Searching for a Diverse Faculty: What Really Works

Faculty renewal and development are inseparable from the process of searching for new faculty. Depending on how institutions carry out this critical practice, they can either stagnate or improve. Institutions aiming to improve will benefit from bringing in new faculty from diverse backgrounds with fresh viewpoints and interests.

More and more colleges and universities have been recognizing that attention to faculty diversity is crucial if they are to honor their institutional values of inclusion and equity and hire outstanding candidates regardless of their gender or ethnicity. Institutions are also recognizing that the benefits of diversity include fostering creativity, innovation, better-functioning groups, and smarter problem solving. In academia, where a diverse faculty brings these benefits to both research and teaching endeavors, it is important to know what practices actually show the most promise. Thanks to a wealth of empirical data from four years of searches for tenure-line faculty, the University of California–Berkeley has been able to zero in on search practices that hold special promise for making a difference.

None of these practices are easy. All may mean doing things in new ways. But each has empirical evidence that points toward its value. Above all, we are now confident that conventional search practices are not enough for consistent success in hiring top women and members of underrepresented minority groups.

Data and Methods

Our study uses data drawn from three sources. First, we used Berkeley’s data from the University of California (UC) online search system, which allowed us to collect information about the gender and ethnicity of our applicants and then to track them through the successive stages of the search process. Second, we drew upon national data from the National Science Foundation’s Survey of Earned Doctorates (2019) to determine the demographics of the pool of available applicants within relevant disciplines or fields. Third, we asked the chairs of search committees to complete a survey stating which commonly recommended practices their respective committees used. We made it clear that we did not have a view about which practices we thought they should use, and we emphasized that we simply needed their help in generating useful data. We were pleased to find that the survey response rate was 91 percent, and the 220 searches for which we received survey data represented 94 percent of the 29,832 applicants for Berkeley positions from academic years 2012–13 through 2015–16.

Our survey asked about 55 practices that fall into four broad areas:

  • position specifications
  • active recruitment
  • reduction of implicit bias
  • departmental commitment to diversity

Using our three sources of data, we compared the demographic composition of applicants for positions at UC Berkeley to the national pool of available applicants and then tracked the demographics of the candidates as searches moved from one stage to the next. The survey data allowed us to look at the demographic patterns for searches that did use a practice and compare them with the patterns for searches that did not use that practice. Through the use of regression analyses and careful review of the data, we assessed the promise of each practice. While our main goal in this article is to present practices that are truly promising, we also describe our less positive findings concerning some practices that are often recommended.

Several caveats are in order. The large number of practices our search committees used in each search means that statistical associations between a single practice and search outcomes will not always be clear. We especially want to stress that studies of this kind do not establish causation; they can only show positive or negative statistical associations.

The Most Promising Search Practices

The really promising search practices we identified involve activities that take place before a search begins or at its very outset. At most universities, the first practice a department takes is thinking about its general priorities, including the research areas in which it wishes to search and the capabilities in other important areas that successful candidates should have. Typically, the next activity would be to appoint the search committee. Finally, one of the search committee’s earliest tasks should be to identify especially promising candidates and encourage them to apply.

Shaping Job Descriptions
Of all the practices we studied, linking job descriptions to issues of gender, race, or ethnicity had the most impressive positive association with greater diversity. A fictional example of using this practice would be to describe a position as focusing on “labor history, including women’s labor history” rather than just on “labor history.”

In figure 1, we show the proportions of women and members of underrepresented minority groups (URMs) in the national pool of available applicants and at progressive search stages, comparing searches that used this practice with those that did not.

Figure 1: Position Description Included Subject Areas or Issues Related to Gender, Race, or Ethnicity

PRFA_19_UCBerk_Fig1a.jpg

The orange lines show percentages for departments that did couple the subject area with diversity issues, and the yellow lines show percentages for departments that did not do this. The top two graphs in figure 1 suggest a clear statistical association between using this practice and increasing proportions of both women and URMs under consideration as searches moved toward completion. By comparison, committees that did not use this practice saw an increasing proportion of white men under consideration as their searches progressed.

