4

Effective Practices for Addressing Gender Disparity in Recruitment, Advancement, and Retention in STEMM

The analysis draws considerably from An Inclusive Academy: Achieving Diversity and Excellence by Drs. Abigail J. Stewart and Virginia Valian. In addition, the chapter builds significantly on the contributions of the Committee on Understanding and Addressing the Underrepresentation of Women in Particular Science and Engineering Disciplines.

A growing body of research literature and an increasing number of examples identify strategies and practices that institutions and organizations can adopt to diversify talent pools, mitigate biases in evaluation and promotion, and create and sustain a positive, inclusive organizational climate (Boxes 4-4 and 4-5). The sections below offer practical guidance on specific steps that institutions can implement to work to improve recruitment, retention, and advancement of women in science, technology, engineering, mathematics, and medicine (STEMM).

ACKNOWLEDGEMENT OF RESEARCH LIMITATIONS

There are certain limitations to the available research and practice that must be acknowledged up front. First and foremost, it is important to acknowledge that most research on promising and effective strategies for improving the recruitment, retention, and advancement of women in STEMM has failed to take into account how the intersection of gender with other marginalized identities (e.g. race, disability status, sexual orientation) may influence the efficacy of these interventions. Efforts supporting women in STEMM have tended to focus on middle-class White women at research intensive universities (Ong et al., 2011). As such, White women have tended to benefit predominantly from these efforts. That is not to say that the interventions outlined in this chapter may not, or cannot, support women of multiple marginalized identities in STEMM, but that we simply do not know.

Second, given that the focus of most of the research on promising practices in the STEMM workplace has been on research universities, information on the

impact of these strategies and practices in a variety of institutional contexts is lacking. Given that most students in the U.S., and most students of color, are not studying at research intensive universities, much more research is needed on the impact of these interventions in a range of institutional contexts, including community college and minority serving institutions (MSIs).

Additionally, maintenance of an inclusive organization climate requires constant awareness of the impact of bias as well as intentional efforts to compensate for ingrained predispositions. As Regner et al. (2019) has shown, in the absence of monitoring, implicit biases continue to drive gender disparities even after evaluators participate in training designed to mitigate bias. Implementing the strategies and practices outlined in this chapter will not produce a process, or people, free from bias. Rather, if implemented intentionally, these strategies and practices can help mitigate biases.

Finally, the practices outlined in this chapter are likely broadly applicable across STEMM disciplines; however, there is certainly room for additional research on the ways these particular strategies and practices can impact recruitment, retention, and advancement of women within particular disciplines and sub-disciplines, as well as within the unique context of different schools and departments (see Chapter 5 for additional discussion). At a high level, the practices related to recruitment into academic computer science and engineering programs and further into industry jobs are particularly important, given the high attrition rates of women from these fields at the undergraduate level and the substantial underrepresentation of women in the technology industry. In the biological and biomedical sciences, particular attention should be paid to the postdoctoral stage and recruitment into the professoriate, as well as to biases that prevent retention and advancement of women into more senior ranks in academia, industry, and medicine.

Given the limitations of the existing research, successful implementation of the strategies and practices outlined in this chapter will be facilitated by the collection and monitoring of data on the recruitment, retention, and advancement of White women and women of other marginalized identities within institutional units (e.g., departments) before, during, and after implementation, as well as consideration of how these practices could work within the particular context of the institution (e.g. mission, size, resources, student needs, faculty expertise and competencies). There is no one size fits all approach that will work within all institutional contexts, and so an iterative pilot stage in which the practices and strategies outlined in this chapter may be “adapted” is likely to be a useful strategy. Chapter 5 offers an overview of a process through which an institutional unit may first “diagnose” its particular issues with recruitment, retention, and advancement of women, then makes use of the information presented in this chapter to take action to “treat” specific issues, and then evaluate whether the treatment worked and institutionalize effective practices. It is through such

a process that many of the institutions highlighted throughout this report have identified the strategies that would work within the particular context of their particular institution.

WHAT WORKS TO IMPROVE RECRUITMENT

Institutions can improve their recruitment of women (and other underrepresented groups) by adopting or “adapting” strategies that enhance their appeal to a broader range of potential applicants and increase the likelihood of fair and effective evaluation of candidates. High levels of retention, a positive climate, and a good record of equitable advancement will support these strategies, but improving recruitment and hiring practices is an essential part of the overall institutional change needed to increase the presence of women scientists in an organization.

STRATEGY 1: Actively recruit year-round and expand networks of candidates.

An academic or other organization can attract a diverse set of candidates if it works continuously to diversify and grow its applicant pool for all positions and adopts proactive strategies to identify qualified candidates. Johnson et al. (2016) found that the odds of hiring a woman were 79 times greater if there were at least two women were in the finalist pool. The research indicates that organizations and institutions should hire new employees who have a documented record of serving as good mentors to women and other underrepresented groups as they will have access to a broader network (Johnson et al., 2016). This effort may include:

• Attending conferences to establish relationships with promising scholars and students from underrepresented groups and their mentors (Stewart and Valian, 2018).
• Requesting referrals from identified mentors of appropriate candidates from underrepresented groups (Stewart and Valian, 2018).
• Using searchable databases, such as PRISM or National Institutes of Health’s Network of Minority Health Research Investigators (see Box 4-1).¹
• Limiting referral hiring of current employee networks and friends, which will likely replicate a lack of diversity (Stewart and Valian, 2018).

