Liberal bias in academia?

images.sodahead.com

by Massimo Pigliucci

A few days ago I was asked by a Washington Times reporter, Emily Esfahani Smith, to comment on a soon to be published paper concerning the issue of liberal (or, rather, anti-conservative) bias in the academy. I am weary of the Washington Times, a paper that is well known (among liberals) to have a decidedly conservative (or, rather, anti-liberal) bias of its own, but agreed to respond in writing to Emily’s questions. The piece was published a few days later, and I was actually quoted pretty much correctly (even though the piece itself did have the predictable slant, featuring a title that goes far beyond the findings of the paper referred).

I have gotten into hot water before concerning this particular issue, with none other than Jonathan Haidt, a sociologist whose research is often interesting, whose data are sometimes questionable, and whose conclusions gingerly reach across the ought / is divide, despite his (too) loud protestations to the contrary. But the new paper was not authored by Haidt (though, predictably, he commented for the Washington Times piece), so let’s take a fresh look. At the bottom of this post I am appending my full exchange with Smith (the reporter), for the sake of the record.

Here is the abstract of the original paper, entitled “Political Diversity in Social and Personality Psychology,” and co-authored by Yoel Inbar and Joris Lammers, both at Tilburg University:

A lack of political diversity in social and personality psychology is said to lead to a number of pernicious outcomes, including biased research and active discrimination against conservatives. In two studies, we investigate the actual and perceived political ideology of a large sample (Study 1: N = 508; Study 2: N = 292) of social and personality psychologists. We find that there is more diversity of political opinion than is often assumed; conservatives are a substantial minority among social and personality psychologists. Second, we find that respondents significantly underestimate the proportion of conservatives among their colleagues. Third, we find that conservatives fear negative consequences of revealing their political beliefs to their colleagues. Finally, we find that conservatives are right to do so. In decisions ranging from paper reviews to hiring, many social and personality psychologists admit that they would discriminate against openly conservative colleagues. The more liberal respondents are, the more willing they are to discriminate.

As I say in the interview, the statements of some of the people interviewed are indeed troubling, and even surprising, considering that they were making them freely in the course of a study that was going to be published. I guess at the least those participants that admitted to anti-conservative bias should be commended for their self-knowledge and candor, though little else.

A few comments on the paper itself, before we move to the broader issue. First off, notice that the authors found “more diversity of political opinion than is often assumed; conservatives are a substantial minority among social and personality psychologists.” So, contra Haidt, there doesn’t seem to be a need for “affirmative action for conservatives” to restore objectivity to the academy.

Second, as usual with these studies, the conclusions ought to be taken with a big grain of salt, for a variety of reasons, starting with the fact that the survey was based on self-selected respondents — which means that there is no way to know to what extent these people are a representative sample of the target population — and that the questions concerned behaviors that the subjects might engage in themselves, not actual data about factual behaviors (i.e., the paper says precisely nothing about actual, as opposed to hypothetical, anti-conservative bias in the academy).

Third, the ideology of the individuals was not assessed independently, but based on self-rating, again with all the caveats that this engenders (for instance, I’m pretty sure that what non-American colleagues meant by terms like “conservative” and “liberal” was quite different from what most American colleagues meant when using the same terms).

Fourth, conclusions such as “The more conservative respondents were, the more they had personally experienced a hostile climate” are a bit haphazard. It could be instead that the more conservative one is the more paranoid he is, and therefore the more likely he thinks he has experienced a hostile climate. Indeed, some level of paranoia can be found also in extreme liberals I know, who sometimes think that everyone is out to get them because of their opinions about politics, or because of their gender, or ethnicity, but then often (not always) come up way short when someone asks them for specific evidence to back up their claims.

Fifth, I find some conclusions of the study to be both predictable and highly uninformative: “For each question, respondents expected their colleagues to be more inclined to discriminate than they themselves were.” Ah yes, we are all better, more objective, more fair, and so on, than other people. If true, of course, the amount of actual bias is going to be minute.

Sixth, “we asked whether they would evaluate papers and grant applications that seemed to take a conservative perspective negatively.” Well, I would. But I would also evaluate negatively a paper or grant that takes a liberal perspective, because I happen to think that scientific papers ought to strive for having no ideological perspective whatsoever (they are not op-ed pieces, or works in political philosophy). And psychology, last time I checked, was presenting itself as a science. Incidentally, the authors immediately admit, in the same phrase: “but we did not ask whether they would evaluate work that seemed to take a liberal perspective negatively.” Well, why on earth not?

Nonetheless, two of the majors findings of the Inbar and Lammers paper confirm my early impression of Jonathan Haidt’s work in this area:

> “In two out of three of the domains we assessed, our sample already meets or exceeds the 10% ‘quota’ that Haidt (2011) suggested as a 10-year target for ideological diversity in social psychology.”

> “Haidt’s (2011) search for conservatives (by show of hands at his talk, web searches, and asking social psychologists to name a conservative colleague) suggested that there are almost no conservative social psychologists. This appears not to be the case.”

Please, in case you missed it, notice the source of Haidt’s data, which drew my original criticism: “by show of hands at his talk, web searches, and asking social psychologists to name a conservative colleague.” And if you thought “Haidt (2011)” referred to a peer review study, think again. It’s just a talk downloaded from his web site. The standards of academic citations in the social sciences must have gone significantly down lately.