Although our data do not allow us to conclude why these patterns appear when job descriptions are coupled with diversity issues, anecdotal evidence suggests several explanations. To return to our fictional example, if women were especially well-represented among scholars of women’s labor, then this search practice would probably encourage more women to apply, thus providing a larger pool of women from which the strongest applicants might emerge as the selected candidate. Indeed, by giving women in traditional subfields of labor history greater confidence that the department thinks inclusively, this type of description might encourage more of them to apply.

Related strategies may be available in many areas where this particular way of shaping job descriptions cannot be used. For example, our past research has shown that women and URMs are better represented in subfields focusing on societal improvement, especially for underserved populations (Goulden, Stacy, and Mason 2009). This can suggest ways of shaping job descriptions. For example, a position might be described as focusing on “architecture and urbanism, including affordable housing” or “infectious diseases, including those affecting vulnerable populations.” While not as dramatic, the statistical associations between diversity and the use of this strategy are clearly positive, as figure 2 shows.

Figure 2: Description of Research Area Included a Focus on Societal Improvement

PRFA_19_UCBerk_Fig2.jpg

In some disciplines—mathematics, for example—it may be hard to see how any of these shaping strategies might be used. Even there, however, a committed department might find creative ways to shape job descriptions. For example, a 2013 report by the National Research Council on the mathematical sciences concludes that their continued academic health depends on their fostering deeper connections with other fields, including engineering, the life sciences, and the social sciences—broad fields within which public or engaged scholarship has an established place. And, of course, in many disciplines, it may be possible to observe that there are sub-areas in which women or URM candidates tend to cluster, perhaps for historical reasons. Job descriptions with explicit links to these sub-areas may help departments increase the proportions of women and URMs whom they can consider.

It is not always easy for departments to think outside of traditional disciplinary structures, but our data strongly suggest that this is an effort well worth making when departments want to be sure they are attracting the broadest pools of strong applicants.

Departmental Priorities
The faculty of a department will usually have recurring opportunities to step back and think strategically about the most pressing needs they hope to meet in future hiring. At UC Berkeley, the exercises that call for this kind of reflection include periodic reviews of departments and annual requests for approval of faculty hiring lines. Some departments may use their annual retreats to discuss their long-term needs.

In addition to considering the research areas in which they hope to search, departments can prioritize other capabilities that candidates may have, for example, readiness to teach a particular kind of course. We found that some departments also explicitly prioritized hiring faculty who will be able to make strong contributions to the departmental goals for advancing diversity, equity, and inclusion. Our data display a pattern for URM candidates that suggests this practice is promising, and we also found that searches that did not use the practice had higher percentages of white men in the later search stages compared with searches that did use the practice. The pattern for women, however, did not show a clear positive statistical association (see figure 3).

Figure 3: Department or Unit Clarified and Prioritized Their Diversity Needs

PRFA_19_UCBerk_Fig3a.jpg

Our study is not able to explain why the outcomes of this practice were different for URMs and women. We speculate that departments find it easier to recognize the absence or near absence of URMs from their faculties than to recognize the underrepresentation of women, including women who are URMs. Clearly, this is an area for further research, and our full report includes more detailed findings and additional nuance (Stacy et al. 2018).

Search Committee Composition
Departments must take many considerations into account when they appoint search committees, including the members’ areas of scholarly expertise. We were curious to explore possible correlations between the committee members’ gender or URM status and the outcomes at successive stages of the searches. We compared committees with at least 40 percent women faculty with those having less than 40 percent, and we also compared committees with at least one URM member with those having none.

Looking at figures 4 and 5, we see somewhat different patterns. Compared with search committees that did not have at least 40 percent women, those that did were more likely to have higher percentages of both women and URMs under consideration at each search stage. Compared with search committees that did not have any URM members, those that had at least one were more likely to have higher percentages of URMs at each search stage. At best, however, a weak statistical association with the use of this practice appeared for women candidates.

