Right now, you can read a preliminary draft of a paper analyzing women’s representation in subfields of mathematics. The abstract:
We use data from papers posted to the Mathematics section of the arXiv to explore the representation of women in mathematics research. We show that women are under-represented as authors of mathematics papers on the arXiv, even in comparison to the proportion of women who hold full-time positions in mathematics departments. However, some subfields have much greater participation than others.
The authors, Dr. Abra Brisbin and Dr. Ursula Whitcher, are both scientists at the University of Wisconsin at Eau Claire. I interviewed Dr. Whitcher about their methodology, findings, and further hypotheses, and about the additional burden of doing diversity work in the sciences.
Question: Can you discuss how your methodology (and Pierson’s) accounts for nonbinary people? Have you tried running the data through any alternative gender-guessing services like http://genderize.io/ ?
Short answer: it doesn’t. We’re guessing gender based on whether native speakers classified a name as masculine or feminine. Names that are typically considered gender-neutral, such as Pat, are treated as “Unknown” in our analysis. So are unusual names, such as Abra. We’re not using any information about how people self-identify; it’s likely that we have misgendered a few individuals whose names are more commonly used by people of a different gender. We haven’t tried alternative gender-guessing services, though that’s a good idea. (We do have plans for future research involving data sets with clearer identification of individuals.)
I’m not sure that drawing generalizations about nonbinary mathematicians by research subfield would be appropriate, anyway, due to privacy concerns—we’re talking about a minority of a small community here.
Question: Anecdotally we hear that women writing in science fiction and fantasy often use their initials to avoid sexist responses from editors and readers. Can you summarize your findings on similar behavior among women working in mathematics?
Our statistical analysis shows that subfields with a lower representation of women tend to have a greater proportion of authors who use only their first initials. We know from personal discussions that some women in math do choose not to use their given names to avoid biased evaluation. It’s worth noting that most journals in pure math do not have a double-blind submissions process– referees know the names of the people whose papers they’re evaluating– so using a name that does not seem feminine could have a practical advantage.
The number of people posting papers with their initials varies hugely by subfield, though (from a low of about 5% for Logic to a high of about 29% for Quantum Algebra); many authors may just be imitating the custom of their peers.
Question: Is there reason to believe that the set of mathematicians who post papers to arXiv would vary substantially, in its gender breakdown, from the set of mathematicians who don’t?
I see no reason to believe that the set of mathematicians who post research papers to the arXiv today is different from the set of mathematicians who publish research papers in pure mathematics as a whole. We do find that women’s postings to the math arXiv started small, and have grown more quickly than the participation of women in mathematics overall, so I wouldn’t want to use arXiv data to evaluate what women were doing in math fifteen or twenty years ago. Also, the arXiv is less widely used in applied mathematics.
I do think there are good reasons to believe that the gender breakdown of “research-active pure mathematicians” is different from the gender breakdown of the mathematical community as a whole. Many mathematicians identify as “primarily a researcher” or “primarily a teacher”, and this categorization interacts with gender stereotypes in complicated ways.
Question: Sometimes, those of us who do volunteer work on diversity and inclusion issues find it takes time away from the research we got hired/admitted to do, and gets in the way of our other work. Have you had to deal with obstacles like this? Have your institutions, colleagues and mentors been supportive of your work on this topic?
We are fortunate to work at an institution (the University of Wisconsin-Eau Claire) that is highly supportive of efforts to promote diversity. Many of our math department colleagues are active in outreach and community-building, and our department has won multiple grants to support broader representation in mathematics. However, our department’s priority is teaching, not research. If I’d needed to maximize my research prestige, I wouldn’t have started this paper before applying for tenure.
Question: When you look at the variation among mathematical subfields (in their fractions of female authorship of arXiv papers), do you have any hypotheses regarding why we see more women publishing papers in some of these fields than others?
Yes, I have many hypotheses!
- Women are more likely to work in subfields with applications to other fields they care about (and therefore, on average, women are more likely to work on problems with applications to biology than on problems with applications to physics or computer science).
- Different countries have different rates of women’s participation in mathematics, and also strengths in different subfields. For example, the percentage of Italian women in mathematics is much higher than the percentage of US women in mathematics, and there’s also a strong tradition of Italian research in algebraic geometry. This predicts there should be comparatively high numbers of Italian women in algebraic geometry.
- A few good mentors could have a strong positive effect on the participation of women in a subfield (or, conversely, a few prominent but toxic people could drive women away from a subfield). When I chat about our project with other mathematicians, this is definitely the hypothesis they favor, perhaps because personal mentoring relations are very important in mathematical culture.
- There’s a fascinating paper that came out in Science at the beginning of 2015, called “Expectations of brilliance underlie gender distributions across academic disciplines”. Their data shows that fields whose practitioners believe you must be brilliant to succeed have lower representation of women (and that this predicts the low representation of women in math compared to women in statistics). If some subfields of mathematics are viewed as more dependent on unique insight, this could affect the participation of women in those subfields, as well as the participation of women in mathematics as a whole.
- Many women in the US start considering academic careers in mathematics after successful undergraduate research experiences, so subfields with many problems accessible to undergraduates might attract more women. (If undergrads consistently made progress in these areas, possibly these subfields would also be viewed as requiring less brilliance?)
I also asked Dr. Brisbin for thoughts on these questions, and she said, “I agree with Ursula’s responses–I couldn’t have said it better!”
Dr. Whitcher’s bio: “I’m an associate professor at the University of Wisconsin-Eau Claire. I earned a mathematics Ph.D. from the University of Washington, where I wrote my dissertation on algebraic geometry problems inspired by questions in string theory. I spent two years as a Teaching and Research Postdoctoral Fellow at Harvey Mudd College before coming to Wisconsin. I’m interested in experimental and computational approaches to mathematics, and am active in the Sage Math open-source community.”
Dr. Brisbin’s bio: “I was a student and a teaching assistant at the George Washington Summer Program for Women in Math, as well as a teaching assistant at the Carleton College Summer Math Program for Women. I earned a Ph.D. in applied math at Cornell University and did a postdoctoral fellowship at Mayo Clinic before coming to UW-Eau Claire in 2012. In addition to this project with Ursula, my primary area of research is statistical genetics.”
Regarding the “Expectations of brilliance underlie gender distributions across academic disciplines” paper mentioned, the following article goes into some detail as to why this is may not be the best conclusion that can be drawn from the data: