> like in countries with better gender equality, where we see more unequal distributions of workers in different fields [0].
I wish the correlation/causation and replication crisis crowd would update their priors on that paper.
They published a correction (notably after all the press coverage) because they weren't measuring what they said they were measuring, but it was still a pretty lousy study, and of course demonstrates nothing about causation. [1] has a commentary, but maybe other people have better coverage.
> What does Stoet and Geary’s propensity ratio measure? Worldwide, women earn more tertiary degrees than men. In Algeria, 62.7% of tertiary graduates, and 53.55% of STEM graduates, are women. Yet Stoet and Geary reported a value of 40.7 for Algeria.
Propensity is when you normalize the value to the base rate of graduates for men and women. 53.55 / 62.7 is the women in STEM divided by women getting degrees, 46.45 / 37.3 is the men in STEM divided by men getting degrees. This gives you the rate of stem degrees normalized by how many of each gender gets tertiary degrees.
So basically they wrote that paper not knowing anything. But I can see why a group of gender studies professors would write that piece anyway for political reasons, as this paradox is very uncomfortable for them.
Yet these adjusted figures more of less reinforce the thesis of the original theory. Algeria and Poland, more conservative countries, actually had higher adjusted representation of women in STEM. Luxembourg has fewer. This is important nuance in statistics, but it looks like it's reinforcing the thesis that more equal societies see greater gender disparities in employment.
Do you really think the point of the paper is they're suggesting an arithmetic error was made? The first section of the paper is literally explains the propensity measure as background so they can discuss it.
> But I can see why a group of gender studies professors would write that piece anyway for political reasons, as this paradox is very uncomfortable for them.
Oh yes, professors of gender studies, anthropology and psychology should stick to their soft sciences and not critique hard science written by *checks notes* professors of psychology who write pop science books and articles.
If you want to measure if a woman chooses to graduate in STEM or some other field, you have to measure the women who got that choice and not the women who weren't given that choice. Their example "reversing" the result changed it to also include all the women who can't go to college at all due to gender discrimination and thus didn't get a choice on the matter, that doesn't help us see whether those women would have gone into STEM or not at college.
That is why you have to compare women who choose STEM to women who choose to study something else. An anti equal society would be expected to push women into feminine fields and away from masculine fields. But as the study showed that didn't happen, gender unequal societies seemed to do the opposite. All this counter study showed is that you can mask this bias by also weighing in that unequal societies also bars many women from entering college at all.
I wish the correlation/causation and replication crisis crowd would update their priors on that paper.
They published a correction (notably after all the press coverage) because they weren't measuring what they said they were measuring, but it was still a pretty lousy study, and of course demonstrates nothing about causation. [1] has a commentary, but maybe other people have better coverage.
[1] https://scholar.harvard.edu/files/srichard/files/richardson_...