Most of you are likely aware of the 2005 American Institute of Physics report on women and minorities that painted an every-so-rosy picture of women in physics...."Good news everyone! not only does the pipeline not leak, but women are actually being preferentially hired over men at every stage of the career ladder!!!".
I have a long story to tell about the fundamental flaw I found in that analysis, and the total lack of response I got from the author (until I climbed the food chain at the AIP to complain). The perceptibly snotty reply I eventually got from the author leads me to believe that the findings of the report reflected not so much the true status of women in physics, but rather the desire of the author to get lots and lots and lots of publicity for herself and for her analysis (her analysis is pretty much the only one out there that paints a rosy picture for women in physics).
But that story is not what this post is about.
This post is about performing that kind of study yourself. You can do so quite easily with the NSF databases. From their database of demographic information of science PhD's you can get the years-from-BSc-to-PhD distributions for males and females. From their degree completion surveys you can determine how many males and females graduated with physics BSc's and PhD's each year in America. You can also just look at degrees earned by males and females who are US citizens.
From the BSc data, you can run a simulation that randomly selects a number from the years-from-BSc-to-PhD distributions for the males and females, and determine the "predicted" number of PhD's we would expect to see in future years.
The AIP report is correct in one respect...around 1993, the number of PhD's awarded in physics to females each year exceeded this predicted value. It looks like females are doing just great at moving on to the next level! Even better than the boys!
But wait, is that the whole story? Absinthe thinks not. Absinthe did the same analysis, this time looking at BSc's and PhD's awarded to US citizens. Over 95% of physics BSc's are awarded to US citizens. But only around 50% of physics PhD's are awarded to US citizens. When we run a simulation that predicts how many US females should be getting PhD's and compare it to the actual number of US female citizens getting PhD's, we find that American females are under-represented at the doctoral level by between 25% to 30% below the expected relative fraction per year.
To summarize, the only reason the physics pipeline for females in America appears not to leak is because foreign women are entering the pipeline at the doctoral level.
Doing the same kind of simulation for US males shows that they are represented at the doctoral level by around 5% over the expected relative fraction per year (this makes sense...when US females are underrepresented in the US citizen doctoral gender pool, obviously the males are over-represented in the relative mix of genders).
Questionnaire data (also available off the NSF website) tell us that only a very small fraction of people receiving physics BSc's plan to continue their studies in another country. Thus it is highly doubtful that the American females with BSc's disappeared off to some foreign country to get their PhD. No dearhearts, instead they leaked. And they leaked big time.
Why are the American females leaking? Why, why, why? What are other countries doing right that helps them to produce enough female physics BSc's that they have more than enough to generously share with America. What is the American system doing wrong?