"Another thing correlated with gender is brain anatomy. There are a bunch; one I remember is a much larger set of communication paths between the two brain hemispheres. Some of those anatomical differences can even plausibly explain some of the higher level psychological differences that have been observed between men and women through the ages. And some of those might explain women preferring to do other things than contribute to open source."
Hmm. I think you need to think again. Look more carefully at the articles you indicated. Differences in gendered behaviour can be entirely explained on a social basis (well, gender itself is a social construct). If you want to look at this in (exhaustive) detail, I'd recommend 'Delusions of Gender' by cognitive neuroscientist Cordelia Fine, which is really a large collection of studies gathered together from a wealth of sources which look at how women and men act according to gender. Fine debunks the flawed studies, including _all_ the ones that make assumptions about behaviour being biological in origin.
The first part of Fine's book is devoted to demonstrating how ideas about, and identification with, gender can be entirely explained as arising from a socially derived basis, with gender identities and associations affecting people even on a subconscious level.
Well, I never had problems with the first idea (socially derived), but until I read Fine's studies I was very sceptical of the degree to which the second (at a subconcious level) could apply.
Having read Fine's studies, and whilst I would disagree with some of the conclusions she comes to, I found it impossible to argue with the logic and sheer volume of many of her arguments.
Subconscious level response is a result of associative memory learning, where one learns about the world around one by building a complex web of associations which include those about gender (e.g. doll = girl, toy truck = boy, housekeeper = woman, breadwinner = man, nurse = woman, physicist = man).
While there are obviously exceptions to these stereotypical associations, because they are based on, in general, real-world social norms they tend to be assimilated without the effort of conscious evaluation regardless of gender (i.e females absorb the female stereotypes as well as the male ones, and vice versa with the male assimilators).
What this means is that women (as well as men) absorb sexist ideas (against themselves) too (and the ideas don't just stop at gender associations).
Fine's arguments display a wealth of research attempting to demonstrate this.
While I think a few of the studies she quotes are of questionable quality and she is not always consistent, the dubious examples she gives and the (few) mistakes I think she makes are not done in such a way that undermine her other studies and discussions, most being well-constructed with totally believable results, and many studies are excellent.
In Fine's words 'Researchers have shown that our implicit representations of social groups are often remarkably reactionary, even when our consciously reported beliefs are modern and progressive.
As for gender, the automatic associations of the categories male and female are not just a few flimsy strands linked to penis and vagina. Measures of implicit associations reveal that men, more than women, are implicitly associated with science, maths, career, heirarchy and high authority. In contrast, women, more than men, are implicitly associated with the liberal arts, family and domesticity, egalitarianism and low authority.'
Fine also argues that our concept of self is malleable, that we have a 'wardrobe' from which we pick a persona depending on who we are interacting with and according to what social context.
This in itself is not particularly surprising, but what is surprising is the degree to which Fine shows a) to what extent we do this subconsciously and b) that we actually subconsciously change our _self-perception and behaviour_ (including absorbing different stereotypes) when we do it, sometimes to our own detriment.
One way which accounts for the appearance of a 'natural' ability deficit for something is via gender priming. This is where people absorb hints about their gender's capabilities at a particular task/role before they carry it out, and this influences their performance.
It is related to, and sometimes works alongside, 'stereotype threat' where people are aware (either subconsciously, consciously or both) that their social group's stereotype implies that they are poor at a specific task/role and this affects their performance.
(these phenomena do not just affect women, also men and members of other socially-contrived identity groups.)
There were some surprising results from Fine's studies:
--- Even ticking a box marked 'male' or 'female' to indicate gender before a questionnaire is started can reduce a candidate's confidence on 'gender conflicting' tasks, i.e. women tend to rate themselves worse at maths but better at verbal skills in a questionnaire than men do (and men do the opposite), compared to control groups where the 'male' or 'female' box is not included (i.e. the control group is not gender primed).
--- Changing the way a question is presented _actually changes that person's performance/ability_ at a task. There are a number of more detailed examples Fine gives, but here's a couple of the shorter ones (the longer ones are more convincing but clearly I am limited as to time and space on an lwn post :-):
(on mental rotation, a 'men-are-better' task:)
'Another...approach was recently devised by Italian researcher Angela Moe.
She described the mental rotation test to her Italian high school participants as a test of spatial abilities and told one group that "men perform bettter than women in this test, probably for genetic reasons".
'The control group was given no information about gender. But a third group was presented with a downright lie. The group was told that "women perform better than men in this test, probably for genetic reasons".
