What Does Bias Look Like?

So my post on the liberal slant in academia has garnered what I believe to be a record number of comments, many, even most of them, pretty angry.  And as I predicted, the positions are very much reversed from the normal take on such things.  Conservatives are explaining how bias can be subtle and yet insidious; and liberal, many of them academics are saying that you can't simply infer bias from statistical underrepresentation, and sarcastically demanding to know whether I really think that people are asking candidates for physics professorships who they voted for in the last election.

They're all right, of course: you can't simply infer bias from statistical underrepresentation, and yet bias can be subtle and yet insidious.  I thought it might clarify the argument a bit to outline how I think bias works in institutions, even though much of it should be old hat, particularly for social scientists.

Most people, when they are accused of being a member of an in-group that is excluding some other set of people, immediately define bias in the narrowest possible terms:  conscious, direct personal discrimination.  Did we make an explicit rule that no person of that persuasion could be hired?  No we didn't.  Well, then, no bias!

Those people offered their own alternate theories, which boiled down to:

  • Smart people are almost always liberal
  • Curiousity and interest in ideas is a liberal trait
  • Conservatives are too rigid and authoritarian to maintain the open mind required of a professor
  • Education erases false conservative ideas and turns people into liberals
  • Conservatives don't want to be professors because they're more interested in something else (money, the military)
  • Conservatives don't want to be professors because they're anti-intellectual
  • Conservatives hold false beliefs that make them ineligible to be professors

Some of these ideas are more plausible than others.  But I'm in the "left" camp on this: while there are certainly reasons to punish this sort of very overt discrimination more harshly than other sorts, that doesn't mean that once you've done away with the actual litmus test,  (or confirmed that you never had one) you cease to have a problem.  That shouldn't be controversial for either the liberals or the conservatives in this argument.  Lets review all of the different ways that bias can operate in a group:

Covert personal discrimination:  we know what this looks like:  "No Irish need apply".  Someone in the group doesn't want blacks, women, fat people, etc around them.  They aren't going to say so, because it's nominally taboo, but they are going to knowingly blackball anyone in the disdained category--or indeed, anyone they suspect might belong.

Unconscious personal discrimination:  this happens when you don't realize that you're letting your discomfort with some characteristic--whether it be race, weight, or voter registration--cause you to blackball an otherwise worthy candidate.  Academics in the comment threads assured me that they were smart, open-minded, and tolerant, and that all they cared about was ideas.

But it's never that simple--which is why, to the disgust of many of my frustrated readers, I will not endorse their assertions that the Tea Party is obviously a bunch of illiterate thugs who are trying to scare ordinary Americans into silence, or that anti-war protesters didn't really care about the war, they just hated America.  Even if I were tempted to agree with either of these sentiments, research shows that our interpretation of the motives of others is very dependent on whether they agree or disagree with us ideologically.

Both liberals and conservatives do this, and the people studied were almost certainly bright Yale students, so don't think that this must be a signature vice of other, dumber people than you.

And perversely, the more you think you are just deciding on the objective facts, like the quality of their work, the more possible it is that you are actually discriminating; research finds that people are most likely to discriminate on the basis of both politics and race when they have some other information with which to generate a plausible excuse.

Studies have shown that racism is so socially proscribed that people exhibit it nowadays only when they can plausibly deny--to themselves and to others--that they are biased. One meta-analysis of studies, for example, found that "discrimination against blacks was more likely to occur when potential helpers had more opportunities to rationalize decisions not to help" by invoking "justifiable explanations having nothing to do with race."
Munro, Lasane, and Leary found the same pattern of behavior in partisanship. The partisan college evaluators were willing to acknowledge that applicants they chose who shared their political loyalties had lower test scores--an objective fact--but they selected the candidates anyway by inflating the importance of the recommendation letters that came with applications. Accepting candidates merely on the basis of low test scores would have shown the evaluators were biased. Accepting candidates on the basis of recommendation letters--and arguing the letters were more important than scores--allowed the evaluators to plausibly deny that they were biased.
Introspection is thus not an adequate method for evaluating whether discrimination happens.