STRATEGY 2: Represent the organization, program, and position in terms that make evident how it might appeal to a broad range of applicants.

___________________

¹ See, for example, the National Institutes of Health Network of Minority Health Research Investigators (NMRI), available at: https://www.niddk.nih.gov/research-funding/research-programs/diversity-programs/network-minority-health-research-investigators-nmri.

Use of language that signals to women that they belong and fit the job description (Smith et al., 2004) is a productive strategy for attracting a more diverse pool of applicants. Specific practices to construct these signals include:

• Writing job advertisements in a way that is appealing to a broad applicant pool, particularly by avoiding gendered wording in job descriptions (Gaucher et al., 2011).
• Placing a minimal number of constraints on the requirements of the job, so that women and men of color are less likely to self-select out of applying (Stewart and Valian, 2018).

• Using gender-neutral terms (e.g., “chair” versus “chairman”).
• Avoiding descriptors that are stereotypically associated with men (e.g., “is a natural leader” or “can succeed in a competitive environment”) (Stewart and Valian, 2018).

Institutions and organizations should signal to all candidates that they will be evaluated based on professional experience, not demographic characteristics, by ensuring that institutional publications, including websites, mention:

• Specific values that support diversity (Stewart and Valian, 2018).
• Fair and inclusive recruitment and hiring procedures.
• Family-friendly policies for faculty, students, post-docs, and staff.

Additionally, institutions and organizations can attract a more diverse pool of candidates by signaling their sensitivity to concerns about community diversity by providing information on the surrounding area, the history of the area, demographic data, and local activities that might be attractive to diverse candidates (Stewart and Valian, 2018).

STRATEGY 3: Examine job and admissions requirements to assess vulnerability to bias, adequacy as accurate indicators of talent, and relevance to success.

Making hiring and admissions decisions can be a time-consuming and complex process, especially when there are many qualified applicants. In many employment settings, hiring managers, admissions officials, and search committees use shortcuts to identify a subset of the most promising candidates. Such shortcuts can introduce bias into the process. For example, the prestige of the institution where the applicant received training is often erroneously used as an unbiased metric to evaluate candidates. In fact, using institutional prestige as an evaluation step is more likely to eliminate African Americans than Whites, Hispanics, or Asian Americans, and more likely to eliminate women than men (Stewart and Valian, 2018). Relying on institutional prestige favors White male applicants, and studies have shown that top students from lower-ranked schools are often as successful as their peers who attended more prestigious schools (Dale and Krueger, 2002, 2011).

Other metrics that are subject to bias are standardized test scores (e.g. GRE [Graduate Record Examinations], SAT), teaching evaluations, number of publications and journal impact factors, and letters of recommendation (Madera et al., 2009; Schmader, et al. 2007; Stewart and Valian, 2018; Trix and Psenka, 2003). Issues to consider in assessing an organization’s evaluation practices include:

• Considering whether the characteristics of existing employees, students, or leaders at the institution are the only or best characteristics necessary

• for success in the job or educational program. Such criteria may be biased against certain groups. As stated by Stewart and Valian (2018) in their book An Inclusive Academy:

  There might be many satisfactory ways to do a job only a subset of which are displayed by current job holders. For example, faculty can be surprised by how well a colleague performs as chair, noticing that his or her focus and style are very different from a previous admired chair’s, but equally, if not more, beneficial to the department. (p. 203)

• Considering the importance of research that has demonstrated that women who exhibit behaviors that are lauded (or at least tolerated) in men—such as ambition, self-promotion, competitiveness, or assertiveness—are evaluated negatively by both men and women. This phenomenon, known as the “backlash effect,” occurs when women who are viewed as competent (i.e., they exhibit agency and authority) are, in turn, viewed negatively in terms of warmth and likeability (Rudman and Phelan, 2008). Such stereotypes can play out through race/ethnicity as well as gender (Fiske, 1999). For example, Asian Americans are often seen as competent, but not likeable, whereas African Americans and Latinx people are commonly viewed as likeable, but not competent (Fiske, 1999). This phenomenon unfairly disadvantages women and people of color, particularly in the context of leadership roles where perceptions of competence are especially important.

STRATEGY 4: Explicitly establish criteria for evaluation before assessing the pool of applicants.

Cognitive psychology research demonstrates that people share tendencies toward systematic errors in judgment that are amplified by implicit biases (Kahneman, 2011). Those making hiring or admissions decisions should be aware of how their biases can affect their judgments and should be required to adopt practices that will mitigate biases in evaluation. As a first step, hiring managers, admissions officers, and search committees can reduce bias by establishing explicit criteria before reviewing any application materials (Brewer, 1996; Norton et al., 2004; Tetlock and Mellers, 2011). This process will allow evaluators to develop relevant rationales for judging applicants that accurately reflect the desirable attributes identified as important by the organization, department, or school in a student, employee, or faculty member (see Strategy 3 above). If qualifications are waived for a specific candidate, the organizational or departmental leadership should require an explanation of why they are not important in that case—and keep track to see for whom requirements are waived.

