In recent years, the absolute number of women earning degrees across science, engineering, technology, mathematics, and medicine (STEMM) fields has increased compared to men. For example, between 2004 and 2014, 2,924,660 women earned bachelor’s degrees in science and engineering compared to 2,890,904 men. Despite this increase, women—especially women of color—are underrepresented relative to their presence in the workforce and the U.S. population (see Figures 1-1 and 1-2) (NSF, 2017). The disparities in number and proportional representation vary by discipline and field (see Figure 1-3), yet, even in professions in which women are at parity or overrepresented, as is the case in certain sub-disciplines within biology and medicine, there remains a dearth of women among the senior ranks in these fields (see Figure 1-4).

FIGURE 1-1

Percentages of the workforce that are men and women by occupational group. SOURCE: U.S. Census Bureau, 2012–2016 American Community Survey (ACS) 5-Year Estimates, available at https://www.census.gov/content/dam/Census/library/working-papers/2019/demo/sehsd-wp2018-27.pdf. (more...)

FIGURE 1-2

Percentage of women in science and engineering occupations, with a bachelor’s degree and above, by discipline. SOURCE: National Science Foundation.

FIGURE 1-3

Percentage of women in science and engineering occupations, with a bachelor’s degree and above, by discipline. SOURCE: National Science Foundation.

FIGURE 1-4

Representation disparity between men and women across STEM disciplines. Relative to men, women are underrepresented at all career stages (bachelor’s, doctorate, postdoctorate, and professor) across nearly every STEM discipline. Representation (more...)

In theory, this underrepresentation of women in senior leadership roles should diminish organically over time, as the number of women earning degrees and entering the workforce increases, but past patterns indicate that time alone may be insufficient to close existing gaps. In medicine, for example, women have for the last quarter-century comprised at least 40 percent of U.S. medical students, yet, as of 2018, women accounted for only 18 percent of hospital chief executive officers and 16 percent of medical school deans and department chairs (Bickel, 2004; Holman et al., 2018; Mangurian et al., 2018). Based on an analysis of 15 years of publication patterns of 36 million authors from more than 100 countries in more than 6,000 journals, Holman et al. (2018) predicted that, absent reforms in education and publishing, gender gaps in certain STEMM specialties may persist for decades (Holman et al., 2018).

The gender gaps that have characterized most U.S. STEMM fields for the past 50 years merit attention because such gaps exact both explicit and opportunity costs for the nation’s scientific enterprise. Multiple components of STEMM fields demonstrably benefit from gender diversity. Using citation analysis to assess impact, for example, Smith-Doerr et al. (2017) demonstrated that gender-heterogeneous problem-solving teams produce more influential scientific papers than do single-gender teams, and more diverse teams generate more innovative solutions to problems (Díaz-García et al., 2012; Page, 2019; Smith-Doerr et al., 2017). The quality of peer review (Murray et al., 2018) and productivity of collaborative science increase with gender diversity as well (Woolley et al., 2010). Moreover, in the 2011 National Research Council rankings of doctoral programs, gender diversity is positively associated with rank (as is racial diversity) (Henderson and Herring, 2013).

Long-standing gender stereotypes that attribute inequities to innate biological differences in aptitude or ability are not reinforced by an abundance of evidence (Barres, 2006; Bian et al., 2017b). Rather, no significant biological differences between the performance of men and women in science and mathematics have been found that can account for the lower representation of women in these fields (NASEM, 2007; Riegle-Crumb et al., 2012). The bulk of evidence indicates instead that underrepresentation of women in STEMM—including at leadership levels—is driven by a wide range of structural, cultural, and institutional patterns of bias, discrimination, and inequity that do not affect men of comparable ability and training (Cortina et al., 2013; Milkman et al., 2015; Moss-Racusin et al., 2012; Rodrigues, In Review).

The underrepresentation of women in STEMM shares many features with underrepresentation of other groups in STEMM, including men of color, LGBTQIA individuals, persons with disabilities, first-generation college students, and the socioeconomically challenged, in that the current culture and structure of STEMM systemically disadvantage members of these groups relative to White and Asian-American males (Cuddy et al., 2007; Dixon and Rosenbaum, 2004; Dovidio et al., 1986; Fazio et al., 1995; Fiske, 2010; Fiske et al., 2002; Gaertner and McLaughlin, 1983; Kay and Jost, 2003; NSF, 2018). Notwithstanding moral, ethical, and justice arguments in favor of equitable participation in STEMM, the lack of diversity in many STEMM fields has consequences for the productivity and long-term sustainability of the enterprise. Specifically, there is a national labor shortage of STEMM professionals in certain disciplines (e.g., computer science)² that cannot be addressed by continuing to rely on the contributions of groups that are currently overrepresented in the STEMM workforce (i.e., White and Asian American men) (see Figure 1-3) (PCAST, 2012).

