Family or Future in the Academy?

Institutional Response to the Pipeline Model: Inadequate at Best

Although tenure-track careers at research-intensive institutions continue to approximate the pipeline model, universities are paying some attention to the work–family concerns of their faculty. An important intervention has been the introduction of family-friendly policies (e.g., family leave after childbirth, a stop-the-tenure-clock policy, dual-career hiring) and institutional support in the form of child care centers (Friedman, Rimsky, & Johnson, 1996; Quinn, 2010; Quinn, Lange, & Olswang, 2004; Raabe, 1997). Despite a rising interest in more family-friendly environments (Friedman et al., 1996; Hollenshead, Sullivan, Gilia, August, & Hamilton, 2005), institutional commitment is inadequate. Even the institutions that acknowledge the need for and adopt some family-friendly policies (e.g., Research I universities) do not conduct periodic evaluations of their work–family policies or longitudinally measure the influence of family-friendly policies on tenure and promotion outcomes (Friedman et al., 1996; Hollenshead et al., 2005). ¹ Institutions document financial and organizational constraints as impediments to building truly family-friendly environments (Friedman et al., 1996). The institutional reluctance and fear of tenure denial at the department level (Lester, 2013; Mason & Goulden, 2004; Thompson, Beauvais, & Lyness, 1999; Williams, Alon, & Bornstein, 2006) is key to why early career academics (assistant professors) are reluctant to make use of work–family policies even when they are available.

Critique of the Pipeline as a Model

The pipeline approach suffers from several fundamental problems. First, the pipeline is linear and unable to account for varying career paths (Xie & Shauman, 2003). The pipeline is also not structured to accommodate young academics with significant responsibilities for caregiving or those from various fields, sectors, and stages (Hammonds & Subramaniam, 2003; Metcalf, 2010). Second, the pipeline model is problematic because it is based on supply-side economics that assume an increased supply of women doctorate holders will automatically result in an increase in the number of women professors, even though “women’s increasing representation in faculty labor supply vastly outstripped resulting representation among faculty” (Kulis, Sicotte, & Collins, 2002, p. 659).

Third, the pipeline model is troublesome because it shifts the focus away from the role of gatekeepers, and the inadequacy of the institutional response in leveling the playing field. It is possible that “fields and departments react to conditions of relative scarcity or abundance” in different ways (Kulis et al., 2002, p. 660). They can create conditions for success to accommodate a more diverse professoriate, or they can divert racial minorities and women into positions that are not on the tenure track (Kulis et al., 2002), which ensures financial savings for the institution. Clark and Corcoran (1986) argue that institutional responses to female academics on the tenure track can be best described by the Salieri effect, that is, that “gatekeepers to career advancement permit access but control achievement” (p. 39). Other scholars have taken a more critical view of the institutional response, noting that “women’s attrition from academia at best represents a leaky pipeline, and at worst, an occupation mired in institutional discrimination” (Marschke et al., 2007, p. 3).

Finally, in a pipeline model there is a “lack of focus on systemic change and power relations, particularly those that are raced, classed, sexed, and gendered” (Metcalf, 2010, p. 3). Some scholars (e.g., Han & Moen, 1999) have argued that a life-course perspective is a better alternative to the pipeline model. A life-course perspective focuses on the whole person, emphasizes connections between individuals (e.g., couples, families), and progress through various life stages. Because of lack of data and adoption of the model in practice, I have refrained from expounding on it in this article.

Literature on Women’s First Transition on the Pipeline

In this subsection, I address the first question, namely, What are the key findings from the literature on how family-related factors influence men’s and women’s first transition? Some of the early work examining the influence of family formation on men’s and women’s transition from doctoral degree recipient to first academic job has suggested a complex relationship between family-related factors (marriage and having children) and men’s and women’s entry onto the tenure track. One consistent finding across disciplines, however, is that women with young children are not welcome on the tenure track. Using cross-sectional data from the 1993 National Study of Postsecondary Faculty (NSOPF-93), Perna (2001) constructed a sample of humanities; social science; and science, technology, engineering, and mathematics (STEM) doctorate holders employed in academia. Using human capital and structural theory lenses to examine the data, she found that, at least in part, the predominance of women in part-time non–tenure-track positions can be attributed to marital and parental status. In contrast, married male faculty members and those with children benefit in employment status both by their marital and parental status. Perna (2001) also showed that sex differences in employment status (tenure track or not) continue to exist, that despite controlling for “differences in race, family responsibilities, human capital, and structural characteristics, women are more likely than men to hold full-time, non tenured positions, positions of lower status in the academic labor market hierarchy” (p. 603).

