The first thing to note, is that both the number of men and women getting CS degrees dropped: the entire field went from ~42k BS degrees to ~24k degrees. There were about 10k less men, and about 8k less women graduating in 1996 compared to 1986. The number eventually rebounded for men, but didn't recover for women until 2003. So something drove men and women out of the field, and women stayed out of it longer.
The next interesting thing is the number of masters and PhDs per gender. Neither of them dropped (so the percentage of BS graduates getting MS and PhD actually increased!). So it was still desirable for men and women in the field to get their masters and doctorates.
So the question isn't why the number of women plunged, it's what drove both men and women out of CS, and what caused it to grow for men? I would probably hypothesis that CS was seen as a risky degree, so while men are generally less adverse to risk (see all the dangerous jobs they do) and got a degree, women choose more stable degrees (though those interested remained, as I think the number of PhD and Master degrees show). Now that CS is now seen as a stable career, we can see there is more interest to join. Of course, that is only looking at the data cited by what you linked, there could definitely be other circumstances.
> I would probably hypothesis that CS was seen as a risky degree, so while men are generally less adverse to risk
I don't know about that. Men consistently study fields with higher income potential than women. The current theory is that when CS started gaining prestige and power, the same thing happened as with all professions that carry power and prestige -- women were pushed out (and by that I don't mean that there was some conspiracy, but society simply started directing women away from CS).
I could definitely be wrong about the risk factor for girls/guys. It was just another possible issue that could be extrapolated from the data
I still feel that saying "women were pushed out" is the wrong way to phrase it. We can see from the data, that both men and women were "pushed out" of the field, with men recovering from the drop earlier. After reading a bit more, it could be that marketing in the 80's (as suggested by the NPR article) negatively effecting both women and men (which was left out of the NPR article) entering the field, but women ended up more effected.
Side note: drops like this have occurred in other fields at different times. Psychology actually ended up losing a lot of men in the 70s, while women increased. I am sure we could find a few other examples as well.
> Side note: drops like this have occurred in other fields at different times. Psychology actually ended up losing a lot of men in the 70s, while women increased. I am sure we could find a few other examples as well.
Absolutely. Many researchers compare those shifts with changing attitudes towards certain professions (say, by counting certain words when they're described in the media etc.), with women participation usually correlated negatively with prestige.
In any case, much of the distinction between masculine and feminine professions is traced back to Victorian times. Of course, similar differences have existed much longer than that and in many cultures, but the Victorians elevated the distinction between gender roles into an elaborate system of social codes (e.g. they had certain rooms in the house more appropriate for men to spend time in, and other for women).
I'm not sure what college has to do with learning to code. If you're waiting until school/college to learn to code, you're probably not going to make a very good coder.
First, college is used as a proxy. The significant drop coincides with the drop in participation in industry, the data is just cleaner.
Second (and unrelated to the discussion, really), I don't know where you get your assumption that learning to code in college is too late. I've been in this business for twenty years and some of the very best developers I know only learned to program in college. If anything, I'd say that not going to school at all and having a weak background in algorithms/mathematics is a much greater stumbling block for some software achievements than not programming before school, but even that is probably a bad generalization. Excellent developers come from all backgrounds.
http://www.nsf.gov/statistics/nsf13327/pdf/tab33.pdf
The first thing to note, is that both the number of men and women getting CS degrees dropped: the entire field went from ~42k BS degrees to ~24k degrees. There were about 10k less men, and about 8k less women graduating in 1996 compared to 1986. The number eventually rebounded for men, but didn't recover for women until 2003. So something drove men and women out of the field, and women stayed out of it longer.
The next interesting thing is the number of masters and PhDs per gender. Neither of them dropped (so the percentage of BS graduates getting MS and PhD actually increased!). So it was still desirable for men and women in the field to get their masters and doctorates.
So the question isn't why the number of women plunged, it's what drove both men and women out of CS, and what caused it to grow for men? I would probably hypothesis that CS was seen as a risky degree, so while men are generally less adverse to risk (see all the dangerous jobs they do) and got a degree, women choose more stable degrees (though those interested remained, as I think the number of PhD and Master degrees show). Now that CS is now seen as a stable career, we can see there is more interest to join. Of course, that is only looking at the data cited by what you linked, there could definitely be other circumstances.