What happened to the women?

Chemjobber recently posted a link to the 2012 NSF statistics for physical science postgraduation work & study. Both Chemjobber and the Inside Higher Ed article he links to looked at the job prospects for new PhDs by field. That data table is also broken down by gender, though, and I think there's a story hidden in there.1


First I looked at my home territory: Chemistry. More men than women earned doctorates in Chemistry in 2012, so a straight comparison of numbers didn't seem too useful. Instead I compared the relative frequencies of men vs women in each of the four categories: "definite postgraduation study," "definite employment," "seeking employment or study," and "other" (a combination of people who are studying something else or have no plans). Here's what I got:

Blue = Men, Red = Women

Blue = Men, Red = Women

It appears that women are having more trouble finding work (either study or employment) than men. It's a difference of about 7%.


What about in Physics? Chemistry is split roughly 60/40 male–female, but Physics is more like 80/20. Does the same gap exist there?

Blue = Men, Red = Women

Blue = Men, Red = Women

Nope. The difference is just 1%. By my totally scientific "let's just eyeball it" measure,2 that's not significant.

Computer and information sciences

Computer science also has a roughly 80/20 gender split, but it has a similar gap as chemistry. The main difference seems to be that Comp. sci. PhDs go get jobs, and chemists go on for postdocs.

Blue = Men, Red = Women

Blue = Men, Red = Women

Looking for patterns

Once I started digging, I just kept on going. I looked at the data for life sciences and engineering, and I kept seeing the same thing, over and over: women were either looking for work with the same frequency as men (Geosciences, Mathematics, Physics, Health), or they were seeking work with greater frequency (the rest). It doesn't seem to matter if women are in the minority (e.g. Physics, Comp. sci.) or the majority (e.g. Bio. & Biomed., Health), either. The only apparent exceptions to this were in some engineering fields with so few women (< 200) that the "let's eyeball it" measure fails.3

Top: PhDs awarded, by field &amp; gender  Bottom: PhD recipients seeking study or employment, by field &amp; gender  (Click for big)

Top: PhDs awarded, by field & gender

Bottom: PhD recipients seeking study or employment, by field & gender

(Click for big)

What's going on?

What happened to the women in Chemistry? Why didn't it happen to women in Physics or Math? With no evidence but anecdata, I'm not about to draw a solid conclusion, but I can think of several possibilities.

Physics and Math are notoriously low on women. Perhaps because of that there are more efforts to get more women in those fields, versus Chemistry, which is nearing an even male–female split and may not see the need for continued recruitment of women.4

Computer science, though lacking women, doesn't match that pattern. There I'd guess that the much higher frequency of postgraduate employment makes a difference. Most Comp. sci. PhDs got jobs, not postdocs, and I suspect that differences in networking could be at play. Comp. sci. PhDs are overwhelmingly male, so it seems logical that these men would mostly know other men, and when it comes to passing along job openings or making hires, they'd hire men. This might also explain some of the patterns in Engineering, too.

That still doesn't fully explain Chemistry, neither does it explain Ag & Nat. Res. or Bio. & Biomed., where the gender difference is smaller, but women are still looking for work with greater frequency. Is this just old-fashioned bias? Is there something else going on? If the NSF numbers are from any point in the year, maybe women graduate later in the year than men, so less time will have elapsed for them between receiving the PhD and being counted in the survey.

Of course, there are plenty of other possibilities. This is just one year's data, and perhaps this has been changing (dare I hope improving?) over time. Whatever the reasons, though, it has me a little worried that in so many fields women are looking for work at higher rates than men. What gives, guys?

Other questions

The other thing I'd love to know is how this breaks down by race & ethnicity (in addition to field and gender). A physicist friend of mine has been looking for faculty jobs, and she's commented a number of times on the utter whiteness (and maleness) of so many physics departments. I didn't see numbers detailing both gender and race, but I hope that non-white, non-male PhDs are finding work (postgrad study, employment, etc.) at better rates than white guys because there is a lot of ground to cover before we reach equality in representation. I think it's really important that the PhDs coming out of graduate school look like the students they teach and the public they encounter.5 We need diversity to become so common it is no longer remarkable and surprising, but rather expected as the norm.

If you'd like to see more, I put together charts of postgraduation status for all6 physical sciences, life sciences and engineering fields. You can also dig through the data yourself. The NSF site has Excel files for each table, and apparently an interactive version is coming in January. There's tons more to learn.

1: I should note that the comparisons I'm making count only the respondents who provided information on their postgraduation status. See note b in any of the following tables: 57, 59, 63.

2: Yeah, maybe not scientific, but the criteria for the eyeball test are simple: (1) are the data sets "large enough" and (2) do the values look different? It turns out that I gauge differences >5% as significantly different. It also turns out that it's a decent approximation: run the pairs through a two-sample z test, and you'll find that the pairs I'd peg as different reject the null hypothesis (i.e. % men seeking = % women seeking) with confidence intervals >95%.

If I've just butchered Statistics, please point me to a place where I can learn more. My background in stats is much weaker than I'd like.

3: When you have only 35 women, as in Aerospace, Aeronautical & Astronautical Engineering, it takes only two women to make a 5% difference in study/employment status.

4: Interestingly, male and female PhDs in Math seem to have made different choices. They have the same frequency of seeking work, but more men went on for postgrad study, whereas more women got jobs.

5: Right now, the are so few black women in Physics that someone keeps a list. It has 56 names as of 2007; that's fewer people than were in my Chemistry cohort at UM when I started. I am not sure if these 56 are just PhDs from US universities, but it's still an incredibly small number. For comparison, the 2012 NSF survey includes 391 women who earned PhDs in Physics and have a postgraduation status reported. Those 56 names are not one year, they go back to 1972.

I was pleased to learn that UM appears to be one of the more supportive universities in the list, having been the alma mater of Willie Hobbs Moore (first black woman to get a Physics PhD) and four other women since. Only MIT has more alumnae on the list.

6: Okay, technically not all of them. I left out the engineering fields with <200 women.