Does Work Pay Psychologically as Well as Economically? The Role of Employment in
Predicting Depressive Symptoms and Parenting among Low-Income Families
Author(s): C. Cybele Raver
Source: Child Development, Vol. 74, No. 6 (Nov. - Dec., 2003), pp. 1720-1736
Published by: Blackwell Publishing on behalf of the Society for Research in Child Development
Stable URL: http://www.jstor.org/stable/3696300 .
Accessed: 30/03/2011 11:14
Your use of the JSTOR archive indicates your acceptance of JSTOR's Terms and Conditions of Use, available at .
http://www.jstor.org/page/info/about/policies/terms.jsp. JSTOR's Terms and Conditions of Use provides, in part, that unless
you have obtained prior permission, you may not download an entire issue of a journal or multiple copies of articles, and you
may use content in the JSTOR archive only for your personal, non-commercial use.
Please contact the publisher regarding any further use of this work. Publisher contact information may be obtained at .
http://www.jstor.org/action/showPublisher?publisherCode=black. .
Each copy of any part of a JSTOR transmission must contain the same copyright notice that appears on the screen or printed
page of such transmission.
JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of
content in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new forms
of scholarship. For more information about JSTOR, please contact [email protected].
Blackwell Publishing and Society for Research in Child Development are collaborating with JSTOR to digitize,
preserve and extend access to Child Development.
http://www.jstor.org
Child Development,November/December 2003, Volume 74, Number 6, Pages 1720-1736
Does Work Pay Psychologically as Well as Economically?The Role of
Employment in Predicting Depressive Symptoms and Parenting
Among Low-Income Families
C. CybeleRaver
This longitudinal study examined quantity and quality of maternalemployment as predictorsof depressive
symptoms and parentingstyle in a sample of 94 low-income motherswhose 4-year-oldchildrenwere enrolled
in Head Startat baseline. Results suggest that answers to the question of whether work "pays"are complex:
Findingssuggest some benefits of greateremploymentparticipationwhile also indicatingthat women holding
lower prestige jobs experiencedincreasesin their use of negative parentingstyle, net of baseline demographic
and psychological characteristics.Sparse evidence for selection processes was found, with cohabitationand
maternaldepressive symptoms modestly predictiveof subsequentmaternalemployment.Implicationsof these
findings for welfare reformand educationallyrelated policies for low-income families are discussed.
Since the passage of the Personal Responsibility and
Work Opportunity Reconciliation Act in 1996, lowincome women have faced increasing policy pressure to work outside the home. An initial wave of
research on the effects of welfare reform focused
primarily on whether programs were successful in
reducing welfare caseloads and in moving lowincome women from reliance on public assistance
into work. A growing second wave of research has
addressed questions of the impact of these employment programs on families' psychosocial well-being
and children's social, behavioral, and academic
outcomes (Chase-Lansdale & Brooks-Gunn, 1995;
Huston et al., 2001; Morris, 2002). For example,
recent findings regarding the positive effects of a
wide range of welfare reform demonstration projects
suggests that benefits accrued to children through
families' use of program-supported child care
(Huston et al., 2001; Morris, 2002). But families'
C. Cybele Raver, Irving B. Harris Graduate School of Public
Policy Studies, University of Chicago.
This research was made possible through a Faculty Scholars
Award from the William T. Grant Foundation, as well as through
support from the National Science Foundation (SBR #961 6454)
and the McCormick Tribune Foundation. Thanks to J. Lawrence
Aber, Lindsey Chase-Lansdale, Thomas DeLeire, Greg Duncan,
Jacqueline Goldsby, Ariel Kalil, Mary Clare Lennon, Phyllis Moen,
and Jane Waldfogel for their helpful comments on previous drafts
of this paper. Thanks to Krista Ham, Mary Spagnola, Paul
Papierno, Heather Warren, Breeze Luetke-Stahlman, my Cornell
University undergraduate research assistants, and the families
and Head Start staff who made this research possible.
Correspondence concerning this article should be addressed to
C. Cybele Raver, Director, the Center for Human Potential and
Public Policy, Irving B. Harris Graduate School of Public Policy
Studies, University of Chicago, 1155 E. 60th Street, Chicago, IL
60637. Electronic mail may be sent to [email protected].
participation in a host of additional work, parenting,
and social service supports was not found to significantly affect their levels of depressive symptoms, or
their use of more effective parenting styles (Huston
et al., 2001; Morris, 2002).
Interestingly, these findings open debate to broader questions regarding whether low-wage work can
reasonably be expected to have a positive or
negative effect on parenting among poor families
(Brady-Smith, Brooks-Gunn, Waldfogel, & Fauth,
2001). Increasingly, a number of investigators have
argued for the utility of asking whether low-income
families benefit materially and psychologically from
increased maternal employment in contexts outside
of experimental welfare-reform-related research
projects (Jackson, Brooks-Gunn, Huang, & Glassman, 2000). Although questions regarding the
impact of maternal employment on child outcomes
have largely been addressed among middle-income
and affluent families (Desai, Chase-Lansdale, &
Michael, 1989; Harvey, 1999), less is known regarding the impact of maternal employment on the
psychological and emotional well-being of lowincome families.
In addition, many low-income families are served
through a second avenue of intervention, namely,
early educational programs that may facilitate
maternal employment as well as provide young
children with preschool experience (Brooks-Gunn,
Berlin, & Fuligni, 2000; Tout, Zaslow, Papillo, &
Vandivere, 2001). Yet, most developmental research
on early educational interventions such as Head
Start have not generally reported on mothers'
( 2003 by the Society for Research in Child Development, Inc.
All rights reserved. 0009-3920/2003/7406-0010
Work and Parenting for Low-IncomeMothers
employment, nor on whether any program effects
may be mediated or moderated by increases in
maternal employment. To address this gap, this
paper briefly reviews extant research on the impact
of maternal employment on family functioning
among low-income families with young children. It
then goes on to report tests of hypotheses regarding
whether work is predictive of improvements or
decrements in low-income, Head Start-enrolled
mothers' depressive symptoms and parenting style
over a 2-year period.
Good,
HypothesizedEffectsof MaternalEmployment:
Bad,or Indifferent?
Working in a financially and psychologically
rewarding job may have clear benefits for lowincome mothers and their families. Prior research
using large nationally representative data sets suggests that increases in employment and exits from
welfare are associated with improvements in poor
families' income, home environments, and long-term
benefits for children (Garrett, Ng'andu, & Ferron,
1994; Meyer & Cancian, 1998; Zill et al., 1995). Young
mothers participating in self-sufficiency activities
such as work, education, or training have been
found to be less emotionally negative and less
controlling than mothers who did not participate in
such activities (Aber, Brooks-Gunn, & Maynard,
1995; Zaslow, Tout, Smith, & Moore, 1998). At first
glance, this suggests that higher levels of low-income
mothers' participation in the workforce should be
associated with improvements in parenting.
To explain these potential improvements, several
mechanisms of work's influence on family functioning have often been suggested. First, employment
may have salutary effects on mothers' emotional and
psychological well-being, providing families with a
sense of daily routine and women with a positive set
of socially organizing and supportive experiences
outside the home (Wilson, 1987). Second, mothers'
employment might be expected to indirectly affect
family well-being by increasing family income,
which might in turn alleviate the psychological
stress of trying to provide for one's family with
limited resources. Lower psychological distress, in
turn, may improve adults' ability to interact in
warm, supportive, and emotionally positive ways
with their children (Conger et al., 1992; Desai et al.,
1989; McLoyd, 1990). It is important to qualify such a
positive view of maternal employment's hypothesized benefits: Increasing family income and decreasing poverty-related stressors may be a difficult
task for low-income working women to accomplish
1721
when their wage rates are low (Ellwood, 1988;
Jackson et al., 2000). With this qualification in mind,
much research and antipoverty policy is premised
on the hypothesis that increased maternal employment will at least modestly benefit poor families.
On the other side of the debate, some recent
research suggests that the effects of low-income
workforce entry on family functioning may be
negative rather than positive. The psychological
benefits of work may be outweighed by its psychological costs: Some low-income women may face
overwhelming work demands at menial, unstable
jobs with little control over their work environment
and little flexibility or support in meeting those
demands (Karasek & Theorell, 1990; Long, 1998;
Rosenfield, 1989). As low-income women face greater work stresses, their ability to parent successfully
may be increasingly compromised (Wilson, Ellwood,
& Brooks-Gunn, 1995). Importantly, employment in
jobs that are more menial and are lower in complexity appears to have negative effects on child
outcomes among educationally disadvantaged, lowerincome families (Menaghan & Parcel, 1995; Parcel &
Menaghan, 1994). One possible mechanism for this
finding may be that women who work longer hours
in more menial jobs may become more depre-ssed as
they struggle to find adequate child care, deal with
the demands of low-wage work, and have
less time to complete routine household and caregiving
tasks (Edin & Lein, 1997; Wilson et al., 1995; Zaslow
et al., 1998). This may not bode well for parenting,
given that prior research suggests that mothers
facing higher levels stress and depression show less
responsiveness and lower consistency in setting
limits (Downey & Coyne, 1990; Wilson et al., 1995).
