Math knows who and what you love

If you go to the website OKCupid you will see that it self identifies as the 'best dating site on Earth' and this is probably an indication of what online dating is like: a little self promotion never hurt anyone. Right?

Another tagline that they use: 'we do the math to get you dates'

The site matches its users by the use of online quizzes which makes it different than the online dating sites that I've seen. The data collected by this site is probably quite interesting to read. A blogger for the site by the name of Christian Rudder is identified as a guitarist and a math guy in an interview on NPR. If you look at his Wikipedia page he is labeled as American musician, humorist, and entrepreneur.

This story was about the analysis that Christian Rudder (and his team) did on the data available through OKCupid. Admittedly it was just for fun, but I bet this data would be of interest to marketers. They first tried to identify the likes of groups that one would normally try to stereotype. They didn't find anything interesting because these lists looked very similar independent of what group one was looking at. But then they tried something interesting and learned a good mathematical definition of what a stereotype is: a 'like' that a group has that is not common to the whole population.

What did they find? From the NPR article:

For example, top interests for white men included Tom Clancy, Van Halen, golfing and Harley Davidson. Top interests for black women were soul food, The Color Purple, gospel and Alicia Keys.

I know that this sounds kind of sad, but really it is identifying what we call a stereotype and why.

The most recent posting to the OKCupid blog was about statistics about gay and straight sterotypes and other statistics from data which seems to involve a lot of really good sociology research.