps the most important of job demands
facing today’s employees. Another important demand is a requirement to be at work
when most people are at home or involved in personal activities -— in other words, the
requirement to work an irregular shift. Research bears out the impact COR theory
predicts work demands should have on the three stress-related outcome variables in this
study, particularly long hours and an irregular schedule.

A heavy workload has been linked to burnout in many studies (e.g. Leiter and
Durup, 1996; Bacharach, Bamberger and Conley, 1991). In fact, in their meta-analytic
review, Lee and Ashforth (1996) found workload to be the strongest predictor of
emotional exhaustion. A heavy workload, working long hours, staying late in the
evenings, or having more work than one can comfortably handle has been consistently
associated with more bidirectional conﬂict and work-to-family conﬂict (e.g. Staines,
Pottick and Fudge, 1986; Cooke and Rousseau, 1984; Bacharach, Bamberger and Conley,
1991; Leiter and Durup, 1996; Frone, Yardley and Markel, 1997 ; Galinsky, Bond and

Friedman, 1996: Judge et al., 1994). Weaker relationships have been found for family-

51

to-work conﬂict (Leiter and Durup, 1996; F rone, Yardley and Markel, 1997). This is
consistent with the large body of research that indicates that work-related demands and
resources are more closely linked to work-to-family conﬂict, while family stressors and
supports tend to have a stronger impact on family-to-work conﬂict (e.g. Frone, Yardley
and Markel, 1997; Ayree, Fields, and Luk, 1999).

Many studies have measured workload with items that asked respondents to
assess the demands of their jobs. Leiter and Durup (1996) for example, used a four-item
measure that included an item asking, “How great is the amount of emotional strain your
job puts on you?” However, because of the subjective nature of such questions and the
similarity of their content to commonly used measures of work/family conﬂict and
burnout, personality inﬂuences, current mood and common method variance may lead to
overestimates of the relationships between these variables. F rone Yardley and Markel
(1997), who used both hours worked and a subjective measure of overload that asked for
agreement on three items such as, “I have too much work to do everything well,” found
stronger relationships to both work-to-family and family-to-work conﬂict for the
subjective measure. To reduce concerns about common method variance, this study will
rely on two very straightforward measures of work demands: total hours worked per
week, and whether the respondent’s job involves a regular daytime schedule or some
other schedule. Working longer hours reduces the time available for family and other
outside activities, while working irregular hours can also lead to work/family conﬂict
(Pleck, Staines, and Lang, 1980) as a nonregular schedule makes it hard to have ﬂee time
when family members, especially children, are at home. Longer hours have been

associated with greater work-to-family conﬂict in several studies (e.g. Thompson et al.,

52

1999; Pleck, Staines and Lang, 1980) and even in a study where the degree of work-to-
family conﬂict was rated by employees’ wives, longer hours were associated with more
conﬂict (Burke, Weir, and DuWors, 1980). Both long hours and irregular hours can also
make it more difﬁcult to ﬁnd child or elder care assistance. In a study investigating what
made workers in dual-career families feel they had been successful at work, family and
balancing the two, irregular schedules were one of the most signiﬁcant negative
inﬂuences (Moen and Yu, 1999). They have also been associated with greater emotional
exhaustion (Demerouti et al., 2001). While the effects are expected to be greater for
work-to-family conﬂict and emotional exhaustion than family-to-work conﬂict, all three
are likely to be increased by longer hours and irregular schedules.
H8a, b, c: Greater work demands (longer hours and an irregular schedule) will be
associated with higher levels of work-to-family conﬂictafamin-to-work conflicf' and
emotional exhaustion”.
Autonomy Allows Workers to Take Control of Problems

Autonomy on the job has been identiﬁed as an important resource that empowers
workers to develop solutions for resolving conﬂicting demands. While the negative
effects of job demands appear to be stronger than the positive effects of work resources,
in a number of studies having greater autonomy has been associated with less
work/family conﬂict (e.g. Greenhaus er al., 1989; Burke, 1993; Galinsky, Bond and
Friedman, 1996) and burnout (Burke, 1993; Demerouti et al., 2001; Lee and Ashforth,
1993; Iverson, Olekalns, and Erwin, 1998). Lambert (1990) offers some reasons why,
suggesting that autonomy helps workers to balance work and family responsibilities by

making it possible for them to schedule work tasks around family activities. She also

53

suggests that autonomy makes it easier to take time off when needed and possibly even
bring children to work if necessary. As she suggests, autonomy may” contribute to greater
ﬂexibility. Autonomy also increases workers’ ability to directly confront sources of
stress and resolve problems and may increase a worker’s sense of control (Lee and
Ashforth, 1993 ), both over work itself and the overlaps between work and family. A
sense of control over work and family issues was strongly associated with lower work-to-
family conﬂict in a study of health care professionals (Thomas and Ganster, 1995). A
level of autonomy consistent with or greater than the level desired was associated with
less anxiety, depression, irritation and fewer stress-related health symptoms in a sample
of university employees (Edwards and Rothbard, 1999).

In their meta-analytical review of burnout predictors and outcomes, Lee and
Ashforth (1996) found only a weak relationship between autonomy and emotional
exhaustion, a mean weighted correlation of -. 13, with zero in the conﬁdence interval for
the relationship after correction for attenuation due to error of measurement and sampling
error. However, a meta-analysis simply involves averaging the results found in previous
studies, and since most studies involved homogeous samples of service workers in
similar jobs, it is quite possible that a low level of variance in autonomy within studies
has affected the results. In a study of workers from a mix of very different professions —
teachers, nurses, assembly line workers, and air traffic controllers — Demerouti et al.
(2001) found that greater job control (measured with items similar to autonomy scales)
was strongly related to emotional exhaustion (r=-.26). An alternate explanation for the
weak results in the meta-analysis might be that excessive autonomy — having much less

direction than one desires — results in negative outcomes (Edwards and Rothbard, 1999)

54

such as a feeling of being lost and lacking control, increasing rather than decreasing
burnout. However, the literature so far does not allow such conclusions to be examined,
at least in part because of the tendency to focus on homogenous samples.

Overall, it appears likely that, especially in broader samples, autonomy is
associated with less work/family conﬂict and emotional exhaustion. Considerable
support for this comes from a study of hospital workers at all levels, which found that
through its inﬂuence on in-role stress autonomy had a signiﬁcant effect on emotional
exhaustion, even after controlling for the personality inﬂuences of positive and negative
affectivity (Iverson, Olkalns, and Erwin, 1998).

Tenure Means Valuable Skills

Another potential resource for workers to draw on is experience in the same
company. Knowing the ropes and better ways to do things can reduce work-related
demands in the form of stresses and overtime. Experienced workers may also have better
access to valuable resources (Grandey and Cropanzano, 1999) and may have developed
better coping methods (Lee and Ashforth, 1993).

Research on the relationship between tenure and work/family conﬂict does not
report consistent results; however, positive effects have been reported for men, who tend
to have longer tenure, particularly in older studies. For example, longer tenure was
associated with lower levels of work/family conﬂict in men studied by Judge et al. (1994)
and Parasuraman et al. (1992). No relationship was found in several studies that included
larger numbers of women (e.g. Thompson et al., 1999; Allen, 2001), but neither included

controls for occupation, which could have made a difference.

55

Although several small studies of teachers found no signiﬁcant relationship
between experience and burnout (Cordes and Dougherty, 1993), in other ﬁelds job
experience and tenure have been found to affect the way that employees differ in their
experience of burnout (Lee and Ashforth, 1983; Maslach, 1982). Authors of these two
studies, both of which involved human service workers, expressed the opinion that more
experienced workers may be better at dealing with difﬁculties and have better skills for
coping with clients. In addition, they note that people who have ﬁnd a particular position
extremely stressful are likely to quit and so would not have long tenure.

H9a, b, c: Having more job—related resources (greater autonomy and more experience
working at one ’s current organization) will be associated with lower levels of work-to-
family corylict“ family-to-work conflict” and emotional exhaustion”.

A Family-Supportive Work Environment

Many organizations today are attempting to supplement the personal and work-
related resources employees have available to manage the multiple demands associated
with their jobs and families. As discussed earlier, they are doing this by increasing
ﬂexibility in when and where work can be done, offering “family-ﬁiendly” beneﬁts, such
as those that help employees ﬁnd and pay for dependent care assistance, and by striving
to create a climate where supervisors and cultural norms are supportive of employee
efforts to balance work and family. Formal beneﬁts and policies combine with informal
expectations and norms in creating an organizational environment that can be more or
less supportive; drawing on COR theory, a more supportive environment can be seen as

important supplementary resource for workers dealing with demands in two domains.

56

Based on previous research, four ways in which organizations can help promote a
more family-supportive environment have been selected as indicators of how supportive
employing organizations are: increasing ﬂexibility in work time and location, providing
dependent care beneﬁts, encouraging supervisor support with work and family issues,
and promoting family-supportive cultural norms. The literature on these types of
supports is described next.

Flexibility in Work Time Reduces Conﬂicts and Stress

Flexibility in when work is done is often viewed as a key resource for workers
with families. Flexibility is seen as easing the difﬁculties that arise when workers face
demands to complete their work, but at the same time need to support their family
members, perhaps by driving an elderly parent to a medical exam or attending a child’s
school activity.

Despite widespread organizational commitment to ﬂexibility, little research has
been published in academic journals regarding its effectiveness (Kossek and Ozeki, 1998;
Baltes et al., 1999). The wide range of conditions and samples studied is one reason for a
lack in consistency in results. In their meta-analytic review of the literature of ﬂextime,
Baltes and colleagues (1999) concluded that, overall, studies to date indicate that ﬂextime
is weakly associated with higher job satisfaction, and somewhat more strongly with lower
absenteeism. Reported effect sizes appeared to be weaker for studies that focused on
managers and professionals, those with less rigorous methods, and in cases where
ﬂextime had been in place for longer. They did not code for other sample characteristics,
but the authors of two studies (Bohen and Viveros-Long, 1981; Shinn et al., 1989)

suggest that ﬂexibility may have the most beneﬁt for single workers and others with low

57

levels of family responsibility, contrary to the assumptions of work/life program
planners. They suggest that ﬂexibility may not help enough to reduce the heavy load
working parents of young children carry. However, they caution that there could be other
factors involved. The ﬁndings of their research are compatible with those of a large-scale
study of working adults which found that ﬂexible time and leave policies did not explain
a signiﬁcant proportion of variance in a bidirectional measure of work/family conﬂict
after controlling for demographic, job, ﬁnancial, and family characteristics, including
recent absences from work due to child care demands, and child care breakdowns
(Galinsky, Bond, and Friedman, 1996). One reason may be that the last two, in
particular, are so closely associated with family-to-work conﬂict. Also supporting a
negative view of the helpfulness of such policies, Dunharn et al. (1987) found that
workers who switched to a ﬂextime schedule reported a statistically insigniﬁcant rise in
interference with activities involving family and ﬁiends and access to needed goods and
services. And in a study of dual-income families with preschoolers in Singapore,
schedule inﬂexibility was not associated with work-spouse conﬂict among both mothers
and fathers; the correlations reported for work-parent conﬂict were larger (.12 for men
and .16 for women) but were not signiﬁcant due to the relatively small size of each
sample group.

However, other studies support a more positive view. Greenberger et al. (1989)
found that supervisor ﬂexibility in allowing workers to take time out for family matters
was associated with lower role strain (a measure mixing work/family conﬂict with in-role
conﬂict and overload) for married mothers, although not for single mothers or married

fathers. Greenhaus et al. (1989) found that, even with autonomy controlled for, ﬂexibility

58

was associated with reduced work-to-family conﬂict. Winnette, Neale, and Williams
(1982) reported that parents of young children working for federal agencies used ﬂextime
to spend more quality time in the evenings with their families and reported less difﬁculty
engaging in familial, recreational, educational, and chore-related activities in a diary-
based study. In a study of workers in several organizations and a women’s business
association, Allen (2001) found that greater use of beneﬁts providing ﬂexibility
(ﬂextime, compressed work week, telecommuting, and part-time work) was associated
with less work-to-family conﬂict, although greater availability was not. Thomas and
Ganster (1995) found that more ﬂexible schedules were moderately related to a sense of
increased control in work/family balancing, but while ﬂexibility was associated with
lower work-to-family conﬂict the relationship was not signiﬁcant, possibly because their
study involved nurses, who have considerable ﬂexibility in switching shifts. Overall
research to the present, then, provides some support for optimistic views that ﬂexibility in
when work is done leads to lower work-to-family conﬂict, but ﬁndings are not
conclusive.

No studies were located that directly measured the relationship of time-based
ﬂexibility to family-to-work conﬂict. However, it seems likely that by allowing workers
to set their own hours, within limits, organizations may enable employees to take care of
family business when they need to and make up for lost time on the job later, reducing
family intrusions on work time as well as a sense of strain that may carry over to affect
concentration on the job. This is not a minor issue; Williams and Alliger (1994) found
that family concerns were much more likely to interrupt concentration and moods at work

than visa versa, likely because workers often need to really focus on the job. Even single

59

workers living alone can beneﬁt from a more ﬂexible approach to work hours, which can
make it easier for them to participate in social activities, schedule visits with health care
providers, and attend events associated with family and other important people or
interests in their lives (Grover and Crooker, 1995) without creating conﬂicts between
their outside lives and work. Supporting that notion are ﬁndings on absences. Dalton
and Mesch (1990) found that introducing ﬂextime reduced absenteeism among utility
company workers, and when regular hours were reintroduced, days of work missed again
climbed. F lextime was associated with fewer absences in several other studies (e. g.
Krausz and F riebach, 1983; Erickson, Nichols and Ritter, 2000), although one study
comparing staggered, ﬁxed starting and quitting times with ﬂexible hours that could be
changed daily found those under the more strict schedule were less likely to miss work
(McGuire and Liro (1987).

The single study found that looked at an exhaustion-based burnout measure in
conjunction with schedule ﬂexibility, however, did ﬁnd a signiﬁcant association for
mothers of preschoolers in dual-career households, but for fathers there was no
relationship, perhaps reﬂecting the strong differences in family role expectations in the
country where the study was conducted: Singapore (Ayree, 1993). Women who reported
a high degree of difﬁculty in changing work hours and days experienced more burnout.

In addition to helping workers balance their work and nonwork lives, ﬂexibility in
work time can reduce another source of stress, commuting. F lextime can allow workers
to adjust their travel time to avoid peak rush periods, reducing the amount of commute
time and related stress, both of which can drain workers of energy. Also, if one doesn’t

feel compelled to arrive at the ofﬁce by exactly 8:00 am, being caught in an accident- or

construction-related trafﬁc jam is bound to be less upsetting (Pierce and Newstrom, 1982,
1983 ). By helping with work/nonwork conﬂicts, reducing concerns about commuting,
and providing workers with a sense of control that may make it easier to tackle problems
that are best resolved during work hours, ﬂexibility is likely to reduce emotional
exhaustion.

Flexibility in Work Location May Be Helpful

Flexibility in where work is done can represent an important resource when
dealing with simultaneous demands related to job and family. For example, being able to
work from home may allow an employee to ﬁnish an important project while also caring
for a sick child. Despite rising interest, academic journals contain few reports of research
on the effects of providing greater ﬂexibility in where work is done.

Where working at home once meant sub-assembly and sewing piecework, in
today’s advanced US economy it is most likely to involve using computers and
telecommunications tools. Most of the information that is available on teleworking or
telecommuting, as it is often called, comes from studies that are descriptive, rely on self-
report measures, and base their results a small number of employees (Hill, Miller,
Weiner, and Culihan, 1998; Feldrnan and Gainey, 1997).

Studies that have looked at whether having the ability to work from home
increases or decreases work-to-family conﬂict found that it did, and it didn’t. Duxbury,
Higgins, and Neufeld (2000) found that a small group of voluntary telecommuting
pioneers reported less work-to-family conﬂict six months aﬁer starting to work one to
three days a week from home, while their colleagues who remained under the same

schedule didn’t experience signiﬁcant change. Some 38 percent noted that

61

telecommuting enabled them to spend more time with their families. In contrast, Hill et
al. (1998) compared IBM workers who were required to operate under a “virtual ofﬁce
plan” that reduced ofﬁce space while giving them electronic tools to work from anywhere
with colleagues who remained in traditional work locations, ﬁnding no difference in
work/family conﬂict between the two goups. The researchers also collected and coded
comments; frequent themes included complaints that the boundary between work and
family life had become blurred, making virtual employees feel as if they were always
working, contrasted with positive comments about how the new program helped
participants fulﬁll household responsibilities and strengthened family relationships.
Mirchirandi (1998) extensively interviewed a small group of female telecommuters in a
variety of jobs and organizations, also ﬁnding that doing work at home sometimes led
workers to feel that their private sanctuaries had been invaded and caused conﬂicts with
their family activities. Most reported actively working to create boundaries between the
work they did for pay and work they did for their families. At the same time, however,
they also found that housework was less of a struggle, since they could handle laundry
and other tasks during breaks, and appreciated being able to participate in signiﬁcant
moments in the lives of their children. Similar beneﬁts were reported by European
women participating in three small studies reported by Bussing (1998). In a large sample
of Canadians working for private employers, Higgins, Duxbury and Lee (1992) found
that employees who reported greater ﬂexibility in where and when work was done were
better able to constructively balance their responsibilities at home and at work.

The few telework studies that have been published in academic journals have

tended to focus on what draws people to telecommuting and on its relationship to

62

productivity, performance, and satisfaction with telecommuting. Gender does appear to
have an impact. As noted by Hundley (2001) in his study comparing self-employed to
organizationally employed workers, while men work for themselves to maximize
earnings, women are attracted options that allow them to work at home and provide
greater ﬂexibility to facilitate caring for their homes and families. Workers with higher
family responsibilities tend to gravitate to jobs that provide more autonomy and
ﬂexibility (F eldman and Gainey, 1997). Results indicate that women are more likely to
choose to telecommute (Belanger, 1999), tend to be happier with their telecommuting
experience (Ragurarn et al., 2001; Belanger, 1999) and report higher productivity under
telecommuting conditions than men (Hill et al., 1998). However, telecommuters as a
group are not necessarily more satisﬁed with their jobs than workers in traditional
environments (Belanger, 1999). The little work that has been done in this ﬁeld seems to
indicate that having the option to work at home may lead to positive outcomes for both
individuals and organizations, but this conclusion is by no means certain.

Research on family-to-work conﬂict is cautiously optimistic. Some 13% of
volunteer telecommuters in a study of Canadian managers who worked at home one to
three days a week (Duxbury, Higgins and Neufeld, 2001) commented that family
interference with work was one of the major disadvantages of such an arrangement.
However, a comparison of mean family-to-work conﬂict for the full sample just prior to
and six moths after the introduction of telecommuting showed somewhat lower levels of
conﬂict as measured by a ﬁve-item scale, while co-workers and co-workers experienced
little change. The female telecommuters in Mirchirandi’s (1998) study reported that the

ability to work from home made it possible for them to continue working when problems

63

cropped up at home, such as a sick child, and it is easy to see how the option to
telecommute could be helpful in reducing family-to-work conﬂict in such ways. On the
other hand, however, telecommuters in a number of studies also reported that when they
worked at home it was sometimes difficult to convince family members that they were
busy and should not be interrupted (e. g. Mirchiradi, 1998; Hill et al., 1998), which would”
be associated with increased family-to-work conﬂict.

Working from home also completely eliminates the stresses and time involved in
commuting, which can be particularly difﬁcult for workers with eldercare responsibilities
or young families, unless they have a spouse or other family member close by who can
respond to emergencies and participate in school activities. Many of the volunteer
telecommuters in Duxbury, Higgins, and Neufeld’s (2001) study were workers who lived
far away from their places of work, and they particularly appreciated the beneﬁts of
greater ﬂexibility in work location. More people selected this as a major advantage of
working from home than any other option, with nearly half agreeing. Although working
from home may make it more difﬁcult to access social support from co-workers and
supervisors, most telecommuters do not work from home everyday (Duxbury, Higgins
and Neufeld, 2000; Mirchirandi, 1998), reducing concerns in that regard. Only 35
percent of the Canadian managers studied selected poor communication with co-workers
as a negative result.

While burnout was not a variable in any of the studies located on telecommuting,
it is expected that, because it may ameliorate conﬂicts between work and nonwork areas
and signiﬁcantly reduce stresses involved with commuting, greater ﬂexibility in work

location should be associated with less emotional exhaustion.

Dependent Care Assistance Contributes to a Sense of Support

Dependent care beneﬁts represent one important type of resource organizations
can offer their employees as way of helping them to manage conﬂicting demands on the
job and home front.

As noted in the introduction, few employers offer direct assistance with dependent
care by operating their own on-site care center or offering ﬁnancial assistance with day
care costs. However, many large companies offer access to information and referral
services that help employees locate child and elder care providers, and about half give
workers the option to set up dependent care spending accounts that allow them to pay for
care with pre-tax dollars. Maternity leave, required by law, is quite common, but
paternity leave is less so (Galinsky and Bond, 1998; Bond, Galinsky and Swanberg,
1998). Another major way employers can be of assistance is to ﬁnd ways that workers
can take time off to care for sick children without giving up pay or vacation time, such as
by trading shifts or using some of their own sick days to care for dependents.

The availability and use of such dependent care support efforts are widely
believed to help employees balance their work and family lives, and some research bears
out that assertion. For example, Kossek and Nichols (1992) found that users of two
moderately subsidized hospital child care centers reported fewer problems with care
arrangements, had fewer care-related problems, and held more positive attitudes towards
managing their child care responsibilities than employees of the same organization using
other care options while on the waiting list. However, the one study on the effects of a
corporate day care center to include a widely used measure of work/family conﬂict in its

design found that, while use of a new facility that charged users the going market rate

65

was not associated with reduced work-to-family conﬂict, satisfaction with child care was
(Goff, Mount and Jamison, 1990.) These results may have been affected by the nature of
the sample. Almost twice as many nonuser as users responded to the relatively small
study. Some 89% of users had spouses with full-time jobs, while only 40% of nonusers
did; the user group was about half female, while men made up 69% of the nonuser group.

The number of formal family-supportive policies used was associated with less
role strain and fewer strain-related health problems among single mothers in a study of
parents of preschoolers (Greenberger et al., 1989). However, among married mothers,
while the number of policies used was related to fewer health symptoms, it was
associated with greater role strain, possibly reﬂecting the fact that those taking advantage
of the programs tended to have more bi-directional conﬂict and greater responsibilities.
In a study comparing employees who used an eldercare referral service with those who
didn’t, Wagner and Hunt (1994) found that users reported more interference with work,
but the authors note that this was likely because of higher demands that made them utilize
the service in the ﬁrst place rather than program ineffectiveness.

Studies that have looked at the availability of employer provided work/family
policies with measures that incorporate dependent care support have found some
evidence that they are helpful. Judge et al. (1994) found that the self-rated extensiveness
of formal and informal work/life support available in their organizations was modestly
associated with lower work-to-family conﬂict among male executives. This was the only
study that speciﬁcally reported on the relationship between such beneﬁts and family-to-
work conﬂict; a small, statistically insigniﬁcant relationship was found, perhaps because

to have achieved this level of success these executives probably had uniformly low levels

of such conﬂict (Kossek and Ozeki, 1998). However, a study analyzing the need and
desire for a day care center at a university found that both male and female employees
expressed an interest in such a service, and that potential users believed it would reduce
tardiness and absence behavior (Mize and Freeman, 1989), both of which represent types
of family-to—work conﬂict. Work/family beneﬁt availability was negatively related to
work-to-family conﬂict in a another study of managers and professionals (Thompson,
Beauvais, and Lyness, (1999). Thomas and Ganster (1995), however, found no
relationship between the availability of supportive policies, including child and elder care
information and referral services and on-site day care, and work-to-family conﬂict or
absence, one indicator of family-to-work conﬂict. They posited that this may have been
because so few of their respondents had access to such services. Similarly, the
availability and use of dependent care beneﬁts were not signiﬁcantly related to lower
work-to-family conﬂict among employees from a technology ﬁrm, utilities ﬁrms, and a
women’s business association surveyed by Allen (2001), although greater ﬂexibility and
total beneﬁts of both kinds used were. Although such beneﬁts are provided with the
intention of helping, then, research has not consistently shown that they reduce conﬂict.
No studies were located that dealt with the relationship between dependent care
beneﬁts and burnout. However, by easing difﬁculties associated with balancing work and
family such beneﬁts may reduce emotional exhaustion. Help in ﬁnding and paying for
care for children and elders may ameliorate some of the emotional drain associated with
burnout. It is also easy to conjecture that having children in a nearby corporate daycare

center rrright afford working mothers an opportunity to spend break time visiting with

67

them, beneﬁcial since, as noted earlier, children appear to serve as a resource that reduces
burnout.

Because this study involves workers with widely varying work and family
circumstances and how organizational efforts to provide a more family-supportive
environment may affect them, policy availability rather than use will be studied. Simply
creating and offering programs may have a beneﬁcial effect on both work/family conﬂict
and emotional exhaustion, even if they are not used, because just knowing that help is
available if ever needed may reduce stress. For example, while working parents may
seldom need a child-care information and referral service or take time off to care for sick
children, knowing that such help is there can help eliminate worries about what to do if
problems crop up. Previous research has shown that perceptions of support availability
(not just actual support received) may act as a buffer between stressful events and
psychological distress because stressful situations appear less threatening when one
believes there is a support system available if needed (Lee and Ashforth, 1993).
Supportive Supervisors Can Promote More Positive Outcomes

Supervisor support has been one of the most researched work resources. COR
theory holds that social support can be helpful for workers seeking to deal with multiple
demands and conﬂicts, and working under a supportive supervisor appears to have a
strong effect on stress-related outcomes, reducing negative consequences. A large
number of studies of work/family conﬂict and burnout have found that supportive
supervisors can make a difference in the attitudes and experiences of employees (e. g.
Iverson, Olekalns, and Erwin, 1998; Leiter and Durup, 1996; Demerouti et al., 2001;

Galinsky, Bond and Friedman, 1996). Support from supervisors is often combined with

68

social support from other sources, including family, friends, and co-workers in the
burnout literature. Lee and Ashforth (1996) report a mean weighted correlation with
emotional exhaustion of -.26 for social support in 6 studies and -.31 for supervisor
support in 13 studies. Supervisor support with work issues had a stronger relationship _
than all other resources except community bonds and met expectations; support from co-
workers and family resources, in contrast, had weighted average correlations of-. 18 and -
.16. A recent study supports the notion that these relationships do not reﬂect simply
personality differences. Demerouti et al. (2001) report a signiﬁcant relationship between
self-reported emotional exhaustion and researcher-rated supervisor support. While a few
studies have failed to ﬁnd relationships (e. g. Greenglass and Burke, 1989, for male
educators; Ray and Miller, 1994, female health care workers) overall, supervisor support
appears to be a resource that may buffer employees from some of the suffering associated
with strong demands.

In studies that focus on the work/family interface, two types of measures have
been used: those that capture simply support with work issues, like the scales most
common in burnout research, and measures that ask about how supportive supervisors are
about family-related concerns. In the literature, there is a strong expectation that both
will be associated with less work/family conﬂict and emotional exhaustion, but stronger
relationships to work/family conﬂict have been found for scales that include or focus on
family-related support. Using general work-related measures of supervisor support,
Frone, Yardley and Markel (1997) found a small but statistically signiﬁcant reduction in

work-to-family conﬂict, but not family-to-work conﬂict. Leiter and Durup (1996),

69

however, found work-related support from supervisors to be signiﬁcantly correlated with
lower levels of both, although the relationships were modest.

Most studies in this area have focused on family-related supervisor support or
combined items measuring that with more general support. Kossek and Nichol (1992)
found a moderate positive relationship between supervisor support with family conﬂicts
and employee attitudes toward managing work and child care responsibilities. Galinsky
et al. (1996) reported that supportive supervisors reduced work-family conﬂict in a large
sample of randomly selected US workers. However, Ray and Miller (1994) reported no
relationship between supportive supervisors and bidirectional work/family conﬂict
Shinn and colleagues (1989) found a modest (r =.11) between supervisors’ support with
family issues and job satisfaction, and an even smaller correlation with family distress.
Using the same scale of supervisor support, Thomas and Ganster (1995) found that
having supervisors who were understanding was associated with less work-to-family
conﬂict among mainly female health care workers with children. Supervisors willing to
listen to and help with family-related problems were also strongly associated with less
bidirectional work/family conﬂict and pretreatment absenteeism among production
facility workers studied by Golf et al. (1990). In a study of managers with graduate
degrees, Thompson et al. (1999) found that managerial support was associated with
greater commitment, but the correlation with work-to-family conﬂict, while fairly large,
was not statistically signiﬁcant, likely because the sample used was relatively small.
Studying a group of employees from a variety of organizations, Allen (2001) found that
family-related supervisor support was associated with greater use of ﬂexibility beneﬁts

and lower work-to-family conﬂict.

70

While there are exceptions in the literature, then, research seems to support the
notion that supervisor support is a resource that may reduce work/family conﬂict and
emotional exhaustion. The current study will focus on the effects of supervisor support
with family issues, rather than overall or work-related support. There are three reasons
for this. First, as is illustrated from the studies discussed above, support with family
issues is likely to have the strongest effect on work/family conﬂict. Second, farnily-
related support can also be viewed as a direct contributor to a family-supportive
organizational environment, where the effects of supervisory support with work issues on
the environment would be more indirect. Third, most of the previous research on
emotional exhaustion has concentrated on the effects of supervisory support with work
issues, ﬁnding evidence of a fairly strong relationship, but has not looked at other types
of supervisor support. By examining the effects of supervisor supportiveness when
family issues crop up, a relatively new area, the study will make a stronger contribution
to the literature.

