From a reader:

Jonah,

I’m a graduate student studying social cognition at a major research university. I’m a Howard Dean liberal, and probably to the left of the Berkeley people who published that study of conservatism. And, I agree with you that the Berkeley study is shameless BS. I recognize that reasonable people can disagree about any political issue, and I would never try to subvert debate by *defining* those who disagree with me as unreasonable.

In the hands of an honest person, meta-analysis can be a tool to iron out the biases inherent to specific studies. One good example is self-esteem. Despite a general bias in our field toward thinking that self-esteem is important, meta-analysis has suggested that self-esteem is not so important (there’s a small correlation between happiness and self-esteem, and little else).

The reason that self-esteem is a good candidate for meta-analysis, and conservatism is a bad candidate, is that self-esteem is well-defined and conservatism is (as you discussed in your article) poorly defined *especially when studying ‘conservatism’ across nations*. In the minds of Jost, Glasser, etc., a conservative is a conservative is a conservative. This fallacy is natural enough. Social psychology speaks of an outgroup homogeneity effect, whereby people think that the members of their own group (the ‘ingroup’) are different from each other but that people in an outgroup are similar. E.g., ‘there is variability among us gentiles, but jews are all the same.’ Or, (to salvage some of my liberal pride) Ann Coulter thinking liberals are all the same. The Berekeley researchers certainly understand outgroup homogeneity, but they obviously couldn’t transcend it.

[Name withheld]