STRATEGY 5: Hold those responsible for admissions and hiring decisions accountable for outcomes at every stage of the application and selection process.

Establishing clear criteria is only useful if evaluators adhere to them when reviewing application materials. To reinforce such adherence, evaluators should be required to complete an application review for each candidate, with the agreed-upon criteria listed (see Box 4-2). In this way, leaders can ask individual evaluators to justify their judgments with reference to the evidence in the file. To facilitate implementing clear hiring criteria, evaluators can, for example:

• Implement a “short-list review” (Billimoria, 2010), whereby the evaluators share their short list of candidates, their efforts to recruit a diverse applicant pool, and the demographic makeup of their list with an administrative office or leader. If their short list is less diverse than the applicant pool, then an admissions oversight body can weigh in and possibly demand a more diverse short list.
• Continue to monitor equity benchmarks throughout the hiring process to gauge whether they are achieving their goals or not by adhering to their clearly defined criteria (Sagaria, 2002).

STRATEGY 6: Educate evaluators to be sensitive when considering “gaps” in a resume.

Resume “gaps” can arise in many ways and immediately dismissing them as disqualifying can inappropriately limit the applicant pool. Women with children, for example, are 79 percent less likely to be hired than an identical candidate without children (Correll et al., 2007). Bias against parents is not only limited to women; according to Rudman and Mescher (2013), men who requested parental leave were viewed as “weaker” and poorer workers who were less deserving of economic rewards.

STRATEGY 7: When possible, evaluate a candidate’s work directly.

In the case of faculty hires in academic settings, search committees should reduce reliance on biased proxies by evaluating a candidate’s scholarship by reading their work directly. Although time-consuming, such direct evaluation of scholarship can increase the accuracy of scholarship assessment (Stewart and Valian, 2018).

• Limiting the number of publications applicants can submit to their top three (or fewer), which will allow candidates to submit the work that they think is most important. This practice also helps level academic capital awarded to those who publish with their advisor (Pinheiro et al., 2014), as men do more frequently than women, and counters biases related to prestige of publications and citation counts, which benefit men over women (Larivière et al., 2013; Maliniak et al., 2013).

• Ensuring that at least one publication per applicant is read by at least two members of the search committee (Stewart and Valian, 2018).
• Encouraging search committees to consider whether candidates achieved more than might be expected given the resources available at the institution(s) where they received their education and training, and avoid reliance on institutional prestige.

STRATEGY 8: Educate those who are assessing applicants about the best practices to mitigate biases in hiring and admissions.

Institutions and organizations should work to ensure that those who are assessing applicants be educated about best practices to mitigate biases in hiring and admissions. While research on one-off implicit bias training program shows that they can backfire, as they have the potential to normalize biases or create a sense of “moral license” among participants, Jackson et al. (2014) also suggest that approaches that emphasize particular practices to reduce bias can have positive effects.

Sekaquaptewa et al. (2019) showed that attending practice-oriented workshops increased positive attitudes toward equitable recruitment practices (such as those described in this chapter) and behavioral intentions to enact those strategies. Importantly, these effects were the result of attendees’ increased belief in the social science evidence presented in the workshop (e.g., evidence of implicit bias in hiring and effects of stereotype threat on performance). Even faculty members who did not personally attend a workshop also showed these outcomes if they were in departments where a high percentage of department members had attended in the past. This finding suggests that the positive influence of the workshop can spread to others in the department, presumably through widespread knowledge and commitment to good practices (Sekaquaptewa et al., 2019).

Related research by Carnes et al. (2015) suggests that bias mitigation education that uses certain practices has the potential to move beyond simply raising awareness of bias to actually reducing gender disparities in hiring and improving organizational climate. Carnes et al. (2015) randomly assigned 46 science, medical, or engineering departments at the University of Wisconsin to receive a bias mitigation workshop intervention and 46 to serve as a waitlist control. Participants took the Implicit Association Test and answered questions indicating behavioral change before and after the intervention. The specific evidence-based strategies used in the workshop included:

• Strategy replacement, where a person recognizes they are stereotyping and challenges the stereotype with data.
• Counter-stereotypic imaging, where a person imagines exemplars of powerful women.
• Individuating, where a person avoids making generalizations about someone.

• Perspective-taking, where a person imagines living as a stereotyped individual.
• Increasing opportunities for contact with a stereotyped group.

These investigators found that stereotype suppression and too strong a belief in one’s objectivity led to the greatest biases. The intervention improved participants’ awareness of their personal biases, their motivation to change, and their self-reported behavior. A follow-up survey of faculty work-life showed that department climate improved and the departments that underwent the intervention had also hired more female and minority faculty than the controls at 2 years following the intervention.

STRATEGY 9: Use structured interviews and avoid asking inappropriate or illegal questions.