Family-Supportive Cultural Norms May Reduce Concerns

Supervisor support is not the only informal resource that an organizational
environment can offer to help support workers coping with work and nonwork demands.
Creating an organizational culture where people are open and understanding about work-
family issues can have a direct positive effect on work/family conﬂicts. Such a culture
can, for example, make parents or eldercare givers feel comfortable about calling to
check on family members during work hours, which may actually allow them to
concentrate better. It can also have an indirect effect by helping employees to feel better

about using the organization’s formal work/life policies. Formal policies that allow for

71

ﬂextime or work at home are not likely to be used if they are not supported by the
organizational culture and those directly above an employee who may want to use them
(Galinsky, Bond, and Friedman, 1996; Kossek, Barber and Winters, 1999). Especially
important are perceptions that employees who don’t always put their work before their
families will suffer in terms of career opportunities. Family-related concerns don’t
disappear simply because employers penalize workers for taking care of family
responsibilities, whether than means taking time off for personal reasons, or using work
time to take care of family needs. Thompson, Beauvais and Lyness (1999) identiﬁed
negative career consequences associated with taking advantage of work/family programs
or putting family before work as an important component of an organization’s
work/family culture. In their sample of managers, they found it was associated with
higher levels of work-to-family conﬂict and an interest in quitting. Allen (2001) found
that global perceptions of how family-supportive an employer is were strongly associated
with reduced work-to-family conﬂict, higher job satisfaction, and less interest in quitting
with demographic and family characteristics controlled. Galinsky et al. (1996) also
reported that a supportive culture was associated with lower levels of conﬂict.

Although there are some gaps and inconsistencies, then, previous studies have
found indications that both the supportiveness of the overall work environment and
individual elements of that support may have positive effects. Particular indicators of a
supportive environment -— such as ﬂexibility in work timing — may have different effects
than, for example, supportive cultural norms. The two may work together to increase the
overall sense of support, or an organization may rely on one to replace the effects of the

other. This implies that examining the effects of the overall environment is just as

72

important as looking at the effects of these environment indicators separately. Therefore,
both will be examined in this study.
H100, b, c: Working in a more family-supportive environment (one with greater flexibility
in the timing and location of work, more access to dependent care assistance, supportive
supervisors, and cultural norms that don ’t penalize workers for putting family first) will
be associated with lower levels of work-to-family conflicf" family-to-work conflict” and
emotional exhaustion”.
Potential Moderators: Income Level and Single Parent Status

The next set of hypotheses deals with the question of whether workers with lower
resources beneﬁt more from a family-supportive organizational environment. In previous
research, the COR model and similar approaches have been used to guide research aimed
at identifying stressors that can increase the chance of negative stress-related outcomes,
like work/family conﬂict and burnout, as well as supports that can help workers to deal
with challenges. Considerable support has been found for the direct effects hypothesized
in this study, as discussed in the previous sections. However, while COR theory has
often used to help develop studies that focus on direct effects, it implies that
combinations of demand and resource levels may also make a strong contribution
towards stress-related outcomes. The best outcomes would be ”expected for people with
few demands and many resources, the worst for those with many demands and few
resources.

It is possible to consider overall demands and resources, but there are two reasons
why it may not make sense to do so. First, while researchers can collect information on a

range of resources and demands, they may miss some important to the individual’s

73

situation. Even a carefully organized study will be limited in its ability to measure
overall demands and resources. Second, such information is not likely to be available all
to policymakers or employers who may want to use information from research to help
their efforts to reduce employee stress. Therefore, to provide the most practical value it
makes most sense to focus on speciﬁc demands and resources that are easy to identify
and likely to have strong effects on the outcomes of interest.

The main focus of this study is to examine the effects of organizational efforts to
be supportive of workers combining responsibilities at work and at home. Therefore, the
study will examine the combination of working in a family-supportive work environment
with resource/demand levels. As noted earlier, many combinations could be tested, but
two personal demand and resource indicators have been identiﬁed as particularly likely to
affect workers’ need for a supportive work environment: single parent status and income
level. Both groups are easily identiﬁable. While they cannot make discriminatory
decisions based on such information, most employers are aware of whether their
employees have children, whether they are married, and they know how much they earn
at their current job. Results of this research, then, can be applied in practical situations if
effects are found.

As noted earlier, workers in high-income families can afford high-quality help on
the home front. If they happen to be employed by an organization that has a less family-
supportive environment, they can often pay for alternative support in the form of more
ﬂexible or better child care and assistance with household tasks. Dependent care
beneﬁts, in particular, may be fairly easy for upper-income workers to replace, since they

don’t have a strong need for ﬁnancial assistance with child care and are likely to be

74

interested in just a few top-notch dependent care assistance options rather than
information on a wide range. Workers from lower-income families, on the other hand,
are less able to pay for help and may ﬁnd that organizational support — whether formal,
like ﬁnancial assistance with child care costs and access to services providing
information on child care and elder care options, or informal, in the form of supportive
cultures and supervisors — represents the main resource they have in managing the
boundaries between work and family.

While potentially valuable, there is little research on how social class or income
levels may affect the relationships between work and family characteristics and between
the work-family interface and other outcomes (V oydanoff, 2002) and the potential effects
of income or class as a moderator have been largely ignored by burnout researchers as
well. One study that examined the moderating effects of class did so by comparing the
relationships between stressors and involvement at home and at work with work-to-
family and family-to-work conﬂict among blue- and white-collar workers. Although
their model ﬁt equally well for the two groups, some of the relationships were different.
While job involvement was strongly associated with work-to-family conﬂict among
white-collar workers, it was not related for blue-collar workers. Work-to-family conﬂict
lead to family distress among blue-collar workers, but not white-collar workers (F rone et
al., 1992). The ﬁrst ﬁnding may well be due to differences in the nature of work, as the
researchers surmised, but the greater relationship between conﬂict and distress at home
found among blue collar workers may be because they have fewer resources to buffer

negative effects.

75

Some support for this view can also be drawn ﬁom a study of parents of
preschoolers, reported in two different papers. For single mothers whose household
income was considerably lower than that of married parents, the number of beneﬁts used
was associated with lower role strain and fewer stress-related health symptoms. Among
married women, beneﬁt use was associated with greater role strain, and among married
men, it was associated with more stress-related health problems (Greenberger et al.,
1989). In a companion study, single mothers were more interested in changing jobs to
get better beneﬁts, and the beneﬁt most desired was ﬁnancial assistance with day care
costs (Goldberg et al., 1988).

Because they have so many alternative resources to draw on, employees with
higher household incomes are less likely to experience strong positive effects from
organizational efforts to create a supportive work environment. Executives in Judge et
al. ’3 (1994) study, in fact, had nearly equal levels of work/family conﬂict in both
directions whether they had a stay-at-home spouse or working partner, probably at least
pmtly because it was easy for them to pay for alternative assistance to meet demands at
home. An important caveat, however, is that organizational assistance not come with
such a heavy price tag that only upper-income employees can afford to use it. This was
illustrated by Lambert’s (1995) study, which found that non-emergency family—oriented
beneﬁts that cost workers money were more likely to be used by higher income
employees. Except for on-site daycare, which is often only partially subsidized by
employers, all of the beneﬁts selected for inclusion in this study do not involve

substantial costs for workers.

76

It is predicted that the positive effects of having a family-supportive
organizational environment will be strongest for employees from low-income
households, less pronounced for employees from middle-income households, and
weakest for employees from high-income households.

H I I a, b, c: The positive eﬂects of a family-supportive environment (more dependent care
beneﬁts, greater flexibility in work timing and location, greater supervisor support, and
more family-supportive cultural norms) on work-to-family con/lief, family—to—work
conﬂict”, and emotional exhaustion” will be strongest for employees ﬁom low-income
households and weakest for employees ﬁom high-income households.

The growing number of single parents in the US represents a group with a high
level of personal demands and no resource in the form of a helpﬁrl spouse. COR theory
would predict that they are prime candidates for experiencing work-to-farnily conﬂict,
family-to—work conﬂict, and emotional exhaustion, unless they have access to alternative
resources that can help them make up for those they are missing.

Dependent care beneﬁts, aside from eldercare assistance, are largely aimed at
working parents and should be most helpful to those most actively involved in raising
children. While ﬂexibility in work time and location, supportive supervisors and a culture
that is understanding about non-work interests may be beneﬁcial for all workers, single
parents are more likely have particularly high family demands, and may therefore
experience more positive effects from assistance. For example, in a study of working
parents with children under 12, Kossek (1990) found that single parents reported more
care problems and care-related absences as well as less positive attitudes about their

ability to balance work and child care than married parents.

77

Previous research in the area of dependent care beneﬁts has looked at whether
one’s current use or likelihood of using them is associated with more positive job
attitudes, with mixed results. Grover and Crooker (1995) and Roehling et al., (2001)
both found that their availability was associated with greater commitment and loyalty
even among employees likely to have little use for such policies, although Roehling et
al., (2001) did note that women with school age children seemed particularly loyal in
organizations with more dependent care support. A family-supportive work environment
is likely to signal to workers that their employer cares about them, with positive results
for employees of all types. However, the effects should be strongest among those who
need such support the most: single parents.

H120, b, c: The positive effects of a family-supportive (one with greater ﬂexibility in the
timing and location of work, more access to dependent care assistance, supportive
supervisors, and cultural norms that don ’t penalize workers for putting family ﬁrst) on
work-to-family conflicf’, family-to-work conflicd’, and emotional exhaustion” will be
stronger for single parents.

Work/Family Conﬂict as a Mediator

As previously noted, considerable research has found work-to-family conﬂict to
be closely associated with emotional exhaustion, particularly among women (e.g. Leiter
and Durup, 1996; Bacharach et al., 1991; Etzion, 1988; Netemeyer et al., 1996) and
although the relationship is weaker, family-to-work conﬂict also appears to be correlated
with emotional exhaustion (N etemeyer et al., 1996; Leiter and Durup, 1996). Where the
direction of effects is considered, researchers have generally conceptualized conﬂict

between work and family as a precursor or predictor of burnout and have found evidence

78

to support this (e.g. Ayree, 1993; Greenglass and Burke, 1988). Dealing with conﬂicts
between work and family can use up valuable resources and leave workers with less
energy for work, contributing to the extreme levels of fatigue associated with emotional
exhaustion. The single longitudinal study to look at this relationship found evidence of
strong effects across time for work-to-family conﬂict and emotional exhaustion. Health
care professional experiencing a high degree of work-to-family conﬂict were more likely
to report higher levels of emotional exhaustion three months later. The effects appeared
to be reciprocal; high levels of emotional exhaustion at time one were associated with
greater work-to-family conﬂict after a three-month lag. Emotional exhaustion at the time
of the ﬁrst survey did not predict later levels of family-to-work conﬂict, nor was family-
to-work conﬂict at time one a strong contributor to variance explained in emotional
exhaustion at time two, although the two variables were moderately correlated within
each time frame (Leiter and Durup, 1996). The strong links found between work-to-
family conﬂict and emotional exhaustion are not surprising; both constructs involve
work-related depletion of resources and insufﬁcient reserves of energy. However,
emotional exhaustion, as commonly measured, refers to an extreme state of fatigue that
reduces the energy one brings to work. While work-to-family conﬂict may also involve a
lack of energy, the focus is not on exhaustion experienced in the work domain, but on
how one’s participation in the family is affected by work demands. Correlations between
the two variables reported in previous research support the view that the two are separate,
but related, constructs; they range from —.03 for a group of female physicians in Israel

(Izraeli, 1988) to .60 for a group of male American police ofﬁcers (Burke et al., 1979). A

79

recent meta-analysis estimated the mean weighted correlation for all published studies at
.42 (Allen et al., 2000).

Given the strong relationship between work/ family conﬂict and emotional
exhaustion, reducing one should logically lead to a reduction in the other. A family-
supportive organizational environment should logically have its strongest impact on
emotional exhaustion by reducing conﬂict between work and family. Although a family-
supportive work environment may have some direct effect on emotional exhaustion by
contributing to an overall sense of support, logically the policies, programs, and attitudes
included in this concept are expected to be much more closely related to work/family
conﬂict, which they were essentially developed to reduce.

There is very little previous research that considers work/family conﬂict as a
mediator of the effects of organizational work/family policies and practices (Kossek and
Ozeki, 1998). Thomas and Ganster, however, did ﬁnd some support for their hypothesis
that the work-to-family conﬂict and a sense of control of the work/family boundary
mediated the relationship between work/family supports and various attitudinal and
health-related outcomes. Mediation models were found to ﬁt for the effects of ﬂexible
schedules and supervisor support on job satisfaction, depression, and stress-related health
symptoms, but not for dependent care-related policies, which had little impact. Ayree
(1993) found that, for women, work-parent and work-spouse conﬂict partially mediated
the effects of work role and family stressors on burnout in a small sample of dual-career
couples with preschoolers, but no signiﬁcant mediation effects were found for schedule

ﬂexibility. There were no signiﬁcant mediating effects for the fathers in the study.

80

To summarize, although previous empirical evidence is limited, logic suggests
that a supportive work environment will impact emotional exhaustion by reducing
conﬂict between work and family.

H13: Work-to-family conﬂict and family-to-work conﬂict will mediate the eﬁbcts of a
family-supportive environment (one with greater flexibility in the timing and location of
work, more access to dependent care assistance, supportive supervisors, and cultural

norms that don ’t penalize workers for putting family ﬁrst) on emotional exhaustion.

81

CHAPTER 4: RESEARCH METHODS

Sample Background

The data to be analyzed come from the 1997 National Study of the Changing
Workforce. For the study, commissioned by the F arnilies and Work Institute with
ﬁnancial support from a group of large employers, Louis Harris and Associates surveyed
a randomly selected sample of 3,552 US. working adults aged 18 to 64 by telephone
during a three-month period in the spring and early summer of 1997. The estimated
response rate was 52.9 percent of eligible households contacted.
Data Collection Procedures

Using a computer, 19,057 calls were made to a stratiﬁed unclustered random
probability sample generated by random-digit—dial methods. A regional stratiﬁcation
variable was used to ensure that the sample distribution across regions would be
proportional to the population. Eligibility to participate in the study was limited to
working adults over 18 in the civilian labor force, living in the continental US. in a non-
institutional residence (i.e. a household with a telephone). Of the numbers dialed, 8,149
were found to be non-residential or non-working numbers and 2,338 were determined to
belong to households without a member who met these requirements. Eligibility could
not be determined in 4,831 cases; 3,739 households were determined to have eligible
members, and telephone interviews were completed for 3,552 of these. Since this
analysis focuses on the effects of organizational supports provided by employers, it is
inappropriate to include self-employed individuals. Therefore, the sample to be analyzed

will consist of the 2,877 surveyed workers employed by others.

82

Interviewers followed a computerized protocol to ask respondents about various
aspects of their lives at home and on the job, and coded the responses to all questions
except for occupation, which was coded by the US Bureau of the Census using criteria
developed for the 1990 Census.

Measures

Since this dissertation relies on preexisting data, measures were constructed ﬁ'om
questions that were asked of respondents. The survey author, the Families and Work
Institute, does not list the sources for items; however, where possible, an effort has been
made to determine how similar these measures are to others used in the literature. A
complete list of measures, including all items and scales, can be found in Appendix A.

To develop scales, the survey instrument was examined for items related to the
constructs of interest. Where multiple-item measures were possible and appropriate
preliminary scales were created and factor analyses conducted. Tables reporting the
factor analysis results summarized here are included in Appendix B. Because the large
number of items used would make a single factor analysis difﬁcult to interpret, two
principal components factor analyses were conducted, one for the four resource scales
(autonomy, ﬂexibility, supervisor support, and cultural norms) and one for the three
scales used to measure the dependent variables (work-to-family conﬂict, family-to-work
conﬂict, and emotional exhaustion). Each set contained the items most likely to be
interrelated due to the similarity of what was being measured; because the constructs
were expected to be interrelated, an equimax rotation was used to make the solutions
more interpretable. Four factors with eigenvalues over 1.0 were found in the ﬁrst factor

analysis. After dropping two items expected to measure ﬂexibility due to ambiguous

83

factor loadings, the remaining items clearly loaded on four scales representing autonomy,
ﬂexibility, supervisor support with family issues, and supportive cultural norms. All
factor loadings were above .50 and the four factors jointly accounted for 58.21 percent of
the total variance. In the second analysis, as expected, three factors with eigenvalues
over 1.0 emerged, representing work-to-family conﬂict, family-to-work conﬂict, and
emotional exhaustion. After dropping one item from the work-to-family conﬂict scale
that had ambiguous factor loadings, all items loaded clearly on a single scale and the
three factors accounted for 66.15 percent of the total variance, supporting the contention
that the three constructs are distinct. Following the factor analyses, alphas were
calculated for each multi-item scale; all were above .70.

The ﬁnal scales and other measures used are described in the next section.
Control Variables

Three important variables were identiﬁed that are not among the primary interests
of this study but could affect results: occupation, industry, and race. Occupation and
industry appear likely to be particularly important as work context has been found in
previous research to affect burnout (Cordes and Dougherty, 1993) and work/family
conﬂict (Frone et al., 1993). While research provides evidence that burnout,
particularly emotional exhaustion, may be experienced in a wide range of jobs, studies
that have included workers in a variety of jobs have found occupation to have signiﬁcant
effects (e.g. Ayree, 1993). Possible implications of race are less clear; however, several
studies have reported differences in the degree of burnout experienced by workers of
different races (Lanaku and Scandur, 1996; Salyers and Bond, 2001). Other research

supports the idea that race may combine with other resources to inﬂuence how stress is

experienced (Broman et al., 1995). Research also indicates that there may be differences
in familial roles, attitudes, values, and expectations among different racial groups
(Broman, 1993; Orbuch and Custer, 1995). Given this, race, industry, and occupation
will be included in the analysis as control variables.

Race. A single measure that directly asked respondents about their racial
background was used to assess race. Although considerable effort was made to include
respondents from a wide range of racial and ethnic backgrounds, the vast majority of
study participants were either white or Aﬁican American, and other racial groups (e.g.
Asian, American Indian) were represented by such small groups that it would be difﬁcult
to conduct useful analyses. Race is not a major focus of the current study and there is no
previous research that indicates differences associated with belonging to any of the
racial/ethnic groups that were represented by a small number of individuals. Therefore,
responses were coded into three categories, white, black, and other (including mixed) and
appropriate dummy variables created for each.

Industry. Earlier research on burnout focused on workers in the service industry,
based on the notion that working closely with people was a particularly strong drain on
personal resources. According to earlier conceptualizations of burnout, similar effects
were not expected for workers whose tasks focused on production-oriented tasks that
involved working more with things than people. While more recent burnout research
(e.g. Demerouti et al., 2001) appear to indicate that this phenomenon may be found in
many different types of jobs, the key differentiation, theoretically, has been between
production-oriented jobs that presumably involve less people work and jobs in the service

industry, which by deﬁnition involve serving people. Therefore, industry was measured

85

by a single-item variable that classiﬁed workers as employed in either the service (coded
0) or goods-producing (coded 1) industry. Coding was done by the US Bureau of the
Census based on information provided by the respondent.

Occupation. Even more important than the type of industry an employee works in
is the kind of job he or she does. The respondents’ occupation was measured using a
single, straightforward item that asked participants what their occupation is, or what type
of work they do. The responses were classiﬁed into seven categories of work: executive,
administrative, and managerial; professional; technical; sales; administrative support;
service; and production, operation, or repair. Coding was done by the US Bureau of the
Census based on information provided by the respondent. For this analysis, a dummy
variable was created to represent each occupational category, and scored (1) where
applicable and (0) in all other cases.
Personal Demands and Resources

Sex. At the end of the interview, the interviewer determined the respondent’s sex
based on questions and responses. Men were coded as (l) and women as (2).

Elder/disabled care. Time spent caring for elders and disabled adults was
measured by asking respondents whether they currently provide special attention or care
for a disabled adult or someone 65 years or older. Those who said they did were asked
much time per week is spent providing assistance in person (such as performing
household work or providing transportation) and in other ways (such as arranging for
care and handling ﬁnances). Time spent in all categories was converted to hours and
summed for each respondent who provided this information; those who said they did not

currently provide any such care were assigned a 0 for this variable.

86

Number of children 0-12, 13-18. The number of children the respondent had in
each age group was determined by combining the results to a series of questions.
Participants were asked if they had a child who was living with the respondent at least
halfof the year or for whom the respondent had guardianship for the full year (such as a
child at boarding school). If the answer was yes, and the respondent was asked if any of
these children were under 18. If so, the respondent was asked to give the age of each
child who ﬁt these qualiﬁcations. Children were divided into two age groups, 1-12 and
13-18, and responses were summed to determine the number of children in each age
group.

While it is possible and potentially meaningﬁrl to compare the experiences of
parents of children in a wide range of smaller age groups (e.g. infants, toddlers,
preschoolers, upper and lower elementary students, driving and non-driving teenagers)
such an approach would result in small numbers of study participants with children in
each age group, particularly during efforts to examine each income group. Initial
analyses indicated that this was the case. Regression equations” were run with children in
one, two, or several age groups. Although the strongest effects in these tests were found
for the total number of children, theoretically it makes more sense to examine two groups
of children, young ones requiring extensive care and older one who are more capable.
Therefore, for this analysis two categories were selected to represent child-related
demands, one for the number of children under 13 and one for the number of children in
their teens. Division was made at this point because children under 13 — whether infants
or elementary students — generally require constant supervision; when they are not in

school, parents must either be at home with them or make arrangements with babysitters,

87

day care centers, family members, or others to ensure that they are properly supervised.
Teenagers also require parental support, but are generally considered more capable of
caring for themselves and even acting as caregivers for younger children. Many are very
busy with their own jobs and activities and they can often even handle their own
transportation. As such, they are considerably more independent and may place different
types of demands on parents, making it more logical to consider them separately rather
that combine them with younger children.

Partner. Whether the respondent was living with partner was assessed by asking
respondents whether they were married, whether they were living with their spouse, and
whether they were currently living with someone as a couple. Those who were living
with their spouse or living with someone as a couple were coded as (1), all others were
coded as (0).

Familial care. For parents, whether most or all needed care for the youngest child
was provided by a family member was again assessed through answers to a series of
straightforward questions about whether some or all needed child care outside school
hours was provided by the respondent’s partner. Respondents without a partner who
provided the majority of care were also asked what the main source of child care they
relied on was — a day care center, older sibling, other relative, etc. Responses were
combined and recoded so that respondents who had a family member that provided most
or all needed care for their youngest child were coded as (1). All others, including those
who did not have children, were coded as (0).

Household income level. Household income for 1996 (the year prior to the study)

was also assessed by directly asking respondents to estimate their household income, and

88

by questions that asked them to estimate the amount, within $10,000 ranges, for the ten
percent who could not provide a more accurate ﬁgure. It was deemed more appropriate
to turn to a reliable external source for income classiﬁcations than to divide the
respondents in the data set into three equal groups for two reasons. First, survey response
rates are likely higher for middle-class families than members of the working poor (who
are less likely to own telephones) or wealthy individuals (who can pay to have their calls
screened or answered for them). Second, because income is fairly normally distributed,
more respondents fall into the middle than at the extremes, so dividing the sample into
thirds would have the effect of including individuals who are still very close the mean
income level in the upper and lower income groups.

An alternative approach would be to deﬁne the groups based on anticipated costs
— e.g., the high-income group should earn enough to hire a housekeeper and/or full-time
nanny. However, it was determined that, due to differences in prices across the US, it
would be unreasonably complex to estimate the cost of high- and low-end assistance with
child care and household chores.

Therefore, income level was derived by using 1996 household income to place the
respondents into three categories created based on US Census data for the year. Those
whose household income fell into the lower 40% of households (incomes below $27,760)
were classed as lower income, those with incomes in the next 40% (827,761-868,015)
were classiﬁed middle income, and those whose incomes placed them in the top 20% of
households (above $68,015) were classiﬁed as higher income. Approximately one-
quarter of the sample fell into the lower-income group, one-half into the middle-income,

and one-quarter into the higher-income group. The cutoffs were viewed as generally

89

appropriate as, below $28,000, the cost of day care for a single child represents a
considerable proportion of household income. At the middle-income level, such
expenses, while high, are still manageable. Families earning above $68,000 should be
able to absorb such costs without much sacriﬁce.

Age. Respondents were asked to give the month and year of their birth; this
information was used to calculate their precise age in years at the time of the survey.
Work-Related Demands and Resources

Work hours. The total number of hours worked a week was measured by asking
respondents how many hours a week they usually do work, both paid and unpaid, at all
jobs. Results were summed by the computer and participants were asked to conﬁrm that
the ﬁgures provided were accurate.

Irregular schedule. Respondents were given a range of work schedule choices
(e.g. regular daytime, split shift, etc.) and asked to choose the one that most accurately ﬁt
their own. While there was a wide range of non-regular schedule options given, there is
no theoretical reason to expect strong differences in the study outcomes between them.
However, working a regular daytime shift is likely to be associated with less work/family
conﬂict because it is generally most congruent with the schedules of children and other
family members, ﬁtting with the most common hours for day care, school, work, and
other activities. Therefore, responses were recoded so that a regular daytime schedule
was coded as (0) and other (irregular) arrangements were coded as (1).

Autonomy. Work autonomy was measured with the average score on a three-item
scale used in the 1977 Quality of Employment Survey (Quinn and Staines, 1979) and

subsequent research. A sample item ﬁ'om the autonomy scale reads: “I have a lot of say

about what happens on my job.” Respondents were asked if they strongly agreed (I)
agreed (2), disagreed (3) or strongly disagreed (4). Responses were reversed and recoded
so that a higher score on the 1-4 likert scale indicates greater autonomy. Coefﬁcient
alpha for the scale was .70.

Tenure. Tenure was measured by a single item that asked respondents how long
they had been working for their current employer. All responses were converted to years.
Organizational Work/Family Environment

Time flexibility. Flexibility in the timing of work was measured using three items
to create a scale that addresses the degree to which the respondent can adjust working
time, including control over scheduling work hours and the degree of ﬂexibility in
starting and ﬁnishing times. A sample item reads: “Overall, how much control would
you say you have in scheduling your work hours - complete control, a lot, some, very
little, or none?” Items were reverse coded as necessary to ensure that a higher scores
indicated greater ﬂexibility, and responses were averaged to create a scale ranging from 0
(low ﬂexibility) to 4 (high ﬂexibility). Alpha for the scale was .72.

Location ﬂexibility. The degree to which employees had ﬂexibility in where they
did their work was measured by combining responses to three items that asked how many
regularly scheduled hours employees worked from home and, for those who answered
none, whether they thought it might be possible to do their work from home and, for
those who answered yes, whether they would be allowed to do so if they asked. Those
who cmrently worked at home or who would be allowed to do so were coded 1, those
who didn’t and thought they would or could not, were coded 0, so that a higher score

indicates greater ﬂexibility.

91

Dependent care beneﬁts. Eight formal dependent care assistance policies were
selected for inclusion in the study based on the questions asked of respondents. They are:
maternity leave, paternity leave, ability to take time off work to care for a sick child
without losing pay or vacation days, child care information and referral services,
eldercare information and referral services, pre-tax dependent care spending accounts,
on- or near-site daycare, and direct ﬁnancial assistance with child care. Each is measured
by a single item asking the respondent whether his or her employer offers the beneﬁt. An
example: “Does your employer have a program or service that helps employees ﬁnd child
care if they need it, or not?” Responses to the eight items were coded 1 (yes, available)
or 0 (not available) and summed to create a single index of beneﬁt availability that
ranged from 0 (no beneﬁts available) to 8 (all of these beneﬁts available). This approach
is similar to those used in other studies in this area (e.g. Grover and Crooker, 1995;
Thompson et al., 1998; Allen, 2001).

Supervisor support was measured using the average of responses to ﬁve items
indicating, on a scale of 14, how supportive the main person the respondent reports to is
in terms of work/family issues (e.g. “My supervisor is understanding when I talk about
personal or family issues that affect my wo ”). A higher score implies greater
agreement that one’s supervisor is supportive. The scale, which had an alpha of .84, is
similar to one used by Thomas and Ganster (1995) to measure supervisor support.

F amily-supportive cultural norms are measured with a ﬁve-item scale (alpha .76)
where a higher score implies greater agreement that one’s organization is sympathetic

and does not penalize employees who put families ﬁrst. A sample item: “At my place of

92

employment, employees who put their family or personal needs ahead of their jobs are
not looked on favorably (reverse scored)” Responses range from 1-4.

Total work environment. Because understanding how much the overall
environment experienced by a worker affects the outcome variables may be more
important than simply examining the indicators individually, a summary variable
combining all ﬁve measures of the work environment was created. To ensure that the
four aspects of a supportive environment included in this study - ﬂexibility, beneﬁt
availability, supervisor support, and cultural norms - were all given equal weight, the
question on location ﬂexibility was averaged with those of the timing ﬂexibility scale and
the mean of the ﬁve items was calculated, resulting in a scale range maximum of four.
The beneﬁts index was divided by two, making the maximum possible value four. No
changes were made to the supervisor support or cultural norm scales, which both had four
as the maximum possible value. The four measures were then summed to create an
estimate of overall organizational support, which ranged from 1-16.