Although pioneering in many ways, the NSOPF is deficient in important respects. First, the cross-sectional nature of the NSOPF-93 data set made it difficult to determine causal direction between employment status and family responsibilities. Second, it is difficult to study a longitudinal process such as tenure and promotion process using cross-sectional data. Third, the NSOPF-93 data set lacked detail on potentially important dimensions of family formation. For instance, the NSOPF-93 data set did not include variables such as ages of dependent children, timing of childbearing, child care arrangements, employment status and occupation of spouse, income of spouse, amount of time spent out of the labor force because of family responsibilities, time devoted to household chores, time devoted to child care, and parent caregiving (Perna, 2001). Finally, although seminal, this study used data from over 20 years ago and there have been significant shifts in policy since then.

In a later article, Perna (2005) analyzed NSOPF-99 data to examine the influence of family ties on tenure and promotion. In her sample of doctorate holders from the humanities, social science, and STEM fields who were employed full-time at 4-year academic institutions, she measured family ties through parental status (number of dependents ranging between 0 and 4 or more). She also used a composite measure of marital status and employment status of partner or spouse (married or living in a marriage-like relationship with partner or spouse at the same institution, married or living in a marriage-like relationship with partner or spouse at another institution, married or living in a marriage-like relationship with partner or spouse outside higher education, separated, divorced, widowed, and single or never married). She found that although family ties continued to be related to better outcomes for men, they were unrelated to women’s employment outcomes.

As with other studies using cross-sectional data, an important limitation of Perna’s (2005) study was her inability to make causal claims. In fact, she acknowledged that her “study focuses on understanding relationships between family ties and the two outcomes at one point in time, rather than on drawing conclusions about causality” (Perna, 2005, p. 287). Moreover, she also noted that NSOPF-99 was missing several important spousal variables—number of years together, number of years spent with a spouse or partner at the same institution, spouse or partner’s pattern of labor market participation, educational attainment and academic discipline of spouse, current rank and tenure status of the spouse or partner’s pattern at the same institution. The data set also had limited variables related to family and family responsibilities (e.g., distribution of household and parental responsibilities; ages of dependent children; responsibilities for caregiving, including care of parents; history or changing nature of parental or family responsibilities, including time spent outside the labor force for childbearing and child rearing) or variables related to departmental, institutional, and national networks. It is important to collect data on such variables because such factors shift the nature of support that institutions can provide to enable all of their faculty to be successful.

To address some of the challenges encountered by scholars using cross-sectional data, such as their inability to establish causal direction between employment and family outcomes, Ginther and Kahn (2006) used event history analysis and analyzed longitudinal survey of doctorate recipients (SDR) data from 1973 to 2001. They offered evidence consistent with Perna’s (2001) work, concluding that “women are less likely to take tenure track positions in science, but the gender gap is entirely explained by fertility decisions” (Ginther & Kahn, 2006, p. 2). They added that “children make it less likely that women in science will advance up the academic ladder beyond their early post-doctorate years, while both marriage and children increase men’s likelihood of advancing” (Ginther & Kahn, 2006, p. 2). They also highlighted that in their sample of scientists and engineers there was no difference in tenure attainment and promotion to full professor after controlling for productivity, employer, demographic, and family variables. In addition, in many cases there was no gender gap in tenure attainment or promotion to full professor even without controlling for these covariates. They did acknowledge, however, that family variables influenced promotion probabilities for men and women differently.

A comprehensive report (Bentley & Wise, 2004) commissioned by the National Science Foundation (also using SDR) found evidence contradictory to (at least in part) Ginther and Kahn’s (2006) work. The report showed that both marriage and having children hurt women’s future among the tenured ranks at each important juncture (transition to tenure-track position and promotion to associate and then full professor) and not just at the point of entry. The apparent contradiction between the two reports is likely attributable to Ginther and Kahn (2006) defining science as life sciences, physical sciences, and engineering, whereas Bentley and Wise (2004) defined science as including the social sciences. It is also important to note that although longitudinal designs are better suited for studying tenure and promotion (Perna, 2001, 2005) and are more amenable for drawing causal claims, they do not take selection into account. Moreover, the predominant reliance of these models on human capital characteristics (e.g., cumulative publication count, years since doctorate, kind of graduate support during doctoral years, time to degree, postgraduate plans, prestige of doctoral-granting institution) can result in omitted variable bias, thereby limiting the identification of important personal and professional variables that can help explain the variation in women’s decision making about work and family.

