Childrearing Stages and Work-Family Conflict: The Role of Job Demands and Resources

Abstract

Objective

We examine the role of job characteristics in influencing variation in mothers’ work-family conflict by childrearing stage.

Background

Although researchers generally contend that having younger children is related to greater work-family conflict, examination of this association is limited.

Method

Using data from the NICHD Study of Early Child Care and Youth Development (N = 774), we conduct fixed effects models to examine variations in mothers’ job characteristics across four waves when their children are 6 months old, 15 months old, third graders, and fifth graders and their links to variations in mothers’ work-to-family conflict and family-to-work conflict across the same waves.

Results

Mothers work fewer hours, but perceive more job pressure, fewer career opportunities, and less supervisor support when children are younger. Because of the countervailing patterns of variations by childrearing stage between work hours and job pressure, there is little difference in mothers’ work-to-family conflict across the four waves. Mothers report more family-to-work conflict when children are younger, but this difference by childrearing stage disappears when perceived job pressure is controlled for.

Conclusion

Life course dynamics of job demands and resources may shape variation in work-family conflict by childrearing stage.

Implications

Workplace policies designed to reduce job pressure while not eliminating career opportunities would be helpful for mothers with young children to adjust to the stressfulness of balancing work and childrearing demands.

Since the late 1980s, a majority of US mothers return to work during the first year of their children’s lives (US Bureau of Labor Statistics, 2009). Over this period, US parents who experience work-family conflict, defined as individuals’ perceptions that their work and family responsibilities interfere with each other, have increased (Nomaguchi, 2009). Work-family conflict involves two directions: work can interfere with family (work-to-family conflict, WFC) and family can interfere with work (family-to-work conflict, FWC) (Bellavia & Frone, 2005). Both WFC and FWC have serious consequences for the well-being of workers and their families (Bellavia & Frone, 2005; Grzywacz & Bass, 2003). Various policies, such as paid parental leave and flexible work scheduling, have been suggested as workplace policies that may help parents with young children to reduce WFC and FWC (Kelly et al., 2014; Landivar, 2017). To better inform policy makers about types of policies that may effectively help parents in specific childrearing stages, it is important to understand how parents’ WFC and FWC vary by childrearing stage.

Although researchers tend to argue that having children under age 6 is related to greater WFC and FWC (e.g., Voydanoff, 2004), empirical findings on differences in WFC and FWC by childrearing stage have been inconsistent (see a review in Schieman, Milkie, & Glavin, 2009). We suspect that this inconsistency is in part because of methodological limitations. First, many studies have used a comparison group that was not ideal. Studies compared employed adults with children under age 6 to all other employed adults (Grzywacz, Almeida, & McDonald, 2002; Voydanoff, 2004), employed adults without any children in the household (Hill et al., 2008; Mennino, Rubin, & Brayfield, 2005), or employed empty-nest parents (Moen & Yu, 2000). In these comparisons, the effect of childrearing stage was mixed with the effect of parental status. Second, almost all studies used cross-sectional data. Relying on cross-sectional data is problematic because mothers who are employed when their children are infants are different from mothers who are employed only when their children started school in their sociodemographic characteristics and family circumstances (Damaske & Frech, 2016; Hynes & Clarkberg, 2005). We need research that compares the same mothers’ WFC and FWC across different childrearing stages, while eliminating the effects of mothers’ background characteristics that are related to both the odds of mothers being employed in a certain childrearing stage and mothers’ WFC or FWC.

Another limitation of prior research is its insufficient attention to variation in job characteristics by childrearing stage as potential sources of variation in WFC and FWC by childrearing stage. Researchers typically argue that having younger children is related to higher WFC and FWC because of higher demands of childcare when children are younger than older (Voydanoff, 2004). Yet, little research has examined this supposition, because the types of child demands vary by childrearing stage and, thus, most surveys do not have comparable measures of child demands across different childrearing stages (Nomaguchi, 2012; Nomaguchi & Milkie, 2017). In contrast, despite the availability of comparable measures of job characteristics across childrearing stages, very few studies, let alone those using longitudinal data, have examined how job characteristics vary by childrearing stage (Hill et al., 2008; Martinengo, Jacob, & Hill, 2010). This is a surprising omission because many studies have emphasized that job characteristics play a critical role in shaping WFC and FWC (Schieman, Milkie, & Glavin, 2009; Voydanoff, 2005).

Using longitudinal data from the National Institute of Child Health and Human Development Study of Early Child Care and Youth Development (SECCYD), we compared mothers’ WFC and FWC across four time points when the focal child was 6 months old, 15 months old, in third grade, and in fifth grade—the waves when the SECCYD collected detailed information about mothers’ job characteristics, WFC, and FWC. We focused on mothers because fathers’ information was limited in the SECCYD. We drew on the Job Demands-Resources (JD-R) model (Bakker & Demerouti, 2007) and a life course perspective on the work-family intersection (Moen & Sweet, 2004) to develop hypotheses on variations in WFC and FWC across these four waves and how such variations are explained by variations in job demands and resources across the four waves. We used fixed effects models that account for unmeasured mothers’ background characteristics which may confound the observed variation in WFC and FWC by childrearing stage (Allison, 2009).

Job Demands and Resources and Their Links to WFC

The JD-R model and other similar approaches have provided insights into understanding the role of job characteristics in shaping WFC (Kelly et al., 2014; Schieman, Milkie, & Glavin, 2009). The basic idea is that work conditions can be categorized into two characteristics: job demands and resources, and that levels and interactions of job demands and resources shape WFC (Bakker & Demerouti, 2007). In this study, we focus on how levels of specific aspects of job demands and resources play a role in shaping WFC (not examining interactions between job demands and resources). Following prior research (e.g., Schieman & Glavin, 2011; Voydanoff, 2004), we organize our discussions from the job demands and job resources perspectives separately:

The job demands perspective highlights that individuals are more likely to experience WFC when job demands are high (Schieman, Whitestone, & Van Gundy 2006). Job demands refer to the physical, socioemotional, and organizational aspects of a job that require sustained physical or mental effort (Bakker & Demerouti, 2007). Researchers have identified broadly two types of job demands that shape WFC. The first type is time- or task-based demands, which we call objective demands, such as long work hours and high status occupations, which may reflect the actual amounts of workload and responsibilities (Schieman, Milkie, & Glavin, 2009; Voydanoff, 2004). The second type is strain-based or subjective measures of demands, such as job pressure defined as perceptions that demands of one’s job for time, energy, and stamina exceed one’s capacities (Schieman, Whitestone, & Van Gundy 2006; Voydanoff, 2004; 2005).