Figure 4: Search Committee Included 40% or More Female Members

PRFA_19_UCBerk_Fig4a.jpg

Figure 5: Search Committee Included at Least One URM* Member

PRFA_19_UCBerk_Fig5a.jpg

These are intriguing results, and we discuss them further in our full report (Stacy et al. 2018). The statistical associations we observed encourage us to recommend that departments aim to diversify the demographics of their search committees. Where this could lead to overburdening women or URM faculty members with service duties, departments should take care to offset search committee service with reductions in other service areas or with course relief.

Targeted Outreach
Our research also confirmed the promise of several kinds of targeted outreach that encourage applications from a small number of unusually strong candidates who also would advance the department’s diversity and equity goals. This effort typically will pay off with better representation of women and URMs in later search stages, rather than in the applicant pool, where a small uptick in applications from targeted women or URMs would not make a significant difference. While the search committee members or chair would often be the ones to reach out to women and URM candidates, other departmental faculty might be asked to make calls or write emails.

Figure 6 provides data concerning searches that did and did not include emails or phone calls from departmental faculty to possible women and URM candidates, encouraging them to apply. As would be expected, the first bump in percentages of women and URM candidates appears at the short-list stage.

Figure 6: Search Committee Called/Emailed Possible Candidates with Diverse Backgrounds and Encouraged Them to Apply

PRFA_19_UCBerk_Fig6a.jpg

One search committee chair commented, “This was a remarkably successful exercise; most people we emailed applied.” Search committees can do considerable homework ahead of time to generate the lists of scholars to whom they want to reach out. Our data did not support traditional practices such as getting the names of new PhDs from a few leading departments or asking those departments to encourage women or URM candidates to apply. Such practices are unlikely to be as effective as, for example, seeking out “underplaced” scholars who might not even be thinking of applying for jobs. Also helpful to committees is reading articles in journals that publish work connected with the subfields the department is using to shape its job description or reading traditional journals with an eye toward strong early-career scholars who are not yet established.

Another way to identify individuals who should be encouraged to apply is to tap an established pipeline. At UC Berkeley, departments had information about the system-wide UC President’s Postdoctoral Fellowship Program (PPFP). This program provides postdoctoral fellowships to “outstanding scholars in all fields whose research, teaching, and service will contribute to diversity and equal opportunity at UC.” Information about past and current fellows is provided in an online directory that is publicly available. There, search committees can learn about highly qualified individuals in an exceptionally broad range of disciplines, with fellowship dates ranging from 1996 onward. Using a pipeline like this proved to be a very promising search practice.

Targeted outreach can be time-consuming, especially for committees that do their homework ahead of time. But the results can be gratifying.

Popular Practices Not Showing Clear Promise

Our data did not provide support for using three search practices that often are recommended or mandated: reviewing comparative data, taking steps to counter implicit bias, and requiring applicants to provide evidence of their commitment to diversity. Still, we believe the success of these practices may depend on how they are implemented, and they deserve especially close study going forward.

Using Comparative Data
Some of our search committees used one or more kinds of comparative data, including their department’s demographic hiring patterns compared with those of peer departments in the same field; their department’s own hiring patterns over time; and the national availability and applicant pools for their recent recruitments.

Our data for using comparisons with peer departments generated the most striking set of negative statistical associations that we found anywhere in our study. These data show a negative statistical association between using the practice and diverse outcomes at every stage of the search process, and they show a positive association between using the practice and hiring white men. In speculating about the reasons for these results, we wonder whether this practice might lead departments to believe they are doing as well as their peers. Such a perspective might, in turn, mean that departments see no reason to invest time and energy in using new practices that might help them diversify their applicant pools and short lists.

We found that the practice of using data about the department’s own hiring patterns over time also had a negative statistical association with gender and URM diversity at some search stages. We speculate that this may reflect a natural reluctance to cast a critical eye on past searches that resulted in the hiring of current colleagues.

The practice of looking at national pools of available applicants did not yield negative associations, but neither did it yield any clearly positive associations. Here we note that faculty members are often uninterested in national pools because they regard their needs or expectations as exceptional. Thus, it may be that search committees felt that considering more women or URM candidates at various search stages would entail slighting important dimensions of their search goals.