Wolfinger et al. (2008) analyzed SDR (1981–1995). Their sample included humanities, social science, and STEM doctorate holders. They used discrete time-event history analysis, which allowed them to disentangle the effects of marital and parental status on women’s transition into tenure-track roles and beyond. Their article showed that single women without children fare better than men in obtaining tenure-track positions. They also found that marriage and the “presence of young children have strong and independent negative effects on the likelihood that women obtain ladder-rank positions” (Wolfinger et al., 2008, p. 398). Once on the tenure track, however, female faculty members were not further penalized for marital or parental responsibilities. The authors also found that older children (between 6 and 18 years) increased the probability of both men and women obtaining tenure-track positions.

The finding that having young children is associated with women’s greater leakage from the pipeline than having older children is consistent with Budig’s (2003) findings. In particular, it corroborated Budig’s finding that pregnancy does not influence fully employed women’s hazard of exiting from the labor market. The presence of preschoolers does, however, increase the likelihood of exit from the workforce. In contrast, having older children increases the likelihood of either part-time or full-time reentry of previously nonemployed mothers in the labor market. Researchers (e.g., van Anders, 2004) have also shown that some women plan their careers and childbearing jointly.

Although important in many regards, the Wolfinger et al. (2008) study is limited in important ways. For instance, they did not account for type of marriage (e.g., spouse’s educational level and whether partner was also an academic or not). This matters because women tend to be married to equally or better educated spouses at higher rates than men. Women also tend to give preference to a male partner’s career, which constrains women’s careers in serious ways (Kulis & Sicotte, 2002).

Some scholars (e.g., Long, 2001) have pointed out that gender inequalities in academia have narrowed over time. Much previous research, however, was focused on women’s career outcomes on the tenure track for a cohort of doctorate recipients from the early 1980s to early 1990s. Morrison et al. (2011) responded to this gap by examining SDR-2006 data for a cohort of humanities and social science PhD holders from the late 1990s, to find out whether women still suffered a marriage and parenting penalty. They also asked a related question: Do women’s careers continue to be constrained due to their spouses’ educational attainment levels? They found that unlike some previous findings, there was no marriage penalty for women. Marriage was no longer negatively associated with entering tenure-track roles, earning tenure, or being promoted to full professor. Although there was no marriage penalty for women, there was still a marriage boon for men. Married men or men with partners without professional degrees were earning tenure faster (than other men) and at the most prestigious institutions in the United States. No such benefits were realized for the women in the sample.

Literature on the Second Transition on the Pipeline

In this subsection, I address the second question, namely, What are the key findings from the literature on how family-related factors shape men’s and women’s second transition? Scholars writing on the relationship between family-related factors and women’s second transition (i.e., earning tenure and subsequently being promoted to full professor) have reported certain findings clearly and consistently. First, tenured female faculty members are more likely to be single than similarly qualified males. For instance, in one study, women were almost three times as likely as their male counterparts to be single and without children (e.g., National Science Foundation, 2010). Another study reported that only 51% of women were married and had children compared with 73% of their male colleagues (Goulden et al., 2011). Second, female faculty reported having fewer children than they desired (Wolfinger et al., 2008). Third, divorce rates among tenured female faculty tended to be almost 50% higher than for similarly situated males (Goulden et al., 2011). Although the aforementioned findings are clear, several studies present mixed findings with regard to the association between family-related responsibilities and earning tenure and subsequent promotion to full professor. In an attempt to resolve the apparent contradiction, scholars have drawn on numerous data sets and used a wide array of research techniques, but they have failed to provide a consistent explanation regarding the influence of family formation and women’s career outcomes.

