Last week I got to attend the always enlightening annual ZEW ICT Economics Conference. One of the Keynotes was from the always insightful, Catherine Tucker. In one part of her talk she related that her team conducted an experiment to place a generic ad for a STEM educational program on social media only to find out it was shown much more often to men/boys than women/girls. Algorithmic bias, right!?
Digging a little deeper, they discovered that their bid lost out on the ad auction for females more often because others would bid higher. Ads are placed based on the results of real time auctions for "eyeballs." It turns out that men are cheap (pun intended). That is, women control so much more discretionary spending that they are more heavily courted by advertisers with higher auction bids. The STEM ad bid was the same for men and women and so lost out more often when it had to compete with stronger bids for female "eyeballs."