Abstract

Despite numerous changes in women’s employment in the latter half of the twentieth century, women’s employment continues to be uneven and stalled. Drawing from data on women’s weekly work hours in the National Longitudinal Survey of Youth (NLSY79), we identify significant inequality in women’s labor force experiences across adulthood. We find two pathways of stable full-time work for women, three pathways of part-time employment, and a pathway of unpaid labor. A majority of women follow one of the two full-time work pathways, while fewer than 10 % follow a pathway of unpaid labor. Our findings provide evidence of the lasting influence of work–family conflict and early socioeconomic advantages and disadvantages on women’s work pathways. Indeed, race, poverty, educational attainment, and early family characteristics significantly shaped women’s work careers. Work–family opportunities and constraints also were related to women’s work hours, as were a woman’s gendered beliefs and expectations. We conclude that women’s employment pathways are a product of both their resources and changing social environment as well as individual agency. Significantly, we point to social stratification, gender ideologies, and work–family constraints, all working in concert, as key explanations for how women are “tracked” onto work pathways from an early age.

Introduction

A majority of women in the United States now participate in the paid labor market (U.S. Bureau of Labor Statistics 2009), spending an increasing amount of their adult lives employed (Bianchi et al. 2006). Yet recent research suggests women’s employment continues to be “uneven and stalled” (England 2010:149). While women’s workforce participation grew rapidly through the late 1990s, this progress appears to have arrested or even decreased slightly since the early 2000s (Boushey 2008). Moreover, these patterns are stratified such that the least advantaged women are also the least likely to work full-time. Approximately one-third of all employed women are employed part-time and are disproportionately working-class, working-poor, and women of color (England et al. 2004; Milgrom and Petersen 2006). In light of these changes, this article looks at inequality among women by comparing the diversity of their labor force experiences. We seek to uncover the different ways that women participate in the workforce and the variables that best predict why some women participate steadily in full-time work while other women do not.

Our goal is to construct a more comprehensive empirical and theoretical understanding of how women’s personal characteristics and structural positions place them on long-term pathways of workforce participation. Prior research suggests that women’s workforce participation varies significantly over time and that many of the same variables that predict women’s employment status—a measure of current employment—also explain their workforce participation over time (Hibbard and Pope 1993; Hynes and Clarkberg 2005; Moen et al. 1992; Pavalko and Smith 1999; Williams and Han 2003). Yet, these same studies are limited by the scope of the data that were available—either shorter time durations or more regionally specific populations than our current study. Thus, we propose a life course approach that draws on existing theories explaining women’s workforce participation and applies these theories to a nationally representative sample of Baby Boomer women as they age.

We use more than three decades (1979–2010) of panel data from the National Longitudinal Survey of Youth 1979 to study variation in women’s work hours through young and middle adulthood. Empirically, we make a novel contribution by estimating group-based developmental trajectories (Nagin 2005) of women’s average weekly work hours from ages 25 to 45, allowing us to build on previous longitudinal studies that had identified 3–6 distinct pathways of women’s workforce participation over time (Damaske 2011; Hibbard and Pope 1993; Hynes and Clarkberg 2005; Moen et al. 1992; Pavalko and Smith 1999; Williams and Han 2003). Theoretically, we build an innovative framework that draws on three existing theories of how and why women participate in paid work: cumulative disadvantage (Dannefer 2003), gender beliefs (Ridgeway and Correll 2004; Risman 1998), and work–family constraints (Gerson 1985; Stone 2007). By integrating these theories, we seek to better understand the uneven effects of the gender revolution on women’s paid workforce participation across adulthood.

The Gendered Life Course and Women’s Work

Gender shapes workforce participation and family life across the life course (Moen 2001), and women’s occupational experiences across young adulthood are shaped by a series of gendered transitions, including entrance into work and family formation (Moen and Han 2001; Williams and Han 2003). Life course researchers investigate the early-life precursors and long-term consequences of pivotal events or turning points, including those related to family formation and workforce participation (Elder 1998). Work stoppages, time out of work, and time spent caring for children often disrupt women’s careers (Hostetler et al. 2007). Yet, what remains unknown are the number and type of longitudinal pathways that characterize women’s overall workforce participation and their precursors early in life.

Much of the early literature on women’s work over the life course focused on the relationship between work and health; this research also provides insight into how the gendered life course influences women’s long-term workforce participation. Using data from the National Longitudinal Survey of Mature Women, Pavalko and Smith (1999) found that older women followed one of three paid workforce pathways: women were employed full-time, intermittently (including women who work part-time or moved in and out of the workforce), or not at all (Pavalko and Smith 1999). Using regional panel study data that interviewed women at two distinct periods, Hibbard and Pope (1993) found that family obligations and childrearing impacted women’s employment stability, which was also associated with marital stability. This study also found that the majority of participants who were employed at the initial interview remained stably employed over the 15 years between interviews, while the majority who were not employed remained outside of the workforce entirely for the 15-year period. Finally, early research established that the gendered life course structures women’s lives so that they may hold different roles at different times in their life, but the number and types of roles that they hold early in life matters for the roles that they will hold in middle age (Moen et al. 1992).

More recently, researchers have investigated work pathways apart from their relationship to health. A study of women in the NLSY79 data that centered on the transition to parenthood enumerated six longitudinal (but relatively short) work patterns in the 12 months prior and 24 months following childbirth (Hynes and Clarkberg 2005). Similarly, a study using retrospective panel data of men and women in New York State identified six distinct work pathways over a 15- to 30-year period (Williams and Han 2003). This study found that half of the pathways were marked by full-time and relatively stable employment, while the other half were marked by changes in employment status as well as employer, markers of relative instability (Williams and Han 2003). Tracing women’s work pathways through analysis of qualitative life histories with 80 randomly sampled women in New York City, Damaske (2011) distinguished three types of workers: (1) steady workers, who work full-time steadily across their adult lives; (2) pulled back workers, who work part-time or not at all; and (3) interrupted workers, whose unemployment experiences shape their labor market participation and who exit and reenter the labor market.

In sum, the existing literature provides important insights into women’s work pathways by drawing attention to the significant differences among women’s longitudinal work patterns, by linking early work and family responsibilities with labor force participation in later years, by identifying how critical transitions points such as motherhood shape work patterns, and by identifying unemployment as a crucial but understudied component of women’s work pathways. Moreover, Damaske’s (2011) research suggests that cumulative disadvantages combine with gender beliefs and work–family opportunities and constraints to shape women’s work pathways. Yet, to our knowledge, none of the prior longitudinal studies have sought to both identify long-term patterns in women’s workforce participation and draw on existing theory to predict which women are most likely to gain access to full-time, stable work. Research also remains incomplete because of the use of retrospective rather than prospective data, limited time spans for follow-up or available waves of data, regional or small samples, or data on older cohorts experiencing different key transition points into adulthood.

