Below is the uncorrected machine-read text of this chapter, intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text of each book. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.
Part II Implications for the Development of Children and Adolescents Part II considers the implications of the employment, family, and child care trends described in Part I. It reviews the science in a few specific areas. Chapter 4 reviews the research on maternal employment and its effects on the family. It describes how maternal employment is associated with a wide range of both positive and negative patterns of development for chil- dren and considers the ways in which it affects different subgroups of children. Chapter 5 looks at early child care and school-age child care settings and considers the effects of care on children. It reviews conceptual and methodological advances that have informed recent research, reviews a set of large-scale studies that have examined the implications of child care quality on child developmental outcomes, and reviews evidence on associa- tions between child care quality and various dimensions of the care. Chapter 6 reviews the evidence on parental employment and a particu- lar group of children--adolescents. It includes an overview of the salient tasks of adolescence, highlighting the opportunities and challenges that adolescents with working parents face. Chapter 7 reviews evidence on the effects of welfare reform on the family, with particular attention to employment, earnings, poverty, fertil- ity, and marriage, as well as the effects on children and adolescents.
4 Maternal Employment and the Family Environment T he evidence presented on trends in work and child care indicates that more mothers work and children and adolescents spend significant time in nonparental care. The committee's next step was to explore the extent to which these trends affect the development and well-being of children and adolescents in these families. This chapter re- views evidence on how maternal employment, particularly employment of low-income families, appears to affect the home environment of children and how that, in turn, affects children. Chapter 5 goes on to focus on the effects of child care environments on young children, while Chapter 6 focuses on the effects of care environments on adolescents. The basis historically for research on maternal employment and chil- dren was premised with a straightforward and negative question: Does a mother's employment harm her children's development (Bianchi, 2000b; Gottfried et al., 1995)? This question emerged when more mothers, espe- cially mothers with young children, began to enter the workforce. There was growing concern that substantial periods of time when a mother is inaccessible to her child, especially a young one, could affect the child's sense of his or her relationship with the mother as a source of comfort and as a safe base for exploring the environment (Ainsworth et al., 1978; Bowlby, 1969). It was hypothesized that not having reliable access to the mother would have unfavorable implications for the child's social, emo- tional, and cognitive development. 67
68 WORKING FAMILIES AND GROWING KIDS A RANGE OF PATTERNS As the research on maternal employment has accumulated over a pe- riod of decades, it has become increasingly clear that the evidence does not support this negative hypothesis. The findings do not fall into a straightfor- ward unidirectional pattern. Rather, the research indicates that: 1. maternal employment is associated with a wide range of patterns of development for children, ranging from positive to neutral to negative; and 2. the differences in developmental status that have been found for children of employed and nonemployed mothers have generally been found for specific and delineated population subgroups--for example, for specific age ranges but not others, for boys but not girls (or vice versa), and for children in families at some socioeconomic levels but not others. Reviews of the research show key patterns (see, for example, Hoffman, 1979, 1984, 1989; Hoffman and Youngblade, 1999; National Research Council, 1982; Zaslow and Emig, 1997). In their recent review, Hoffman and Youngblade (1999) found evidence to support the following patterns: on the positive side, school-age and adolescent daughters of employed moth- ers show higher academic aspirations and achievement and are more likely to make nontraditional role choices than are daughters of nonemployed mothers, and both sons and daughters of employed mothers have less tradi- tional attitudes about gender roles. The evidence further indicates that when preschool and school-age children in poverty show differences in development in light of their mothers' employment status, they also show more favorable cognitive and socioemotional outcomes. On the negative side, some findings indicate lower school performance and academic achievement during middle childhood for middle-class sons of employed mothers. Hoffman and Youngblade (1999) point to an emerg- ing pattern of findings suggesting that maternal employment may be associ- ated with unfavorable developmental outcomes for children when the em- ployment is resumed in the child's first year and is extensive (full time rather than part time). Findings providing further evidence of such a pat- tern have continued to emerge since the publication of that review (Brooks- Gunn et al., 2002; Han et al., 2001; Waldfogel et al., 2002). As the research on maternal employment has evolved, the possible reasons for this more complex patterning of results have become increas- ingly clear. Three potential explanations have been identified: · Maternal employment influences family life in multiple ways simul- taneously, with influences sometimes in counterbalancing directions. A clearly articulated hypothesis in the research on maternal employment is
MATERNAL EMPLOYMENT AND THE FAMILY ENVIRONMENT 69 Parents not only affect their children's psychological development, they also introduce them to the world of work (Galinsky, 1999:355): You have to work and teach your children how to work. Also, let them know how to do something they'll enjoy. You have to work to get what you want in life. that this "status" variable (i.e., whether mothers are employed or not) does not affect children directly, but rather affects them to the extent that it brings about changes in their immediate experiences (Gottfried et al., 1995). Furthermore, maternal employment appears to affect multiple aspects of children's environments simultaneously, and it may do so in contrary direc- tions. In recent ethnographic work, mothers making the transition from welfare to work themselves articulated this idea of multiple counterbalanc- ing influences of their employment on their children (London et al., 2000). For example, mothers moving from welfare to employment see themselves as providing resources for the family and better role models for their chil- dren. At the same time, they perceive themselves as less available to their children and express concern about their ability to supervise them. Find- ings of neutral or small associations of maternal employment with child outcomes may actually reflect counterbalancing influences in the family rather than an absence of influences. · Families actively adapt to the mother's employment patterns. The evidence suggests that families do not respond passively to the mother's employment, but rather actively compensate for hours the mother is away from the child, for example, through a reallocation by mothers of time spent in leisure to time spent with children and a redistribution of house- hold tasks between parents (as explained in Chapter 2). The extent to which maternal employment is associated with children's outcomes (or instead shows limited or neutral patterns of association) may reflect the resources that the family has to make such active adaptations. Families with low incomes and complex work schedules that do not permit flexibility, or single-mother families with fewer resources to draw on, may have less capacity to make active adaptations. · Maternal employment is not a unitary variable in itself, but rather reflects key variations in employment circumstances. Studies that distin- guish simply whether a mother is employed or not (or employed full time, part time, or not at all), may overlook other key variations in employment circumstances that are important to children. Recent research suggests, for
70 WORKING FAMILIES AND GROWING KIDS example, whether hours of employment are standard or nonstandard and such job characteristics as degree of autonomy on the job may be linked to family processes and child outcomes (Menaghan and Parcel, 1995; Han, 2002a). It is becoming increasingly clear that we need to move beyond the simple identification of employment status to capture such variation in job characteristics in order to understand influences on children. STRENGTHS AND LIMITATIONS OF THE RESEARCH Recent research has made substantial strides in examining maternal employment and child outcomes in low-income families. Whereas earlier research focused mostly on middle-class families, the focus more recently has been on families participating in the transition from welfare to work (Grogger et al., 2002; Morris et al., 2002; Zaslow et al., 2002) as well as low-income families in general (Phillips, 2002; Tout et al., 2002). The research on welfare-to-work programs studies what happens to children and families in light of whether the mother participated in a welfare-to- work program, rather than her transition to employment per se. Across a range of program approaches (for example, programs that mandate work without providing strong supports for employment, programs that combine work mandates with financial incentives for working, programs with time limits on welfare receipt), these programs have brought about increases in maternal employment. As such, these studies provide a context for consid- ering child outcomes when employment increases as a result of welfare-to- work programs. Another important development in the research is an explicit focus on the processes underlying associations between maternal employment and child outcomes, such as family economic resources, maternal parenting behavior, father involvement, and maternal psychological well-being. While this is an important step in the research, it is necessary to acknowledge that this approach is as yet limited. There is not a literature that can point to key underlying processes across the full range of child ages and population subgroups. And studies tend to provide fragments of the picture, linking maternal employment to underlying processes, and maternal employment to child outcomes, but not completing the picture by examining whether and how the underlying process helps to explain the association between maternal employment and child outcomes (or even a step further, consider- ing how multiple processes function simultaneously). For example, a study may consider whether father involvement differs in families with and with- out an employed mother, but not whether father involvement is a key process in explaining the link between maternal employment and child outcomes. We are limited to a small set of studies that explicitly test the
MATERNAL EMPLOYMENT AND THE FAMILY ENVIRONMENT 71 role of specific underlying processes in explaining the associations between maternal employment and child outcomes.1 A persisting issue in this body of work is whether differences by em- ployment status in family processes or outcomes for children reflect differ- ences in the characteristics of families in which the mother did and did not become employed rather than differences due to employment (see the dis- cussion of these "selection effects" in Blau, in press; Vandell and Ramanan, 1992). Zaslow and colleagues (1999) note the substantially differing con- clusions that are reached when a study simply describes differences in child outcomes in light of mother's employment status, or whether it seeks to control for the child, family, and broader social context factors that can predict to both maternal employment and child outcomes. In this chapter, we reserve the terms "effects" and "impacts" to de- scribe the results of studies using experimental designs (see Box 4-1 for elaboration on research terms used in this chapter and the rest of the report), specifically evaluations of welfare reform programs that sought to encourage or require employment. When discussing findings from non- experimental studies of maternal employment, because of concern with variation across studies in how well selection effects are accounted for, we do not use the terminology of "effects" of maternal employment on families or children, but rather restrict ourselves to describing "associations" of maternal employment or of "implications" of maternal employment for families and for children. And we restrict our focus to studies that, at the least, control for background characteristics that may predict both mater- nal employment and child outcomes. MATERNAL EMPLOYMENT AND THE FAMILY ENVIRONMENT Hoffman and Youngblade's review of the research (1999) hypothesizes that there are three key aspects of the family environment that differ in light of the mother's employment status and that in turn may be important to children's development: parenting behavior and the home environment, father involvement, and mother's psychological well-being. In addition, the work on maternal employment in the context of welfare reform adds a fourth key element to this list: family economic resources, which may in turn affect any of these three factors. We provide a brief overview of the evidence here, starting with the findings on family economic resources, and returning to the set of factors hypothesized by Hoffman and Youngblade. 1These studies use the statistical approach to studying mediation developed by Baron and Kenny (1986).