A third possibility is that work has no demonstrable effects on mothers' depressive symptoms and
parenting. Recent reanalyses of the National Longitudinal Survey of Youth (NLSY) considered the
effects of timing, extensiveness, and intensity of
mothers' work involvement on children and found
few significant effects (Harvey, 1999). Similarly,
results from studies of mothers leaving welfare for
work suggest that, although the benefits of mothers'
participation in a host of employment-related services are open to question, it appears that participation is not, at least, causing serious harm (Duncan &
Chase-Lansdale, 2001; Dunifon, Kalil, & Danziger,
2003; Morris, Huston, Duncan, Crosby, & Bos, 2001).
Hypothesized Effects of Work,Reconsidered
How is it that some studies conclude that work
has a positive effect, other studies conclude that
1722
Raver
work has a negative effect, and a third set of studies
find no effect? Differences in findings may be due
to the ways in which women's personal and household characteristics may serve as unmeasured or
omitted variables that are predictive of both their
workforce participation outside the home and their
use of differing styles of parenting within the home
(Waldfogel, Han, & Brooks-Gunn, 2002). Mothers'
educational level, age, and ethnic minority status
have been consistently identified as unmeasured
family background characteristics that are likely to
be associated with women's parenting styles and
their access to and participation in more psychologically and financially rewarding jobs (Brady-Smith
et al., 2001). The effects of juggling work and family
responsibilities are likely to be different for mothers
residing in one- versus two-parent households as
well, with single mothers experiencing significantly
higher financial, psychological, and caregiving strain
than mothers in two-parent households (Ali &
Avison, 1997; Jackson et al., 2000). Single mothers
may have less opportunity to leave a "bad" job in
the absence of financial support from a spouse or
partner (Menaghan & Parcel, 1995). When researchers control for these potential confounds in their
analyses, associations between employment and
parenting outcomes drop and are often no longer significant (Brady-Smith et al., 2001).
Second, some investigators have suggested that
women's psychological characteristics (such as their
levels of depressive symptoms and their cognitive
and emotional responses to stressful situations) may
also serve as unmeasured or omitted variables that
could explain observed associations between
mothers' employment and their parenting styles
(Kalil, Dunifon, & Danziger, 2001; Mayer, 1997;
Yoshikawa, Rosman, & Hsueh, 2001; Zaslow et al.,
1998). Mothers who struggle with high levels of
depressive symptoms have been found to face more
difficulty in finding and keeping jobs as well as
having more difficulty in using optimal parenting
styles (Danziger & Kalil, 2002; Ginexi, Howe, &
Caplan, 2000; Yoshikawa, 1999). Research on occupational health suggests that although lower prestige work is associated with employees' more
negative mood, the direction of influence between
job quality and employees' experience of negative
emotion remains unclear (De Jonge et al., 2001; Long,
1998). In short, by virtue of these individual
differences in age, skills, mental health, and disposition, several investigators have argued that some
mothers may select (or be constrained by greater
personal or structural obstacles) into more materially
disadvantaged circumstances; less employment at
more menial, unstable jobs; and patterns of more
harsh and negative parenting (e.g., Mayer, 1997).
This study responded to these concerns by using a
longitudinal design and a set of time-lagged models,
offering a conservative test of the impact of maternal
employment on family outcomes by controlling for
earlier levels of those outcomes (Duncan, Yeung,
Brooks-Gunn, & Smith, 1998; Menaghan, Jekielek,
Mott, & Cooksey, 1998). To address concerns for
selection bias, this study considered maternal employment as an endogenous, mediating variable.
Correspondingly, earlier assessments of (a) family
background characteristics and (b) psychological
characteristics such as mothers' depressive symptoms and emotionally negative parenting style at
Time 1 are modeled as exogenous variables that
could potentially predict both quantity and quality
of mothers' employment and mothers' psychological
and parenting outcomes at Time 2. As illustrated in
Figure 1, this model extends prior models of income
loss, psychological distress, and negative parenting
(Conger et al., 1992; Jackson et al., 2000; McLoyd,
1990) to test hypotheses regarding the impact of the
quantity and quality of mothers' employment on
their depressive symptoms and parenting style,
among families with preschool-aged children.
An alternate approach would be to carry out
fixed-effects models whereby possible omitted variables that threaten claims of causality are essentially
removed from analyses by regressing changes in an
outcome variable (such as depressive symptoms or
parenting) on changes in the predictor variable (such
as employment; see Dearing, McCartney, & Taylor,
2001; Duncan et al.,1998). Fixed-effects models offer
the opportunity to make stronger claims regarding
causality between dependent and independent variables. Fixed-effects models have several drawbacks
as well as strengths, however: They require a large
sample size, do not allow the investigator to address
endogeniety or selection, and may be less useful
when capturing lagged developmental effects over
time (Chase-Lansdale et al., 2003). Given this study's
focus on selection processes into employment and its
inclusion of a relatively small group (N < 100) of lowincome mothers, tests of mediation were selected
over a fixed-effects modeling approach (for additional discussion of selection and the omitted
variables problem, see Burchinal & Nelson, 2000;
Foster, 2002).
Differences in the conclusions drawn in the extant
literature on the impact of mothers' work on child
and family functioning may also have been due to
differences in definitions of maternal employment
and to different approaches taken toward the
Work and Parenting for Low-IncomeMothers
Depression
,I
T1
Age
Ethnicminoritystatus
~
1723
Depression
T2
Quantityof employment
Qualityof employment
Cohabitation
Negative
T1
parenting
Negative
T2
parenting
Fig. 1. Theoretical model of demographic characteristics, mothers' employment, depressive symptoms and negative parenting style from
Time 1 to Time 2.
measurement of parenting style. The availability of
good-quality child care for working mothers has
represented an additional confound in prior research. It is to these additional potential confounds
that this paper now turns.
and Parenting
ExpandingDefinitionsof Employment
Early research on the impact of maternal employment on child outcomes considered work in categorical terms, comparing outcomes for unemployed
versus employed mothers (Harvey, 1999). Later
definitions of work were expanded to consider
timing of work (such as within the first year of the
child's birth; e.g., Belsky, 1988), duration of work
(e.g., number of years of early childhood that a
mother works; see Harvey, 1999, for review), and
quantity or intensity of work involvement (e.g.,
considering earnings as an index of low-income
women's quantity of employment; Morris, Bloom,
Kemple, & Hendra, 2003).
Broadening the focus from quantity to quality of
work, Menaghan, Parcel, and colleagues have
focused extensively on the type of jobs held by
employed mothers. Specifically, mothers' job entry
into low-wage, menial jobs has been found to exert a
negative effect on the family's home environment, as
measured using Caldwell and Bradley's (1984)
Home Observation of Measurement of the Environment (HOME) assessment (Menaghan & Parcel,
1995). There was less of a drop in caregiving quality
for women entering better paying, higher quality
jobs. Occupations that require low skill, low levels of
autonomy, and high levels of repetition have been
argued to have a negative rather than positive
impact on mental health (Gecas, 1989; Zill et al.,
1995). Alternately, participation in a rewarding and
relatively supportive workplace may have clear,
salutary effects on low-income mothers' mental
health (Klebanov, Brooks-Gunn, Chase-Lansdale, &
Gordon, 1997). In this study, ratings of occupational
prestige were used as a rough index of the extent to
which occupations are more or less menial (Mortimer & Lorence, 1979) to test whether a "good" job is
better than a "bad" job in psychological and family
functioning terms (see also Jackson et al., 2000).
In addition, this study took a comprehensive look
at the style of parenting used by working and
nonworking mothers in low-income families. Following developmental research that suggests that
parents resort to more power-assertive and coercive
strategies of discipline when under greater financial
strain (Dodge, Pettit, & Bates, 1994; Elder, Eccles,
Ardelt, & Lord, 1995; McLoyd & Wilson, 1991), this
study focused on mothers' use of firm limit-setting
versus coercive practices as an important index of
optimal parenting. Recently, several investigators
have argued that firm limit setting and high
monitoring represent a normative form of "nononsense" parenting that is associated with optimal
child outcomes, particularly for African American
children (Brody & Flor, 1998; Deater-Deckard &
Dodge, 1997).
Recent theoretical models suggest that parents
resort to more harsh and coercive parenting styles as
a result of their own rising feelings of anger,
irritation, and distress, particularly as economic
and psychological pressures mount (Brody et al.,
1994; Conger et al., 1992; Mistry, Vandewater,
Huston, & McLoyd, 2002; Raver, 2002). Parents'
emotional expressiveness appears to influence not
only their parenting styles but also children's
1724 Raver
development of later emotional and social competence, among both middle- and low-income samples
(Cassidy, Parke, Butkovsky, & Braungart, 1992;
Denham & Grout, 1993; Garner, Jones, & Miner,
1994; Isley, O'Neil, Clatfelter, & Parke, 1999; Raver &
Spagnola, 2002). In light of these findings, this study
included measures of mothers' expressiveness of
feelings of anger, frustration, and irritation, both in
the context of caring for their children and in
creating a general family emotional climate. In so
doing, this study aimed to fill an empirical gap
between the developmental literature on family
emotional processes, which often leaves questions
of maternal employment unaddressed, and the
literature on maternal employment, which must
often rely on relatively few survey items (e.g., from
Caldwell and Bradley's 1984, HOME assessment) as
rough proxies of parenting style.