Given the existing literature, we first hypothesize:

  • Hypothesis 1 Women’s workforce participation, measured by average weekly hours of paid employment per year across ages 25–45, will follow at least three group-based pathways, including a pathway of steady continuous full-time work, a pathway of continuous unpaid work over time, and one or more pathways of pulled back intermittent work (i.e., work that is part-time and that may include workforce entry and exit across the life course).

A Life Course Approach to Women’s Work Pathways

Cumulative Advantages/Disadvantages

The cumulative advantages/disadvantages perspective suggests that early socioeconomic inequalities may restrict poorer women’s opportunities to engage in full-time, steady work as they enter into adulthood. Early advantages and disadvantages related to race/ethnicity, family poverty, educational attainment, single parenthood, and local labor markets accumulate over time, affecting socioeconomic status, health, and well-being in adulthood (Dannefer 1987; Elder 1998; O’Rand 2006; Willson et al. 2007). Despite research demonstrating the ways that early socioeconomic status (SES) plays an important role in women’s life chances (Frech and Damaske 2012; O’Rand 2006), we know of no research that has explicitly examined the relationships between women’s longitudinal work pathways and a range of early socioeconomic, racial/ethnic, family formation, and other advantages/disadvantages. In other words, women’s ability to work full-time across the life course may be an accrued advantage.

There is substantial evidence of socioeconomic variation in adult women’s workforce opportunities and participation. Middle-class women are more likely to be employed full-time than are working-class women, differences often attributed to variation in educational attainment (England et al. 2004; Percheski 2008). Educational attainment also has recently been shown to be associated with better marital prospects (McClendon et al. 2014). Experience of poverty early in life may also limit women’s opportunities to find and keep stable employment (Damaske 2011). Working-class, black, and Latina women are also more likely to have interruptions to their paid labor—including workforce exits and time spent unemployed—which may have long-term effects on wages and average hours spent in the workforce (Alon and Haberfeld 2007; Reid and Padavic 2005). Furthermore, white women are less likely to enter motherhood at a young age than are black and Latina women, and age at first birth may truncate women’s anticipated educational attainment and curtail attachment to the labor force by disrupting a work trajectory before it begins (Hoffman and Maynard 2008). Women’s wages are negatively affected by high unemployment levels (McCall 2001), which suggests that women’s labor force participation also may be negatively impacted by higher unemployment levels.

Early SES may not only play a role in these early life transitions but may also continue to have repercussions across the life course. Cumulative advantage/disadvantage theory (CAD) posits that early differences in structural location, skills, and resources accumulate over time, widening the gap between the more and less advantaged (Dannefer 2003). As individuals age and face pivotal transitional periods, such as marriage, childbearing, educational opportunities, and workforce participation, those with fewer resources face greater constraints in their choices and the greatest “life course risks” (O’Rand 2006:149). The possible cumulative effect on work pathways of these early disadvantages leads to our second hypothesis:

  • Hypothesis 2 Consistent with literature on cumulative advantages and disadvantages, we expect that socioeconomically disadvantaged women (less-educated women, those whose mothers did not graduate high school, those living in nonintact families during adolescence, those living in poverty during young adulthood, and those in counties with higher unemployment as young adults) will be less likely to engage in continuous full-time work (steady work) relative to more-advantaged peers.

Gender Beliefs

To understand how and why women participate in paid work over the course of their lives, prior research has also investigated gender beliefs (Correll 2004; Gerson 1985; Risman 1998). The gender beliefs perspective suggests that women’s gendered perceptions of the workforce, expectations about gendered occupations, and expectations about motherhood shape their continued employment, making some women more likely to track into unpaid or part-time work (Blair-Loy 2003; Correll 2004; England 2010). Gender beliefs guide decisions about workforce participation, marriage, and family formation (Ridgeway and Correll 2004; Risman 1998). Beliefs about gender are particularly influential during transitional periods in the life course, such as entry into the paid labor market, marriage, or childrearing (Damaske 2011; Vespa 2009), suggesting that they may also influence women’s work pathways.

Gender beliefs may lead women to enter occupations that are stereotyped as (often lower-paying) female fields, such as clerical or service work (England 2010), leading them to stagnate in or leave low-wage and lower-skill work. Women who anticipate low job prospects or who see women’s jobs as ancillary may anticipate working only occasionally (Damaske 2011). Women’s beliefs about who is most responsible for caring for children may also influence their workforce patterns: the notion that women are the preferred caregivers for young children continues to be almost universally held (Folbre 2001). Women who believe that they will leave the workforce when they become mothers may choose not to enter the workforce or may enter into more transitional employment because they anticipate leaving in the future. Some women prefer to work until they have children, when they may decide to leave work to focus on their children or they might move to part-time work (Stone 2007; Webber and Williams 2008). It is important to note that the “choice” to stay at home with children is one that is often made within a constrained framework and one made in a society in which women learn at a young age to prioritize caregiving (Folbre 2001; Stone 2007; Webber and Williams 2008). Women’s expectations about their future in the workforce and their plans to prioritize childrearing do contribute to the decisions they make about their education, their marital preferences, and their childbearing, although structural position may also shape women’s ability to act upon this agency (Damaske 2011)

  • Hypothesis 3 When adding women’s workforce ideologies to our previous models, we expect that women who (as young adults) held more-traditional gender beliefs or who did not have expectations of long-term participation in full-time paid work will be more likely to follow stay-at-home pathways and less likely to work steadily full-time across adulthood than women who held more egalitarian gender beliefs or early expectations of working full-time throughout adulthood.

Work and Family Opportunities and Constraints

Women’s dual responsibilities in the home and at work may explain whether they remain in long-term stable employment, with women most likely to stay employed when they find both opportunities in the labor market and support at home for their paid work (Gerson 1985; Stone 2007). In other words, micro-level experiences in the workplace and at home may shape women’s work pathways (Cha 2010; Gerson 1985; Stone 2007). Indeed, many researchers have noted that the conflict between work and family begins following the transition to parenthood and is most intense when children are young (Bianchi et al. 2006; Moen 2001). Additionally, the work and family responsibilities that women take on in young adulthood is associated with their employment later in life (Hibbard and Pope 1993; Moen et al. 1992).