The job resources perspective emphasizes that individuals with fewer job resources experience greater WFC (Kelly et al., 2014). Job resources refer to the physical, socioemotional, and organizational aspects of a job that are functional in achieving work goals or stimulate personal growth and development (Bakker & Demerouti, 2007). As in the case of job demands, researchers have identified objective (e.g., earnings, availability of flexible scheduling policy) and subjective (e.g., perceptions of supervisor support, task significance, or how flexible one’s job is) types of job resources. Job resources are related to lower WFC in part through providing workers with motivation, such as task significance (e.g., realizing personal growth through one’s work), decision-making latitude, and greater autonomy (Mennino et al., 2005; Schieman, Milkie, & Glavin, 2009; Voydanoff, 2004). Job resources also help workers reduce WFC by providing them with tools that facilitate their ability to complete their tasks and achieve their career goals (e.g., career opportunities, supportive supervisor) or to negotiate conflicts between work and family (e.g., scheduling flexibility) (Kelly et al., 2014; Mennino et al., 2005). Material resources, such as higher earnings, also help reduce WFC (Schieman, Milkie, & Glavin, 2009).

The association between job resources and WFC is not straightforward in part because jobs that provide workers with more resources typically impose them with greater demands as well. For example, workers in higher status occupations are more likely than those in lower status occupations to have higher earnings, career opportunities, workplace support, and scheduling control, but also likely to work long hours (Schieman, Milkie, & Glavin, 2009). Thus it is imperative to control for work hours and occupational status—objective job demands—in order to eliminate the confounding effects of job demands that would obscure the association between job resources and WFC. Prior research has found that, with work hours and occupational status held constant, earnings, career opportunities, supportive supervisor, scheduling flexibility, and task significance are related to lower WFC (Kelly et al., 2014; Schieman, Milkie, & Glavin, 2009; Voydanoff, 2004).

Job Demands and Resources and Their Links to FWC

Turning to FWC, the other direction of work-family conflict, little research has explicitly used the job demands and job resources perspectives to identify determinants of FWC. This may be because earlier research tended to focus on family-related characteristics as predictors of FWC (e.g., Frone et al., 1992). Yet, a number of empirical studies have shown that job characteristics cannot be ignored in order to fully understand factors related to FWC. For example, research has shown that job pressure, a subjective measure of job demands, is related to greater FWC (Hill et al., 2008; Mennino et al., 2005; Nomaguchi 2012; Voydanoff, 2005). This makes sense because a mother may not feel that her childcare responsibilities affect her job performance unless she feels that her job demands go beyond her capacity. In regard to work hours, empirical findings have been mixed. Some studies found a positive association between work hours and FWC (Dilworth, 2004), whereas other studies found no relationship (Mennino et al., 2005; Nomaguchi, 2012), or a negative association between work hours and FWC (Byron, 2005; Hill et al., 2008; Voydanoff, 2005). We suspect that the inconsistent findings may be because of differences in control variables across the studies. It appears that work hours were positively related to FWC when job pressure was not controlled for, whereas work hours had no association or were negatively related to FWC when job pressure was controlled for. With job pressure held constant, longer work hours may mean that greater time availability—mothers have more time to deal with a similar amount of work.

Job resources, too, can shape FWC. If a mother felt that her supervisor was not supportive or that her expected promotion was foregone, she might try to increase the effort she puts into her work, which may in turn increase her perception that her family responsibilities interfere with her work (Voydanoff, 2005). Qualitative studies have suggested that feelings of receiving less supervisor support and fewer career opportunities than they believe they deserve may fuel mothers’ doubts about having returned to work, which, in turn, may increase FWC (Blair-Loy, 2003; Stone, 2007; Williams, Blair-Loy, & Berdahl, 2013). Quantitative studies have shown that earnings, supervisor support, task significance, and scheduling flexibility are related to lower FWC (Hill et al., 2008; Mennino et al., 2005; Nomaguchi 2012; Voydanoff, 2005). As in the case of WFC, these associations should be examined controlling for work hours and occupational status.

Variations in Job Demands and Resources by Childrearing Stage and Their Links to WFC and FWC

A life course perspective on work-family conflict suggests that the amounts of job demands and resources typically change across life stages (Moen & Sweet, 2004). Because job demands and resources are closely related to WFC and FWC as discussed above, examining how they are distributed across different childrearing stages can increase understanding of the processes shaping change in WFC and FWC over the course of childrearing years.

We focus on two broad childrearing stages—when children are very young (6 and 15 months old) and when children are late elementary school ages (third and fifth grades). Moen and colleagues have noted that childrearing stages are closely linked with career stages (Moen, Waismel-Manor, & Sweet, 2003). When their children are very young, mothers are likely to be in their early career stage that typically involves high demands (e.g., learning new rules, routines, and skills, building support networks, and achieving job security) and low resources (e.g., earnings). When children are late elementary school-age, mothers are likely to be in their mid-career stage where job demands may remain high, but resources may be greater (e.g., earnings, work experience, support networks) than in their early career stage. Below, drawing on this model, we discuss how specific job demands and resources may vary between these two childrearing stages. Then, on the basis of prior research on how these specific job demands and resources are related to WFC and FWC, we state predictions as to how WFC and FWC may vary between these two childrearing stages.

Job demands

We focus on two objective and one subjective measures of job demands. First, we consider work hours. Research has shown that mothers are more likely to work fewer hours when their children are younger because mothers continue to shoulder more child care responsibilities than fathers especially before their children enter school (Bianchi, 2011; Milkie, Raley, & Bianchi, 2009). This means that mothers have fewer job demands, at least in terms of time demands, when their children are younger than when their children are older. Second, we examine higher status occupations, meaning managerial and professional jobs, compared to other jobs. Prior research has indicated two contrasting predictions as to variation in the percentage of mothers holding managerial or professional jobs by childrearing stage. On the one hand, mothers with higher status occupations are more likely to go back to work soon after childbirth in part because they have better resources that allow them to do so, such as higher earnings or flexible scheduling policies and in part because their opportunity cost of being out of the labor force is higher (Damaske & Frech, 2016; Landivar, 2017). On the other hand, mothers may be more likely to hold a managerial position in their mid-career than early-career stages (Martinengo, Jacob, & Hill, 2010). With these countervailing effects, there may be little difference in the percentage of mothers holding higher status occupations by childrearing stage.

Finally, our third measure of job demands is job pressure, a subjective measure of job demands. In contrast to objective measures of demands, mothers may experience greater job pressure when children are younger. Past research has shown that early-career workers are more likely than mid-career workers to struggle with challenges of demanding tasks, employment insecurity, and uncertainty regarding their occupational choices (Hargreaves, 2005). Early-career workers tend to be energetic and may place unrealistic expectations on their daily achievements which may not be met (Klassen & Chiu 2010). These unrealistic expectations can reflect in greater job pressure. Because mothers are more likely to be in the early-career stage when their children are younger, mothers may be more likely to report greater job pressure when their children are younger than when their children are older.

In sum, when looking at variation in job demands by childrearing stage, we expected that mothers would work fewer hours, but would perceive greater job pressure when their children were younger (H1a). Because prior research suggested that both work hours and job pressure were strong predictors of WFC (Schieman, Milkie, & Glavin, 2009; Voydanoff, 2004; 2005) and these factors had countervailing associations with childrearing stage, there would be little variation in WFC across childrearing stages at the descriptive level; and this would be explained by more work hours but less job pressure that mothers experience as children get older (H2a). For FWC, as discussed in the previous section, job pressure is a strong predictor, but work hours were not (e.g., Mennino et al., 2005). Thus, we expected that FWC would be greater when children were younger than older, and this would be explained by greater job pressure that mothers experienced when their children were younger (H3a).