The cost, quality, and availability of child care
have been identified as a third set of confounds in
previous research on maternal employment, parenting, and child outcomes. To control for this type of
confound, this study was carried out among women
who enrolled their children in Head Start programs,
which provides fully subsidized, high-quality child
care as well as transportation for participating
children to eligible families. Previous research
indicates that Head Start families are no more
advantaged, and may be more disadvantaged, than
nonenrolled families (General Accounting Office,
1994; Hofferth, 1994; Schnur, Brooks-Gunn, & Shipman, 1992). However, given that Head Start serves
only 60% of the eligible poor children and their
families, the trade-off in using a Head Start-enrolled
sample is that family characteristics that lead
families to use Head Start may make them seriously
nonrepresentative of low-income families as a
group. Therefore, substantial caution is taken when
making any broad claims regarding work, parenting,
and the study's findings.
Study Hypotheses
First, in keeping with models of poverty, psychological distress, and parenting (Jackson et al., 2000;
McLoyd, 1990), mothers' increased participation in
the workforce was expected to be associated with
improvements in parenting and maternal mental
health over time. That is, increases in mothers' work
hours and higher levels of maternal earnings were
expected to be associated with lower depressive
symptoms and lower use of emotionally negative
parenting style at Time 2, even after taking into
account mothers' demographic and psychological
characteristics (such as age, ethnic minority status,
residence in a one- versus. two-parent household,
prior levels of depressive symptoms, and prior
parenting style).
Second, considering quality, as well as quantity of
mothers' employment may strengthen our understanding of the impact of maternal employment on
mental health and parenting. Although mothers'
increased quantity of employment might be expected to have salutary effects on parenting and
mental health, lower quality of employment was
expected to be predictive of decrements in mothers'
mental health and parenting over time. This study
examined whether mothers' participation in "good"
versus "bad" jobs is associated with these family
outcomes, net of their demographic characteristics,
earlier levels of depressive symptoms, and parenting
styles at Time 1.
Third, this study tested the hypothesis that some
mothers face more barriers than others in the
amount they can work and in the types of jobs they
can hold, and that these barriers also pose problems
for mothers' mental health and parenting. In short,
models of the role of work in predicting mothers'
depressive symptoms and parenting were tested
using conservative forms of estimation, where
possible selection processes into more work, greater
mental health, and optimal parenting were included
(e.g., Mayer, 1997). To test this hypothesis, mothers'
quantity and quality of employment were modeled
as endogenous to, rather than exogenous to, demographic characteristics, maternal depressive symptoms, and parenting over time.
It is important to qualify these hypotheses in light
of the study's small sample and the risk of Type II
error. Any null findings could be due to lack of
statistical power, and therefore the following analyses are considered exploratory rather than definitive. With that caveat, this study aimed to contribute
to the research literatures outlined earlier by
providing a more complex characterization of the
psychological benefits and costs of work for a small
group of low-income, Head Start-enrolled mothers.
Method
Sample
At Time 1 (1997-1998), 146 mothers with low
incomes (as defined by Head Start eligibility guidelines) and a target child, ages 3.8 to 4.6, were
enrolled in a related study on parenting and
children's social and emotional development from
Head Start sites in urban and rural settings in
Work and Parenting for Low-Income Mothers
upstate New York.These settings included (a) set of
working-class and poor neighborhoods, primarily
composed of African American families in a large,
industrial Northeastern city (Rochester,New York)
and (b) a set of low- to middle-income, isolated
farming communities,primarilycomposed of White
families, in a sparsely populated, rural county
(TompkinsCounty, New York) 100 miles south of
the urban community described previously. Two
years later at Time 2, 100 of the 146 families were
reinterviewed(yielding a follow-up response rate of
68%),and extensive questionnaireand observational
data were collected during a second home visit (see
the following for more extensiveanalysesof attrition).
For each family visit, families were debriefed,
thanked, and reimbursed $20. Complete longitudinal data were available for 94 families, where the
same female caregivercompleted all assessments at
both time points. Family incomes at Times 1 and 2
were deflated to 1997 values. Average income at
both time points was less than or equal to $20,000
(MTI
= $17,568 and MT2 = $20,604) for families that
averaged two adults and three children. Families'
reported incomes as well as use of Head Start and
public assistance (with slightly more than one third
of mothers reportingthat their households had gone
"on" or "off of welfare" in the past year) suggests
that the families included in this study can be fairly
characterizedas low-income at the time of assessment.
Between-sitecomparisonssuggest that samples in
rural versus urban sites were strongly racially
segregated, with few African American families
enrolled in the rural Head Start centers and few
White families enrolled in the urban Head Start
centers. Mothers were, on average, 29.93 years of
age, with no significant difference in age found
between sites. A majorityof mothers in both settings
had worked during some portion of the 2 years
before their participationin this study. Analyses of
between-group differences on all exogenous and
endogenous variables suggest that rural and urban
mothers did not differ on most demographic
characteristics, except that rural mothers were
significantly more likely to be married than urban
mothers (Pearson chi-square= 15.87, p<.001). Because few differences were found, based on mothers'
residence, rural and urban subsamples were combined for all analyses.
Attrition
Two years after the Time 1 interviews, follow-up
interviews were conducted with 94 of the 146 female
1725
caregiversat Time2. Analyses of participantattrition
suggest that there were no significant differences
between followed and nonfollowed women on Time
1 measures of income, ethnic minority status, age,
employment status, likelihood of urban versus rural
residence, or family size. To detect whether relations
among demographic, psychological characteristics,
and work involvement differed with the full versus
smaller, followed-up subsample, employment (current hours worked per week) at Time 1 was
regressed on all demographic characteristics,depressive symptoms, negative parenting, and followed-up versus attrited status as a moderator
(including followed-up status and age, ethnicity,
cohabitation, education, income, depressive symptoms, and negative parentingas two-way interaction
terms). Overall regression equations were not
statistically significant, and of all seven interaction
terms entered, only one interactionterm (FollowedUp Status x Negative Parenting) approached significance at the .08 level. Post hoc examination of
correlationsfor larger and followed-up subsamples
revealed weak correlationsof similarmagnitude and
direction, suggesting that this marginal finding was
spurious. In all other cases, correlations among
demographic characteristics,psychological characteristics, and employment did not differ for the
followed-up sample versus the larger sample.
Procedures
Familiesfrom eight Head Startsites were initially
asked for their permission to be contacted by the
research staff through a multistep process (e.g.,
mailings, discussions with Head Startoutreachstaff,
and research staff's presentations at Head Start
parent orientationmeetings and potlucks). Of those
families initially agreeing to be contacted, 80%
enrolled in the study by scheduling and completing
a 2-hr home visit at Time 1. Reasons for nonenrollment following families' initial contact with
researchstudy staff included parents'lack of interest
or time, family relocation, and child's withdrawal
from the Head Startprogram.All data reportedwere
collected during parent interviews conducted at
Time 1 in the fall of the families' enrollment into
Head Start (in 1997-1998) and again at Time 2
(when most children were entering first grade, in
1999-2000). Interviews included a demographic
questionnaire (assessing demographic characteristics, family income, and parents' history of employment over the past 2 years), parents' depressive
symptoms (Center for Epidemiological Studies
Depression Scale [CES-D], Radloff, 1977), expres-
1726 Raver
siveness of negative emotions such as anger and
irritation (Halberstadt, Cassidy, Stifter, Parke, & Fox,
1995; Raver, 2002), and parenting (Gerard, 1994).
After the parental interview was completed, the
parent was debriefed, thanked, and reimbursed $20
as a token of the interviewer's appreciation.
Measures
Following Garcia Coll et al. (1996) and others, all
measures were screened for their appropriateness
for use with both White and ethnic minority
families. Most of those listed in the following
discussion have a solid track record in their use
with low-income communities and were expected to
be culturally sensitive and appropriate for use in this
study. As an index of each measure's internal
consistency, Cronbach's alphas are provided for all
psychological and parenting scales in Table 1.
At Time 1, mothers' current employment status
(e.g., whether mothers were currently working, and
hours per week currently working) was collected
during interviews. At Time 2, more extensive
indexes of mothers' employment entry, exits, and
length of work participation were collected during
parent interviews, including start and end dates for
each job, job title, description of each job, and
number of hours worked per week for the last job
held, for the 2 years before the family visit.
