There are so many ways to go wrong in evolutionary psychology. Kate Clancy does us the tremendous favor of specifying how to do evo psych right. Invaluable reading. Here's the pith:

For something to be an evolutionarily stable strategy it has to fit a few conditions:

1. You need clear evidence it is an adaption, which means it has to conform to the conditions from the previous section: being heritable, variable, and producing differential reproductive success.
2. You would also need evidence that what you’re seeing isn’t simply a correlated response from another, linked behavior being selected.
3. You would need to demonstrate that the behavior is at least equivalent to, if not resistant to, alternative strategies, in terms of its rate of success.

I laid out how hebephilia fails these tests in a post last year. The problem with demonstrating natural selection, and in particular evolutionarily stable strategies, is that the burden of evidence is incredibly high. Which means most stories that buttress bad evolutionary psychology work will ultimately lead to that study’s collapse, when we see the stories are made of thin air.

If you design your study really well, finding ways to anticipate and control for cultural bias, and still find a correlation, I’m quite happy for you! But chances are good you don’t have enough to contend what you’re seeing is an evolutionarily stable strategy. So hold the storytelling. Just for a little while.

I tried to get at some of this a while back in Enough With the 'Slut Gene' Already: Behaviors Ain't Traits. Clancy does a broader, more thorough job here. Bookmark this one and come back to it when you need to hip-check some evo psych mojo that's Just Wrong.

5 Ways to Make Progress in Evolutionary Psychology: Smash, Not Match, Stereotypes | Context and Variation, Scientific American Blog Network.