72 WORKING FAMILIES AND GROWING KIDS BOX 4-1 Research Terms Experimental design: involves the random assignment of individuals to either a treatment group (in this case, participation in the program being assessed) or a control group (a group that is not given the treatment). Many believe that the experimental design provides some of the strongest, most clear evidence in re- search evaluation. This design also affords the highest degree of causal infer- ence, since the randomized assignment of individuals to an intervention condition restricts the opportunity to bias estimates of the treatment effectiveness. Nonexperimental design (also known as correlational methods): does not in- volve either random assignment or the use of control or comparison groups. These designs gather information through such methods as interviews, observations, and focus groups, and then examine relations or associations among variables in an effort to learn more about the individuals receiving the treatment (participating in the program) or the effects of the treatment on these individuals. Nonexperimental studies sometimes use statistical techniques to control for such factors as matura- tion, self-selection, attrition, or the interaction of such influences on program out- comes. The concern, however, is that unmeasured factors or variables may ac- count for obtained relationships. Multivariate analysis: any analysis in which two or more dependent variables are included in a single analysis. Hierarchical regression: predictors are entered in analyses in a sequential order in which "control" variables are entered first followed by the selected variables of interest. Researchers are seeking to answer the question, "Does variable x predict outcome y after variables a, b, and c are controlled?" Psychometrics: the branch of psychology that evaluates the reliability and validity of different measurement techniques. Reliability refers to the consistency of mea- surement over time, across raters or observers, or across individual items of a survey. Validity refers to whether the measure is assessing the construct of inter- est. Effect size: calculated as the difference in means between the treatment and the control group divided by the standard deviation of the control group or the differ- ence between the value specified in the null hypothesis and the research hypoth- esis. The larger the effect size, the more powerful the test because the difference between the sample and the null hypothesis mean will be farther apart, thus in- creasing the probability of rejecting the null hypothesis. Econometrics: The branch of economics that applies statistical methods to an- alyze data.
MATERNAL EMPLOYMENT AND THE FAMILY ENVIRONMENT 73 Family Economic Resources A mother's employment may affect children through the economic re- sources she provides for the family (Huston, 2002). Researchers hypothesize that the relative importance of the economic contribution from maternal employment is greater in lower-income than higher- income families (Desai et al., 1989), and that this may help account for the tendency of maternal employment to have positive implications in lower income families (Zaslow and Emig, 1997). Recent work (Dearing et al., 2001) indeed provides evi- dence that changes (both increases and decreases) in families' income-to- needs ratio (the ratio of total family income to poverty threshold for the appropriate family size) are much more important for cognitive and social outcomes of children in poor than nonpoor families. A child warns parents that neglected children could become problems to society now or in the future (Galinsky, 1999:350): It is good to work, and it definitely makes finances better for a family with two sources of income. Just don't alienate your children or let them do whatever they want whenever they want because that could get them in trouble. Economic resources can derive not only from earnings, but also through benefits intended to support employment, such as financial work incen- tives, including the federal and state earned income tax credits and child care subsidies (Zedlewski, 2002). The mother's contribution to overall family income may be of importance to children by influencing the ad- equacy of food, clothing, and housing; safety from injury and from danger- ous elements in the physical environment (for example, from environmental toxins and violence); and by ensuring health care services (Huston, 2002). Family economic resources also contribute to the number and variety of toys and books available to the child in the home and the extent to which the family can engage in stimulating outings (e.g., Bradley, 1995; Bradley and Caldwell, 1984b; Bradley et al., 1988). While there is substantial research looking at the link between family economic resources and child outcomes (Duncan and Brooks-Gunn, 2000; Duncan et al., 1994; Chase-Lansdale et al., 2003), very little research has looked at the role of economic resources in transmitting the implications of maternal employment to children, despite the fact that this is one of the main reasons for working and may help to shape other changes in the family (such as changes in maternal mental health).
74 WORKING FAMILIES AND GROWING KIDS Experimental studies of welfare reform programs provide evidence sug- gesting the importance of the economic implications of employment for children. Evidence looking across multiple evaluation studies of programs to encourage or require employment among families receiving welfare (Grogger et al., 2002; Morris et al., 2002; Zaslow et al., 2002) concludes that favorable impacts on children's development tend to occur in pro- grams in which an increase in maternal employment was accompanied by an increase in family income. This pattern of increases in both employment and income occurred most consistently in programs that provided strong financial incentives for working (for example, through earned income disre- gards, which allow parents to keep more of their welfare benefits while working). Examples of programs with strong financial work incentives with positive impacts on young school-age children include the Minnesota Family Investment Program (Gennetian and Miller, 2000), New Hope (Bos et al., 1999), and the Canadian Self-Sufficiency Program (Morris and Michalopoulos, 2000).2 This pattern of favorable impacts on children did not typically occur in programs that increased employment without increasing overall family in- come (except in instances in which there was a program impact involving an increase in maternal educational attainment, for example, in a subset of the six JOBS programs, studied in the National Evaluation of Welfare-to- Work Strategies Child Outcomes Study; McGroder et al., 2000). Beyond the studies in which mothers increased their educational attainment, there are also indications that in programs in which the families did not make economic progress or actually experienced a setback on one or more of these economic outcomes, impacts for children fell in the neutral to unfa- vorable range (for example, in the New Chance Demonstration and se- lected sites of the Teenage Parent Demonstration, welfare reform programs for adolescent mothers; Quint et al., 1997; Kisker et al., 1998). Recent reviews of research on welfare reform programs noting links between the economic impacts for families and the impacts for children have particularly found outcomes for children related to cognitive develop- ment and academic achievement, but also for behavioral outcomes. There were few impacts at all in these evaluations on outcomes related to children's health. Although health outcomes were studied in the least detail in these evaluations, the pattern of findings in these evaluations parallels the pattern of findings linking economic resources and child outcomes directly: out- comes related to intellectual achievement are most consistently found to be 2Experimental evaluations of welfare reform programs done in five of the states that were granted welfare waivers in the years prior to the 1996 welfare reform (Connecticut, Florida, Indiana, Iowa, and Minnesota) reported similar findings.
