Report comment

Or is the evil that people use apriori information (aka “stereotypes”) to construct priors? Wait, I thought you were a lesswrong rationalist? Aren’t people supposed to be Bayesian reasoners?

Are you saying that it is *correct* to assume that male scientists write better scientific studies than female scientists? Or is it that you don’t think widespread bad priors are a bad thing, so long as one is using them to do proper Bayesian reasoning? Or am I misunderstanding you completely?

Report comment

Report comment

Report comment

So how would one design an experiment to distinguish between correct Bayesian reasoning and incorrect reasoning? I’d suggest a multiple stage approach. First you give your subjects the 150-word abstracts (that is, a small amount of information). Then you sample their opinions. Second, you give them the full paper. (More info.) Then you sample again. Ideally you might then add a third stage, perhaps a second paper by the purported authors. And again sample opinion. The bigotry theory of gender norms suggests that more information will not change the opinions of the subjects: they know that women should not do math, dammit, because math is hard, even though this paper really is a pretty good result! The Bayesian theory suggests that more information will change their opinions.

Report comment

Report comment

both women and men working together to enforce norms that some areas should be male and others should be female. Which of course is pretty bad in itself.

Sustaining gender norms is bad? Why? Is it evil that men are dominant in the fields of “political communication, computers, news and journalism”? Is it wrong that women are more interested in “children, parenting and body image”? Or is the evil that people (or rather graduate students in communication) have noticed these aspects of the real world? Or is the evil that people use apriori information (aka “stereotypes”) to construct priors? Wait, I thought you were a lesswrong rationalist? Aren’t people supposed to be Bayesian reasoners?

I might also point out that in asking people to rate a paper using a 150 world abstract, the researchers were dialing down the signal to a very low level. This comment is more than 150 words. Certainly one can say something in 150 words, but not that much. And the article you linked, at least, gives no hint as to how strong the effect was. Just that people rate male-authored articles in “male” areas “significantly higher”. “Significance” is a very low bar. Is this a little higher, or a lot?

Report comment

Report comment

Report comment

Report comment