Wednesday, February 09, 2011

Political bias in academia, revisited yet again

Lots of people who should know better seem to be excited about this silly John Tierney article on this Jonathan Haidt presentation about political bias in the academy.

Haidt has taken things that surely by now everyone knows about politics and the academy-- e.g. that the professoriate leans left compared to the American populace as a whole-- and dressed them up in various bits of metaphor and jargon. Some of the metaphors are good ones; I like the magnet image. But they're metaphors, not evidence. Throw in what on usenet we once would have called ObLarrysummers, and the cute fact that it's social psychologists in the audience-- people who think of thesmelves as good at analyzing patterns of bias!-- and we're done. We have a just-so story about the evolution of taboos around topics having to do with race and gender straight out of the political correctness wars circa 1992, retold in a way that emphasizes what's supposed to be clear as soon as one says "taboo"-- that there's something magical and superstitious rather than rational about it.

But here's what we don't have:
1) Any meaningful new evidence about the political imbalance in the academy.
2) Any meaningful evidence at all of bias.
3) Any attempt whatsoever to sort out the competing explanations for the political imbalance that are the heart of any serious conversation on the topic.

Stipulating that the academy skews left compared with a random population sample (which is what Haidt's facially meaningless phrase "statistically impossible lack of diversity" actually means), we are always faced with the question of why. Random population samples are, after all, not something one usually comes across in daily life; even professional opinion pollsters trying to get such samples have real trouble doing so, because, e.g., the people who own landline phones, as opposed to no phones or only cell phones, aren't a random population draw, and if you survey using the phone book you'll get a skewed sample.

The common explanations:

1) Initial self-selection. Different careers reflect different mindsets and values, and the sorting into those careers early in means that there's no reason to expect them to look like random population draws. No one is surprised when the military skews right or the Peace Corps skews left. There's probably some sorting as between, say, those who go into business or finance of various kinds out of college and those who pursue PhDs. This point can be put in a value-neutral way, or said with whatever sanctimonious inflection one likes. "Conservatives are practical, and practical people go into business, while head-in-the-cloud unrealistic people want to go into an ivory tower." "Liberals believe in critical thought and reasoning, so they are naturally attracted to careers that value it; conservatives don't like having their assumptions questioned, so they naturally turn away from intellectual careers." Note that one has to go through another round of this in order to think about why any one discipline is politically skewed: economists, engineers, philosophers, and literary theorists are plausibly different groups of people to begin with, and they sort themselves out accordingly.

2) Selection and screening mechanisms. The favorite sanctimonious left-wing explanation for why there aren't conservatives in the academy is that the academy's various hurdles, from grad school admission onward, screen for intelligence, and conservatives are less intelligent people.

3) Change over time. The second-favorite sanctimonious left-wing explanation is that academics, who might start out with no real political views at all, are influenced by evidence and argument, and these favor movement toward the left, or at least toward a point farther left than the American median. Sometimes this is distinctively about the post-2001 years, but sometimes not. Common symptoms of this argument are the mindless repetition of John Stuart Mill's comments about conservatives being stupid or Lionel Trilling's "irritable mental gestures." Equally uninteresting is something like this: university professors, with their tenure and their taxpayer-funded salaries become left-wing because they resent their dependence on hard-working productive people who have real jobs; they spend their whole lives cut off from and failing to learn about markets, competition, and business, so what do you expect?

4) Bias-- whether unintended (the hostile environment "locker room talk" Haidt discusses) or intended ("don't tenure him, he's a Republican."). These are meaningfully different from each other, but both are problematic.

Sorting these out requires hard longitudinal work. When, in the decades from freshman year of college to tenure, does the political skew get introduced? Are conservatives disinclined to enter a given discipline at all, maybe just because they're disproportionately more interested in other things? Trying to enter the discipline but failing because they're not smart enough? Entering the discipline but becoming more left-wing over time? Trying to enter but dropping out because they're discouraged by groupthink? Or trying to enter but being kept out by overt bias? The answer may be "some of each," of course, and the dynamics may feed on each other-- but that's a hypothesis in its own right.

