Am I a boy or a girl?

Mike Kruger pointed me to this site which estimates the probability you’re female and the probability you’re male based on your browsing history. It estimates the probability that I’m male as 66%. I don’t really know what these “probabilities” are, though. They’re between 0 and 1, but I doubt they’re calibrated to give a direct probability interpretation. (For example, if you took everybody whose claimed probabilities were 66%, would 66% of these people actually be male?)

And here are Mike’s comments. He calls the method “Bayesian” but I’m not so sure.

5 thoughts on “Am I a boy or a girl?

  1. It's naive Bayesian, isn't it? At least, that's my interpretation of it — it's Bayesian, but assuming that visiting various web sites is independent. This is, of course, very far from being true, so I doubt the results are correctly calibrated.

  2. Kind of interesting. It appears to assign an OR to each site, and then multiply them to get an OR.

    I was 79% male, but Slashdot had a big effect on that (OR = 1.7) – without that, I'm only got a 68% chance of being male.

    The question is where do the ORs come from – some of them seem a little curious. Slashdot probably is mostly male, but truecredit.com is my most female site – with 0.74.

  3. Hmmm…exactly 50% chance of female, 50% chance of male. What is that telling me?! On a side note, Jeremy's commentary (prior comment) makes me want to browse various web sites and see what subsequently happens to my score.

  4. The independence assumption is quite unrealistic, but, with that assumption, this does appear to be an bona fide bayesian statement. Well, at least modulo numerical errors; mine returned 100% male. Appears to have been driven by newegg, slashdot, wired, etc. I'm also somewhat surprised that the male/female ratio for gmail is .9; I would have expected closer to one.

  5. I know the guy who did this. The likelihood is based on the probabilities stated in Quantcast data so it is P(M | S1 and S2 and … Sn) where the prior probabilities are given by Quantcast.

Comments are closed.