Bryan Caplan  

Diversity of the Mind

A penny for our thoughtless so... Henderson on Weidenbaum...
A bunch of my favorite social scientists, including Philip Tetlock, Jonathan Haidt, Lee Jussim, and Charlotta Stern, have co-authored an amazing article on the scarcity and value of political diversity in social psychology*, forthcoming in Behavioral and Brain Sciences.  Full pre-print version here.  Highlights:

1. Ideological imbalance is massive.
Inbar and Lammers (2012)... set out to test Haidt's claim that there were hardly any conservatives in social psychology. They sent an email invitation to the entire SPSP [Society for Personality and Social Psychology] discussion list, from which 2923 individuals participated. Inbar & Lammers found that 85 percent of these respondents declared themselves liberal, 9 percent moderate, and only 6 percent conservative (a ratio of 14:1). Furthermore, the trend toward political homogeneity seems to be continuing: whereas 10% of faculty respondents self-identified as conservative, only 2% of graduate students and postdocs did so.
2. Ideological diversity has a more scientific benefits than mere demographic diversity:
Research in organizational psychology suggest that: a) the benefits of viewpoint diversity are more consistent and pronounced than those of demographic diversity (Menz, 2012; Williams & O'Reilly, 1998); and b) the benefits of viewpoint diversity are most pronounced when organizations are pursuing open-ended exploratory goals (e.g., scientific discovery) as opposed to exploitative goals (e.g., applying well-established routines to well-defined problems; Cannella, Park & Hu, 2008).

Seeking demographic diversity has many benefits (Crisp & Turner, 2011), including combating effects of past and present discrimination, increasing tolerance, and, in academic contexts, creating bodies of faculty who will be more demographically appealing to students from diverse demographic backgrounds. However socially beneficial such effects may be, they have little direct relation to the conduct or validity of science. Viewpoint diversity may therefore be more valuable than demographic diversity if social psychology's core goal is to produce broadly valid and generalizable conclusions. (Of course, demographic diversity can bring viewpoint diversity, but if it is viewpoint diversity that is wanted, then it may be more effective to pursue it directly.) It is the lack of political viewpoint diversity that makes social psychology vulnerable to the three risks described in the previous section. Political diversity is likely to have a variety of positive effects by reducing the impact of two familiar mechanisms that we explore below: confirmation bias and groupthink/majority consensus.
3. Ideological diversity helps defuse confirmation bias:
Confirmation bias can become even stronger when people confront questions that trigger moral emotions and concerns about group identity (Haidt, 2001; 2012). Further, group polarization often exacerbates extremism in echo chambers (Lamm & Myers, 1978). Indeed, people are far better at identifying the flaws in other people's evidence-gathering than in their own, especially if those other people have dissimilar beliefs (e.g., Mercier & Sperber, 2011; Sperber et al., 2010). Although such processes may be beneficial for communities whose goal is social cohesion (e.g., a religious or activist movement), they can be devastating for scientific communities by leading to widely-accepted claims that reflect the scientific community's blind spots more than they reflect justified scientific conclusions (see, e.g., the three risk points discussed previously).
4. Why do liberals predominate?  The paper considers all the main hypotheses.  Results: Homophily and differences in interests are big parts of the story; differences in intellectual ability are not.  But there is also strong evidence that (a) hostile climate and (b) conscious discrimination play big roles too.  The evidence on conscious discrimination shocks even me:
Inbar and Lammers (2012) found that most social psychologists who responded to their survey were willing to explicitly state that they would discriminate against conservatives. Their survey posed the question: "If two job candidates (with equal qualifications) were to apply for an opening in your department, and you knew that one was politically quite conservative, do you think you would be inclined to vote for the more liberal one?" Of the 237 liberals, only 42 (18%) chose the lowest scale point, "not at all." In other words, 82% admitted that they would be at least a little bit prejudiced against a conservative candidate, and 43% chose the midpoint ("somewhat") or above. In contrast, the majority of moderates (67%) and conservatives (83%) chose the lowest scale point ("not at all").

Inbar and Lammers (2012) assessed explicit willingness to discriminate in other ways as well, all of which told the same story: when reviewing a grant, 82% of liberals admitted at least a trace of bias, and 27% chose "somewhat" or above; when reviewing a paper, 78% admitted at least a trace of bias, and 21% chose "somewhat" or above; and when inviting participants to a symposium, 56% of liberals admitted at least a trace of bias, and 15% chose "somewhat" or above. The combination of basic research demonstrating high degrees of hostility towards opposing partisans, the field studies demonstrating discrimination against research projects that are unflattering to liberals and their views, and survey results of self-reported willingness to engage in political discrimination all point to the same conclusion: political discrimination is a reality in social psychology.
If you're already telling yourself, "Discrimination can't be a serious problem if people are so quick to admit their own failings," you have entered Monty Python territory.

The piece ends with some wonderfully quixotic proposals to fix social psychology.  If anyone thinks any of these proposals will be seriously adopted, I'm ready to bet against you.

* Including personality psychology.

COMMENTS (28 to date)
Bob Knaus writes:

1) Surely experience would make 2% conservative grads turning into 10% conservative faculty as they age seem reasonable, no?