Dependent Variables

Work—to—family conﬂict was measured using the mean response to four items to
create a scale (alpha .81) with a range of 1-5. A higher score reﬂects more experience
with conﬂicts over the past three months. A sample item: “In the past three months, how
often have you not had enough time for your family or other important people in your life
because of your job? Would you say very often, often, sometimes, rarely, or never? ”

Family-to-work conﬂict was measured using a similar ﬁve-item scale (alpha .85),
with a higher score indicating more frequent experience with conﬂict in the preceding

three months. “How often has your family or personal life kept you from doing as good a

93

job at work as you could?” is one item included. Both work/family conﬂict scales
contain items similar in content to those in other scales commonly used to measure
work/family conﬂict and spillover (e.g. Koppelman, Greehaus and Connolly, Gutek,
Searle and Klepa, 1991).

Emotional exhaustion was measured using four items from the emotional
exhaustion component of the Maslach Burnout Inventory (Maslach and Jackson, 1981),
the scale most commonly used in this ﬁeld. A sample item: “How often during the past
three months have you felt used up at the end of the workday?” Alpha for the
abbreviated 4-item scale was .87, which is comparable to reliability estimates reported in
other studies for the full 8-item scale (Lee and Ashforth (1986) report a weighted mean
reliability of .86 for 47 studies).

Data Analysis

The main goals of this study were to estimate the effects of different demands and
resources on the three outcomes —— work-to-family conﬂict, family-to-work conﬂict, and
emotional exhaustion — and to examine in greater detail any differences in the impact of a
supportive work environment for individuals with fewer personal resources, namely,
single parents and those from lower income households. The emphasis was on the effects
of the organizational environment indicators within the context of other personal and
work-related resources and demands rather than the overall ﬁt of the model as a whole
for each demographic group. Therefore, the effects of the independent variables on the
set of dependent variables as outlined previously in the hypotheses section were
estimated using hierarchical multiple regression. For each equation, the dependent

variable (work-to-family conﬂict, family-to-work conﬂict, and emotional exhaustion)

was regressed on the full set of predictors. In all cases, the ﬁrst three steps were

identical. The control variables (dummies representing race, industry and occupation)
were entered at the ﬁrst step. Because it had the most variety and least clear implications,
the “other” racial group was used as the omitted category for race. “Service” was used as
the omitted category for occupation because many of the occupations that have been
studied in the burnout and work/family literature fall within it.

At the second step, variables representing personal demands and resources were
entered. At the third step, job-related demands and resources were entered. Variables
representing organizational efforts to create a supportive work environment - ﬂexibility
in work time and location, dependent care beneﬁts available, supervisor support, and
cultural norms - were entered at the fourth step. Separate regression equations were
conducted to examine the effects of the overall environment and the indicators separately.
Steps 1-3 remained the same, but in the ﬁrst regression equation for each dependent
variable (labeled equation a Tables 2-4), the total environment summary variable was
entered at step four. In the second regression equation for each dependent variable, the
individual environment indicators were entered together at step four (labeled equation b
in Tables 2 and 3, which report the results for work-to-family and family-to-work conﬂict
regressions, and equation c in Table 4, which reports the results for emotional
exhaustion).

Two dummy variables were also included at step four in all regression equations
to control for the effects of missing data. There were a number of missing responses for
some of the dependent care assistance scale items. The question about being able to miss

work to care for a sick child without losing pay or vacation time was only asked of

95

parents with children under 18, resulting in missing data for all nonparents on this issue.
Many respondents appeared to be unclear about paternity leave, also, as that item was
missing for 519 respondents. However, these represent important independent variables
for the study, and many of these people reported information on other dependent care
beneﬁts. In order to retain them in the analysis, missing data was replaced with the mean
for the sample for that beneﬁt and the results were combined for the index. The effects of
replacing the missing data on the question regarding time off to care for sick children are
already controlled for because having children under 18 is part of the regression equation.
However, to control for any common source of variance in non-response to other items, I
created a dummy variable that coded those noncontingently missing responses in the
index as 1 and those without missing data as 0. This dummy was included in the
regression analysis at the same step as the dependent care index to control for any bias
that might be associated with not providing answers. Such a procedure is recommended
by Cohen and Cohen (1983) as a way to keep individuals who did not provide
information on all included variables in the analysis without affecting estimates of
relationships. As noted by Cohen and Cohen (1983), such an approach also makes it
possible to examine the likelihood that individuals missing data on a particular scale may
be different in some systematic way and to identify the effects.

Similarly, a large number of study participants were not asked about supervisor
supportiveness because they could not identify a single individual who they reported to as
a supervisor. Again, missing data for this scale was replaced by the mean and a dummy
variable created. Those who did not have a supervisor and thus did not respond to these

questionswerecodedasalandallothersasao. Inallothercaseswheredatawas

missing, listwise deletion was used. This approach is similar to the one adopted by
Roehling et al. (2001) in their analysis of data from the 1992 NSCW and again follows
the recommendations of Cohen and Cohen (1983). It allows us to include as many
individuals as possible in the analysis and examine the effects of not having a supervisor.

The moderated relationships predicted involving workers from low-income
households and single parents were tested by entering the interactions predicted by the
hypotheses (income level x environment, and single parent x environment) into the
regression equations at step 5. In the ﬁrst regression equation for each dependent
variable (labeled equation a in Tables 2,3, and 4, where results are reported), interactions
between the total environment summary measure and the two moderators — income and
single parent status — were entered at this ﬁfth and ﬁnal step. In the second regression
equation for each dependent variable (labeled equation b in Tables 2 and 3, where the
work-to—family and family-to-work regressions are reported), interactions were tested
separately for each of the ﬁve indicators — ﬂexibility in work time, ﬂexibility in work
location, dependent care beneﬁts, supervisor support with family issues, and cultural
norms.

V In addition, the effects of income level were regarded as most important to
increasing understanding in the ﬁeld, so these were also illustrated by running separate
regression equations for each of the three income groups where signiﬁcant interaction
effects were found. For each income group, the effects of the total environment summary
variable and the individual impact of each environment indicator were again examined by
estimating two separate regression equations, again, labeled a (total environment

summary variable) and b (ﬁve individual environment indicators) in the results tables.

97

In the case of emotional exhaustion, hypothesis 13 predicts that work-to-family
conﬂict and family-to-work conﬂict may mediate the relationship between organizational
efforts to create a supportive work environment and the dependent variable. This was
examined in an additional pair of regression equations for that dependent variable. In the
ﬁrst (labeled equation b in Table 4, where the results for emotional exhaustion were
reported), the total environment variable was entered at step four and the two conﬂict
variables were entered at step ﬁve. In the second, (labeled equation c in Table 4) the
individual environment indicators were entered at step four and the two conﬂict variables
were again entered at step ﬁve.

Criteria established by James and Brett (1984) were used to determine if
mediation effects exist. They specify that, ﬁrst, the independent variable must be
signiﬁcantly related to the mediator variable. Second, the mediator variable must be
signiﬁcantly related to the dependent variable. Third, when the inﬂuence of the mediator
variable is held constant, the effect of the independent variable on the dependent variable
should be nonsigniﬁcant. To test for the ﬁrst two criteria, partial correlations for
organizational work/life supports controlling for the inﬂuence of other predictors were
examined. Because several signiﬁcant relationships were found, emotional exhaustion
was again regressed on the set of predictors, this time with the conﬂict and spillover
variables included in the equation in a ﬁfth and ﬁnal step. According to James and Brett
(1984) the degree to which work-to-family and family-to-work conﬂict mediate the
effects of job characteristics is estimated by assessing the degree of change in the beta-
weights after these additional variables are included. If they become non-signiﬁcant, a

hill mediation model would be supported. Signiﬁcant change can be interpreted as

98

partial mediation, and no real change is an indication that the effects are felt directly and

not mediated by conﬂict between work and family.

99

CHAPTER 5: RESULTS

Outline of the Chapter

The results of the study are summarized in Tables 1-5. The means, standard
deviations, and intercorrelations between the main study variables are reported in Table
1. This chapter will begin with an overview of the sample and descriptive statistics for
the main variables, followed by an examination of the hypotheses and results.
Hypotheses 1-10 dealt with the direct effects of the personal and organizational demand
and resource measures on the dependent variables: work-to-family conﬂict, family-to-
work conﬂict, and emotional exhaustion. They were tested by examining the simple
Pearson correlations and the standardized regression coefﬁcients for the full sample.
Hypotheses 10-13 dealt with mediating and moderating effects. The interactions were
examined in two ways, ﬁrst, by including them in the regression equations for the full
sample, and second, by running separate regressions for the three income levels where
signiﬁcant interactions were found. Tables 2, 3, and 4 show the results of the hierarchical
regression analyses for the three dependent variables, both for the ﬁrll sample and for the
lower, middle, and high income groups separately. A summary of the hypotheses and
study ﬁndings can be found near the end of this chapter in Table 5. Although the
discussion of the results of the other hypotheses will focus on the regression analyses for
the full sample, where interesting differences in these relationships were found between

the income groups, they will also be noted.

100

Table 1: Means, Standard Deviations and Intercorrelations

 

 

Mm 5.12. .1. 3 3. .4. 5 9

1. Manager .16 .37

2. Professional .18 .39 -.20"

3. Technical .05 .22 -.10" -.11"

4. Sales .11 .31 -.15" -.16“ -.08**

5. Admin. .16 .36 -.17" -.20" -.10" -.15"

6. Service .11 .31 -.15" -.16"”" -.08 -.12" -.15**

7. Prod/op/repair .24 .43 -.25“ -.27" -.13" -.19" -.24" -.20"”"
3,1ndustry‘ 1.78 .42 .00 .12" 04* .14" .13" .18"
9. White .80 .41 05* .07" .00 .01 -02 -.08"
10. Black .12 .32 -.04* -.O6"““ .01 -.02 .01 09"
11. Other race .09 .28 -02 -03 -.01 .02 .02 .01
12. Sex 1.52 .50 -02 07" .03 .02 .26“ .06”
13, Bid/0th cm} 1.75 6.52 -05" -01 .00 —04* .01 07"
14. # kids < 13 .58 .91 -01 -02 -.01 -01 -02 -.01
15. # kids 13-18 .25 .57 .00 .00 .00 .00 -05* .00
16. Partner .61 .49 .05" .02 -02 -04* -02 -.06”
17, Family We .23 .42 -03 -02 -.06" .02 -05" .03
18. Age 40.18 11.9 .06" .03 -03 -04* 04* -04"
19. Income level 2.0 .71 .19" .17” .02 -.01 -.04" -.19"
20.w6rk hrs/wk 45.97 13.6 .13" .06" -02 -.05* -.14" -.13"
21. Irreg. sched. .28 .45 -.10** -.10" -04* .11" -.10" .17"
22. Autonomy 3.03 .74 .18" .08" -.04* .01 -.06" -07"
23. Tenure 7.54 8.15 .08" .03 .02 -09" -.01 -.08“
24. Flex time 1.55 1.38 .22M .07" .01 .08" -.07** -05"
25. Flex place .25 .43 .15" .15” .02 .06” -.08" -.07"
26, Dep- cm‘ 3.22 1.48 .09" .10" .06" -.10" .03 —.O6"‘*
27. Sup. support 3.34 .66 05* .03 -03 .03 .04 -03
28. Cult. norms 2.96 .73 .11" 07" .03 .02 .02 -.O8"
29, To, cm} 9.33 2.22 .21" .13" .03 .03 -02 -09"
30, we” 2.92 1.04 04* 05* .01 -02 -02 -.06”
31, NC7 1.93 .70 .02 .09" -03 -03 -02 -03
32, 558 2.97 1.04 .00 .03 -.01 -02 -01 -01

 

 

Pearson zero—order correlations between major study variables. 1. Goods-producing =1, service =2.
2. Hours per week caring for elders and disabled adults (Continued on the next page)

101

Table 1: Means, Standard Deviations and Intercorrelations (Continued)

 

Z .8. 2 L9 LI. 2 1.3. .1.! L52
1. Manager
2. Professional
3. Technical
4. Sales
5. Admin.
6. Service
7. Prod/op/rep
8. Industry -.47“‘
9. White -.03 -.02
10. Black .03 .03 -.72"
11. Other race .01 -.01 -.61" -.ll"
12. Sex -.34" .23" -.03 .07" -.03
13. Eld/oth care .02 .01 ~05" .07" .00 .06M
14. # kids < 13 .06" -.05* -.06"“' .04“ .05“ -.O6“ -.07*"
15. # kids 13-18 .02 .00 -.03 .06“ -.02 .03 .03 .01
16. Partner .03 ~07" .09" -.10"”" -.02 -.10** -.02 -.25** .12"
17. Family care .07 -.07" -.06" .03 .06" -.12*"' -.06" .67" .06"
18. Age -.04"' -.01 .10" -.03 -.10** .07" .08" -.24" .09"
19. Income -.15" -.01 .13” -.13" -.03 -.11** -.04" .02 .05‘I
20.Work hrs/wk .10" -.11" -.02 .02 .00 -.26" -.01 .06" -.01
21. Irreg. sched. .07" .02 -.03 .02 .02 -.07" .02 .00 -.O3
22. Autonomy -.11** .01 .08" -.12"'* .02 -.09"”" -.03 .00 .00
23. Tenure .03 -.07** .05“ .00 -.07" -.05" .01 -.10" .02
24. Flex time -.22** .07" .06" -.07** -.01 -.07"‘"‘ -.03 -.03 -.01
25. Flex place -.19" .09" .01 -.01 .00 -.03 -.01 04* .02
26. Dep. care -.11" .09" -.05"' 04* .02 -.02 -.02 -.01 -.01
27. Sup. support -.09** .06" .06" -.05"‘ -.02 .03 -.Ol -.03 .02
28. Cult. norms -.14“ .03 .13" -.08** ~09" .06" -.05‘ .01 .00
29. Tot. env. -.25** .11" .07" -.06** -.04 -.01 -.04" -.Ol .00
30. WFC -.01 .00 .00 -.01 .02 .04 .02 .13" .07”
31. FWC -.05** .00 -.02 -.01 .04 .03 .04 .11" .04"
32. EB -.01 .01 -.02 .02 .01 .07" .01 .04“ .00

 

 

 

3. Most child care provided by a family member 4. Number of dependent care benefits 5. Total
environment 6. WFC = work-to-family conﬂict 7. FW C = family-tework conﬂict 8. EE = emotional
exhaustion. Due to listwise deletion for missing data, Ns range from 2,553 to 2,877. * signiﬁcant at the
.05 level. “ signiﬁcant at the .01 level.

102

Table 1: Means, Standard Deviations and Intercorrelations (Continued)

 

 

1.9 ﬂ

1. Manager

2. Professional

3. Technical

4.Sﬂes

SHAdnnn.

6. Service

7.anﬂophepan

8. Industry

9. White

10. Black

11. Other race

12. Sex

13. Eld/oth care

14. # kids < 13

15. # kids 13-18

16.Panner

17. Family care .27“

18. Age .14" -.18"
19. Income level .37" .01
20.Work hrs/wk .04 .03
21. Irreg. sched. -.10** .04
22. Autonomy .08" .02
23. Tenure .13" -.08“
24. Flex time 04* -.01
25. Flex place .05" .04
26. Dep. care .02 -.02
27. Sup. support 05* -.02
28. Cult. norms .04“ .01
29. Tot. env. .06“ -.01
30. WFC .06" .09"
31. FWC .01 .06"
32 EE, a04 .02

18

.11”
a04
{14“
.07"”'K
.48""'I
a0]
204’*
a0]
.06**
.02
.01
nlO**
206*‘
-.12"“'l

12

.18”

-.10"“"

.15**
.23**
.21**
.l6‘”'I
.l6**
.03
.l3**
.23**
.O7**
.05H
.02

go

.02

.08”

.09"
.05"I

.12"

.09"
--.04"I
-.04*
.04
.24"
.07"
.19”l

21

-.05**
-.10"
.02
-.01
-.O3
_'06*#
~08"
-.04
.07M
.00
.02

g;

(70)
.06.!
.38**
.21**
.16**
.31**
.27**
.45**
414**
402
423**

23

.03

.12"

-.Ol
.03
-.04"'
.01
-.O4"I

 

103

 

Table 1: Means, Standard Deviations and Intercorrelations (Continued)

 

 

2.1 2.5 A5 .21
1. Manager

2. Professional

3. Technical

4. Sales

5. Admin.

6. Service

7. Prod/op/rep

8. Industry

9. White

10. Black

1 1. Other race

12. Sex

13. Eld/oth care

14. # kids < 13

15. # kids 13-18

16. Partner

17. Family care

18. Age

19. Income

20.Work hrs/wk

21. Irreg. sched.

22. Autonomy

23. Tenure

24. Flex time (.72)

25. Flex place .30"
26.Dep.care .19" .11"
27. Sup. support .19" .08" .20" (.89)
28. Cult. norms .25" .09" .18" .43"

29. Tot. env. .72" .42" .56” .60‘“I
30. WFC -.O9** .05‘ -.10" -.23‘”'I
31. FWC .00 .12" .01 -.ll"
32. EB -. 14" .02 -.04“ -.31"

g;

(76)
.65**
426**
418**
428*3

12

-.23"'*
-.O8"
-.29"

30

(81)
.42**
.59"

g;

(85)
.35rr

3;

(87)

 

104

 

Table 2: Results of Multiple Regression Analysis - Work-to-Family Conﬂict

 

 

Predictors Full Sample
Step 1 Controls Sun; Steal 58.2.2 Strut m £29.41: 8:11.512
Managerial/executive .09" .09 .09" .13" .12" .12" .12*"I
Professional .09" .08"I .08" .11" .11" .10" .10"
Technical 06* .05 .05“ .07" .07“ .06" .07"
Sales .04 .04 .04 .05“ .05“ .05 .05
Administrative .05 .04 .06‘ .07" .08” .08" .08"
Production/operator/repair .07 .08“ .04 .02 .02 .04 .04
Industry .00 .00 .00 .01 .02 .01 .01
White -.01 .00 .00 .01 .02 .03 .03
Black -.01 .00 -.O4 -.03 .Ol -.02 -.02
Step 2 Pars. Dem. & Res.
Sex .07" .11M .10" .10" .12" .12"
Elder/disabled care .03 .02 .02 .02 .01 .01
# kids <13 .11" .09" .08‘“I .08" .08" .08"
# kids 13-18 .06“ .07" .06" .06" .06" .06"
Partner .02 .05" .05“ .05“ .05“ .05“
Family child care -.01 .00 .00 .00 .00 .00
Age -.08" -.05* -.06" -.O7" -.06‘I -.06“
Income level .05“ .03 .06" -.14 .05“ -.08**
Step 3 Work Dem. & Res.
Work hours .26“ .25" .25" .24" .24“
Irregular schedule .09“ .10" .09" .09" .08“
Autonomy -.17"”" -.07" -.07" -.07" -.07"”"
Tenure ~02 -.02 -.02 -.02 -.03
Step 4 Eq. :1 Tot. Env.
Total environment -.26” ~40"
No supervisor -.01 -.01
Missing dependent care .00 .00
Step 5 Eq. 11 Interaction
Environment x income .27"
Step 4 Eq. b Indicators
Flex time -.05* -.05"‘
Flex place .04“ .04“
Dependent care beneﬁts -.O7"* -.24"
Supervisor support -.10" -.10"
Cultural norms -.20“ —.20**
No supervisor -.02 -.02
Missing dependent care -.02 .00
Step 5 Eq. [3 Interaction
Dependent care x income .23"
Model Summary
N 2,796 2,61 1 2,571 2,453 2,453 2,453 2,453
R .08 .19 .36 .42 .43 .45 .46
n2 .01 .04 .13 .18 .18 .20 .21
Adj R’ .00 .03 .12 .17 .17 .20 .20
Change R2 .01 .03M .09" .05M .00" .08" 00"

 

 

Notes: Standardized regression coefﬁcients shown. Two equations were estimated; steps 1-3 were
identical in both. Service and other race were the omitted categories for occupation and race. In equation
a, the effects of the total environment variable and its interaction with income level were estimated at
steps 4 and 5. In equation b, the effects of each environment indicator were examined separately at step
4. At step 5, the interaction of each with income was tested separately; the only signiﬁcant interaction
found is reported here in the step 5b column. "‘ signiﬁcant at .05 level " signiﬁcant at .01 level

105

Table 2: Results of Multiple Regression Analysis — Work-to-Family Conflict

 

(Continued)

Predictors Lower Income Sample

Step 1 Controls $511.1 $222 $211.: mm m
Managerial/executive .01 .00 .03 .05 .04
Professional .00 .01 .03 .07 .06
Technical .07 .06 .08 .10“ .10"
Sales .02 .03 .02 .04 .03
Administrative .02 .03 .03 .04 .05
Production/operator/repair .09 .07 .04 .05 .08
Industry .01 .00 .00 .02 .03
White .02 .06 .03 .06 .06
Black -.01 .00 -.04 -.01 -.01

Step 2 Personal

Demands & Resources
Sex -.01 .01 .03 .05
Elder/disabled care .01 .01 .00 -.01
# kids <13 .04 .03 .01 .01
# kids 13-18 .09" .07 .05 .04
Partner .03 .04 .04 .03
Family child care .05 .08 .09 .08
Age -.11" -.08 -.10* -.09'

Step 3 Work Demands

& Resources
Work hours .20" .18" .18“
Irregular schedule .14" .13" .12”
Autonomy -.22** -.10“ -.12"
Tenure .05 .05 .06

Step 4 Equation a

Total Environment
Total environment -.29"
No supervisor -.02
Missing dependent care .01

Step 4 Equation b

Environment Indicators
Flex time -.02
Flex place .02
Dependent care beneﬁts -. l6"
Supervisor support -. 13"
Cultural norms -. 16"
No supervisor -.02
Missing dependent care .01

Model Summary
N 708 698 687 643 643
R .10 .20 .37 .45 .47
R2 .01 .04 .14 .20 .22
Adj 1?.2 -.oo. .01 .11 .17 .19
Change R’ .01 .03" .10" .06” .09"

 

 

 

Notes: Figures are standardized regression coefficients. Two regression equations were estimated; steps
1-3 were identical in both. Service and other racial background were the omitted categories for
occupation and race. In equation a, the effects of the total environment measure were estimated at step 4
and are reported in column 4a. In equation b, the effects of each individual environment indicator were
examined separately at step 4 and are reported in column 4b. "‘ signiﬁcant at .05 level "”' signiﬁcant
at .01 level

106

Table 2: Results of Multiple Regression Analysis — Work-to-Family Conﬂict

 

 

(Continued)

Predictors Middle Income Sample

Step 1 Controk $211.1 Sin; §t2L3 Stems: M
Managerial/executive .06 .08 .09‘ .12” .10‘I
Professional .04 .05 .07 .09 .08
Technical .04 .04 .05 .06 .05
Sales .07 .08 .07 .07 .06
Administrative .06 .05 .09“ . 10 .09
Production/operator/repair .04 .07 .06 .03 .04
Industry -.01 .00 .00 .01 .00
White .00 .00 .00 .01 .02
Black .03 .02 -.03 -.Ol .02

Step 2 Personal

Demands & Resources
Sex .10" .16" .15" .16”
Elder/disabled care .04 .03 .03 .03
#kids <13 .15“ .12" .11" .12"
#kids 13-18 .09" .10“ .11” .10"
Partner .01 .04 .05 .05
Family child care -.05 -.02 -.03 -.02
Age -.06* -.03 -.04 -.04

Step 3 Work

Demands & Resources
Work hours .28" .27 " .26"
Irregular schedule .09M .09" .08"
Autonomy -.16*"‘ -.O7"‘ -.07"
Tenure -.05 -.06* ~07"

Step 4 Equation a

Total Environment
Total environment -.24"
No supervisor .02
Missing dependent care -.03

Step 4 Equation b

Environment Indicators
Flex time -.05
Flex place .04
Dependent care beneﬁts -.06"'
Supervisor support -.07"‘*
Cultural norms -.22"‘*
No supervisor .01
Missing dependent care -.03

Model Summary
N 1,299 1,288 1,268 1,212 1,212
R .07 .21 .38 .44 .47
R’ .00 .04 .15 .20 .22
Adj R’ .00 .03 .13 .18 .21
Change R2 .00 .04" .11M .05" .08"

 

 

Notes: Figures are standardized regression coefﬁcients. Two equations were estimated; steps 1-3 were
identical in both. Service and other racial background were the omitted categories for occupation and
race. In equation a, the effects of the total environment measure were estimated at step 4 and are reported
in column 4a. In equation b, the effects of each individual environment indicator were examined

separately at step 4 and are reported in column 4b.

107

" signiﬁcant at .05 level " signiﬁcant at .01 level

Table 2: Results of Multiple Regression Analysis - Work-to-Family Conflict

 

 

(Continued)

Predictors Upper Income Sample

Step 1 Com-I- ﬁrst; m2 Etta; m m
Managerial/executive .19“ .23" . 15 .21" .10“
Professional .20‘ .24" .17 .19" .08
Technical .06 .07 .05 .08 .05
Sales -.01 .01 .00 .02 .06
Administrative .06 .07 .06 .07 .09‘
Production/operator/repair .07 . 10 .00 -.04 .04
Industry .03 .01 .03 .03 .00
White -.06 -.03 -.02 -.02 .02
Black -.05 -.05 -.05 -.06 -.02

Step 2 Personal

Demands & Resources
Sex .07 .1 1‘ .09" .10"
Elder/disabled care .05 .03 .03 .02
# kids <13 .08 .09 .09‘ .08
# kids 13-18 .00 .01 .00 .00
Partner .00 .01 .00 .01
Family child care .03 .00 .01 -.01
Age -. 10* -.09 -.09 -.08

Step 3 Work Demands

& Resources
Work hours .28" .31" .28“
Irregular schedule .01 .05 .04
Autonomy -.1 1” .00 .01
Tenure -.0] .01 .00

Step 4 Equation a

Total Environment
Total environment -.27"
No supervisor -.07
Missing dependent care .04

Step 4 Equation b

Environment Indicators
Flex time -.09*
Flex place .04
Dependent care beneﬁts -.02
Supervisor support -. 17"
Cultural norms -.17"
No supervisor -.07
Missing dependent care .05

Model Summary
N 629 625 616 598 598
R .17 .25 .37 .44 .47
R2 .03 .06 .13 .19 .22
Adj 1?.2 .02 .04 .10 .16 .19
Change a“ .03* .03" .07** .06" .09**

 

 

Notes: Figures are standardized regression coefficients.

Two equations were estimated; steps 1-3 were

identical in both. Service and other racial background were the omitted categories for occupation and

race. In equation a, the effects of the total environment

measure were estimated at step 4 and are reported

in eohrmn 4a. In equation b, the effects of each individual environment indicator were examined

separately at step 4 and are reported in column 4b.

108

" signiﬁcant at .05 level " signiﬁcant at .01 level

Table 3: Results of Multiple Regression Analysis — Family-to-Work Conflict

 

 

Predictors Full Sample

Step 1 Controls Steel Steel m2 5.1192111 5.11215:
Managerial/executive .03 .03 .04 .05 .05
Professional .09" .09“ .10" .11" .12“l
Technical -.02 -.02 -.02 -.01 -.01
Sales .02 .02 .02 .04 .04
Administrative .01 .01 .02 .02 .03
Production/operator/repair -.03 -.03 -.04 -.04 -.04
Industry -.02 -.02 -.02 -.01 -.01
White -.04 -.03 -.03 -.02 -.02
Black -.03 -.03 -.04 -.04 -.04

Step 2 Personal

Demands & Resources
Sex .01 .02 .02 .02
Elder/disabled care .06" .06" .06" .06"
#kids <13 .13“ .12" .12“ .12"I
# kids 13-18 .05‘ .05" .05‘ .05"
Partner -.04 -.03 -.03 .03
Family child care -.02 -.01 -.01 -.01
Age -.03 -.05* -.05" -.06"'
Income level .03 .02 .04 -.22”

Step 3 Work

Demands & Resources
Work hours .05“ .05" .04“
Irregular schedule .03 .03 .03
Autonomy -.04* .00 .00
Tenure .05" .05‘ .05‘

Step 4 Equation a

Total Environment
Total environment -.12" -.29"
No supervisor .04" .04“
Missing dependent care -.01 -.01

Step 5 Equation a

Interaction
Environment x income .35"

Model Summary
N 2,816 2,627 2,588 2,464 2,464
R .11 .18 .20 .22 .23
R2 .01 .03 .04 .05 .05
Adj R’ .01 .03 .03 .04 .04
Change R’ .01" .02" .01** .01** .00**

 

 

Notes: Figures are standardized regression coefﬁcients. Two regression equations were estimated. Steps
1-3 were identical in both. Service and other racial background were the omitted categories for
occupation and race. In regression equation a, the effects of the combined total environment measure and
its interaction with income level were estimated at steps 4 and 5. In regression equation b, the effects of
each individual environment indicator were examined separately at step 4. At step 5, the interaction of
each with income was tested separately; when entered individually, two were signiﬁcant, and the results
are reported in columns Sbl and Sb2. When both interactions were entered together, neither beta weight

was signiﬁcant, although the step as a whole was.