Two bodies of literature are particularly relevant regarding the study of women’s scholarly journeys on the tenure track. One group reported on the influence of family formation on faculty-research productivity (e.g., Fox, 2005; Hunter & Leahey, 2010; Long, Allison, & McGinnis, 1993; Sax, Hagedorn, Arrendondo, & Dicrisi, 2002) and the other on the influence of family-related factors on tenure and promotion (e.g., Bentley & Wise, 2004; Ginther & Kahn, 2006; Morrison et al., 2011; Wolfinger et al., 2008). A close examination of faculty-research productivity is important because scholarly productivity can help explain as much as half the variation in tenure attainment and promotion between men and women (Long et al., 1993). Other scholars have argued that research productivity “operates both as a cause and effect of status in science” (Fox, 2005, p. 131). Unless one understands how various factors such as family formation contribute to scholarly productivity, it will be very difficult to assess or rectify inequity of rank, promotion, or salary (Fox, 2005). Understanding the influence of family formation on research productivity is akin to studying the indirect effects of family factors on tenure and promotion. An equally important body of literature examines the influence of family-related factors on tenure and promotion directly.

Literature on Institutional Work–Family Policies and Practice

In this subsection, I address the third question, namely, What is the state of the literature on the prevalence and use (or underuse) of work–family policies in U.S. colleges and universities? Despite a 2001 call for action from the American Association of University Professors, comprehensive macro and micro accounts of existing work–family policies have not been conducted. I found only two comprehensive accounts of work–family policies in colleges and universities in the United States (Friedman et al., 1996; Hollenshead et al., 2005). Research on work–family policies is an important first step for moving in a direction in which “the priorities, workloads, rewards structure, and values of the academy” are aligned with the whole person (American Association of University Professors, 2001, p. 220). Such research can guide successful integration of work and family both in principle and in practice. Finally, I want to note that even though studies have reviewed work–family policies across various types of colleges and universities in the United States, I primarily focus on findings that are directly related to research-intensive institutions.

Limited Use of Explicit Theoretical Frameworks

Many scholars use human capital and structural theories to frame their studies. Yet these theories are limited in important ways. I propose an alternative—an opportunity-cost framework—to address some of those limitations. Studies using longitudinal data draw on human capital characteristics (to a large extent) and structural perspectives (to a lesser extent). Part of the heavy reliance on human capital measures is due to lack of other measures in nationally representative data sets. However, given the omitted variable bias, it is logical that researchers should call for a broader range of data to be collected (e.g., greater number of structural variables, especially related to departmental and institutional work–family measures; child care arrangement measures; personal and professional network measures) and draw on new and integrative theoretical frameworks (e.g., an opportunity-cost framework).

Human capital theory and structural theory, although valuable, are unable to fully explain the lower proportion of women among the tenured ranks. A framework that enables researchers to understand how individual faculty choice interacts with structural aspects of higher education institutions is desirable. Opportunity-cost is one such framework. Opportunity costs could be forgone earnings, lost time, pleasure, or any benefit that is of value to an individual (Brealey, 2012). In the case of tenure-track faculty members, choosing a career in research and teaching over other available occupations means losing the rewards offered by those other occupations: “Individuals whose opportunity costs outweigh the rewards gained from teaching will be more likely to leave” the profession (Guarino et al., 2006, p. 176).

At the same time, it is important to examine how individual circumstances (e.g., higher financial benefits in industry vs. the academy) interact with structural aspects of universities and colleges (e.g., availability of an on-campus child care center). Moreover, opportunity costs can vary based on a faculty member’s discipline, personal and professional values, family situation, and other labor market opportunities available to the individual as well as career stage. Finally, ease of entry, monetary rewards, working conditions, health care benefits, child care benefits, a family-friendly environment, and overall job satisfaction can be manipulated by administrators and policy makers to attract/retain high-quality employees (Guarino et al., 2006).

Examining the career decision making of women who have a doctorate and a young family with an opportunity-cost framework should be beneficial, because it would allow scholars to capture the richness of individual decision making. These educated women have various choices available to them, including staying at home, pursuing a non–tenure-track role in academia, pursuing a tenure-track role, or pursuing a career in industry or government. The opportunity costs may differ across academic fields, however, and within and across employment sectors. For instance, some disciplines such as science and engineering have very close connections with industry; industry offers better pay (D’Este & Perkmann, 2011; Reis, 2012), and there are relatively permeable boundaries between academia and industry. Such permeable boundaries between academia and industry may not exist in other disciplines, such as the humanities or social sciences.