When women find “good jobs”—that is, full-time, year-round jobs with benefits—they are more likely to remain employed (Nelson and Smith 1999). On the other hand, women are more likely to leave work if they perceive that there are constraints to finding work or advancing within the company and if they work in fields that demand “overwork” and have inflexible schedules (Cha 2013; Gerson 1985; Stone 2007). Women’s labor force participation also appears to be related to their own experience of job loss, with job losses decreasing women’s likelihood of participating steadily in the labor market and increasing their likelihood of pulling back from work (Damaske 2011). Moreover, time spent unemployed may be an indicator of strong local work constraints, with a high number of weeks spent looking for a job suggestive of a poor local economy (Hout et al. 2011; McCall 2001). Responsibilities at home factor into how and whether women work for pay. When husbands work more than 50 hours per week, wives are less likely to participate in the labor market because their husbands do not participate in second-shift tasks at home (Cha 2010). A husband’s high wages may decrease women’s workforce participation, particularly when the husband’s earnings are significantly higher than the wife’s (Shafer 2011).

  • Hypothesis 4 After adding variables related to cumulative advantage and disadvantage, we expect that women’s early work experiences and family opportunities and constraints at age 25 will structure their long-term pathways of work such that women with the fewest work constraints (i.e., those reporting few barriers to work, a spouse not working long hours, fewer or no children, and no experience with job loss or long-term unemployment as young adults) and the greatest work opportunities (i.e., work in professional occupations or continuing higher education at age 25), will be most likely to work steadily.

Data, Measures, and Methods

Data and Sample

We identify longitudinal group-based pathways of women’s workforce participation using the National Longitudinal Study of Youth–1979 (NLSY79). The NLSY79 is a nationally representative sample of more than 12,000 late Baby Boomers born between 1957 and 1964, and began when respondents were ages 14–21 in 1979. The NLSY79 women came of age during a period of remarkable change. Women’s labor force participation grew significantly from the 1960s through the 1990s, particularly for the college educated; overall working hours for women with young children increased during the 1970s and 1980s; the divorce rate doubled from 1960 to 1977; and nonmarital births increased from 5.3 % in 1960 to 30.1 % in 1992 (Coleman and Pencavel 1993; Klerman and Leibowitz 1994; Lerman and Schmidt 1999). NLSY79 respondents were interviewed annually through 1994 and biennially thereafter.

Our sample is first limited to the 4,930 women who were not part of the military or low-income oversamples that were not followed after 1984. We further limit the sample to the 4,713 women who provided at least three cross-sectional measures of workforce participation (or nonparticipation) between the ages of 25 and 45.

A significant strength of the NLSY79 is its high retention rate: roughly 80 % of the baseline sample interviewed in 2010, the last year from which we draw employment data. We impute missing values resulting from item nonresponse for explanatory variables (but do not impute information related to weekly paid work hours, our dependent variable; see von Hippel 2007) using the ice command in Stata 12. For all imputed measures with the exception of spousal wages, well under 15 % of data are missing (about 17 % of spousal income data are missing and imputed among married women at age 25.)

Measures

Dependent Variable

Weekly work hours

We use the created Work History File1 to calculate women’s average weekly work hours in the paid labor force across ages 25–45. The NLSY79 Work History Files include created variables describing women’s total hours worked in the last calendar year, weeks worked in that year, and time spent out of the labor force or unemployed across all jobs using week-by-week arrays that began tracking work in 1978 and continued through the most recent round of data collection.2 We use data on women’s work hours beginning in 1982, when the oldest women in the sample turned 25, continuing through 2010, when the youngest women in the sample turned 45. We structure the data by age so that women born in 1957 (the oldest women in the sample) provide “age 25” data in 1982, women born in 1958 provide “age 25” data in 1983, and so on, through age 45, when women born in 1964 or 1965 (the youngest women in the sample) provided the last round of work hour data in 2010, when they were 44 or 45 years old, respectively. To calculate women’s average weekly work hours, we divide the NLSY-created variable for total number of hours worked in the last 12 months by the NLSY-created variable for number of weeks worked over the last 12 months, giving us the average number of hours in paid work per week during employed weeks. Scores of 0 are assigned only to women who reported 0 weeks of paid work in the last year. Nearly 90 % of our sample provides eight or more cross-sectional reports of average weekly work hours between ages 25 and 45 (or calendar years 1982–2010), with a total range of 3 waves of data provided (n = 28 women, less than 1 % of the sample) to 11 waves of data (n = 2,656 women, 55 % of the sample). Table 1 details the number of person-year observations at each age, the calendar years corresponding to women’s age in years when data were collected, the means and standard deviations of women’s work hours by age, and the range of hours by age.3

Explanatory Variables

Explanatory variables predicting women’s pathways of workforce participation are presented in Table 2.4 Variables related to cumulative advantage and disadvantage include individual, community, and family-of-origin characteristics: race/ethnicity and U.S. nativity, household poverty status at ages 19–22 (1 = household falls below poverty threshold), respondent’s years of completed education by age 25 (centered on 12 years), whether the respondent ever becomes a parent (included to allow for interactions), teen birth (1 = gives birth before age 18, conditional on becoming a parent), and nonmarital birth (1 = not married at first birth, conditional on becoming a parent). At the community level, we control for job availability and employment context using rural/urban residence (1 = rural) and the county-level unemployment rate when the respondent was 19–22 years of age (and potentially just entering or having just entered the workforce).5 We use two proxies to measure family of origin socioeconomic status: (1) whether a residential mother did not graduate from high school, because maternal education levels are associated with offspring cultural capital (Domina and Roksa 2012); and (2) whether a respondent lived with two biological parents as an adolescent, because single parenthood and economic hardship are strongly associated (McLanahan and Percheski 2008).

Between 19–22 years of age, women are asked about their gender beliefs (higher scores indicate more-traditional gender beliefs), whether they believe gender is a barrier to a “good” job (1 = perceives gender as barrier), and whether a woman expects to work at age 35 (the reference and modal category), work and raise a family, or raise a family.

Early work experiences and family opportunities and constraints are measured at age 25 and include perceived barriers to a good job (including those related to transportation, race/ethnicity, nationality, language, or age), unemployment experiences by age 25 (spent at least four weeks unemployed in a year or experienced involuntary job loss via firing or layoff),6 current occupation at age 25 (including categories for students, homemakers, and the unemployed, with service-sector work as the reference category), marital status (to account for spousal characteristics using internal moderators (Mirowsky 1999), spousal wages (among the married, in thousands, logged), spousal work hours (centered on 40 hours per week, among the married), and number of residential children by age 25 (conditional on ever becoming a parent).

