Monday, April 20, 2015

Hiring bias

A study hit the news a few days ago claiming that in STEM, if a candidate was a woman she had a 2:1 advantage over a man in the probability of being hired. The direction of bias matches what a blurbs for job openings suggest; and there seems to be pressure to overcome alleged biases by hiring more women--if various statements from university committees are anything to go by. I get no pressure--an advantage of a low-level position. I was on only one hiring committee in all the years so far at UW, and of the 5 candidates that time the winner stood out--there wasn't a lot of judgment call involved. He worked out pretty well, too.

This doesn't match the preferred narrative, which is one of Ed Yong's hobbyhorses, so he triumphantly linked to a couple of reports that refuted the study, like this one which features gems like "The researchers’ own biases lead them to believe that women and men belong to two discrete groups (making genderqueer and transgender scientists invisible).". It is quite true that the final selection is based on different criteria than the test study, but it is also true that Harvard does not interview all the applicants. Or even most. If their hiring committee tells you they read CV from all 3000 applicants for a job, remind them what happens in the 8'th circle. The pre-selection biases matter, and that is where the original study really matters.

1 comment:

bs king said...

I liked this take on it:
http://slatestarcodex.com/2015/04/15/trouble-walking-down-the-hallway/

Points out the interesting fact that the authors of this study had lobbied against the idea that there was a disparity. OTOH, the author of the biggest counter study has lobbied for her side before doing the study as well.

Interesting that everyone found just what they were looking for.