Quantity of mothers' employment was coded two
ways. Following Menaghan and Parcel (1995),
"changes in hours currently employed" across the
two time periods were calculated from hours
worked per week (including 0) at mothers' currently
held job at Time 2 minus hours worked per week
(including 0) at mother's currently held job at Time
1. One concern might be that such an index may not
fully capture the work involvement of women who
may have worked substantially during the interim
but may be currently unemployed. A second, less
stringent assessment of quantity of mothers' employment, in terms of their yearly earnings from Time 1
to Time 2 was calculated to credit women with any
recent work held in the last 2 years (Morris et al.,
2003; Morris et al., 2001). "Yearly earnings from Time
1 to Time 2" was calculated by multiplying the
number of months worked in that interval (from 0 to
24) by the number of hours worked per week in the
last job held and the dollar amount of hourly pay
received by mothers at their last job. This value was
multiplied by 4 (weeks per month) and divided by 2
(years elapsed between Time 1 and Time 2) to yield a
rough estimate of mothers' yearly earnings. Quality of employment was also assessed. Specifically,
General Social Survey scores of job prestige were
coded for mothers' most recently held jobs at Time 2
and were used as a rough proxy for the extent to
which a given job was considered menial.
Mothers' depressive symptoms were assessed
using the CES-D scale for depressive symptomatology (Radloff, 1977). This scale has demonstrated
reliability and validity with middle- and low-income
populations (Radloff, 1977) and has been used
extensively in studies of associations among financial strain, employment, and low-income women's
depressive symptoms (Lennon, Bloom, & English,
2001). Mothers' mean scores at Time 1 and 2 were in
the elevated but not clinical range (MT1 =11.65,
SD= 7.69, range= 0-37; MT2= 12.07, SD=9.35,
range = 0-48), with 26% of mothers scoring in the
clinical range (scores of 16 or above) at Time 1 and
20% of mothers scoring in the clinical range of
symptomatology at Time 2. The scale yielded high
internal consistency at both time points (see Table 1).
Parents' expressiveness of negative emotions such
as anger and irritation was assessed using two
measures. The negative dominant subscale of the
Self-Expressiveness in the Family Questionnaire
(SEFQ; Halberstadt et al., 1995) taps parents' proneness to expressions of anger and hostility and was
used to assess the extent to which parents maintained an angry emotional climate in the home. The
SEFQ is a parental report measure of the predominant style of emotional expressiveness used by the
respondent in the context of everyday family events,
correlates moderately with observed measures of
parents' expressed affect during laboratory tasks
(Cassidy et al., 1992; Halberstadt et al., 1995), and
has been found to be reliable and valid with lowincome samples (Garner et al., 1994; Greenberg,
Lengua, Coie, & Pinderhughes, 1999). An additional
11-item questionnaire was added to assess parents'
ability to maintain emotional equilibrium when
faced with childrearing challenges, with items such
as: "Bedtimes are a quiet, peaceful time for us,"
"I get angry when I have to tell my child the same
thing, over and over again," and "I feel 'rattled'
when the family gets loud or noisy." High scores on
the scale represent mothers' reports of more emotional equilibrium or low levels of anger and
irritation during routine caregiving situations. Scale
scores were reversed for this study to correspond
with the negative dominant subscale of the SEFQ
(Halberstadt et al., 1995, see the following). Both
scales yielded good internal consistency across both
time points (see Table 1).
Parents' use of coercive versus firm limits was
assessed using the 12-item limit-setting subscale of
Work and Parenting for Low-IncomeMothers
1727
Table 1
and Employment
Characteristics
DescriptiveDatafor Mothers'Demographic,
Psychological,
Variable
Demographic characteristics
Mothers' age
% African American
% cohabiting or married
Psychological characteristics
Depressive symptoms at Time 1 (c = .85)
Depressive symptoms at Time 2 (a = .88)
Negative parenting at Time 1
Mean SEFQ ND subscale score T1 (a = .72)
Emotional equilibrium T1 (oc= .68)
T score, PCRI limit-setting subscale Ti (a = .80)
Negative parenting at Time 2
Mean SEFQ ND subscale score T2 (ot= .83)
Emotional equilibrium T2 (a = .70)
T score, PCRI limit-setting subscale T2 (a = .77)
Employment characteristics
% mothers currently working, T1
Hours per week worked at job, T1
% mothers currently working, T2
Hours per week worked at job, T2
Change in hours worked at current job, T2-T1
Months worked, T1 to T2
Average pay per hour at last job, T2
Mothers' yearly earnings, T1 to T2
Average occupational prestige score at last job, T2
M (SD)
Range
29.93 (7.90)
46%
62%
18.00-55.00
11.65 (7.69)
12.07 (9.35)
0.00 (2.35)
3.83 (1.06)
40.83 (5.72)
50.11 (9.49)
0.00 (2.30)
3.89 (1.23)
41.38 (6.09)
50.53 (8.16)
1.00-37.00
0.00-48.00
-4.81 -6.84
2.00 -6.80
19.00-59.00
29.00- 76.00
- 4.96-6.88
1.80-8.00
28.00 -57.00
33.00-76.00
52%
18.58 (19.86)
65%
32.12 (16.39)
5.43 (21.53)
15.97 (9.18)
7.65 (4.40)
10,872 (8,891)
36.83 (6.92)
0.00-80.00
0.00-60.00
- 48.00-52.00
0.00-24.00
0.00-19.30
0-37,060
22.00-57.00
Note. SEFQ ND = Self-Expressiveness in the Family Questionnaire negative dominant quadrant; PCRI = Parent - Child Relationship
Inventory.
the Parent-Child Relationship Inventory (PCRI;
Gerard, 1994). The subscale includes such items as:
"I have trouble disciplining my child" and "I often
threaten to punish my child but never do." Previous
research with this measure suggests that it is a valid
index of parenting with high-risk, ethnic minority
populations, and it correlates well with other
measures of childrearing (MacPhee, Fritz, & MillerHeyl, 1996). Alpha and test- retest reliabilities have
been high with White and ethnic minority families
(MacPhee et al., 1996), and the inventory yielded
adequate internal consistency across both time
points (see Table 1).
The three measures of family emotional climate
and disciplinary style (high expressiveness of anger
and irritation, low ability to maintain emotional
equilibrium during routine caregiving, and high use
of coercive versus. firm limit setting) were positively
associated (ranging from r =.22 to r =.47, p<.05),
suggesting that they represent three important
components of emotionally negative parenting style.
Following Rushton, Brainerd, and Pressley (1983),
the three measures were therefore standardized
within Time 1 and Time 2 assessments and were
aggregated so that the number of parameters to be
estimated could be reduced when using structural
equation modeling (SEM) techniques (see also Kline,
1998).
Results
Overviewof PlannedAnalyses
First, descriptive statistics are provided on the
quantity of employment (in terms of hours worked
at current jobs at Times 1 and 2, changes in hours
worked at current employment from Time 1 to Time
2, and yearly earnings from Time 1 to Time 2) and
work quality (in terms of occupational prestige) in
Table 1. Additional descriptive statistics regarding
mothers' depressive symptoms and their parenting
styles are also provided for both time points.
Then, simultaneous SEM-fitting techniques were
used to test hypotheses regarding work's impact on
mothers' depressive symptoms and parenting style
at Time 2, net of prior parenting and mental health
and a set of demographic characteristics. Structural
1728 Raver
equation models were estimated using Amos software (Arbuckle & Wothke, 1999, version 4.01) to
examine the goodness of fit for a model that
included the two exogenous variables of mothers'
negative parenting style and depressive symptomatology at Time 1 as well as the three exogenous
demographic characteristicsof mothers' age, ethnic
minority status, and cohabiting status. These five
variables were modeled as predictors of the endogenous variables of negative parenting style and
depressive symptomatologyat Time 2. In each of the
three sets of analyses reported next, work quantity
or quality was hypothesized to function as a third,
mediating endogenous variable.
Best-fitting models are represented in Figure 2
with disturbance terms and nonsignificant path
coefficients left off of figures for ease of interpretation. Associations between Time 1 variables and
demographic characteristics are also left off of
figures for ease of interpretation(and are available
in Table 2). Because small sample size limits statistical power to detect effects of substantive interest
(e.g., associations between demographic characteristics, work, and mothers' outcomes), alpha levels
were set at .05 to detect statisticallysignificanteffects
and were relaxed to .10 to detect and interpret
statisticallymarginal effects.
For each SEM analysis of the role of a given
employment variable (e.g., mothers' yearly earned
income), a set of nested models was tested for
relative improvements in fit. A baseline model
included paths from all exogenous variables (e.g.,
Time 1 depression, Time 1 parenting,plus the three
demographic control variables) to both outcome
variables (depression at Time 2 and parenting at
Time 2), with paths from exogenous variablesto the
given employment variable constrained to equal 0.
Additional models estimated relative improvement
in model fit by freeing paths from demographic
control variablesto the employment variableand by
freeing paths from Time 1 depressive symptoms and
Time 1 parenting style to the employment variable.
What can these additional models tell us? If
selection processes play a role in leading some
mothersboth to be working less and to increasetheir
use of more negative parenting, for example, these
more complex models should yield significantly
lower chi-square values. They should also yield
statisticallysignificantparameterestimates for paths
from mothers' psychological and demographic
characteristics at Time 1 to both the mediating
employment variable and the Time 2 outcomes. If,
on the other hand, freeing additional parametersin
these more complex models does not yield significant changes in the model's chi-square value, the
baseline model would be consideredto offer the best
fit. Best-fittingmodels are presentedlater in the text,
afterthe studies descriptiveemploymentstatisticsare
reported.
and MaternalEmployment
Characteristics
Demographic
BetweenTime1 and Time2
Descriptive data for mothers' demographic, psychological, and employment characteristics are
provided in Table 1. On average, mothers were
significantly more likely to be currentlyworking at
Time 2 than at Time 1, X2(1, 93) = 8.47, p <.01, with
one half of the mothers employed at the time of the
initial interview, and two thirds of the mothers
employed at Time 2. When asked if they had ever
worked in the 2 years between assessments,82 of the
94 mothers had held a job for 5 hr a week or more
sometime in the past 2 years.
