Abstract
In this article, we examine trends in women’s mobility among male-dominated, gender-neutral, and female-dominated occupations. Earlier research, largely employing data from the 1970s and early 1980s, showed that along with significant net movement by women into male-dominated fields, there was also substantial attrition from male-dominated occupations. Here, we build on previous research by examining how “gender-type” mobility rates have changed in recent decades. The findings indicate that while still quite high, levels of women’s occupational mobility among female, gender-neutral, and male occupations have decreased considerably over time. We suggest that this is the result of increasing differentiation among women. In particular, many women, especially those in high-status occupations, plan to pursue employment in a male-dominated field, succeed in gaining entry, and tend to remain in these fields more often than their counterparts in previous decades. We interpret these findings as evidence that gender segregation is maintained by an enduring but imperfect system of social control that constrains women’s choices before, during, and after entry into the labor market. The evidence presented here underscores the importance of studying gender-type mobility as a distinct dimension of labor market inequality.
Keywords
Occupational gender segregation is a system of inequality in which working in female-dominated occupations is associated with lower pay and promotion opportunities (Cohen and Huffman 2003; Glass 1990; Levanon, England, and Allison 2009; Mandel 2012). After remaining essentially fixed until 1970, gender segregation in the United States declined through the 1970s and 1980s and has restabilized since the mid-1990s (Blau, Brummund, and Liu 2013; Cohen 2013; Hegewisch et al. 2010; Levanon and Grusky 2016; Stainback and Tomaskovic-Devey 2012).
Sociologists (along with other social scientists) have long recognized that mobility and immobility are fundamental features of any system of inequality. That is, systems of economic inequality are understood not simply in terms of the levels of inequality but also in terms of how often individual people circulate among various strata. Blau and Duncan (1967) developed this idea in studies of socioeconomic stratification, and Duncan (1984) and Chetty et al. (2017) did the same for economic inequality and mobility across economic strata. Occupational segregation, like all systems of inequality, can be defined in part by the degree of openness or mobility among individuals employed within this system. In this study, we focus on the extent of movement among male-dominated, gender-neutral, and female-dominated occupations. We refer to such moves as “gender-type mobility.”
Prior research on career trajectories has documented the movement of workers over the course of their time in the labor market. In these studies, careers are often conceived of as steady movements from entry-level positions to higher levels of responsibility, authority, and earnings (Kalleberg and Mouw 2018). However, the experiences of many workers do not fit this idealized trajectory. Gender-type mobility patterns are especially at odds with this conventional view of career paths. For example, women employed in male-dominated occupations such as skilled trades, restaurant chefs, and computer coders often leave these jobs (Bilginsoy 2003; Harris and Guiffre 2015; Hegewisch and O’Farrell 2015; National Science Foundation 2019), despite the short-term and long-term cost of such moves. These exits, in turn, reinforce gender segregation (Moore 1988; Torre 2017). Indeed, policy analyses of specific male-dominated fields often cite retention along with recruitment of female employees as a consistent challenge and focus of reform (Hegewisch and Hartmann 2014). At the other end of the gender spectrum are many female-dominated fields with high levels of turnover, such as teachers, restaurant servers, salesclerks, and home health aides (Carver-Thomas and Darling-Hammond 2017; Espinoza 2017; Gandhi, Yu, and Grabowski 2021; Sikora 2021; U.S. Bureau of Labor Statistics 2021). Although these positions often have limited opportunities for promotion, some women nonetheless find their way from these jobs into gender-neutral and even male-dominated fields. In short, occupational segregation is not a static phenomenon that is fixed in place at the point of hiring. Rather, there is considerable movement across gender-type boundaries that warrants careful investigation.
In this article, we build on the existing literature on occupational gender segregation in three ways. First, we examine levels of intragenerational mobility across gender-type boundaries over the period 1970–2018. This long-term view will enable us to see whether or not gender occupational mobility patterns move in step with levels of segregation. We also document the extent of change net of compositional factors over time—that is, changing levels of education and changes in family composition.
In a second contribution, we study the intersectionality of gender and class over the last four decades. Several studies have noted that the decline in gender segregation has been sharper among white-collar positions than blue-collar ones (Bergmann 2011; England 2010; Mandel 2012; Torre 2019). Therefore, we examine how this point of class differentiation intersects with women’s mobility patterns over time.
Finally, we conduct a set of cohort-specific analyses. As new cohorts of women increasingly enter the labor market, they may approach their lives and their futures in a distinctive way. As a result, they might follow different career trajectories than their predecessors. Here, we study how the paths of earlier and later cohorts of women workers have changed over time.
Our research draws on the March Supplement of the Current Population Survey (CPS) data set to examine women’s occupational mobility at six time points between 1970 and 2018. Whereas the CPS data are typically used for cross-sectional analyses, we employ a special set of retrospective questions included in the CPS’s March waves that makes it possible to analyze intragenerational occupational mobility. We estimate pooled logistic regression models with year fixed effects to account for changes in the probability of occupational transitions among male-dominated, gender-neutral, and female-dominated occupations varying over time.
Our analyses make theoretical and empirical contributions to the existing literature. Theoretically, we view mobility as a defining feature of any system of inequality. A labor market with a given level of segregation and no occupational mobility is very different from a similarly segregated labor market that allows for considerable movement between male-dominated and female-dominated fields. Examining mobility across gender-type boundaries does not resolve the debate over the role of supply- and demand-side factors in occupational segregation; rather, focusing on mobility broadens this debate by including experiences and behaviors that occur after the initial hiring decisions have been made.