A slice-of-time show of hands at an academic conference in response to the question "who here is a conservative?" does absolutely nothing to sort any of this out. Neither does a google search on the phrases "liberal social psychologist" and "conservative social psychologist." Those two data only show what we already knew: the professoriate skews left. The two solicited anonymous e-mails complaining about uncomfortable environments provide anecdotal support for "unintended bias," but, well, not very much. And the cute fact that the presentation is being offered to social psychologists who study the emergence of bias in groups has some nice rhetorical effect in the room, but still doesn't add any insight.

I freely admit that sorting these questions of causation out is extremely hard. But anyone who claims to be talking about this subject and doesn't even acknowledge them, to say nothing of trying to solve them, hasn't added to our knowledge, and certainly hasn't provided any reason for supporters of the bias hypotheses to run around claiming vindication. The fact that the presentation (note: not an article, not a paper, a power point presentation at a conference) got a NYT write-up from a sympathetic columnist doesn't make it any more of a contribution to our knowledge.


Brian Leiter said...

Jacob, it' hard to see why you wouldn't want to sign on to #2--after all, libertarians are not conservatives, and there are way more libertarians in the academy than in the population at large too. Doesn't that also fit with the thought that to be an Amercian-style "conservative" you really do need to be a dumb-dumb?

Jacob T. Levy said...

First of all, I didn't say the argument was untrue; I said it was sanctimonious. Its truth remains, as far as I can tell, empirically untested; but quick recourse to it always seems to me smug, self-satisfied, and question-begging.

Second, at a hunch, the American professoriate is to the left of the American bar, and of the finance industry, and of the information technology industries. Certainly I'll bet that the liberal arts academy is to the left of those sectors. And I wouldn't be willing to give odds that we are in any meaningful systematic way smarter than all of those folks.

Third, I decline the invitation to "want to sign off" on an argument just because it would include me on the right side of a divide between smart people and dumb people! What a dreadful invitation to offer in the first place.

There are particular doctrines about which I'm willing to say #2, e.g. creationism. But I also believe in the phenomenon of "ideas so dumb that only an intellectual could believe them;" that is, I think there are characteristic mistakes and pathologies of academic mindsets. That alone counsels against leaping to #2, and embracing self-satisfaction.

I unscientifically treat #1 as the null hypothesis for the overall pattern, even when I know particular cases of left-wing bias or right-wing stupidity. I think that there's a broad kind of fairness to academic judgments (albeit with plenty of individual exceptions), so bias doesn't explain much; and that the ideological gaps from one discipline to another, or from academia to the other high-intelligence professions, mean that simple intelligence doesn't explain much.

John Quiggin said...

I thought we sorted this one out a while ago. Holding other things (including education) equal, higher income produces more right wing voting, for obvious reasons, while (holding income constant), higher education produces more leftwing voting, for which sanctimonious lefties have an obvious explanation (I'm sure there was some data on this, but I can't find it now).

So, there is no puzzle about the academy leaning left - academics vote the way you would expect, given very high education and not so high income.

alex said...

To John Quiggin: income and education are insufficient to explain the liberal leanings of academics. One way to see this is to observe that there is a lot of variation in the number of conservatives by field in academia, whereas the educational attainments do not vary much from field to field, and salaries only vary somewhat. For example,

1. While 61% of education professors label themselves as liberal, 88% of english literature professors do. I do not believe professors in education earn much more than professors in education.

2. While 55% of economics professors label themselves as liberal, 80% of philosophy professors do. However, economics is the higher paying discipline among the two.

Source for the data:

Brian Leiter said...

Jacob, thanks for a more serious reply than my tongue-in-cheek question deserved!

ChrisP said...

I think a lot of these explanations are out of date. Academia today isn't leftist so much as left-liberal. Which is to say, like the New York Times rather than The Nation.