Table2
BetweenKeyVariables
Intercorrelations
Variable
1. Age
2. Cohabiting with partner
3. Ethnic minority status
4. Change in hours worked per week, T1 to T2
5. Yearly earnings, T1 to T2
6. Prestige, last job, T2
7. Depressive symptoms, T1
8. Negative parenting, T1
9. Depressive symptoms, T2
10. Negative parenting, T2
1
2
.11
-.20*
-.03
.04
- .11
-.26**
-.25*
- .03
-.37**
.25*
.10
.09
-.05
-.19*
.03
-.15t
-.15t
Note: N= 94 (except for correlational values for prestige, N = 82).
tp < .10 (one-tailed).
(one-tailed).
.05 (one-tailed). **p
*p_
_.01
3
4
5
-.11
.07
.01
.12
.20*
.12
.12
-.06
.04
-.02
-.16+
- .15+
.27**
-.17*
-.03
- .08
-.11
-.14[t
6
7
8
.05
-.04
.06
-.24*
.45**
.47**
.33**
.21*
.64**
9
.39**
10
Work and Parenting for Low-Income Mothers
0.48***
0.4*** (0.57)
(0.5)
Depression
1729
Depression
T1
T2
0.25**
(11.02)0.28***
T.,
worked/wk
Cohabitation
Age
Ethnicminoritystatus
-0-16 (-0.07) (0.07)
-0.16t
2vs._T1
Negative
parentingT1
Negative
parentingT2
***(0.5)
NFI= 0.98, RMSEA= .00.
Figure2a. N = 94; chi-square= 3.14, df= 6, p = 0.79, Bentler-Bonnet
0.49***(0.60))
Depression
Depression
T1
-0.17t
T2
(-0.20)
Yearly
Age
Ethnicminoritystatus
Cohabitation
Negative
parentingT1
0.27***
earnings
(approximate)
T1
from
to
(Negative
0.58***(0.57)
T2
-0.13
(0.07)
(-0.03)
parentingT2
NFI= 0.96, RMSEA= .00.
Figure2b. N = 94; chi-square= 5.26, df= 6, p = 0.51, Bentler-Bonnet
Depression
T1
Job
Prestige
Age
Ethnicminoritystatus
Cohabitation
Negative
parentingT1
Depression
T2
0.41** (0.52)
0.57***(0.55)
-0.23**
(-.08)
0.25***
(0.06)
Negative
parenting 2
Figure 2c. N = 82; chi-square = 3.05, df = 5, p = 0.69, Bentler-BonnetNFI = 0.99, RMSEA= .00.
Fig. 2. Structural equation models showing quantity and quality of mother's employment. Standardized parameter estimates provided for
significant paths (unstandardized coefficients provided within parentheses). Analyses of quantity of employment include all mothers
(N = 94). Analyses of quality of employment are restricted to the subsample of mothers with some employment between Time 1 and Time
2 (n = 82). NFI = normed fit index; RMSEA = root mean square error of approximation. t = p <.01. *p < .05. **p <.01. ***p<.001.
Analyses of mothers' employment histories
yielded a complex portrait of the quantity and quality
of work that low-income mothers engaged in over the
2 years. Mothers worked on average 32 hr a week, a
little less than 16 months out of the 24 months
between visits, and were paid $7.65 an hour on
average. Regarding occupational prestige, the bulk of
mothers held blue-collar jobs in service, manufacturing, and clerical sectors. Typical jobs held by mothers
in this sample included cashier, certified nurse's
assistant, child care provider, and assembler. Mothers
reported considerable instability or cycling in and out
1730 Raver
of unemployment, with 39% of mothers in the sample
experiencing at least one episode where they had lost
or left a job. In addition, 44% of mothers had worked
continuously and the remaining 17% moved from
unemployed to employed status.
Generally, mothers reported elevated, but not
clinical, levels of depressive symptomatology at both
Time 1 and Time 2. Twenty-six percent of mothers
scored in the clinical range of depressive symptoms
at Time 1, and 20% of mothers scored in that same
range at Time 2. These findings suggest that this
group of mothers may be relatively more advantaged than women who have recently left welfare,
where percentages of women who fall above clinical
cutoff scores for depressive symptomatology range
from 40% to 60% of women surveyed. Zero-order
correlations of all key study variables included in
subsequent SEM analyses are provided in Table 2.
Mothers'
Analysesof Quantityof MaternalEmployment,
DepressiveSymptoms,and ParentingStyle
For the first set of structural equation models,
changes in hours employed per week (hours
employed per week at Time 2 minus hours employed per week at Time 1) was included as the
employment variable to be tested. Results suggest
that a trimmed model with Time 2 outcomes
predicted by (a) all demographic variables, (b)
depressive symptoms and parenting at Time 1, (c)
the employment variable, and (d) work predicted by
mothers' cohabiting status, fit the data best,
X2= 3.14, df = 6, p= .79 (AX2= 6.01, Adf= 1, p<.01
over the baseline model). Final model fitting indexes
suggest a reasonable fit to the model (illustrated in
Figure 2a), with Bentler-Bonnet normed fit index
(NFI) = .98 and root mean square error of approximation (RMSEA) = 00. Figure 2a shows that parameter estimates for change in hours employed as a
predictor of mothers' depressive symptoms at Time
2 were marginally significant, with increased hours
worked associated with slight decrease in mothers'
depressive symptoms from Time 1 to Time 2
(beta = -.16, p<.10). It is important to note that
mothers' depressive symptoms and mothers' parenting style demonstrated considerable stability over
time (betas = .48 and .58, p<.001, respectively) and
that mothers' increase in work hours predicted
variance in their Time 2 depressive symptoms net
of mothers' prior history of depressive symptoms,
parenting style, and demographic characteristics at
Time 1. Figure 2a also illustrates that cohabitation
played a significant role in predicting changes in
mothers' work hours over time, with cohabitation
associated with working more hours per week from
Time 1 to Time 2 (beta = .25, p<.01). When additional models were estimated to test whether
changes in work hours were endogenous to mothers'
depressive symptoms and parenting style at Time 1,
these models did not yield a significant improvement in fit and were rejected.
A second set of analyses was then conducted with
mothers' yearly earnings included in the model as
the endogenous measure of work quantity (see
Figure 2b). A more complex model with a single
freed path from mothers' depressive symptoms at
Time 1 as a predictor of mothers' yearly earnings
between Time 1 and Time 2 yielded the best fit to the
data, X2= 5.26, df= 6, p = .51 (AX2= 2.75, Adf= 1,
p<.10 over the baseline model). Fit indexes for this
final model were good, Bentler-Bonnet NFI = .96 and
RMSEA = .00. As illustrated in Figure 2b, mothers'
higher yearly earnings were modestly predictive of
decreases in their use of a negative parenting style
(beta = -.13, p <.10). In addition, mothers experiencing higher levels of depression at Time 1 earned less
from Time 1 to Time 2 (beta = -.17, p<.10). A third
model with additional paths from demographic
characteristics at Time 1 as predictors of mothers'
yearly earnings from Time 1 to Time 2 did not yield
significant improvement in fit.
In sum, parameter estimates yielded from these
two models of mothers' quantity of employment
suggest benefits of maternal employment, with
higher levels of work participation predicting (a)
mothers' lower depressive symptoms at Time 2, net
of their prior histories of symptomatology, and (b)
mothers' lower use of negative parenting styles, net
of their prior use of emotionally negative parenting.
Inspection of parameter estimates also suggests that
there was some sparse evidence for the endogeneity
of quantity of employment, in that cohabiting
women were more likely to have increased their
work hours from Time 1 to Time 2 and more
depressed mothers at Time 1 reported lower earnings in the 24 months from Time 1 to Time 2 than did
mothers who were less depressed at Time 1. In
contrast, negative parenting style at Time 1 was not
significantly associated with changes in amount of
employment nor with mothers' earnings over time.
Predicting Mothers' Depressive Symptoms and Negative
Parenting Style at Time 2 From the Quality of Mothers'
Employment
Next, a third set of analyses were undertaken to
examine whether the quality of the last job that
mothers held at Time 2 was a significant predictor of
Work and Parentingfor Low-IncomeMothers
mothers' depressive symptoms and parenting at
Time 2, net of demographic characteristics and
mothers' initial psychological and parenting status
at Time 1. This model was estimated for all mothers
who had held at least one job in the past 2 years
(N= 82). Structural equation models were conducted with job prestige as the work variable of
interest.
