Dr. Aaron Sell recently wrote an interesting piece about political incentives and shoddy statistical work in science. In particular, he was highly critical of the much-publicized Conley article in the Journal of Personality and Social Psychology which attempted to demonstrate that "large gender differences... [in willingness to engage in] casual sex [have] more to do with perceived personality characteristics of the female versus male proposers than with gender differences."
[[The full article can be found here. Please note - I have deliberately chosen to omit quoting any material which Less Wrongers may find mindkilling. It would be nice if we could keep it that way.]]
He points out that Conley's results were obtained by hypothetical pseudo-experiments, and spurious controls. For example, when Conley found that men were more willing to sleep with their friends than women, she made this difference "evaporate" by:
control[ling] for “sexual capabilities.” The sexual capabilities covariate contained two items: “the proposer would be a great lover” and “would provide you with a positive sexual experience.” Keep in mind, these items were rated by the subjects just after deciding how likely they would be to have a sexual encounter with the person! If you control for two tightly correlated variables any effect disappears. The gap in male and female pay completely disappears when you control for testicle number... So how tightly correlated is “likelihood of agreeing to sex” with “would provide a positive sexual experience”? We don’t know; it’s not reported. JPSP. Seriously.
Yet for bottom line reasons, we now find Eagly and Woods reading Conley uncritically, and even claiming that she has overturned Clark and Hatfield's classic (and truly experimentally based) article demonstrating that women are radically less willing than men to have casual sex.
This sort of nonsense is an indictment of our science... As long as the most prestigious journals of our discipline publish this kind of political masturbation , we have no right to demand that the public take us seriously. When politics and good science collide there is no reason the public should bet on science. They are better off trusting their uninformed intuitions. An imbecile... know[s] damn well that 80% of women don’t want casual sex with random men who are “reputed to be sexually skilled.”
While empiricism is great, I have long believed that the social and organisational structures in which science is practised makes it especially vulnerable to political capture, so this plays right into my biases. Am I missing something important?