This is a good point worth keeping in mind. We're talking about a statistical property of groups of people, and one that is relatively uncommon in both groups.
The interesting thing about medicine and law is that they are professions that deal with human needs and social relationships. As formal barriers were removed, the percentage of law and medical degrees awarded to women steadily increased to parity:
The graph doesn't show the last few years, but I believe medicine is over 50% women now.
Compare this to computer science. I can't find easily find data for Ph.D.s or master's degrees, so here is bachelor's degrees over roughly the same period:
(NB: technically enrollment rather than completion.) That looks to me pretty clearly like a thumb lifting off the scale.
I do not think the quantity of sexism in CS is zero. Yet from my experience of CS and law, I have a hard time believing that there is more bad behavior, (on the order of a 200-300% difference) among computer scientists than among lawyers. Maybe I'm wrong about that?
In any case I think comparisons to medicine and law actually raise more questions than they answer.
At some level of scale medicine and law are about human and social needs, but at that same level of scale, programming is about social and consumer needs. Yet by that chart you posted, the level of CS degrees awarded to women in the era of Twitter is half of what it was in the era of Lotus 123. And what about accounting? Big 4 accounting firms are close to parity. And I'm not sure how different the professions are at the lowest levels. Poring through a spreadsheet trying to tie-up a number that's off isn't much different than poring over a core dump. Yet half the people who do the former are women.
Your first point is a non-explanation. If the conjecture is that the preference for CS is based on inherent factors, that doesn't explain why the preference ratio would change over time. Did the nature of programming change in 1985 and 2003 to make it more attractive to men?
Disclaimer: the sources below are controversial. The below is not necessarily an endorsement of these conclusions, but an attempt to address them on their own terms.
Regarding your second point. At the ranges in question, the male-female disparity is not enough to explain the observed results. I'm going to rely on SAT Math data, because that's more rigorously studied than what you posted. The male-female disparity among people with perfect SAT Math scores is less than 2-1: http://www.aei.org/publication/2013-sat-test-results-show-th.... So that might explain why only 40%+ of math majors are women. It doesn't explain why less than 20% of CS majors are women, or why that ratio has fallen by half even as the field has become less mathematically rigorous.
Also, there is quite a lot of evidence that women outperform their SAT Math scores relative to men: http://esd.mit.edu/Headline/widnall_presentation.html ("He found that women outperform their predictions. That is, that women perform better as students than their math SAT scores would predict. The effective predictive gap is about 30 points.") It is interesting to note that men also outperform women at the upper range of MCAT and LSAT scores, by similar margins. Yet, differences in observed performance in medical and law school by gender are slim to non-existant, and those professions have an even number of men and women, at least at the degree and entry level.
medicine is over 50% women now (everyone at every level) or doctors are over 50% women now? If it's the former, I'm wondering if it has more to do with the industry becoming more equal or if the demand for nurses has gone up much faster than the demand for doctors.
This is not to say that doctors are more likely to be men than women (I happen to think it's roughly equal based on my own experiences), but that nurses are far more likely to be women because nursing is a profession that has been overwhelmingly female, and one in which men are discouraged from participating.
Worldwide, it varies from country to country, but the are more male physicians than female physicians in every country except Algeria, Cabo Verde, Czech Republic, Estonia, Guinea and Mongolia.
All jobs require some dealing with people and some dealing with things. Does that imply no difference between the fractions of the day that software engineers and social workers spend systematizing vs. empathizing?
But let's examine the point more closely: what is the name for the department that is explicitly in charge of managing human needs and social relationships at a large software development company? Who tends to prefer those jobs?
I am cringing. Gould was a classic case of projection, having been guilty of everything he accused his opponents of: misreading one's opponents, proneness to ideological bias, and experimental technique so sloppy that deliberate fraud starts to look like the simpler explanation. (http://www.plosbiology.org/article/info%3Adoi%2F10.1371%2Fjo...)
There are other bones to pick with Gould, but these are the ones that make him impossible to read as an interested layperson without personally verifying every sentence.
What is your preferred explanation? That a doctor suspected ebola but, because Duncan was an African without insurance, thought it better that he walk around for four more days?
Girlfriend's family is from Flanders, the Dutch-speaking region of Belgium. Her sister speaks fluent English despite very modest formal language training. (On the order of the n<=3 years of Spanish or French that most American students take without much to show for it.) She did, however, watch a lot of subtitled TV.