109

" signiﬁcant at .05 level ” signiﬁcant at .01 level

Table 3: Results of Multiple Regression Analysis — Family-to-Work Conflict

 

 

(Continued)
Predictors Full Sample (Continued)
Step 1 Control: 5122112 $29.51;; Steam
Managerial/executive .04 .04 .04
Professional .10" .10" .11“
Technical -.02 -.02 ~01
Sales .03 .03 .03
Administrative .03 .03 .03
Production/operator/repair -.02 ~02 ~02
Industry -03 ~02 -.02
White -.01 -.01 -.01
Black -.03 -.03 -.03
Step 2 Personal Demands & Resources
Sex .04 .04 .04
Elder/disabled care .05" .05" .05”
#kids <13 .12" .12" .12"
# kids 13-18 05* 05“ 05’
Partner -03 -03 -04
Family child care -01 -01 -.01
Age -.04 -.04 -05"
Income level .03 -. l9 -. 13
Step 3 Work Demands and Resources
Work hours .03 .03 .03
Irregular schedule .02 .02 .02
Autonomy -01 -01 ~01
Tenure 05“ 05* .05
Step 4 Equation b Environment Indicators
Flex time .01 .01 .01
Flex place .12“ .12" .12"
Dependent care beneﬁts -.02 -.02 -02
Supervisor support -03 -.15” -.04
Cultural norms -.20" -.20" -.30"
No supervisor .04 .04 .04
Missing dependent care -.01 -.Ol -01
Step 5 Equation b Interaction 1
Supervisor support x income 26*
Step 5 Equation b Interaction 2
Cultural norms x income .21“
Model Summary
N 2,464 2,464 2,464
R .30 .3 l .31
It2 .09 .09 .09
Adj 1?.2 .08 .08 .08
Change R2 .05** 00* 00*

 

Notes: Two regression equations were estimated. Steps 1-3 were identical in both. Service and other
racial background were the omitted categories for occupation and race. In regression equation a, the

effects of the combined total environment measure and its interaction with income level were estimated at
steps 4 and 5. In regression equation b, the effects of each individual environment indicator were
examined separately at step 4. At step 5, the interaction of each with income was tested separately; when
entered individually, two were signiﬁcant, and the results are reported in columns Sbl and Sb2. When
both interactions were entered together, neither beta weight was signiﬁcant, although the step as a whole

W88.

" signiﬁcant at .05 level " signiﬁcant at .01 level

110

 

Table 3: Results of Multiple Regression Analysis — Family-to-Work Conflict

 

(Continued)

Predictors Lower Income Sample

Step 1 Controls Step 1 Sign; Step 3 m M
Managerial/executive .00 .00 .02 .03 .03
Professional .01 .01 .02 .05 .04
Technical ~03 ~03 ~02 ~01 .00
Sales ~04 ~03 ~03 ~02 ~02
Administrative ~02 ~01 .00 .01 .02
Production/operator/repair ~05 ~05 ~05 ~05 ~01
Industry ~05 ~06 ~05 ~04 ~04
White .02 .05 .04 .06 .06
Black .02 .02 .00 .02 .03

Step 2 Personal

Demands & Resources
Sex .00 .00 .01 .04
Elder/disabled care 09" 09* .08 .07
# kids <13 ~01 .00 ~01 ~01
# kids 13-18 08* .08 .07 .06
Partner .02 .02 .02 .01
Family child care .07 .07 .08 .08
Age -.07* -.09* ~.10 ~09‘

Step 3 Work

Demands & Resources
Work hours .03 .02 .02
Irregular schedule 09* .08 .07
Autonomy ~09" ~03 ~05
Tenure .lO" .10“ .11"

Step 4 Equation a

Total Environment
Total environment -.15“""
No supervisor .02
Missing dependent care ~01

Step 4 Equation b Environment

Indicators
Flex time .07
Flex place .06
Dependent care ~07
Supervisor support ~07
Cultural norms ~.19"
No supervisor .02
Missing dependent care ~01

Model Summary
N 719 707 696 648 648
R .06 .16 .22 .26 .32
R2 .00 .03 .05 .07 .10
Adj R2 ~01 .00 .02 .03 .06
Change R2 .00 02* .02" 02** .05**

 

 

 

Notes: Figures are standardized regression coefﬁcients. Two regression equations were estimated. Steps
1~3 were identical in both. Service and other racial background were the omitted categories for
occupation and race. In regression equation a, the effects of the combined total environment measure
were estimated at step 4 and are reported in column 4a. In regression equation b, the effects of each
individual environment indicator were examined separately at step 4 and are reported in column 4b.

"‘ significant at .05 level " signiﬁcant at .01 level

111

Table 3: Results of Multiple Regression Analysis - Family-to-Work Conflict

 

 

(Continued)

Predictors Middle Income Sample

Step 1 Control: 491.8 1 ﬁne; $99.3. mu: m
Managerial/executive .04 .05 .03 .04 .03
Professional .15" .16" .15" .16" .14"
Technical .01 .01 .00 .01 ~01
Sales .08 09" .08 08" .07
Administrative .07 .07 .07 .07 .06
Production/operator/repair .03 .03 .02 .01 .01
Industry ~01 -.01 ~01 .00 ~02
White ~02 ~02 ~01 ~02 .01
Black ~01 ~02 ~02 .03 ~03

Step 2 Personal

Demands & Resources
Sex .01 .03 .03 .04
Elder/disabled care 06* 06“ 06* .05
#kids <13 .17" .16" .16” .16”
# kids 13~18 .03 .04 .05 .04
Partner ~06" ~06 ~05 ~05
Family child care -.08‘ ~07 ~07 ~06
Age ~01 ~02 ~02 ~01

Step 3 Work

Demands & Resources
Work hours 07* 06* .05
Irregular schedule ~01 ~01 ~02
Autonomy .00 .04 .03
Tenure .01 .01 .01

Step 4 Equation a

Total Environment
Total environment ~12"
No supervisor .06
Missing dependent care ~03

Step 4 Equation b

Environment Indicators
Flex time ~01
Flex place .12"
Dependent care .00
Supervisor support ~03
Cultural norms ~22"
No supervisor .04
Missing dependent care ~03

Model Summary
N 1,305 1,294 1,274 1,216 1,216
R .13 .20 .21 .24 .33
R2 .02 .04 .04 .06 .11
Adj R’ .01 .03 .03 .04 .09
Change R2 02* .02" .00 06" .06"

 

 

 

Notes: Figures are standardized regression coefficients. Two regression equations were estimated. Steps
1~3 were identical in both. Service and other racial background were the omitted categories for
occupation and race. In regression equation a, the effects of the combined total environment measure
were estimated at step 4 and are reported in column 4a. In regression equation b, the effects of each
individual environment indicator were examined separately at step 4 and are reported in column 4b.

" signiﬁcant at .05 level " signiﬁcant at .01 level

112

Table 3: Results of Multiple Regression Analysis - Family-to-Work Conflict

 

 

(Continued)
Predictors Upper Income Sample
Step 1 Controls am $120.2 m3 m1: 51:11.41:
Managerial/executive .03 .06 .06 .08 .08
Professional .06 .08 .08 .09 .09
Technical ~05 ~05 ~05 ~05 ~04
Sales ~02 ~01 ~01 .00 ~01
Administrative ~08 ~06 ~05 ~04 ~01
Production/operator/repair ~. 12 ~. 10 ~. 12 ~. 1 3 ~. 1 1
Industry .00 ~01 ~01 ~02 ~03
White ~14" ~14" ~.13 ~.13"' ~.11‘
Black -.11* -.13* ~.l3 ~.l3* -.12"'
Step 2 Personal
Demands & Resources
Sex .05 .06 .05 .06
Elder/disabled care .03 .03 .03 .02
# kids <13 .20" .20" .20" .19"
# kids 13~18 .07 08“ .07 .07
Partner ~05 ~05 ~05 ~05
Family child care .05 04 .04 .03
Age ~04 ~08 .08 ~09
Step 3 Work
Demands & Resources
Work hours .04 .05 .02
Irregular schedule .03 .04 .04
Autonomy ~04 ~02 ~02
Tenure .08 .10“ .10‘
Step 4 Equation a
Total Environment
Total environment ~08
No supervisor .02
Missing dependent care .06
Step 4 Equation b
Environment Indicators
Flex time ~05
Flex place .16"
Dependent care ~03
Supervisor support ~01
Cultural norms ~.12"
No supervisor .03
Missing dependent care .06
Model Summary
N 630 626 618 600 600
R .20 .32 .33 .34 .39
1?.2 .04 .10 .11 .12 .15
Adj R2 .03 .08 .03 .08 .11
Chagge 1?.2 .04" .06" .01 .01 .04"

 

 

Notes: Figures are standardized regression coefficients. Two regression equations were estimated. Steps
1~3 were identical in both. Service and other racial background were the omitted categories for
occupation and race. In regression equation a, the effects of the combined total environment measure
were estimated at step 4 and are reported in column 4a. In regression equation b, the effects of each
individual environment indicator were examined separately at step 4 and are reported in column 4b.

"' signiﬁcant at .05 level ** signiﬁcant at .01 level

113

Table 4: Results of Multiple Regression Analysis - Emotional Exhaustion

 

 

Predictors Full Sample
Step 1 Contmls 5.1.92.1 51:11.2 Stan; 5.19m 5129.511 ﬁts-11.512
Managerial/executive .04 .04 .04 .08” .03 .02
Professional .05 .04 .03 07’ .04 .00
Technical .02 .02 .01 .03 .02 .00
Sales .01 .00 .00 .02 .00 ~01
Administrative .01 .00 .00 .02 .01 ~02
Production/operator/repair .05 07* .02 .00 .04 .00
Industry .02 .01 .02 .03 .02 .02
White .01 .02 .02 .03 .01 .03
Black .03 .03 ~01 .00 .02 .02
Step 2 Personal
Demands & Resources
Sex .09“ 11" .11" .11" .06"
Elder/disabled care .01 .01 .00 .00 ~02
# kids <13 .01 ~01 ~02 ~02 ~07"I
# kids 13-18 .01 .01 .00 .00 ~04"
Partner ~ .02 ~01 ~01 ~01 ~02
Family child care .00 .01 .02 .02 .02
Age -.12*"' -.10** ~.11" ~.l2*" ~08"
Income level 05* .03 07" ~08 .03
Step 3 Work
Demands & Resources
Work hours .22“ .22" .22" .09”
Irregular schedule .01 .02 .01 ~03"I
Autonomy ~24“ -.13“ ~14" ~.10"
Tenure .01 .01 .01 .01
Step 4 Equations a & b
Total Environment
Total environment ~28" ~38" -.14"
No supervisor ~03 ~03 ~03
Missing dependent care .00 .01 .00
Step 5 Equation a Interaction
Environment x income .20
Step 5 Equation b
Conflict Variable Mediation Test
Work-to-family conﬂict .47"
Family-to-work conﬂict .14"
Model Summary
N 2,814 2,627 2,587 2,464 2,453 2,453
R .04 .15 .35 .42 .42 .65
R2 .00 .02 .12 .18 .18 .42
Adj R2 -.00 .02 .11 .17 .17 .42
Change it2 .00 .02** .10** 06" .00 .25**

 

Notes: Figures are standardized regression coefﬁcients. Three equations were estimated. Steps 1~3 were
identical in all; step 4 was the same for equations a and b. Service and other racial background were the
omitted categories for occupation and race. In equation a, the effects of the total environment measure
and its interaction with income were estimated at steps 4 and 5 and are reported in columns 4ab and 5a.
In equation b, the total environment measure was also entered at step 4. Possible mediation of the total
environment effect by the work/family conﬂict variables was tested by entering them at step 5; results are
reported in column 5b. Regression equation c is reported and explained on the page that follows. "'
signiﬁcant at .05 level " signiﬁcant at .01 level

114

 

Table 4: Results of Multiple Regression Analysis - Emotional Exhaustion

 

 

 

Predictors Full Sample
Step 1 Controls Steal 51211.2 Stan}: mus Strait
1 Managerial/executive .04 .04 .04 .06 .00

Professional .05 .04 .03 .05 ~01
Technical .02 .02 .01 .02 ~01
Sales .01 .00 .00 .01 ~01
Administrative .01 .00 .00 .02 ~02

Production/operator/repair .05 .07 .02 .02 .01
Industry 02- .01 02- .02 02-
White .01 .02 .02 .04 .03
Black .03 .03 -.01 .00 .02

Step 2 Personal

Demands & Resources
Sex .09" .11" .12 .07"
Elder/disabled care .01 .01 ~01 ~02 ,
# kids <13 .01 ~01 ~01 ~07"
# kids 13-18 .01 .01 .00 ~.03*
Partner ~ .02 .01 .00 ~ 0.2
Family child care 00.01.01 02
Age ~.12" ~ .10** ~ .10" ~ 07"
Income level 05" .03 05* .02

i Step 3 Work Demands & Resources

Work hours .22" .20" 09"
Irregular schedule 01 .00 ~04"
Autonomy - .24" -.13" ~.10"
Tenure .01 .00 .01

Step 4 Equation c

Environment Indicators
Flex time ~05“ ~03
Flex place .05" .02
Dependent care ~05" ~01
Supervisor support ~.18‘”'I ~.13"
Cultural norms - .18" ~06"
No supervisor ~ .08 ~03
Missing dependent care .01 .01

Step 5 Equation c

Conflict Variable Mediation Test
Work-to-family conﬂict .46"
Family-to-work conﬂict .13"

Model Summary
N 2,814 2,627 2,587 2,464 2,453
R .05 .15 .35 .46 .66
R2 .00 .02 .12 .21 .44
Adj R2 .00 .02 .11 .21 .43
CMe It2 .00** .02" .10** .09** .22**

 

Notes. Figures are standardized regression coefficients. Three regression equations were estimated.
Steps 1~3 ”were identical in all but are repeated here for ease of comparison. Service and other racial
background were the omitted categories for occupation and race. Regression equations a and b are
reported and explained on the preceding page. In equation c, the effects of each individual environment
indicator were examined separately at step 4 and are reported in column 40. Possible mediation of the
environment indicator effects by the work/family conﬂict variables was tested by entering them at step 5,
with results reported in column 5c. Potential interactions of each indicator with income were tested but
none were signiﬁcant, so they are not reported here. "' signiﬁcant at .05 level ** signiﬁcant at .01 ievel

115

Sample Characteristics and Descriptive Statistics for Major Variables
Demographics

While there are some small differences, comparisons of the current sample and
government statistics for US workers reveal that the workers who participated in the
study are fairly representative of the US workforce.

The demographics of the sample used largely reﬂect those of the US. as a whole;
79.2 percent were Caucasians, 11.4 percent African Americans, and 8.5 percent other
races (including mixed). Minority representation in the sample is slightly high, as an
estimated 82.7 percent of the US population is white, 12.7 percent African American and
4.5 percent falls into other categories (US Bureau of the Census, 1998). In terms of
occupation, 15.7 percent of the sample were classiﬁed as executives, administrators, or
managers and 18.0 percent as professionals. Sales was the occupation for 10.4 percent of
participants, technical work for 5.0 percent, administrative support for 15.4 percent, and
service for 10.7 percent. Some 23.9 percent were classiﬁed as involved with production,
operating equipment, or conducting repairs. Government estimates for the population of
US employees are similar; an estimated 15.3 percent of US workers are in managerial,
executive, and administrative positions; 16.1 percent professional specialties, 3.6 percent
in technical and related support positions, 10.0 percent in sales, 14.5 in administrative '
support jobs, 11.1 percent in service jobs, and 27.8 percent employed in production and
repair work (US Census Bureau, 2000).

Just under a quarter of the workers in the sample were classiﬁed as working in a
goods-producing ﬁrm, while 76.6 percent worked in the service industry, again similar to

Census estimates for the US as a whole; government ﬁgures for 1997 show 20.4 percent

116

 

of US employees in goods-producing industries and 79.7 percent in the service sector
(US Census Bureau, 2000).
Personal Demands and Resources

As can be seen in Table 1 on the previous page, females made up 51.6 percent of
the current sample and males 48.4 percent, similar to Census estimates of 51.1 and 48.9
percent for that year (US Census Bureau, 2000). About 85% of participants lived with
family members. A little over two-thirds of were married or living with a partner, and
nearly half had at least one child living at home at the time of the survey. In just under a
quarter of households represented, most or all care required for children was provided by
a partner or other family member. Of the respondents, some 456 or 15.8 percent were
currently providing assistance to an elderly or disabled adult. Time spent ranged from 15
minutes to 99 hours per week; the mean for the entire sample was 1.75 hours per week,
and the mean for the affected group was 11.06 hours.

Because the study was limited to working adults, the mean age was slightly
higher, at 40.2, than the mean for the US, 36.1. Just over half of the respondents belonged
to the baby boom generation (born 1933-1951), 15% were from the pre-war generation
and approximately one-third younger workers were part of “Generation X”.

In terms of household income, the study group was somewhat better off than US
households in general, probably because only working adults were surveyed, which
meant that households with no working adults were not included. The median income for
the all US households in 1996 was $35,492; the mean was slightly higher, at $47,123.
For the current sample, the median and mean were $42,000 and $54,517, respectively.

The sample is large and quite representative, with response rates that compare

117

quite favorably for social research. However, as noted earlier, because of its reliance on
telephone interviews the study may have somewhat lower response rates for workers at
the lowest and highest ends of the ﬁnancial spectrum than those in the middle. The
survey involved phoning participants, ﬁnding them at home and persuading them to
participate. The poorest Americans, whether working or not, are less likely to have
telephones, which represent an additional expense. The richest Americans are less likely
to answer the telephone themselves or to come to the telephone and spend the time
required for an interview if informed of the request by an employee.
Work Demands and Resources

The respondents worked an average of 45.97 hours per week. Reported weekly
work hours ranged from 4 to 144, with a median of 45 hours and a mode of 40. This is
slightly higher than the US government’s estimate of mean work week for non-
agricultural US employees in 1997, 39.4 hours (U S Census Bureau, 2000). Nearly three-
quarters of respondents reported having a regular daytime schedule. Average tenure at
their cunent organization was 7.54 years, but there was considerable variance in this;
estimates ranged from less than a year to 46 years. Those who had been with their
current employer less than one year made up 14.4 percent of the sample, while
approximately 10 percent had been employed for 20 years or more. The median reported
tenure was 4 years. This is again similar to ﬁgures reported by the US government; the
median reported for US employees was 4.0 years, with 24.8 percent of workers 20 and
over employed at their current organization for less than one year and 9.5 percent

employed for over 20 years (US Census Bureau, 2000).

118

Overall, the workers in the study reported having a fairly high degree of
autonomy; the sample mean was 3.03 on a four-point scale.
Organizational Work/Family Environment

The mean on the composite scale measuring the degree of ﬂexibility in work time
was 1.55 on a scale of 0-4, representing a low to moderate degree of ﬂexibility. This
reﬂects the fact that while 43.8 percent reported having ﬂexibility in their starting and
quitting times, only 24.7 percent were able to make changes in when they started and
ﬁnished work on a daily basis. Respondents also reported only a moderate degree of
control over the scheduling of their work hours, with a mean of 1.9 out of a possible four
points. Just over a quarter of the sample reported that they currently worked some of their
regular hours from home or would likely be able to do so.

Respondents reported having access to a mean of 3.22 out of 8 potential
dependent care beneﬁts. Of those who provided information, 94 percent reported that
their organizations offered maternity leave and 79.8 percent said that paternity leave was
available. Information and referral services for child and elder care were offered by the
employers of 24.9 percent and 19.7 percent of respondents, respectively. On- or near-site
daycare was offered by the employers of 11.3 percent of the workers sampled, and 13.6
percent reported that their employers contributed a portion of child care expenses for
qualifying workers. Some 29.2 percent of respondents said that dependent care Spending
accounts were available at their workplaces, and 49.1 percent of working parents said that

they were able to take time off to care for a sick child without losing pay or vacation

days.

119

Overall, supervisors were seen as quite supportive of workers’ efforts to balance
their jobs and their families; the mean on the supervisor support scale was 3.34 on a scale
with 4.0 as the highest possible value. Organizational norms, while still fairly supportive,
were not quite as understanding. The mean on the cultural norms scale was 2.96 out of a
possible 4, indicating that, in general, workers considered their organizations to be
supportive and believed they did not penalize the careers of workers who used ﬂexibility
beneﬁts or put their families ﬁrst on occasion.

Reﬂecting these results, the summary index of the overall environment which was
equally comprised of ﬂexibility, beneﬁt availability, supervisor support, and cultural
norms had a mean of 9.33 for the sample out of a possible 16.

Results of Regression Analysis and Evaluation of Hypotheses
Control Variables

Although they were not central to the interest of this paper, individual results are
reported for the three sets of dummy variables created to represent the controls: race,
industry, and occupation. As discussed previously, the “other racial background” group
was selected as the omitted variable for race, and “service” was the omitted category
form occupation; beta weights thus reﬂect the difference between the groups presented
and these contrast groups. As can be seen in Tables 2 and 3, together these three sets of
categorical control variables explained one percent of the variance in work-to-family
conﬂict (adjusted R2 .00) and family-to-work (adjusted R2 .01) conﬂict, and even less of
the variance in emotional exhaustion, as reported in Table 4. The minimal amount of
variance explained by these variables supports the notion that work/family conﬂict and

burnout can be found across all types of jobs and among workers of all races.

120

Personal Demands and Resources

Personal demands and resources, entered into the regression equations at step 2,
together explained an additional 3 percent of work-to—family conﬂict, as can be seen by
examining the change reported in R2 in Table 2. An additional 2 percent of the variance
in family-to-work conﬂict was explained by these factors, as is illustrated in Table 3, and,
as can be seen from Table 4, this step also added 2 percent to the variance explained in
emotional exhaustion.
Gender

H1 predicted that gender would be associated with all three of the dependent
variables, with women more likely to experience higher levels of conﬂict and exhaustion.
This was generally supported for work-to-family conﬂict and emotional exhaustion, but
not family-to-work conﬂict. Although the Pearson correlation between gender and work-
to-family conﬂict was not signiﬁcant at the .05 level (see Table 1), the relationship was
signiﬁcant when occupation, industry and race were controlled. As can be seen from
Table 2, being female was a statistically signiﬁcant predictor in the regression equations
for work-to-family conﬂict at all steps, except in the low-income group. Being female
was also associated with experiencing a higher degree of emotional exhaustion, with a
Pearson correlation of .07 (signiﬁcant at the .00 level), and a beta weight that remained
statistically signiﬁcant at the .05 level or better in the regression equation for the full
sample (see Table 4). It appears that experiencing conﬂict between domains is more
important than gender. Contrary to expectations, the weakest results were found for
family-to-work conﬂict. The Pearson correlation between the two was not signiﬁcant,

and gender was not statistically signiﬁcant in the regression results reported in Table 3.

121

Time caring for elderly and disabled adults

H2 predicted that more hours spent caring for elders and disabled adults would be
associated with more work-to-family conﬂict, family-to-work conﬂict, and emotional
exhaustion. However, only in the case of family-to-work conﬂict was this hypothesis
supported. The Pearson correlation between the two (.04; see Table 1) was small but
statistically signiﬁcant at the .05 level, and as a predictor in the regression equation for
the full sample reported in Table 3 it remained signiﬁcant at the .00 level through all
steps. The beta coefficient for time on such care was not signiﬁcant in the regression
equation for workers from hi gher-income households, perhaps reﬂecting their ability to
purchase care. It was also not signiﬁcant in the regression equations for the lower and
middle-income workers once the work environment indicators were included.

As can be seen from Tables 2 and 5, spending more time on caring for elders and
adults was not associated with experiencing more work-to-farnily conﬂict or emotional
exhaustion, perhaps because family variables tend to be more closely associated with
family-to-work conﬂict (F rone, Yardley and Markel, 1997). Also, while these demands
no doubt have a strong impact on the lives of those who are experiencing them, only
about 15 percent of the sample was affected, and most of them spent less than 10 hours a
week on care.

Children

H3 dealt with the effects of children. Having more and younger children was
expected to be associated with more negative outcomes; however, this was supported
only for work-to-family and family—to-work conﬂict. As can be seen from the correlation

coefficients in Table 1, having more children under 13 had a moderately strong

122

correlation with both more work-to-farnily (.13, si g. at .00 level) and more family-to-
work (.11, sig. .00) conﬂict. Having younger children was also weakly associated with
greater emotional exhaustion (r=.04, sig. 05). For the number of children 13-18, who
would not require constant supervision but still represent a source of demands, the
correlations were .06 (sig. .01) for work-to-farnily conﬂict, .04 (sig. .05) for family-to-
work conﬂict, and .00 for emotional exhaustion. The patterns in the regression equations
were similar. As can be seen in Tables 2 and 3, having more children, especially young
children, contributed to explained variance in both work-to-family and family-to—work
conﬂict in the regression equations for the full sample, although the beta weights were
not consistently statistically signiﬁcant in the lower and higher income groups, which had
smaller samples. In the regression equation for emotional exhaustion reported in Table 4,
the beta coefﬁcients for having children were generally small and insigniﬁcant, at least
until the work/family conﬂict variables were entered at step ﬁve in the equations used to
examine potential mediating effects. In those equations, having children, especially
young children, was associated with a small but signiﬁcant decrease in emotional
exhaustion. Beta coefﬁcients were -.07 (sig. .01) for the number of 0-12-year-olds and
0.04 (sig. .05) for the number of teenagers in the regression equation examining
mediation of the effects of the total environment summary variable. Results in the
equation that included the ﬁve environment indicators separately as well as the two
conﬂict variables were similar. Children, then, may contribute to reduced emotional

exhaustion — once the conﬂict associated with them is controlled for.

123

Partner

H4 predicted that living with a partner, whether married or not, would be
associated with less work-to-family conﬂict, family-to-work conﬂict and emotional
exhaustion. Evidence of a relationship was only found in the case of work-to-family
conﬂict. In terms of the Pearson correlations reported in Table 1, the only statistically
signiﬁcant relationship found was for work-to-family conﬂict (.05, sig. at .01 level) and
not in the predicted direction —— the correlation indicates that having a partner was actually
associated with more work-to-family conﬂict. In the regression equations for work-to-
family conﬂict reported in Table 2, the standardized beta coefﬁcient for having a partner
was also positive and signiﬁcant when job characteristics were included in the regression
equation for the full sample (-.05, sig. at .05). However, the beta coefﬁcient for having a
partner was not signiﬁcant in the separate samples for each income level. This is at least
partially due to sample size, as patterns similar to the one found for the full sample could
be seen for the lower and middle income groups. For workers in these groups, once job
characteristics were controlled for, the beta weights for having a partner increased in size;
however, they were not signiﬁcant in due smaller sample sizes. Having a partner
appeared to have little impact on upper-income workers, perhaps because they could pay
for alternative assistance. Having a partner was generally not a signiﬁcant predictor in the
equations for family-to-work conﬂict and emotional exhaustion.
Family member provides child care

Being able to rely on a family member to handle most needed child care for the

youngest child (presumed to be an indicator for care arrangements for all children) was

124

predicted to be associated with less conﬂict and emotional exhaustion in H5. This
hypothesis was not supported.

About one-ﬁfth of the sample had such child care arrangements, and family care
was correlated with both work-to-family and family-to-work conﬂict, but not as
expected. As can be seen in Table l, the positive correlations of .08 and .06 (both sig. at
.00 level) indicate that having someone in the family care for the children was actually
associated with more conﬂict. This may reﬂect the fact that the person involved is oﬁen
the spouse and/or represents a strong set of demands of his or her own. In the regression
equations reported in Tables 2, 3, and 4, where having a partner was included at the same
step, the standardized beta coefﬁcients were not signiﬁcant, lending some support to this
interpretation.

Household Income Level

H6 predicted that a higher household income would serve as a resource, reducing
conﬂict and emotional exhaustion, particularly when industry and occupation were
controlled for. There was little support for this hypothesis.

As expected, a higher income (often associated with a more demanding job) was
actually positively correlated with both work-to-family (.06) and family-to-work (.05)
conﬂict (both si g. at .00 level) when the Pearson correlations reported in Table l were
examined. However, contrary to prediction, the beta coefﬁcients for household income
level in the regression equations for work-to-family conﬂict and emotional exhaustion
reported in Tables 2 and 4 were signiﬁcant and positive even when occupation and
industry were controlled for. Even when the job characteristics considered as work

demands and resources were included at step 3, the beta coefﬁcients for income level

125

were positive and statistically signiﬁcant at the .01 level for work-to-family conﬂict and
emotional exhaustion. It seems likely the variables used to measure job demands did not
fully capture the differences. When the interaction effects were included at step 5, in all
three full-sample regression equations the beta coefﬁcients for income level became
larger, highly signiﬁcant, and negative. With the interactions between income and
organizational environment resources partialed out, a higher income level was associated
with less conﬂict and emotional exhaustion, but in general there was little support for
H6, likely because the other variables included in the equation did not adequately capture
and control for job characteristics.

Age

H7, the ﬁnal hypothesis regarding personal demands and resources, predicted that
older workers would experience less conﬂict and burnout, at least partly because they
would have developed better coping skills through life experience. This hypothesis was
generally supported by the data.

As can be seen in Table 1, the Pearson correlations were all negative and
signiﬁcant at the .00 level, (-. 10 for work-to-family conﬂict, -.06 for family-to-work
conﬂict, -. 12 for emotional exhaustion). The beta coefficients for age were consistently
signiﬁcant in the three regression equations for the full sample reported in Tables 2, 3,
and 4. Similar patterns were observed in the separate equations by income group,
although the effects were not always signiﬁcant, in many cases due at least partly to the
smaller sample size. The standardized beta weights for age were not signiﬁcant for
middle- and upper-income workers in the work-to-family equations once the work

demands and resources and environment variables were included in the regression (see

126

Table 2). In the family-to—work equations reported in Table 3, the coefﬁcients for age
were also generally not signiﬁcant for the middle- and upper-income workers. In the
regression equations for emotional exhaustion reported in Table 4, the beta coefﬁcients
for age were strong and highly signiﬁcant for the upper and lower income groups, but not
for middle income earners.