Once a short list has been established, organizations and institutions should ensure that interviews are fair, and recognize that the interview will influence a candidate’s decision to accept an offer. Fair interviews adhere to the following guidelines:

• To verify that every candidate is treated equitably, employers should hold structured interviews with candidates, which are more resistant to biasing factors in evaluation than unstructured interviews (Levashina et al., 2014). For a thorough assessment of the elements of a good structured interview, see (McCarthy et al., 2010).
• Interviewers should be made aware that casual conversations over a meal at the end of an interview day count as part of the interview. Therefore, interviewers should not inquire about any personal information that candidates do not offer up themselves (e.g., “Are you currently pregnant or planning to have children?”), as these questions are illegal in interview settings, cannot be used in the discussion of candidates, and may be upsetting or off-putting to a candidate.
• Information on resources that may seem relevant to only some candidates, such as childcare policies, should be provided to all candidates.
• Interviewers should ask performance-based questions or behavioral interview questions (e.g., “Tell me about a time . . .”), because answers to such questions are strong predictors of future job performance and reduce reliance on perceived “potential,” which tends to unduly advantage White men (Bock, 2015).
• If “culture fit” is viewed as important for success, organizations should provide a specific definition of what is meant by the culture of the specific organization or institution. Research has shown that evaluation of fit can disadvantage women and people of color if it is vague or inappropriately

STRATEGY 10: Offer students, trainees, and employees a living wage.

In academia, research indicates that many women leave at the transition from graduate school to the assistant professor stage (NRC, 2010). Although few studies have determined the specific underlying motivations and ultimate destinations of women who leave at this stage, among the testable hypotheses is that departures are due to economic constraints, particularly those arising from family needs. The low pay and long training periods for scientists, engineers, and medical professionals means they may face extreme financial hardship as they attempt to start a family while simultaneously continuing their training. The tradeoffs may be particularly challenging for those from disadvantaged backgrounds.

To address this issue:

• Offer salaries commensurate with the training STEMM professionals have received, combined with their value to the research endeavor. For example, paying a Ph.D. postdoc with 10 years of post-secondary education $50,000 for a 50-60-hour work-week—in some cases while they also raise children and pay for childcare—means only those with outside income sources can realistically pursue this career path.
• Do not defer contributions to retirement plans for early-career positions, such as postdoctoral positions. Deferring such contributions has major long-term financial implications and represents a further impediment to those from less advantaged backgrounds. This is especially true for some women who may also absent themselves from the formal workforce for an additional period for family reasons.

WHAT WORKS TO IMPROVE ADVANCEMENT

Background

Biases in evaluation for promotion are well documented and lead to different outcomes for women and men as they advance through their careers. In one recent study of performance evaluations in the technology sector, 75.5 percent of women’s performance evaluations included language that was critical of their personality or behavior (e.g., “you come off as abrasive”) compared with 1 percent of men’s performance evaluations (Snyder, 2014). Several other studies have found that men are more likely to be granted tenure than women in the professoriate (Box-Steffensmeier et al., 2015; Di Fabio et al., 2008). Women spend longer within a particular rank than men (Ash et al., 2004; Geisler et al., 2007; Poor et al., 2009; Valian, 2000) (see Boxes 4-4, 4-5, and 4-6 for additional commentary

on tenure). Women at 4-year colleges and universities are 10 percent less likely than men to attain full professor promotion, even after controlling for productivity, educational background, institution type, race, ethnicity, and nationality, suggesting that external factors may drive the differences in representation between men and women at the faculty level (Perna, 2001). Additionally, women tend to hold appointments at lower ranks, have positions with less prestigious affiliations, and receive less recognition than their male colleagues across academic science (Bailyn, 2003; Fox, 2001; Holton and Sonnert, 1995; Lincoln et al., 2012; NRC, 2001; Sonnert and Holton, 1995; Xie, 2003). There are particular issues with lack of advancement in the life sciences, where women are well represented at lower ranks but poorly represented at higher levels (NSF, 2019).

Many of the same kinds of practices that mitigate bias in evaluations for recruitment also mitigate biases in evaluations of performance that are associated with advancement. Here, the committee highlights a range of evidence-based practices that can mitigate bias in the advancement of women in STEMM.

STRATEGY 1: Sponsor women in senior leadership.

Sponsorship² is an important avenue to improve numbers of women in senior leadership. Neither formal mentoring programs nor executive coaching led to increases of women in top leadership, but sponsorship programs designed to accelerate the careers of women as leaders had a positive impact; in contrast with mentors, who work closely with faculty to enhance their research and education skills, sponsors have the position, power, and influence needed to advocate publicly for advancement of talented women to senior leadership positions (Helms et al., 2016; Travis et al., 2013). Both male and female employees are more likely to report that they are advancing in their careers, and feel more comfortable asking for raises, when they have sponsors in their company or institution compared with when they do not have sponsors (Hewlett et al., 2010). Connecting women to sponsors through formal programming is crucial for the advancement of women in STEMM (Huston et al., 2019; Sherbin, 2018).

Sponsorship also might help address another factor hindering women from reaching top leadership roles in technology companies—their lack of visibility within an organization and/or shortage of opportunities to showcase their skills and value to the organization (Correll and Mackensie, 2018; Simard et al., 2008). Specifically, sponsors can work to ensure that women have access to high-profile projects to propel them into more prestigious leadership positions.