There are about 20 million Dutch speakers, vs. perhaps 400 million English speakers in the US and UK, which have large markets for TV shows and movies with correspondingly large budgets. For the average Dutch speaker, the subtitled offerings in English are apparently much richer than the home-grown stuff.
This, at least, was how it was explained to me by multiple independent Dutch-speaking Belgians who all had an embarrassingly firm command of idiomatic English.
This pair of rhetorical questions actually has a non-rhetorical answer, and it is far more interesting (though less precious) than the one implied here.
It is more correctly circular reasoning. The psychometric measure g does not break this circle at all. I will never believe in "intelligence tests" until this circular reasoning can be broken.
I do not know why the positive manifold is as it is. But it still seems like it is a precise measure of an inaccuracy.
What I am getting at in a roundabout way is that I cannot find any credible definition of intelligence that is not dependent on the these tests. If intelligence were able to measured some other way that corroborated these tests, then I will be more likely to believe them.
Two points about this. First, a consensus developed over more than a hundred years of study requires a little more than an off-handed dismissal-- do you understand _why_ a general factor of intelligence is acknowledged by practically everyone who has studied the issue, or why it is impossible to devise a test that purports to measure cognitive ability but varies independently of that general factor?
Secondly, I gladly accept your reductio of a "general factor of fitness". If you restrict yourself to the range of world-class athletes, I grant that performance in archery will correlate negatively with performance in sprinting. This is just Berkson's paradox, or the "restriction of range" phenomenon. If you consider the entire population, however, I would be very surprised if a "general factor of fitness" didn't fall right out of a battery of athletic tests. In general, people who are faster than average will also be stronger than average and probably more dextrous too, and it makes sense to call that factor "physical fitness".
Intelligence and fitness are two very different things. Fitness has a lot to do with make up of muscle fibres in the body, fast twitch (anerobic) vs slow twitch (aerobic).
This is why marathon runners and weight lifters have very different body types , training a lot in one discipline will make you worse at the other. Genetics plays a role in which type of muscle you can most easily develop.
Intelligence simply doesn't work the same way, learning to draw well won't make you worse at computer programming for example.
Yes, maxing out one dimension of performance may degrade it in another. This is just Berkson, again. My point is that along the entire range of variation, all reasonable measures of athletic performance are likely to be positively correlated. If you doubt this, consider the following thought experiment:
1. Pick an arbitrary athletic contest: 100m dash, bench press, obstacle course, marathon, whatever.
2. You pick a member of the population at random.
3. I pick a member of the German national football team at random.
4. If Joe Schmoe wins, I pay you a dollar. If Dieter Schmieter wins, you pay me a dollar.
5. Repeat until convinced.
Who do you think will win more money in the long run? Will the Germans tend to win because of their high general fitness, or will it be a toss-up because the Germans merely have "high soccer fitness" which does not help (or perhaps even disadvantages) them at non-soccer tasks?
That is what I mean when I speak of a "general factor of fitness". Over the whole range of population, people who do better at one athletic task will tend to do better at any other. If you take a sample from the population, measure their performances on whatever athletic tasks you choose, and drop the results into PCA, you will find one factor which is positively correlated with every test and explains, I'm guessing, at least 3/4 of the variance. What happens at the extreme tails is already acknowledged, and does not bear on this general point.
I tried to find data for non-elite-athletes taking something like the NFL combine and couldn't find anything, but there, at least, is a testable prediction for you.
My only point about a "general factor of fitness" was that I am unfazed by the comparison to IQ, because physical fitness is a perfectly reasonable concept that captures something real in the world. So too IQ. If you think the analogy is inapt, remember who brought it up :)
You're comparing a trained athlete against a person who is untrained. It is not surprising that fitness related tasks are very amenable to training in a way that is not quite so clear with intelligence. A more interesting comparison would be between a sedentary Dieter Schmieter who worked as a computer programmer (or a Dieter Schmieter who had trained purely for strength) and a random member of the population who was trained by Dieter Schmieter's coach for 10 years.
Amongst the general untrained population you will see very large variance between performance on different fitness tests. Lots of broadly built overweight people who are very strong despite never having done any weight training but can't walk up a flight of stairs without breaking a sweat and skinny people who can run well without much training but barely lift anything. So I would doubt that there would be a strong general factor for physical tasks.
