From Simon Oxenham's excellent weekly column in New Scientist comes word of another social psychology study that will undermine our faith in social psychology studies. We can never get enough of those.

The new study, to be published next month, bears the unappetizing title "Interpretations and Methods: Toward a more effectively self-correcting social psychology." You can read an early version of the paper here.

Its lead author, Lee Jussim of Rutgers University, has emerged as one of the great intellectual heroes of the age. For reasons that will become clear, he is mostly uncelebrated. Yet he is absolutely fearless in his determination to think for himself. As a social scientist, he manages to resist the enormous pressure to conform to the lowing herd of his peers.

That pressure, as it happens, is the phenomenon at the center of his new study. Social pressure to conform, Jussim says, is a fact of human behavior that the scientists who study human behavior too seldom take into account—at least when it applies to them. "Confirmation bias" is the fancy-pants social science term for the obvious truth that people tend to believe what we want to believe, focusing on evidence that appeals to us at the expense of evidence to the contrary. Social psychology is riddled with confirmation bias. And we are beginning to see that it has rendered whole sections of the field nearly worthless.

This is especially true, says Jussim, when the subject is politics. It's long been understood that left-wingers outnumber right-wingers in social psychology by at least ten to one. Among other things, the lack of diversity explains the field's obsessions with race, gender, environmentalism, and the other enthusiasms of the left. Jussim's interesting gloss is that the bias is built into the way the studies themselves are put together. Researchers are reluctant to consider conclusions they don't like even when their own data make alternative conclusions more plausible.

One table produced by Jussim and his co-authors looks at a famous example. In 2003, a study allegedly demonstrated that "conservatism is a syndrome characterized by rigidity, dogmatism, prejudice, and fear"—a slander undoubtedly dear to the hearts of nine out of ten social scientists. In 2010, another group of researchers returned to the original study's data and combined it with more recent data of their own. They found almost no evidence of such a "syndrome." Prejudice, fear, and dogmatism are pretty evenly distributed among conservatives and liberals alike.

Here's the kicker, though: Researchers have cited the older, anti-conservative study more than 1000 times in their own studies even since it was discredited in 2010. The second study has been cited a grand total of 60 times. Social scientists tend to believe what they want to believe. Maybe they're real people after all.

As Jussim points out, experimental science is supposed to be self-correcting: when a finding can't be replicated, scientists have to go back to the drawing board. Very slowly, thanks to Jussim and a handful of other brave researchers, social science has begun at last to correct itself. As a result, some of the most cherished building blocks of the entire enterprise—the value of "priming," the effects of "stereotype threat"—have gone from settled truths to shaky hypotheses. Jussim's paper is another step in a long process of intellectual hygiene.

The self-correction has powerful enemies in the social-science establishment. But if it's allowed to continue, it may at last disenthrall the general public from our addiction to studies by unscientific scientists.

Who knows? Maybe even science journalists will catch on eventually.