MATERNAL EMPLOYMENT AND THE FAMILY ENVIRONMENT 75 related to family income, with less evidence of a link with behavioral ad- justment or health and safety outcomes (Dearing et al., 2001). Effect sizes of the impacts on children in the experimental evaluations of welfare re- form fell in a range from about 0.10 to 0.80, with most falling at the lower end of the range. As discussed by Zaslow and colleagues (2002), while the effect sizes tended to be smaller than those found in the most successful programs aimed directly at improving the development of young children (such as the Abecedarian and High/Scope programs), they are comparable to the effect sizes for other programs focusing on young children (such as Early Head Start and the Tennessee STAR class size reduction program). These findings suggest that, beyond employment per se, the circumstances that surround and follow from it, including the implications of employment and associated benefits for families' overall economic circumstances, are important for children. Zedlewski (2002) finds the evidence on economic resources of families following welfare reform to provide a complicated picture. Labor force participation has increased among single mothers with young children over- all, among welfare recipients, and among families leaving welfare. Con- cerning the economic well-being of these families, however, findings differ according to what elements are included in the calculation of family eco- nomic resources. With only cash income taken into consideration, poverty has declined in the years since welfare reform. However when total family income is considered, including noncash benefits as well, studies suggest that a portion of families are faring worse in the years since welfare reform. The evidence indicates that while participation in the earned income tax credit is strong among eligible families, a substantial proportion of eligible families are not receiving food stamps, Medicaid, or child care benefits for which they are eligible. An important step for the research on maternal employment will be to take a closer look at how family economic resources are defined, in order to determine which approach best helps to explain associations be- tween maternal employment, family resources, and children's outcomes. Parenting Behavior In studying the implications of employment for family life, particular emphasis has been placed on parenting behavior. This appears to be the case for two reasons: first, parenting behavior and the home environment appear to serve as conduits through which a broader set of influences on the family are conveyed to the child. For example, McLoyd (1990) summarizes evidence from a range of studies indicating that family economic stress is conveyed to children partially through the psychological distress it creates in parents and a resulting tendency to show harsher and less supportive parenting, which in turn predicts children's social and emotional outcomes.
76 WORKING FAMILIES AND GROWING KIDS Second, in studies that simultaneously examine the role of the home envi- ronment and of children's experiences in child care, results point to a relatively greater influence of the home environment (National Institute of Child Health and Human Development, 2001b). Thus, if maternal em- ployment leads to changes in children's relationships and interactions with parents, this is likely to be an important pathway through which their development is influenced (Huston, 2002). Much of the research on parenting has focused on maternal rather than paternal behavior. We next review evidence on the links between maternal parenting behavior and the quality of stimulation and support available to children in the home environment. The issue of father involvement is covered in a subsequent section. Maternal Parenting Chase-Lansdale and Pittman (2002) identify six dimensions of parenting: (1) gatekeeping, (2) warmth and responsiveness, (3) control and discipline, (4) cognitive stimulation, (5) modeling, and (6) family routines and tradi- tions. In a review of the evidence on parenting behavior in light of welfare reform, these researchers note that initial expectations were that there would be a range of effects on parenting behavior, including an increase in family routines when mothers were employed and more positive role modeling. Focusing on the findings on parenting in the experimental evaluations of welfare reform programs, they find that employment in the context of reform has actually had limited effects on parenting. Rather than cutting across the differing dimensions of parenting, Chase-Lansdale and Pittman see the effects for preschool and school age children in these studies as concentrated prima- rily in terms of an enhancement in the gatekeeping aspects of parenting, which involve not direct interactions or structuring of the home environment, but rather oversight and guidance of the child's activities outside the home. For example, gatekeeping encompasses selection of child care contexts and out of school lessons and activities, as well as monitoring the child's where- abouts, activities, and friendships. Their review finds that mothers of young children in a number of welfare reform programs (though not all) were more likely to make use of formal child care arrangements for their young children. They note that use of more formal child care may be related to better academic school readi- ness and, for school-age children, adjustment and progress in school. For example, mothers in the New Hope program were more likely to place their children, especially their sons, in structured after-school programs; sons in particular in this program showed positive program impacts in terms of teacher-rated measures of behavior in the classroom and the children's own educational and occupational aspirations (Bos et al., 1999).
MATERNAL EMPLOYMENT AND THE FAMILY ENVIRONMENT 77 An important extension of the gatekeeping hypothesis for preschool and school-age children comes from the recent work of Gennetian and colleagues (Gennetian et al., 2001; Crosby et al., 2001). These researchers distinguish between two types of welfare reform programs in terms of child care benefits: those that provide enhanced child care supports for families in the program group, and those that simply offer more of the same child care supports and benefits as employment increases, not distinguishing the nature of supports for those in the program and the control groups. En- hanced child care supports in the former group include any one (or a combination) of the following: (1) provide information and support specific to child care through case management or access to resource and referral agencies, (2) improve the process of reimbursement for child care, (3) con- tinue eligibility during the transition off welfare, (4) promote the use of formal child care through financial or other means, and (5) restrict the provision of subsidies to regulated care. In analyses of data from multiple welfare reform programs, this re- search finds that while such programs generally increase reliance on child care (as mothers increase their work preparation and employment activi- ties), it is specifically in the programs with enhanced child care supports that mothers increased their reliance on formal child care settings, such as child care centers and organized after-school programs (showing larger absolute impacts on use of formal care and larger relative impacts of center compared with home-based child care). The pattern of differential increase in center compared with home-based care was apparent for preschool-age and school-age children, although the evidence for the older children was less extensive. These findings are important in indicating that the nature of the sup- ports available to parents matter for their choice of type of care; that is, resources and information help parents guide their children's experiences. While some researchers have noted that the relatively greater reliance of low-income families on home-based child care may reflect a preference for such care, these new findings raise the possibility that this pattern reflects, at least in part, a resource constraint. Lowe and Weisner (2001) provide ethnographic data suggesting that low-income families recognize the strengths of formal care arrangements but also have some concerns about this type of care (e.g., suspicion about care provided by someone without an existing relationship with the family). The results of the analyses using data from experimental studies suggest that when resources and informa- tion are available, low-income families will make relatively more use of formal child care and activities. Turning to the dimensions of parenting other than gatekeeping, pro- gram impacts on the more dyadic/interactional aspects of parenting (cutting across the dimensions noted by Chase-Lansdale and Pittman of warmth and
78 WORKING FAMILIES AND GROWING KIDS responsiveness, control and discipline, and cognitive stimulation) did occur in families with preschool and school-age children, albeit in a less concentrated pattern when looking across multiple measures of parenting in a particular study. Zaslow and colleagues (2002) report that of 20 analytic comparisons (involving program variations or different subgroups of families participating in 7 experimental evaluation studies), impacts on at least 1 in 16 parenting measures were found, although this usually involved an impact on only 1 or a few of multiple parenting measures examined. There is some evidence that the direction of impacts on these more dyadic aspects of parenting (whether favorable or unfavorable) corre- sponded to the pattern of economic progress for families. The strongest evidence for this comes from looking at patterns of parenting in the evalu- ations of specific programs in which economic impacts differed for sub- groups or variants of the program. In the Minnesota Family Investment Program (Gennetian and Miller, 2000), for example, long-term recipients of welfare assigned to the program group showed increases in income as well as employment. For this subgroup, the version of the program that involved financial incentives without a work requirement decreased harsh parenting of school-age children as reported by mothers. In contrast, the version of the program used with recent applicants resulted in increased employment but not increased family income. For this subgroup, the result was an increase in harsh parenting. McGroder and colleagues (2002) provide exploratory evidence that the impacts on dyadic aspects of parenting, when they did occur, helped to explain program impacts on child outcomes for the school-age children in the JOBS program. For example, two years after assignment to the Atlanta labor force attachment JOBS program (a program emphasizing work first rather than starting with education or training before attempting to locate employment), there was a positive program impact on 5- to 7-year-old children's school readiness scores. Positive parenting and mothers' verbal interactions with the child, both of which improved as a result of the program, were found to partially mediate this favorable child impact. While the labor force attachment program in Atlanta increased mothers' employ- ment and decreased the proportion of families in the program group rela- tive to the control group living in deep poverty, the same program approach at another site, Grand Rapids, Michigan, had different economic impacts, with no increase in employment and a decrease in the proportion of pro- gram group families living at or above the poverty line. The Grand Rapids labor force attachment program increased children's antisocial behavior problems at the two-year follow-up, and McGroder and colleagues found this impact to be completely mediated by decreases in maternal reports of warmth in parenting as well as increases in maternal depression. McGroder and colleagues caution that these mediational analyses, carried out in an