Work Demands and Resources

The four variables selected to represent work demands and resources and entered
at the third step of the regression equations explained an additional 9 percent of the
variance in work-to-family conﬂict, as illustrated in Table 2. The change in R2 associated
with this step in the regression equation for family-to-work conﬂict was only 1.0 percent,
as is shown in Table 3. Work demands and resources added the most to the explanation
of emotional exhaustion; as can be seen in Table 4 R2 increased by 10 percent.

H8 dealt with work demands, predicting that longer hours and an irregular
schedule would be associated with more negative outcomes. This was generally
supported for work hours but not for having an irregular schedule. H9 predicted that
having greater autonomy and longer tenure or experience in the same line of work would
be associated with lower levels of conﬂict and emotional exhaustion. This was supported
for autonomy in the case of work-to-family conﬂict and emotional exhaustion, and for
tenure in the case of family-to-work conﬂict.

Work hours

As can be seen in Table l, the Pearson correlations for total work hours were

strong and signiﬁcant at the .00 level: .24 for work-to-family conﬂict, .07 for family-to-

work conﬂict, and .19 for emotional exhaustion. In the full-sample and income group

127

regression equations for work-to-family conﬂict reported in Table 2, hours worked was
the strongest predictor, except for the summary measure of how supportive the work
environment is. Work hours was also a powerful predictor in the regression equations for
emotional exhaustion, as can be seen in Table 4. While the beta coefﬁcient for work
hours was also signiﬁcant at the .05 level in the full-sample regression equation for
family-to-work conﬂict (Table 3), it did not have as much impact and was not always
signiﬁcant in the equations for the three income groups, congruent with previous ﬁndings
that work stressors have their strongest impact on work-to-family conﬂict (e. g. Frone,
Yardley and Markel, 1997).
Irregular schedule

As can be seen from the correlation matrix reported in Table 1, having an
irregular schedule was associated only with work-to-family conﬂict, (.07, sig. at the .01
level), and its standardized beta in the full-sample regression equation for that variable
reported in Table 2 was also signiﬁcant at the .01 level, although it was not as powerful a
predictor as hours worked. Similar patterns were observed in the separate regression
equations for the lower and middle income groups, but not for the high income group,
which is less likely to have such arrangements. The beta coefﬁcients for having an
irregular schedule were not signiﬁcant in the equations for farnily-to-work conﬂict and
emotional exhaustion reported in Tables 3 and 4.
Autonomy

Examining the Pearson correlations reported in Table l, autonomy had a
signiﬁcant negative relationship with work-to-family conﬂict (-. 14, sig. at .01 level) as

well as emotional exhaustion (-.23, sig. at .01). No relationship was found for family-to-

128

work conﬂict. In the regression equations for work-to-family conﬂict (Table 2) and
emotional exhaustion (Table 4), the beta coefﬁcients for autonomy were generally strong
and signiﬁcant, although they shrank when ﬂexibility was added at step 4 in the separate
equations for each income group, implying that some of the beneﬁts of autonomy are
associated with the ﬂexibility it makes possible.
Tenure

As can be seen in Table 1, tenure was correlated only with emotional exhaustion
(-.O4, si g. at .05). In the regression equations, it did not have a statistically signiﬁcant
impact on work-to-family conﬂict, as can be seen in Table 2. Tenure added the most to
explaining family-to-work conﬂict, and was a signiﬁcant predictor in the equations
shown in Table 3 for the full sample and lower income groups, but still contributed very
little to variance explained. As can be seen in Table 4, the beta coefﬁcient for tenure was
small and insigniﬁcant in the regression equations predicting emotional exhaustion.
Overall, then, experience at the employing organization contributed little to
understanding variance in the three outcomes.
Fondly-Supportive Organizational Environment

Hypothesis 10 predicted that a family-supportive work environment would be
linked to lower levels of conﬂict and emotional exhaustion, and this was supported by the
results.

In the full sample, the effects for the summary environmental variable were quite
strong. The Pearson correlations, reported in Table l, were negative, and all were
signiﬁcant at the .01 level. The relationship between a supportive environment and

work-to-farnily conﬂict was -.23, with a correlation of -.08 for family-to—work conﬂict,

129

and -.29 for emotional exhaustion. In the full-sample regression equations, this step
made the second largest contribution to explained variance in family-to-work conﬂict,
just behind personal demands and resources, as can be seen in Table 3 (step 4a). It was
also just slightly behind work demands and resources in explaining work-to-family
conﬂict (see Table 2 step 4a) and emotional exhaustion (see Table 4 step 4a). A
supportive environment added 5 percent to variance explained in work-to—family conﬂict,
1 percent in the case of family-to-work conﬂict, and 6 percent for emotional exhaustion.

The effects of the individual environment indicators were also fairly strong in the
full sample. Looking at the simple correlations, ﬂexibility in work time was associated
with lower levels of work-to-family conﬂict (-.09, sig. at .01) and emotional exhaustion
(-.14, si g. at .01), but not family-to-work conﬂict. A similar pattern emerged for access
to dependent care beneﬁts, which lmd a -. 10 (sig.at .01) correlation with work-to—family
conﬂict and a -.04 (sig. at .05) correlation with emotional exhaustion. Supervisors who
were supportive when family issues arise and supportive cultural norms were all
signiﬁcantly and negatively related to all three outcome variables; correlations ranged
from -.1 l for supervisor support and family-to-work conﬂict to -.31 for supportive
cultural norms and emotional exhaustion (see table 1). However, the results for
ﬂexibility in work location were quite different from what was expected. The ability to
work from home was actually associated with more work-to-family conﬂict (r=.05, sig. at
.05) and even greater levels of family-to-work conﬂict (r=.12, sig. at .01). There was no
relationship between ﬂexibility in work location and emotional exhaustion.

The patterns found in the regression equations for the full sample were similar;

after controlling for demographics, personal demands and resources, and work demands

130

and resources, ﬂexibility in work time, greater access to dependent care beneﬁts, family-
supportive supervisors and supportive cultural norms were all associated with less work-
to-family conﬂict and emotional exhaustion, with the strongest effects found for cultural
norms that don’t penalize workers for putting family before their jobs when necessary
(see results reported in Table 2 step 4b and Table 4, step 40). In the case of family-to-
work conﬂict, only supportive cultural norms had a statistically signiﬁcant beta weight.

Overall, when the supportive environment indicators were added separately at the
fourth step of the regression equations, they explained an additional 8 percent of the
variance in work-to-family conﬂict, an additional 5 percent of the variance in family-to-
work conﬂict, and 9 percent of the variance in emotional exhaustion. The change in R2,
then, was greater when the ﬁve indicators were entered separately than when the total
environment summary variable was used.
Interaction Eﬂects

H11 suggested that a supportive work environment would have the most positive
impact for those with lower ﬁnancial resources, namely those from lower-income
households, while H12 predicted that and single parents would beneﬁt more from a
supportive environment than others. There was a fair amount of support for H11, which
dealt with income, but not for H12.
Income level and work environment

The environment x income level interactions were signiﬁcant and the change in
variance explained by this step, while very small, was statistically signiﬁcant at a .01
level in the full-sample work-to-family conﬂict regression equation reported in Table 2

(see equation a), as well as the family-to-work conﬂict regression equation reported in

131

Table 3 (equation a). In the case of emotional exhaustion, the interaction effect was
signiﬁcant at the .06 level, just missing the standard cutoﬁ“ (see Table 5 equation a).
Thus, some support was found for the idea that a family-supportive environment has the
most positive effects for those who need it most due to lower resources, although this was
less supported in the case of burnout. In separate regression equations for each income
level (see Tables 2 and 3, equation a), the standardized beta coefﬁcients for the summary
total environment variable were consistently larger for the lower-income group than for
higher and middle income workers.

The interactions between income level and the ﬁve supportive environment
indicators were also examined in separate equations for each dependent variable. Those
which were signiﬁcant at the .05 level or better are reported in the results tables. Only
the interaction between income level and dependent care beneﬁt availability bad a
signiﬁcant standardized beta coefﬁcient in the work-to-family conﬂict regressions (see
Table 2 equation b). In the case of farnily-to-work conﬂict, the interactions between
income level and both supervisor support and cultural norms were signiﬁcant when
entered separately at step 5, and those results are reported in Table 3 (equations b1 and
b2). When both interactions were entered together at the ﬁfth step, neither interaction
was signiﬁcant, but the step as a whole did add to variance explained at a statistically
signiﬁcant level. Similar to the ﬁndings for the overall enviromnent, the strongest
interaction effect observed for emotional exhaustion, the interaction between income
level and supervisor support, was signiﬁcant at the .06 level, just missing the standard .05

cutoff. Therefore, it is not presented in the results.

132

In the case of work-to-family and family-to-work conﬂict, differences between
workers at the three income levels are examined more closely in additional regression
equations included in Tables 2 and 3. Because none of the interactions were signiﬁcant
in the case of emotional exhaustion, separate tables for each income group are not
presented The signiﬁcant interactions have been graphically illustrated by plotting the
regression lines for each income group; these are shown in Figures 3-7, included in
Appendix B. In each case, the effects of organizational support are most dramatic for
lower income workers.

As can be seen from these ﬁgures and from the low additional amount of variance
explained by the interactions (which ranged from 0 to 1 percent), the interactions,
although statistically signiﬁcant, are not particularly dramatic. Part of this may have to
do with the fact that the middle income group is the most strongly represented here,
partially because middle class workers are more numerous and partly because, as noted
earlier, they are easier to contact by telephone than either very low- or hi gh-income
employees. Another potential reason the results are not so strong may be the cutoff
points chosen for each income group. While they were logically selected using a reliable
external data source and appear reasonable, it may well be that the classiﬁcations are not
optimal. The higher income group in particular may be inappropriate; while workers
with an income over $68,000 likely can easily afford quality day care as well as
bimonthly lawn and cleaning help, likely an income of $100,000 or more is required to
truly be capable of hiring full-time help to handle most tasks on the home ﬁ'ont. Full-
time help tlmt almost completely relieves workers of such demands would have the

strongest effects. The ﬁgures in Appendix B provide some support for this interpretation,

133

as the regression lines for the middle and upper income group are quite similar to each
other, with more of a difference being seen for the lower income group in most cases.
Single parent status and work environment

The environment x single parent interactions were not signiﬁcant when entered
into the regression equations for work-to-family conﬂict, farnily-to-work conﬂict, and
emotional exhaustion. Interactions involving the ﬁve environment indicators were also
not signiﬁcant. They were dropped from the model and are not presented in the
accompanying tables.
Mediating Eﬂects for Full Sample

Hypothesis 13 predicted that work-to—family and family-to-work conﬂict would
mediate the effects of a family-supportive work environment on emotional exlmustion.
That is, a supportive environment would reduce emotional exhaustion by decreasing the
amount of work/family conﬂict experienced. This hypothesis was examined in two V
regression equations relying on the full sample, one that looked the effects of the total
environment composite variable and one that included all of the ﬁve environment
indicators. Both provided evidence of Mal mediation. In ﬁrst regression equation, the
beta coefﬁcient for the total environment summary variable was strong and highly
signiﬁcant at -.28 when entered at step 4 (see Table 4 equation a). When work-to-family
and family-to-work conﬂict were entered into the equation at step 5, the total
environment variable coefﬁcient remained signiﬁcant at the .01 level but shrank
considerably to -.14. In the second equation (see Table 4 equation c), All of the
indicators were negative and statistically signiﬁcant when entered at step 4, except for

ﬂexibility in work location, which was positive and signiﬁcant. Beta coefﬁcients ranged

134

in size from -.05 (ﬂexibility in work time) to -. 18 (supervisor support and cultural
norms). After the work/family conﬂict variables were entered at step 5, only the
supervisor support and cultural norms coefﬁcients remained signiﬁcant, and these had
shrunk to -. l3 and -.06, respectively, showing indications of partial mediation. Greater
evidence of mediation was seen in the case of the ﬂexibility variables and the number of
dependent care beneﬁts available, which were no longer signiﬁcant once work/family
conﬂict was included in the equation. The work/family conﬂict step added over 20
percent more to variance explained in emotional exhaustion, raising the adjusted R2 to .43
where the indicators were all entered separately. The beta coefﬁcient for work-to-family
conﬂict, at .46 in that equation, was nearly three times the .13 estimated for family-to-
work conﬂict, indicating that work-to-family conﬂict was is more closely related to
burnout.
Overall Variance Explained

In the full-sample regression analyses, the variables included, controls, personal
demands and resources, work demands and resources, supportiveness of the work
environment, and the interaction between the work environment indicators and income
level, explained 18 percent of the variance in work-to-family conﬂict (adjusted R square
= .20), as shown in Table 2 (equation b). As can be seen in Table 3 (equation b), they
explained just 9 percent of the variance in family-to-work conﬂict (adjusted R square =
.08), but 21 percent of the variance in emotional exhaustion (adjusted R square = .21), as
reported in Table 4 (equation c). The best results for emotional exhaustion were achieved

by including the work/family conﬂict variables in the regression equation that examined

135

the effects of the environment indicators separately (see Table 4 equation 0). This model
explained 44 percent of the variance in emotional exhaustion (adjusted R square = .43).

The model was much stronger, apparently, in predicting work-to-family conﬂict
and emotional exhaustion, perhaps because both of these clearly have their roots in the '
work domain. The present model appears to have done a fairly good job of capturing
work demands and resources. However, the amount of variance explained for family-to-
work conﬂict was much lower. Since this type of conﬂict has its roots in the nonwork
realm, presumably important personal demands were not well measured or may not have
been included. Measures that include how supportive a spouse is or how many hours are
devoted to home and child care tasks might have contributed to stronger predictions.
Another reason the amount of variance explained for farnily-to-work conﬂict is lower
than in previous research is the reliance on straightforward indicators of demands and
resources rather than scales that involve an attitudinal component. Studies that have
included measures of inrole conﬂict (e. g. conﬂict within the family) or ambiguity of
family roles have been able to explain more variance in this outcome.

A summary of the hypotheses included in the study and ﬁndings related to them is

included in a table that begins on the next page.

136

Table 5: Summary of Hypotheses and Results

 

 

H1

H4

H5

H6

H7

H8

Being female will be associated with higher
levels of work-to-family conﬂict, family-to-work
conﬂict, and emotional exhaustion.

Spending more time caring for elders and
disabled adults will be associated with higher
levels of work-to—family conﬂict, family-to-work
conﬂict, and emotional exhaustion.

Having more children, particularly young
children, will be associated with higher levels of
work-to-family conﬂict and family-to-work
conﬂict and emotional exhaustion.

Living with a partner will be associated with
lower levels of work-to-family conﬂict, family-
to-work conﬂict and emotional exhaustion.

Having most or all needed child care provided by
a family member will be associated with lower
levels of work-to-family conﬂict, family-to-work
conﬂict, and emotional exhaustion.

When occupation and industry are controlled, a
higher household income level will be associated
with lower levels of work-to-family conﬂict,
family-to-work conﬂict, and emotional
exhaustion.

Being older will be associated with lower levels
of work-to-family conﬂict, family-to-work
conﬂict, and emotional exhaustion.

Greater work demands (longer hours and an
irregular schedule) will be associated with higher
levels of work-to-family conﬂict, family-to—work
conﬂict, and emotional exhaustion.

Having more job-related resources (greater
autonomy and more experience working at one’s
current organization) will be associated with
lower levels of work-to-family conﬂict, family-
to-work conﬂict, and emotional exhaustion.

Supported
f r WF

Yes

No

Yes

No

No

No

Yes

Yes

Autonomy
yes, tenure
no

Supported
ELM

No

Yes

Yes

No

No

No

Yes

Hours yes,
schedule
no

Tenure
yes,
autonomy
no

Supported
ME

Yes

No

No

No

No

No

Yes

Hours yes,
schedule
no

Autonomy
yes, tenure
no

 

137

 

Table 5 (continued)

 

 

HlO

H11

H12

H13

Supported Supported
fgr EFQ for 2w;
Working in a more family-supportive
environment (one with greater ﬂexibility in the
timing and location of work, more access to Yes Yes
dependent care assistance, supportive supervisors,
and cultural norms that don’t penalize workers
for putting family ﬁrst) will be associated with
lower levels of work-to-family conﬂict, family-
to-work conﬂict, and emotional exhaustion.

The positive effects of a family-supportive

environment (more dependent care beneﬁts,

greater ﬂexibility in work timing and location, Yes Yes
greater supervisor support, and more family-

supportive cultural norms) on work-to-family

conﬂict, family-to-work conﬂict, and emotional

exhaustion will be strongest for employees from

low-income households and weakest for

employees from high-income households.

The positive effects of a family-supportive work

environment (one with greater flexibility in the

timing and location of work, more access to No No
dependent care assistance, supportive supervisors,

and cultural norms that don’t penalize workers

for putting family ﬁrst) on work-to-family

conﬂict, family-to-work conﬂict, and emotional

exhaustion will be stronger for single parents.

Work-to-family conﬂict and family-to—work
conﬂict will mediate the effects of a family-
supportive environment (one with greater
ﬂexibility in the timing and location of work,
more access to dependent care assistance,
supportive supervisors, and cultural norms that
don’t penalize workers for putting family ﬁrst) on
emotional exhaustion.

Supported
for EE

Yes

No

Partially

 

138

 

Why Income Matters

The ﬁnding that workers at lower income levels may beneﬁt more from working
in a family-supportive environment is one of the major contributions of this study, and
further analyses were conducted to examine why income matters. Not much is known
about how the effectiveness of different ways employers can support their workers in
balancing their roles in the two domains may differ based on income level. Yet there are
strong practical implications for employers; this analysis implies that stress will be more
greatly reduced by efforts to create a family-supportive work environment among lower-
income workers. It is important to understand why and in what ways these workers differ
from colleagues with greater disposable income. The separate regression analyses and
correlations reported here for workers in three different income groups provide some
information.
How Workers Diﬂer

As can be seen by the correlations reported in Table 1 and the means reported in
Tables B2-4, found in appendix B, lower income workers are somewhat more likely to be
younger and female, to spend more time caring for elders and disabled adults, and to have
teenage children They are considerably more likely (r=.37) to be single. Lower income
workers are more likely to work under irregular schedules, have less autonomy, and
shorter tenure, but they work fewer hours. Higher incomes are associated with more
workplace support, including ﬂexibility in work time and location, dependent care
beneﬁts, and cultural norms, although less difference was found in the degree to which
supervisors were supportive about family issues. Workers at higher income levels tend to

experience more conﬂict between work and family, presumably because they tend to

139

have more demanding jobs. There is not much difference in the level of emotional
exhaustion experienced, but the mean for the middle income group is the highest,
followed by the upper and lower income groups.

Differences in Environment Indicator and Outcome Relationships

Flexibility in work time and location was considerably less common among
lower-income workers than their more moneyed counterparts. However, the correlation
between time-related ﬂexibility and work-to-family conﬂict was strongest for lower-
income workers (r=-. 13; see Table B2 located in appendix B), and about equal for middle
and higher income workers (r=-. 10; see Tables B3 and B4). The correlation between time
ﬂexibility and emotional exhaustion, however, was strongest for higher income workers
(-. 17; see Table B4), followed by lower (r=-.14; see Table B2) and middle income
employees (r=. 12; see Table B3). Although it might be expected to reduce instances
where family demands forced one to be late or leave early, formal time-based ﬂexibility
had no relationship to family-to-work conﬂict in any group. Flexibility in work location
was associated only with family-to—work conﬂict, and only among middle (r=. 14 see
Table BZ) and upper income (r=. 16 see table B3) workers; as in the full-sample
correlations, greater ﬂexibility was related to experiencing more conﬂict.

Differences were also found in the relationships between access to dependent care
beneﬁts and the outcome measures. Among lower income workers, having more beneﬁts
was strongly and signiﬁcantly associated with experiencing less work-to-family conﬂict
(r=-.20), family-to-work conﬂict (r=-.1l) and emotional exhaustion (r=-.15), as can be
seen in Table B2. For middle income workers, the relationship was not signiﬁcant in the

case of family-to-work conﬂict, and was weaker (r=-. 12) for both work-to-family conﬂict

140

and emotional exhaustion. None of the much smaller correlations were signiﬁcant
among the higher income group. Beneﬁts appear to be one area where those with greater
monetary resources may have more alternatives than lower income workers, for whom
such beneﬁts represent a powerful form of assistance in reducing conﬂict and stress.

As can be seen in Table BZ, infomal support in the form of understanding and
helpful supervisors was strongly associated with less work-to-family conﬂict (r=-.29) and
emotional exhaustion (r=-.37) for lower income workers, and more modestly linked to
less family-to-work conﬂict (r=-. 18). For middle income workers (see Table B3), the
pattern was similar but the correlations were smaller; r=~.21 for work-to-family conﬂict,
r=-.29 for emotional exhaustion, and r=-.08 for family-to-work conﬂict. The correlation
between supervisor support and family-to-work conﬂict was the same for the upper
income group (see Table B4), but was only signiﬁcant at the .05 level due to the smaller
sample size. Supportive supervisors were again strongly associated with less work-to-
family conﬂict (r=-.25) and emotional exhaustion (r=-.29).

Supportive cultural norms were strongly associated with less emotional
exhaustion at all three income levels, with the closest relationship found among higher
income workers (r=-.33; see Table B4), followed by lower income workers (r=-.29; see
Table B2) and middle income workers (F-.27; see Table B3). It appears that cultural
norms that don’t penalize employees for failing to put work ﬁrst at all times are
associated most strongly with less burnout among the high income workers who likely
care deeply about promotion opportunities. Such norms, however, are also linked to less
stress among lower-income workers. All three income groups report similarly strong

relationships between supportive cultural norms and work-to-family conﬂict (r= -.26 or

141

-.27; see Tables BZ-4). The strength of the relationship between supportive norms and
family-to-work conﬂict depends on income level; as COR theory would predict, the
correlation is strongest for lower income workers (r= -.22; see Table B2), followed by
middle income workers (r= -. 19; see Table B3) and upper income workers (r= -.17; see
Table B4).

Reﬂecting the relationships discussed above, the total environment summary
variable is most strongly associated with reduced emotional exhaustion, perhaps because
the overall environment has direct effects on emotional exhaustion as well as indirect
effects via work/family conﬂict. The correlation, again, is strongest among lower income
workers (r= -.34; see Table B2), with the relationship found among middle income
workers (r-= -.28; see Table BB) and higher income workers (r= -.29; see Table B4)
slightly weaker. The overall environment was also strongly linked to reduced work-to-
family conﬂict in all three samples. For the lower income sample, the relationship was -
.33 (see Table B2), while the correlation was a slightly smaller -.24 in the middle income
sample (see Table B3) and -. 19 in the upper income sample (see Table B4).

Regression Results By Income Level

The correlation patterns discussed in the previous section are also apparent in the
separate work-to-family and family-to-work regression analyses conducted for each
income group to further examine the signiﬁcant interaction results found for income. In
these analyses, the effects of the control variables, as well as personal and work-related

demands and resources are controlled for and examined.

142

Income and Work-to-Family Conﬂict

In the separate regression equations for work-to-family conﬂict reported in Table
2, every step except the controls adds signiﬁcantly to variance explained for all three
income groups. The largest change in R2 was associated with work demands and
resources for the lower and middle income groups, followed closely by the work
environment indicators. For the higher income group, the environment indicators added
the most to variance explained In the lower income group, three of the ﬁve indicators -
the dependent care beneﬁt index, supervisor support and cultural norms that don’t
penalize workers for putting family ﬁrst — had beta coefﬁcients signiﬁcant at the .01
level, all of similar size. The same environment indicators had statistically signiﬁcant
beta coefﬁcients in the middle income group, although the estimated effect size for
cultural norms was much higher than for supervisor support and the beneﬁts index. For
workers in the high income group, the beta estimates for dependent care beneﬁts were not
signiﬁcant, although the beta associated with ﬂexibility in work time was, perhaps
because there was greater variance as more people would be likely to have a high degree
of ﬂexibility.

In all three income groups, the total environment summary variable added
signiﬁcantly to variance explained (see Table 3 equation a), but not as much as including
the environment indicators separately (see Table 3 equation b). The largest beta
coefficient was found in the lower income group (-.29), followed by the higher income
group (-.27) and the middle income group (-.24). This is partially consistent with
expectations, although the smallest coefﬁcient would have been expected to be associated

with the high income group. Although it is not completely clear from the analyses, one

143

plausible explanation is that those in the lower income group experience less emotional
exhaustion due to less extensive job hours and demands. The middle income group may
face a higher level of demands without access to the more extensive resources available
to higher income workers.

In general, however, the overall results are consistent with the prediction that
these beneﬁts are most valuable to workers with fewer ﬁnancial resources, who are less
likely to be able to pay for assistance or alternatives on their own. Overall, the model did
a slightly better job of explaining work-to-family conﬂict in the middle-income group,
where adjusted R2 was .21, than for the upper and lower income groups, where it was .19.
However, R2 was the same for all three groups; the differences reﬂect sample size.
Income and F amily-to- Work Conﬂict

In the equations for family-to—work conﬂict by income level reported in Table 3,
the control variables again explained only a small amount of variance — 0 percent in the
low income group, 2 percent in the middle income group, and 4 percent in the high
income group. This is likely because income is so closely associated with occupation.
The steps representing personal demands and resources, work demands and resources,
and the family supportiveness of the work environment all added signiﬁcantly to the
relatively small amount of variance explained in the lower income group. In the middle
and high income groups, personal demands and resources and work environment added
signiﬁcantly, but work demands and resources did not. In the lower and middle-income
groups, the largest amount of change in R2 was associated with adding the environment
indicators separately at step 4, and this was second only to personal demands and

resources in explaining variance in family-to-work conﬂict for the higher income group.

144

As was the case with the full-sample analyses, the regression equation that
examined the effects of the composite environment variable (labeled equation a in Table
3) explained less variance than the equation that included each of the ﬁve indicators
(labeled equation b) in all samples. The conﬂict-reducing effects of the total environment
on family-to-work problems were strongest for the low income workers, with a beta
coefﬁcient of-. 15 (sig. at .01), and still visible for the middle income group, which had a
beta coefﬁcient of -. 12 (si g. at .01). However, the beta coefﬁcient for the higher income
group (-.08) was considerably smaller and not statistically signiﬁcant.

Looking at the individual beta coefﬁcients for the environment indicators, only
cultural norms was signiﬁcant for workers from lower income households. At the middle
and higher income levels, only non-penalizing cultural norms and ﬂexibility in work
location were signiﬁcant. Contrary to predictions, being able to work from home
appeared to be associated with more conﬂict, rather than less. This result is likely
inﬂuenced by the fact that many of those who indicated they had this beneﬁt were
actually using it. They may have sought such an option because they already had high
levels of conﬂict, or it may be that working at home makes conﬂicts more likely and
more salient.

The results for cultural norms could be interpreted to mean that a more forgiving
culture actually reduces family intrusions on work. As noted earlier, allowing workers to
call and confrrrn that their children or parents are o.k. can make it easier for them to relax
and concentrate. An equally compelling interpretation, however, is that workers in
organizations that expect work to always come ﬁrst and penalize workers for using

family-related beneﬁts, like ﬂextime, are more sensitive when their families affect their

145

work, and thus report a higher degree of family-to-work conﬂict. The cunent study does
not allow us to determine which of the two explanations is more accurate; probably both
are somewhat true.

Overall, the model was most successful in explaining variance in family-to-work
conﬂict among workers from upper-income families, with an adjusted R2 of .11, than
among middle income workers, with an adjusted R2 of .09, or lower-income employees,
with an adjusted R2 of .06.

Higher Income Associated with Smaller Environment Impact

While there are some discrepancies, overall, these analyses underline the fact that
there are differences in how a family-supportive work environment affect workers at
different income levels. As the conservation of resources theory would predict, they are
differences in terms of degree, not direction. Working in a more supportive environment
is associated with less work/family conﬂict and emotional exhaustion for all workers.
But dependent care beneﬁts, particularly, appear most valuable to workers from lower
income households, who are less able to afford alternative assistance. Informal support
in the form of nonpunative cultural norms and understanding supervisors is beneﬁcial for
all workers with less difference due to income levels.

Summary of Results

While not all of the variables included in the study had statistically signiﬁcant
effects, when examined as sets, personal demands and resources, work-related demands
and resources, and a family-supportive work environment all added signiﬁcantly to

variance explained in the three outcome variables for the full sample analyses.

146

In the case of work-to-family conﬂict, being a woman, being younger, and having
more children (especially children under 13) were associated with greater conﬂict. Long
work hours and an irregular schedule were also associated with more work-to-family
conﬂict, while greater work autonomy and a supportive work environment were
associated with less. There was support for the notion that a working in a supportive
environment had a greater effect on workers from lower-income households.

A supportive environment also appeared to have a stronger impact on family-to-
work conﬂict for lower- and middle-income workers than those from higher income
households, although it was associated with less conﬂict for workers in all three groups .
Greater family-to-work conﬂict was associated with spending more time caring for
elderly and disabled adults, having children, especially young children, being younger,
and working longer hours.