In terms of identifying institutional changes that can reverse entrenched patterns of underrepresentation of women in STEMM fields, insights can be gained from examining policies and practices implemented by institutions that have in fact succeeded in narrowing gender gaps. Two early examples are Harvey Mudd College in Claremont, California, and Carnegie Mellon University in Pittsburgh, Pennsylvania: Against a pattern of steadily declining numbers of women earning bachelor’s degrees in computer science, these schools substantially increased the number of women graduates in this field (Fisher et al., 1997). At Harvey Mudd College, half of students in 2016 with undergraduate degrees in computer science, engineering, and physics were women compared with 18 percent of computer science graduates nationwide (Weisul, 2017). Similarly, in the same year, nearly 50 percent of incoming classes at Carnegie Mellon University’s School of Computer Science and College of Engineering were women (CMU, 2016) (see Box 1-1).

BOX 1-1

Success Stories in Computer Science and Engineering.

Similarly, at the University of Michigan in Ann Arbor, sustained institutional support for a range of interventions developed through the National Science Foundation ADVANCE³ program led to an increase in the percentage of women hired (as a proportion of all new faculty hired) from 13 percent in 2003 to 31 percent in 2016 (see Box 1-2).

BOX 1-2

Example: University of Michigan STRIDE Program: Strengthening Recruitment and Hiring Practices for Increasing Diversity.

Success stories at research intensive universities such as Carnegie Melon, Harvey Mudd, and University of Michigan offer valuable lessons learned; however, it is important to acknowledge that the vast majority of students in the United States, including women students, are post-traditional⁴ students pursuing education at other kinds of institutions. Post-traditional students tend to be older, live off campus, have children and jobs, and earn their degrees over longer time frames. Only 26 percent of students today fit the “traditional” profile of the student who enrolls in college or university full time in the fall after high school graduation, lives on campus, does not work while enrolled in school, and completes a bachelor’s degree in 4 years (Brown, 2017) (see Box 1-3 for additional discussion of post-traditional students).

BOX 1-3

Efforts by Community Colleges to Increase Representation of Women in STEMM.

Community colleges are often well prepared to serve the needs of post-traditional students (NASEM, 2016) and many are taking an active role in increasing the number of women in STEMM. Women make up the majority of community college students (56 percent) (Horn et al., 2006) and research has indicated that community colleges can be a good training ground for women interested in entering STEMM fields (AAUW, 2013). In addition, racial and ethnic minorities are more likely to be enrolled at community colleges than in 4-year institutions (Horn et al., 2006; NASEM, 2016). This is particularly true for Hispanic students. In 2016, nearly half of all Hispanic students were enrolled in a community college, compared to 30 percent of White students (Pew Research Center, 2016). Therefore, community colleges are a critical pathway to advance women, particularly women of color, in STEMM. See Box 1-3 for several examples of efforts by community colleges to increase representation of women in STEMM.

In addition to community colleges, minority serving institutions (MSIs), which have the most diverse student bodies in the nation, are another group of institutions that are well-equipped to prepare post-traditional students for careers in STEMM (NASEM, 2019a). The 2019 National Academies report on MSIs found that a slightly higher percentage of undergraduate students are enrolled in STEM fields at 4-year MSIs than at 4-year non-MSIs (NASEM, 2019a). Historically Black Colleges and Universities (HBCUs) are well known for producing a large number of African American scientists (NASEM, 2019a), many of whom are women. Between 1995 and 2004, 46 percent of Black women who earned STEM degrees received their degree from an HBCU (Arroyo, 2009). Additionally, among Black women students who earned doctorates in science and engineering, over 30 percent began their education at an HBCU (Arroyo, 2009). These institutions are uniquely prepared to educate and prepare a diverse STEMM workforce (see Box 1-4 for several examples at Historically Black Colleges and Universities and Box 1-5 for an additional example).

BOX 1-4

Women in STEMM at Historically Black Colleges and Universities.

BOX 1-5

University of Maryland Meyerhoff Scholars Program.