Academics are all fundamentally afraid of one thing: looking stupid. It is easier to avoid looking stupid if you are non-judgmental and permissive -- especially with regard to people from other cultures about which you know little. This cultural relativism then erodes any overly emphatic assertion of classical American virtues -- either the person pronouncing them is likely to step onto turf he doesn't know well, or he is unwilling to offend people from other cultures. Since most universities are highly multinational these days, the safest path for an academic is therefore to be a liberal of a permissive and cosmopolitan type.

Jacob T. Levy said...

Setting aside the condescending "explanation," I don't see that any of the discussion here or in the original article is contradicted by your leftist/ left-liberal distinction. In my post I said "leans left compared to the American populace as a whole," which is agnostic about whether they're NYT left-liberal, or center-left-only-by-American-standards-which-is-really-center-right, or unreconstructed socialists, or far left cultural relativists, or whatever, under whatever approving or disapproving description. Haidt just used a contrast between "liberal" and "conservative," which doesn't make any claim about whether "liberal" encompasses or excludes progressives, socialists, greens, etc., etc.

Bill Harshaw said...

Are you maybe missing part of the point? Instead of worrying about why academics are liberal, isn't the thrust of the Tierney/Haidt piece that academics have become a supercilious, self-satisfied in-group which is maintaining itself by dissing the minority? IMHO you don't need any big theories why an in-group looks down on an out-group, it's what people do, whether it's country-club conservatives, fraternities, or whatever.

Aretae said...

Mr. Levy,

There's one paragraph/line in there that is important that you don't seem to have hit:

“Anywhere in the world that social psychologists see women or minorities underrepresented by a factor of two or three, our minds jump to discrimination as the explanation,” said Dr. Haidt, who called himself a longtime liberal turned centrist. “But when we find out that conservatives are underrepresented among us by a factor of more than 100, suddenly everyone finds it quite easy to generate alternate explanations.”

He's made a fabulous claim here, in the social scientists language. EITHER give up the discrimination line on Minorities, or keep it on Conservatives in Academia. You can't keep it in one place and ignore it elsewhere.

He's given a dilemma with two horns, and no obvious way out, and in a relatively unignorable fashion. Really well done. It didn't need new evidence. It needed insider status to call out the hypocrisy.

John Quiggin said...

"While 55% of economics professors label themselves as liberal, 80% of philosophy professors do. However, economics is the higher paying discipline among the two."

There's a misplaced "However" here. That's exactly what the standard model predicts.

I don't suggest that everything about voting behavior is explained by just two variables. But they seem sufficient to explain the main results of interest.

alex said...

To John Quiggin: Err, yes, the however there was misplaced. Nevertheless, despite my mistake, I believe the overall point is correct. Compare for example education vs mathematics vs, philosophy, which according to this data

have nearly identical salaries, but for which percentages of conservatives according to the Rothman-Lichter-Nevitte survey are, respectively, 29, 17, 5.

As for whether "main results of interest" are explained by the model you propose, I would submit to you that something additional is needed to explain the enormous imbalance between conservatives and liberals in english literature: 88% vs 3%.

Anonymous said...


The unspecified premise in the argument (technically, it's a tu quoque, not an argument) you're so impressed with is that political ideology is the same kind of distinction between people as race, gender, and so one. This is a controversial claim. If he's got a good argument for treating them similarly, he should clue us in.

TGGP said...

Razib compares moderates, conservatives and liberals for education and vocabulary (the General Social Survey's proxy for IQ) here.

Daniel Klein writes a lot about this topic, which is a bit odd since his main axe to grind his libertarianism which if anything is likely overrepresented in academia. He did have one paper with a cluster analysis distinguishing various "left" groupings, as in the NYT vs Nation comment. I recall he had another one showing that among phd holders within a discipline, conservatives were least likely to work in academia.
Actually, on further inspection it may just be one paper I was recalling.

I was a bit surprised to the reference to the "American bar", since I associate lawyers with liberalism. But maybe I don't hear as much about business lawyers. On the other hand, Andrew Gelman's analysis in "Red State, Blue State" showed professionals have shifted overwhelmingly to the left, and lawyers seem to be a prime example of that. I.T is different, I work in software and didn't even spend the full four years at uni.