The variable with the strongest association with emotional exhaustion was work-
to-family conﬂict. F amily-to-work conﬂict, longer work hours, and being female were
also associated with greater emotional exhaustion; autonomy on the job and a family-
supportive work environment were associated with less. A supportive work environment
had a relatively association with emotional exhaustion, and was connected with less
burnout. The effects of the supportive environment indicators on emotional exhaustion

appeared to be partially mediated by work—to-family and family-to-work conﬂict.

147

CHAPTER 6: DISCUSSION

Contributions of the Study

The purpose of this study was to examine the effects of a family-supportive work
environment on emotional exhaustion and conﬂict between work and family in the
context of other relevant demands and resources in both domains, work and home.
Consistent with the notion, drawn from COR theory, that a family-supportive work
environment can be viewed as an important resource for reducing stress, working in a
more supportive environment was associated with less work-to-farnily conﬂict, as well as
lower levels of family-to-work conﬂict and emotional exhaustion. This was particularly
true for workers from lower income households.

The results found for work-to-family conﬂict are congruent with the ﬁndings of
earlier studies tlmt involved more homogenous samples of workers (e. g. Allen, 2000;
Thompson et al., 1999). The only previous study to examine the relationship between
indicators of a family-supportive environment and family-to-work conﬂict was based on
a sample of male executives and found no connection between the two (Judge et al.,
1994). However, male executives represent a unique group of workers that likely has
uniformly low levels of family-to-work conﬂict, as excessive family interference would
probably have derailed their careers (Kossek and Ozeki, 1998). They also have greater
ﬁnancial resources to draw on and are in a better position to purchase assistance in
dealing with home and family responsibilities that might cause family-to-work conﬂict.
Thus, the ﬁndings in the current study, which included a wider variety of workers, can be
viewed as more applicable to understanding the overall effects of a family-supportive

environment. While previous work had found that supportive supervisors were

148

associated with less emotional exhaustion (Lee and Ashforth, 1996), this was the ﬁrst
study to look at how it might be related to other aspects of a family-supportive
organizational environment.

There was no evidence that working in a family-supportive environment had a
stronger impact on conﬂict and emotional exhaustion among single parents. However,
congruent with implications drawn from social exchange and COR theory, there was
evidence that lower-income workers, who may have the greatest need for organizational
support due to economic limitations on their ability to pay for alternative assistance, may
derive more beneﬁt from working in a family-supportive environment. The relationships
between dependent care beneﬁts and the outcomes were strongest for workers from
lower-income and middle households. They were associated with reduced work-to-
family conﬂict for these groups and with emotional exhaustion for lower-income workers
only; no signiﬁcant relationship to family-to-work conﬂict was found.

Some of the strong effects found for beneﬁts and other supportive environment
indicators may be related to the fact that organizations with family-supportive
environments are also likely supportive in other respects — and workers are also likely to
experience less stress when they have access to better working conditions and more
comprehensive traditional beneﬁts, like paid vacation and retirement plans. Those in the
lower echelons of America’s workforce are not as well treated in these respects, so the
effects of broad-based organizational would be particularly strong for them.

Workers of all income levels experienced more positive outcomes when they had
supervisors supportive of their efforts to do well in both work and family roles, and a

culture that did not penalize them for occasionally putting family needs before work,

149

ﬁndings that are compatible with previous studies (Roehling er al., 2001; Allen, 2000;
Thompson et al., 1999).

The weak results found for ﬂexibility in work timing compound the conﬁrsion
created by contradictory ﬁndings in previous research. Time-based ﬂexibility was
signiﬁcant only in the work-to-family conﬂict regression equation for higher-income
workers, perhaps reﬂecting the formal nature of the ﬂexibility measure and the fact that
ﬂextime policies are more likely to cover workers in better-paying jobs. Supervisors who
allow workers to handle nonwork problems during work hours, a form of informal
ﬂexibility, may have more impact than formal ﬂextime programs. Flexibility in work
location, a type of support that has attracted considerable attention in recent years but has
been little studied, was actually associated with more family-to-work conﬂict in the
middle- and upper-income groups more likely to utilize this beneﬁt, and contributed little
to understanding work-to-family conﬂict or burnout. As noted earlier, many of those
who reported access to this beneﬁt were actually using it. The ﬁnding that working from
home may be associated with greater family-to-work conﬂict is understandable in the
light of the qualitative studies discussed in Chapter 3. However, more research is needed
to evaluate this conclusion.

Overall, a supportive environment was one of the strongest contributors to
variance explained for workers of all income levels for all three outcomes. Part — but not
all - of the impact of such an environment on emotional exhaustion was mediated by
work/family conﬂict, particularly work-to-family conﬂict. While more complete
mediation was expected, this ﬁnding is consistent with the results of previous research,

supporting other studies that have concluded that a family-friendly environment may act

150

as a signal of overall support for workers (Grover and Crooker, 1995; Allen, 2000) and
may reduce stress even if beneﬁts are not used (Lee and Ashforth, 1993) or if they are
more extensive than a worker’s family situation would require (Edwards and Rothbard,
1999)

While the major focus of the study was to examine the effects of organizational
support for workers balancing home and job responsibilities, it also examined
relationships between the outcome variables and other important personal and work
demands and resources. Although ﬁndings regarding the relationship between gender
and both work-to-farnily conﬂict and emotional exhaustion have been mixed, sex
emerged as a signiﬁcant inﬂuence on both. Contrary to expectations, however, it did not
have strong effects on family-to—work conﬂict, possibly because workers of both genders
consciously attempt to learn the accepted limits within their organizations, and either stay
within them or ﬁnd more compatible workplaces. In this study, as in others, reported
levels of work-to-family conﬂict are higher than those of family-to—work conﬂict,
indicating that workers may be more protective of the boundary between work and family
when job performance is at stake (Bond, Galinsky and Swanberg, 1998).

Consistent with previous research and COR theory, having more children,
especially young children, was associated with more work/family conﬂict in both
directions. Little research that reported on the relationship between children and burnout
could be located; the single study that reported a statistically signiﬁcant relationship had
found that parents who served the public at a government agency experienced less
emotional exhaustion than nonparents, indicating that they might serve primarily as a

resource rather than a source of demands (Maslach and Jackson, 1985), contrary to

151

thinking in work/family research. There are important differences between that study and
this one; the sample used here is broader and more variables are included in the analysis.
In the current study, no relationship was found until work/family conﬂict was controlled
— at which point the number of children appeared to have a weak association with lower
burnout. This is consistent with the view that children may be considered both a source
of demands and a resource. The same may hold for partners, which contributed little to
explaining variance in all three outcomes. While some other studies have found partners
to serve primarily as a resource, the strongest effects are, perhaps naturally, reported for
those that include measures of how supportive the spouse is.

The ﬁndings of the current study highlight the importance of considering age and
income level, which have often been ignored in previous research or treated as controls.
The greater knowledge and experience associated with being older represents a strong
resource, as is indicated by the signiﬁcant effects found for all three outcomes.

As anticipated, longer work hours exerted a strong inﬂuence on both conﬂict and
burnout Autonomy also appeared important, at least in understanding work-to-family
conﬂict and burnout, consistent with previous research (Frone et al., 1998; Lee and
Ashforth, 1996). Consistent with previous research driven by role theory, these work-
related inﬂuences added less to the explanation of family-to-work conﬂict. Researchers
seeking to explain this phenomenon need to focus on non-work factors.

Irregular schedules, which previous research has associated with increased
burnout (Demerouti et al., 2001), here had their only effect on work-to-family conﬂict.
As anticipated, a need to be on the job during the hours that most people are home with

their families was associated with greater conﬂict, likely because it reduced the overall

152

amount of time that could be spent together. Contrary to the ﬁndings of some other
studies, longer tenure had only a weak effect, and only on farnily-to-work conﬂict. The
implication is that greater experience may enable workers to better manage the demands
of their private lives without impacting their work, but does not necessarily mean they
won’t let their work interrupt family activities.

The demands and resources included in the model did a better job of predicting
levels of work-to-family conﬂict and emotional exhaustion than family-to-work conﬂict.
Likely this reﬂects the fact that work-related demands and resources, including a family-
supportive organizational environment, are most closely related to these outcomes, while
family-to-work conﬂict is more closely associated with demands and resources off the
job. Although eight measures of personal demands and resources were included in this
study, they explained very little variance in farnily-to—work conﬂict, perhaps because they
were not speciﬁc enough. Future research of family-to-work conﬂict in particular needs
to include detailed information on a very broad range of personal demands and resources.

Overall, the results of the study highlight the importance of including both work
and personal factors when studying work/family conﬂict and burnout. Although the
major emphasis was on ﬁlling the gaps in knowledge about the effects of a supportive
work environment on these outcomes, the ﬁndings show the value of linking the
perspectives that underlie research in all of these areas, a major contribution of this study.
Implications for Future Research

The study has two important implications for future research on burnout. First,
occupation and industry explained relatively little variance in emotional exhaustion,

implying that this part of the burnout syndrome, at least, is applicable to workers in wide

153

range of jobs, rather than affecting mainly the human services workers previous research
has focused on. The current study validates previous ﬁndings in more homogenous
groups that strong work demands contribute most to understanding emotional exhaustion
(Lee and Ashforth, 1996). Similar to studies of human service workers, autonomy was
found to be a valuable resource, but to have less impact than work demands (Lee and
Ashforth, 1996). More research is needed, however, to evaluate whether all of the
relationships supported by COR theory and the extensive amounts of previous research
also hold for all workers.

Second, the study indicates that family composition may be less important in
studying burnout than the way than the boundary between work and family and how well
it is managed. Conﬂict between the two domains and a supportive work environment,
which appears to help workers in boundary management, both had signiﬁcant effects,
while marital status and the number of children did not. The implication is that family
may have both positive and negative effects, but when conﬂicts arise between work and
family the overall impact is strongly negative. Despite the lack of signiﬁcant results for
having a partner or children, then, nonwork inﬂuences may still be important. Conﬂict
that ﬂows from work to family is more destructive than conﬂict from family to work.

For researchers in the area of work and family, one of the most important
implications that this study has is that the nature of the sample matters, particularly where
income is concerned With a few notable exceptions, previous research in the area of
work and family has often focused on fairly well educated groups of upper-middle-class
workers, such as university employees (e.g. Edwards and Rothbard, 1999, Grandey and

Cropanzano, 1999) or graduates (e. g. Thompson et al., 1999). This is understandable, as

154

such workers are easy for researchers to survey and tend to have experiences similar to
their own. However, as F rone et al. (1992) found in their examination of the differences
between blue- and white-collar workers, it is likely that income and social class may be
associated with differences in the work/family interface. Thinking regarding important
demands and resources in current research often reﬂects a middle-class perspective. This
study ﬁnds support for the notion that the strength of effects of a supportive work
environment may not be the same for everyone. It is likely true that the effects of other
inﬂuences that have been viewed as important may differ by income group as well. For
example, spouse work hours have been posited as a “demand” tlmt may lead to greater
conﬂict between work and family, and support found among upper-middle-class dual-
income couples (e.g. Ayree, 1993). However, for low-income workers, having a spouse
who works longer hours may mean the money to pay for decent day care, and thus reduce
conﬂict. Spouse work hours are unlikely to have a strong impact on high-income
workers, because it is easy for them to purchase assistance with household tasks. Indeed,
spouse work status had little impact on the level work/family conﬂict experienced by the
executives in Judge et al. ’5 (1994) study. Future research should deﬁnitely note the
implications associated with choosing a sample that is largely similar in income levels,
and more research that compares differences in the work/family interface between
income groups is needed. Studies that involve broad samples of workers can contribute
much to our understanding of how generalizeable the ﬁndings of previous research are.
Overall, the implication is that research should consider a broad array of factors,
and may want to draw extensively on other literatures to identify potential inﬂuences.

Researchers in the area of HR practices and policies, particularly, may beneﬁt from

155

examining a wide range of theoretical perspectives, as work in this area has tended to be
more driven by practical considerations.

Implications for Practice

In today’s weakening economy, companies are re-examining the beneﬁts and
programs they offer, including those aimed at supporting families. Programs that cannot
show their worth are unlikely to be kept. The current study provides some evidence that
such policies are effective. In ﬁnding that a family-supportive work environment is
associated with lower levels of work-to-family conﬂict, family-to-work conﬂict, and
emotional exhaustion, this study conﬁrms that organizational efforts to be more “family
ﬁiendly” have effects beyond positive work attitudes. Supportive environments appear to
actually reduce conﬂict and burnout, increasing their potential value for employers, since
these outcomes have been associated with negative consequences for employers,
including reduced job satisfaction, commitment, and performance (Kossek and Ozeki,
1998; Lee and Ashforth, 1996).

Workers with the highest levels of work and family demands experienced the
most conﬂict and burnout. The study showed that demographic characteristics can be
used as rough indicators of family demands. While it is illegal to discriminate against
workers because of age, gender, marital status, or because they have several young
children, most employers have at least some of this information in their personnel ﬁles.
This information can be used to get some idea of the kinds of resources and demands
their employee have at home. More extensive information is available on work demands
through job descriptions. Concerned employers can reduce work demands and help those
with high personal demands by supplementing personal resources with a family-

supportive environment.

156

When looking at workers of all income levels, the most effective forms of support
were not formal beneﬁts. Supervisors who are supportive when workers experience
family-related problems and cultural norms that do not penalize workers for putting
family ﬁrst had the strongest impact on work/family conﬂict and bumout. This
highlights the importance of working to create a supportive culture in addition to
introducing policies and beneﬁts. This may be accomplished through supervisor training
and a careful re-examination of the criteria commonly used for promotion decisions. Are
workers encouraged to keep quiet about family problems, or helped to resolve them?
Does turning down a transfer because of a spouse’s career signiﬁcantly reduce promotion
opportunities? If so, companies are unlikely to reap beneﬁts from other work/family
programs, as workers will not use them (Thompson et al., 1999). On the other hand,
once supportive attitudes are entrenched in the company’s supervisors and culture, they
may provide a relatively low-cost source of competitive advantage, in that such support
reduces burnout and family-to-work conﬂict and is associated with more positive work
attitudes (Thompson et al., 1999; Allen, 2000).

Formal ﬂextime policies and the ability to select one’s work schedule did not
have strong effects. It is likely that having a supervisor who is accommodating when
workers need to take time off during the work day for personal or family reasons,
represents a more useful source of ﬂexibility for workers, again underscoring the need for
making sure that supervisors are supportive. Having the ability to work from home was
did not have an impact on work-to-family conﬂict or burnout, but was associated with
more family-to-work conﬂict. Employers, already leery about allowing workers without

strong track records to participate in such programs, may want to continue examining the

157

effects of telecommuting and other work-from-home programs and should make an effort
to set them up in a way that limits potential interruptions.

The effects of working in a family-supportive environment were stronger for
workers most likely to need support: those from lower-income households. Lower-
income workers particularly showed the highest beneﬁts from greater access to formal
dependent care assistance policies. Unfortunately, as the correlation between the beneﬁts
index and income level shows, they also reported the least extensive beneﬁts. Given that
they do appear to be associated with less negative outcomes, organizations that employ
large numbers of low-paid workers in particular may seriously want to consider making a
strong effort to create a supportive environment.

Study Limitations and Suggestions for Future Research

This study has both strengths and weaknesses that may affect its contribution to
answering the questions outlined in the introduction. A major strength is the size and
diversity of the sample used. In general, research in the area of work and family has
relied upon small studies, with only a few based on nationally representative samples
(Grzywacz, Almeida, and McDonald, 2002). Much of what is accepted as knowledge in
this area comes from studies that have drawn smaller samples from homogenous groups
with the same occupations, looked at workers in a single organization, or limited their
analyses to special groups like parents of preschoolers, dual-career couples, male
executives, or career women, making it more difﬁcult to generalize from the results
(Kossek and Ozeki, 1998). The large, national sample included in the NSCWF provides
for more naturally occurring restriction in variable range, offers sufﬁcient power to detect

weaker relationships, and allows for generalization to all types of workers in North

158

America. In addition, the large number of issues that information was collected on makes
it possible to study the relationships between many different factors in work and family
life. The large number of study participants at varying income levels made it possible to
investigate how effects differ for each group.

There are also some weaknesses associated with using the NSCWF data set,
however. First, all of the information for all of the variables measured was collected at a
single point in time from a single source. Cross-sectional analyses like this one can
imply, but not prove, causal effects. They also may suffer from common method
variance, which can inﬂate the relationships found. In this sense, the study does not
improve upon previous work in the area, much of which has relied upon cross-sectional
analyses of self-reported estimates provided in response to questions on a single survey.
However, the content of the questions somewhat allays these concerns regarding this
analysis. Many of the items dealt with straightforward issues like the existence or
nonexistence of personnel policies, demographic characteristics and estimates of work
hours and family ﬁnances, which are unlikely to be affected by attitudes or other biases.
Indeed, as far as possible, demand and resource variables in this study were selected
because they could be measured in ways that would not be affected by personality, mood,
or a common response pattern.

While there is greater concern that common method variance or another factor,
such as personality, may inﬂate the relationships between scales that have an attitudinal
component (i.e. ratings of supervisor support, cultural norms, work-to-family and family-
to-work conﬂict, and emotional exhaustion), it is difﬁcult to think of good alternative

sources for many of these variables. Information on supervisor support, cultural norms,

159

beneﬁt availability and the degree of ﬂexibility in work location and timing could be
collected from other sources, such as co-workers employed in the same unit. There are
tradeoffs, however, in adOpting such an approach, because it would logically limit the
number of cases studied due to costs and other constraints. Also, it may well be that
employee perceptions of how supportive their work environment is are more important
than the actual level of support provided. Somewhat allaying concerns about either
approach, previous studies which have used multiple sources for some of this information
(e.g. Thomas and Ganster (1995) for beneﬁt availability and Demerouti et al. (2001) for
supervisor support) have found a strong correspondence between individual and outside
ratings.

In the case of the three dependent variables, alternate rating sources are even
more limited. While it may be possible to get family members to evaluate the degree of
work-to-family conﬂict (as Burke, Weir and DuWors, 1980 did) such an approach may
result in unwanted effects, such as underestimates for workers who hide their difﬁculties
and emotions better or perceptions of conﬂict that incorporate the degree to which family
desires or expectations are fulﬁlled. Multiple raters may be useful, but an accurate
measure of the degree of work/family conﬂict and emotional exhaustion probably
requires asking the person involved, not just inferring it from another source. In defense
of one-time general self ratings like those used in this study, there are indications that
such ratings of work/family conﬂict do accurately represent changes in mood and energy
levels experienced at different times and by moving between the two domains. Two
studies that used multiple measures of different types, found strong correspondence

between them (Williams and Alliger, 1994; Grzywacz et a1. , 2002).

160

Based on this information, it appears likely that the advantages of using self-
report measures from a single survey may make up for many of the drawbacks,
particularly in a large study of this type. However, future research that does incorporate
measures from different sources, such as family members, co-workers, and observers, is
also needed to round out the literature.

Personality effects also represent a major concern for this and other studies, as
well as an important area for future research. Research on emotional exhaustion has
found it to be strongly related to the personality measure of negative affect and also
signiﬁcantly related to positive affect. These measures are intended to capture the
tendency to view situations negatively and positively (Wright and Cropanzano, 1998;
Iverson, Olekalns, and Erwin, 1998). Personality may be one reason for the close
relationships found between the perceptual variables included in this study (supervisor
support, cultural norms, work-to-family and family-to-work conﬂict, and emotional
exhaustion), both because it affects the tendency to provide negative or positive analyses
of the situation, and because it may affect how people respond to stress. As Iverson er a1.
(1998) noted, negative affectivity may lead people to interpret situations negatively and
emphasize negative information, which would mean those with more negative outlooks
may select negative responses to all items regardless of the actual situation, increasing
relationships found. People who are more negative in general may also make less use of
direct coping strategies (Iverson et al., 1998), increasing the ill effects of stresses and
reducing the effectiveness of support. Future research in this area, then, should deﬁnitely
make an effort to incorporate these variables and estimate their inﬂuence on the

relationships found here.

161

Conclusions

A single study such as this one cannot and should not be considered a means of
providing a deﬁnitive answer to the questions posed. Rather, it should be viewed as
contributing to our understanding in the context of a program of research employing a
variety of methods. The goal of this dissertation is to help ﬁll in gaps in our
understanding of the effects of working in a family-supportive work environment by
looking at how this is related to work-to—family conﬂict, and family-to-work conﬂict and
burnout, within the context of other important inﬂuences at work and at home. The study
found that more positive outcomes were associated with working in supportive
environments. The results agree with previous ﬁndings that informal support, such as
understanding supervisors and a culture that doesn’t penalize workers for putting family
ﬁrst, are perhaps the most important elements of a supportive environment for the
majority of workers (Allen, 2001; Thompson et al., 1999; Thomas and Ganster, 1995).
While childless workers and those from higher-income families may have less use for
dependent care beneﬁts, they appear to have beneﬁts beyond reducing work/family
conﬂict and are among the most effective resources for those who have lower ﬁnancial
resources. Employers who draw their workers from the ranks of lower-income families
in particular may want to seriously consider working to create a family-supportive
environment.

Why should employers care? Emotional exhaustion and work/family conﬂict
have been closely associated with reduced job satisfaction (Kossek and Ozeki, 1998; Lee
and Ashforth, 1996) lower organizational commitment (Wiley, 1987; Netemeyer et al.,

1996; Lee and Ashforth, 1996) and a greater intention to quit (Netemeyer et al., 1996;

162

Lee and Ashforth, 1996). Family-supportive beneﬁts and informal support, on the other
hand, have been associated with higher job satisfaction (Allen, 2001), greater loyalty
(Roehling et al., 2001) increased organizational commitment (Grover and Crooker, 1995;
Allen, 2001 ) and lower turnover intentions (Allen, 2001; Grover and Crooker, 1995;
Thompson et al., 1999) in addition to the effects found here. This study supports others
in ﬁnding that family-friendly work environments have beneﬁts for both workers and

organizations.

163

APPENDIX A: SURVEY ITEMS AND SCALES

164

APPENDIX A: SURVEY ITEMS AND SCALES
Control Variables
Race
What is your race?
White ..
Black, African American
Other, including mrxed

MN!—

Occupation
What kind of work do you do in this job? That is, what is your occupation?

Executive/administrative/managerial...
Professional.......................
Technical............
Sales
Administrative support
Servrce
Production/operator/repair...

\rosmlrszu—a

Industry
What kind of business or industry is this company or organization involved in?
What do they make or do where you work? What is the main thing?

Goods Producing...... 1
Service..................... 2

Personal Demands and Resources

Gender

(This was coded by the interviewer based on responses to questions.)

Elder/Disabled Care Time Demands

Do you currently provide special attention or care for someone 65 years old or older?
About how much time do you spend per week providing care or assistance in person —

such as meal preparation, household work, physical care, transportation to medical
services, etc.?

165

And about how much time do you spend per week doing other things — such as calling on
the telephone to see whether everything’s alright, arranging for services, handling
ﬁnances, etc?

Do you currently provide special assistance or care for a disabled non-elderly adult?
About how much time do you spend per week providing this special care or assistance?
(Responses were summed to estimate total hours spent per week)

Number of Children 0-12 and 13-18

Do you have any children who live with you for a least half the year - including natural,
adopted, foster, or stepchildren for whom you are responsible, or not? How many?

Do you have any children who live with you for less than half the year, but for whom you
are the legal guardian for the entire year — such as a child away at school or a disabled
child in long-term care? How many?

Of these children, how old is your youngest child? How old is your next-youngest child?
(Responses used to compute the number of children in each age group.)

Partner

Are you presently married, living with someone as a couple, single and never married,
divorced, widowed or separated? Do you live with your spouse, or not?

(Those who are married and living with their spouse or living with someone as a couple
were coded as I , having a partner, others were coded as 0, no partner)

Familial Child Care

What is the main child care arrangement you use for your youngest child while you are
working at your job?

Age
In what month and year you were born?
Household Income Level

What was your family’s total income from all sources before taxes in 1996? A rough
guess will be ﬁne.

(Household income estimates were used to place respondents into three categories: 1,
low income ($2 7, 760 or below, in the bottom 40% of US households for I 996 as

166

estimated by the UC Census Bureau) 2, middle income (between $27,761 and $68,015, or
between the 40‘" and 80’” percentiles for US households) and high income (above
$68,015, or in the top 20% of US households).

Work Demands and Resources

Total Weekly Hours Worked

How many paid hours a week are you regularly scheduled to work? Don‘t include any
paid or unpaid extra hours that you put in beyond your ofﬁcial work week.

On average, how many additional hours a week do you do any work related to this job -—
including both unpaid and paid overtime worked at any location?

On average, how many hours a week do you work at your other job (or jobs)?

(Total hours is the sum of responses to these 3 items; respondents were told the totals and
asked “Does this seem about right for your typical week? ")

Regular/Irregular Schedule

Which of the following best describes your work schedule at your main job — a regular
daytime schedule, a regular evening shift, a regular night shift, a rotating shift — one that
changes periodically from day to evening or night, a split shift consisting of two distinct
periods each workday, or a ﬂexible or variable schedule with no set hours?

(Responses recoded so that a regular daytime shift =0, all others = I )

Autonomy

I have the freedom to decide what I do on my job. Do you strongly agree, somewhat
agree, somewhat disagree, or strongly disagree?

It is basically my own responsibility to decide how my job gets done.
I have a lot of say about what happens on my job.

Tenure

How long have you worked for this employer?

Family-Supportive Organizational Environment

Time ﬂexibility

Are you allowed to choose your own starting and quitting times within some range of
hours, or not?

167

Are you allowed to change your starting and quitting times on a daily basis or must you
stick to the times you choose?

Overall, how much control would you say you have in scheduling your work hours -
complete control, a lot, some, very little, or none?

Location Flexibility

Including only regularly scheduled hours — not overtime or extra time worked — how
many hours a week do you usually work at home? (Recoded so that those who spend any
time working at home coded as 1, others coded as 0.)

Would you be allowed to spend part of your regular workweek working at home if you
asked? Do not include extra hours that are either paid or unpaid. (Recoded so that no =
0, maybe or yes =1, and combined with responses to previous question, recoded so 0=not
available and 1 = maybe, yes, or currently work at home =1 . )

Dependent Care Programs

Does your employer have a program or service that helps employees ﬁnd child care if
they need it, or not?

Does your employer have a program that helps employees get information about elder
care or ﬁnd services for elderly relatives if they need them, or not?

Does your employer operate or sponsor a child care center for the children of employees
at or near your location, or not?

Does your employer provide employees with any direct ﬁnancial assistance for child care
- that is, vouchers, cash, or scholarships, or not?

Does your employer have a program that allows employees to put part of their income
before taxes in an account that can be used to pay for child care or other dependent care?
These programs are sometimes called "cafeteria plans" or "dependent care assistance
plans."

Are women who work for your employer able to take time off work to recuperate from
childbirth without endangering their jobs, or not?

Are men who work for your employer able to take time off work when they become
fathers without endangering their jobs, or not?

Are you allowed to take a few days off to care for a sick child without losing pay, without

using vacation days, and without having to make up some other reason for your absence,
or not?

168

Supervisor Support

My supervisor accommodates me when I have family or personal business to take care of
— for example, medical appointments, meeting with child's teacher, etc. Do you strongly
agree, somewhat agree, somewhat disagree, or strongly disagree?

My supervisor is understanding when I talk about personal or family issues that affect my
work.

I feel comfortable bringing up personal or family issues with my supervisor.

My supervisor really cares about the effects that work demands have on my personal and
family life.

My supervisor is fair and doesn't show favoritism in responding to employees' personal or
family needs.

Cultural Norms

At my place of employment, employees have to choose between advancing in their jobs
or devoting attention to their family or personal lives. Do you strongly agree, somewhat
agree, somewhat disagree or strongly disagree?

At the place where you work, employees who ask for time off for family reasons or try to
arrange different schedules or hours to meet their personal or family needs are less likely

to get ahead in their jobs or careers.

At my place of employment, employees who put their family or personal needs ahead of
their jobs are not looked on favorably.

If you have a problem managing your work and family responsibilities, the attitude at my
place of employment is: "You made your bed, now lie in it!"

There is an unwritten rule at my place of employment that you can't take care of family
needs on company time.

Dependent Variables

Work-to—Famiiy Conﬂict

In the past three months, how often have you not had enough time for your family or
other important people in your life because of your job?

In the past three months, how often have you not had the energy to do things with your
family or other important people in your life because of your job?

169

In the past three months, how often have you not been able to get everything done at
home each day because of your job?

In the past three months, how often have you not had enough time for yourself because of
your job? Would you say very often, often, sometimes, rarely, or never?

F amily-to- Work Conflict

How often has your family or personal life kept you from getting work done on time at
your job?

How often has your family or personal life kept you from taking on extra work at your
job?

How often has your family or personal life kept you from doing as good a job at work as
you could?

How often has your family or personal life drained you of the energy you needed to do
your job?

How often has your family or personal life kept you from concentrating on your job?
Emotional Exhaustion

How often during the past three months have you felt used up at the end of the workday?
Would you say very often, often, sometimes, rarely, or never?

During the past three months, how often have you felt emotionally drained from your
work?

How often during the past three months have you felt tired when you got up in the
morning and had to face another day on the job?

How often during the past three months have you felt burned out or stressed by your
work?