Programs that support students as they transition from an MSI to a predominantly White institution, where they often experience a significant shift in diversity and culture, are important for improving retention in STEMM (Ong et al., 2011). The Fisk-Vanderbilt Master’s to Ph.D. Bridge Program (FVBP) offers one example of an intentional bridge program between an MSI and a predominantly White institution. The FVBP is currently “on pace to become the nation’s top producer of underrepresented minority Ph.D.s in physics, astronomy, and materials science,” (Stassun et al., 2011) and currently leads the nation in master’s degrees in physics for African Americans. As of 2011, the number of Ph.D.s awarded to underrepresented minorities through the FVBP was up to an order of magnitude above the U.S. average—by a factor of 10 in astronomy, 9 in materials science, 5 in physics, and 2 in biology (Stassun et al., 2011).

Given that in all STEMM disciplines advancement of women into leadership roles is an issue, some institutions are taking steps to explicitly address this particular issue of underrepresentation. For example, in medicine, an intentional sponsorship program at University of Texas MD Anderson Cancer Center has led to improvements in the diversity of senior leadership. In 2007, when the program was initiated, 33 percent of the faculty and 14 percent of department chairs were women. As of 2017, 39 percent of the faculty and 29 percent of department chairs were women (Travis and Dmitrovsky, 2017) (also see Box 1-6).

BOX 1-6

MD Anderson Cancer Center Leaders’ Sponsorship Program: Improving the Representation of Women and Minorities in Academic Medicine.

In this report, the committee provides a review of the scholarship and practices that have underpinned the achievements of institutions that have succeeded in narrowing the gender gap. This review has also allowed us to define knowledge gaps, particularly with respect to identifying specific practices that can improve the recruitment, retention, and advancement of women of color and women of other intersecting identities. A persistent pattern that is widely recognized is that interventions designed to increase the representation of women in STEMM disproportionately benefit White women over women of color (Ong et al., 2011). Thus, another priority for the committee was to determine the extent to which limitations on the available body of scholarship focused on women of color in STEMM constrain the development and adoption of interventions conducive to achieving full and equitable participation of women of color and women of other intersecting identities in STEMM. The committee also compared the efficacy of interventions across disciplines within science, engineering, medicine, and mathematics and throughout career life cycles of women in STEMM to determine whether the success of particular programs and practices is context dependent.

Our goal in presenting an overview of the research on effective and promising practices and identifying factors contributing to the successes of exemplary programs is to provide guidance to institutions on how to adopt these strategies and practices and tailor them to suit their unique institutional contexts. That there are examples of successful systemic change resulting in achieving gender equity in STEMM fields across a variety of institutions that differ in size, composition, mission, and geographic location suggests that broader positive change in the representation of women in STEMM is an attainable goal.

STUDY APPROACH

Task

To conduct this study, the National Academies formed a committee of practicing scientists, administrators, scholars of women’s issues, and authorities on and advocates for equity, diversity, and inclusion in STEMM to address the statement of task (see Box 1-7).

BOX 1-7

Statement of Task.

Approach

In response to this task, the committee developed a set of findings and recommendations based on the evidence available. The findings and recommendations were informed by two extensive commissioned literature reviews, a series of focus groups, research presentations, and the committee’s own expertise and experience. The first commissioned paper, by Drs. Michelle Rodrigues and Kathryn Clancy, focused on a comparative examination of research on why women are more underrepresented in some STEMM disciplines compared to others. The second commissioned paper, by Drs. Evava Pietri, Leslie Ashburn-Nardo, Corinne Moss-Racusin, and Jojanneke van der Toorn, focused on interventions for improving the recruitment, retention, and advancement of women in STEMM. Lastly, to help address the third objective in the statement of task, the National Academies commissioned a series of focus groups, carried out by Tasseli McKay and Dr. Christine Lindquist of RTI International. These focus groups discussed why effective interventions have not been scaled up or adopted broadly by many institutions.

Throughout the report, the committee presents data and findings disaggregated by race and gender as much as possible. Unfortunately, in many instances the research conducted on promising and effective practices in recruitment, retention, and advancement in education and the STEMM workplace fails to consider the intersectional experiences of women of color or women of other marginalized identities (e.g., LGBTQIA women, women with disabilities). In some instances, the available research is limited to the experiences of White women only, in others gender and race are treated as distinct variables.