Tests of nested structural equation models suggest that the baseline model (with demographic
characteristics, depressive symptomatology, and
negative parenting style at Time 1 included as
predictors of Time 2 outcomes only but not as
of job prestige) offered the best fit,
predictors
S
2 = 3.05, df= 5, p = .69, Bentler-Bonnet NFI = .99,
RMSEA = .00. Inspection of the parameter estimates
yielded from this final model yields a conservative
estimate of the associations among job quality,
depressive symptoms, and parenting at Time 2, net
of demographic and psychological factors. As illustrated in Figure 2c, mothers' employment in more
menial, lower prestige jobs is associated with increases
in mothers' use of more angry, coercive parenting
styles from Time 1 to Time 2 (beta = -.23, p <.001). No
evidence for selection processes was found in predicting the quality of mothers' employment from
Time 1 to Time 2 in this sample, as indicated by the
lack of fit offered by more complex models where job
prestige was modeled as endogenous to mothers'
demographic and psychological characteristics.
Discussion
Does work pay, in psychological as well as economic
terms, for the mothers of preschool-aged children in
this study? The first two sets of analyses suggest that
there may be benefits of higher levels of workforce
participation for mothers' mental health and parenting over time. Increases in mothers' work hours
were predictive of decreases in mother's depressive
symptoms over time. In addition, mothers' higher
earned income from Time 1 to Time 2 was predictive
of decreases in mothers' use of angry and coercive
parenting style at Time 2, net of their demographic
characteristics and their earlier mental health and
parenting histories. These findings are in keeping
with other recent reports on the positive impact of
work and self-sufficiency activities on depressive
symptoms and parenting in other, larger, randomized studies (Duncan & Chase-Lansdale, 2001).
It is important to note that effect sizes for the
quantity of employment in predicting family wellbeing in this study are small, suggesting that the
benefits of work are modest, at best. Conditions of
1731
financial hardship may be so substantial for working
poor families that mothers are hard pressed to meet
their families' needs even when working, given their
wage rates (Danziger & Kalil, 2002; Edin & Lein,
1997; Gyamfi, Brooks-Gunn, & Jackson, 2001; Jackson
et al., 2000). The findings of this study support others'
recent welfare-related research suggesting that the
effort to strengthen family functioning through
mandated employment for low-income mothers
may be relatively ineffective unless policy initiative
is taken to make work pay in economic terms.
In light of this study's positive findings regarding
mothers' workforce participation, is it fair to
conclude that any job is a good job for low-income
mothers and their families? This study's third set of
analyses suggest that claims of the benefits of
maternal employment should be strongly qualified
by considering the quality of the jobs that lowincome women must take. Results from this study
suggest that mothers who worked in lower prestige
jobs were likely to become significantly more angry
and coercive in their parenting style over time. This
association between occupational prestige and parenting held, net of contributions made by mothers'
demographic and psychological background characteristics. Reviews of studies across a wide range of
welfare reform demonstration programs are equivocal regarding whether a human capital investment
approach, emphasizing mothers' further education
and training, is better or worse than a work-first
approach, encouraging women to take the first job
available (Morris, 2002). These findings suggest that
there are potential negative ramifications of unstable, low-paying, and stressful jobs on family
functioning (see Jackson et al., 2000, for similar
cross-sectional findings; Greenberger, O'Neil, &
Nagel, 1994). Women who quickly cycle back out
of work and onto public aid may be doing so
because of the costs that low-wage, unstable,
stressful work may have for both mothers and their
families in psychosocial as well as economic terms
(Edin & Lein, 1997). Questions of the value of human
capital investments in mothers' education and
training may become more salient in the next few
years, given the possibility of higher rates of
unemployment, more difficulty placing low-income
women in jobs, and slowing U.S. economic growth.
How can these two sets of findings (of benefits
associated with work participation on the one hand
and worrisome risks associated with lower job
quality on the other) be reconciled? Work may have
positive effects on mental health by reducing
financial strain for women who are able to find
and keep higher paying, less stressful, higher
1732 Raver
prestige jobs. For mothers entering low-quality jobs,
however, the psychological costs may outweigh the
financial benefits. Specifically, quantity of work may
have been moderated by the type of job that mothers
held: Long hours at a stressful, menial, or physically
taxing job may contribute to mothers' feelings of
depression, frustration, and fatigue, whereas long
hours at a psychologically and financially rewarding
job may have more positive effects. These questions
of moderation must be saved for future analyses
with larger and more nationally representative data
sets.
Because it was nonexperimental in design, this
study cannot provide definitive evidence that work
causally affects mothers' mental health and parenting over time. What this study can do is provide
relatively conservative estimates of work as a
predictor of family processes by controlling for a
set of demographic and psychological factors that
might seriously bias our estimates. In addition, work
as a predictor of family functioning was tested
against a selection hypothesis that demographic and
psychological characteristics such as age, ethnicity,
single-parent status, depressive symptoms, and
proneness to angry interpersonal style pose significant obstacles to some mothers' ability to find and
keep good jobs.
In keeping with prior research, sparse evidence
for selection processes was found, with mothers'
depressive symptoms appearing to serve as a
significant barrier to the amount that mothers
earned over time (Danziger, Corcoran, Danziger, &
Heflin, 2000; Jackson et al., 2000; Yoshikawa, 1999).
Mothers' prior history of depressive symptoms was
not found to be predictive of the quality (e.g.,
prestige) of their work, however, congruent with
findings by other investigators in the area of
occupational strain (De Jonge et al., 2001). Among
the demographic characteristics considered as predictors of mothers' workforce participation, cohabitation with a spouse or partner predicted modest
increases in mothers' weekly work hours. These
findings suggest that further research is needed on
the structural factors inside and outside the family
that may influence mothers' workforce participation.
This study provides an important portrait of the
employment experiences of low-income mothers
who participate in Head Start. Findings suggest that
most low-income mothers of Head Start-enrolled
children in this study were working outside the
home in a range of service, clerical, manufacturing,
and entry-level health care jobs. One half of the
mothers in rural and urban sites were working when
children were enrolled and two thirds were working
2 years later, with almost all mothers having held at
least one job in the intervening 2 years. Regarding
their mental health and parenting, fully 26% of
mothers scored in the clinical range of depressive
symptoms at Time 1 and 20% of mothers scored in
that same range at Time 2. These rates of depressive
symptomatology are lower than has been found in
other studies of low-income mothers (Danziger &
Kalil, 2002; Derr, Douglas, & Pavetti, 2001; Lennon et
al., 2001) but are nonetheless likely to have significant negative ramifications for family and child wellbeing over time (Alpern & Lyons-Ruth, 1993; Belle,
1990; Leadbeater, Bishop, & Raver, 1996; Petterson &
Albers, 2001). In light of other recent post-welfarereform studies, it is clear that this group of mothers
may be relatively more advantaged and less depressed than similarly aged women who have
recently left welfare (Coiro, 2001; Danziger & Kalil,
2002; Gyamfi et al., 2001). With that qualification in
mind, these mothers clearly face many of the same
economic and psychological pressures to make ends
meet by working in low-wage jobs, as has been
reported elsewhere (e.g., Edin & Lein, 1997).
In summarizing the implications of this study's
findings, several additional caveats are in order. This
study's use of a small and nonrepresentative sample
suggests that the generalizability of these findings
are limited and that null findings regarding relations
between mothers' demographic characteristics and
employment may have been due to inadequate
statistical power. In addition, this study emphasized
possible parenting mechanisms that may link maternal employment and child outcomes, based on
McLoyd's economic stress model of financial hardship and family process (see McLoyd, 1990; Mistry et
al., 2002). This study did not examine other equally
important investments models of influence, whereby
parents balance trade-offs between being able to
spend time and earned income on their children
with parenting assessed in terms of monitoring,
supervision, and the provision of cognitively stimulating materials and activities (Bradley, Corwyn,
McAdoo, & Garcia Coll, 2001; Duncan et al., 1998;
Gennetian & Miller, 2002; Gershoff, Aber, & Raver,
2002; Han, Waldfogel, & Brooks-Gunn, 2001).
Finally, small sample size precluded this study
from a closer examination of several additional
moderating factors that may play a key role in
relations between work and parenting. This study's
small sample size precluded tests of important
additional moderating influences of mothers' residence in one- versus two-parent households, ethnic
minority membership, or residence in an urban
versus rural setting. For example, given that African
Work and Parenting for Low-IncomeMothers
American and White families reside in spatially
segregated neighborhoods with widely differing
rates of joblessness (Brooks-Gunn, Duncan, & Aber,
1997), that African American women face significantly greater barriers to higher paying employment
and have a much longer history of greater labor force
participation than do White women (Dill, 1987),
models would be expected to fit differently for these
two groups. Multiple-group path analyses with a
much larger sample would yield clearer answer to
these questions.
Thus, findings of this study raise a set of
additional questions that can be best pursued in
additional research with larger, more nationally
representative data sets. These findings highlight
important questions regarding the quality as well as
quantity of work that low-income mothers of young
children are engaged in, and the importance of
considering families' emotional climates and
mothers' mental health when assessing the impact
of work on parenting. Answers to these questions
will provide a more complete empirical portrait of
mothers' struggles and successes in balancing the
complex demands of work and family while also
under the pressure of economic disadvantage.