Analytic Strategy: Group-Based Developmental Trajectories

We use group-based developmental trajectory models, a type of finite mixture modeling (Nagin 2005) in Stata 12 (Jones and Nagin 2013) to identify group-based pathways of women’s average weekly hours across ages 25–45. Group-based trajectory models identify clusters of women following common pathways, or developmental trajectories of work hours as they age; these models can also identify risk factors for entering these pathways and their associated outcomes (Nagin 2005). This analytic approach differs from hierarchical or growth curve modeling because of its focus on group-specific, rather than within-individual, trajectories of change over time. Developmental trajectories are better aligned with our hypotheses and with recent scholarship identifying three to six distinct patterns in women’s work pathways (see Damaske 2011; Williams and Han 2003). After identifying pathways of work hours, we estimate stepwise models predicting women’s entry into these pathways, adding in variables from our cumulative advantage/disadvantage, gender beliefs, and work–family hypotheses. Although these theories historically have been considered distinct, we argue that the variables used to test these theories are interrelated under a life course framework.

Results

Identifying Group-Based Pathways of Women’s Workforce Participation

Our first research aim is to identify the number and shape of women’s pathways of workforce participation (measured by average hours spent per week in paid work) across ages 25–45. Nagin (2005) argued that a model of best fit is parsimonious and both empirically and theoretically sound. We use existing research describing women’s workforce participation across the life course and conduct post hoc tests of model fit to identify the optimal number of groups and shape of pathways describing women’s average weekly work hours across ages 25–45. Table 3 describes the model parameters for a six-group model; Table 5 in the  appendix includes the average predicted probability of group membership (APP) for women in each group-based pathway, along with Bayesian information criterion (BIC) statistics for models tested with as few as three and as many as eight pathways of workforce participation. Consistent with Nagin’s (2005) recommendation that each group average an APP of at least .700, our six-group model has APPs of .946, .897, .928, .833, .914, and .856. Although seven- and eight-group models have lower BIC statistics and high APP values, the additional groups do not meaningfully vary from those depicted in the six-group model. Specifically, the eight-group model breaks out the pathways of gradually increasing hours and decreasing hours into two separate pathways each, and the seven-group model identified two groups of women who gradually decreased work hours over time but began this decline in hours at different ages (mid-30s vs. earlier). Thus, we select the six-group model as the most parsimonious model depicting common group pathways of women’s workforce participation across ages 25–45. We plot the six-group model in Fig. 1.

Figure 1 provides strong support for Hypothesis 1, predicting a pathway of steady full-time workforce participation, a pathway of workforce nonparticipation across adulthood, and one or more categories of pulled back (either intermittent or part-time) participation. First, we see evidence of two distinct categories of full-time working women: one category including women averaging work hours persistently at or above 40 hours per week (which we term “Overwork”), and a modal category of women steadily working at or near 40 hours per week during weeks spent employed across ages 25–45 (termed “Steady”). Also in line with previous research, we identify 8.2 % of women engaging in continuous nonparticipation in paid work (“Stay-at-Home”). We identify three distinct pathways of intermittent or part-time workforce participation among women. Figure 1 shows women engaging in pathways of “Increasing Hours” (10.5 %), “Decreasing Hours” (8.6 %), and “Workforce Reentry” (10.6 %). In sum, women’s paid workforce participation varies over time and across women, likely reflecting women’s dual roles as workers and primary caregivers.

Predicting Entry Into Pathways of Women’s Workforce Participation

The second stage of our analyses uses existing explanations of women’s work to build stepwise models predicting women’s entry into pathways of weekly work hours. In Models 1–4 of Table 4, log-odds coefficients report women’s increased or decreased likelihood to experience the group pathway under investigation, relative to the pathway of Steady work (our modal and reference category).

Cumulative Advantages and Disadvantages

The life course concept of cumulating advantages and disadvantages (see Elder 1998) emphasizes the long-term role of early-life conditions for later-life socioeconomic attainment and well-being. Indeed, we find in Model 1 of Table 4 that the least-advantaged women in our sample had the greatest difficulty securing Steady work, supporting Hypothesis 2. Education was associated with steady workforce participation for women. In contrast, early poverty, low maternal education, parenthood, and higher county-level unemployment were each associated with intermittent or part-time work across adulthood. Specifically, as women’s education increased, they were less likely to follow a pathway of Stay-at-Home or Increasing Hours and more likely to follow an Overwork pathway. Women who became mothers were more likely to follow Stay-at-Home, Increasing Hours, Decreasing Hours, or Reentry pathways and were less likely to follow a pathway of Overwork. The exception to this is single mothers, who were more likely than mothers with marital births to follow an Overwork pathway across adulthood. Non-Latina black respondents and Latina respondents were less likely to report Reentry pathways, and non-Latina black respondents were less likely to follow Decreasing Hours pathways than non-Latina, nonblack peers. Higher county-level unemployment discouraged steady work from an early age, making Stay-at-Home or Increasing Hours pathways more likely relative to a Steady pathway. Women in rural areas were less likely to follow a Reentry pathway relative to Steady work.

Gendered Beliefs and Expectations

Model 2 of Table 4 adds variables related to women’s gendered beliefs and early expectations (around ages 19–22) of participating in paid versus unpaid work. We find some support for Hypothesis 3 in this model: more-traditional gender beliefs and early expectations about balancing work and family were associated with intermittent (Increasing Hours, Decreasing Hours, or Reentry) and Stay-at-Home pathways of workforce participation relative to Steady work. Perceiving gender as a barrier to finding a good job selected women out of Reentry pathways relative to Steady work, which was not consistent with Hypothesis 3. Cumulative advantage and disadvantage variables generally retained their significance after these Model 2 variables were added.

Early Work Experiences and Family Opportunities and Constraints

Model 3 of Table 4 adds our work–family variables to the variables included in Model 2 to assess their role net of early advantages and disadvantages and women’s beliefs and expectations. As with Model 2, most previously significant variables remained so after work–family variables were added to Model 3 of Table 4, indicating that existing theories of women’s work are best studied together rather than separately. In support of our fourth hypothesis, Model 3 indicates that both family and work circumstances at age 25 were strongly associated with women’s weekly work hour pathways after adjusting for other variables. Practical barriers to a good job and women’s own work and early unemployment experiences were associated with entry into work pathways, indicating that experiencing these barriers early in a woman’s career could have long-term consequences. Women who reported four or more weeks of unemployment in any one year at age 25 or earlier were less likely to follow Stay-at-Home or Increasing Hours pathways and were more likely to follow the pathway of Decreasing Hours across adulthood, suggesting a need for paid work that was not consistently met. Women who had been fired prior to age 25 were less likely to follow an Overwork pathway, likely reflecting the relatively advantaged status of women on Overwork pathways. Occupation and non-employment activities at age 25 also mattered for long-term pathways of workforce participation: schooling was associated with movement into the paid workforce, and clerical work, unemployment, or homemaking at age 25 were associated with intermittent or Stay-at-Home pathways. Women working in a professional field at age 25 were also less likely than women in the service sector to experience an Increasing Hours pathway and more likely to experience Overwork and Decreasing Hours pathways relative to Steady pathways. A surprising finding is that women who were homemakers at age 25 were also more likely to overwork—a finding that is likely unreliable because of small cell sizes.