For individuals focused on research and teaching, the opportunities in humanities or some social sciences may constrain them to academia, and they may choose to pursue either tenure-track or non–tenure-track roles within academia. The opportunity-cost framework should be extremely useful in understanding individual decision making across academic fields, across employment sectors, and over time. Using the opportunity-cost framework to understand how individual choices interact with structural characteristics of the university is critical for deepening an understanding of women’s decision making around tenure-track careers.

Although the opportunity-cost framework is one critical way to analyze women’s decision making, there are also other powerful alternative theoretical frameworks (e.g., feminist perspectives, critical race perspectives). Alternate frameworks question, for example, why a higher proportion of women of color who are academics do not have children or why progress has been so slow in spite of attempts at policy implementation and in spite of the progress women have made. In addition, alternate frameworks also raise fundamental questions about how power, privilege, and ideal worker norms tend to dominate the narrative on gender. Until there is substantive change in how people think about gender itself, work–family issues will remain largely the domain of women and marginalized by institutions.

Limited Set of Family-Related Variables

The majority of articles examining the influence of marriage and family formation on women’s career prospects in academia rely heavily on rigorous quantitative methods and nationally representative data sets. Scholars (e.g., Perna, 2003) have pointed out that though nationally representative data sets have important advantages, such as large sample sizes and fairly high response rates, they are not without major limitations. For instance, despite the finding that women are most likely to leak out of the tenure pipeline soon after receiving a PhD, there is a dearth of national longitudinal data on predoctoral experiences. Many national data sets are rich in human capital measures but lacking in other important measures, such as those that relate to structural features. Moreover, several nationally representative data sets are also lacking in measures of institutional support (e.g., presence and utilization of family-friendly policies, presence and utilization of child care centers, policies concerning dual-career couples).

Another important limitation of national data sets such as the NSOPF-93 and NSOPF-99 is that they only include doctorate holders who are employed within academia. Examining a population of doctorate holders employed in academia limits researchers from understanding selection biases mentioned in some studies. Also, there is no more recent NSOPF data. In contrast, some national data sets such as SDR do include data on career trajectories of doctorate holders beyond academia. Such data have not been fully used by scholars. Not examining career trajectories of those who leave academia is problematic, given that career decision making can look widely different across academic fields and at different career points.

Lack of Attention to Multicultural and Diversity Issues

There are few articles that examine gender, racial, and ethnic influences when assessing the association between family and tenure and promotion outcomes. This is primarily because the data set is predominantly White and middle-class and there is a small number of faculty of color in national samples. Once a researcher splits the sample by discipline, gender, and faculty rank, a much smaller proportion of faculty members from certain racial and ethnic groups remain. Alternately, the resulting sample is too small for detailed analysis. There is a need to oversample faculty members from various racial and ethnic groups, as it is likely that both gender and racial/ethnic background will provide an important context for understanding how family-related factors shape tenure and promotion outcomes.

Lack of Department-Level Data in Comprehensive Accounts

Although the work–family literature I have reviewed provided some data on the availability and utilization of work–family policies by faculty, authors failed to collect department-level data at higher education institutions. This is an important gap in the literature because in order for faculty to use work–family policies, availability of policies is a necessary but not a sufficient condition (Ward & Wolf-Wendel, 2007). A clear signal at the departmental level (e.g., supportive work environment, support of the chair or dean) that using work–family policies will not result in any professional penalization (e.g., tenure denial) is necessary for faculty to make use of family-friendly policies. Collecting data at the departmental level can be problematic, however, because departments within a university may vary widely in terms of norms and expectations of faculty. Historically, institutions have preferred to assist faculty with work–family challenges on a case-by-case basis (Friedman et al., 1996).

Even in institutions where work–family policies formally exist (i.e., there is a written policy document), administrators implement work–family policies to varying degrees (Bird, 2011; Drago & Colbeck, 2003; Friedman et al., 1996; Quinn et al., 2004; Raabe, 1997). Depending on the type of departmental culture, faculty response to policies varies as well (Bird, 2011; Drago et al., 2001; Friedman et al., 1996; Quinn, 2010; Sallee, 2013). A negative or hostile departmental environment is a major deterrent to work–family policy use among faculty, who are afraid of appearing to be not sufficiently committed or professional (Drago & Colbeck, 2003; Quinn, 2010).