Concentrating on mobility patterns can point the way to policy strategies for reducing occupational segregation. For example, if segregation were attributable to women self-selecting into female-dominated occupations, then policy makers should turn their attention to gender-type socialization and entry barriers. If, however, the evidence points to women exiting male-dominated occupations at high rates, then reform would need to include tactics designed to reduce this attrition. In other words, increasing retention and reducing attrition in male-dominated fields become central to the policy agenda in this area. Our research also demonstrates the need to create pathways for women in female-dominated fields to move (or advance) into gender-neutral and male-dominated fields. Finally, by examining class-based differences, this study helps to explain the persistence of gender segregation in the most disadvantaged and precarious occupations.
The mobility patterns we document here are important in multiple respects, first and foremost because gender-type mobility continues to be remarkably common. Although our analysis indicates a significant decline of women’s mobility among male-dominated, gender-neutral, and female-dominated occupations, it remains far higher than is generally assumed. Second, results are consistent with a growing degree of differentiation among women. On one hand, a distinct group of women has succeeded in gaining entry to, and remaining in, higher-status male-dominated occupations. On the other hand, women with fewer labor market credentials and who work in less prestigious occupations continue to be concentrated in female-dominated fields; despite high turnover rates, their career mobility patterns have changed less over time than those of women in high-status professions. However, our analyses of the most recent and earlier cohorts of women yield few notable differences in terms of mobility.
We interpret the findings reported in this article as consistent with an evolving pattern of social control. This framework expands the focus on supply and demand factors into the period after women have entered the labor market. We suggest that the some of the same factors that have contributed to a decline in gender segregation have paradoxically resulted in a decline in gender-type mobility.
The Reproduction of Occupational Gender Segregation: Static Versus Dynamic Explanations
Demand- and Supply-Side Explanations
Numerous researchers have attempted to apportion the explanation of gender segregation between supply-side and demand-side factors. The former includes gendered socialization and choice (Antecol and Cobb-Clark 2013; Barone and Assirelli 2020; Cech 2013; Goldin 2006; Thébaud and Taylor 2021), whereas the latter stresses barriers to opportunities for women (Cha 2013; Heilman and Eagly 2008). Much of the recent empirical research on gender segregation continues to focus explicitly or implicitly on women’s choices. For example, Cech (2016) examines whether women’s decisions to pursue employment in STEM (science, technology, engineering, mathematics) fields reflect their future plans to have children and a family. Also, Levanon and Grusky (2016) emphasize “gender essentialism” as an enduring dimension of gender segregation, yet detailed studies of occupational choice (Barone and Assirelli 2020; Ochsenfeld 2016) find limited evidence of this essentialism in women’s career plans and choice of college majors. Others, however, continue to view gender segregation as a constraint on women’s occupational choices (Hegewisch and Hartmann 2014).
The main polarity of research oscillates thus between issues of labor market supply on one hand and labor market discrimination on the other. We should note, though, that studies of the labor force as a whole typically lack measures of many key variables, such as workplace culture, sexual harassment on the job, and the extent of support for “nontraditional” occupational choices on the part of friends and family. Consequently, while many studies point to the limited role of supply-side factors, they are often not able to completely discount the possibility that unmeasured factors may play a role. Perhaps a more fundamental limitation is that this debate is constrained by its static focus. Analysts typically focus on the assortment of women and men at a single point in time rather than viewing gender segregation as a dynamic process.
One prominent exception is a recent study by Guinea-Martin, Mora, and Ruiz-Castillo (2018), who conceptualize segregation from a life-course perspective. Their more dynamic framework differs from the majority of research in this area (see also Vespa 2009; Zhou 2017). Empirically, however, their research examines cross-sectional data and does not explicitly delve into questions of mobility across gender-type boundaries.
The Mobility Paradox
It is natural to assume that high levels of gender segregation must make mobility between male-dominated and female-dominated occupations difficult, if not impossible. A starkly segregated occupational landscape must be built upon formidable barriers to entry. Thus, if most women work in female-dominated jobs, one might expect women to move from one female-type job to another, and career mobility into male-dominated fields would be rare. Similarly, men who change occupations would be most likely to move into another male-dominated job. Yet, against these expectations, evidence from multiple large-scale data sets covering a variety of pre-labor market, educational, and labor market stages have documented remarkably high levels of mobility across gender-type boundaries in the 1970s and 1980s (Jacobs 1989). Among those workers who changed occupations, the connection between the gender type of origin field and destination field was surprisingly weak. 1 Moreover, gender-type movement was not restricted to mobility across occupations: The same general pattern was evident with respect to occupational aspirations, as well as to movement across typically male- or female-dominated college majors. Later, Levine and Zimmerman (1995) reexamined this issue and argued that the extent of women’s gender-type mobility had actually been understated by previous studies (see also Polavieja and Platt 2014; Riegle-Crumb, King, and Moore 2016). In particular, they reported only the slightest connection between gender-type aspirations and subsequent career outcomes. Moreover, they found little change in the overall association across gender-type jobs between the 1968 cohort and the 1979 cohort of young women. However, in some of their analyses, Levine and Zimmerman (1995) suggest that there might be an increased tendency for women in the later cohort to aspire to, and succeed in, a career in a male-dominated occupation.