Job resources

We examine one objective and four subjective measures of job resources. Mothers’ earnings, an objective measure of resources, may be lower when their children are infants than when their children are school-age because wages typically increase with job experience (Budig & Hodges, 2010). Mother may also perceive less supervisor support and fewer career opportunities when their children are younger. US workplace culture continues to regard the ideal worker archetype as someone who devotes herself to her job without distractions by family responsibilities (Kelly et al., 2010; Moen & Sweet, 2004). In their laboratory experiments, Correll et al. (2007) found that mothers were perceived less competent in the workplace than non-mothers. Qualitative studies have documented that mothers with young children reported being assigned less challenging tasks, passed over for standard promotions, and faced demotions (Blair-Loy, 2003; Moen, Lam, Ammons, & Kelly, 2013; Stone, 2007).

In regard to two other types of subjective job resources, perceived scheduling flexibility and task significance, levels of these job resources may not vary by childrearing stage. Scheduling flexibility is critical for mothers to go to work when their children are school-age as well as when their children are younger. Childcare responsibilities could spill over into a regular work schedule at a moment’s notice—children get sick; childcare plans fall apart; school events, medical check-ups, and extracurricular activities are often scheduled during the regular work hours (Blair-Loy, 2003; Damaske, 2011). Mothers who do not have access to scheduling flexibility are likely to end up quitting their jobs when child demands are heavy (Budig & Hodges, 2010; Damaske, 2011; Landivar, 2017), which means that they are not in the sample of the present analysis. Task significance may also vary little by childrearing stage. Given the high cost of childcare and the persistent cultural belief in the importance of maternal care for a child’s proper development, mothers do not return to work unless they perceive their work as meaningful enough for them to be away from their children (Damaske, 2011).

In sum, we expected that mothers would perceive less supervisor support and fewer career opportunities when their children were younger, whereas there would be little difference in perceived scheduling flexibility and task significance by childrearing stage (H1b). Because perceived less supervisor support and fewer career opportunities are negatively related to WFC and FWC (Mennino et al., 2005; Voydanoff, 2005), we expected that mothers would experience greater WFC and FWC when children were younger than older, and these variations would be explained by perceived less supervisor support and fewer career opportunities that mothers felt when their children were younger (H2b for WFC; H3b for FWC). Prior research has shown that because jobs that provide more workplace support and career opportunities tend to be higher status occupations that require longer work hours, work hours and occupational status should be held constant when examining the association between job resources and WFC as well as FWC (Schieman, Milkie, & Glavin, 2009).

Summary of Hypotheses

In this paper, we examined variations in mothers’ job demands and resources by childrearing stage and whether such variations have implications for variations in mothers’ WFC and FWC by childrearing stage. We stated the following hypotheses:

Variations in Job Demands and Job Resources by Childrearing Stage

• H1aMothers will have fewer objective job demands, such as fewer weekly work hours, but greater subjective demands, such as greater job pressure, when their children are younger.
• H1bMothers will have lower earnings and perceive less supervisor support and fewer career opportunities when their children are younger, but there will be no differences in mothers’ perceptions of task significance and scheduling flexibility by childrearing stage.

Variation in WFC by Childrearing Stage

• H2aBecause mothers work fewer hours, but experience more job pressure when their children are younger, there will be no net differences in WFC across childrearing stage. If it were not for their fewer work hours when children are younger, mothers would report more WFC when their children are younger than when their children are older. If it were not for their higher job pressure when children are younger, mothers would report less WFC when their children are younger.
• H2bMothers will report more WFC when their children are younger than when their children are older, because mothers have lower earnings, perceived less supervisor support, and fewer career opportunities when their children are younger. If it were not for their lower earnings, perceptions of less supervisor support, and perceptions of fewer career opportunities when children are younger, there will be little difference in WFC across childrearing stage.

Variation in FWC by Childrearing Stage

• H3aBecause mothers experience more job pressure when their children are younger, mothers will report more FWC when their children are younger. If it were not for their higher job pressure when children are younger, there will be little difference in FWC across childrearing stage.
• H3bMothers will report more FWC when their children are younger than when their children are older, because mothers have lower earnings and perceive less supervisor support and fewer career opportunities when their children are younger. If it were not for their lower earnings, perceptions of less supervisor support, and perceptions of fewer career opportunities when children are younger, there will be little difference in WFC across childrearing stage.

All multivariable analyses were controlled for time-varying family demands and resources that prior research has suggested are related to the odds of a mother being employed in a certain childrearing stages as well as WFC or FWC. These include: family income, the number of children, and relationship status (Grzywacz et al, 2002; Hill et al, 2008; Martinengo, Jacob, & Hill, 2010; Mennino et al., 2005; Moen & Yu, 2000; Nomaguchi, 2009, 2012; Voydanoff, 2004, 2005).

METHOD

Data

The SECCYD is a longitudinal study of 1,364 children and their families, which was originally designed to study the association between children’s early childcare experiences and developmental outcomes. This study began in 1991 when families of newborns were recruited from hospitals in 10 cities in 9 states in the United States (NICHD ECCRN, 2005). The SECCYD collected information about the focal child’s primary parent’s experiences of WFC and FWC when the child was 6, 15, 36 months old and then in third and fifth grades. Because questions regarding job characteristics were not asked in the 36-month survey, we focused on the four waves including 6 months (T1), 15 months (T2), third grade (T3), and fifth grade (T4). Supplemental analyses (not shown) suggested that the prevalence of WFC and FWC in 36 months was very similar to that in 15 months, respectively. The SECCYD has unique advantages that serve the purpose of the present study well. Mothers answered a wide range of questions on job characteristics, WFC, and FWC across different childrearing stages. Very few studies with a sample of mothers with diverse background characteristics have collected such detailed information using the same measures across different childrearing stages.

Of the 1,364 children and their mothers who initially participated in the study, 1,243 mothers in 15 months, 1,053 mothers in third grade, and 970 mothers in fifth grade were re-interviewed. We selected the cases in which mothers participated in all of the four waves of interviews (n = 913). Then we restricted the sample to mothers who worked for pay at least 1 hour per week in two or more waves of the four waves (N = 774). Basic demographic characterisitcs of mothers in the analytical sample are presented in Table 1. The average age of mothers at birth of the focal child was 29 years old with a range from 18 to 46 years. Eighty-three percent of mothers were non-Hispanic White; 42 % had a bachelor’s degree or higher; and 83 % were married at birth of the child.

Table 1.