Much of the research presented in this report is qualitative in nature, which is appropriate given the topic of study, and critically important for understanding the factors that affect women. This is particularly true for research on women of color, given that the sample sizes are usually too small for quantitative analysis. Data in many of the peer-reviewed studies on the underrepresentation of women in STEMM throughout the report are drawn from focus groups, surveys, interviews, case studies, and similar methods used frequently in the social sciences, where the results are not easily translated into numbers (Bhattacherjee, 2012). This research adheres to rigorous standards for protocol and sample design, objectivity of the questionnaire or survey instrument, protection of human subjects, and methods used for data collection and follow-up to ensure that the results are as representative of the target population, complete, and accurate as possible, with a minimum of response bias. The analysis is also rigorous—requiring explanation of ambiguous results, aggregation of data where cell sizes are too small, and careful wording of conclusions to avoid going beyond what the information will support. Information gathering that does not allow for this rigorous methodology can still yield important information from the “field,” as it were, and can provide useful indications of concerns, themes, or directions; it is less useful in supporting definitive conclusions about an entire population.

In its review of the research, the committee describes unique distinctions among STEMM disciplines that create different barriers and challenges for women in STEMM. However, the research literature on promising and effective practices in different STEMM disciplines is not extensive enough to demonstrate which specific interventions work in certain disciplines and which do not. There is, however, reason to suspect that many of the interventions described throughout the report could benefit students and professionals in a range of specific STEMM disciplines and sub-disciplines, because most of these interventions have been tested across different disciplines with similar outcomes.

Further, each discipline is not a monolith. Institutional context, as opposed to disciplinary culture, may be an equally useful way to predict which policies and practices would be successful within a specific department or institution, rather than examining these interventions by discipline. Several of the recommendations in this report support institutions taking action to collect and examine demographic data, disaggregated by gender and race/ethnicity, of students, faculty, and staff at the departmental level in order to understand their own institutional context. By examining these data, departments can diagnose where there are specific problems and use this report as a resource to implement best practices to address those challenges.

For readers particularly interested in the current state of knowledge on the impact of specific interventions in the context of a specific discipline, the report contains a table in Appendix A that provides an extensive review of interventions shown to improve recruitment, retention, and advancement of women in STEMM. The table provides details on whether the intervention has been tested in STEMM and, if so, in which disciplines.

The committee largely limited the scope of the report to academia. However, the project staff reviewed research on promising practices in industry settings and found that many of the same general strategies are employed in industry as in academia (e.g. writing job advertisements inclusively, using structured interviews, bias mitigation training), although adjusted for industry settings. Most of the evidence on promising and effective practices in industry settings is not peer-reviewed, but rather exists in the form of case studies and so-called “gray literature.” It is the committee’s view that many of the recommendations offered in the report could be usefully adapted to industry settings, and should be used as a resource for industry leaders, as appropriate, given that in certain STEMM fields (e.g. computer science, engineering) the majority of STEMM professionals work in industry settings—settings with very poor representation of White women and extremely low numbers of women of color.

Report Organization

The committee has structured the report to align closely with the components of the statement of task. The report offers an overview of the barriers faced by women across a range of scientific, engineering, and medical disciplines (Chapter 2); describes evidence-based strategies that have positive impacts on the recruitment, retention, and advancement of women in STEMM education and careers (Chapters 3 and 4); and reviews common barriers to institutional change and factors that can overcome barriers and facilitate and support such change (Chapter 5). The report concludes with a set of actionable recommendations for a range of stakeholders on how to effect substantive change in women’s experiences, representation, and leadership in STEMM (Chapter 6).

Footnotes

1

This chapter builds on the significant contribution of the Committee on Understanding and Addressing the Underrepresentation of Women in Particular Science and Engineering Disciplines.

2

Labor shortages in STEMM fields are found in some disciplines and not others and change over time. In general, the academic sector is oversupplied, but there are labor shortages in government and industry in certain disciplines and sub-disciplines (e.g. cybersecurity), available at: https://www​.bls.gov/opub​/mlr/2015/article​/stem-crisis-or-stem-surplus-yes-and-yes.htm.

3

The NSF ADVANCE program provides grants to enhance the systemic factors that support equity and inclusion and to mitigate the systemic factors that create inequities in the academic profession and workplaces. The goal of this program is to broaden the implementation of evidence-based systemic change strategies that promote equity for STEM faculty in academic workplaces and the academic profession.

4

The U.S. Department of Education uses the term “nontraditional” to refer to these students. However, this committee prefers the term “post-traditional” to signal the value these students bring to their colleagues.