Stereotyping  People assume that if you're a member of some group, you probably have some trait that they have (accurately or not) associated with that group.  Thus, the astronomy department at the University of Kentucky apparently decided that anyone who attempted to reconcile Genesis with recent astronomical findings might well be a creationist.  They didn't hire him, even though he says he's a believer in standard evolution.

Reverse bias:  The out-group reacts to discrimination by refusing to join the in-group, and punishing those who do: "don't act white", "what, you think you're better than us?", "Professors are a bunch of self-satisfied hacks--who'd want to be one?"  They thus self-select out of the hiring pool.

I don't have to tell you that this sort of discrimination operates not just in interviews, but before and after: who does well in school, who gets selected for promotion or tenure.

But of course, institutions can be biased too.  Institutional arrangements can exacerbate individual bias, but they can also have structural features of their own that enhance it.  Numerically, the more people there are of a given persuasion, the more likely it is that at least one of them is going to discriminate against others.  The more skewed, the more likely--you have a cosy group dynamic that an outsider would disrupt, and you're more likely to develop a shared understanding that, well, you guys are better than other people.

A very tight job market is also going to make things worse.  The more candidates you have for the fewer jobs, the more eager you are to find reasons to get rid of otherwise qualified candidates.  I don't have to tell academics that their job market is about as brutal as it gets.

But the problems can go beyond this into institutional bias, which I also called structural bias in my original post.  These are systemic characteristics of an institution that cause it to discriminate against out groups.  Sometimes these features are even valid--personally, I'm not sure I favor relaxing the size and strength rules for firefighters so that more women and shorter minorities will pass the exam.  Live survivors of fires seem more important than parity.

However, there are institutional biases that are, to put it mildly, much more problematic.  I'll take a stab at listing many of them, most offensive to least:

  • Hostile work environment:  the group is actively hostile to you, ranging from behavior clearly designed to get you to leave, to disparaging remarks and demands that you account for anything bad recently done by a member of [insert group here]

  • Inhospitable work environment:  the group has a common bond that excludes you.  Often this means it's hard to get ahead--for example, a firm I worked for regularly took clients to strip clubs.  Since managers were supposed to groom clients, women didn't end up as senior managers.

  • Covering:   Ann Althouse's review of Kenji Yoshino's terrific book on the topic described this best: "Nevertheless, when he returns to Yale as a professor and deputy dean, he still feels the need to play down his sexual orientation. He avoids "gay examples" when teaching constitutional law. He attends law school functions without bringing the man he is dating. And he takes it to heart when a colleague remarks that he should be a "homosexual professional" and not a "professional homosexual. . .  Despite coming out to his friends, his parents and his colleagues, Yoshino still feels afflicted by the pressure to act as though being gay does not have much effect on his life. That is, he is required to "cover."  In order to survive, you must hide parts of your life that other people get to integrate with their work persona; if you're not willing to do this, you may be penalized--or you may decide not to risk it.

  • Affinity bias:  People like people who are like them.  They collect subordinates or students with whom they have common bonds.  They mentor those people, help them move up through the system, reproduce themselves throughout the organization.  People can't control this, really, but it makes a big difference.

  • Social Capital   Organizations naturally recruit from people who are close to the organization--if your friends and relatives do it, you understand the system, and it seems comforting and familiar rather than strange and scary. Those relationships often also give you an entree into the system.

  • Hidden tripwires Usually the dominant group doesn't even realize they are there. For example, the low pay (and increasing reliance on unpaid internships for entree) of journalism often excludes people who don't have, at the minimum, a family that could take them in and help cover the bills if disaster struck.  It's not surprising that the profession is so predominantly white and affluent even though everyone talks a lot about diversity.

As you can see, there are a lot of ways short of covert discrimination that a system can shut out diversity.

So what's happening in academia?  Is it simply that conservatives are less interested or qualified, or are being turned into liberals by the process of education, as my interlocutors have argued, or is there more going on here?