Means (SD) for Mothers’ Background Characteristics (N = 744)

Mother’s age at childbirth	29.00 (5.34)
Mother’s race/ethnicity
White	0.83
Black	0.10
Hispanic or other	0.08
Mother’s education
Less than high school	0.05
High school	0.20
Some college	0.32
College	0.24
Advanced degree	0.18
Mother’s marital status
Married	0.83
Cohabiting	0.06
Single	0.11
Child’s gender (1 = girls)	0.49
Child’s birth order (1 = first)	0.45

Mothers in the present analysis were more economically advantaged than their counterparts in the general population for three reasons. First, mothers in the SECCYD were more economically advantaged than mothers in the general population (NICHD ECCRN, 2005). Second, mothers with higher levels of education, Whites, and the married were more likely to be retained in the later waves. Third, as mentioned earlier, economically advantaged mothers are more likely to be in the labor force when they have young children (Damaske & Frech, 2016; Landivar, 2017). Thus, we were careful not to generalize the findings to those who were economically disadvantaged. We will come back to this caveat in the discussion section.

Measures

Work-to-family conflict (WFC) was measured as the average of the four questions (α = .82, .85, .89, .89 at T1, T2, T3, T4, respectively), including (a) “Working leaves you with too little time to be the kind of parent you want to be,” (b) “Working causes you to miss out on some of the rewarding aspects of being a parent”, (c) “Working leaves you with too little energy to be the kind of parent you want to be,” (d) “Because of the requirements of your job, you have to miss out on home or family activities that you would prefer to participate in,” and (e) “Because of the requirements of your job, your family time is less enjoyable and more pressured” (1 = not at all true, 2 = somewhat true, 3 = fairly true, and 4 = very true). Family-to-work conflict (FWC) was measured as the average of the four questions (α = .65, .66, .65, .72 at T1, T2, T3, T4, respectively): (a) “Thinking about your children interferes with your performance at work,” (b) “Because of your family responsibilities, you have to turn down work activities or opportunities that you would prefer to take on,” (c) “Because of your family responsibilities, the time you spend working is less enjoyable and more stressed,” and (d) “When you spend time working, you’re bothered by all the things at home that you should be doing” (1 = not at all true, 2= somewhat true, 3 = fairly true, and 4 = very true).

Childrearing stage was measured as the four waves of the SECCYD, including when the focal child was 6 months old, 15 months old, third grader, and fifth grader.

Job demands

Two objective and one subjective measures of job demands were examined. Weekly work hours was measured as mothers’ self-report of usual work hours. In supplemental analyses (not shown), results were replicated using four dummy variables including 1 – 20 hours (reference), 21 – 34 hours, 35 – 42 hours, and 43+ hours per week with no notable difference from the results presented here. Occupational status was measured as three dummy variables including management, professional, and other (reference). Job pressure, a subjective measure of job demands, was the average of three items (α = .76, .77, .84, .80 at T1, T2, T3, T4, respectively) asking mothers about concerns regarding their jobs including: (a) “Having to juggle conflicting tasks or duties?” (b) “Having too much to do?” and (c) “The job’s taking too much out of you?” (1 = not at all a concern, 2 = somewhat of a concern, 3 = of considerable concern, and 4 = of extreme concern).

Job resources

One objective measure of job resources was mother’s annual earnings, which was measured in ten thousands of dollars. The Consumer Price Index (https://www.bls.gov/cpi/) was used to adjust 1991 and 1992 dollars (T1 & T2) to 2002 dollars (T4). We examined four subjective measures of job resources. Perception of few career opportunities was the average of three items (α = .68, .68, .54,.55 at T1, T2, T3, T4, respectively), including (a) “Having little chance for the advancement you want or deserve?”; (b) “Making less money than you feel you deserve?”; and (c) “Lack of respect at your workplace for people who do your job?” (1 = not at all, 2 = somewhat, 3 = considerably, 4 = extremely). Perception of supervisor support was the average of four questions (α = .65, .67, .76, .78 at T1, T2, T3, T4, respectively) regarding job concerns and job rewards, including: (a) “Lack of support from your supervisors?” (reverse-coded); (b) “Your supervisor’s lack of appreciation for your work” (reverse-coded); (c) “Your supervisor paying attention to what you have to say?”; and (d) “Your supervisor’s respect for your abilities?” (1 = not at all to 4 = extremely). Task significance, which have also been called meaningful work or challenging work (Nomaguchi 2009; Voydanoff 2004), was the average of the 9 items (α = .87, .87, .91,.88 at T1, T2, T3, T4, respectively), (a) “Doing work you consider important?”; (b) “The sense of accomplishment and competence you get from doing your job?”; (c) “Having an impact on other people’s lives?”; (d) “The job’s fitting your skills?”; (e) The appreciation you get?; (f) The recognition you get?; (g) Being able to work as part of a team or group? (h) “Being able to make decisions on your own?”; (i) “Having the authority you need to get your job done?” (1 = not at all to 4 = extremely). The last two items of task significance could indicate job autonomy, but factor analyses (not shown) demonstrated that these 9 items form one construct. Perceived job flexibility was measured by one question that was phrased in slightly different ways between the first two and the latter two waves. In 6 months and 15 months, the question was: “How flexible are your work hours?” (1 = not at all, 2 = only little; I can be gone if it’s an emergency, 3 = fairly flexible, 4 = completely flexible). In third and fifth grades, mothers were asked: “Sometimes people need to take time away from their job for family responsibilities. How flexible would you say your job is in this situation?” (1 = very inflexible, 2 = somewhat inflexible, 4 = somewhat flexible, 5 = very flexible). Because the number of response categories were different, we standardized scores of each question.

Controls

Annual household income by other members was created by subtracting mothers’ earnings from the total household income and was measured in thousands of dollars. Dollars in 1991 and 1992 (T1 and T2) were adjusted to 2002 (T4) dollars. When mothers’ earnings were equal to the total household income, 0 dollar was assigned. Partnership status was measured as three dummy variables including married (reference), cohabiting, and single. The number of children in the household was an ordered variable ranging from 1 to 8.

Analytical Plan

We first compared mean scores of each measure of job demands and job resources, WFC, and FWC by childrearing stage. We used t-tests to assess whether differences in means were significant. We then conducted multivariate analyses, using fixed effects regression models (Allison, 2009), which utilized pooled data of the four waves (N = 3,096 person-year data). Fixed effects models focus on within person variation and examine the associations between differences in explanatory variables across waves and differences in outcome variables across the same waves (Allison, 2009). The results of Hausman tests (not shown) indicated that there were unmeasured individual characteristics correlated with WFC and FWC, suggesting the need for fixed effects models for the current analyses (Allison, 2009). For missing cases, we conducted multiple imputation using PROC MI in SAS with 25 iterations for each wave and combined the four waves of imputed data to create the pooled data set (Allison, 2001).

To assess H1a and H1b, we examined how each of the eight job characteristic varied by childrearing stage controlling for family characteristics. Then we used hierarchical analyses (Cohen, Cohen, West, & Aiken, 2013) to examine how the association between childrearing stage and WFC or FWC would change when job demands and resources were controlled for. We presented eight models. Model 1 (baseline) included childrearing stage and control variables. Models 2 to 4 examined the role of job demands (H2a, H3a), with Model 2 adding objective job demands to Model 1, Model 3 adding subjective job demands to Model 1, and Model 4 adding all job demands. Models 5 to 7 examined the role of job recourses (H2b, H3b), Model 5 adding an objective measure of job resources to Model 2, Model 6 adding subjective job resources to Model 2, and Model 7 adding all job resources. Note that we included objective job demands in Models 5 to 7 as controls as discussed in the background section. Model 8 included all job demands and resources.