...In both the men-are-better and the control group, men out-performed women with the usual size of gender difference.
But women in the women-are-better group, the recipients of the little white lie, performed just as well as the men.'
'When, in one study, participants were told that performance on mental rotation is probably linked with success on such tasks as "in-flight and carrier-based aviation engineering...nuclear propulsion engineering, undersea approach and evasion, [and] navigation", the men came out well ahead.
'Yet when the same test was described as predicting facility for "clothing and dress design, interior decoration and interior design...decorative creative needlepoint, creative sewing and knitting, crocheting [and] flower arrangement", this emasculating list of activities had a draining effect on male performance.'
And another, this time performed by 'more than 100 university students enrolled in a fast and difficult calculus class that was a pipeline to the hard sciences':
'...The test packet handed out to each student included some information about the test. Students in the stereotype threat condition were told that the test was designed to measure their maths ability, to try to better understand what makes some people better at maths than others. This kind of statement can on its own create stereotype threat for women, who are well aware of their own stereotyped inferiority in mathematics.
'But added to this, in the non-threat condition, was the information that despite testing on thousands of students no gender difference had ever been found. So what was the effect of this extra information?
'The men and women in the two groups had, on average, all received much the same course grades. You'd expect then, given their apparently equivalent ability, that males and females in the threat and non-threat condition would perform at about the same level on the test.
'Instead, the researchers found that females performed better in the non-threat condition, and this was particularly striking among Anglo-American participants, who generally show the greatest sex difference in maths performance. Among these participants, men and women in the threat condition, as well as men in the non-threat condition, all scored about 19 percent on this very difficult test.
'But women in the non-threat group scored an average of 30 percent correct, thus out-performing every other group - including both groups of men.
In other words, the standard presentation of a test seemed to supress women's ability, but when the same test was presented to women as equally hard for men and women, it "unleashed their mathematics potential".'
--- the more men there are taking a maths test in the same room with a solo woman, the lower women's performance becomes
--- 'Stereotype threat hits hardest those who actually care about their maths skills and how they do on tests, and thus have the most to lose by doing badly, compared with women who don't much identify with maths.'
The reason Fine gives for this is that stereoytpe threat exaggerates performance anxiety in women who are particularly keen to do well - the most productive parts of their thought processes get 'used up' trying to manage the anxiety and stress, and they keep getting distracted trying to manage negative thoughts. Of course this is a phenomenum that can affect anyone in a high pressure situation (regardless of gender), but in the woman's case it is more extreme, and the examples Fine gives point to it having worse consequences for women the 'higher up' they are with their (male stereotyped) specialism.
I've only included one or two study examples herei. Fine has _loads_ more examples, much more thoroughly-treated, pithy ones, to illustrate her points. She goes on to present more and more detailed cases which combine together to overwhelmingly demonstrate that the difference, not just in competence, but even for inclination for gender roles and gendered occupations is _entirely_ socially constructed.
(you can read the book or you can take it from me that Fine's treatment of her subjects is highly detailed, comprehensive, exhaustively sourced and extremely well-referenced)
Included in Fine's studies are some CS related examples but sadly these are amongst the longer ones.
Interestingly she refers to some studies that indicate that the male to female ratio of computer science participants _varies_ geographically from one country to another and is related to that country's society and societal values. It tends to be Western, developed countries that have a very low female percentage of participants, but this does not hold true for, example, the Republic of Armenia:
'In the 1980's and 90's, the percentage of women in the largest computer science department in the country did not fall below 75 percent. Today, thanks to its increasing popularity among men (rather than declining popularity among women), Armenian women still make up close to half of computer science majors (and, anecdotally, their numbers appear to be high in many other former Soviet Republics).'
(In a note in one of her references, as an aside, she also states that a 2002 study showed that 'Computer science is not male dominated in Singapore or Malaysia either' giving the reference for this too.)
Fine suggests (from an interview of young Armenians of both genders) that the comparative gender difference in CS in Armenia is down to a much lower cultural emphasis on one's career as personal fulfillment, a profession instead being seen as a source of financial stability and a good living, and happiness being instead derived from 'their family and friendships'.
This is a further demonstration that participation in computer science is not biologically determined according to sex.
Fine uses many of these studies in conjunction with tests made incorporating sex-related biological factors (e.g. genetic/hormonal/neurological determinants - she gives a whole range of studies carried out not just on adults but also on children, even including young babies, which are really interesting) to clearly demonstrate that these biological factors are completely irrelevant.