MATERNAL EMPLOYMENT AND THE FAMILY ENVIRONMENT 79 experimental framework, should be viewed as exploratory, in that while covariates controlled for selection factors (and the extensive set of baseline variables made it possible to control for a broad range of initial character- istics), the possibility nevertheless remains that further (unobserved) vari- ables should have been taken into account. Regarding these more dyadic aspects of parenting, Chase-Lansdale and Pittman (2002) also note that a substantial number of states are allocating temporary assistance for needy families (TANF) funds to provide parenting education programs or home visitation to families. It is interesting to note that one of the demonstration programs, New Chance, involves a parenting education component. One study (Zaslow and Eldred, 1998) found posi- tive program impacts on parental support and cognitive stimulation of mother-child interaction. However, the program group children did not, over time, show positive program impacts on outcomes, and behavior at the final follow-up in the evaluation showed some negative impacts. While positive parenting behavior did predict more favorable child development outcomes in this sample over time, there were important negative influences on development that appeared to counterbalance positive parenting, in- cluding very low income, difficult life circumstances, residential mobility, and isolation or lack of social support for the very young welfare-receiving mothers in this study. These results suggest that parenting education has the potential to influence parenting behavior positively, even among very disadvantaged welfare-receiving families (in this case, adolescent mothers who had dropped out of school), yet that the broader social context of these families in contributing to child outcomes needs to be taken into account as well. Chase-Lansdale and Pittman further caution that features of parenting edu- cation programs, such as the skill level and training of program providers, the dosage (amount of time spent in care) of the program, and the motiva- tion of the parents to participate, are all likely to be important to the impacts on parenting of such programs. It is also important to consider the implications of maternal employ- ment for parenting in relation to differing developmental periods as well as the challenges and resources maternal employment may provide in specific socioeconomic contexts. Parenting of Adolescents Experimental studies of welfare reform and children have had an inten- sive focus on preschool-age children (with follow-up into their school years) rather than on older or younger children. As in the broader set of studies on maternal employment, the research focusing specifically on transition to work in the context of welfare reform has followed (and sometimes raced
80 WORKING FAMILIES AND GROWING KIDS to catch up with) changes in demographics and policy. The focus on preschool-age children in studies conducted in the 1990s occurred because the wave of welfare reform in 1988, for the first time, required mothers of children age 3 years and older (and younger only at state option) to partici- pate in work-related activities. There was concern that changes in employ- ment would necessitate changes in the care routines of these children to a greater extent than children already in school, and that any changes in the home environment would also be of greater significance for younger than older children. Only a limited number of experimental evaluations have included mea- sures of adolescent development, and the measures included have generally been brief indicators of adjustment rather than the kind of in-depth mea- sures used for younger children. There was little anticipation that changes in mothers' employment status or assignment to a welfare reform program could change adolescents' experiences or development in a substantial way. But results from these studies have moved the focus on this age group from peripheral to central. Across a number of very different programs (in which employment and income impacts varied, and for whom impacts for younger children tended to correspond with these economic patterns), findings for adolescents, when they did occur, were consistently unfavorable (Brooks et al., 2001; Gennetian et al., 2002b). These occurred on such outcomes as parent report of adolescent school achievement, behavior problems in school, adolescent participation in delinquent activities, and substance use (see, for example, Bloom et al., 2000; Gennetian and Miller, 2000; Hamilton et al., 2001; Morris and Michalopoulos, 2000). The unfavorable impacts for adolescents have led researchers to ask what the underlying processes for such a pattern might be and to consider the possibility that relationships and roles in the family may change when mothers become employed during the transition from welfare to work. Brooks and colleagues (2001) view the findings in the context of particular developmental tasks of the adolescent period, especially the need to move toward greater autonomy and assumption of responsibility, yet with an appropriate range of parental oversight and monitoring (Eccles et al., 1993). The possibility exists that the assumption of a demanding employment role by the mother alters relationships in the family in such a way that the balance of autonomy, responsibility, and parental oversight is pushed be- yond an appropriate range, especially in school and neighborhood contexts in which increased autonomy may also involve increased exposure to nega- tive influences (see Chapter 6). While this work is in early hypothesis-building phases, Brooks and colleagues examined preliminary evidence for three different patterns that might be indicators that the balance has tipped beyond an appropriate
MATERNAL EMPLOYMENT AND THE FAMILY ENVIRONMENT 81 range: (1) an increase in harsh parenting in families participating in a welfare reform program, hypothesized to occur if there is increased friction surrounding adolescent autonomy; (2) diminution in parental monitoring among mothers in welfare reform program groups as a result of employ- ment demands, which could result in too great autonomy, particularly in high-risk neighborhoods; and (3) increased assumption of adult-like roles by adolescents to help the family function after the mother becomes em- ployed, but to an extent that exceeds a positive range (that is, regularly caring for siblings for prolonged periods). Some evidence exists for each of these possibilities in the experimental evaluation studies, although the evidence regarding monitoring is mixed. For example, regarding the first hypothesis, in the Canadian Self-Sufficiency Project, one of the programs in which adolescents showed negative program impacts, program group mothers reported an increase in their use of harsh parenting with their adolescents ages 15 to 18 (Morris and Michalopoulos, 2000). Regarding monitoring, studies to date have examined this aspect of parenting for the younger children in the family but not adolescents. In order to be seen as pertaining to adolescents, it must be assumed that there is a general tendency in families for mothers to show increased or decreased monitoring across all the children in the family when they participate in a welfare reform program (whereas issues of monitoring may well be specific to age groups). While one program (the Florida Transition Program) did lead to a slight decrease in the monitoring of younger children (Bloom et al., 2000), another (the Minnesota Family Investment Program; Gennetian and Miller, 2002) increased monitoring of younger children for the group of long-term welfare recipient families. Regarding the hypothesis of more adult-like roles, in the Canadian Self-Sufficiency Project, adolescents were more likely to be working 20 or more hours a week and performed household chores more frequently when their mothers were in the program group. In the Florida evaluation, mothers were more likely to report that their younger children were being cared for by a sibling (presumed to be an adolescent). Gennetian and colleagues note that the unfavorable impacts on adolescents in the ex- perimental evaluations were concentrated among the adolescents with a younger sibling, suggesting that responsibility for sibling care may be one contributing factor. The implications of these differing hypotheses for programs and poli- cies differ. For example, the monitoring hypothesis suggests the need for out-of-school youth activities to provide supervision and meaningful activi- ties to adolescents. Yet the hypothesis of assumption of adult-like roles suggests that the adolescents may not be able to participate in youth activi- ties if they have substantial responsibilities at home during the hours that younger siblings are out of school. Further research is needed, focused on
82 WORKING FAMILIES AND GROWING KIDS which one or more of these hypotheses helps to explain the unfavorable program impacts on adolescents, to help clarify appropriate program and policy responses. Recent work by Brooks and colleagues (2001) is taking a first step in this direction. Research is considering mothers' employment status, family processes, and adolescent outcomes with the new cohort of the National Longitudinal Survey of Youth (NLSY, the 1997 cohort). This work focuses on single mothers in low-income families (below 200 percent of the federal poverty level) with 12- to 14-year-olds. These are nonexperimental analy- ses seeking further insight into the pattern found in the experimental evalu- ations. Findings thus far underscore the importance of examining maternal employment separately in three groups of families: those with current or recent welfare receipt, those with welfare receipt in the past five years, and those with no previous welfare receipt. Maternal employment was related to teens' reports of more favorable relationships with their mothers and lower delinquency in families currently or recently receiving welfare. In contrast, mothers' employment was related to adolescents' reports of lower quality relationships with their mothers and less maternal monitor- ing--although not to differences in adolescent outcomes--in families with previous but not recent welfare receipt. In considering why maternal em- ployment might have differing implications for families with differing histo- ries of contact with the welfare system, Brooks and colleagues (2001) raise the possibility that those with current receipt may have more access to resources and caseworker support than those previously but not currently associated with the welfare system, though they note that it is impossible to rule out differences in selection into employment and welfare as helping to account for these patterns. With this new focus on maternal employment and adolescent develop- ment, Chapter 6 reviews the evidence on adolescent development in light of mothers' employment status for more economically heterogeneous samples. Parenting of Infants The experimental studies of welfare reform and children have another important gap with respect to child age: these studies rarely considered program impacts on infants and, when they did so, did not use extensive or in-depth measures (see the review of these measures and findings in Zaslow et al., 2002). This is a serious gap, given the work requirements in the 1996 welfare reforms imposed on mothers with infants and toddlers. A recent set of studies using data from major longitudinal studies with in-depth mea- sures of young children's development as well as parenting suggest that early and extensive maternal employment may have negative implications for the cognitive development of children in specific population subgroups.