1733
Brody, G. H., & Flor, D. L. (1998). Maternal resources,
parenting practices, and child competence in rural,
single-parent African American families. Child Development, 69, 803-816.
Brody, G. H., Stoneman, Z., Flor, D., McCrary, C., Hastings,
L., & Conyers, 0. (1994). Financial resources, parent
psychological functioning, parent co-caregiving and
early adolescent competence in rural, two-parent,
African-American
590-605.
families.
Child Development, 65,
Brooks-Gunn,J., Berlin, L., & Fuligni, A. (2000). Early
childhood intervention programs: What about the
family? In J. Schonkoff.& S. Meisels. (Eds.), Handbook
of early childhood intervention (2nd ed., pp. 549-588).
New York:CambridgeUniversity Press.
Brooks-Gunn, J., Duncan, G., & Aber, J. L. (1997).
Neighborhoodpoverty: Context and consequencesfor chil-
dren.New York:Russell Sage Foundation.
Burchinal,M. R., & Nelson, L. (2000).Familyselection and
child care experiences:Implicationsfor studies of child
outcomes. Early Childhood Research Quarterly, 15,
385-411.
Caldwell, B., & Bradley, R. (1984). Home Observationfor
Measurement of the Environment. Unpublished manu-
script, University of Arkansasat Little Rock.
Cassidy,J., Parke,R. D., Butkovsky,L., & Braungart,J. M.
(1992).Family-peerconnections:The roles of emotional
expressiveness within the family and children'sunderstanding of emotions. Child Development,63, 603-618.
References
Aber,J. L., Brooks-Gunn,J., & Maynard,R. (1995).Effects
of welfarereformon teenage parentsand their children.
Future of Children,5, 53-71.
Ali, J.,& Avison, W. R. (1997).Employmenttransitionsand
psychological distress: The contrasting experiences of
single and married mothers. Journal of Health and Social
Behavior,38, 345-362.
Alpern, L., & Lyons-Ruth,K. (1993).Preschoolchildren at
social risk: Chronicityand timing of maternal depressive symptoms and child behavior problems at school
and at home.
371-388.
Development and Psychopathology, 5,
Arbuckle,J. L., & Wothke,W. (1999).Amos4.0. Chicago:
SmallWaters.
Belle, D. (1990). Poverty and women's mental health.
Chase-Lansdale,P. L., & Brooks-Gunn,J. (1995). Escape
frompoverty.New York:CambridgeUniversity Press.
Chase-Lansdale, P. L., Moffitt, R. A., Lohman, B. J.,
Cherlin, A. J., Coley, R. L., & Pittman, L. D., et al.
(2003). Mothers' transitions from welfare to work and
the well-being of preschoolersand adolescents. Science,
299, 1548-1552.
Coiro, M. (2001). Depressive symptoms among women
receiving welfare. In M. Lennon (Ed.), Welfare,work
and well-being(pp. 1-23). Binghamton,NY: Haworth
Press.
Conger, R. D., Conger, K. J., Elder, G. H., Lorenz, F. O.,
Simons, R. L., & Whitbeck, L. B. (1992). A family
process model of economic hardship and adjustment
of early adolescent
526- 541.
boys.
Child Development, 63,
American Psychologist, 45, 385-389.
Danziger,S., Corcoran,M., Danziger,S., & Heflin, C. M.
(2000). Work, income, and material hardship after
reconsidered.
235-272.
Danziger,S., & Kalil, A. (2002).Welfarereform:Effectsof
TANFon family well-being.In R. M. Lerner,F.Jacobs,&
Belsky, J. (1988). The "effects" of infant day care
Early Childhood Research Quarterly, 3,
Bradley,R. H., Corwyn, R. F., McAdoo, H. P., & Garcia
Coll, C. (2001).The home environmentsof children in
the United StatesPartI:Variationsby age, ethnicity,and
poverty status. Child Development, 72, 1844-1867.
Brady-Smith,C., Brooks-Gunn,J., Waldfogel,J., & Fauth,
R. (2001). Work or welfare?:Assessing the impacts of
recent employment and policy changes on very young
children. Evaluation and ProgramPlanning, 24, 409-425.
welfare reform. Journalof ConsumerAffairs, 34, 6-30.
D. Wertlieb (Eds.), Handbook of applied developmental
science: Promoting positive child, adolescent, and family
development(pp. 395-420). Thousand Oaks, CA: Sage.
Dearing,E., McCartney,K., & Taylor,B. A. (2001).Change
in family income-to-needs matters more for children
with less. Child Development, 72, 1779-1793.
Deater-Deckard,K., & Dodge, K. A. (1997). Exteralizing
behavior problems and discipline revisited: Nonlinear
1734
Raver
effects and variations by culture, context and gender.
PsychologicalInquiry,8, 161-175.
De Jonge, J., Dormann, C., Janssen, P. P., Dollard, M. F.,
Landeweerd, J. A., & Nijhuis, F. J. (2001). Testing
reciprocal relationships between job characteristics and
psychological well-being: A cross-lagged structural
equation model. Journal of Occupational and Organizational Psychology, 74, 29-46.
Denham, S. A., & Grout, L. (1993). Socialization of
emotion: Pathways to preschoolers' emotional and
social competence. Journal of Nonverbal Behavior, 17,
205-227.
Derr, M. K., Douglas, S., & Pavetti, L. (2001, September).
Providing mental health services to TANF recipients:
Program design choices and implementationchallenges in 4
states. Paper presented at the Joint Center for Poverty
Research September Research Institute, "The Hard to
Employ and Welfare Reform," Washington, DC.
Desai, S., Chase-Lansdale, P. L., & Michael, R. T. (1989).
Mother or market?: Effects of maternal employment on
the intellectual ability of 4-year-old. Demography, 26,
545-561.
Dill, B. T. (1987). The dialectics of black womanhood. In S.
Harding (Ed.), Feminism and methodology(pp. 97-108).
Bloomington, IN: Indiana University Press.
Dodge, K. A., Pettit, G. S., & Bates, J. E. (1994).
Socialization mediators of the relation between socioeconomic status and child conduct problems. Child
Development, 65, 649-665.
Downey, G., & Coyne, J. C. (1990). Children of depressed
parents: An integrative review. Psychological Bulletin,
108, 50 - 76.
Duncan, G. J., & Chase-Lansdale, P. L. (2001). Welfare
reform and child well-being. In R. B. Blank & R. T.
Haskins (Eds.), The new world of welfare (pp. 391-417).
Washington, DC: Brookings Institution Press.
Duncan, G. J., Yeung, W., Brooks-Gunn, J., & Smith, J. R.
(1998). Does poverty affect the life chances of children?
American SociologicalReview, 63, 406-423.
Dunifon, R., Kalil, A., & Danziger, S. K. (2003). Maternal
work behavior under welfare reform: How does the
transition from welfare to work affect child development? Childrenand YouthServices Review,25, 55-82.
Edin, K., & Lein, L. (1997). Making ends meet: How single
mothers survive welfare and low-wage work. New York:
Russell Sage Foundation.
Elder, G. H., Eccles, J. S., Ardelt, M., & Lord, S. (1995).
Inner-city parents under economic pressure: Perspectives on the strategies of parenting. Journal of Marriage
and the Family, 57, 771- 784.
Ellwood, D. (1988). Poor support: Poverty in the American
family. New York: Basic Books.
Foster, M. (2002). How economists think about family
resources and child development. Child Development,73,
1904-1914.
General Accounting Office (1994). Early childhoodprograms:
Parent education and income best predict participation.
Report GAO/HEHS-95-47.
Garcia Coll, C., Lamberty, G., Jenkins, R., Pipes McAdoo,
H., Crnic, K., & Hanna Wasik, B., et al. (1996). An
integrative model for the study of developmental
competencies in minority children. Child Development,
67, 1891-1914.
Garner, P. W., Jones, D. C., & Miner, J. L. (1994). Social
competence among low-income preschoolers: Emotion
socialization practices and social cognitive correlates.
Child Development,65, 622-637.
Garrett, P., Ng'andu, N., & Ferron, J. (1994). Poverty
experiences of young children and the quality of their
home environments. Child Development, 65, 331-345.
Gecas, V. (1989). The social psychology of self-efficacy.
Annual Review of Sociology,15, 291-316.
Gennetian, L. A., & Miller, C. (2002). Children and welfare
reform: A view from an experimental welfare program
in Minnesota. Child Development, 73, 601-620.
Gerard, A. B. (1994). The parent-childrelationshipinventory:
Manual. Los Angeles: Western Psychological Services.
Gershoff, E. T., Aber, J. L., & Raver, C. C. (2002). Child
poverty in the U. S.: An evidence-based framework for
programs and policies. In R. M. Lerner, F. Jacobs, & D.
Wertlieb (Eds.), Handbookof applied developmentalscience:
Promoting positive child, adolescent,and family development
(pp. 81-136). Thousand Oaks, CA: Sage.
Ginexi, E. M., Howe, G. W., & Caplan, R. D. (2000).
Depression and control beliefs in relation to reemployment: What are the directions of effect? Journal of
OccupationalHealth Psychology,5, 323-336.