___________________

² According to the National Academies report The Science of Effective Mentorship in STEMM (2019b), sponsorship is defined as “a potential career support function that involves a senior person publicly acknowledging the achievements of and advocating for a mentee.”

STRATEGY 2: Establish clear, unbiased metrics for promotion and advancement.

To ensure a fair review, evaluations should be based on criteria that are clearly defined and consistently applied. This set of criteria should be consistent with the mission and goals of the company or university (Perna, 2001) and reward the types of behaviors that the employer or institution values (Stewart et al., 1996). Criteria should include the full range of activities that may be of interest to women in academia, including teaching, service, and research, and should provide incentives for working to achieve the company or institutional goals (Tierney and Bensimon, 1996).

• Evaluations should be conducted by more than one individual, as such evaluations are trusted and seen as fairer than those done by a single individual; this also builds in accountability, when evaluators know their opinion will be assessed alongside someone else’s (Thorngate et al., 2009).
• Ensure that the review process is transparent and accountable by:
  • making sure administrators and tenure committee members are transparent about the procedure for evaluation, the steps in the process, and the timeline; and
  • reviewing the process periodically to determine if the process is fair or needs to be changed in any way.
• Collect data to answer questions such as:
  • Do your performance evaluations show consistently higher ratings for majority men than for women, people of color, or other underrepresented groups?
  • Do ratings decline after women employees have children, take parental leave, or adopt flexible work arrangements?
  • Do the same performance ratings result in different promotion or compensation rates for different groups?

The Center for WorkLife Law at the University of California at Hastings College of the Law offers a useful set of evidence-based tools and worksheets for employers. Among the practices recommended by the center are the following:

• Appoint a bias interrupter. Have team members or human resource business partners who have been trained to spot bias involved at every step of the evaluation process.
• Begin with clear and specific performance criteria directly related to job requirements.
• Require evidence in the form of specific examples from the evaluation period that justifies the rating.

• For each candidate, consider their demonstrated performance and potential separately. Performance and potential should be appraised separately, given the tendency for a majority of men to be judged on potential whereas others may be more often judged on performance (University of California at Hastings, 2019).
• Separate personal styles from skill sets for each candidate. Personal style should be appraised separately from skills, because a narrower range of behavior often is accepted from women and people of color. For example, women may be labeled “difficult” for actions that are accepted in majority men (University of California at Hastings, 2019).
• Level the playing field by ensuring all candidates for advancement possess skills for self-promotion and for articulation of expectations.
• Offer alternatives to self-promotion. Encourage or require managers to set up more formal systems for sharing successes, such as a monthly email that lists employees’ accomplishments.
• Provide a “bounceback.” Managers whose performance evaluations show persistent bias should have their evaluations returned to them for further analysis.
• Have bias interrupters play an active role in calibration meetings. In many organizations, managers meet to produce a target distribution of ratings or cross-calibrate rankings. Have managers read the Identifying Bias in Performance Evaluations Worksheet before they meet. Have a trained bias interrupter in the room.
• Retain a formal performance appraisal system. Eliminating formal performance evaoluation systems and replacing them with feedback-on-the-fly creates conditions for bias to flourish.

For more information and resources from the Center for WorkLife Law, see https://biasinterrupters.org/toolkits/orgtools/.

STRATEGY 3: Recognize and reward outstanding contributions to STEMM.

There are many ways to recognize and reward people for their outstanding contributions outside of formal promotion and evaluation processes. These include institutional awards for teaching, mentoring, service, and honorary degrees, as well as recognition by honorary societies, such as the American Academy of Arts and Sciences, the American Association for the Advancement of Science, and the National Academies of Sciences, Engineering, and Medicine. The fact that women have been historically underrepresented, marginalized, or overlooked in many scientific disciplines has contributed significantly to their underrepresentation in honorific societies. Differential failure to recognize and reward accomplishments by members of particular groups contributes not only to lack of advancement but also to attrition and to an overall negative climate for underrepresented groups.

STRATEGY 4: Encourage and Reward Contributions to Diversity, Equity, and Inclusion.

Chapter 2 presents research demonstrating that women in academia shoulder the burden of teaching, mentoring, and service (Armstrong and Jovanovic, 2015; Hermanowicz, 2012; Kulis et al., 2002; Madge and Bee, 1999; Urry, 2015), activities that can have a positive impact on diversity, equity, and inclusion (see Chapter 3). In fields where women are over- or well-represented at early career stages, but remain underrepresented at higher ranks, this gendered division of labor could contribute to this disparity by penalizing women for devoting time and energy to service efforts if they take time away from other activities the institution prioritizes and rewards, such as securing research grants and publishing peer-reviewed papers.

