There are a number of issues with this analysis but I found one that I haven't seen anyone else mention: Kaufmann estimates gender ID from the FIRE data without using the survey weights
When the weights are used as designed, the evidence for his core claim disappears
add a skeleton here at some point
3 months ago