This pattern seems paradoxical: How can high segregation coexist with high rates of mobility? Why do high rates of mobility not “whittle away” at the edifice of segregation and ultimately undermine this enduring structure? The answer is that gender segregation persists because of the substantial attrition of women from male-dominated fields.
The exodus of women from typical male jobs has been interpreted from the perspective of enduring social control. Gender segregation not only rests on gender socialization and educational experiences but also reflects structural and workplace factors that endure after women have entered the labor market (Jacobs 1989). The notion of enduring social control is not fundamentally different from a perspective that focuses on issues of supply and demand. The key difference is that unlike the more static applications of the supply and demand framework, the social control perspective views the allocation and assortment of women into female-dominated fields as an extended process that continues to operate via structures and pressures even after women and men have entered the labor market (see also Gershon 2019). Indeed, efforts to promote women in male-dominated fields often emphasize retention and recruitment (for example, see Hegewisch and O’Farrell 2015). Social control emphasizes the importance of retention processes—along with pre-market preferences, educational investments, and employer decisions.
Piecing the Story Together
To sum up, both supply-side and demand-side factors no doubt contribute to occupational segregation, but neither employee choice nor employer discrimination is a perfectly efficient sorting process. There is slippage along the way, and the cumulative extent of such “slippage” is considerable. After all, even the most sophisticated individual-level analyses of gender inequality, conducted with the most extensive set of control variables, can explain only a modest portion of the observed variance in labor market outcomes. People change their minds, they stumble into unexpected opportunities, or they may find their dream workplace is located in an inhospitable setting. Therefore, we argue that gender-type mobility warrants examination as a distinct dimension of occupational gender segregation.
There has been relatively little recent research on the career dynamics associated with gender segregation. Apart from works on attrition from STEM fields known as the “leaky pipeline” (Alper 1993; Berryman 1983; Glass et al. 2013; Shauman 2017; Xie and Killewald 2012; Xie and Shauman 2003) and other selected studies (Torre 2014, 2017), few scholars have focused on the link between women’s gender-type mobility and occupational gender segregation since the 1980s.
The questions we pose here focus on trends over time in gender-type mobility. The 1970s to 1980s, when the pioneering studies in this area were undertaken, was a period of rapid change for gender segregation and for women’s employment experiences in general. Gender segregation levels subsequently declined and then stabilized, prompting us to ask whether the mobility patterns documented during the 1970s and 1980s were unique to that period and how they may have evolved since that time. Consequently, we examine how the level of mobility between male-dominated, gender-neutral, and female-dominated occupations has changed over time (1970–1971 through 2017–2018), and how these patterns vary across different strata of the labor force. By covering the period 1970–2018, the analysis enables us to compare periods of high gender segregation (1970s), declining levels (1980s), and moderate-but-stable levels of gender segregation (1990–present).
Research Questions
Our study is organized around three research questions. First, has the extent of gender-type mobility—that is, among male-dominated, gender-neutral, and female-dominated occupations—increased or decreased since the 1970s? Second, does the pattern of gender-type mobility vary by social strata? More specifically, are any observed changes concentrated among those with higher educational levels and those in professional and managerial positions? Finally, do observed trends differ across cohorts? The rationale for investigating each of these questions is developed next.
Research Question 1: Trends in Gender-Type Mobility
We examine the extent to which gender-type mobility has changed during the period 1970–2018. Concretely, we document the extent to which this mobility has increased or declined over that period. One logical possibility is that declines in gender segregation coincide with increased movement between male-dominated and female-dominated occupations. If gender has become less salient as a defining feature of occupations, then it stands to reason that gender composition might also become less salient. Thus, barriers to gender-type mobility could be expected to ease as the level of gender segregation in the labor market declined during the 1970s and 1980s. Similarly, as occupational gender segregation stabilized during more recent decades, the level of gender-type mobility would also stabilize.
A competing scenario would be that mobility between male-dominated and female-dominated occupations declines coinciding with declining gender segregation levels. This pattern could result from two factors: on one hand, the increased commitment of a segment of women to pursue careers in male-dominated fields; on the other hand, less pressure to exit such fields. Women could face less discrimination from co-workers or management and could benefit from more flexible workplace arrangements and family-friendly workplace policies. In other words, it is possible that the very trends that contribute to women’s increased representation in male-dominated fields also result in lower mobility from these occupations, contributing to a lower overall rate of gender-type mobility.
While the rate of occupational change is not the central focus of this research, it is an issue that must be addressed to make sense of gender-based patterns of mobility. Analyses of gender-type mobility between occupations are meaningful only for those who change occupations. Although in any given year only a minority of individuals change occupations, the cumulative effect is a substantial degree of occupational shifts over the course of one’s time in the labor market. In a sense, this pattern creates the potential or “room” for individuals to shift between male-dominated and female-dominated occupations. A variety of studies have highlighted increased instability in the workplace and declines in job security (e.g., Kalleberg 2013), especially for workers in their 20s and 30s. Taken together, these empirical studies point to increased mobility in recent decades (Jarvis and Song 2017). We run supplementary analyses to examine these trends in order to provide the context in which gender-type mobility is being examined.