Making Efforts to Counter Implicit Bias
Our survey asked whether the search committee discussed implicit bias and, in a separate question, whether it encouraged committee members to take campus training to counter implicit bias. Our data for both practices showed no notable differences between searches that did and did not make one or both of these efforts to counter implicit bias. Thus, the patterns displayed by our data are in line with those found in some other studies (Dobbin and Kalev 2018), and they point to the conclusion that these training practices, at least as they were carried out at UC Berkeley during the four-year period of this study, do not show clear promise.

These results are perhaps unsurprising given the lack of convergence in the research literature on conclusions about implicit bias, training designed to counteract it, and effects on hiring from a diverse group of candidates. Different kinds of training currently are being offered at campuses across the UC system and across the United States. Classifying and studying the different types of training may reveal that some are promising while others are not. For example, some anecdotal evidence suggests that training by faculty peers is more likely to be associated with greater diversity at various search stages than training by non-faculty individuals.

Requiring Applicants to Demonstrate Contributions to Diversity, Equity, or Inclusion
This practice specifies that in assessing applicants, departments will consider an individual’s promise or accomplishments in making contributions to diversity, equity, or inclusion. While there are many ways to do this, a popular one is to require applicants to provide “equity and inclusion statements” with their other application materials. Versions of this practice are increasingly being recommended and adopted nationwide, and indeed most UC campuses (though not UC Berkeley at the time of the study) now require diversity, equity, and inclusion statements from candidates.

At the application stage of the process, our data do show some differences between searches that used this kind of practice and those that did not. Beyond that stage, however, we found no clear and consistent patterns in the data that would suggest a positive statistical correlation between this practice and diversity.

We suspect that the promise of this practice depends upon how search committees actually assess and use the evidence that their candidates present. For example, one department that requires a diversity statement might treat it as only a single document among many in a dossier; another might treat it as a tiebreaker; and yet another might put candidates on their short list only if their statements provide strong evidence about their ability to advance diversity, equity, and inclusion. In the future, we hope to be able to look separately at different ways of implementing this general kind of practice to see whether positive associations with diversity emerge for some versions of the practice and not others.

Conclusion

Departments and hiring search committees invest large amounts of time and effort in identifying the scholars they wish to recruit. We hope the study we conducted will help to identify practices that can move departments closer to their goals for a strong and diverse faculty. In this way, institutions will be able to accelerate their processes of renewal and development. 

 

Readers who would like to see the authors’ full report can find it at
http://www.ofew.berkeley.edu/sites/default/files/searching_for_a_diverse_faculty-_data-driven_recommendations.pdf

 

References

Dobbin, Frank, and Alexandra Kalev. 2018. “Why Doesn’t Diversity Training Work? The Challenge for Industry and Academia.” Anthropology Now. October 27, 2018. http://anthronow.com/uncommon-sense/why-doesnt-diversity-training-work.

Goulden, Marc, Angelica Stacy, and Mary Ann Mason. 2009. “Assessment Denied: The National Research Council’s Sins of Omission.” Chronicle of Higher Education. August 31, 2009. https://www.chronicle.com/article/Sins-of-Omission/48233.

National Research Council. 2013. The Mathematical Sciences in 2025. Washington, DC: National Academies Press. https://doi.org/10.17226/15269.

National Science Foundation. 2019. “Survey of Earned Doctorates.” https://www.nsf.gov/statistics/srvydoctorates/#sd.

Stacy, Angelica, Marc Goulden, Karie Frasch, and Janet Broughton. 2018. Searching for a Diverse Faculty: Data-Driven Recommendations. University of California–Berkeley. http://ofew.berkeley.edu/sites/default/files/searching_for_a_diverse_faculty-_data-driven_recommendations.pdf.


Marc Goulden, Director of Data Initiatives, Office for the Faculty; Angelica Stacy, Associate Vice Provost for the Faculty and Professor of Chemistry; Karie Frasch, Director of the Office for Faculty Equity and Welfare; and Janet Broughton, Professor of Philosophy Emerita, Vice Provost for the Faculty Emerita, and Executive Dean of the College of Letters and Science—all of University of California–Berkeley

Select any filter and click on Apply to see results