So, how bad is the gender gap in CS?

29 May 2011

Via Sociological Images (who in turn got it from Kieran Healy) I stumbled across the eye-opening chart below:

Percentage of PhDs awarded to women in the U.S. in 2009, selected disciplines

(Click the image to get a larger version.) See that? That's the percentage of U.S. PhDs awarded to women in 2009, broken out by discipline. Actually, that's only a small fraction of the disciplines included in the original data set-- you can go here to see the original data, or here to see a similar but much larger chart for all disciplines.

(For those of you who can't see the chart, it shows about 20 disciplines, sorted from most female-skewed to most male-skewed. At the top are things like Psychology, Linguistics and Sociology, which are 60% to 70% female. In the middle are things like BioChem, Statistics and History with 40% to 50% female-PhDs. At the bottom three are (in order) Engineering, CS, and Physics at around 20% women PhD recipients. And there's actually a sharp break between those three and the next three: Philosophy, Mathematics and Religious Studies, who come in at 30%. So not only is CS second from the bottom, it is materially worse than the seminary.)

I honestly don't know what to make of this chart, other than a few shallow thoughts:

  • An 80-20 split among CS PhDs feels right, actually, based on anecdote and subjective impressions. But I didn't realize how bad that gap was compared to other disciplines.
  • The full chart is worth a look, as it breaks out these disciplines into sub-disciplines. It shows there's actually a lot of variation between sub-fields of mathematics, for example, and for physics. But these may be artifacts of small sample-sizes-- I'm not sure if the absolute numbers at this level of granularity are large enough to support any conclusions. But the chart does support another impression of mine: Theory of Computation (my particular field) is doing much better than general CS. Well, relatively speaking.
  • As I've said before, though, I don't care about this issue for social-justice reasons. I care about this issue because I want more people to go into computer science research. So what I'd really like to know is whether the gender gap correlates in any way with overall population of researchers / PhD students. For example, Mathematics used to have a gap like ours not too long ago, I'm told. They still have work to do, of course, but did the number of PhD students in Math go up as the gender gap closed?
  • While I'm thinking about math, I like how this diagram refutes the argument that women don't go into CS or engineering because women can't hack math. Bull crap. Look at the numbers for Statistics. Look at the numbers for BioChem or Economics. Heck, look at the numbers for Mathematics. They're all doing better than CS. Is anyone seriously going to argue that CS is more math-heavy than Math?

Religious studies !=

Religious studies != seminary. (I know a woman getting a Ph.D. in BU's religion program, hoping for a professorship, not the least interest in ministry.) OK, nitpicking over, time to finish reading the post :).


Yeah, that was mostly tongue

Yeah, that was mostly tongue in cheek. But I still suspect that the gender gap in the field is caused by the requirement by many religions that officiants be male.


Was math's discrepancy really

Was math's discrepancy really that bad? I was under the impression that the ratio in math historically had been *better* than that in most STEM disciplines (though maybe that was just in the undergraduate world, in which case there's interesting pipeline questions). Ah, yes, this chart confirms that impression -- http://www.nsf.gov/statistics/databrf/sdb97326.htm . And also reminds me that the ratio in CS sucked a lot less in the 80s, and what's up with that?