170

APPENDIX B: SUPPLEMENTARY TABLES

171

 

 

mo.
n..
...
8.
..o.
n..

n..
8.

a.
8.
and

v .30!—

do. an:
e.. me.
.0. en .
8.- mm .
0o. en .
8. mm.
no. 3.
no. n..
mo. ....
8. e..

.33—=0

n88...— " 88am

2.
...
om.
n..
e..
K.

Nu .
an.

3.

as.
man

astonam
_ ..euonh

€00.30 8 2.0.. ...0... a. .80.... 8m

9 38... mmmA 03 800: >..Ea.. .8 3.880.. :05 808 3 v.59. 8 3.5.02.3 80.0%?
on; 9 .9 .8 338. >35... 8.. .mo 08.. 8.. 0.3 2.3 808880 183 90>
08:3 08... 05 a ”82.88% 3.328.. 0... 5.3 00%.. :0> .838 32. 08 :0. 83...

.08.. 339:8 .8 300: >35... .8 83 00.3
redo =o> a... Eon—>295 ..o 03... >8 8 e..: 85.35 5 a. 0.2....

...u. a. 0.. 3o: ...2. 50> 0.88 8?. a. Eon—>295 .8 08... >8

8 0.03:8 05 .m0........neo%0e 3.8.... .23 e..—o3 88> mega 80.98.. a 03:. 30> ...
.83. .5882. 8 >..En.. ...08 o. 8.809. $.83... 8 2.8..

:05 a. 9.8533 5038.. 88:0 3 03:. $08.95 8088880 .8 08:. >8 2.

>398}. .8 .0888. 8: 03 38.. ..05 ..o
.805 800: .832 8 >55. ..05 .2. 2.3 808.95 8088.980 .8 08... >8 2

.20 .858. «.230 5.3 $5008 £588.83 .3808 .08—5x0 8.. r .8
0.80 00.5 o. anon—.25 Econ—0n. 8 >..Ea.. 0.8.. . :33 08 88.88883 c8389.... >2

.308 >28... .8
.885. 808880 9 3.9898. c. 83.8.6. 39.» 0.800.. .8» ...e. a. 8....3093 >2

8mg?“ >8 5.3 88.... >35... 8 8:88.. a: maﬁa—.5 0.8.5.88 .00.. .

e.. 2.5 ea
.8022. >8 .8 05.. 3550.. e..—o3 .0... 800...». 05 Scan «080 >..a0._ “8.32.3 >2

e..—o3 >8
80.5 an... 38.... 2.80.. .8 .3088 59... 0:5 . :0..3 $8550.03 a. 5.3.0....» >3.

53—

 

 

.eoaeehgem .8: 82.8.3— 383 ”83.23. 888m n..—2.2.880 .38—...:— .8 8.30:

3m 033—.

172

 

 

8......30 50:00.0 00.00000. 80.00300
6.0.00.

0.0.00...
0..
0..
0..

n..
hm.
mo.

co.

0 .800...

ON.
8.
.N.

a:

.3:

Va .

.0».

n .880.—

.8..E&m 38> .88. 0c 008000..

.000. ... .0 .000. 8.0 008550.. ... 0.32 8.080 3.2.800.
8. .N. .00.. >8 .8 00009... 00.3 8000 >00 ..0 8. 0 0.5.. .
.8. No- .0000 0.0» 00.. >8 30.. 00.000 0. 5......000000. :30 >8 3.00.02. 0. 0.
... 0.. ...o..a=o8.§3§809§§0050§.

0080.. 80....

8.- 3. 080.. 0.83 00.000000 900.000.. 80> ..0 088 0.83 .0. 09.0.5 00> 0.000 .8. 00> 0Q
0000.. .8 .05... >..0> .088 .00. 0 8.0000 80.9000

0.. n.. I 0.80... 0.83 80> m0..000..00 ... 02... 00> >00 00> 0.003 .8800 ..008 30.. ...805
...80 8 0.80..

N.. 3. ..0 0000.. 088 0.5.3 008.. 950.00 000 0.5.0.0 030 80> 0009.0 00 0030.... 00> 02.
00002.0 00> 008.. 05 8 0.0.00 00>

00. 8. .008 8 0.00.. >000 0 00 008.. 00.5.0.0 000 0.5.000 80> 09.8.0 0. 0030.... 00> 0.2.

08-8-0. 08000.
.2330 .3288

N 0800.. . .880.— 80.—

 

 

0.3588. ... 2......

173

 

 

625 Ed
320 2.5

a».
R.
mm.

3..
:.
mm.
3.
2.

co.

om.
3.
mm.
«N.
38
25.3...
6.31.3 3
n 5.05,.—

553 8.2
and 03

t.
2.
S.

2.
>0.
3.
E.
vs.

Na.

0..
N_.
3.

m _ .
380
53>»

.3585.
N 5.00.,—

333 8.?

A85 0%
S.
0m.
2.

on.
a.
3.
mm.
no.

N_.

a .
a .
an .

M».
a
150:08H

— 5300."—

50.55.98 00§5> .305 .00 05080205

300055055 8 05500.. 505305 055350me

30> 50> .50 005800 0.:— 50> 8 08005 5555055 0050 .50

>=8£ 50> 555 $85 05 05 >950 055 5.5 H02 50> 028 5050 30: .5508 00.55 605 05 5

Son 5030 0.5300 >5
5000 080: 50 0505 m85>50>0 50m 05 030 5000 H02 50> 025.— 5050 30..— .05508 00.5 5005 05 5

$9.
50> .50 050000 285% 50.0 085 35050 50: 50: 50> 0.8: 5050 30: .5808 005 $05 05 5

30.. 50> .«0 005800 0.5 50> 8 0.5005 58.555

5050 50 >=8£ 50> 5.5 085 535050 55 HOZ 50> 028 coco 30: .5808 00.55 805 05 5
30.. 50> :0 558800500 80¢ 50> 80x 0.:— 850800 ..0 >555 50> an: 5050 303

30> 50> «a 503 058 50 55.3 805 50> 80x 0.:— 850800 ..0 >=85 50> 05 50:0 303

20> 50> 05 05 505005 50> >505 05 50 50> 50805 0.:— 850805 50 >=8£ 50> v.0: 50c0 30m
«50> 50> 8 085 50 0505 503 550m 80¢ 50> 800. 0.5 350805 ..0 >555 50> 02 50:0 30:
52500

50> 8 503 55 00> 5 500m 8 9505 805 50> 500. 0.:— 580805 .0 >=85 50> 05: 50:0 30:

50> 05 50 >5 55050 005 05 55:
5:5 55:08 05 8 5 8m 50> 50:3 505 :05 50> 025. £808 085 5555 05 @555 5050 30m

$5503 05 .5 550 05 .0 55 505 :00 50> 055. 05508 025 585 05 5555 50¢0 303
$53 50> 805 50805 25805080 20.5 50> 025. «5508 005 #05 05 5555 50.50 30:
3.03 50> >0 503.05 50 50 50550 :00 50> 025. 05508 00.55 5055 05 5555 50:0 30:

80:

 

 

0025:; 080850 "000323. 5300."— 350505800 .58058& >0 025.0%

€055.50“! :— 035,—.

174

Appendix Table BZ: Intercorrelations for Lower-Income Group

 

£211 £22, 1 2 3 1 § 9
1. Manager .07 .26
2. Professional .09 .29 -.09"‘
3. Technical .04 .19 -.O6 -.O6
4. Sales .11 .31 -.10** -.11" -.07
5. Admin. .17 .38 -.13" -.15" -09" -.l6"
6. Service .21 .40 -.14" -.16" -10** -.18” -23”
7. Prod/op/repair .31 .46 -.19" -.22" -. 14” -.24" -.31" -.34"
8.1ndustryl .79 .41 .12M .11" .03 .17" .15M .25"
9. White .71 .46 .04 09* .05 -.02 -.01 -.07
10. Black .19 .39 -.04 -.04 -.O6 .04 .00 08*
11. Other race .10 .31 -.01 -.08" .00 -.03 .01 -.01
12. Sex 1.6 .49 .07 .07 .01 .10** .24" .05
13. Eld/oth cm2 2.3 7.38 -.02 .07 .02 -.04 -.04 .05
14. # kids < 13 .51 .90 -.01 -.06 —.01 -.05 .03 -.04
15. # kids 13-18 .21 .53 .06 -.O6 -.01 -.01 -.05 .02
16. Partner .32 .47 -.O6 -.11" .00 -.07 -.12 .04
17. Familial care3 .21 .41 .02 -.05 -.07 -.05 -.02 .03
18. Age 38.73 13.51 .01 -.01 -.07 .02 .02 .02
19.w6rlt hrs/wk 42.27 12.91 .03 .01 .03 -.09* .03 -.13"
20. Irreg. sched. .35 .48 -.O9"' -.O7 -.O4 .15" -.12"“" .15"
21. Autonomy 2.87 .75 .11" .07 .07 -.02 -.04 -.Ol
22. Tenure 4.75 6.36 .01 .00 -.03 -.07 .03 -.01
23. Flex time 1.23 1.23 09* .12" .05 .05 -.05 .03
24. Flex place .16 .37 .12" .19M .07 .03 -.02 -.06
25. Dep. carc‘ 2.92 1.48 .06 07* .14" -.08“ .07 -.06
26. Sup. support 3.30 .68 .04 .06 -.03 .05 .06 -.02
27. Cult. norms 2.81 .76 .06 .11M .10** .03 .02 -.11"
28. Tot. cnvircn.’ 8.67 2.07 .10** .17" .11" .02 -.02 -.07
29. WFC‘ 2.82 1.06 -.01 -.02 .05 -.02 .01 -.05
30. ch7 1.90 .76 .00 .00 -.01 -.03 .01 .01
31. EB“ 2.93 1.12 .06 .06 .03 -.07 -.01 .00

 

 

 

Pearson zero-order correlations between major study variables. 1. Goods-producing =1, service =2.

2. Hours per week caring for elders and disabled adults 3. Most child care provided by a family member
4. Number of dependent care beneﬁts 5. Total environment 6. WFC = work-to-family conﬂict 7. FWC =
family-to-work conﬂict 8. E = emotional exhaustion. Due to listwise deletion for missing data, Ns
range from 670 to 734. " signiﬁcant at the .05 level. " signiﬁcant at the .01 level.

175

Appendix Table BZ: Intercorrelations for Lower-Income Group

 

 

(Continued)

1 3 2 L9 .1_1 Q .1.; .11
1. Manager
2. Professional
3. Technical
4. Sales
5. Admin.
6. Service
7. Prod/op/repair
8. Industry -.59**
9. White -.02 .02
10. Black -.03 .03 -.75“
11. Otherrace .07 -.06 -.53" -.17**
12. Sex ~40" .26" -.04 .11"”‘l -.08"'
13. Eld/oth care -.03 .01 -.04 .06“ -.03 .05
14. # kids < 13 .09 -.O6 -.15"'* .08" .13" .Ol -.05
15. # kids 13-18 .02 .03 -.O8" .08“ .02 .09" .02 .07
16. Partner .22” -.16" .09" —.06 .10 -.24‘”'I .00 .26"
17. Familial care .07 -.05 —.02"”" .09‘ .11" -.06 -.05 .69"
18. Age -.02 -.02 -.15" -.06 -.09* .19" .07 -.24"
19.Work hrs/wk .11" -.O7 -.03 .03 .00 -.19" -.01 .02
20. [meg sched. -.O3 .07” -.01 -.02 .05 -.08‘ -.02 -.04
21. Autonomy -.08 .06 .Ol -.10** .11" -.07 .02 -.04
22. Tenure .04 ~10" .07‘ -.O4 -.05 .05 .01 -.15"
23. Flex time ~16" .10" .07 -.O6 -.02 -.O3 .04 -.O7
24. Flex place -.16” .10" -.01 .01 .OO .01 .00 .01
25. Dep. care -.08"' .10" .02 .02 -.05 .00 -.04 -.O9
26. Sup. support -.06 .11" .02 -.01 -.O3 .06 -.06 -.02
27. Cult. norms -.12" .05 .06 -.Ol -.08"‘ .16"I -.04 .00
28. Tot. environ. -.19“"" .16" .07 -.04 -.06 .06 -.04 -.06
29. WFC .07 -.04 .03 -.O3 .00 -.O4 .00 .11"
30. FWC .00 -.03 -.02 -.OO .00 .00 .07 .08“
31. EB .00 - 01 .01 .00 -.Ol .03 -.Ol .03

 

176

 

Appendix Table BZ: Intercorrelations for Lower-Income Group

 

 

(Continued)

l§ lé 1.7. .113. .12 2.0 ll 2.;
1.Manager
2. Professional
3.Technical
4. Sales
5. Admin.
6. Service
7. Prod/op/repair
8. Industry
9. White
10. Black
11.0therrace
12. Sex
13.Eld/oth care
l4.#ldds<l3
15. # kids 13-18
16. Partner .07
17. Familial care .06 .31"
18. Age .07 .06 -.18"
l9.Work hrs/wk .03 .02 -.03 ~14"
20.1rreg.sched. .00 -.03 -.Ol -.13"”" -.02
21. Autonomy -.Ol .05 -.01 .04 .03 -.02
22. Tenure .00 .04 -.14“ .44" -.02 -.08" .05
23. Flex time .05 .04 -.03 .06 -.06 .02 .36" .08
24.Flexplace .00 .01 .05 -.11” .11" .OO .13" -.02
25.Dep.care -.O8* -.07 -.10*" -.08“' .10" -.05 .15“ .04
26. Sup. support -.05 .01 -.01 .09" -.11"‘ -.08"‘ .29" .05
27. Cult. norms -.05 -.09* .02 .01 -.05 -.08" .23" .03
28. Tot. environ. -.09"' -.06 -.O3 -.05 -.03 -.O7 .41" -.Ol
29.WFC .09" .06 .11" -.12" .18" .11" -.19“ -.O3
30. FWC .08“ .03 .08" -.O7"' .03 .06 -.O7 .02
31.EE .06 -.07* .02 -.14” .17“ .02 -.23" .02

 

177

 

Appendix Table 32: Intercorrelations for Lower-Income Group

(Continued)

 

 

Q 2:!
1. Manager
2. Professional
3. Technical
4. Sales
5. Admin.
6. Service
7. Prod/op/repair
8. Industry
9. White
10. Black
11. Other race
12. Sex
13. Bid/0th care
14. # kids < 13
15. # kids 13-18
16. Partner
17. Familial care
18. Age
19.Work hrs/wk
20. Irreg. sched.
21. Autonomy
22. Tenure
23. Flex time
.15"
.14" .07
.17" .06
.22" .07

24. Flex place
25. Dep. care
26. Sup. support
27. Cult. norms
28. Tot. environ. .67" .30"
29. WFC -.13" .01
30. FWC .01
31. EB -.l4"

'2?

2_5,

.21"
.17M
.56"
~20"
-. l l "

-.15"

_2_6

.40“
.64"
-.29”
-.18"
-.37"

g

.67"
-.26"
-.22"
-.29"

25. L9 E

-.33"
-.l6*"
-.34"

.42"

.56" .38"

 

178

 

Appendix Table B3: Intereorrelations for Middle-Income Group

 

 

M_earl an. 1 .2. § :1 § .9
1.Manager .14 .35
2. Professional .18 .39 -.20‘“"
3. Technical .06 .23 -.10*"' -.12"
4. Sales .11 .31 -.14" -.17"”" -.08**
5. Admin. .16 .36 -.18** -.21** -.10** -.15**
6. Service .08 .28 -12** -.14" -.07** -.11** -.13**
7. Prod/op/repair .27 .44 -.25*"' -.29"”" -.15” -.21"‘* -.26" -.18"
8. Industry‘ .77 .42 .02 .14" .03 .15** .15** .16“
9. White .83 .38 .02 .04 .01 .02 .01 .05
10. Black .10 .30 -.02 -.04 .03 -.06* .02 .04
11. Other race .08 .27 .00 -.01 -.02 .04 .00 .02
12. Sex 1.49 .50 .03 .13** .07" .00 .25" .02
13.Eld/oth care2 1.56 6.25 -.05 -.02 -.01 -.05 .01 .07*
14.#kids<13 .65 .95 -.01 -.O6* -.01 -.01 -.03 .03
15.#ldds13-18 .25 .58 -.02 .02 -.02 .06* -.06* -.01
16. Partner .64 .48 -.04 -.08” -.03 -.01 06* -.01
17. Familial care3 .25 .43 -.03 -.03 -.06"' .05 -.07** .03
18. Age 39.54 11.13 .00 -.02 -.01 -.04 .08** -.07*
l9.Work hrs/wk 46.65 13.35 .09" .01 -.01 .01 -.18" -.06*
20.1rreg.sched. .27 .45 -.09** -.11** -.06* .10** -.10** .18“
21. Autonomy 3.03 .74 .15" .08** -.06* .02 -.04 -.O6"
22. Tenure 7.75 7.98 .03 -.04 .03 -.1 1** .03 -.08**
23. Flex time 1.47 1.46 .15** .05 .02 .09** .03 .03
24. Flex place .24 .68 .10** .15** .05 .04 -.07* -.04
25. Dep. care‘ 3.31 .73 .05 .10** .01 -.11** .07* -.01
26. Sup. support 3.34 2.17 .05 .01 ..04 .04 .06* .00
27. Cult. norms 2.97 1.03 .10** .08" .00 -.01 -.01 -.01
28. Tot environ.’ 9.26 .70 .15** .12** .01 .02 .01 -.03
29. WFC6 2.95 1.03 .02 -.01 .01 .03 .01 -.05
30. FWC7 1.92 .70 -.02 .11" -.03 .03 .01 -.05
31. EE” 3.02 1.01 .01 .02 .00 .03 -.03 -.03

 

 

Pearson zero-order correlations between major study variables. 1. Goods-producing =1, service =2.

2. Hours per week caring for elders and disabled adults 3. Most child care provided by a family member
4. Number of dependent care beneﬁts 5. Total environment 6. WFC = work-to-family conﬂict 7. FWC =
family-to-work conﬂict 8. EE = emotional exhaustion. Due to listwise deletion for missing data, Ns
range ﬁ'om 1,264 to 1,329. "' signiﬁcant at the .05 level. ** signiﬁcant at the .01 level.

179

Appendix Table B3: Intercorrelations for Middle-Income Group

 

 

(Continued)

1 a _9_ 10 11 12 13 14
1. Manager
2. Professional
3. Technical
4. Sales
5. Admin.
6. Service
7. Prod/op/repair
8. Industry -.48**
9. White .01 .04
10. Black .04 .04 -.72"
11. Other race -.02 .01 -.63"““ -.10"
12. Sex -.40"”" .22" -.01 .04 -.O3
13. Eld/oth care .04 -.Ol -.06"‘ .06“ .02 .05
14. # kids < 13 .07“ -.O8"“I -.03 .02 .02 -.O8" -.06‘
15. # kids 13-18 .02 .00 -.O4 .05 .00 .03 .03 -.01
16. Partner .07“ -.05 .10" -.O7* -.O6"‘ -.03 -.O3 .28"
17. Familial care .08" -.10" -.02 .01 .02 -.15" -.06" .66"
18. Age .02 .01 .05 .02 -.O9""" .06" .10** -.24"”"
19.Work hrs/wk .11" -.O9*"' -.07"' .07” .02 -.26“ -.Ol .07“
20. Irreg. sched. .09" -.Ol -.02 .03 -.Ol -.07" .06“ .02
21. Autonomy -.11" .03 .07“ -.11" .03 -.04 -.03 .01
22. Tenure .10" -.O6* -.01 .07“ -.06‘ -.05 .02 -.10"
23. Flex time -.19"”" .09" .03 .03 .01 .00 .04 .04
24. Flex place -.18" .10" -.03 .05 -.OO .01 -.01 .02
25. Dep. care -.10" .13" -.l4” .11" .07" .02 .00 -.04
26. Sup. support -.10" .06“ .06“ -.08" .01 .04 .00 -.O3
27. Cult. norms -.13*"‘ .04 .10" -.07M -.O6* .06“ -.O4 .03
28. Tot. environ. -.24" .14" .01 -.02 .01 .05 -.03 -.03
29. WFC -.02 .00 -.O3 .02 .01 .09" .03 .12
30. FWC -.O6"' .01 -.01 -.Ol .03 .04 .05 .08"
3]. EB .01 -.01 -.Ol .03 -.02 .07“ .02 .01

 

180

 

Appendix Table B3: Intercorrelations for Middle-Income Group

 

 

(Continued)

.1_5. 19 L2 1.3 1.2 L1! 2.1. 22
1.Manager
2. Professional
3.Technical
4. Sales
5. Admin.
6. Service
7. Prod/op/repair
8. Industry
9. White
10. Black
11.0therrace
12. Sex
l3.Eld/othcare
14.#kids<l3
15. # kids 13-18
16. Partner .11"
17. Familial care .07“ .32"
18. Age .09" .07" -.19“
19.Workhrs/wk -.02 -.10** .03 -.O3
20.Irreg.sched. -.04 -.04 .05 -.08" .O7*
21. Autonomy .00 -.Ol .04 .04 .02 -.07*
22. Tenure .01 -.01 —.O6* .45“ .08" -.O6"‘ .03
23.Flextime -.O3 -.05 .01 -.07* .Ol .02 .33“ -.10"
24.Flexplace .03 -.05 -.01 -.09" .05 -.02 .19" -.06"‘
25. Dep. care .03 -.05 -.02 -.01 .02 -.04 .12** .10“
26. Sup. support .04 .04 -.O3 .03 -.05 -.06* .32" -.05
27. Cult. norms .02 .03 .02 -.04 -.O7* -.O9" .25" -.06*
28. Tot. environ. .02 -.O3 -.01 -.05 -.02 -.O6 .42" -.O6"
29.WFC .09" .03 .06‘ -.08** .25" .09" -.15"‘* -.06*
30. FWC .02 -.05 .00 -.04 .07“ -.03 .01 .00
31.EE .00 -.04 .00 -.O7“ .20" .03 -.25** -.04

 

181

 

Appendix Table B3: Intereorrelations for Middle-Income Group

(Continued)

 

 

1.Nﬁunqnn

2. Professional
3.Technkxu

4. Sales
SHAdnun.
support
6.Sendce

7. Prod/op/repair
8.1ndustryl
9.Vthe

10. Black

11. Other race
12.Sex

l3. Eld/oth care2
14. # kids < 13
15. # kids 13-18
l6.Panner
17.Fanuhalcan?
18. Age
l9.Work hrs/wk
20. Irreg. sched.
21. Autonomy
22.TEnun:

23. Flex time
24. Flex place
25.1nqxlcan9
26. Sup. support
27 . Cult. norms
28. Tot. environ.’
29.VVFcf
30.Fch9
31.FI?

.2;

.31"
.l3""'I
.18“I
.22**
.70**
--.10"“'I
.01
-.12“"'I

25

.08**
.09**

.07*

.42"

.03

.14MI

.01

.2;

.18"
.17"
.53"
-.12"
-.03
-.12"

29

.43”
.62"
-.21"”"
-.08“
-.29"

21

.64##
-.27"
-.l9"‘*
~27"

Z§

-.24"'*
-.O7*
-.28“

.22 19.

.40"

.61“ .32"

 

182

 

Appendix Table B4: Intercorrelations for Higher-Income Group

 

Mm 5.2. l 2 .3. 5. § .6
1. Manager .27 .44
2. Professional .28 .45 -.37"'"‘
3. Technical .05 .22 -.14** -.15**
4. Sales .10 .30 -.20** -.21** -.08*
5. Admin. support .13 .34 -.23" -.24" -.09* -.13"
6. Service .05 .21 ..13** -.14** ..05 -.08 -.09*
7. Prod/op/repair .13 .33 -.23" -.24" -.O9"‘ -.13"“" -.15" -.08"'
8. Industry‘ .78 .41 -.08* .11** 09* .10 09* .10*
9. White .85 .36 .01 .05 -.06* .03 -.06 -.01
10. Black .07 .26 .01 -.02 .05 -.08 .00 .03
11. Other race .08 .28 -.03 -05 .03 .03 .07 -01
12. Sex 1.45 .50 .05 .06 -02 -.04 .29** .01
13. Eld/oth care2 1.55 5.32 -.07 -02 .01 -04 .03 .15**
14. # kids < 13 .56 .86 -01 .03 .02 .03 -.O6 .00
15. # kids 13-18 .29 .61 .00 .01 .05 -05 -.03 .02
16. Partner .83 .37 .06 .03 -.01 -.10* .00 .02
17. Familial care3 .21 .41 -.04 -.03 -.04 .05 .03 08*
18. Age 42.33 10.57 .12** 09* -07 .12** -02 -04
l9.Work hrs/wk 49.21 13.63 .15** .03 -.11** -.11** -.20** -.11**
20. Irreg. sched. .22 .41 .10* -.02 .00 .10 -.14** .03
21. Autonomy 3.19 .71 .20** -.02 -.13** .06 -.07 -.07
22. Tenure 9.83 9.02 08* .04 .04 -.10* -11** .00
23. Flex time 2.06 1.44 .28** -.05 -05 09* -12** -03
24. Flex place .36 .48 .16** .03 -.06 .15** -.15** -04
25. Dep. care4 3.57 1.53 09* .04 .07 -.08 -05 -04
26. Sup. support 3.36 .61 .03 .05 -.Ol .01 .03 -.OS
27. Cult. norms 3.07 .67 09* -04 .01 08* .02 -.09*
28. Tot. environ.5 10.11 2.21 .23** -.01 -01 08* -.08 -.08
29. WFC‘5 3.01 1.0 .05 .11** .01 -.11** ..03 -.07
30. FWC7 2.00 .62 .06 .10** -05 -01 -.O6 .00
31.13138 2.98 .97 -04 .07 .00 -07 .00 -.01

 

 

 

Pearson zero-order correlations between major study variables. 1. Goods-producing =1, service =2.

2. Hours per week caring for elders and disabled adults 3. Most child care provided by a family member
4. Number of dependent care beneﬁts 5. Total environment 6. WFC = work-to—farnily conﬂict 7. FWC =
family-to-work conﬂict 8. EE = emotional exhaustion. Due to listwise deletion for missing data, Ns
range ﬁ'om 613 to 640. "‘ signiﬁcant at the .05 level. ” signiﬁcant at the .01 level.

183

Appendix Table B4: Intercorrelations for Higher-Income Group
(Continued)

 

 

Z 3 2 1.9 .1.! ll .12 11

1. Manager

2. Professional

3. Technical

4. Sales

5. Admin.

6. Service

7. Prod/op/repair

8. Industry -.34**

9. White -.01 -.03

10. Black .02 .03 -.65“

11. Other race -.01 .01 -.70"‘* -.08"‘

12. Sex -.26" .23" -.03 -.02 .05

13. Eld/oth care .03 .06 -.06 .08‘ .01 .04

14. # kids < 13 .00 .03 -.02 .05 -.02 -.O7 -.10"

15. # kids 13-18 .02 -.02 .01 .08 -.09* .02 .02 -.O4
16. Partner -.03 -.02 .04 -.02 -.03 .08" .01 .23"
17. Familial care .06 -.02 -.04 .03 .02 -.11” -.04 .66"
18. Age -.O9"' .01 .11" -.02 -.12" -.03 .07 -.22"
l9.Work hrs/Wk .20" -.17" -.01 .03 -.01 -.27" .06 .06
20. Irreg. sched. .12" .01 -.Ol .01 .00 -.O7 .00 .01
21. Autonomy -.O9* -.05 .11" -.06 -.O9"‘ -.l3" -.05 .00
22. Tenure .02 -.02 .03 .04 -.07 -.09"' .04 -.08*
23. Flex time -.22” .01 .03 -.05 .00 -.13“ -.O4 .02
24. Flex place -.18"‘ .06 .04 .04 -.Ol -.06 .04 .06
25. Dep. care -.07 .03 -.O7 .05 .04 -.O3 -.03 .11"
26. Sup. support -.10“ .01 .07 -.01 -.O7 -.01 -.03 -.Ol
27. Cult. norms -.12" -.O4 .17" -.ll" -.12" .03 -.04 -.O3
28. Tot. environ. -.24" .02 .08’ -.05 -.06 -.08 -.03 .06
29. WFC -.O4 .02 -.02 -.02 .04 .07 .03 .12"
30. FWC -.12"”" .02 -.O6 -.02 .10" .06 -.Ol .22"
31. EB .03 .04 -.O8* .04 .07 .15" .05 .05

 

184

 

Appendix Table B4: Intercorrelations for Higher-Income Group

 

 

(Continued)

1.5 .19 11 L3 1.2 19 2.1. 22
1. Manager
2. Professional
3. Technical
4. Sales
5. Admin.
6. Service
7. Prod/op/repair
8. Industry
9. White
10. Black
11. Other race
12. Sex
13. Eld/oth care
14. # kids < 13
15. # kids 13-18
16. Partner .11**
17. Familial care .05 .21"
18. Age .10" .27" ~14"
l9.Work hrs/wk -.04 .02 .10“ .01
20. Irreg. sched. -.O4 -.11“ .04 -.l9" .04
21. Autonomy .00 .06 -.01 .10" .10“ .01
22. Tenure .03 .16" -.O7 .50" .04 -.09‘ .01
23. Flex time .01 .01 .04 .02 .08 .08 .40" -.O4
24. Flex place .00 O4 .08“ .06 .15" .04 .24" .00
25. Dep. care -.03 00 .06 -.01 .12" .11" .13" .13“
26. Sup. support .03 .06 -.05 .08 .01 .01 .30" .01
27. Cult. norms -.O3 04 -.02 .08 -.03 .02 .30" .01
28. Tot. environ. .00 O4 .04 .08 .10“ .09* .46" .05
29. WFC .00 .04 .08‘ -.10"' .25" .00 -.10" -.04
30. FWC .04 .02 .16" -.11"”" .04 .02 -.05 .00
31. EB -.08* -.01 .03 -.18"”" .19" .00 -.23" -.10‘

 

185

 

Appendix Table B4: Intercorrelations for Higher-Income Group

(Continued)

 

 

1. Manager

2. Professional
3. Technical
4. Sales

5. Admin.

6. Service

7. Prod/op/repair

8. Industry

9. White

10. Black

11. Other race
12. Sex

13. Eld/oth care
14. # kids < 13
15. # kids 13-18
16. Partner

17. Familial care

18. Age
l9.Work hrka

20. Irreg. sched.
21. Autonomy
22. Tenure

23. Flex time
24. Flex place
25. Dep. care
26. Sup. support
27. Cult. norms

28. Tot. environ.

29. WFC
30. FWC
31. BE

23.

.29"”‘I
.20"”‘l
.19‘""I
.25“I
.74"“'I
a10**
.00
al7“

2_4

09*
.10*
08*
.44**

.16""'l
-.Ol

2_5

.21"
.18"
.56"
-.Ol
.01
-.07

.2_6_

.45"

.59"
-.25*"‘
-.O8"'
~29"

AZ

.63"
-.26"
”17*.
-.33"

22 2.2 9.2

-.19"‘"'
-.02
-.29"

.43“

.62" .34"

 

186

 

Work-to-Family Conﬂict

Figure 3: Interaction of Income and Total Environment

 

 

 

 

 

Work-to-Family Conﬂict
6
54
11
O f r v r r w r
0.0 2.0 4.0 6.0 8.0 10.0 12.0 14.0

Total Environment

Income Group

#-

3.00

2.00

1.00

16.0

Figure 4: Interaction of Income and Dependent Care Beneﬁt Index

Work-tO-Family Conﬂict

 

 

 

 

Work-to-Family Conﬂict
6
5 .
4 4
3T“ " “heir—:5» —————
V Income Group
2‘ 3.00
l- 2.00
0 ‘ r - 1.00
0.0 2.0 4.0 6.0 8.0
Dependent Care Beneﬁts

187

F amily-to-Work Conﬂict

F amily-to-Work Conﬂict

Figure 5: Interaction of Income and Total Environment

 

 

Family-to-Work Conﬂict
6
5 .
4 .
3 .
Income Group
3.00
1 - 2.00
0 1.00

 

 

 

0.0 20 4:0 6:0 870 10.0 120 1H) 16.0

Total Environment

Figure 6: Interaction of Income and Supervisor Support

Family-to—Work Conﬂict

 

Income Group

————-

 

3.00

 

 

l~ 2.00
0 1.00

 

1 V I r V T j

.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0 4.5

Supervisor Support

188

Family-to-Work Conﬂict

Figure 7: Interaction of Income and Cultural Norms

Family-to-Work Conﬂict

 

 
 

'ﬁ-“_

 

-«
‘1

  

 

 

T Y 1' I I V '

.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0

Cultural Norms

189

Income Group

3.00

2.00

1.00

APPENDIX C: BACKGROUND ON THE NATIONAL STUDY OF THE

CHANGING WORKFORCE

190

APPENDIX C: BACKGROUND ON THE
NATIONAL STUDY OF THE CHANGING WORKFORCE

Beginning in 1969, the Department of Labor funded three national surveys of the
United States workforce as part of the Quality of Employment Survey (QES). The last
survey in this series, which was conducted in 1977, marked the ﬁrst time that research on
a large, representative sample of US. workers collected information about not only the
work lives of employees, but their personal lives as well. Following the halting of the
QES program in 1977 there were no large-scale, nationally representative surveys of
workers on life on and off the job for 15 years, although numerous small studies were
conducted by researchers interested these areas.

In 1990, the Families and Work Institute obtained private support for the National
Study of the Changing Workforce (NSCW) as an ongoing research program. The
institute's program is more explicit and comprehensive than the QES in addressing issues
related to both work and personal life. It also reﬂects a strong business perspective, in
addition to the broader social and economic perspectives that shaped the QES. Financial
support for the core NSCW research program comes from private-sector sources. The
1997 survey was cosponsored by KPMG Peat Marwick LLP (lead sponsor); Allstate
Insurance Company; The Boeing Company; Ceridian; Citibank; The Commonwealth
Fund; Fannie Mac; The GE Fund; IBM Corporation; Johnson & Johnson; Merck & Co.,
Inc.; Mobil Corporation; NCR Corporation; Salt River Project; and Xerox. The 1992
survey was co-sponsored by Salt River Project (lead sponsor); Allstate Insurance
Company; American Express Company; AT&T; Commonwealth Fund; DuPont

Company; General Mills Foundation; IBM Corporation; Johnson & Johnson; Levi

I91

Strauss & Co.; Merck & Co., Inc; Mobil Corporation; Motorola, Inc.; The Rockefeller
Foundation; and Xerox.

The NSCW surveys representative samples of currently employed US. workers
every ﬁve years, with ﬁndings on important and timely issues released during the
intervening years through institute reports, publication in academic journals, books,
media coverage, and presentations to audiences of private- and public-sector decision-
makers. The ﬁrst NSCW survey was conducted in 1992; the second, in 1997, and the
third in 2002. Sample sizes are large enough to support analyses of many subgroups of
interest: 1997 NSCW total sample = 3,552 (2,877 wage and salaried workers); 1992
NSCW total sample = 3,718 (2,958 wage and salaried workers). The 1977 QES total
sample was 1,515 (1,298 wage and salaried workers).

AS was the case with the QES, the NSCWF studies provide information on a wide
range of topics, allowing researchers to study many different relationships in the areas of
work, family, and community, with sample sizes larger and participants more diverse
than many researchers who rely on their resources can manage. The Families and Work
Institute publishes a summary of the results of each study with Simple data analyses
describing the main ﬁndings, and makes the data available to academic researchers.
Following this discussion is a list of published studies in this area that rely on data from
the 1992 and 1997 NSCWF.

As should be clear from the titles, these studies use the data to answer questions
about a variety of different issues. Hundley (2001a) examined differences in job
satisfaction between the self and organizationally employed. Overall, self-employed

workers were more satisﬁed than those who worked for others, but this was largely

192

because they tended to have more autonomy, ﬂexibility, job security, and opportunities to
use their Skills. Self-employed workers were not more satisﬁed than employed workers
whose jobs provided them with the same advantages. Hundley (2001b) also used NSCW
data to explore gender differences in self employment, ﬁnding that men were more likely
to choose to work for themselves because they could earn more money, while women
were often seeking greater ﬂexibility and more time with their families. Moen and Yu
(1992) examined a subsample of the 1992 NSCW composed of those in dual-career
families to try to uncover What made them feel most successful at home, at work, and in
balancing the two. Life stage had important effects; older workers were more likely to
feel successful.

Two studies share this dissertation’s interest in ways that employers can help
workers cope with demands in two domains. Galinsky, Bond, and Friedman (1996)
examined data from the 1992 survey, ﬁnding that parents overall experienced more stress
and overall conﬂict between work and family, and also reported having more diﬂiculty
coping than nonparents. Galinsky er al. included a Wide range of variables in a
subsequent regression analysis that sought to examine what might inﬂuence these
outcomes among parents only, ﬁnding that, while informal support appeared to help,
access to formal beneﬁts of all kinds did not. Roehling et a1. (2001) focused on how
employee loyalty is affected by life stage and work/life beneﬁts using the 1992 NSCW.
Employees at all stages of life were more loyal when they had greater ﬂexibility, and
dependent care assistance was associated with greater loyalty among mothers of school-

age children.

193

Like these studies, this dissertation reﬂects an interest in the intersection of work
and family life. Building on the work of these researchers and others, it uses previous
ﬁndings and the conservation of resources model to identify some of the resources and
demands at work and at home most likely to affect work-to-family conﬂict, family-to-
work conﬂict, and emotional exhaustion. Focusing approaches organizations can use to
create a more family-supportive environment, it ﬁnds that such support is associated with
less work-to-family conﬂict and, extending previous ﬁndings, less family-to-work
conﬂict and burnout. It takes a different approach than the Roehling et al. (2001) and
Galinsky et a1. (1996) studies by examining the how the effects of a supportive
environment and other factors may differ by income group, and is the ﬁrst study to
examine Whether conﬂict between work and family mediates the effects of a supportive
work environment on a work outcome. By asking new questions of the data collected for
the NSCWF, it joins previous NSCW studies in adding to our understanding of the

relationship between work and home.

194

Previously Published Studies Using N SCWF Data

Galinsky, E., Bond, J. and Friedman, DE. (1996). The role of employers in addressing
the needs of employed parents. Journal of Social Issues, 52, 3, 111-136. (1992 NSCWF)

Hundley, G. (2001a) Why and when are the self-employed more satisﬁed with their
work? Industrial Relations, 40 (2), 293-316. (1997 NSCWF)

Hundley, G. (2001b). Domestic division of labor and self/organizationally employed
differences in job attitudes and earnings. Journal of Family and Economic Issues, 22 (2),
121-139. (1997 NSCWF)

Moen, P., and Yu, Y. (1999). Having it all: Overall work/life success in two earner
families. In T. Parcel (ed) Research in the Sociology of Work: Vol. 7. Greenwich, CT:
JAI Press. (1992 NSCWF)

Roehling, P.V., Roehling, M.V., and Moen, P. (2001). The relationship between work-

life policies and practices and employee loyalty: a life course perspective. Journal of
Family and Economic Issues, 22 (2), 141-170. (1992 NSCWF)

I95

REFERENCES

196

REFERENCES

Adams, G.A., Kings, L.A., & King, D.W. (1996). Relationships of job and family
involvement, family social support, and work-family conﬂict with job and life
satisfaction. Journal of Applied Psychology, 81, 411-420.

Allen, TD. (2001). F amily-supportive work environments: the role of organizational
perceptions. Journal of Vocational Behavior, 58, 414-435.

Allen, T.D., Herst, D.E.L., Bruck, CS, and Sutton, M. (2000). Consequences
Associated with work-to-family conﬂict: a review and agenda for future research.
Journal of Occupational Health Psychology, 5 (2), 278-308.

mm, J. W., Gibeau, J .L., and Larson, P.J. (1990). Working families and eldercare: a
national perspective in an aging America Social Work 35: 405-411

Ayree, S. (1992). Antecedents and outcomes of work-family conﬂict among married
professional women: evidence from Singapore. Human Relations, 45, 813-837.

Ayree, S. (1993). Dual-earner couples in Singapore: An examination of work and
nonwork sources of their experienced burnout. Human Relations, 46 (12), 1441-1459.

Ayree, S., Fields, D., and Luk, V. (1999). A cross-cultural test of a model of the work-
family interface. Journal of Management, 25, 491-51 1.

Bacharach, S.B., Bamberger, P., & Conley, S. (1991). Work-home conﬂict among nurses
and engineers: mediating the impact of role stress on burnout and satisfaction at work
Journal of Organizational Behavior, 12, 39-53.

Baltes, BB, Briggs, T.E., Huff, J .W., Wright, J .A., and Neuman, GA. (1999). Flexible
and compressed workweek schedules: A meta-analysis of their effects on work-related
criteria. Journal of Applied Psychology, 84 (4), 496-513.

Barnett, RC, and Hyde, J .S., (2001). Women, men, work, and family: an expansionist
theory. American Psychologist, 56 (10), 781-796.

Belanger, F. (1999). Workers’ propensity to telecommute: an empirical study.
Information & Management, 35, 139-153.

Block, R (2002). Personal communication, May 6, 2002.

Bohen, J .J ., and Viveros-Long, A. (1981). Balancing Jobs and Family Life.
Philadelphia: Temple University Press.

Bond, J .T., Galinsky, and Swanberg, J .E. (1998) The 1997 National Study of the
Changing Workforce. New York: Families and Work Institute.

197

Bond, J .T. and Galinsky, E. (1998). The 1998 Business Work-Life Study. New York:
Families and Work Institute.

Bretz, R.D., Boudreau, J .W., & Judge, TA. (1994). Job search behavior of employed
managers. Personnel Psychology, 47, 275-301.

Burke, R.J., (1988). Some antecedents and consequences of work-family conﬂict
Journal of Social Behavior and Personality, 3, 287-302.

Burke, R.J. (1993). Toward an understanding of psychological burnout among police
ofﬁcers. Journal of Social Behavior and Personality, 8, 425-43 8.

Bussing, A. (1998). Teleworking and quality of life. In Jackson, RI and Van Der
Wielen, J. M (eds). T eleworking: International Perspectives ﬁ'om Telecommuting to the
Virtual Organization. London: Routledge, pp 144-165.

Broman, CL. (1988). Household work and family-life satisfaction of blacks. Journal of
Marriage and the F amily, 50 (3) 743-748.

Broman, CL. (1993). Race differences in marital well-being. Journal of Marriage and
the Family, 55: 724-732.

Broman, CL, Hamilton, V L., Hoffman, W. S, and Mavaddat, R. (1995). Race, gender,
and the response to stress: Autoworkers' vulnerability to long-term unemployment.
American Journal of Community Psychology, 23 (6) 813-832.

Carlson, D.S., Kacmar, KM, and Williams, LJ. (2000). Construction and initial
validation of a multidimensional measure of work-family conﬂict. Journal of Vocational
Behavior, 56, 249-276.

Cohen J. and Cohen P. (1983 ). Applied Multiple Regression and Correlation Analysis for
the Behavioral Sciences. Hillsdale, NJ: Erlbaum.

Cooke, R.A. & Rousseau, BM. (1984). Stress and strain from family roles and work-
role expectations. Journal of Applied Psychology, 69, 252-260.

Crouter, A. C. (1984). Spillover from family to work: the neglected side of the work-
family interface. Human Relations, 37 (6), 425-442.

Dalton, DR, and Mesch, DJ. (1990). The impact of ﬂexible scheduling on employee
attendance and turnover. Administrative Science Quarterly, 35, 370-3 87.

Demerouti, E., Bakker, A., Nachreiner, F., and Schaufeli, W. (2001). The job demands-
resources model of burnout. Journal of Applied Psychology, 86 (3), 499-512.

Drory, A. & Sharnir, B. (1988). Effects of organizational and life variables on job
satisfaction and burnout. Group & Organization Studies, 13, 441-455.

198

Dunham, R.B., Pierce, J .L., and Castaneda, MB. (1987). Alternative work schedules: to
ﬁeld quasi-experiments. Personnel Psychology, 40, 215-242.

Duxbury, L. Higgins, C. and Neufeld, D. (2001). Telework and the balance between
work and family: Is telework part of the problem or part of the solution? In Johnson, N.J.

(ed) Telecommuting and Virtual Ojﬁces: Issues and Opportunities. Hershey, pp. 218-
255.

Feldrnan, DC, and Gainey, T.W. (1997). Patterns of telecommuting and their
consequences: framing the research agenda. Human Resource Management Review, 7
(4), 369-388.

Frone, M.R., Russell, M. and Barnes, GM. (1996). Work-family conflict, gender, and
health-related outcomes: A study of employed parents in two community samples.
Journal of Occupational Health Psychology, 1, 57-69.

Frone, M.R., Yardley, J .K., and Markel, KS. (1997). Developing and testing an
integrative model of the work-family interface. Journal of Vocational Behavior, 50, I45-
167.

Galinsky E., and Bond, J. (1988). The 1998 Business Work-Life Study: a Sourceboolc
NY: Families and Work Institute.

Galinsky, E., Bond, J. and Friedman, DE. (1996). The role of employers in addressing
the needs of employed parents. Journal of Social Issues, 52, 3, 111-136.

Goff, S.J., Mount, .K., & Jamison, KL. (1990). Employer supported child care,
work/family conﬂict, and absenteeism: a ﬁeld study. Personnel Psychology, 43, 793-
809.

Golembiewski, R.T, Munzenrider, RF, and Stevenson, J .G. (1986). Stress in
Organizations: Toward a Phase Model of Burnout. New York: Praeger.

Goldberg, W.A., Greenberger, E., Koch-Jones, J ., O’Neil, R., and Hamill, S. (1989).
Attractiveness of child care and related employer-supported beneﬁts and policies to
married and single parents. Child and Youth Care Quarterly, 18(1), 23-37.

Good, L.K., Sisler, G. & Gentry, J .W. (1988). Antecedents of turnover intentions among
retail management personnel. Journal of Retailing, 64, 295-314.

Googins, B. & Burden, D. (1987). Vulnerability of working parents: balancing work and
home roles. Social Work, July-August 1987, 295-299.

Grandey, A. and Cropanzano R. (1999). The conservation of resources model applied to
work-family conflict and strain. Journal of Vocational Behavior, 54, 350-370.

199

Greenberger, E., Goldberg, W.A., Hamill, S., O’Neil, R., and Payne, CK. (1989).
Contributions of a supportive work environment to parents’ well-being and orientation to
work. American Journal of Community Psychology, 17 (6), 755-783.

Greenglass, ER, and Burke, R.J. (1988). Work and family precursors of burnout in
teachers: sex differences. Sex Roles, 18 (3/4), 215-229.

Greenhaus, J. & Beutell, N. 1985. Sources of conﬂict between work and family roles.
Academy of Management Review, 10: 76-88.

Grover, 8L. and Crooker, K.J. (1995). Who appreciates family-responsive human
resource policies: the impact of family-friendly policies on the organizational attachment
Of parents and non-parents. Personnel Psychology, 48, 271-288.

Grzwacz, J .C. (2000). Work-family spillover and health during midlife: is managing
conﬂict everything? American Journal of Health Promotion, 14 (4), 236-243.

Grzywacz, J .C., Almeida, OM, and McDonald, DA. (2002). Work-family spillover and
daily reports of work and family stress in the adult labor force. Family Relations, 51, 28-
36.

Gutek, B.A. Searle, S., and Klepa, L. (1991). Rational versus gender role explanations
for work-family conﬂict. Journal of Applied Psychology, 76, 560-568.

Hartman, R.I., Stoner, CR, and Arora, R. (1991). An investigation of selected variables
affecting telecommuting productivity and satisfaction. Journal of Business and
Psychology, 6 (2), 207-225.

Higgins, CA, and Duxbury, LE. (1992). Work-family conﬂict: a comparison of dual-
career and traditional-career men. Journal of Organizational Behavior, 13, 389-41 1.

Hill, J. E., Miller, B. C., Weiner, S. P. and Colihan, J. (1998). Inﬂuences of the virtual
ofﬁce on aspects of work and work/life balance. Personnel Psychology, 41, 667-683.

Hobfall, S. E. (1989). Conservation of resources: A new attempt at conceptualizing
stress. American Psychologist, 44, 513-524.

Hundley, G. (2001). Domestic division of labor and self/organizationally employed
differences in job attitudes and earnings Journal of Family and Economic Issues, 22 (2),
121-139.

Iverson, R.D., Olekalns, M. and Erwin, RI (1998). Affectivity, organizational stressors,

and absenteeism: a causal model of burnout and its consequences. Journal of Vocational
Behavior, 52, 1-23.

200

Izraeli, D.N. (1988). Burning out in medicine: a comparison of husbands and wives in
dual-career couples. Journal of Social Behavior and Personality, 3, 329-346.

James, LR, and Brett, J .M. (1984). Mediators, moderators, and tests for mediation.
Journal of Applied Psychology, 69(2), 307-321.

Judge, T.A., Boudreau, J .W. & Bretz, RD. (1994). Job and life attitudes of male
executives. Journal of Applied Psychology, 79, 767-782.

Katz, D. Kahn, R. L. (1978). The Social Psychology of Organizations. New York:
Wiley.

Kopelman, R.E., Greenhaus, J .H., & Connolly, T.F. (1983). A model of work, family,
and interrole conﬂict: a construct validation study. Organizational Behavior and Human
Performance, 32, 1980215.

Kossek, BE. (1990). Diversity in child care assistance needs: employee problems,
preferences, and work-related outcomes. Personnel Psychology, 43, 769-791.

Kossek, E.E., Barber, AH, and Winters, D. (1999). Using ﬂexible schedules in the
managerial world: the power of peers. Human Resource Management, 38, (1), 33-46.

Kossek, E.E., & Nichols, V. (1992). The effects of on-site child care on employee
attitudes and performance. Personnel Psychology, 45, 485-509.

Kossek, E..,E Noe, RA, and DeMarr, BJ. (1999). Work-family role synthesis:
Individual and organizational determinants International Journal of Conﬂict Resolution,
10, 102-129.

Kossek, E. E. and Ozeki, C. (1998). Work-family conﬂict, policies and the job-life
satisfaction relationship: a review and directions for OB/HR research Journal of Applied
Psychology, 83(2), 139-149.

Kossek, BE, and Ozeki, C. (1999). Bridging the work-family policy and productivity
gap: a literature review. Community, Work, and Family, 2 (1), 7-32.

Krausz, M. and Freibach, N. (1983). Effects of ﬂexible working time for employed
women upon satisfaction, strains, and absenteeism. Journal of Occupational Psychology,
56, 155-159.

Lambert, SJ. (1990). Processes linking work and family: a critical review and research
agenda. Human Relations, 43 (3), 239-257.

201

Lambert, SJ. (1995). An investigation of workers' use and appreciation of supportive
workplace policies. Academy of Management Journal, Best Papers Proceedings 1995,
136-142.

Lambert, SJ. (2000). Added beneﬁts: the link between work-life beneﬁts and
organizational citizenship behavior. Academy of Management Journal, 43 (5), 801-815.

Lee, RA. (1983). Flextime and conjugal roles. Journal of Occupational Behavior, 4,
297-315.

Lee, R. T., and Ashforth, B. E. (1993). A longitudinal study of burnout among
supervisors and managers: comparisons between the Leiter and Maslach (1988) and
Golembiewski et al. (1986) models. Organizational Behavior and Human Decision
Processes, 54: 369-398.

Lee, R. T., and Ashforth, B. E. (1996). A meta-analytic examination of the correlates of
the three dimensions of job burnout. Journal of Applied Psychology, 31, 123-133.

Leiter, MP. (1990). The impact of family resources, control coping, and skill utilization
on the development of burnout: A longitudinal study. Human Relations, 1 I, 1067-1083.

Leiter, M.P., & Durup, M.J. (1996). Work, home and in-between: a longitudinal study of
spillover. Journal of Applied Behavioral Science, 32, 29-47.

Lyness, SS, and Thompson, DE. (1997). Above the glass ceiling? A comparison of
matched samples of female and male executives. Journal of Applied Psychology, 82,
359-375.

Marks, SR (1977). Multiple roles and role strain: some notes on human energy, time
and commitment. American Sociological Review, 42, 921-936.

Martire, L.M. and Stephens, MAP. (2003). Juggling parent care and employment
responsibilities: the dilemmas of adult daughter caregivers in the workforce. Sex Roles,
48 (3/4), 167-173.

Maslach, C. and Jackson S. (1981). The measurement of experienced burnout Journal
of Occupational Behavior, 2, 99-113.

Maslach, C. and Jackson S. ( 1981b). The Maslach Burnout Inventory, Consulting
Psychologists, Palo Alto.

Maslach, C. and Jackson S. (1985). The role of sex and family variables in burnout Sex
Roles, 12 (7/8) 837-851.

Maslach, C. (1982). Burnout: the Cost of Caring. Englewood Cliffs, NJ: Prentiss Hall.

202

McGuire, J .B. and Liro, J .R. (1986). Flexible work schedules, work attitudes, and
perceptions of productivity. Public Personnel Management, 15 (1), 1986.

Milkovich, G.T., & Gomez, LR. (1976). Day care and selected employee work
behaviors. Academy of Management Journal, 19, 1 1-1 15.

Mirchirandi, K. (1998). Protecting the boundary: Teleworker insights on the expansive
concept of work. Gender & Society, 12 (2), 168-187.

Mize, J. and Freeman, LC. (1989). Employer-supported child care: assessing the need
and potential support. Child & Youth Care Quarterly, 18 (4). 289-301.

Moen, P., and Yu, Y. (1999). Having it all : Overall work/life success in two earner
families. In T. Parcel (ed) Research in the Sociology of Work: Vol. 7. Greenwich, CT:
JAI Press.

Narayanan, V.K. and Nath, R. (1984). The inﬂuence of group cohesiveness on some
changes induced by ﬂexitime: a quasi-experiment Journal of Applied Behavioral
Science, 20 (3), 265-276.

Netemeyer, R. Boles, J ., and McMurrian, R. (1996.) Development and validation of
work-family conﬂict and family -work conﬂict scales. Journal of Applied Psychology.
81: 400-409.

Orbuch, TL and Custer, L. (1995). The social context of married women's work and its
impact on black husbands and white husbands. Journal of Marriage and the Family, 5 7
(2), 333-341.

Parasuraman, S., Greenhaus, J .H., & Granrose, CS. (1992). Role stressors, social
support, and well-being among two-career couples. Journal of Organizational Behavior,
13, 339-356.

Perry-Smith, J .E., and Blum, TC. (2000). Work-family human resource bundles and
perceived organizational performance. Academy of Management Journal, 43 (6), 1107-
1117.

Pierce, J.L., and Newstrom, J .W. (1982). Employee responses to ﬂexible work
schedules: an inter-organization, inter-system comparison. Journal of Management, 8
(1), 9-25.

Pierce, J .L., and Newstrom, J .W. (1983). The design of ﬂexible work schedules and
employee responses: relationships and process. Journal of Occupational Behavior, 4,
247-262.

Quinn, R. P., and Staines, G.L. (1979). The 1977 Quality of Employment Survey. Ann
Arbor, Michigan: Institute for Social Research.

203

Raghuram, S. Garud, R., Wiesenfeld, B. and Gupta, V. (2001). Factors contributing to
virtual work adjustment. Journal of Management, 27, 383-405.

Ray, E.B., & Miller, K. (1994). Social support, home/work stress, and burnout: who can
help? Journal of Applied Behavioral Science, 30, 357-373.

Roehling, P.V., Roehling, M.V., and Moen, P. (2001). The relationship between work-
life policies and practices and employee loyalty: a life course perspective. Journal of
Family and Economic Issues, 22 (2), 141-170.

Rothausen, T.J. (1994). Job satisfaction and the parent worker: the role of ﬂexibility and
rewards. Journal of Vocational Behavior, 44, 317-336.

Rothausen, T. J. (1999). ‘Family’ in organizational research: a review and comparison of
deﬁnitions and measures. Journal of Organizational Behavior, 20, 817-836.

Rothbard, N. P. (2001). Enriching or depleting? The dynamics of engagement in work
and family roles. Administrative Science Quarterly, 46: 655-685.

Salyers, M. P. and Bond, GR. (2001). An exploratory analysis of racial factors in staff
burnout among assertive community treatment workers. Community Mental Health
Journal, 37 (5) 393-404.

Scharlach A., and Boyd, S.L. (1989). Caregiving and employment: results of an
employee survey. The Gerontologist, 29, 382-87.

Shinn, M. Wong, N.S., Sirnko, RA, and Ortiz-Torres, B. (1989). Promoting the well-
being of working parents: coping, social support, and ﬂexible job schedules. American
Journal of Community Psychology, 17 (1), 31-55.

Staines, G.L., Pottick, KI, and Fudge, DA. (1986). Wives’ employment and husbands’
attitudes toward work and life. Journal of Applied Psychology, 71 (I), 118-128.

Tennstedt, S.L. and Gonyea, J .G. (1994). An agenda for work and eldercare research
Research on Aging, 16, 85-108.

Thomas, L.T., & Ganster, DC. (1995). Impact of family-supportive work variables on
work-family conﬂict and strain: a control perspective. Journal of Applied Psychology,
80, 6-15.

Thompson, C.A., Beauvais, LL, and Lyness, KS. (1999). When work-family beneﬁts
are not enough: The inﬂuence of work-family culture on beneﬁt utilization,

organizational attachment, and work-family conﬂict. Journal of Vocational Behavior,
54, 392-415.

204

US Bureau of the Census (1998). Current Population Reports, Series P23-194,
Population Proﬁe of the United States: 199 7. Washington, DC: Author.

US Census Bureau (2000). Statistical Abstract of the United States. Washington, DC:
Author.

Voydanoff, P. (1988). Work and family: a review and expanded conceptualization.
Journal of Social Behavior and Personality, 3 (4), 1-22.

Voydanoff, P. (2002). Linkages between the work-family interface and work, family,
and individual outcomes: an integrative model. Journal of Family Issues, 23 (1), 138-
164.

Wagner, D.L. and Hunt, GO. (1994). The use of workplace eldercare programs by
employed caregivers. Research on Aging, 16, 69-84.

West, H. (2000). Comments on the rise of computing technology and working at home:
data from the current population survey supplements In Telework and the New
Workplace of the 21" Century, US Department of Labor, available at
http://www.dol.gov/aspltelework/tochtml.

Williams, KI, and Alliger, GM. (1994). Role stressors, mood spillover, and perceptions
of work-family conﬂict in employed parents. Academy of Management Journal, 37 (4),
837-868.

Wright, T. A. and Cropanzano, R (1998). Emotional exhaustion and a predictor of job
performance and voluntary turnover. Journal of Applied Psychology, 83, (3 ), 486-493.

Youngblood, S.A., & Chambers-Cook, K. (1984). Child care assistance can improve
employee attitudes and behavior. Personnel Administrator, February 1984, 45-95.

205