RESULTS

Table 2 presents means for all variables for the pooled sample in the first column and by childrearing stage in the next four columns. Means for job demands and resources varied by childrearing stage and patterns of variation differed depending on aspects of these characteristics. On average, mothers worked fewer hours when their children were 6 or 15 months old than when their children were in third or fifth grade. There were no significant differences in the percentage of mothers who held managerial or professional occupations by childrearing stage. Mothers experienced greater job pressure when their children were 6 or 15 months old than when their children were in third or fifth grade. Mothers also felt more job pressure when their children were in third grade than when their children were in fifth grade. Turning to job resources, mothers had lower earnings when their children were 6 or 15 months old than when their children were in third or fifth grade. Mothers perceived fewer career opportunities when their children were 6 months old than when their children were in third or fifth grade and when their children were 15 months old than when their children were in third grade. Mothers felt that their supervisors were less supportive when their children were 6 or 15 months old than when their children were in third or fifth grade. Mothers perceived more task significance when their children were 15 months old than when their children were in third or fifth grade. There were no differences in mothers’ perceptions of job flexibility by childrearing stage.

Table 2.

Means (SD) for Variables in the Analyses for the Total Sample and by Childrearing Stages (N = 744).

	Total Sample	Childrearing Stage
		6 months	15 months	Third grade	Fifth grade
Job demands
Weekly work hours [1 – 91]	28.32 (17.81)	24.75 (18.04)	26.23 (18.31)	30.55 (17.10) ***c	31.75 (16.82) ***c
Managers [0 – 1]	0.12	0.11	0.12	0.12	0.14
Professionals [0 – 1]	0.28	0.27	0.26	0.28	0.29
Other occupations [0 – 1]	0.60	0.62	0.62	0.60	0.57
Job pressure [1 – 4]	1.73 (0.77)	1.89 (0.74)	1.84 (0.73)	1.64 (0.79) ***c	1.56 (0.76) ***cd
Job resources
Mother’s annual earnings [0 – 30]	2.90 (2.78)	2.30 (2.16)	2.47 (2.42)	3.25 (2.90) ***c	3.60 (3.30) ***cd
Less career opportunity [1 – 4]	1.61 (0.70)	1.72 (0.72)	1.71 (0.71)	1.49 (0.67) ***c	1.52 (0.68) ***
Supervisor support [1 – 4]	3.15 (0.70)	3.00 (0.67)	3.06 (0.69)	3.27 (0.69) ***c	3.27 (0.71) ***c
Task significance [1 – 4]	2.87 (0.70)	2.86 (0.65)	2.92 (0.63)	2.83 (0.76) *b	2.84 (0.76) a
Job flexibility [−3.78 – 4.06]	−2.E-16 (1.00)	−1.E-14 (1.00)	2.E-15 (1.00)	4.E-15 (1.00)	6.E-15 (1.00)
Control variables
Other household income [0 – 497]	4.64 (4.94)	4.43 (4.01)	4.33 (3.92)	4.67 (5.59)	5.15 (5.89) **b
Number of children [1 – 8]	2.12 (1.01)	1.86 (1.04)	1.89 (1.02)	2.35 (0.90) ***c	2.40 (0.95) ***c
Married [0 – 1]	0.79	0.82	0.82	0.77***c	0.77***c
Cohabiting [0 – 1]	0.06	0.07	0.06	0.04*a	0.05
Single [0 – 1]	0.15	0.11	0.12	0.19***c	0.18***c
Work-family conflict
Work-to-family conflict (WFC)	1.82 (0.77)	1.82 (0.75)	1.84 (0.75)	1.85 (0.78)	1.78 (0.78)
Family-to-work conflict (FWC)	1.47 (0.51)	1.52 (0.52)	1.49 (0.49)	1.45 (0.51)*	1.42 (0.51) ***c

Differences from 6 months were significant at *p < .05; **p < .01; ***p <.001 level. Differences from 15 months were significant at ^ap <.05; ^bp <.01; and ^cp < .001 level. Differences from third grade were significant at ^dp <.05; ^ep <.01; and ^fp < .001 level.

Numbers in [ ] indicate the range of response categories.

Differences in means for mothers’ WFC across the four waves were not significant. For FWC, mothers on average reported less FWC when children were older than when children were younger. Specifically, mothers reported less FWC when their children were in third grade than they did when their children were 6 months old, and reported less FWC when children were in fifth grade than when their children were 6 or 15 months old. There were no significant differences in FWC between 6 and 15 months, 15 months and third grade, or between third and fifth grades, respectively.

In supplemental analyses (not shown), Pearson’s correlation analyses indicated that job characteristics were related to one another. For example, managerial positions were positively related to other measures of job demands, such as work hours and job pressure, but also positively related to measures of job resources, such as earnings, perceived career opportunities, and task significance. The measures of job demands and resources were related to WFC as discussed in the background section except for professional and non-professional jobs. The measures of job demands and resources were also related to FWC except for work hours, managerial jobs, and earnings.

Next, we examined the association between childrearing stage and each of the eight job characteristics using fixed effects models. For each model, we examined supplemental analyses in which we rotated the reference group across the four waves and showed the results for differences in the coefficients with superscripts. As shown in Table 3, the patterns of variation in each characteristic by childrearing stage were mostly similar to those found in the descriptive results with a few exceptions. When control variables were in the model, mothers were more likely to have managerial positions when their children were in fifth grade than when their children were 6 months old, and more likely to have professional jobs when their children were in third or fifth grade than when their children were 6 months old. There was no significant difference in mothers’ report of task significance when children were in third grade and when their children were younger. Altogether, supporting H1a, mothers worked fewer hours but reported greater job pressure when their children were younger. Also supporting H1b, mothers reported lower earnings and perceived less supervisor support and fewer career opportunities when their children were younger.

Table 3.

Fixed Effects Models for Association Between Childrearing Stages and Job Characteristics (N = 744; 3,096 observations).

