I'm interested in group means for their own sake. Or rather, I'm interested in them because they hint at underlying mechanisms, and I'm interested in mechanisms for their own sake.
FWIW, I'm asexual and do not regard homosexuality as some kind of defect. Otherwise, what would I have to think of myself?
Additionally, I don't think you can "lack" masculinity in the sense that you "need" some amount and are deficient otherwise. There are some people who are closer to one extreme than to the other, and if likelihood of attraction to a specific sex changes somewhat monotonically with position on this axis, that tells you very little about individuals, but a bit more about biological interactions, which, again, I find interesting for their own sake.
And this study and other similar ones would seem to prove you wrong?
>You'd think that the idea that gay men are sexually attracted to men because they are more feminine than other men would have died in the face of all the obvious counterexamples
A few exceptions don't disprove a statistical correlation.
Besides it's more complicated than just "gay people are more feminine". There are various hormones and biological interactions other than just testosterone. And the effects happen early on in development; Whether an adult has lots of testosterone and muscles is irrelevant.
As this study shows, humans are much worse than this algorithm at identifying gay people. It's not merely a matter of "that face looks feminine" or human scorers would have done much better. Presumably many of the correlations are very small or subtle, but in aggregate they can get surprising accuracy.
A large number of exceptions to a generalization absolutely is sufficient to undermine a causal hypothesis. If X causes Y, then then you shouldn't frequently find Y in the absence of X. You do frequently find gayness in the absence of above average femininity.
>Besides it's more complicated than just "gay people are more feminine".
Of course it is more complicated that that. That's what I've been saying. But in the media, and on HN, the idea of relating homosexuality directly to a masculine-feminine continuum is very popular. Many do not realize that this is not science, but just old-school homophobia dressed in pseudoscientific garb.
Only if it's a deterministic causal hypothesis. Otherwise, the proportion of counterexamples only serves to bound the effect size, which is obviously quite small in this case.
I don't think anyone in this thread seriously believes that "almost all gay men [are] unusually feminine". If they were, they wouldn't be called "gay men" but "women with a penis" or something like that.
> That's what I've been saying. But in the media, and on HN, the idea of relating homosexuality directly to a masculine-feminine continuum is very popular. Many do not realize that this is not science, but just old-school homophobia dressed in pseudoscientific garb.
A direct relation would certainly be surprising, but that doesn't mean you can't study correlations between sexual orientation and certain measurements of the body. That's science and might lead to insights how and when sexual orientation develops. I don't really care what people make of it who confuse correlation for causation or don't understand that statistical significance doesn't imply a large effect size.
What a bizarre thing to say. People used to really think that all gay men were unusually feminine, but they still didn't call them women with penises, because they aren't.
>Otherwise, the proportion of counterexamples only serves to bound the effect size, which is obviously quite small in this case.
But why the interest in an effect which is small if it exists at all, and for which there is little evidence? There are probably hundreds of other features which correlate equally strongly (i.e., not very strongly at all) with homosexuality. This one happens to tap into pre-existing stereotypes.
>A direct relation would certainly be surprising, but that doesn't mean you can't study correlations between sexual orientation and certain measurements of the body. That's science and might lead to insights how and when sexual orientation develops.
The topic of the article isn't a scientific study. It's a couple of people in a business school playing around with deep learning for marketing purposes. They are not doing it because they're genuinely interested in uncovering the mechanisms underlying the development of human sexuality.
>I don't really care what people make of it who confuse correlation for causation or don't understand that statistical significance doesn't imply a large effect size.
In other words, you don't care what the majority of the population make of this research. That's fine for you if you're not gay and you're not affected by it. But the question of how this research is marketed, and how it's perceived by the general public, is potentially very important to gay people. It matters to us what the general public thinks about us.
A team reported their results with a model they trained. The results are accompanied by some hopeful theory about things that the authors have no way to know. There's no proof of anything in there.
Ok. I think it's also quite likely that they're not.
>(most feminine woman) (average straight woman) (average gay woman) (most feminine man) (most masculine woman) (average gay man) (average straight man) (most masculine man)
Are you serious about this? Gay men are just men who are sexually attracted to men. You'd think that the idea that gay men are sexually attracted to men because they are more feminine than other men would have died in the face of all the obvious counterexamples, now that fewer people are hiding their sexuality, but it seems surprisingly persistent.
The problem with this kind of thing is that despite representations to the contrary, no-one is really interested in the group means for their own sake. They are interested because they want to pedal the tired old theory that gayness is a defect resulting from a lack of masculinity. People are often too polite (or afraid?) to say that now, but gay people are capable of reading between the lines.