Yet another exciting story about gender differences: Men’s brains are more responsive to video games than women’s brains. I’ve seen a lot of these studies and most of them are crap – “Look! We took pictures of BRAINS! Any differences we found must be genetic! Because brains aren’t influenced by environment! That’s why it’s impossible for humans to learn anything new!” I read the original paper hoping to find different mistakes than usual, but didn’t find any deviation from the script.
The basic fallacy goes like this: If we measure a difference between two groups that differ genetically, then the cause of the difference must be genetic (and therefore it’s pointless to, say, write video games that appeal to women). If you do this with something familiar and everyday, like, oh, hair length (“Men have shorter hair than women because men are innately short-haired”), most people will be able to tell that environment has an effect too (“Wait a minute… that guy’s hair is short because someone cut it with scissors!”). But as soon as you measure the length of someone’s hippocampus, suddenly people are willing to assume that it’s all genetic because they’ve never seen a hippocampus in their lives, much less observed it growing and changing size in response to the environment. When we measure brains or anything else, we are comparing the combined output of both genes and environment, whether we are using wooden rulers or fMRI machines.
To be clear, I believe (know for certain) that some of the differences between men and women are genetic in basis. I just don’t know which ones they are, especially when it comes to brains. Given the current quality of research on the topic, I doubt I’ll find out in my lifetime.
So, want your poorly constructed study in the national news? Here’s how to do it in 6 easy steps!
- Pick some popular yet politically incorrect prejudice about some group of people – black people are stupider than white people, or women don’t make good leaders – and postulate that the cause is genetic.
- Get your hands on some impressive whiz-bang piece of scientific equipment that only 2 or 3 other labs in the world have. Or get a few dead human beings and cut them open, that will work too.
- Select some incredibly tiny non-representative sample of college students from the nearest campus and apply some bogus method to screen out environmental differences (I suggest a self-administered questionnaire, those are fast, cheap, and inaccurate).
- Make some fancy whiz-bang measurements. Whoa! The groups are different! Even if they aren’t, run the results through obscure pseudo-statistical functions and put in lots of p-values.
- Write a paper which says, “And because we measured BRAINS with fancy whiz-bang equipment, obviously, the cause of the differences must be [wave hands wildly] genetic!”
- Call the press. Even if they don’t believe you’re right, they’ll print it anyway – controversy sells, even if it’s manufactured.
I have about a thousand more stories and examples of scientifically bankrupt “studies” on gender differences, but I have to get some code done. I’ll close by saying that I once requested the data for a study on gender-based cognitive differences and found that the authors’ claim of negative correlation between two (fairly bogus) cognitive measures was absent in the male population and positive (reversed) in the female population. How did I perform this statistical feat? I used the “CORR()” function in Excel. (The authors had constructed their own pseudo-statistical function which would have found “correlation” in a random data set.) And yet this was published in a respected scientific journal. This is the level of competency we are dealing with here, folks.