	Work Hours	Occupation		Job Pressure	Mothers’ Earnings	Perceived Less Opportunity	Perceived Supervisor Support	Task Significance	Schedule Flexibility
		Managersi	Professionalsi
Childrearing stage
15 monthsg	1.476	.116	.034	−.050	.180	−.014	.054	.062	.001
	(.865)	(.087)	(.034)	(.051)	(.138)	(.045)	(.047)	(.044)	(.062)
Third gradeg	7.114***c	.324a	.293***b	−.241***c	1.059***c	−.241***c	.273***c	−.026	−.020
	(.883)	(.237)	(.088)	(.046)	(.145)	(.042)	(.043)	(.043)	(.064)
Fifth gradeg	8.718***c	.538*a	.311**	−.310***c	1.383***cd	−.200***c	.276***c	−.027a	−.033
	(.888)	(.224)	(.098)	(.045)	(.150)	(.042)	(.044)	(.044)	(.063)
Controls
Other income	−.630***	.116***	.107***	−.002	.075***	−.019***	.009**	.014***	.022***
	(.070)	(.020)	(.016)	(.003)	(.012)	(.003)	(.003)	(.003)	(.004)
Number of children	−2.377***	−.322**	−.092	−.023	−.249***	−.007	−.019	−.002	.026
	(.315)	(.099)	(.068)	(.016)	(.052)	(.015)	(.015)	(.014)	(.021)
Cohabitingh	1.259	−1.435**	−1.106***	−.066	−.207	.064	−.003	−.056	.147
	(1.368)	(.494)	(.240)	(.069)	(.223)	(.063)	(.064)	(.061)	(.087)
Singleh	.375	−.104	−.759***	−.009	−.193	.179***	−.045	−.084*	.081
	(.926)	(.218)	(.184)	(.045)	(.152)	(.041)	(.043)	(.042)	(.058)
Intercept		−1.725***	−1.175***
		(.294)	(.197)
R2	.093***			.042***	.095***	.070***	.043***	.022***	.019***
Likelihood Ratio		308.378*** (df = 14)

^*p < .05;

^**p < .01;

^***p < .001.

Standard errors in parentheses.

Differences from 15 months were significant at ^a p <.05; ^b p <.01; ^c p <.001.

Differences from third grade were significant at ^d p <.05; ^e p <.01; ^f p <.001.

The omitted reference groups are: ^g6 months, ^hmarried, and, ⁱother occupations.

Finally, we examined whether variations in job demands and resources by childrearing stage had implications for variations in WFC and FWC by childrearing stage. We first looked at results for WFC (Table 4). The baseline model (Model 1) suggests that there was little difference in WFC by childrearing stage controlling for family characteristics. Model 2 shows that work hours and higher status occupations were associated with greater WFC, and when these objective job demands were controlled for, the coefficient for fifth grade became negative and significant. If it were not for their longer work hours and higher occupational positions when their children were in fifth grade, mothers would have reported lower WFC when their children were in fifth grade than when their children were younger. In contrast, when job pressure was controlled for, the coefficients for third and fifth grades became positive and significant (Model 3). This means that mothers would have reported higher WFC when their children were in third and fifth grades than when their children were younger if it were not for their lower job pressure when their children were older. When all job demands were controlled for (Model 4), coefficients for childrearing stages became smaller and non-significant. These findings support H2a.

Table 4.

Fixed Effects Models for Association Between Childrearing Stages and WFC (N = 744; 3,096 observations).

	Model 1	Model 2	Model 3	Model 4	Model 5	Model 6	Model 7	Model 8
Childrearing stage
15 monthsg	.017	−.005	.040	.022	−.005	.009	.009	.028
	(.048)	(.047)	(.042)	(.041)	(.047)	(.044)	(.044)	(.040)
Third gradeg	.038	−.070	.149***b	.063	−.070	.011	.006	.074
	(.044)	(.043)	(.041)	(.041)	(.043)	(.041)	(.041)	(.040)
Fifth gradeg	−.026	−.160***cd	.116**	.010	−.160***cd	−.087*ae	−.093*ae	.010
	(.043)	(.042)	(.040)	(.039)	(.042)	(.041)	(.041)	(.039)
Job demands
Work hours		.015***		.010***	.015***	.012***	.012***	.009***
		(.001)		(.001)	(.001)	(.001)	(.001)	(.001)
Managerh		.102*		.008	.101*	.182***	.165***	.070
		(.045)		(.042)	(.046)	(.043)	(.045)	(.043)
Professionalh		.096**		−.020	.095**	.128***	.117***	.021
		(.033)		(.031)	(.034)	(.032)	(.033)	(.031)
Job pressure			.459***	.395***				.342***
			(.019)	(.019)				(.020)
Job resources
Earnings					.001		.009	.002
					(.007)		(.006)	(.006)
Less opportunity						.230***	.233***	.134***
						(.023)	(.023)	(.022)
Supv support						−.046	−.045	.023
						(.025)	(.025)	(.023)
Task significance						−.077***	−.080***	−.109***
						(.023)	(.023)	(.021)
Flexibility						−.079***	−.078***	−.060***
						(.014)	(.014)	(.013)
Controls
Other income	−.026***	−.019***	−.026***	−.019***	−.019***	−.014***	−.015***	−.016***
	(.003)	(.003)	(.003)	(.003)	(.003)	(.003)	(.003)	(.003)
N. of children	.009	.046**	.019	.042**	.046**	.044**	.045**	.043**
	(.016)	(.015)	(.015)	(.015)	(.015)	(.015)	(.015)	(.014)
Cohabitingi	.019	.024	.050	.030	.024	.031	.030	.034
	(.069)	(.066)	(.061)	(.060)	(.066)	(.065)	(.065)	(.060)
Singlei	.015	.021	.019	.012	.021	−.017	−.017	−.010
	(.045)	(.043)	(.041)	(.040)	(.043)	(.042)	(.042)	(.040)
R2	.037***	.152***	.239***	.284***	.151***	.229***	.230***	.313***

^*p < .05;

^**p < .01;

^***p < .001.

Standard errors in parentheses. Differences from 15 months were significant at ^a p <.05; ^b p <.01; ^c p <.001. Differences from third grade were significant at ^d p <.05; ^e p <.01; ^f p <.001. The omitted reference groups are: ^g6 months, ^hother occupations, and ⁱmarried.

Turning to job resources, earnings were not related to WFC (Model 5). Perceived career opportunities, task significance, and scheduling flexibility were negatively related to WFC, whereas supervisor support was not related to WFC (Model 6). When these subjective job resources were included in the model, the magnitude of mothers’ lower WFC when their children were in fifth grade than when their children were 6 and 15 months old was reduced, although remained significant (Models 2 vs. 6). Supplemental analyses (not shown) suggest that variation in perceived fewer career opportunities contributed most to these changes in the coefficients for childrearing stages. Model 7 included all job resources, which did not change the coefficients for childbearing stages from Model 6. Unlike H2b, after controlling for job resources, differences in WFC by childrearing stage were reduced but remained. In the full model (Model 8), there were no differences in WFC across childrearing stages. All in all, job demands—work hours and job pressure—appeared to contribute to variation in WFC by childrearing stage than did job resources.

Next, we examined FWC (Table 5). In the baseline model, mothers’ FWC was lower when their children were in third or fifth grade than when their children were 6 months old. Mothers’ FWC was also lower when their children were in fifth grade than when their children were 15 months old. As shown in Model 2, high status occupations, but not work hours, were positively associated with FWC. When these objective job demands were controlled for, differences in the coefficients between 6 months and third grade, between 6 months and fifth grade, and between 15 months and fifth grade became larger, and differences in coefficients between 15 months and third grade became significant. When job pressure, which had a strong, positive association with FWC, was controlled for (Model 3), differences in FWC by childrearing stage were no longer significant. H3a was supported. Note that as we discussed in the background section, the association between work hours and FWC differed whether or not job pressure was controlled for (Models 2 vs. 4).

Table 5.

