It’s no secret that women are underrepresented in STEM (science, technology, engineering and mathematics) occupations. There’s several possible reasons for this: not much encouragement for girls or women to take courses in these subjects, lack of visible role models, and lack of support in the workplace. But a set of studies documented in a new article in Administrative Science Quarterly raises another potential problem for women in STEM occupations: gender-related discrimination in co-workers’ evaluations of their expertise.

The three studies described in the article, by Penn State professor Aparna Joshi, looked at the effect of gender and education on evaluations of the expertise of women working in multidisciplinary teams in science and engineering laboratories. I was really impressed with the thoughtful design of these studies. The research methodology includes measures of  visible “cues” such as the gender of the person being evaluated, which could affect evaluations immediately, and measures of other “cues” such as research productivity, that might only start to affect evaluations as the research output is produced or as the team members get to know more about each other. So the data are more likely to capture the changes in team members’ perceptions of each other across time. The research methodology also recognizes that evaluations can be influenced by the context they take place in, by including factors such as the genders of the evaluator and evaluatee, and the gender distribution in the workplaces where the teams are located.

I’m going to summarize the methodology and results of each the three studies that the paper reports on, but I really encourage you to read the full article to get all the details.

 

The first study involved 215 members of 32 multidisciplinary teams working in engineering laboratories affiliated with a single university. The team members completed two online surveys administered four weeks apart; at least four members of each team had to respond to the surveys for the team’s data to be included in the analysis. On average, 23% of the teams’ members were women, with the actual percentages of female team members ranging from 0% to 66%. The data that were collected, in addition to demographic information, included the respondents’ level of education, and their evaluation of each of their other team members’ research expertise. The data analysis included not looking just at the evaluations, but also at the characteristics of the actor (the evaluator) and the target (the evaluatee).

The analysis of these data showed that female actors’ ratings of female and male targets in their team increased as the targets’ level of education increased – but “highly educated female targets receive[d] the…lowest ratings from male actors in the team. Male actors did not distinguish between less-educated and more-educated male targets; they simply rated all male targets higher than all female targets” (p. 214).

The second study was conducted at another workplace that was structured similarly to the site of the first study. The respondents were 192 members of 31 research teams; the percentage of women on the teams ranged from 0% to 100%, with an average of 48% female team members. The respondents answered the same surveys used in the first study, and the surveys were administered at the same intervals as in the first study. However, the first survey included a set of questions to test “gender identification” – the importance of social group membership to the respondent’s personal identity. (For example, one of the questions in the set was “I am proud to be a member of my gender group”.)

The analysis of these data showed that female actors identified more strongly with their gender than male actors did – but male actors who identified highly with their gender were more likely to rate male and female targets differently. Male actors who identified highly with their gender tended to decrease their ratings of female targets as the female targets’ level of education increased – in other words, the more educated the female target, the lower her expertise was rated. Male actors in general did not differentiate their ratings of male targets based on the targets’ level of education.

The third study collected data from the two workplaces that were the sites of the first and second studies, eight to sixteen months after the initial research at each site. Joshi collected data on the research productivity of the participating teams and team members. Research productivity was measured by the number of the team’s research publications, book chapters, conference presentations, and publications in conference proceedings. Joshi also calculated a measure of the quality and prestige of the journals in which the research was published, a measure of how often the team’s work was cited by other researchers, and the amount of research funding obtained by the team (which could affect the team’s ability to be productive). Additionally, Joshi obtained the number of male and female faculty members in the university faculties the teams were affiliated with.

The analysis of these data showed that the perceived expertise of a team member was a strong predictor of how that person’s expertise was used in the team. In teams with a high proportion of women members (more than 60% women), women with more education were more involved with the team’s research output; the same was true for men in male-dominated teams. The teams with more women members were more productive when their affiliated university departments had more female faculty members.

 

Joshi suggests several theoretical and practical implications of these findings. One is the importance of understanding exactly why male and female team members might rate each other’s expertise differently. Another is the rather troubling evidence that higher levels of education may not help women be more favourably perceived by their male co-workers or team members – and the potentially negative impact of an evaluator’s strong identification with their own gender. However, it also appears that as the number of women in a team’s context increases, so does the participation of women in the team’s output. (Joshi acknowledges that since the study took place only in academic workplaces and looked only at teams working in science and engineering, the results might not necessarily hold true in other types of workplaces.)

I think there are some additional issues that these findings raise. Many workplaces address diversity by focusing on hiring “diverse” employees – but then don’t do much to support those employees afterwards. So retention is often a more critical issue in workplace diversity than hiring. If employees who are in the minority perceive that their performance is being evaluated unfairly – i.e. for reasons other than the quality of their work – that may affect not only their job satisfaction or organizational commitment, but also may make them want to find a job somewhere else. And that may make it difficult in the long term to establish a more diverse work context, which these studies suggest can lead to more team participation by members of traditionally underrepresented demographic groups.

The results of these studies offer lots of possibilities to build on for future research, and I really hope that other researchers will investigate some of the directions that these results suggest. It would be fascinating to see, for example, if the same effects occur in workplaces that aren’t focused on science and engineering, or if the same effects occur in occupations that are female-dominated rather than male-dominated. That sort of information could go a long way towards identifying and overcoming the barriers to more equitable participation in STEM occupations, and potentially in other occupations as well.