Bias in Bias

Writing in the NYT, John Tierney takes note of new efforts within the social psychology fields to explore bias against conservatives in academia. In doing so, he takes a bunch of gratutious shots at explorations of implicit bias in other fields that I think are wildly out of line.

As a liberal who has been flagging the issue of anti-conservative bias in academia as a serious problem that should be addressed for well over a half decade, I think I have credibility to speak on this issue. I absolutely agree with Mr. Tierney that this is something that colleges and universities should work to redress, including via affirmative action policies where appropriate -- that was my position six years ago, and that's my position today.

But what I don't get, and think is exceptionally difficult to justify, is Tierney's claim that the bulk of the social psychology evidence indicates that implicit bias and other structural barriers against women are gone.

Similarly, Larry Summers, then president of Harvard, was ostracized in 2005 for wondering publicly whether the preponderance of male professors in some top math and science departments might be due partly to the larger variance in I.Q. scores among men (meaning there are more men at the very high and very low ends). “This was not a permissible hypothesis,” Dr. Haidt said. “It blamed the victims rather than the powerful. The outrage ultimately led to his resignation. We psychologists should have been outraged by the outrage. We should have defended his right to think freely.”

Instead, the taboo against discussing sex differences was reinforced, so universities and the National Science Foundation went on spending tens of millions of dollars on research and programs based on the assumption that female scientists faced discrimination and various forms of unconscious bias. But that assumption has been repeatedly contradicted, most recently in a study published Monday in The Proceedings of the National Academy of Sciences by two Cornell psychologists, Stephen J. Ceci and Wendy M. Williams. After reviewing two decades of research, they report that a woman in academic science typically fares as well as, if not better than, a comparable man when it comes to being interviewed, hired, promoted, financed and published.

“Thus,” they conclude, “the ongoing focus on sex discrimination in reviewing, interviewing and hiring represents costly, misplaced effort. Society is engaged in the present in solving problems of the past.” Instead of presuming discrimination in science or expecting the sexes to show equal interest in every discipline, the Cornell researchers say, universities should make it easier for women in any field to combine scholarship with family responsibilities.

That's just not an accurate summary of the state of the literature. To talk as if all the social science research explodes the myth that bias continues to exert sex-differentiated impacts, and the theory simply carries on via blind political correctness, is wrong. (see, for example, the research on stereotype threat). Ceci and Williams' study is interesting, but it is nowhere near as robust as Tierney (or even the authors themselves, though they're more circumspect about it) claims.

Most notably, the authors concede the existence of sex-bias in math intensive fields. And in general, they attribute much sex-differentiation to the "resources" available to the academics, which encompasses their position (junior or senior, or non-tenure track), teaching loads, and the prestige of the institution they teach at. But their hard delineation that differences in who gets these positions are attributable to "personal choices" (they agree that these choices might be "constrained", and their policy suggestions are centered around removing these constraints by, for example, improving work/life balance concerns) is difficult to warrant -- after all, that women are channeled to adopt social roles which are, at the very least, partially home-based may itself be attributable to implicit bias. And, once marginal candidates are filtered out, we might expect that the remaining women -- who are the most committed and probably the most talented (if we expect that exceptionally talented candidates are likely to face pressure to "stay in the game") -- should outperform their male peers, at least by a little bit. It's a bit weird to recognize that women are systematically ending up in low ranked position, but reject a hypothesis that implicit bias might play a role because the women who survive the gauntlet do slightly better than men at the final stage.

I also have no idea what Tierney even means in the following passage: "the taboo against discussing sex differences was reinforced, so universities and the National Science Foundation went on spending tens of millions of dollars on research and programs based on the assumption that female scientists faced discrimination and various forms of unconscious bias."

What work is "based on the assumption" doing here? If it simply means that the research seeks to take that theory and put it to an empirical test, Tierney can't have an objection -- that covers the Ceci and Williams study as well. That we have a well-developed theory explaining sex difference (implicit bias) is an excellent reason for running studies based on that theoretical orientation. Obviously, if the studies consistently were disproving the thesis, we shouldn't beat a dead horse, but the problem is that Tierney is flatly inaccurate if he thinks that's what is going on. If the claim is that we're doing research that presupposes implicit bias to exist, and uses that presumption as some sort of assumed-but-unproven axiom in another study, that'd be more problematic (except for the fact that, contra Tierney, there is a substantial amount of literature backing up implicit bias' existence), but I also can't think of any sort of research that takes that approach. The objection strikes me as incoherent.

I certainly understand the frustration some conservatives feel that the question of implicit bias against them in the academy is not given full hearing. I've been an acolyte of long-standing of the theory that implicit bias is a problem worthy of correction, and it very quickly became evident to me that, if I were to consider my position in any way principled, I couldn't discount potential anti-conservative bias in the structures I work in. But my frustration is that everyone seems to be using the concept opportunistically -- if liberals are dismissive of anti-conservative bias, conservatives are dismissive of the prospect of racist or sexist bias. The red flags (Daniel Patrick Monyihan! Taboos! Larry Summers!) are all here, and it's very difficult to take Tierney's complaint as a serious one insofar as he immediately turns around and starts deriding the very same field as political correctness run-amok in other contexts.