I can't help but think that the article/paper are overstating a statistical effect here by attempting to reduce a multicausal phenomenon to a monocausal one. I

I can't help but think that the article/paper are overstating a statistical effect here by attempting to reduce a multicausal phenomenon to a monocausal one. If you look just at rich western countries, most of them occupy a fairly narrow strip on the graph. There are outliers, but they can go either way; Finland has a very low STEM graduation rate for women, Denmark a very high one (Figure 3b in the paper). Austria and Germany have fairly different outcomes, despite being culturally more similar than other countries that differ less. This is sort of obscured by throwing in many countries that honestly have very different cultures, very different academic systems, and different constraints for people of different genders that seem to be difficult to reduce to just a single number. For a simple example, there are countries where nursing is primarily taught at colleges and others where nursing is primarily taught through vocational programs. Given the gender disparity in nursing, that would affect the gender breakdown for STEM graduation rates (depending on whether you count nursing as part of the "M" in STEM or not, one way or the other). I'm honestly surprised that for most affluent western countries the numbers are not more spread out. If I had to guess – and we really have too few data points for that – I'd say we're simply looking at a normal distribution here, especially as (with the exception of the UK) the larger countries tend to be closer to the midpoint of the 20%-30% range and it's smaller countries that are at the fringes of the range.