I think it's certainly possible that different career preferences is true and that the current distribution in practice is still skewed. That is, perhaps the "right" distribution is something like 60/40 men/women, in which case both 50/50 men/women and 90/10 men/women would be skewed and pushing people out of careers they'd be happy with.
As to distribution, there are some fundamental biological differences that explain the skew that have been discovered. The largest pertinent difference being aptitude for mechanical reasoning in the general population.
> 3233 young and old adolescents representative of the population
> For the young adolescents, the observed difference in Mechanical Reasoning is equivalent to 10 IQ points, and this difference increases to 13 IQ points for the old adolescents.
> Beyond the observed small average sex difference in the general factor of intelligence (g), the boys' large advantage in mechanical reasoning (MR) must be strongly underscored. This sex difference is not explained by g, and therefore the probable contributions of what is measured by relevant subtests such as abstract reasoning (AR) or spatial relations (SR) can be excluded. The MR difference is still present with almost the same magnitude when the general factor of intelligence (g) is removed. It is also noteworthy that, for the old adolescents, more than half of the variance associated with numerical reasoning (NR) cannot be attributed to g. Thus, we suggest that mental processes captured by these psychological measures are behind the documented male advantage in STEM disciplines
This means that, at the average point in the distribution, the average everday man has a one standard deviation edge in measurements of mechanical reasoning over the average everyday woman. This explains why you would see such a large skew in the field of engineering.
What is "mechanical reasoning"? That paper gives examples of two standardized test questions about pulleys and levers - intuition about that sort of thing is important for some parts of STEM, sure, but I'm not sure it's a skill I've ever used in software engineering, except maybe when plugging in a second monitor on a crowded desk.
There are fundamental biological differences, sure. What do those fundamental biological differences mean for specific career fields and why - and given the mean and standard deviation, which presumably we have, what's the right number, if it's not 50/50?
Sure, one standard deviation makes the ratio like 63% to 37%? That's not what we see in tech job ratios, it's worse than that. Everything past that is unjust and deserves effort to rectify
That's just at the mean. It skews even more once you start heading toward the tails, the high end of which you would expect to see in the engineering labor force.
How many standard deviations past the mean do you need to be to work in STEM?
One way to bound this is to look at the number of people in STEM careers. 6.2% of US jobs were in STEM in 2015 (https://www.bls.gov/spotlight/2017/science-technology-engine...), meaning that STEM employment requires at most being just under 2 standard deviations above the mean—and that only if you assume that society optimizes perfectly to fill all of those careers with the people most suited to them in terms of base biological aptitude. Once you throw in other factors like location and access to education, it probably gets significantly lower than that.
Would you say that all your coworkers at all STEM jobs you've been at have been at least one standard deviation above the mean at aptitude for the job?
It is possible. But it's also possible that it's not skewed, or even skewed the other way. It's as though someone should investigate this question before embarking on programs to shift the distribution to arbitrary places.
> It's as though someone should investigate this question before embarking on programs to shift the distribution to arbitrary places.
If the status quo is 90/10 or worse, I don't agree: 50/50 is the null hypothesis. The null hypothesis can certainly be disproven, but the amount of evidence needed to hold a belief that 50/50 is probably correct is much smaller than the amount of evidence needed to hold a belief that 90/10 is probably correct. I don't think the right approach is "We should let the status quo stay indefinitely until we've convinced ourself of what the right number is" - if we have evidence that the status quo is potentially right, that evidence should be plenteous and easy to gather.
(That said, I do agree that programs focused on improving "the pipeline" and getting more girls interested in STEM are misguided and it would be better to focus on removing the barriers that cause women already interested in STEM careers to leave. But that's a little bit of a different subject.)
I said that someone should investigate it. As in, actively. 50/50 is a null hypothesis. Another perfectly reasonable null hypothesis is the status quo.
For instance, do you know that the proportion of female software engineers is approximately the same as the proportion of females that pass the AP computer science exam in high school? This implies strongly that if there is some effect that's shifting the distribution against women, it happens before high school. That is to say, it has nothing to do with workplace culture or discriminatory hiring practices. Of course, that doesn't mean that those things aren't important - they are. But they likely aren't causing under-representation of women in CS.