MATERNAL EMPLOYMENT AND THE FAMILY ENVIRONMENT 83 They suggest that a key aspect of parenting for the infancy period, maternal sensitivity, may be affected over time when maternal employment is both early and extensive. It is important to note that these new studies do not focus specifically on the transition from welfare to work. Brooks-Gunn and colleagues (2002), in a study focusing on white non- Hispanic families in the Study of Early Child Care of the National Institute of Child Health and Human Development (NICHD), found that when mothers worked 30 or more hours per week by the 9th month of a child's life, the mothers scored lower on sensitivity to the child at 36 months than those who did not work this extensively early on. This pattern held even when controlling for previous maternal sensitivity. Furthermore, children of mothers who worked 30 or more hours per week by the 9th month scored lower on a measure of cognitive school readiness at 36 months than children of mothers who did not work full time early on. Maternal sensitiv- ity and the quality of the home environment as well as the quality of child care helped to explain the relationship between extensive early maternal employment and children's scores on the school readiness measure. In analyses with another dataset, the National Longitudinal Survey of Youth-Child Supplement, Waldfogel and colleagues (2002) found nega- tive implications of maternal employment of 21 or more hours per week in the first year of life for the cognitive development of children in white (though not Hispanic or black) families. Negative implications persisted through age 7 or 8 and were stronger for mothers who worked more hours. The pattern differed by income group, with the strongest pattern found for children in the lowest income families. While sensitivity was not measured in this survey, a measure of the home environment (includ- ing supportiveness of the mother toward the child) was included. Taking into account the quality of the home environment reduced but did not eliminate the association of early maternal employment with children's cognitive outcomes. These findings suggest that maternal employment, if resumed early and extensively, may for some families hinder the emergence of mother-infant sensitivity. There may be insufficient time for mothers to learn their in- fants' cues and develop patterns of responding to them. It is important to note that the findings in these studies do not cut across subgroups but tend to pertain to specific and delimited groups (as, for example, white but not Hispanic or black families in the NLSY analyses). It is important to pursue further the question of why these patterns are occurring in some subgroups and not others and in general to test the generalizability of the pattern. A study focusing explicitly on length of maternity leave during the first year of life and mother-child interaction (Clark et al., 1997) found differ- ences according to whether the return to employment occurred at 6 weeks
84 WORKING FAMILIES AND GROWING KIDS or at 12 weeks. Interestingly, the study for this sample was predominantly white (as well as middle class). In observations of mother-child interaction when the babies were 4 months old, mothers with the shorter (6 week) maternity leave showed more negative affect and behavior with their in- fants than mothers with the longer (12 week) maternity leave. While there was a direct relationship between the timing of leave and negative maternal behaviors with the infant, this was not the case for positive maternal behav- iors. However, statistically significant interactions were found between mothers' depressive symptoms and length of leave, as well as between infant temperament and length of leave, for positive maternal behaviors. Among mothers with higher levels of depressive symptoms, those who had taken a shorter maternity leave were observed to show less positive behav- ior with their infants than those who had taken longer leaves. In contrast, there was no difference in the positive behaviors of mothers according to the timing of leave when depressive symptoms were low. Similarly, when infants had more difficult temperament, the extent of maternal positive behaviors in interactions varied by length of leave, but this was not the case when infants did not show difficult temperament. While the study by Clark and colleagues (1997) suggests that the timing of return to employment even in the first months of the infant's life may be important to the quality of mother-infant interaction, other work suggests that time together may continue to be important to the quality of mother- child interaction even beyond the first year. The NICHD Study of Early Child Care reports findings suggesting that "the amount of time that moth- ers and children spend together is associated with the ease of their interac- tion and communication" (1999a:1410). It also appears that the presence of a husband/partner is positively associated with maternal sensitivity. However, the pattern reported in the NICHD study, of more hours in child care predicting less maternal sensitivity in interactions with the child, held across the first three years (rather than pertaining specifically to extensive use of child care during the first months of life). Thus, further work is needed examining whether time together contributes to maternal sensitivity and the affective quality of mother-child interaction specifically (or perhaps more strongly) during the first months of a child's life, or beyond this period as well. Parenting in Very-Low-Income Families The findings we have reviewed to this point, focusing very heavily on low-income families, with much of the evidence reviewed coming from the welfare reform experimental evaluations, underscore the importance of taking child age into account in considering the implications of maternal employment and parenting. There are, perhaps, the rough outlines of an
MATERNAL EMPLOYMENT AND THE FAMILY ENVIRONMENT 85 extension of the "stage environment fit" hypothesis proposed by Eccles and colleagues (1993) regarding the development of young adolescents. This hypothesis suggests that development in early adolescence will depend on the fit of the environment with the developmental tasks of the period, such as the degree to which the social context permits adolescents an increasing role in decision making (on which middle school classrooms are found to vary substantially). Extending this rubric of fit to the maternal employment research, such employment may provide a context that poses challenges to the salient parenting tasks of some developmental periods but fosters the salient tasks of other developmental periods, depending on the broader context in which parenting occurs. In very-low-income families, maternal employment may foster the tasks of parenting that involve directing the child to positive out-of-home care environments during the preschool years and middle childhood (gatekeep- ing) when employment results in increasing resources or information for identifying such care. However, for parenting in the earliest years, the possibility exists that extensive hours of employment begun early may hinder the emergence of sensitive responding to the infant in some groups of families and that adequate time together is necessary for the parenting task of establishing early reciprocal responsiveness. For parenting during adolescence, the possibility exists that maternal employment may, for some families, result in too fast or too extensive movement toward autonomy and assumption of responsibility, either granted by the mother because of needs for adolescent participation in responsibilities in the family, or taken by the adolescent in the mother's absence and a source of friction. It is important to note that the findings on which this rough hypothesis is based are drawn from studies of very-low-income families (the experi- mental studies of welfare reform programs) or are particularly strong in the lowest income families studied in broader samples (the findings for infants). There may be particular challenges facing these families, for example, in terms of economic resources and the neighborhood context, that contribute to the patterns of parenting noted, perhaps posing other challenges that make it difficult to establish mutual responsiveness with infants, or con- fronting adolescents with dangerous environments when they push for greater autonomy. It is important to complement these findings with results from further studies of maternal employment and parenting that look at patterns in a wider range of economic groups. In these further studies, the low-income and working-class families are a more heterogeneous group than the more economically restricted samples of families studied in the welfare reform experiments.
