This is definitely very interesting data. But there are several issues with the analysis: > Using a statistical measure called r-squared, we determined that th

This is definitely very interesting data. But there are several issues with the analysis: > Using a statistical measure called r-squared, we determined that the male/female mix within a career is on average about 31% explained by the fact that men and women have different interests. What this means is that about two-thirds of discrimination is explained by other factors. Not "discrimination" in that last sentence, but "difference". Perhaps this was unintentional, and "discrimination" is meant in some obtuse technical sense. But to most people "discrimination" means intentionally unfair treatment. The data they show - again, very interesting data - does not show anything about discrimination. edit: a second major issue is that they don't normalize the likes, or at least don't mention that they do. The likes are not limited, in other words, one can like 1,000 careers, and work in just 1. That would appear to show that person wanted to work in all those careers, but didn't, showing a "gap". A more normalized measure could let each person only like one career, say. Again, maybe they did this, but it isn't stated, and generally "likes" are not limited.