These kinds of studies are ridiculous.<p>First there's the element of subjectivity and discretion in the "trained coders" that rate a person's attractiveness. On a single overall number scale no less.<p>But more importantly, how do they even hope to attain any semblance of ceteris paribus ("all things being equal") in this study? Of the millions of factors that contribute to pair selection and how couples interact, the researchers want to isolate disparities in attractiveness and test whether this is significant. Short of cloning a whole bunch of people and pairing them up with ugly and beautiful partners, AND THEN doing the tests, the sheer number of conflating variables makes this kind of research worthless. I don't think even the clones would make the results valid given the variables that aren't genetic eg. upbringing, mood on the day, individual preference etc.<p>It comes accross as if they came up with the conclusion first, and then fit a flimsy experiment to prove the point. The correlations are no doubt statistically significant, but to imply that they mean anything without even attempting to isolate any other conflicting factors is a joke.<p>Human interaction is a complex system. There's every possibility that attractiveness does contribute to pair selection as the researchers assert, but what is the proportion of that influence? What billion other factors contribute? My guess is that you'll never be able to isolate just one factor that accounts for the bulk of the influence.<p>This is not to say that social science experiments can't be done right, but it's not easy. This kind of reductionist tripe only serves to muddy the really good research out there.