Research Question 2: Variation Across Occupational Strata
In addition to investigating trends in mobility patterns over time, we also explore variation in trends across occupational strata. Recent research has shown increasing differences among women in higher status occupations (i.e., managerial and professional occupations) and lower status occupations (i.e., service, clerical, and blue-collar occupations). These studies demonstrate that the upward occupational mobility among women is attributable largely to the relative success of women in managerial and professional occupations compared with those in less prestigious occupations (Bergmann 2011; England 2010, 2011; Mandel 2012; Xie, Killewald, and Near 2016). Furthermore, women in high-status occupations have been found to be more likely to enter male-dominated occupations than are women in low-status occupations (Bergmann 2011; England 2010, 2011; Torre 2019). There are also several reasons to expect that attrition rates for women in high-status male-dominated fields would be lower than for their counterparts in low-status occupations. First, high-status (and better educated) workers have generally adopted more gender-egalitarian attitudes than do low-status workers (Charles and Grusky 2004; Cotter, Hermsen, and Vanneman 2011). Second, women in high-status occupations draw on significantly more resources than low-status workers when it comes to balancing work and family life (Pettit and Hook 2009). Unlike managerial and professional women, low-status workers are not always covered by the Family and Medical Leave Act 2 and frequently experience difficulties in taking unpaid leave from work or paying for private childcare (Gerstel and McGonagle 1999; Pettit and Hook 2009). Taking all these factors into consideration, we examine whether women in high-status occupations might experience different mobility patterns than women employed in low-status occupations.
In addition, we conduct supplementary analyses on the gender-type mobility of men. The comparison of both women’s and men’s mobility patterns over time will allow us to discern whether observed changes over time are related to gender or to occupational positions.
Research Question 3: Cohort-Specific Trends
Finally, we examine the extent to which changes in mobility patterns can be explained by the unique experiences of new cohorts of women entering the labor market. Women are increasingly training for and entering male-dominated occupations whose career trajectories tend to project them into other male-dominated occupations. Changes in gender segregation, then, might be driven by changes in the careers for which women train and hence their entry occupations. These first occupations, which are increasingly male-dominated, channel women into careers that continue to unfold largely within male-dominated occupations. This phenomenon might be particularly relevant in the case of high-status women because women trained in male-dominated higher education classrooms are better equipped than women in low-status occupations to manage workplace gender challenges (Bergmann 2011; Torre 2017). We ask, therefore, whether the observed trends are different for the youngest age cohorts of women. To address this point, we compare the career trajectories of young women (below 35 years old) with older women (above 35 years old). This analysis enables us to examine potential differences in mobility between the earlier and later cohorts.
Data and Methods
Data
Our main data source is the CPS, a nationally representative monthly sample survey of 60,000 eligible households conducted by the U.S. Census Bureau for the Bureau of Labor Statistics. More specifically, we employ the Annual Social and Economic Supplement (ASEC) of the CPS, also known as the March CPS Supplement. In addition to relevant sociodemographic characteristics on individuals, the March Supplement includes information on each respondent’s longest held job during the previous year and data on their current job, permitting analysis of occupational mobility over a 1-year period. 3 It provides detailed (three-digit) occupational data codes, making it possible to capture occupational shifts that would be missed if only a limited set of broadly defined occupations were measured. In this article, we employed pooled data from six different time points, namely, 1971, 1981, 1991, 2001, 2011, and 2018. These 1-year retrospective data cover the occupational transitions in 1970–1971, 1980–1981, 1990–1991, 2000–2001, 2010–2011, and 2017–2018. Appended to the March CPS Supplement is the gender composition of three-digit Census occupations for 1970, 1980, 1990, 2000, and 2010. Following Blau, Brummund, and Liu (2013), we use U.S. Census data to examine the level of occupational segregation over time because it provides large sample sizes and reports occupations in the most detailed categories available (around 500). Changes between the 2000 and 2010 Census categories were relatively minor, whereas the changes between the 1990 and 2000 categories were more substantial. All the data were reclassified based on 1990 occupational categories to determine whether any of the changes observed might be attributable to changes in occupational coding rather than changes in women’s behavior. The results did not differ substantially using the recoded data. Agricultural occupations are excluded from the analysis, and only salaried employees (both full- and part-time) are considered.
Dependent Variables
The analyses in this article examine the trends of women’s mobility among male-dominated, gender-neutral, and female-dominated occupations between 1970 and 2018. First, we classify occupations as male-dominated when female presence in the occupation is below 33.3 percent and as female-dominated if women’s representation is 66.6 percent or above; all other occupations are considered gender-neutral. We choose three equal-frequency categories to make the results comparable with previous findings (Jacobs 1989; Sheridan 1997; Torre 2014). However, the results remain robust to the use of alternative classification schemes (i.e., defining “gender-neutral” within a narrower band, 45–55 percent, and defining a strong female-dominated occupation as one with a 75 percent female presence). Next, dependent variables are defined based on the gender type of both the origin and the occupation of destination. Nine transition types are examined, that is, from each of the three origin categories (male-dominated occupations, gender-neutral occupations, and female-dominated occupations) to each of these destination categories.
Supplementary analyses examine the overall rate of occupational change from 1970 to 2018, as well as men’s mobility. The dependent variable occupational change is coded as 1 when the person changes occupations from year t to t + 1, and 0 otherwise. Finally, we replicate the analysis of mobility across gender-type boundaries for men.