Fixed Effects Models for Association Between Childrearing Stages and FWC (N = 744; 3,096 observations).

	Model 1	Model 2	Model 3	Model 4	Model 5	Model 6	Model 7	Model 8
Childrearing stage
15 monthsg	−.032	−.034	−.020	−.017	−.034	−.027	−.027	−.014
	(.036)	(.036)	(.033)	(.033)	(.036)	(.035)	(.035)	(.033)
Third gradeg	−.087**	−.099**a	−.028	−.012	−.096**	−.050	−.050	−.003
	(.033)	(.033)	(.031)	(.031)	(.033)	(.033)	(.033)	(.032)
Fifth gradeg	−.116***a	−.132***b	−.039	−.020	−.128***b	−.087**	−.087**	−.015
	(.031)	(.030)	(.030)	(.029)	(.030)	(.030)	(.031)	(.029)
Job Demands
Work hours		.001		−.002*	.002*	.000	.000	−.002
		(.001)		(.001)	(.001)	(.001)	(.001)	(.001)
Managerh		.066*		.004	.078*	.112***	.113***	.047
		(.032)		(.030)	(.033)	(.031)	(.032)	(.031)
Professionalh		.073**		−.003	.080**	.095***	.096***	.029
		(.025)		(.024)	(.025)	(.024)	(.024)	(.024)
Job pressure			.247***	.259***				.236***
			(.013)	(.013)				(.014)
Job Resources
Mother’s earnings					−.006		−.001	−.006
					(.005)		(.005)	(.004)
Less opportunity						.141***	.141***	.072***
						(.016)	(.016)	(.016)
Supv support						−.032	−.032	.014
						(.018)	(.018)	(.017)
Task significance						−.035*	−.035*	−.055***
						(.017)	(.017)	(.016)
Flexibility						−.022*	−.022*	−.010
						(.010)	(.010)	(.010)
Controls
Other income	.003	.003	.004	.003	.003	.005*	.005*	.005*
	(.002)	(.002)	(.002)	(.002)	(.002)	(.002)	(.002)	(.002)
N of children	.028*	.033**	.034**	.030**	.033**	.032**	.032**	.031**
	(.011)	(.011)	(.010)	(.010)	(.011)	(.010)	(.010)	(.010)
Cohabitingi	.074	.090	.091*	.094*	.090	.090	.090	.093*
	(.047)	(.047)	(.043)	(.043)	(.047)	(.046)	(.046)	(.043)
Singlei	.084**	.092**	.086**	.086**	. 092**	.068	.068*	.073*
	(.030)	(.030)	(.028)	(.028)	(.030)	(.030)	(.030)	(.028)
R2	.020***	.027***	.155***	.159***	.028***	.082***	.082***	.175***

^*p < .05;

^**p < .01;

^***p < .001.

Standard errors in parentheses. Differences from 15 months were significant at ^a p <.05; ^b p <.01; ^c p <.001. Differences from third grade were significant at ^d p <.05; ^e p <.01; ^f p <.001. The omitted reference groups are: ^g6 months, ^hother occupations, and ⁱmarried.

Looking at job resources, mothers’ earnings were not related to FWC (Model 5) and adding this variable changed the coefficients for childrearing stages little from Model 2 except that the lower FWC when children were in third grade than when children were 15 months old was no longer significant. Perception of career opportunities, task significance, and scheduling flexibility were negatively related to FWC, whereas supervisor support was not related to FWC (Model 6). Adding these subjective measures of job resources to the model changed the coefficients for childrearing stages from Model 2. Specifically, the lower FWC when their children were in third grade than when their children were 6 months old and the lower FWC when their children were in fifth grade than when their children were 15 months old disappeared. The lower FWC when children were in fifth grade than when children were 6 months old was reduced although remained significant. Supplemental analyses (not shown) indicated that perceived career opportunities contributed most to these changes in the coefficients for childrearing stage. Model 7 added all job resources, which did not change the childrearing stage coefficients much from Model 6. Findings from Models 5 to 7 suggest that H3b was not supported. Finally, when all job demands and resources were controlled for (Model 8), there was no significant difference in FWC by childrearing stage. Similar to the findings for WFC, job demands, especially job pressure, appeared to contribute to variation in FWC by childrearing stage more than did job resources.

To examine robustness of the findings, we conducted several supplemental analyses (not shown). First, we assessed whether the patterns of findings would vary by the age of youngest child at home. The focal child was the youngest child in all cases in the 6 months interview and almost all cases in the 15 months interview. In the third and fifth grade interviews, the average age of youngest child was 6.23 (SD = 2.31) and 8.65 (SD = 2.96) years respectively. We found that interaction terms between childrearing stage and the age of the youngest child on WFC and FWC were not significant. Second, we examined whether the number of waves mothers worked would make differences in the patterns of findings and found this was not the case. Lastly, we conducted the same models for the sample of mothers who were employed in all four waves (N = 405). The patterns of findings were very similar with one exception: mothers reported lower WFC when their children were in fifth grade than when their children were 6 or 15 months old at the descriptive level. This could be because differences in the average weekly work hours by childrearing stage were smaller when the sample was restricted to mothers who were employed in all four waves.

DISCUSSION

Using longitudinal data, this paper examined variations in job demands and resources by childrearing stage and whether these variations relate to variations in WFC and FWC by childrearing stage. Present analysis produced key findings that clarify prior suppositions regarding variations in WFC and FWC by childrearing stage, contribute to the theoretical development of predictors of WFC and FWC, and stimulate further investigations in this area.

We found that variations in job demands and resources by childrearing stage depend on aspects of these characteristics. Mothers work fewer hours but feel more job pressure when their children are younger. Mothers have lower earnings and perceive fewer career opportunities and less supervisor support when their children are younger, but report little difference in scheduling flexibility and task significance across childrearing stages. With these conflicting patterns of variation in job demands and resources across the four waves, there is little difference in WFC across these waves. Controlling for work hours, mothers have lower WFC when their children are in fifth grade than when their children are infants, toddlers, or in third grade. When job pressure, which is lower when their children are older than when their children are younger, is controlled for, mothers have higher WFC when their children are older than when their children are younger. These findings generally support Moen et al.’s (2003) idea that dynamics of job demands and resources across life stages play an important role in shaping WFC. Although researchers tend to focus on child demands when they predict variation in WFC by childrearing stage (Voydanoff, 2004), our findings suggest the need for more attention to nuanced patterns of variation in job characteristics and WFC by childrearing stage.

For FWC, mothers report greater FWC when children are infants or toddlers than when their children are in third or fifth grade. When job pressure, which is greater when their children are younger, is controlled for, mothers’ greater FWC when their children are younger disappears. Perceptions of fewer career opportunities are also related to mothers’ greater FWC when their children are 6 months old than when their children are older, although the greater FWC when their children are 6 months old than their children are in fifth grade does not disappear after controlling for this factor. All in all, higher job pressure and, to the lesser extent perceived fewer career opportunities, that mothers tend to experience when children are younger than when their children are older contribute to mothers’ higher FWC when children are younger. Consistent with prior studies (Grzywacz & Marks, 2000; Nomaguchi, 2012; Voydanoff, 2005), our findings suggest merit for expanding the use of the JD-R model to understanding FWC, not just WFC.