Regarding marital status, spousal characteristics, and parenthood, married women were less likely to follow a Reentry pathway relative to Steady work, and divorced women were less likely to follow the Stay-at-Home pathway. Higher levels of spousal wages were associated with an increased likelihood to engage in a Reentry pathway, supporting the idea that the Reentry pathway is associated with economic advantages in the home. Married women who had spouses working long hours were themselves more likely to follow the Overwork pathway than the pathway of Steady work. Although parenthood per se continued to be associated with women’s greater likelihood to participate in pathways of Decreasing Hours or Reentry relative to Steady work, the interaction between parenthood and number of children was negative for some groups, indicating that greater numbers of children made it progressively less likely for women to decrease hours or reenter work relative to engaging in steady work, but more likely that women would stay at home or report increasing hours. The trends suggest that family characteristics do not uniformly shape women’s work participation.

Discussion and Conclusions

Understanding why some women work steadily throughout their lives while others work part-time or not at all has long been of interest to gender and work–family scholars (see Gerson 1985; Jacobs 1989; Risman 1998; Stone 2007). This study makes two significant contributions to these research aims. First, we draw from repeated observations of women over time to demonstrate significant variation in their group-based pathways of work hours across adulthood. We find two pathways of stable full-time work for women, three pathways of part-time or intermittent employment, and a pathway of unpaid labor. This study provides strong support for the work pathways identified in Damaske’s qualitative research (2011) and builds on prior quantitative studies (Hynes and Clarkberg 2005; Williams and Han 2003) by using a nationally representative data set over an extended period.

Second, we expand on existing theoretical and empirical explanations of women’s workforce participation by using a life course perspective to combine three theories of women’s work to predict entry into pathways of women’s workforce participation during their 20s, 30s, and 40s. We find strong support for the lasting effects of cumulative disadvantages for women’s work hours, with women’s race, early experiences of poverty, educational attainment, and early family characteristics playing significant roles in shaping women’s work careers. Women’s gendered beliefs and expectations as well as their work–family opportunities and constraints further shaped their work hours. Although prior research has documented the importance of work–family opportunities and constraints on workforce participation (Gerson 1985; Jacobs 1989) and has suggested a role for gender beliefs (England 2010; Risman 1998), our models add early cumulative disadvantages, which appear to have profound influences on women’s work pathways. The role of cumulative disadvantages in women’s workforce participation has received scant attention in past research. Yet, our findings suggest that early disadvantage is best studied in concert with the later correlates of that disadvantage to best understand how women’s long-term work pathways are shaped early on.

Identifying women’s long-term pathways of work refines our understanding of how prevalent each pathway of work is for women. Even though annual data show that about 20 % of mothers are not employed in a given year (Bianchi et al. 2006), only 8.2 % of the women in our sample followed a long-term pathway of stay-at-home mothering. Consistent with prior research, we find that while some women do leave the labor market upon becoming a parent, they generally do not leave the paid workforce permanently (see Hynes and Clarkberg 2005).

This is not to say that women are as likely as men to work continuously full-time in paid labor. The prevalence of women following the pathways of Increasing Hours, Decreasing Hours, and Reentry into the workforce (29.7 % total), in addition to those 8.2 % following the pathway Stay-at-Home, means that 37.9 % of women have weaker labor force attachment than full-time workers through their mid-40s. This supports previous research finding that women’s overall participation in the workforce is uneven (see England 2010) and that pulled back, intermittent or part-time workers are heterogeneous when examined over time (Damaske 2011). Yet, a majority of women in our sample (62.1 %) worked steadily, with full-time or overwork hours across adulthood. On average, then, a majority of women are participating in the labor market at fairly high levels through age 45.

Overall, women on Steady pathways were more advantaged both during their childhood and throughout young adulthood relative to the majority of their counterparts. Women who followed Overwork pathways also experienced early advantages, including higher education and lower poverty. Finding an association between socioeconomic advantage and greater workforce attachment builds on previous cross-sectional research indicating that higher-SES women are more likely to work (see England et al. 2004; Percheski 2008), and shows that this pattern appears to persist over the course of women’s working lives. Steady workers did anticipate greater gender barriers, which might be explained by the possibility that women who anticipated steady work also anticipated entering more male-dominated jobs, for which they saw their gender as a barrier.

The intersection of race and gender also appears important. Black and Latina women were less likely to follow Decreasing or Reentry pathways relative to the Steady pathway, which could reflect the greater need for two incomes in nonwhite households. It could also indicate cultural differences across race in preferences about working or expectations about women’s work (Damaske 2011). Supplemental analyses (not shown but available upon request) in which work pathways were estimated separately by race for whites, blacks, and Latinas, show four of the longitudinal patterns to be quite consistent across race: Steady, Increasing Hours, Decreasing Hours, and Stay-at-Home. The Overwork pattern was not identified among Latinas, and the Reentry pattern was apparent only for white women. These supplemental analyses suggest that the model presented captures the primary pathways for whites, blacks, and Latinas, and also that future research should consider more fully how race matters for women’s workforce participation longitudinally.

Gender beliefs also influenced women’s working lives. As we might anticipate, more-traditional gender beliefs and aspiring to raise a family rather than work or balance work and family at age 35 were associated with an increased likelihood to follow a Stay-at-Home or Increasing Hours pathway relative to Steady work. Those holding the expectation that they would balance work and family at age 35 were less likely to follow an Overwork pathway and were more likely to follow a Stay-at-Home, Increasing Hours, or Reentry pathway. Holding more-traditional gender ideologies, then, decreased the likelihood that women would work steadily or would overwork. Aspiring to work and raise a family significantly decreased women’s likelihood of following the Overwork pathway compared with the Steady one, suggesting that women may anticipate the seeming incompatibility of highly demanding positions and childrearing. Yet, work constraints and opportunities also appear to shape the likelihood that women will stay at home: women who were unemployed at age 25 and looking for work were more likely to follow a Stay-at-Home pathway than a Steady one, which may indicate that women with low levels of labor force attachment were discouraged by poor employment prospects early in the life course. Moreover, finding professional employment appears to increase women’s likelihood of engaging in an Overwork pathway. Surprisingly, being married did not differentiate the majority of women, although women married to spouses working long hours were more likely to overwork, which may reflect marital homogamy in education and profession among the most advantaged women. Finally, the increased likelihood of following a Reentry pathway for women with higher-earning spouses likely reflects the seeming advantages that this group of women holds.

Our findings strongly support the use of a life course framework for understanding women’s work and indicate that women’s likelihood of working steadily is shaped by their likelihood of growing up in poverty, living in a community with high unemployment, having a mother with low education, having lower education levels themselves, holding traditional gender ideologies, and facing a variety of barriers to good work. Important for future research is the interrelated nature of these variables, particularly the relationships between early class position and later-life educational attainment and labor market attachment. Although we cannot disentangle women’s likelihood of pursuing higher education from the other cumulative disadvantage variables or from their gender ideologies, it is likely that they are linked. Working-class men’s disinterest in pursuing higher education (or even completing high school) has long been understood to be a product of their class position (see Willis 1977), yet it is only recently that working class girls’ disenfranchisement from secondary and postsecondary schooling has been considered as much a response to their class position as a result of their gender ideologies (see Bettie 2003; Damaske 2011). Our research suggests that gender ideologies and class background may be deeply linked in the process of shaping women’s work. This subsequently implies that access to steady work may be more available to more-advantaged women. However, advantaged women may, as Stone (2007) found and our research supports, leave the workforce for a time and return to work: the Reentry pathway, like the Steady one, appears to be followed by women who are more advantaged. Although research has long noted the transmission of social class advantage (particularly for men) through education and occupational status, our research suggests that for women, class advantage may also play out through access to continual work.

Life course scholars emphasize the role of cohorts’ shared experiences as they age, and the women in our study came of age during a time of remarkable economic and social transitions in the United States. Our study shows that these transitions had different impacts on different groups. During the 1970s, 1980s, and 1990s, college-educated women and women married to men in the highest earning quintiles saw the greatest growth in their labor force participation and the greatest increase in working hours (Coleman and Pencavel 1993; Klerman and Leibowitz 1994). They are also the women in our study who were the most likely to work steadily, suggesting a sustained change in the labor force participation of the most-advantaged women. On the other hand, during this same period, the increase in nonmarital births was more common among women of color, women living in poverty, and younger mothers (Bratter and Damaske 2013; Lerman and Schmidt 1999). In our study, early poverty and nonmarital births were more likely to be associated with intermittent work over time. Thus, the women in our study came of age during a time of marked social and economic transitions, but not all of these transitions led to increased access to steady full-time work. Significantly, most of these changes (increased labor force participation of college-educated women and increased nonmarital births) remain hallmarks of today’s labor force, suggesting that many of our findings may extend to more recent cohorts.

Our methods are an appropriate and innovative way to identify and describe the broad range of experiences that women have in the workforce across their prime childbearing and wage-earning years, as well as to describe the social conditions and individual circumstances early in life that make each of these pathways more or less likely. Unlike fixed-effects models, which would estimate individuals’ changes in work hours over time and across key transition points, group-based developmental trajectories are specifically aimed at identifying groups of individuals with shared experiences—and shared risk factors—over time. Indeed, a major strength of our approach is that heterogeneity among women is assumed and can be explicitly modeled in ways that are consistent with prior research and theory. Analytical approaches that focus on individuals (e.g., growth curve modeling) are limited to assuming a single intercept and slope of work hours over time and can allow women to vary only year by year from this single intercept and slope. For our study, the tradeoff is that we are limited in our ability to make claims related to individuals rather than groups. For example, group-based developmental trajectories cannot show how individuals alter their work hours in response to common life course events, such as divorce, the birth of an additional child, or the completion of additional education.

We note several limitations. First, data availability limits our analyses. The gender beliefs measures were collected when women are about 19–22 years old, and the NLSY79 does not collect regularly-spaced, time-varying data on gender beliefs. Because gender beliefs change as people age and experience key transition points (see Vespa 2009), we may also be underestimating the role of gender beliefs to the degree that they change over time and in relation to work or family experiences. Additionally, we may also be underestimating the importance of work–family opportunities and constraints, which are measured only once, at around age 25. This limitation keeps us from estimating the roles of gender beliefs or work–family constraint dynamically as time-varying variables, but it is a product of data availability and our modeling strategy. We cannot predict entry into workforce pathways without sacrificing the potential inclusion of explanatory variables that may act as mechanisms during the time the pathway data are collected. We are also unable to estimate the impact of changing social policies on women in our study. The passage of the Family and Medical Leave Act in 1993 and the Personal Responsibility and Work Opportunity Reconciliation Act of 1996 likely affected women in this sample by offering new opportunities for maternity leave and new work requirements. Yet, we cannot account for the different impact these policies may have had on women’s work hours.

Finally, endogeneity and the interrelatedness of our variables across theories of women’s workforce participation cannot be accounted for. It is quite likely that women who anticipate working full-time may exercise agency early in life about whom they marry and what level of education to pursue, and that these choices will differ from the choices made by women who anticipate a life of part-time or non-employment or women who unexpectedly experience an early pregnancy or other event that limits their ability to work or plan a future of paid work. Prior studies suggest that women’s expectations about work often do not match reality (Gerson 1985; Risman 1998). Given the gap between ideals and reality, it is likely that women may try to plan their lives according to their preferences surrounding work and family but may not be able to follow those plans to the extent that they anticipated. We cannot use these data to account for changing preferences of the women—or their spouses. However, we find ample evidence to support prior findings that such preferences are enacted within constraints, and our findings suggest that some women’s choices seem far more constrained than others.

Women’s workforce participation has increased dramatically since the 1960s, and women now enjoy a broad range of opportunities to earn a wage, continue education, delay marriage and childbearing, remain at home as a caregiver, or cycle in and out of work to meet the changing demands of families. However, as scholars have long articulated, women’s workforce decisions are not made freely based on personal preference for paid or unpaid work. Rather, women’s long-term pathways of work are structured by the early home environment and socioeconomic disadvantage, women’s own beliefs about how and whether they should work, and opportunities and constraints in work and family life. Moreover, work conditions vary significantly and likely shape women’s ability to remain employed; having access to paid sick leave, schedule control, and consistent total working hours are all hallmarks of “good jobs,” which may make continued workforce participation possible. Future research should continue to use longitudinal data to determine whether these patterns replicate for younger cohorts. Despite vast social and economic change, women in younger cohorts than the NLSY79 remain more responsible for childrearing, face barriers to continued employment, and have seen growing income inequality among women, which suggests that our findings will remain relevant for current and, perhaps, future generations. Our study provides evidence that women’s employment pathways are largely a product of their changing social environment and the resources available to them. Significantly, we point to social stratification and work–family constraints as key explanations for how women are tracked into work pathways from an early age, allowing some women to cumulate advantages through access to good and continued work while other women experience repeated workforce exits.

