I had a discussion today with several other grad students over whether math grad school ought to give special benefits to women who want to take time off for children, i.e., ought they to suffer less for an extended child-raising absence than someone who takes an extended vacation. Despite my belief that in some sense such a procedure is unfair and my doubt over whether this would increase the amount of math done I was convinced that it would improve the happiness of mathematicians by attracting more women to the field. Since psychologically it would not be viewed as a cost or unfair at worst it would simply be trading some mathematical accomplishment for happier mathematicians. Something I’m entirely behind.

However, what initiated this discussion is the question of whether increasing the number of female professors is an effective way to increase further female participation in fields like mathematics. I argued that most of the harm is done far before college and thus the ratio of college professors only has a small direct effect on the percent of females pursuing advanced degrees in mathematics and related fields. Well here is some hard data from the BPS cohort study to back up my point. If you will take a look at page 98 it shows that the difference in male vs. female enrollment in hard science courses was essentially constant from freshman to senior. If girls were getting discouraged by the gender gap among their professors one would expect them to be most equal with the boys the first year and gradually decreasing over time.

Ultimately this data isn’t entirely compelling but it is suggestive of the conclusion that primary factors affecting female under-representation in the hard sciences and math occur before college. At the very least it suggests anyone trying to justify a policy based on a role model effect from female college professors has a burden to meet. Of course it is still plausible that some role model effect occurs before college but this would suggest a very different set of policies than a role model effect in college. If the role model effect is primarily based on famous household names then a goal of increasing the number of women in hard science would want to do its utmost to prevent any exceptional woman from taking time off to raise a family especially if women who raise families are less likely to continue in the profession than people who don’t have children. Alternatively if it is based on high-school teachers perhaps an effective means of encouraging more women in math and science would be to discourage all but most smart women from actually doing research and instead guide them into teaching.

Obviously I’m not endorsing either of these two policies, especially the later one. However, I wished to illustrate how slightly different mechanisms could recommend very different policies. Besides responding to the people I was arguing with this is really the point of this post. Whether the statistic I cited turns out to be misleading or not it illustrates the fact that the sort of feedback loops used as a justification for artificially increasing diversity can behave very counterintuitively. Thus we should find no plan whose justification rests on an indirect increase in diversity (like affirmative action) plausible until we have examined the statistical facts ourselves.

To reassure the suspicious out there I just searched the web until I found some credible data which would suggest something about the percent of women lost to math and science in college. I didn’t look until I found something that illustrates my point. But if you want to look at an interesting/scary/huh paper I found along the way about women in CS you can. Since it basically argues that we need to make the CS culture more normal to attract women I thought it might interest some people. Though I’m not sure how I feel about it.