Why mothers perceive greater job pressure, less supervisor support, and less career opportunity when their children are younger is unclear. Although we were unable to investigate this question with our data, other research provides some speculations. One possible reason may deal with mothers’ career stage. Mothers are in their early career stages when their children are younger. Early career workers tend to be enthusiastic and eager to prove themselves, while they might lack the tools of managing problems that arise at work (Hargreaves, 2005). It would be ideal if we could examine whether mothers’ workplace seniority or job experience would explain these variations in mothers’ subjective demands and resources by childrearing stage. Another possibility is discrimination against mothers with young children. Prior qualitative research has reported that mothers experience less supervisor support, miss standard promotions, and are assigned less challenging tasks when they have very young children (Blair-Loy, 2003; Stone, 2007). Such unfair treatment from supervisors may signal the need for greater work effort to prove their commitment to their job, which may increase experiences of job pressure (Voydanoff, 2005). Future research should compare perceptions of job demands and resources between mothers and non-mothers who are in the same career stage.

Our findings have several policy implications, two of which we discuss here in detail. First, policies that help reduce job pressure, such as paid parental leave or job sharing benefits, would be helpful for parents with very young children. Since the early 1990s when mothers in the present analysis were initially sampled, the U.S. government introduced the 1993 Family and Medical Leave Act (FMLA) (Gault et al., 2014). Yet, the FMLA affords unpaid leaves only and the eligibility for the FMLA leave is limited to workers who are employed by an organization with 50 or more employees and have worked 1,250 hours or more in the previous 12 months at their current jobs (U.S. Department of Labor, 2017). Behind the media headline that a handful of progressive companies have extended their paid parental leave benefits to cover 18 weeks, 6 months, or even one year after childbirth (Smolkin, 2016), only 12 % of U.S. private sector workers had access to paid parental leave through their employers in 2013 (Gault et al., 2014). More efforts are needed to make paid parental leave available for more workers. Keep in mind that our findings suggest that with the same amount of work hours, mothers’ WFC does not decline until their children reach fifth grade. Policies that allow parents to choose when they take a leave may be more effective than paid leave policies that are fixed to the period right after birth or adoption of a child. Second, paid parental leave or flexible work schedule policies may not be effective in reducing parents’ work-family conflict if the use of these benefits is associated with loss of career opportunities. Recent qualitative work has documented that, even when family-friendly policies are available, employees, especially men, do not utilize them because of the fear that they will be seen as uncommitted to their careers (Moen, Lam, Ammons, & Kelly, 2013; Williams, Blair-Loy, & Berdahl, 2013). Changing workplace culture is necessary so that reducing work hours, taking paid leaves, or using flexible schedules when needed is considered as a part of the ideal worker archetype.

Mothers in the present analyses gave birth in 1991, a majority of whom were born in the 1960s. We used this sample because this is the best available data set that serves for the purpose of the present study. Twenty-five years later, combining motherhood with paid work seems to be more normative as a general idea than before (Pepin & Cotter, 2018). Yet, in practice, policies and culture in the workplace have changed little to accommodate for mothers’ participation as regular employees (Moen, 2011; Stone, 2007). The U.S. is the only industrialized country that does not mandate any paid parental leave (Gault et al., 2014). Workplace culture continues to assume workers’ intensive commitment to their jobs and uses a career development model that only suits individuals without childcare responsibilities (Moen, 2011). Parenting is still highly gendered with mothers being more likely than fathers to shoulder childcare responsibilities and adjust their paid work activities around the needs of their children (Bianchi, 2011; Damaske, 2011; Milkie, Raley, & Bianchi, 2009). Thus, not surprisingly, the stereotype that mothers are less competent in the workplace remains even among recent cohorts of young adults (Benard & Correll, 2010). Together, our results are relevant to today’s U.S. mothers.

Mothers in the current study are more economically advantaged than their counterparts in the general population. The findings and their policy implications we discussed here may not be applicable to mothers with low socioeconomic status (SES). Researchers have argued that the most relevant challenges for parents to balance work and family responsibilities differ by SES (Bianchi, 2011). For low SES mothers, staying employed is a primary challenge because of unpredictable work schedules set by employers in addition to the lack of access to child care, paid parental leave, or paid sick leave (Damaske, 2011; Landivar, 2017). Investigating whether the degree of these issues that low SES mothers tend to encounter differs by childrearing stage is important in order to better inform policy makers about effective programs that would help low SES mothers balance work and family responsibilities.

This study has other limitations. First, although fixed effects models are useful in eliminating unobserved heterogeneity in time invariant variables, they are unable to detect the influence of time varying characteristics that we did not have information on. Also, we did not assess the causal direction of the associations or formal mediations. Although we used longitudinal data, a causal analysis would require strong assumptions about absence of confounding variables at various stages. Future research that employs longitudinal causal mediation models (MacKinnon, 2008) is needed. Second, some of the measures should be improved. The measures of WFC and FWC in the present analyses focused on work-parenting conflict, whereas measures of other studies cover a wider range of family responsibilities. Job flexibility was measured by a single question and its wording was slightly different across childrearing stages. Some potentially important measures of job demands and resources were not available such as job seniority. Finally, although collected in 10 cities in various regions of the U.S., the sample was not nationally representative of employed mothers with children of equivalent ages. Future research that utilizes a representative sample would be ideal.

The findings presented here suggest potential venues of the future research. First, our findings encourage future theoretical development of the JD-R model to investigate FWC not just WFC. Second, it is important to examine fathers’ WFC and FWC as well as mothers’. Prior research has shown that fathers today experience WFC as much as mothers do (Nomaguchi, 2009) and that WFC has negative consequences for fathers’ well-being (Nomaguchi & Johnson, 2016). Third, as Kurz (2000) argued, we need to know more about specific issues of integrating work and family life for parents with middle-school or high-school children. Finally, although the current analysis focused on demands and resources in the job domain only, future research should expand the scope to examine changes in the demands and resources both in the job and family domains across life stages. Prior research has found that fathers tend to shoulder more child care when their wives are employed (Raley, Bianchi, & Wang, 2012) and that father involvement in child care is related to less parenting stress for mothers (Nomaguchi, Brown, & Leyman, 2017). Fathers may do more child care when children are older, which may contribute to lower WFC and FWC of mothers when children are older. Such examinations may help us to better understand the life stage dynamics of mothers’ and fathers’ experiences of integrating work and family life.

In conclusion, using longitudinal data, we found that on average mothers report greater FWC when their children are infants or toddlers than when their children are late elementary school age, whereas there is little difference in mothers’ WFC across these childrearing stages. The observed non-difference in WFC by childrearing stage is largely because mothers tend to reduce work hours but experience greater job pressure when their children are younger. Greater job pressure that mothers experience when their children are younger explains their higher FWC when children are younger. These findings suggest the merits of further investigations on the role of workplace policies and culture in shaping variations in mothers’ work-family conflict—both WFC and FWC—by childrearing stage.