Covariates
Models include both relevant sociodemographic and work-related covariates. Sociodemographic variables include controls for age, education, marital status, parental status, and race/ethnicity. Education is measured as a dummy variable for educational attainment (1 if college or more, 0 otherwise). Marital status scores 1 when the individual is married and 0 otherwise, and parental status captures the presence of children living in the household (1 if there are children younger than 18 years in the household, 0 otherwise). Finally, the variable ethnic group distinguishes among white, Black, and other ethnic backgrounds. Work-related covariates include controls for weekly hours worked before occupational change (1 if full-time, 0 otherwise). Occupations are classified as high status (managers and professionals) and low status (service, clerical, and blue-collar occupations). Finally, the main interest of this study lies in the variable time. This indicator is represented by six dummy variables, one for each of the six waves of data analyzed. This approach treats each time point discretely to capture the nonlinear time trends that we observe for gender segregation.
Table 1 summarizes the main descriptive statistics for the variables included in the analysis. Values for women in each of the six time points covered by the study are included, allowing us to assess the extent to which women’s occupational distributions have changed across years. In addition, this table displays the proportion of women working in male-dominated, gender-neutral, and female-dominated occupations.
Descriptive Statistics: Mean (Standard Deviation)
Analytical Strategy
The empirical portion of the article is divided into two sections. First, we use pooled logistic regressions with year fixed effects to estimate the probability of transitioning among gender-type occupations. Next, to observe whether segregation is driven by changes among the youngest age cohort, we split the sample between women below 35 years old and women above 35 years old. All the analyses are conducted separately for high-status workers and low-status workers.
We have also conducted supplementary analyses that address two issues related to this article’s main theme. First, we examine the overall rate of occupational change and assess whether observed trends are attributable to changes in the composition of the labor force. The probability of occupational change (Pi) is related to the covariate vectors by a year fixed effects logistic regression equation. Second, we replicate the analysis of women’s gender-type mobility for men.
Results
Trends in Gender-Type Mobility
Occupational segregation has declined significantly over the last 50 years. In 1970, nearly 70 percent of women would have had to change occupations in order to be distributed in the same manner as men, as measured by the Index of Dissimilarity (D) (Duncan and Duncan 1955). 4 This index fell 20 percentage points between 1970 and 2020 (D = 69 in 1970; D = 49 in 2010), with the steepest decrease occurring between 1970 and 1990. After 1990, occupational segregation has remained rather stable, as several researchers have documented (Cohen 2013; Hegewisch et al. 2010). In addition, segregation has remained significantly higher in low-status occupations (D = 57 in 2018) than in high-status occupations (D = 43).
We now turn to the question of how gender-type mobility patterns have changed given the initial decline and subsequent stability in the level of gender segregation from 1970 onward. We estimate the probability of transition among gender-type occupations using pooled logistic regressions with year fixed effects. The results for the time trends for each of the nine transitions are presented in Table 2. These changes are net of any observed trends in educational, marital status, and other demographic covariates listed in Table 1. The full tables, including the coefficients for all the explanatory variables, are given in the Online Appendix (see Table A1). The results clearly show a significant decline in mobility rates while also revealing that mobility levels continue to be surprisingly high.
Logistic Regression of Women’s Gender-Type Mobility by Occupational Group Year Coefficients
Note: Values are odds ratios, with p values in parentheses. Year coefficients are controlled by age, education, children in household, marital status, ethnic group, and hours worked.
p < .05. **p < .01. ***p < .001.
Although the year coefficients reported in Table 2 do not constitute a linear trend, it is nonetheless clear that mobility rates in recent decades are lower than they were in the 1970s and 1980s. To help visualize the results, Figure 1 charts the predicted probability of transitioning between male- and female-dominated occupations over time (the predicted probability of the nine transitions is plotted in Figure A1 in the Online Appendix). The results corroborate that gender-type mobility has tended to decline over time.
Women’s Probability (Pr) of Transitioning Between Male- and Female-Dominated Occupations, 1970–2018
Figure 1 documents a remarkable degree of mobility across gender-type boundaries, albeit at lower levels than observed during the 1970s. Focusing on the results for the 2017–2018 transition, we see that for women occupation changers who start out in female-dominated fields, only about half (55.6 percent) move to another female-dominated occupation, with the balance moving to gender-neutral (34.5 percent) and male-dominated (9.9 percent) fields. For women starting out in male-dominated occupations, more than two thirds of those who change occupations (69.2 percent) move out of this domain, ending up either in gender-neutral (43.4 percent) or female-dominated (25.8 percent) fields. These rates of mobility, though lower than they were in previous decades, remain exceptionally high. Consequently, any complete account of gender segregation must take this substantial, and persistent, mobility among male-dominated, gender-neutral, and female-dominated occupations into account.
Turning to the historical trajectories, the percentage of women moving from a male-dominated occupation to a female-dominated field has decreased from 53.2 percent in the 1970s to 25.8 percent in 2018, as shown in the upper-right corner plot of Figure 1. The decreasing degree of mobility from male-dominated to female-dominated occupations has been accompanied by a notable increase of the probability of moving to occupations with a similar gender composition. In recent years, for example, career mobility among women pursuing jobs in male-dominated fields most often involves moving to other male-dominated occupations rather than switching to female-dominated fields (upper-left plot). More specifically, the probability of changing occupations within the male-dominated field rises from 17.6 percent in 1970 to 29.8 percent at the end of the observation period.