This issue was raised in the focus group research carried out by RTI International on behalf of the National Academies. Listening session participants suggested that once women achieve tenure, they encounter limited support for their ongoing advancement. They observed that many women’s careers stall at associate professor, with little institutional support for achieving promotion to full professor. In the words of one participant:

In addition to a lack of programs or policies aimed specifically at supporting or recognizing the intellectual contributions of mid-career women, participants noted that mid-career faculty often face the heaviest service burdens. When inequitably distributed by gender and race (as it is perceived to be at most institutions), the burden of service effectively curtails women’s other contributions and hinders their advancement to full professorship. To quote one participant:

While some listening session participants felt that formal efforts to ensure equal distribution of service labor could help to address this challenge, others proposed that recognizing faculty contributions to diversity in their departments would better help to preserve these important contributions (for example, access to effective mentorship for underrepresented students) while still righting the imbalances they created. By formally assessing and valuing contributions to diversity in the context of performance review, merit increase, and promotion decisions, universities could avoid inadvertently penalizing those whose time was disproportionately devoted to it. Some institutions have also begun to take concrete steps in this direction. For example, several institutions have formally

incorporated contributions to diversity, equity, and inclusion in tenure and promotion criteria. Box 4-4 provides several examples, and Box 4-5 offers an example rubric for how a department might consider the multiple ways a faculty member could work to promote diversity, equity, and inclusion.

WHAT WORKS TO IMPROVE RETENTION AND ORGANIZATIONAL CLIMATE

Background

Creating positive and collegial work environments is an essential element in retaining employees and thus is critical for achieving gender parity. STEMM professionals, regardless of gender or race, require similar conditions to thrive: inclusion, full participation, and community respect. These conditions are more available to majority demographic groups in university and business settings than to those that are from underrepresented groups, who often feel different from, or less comfortable with, colleagues (Ackelsberg et al., 2009; Gutierrez et al., 2012; Guzman et al., 2010; Harris, 2007; Thompson, 2008; Tokarczyk, 1993; Urry, 2008).

To improve retention and promote a positive organizational climate, employers and educators should provide a supportive working environment for a diverse group of students and employees by:

• Devoting resources to support research, teaching, advancement, and career development,
• Creating structures that promote fairness and transparency, and
• Treating employees with respect in both their personal and professional lives.

A positive organizational climate will also serve to prevent sexual harassment (Buchanan et al., 2014; Fitzgerald et al., 1997; Glomb et al., 1997; Wasti et al., 2000).

As stated in a National Academies report on this topic:

Preventing sexual harassment, rather than waiting to address it once it occurs, is critically important to retaining women in STEMM, not only because of the damaging effects of sexual harassment on women’s personal and profes-

sional health and well-being, but also because the research shows that women are unlikely to report sexual harassment because of an accurate perception that reporting sexual harassment will lead to a process that will re-victimize them (NASEM, 2018b). Scholars and focus group participants described a number of specific ways for organizations and institutions to improve the overall organizational climate, several of which are described below.

STRATEGY 1: Ensure fair and equitable access to resources for all employees and students.

To create fair and equitable access to resources for all employees and students, organizations should provide:

• Equal knowledge of institutional resources. Often, information about these resources is obtained through informal communications; however, these informal communication networks flow more smoothly among similar groups, which often leave underrepresented groups in the dark (Kanter, 1977; Rankin et al., 2007).
• Transparency about how resources are allocated. Institutions and organizations should repeatedly communicate information (Collins, 2001).
• A dedicated employee whose job is to disseminate necessary information, complete with relevant supplemental information and links.
• Direct assistance to employees including:
• • technology assistance, and
  • skill development
• Policies and resources for programs that recognize the diversity of families that exist and how family situations can evolve over time and throughout individuals’ careers.
• Institutionalization of effective policies and practices so that they will survive a transition in leadership. One of the key messages from the focus group research carried out for this report (see Chapter 5 and www.nap.edu/catalog/25585) was that leadership transitions are a point of vulnerability for efforts to promote diversity and inclusion because new leaders may have different priorities. Standing policies make it less likely a change will be undone (see Chapter 5 for additional discussion).

STRATEGY 2: Create a respectful and equitable organizational climate.

Institutions and organizations should create respectful work climates that are perceived as equitable and fair by employees and students (see Box 4-5). When women feel that they are not receiving fair treatment or sufficient resources from institutional leaders, they tend to experience lower levels of job satisfaction and organizational commitment (Branscombe and Ellemers, 1998; Ensher et al., 2001; King et al., 2010). Organizations can create a fairer environment by being transparent about:

• The decisions that are likely to be made in the near and far future
• The rationale behind those decisions
• Who will participate in making these decisions
• What criteria will be used in the decision making (Stewart and Valian, 2018).

This level of transparency helps employees feel that organizations are operating according to some general principles and that, even if the decision is unfavorable, their preferences were heard.

Changing a workplace climate requires a coordinated, cooperative effort (Jordan and Bilimoria, 2007). In a case study of an academic science work envi-

ronment conducive to the advancement of both men and women scientists, Jordan and Bilimoria (2007) identified several conditions and factors that supported a cooperative, inclusive, and productive work culture. The foundation of this culture was a set of values and beliefs about the goals of science (the pursuit of meaningful, significant advancements of knowledge) and qualities of successful scientists (the ability to learn rapidly, effective communication skills, an abundance of creativity, and a strong work ethic) reflected in the cooperative interactions within the department. Department chairs were helpful in creating and sponsoring activities, but support and ongoing leadership in maintaining these practices came from faculty within the department. Over time, these processes embedded values and beliefs held by a majority of department members as shared values and beliefs of the department, which ultimately shaped the department culture.