Perhaps the most problematic aspect of the Inbar and Lammers paper, however, is the above mentioned lack of the obvious control: they didn’t ask conservatives about their biases (nor, for that matter, did they ask another obvious control group: politically neutral or middle of the road faculty). Of course, even if they had, one would still have to make a good number of more or less debatable assumptions (such as that the various groups are all equally trustworthy in their self-reporting, for instance), but it would have been better than nothing. As it is, the paper’s findings are next to impossible to put into proper perspective.

The broader issue, of course, is one that the Washington Times article didn’t even raise, despite the fact that I did in my response to the reporter’s questions (see transcripts below): I’m pretty confident that if one did a similar study of, say, the Wall Street culture one would find a huge anti-liberal “bias,” and one that likely isn’t even close to Haidt’s 10% bar for liberals. But such findings wouldn’t surprise anyone, and possibly wouldn’t even demonstrate a bias in the ominous sense implied (ok, shouted) by the Washington Times headline. Most (obviously, not all) intelligent liberals are simply much more likely to go to work elsewhere because they do not share the values around which Wall Street culture is built (extreme levels of competition, individualism, and greed). There is no need to invoke any conscious systematic anti-liberal discrimination at banks and investment firms, because most people who find themselves on the left side of the political spectrum are just not going to be interested in spending their lives that way (again, obviously, there are plenty of exceptions, but the Inbar and Lammers paper shows that this is true also about conservatives in the academy).

Similarly, I bet dollars to donuts that bright conservative types are simply going to find academic life unattractive compared to what else they could do with their talents. After all, they would make relatively little money, they would need to spend a large number of years in extremely low paid and temporary positions (graduate student, postdoc, untenured professor), they would have to withstand constant razor sharp criticism of their ideas (for grant proposals and publications), and they would find themselves in an institution that values things for which they often have little sympathy, such as open discourse, critical thinking, equality, questioning of authority, and so on. (I’m not making this up from the particular standpoint of my liberal bias, there is now plenty of evidence that the conservative mind doesn’t like that stuff.) Why bother?

So, while actual discrimination against certain political points of view ought to be resisted within the academy — on the sole ground that it betrays one of its fundamental values — and while academic liberals have a scholarly duty to respectfully engage with colleagues of different ideological stripes, let’s be very careful before shouting “bias!” on the basis of questionable evidence (Haidt) or highly incomplete experimental designs (Inbar and Lammers). As for newspapers, well, that’s a whole different can of worms.

_____

Appendix: My responses to Emily Esfahani Smith’s questions for the Washington Times article.

EES: What do you think about the study’s conclusion that “In decisions ranging from paper reviews to hiring, many social and personality psychologists admit that they would discriminate against openly conservative colleagues. The more liberal respondents are, the more willing they are to discriminate”?

MP: Taken at face value the results would be disturbing. But they cannot be taken at face value, for a variety of reasons. First, the paper has not actually gone through peer review, and as such it is actually improper — by academic standards — to discuss it in the press. [The WT article claims that the paper is now in press in Perspectives on Psychological Science, though neither the download web site nor the manuscript itself carry that information.]

Second, the methodology of the paper has a number of flaws, acknowledged by the authors themselves. To name but a few: the study was conducted via a self-selecting electronic survey, which means that we do not know the extent of bias in the sample — it could be huge. Moreover, the authors crucially did not investigate the extent of bias against liberal-sounding papers and grant proposals, only the conservative ones. The academy has a strong ethics of reducing bias as much as possible, so it is to be expected that people would reject papers and grant proposals that smacked of clear ideological bias — one way or the other. As for hiring practices, I have never ever seen any job candidate being asked about his/her ideologies, so it seems to me next to impossible that conservatives would be thus discriminated against. Interestingly, the study investigates likely behaviors, not actual outcomes, and one should start by demonstrating that there is a problem first, then inquire into its possible causes. As it is, the authors addressed the causes of a phenomenon of which they have not established the existence. Peculiar, no?

EES: What do you think of Haidt’s suggestion of having affirmative action in the academy for underrepresented conservative scholars?

MP: Not much. First off, even the authors of this paper acknowledge that Haidt’s “quota” is already met in most cases. Second, political ideologies do not constitute protected categories as far as affirmative action is concerned (unlike, say, gender and ethnicity) and for good reasons. Should we have protection for feminists in the academy? Marxists? I’m sure if the authors conducted studies on those groups they would find even more extreme results than those they report in this paper.

Incidentally, the paper explicitly contradicts some of Haidt’s “findings” about conservatives being a tiny minority in the academy. Which is no wonder, considering that Haidt reached that conclusion based on “show of hands at his talk, web searches, and asking social psychologists to name a conservative colleague” — hard to imagine a more unscientific way of collecting data.

EES: How many conservative academics / scholars do you know? How many are in your department?

I actually don’t ask my colleagues about their ideologies, so I don’t know. I know indirectly of both conservatives and progressives, and yes the latter probably are in a majority. But there may be countless reasons other than discrimination for this. How about a survey of Wall Street high level employees? Wanna bet that you would find evidence for “discrimination” against liberals there?

Fourth, have you ever witnessed a conservative being discriminated against because of his political affiliation?

No, never in my 30 years of service in 8 scientific and humanistic departments.

EES: In choosing between two equally qualified job candidates for a professorial opening, would you be inclined to vote for the more liberal candidate?

No, and as I said above, I wouldn't even know. Campus Human Resources are very careful in regulating what sort of questions you can ask a candidate during a job search (I know because I am conducting one now, as a department Chair), and you have to ask the same questions to every candidate.