It is important to note that the patterns of gender-type mobility documented here are not driven by motherhood (see Table A1 in the Online Appendix). Unlike previous studies following a life-course approach, we find no systematic pattern of mothers fleeing male-dominated occupations and childless women fleeing female-dominated occupations. Indeed, for eight of the nine occupational moves examined, the coefficient on children in the household is not statistically significant.
A complementary approach is to examine the association between the gender composition of origin and destination occupations, treating these as a continuum from 0 percent female to 100 percent female. We calculated the Pearson correlation between the percent female at time t − 1 and the percent female in the occupation at time t. This correlation is .10 or less during the 1970s and early 1980s but rises over .20 during the 1990s and 2000s. It remains higher for high-status workers for the whole period.
Differences by Occupational Status and Cohort
The results in Figure 1 also differentiate between the mobility patterns of high-status and low-status workers. A number of distinctive patterns are evident. First, attrition rates for women in male-dominated fields continue to be remarkably higher for women in low-status occupations (29.1 percent) than for women in high-status occupations (15.8), although both rates have dropped substantially over the years. Second, the percentage of women moving from one male-dominated high-status occupation to another (upper-left corner plot) rises from 20.6 percent in 1970 to 33.9 percent at the end of the observation period. Women in low-status occupations are less likely to move to other male-dominated fields, although there has been a mobility increase for this group as well (rising from 16.7 to 27.9 percent).
As for the occupational shifts of women in female-dominated occupations, the results indicate that women entered traditionally male jobs at higher rates in the 1980s (14.8 percent), and then with less frequency after 1990 (decreasing nearly to the initial 1970 level of 9.9 percent). This trend is consistent with the prior literature (e.g., Shaw 1983), which suggested that during the 1970s and 1980s, women actually had access to a much broader set of opportunities than they had thought possible when growing up, thereby increasing gender-type mobility during that period.
The bottom-right corner plot indicates a slow decrease of moves within the female sector, from about 71.3 percent in the 1970s to 64.6 percent in 2000 (coinciding with substantial exits into male fields), followed by a significant decrease after 2000 to 55.5 percent. This pattern is even clearer once we split the sample by occupation levels. As we observe, women in high-status occupations showed a decline in mobility from the 1970s to the 1990s, followed by a modest recovery. Mobility rates for women in low-status female-dominated occupations remained rather stable during the 1970s, 1980s, and 1990s. That could suggest that opportunities in male-dominated occupations were more prevalent for professional women, whereas women in lower-status occupations continued to have fewer opportunities to move into male-dominated fields (Bergmann 2011; Torre 2019). Also striking, the probability of women remaining in gender-neutral jobs was more than 20 percentage points higher in 2018 (48.5 percent) than it was in 1970 (23.5 percent).
We examined the possibility that these results differ by cohort, given that the observed trends could be driven by changes in the career types for which young women train. As new cohorts of women enter the labor market, they may have different values, experiences, and goals than earlier generations of women workers. We, therefore, ask the following: Are the trends documented here attributable solely to the behavior of the youngest cohorts of women? Figure 2 displays the probability of entering and exiting male-dominated occupations for women younger than 35 years and compares it with women older than 35. The full tables are given in Tables A2 and A3 in the Online Appendix.
Women’s Probability (Pr) of Entering and Exiting a Male-Dominated Occupation, 1970–2018
Not surprisingly, young women are more likely to have a job in a male-dominated field in 2018 than in previous years. However, a large number of young women continue to exit the male-dominated fields when they change occupations. Indeed, there is a similar time trend for women younger and older than 35. This suggests that changes in women’s mobility between 1971 and 2018 are not simply a case of the unique experiences of new cohorts of women entering the labor market. In other words, younger and older women alike continue to exit male-dominated fields in considerable numbers, and both the levels and the trends are not limited to the experiences of younger women.
Supplemental Analyses
Probability of Changing Occupations
We also examine whether career mobility rates increased or decreased between 1970 and 2018. The rate of occupational mobility sets the upper bound for movement between male-dominated and female-dominated occupations, so understanding these patterns helps to establish the context for the previous analysis.
To address this question, Figure A2 in the Online Appendix displays the probability of changing occupations over time after adjusting for the demographic attributes of the female labor force. The data indicate that the rate of occupational change was higher in 2018 than in 1971 for women similar in terms of age, education, marital status, presence of children in the household, race, ethnicity, and full-time employment status. The full regression models are provided in Table A4 in the Online Appendix.
Although our analysis focuses on occupational mobility over a short (1-year) time period, the rate of occupational change is nonetheless quite substantial: At least 9.0 percent of women change occupations within a 1-year period. Moreover, the trend data also indicate a general increase in the rate of occupational mobility between 1970–1971 and 2017–2018. A significant dip occurred following the 2008 recession, and the rate of occupational mobility has since rebounded almost to the levels observed during the 1990s and 2000s. Overall, the rate of occupational mobility for women increased from 9.0 percent per year in 1971 to 19.9 percent in 2018. This finding suggests that there is greater potential for gender-type mobility in recent decades than in the 1970s and 1980s.