The focus groups also identified a number of facilitators of positive organizational climate, including:

Gender balance affects climate. Focus group participants perceived that, as changes in the gender composition took place in some departments over time, the culture changed accordingly and improved subsequent efforts at climate improvement (such as the success of mentoring programs).

Mentoring and social networks affect climate. Interpersonal connections built through mentoring and networking ameliorated negative organizational climates.

• Networks composed of women in their fields that provided both instrumental support (e.g., guidance on the university tenure process) and emotional support (e.g., an outlet for venting). Interdisciplinary collaborations offered an important avenue for intellectual work stifled by hostile departmental climates.
• Networking among National Science Foundation ADVANCE programs brought people together across institutions to exchange ideas. Cross-institution networking promoted both the sustainability and influence of efforts initiated with ADVANCE funding.

Individuals facilitated a positive climate. Participants identified several ways in which individuals’ involvement acted to facilitate positive climates, including:

• Internalization of responsibility for contributing to positive climate.
• Bystander intervention and peer-to-peer learning (see Box 4-7).
• Availability of particular individuals for informal, confidential conversations about equity.

The specific disciplines can promote positive climates. The final set of facilitators pertained to the broad influence of disciplines as a whole. For example, as one focus group participant noted:

Climate assessments can drive positive change in disciplines. Some participants commented on the particular influence that external climate assessments or reviews conducted by various scientific or professional societies have within their disciplines. They emphasized the positive change within departments or institutions that can result from this external force.

STRATEGY 3: Policies and resources should work to address the family-related needs of students and employees.

Central to the challenges of retention are the family-related needs that an employee or student may have over the course of their career and education. Without a federal policy for paid family leave (Box 4-8), it is at the employers’ discretion to determine fair family leave policies. However, these policies differ from institution to institution and even vary within a single institution (see Box 4-9).

Due to the funding structure supporting most researchers—that is, they are supported partially by their institutions and partially by federal grant money—the employer and the funding agency supporting the salary and work of an employee share responsibilities for family leave accommodations. A major finding of the focus group research conducted for this report was that family leave-related policies and practices within extramural funding agencies are a major barrier to gender equity among STEMM faculty, even when university-level policies are supportive. For example, participants explained that the fact that faculty with National Institutes of Health funding drain their grants during paid family leave forces individuals to choose between having their grant funds expended while work is paused or avoid taking leave altogether. Similarly, the failure to educate study sections on assessing timeline for researchers who have taken family leave can deter women from taking leave and potentially harm their career advancement if they do, due to perceived low productivity.

However, extramural funding agencies can also be a facilitator for supporting parents in STEMM. For example, the National Institutes of Health grants allow for reimbursement of childcare costs, parental leave, or additional technical support when such costs are incurred under formally-established institutional policies (National Institutes of Health, 2020). Grantee institutions can also request administrative supplements to cover these types of costs, if necessary. Additionally, the National Institutes of Health now provides up to 8 weeks of paid parental leave for intramural trainees and extramural trainees who have a Ruth L. Kirschstein National Research Service Award.

As discussed in Chapter 2, it is important to acknowledge that the “ideal worker norm” 24/7 work culture of certain STEMM disciplines disadvantages women because they tend to bear a greater share of parenting and domestic responsibilities relative to men. In one study, married or partnered female physician researchers with children reported spending 8.5 hours more per week on parenting and/or domestic activities than male physician researchers (Villablanca et al., 2011). Thus, women are more likely to view work-life balance as a key priority (DeCastro et al., 2014). Nevertheless, real and perceived barriers to the utilization of family-supportive policies are evident; even when career flexibility policies are available, cognitive dissonance can arise over their use due to concern for negative personal or professional repercussions,

including being perceived as being less committed to one’s career (Carr et al., 2017; Villablanca et al., 2011).

The remedy for this situation is well-articulated, broadly communicated, and consistent policies related to family caregiving and childbearing, accompanied by culture change efforts aimed at normalizing the use of such benefits and resources. Institutions and organizations should create policies, and set aside resources, to support employees during times when family and personal life demands are heightened. Practices should include:

• Making stop-the-tenure-clock and modified duty policies (which should be available to as wide a group as possible) a genuine time-out from work

for caregiving and medical recovery (not as an opportunity to publish),⁴ and should not penalize those who take advantage of the policies.

• Providing private spaces with appropriate equipment for parents to feed infants and (if needed) to express and store milk,

___________________

⁴ Research on the impact of “stop-the-clock policies” has been mixed, with one study showing such policies benefiting men who used the time out of the office to publish papers and book chapters (Antecol et al., 2018).

• Limiting department meetings and functions to specified working hours that are consistent with family-friendly workplace expectations.

In general, policies and practices that address workers’ need to balance work and family roles, and that recognize that these policies increase productivity and enhance work performance, will serve to support the creation of a positive, inclusive organizational climate.