86 WORKING FAMILIES AND GROWING KIDS Parenting Beyond the Infancy Period Focusing on results beyond the infancy period in studies of more socioeconomically diverse samples, two further patterns regarding mater- nal employment and parenting behavior can be seen: (1) studies find more positive patterns of parenting in families with employed than nonem- ployed mothers, with only slight indications of this patterning in middle- class families, but stronger indications of the pattern in groups of work- ing-class and lower income families. This patterning of results anticipates findings in the area of maternal employment and mothers' mental health (summarized below) and maps onto the pattern of child outcomes that Hoffman and other reviewers have summarized for low-income families (of outcomes for children of employed mothers falling in a neutral to positive range); and (2) there are patterns of association of parenting and the home environment among employed mothers according to the charac- teristics of employment. Regarding the first pattern, a longitudinal study of maternal employ- ment in middle-class families that followed a sample of children from in- fancy to adolescence (Gottfried et al., 1995) found evidence that maternal employment showed only a few limited associations with parenting behav- ior. In infancy, employed mothers engaged in more attempts at toilet training than nonemployed mothers. During the early elementary school years, employed mothers had higher educational aspirations for their chil- dren, their children were engaged in more out of school lessons, and they and their children watched less television. However, this study did not find evidence of differences by employment in the extent of nurturance or stimu- lation that children received. The researchers emphasize the active adapta- tions to the mothers' employment in the families in this sample, including greater involvement of fathers. A study encompassing both middle income and lower income families with 3rd and 4th grade children extends this set of findings, suggesting that associations of maternal employment and parenting practices may be stron- ger in lower than higher income families (Hoffman and Youngblade, 1999). Overall, employed mothers in this study were found to rely less on authori- tarian, power-assertive disciplinary styles. Less coercive discipline styles and more overt affection toward the child were linked with more positive social adjustment for children of employed mothers, particularly among working-class families in this study. Employed mothers' less authoritarian parenting was linked to higher scores for the 3rd and 4th grade children on tests of reading and math achievement and to teacher ratings of more effective learning patterns. Linkages to children's school outcomes were stronger in single-parent than two-parent families. Work by McLoyd and colleagues (1994) also found an association between maternal employment and disciplinary approach. This work, fo-
MATERNAL EMPLOYMENT AND THE FAMILY ENVIRONMENT 87 cusing on black families headed by single mothers, found more authoritar- ian parenting among unemployed than employed mothers of 7th and 8th graders. These findings are discussed in further detail below in relation to mothers' psychological well-being in relation to employment. As in the work of Hoffman and Youngblade, disciplinary approach was found to be related to children's outcomes, with the more power assertive discipline shown by unemployed mothers predicting less favorable adjustment in the young adolescents. While the studies noted above contrast families with employed and nonemployed mothers, other studies do not contrast employment catego- ries but look at variations in parenting in light of the nature or extent of employment. For example, focusing on a demographically diverse sample of families, Menaghan and Parcel (1995) provide evidence that maternal employment has differing implications according to the work circumstances of the mothers. Building on the work of Kohn and Schooler (1983), these researchers present the hypothesis that specific features of parents' jobs influence the behaviors that they value and encourage in their children (Menaghan and Parcel, 1995; Parcel and Menaghan 1990, 1994; Rogers et al., 1991). For example, they hypothesize that jobs that are repetitive and unstimulating and permit little self-direction will be associated with paren- tal childrearing values that emphasize obedience rather than initiative. Such jobs provide limited cognitive stimulation to the parent, which may, in turn, influence the extent of stimulation in parent-child interaction. In contrast, when parents have jobs that involve greater variety, stimulation, and self-direction, they may be more likely to use strategies of reasoning in discipline with their children and to expect self-direction and internaliza- tion of adult norms. Some of the existing research designs do a better job of handling selec- tion effects than others. One of the designs that is particularly helpful is one used in the study by Menaghan and Parcel (1995), looking at changes in employment status and related changes in the home environment. This design takes into account the initial characteristics of the mother and more fully isolates changes that occur with changes in employment status by the same mothers. (This approach found deterioration in the home environ- ment particularly when single mothers started jobs low in complexity and wages). Other studies have been able to take into account (control for) cognitive or literacy test scores administered to the mothers at the start of a welfare reform evaluation or survey data collection. For example, mothers were administered the AFQT in the NLSY; this is often used as a control variable in analyses examining employment in this dataset. Mothers were administered tests of literacy at the start of the Child Outcomes Study of the National Evaluation of Welfare-to-Work Strategies; this measure is used as a control variable in many analyses with these data.
88 WORKING FAMILIES AND GROWING KIDS In a further examination of work circumstances of employed mothers, Han (2002a) focused on the issue of nonstandard work hours (work that occurs during the evenings, nights, and weekends; see Chapter 2 for more information on trends around nonstandard work schedules). Using data from the NICHD Study of Early Child Care, Han found less positive cogni- tive and social developmental outcomes over time for children whose moth- ers had ever worked nonstandard hours by the child's third year of life in comparison with those who had not, controlling for extent of employment. This study also found differences in parenting and maternal psychological well-being in relation to work schedule. The quality of the home environ- ment was less optimal at 36 months when the mother had ever worked nonstandard hours, mothers experienced more depression at specific time points in the longitudinal study (though not others), and children had less exposure to center care when mothers had worked this schedule. The relationship between work schedule and children's developmental outcomes was attenuated, although it did not disappear, when the home environment and type of child care was taken into account. When he entered school, Nancy's son was often left home alone either before or after school. Nancy describes one particular week (Heymann, 2000:42): My boss made me work the six o'clock shift while Andrew was six or seven--maybe seven. I would leave him in the morning, and he got up that week and he was on his own. He was scared. And he got in trouble a couple of times that week . . . arguing with a teacher, fighting with a classmate. I shouldn't have did what my boss wanted. . . . They changed my hours without any notice. . . . It didn't work for my son because he couldn't handle being in the house alone at that age. Other research focusing on nonstandard work hours provides mixed findings. On one hand, there is evidence that for some occupations, work- ing nonstandard hours allows parents to spend more time with their chil- dren, providing increased supervision and involvement (Garey, 1999; Grosswald, 1999; Hattery, 2001). Further research finds working non- standard hours to be associated with higher proportion of employed par- ents being home when children are leaving and returning from school. Depending on the specific work schedule, gender of the parent, and activity considered, parent-child interaction may be greater when parents work nonstandard schedules (Presser, in press). On the other hand, Heymann (2000), in her research examining the circumstances of a range of working
MATERNAL EMPLOYMENT AND THE FAMILY ENVIRONMENT 89 families in the United States, found work during nonstandard hours and work by spouses during different shifts (strategies used by many families to juggle the needs of work and child rearing) had unfavorable implications for the well-being of the children. Further work is needed looking at the specific circumstances in which nonstandard work hours support or hinder supervision of children and children's development. Such work should consider marital status, whether one or both spouses are working non- standard hours, and the specific nonstandard schedule worked. Father Involvement While there has been much focus on maternal parenting behavior and the home environment, existing research extends the picture of parenting to father involvement as well. As was described in more detail in Chapter 2, father involvement in parent and family household tasks may change as a result of maternal employment. Hoffman (1989) observed that "probably the most clearly demonstrated effect of maternal employment is a modest increase in the participation of fathers in household tasks and child care" (p. 286). Some research has found father involvement to increase not only in keeping with the mothers' employment status, but also with their hours of employment (Gottfried et al., 1995). These findings generally pertain to father involvement in dual-earner families. We note that research to date focusing on father involvement in single-parent families in welfare reform programs has tended to examine the economic contributions of the father, through formalized or informal child support (and paternity establishment as a prerequisite to formalized child support), rather than involvement in the care of the child or the household (McLanahan and Carlson, 2002). Existing studies show sub- stantial increases in paternity establishment and child support payments in the years following welfare reform. McLanahan and Carlson note that there is a new generation of programs aimed at improving not only employ- ment and the economic contribution that low-income fathers have the po- tential to make, but also parenting skills and direct involvement of nonresi- dent fathers with their children. To date, however, there is little evidence on the efficacy of such programs or on the direct involvement of fathers in the care of their children when they do not reside with them, in light of the mothers' employment. Accordingly, we focus here on father involvement in the household in dual-earner families. In the research on dual-earner families, studies distinguish between contributions to housework and child care on one hand, and playful or educational joint activities or interaction with the child on the other. Two recent studies (Crouter et al., 1999; Hoffman et al., 1999) found mothers' work hours to be related to the division of labor in two-parent families in
90 WORKING FAMILIES AND GROWING KIDS terms of household tasks and physical care and supervision of children, but not in terms of educational or playful interactions with children. The NICHD Study of Early Child Care also found differing predictors of father involvement in caregiving and of father sensitivity in play with the child. Greater work hours by the mother and fewer work hours by the father predicted greater paternal involvement in caregiving, but work hours were not associated with paternal sensitivity during play (National Institute of Child Health and Human Development, 2000c). This distinction between caregiving and household tasks and the qual- ity and quantity of nonobligatory interactions with the child perhaps helps to explain the findings of observational studies of fathers and their very young children, which have not reported increased interactions of fathers with their children in dual-earner families when family interaction is sampled for discrete periods of time (e.g., Stuckey et al., 1982), or a differ- ence in the quality of observed parenting behavior by fathers in light of the mother's employment status (e.g., Grych and Clark, 1999). Findings from the observational study of father-infant interaction by Grych and Clark (1999) suggest that while maternal employment during the first year of a child's life does not seem to affect the quality of father- infant interaction directly, it may do so indirectly by influencing the context in which fathers and their infants interact. For fathers in this study whose wives were not employed, increased involvement in caregiving both early and late in the infant's first year (at 4 and 12 months) was accompanied by greater expression of positive affect during interactions with the infant. A similar pattern of greater positive affect occurring with greater paternal involvement in caregiving was also found for fathers of wives employed part time, although only at the later point during the infant's first year. In contrast, for husbands whose wives were employed full time, greater in- volvement in caregiving was accompanied by more negative interactions with the infant at the earlier time point. The findings suggest the hypothesis that when the mother is not em- ployed, greater father involvement in caregiving is at the volition of the father and is pleasurable, while this is not the case, at least early on, for fathers of wives employed full time (who may feel that the caregiving is obligatory and not pleasurable). This hypothesis is consistent with results reported by Vandell and colleagues (Vandell et al., 1997) that fathers whose wives were employed reported more anger when they were more involved in the caregiving of their 4-month-old infants. Grych and Clark (1999) caution that the sense of being pressed into greater responsibility for child care early on by fathers of wives employed full time does not appear to be sustained through the end of the first year in their sample, "suggesting that they may have become more proficient at balancing work and family re- sponsibilities" (p. 900).