Greenberg, M. T., Lengua, L. J., Coie, J. D., & Pinderhughes, E. E. (1999). Predicting developmental outcomes at school entry using a multiple-risk model: Four
American communities. Developmental Psychology, 35,
403-417.
Greenberger, E., O'Neil, R., & Nagel, S. K. (1994). Linking
workplace and homeplace: Relations between the
nature of adults' work and their parenting behaviors.
DevelopmentalPsychology,30, 990-1002.
Gyamfi, P., Brooks-Gunn, J., & Jackson, A. (2001).
Associations between employment and financial and
parental stress. Womenand Health, 32, 119-135.
Halberstadt, A. G., Cassidy, J., Stifter, C. A., Parke, R. D., &
Fox, N. A. (1995). Self-expressiveness within the family
context: Psychometric support for a new measure.
PsychologicalAssessment, 7, 93-103.
Han, W., Waldfogel, J., & Brooks-Gunn, J. (2001). The
effects of early maternal employment on later cognitive
and behavioral outcomes. Journal of Marriage and the
Family, 63, 336-354.
Harvey, E. (1999). Short-term and long-term effects of early
parental employment on children of the National
Longitudinal Survey of Youth. DevelopmentalPsychology,
35, 445-459.
Hofferth, S. L. (1994). Who enrolls in Head Start? A
demographic analysis of Head Start eligible children.
Early ChildhoodResearchQuarterly,9, 243-268.
Huston, A. C., Duncan, G. J., Granger, R., Bos, J., McLoyd,
V., & Mistry, R., et al. (2001). Work-based antipoverty
Work and Parenting for Low-Income Mothers
programs for parents can enhance the school performance and social behavior of children. Child Development, 72, 318-336.
Isley, S. L., O'Neil, R., Clatfelter, D., & Parke, R. (1999).
Parent and child expressed affect and children's social
competence: Modeling direct and indirect pathways.
DevelopmentalPsychology,35, 547-560.
Jackson, A. P., Brooks-Gunn, J., Huang, C., & Glassman, M.
(2000). Single mothers in low-wage jobs: Financial
strain, parenting and preschoolers' outcomes. Child
Development, 71, 1409 - 1423.
Kalil, A., Dunifon, R. E., & Danziger, S. K. (2001). Does
maternal employment mandated by welfare reform
affect parenting behavior? In G. J. Duncan & L. ChaseLansdale (Eds.), For betteror worse: Welfarereformand the
wellbeing of children and families (pp. 154-178). New
York: Russell Sage Foundation.
Karasek, R. A., & Theorell, T. (1990). Healthy work: Stress,
productivity, and the reconstruction of working life. New
York, Basic Books.
Klebanov, P., Brooks-Gunn, J., Chase-Lansdale, P. L., &
Gordon, R. A. (1997). Are neighborhood effects on
young children mediated by features of the home
environment? In J. Brooks-Gunn, G. L. Duncan, & J. L.
Aber (Eds.), Neighborhoodpoverty, Volume I (pp. 119145). New York: Russell Sage Foundation.
Kline, R. B. (1998). Principles and practice of structural
equation modeling. New York: Guilford Press.
Leadbeater, B. J., Bishop, S. J., & Raver, C. C. (1996).
Quality of mother-toddler interactions, maternal depressive symptoms, and behavior problems in preschoolers of adolescent
mothers. Developmental
Psychology,32, 280-288.
Lennon, M. C., Bloom, J., & English, K. (2001). Depression
and low-incomewomen:Challengesfor TANFand welfare-towork policies and programs. New York: Research Forum
on Children, Families and the New Federalism, National Center for Children in Poverty.
Long, B. C. (1998). Coping with workplace stress: A multiple-group comparison of female managers and clerical
workers. Journalof Counseling Psychology,45, 65-78.
MacPhee, D., Fritz, J., & Miller-Heyl, J. (1996). Ethnic
variations in personal social networks and parenting.
Child Development,67, 3278-3295.
Mayer, S. (1997). What money can't buy: Family income and
children's life chances. Cambridge, MA: Harvard University Press.
McLoyd, V. C. (1990). The impact of economic hardship on
Black families and children: Psychological distress,
parenting, and socioemotional development. Child
Development,61, 311-346.
McLoyd, V. C., & Wilson, L. (1991). The strain of living
poor: Parenting, social support, and child mental health.
In A. C. Huston (Ed.), Children in poverty: Child
developmentand public policy (pp. 105-135). Cambridge,
England: Cambridge University Press.
Menaghan, E., Jekielek, S., Mott, F., & Cooksey, E. (1998,
May). Workand family circumstancesand child trajectories:
1735
When (and for what) does AFDC receipt matter? Paper
presented to the preconference on family process and
child development in low-income families, Joint Center
for Poverty Research, Chicago.
Menaghan, E. G., & Parcel, T. L. (1995). Social sources of
change in children's home environments: The effects of
parental occupational experiences and family conditions. Journalof Marriage and the Family, 57, 69-84.
Meyer, D. R., & Cancian, M. (1998). Economic well-being
following an exit from Aid to Families with Dependent
Children. Journal of Marriage and Family, 60, 479-492.
Mistry, R. S., Vandewater, E. A., Huston, A. C., & McLoyd,
V. C. (2002). Economic wellbeing and children's social
adjustment: The role of family process in an ethnically
diverse low-income sample. Child Development, 73,
935-951.
Morris, P. A. (2002). The effects of welfare reform policies
on children. Social Policy Report, 16, 4-19.
Morris, P., Bloom, D., Kemple, J., & Hendra, R. (2003). The
effects of a time limited welfare program on children:
The moderating role of parents' risk of welfare
dependency. Child Development, 74, 851-874.
Morris, P., Huston, A., Duncan, G., Crobsy, D., & Bos, H.
(2001). How welfare and work policies affect children: A
synthesis of research.New York: Manpower Demonstration Research Corporation.
Mortimer, J. T., & Lorence, J. (1979). Occupational
experience and the self-concept: A longitudinal study.
Social Psychology Quarterly,42, 307-323.
Parcel, T. L., & Menaghan, E. G. (1994). Parents' jobs and
children'slives. New York: Aldine de Gruyter.
Petterson, S. M., & Albers, A. B. (2001). Effects of poverty
and maternal depression on early child development.
Child Development, 72, 1792-1813.
Radloff, L. S. (1977). The CES-D scale: A self-report
depression scale for research in the general population.
Applied PsychologicalMeasurement,1, 385-401.
Raver, C. C. (2002). Psychological well-being, negative
emotional expressiveness, and optimal parenting among
low-incomefamilies. Unpublished manuscript.
Raver, C. C., & Spagnola, M. (2002). "When my mommy
was angry, I was speechless:" Children's perceptions of
maternal emotional expressiveness within the context of
economic hardship. Marriage and Family Review, 34,
63-88.
Rosenfield, S. (1989). The effects of women's employment:
Personal control and sex differences in mental health.
Journal of Health and Social Behavior,30, 77-91.
Rushton, J. P., Brainerd, C. J., & Pressley, M. (1983).
Behavioral development and construct validity: The
principle of aggregation. Psychological Bulletin, 94,
18 - 38.
Schnur, E., Brooks-Gunn, J., & Shipman, V. C. (1992). Who
attends programs serving poor children? The case of
Head Start attendees and. Journal of Applied Developmental Psychology, 13, 405-421.
Tout, K., Zaslow, M., Papillo, A. R., & Vandivere, S. (2001).
Early care and education: Work support for families and
1736
Raver
developmental opportunity for children. Working paper, Urban Institute, Washington, DC.
Waldfogel, J., Han, W., & Brooks-Gunn, J. (2002). The
effects of early maternal employment on child cognitive
development. Demography,39, 369-392.
Wilson, J. B., Ellwood, D. T., & Brooks-Gunn, J. (1995).
Welfare-to-work through the eyes of children. In P. L.
Chase-Lansdale & J. Brooks-Gunn (Eds.), Escape from
poverty:Whatmakesthe difference
for children?(pp. 6386). New York: Cambridge University Press.
Theinnercity,
Wilson, W. J. (1987).Thetrulydisadvantaged:
the underclassand publicpolicy.Chicago: University of
Chicago Press.
Yoshikawa, H. (1999). Welfare dynamics, support services,
mothers' earnings, and child cognitive development:
Implications for contemporary welfare reform. Child
70, 779-801.
Development,
Yoshikawa, H., Rosman, E. A., & Hsueh, J. (2001).
Variation in teenage mothers' experiences of child care
and other components of welfare reform: Selection
processes and developmental consequences. Child De-
velopment,72, 299-317.
Zaslow, M. J., Tout, K., Smith, S., & Moore, K. (1998).
Implicationsof the 1996 welfarelegislationfor children:A
research
SocialPolicyReport,3. Ann Arbor,MI:
perspective.
Society for Research in Child Development.
Zill, N., Moore, K. A., Nord, C. W., Smith, E. W., Stief, T., &
Coiro, M. J. (1995). The life circumstances and development of children in welfare families: A profile based on
national survey data. In J. Brooks-Gunn & P. L. ChaseLansdale (Eds.), Escape from poverty (pp. 38-59). New
York: Cambridge University Press.