Maggie Koerth-Baker at 8:33 am Mon, Jun 4, 2012
ADVERTISE AT BOING BOING!
Well, it’s better by a degree that is statistically significant. That’s still pretty interesting.
Yeah, I think that the 60% figure (instead of 50%) quoted in the article is fairly significantly larger than “nobody’s gaydar is really that much better than random guess work.” The article concludes this as “This demonstrates gaydar ability — which is far from judgment proficiency,” but you are underselling the study much further, I think.
That said, though, I’m a little confused about their methodology. To say that “random guesswork” should yield 50% accuracy should imply both that 50% of the faces were gay and that the participants knew this fact, neither of which seems true in the paper.
If 10% of Americans are gay (say) and I randomly flip a coin to decide if people are gay, I’m going to score 50% accuracy. But that’s terrible: if I just assume everyone is straight without even looking at them, then I’ll actually have 90% accuracy. Far, far better than chance.
I don’t quite see how the article took this into account (but I’m assuming I just overlooked it).
I was hoping this study would do more than just look at photographs of people’s faces (a la the same data that OK Cupid has been gathering through their informal quizzes for at least the last couple of years) as I suspect the accuracy of gaydar would increase if, rather than head shots, participants could see video of the subject walking towards them or speaking, as you would in a more true-to-life setting. Although granted, it would be far more difficult to control for clothing, styling, speech patterns or other cultural clues.
I know my gaydar is terrible. Both Type I and Type II errors. I’ve learned to just treat people’s sexual orientation as “?” until actual evidence comes in.
I wonder how I’d perform on the test described in the article — whether my gaydar mistakes are caused by “noise” from clothing, accessories, speech, movements, etc., and I’d get it right if I just looked at faces, or whether my processing of sexual orientation from facial features is itself wrong most of the time.
I also wonder if the fact that I’m straight has something to do with it. Has anyone looked at gaydar accuracy broken out by the sexual orientation of the observer?
nobody’s gaydar is really that much better than random guess work
Few people have gaydar. Most gays I’ve known don’t have it.
Looking at a portrait still and then consciously guessing that persons sexual orientation has little to do with gaydar.
People with gaydar see a person and just know, with a pretty high degree of certainty. In many ways just like how anyone “knows” if the person they see is a woman or a man.
Gaydar even works from a distance and doesn’t rely on obvious clues that are easy to put in words.
I have anti-gaydar. I’ve done online tests and my accuracy is usually below 5%. It’s weird because, in most other things, I’m extremely good at reading people. I can usually tell if you’ve been in jail, in the military, are on anti-depressants, any number of other things with a few seconds observation. I’m a freakishly good early pregnancy detector. Just not in the one area that might be useful.
Consider inverting your results, and then you’ll have 95% gaydar.
Yes — try Ashley’s idea. Kinda like when George Costanza was in the diner with the other felon…er….peeps and resolved to consult his instinct – then do the very opposite.
It worked, at least for awhile, but, as being a self-confessed, slow-witted bald-man, he figured out how to fcuk it up.
I appreciate the randomness of the original article´s header pic.
The study doesn’t establish that *nobody’s* gaydar is that much better than random, it says the *average* gaydar, applied to highly artificial images, is slightly better than random. Their participants were just a random set of students, likely mostly straight. If they’d chosen gay students or students with self-professed gaydar, their numbers would have been higher. If they’d used videos of gays and straights in the wild, the numbers would have been higher still.
Mail (will not be published) (required)