Assortative mating

I've argued before that people are Lancasterian goods and that we all get the best spouse we can afford given our particular bundles of characteristics.

Jeff Ely finds additional evidence:

I visited the Cowles Foundation at Yale for the winter of 2006, and taught a senior elective course. Seven fortunate students took my seminar in information economics. One impressive woman student — who organized the gay and lesbian social scene — asked whether the shallow view of Becker’s model was so unrealistic. Did babes match with hunks?

We brainstormed on data sources and settled on two new web sites: facebook.com and hotornot.com. Facebook allowed users to indicate with whom they were “in a relationship with”. Facebook was still new, and not yet open to all email addresses. So the student asked her friends at various campuses across America for their logins. And so began our stealth project. Hundreds of photos of matched men and women were downloaded, and then uploaded to HotOrNot, all on the sly. HotOrNot afforded us the average evaluation of about 200 women for every man, and 2000 men for every woman.

The result: Regressing straight men’s or women’s hotness on their partner’s hotness gave a highly significant fit, with a slope of about 0.7 — so that a man rising in hotness from 7 to 8 expects his partner to rise by 0.7 points. But sorting was far closer for gays and lesbians, with a slope for each of about 0.9. As Becker implied, beauty is income in this meat market, and the “richest” men match with the “richest” women.

The fit would be even better if you could adjust for the other Lancasterian characteristics.

I love the gonzo approach to this project. The human subjects review panels there must be more sane or have fewer teeth. Excellent either way.

And Robin Hanson would (and has) asked why we care so much about income inequality when this form of inequality may matter more....