"Gaydar" is not to be trusted

The headline in The New York Times read The Science of "Gaydar". In the story I read "gaydar is indeed real." "Hmmm. Interesting," I thought.

"Gaydar judgments were about 60 percent accurate," the researcher wrote. He defended this seemingly low number writing "chance guessing would yield 50 percent accuracy." I was perplexed. I thought chance guessing would yield a result close to 50 percent but usually off by just a little. I thought a little "chance error" was to be expected.

To minimize chance error researchers work with large numbers. Toss a coin twice and there is no surprise if all tosses, all two, are the same. Toss a coin a thousand times and get a result close to a thousand heads or a thousand tails and something is wrong. Maybe the coin is "loaded."

I looked up the research report. I learned:

Only 19 students participated successfully. Seven additional participants were excluded.
All the students taking part in the research were women.
The images used came from Facebook and were self-identified as straight or gay. Self-identified?
Photographs included 111 gay men, 122 straight men, 87 gay women, and 93 straight women.

The story in The Times reported in the tests, "Gaydar judgments were about 60 percent accurate." I wondered what the expected range of chance error was for this research. What seemed most important to me was the 40 percent failure rate.

With such a high failure rate, it didn't seem one should act on one's gaydar feelings. Too big a chance for error.

Then I read the last lines in The Times article:

"Should you trust your gaydar in everyday life? Probably not. In our experiments, average gaydar judgment accuracy was only in the 60 percent range. This demonstrates gaydar ability — which is far from judgment proficiency. 

But is gaydar real? Absolutely."

Yes. It is absolutely real. And absolutely not to be trusted.