MATERNAL EMPLOYMENT AND THE FAMILY ENVIRONMENT 91 Turning to families with older children, Hoffman and colleagues (1999) provide the most detailed examination of whether and how increased father involvement in dual-earner families is associated with child outcomes. In a study of stable maternal employment (consistent employment status over a three-year period) in low- and middle-income families with 3rd and 4th grade children, they found that the greater the father's involvement in household tasks and child care, the less stereotyped were the children's attitudes about appropriate roles for men and women. Children's less stereotyped gender roles, specifically their perception of women's compe- tence in traditionally male domains, predicted achievement test scores in the 3rd and 4th grade for both boys and girls. These researchers also tested a model regarding daughters' scores on tests of academic achievement. They found that maternal employment was associated with greater partici- pation in household and child care tasks by fathers, which in turn predicted daughters' less stereotyped attitudes about women's competence. This in turn predicted a greater sense of efficacy and higher scores on tests of reading and math achievement. Thus, greater father involvement in dual- earner families may help to explain the findings for girls of greater aspira- tions and achievement. The researchers note the key limitation of their work is that the examination of interrelationships involved concurrent rather than longitudinal data. We have noted the paucity of work laying out and testing such models. While the research of Hoffman and colleagues is clearly a step forward, there is a need for more work of this kind, using longitudinal data and examining patterns across key subgroups (for example, considering whether models are similar or different according to gender, race/ethnicity, and socioeconomic status). Maternal Psychological Well-Being Mothers who are employed have been found to show better psycho- logical well-being on measures of depression, stress, psychosomatic symp- toms, and life satisfaction (Kessler and McRae, 1982; McLoyd et al., 1994; Repetti et al., 1989). Mothers' psychological well-being, in turn, has been shown to be important to children's development, influencing development through the quality of mother-child interactions (Downey and Coyne, 1990; Goodman and Brumley, 1990; Harnish et al., 1995, Hair et al., 2002; National Institute of Child Health and Human Development, 1999c). Some work suggests that this pattern holds only or more strongly for lower income than middle-class families (Warr and Parry, 1982). In recent research, Hoffman and Youngblade (1999) found evidence of better mater- nal psychological well-being for employed than nonemployed working- class mothers of school-age children, using measures of depressive symp-
92 WORKING FAMILIES AND GROWING KIDS toms and morale. However, no parallel difference by employment status was found for middle-class mothers. Maternal psychological well-being has also been shown to be related to job characteristics. Fuller and colleagues (2001) found maternal depressive symptoms to be lower among low-income mothers working in higher qual- ity jobs (as indexed by the provision of health benefits). Han (2002b) found that mothers who worked nonstandard hours (evenings, nights, or rotating shifts) had higher scores on a measure of depression by the time their child was 15 months old than mothers working standard hours. McLoyd and colleagues (1994) found maternal unemployment to be associated with greater depression among black mothers of adolescent chil- dren in single-parent families. Employed mothers perceived less financial strain and greater instrumental social support. Their 7th and 8th grade children perceived their relationships with their mothers to be more posi- tive, perceived less economic hardship, and had lower anxiety levels. The increased depressive symptomatology among unemployed mothers was as- sociated with increased use of harsh punishment with their adolescent chil- dren, which in turn predicted greater difficulty concentrating and more depression among the adolescents. A similar set of linkages is reported for 3rd and 4th grade children in the work of Hoffman and Youngblade (1999). For example, among work- ing-class mothers, employment was predictive of fewer depressive symp- toms, which in turn were found to be associated with more authoritative (firm but warm), rather than power-assertive, parenting. Such parenting in turn was predictive of children's higher achievement test scores in reading and math, fewer learning problems as rated by teachers, and more positive social skills on teacher ratings of peer social skills and acting out behavior. Maternal depressive symptoms partially mediated the relationship between maternal employment and parenting style in these analyses. We note the important caution that causal direction is not entirely clear in this work and in other work showing an association between maternal employment and maternal mental health. It is indeed possible that mothers may derive a sense of competence from their work, that contact with co- workers serves as a source of social support, and that the income derived from employment may reduce anxiety about family economic resources. However, it is also possible that mothers with poor mental health may find it more difficult to find or maintain employment, and that this is the source of the employment-mental health link (see findings in Vandell and Ramanan, 1992). Indeed, in recent research with families with a history of welfare receipt, mother's depressive symptoms were found to predict subse- quent employment (Hair et al., 2002). If employment is the cause of improved maternal psychological well- being, then one might expect that mothers would show improved well-
MATERNAL EMPLOYMENT AND THE FAMILY ENVIRONMENT 93 being in the welfare-to-work evaluations in which mothers showed an in- crease in employment. In an overview of the findings on maternal psycho- logical well-being in these evaluations, Ahluwalia et al. (2001) examined findings regarding impacts on depressive symptoms in 20 analytic groups in 7 programs (with separate analytic groups in some of the evaluations for families with children of differing ages, for variations in the programs, and for subgroups of families such as recent or long-term welfare recipients). Impacts on maternal depressive symptoms were found in only seven of these groups. In most instances (for five of the programs in which statisti- cally significant impacts were found), these impacts were unfavorable rather than favorable. Interestingly, in most of the programs in which the unfavorable impacts (increases in depressive symptoms) occurred, overall family income did not increase despite the family's participation in a welfare reform program. In some of these programs, employment increased while income did not in- crease; in others, neither employment nor income increased. The possibility exists that in some programs and for some families, maternal psychological well-being may decline when participation in a welfare reform program intended to increase employment does not result in employment or an improvement in the family's economic situation. It is also possible that employment in the context of a welfare reform program differs substantially from employment in other circumstances. Most of the programs evaluated involved mandatory participation in em- ployment-related activities (with the possibility of sanctions for noncompli- ance). The link between mothers' mental health and employment may well exist and follow a causal sequence in which employment results in better maternal well-being, but only when the mother can choose the timing and nature of the employment. Employment in the context of a mandate may not show the hypothesized benefits. Yet the results from the welfare reform evaluations suffice to caution that the causal direction of the maternal employment-psychological well-being link needs closer examination. Huston (2002) suggests that such an examination encompass the possibility of a recursive relationship, with maternal psychological well-being perhaps helping to determine employment outcomes, which in turn may contribute to mothers' psychological well-being. ADAPTATION TO MATERNAL EMPLOYMENT Bianchi (2000b) observes that despite the increases in rates of maternal employment in recent decades, time use studies show substantial consis- tency in maternal time with children. Greater demands of household tasks in earlier decades limited the time that mothers at home actually spent in interactions with their children. Also, families had more children in the
94 WORKING FAMILIES AND GROWING KIDS past, so time per child has changed less. In addition, the evidence indicates that families actively adapt to employment in ways that maximize parental time with children, for example, through mothers choosing part-time em- ployment, a reallocation of mothers' time away from leisure activities and toward time with children, and greater father involvement (as noted above) when the mother is employed. Recent research continues to provide a picture of active adaptation to maternal employment in which families seem to protect parental time with children. At the same time, new work poses the possibility that there may be constraints in some families in making such adaptations. Aronson and Huston (2001) examined time use data for mothers with infants in the NICHD Study of Early Child Care. They found that em- ployed mothers did spend less time overall with their infants. However, employed mothers were more likely to make reductions in other activities than infant care and to compensate for time away from their infants through time use on weekends. The tendency to maximize time with children when the mother is em- ployed emerges especially in recent findings on two-parent low-income families. National survey data indicate that there have been increases in these families in the percentage of young children cared for only by parents when the parent most involved in the care of the child (almost always the mother) is employed. Data from the National Survey of America's Families (NSAF) indicate that from 1997 to 1999 the percentage of children under age 5 in such families cared for only by parents increased from 28 to 33 percent. This increase is not found in single-parent low-income families or in higher income families (Zaslow and Tout, 2002). Data on parental activities with young children suggest that active adaptation may be more difficult in single-parent than two-parent low- income families. Phillips (2002), also using NSAF data, found that full- time employment in single-parent low-income families was associated with a reduction in parent involvement in reading and outings with preschool children. In two-parent families, however, high levels of parental work were not found to be associated with diminished parent involvement in activities with preschoolers. The evidence suggests that maternal employment is not a circumstance to which families respond passively, but rather one that they actively seek to shape (Gottfried et al., 1995). A hypothesis that seems to be emerging in the research is that there may be some groups of families, such as low- income single-parent families, who are more constrained in this adaptation process.