The committee acknowledges that the issue of supporting working parents is a complex multi-stakeholder issue, but without a national policy, institutions and organizations must work to support working families using whatever resources are available. In the private sector, many companies have begun offering generous, paid maternity and paternity leave for birth and adoption (Greenfield, 2018).

FINDINGS: CHAPTER 4

FINDING 4-1. Although many educational institutions and employers have adopted programs and policies aimed at improving equity and diversity in STEMM, such interventions typically fail to consider the complex, cumulative ways in which the effects of multiple forms of discrimination (e.g., racism, sexism) intersect in the experiences of women of multiple marginalized identities (e.g., women of color, women with disabilities, sexual minorities, etc.). Programs aimed at improving the representation of women in STEMM have largely benefited White women and have not paid enough attention to the experiences of women with multiple intersecting identities.

FINDING 4-2: There are a number of recruitment strategies identified by institutions and companies that have been proven effective in increasing the number of women entering academic programs and STEMM jobs. Some of these strategies include:

1. Working continuously to identify promising candidates from underrepresented groups and expanding the networks from which candidates are drawn.
2. Writing job advertisements in ways that would be appealing to a broad applicant pool.
3. Interrogating the requirements and metrics against which applicants will be judged to make sure they are not biased and are not poor predictors of success.
4. Deciding on the relative weight and priority of different admissions or employment criteria before interviewing candidates or applicants.
5. Holding those responsible for admissions and hiring decisions accountable for outcomes at every stage of the application and selection process.
6. Educating evaluators, including reviewers, to be thoughtful when considering “gaps” in a resume.

7. When possible, evaluating a candidate’s work directly.
8. Using structured interviews.
9. Educating hiring and admissions officials about bias and strategies to mitigate biases.
10. Increasing stipends and salaries for graduate students, postdocs, nontenure track faculty, and others to ensure all STEMM trainees and employees are paid a living wage.

It is important to acknowledge that most of the research on these strategies has not disaggregated data by gender and intersectional identity (e.g., race, disability status, sexual orientation) and has tended to focus on improvements for White women at research universities. Thus, it is difficult to know the extent to which such strategies and practices can benefit women of multiple marginalized identities studying and working in a range of institutional contexts (see Chapter 5 for a discussion of how local, disaggregated data collection and evaluation can offer a process through which a unit (e.g., department, school) can work to develop targeted, data-driven strategies and practices to support all women).

FINDING 4-3: There are several approaches that have been proven effective in mitigating biases that serve to hinder the advancement of women in STEMM along their educational and career trajectory. Such approaches include:

1. Establishing clear metrics for success and advancement and avoid reliance on metrics that are known to be biased (e.g. teaching evaluations, impact factor of publications, external letters of support, or appraisal of “potential”).
2. Mitigating bias in performance evaluations, promotion decisions, and in selections for awards and special recognitions. For example:
   1. having someone who has been trained to spot bias involved at every step of the evaluation process.
   2. beginning with clear and specific performance criteria directly related to job requirements.
   3. considering performance and potential separately for each candidate.
   4. separating personality issues from skill sets for each candidate.
   5. leveling the playing field by ensuring everyone knows how to promote themselves effectively and sending the message they are expected to do so.
   6. offering alternatives to self-promotion to communicate accomplishments.
   7. ensuring that performance reviews are conducted by more than one individual so decisions are based on more than one perspective.

Again, it is important to acknowledge that most of the research on these strategies has not disaggregated data by gender and intersectional identity (e.g., race, disability status, sexual orientation) and has tended to focus on improvements for White women at research universities.

FINDING 4-4: There are a range of strategies and practices that institutions and organizations can adopt or “adapt” to improve the retention of women within STEMM educational programs and careers, including:

1. Ensuring that there is fair and equitable access to resources for all employees and students, including equal knowledge of institutional resources and transparency of how resources are allocated.
2. Revising policies and resources to reflect the diverse personal life needs of employees and students at different stages of their careers and education and advertise these policies and resources so that all are aware of and can readily access them.
3. Monitoring use of policies and revising them when necessary to meet the needs and access for all groups.
4. Creating programs and educational opportunities that encourage an inclusive and respectful environment free of harassment.
5. Setting and widely sharing standards of behavior, including sanctions for disrespect, incivility, and harassment. These standards should include a range of disciplinary actions that correspond to the severity and frequency for perpetrators who have violated these standards.
6. Creating policies that support employees during times when family and personal life demands are heightened. For example, stop-the-clock and modified duty policies, which should be available to as wide a group as possible, should be a genuine time-out from work, and should not penalize those who take advantage of the policies.
7. Providing private space with appropriate equipment for parents to feed infants and (if needed) to express and store milk.
8. Creating policies and practices that address workers’ need to balance work and family roles, recognizing that these policies increase productivity and enhance work performance.
9. Limiting department meetings and functions to specified working hours that are consistent with family-friendly workplace expectations.

It is important to acknowledge that most of the research on these strategies has not disaggregated data by gender and intersectional identity (e.g., race, disability status, sexual orientation) and has tended to focus on improvements for White women at research universities.

This page intentionally left blank.