60% gaydar accuracy after viewing a face for 50ms! That's really amazing. If the findings are backed up by additional studies, this could be a blow to argument that gayness is a choice and is not (much) affected by physiological factors.
It would be interesting to see is there's any difference recognizing gay men and women by straight men and women.
You can get at that 60% if you can reliably identify the 'easiest' 10% at either side of the distribution and guess the rest.
Also, the sample images were "of individuals [...] who self-identified as straight or gay". It would not surprise me if that overselected for those edge cases. Or am I showing my ignorance and is it common to express one's sexual preference on one's public image on Facebook?
I don't know.. there is a saying in German that "you earn your face" - and it's obvious that hormonal balance, face muscle tone and predominant facial expressions influence our face's appearance in the long term. 60% accuracy does not sound SO obvious to me, I bet it's easier to distinguish between religious extremists (ascetics) and non-extremists.
I’m extremely, extremely, extremely wary of the methodology employed here. The author selected the photographs from individuals on Facebook “who self-identified as straight or gay” living in “11 major cities,” all of whom are white.
I.e., this experiment is already selecting for individuals who a) have a high degree of control over their appearance to the author; b) exist within a system—Facebook—which rewards personal disclosure of one’s sexual orientation in vastly different ways. Identifying as gay on a public forum like Facebook—in a manner so that this information is even available to strangers—requires far more deliberation, and often presents far more social costs, than identifying as straight. These disclosures do not bear the same weight—even in a major city.
What the author should have—and could have—done is gather photographic data without selecting for subjects who self-disclose their sexuality.
Right now, the one evident conclusion this paper supports is that certain the certain details of the way one represents himself on Facebook correlate with a propensity for self-disclosure of sexual orientation on Facebook. (Which, remember, doesn’t require that self-disclosure—or a profile photo, for that matter.)
However: if we accept (as this author does) that sexuality correlates with facial geometry in a certain way, then it is quite possible that the facial photographs employed by the experiment are taken from individuals who are very aware of their appearance, and choose to self-disclose their sexuality based on it.
In fewer words, this study seems to suggest that humans can determine if an individual is gay or straight, based on facial features alone, when said individual elects to disclose his sexuality, and maintains personal control over his outward appearance, regardless of that appearance’s fidelity to reality.*
*What I mean here is that people have extraordinary control over which photos represent them on Facebook. The author, I’m guessing, has no way of knowing if any of these photos were manipulated in any way.
I was apart of a similar study once. I'm gay, and they were unable to tell. During their study it was actually an under chance of 40% which was pretty bad. Needless to say, 60% isn't much of chance either.
This is the only thing that made any sense in that article:
"We speculate that people overzealously interpret whatever facial factors lead us to classify men as gay. That is, it may be that straight men’s faces that are perceived as even slightly effeminate are incorrectly classified as gay, whereas straight women’s faces that are perceived as slightly masculine may still be seen as straight. That would be consistent with how our society applies gender norms to men: very strictly."
You should probably review a statistics test -- under 40% is great. In fact, under 40% is better than 60% (assuming statistical significance). All that would mean is that people see homosexuals as more heterosexual than heterosexuals (which would be quite funny). Any statistically significant deviation from 50% implies successful visual differentiation abilities.
The 60% accuracy is for a 50ms glimpse of a photo "mostly devoid of cultural cues". In real life you get much richer information for much more time, so it seems reasonable to conclude that 60% is a lower-bound for real-world gaydar accuracy.
Unless the extra information is somehow misleading.
I'm thinking here of the example of lie detection. Humans generally do pretty well at detecting lies when they're just reading text. They do slightly less well if they can hear the person talking. And if the person is visible, we don't do any better than chance.
It would be interesting to see is there's any difference recognizing gay men and women by straight men and women.