generalizations from ME

I've been reading the comments for this post about a Dilbert cartoon based on a joke that only women like shopping, and the comments keep reminding me of Larry Summers's latest adventure in stupidity, in which he tried to justify not having a lot of women in top science professorships while claiming he was just explaining.

Both the people commenting on flea's blog and Larry Summers are making vast generalizations from their personal experiences. "I like shopping, but I can barely get my husband to buy his own underwear." "My daughter got two trucks and named them daddy truck and baby truck." They see these personal experiences, they see stereotypes and heuristics which validate them, and they conclude that the stereotypes are accurate and useful ways of seeing the world. And you know, I bet they're confused, because flea and people who agree with her are also arguing from their personal experience (I hate shopping); so is Nancy Hopkins, an MIT researcher who did a major study on gender bias at MIT and has been very clear about her distaste for Summers's remarks (I'm offended because I'm a woman who likes to work 80-hour weeks). But there's a difference. Flea and Hopkins are looking at stereotypes, and saying, "The world is more complicated than that." They're not saying that there are no women who like shopping, or that there are no women who would rather raise children than study physics: their experience provides no basis for stereotypes, just a way to refute them.

For anyone who's forgotten seventh grade science (apparently including the president of Harvard University!) the scientific method works like this: you look around you for a while, and come up with a hypothesis. Your hypothesis lets you make predictions, and then you design an experiment. If the stuff your hypothesis predicted happens when you do the experiment, your hypothesis is doing ok. But if your hypothesis fails - if there are situations where your predictions are wrong - you have to get a new hypothesis, either by narrowing the scope of your old hypothesis or by finding a new explanation with new predictions. You can't just keep bringing up situations where your predictions were right, because there's a counter-example.

This is what flea and Hopkins are doing. They're giving a counter-example. And Summers and the commenters don't have theories that explain flea, or Hopkins, or me. They just keep bringing up times when they were right. Of course, nothing flea or Hopkins is saying means that Summers's daughter has to grow up to be a brilliant physicist or that no women can like shopping, so using that as a response to their arguments makes no sense whatsoever.

The rules of social science evidence are a little different, of course. I know that. One counter-example doesn't disprove the whole thing, because people are ridiculously complicated and varied. There's a lot to say about the real social science evidence that Summers is ignoring, or the way the kind of evidence you allow changes the results you get. But when you look at the logical structure, Summers and the commenters are making nonsensical arguments; and despite their use of the same kind of evidence, flea and Hopkins are making arguments that make plenty of sense.