2 ms·
OKCupid's usually pretty good about data, but they blew it on a couple of these. For example: Their method of controlling for income only takes into account z
by nostrademons 16y ago
OKCupid's usually pretty good about data, but they blew it on a couple of these. For example:
Their method of controlling for income only takes into account zip code. There's another obvious source of bias: the fact that these people are on an online dating site probably indicates that they are time-poor, have decent computer skills, and are in professions that value introverts. Many of these jobs are in high-paying fields, eg. software. People who work 9-5 in blue collar jobs often go out to bars to pick up women instead. It's no big surprise that the average OKCupid user makes more money than the average person in general.
Side note: there're a couple interesting things in the raw data for income. One is the big dip in response rates for 23-24 year olds that say they make $100K+. The other is the same dip in response rates for 29 year olds. I wonder if those are people assuming that the other person is lying.
The bisexual angle is also easily explained by looking at the profiles of the people you're aggregating. Many girls say, as the top line of their profile "NO MEN! I'm married, but looking for a bi or lesbian girl to have fun with on the side." They are bisexual, but they found the partner they want for one gender already, and are only looking for someone of the opposite gender.
The height stuff was interesting though, particularly since I listed myself as 5'11" despite thinking that I'm only 5'9". I finally revised my estimation of my own height upwards last year, when I found out I was at least an inch taller than my sister's boyfriend, who she swears is 5'10".