MATERNAL EMPLOYMENT AND THE FAMILY ENVIRONMENT 95 SIMULTANEOUS INFLUENCES ON MULTIPLE ASPECTS OF FAMILY LIFE We have considered the implications of maternal employment for as- pects of family life separately, without considering how influences on dif- ferent aspects of family life might operate jointly to affect children's devel- opment. Yet it may be important to take into account how multiple influences of employment on family life operate simultaneously. McGroder and colleagues (2000) examined the mediators of the im- pacts on young children of mothers' assignment to JOBS welfare-to-work programs. They found that these programs tended to affect families in multiple ways, and not always in the same direction. Impacts on children reflected the net effect of these influences. For example, one of the six programs examined in this evaluation, the JOBS labor force attachment program in the Atlanta site of the study, had a favorable impact on children in reducing their externalizing behavior problems. This program had a positive impact on a summary rating of mothers' parenting behavior. However, the program also increased moth- ers' feelings of time stress and perceptions that the welfare office pushed parents to go to school or get training. While positive parenting predicted fewer externalizing behavior problems, time stress and feelings of being pushed by the welfare office predicted more such problems. The favorable impact of the program on children's externalizing behavior reflected the balance of these influences: the impact on children's externalizing behavior was mediated by favorable parenting, but would have been even more favorable without the counterbalancing influence of mothers' subjective sense of time stress and pressure. McGroder and colleagues note that while these analyses controlled for a range of family characteristics prior to ran- dom assignment in this experimental evaluation, the possibility nevertheless exists that further (unobserved) factors were contributing to the patterns noted; thus the findings should be viewed as exploratory. Our understanding of how maternal employment influences family life and children's development would be deepened by further research looking at multiple aspects of family life simultaneously, taking into account the possibility that these may have counterbalancing influences. SUMMARY On the basis of evidence presented in this chapter, we conclude that the effects of maternal employment depend on a range of factors and may vary by subgroup. Very young children may be particularly affected by maternal employment. For newborns, outcomes for mothers and children are better when mothers are able to take more than 12 weeks of leave, and outcomes
96 WORKING FAMILIES AND GROWING KIDS for children may be better when mothers are able to return to work part time or to delay returning to work full time until after the first year. A family's income also appears to affect material well-being, which in turn affects children and adolescents. The research on maternal employ- ment and the family environment for children has recently been extended by studies of families enrolled in a range of programs to support the transi- tion from welfare to employment. Findings regarding the impact of the family environment of children from studies of low-income families differ in a number of ways from studies of maternal employment in more hetero- geneous samples of families. For example, while findings in broader samples suggest that maternal employment is associated with better mental health for mothers, this pattern is not found with any consistency among mothers participating in welfare-to-work programs. Findings in broader samples indicate that employed mothers tend to use less power-assertive discipline, and that this has favorable implications for children's development. In the wel- fare-to-work evaluations, there is limited evidence of effects on dyadic aspects of parenting (like expression of warmth or disciplinary practices), although when these occur they appear to play a role in shaping program impacts on children. Instead, impacts on parenting in the welfare-to- work evaluations are concentrated in the gatekeeping aspects of parenting, such as enrollment of children in child care and after-school activities. There are indications in the welfare-to-work evaluations of the particular importance of economic resources associated with employment in shaping positive impacts for young children of mothers making the transition to work (although these same factors do not seem to contribute to positive impacts for adolescent children in these families, who show a pattern of unfavorable impacts irrespective of whether increased employment was associated with increased family income). The role of economic resources has been hypothesized as important in explaining the implications for children in a broader range of families, but little work has been carried out focusing explicitly on this issue in more heterogeneous samples. While there are some indications that relationships and roles in families with adolescents are affected negatively during the transition from welfare to work, there is no parallel pattern for adolescents in broader samples of low-income families, and indeed there are indications that employment is related to more positive patterns of mother-adolescent relations. Studies of maternal employment vary substantially in how well they have ad- dressed selection effects. One possible interpretation is that the differences in findings primarily reflect methodological differences across the studies of welfare and nonwelfare families. The former have been studied in experimental evaluations of wel-
MATERNAL EMPLOYMENT AND THE FAMILY ENVIRONMENT 97 fare-to-work programs, while the more heterogeneous samples have been studied in descriptive research that looks at associations of employment with family life and child outcomes either concurrently or over time. Perhaps the experimental evaluations have more fully isolated the effects of employment from selection effects, and a truer picture emerges of the implications of employment in these studies. A number of further differences across the sets of studies need to be kept in mind as possibly contributing to the differences in findings noted. The families studied in the welfare evaluation studies are more disad- vantaged than the low-income families in broader samples. In the latter, "low income" may be defined as including families up to 200 percent of the poverty line (as in analyses in light of income in the National Survey of America's Families, Phillips, 2002). Families in the welfare reform evaluations, in nearly all of the studies, did show increases in employ- ment on average, but the evaluations reflect the impacts of assignment to a welfare-to-work program rather than the impacts of employment per se. Families making the transition to work in the welfare reform context were experiencing mandates to work or incentives to work that affected the speed with which they needed to find employment and the benefits from employment. In broader samples, while there are clearly constraints operating, mothers are somewhat freer to choose the timing of employment and the nature of the job. They may, for example, take into account to a greater extent their own satisfaction with a child care arrangement, the availability of other adults to help, job characteristics, and issues concerning their children's well-being, such as health. In future work, it would be particularly helpful to look systematically in heterogeneous samples at whether maternal employment is associated with different family processes and child outcomes in light of history of welfare receipt and socioeconomic circumstances. In addition, while all of the studies included in this review took background characteristics of the families into account, future work would be particularly informative if it grappled more fully with selection effects. Even given these needs for further work, the set of studies reviewed here does provide some guidance as to where further supports for low- income working families might be targeted. Those instances in which unfa- vorable associations of maternal employment and family life occurred can help to identify contexts in which supports might be helpful. In the work reviewed, these include: a very early and extensive resumption of employ- ment after the birth of a child for some groups of families (although, as noted above, there is a need to understand why this pattern is occurring for some subgroups of families but not others), and employment (especially by single mothers) in jobs that involve low complexity, lack benefits, or in-
98 WORKING FAMILIES AND GROWING KIDS volve working nonstandard hours. For families making the transition from welfare to work, the research suggests that supports might be helpful in connecting families with the full set of benefits (such as child care subsidies) for which they are eligible and targeting families struggling to make the transition to work. The research also suggests that a particular focus be given to the needs of adolescent children in these families.