Acknowledgments

Both authors contributed equally to the article. We deeply appreciate the critical feedback from the Demography Editor and reviewer, and also the thoughtful comments on the revision from Gordon De Jong and Michelle Frisco. We also thank Jamie Lynch, Bobby Jones, Natasha Sarkisian, Kristen Schultz Lee, Heather Jacobson, Richard Petts, and Emily Greenman for their helpful comments and critiques on earlier drafts of this article. We acknowledge assistance provided by the Population Research Institute at Penn State University, which is supported by an infrastructure grant by the National Institutes of Health (2R24HD041025-11). This research was conducted with restricted access to Bureau of Labor Statistics (BLS) data. The views expressed here do not necessarily reflect the views of the BLS.

Appendix

Notes

2

The created Work History files have the same variables calculated “since the last interview,” but we use the calendar year variables to match interview year with year of age. Because of the low attrition in the NLSY, this excludes very little employment data.

3

Work hours are topcoded at 80+ hours per week to aid in model convergence. Less than 1 % of women worked more than 80 hours per week at a main job at each wave. Results do not change when work hours remain continuous, but some models do not achieve convergence.

4

Individuals are assigned to groups with varying probabilities of placement. As such, descriptive statistics by group are not precise unless they are weighted to adjust for each individual’s probability of correct placement (Nagin 2005:91). Thus, we do not provide descriptive statistics according to pathways of workforce participation, but they are available upon request.

5

To construct county-level unemployment rates during young adulthood, we use the Integrated Public Use Microdata Series (IPUMS) versions of the Current Population Survey from 1979–1984 (King et al. 2010) and historical Bureau of Labor Statistics (BLS) reports of employment status by state and county. We merge these data with the restricted NLSY79 Geocode data identifying respondents’ states and counties of residence at each interview to calculate variables for women’s labor market opportunities between ages 19 and 22.

6

Strully (2009) argued that job loss and unemployment should be measured as two distinct experiences.

