Last July, I was on a panel at SkepchickCon on gender differences. Someone from the audience asked, “Is there any good evolutionary psychology out there?”
After everyone was done laughing, I gave my real answer. I pointed out that evolutionary psychology is incredibly difficult to do well. It requires the accumulation of a lot of information that isn’t always easy or inexpensive to get. It requires lines of evidence from several disciplines to be considered together to even approach the proof required to say that the brain contains mechanisms to determine behavior that evolved in humanity’s pre-agricultural history.
Kate Clancy (of the vary appropriately named Context and Variation) has a wonderful post up today about five ways in which evolutionary psychology as usually practiced fails to meet the standards required to be considered good. She hits one of my peeves–methodology geek that I am–operationalizing variables:
In some studies of evolutionary psychology, a never-before-used variable is often created to serve as a proxy for what they really want to know. Not too long ago I took issue with a “maternal tendencies” variable. Because they couldn’t assess maternal behavior in these young, childless undergraduate women, they asked them how many children they wanted to have. The more children these eighteen and nineteen year olds wanted, the more maternal they were.
Yet desired family size at eighteen, and maternal tendencies as a future mother, are very, very different things. As I pointed out in my post on this, there is too much context-dependence embedded in when you ask women how many kids they want for it to tell you anything with much biological meaning.
So, make sure you’re measuring what you think you’re measuring. And validate the heck out of any new proxy you come up with.
Kate is always wonderful on the topic of overly simplified variables, particularly when it comes to the tendency of researchers to assume fertility rather than measuring it.
In today’s post, she also makes a great post about using undergrads for this research that had never jumped out at me before:
The two reasons oversampling from WEIRD people is bad is first that oversampling in general is bad, but second that being WEIRD puts you about as far removed from the conditions in which we evolved as you can get. WEIRD stressors are chronic and psychosocial (which makes them great if that’s your research interest, otherwise not so much). They have a lot of weird (ha ha) immune problems, possibly related to under-challenging their immune systems when young. They tend to survive the major childhood illnesses but then die of heart attacks, strokes or cancer. Many of them delay childbearing into well into their reproductive years and breastfeed for a short duration if at all, meaning they have eight to ten times as many menstrual cycles as the average forager. And they tend to have nuclear families, rather than breed cooperatively in large groups, sharing the parenting load among peers and across generations.
This makes WEIRD (Western, Educated, Industrialized, Rich, Democratic) research subjects excellent for using as a contrast with, say, modern hunter-forager societies, but ought to tell you why projecting backward without a lot more evidence is a big problem.
Go read all of Kate’s post. It will go a long way toward helping you understand why a lot of us don’t find the vast majority of evolutionary psychology terribly convincing.