I'll also note that there are plenty of other options that have been discussed and that may be turn out to be one of the root causes, not just the two you are considering.
Then they both fell dramatically through the mid '90s. Women dropped in half, and men dropped by about 36%.
Then men exploded, and by the mid '00s were about 60% above their mid '80s peak, whereas the women were only back to about the same as their mid '80s peak.
Then they both declined, and again the women's decline was more. Men fell about 25%, and women fell about 50%, by the end of the '00s.
It's that explosion of men from the mid '90s through the mid '00s that is the biggest contributor to the percentage of undergraduate CS degrees that go to females dropping in half from its mid '80s peak.
I've not heard any good theories on why there was such a surge of men coming to CS from the mid '90s through the mid '00s.
For master's degrees, the situation is quite different. From the early '70s through the mid '00s, both men and women rose slowly and steadily, with the women rising faster until the mid '80s when they then rose at about the same rate. This resulted in the percentage of master's degrees going to women rising steadily to around 32%. Then there was a slight decline for both men and women, with the women declining slightly more.
For PhDs, there was no decline. Both men and women rose from the early '70s through the end of the '00s (when my data ends), with the women rising faster so the percent of PhDs going to women in CS rose steadily from about 7% to about 22%.
Data if you want to play around with the numbers: http://nces.ed.gov/programs/digest/d12/tables/dt12_349.asp
I completely agree and was just using one possibility to illustrate my point that this topic seems to be crippled by this "Yes tech!" "No tech!" paradigm. Did you have any specific alternatives you were thinking of? Also if you know of any articles on the subject I'd love to read them. I think one possibility is that gender roles have gotten worse over time. In the 70s and 80s we had a lot of women joining the work force as families switched from single income to dual income households. I wonder if the result of things "settling" in the decades that passed was a more pronounced reflection of social values (gender roles) in the work force.
There are several.
For example, it is known that gender bias in groups has a tendency to not only perpetuate itself, but to also reduce participation by the minority gender (even without any hostile action by the majority; it seemingly suffices often to simply feel like an outsider).
Second, the power of stereotypes is strong, even where they are wrong. For example, in Liechtenstein's 1968 referendum on women's suffrage, even nearly half the women wanted to deny themselves the vote [1]. A bad case of Stockholm Syndrome, perhaps, but indicative of how powerful established stereotypes can be. In the early 1980s, working with computers was still often considered to be something more like a secretarial job.
Third, computer science started to pay more and grant higher status, making it more attractive to men (especially in a society where men are still predominantly the breadwinners and women responsible for childcare [2]).
And that's just off the top of my head.
I have no idea what the actual causes are (I am fairly convinced that there are several of them, working in tandem). What I do know is that gender roles are shaped pretty much from birth, so I'm extremely skeptical of a theory that attributes this to preference alone (preference may well play a role, but it's so very difficult to tease out the difference between nature and nurture when it comes to these things).
[1] https://en.wikipedia.org/wiki/Liechtenstein_referendums,_196...
[2] Ironically, IT jobs are often more likely to allow for working from home, making it easier to reconcile a career with childcare.