References

References
Alon, S., & Haberfeld, Y. (
2007
).
Labor force attachment and the evolving wage gap between white, black, and Hispanic young women
.
Work and Occupations
,
34
,
369
398
. 10.1177/0730888407307247.
Bettie, J. (
2003
).
Women without class: Girls, race, and identity
.
Berkeley
:
University of California Press
.
Bianchi, S. M., Robinson, J. P., & Milkie, M. A. (
2006
).
Changing rhythms of American family life
.
New York, NY
:
Russell Sage Foundation
.
Blair-Loy, M. (
2003
).
Competing devotions: Career and family among women executives
.
Cambridge, MA
:
Harvard University Press
.
Boushey, H. (
2008
).
“Opting out?” The effect of children on women’s employment in the United States
.
Feminist Economics
,
14
(
1
),
1
36
. 10.1080/13545700701716672.
Bratter, J. L., & Damaske, S. (
2013
).
Poverty at a racial crossroads: Poverty among multiracial children of single mothers
.
Journal of Marriage and Family
,
75
,
486
502
. 10.1111/jomf.12012.
Cha, Y. (
2010
).
Reinforcing separate spheres
.
American Sociological Review
,
75
,
303
329
. 10.1177/0003122410365307.
Cha, Y. (
2013
).
Overwork and the persistence of gender segregation in occupations
.
Gender and Society
,
27
,
158
184
. 10.1177/0891243212470510.
Coleman, M. T., & Pencavel, J. (
1993
).
Trends in market work behavior of women since 1940
.
Industrial & Labor Relations Review
,
46
,
653
676
. 10.1177/001979399304600405.
Correll, S. J. (
2004
).
Constraints into preferences: Gender, status, and emerging career aspirations
.
American Sociological Review
,
69
,
93
113
. 10.1177/000312240406900106.
Damaske, S. (
2011
).
For the family? How class and gender shape women’s work
.
New York, NY
:
Oxford University Press
.
Dannefer, D. (
1987
).
Aging as intracohort differentiation: Accentuation, the Matthew effect, and the life course
.
Sociological Forum
,
2
,
211
236
. 10.1007/BF01124164.
Dannefer, D. (
2003
).
Cumulative advantage/disadvantage and the life course: Cross-fertilizing age and social science theory
.
Journals of Gerontology, Series B: Psychological Sciences and Social Sciences
,
58
,
S327
S337
.
Domina, T., & Roksa, J. (
2012
).
Should mom go back to school? Post-natal educational attainment and parenting practices
.
Social Science Research
,
41
,
695
708
. 10.1016/j.ssresearch.2011.12.002.
Elder, G. H., Jr (
1998
).
The life course as developmental theory
.
Child Development
,
69
,
1
12
. 10.1111/j.1467-8624.1998.tb06128.x.
England, P. (
2010
).
The gender revolution: Uneven and stalled
.
Gender and Society
,
24
,
149
166
. 10.1177/0891243210361475.
England, P., Ross, M., & Garcia-Beaulieu, C. (
2004
).
Women’s employment among blacks, white and three groups of Latinas: Do more privileged women have higher employment?
.
Gender and Society
,
18
,
494
509
. 10.1177/0891243204265632.
Folbre, N. (
2001
).
The invisible heart: Economics and family values
.
New York, NY
:
New Press
.
Frech, A., & Damaske, S. (
2012
).
The relationships between mothers’ work pathways and physical and mental health
.
Journal of Health and Social Behavior
,
53
,
396
412
. 10.1177/0022146512453929.
Gerson, K. (
1985
).
Hard choices: How women decide about work, career, and motherhood
.
Berkeley
:
University of California Press
.
Hibbard, J. H., & Pope, C. R. (
1993
).
Health effects of discontinuities in female employment and marital status
.
Social Science & Medicine
,
36
,
1099
1104
. 10.1016/0277-9536(93)90128-Q.
Hoffman, S. D., & Maynard, R. A. (
2008
).
The consequences of teenage childbearing on the mother and their spouse
.
Washington, DC
:
Urban Institute
.
Hostetler, A. J., Sweet, S., & Moen, P. (
2007
).
Gendered career paths: A life course perspective on returning to school
.
Sex Roles
,
56
(
1–2
),
85
103
. 10.1007/s11199-006-9150-8.
Hout, M., Levanon, A., & Cumberworth, E. (
2011
).
Job loss and unemployment
. In Grusky, D. B., Western, B., & Wimer, C. (Eds.),
The great recession
(pp.
59
81
).
New York, NY
:
Russell Sage Foundation
.
Hynes, K., & Clarkberg, M. (
2005
).
Women’s employment patterns during early parenthood: A group-based trajectory analysis
.
Journal of Marriage and Family
,
67
,
222
239
. 10.1111/j.0022-2445.2005.00017.x.
Jacobs, J. A. (
1989
).
Revolving doors: Sex segregation and women’s careers
.
Stanford, CA
:
Stanford University Press
.
Jones, B. L., & Nagin, D. S. (
2013
).
A note on a Stata plugin for estimating group-based trajectory models
.
Sociological Methods & Research
,
42
,
608
613
. 10.1177/0049124113503141.
King, M., Ruggles, S., Alexander, J. T., Flood, S., Genadek, K., Schroeder, M. B., . . . Vick, R. (
2010
).
Integrated Public Use Microdata Series, Current Population Survey: Version 3.0
[Machine-readable database].
Minneapolis
:
Minnesota Population Center [producer and distributor]
.
Klerman, J. A., & Leibowitz, A. (
1994
).
The work-employment distinction among new mothers
.
Journal of Human Resources
,
29
,
277
303
. 10.2307/146099.
Lerman, R., & Schmidt, S. (
1999
).
An overview of economic, social, and demographic trends: Introduction
.
Washington, DC
:
Urban Institute
.
McCall, L. (
2001
).
Complex inequality: Gender, class and race in the new economy
.
New York, NY
:
Routledge
.
McClendon, D., Kuo, JC-L, & Raley, R. K. (
2014
).
Opportunities to meet: Occupational education and marriage formation in young adulthood
.
Demography
,
51
,
1319
1344
. 10.1007/s13524-014-0313-x.
McLanahan, S., & Percheski, C. (
2008
).
Family structure and the reproduction of inequalities
.
Annual Review of Sociology
,
34
,
257
276
. 10.1146/annurev.soc.34.040507.134549.
Milgrom, E. M., & Petersen, T. (
2006
).
The glass ceiling in the United States and Sweden: Lessons from the family-friendly corner of the world, 1970 to 1990
. In Blau, F. D., Brinton, M. C., & Grusky, D. B. (Eds.),
The declining significance of gender?
(pp.
156
211
).
New York, NY
:
Russell Sage Foundation
.
Mirowsky, J. (
1999
).
Analyzing associations between mental health and social circumstances
. In Aneshensel, C. S., & Phelan, J. C. (Eds.),
Handbook of the sociology of mental health
(pp.
105
124
).
New York, NY
:
Springer
.
Moen, P. (
2001
).
The gendered life course
. In George, L., & Binstock, R. (Eds.),
Handbook of aging and the social sciences
(pp.
179
196
).
San Diego, CA
:
Academic Press
.
Moen, P., Dempster-McClain, D., & Williams, R. M., Jr (
1992
).
Successful aging: A life-course perspective on women’s multiple roles and health
.
American Journal of Sociology
,
97
,
1612
1638
. 10.1086/229941.
Moen, P., & Han, S-K (
2001
).
Gendered careers: A life-course perspective
. In Hertz, R., & Marshall, N. (Eds.),
Working families: The transformation of the American home
(pp.
42
57
).
Berkeley
:
University of California Press
.
Nagin, D. (
2005
).
Group-based modeling of development
.
Cambridge, MA
:
Harvard University Press
.
Nelson, M. K., & Smith, J. (
1999
).
Working hard and making do
.
Berkeley
:
University of California Press
.
O’Rand, A. M. (
2006
).
Stratification and the life course: Life course capital, life course risks, and social inequality
. In Binstock, R., & George, L. (Eds.),
Handbook of aging and the social sciences
(pp.
145
162
).
New York, NY
:
Academic Press
.
Pavalko, E. K., & Smith, B. (
1999
).
The rhythm of work: Health effects of women’s work dynamics
.
Social Forces
,
77
,
1141
1162
. 10.1093/sf/77.3.1141.
Percheski, C. (
2008
).
Opting out? Cohort differences in professional women’s employment rates from 1960 to 2005
.
American Sociological Review
,
73
,
497
517
. 10.1177/000312240807300307.
Reid, L. L., & Padavic, I. (
2005
).
Employment exits and the race gap in young women’s employment
.
Social Science Quarterly
,
86
,
1242
1260
. 10.1111/j.0038-4941.2005.00344.x.
Ridgeway, C. L., & Correll, S. J. (
2004
).
Unpacking the gender system: A theoretical perspective on gender beliefs and social relations
.
Gender and Society
,
18
,
510
531
. 10.1177/0891243204265269.
Risman, B. J. (
1998
).
Gender vertigo: American families in transition
.
New Haven, CT
:
Yale University Press
.
Shafer, E. F. (
2011
).
Wives’ relative wages, husbands’ paid work hours, and wives’ labor-force exit
.
Journal of Marriage and Family
,
73
,
250
263
. 10.1111/j.1741-3737.2010.00802.x.
Stone, P. (
2007
).
Opting out? Why women really quit careers and head home
.
Berkeley
:
University of California Press
.
Strully, K. W. (
2009
).
Job loss and health in the U.S. labor market
.
Demography
,
46
,
221
246
. 10.1353/dem.0.0050.
U.S. Bureau of Labor Statistics
. (
2009
).
Employment and earnings, 2009 annual averages and the monthly labor review
.
Washington, DC
:
U.S. Department of Labor
.
Vespa, J. (
2009
).
Gender ideology construction: A life course and intersectional approach
.
Gender and Society
,
23
,
363
387
. 10.1177/0891243209337507.
Von Hippel, P. T. (
2007
).
Regression with missing Ys: An improved strategy for analyzing multiply imputed data
.
Sociological Methodology
,
37
,
83
117
. 10.1111/j.1467-9531.2007.00180.x.
Webber, G., & Williams, C. (
2008
).
Mothers in “good” and “bad” part-time jobs: Different problems, same results
.
Gender and Society
,
22
,
752
777
. 10.1177/0891243208325698.
Williams, S., & Han, S-K (
2003
).
Career clocks: Forked roads
. In Moen, P. (Ed.),
It’s about time: Couples and careers
(pp.
80
97
).
Ithaca, NY
:
ILR Press
.
Willis, P. E. (
1977
).
Learning to labor: How working class kids get working class jobs
.
New York, NY
:
Columbia University Press
.
Willson, A. E., Shuey, K. M., & Elder, G. H., Jr (
2007
).
Cumulative advantage processes as mechanisms of inequality in life course health
.
American Journal of Sociology
,
112
,
1886
1924
. 10.1086/512712.