Furthermore, the data indicate that women in high-status occupations switch fields less often than women in low-status occupations. This finding is consistent with the results of earlier research (Donohue 1988; Royalty 1998). The trend toward higher rates of occupational change, however, is evident for women in both high- and low-status occupations. Specifically, the rate of occupational mobility for women employed in managerial and professional occupations increases from 7.1 to 16.3 percent between 1971 and 2018; the corresponding figures are 9.6 and 21.0 percent for women working in clerical, service, or blue-collar occupations.
To summarize, the decline in gender mobility cannot be attributed to a decline in overall occupational mobility patterns. On the contrary, gender-type mobility has declined even though mobility has increased.
Men’s Gender-Type Mobility
We explored the trends in gender-type mobility for men as well to provide further insights into these occupational processes. The evidence suggests that the patterns of gender-type mobility documented here are starkly different for men and women (see Figure A3 in the Online Appendix). Whereas for women, moving between female-dominated occupations is most common, men most often move from one male-dominated occupation to another. 5 This finding reflects the gender-segregated nature of labor markets. Moreover, the chances of transitioning between male-dominated fields have increased over time for women but have remained steady for men. Moreover, the probability of men transitioning between female-dominated occupations increased for men but not for women. The details of these patterns suggest changes in how both men and women view, invest in, and commit to careers in nontraditional occupations (i.e., those that defy traditional gender expectations). In other words, these patterns do not inhere in the occupations themselves but in the choices, attitudes, and experiences of the women and men entering them. The time trend reflects declining mobility for both men and women, but these changes are located in different occupational groups.
Conclusion
The main findings of this article can be summarized as follows. First, the data indicate that occupational mobility among female-dominated, gender-neutral, and male-dominated occupations has declined significantly over time. Despite these substantial declines, though, gender-type mobility remains far higher than is generally understood. Among occupation changers, a substantial majority of women who start out in male-dominated fields do not move to another male-dominated field but rather to a gender-neutral or a female-dominated occupation. Movement out of female-dominated fields is also quite substantial. Second, we clearly see that women in high-status male-dominated occupations are increasingly persisting in these fields. The data reveal a consistent and increasing differentiation among women in high- and low-status occupations. Finally, the results indicate that changes over time are not restricted to the distinctive experiences of the most recent cohorts entering the labor market.
The findings are consistent with a combination of supply- and demand-side factors that we describe with the term enduring social control, which emphasizes the constraints that women face in pursuing a full set of occupational choices that occur before, during, and after the point of labor market entry. This framework acknowledges individual agency as well because the mobility of individuals within the system is a central focus. Declining social control is consistent with declining occupation segregation if women begin to see entering male-dominated occupations as a less daunting prospect and if their experiences in such fields become more positive. Lowering barriers to entry could result in more investment in occupation-specific skills, thus expanding the pool of women prepared to pursue male-dominated fields. Declining social control is also consistent with less pressure for women to exit male-dominated fields. In other words, gender-type mobility may have declined in part because a minority of women have pursued careers exclusively in male-dominated fields and increasingly succeeded in their attempts.
The overall pattern can be summarized as follows. During periods of high gender segregation, few women plan careers in male-dominated fields, and those who do make such plans often face unexpected obstacles, ranging from a male-dominated workplace culture and active harassment to inadequate structural supports, which can in turn result in disproportionate exits from these fields. At the same time, the existence of both male-dominated and female-dominated fields for workers with similar educational levels enables a relatively modest number of women to “stumble into” male-dominated fields. The 1970s and 1980s were representative of this pattern, with one major exception: The flow of women streaming into male-dominated fields temporarily overwhelmed the number of women leaving these fields, thus reducing the degree of occupational gender segregation. By the 1990s, women who entered male-dominated fields early in their careers were more likely to persist in their choice, even when changing occupations. Consequently, the trajectories of women in male-dominated fields were increasingly distinct from those of women in female-dominated occupations. A lower “equilibrium” level of gender segregation was reached, accompanied by a lower level of mobility between male-dominated and female-dominated fields.
We think it is important to emphasize that gender-type mobility continues to be higher than commonly thought. About 45 percent of women who change occupations, exit from male-dominated to gender-neutral or to female-dominated occupations. While this is a lower rate than was observed during the 1970s and 1980s, it remains a very high rate of attrition that contributes to reproducing segregation levels. Attrition from male-dominated occupations remains substantial even among the youngest cohort, even if these women are more prepared to deal with the gender challenges of male-dominated workplaces.
Furthermore, these findings indicate that mobility patterns vary across different strata of the labor force and across different historical periods. Although all women may now theoretically be able to plan professional careers with more confidence and commitment, a detailed analysis offers evidence that the level of gender-type mobility clearly depends on professional background. Results are in line with recent research revealing an increasing polarization among women in the labor market (Bergmann 2011; England 2010; Torre 2019). On one hand, female professionals and managers have improved their professional trajectories and broadened their career opportunities. Elite education has been central to women’s ability to access high-status occupations, unlock the door to traditionally male fields, and maintain employment in these positions. As we have seen, this group of women displays higher rates of gender-type occupational mobility (probably due to having more resources at their disposal to care for children and handle other domestic tasks without needing to leave the labor market). Over time, their career trajectories have tended to move in the direction of convergence with men’s, although full convergence remains a distant goal. On the other hand, women who fill service and blue-collar jobs do not have the same opportunities. These women often hold part-time jobs, and without support for childcare, they are more vulnerable to rigidity in the labor market. In addition, low-status women who enter male-dominated fields are more likely to reconsider their decision. Indeed, results indicate that their probability of moving from a male- to a female-dominated occupation continues to be significantly higher than for women in high-status occupations. Moreover, when changing jobs, they are more likely to move to gender-neutral occupations rather than to other male-dominated ones.