The evidence offered for proposition that it's all the hiring pool is pretty weak.  There's an awful lot of gross stereotyping: there are no conservatives because creationists can't be professors; because conservatives are hostile to new ideas; because reflective thinking inevitably turns you into a liberal.

Yet many of these are clearly generated by consulting lurid mental images of conservatives generated in liberal web fora, not any reliable data.  For example, multiple people argued that conservatives probably weren't in academia because they liked money and power, which caused them to abandon academia for think tanks.  This is an obviously crazy statement to anyone who knows anything about conservative think tanks, which may pay already famous people well, but which are not really that much more lucrative than professorships for the analysts who exit grad school to toil in their research arms.  And the think tankers don't get tenure.

Even if some of the theorized traits do recur at a higher frequency in the group--as creationism, obviously does--it's mindless prejudice to generalize from many to all.  And that's the kind of generality you'd need to explain a 266-to-1 disparity in social psychology, given that the population is 40% conservative and only 20% liberal.  In the professorship as a whole, the breakdown seems to be about 80% liberal-to-center-left, 20% everything else, which is still a remarkable skew.

Blacks are slightly more likely than Republicans to attend church weekly, and black churches are about as likely as white evangelical churches to be creationist.  Yet if someone told you that the reason there are too few black professors is that creationists don't make good professors, you'd think they were bonkers.

Of course, one might argue that we know blacks have been discriminated against, and can presume that this persists.  Of course this is true.  My point is that just because you can generate a story to explain the skew towards liberalism, does not mean that this story is particularly convincing.

One of the things the legacy of racism has taught us is just how good dominant groups are at constructing narratives that justify their dominance.  Somehow, the problem is never them.  It's always the out group.  Maybe the out group has some special characteristic that makes them not want to be admitted to the circle--blacks are happy-go-lucky and don't want the responsibility of management, women wouldn't deign to sully themselves in commerce, Jews are too interested in money to want to attend Harvard or go into public service.  These explanations always oddly ignore the fact that many members of the out-group are complaining about being excluded.

More troubling is that these volitional arguments are almost always combined with denigration: the out group is stupid, greedy, mean, violent, overemotional, corrupt . . . whatever. As indeed these arguments are when they're deployed against conservatives in my comment threads.  In fact, it seems clear to me that many commenters have taken the underrepresentation of conservatives in academia as vindication of their beliefs--if conservatives can't make it in academia, that proves that conservatives are not smart, and liberal ideas must be better.  This is possible, of course.  It's also possible that academics are validating their own bias by systematically excluding those who disagree with them.

So while in theory, it's true that you can't simply reason from disparity to bias, I have to say that when you've identified a statistical disparity, and the members of the in-group immediately rush to assure you that this isn't because of bias, but because the people they've excluded are all a bunch of raging assholes with lukewarm IQ's . . . well, I confess, discrimination starts sounding pretty plausible.

When that group of people is assuring you that the reason conservatives can't be in charge is that they do not have open minds . . . when the speed and sloppiness of their argument make it quite clear that they rejected the very possibility of discrimination without giving it even a second's serious thought . . . well, I confess, it starts sounding very plausible.  More plausible than I, who had previously leaned heavily on things like affinity bias to explain the skew, would have thought.

Moreover, what evidence we have does not particularly support many of the alternative theories. For example, the liberal skew is strongest at elite universities.  This is not consistent with the notion that education is turning all the conservatives into liberals, or that they're not interested in becoming professors.  I'd say it's more consistent with the possibility that they're disproportionately having a hard time getting hired, or retained.

And the skew goes way back--it shows up during the Eisenhower administration, and seems to have been locked in during the mid-1970s.  That's not really consistent with a story about how Republican anti-intellectualism has driven professors out of the conservative camp.  It's rather more consistent with conservatives reacting to exclusion by becoming anti-intellectual, though I don't know that this is actually the case.