It is important to collect data at the departmental level because sometimes department chairs and faculty are not even aware that various work–family policies exist (Bird, 2011; Quinn et al., 2004; Waltman & August, 2005). If faculty and chairs are not aware of some policies, they certainly cannot promote their use or use the policies themselves. Sometimes communication about work–family policies is inconsistent. For instance, Quinn et al. (2004) found in their interviews with University of Washington faculty members and departmental chairs that the availability of a part-time tenure track was not widely known. In some cases in which respondents were aware of the existence of a policy, the details were not well understood. There were also instances in which work–family policy use was considered a special accommodation and not a legitimate career choice. Unless there is a systematic way of collecting reliable data about the awareness and attitudes of deans, chairs, and faculty toward work–family policies, departmental cultures are unlikely to become more family friendly.

Lack of Standardization of Work–Family Policy Terms

Standardization of work–family policy terms is important so that terminology is consistent across studies making comparisons across various institutions possible, which would better enable scholars and policymakers to follow the evolution of the work–family policy debate. Some articles highlighted lack of a standardized definition of work–family policy terms across institutions (Friedman et al., 1996; Hollenshead et al., 2005; Quinn, 2010). For instance, tenure-clock-stopping policy may be referred to as “tenure clock extension,” “stopping the tenure clock,” “waived years,” or “tenure rollback” (Quinn, 2010). Similarly, Hollenshead et al. (2005) pointed out that leave for female faculty members recovering from childbirth can have many different names and can be defined in different ways. Policies may have the same name but very different meanings or eligibility requirements across campuses, making it difficult to compare institutions. Having standard names and consistent definitions of policies is essential for clarity on the prevalence, use, and influence of work–family policies on men’s and women’s career outcomes in the academy.

Lack of Longitudinal Data

Although some studies reviewed here have included longitudinal data sets (e.g., Ward & Wolf-Wendel, 2012), scholars have also pointed out that there is no systematic institutional effort to gather data related to work–family policy over a period of time across different types of higher education institutions (Friedman et al., 1996; Hollenshead et al., 2005; Quinn, 2010). For instance, Friedman et al. (1996) showed that only 26% of leadership campuses conducted an evaluation of their work–family policies and programs, and only 2% of campuses that were not leadership campuses conducted such an evaluation. Gathering longitudinal data and conducting evaluations is essential for understanding whether or not policies are working.

Selection Effects

A majority of the quantitative studies I examined suffer from the problem of “selection bias,” commonly defined as “an error in choosing the individuals or groups to take part in a study” (National Cancer Institute, National Institutes of Health, n.d.). Selection effects can operate in a number of ways. For instance, selection biases may come into play when participants self-select into doctoral programs across various academic disciplines, employment sectors, and type of job within a sector (e.g., tenure-track academic jobs, non–tenure-track academic jobs). For example, Morrison et al. (2011) pointed out that female faculty members in their samples may be individuals who prioritize a career in the academy over all else or only those females who are highly secure in their careers decide to have children. Other women with doctorates may prioritize family formation over academic careers and opt out of academia. The authors suggest that it is difficult to say whether the observed work–family pattern results from a life-course trajectory already in motion prior to tenure-track roles when parenting could have mattered or from women adjusting their fertility during their time on the tenure-track or from selection effects in which only women who felt highly secure in their careers chose to have children and therefore to rise on the academic ladder.

Selection biases may also come into play when researchers who want to examine links between family-related factors and career outcomes include only doctorate holders employed in academia. Even in cases where data about nonacademic career trajectories were available (e.g., SDR), a majority of researchers chose to exclude such data from study samples, thereby eliminating doctorate recipients who have left an academic position (possibly because of failure to attain tenure or even after being promoted). Moreover, selection bias is problematic because the doctoral recipients included in the analyses were nonrandomly assigned to the sample. In addition, presence of selection bias tends to distort the statistical analyses, and unless the reader is extremely cautious in drawing conclusions from the analyses, the interpretation is likely to be misleading. Given that literature discussed in this review is already full of contradictory findings, the notion that researchers cannot even assert that the findings are based on the correct sample undermines the attempt to understand the influence of family formation on women’s career trajectories on the tenure track.