Conservatives are from Mars, liberals are from Venus. So announced the headline of an article in The Atlantic back in 2012 that was based on a number of peer-reviewed studies. But are they really?

The preferred narrative in academic and media circles would have you believe they are. Thus, in 2012, a study the article did not cite even claimed that conservatives score higher than liberals on “authoritarianism” and “psychoticism,” while liberals scored higher than conservatives on “social desirability” and “neuroticism.” So the overall weight of “scientific” evidence fed the narrative that conservatives are hard and cold, while liberals just want everybody to get along.

But there’s a bit of a problem: that latter study read its own data exactly backwards.

That was announced in January of this year by a prominent social-science journal, the American Journal of Political Science, which contains this: “Erratum to ‘Correlation not Causation: The Relationship between Personality Traits and Political Ideologies’ American Journal of Political Science 56 (1), 34–51, by Brad Verhulst, Lindon Eaves, and Peter K.Hatemi.”

Translation? We messed up.

So it appears, according to the actual data the authors collected, that liberals are more likely to exhibit authoritarian traits and psychotic tendencies, and conservatives more likely to exhibit socially desirable traits and neurotic tendencies.

Or are they? That depends on how seriously to take research of this kind. To learn how seriously one should take it, one does well to read the commentaries on the present howler.

I’ve read three. My attention was drawn to the above erratum by this NR Online post last Thursday from the redoubtable David French, who got it from Steven Hayward at Powerline. The latter learned about it from the academic sleuth site Retraction Watch, which he rightly calls “indispensable.” The RT post also exposes errors in other, related papers. All three of the pieces I’ve linked are caustically entertaining reads.

Curiously, all the admitted errors tended to make conservatives look worse than liberals. Is there a lesson here?

As social psychologist Jonathan Haidt (who was cited in the 2012 Atlantic piece) has pointed out, social scientists are overwhelmingly liberal politically. So it makes sense that their “confirmation bias” would tend to make liberals look better than conservatives.

That’s not a problem in itself: We’re all human, and therefore we all have confirmation biases of one sort or another. The real problem is that, far too often, the bias leads the researchers not only to make mistakes, but also to interpret their own data, when it’s unfavorable to their preferred narrative, as though it were actually favorable. That leads to howlers like the one AJPS recently admitted.

That ought to embarrass social scientists. But it’s more likely, as French suggests, that “the whole thing will now likely disappear down the memory hole. Everyone knows conservatives are the real authoritarians, so this wrong study has to be wrong. Or was the wrong study right? It’s hard to keep up when the “science” keeps shifting.”

Indeed. Be very careful when research confirms your prejudices.

H/T Retraction Watch

[Photo Credit: Alex E. Proimos/Flicker, CC BY 2.0]