And with apologies to all the brilliant, open-minded, scientifically grounded liberal academics who suggested this one, it's also not because academia simply weeds out illogical people who can't handle the scientific method.  Professors are overwhelmingly pro-choice, pro-gay-marriage, anti-military, and in favor of redistribution and regulation programs.  These aren't a matter of logic and scientific evidence; they're value judgements.  Moreover, they cluster in a way that suggests something other than rational analysis driving the decision--why should your views on military operations in Iraq, or climate change, be correlated with your views on abortion?

Note that this also excludes the thesis that professors aren't really that liberal, but just self-identify that way because conservatives are so terribly anti-intellectual.  Professors are mostly of one mind on most of the major liberal political issues.

There are some theories I find more plausible; Jonathan Haidt notes that conservatives tend (on average) to score lower on certain personality traits that may be required in an academic--though even then, I'd note that tests designed by liberals are going to implicitly validate liberal norms.  And, of course, there are skews in places like the military, though not as strong ones; if one profession skews one way, then definitionally, some other profession has to skew in the opposite direction.

But overall, it seems unlikely to me that the skew we're seeing is simply a result of conservatives all suddenly wanting to be elsewhere for the last thirty years.  I find it even less likely that it's simply because all conservatives are idiots, or uneducated.

But is this even a problem?  Do conservatives have a right to a place in academia?

There are three potential arguments for why it's a problem.  One if the harm it does to conservatives.  But the others are the harm it does to academia, and to the rest of us.  I think the latter are by far the bigger problems.  Not to trivialize the conundrum faced by conservatives who want to be professors . . . but it's not like they're ending up as longshoremen.  The other two problems are much more broadly harmful.

Excluding conservatives means that academia is losing the trust of the more conservative members of society.  Academics are incredulous and angry about this--the way that many whites are when they hear rumors are spreading in the black community that AIDS was deliberately created and released by the government to destroy them. But the mistrust of the government in the black community is not crazy--not after things like the Tuskeegee Syphilis experiment.  Nor is it crazy to think that academia wields its prestige like a club against conservative ideas--or even conservatives themselves, as with the steady stream of studies that characterize conservatives as authoritarian, less open to experience, and so forth.  Conservatives point out that the questions are more than occasionally loaded, and the studies are done on psychology students, an overwhelmingly liberal group whose few conservatives may not look much like conservatives in the wild.  Yet the academics in question more than occasionally participate in the denigrating connotations this information is given in the media.

Which hints at the other problem with excluding conservatives: it makes scholarship worse.  Unless we assume what to many liberals is "proven" by their predominance in academia--that conservative ideas have no merit--leaving conservatives out means that important viewpoints are excluded.  We are never the best interrogators of our ideas.  It requires motivated critics to lay bare our hidden assumptions, our misreading of the data, our factual inaccuracies.  No matter how scrupulously honest you try to be, you are no substitute for an irritated opponent thinking, "That can't possibly be right!"

If you build a group with the same assumptions, you can all too easily go wrong.

Moreover, as Haidt points out, that group develops sacred taboos that can't be violated.  If facts threaten the sacred space, facts get jettisoned.  Think of the creationist museum--or Larry Summers getting attacked by a swarm of angry critics for suggesting that it was possible that inborn differences in the distribution of intelligence might explain why men--who have a higher observed variance in math ability--are more likely to be found in the sciences.

Now, Summers could be wrong---as I say, I am inherently suspicious of narratives that offer neat explanations for why the dominant position of one group can't be changed.  But the critics did not rush to explain why this was unlikely to be right.  They furiously rushed to punish him for having said it.  They were angry about sexism, not science.

Yet one more reason why I am suspicious that liberals' unshakeable commitment to scientific rigor is what forces them to exclude conservatives tainted by association with creationism.

I don't say this to denigrate liberals--obviously, conservatives have their taboos too.  But it's healthier if different groups, with different taboos, all have a place in the quest for truth.  Monoculture is as unhealthy for ideas as it is for agriculture.