Michael Giberson said...

Jacob, it seems clear that Haidt didn't intend to prove bias in social psychology, so your observation that Haidt didn't prove bias doesn't appear relevant.

The failure to examine why bias exists in the field of study may be more interesting.

I take Haidt's address as trying to sensitize social psychologists to their shared biases, draw their attention to the likely pernicious effects of shared bias among social psychologists, and then to support a 'policy proposal' within the Society of Personality and Social Psychology intended to help remedy the problem.

Haidt's lack of diagnosis raises questions about the adequacy of the proposed intervention. Without a firm understanding of why the academy is working the way it is working (to produce very few conservative social psychologists), I'm not sure why the SPSP should want to buy into the proposed solution of affirmative action for conservatives.

Of course, if we thought like most social psychologists do, maybe it would be obvious to us that affirmative action is the correct approach -- maybe he didn't think his audience needed convincing on this point (or maybe his own biases led him not to realize that the issue is problematic).

Personally, as a way to root out some of the pernicious effects of shared biases, I'd recommend a more disciplined attempt at inter-disciplinary research on frontiers where social psychology overlaps with political science, economics, and other fields with somewhat different shared biases.

Anonymous said...

Brian Leiter's comment more or less illustrates a solution: when you have the sort of atmosphere where someone can rise to be a gatekeeper in their profession when their reflexive response to questions about group X is to call them juvenile names, you can be certain that people from group X will generally move away from that profession. Maybe they shouldn't, or maybe it's good that they do, but there's nothing really complicated about it. At the same time, though, Leiter's comment gets at a point which is implied in the response of a previous anonymous poster to aretae. The point here is that what many academics, even if they're much less obnoxious than Leiter, do sincerely believe that the reason for the ideological skew of academia is just that conservatives are just idiots. But if they're correct, it raises some interesting questions: for example, does it mean that higher education is necessarily "liberal indoctrination", and should this fact be celebrated? Should a university education basically be advertised in these terms? How would this effect universities as a whole and their relation to society? Or do academics have to keep quiet about their true activity? It's an especially important question if you're claiming that academic knowledge is transferable to politics. If we know, after all, that academics are liberals because academics are smarter, and that conservatives must therefore be kept out of academia, can we really seriously justify allowing conservatives to, say, edit newspapers? Or even vote?

Dave said...

I feel like I'm missing something important, which is how people define conservative or liberal. Are we asking whether LIBERAL arts majors describe themselves as liberal? Is there more to this than a label, like a survey of values? There is alot about working for a university that is liberal by definition, too, right (state funding, affirmative action, tenure)?

Jacob T. Levy said...

The question is how one subjectively aligns with extramural political labels, whether ideological or partisan. "We" aren't defining "liberal" at all. The people identifying one way or another do the definitional work in their heads.

As for the rest: even if tenure were somehow inherently liberal, which seems like a pretty weird idea to me, you can't generalize from the characteristics of the institution to the subjective views of the employees. "At-will employment without benefits is inherently right-wing, so we should expect all temp workers to be right-wing"-- no. And it'd be impossible to really nail down what was "inherently" left or right at that level of abstraction. Elite private universities are incredibly wealthy institutions that maintain the class structure through affirmative action for underqualified children of wealthy alumni, and by multiplying the consequences of good or bad primary and secondary education so that those who come from bad school districts get left ever-further behind. Under that description, they're "inherently" conservative. How does that balance against the traits you listed? What would balancing them even mean?

It's simply a mug's game to try to map the supposedly objective characteristics of institutions onto the subjective views of their members.

Noah said...

I'm increasingly worried that the underrepresentation of conservatives in science is bad for science. It's quite dangerous to have one of the major parties be hostile to science. It seems likely that one of the causes of republican's anti-science stance is the scarcity of conservative scientists. If scientists are immediately thought of as "liberal" then there's less social discomfort for conservatives who disregard scientific conclusions.

I'm not sure what the right solution to this problem is, but it does suggest that some sort of outreach to promote more conservatives becoming scientists would benefit science.