Altogether, the findings here point to the importance of gender-type mobility as a distinct dimension of gender inequality. Distinguishing between mobility trends and other indicators of opportunity for women in the labor market is important because not all trends move in the same direction. Consequently, gender-type mobility needs to be included among the key indicators of trends in women’s employment experiences. The evidence clearly shows that gender-type mobility does not follow the same historical patterns as does overall gender segregation, and thus it warrants separate attention as a distinct dimension of gender inequality.
In addition to the theoretical importance of these findings, the results presented here also have important policy implications. They underscore the continuing attrition of women from male-dominated fields. Issues such as sexual harassment in the workplace, promoting more female-friendly workplace cultures, and instituting more family-friendly workplace policies may prove fruitful in reducing women’s attrition from male-dominated fields. The data also underscore the need to devote more attention to building pathways for women into male-dominated fields. These nonstandard pathways appear to be even less available to women in recent years than they were in the 1970s and 1980s. Given the high level of turnover in many female-dominated fields, there is a pool of women who could benefit economically from moving to gender-neutral and female-dominated occupations.
Increasing pay in culturally undervalued female-dominated occupations remains an important objective. There may be some progress in this area as a result of efforts to increase the minimum wage and the Biden administration’s focus on investing in the “infrastructure” of care work, much of which is done by low-paid women. These factors could produce uncertain and even contradictory effects on mobility patterns, but they would improve the economic position of women and make the gender segregation of occupations less costly to women and to society in general.
A number of lines of inquiry for future research remain open. Thus far, we have examined the intersection of gender segregation and class (as indexed by occupational level). Intersectionality also extends to race in some but not all dimensions. Women of color are more likely to be concentrated in lower status occupations. Moreover, they are more likely than white women to change occupations. This is evident in the analyses presented in the Online Appendix Table A4. In additional analyses (not shown), we found that the racial gap is even greater before controlling for educational differences. The patterns of gender-type mobility, however, appear to be driven more by gender and occupational level than by race and ethnicity. We were not able to document clear and consistent patterns for women of color that differed from the overall patterns documented here. In some cases, this may be attributable to small sample sizes, such as the few women of color in high-status male-dominated occupations, especially during the earlier periods included in this study.
Unfortunately, available data do not allow us to fully identify all of the specific mechanisms underlying gender-type mobility. Thus, we are not able to pinpoint how much of the change observed can be traced to declining workplace harassment and discrimination; to the lowering of structural barriers for women brought about by access to family and medical leave; or to shifting career investments and commitments. Future studies should look at these issues in more detail. Also, these studies should analyze data on multiple occupational moves over a longer period of women’s careers, examine data within individual firms and particular career contexts, and link mobility patterns with data on other important career measures such as work–family support mechanisms and the availability of parental and sick leave.
A final point concerns gender segregation within ostensibly gender-neutral occupational categories. A number of researchers (e.g., Stainback and Tomaskovic-Devey 2012) have shown that gender segregation at the level of job title can be substantial even within detailed (three-digit) census occupations. The implication for this article is that highly gender-type occupations have little measurement error in terms of the actual gendered employment experience, but gender-neutral occupations may be underestimating male-to-male and female- to-female career transitions. Further research on career patterns in this set of occupations is warranted.
Supplemental Material
sj-docx-1-gas-10.1177_08912432211046328 – Supplemental material for The Gender Mobility Paradox: Gender Segregation and Women’s Mobility Across Gender-Type Boundaries, 1970–2018
Supplemental material, sj-docx-1-gas-10.1177_08912432211046328 for The Gender Mobility Paradox: Gender Segregation and Women’s Mobility Across Gender-Type Boundaries, 1970–2018 by Margarita Torre and Jerry A. Jacobs in Gender & Society
Footnotes
Authors’ Note:
Margarita Torre acknowledges support from the Spanish Ministry of Science and Innovation through the research grant RTI2018-098182-A-I00. Thanks go to Barbara J. Risman and Irma Mooi-Reci for their assistance, and to Maria Charles, Daniel Oesch, Paula England, Kathleen Gerson, Alexandra Killewald, Pilar Gonalons-Pons, and Juan Díez Medrano for their very helpful comments on the previous drafts of this article.
Notes
Margarita Torre is associate professor of sociology at the University Carlos III of Madrid and a fellow at the Carlos III-Juan March Institute (IC3JM). Her research centers on gender and work, with a focus on the mechanisms of exclusion encountered by both men and women seeking nontraditional jobs. Her current projects examine the evolution of gender inequality in scientific collaborations across countries and disciplines, and the performance of gender in social media.
Jerry A. Jacobs is professor of sociology at the University of Pennsylvania. He has written about many aspects of women’s careers, including gender gaps in earnings, authority and time use, and gender inequality in higher education. His six books include The Time Divide: Work, Family and Gender Inequality (2004) with Kathleen Gerson and The Changing Face of Medicine: Women Doctors and the Evolution of Health Care in America (2008) with Ann Boulis. His current projects include research on the future of work and a study of technology and the care work needs of older adults.
References
Supplementary Material
Please find the following supplemental material available below.
For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.
For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.