2) The survey should have used neutral framing, with "Favor liberal" and "Favor conservative" as endpoints. People tend to midpoints in surveys... this one almost certainly exaggerates the discrimination.

These points seem obvious to me as a layman, perhaps they address them in the paper.

Graham Peterson writes:

Bob Knaus, despite the neutral framing issue, I think this probably understates discrimination. Discrimination is usually socially unacceptable, so there are probably a lot of people who claim not to discriminate but really do. The fact that so many admitted to discriminating is surprising.

Tom West writes:

I don't find this terribly surprising.

The outcomes of social sciences often have policy implications, and its a no-brainer that ideological bent will influence at least topics of study, if not results (usually by discarding non-conforming results as "obviously wrong").

Hence allowing non-ideologically aligned members on one's faculty is active *harming* your ideology.

I would also expect that any area of study that has significant real-life policy implications and thus is primarily seen by powerful outsiders as a method of providing academic justification for preferred policies to have severe difficulties in staying unbiased.

And lastly, perhaps non-economic social sciences attract left wingers because study of those areas keeps producing results that conform with left-wing expectations in the same manner that studies of economics often produce results that conform with libertarian expectations and thus attract libertarian-leaning academics.

John writes:

I wonder to what extent the surprising responses to the bias questions is explained by greater awareness of unconscious bias among social scientists.

If someone asked me "if two job candidates (with equal qualifications) were to apply for an opening in your department, and one was black, do you think you would be inclined to vote for the white one?" I would be quite tempted to say "absolutely not! Racism is wrong."

However, I also know that essentially everybody shows some level of unconscious racial bias. So if I answered with complete honesty, I might be among the 82% who "admit prejudice"--not because I approve of it or because I would do it consciously, but because I'm aware of it.

kczat writes:

In reference to the "quixotic proposals" in the paper, would it be ethical to actually submit the "ideological mirror image" version of the paper and, only after it was reviewed, reveal the actual version?

I suppose you could randomly swap half of a subset of papers for their ideological mirror images and see whether there is a statistically significant difference in acceptance rate. That would be a paper by itself.

Mark V Anderson writes:
However, I also know that essentially everybody shows some level of unconscious racial bias. So if I answered with complete honesty, I might be among the 82% who "admit prejudice"--not because I approve of it or because I would do it consciously, but because I'm aware of it.

Yes but you only refer to unconscious bias that most people would deny. It could be that these researchers simply see nothing wrong with being biased against conservatives, so they don't feel any need to deny it. Unfortunately I am afraid that is the real answer. It appears to me an increasing trend for leftists to truly believe that a right-wing ideology means a lower intelligence. This is especially true for those in ideological bubbles, as most social scientists exist in.

MikeDC writes:

So... there's a downside to bubbles?

One might even say there's a negative externality at play...

Brian writes:

"perhaps non-economic social sciences attract left wingers because study of those areas keeps producing results that conform with left-wing expectations..."

Tom West,

So you are saying that social reality has a leftward bias? RIGHT.... Nice rationalization there.

TMC writes:

[Comment removed for supplying false email address. Email the to request restoring your comment privileges. A valid email address is required to post comments on EconLog and EconTalk.--Econlib Ed.]

James in London writes:

Would have been interesting to ask whether the group favoured bigger or smaller government, perhaps, rather than conservative or liberal. The former question has real practical consequences, the latter is a bit wooly and can mean different things over time.

Kevin writes:

Interesting article and comments.

James makes a good point about Big vs. Small Gov't, but I wonder if Social Conservatives are an even more despised minority among Social Psychologists than are Small Gov. (or Libertarian) Conservatives.

John writes:

Mark V Anderson, I agree that "bias against conservatives" could be (and probably is) an explanation for the result. I also think "social psychologists are more aware of ample research demonstrating that unconscious bias is going to make them 'less likely' to accept a conservative candidate, making them more likely than the general population to admit unconscious bias in a survey" could be an explanation for the result. I think the former is probably a larger effect than the latter, but I do wish a control question had been posted to help distinguish between the two. Rationality is looking for evidence to disprove our beliefs first.

Phil H writes:

Terrible framing here: "Ideological imbalance". What the researchers have found is that those who are educated in the social and political sciences (may) be more liberal than the general population; and that this shift becomes greater the more education one has.

So there is a difference. But to call it an "imbalance" is to massively beg the question. Why on earth should one think that the ideological distribution of the American population is "balanced"?

Someone who believed in education would be tempted to suggest that if learning more makes you more liberal, then there must be something in it...

The problem here is not really conservatism. The problem is that conservatism has been arrogated by a political party which prides itself on ignorance. In particular, the question about a job candidate who is "known" to be "politically conservative". That basically means affiliation with the Republicans, who are at present notorious for meddling with academia to promote their ideological ends. Good reason for thinking twice about that candidate.

KevinH writes:

What I'm about to say probably doesn't erase actual discrimination and bias on the part of social psychologists (I've known a few), however, as we all know correlation is not causation, and there are some reasons why conservatives make poor social psychologists.

Two very fundamental findings from social psych over the past 50 years have been 'moral relativism' and 'socially constructed power'. There are hundreds of studies confirming that these are very real phenomenon. One could could certainly argue about how much importance they play in situation x or y, but main-stream conservatives tend to underestimate these effects quite heavily, especially when there are protestant or baptist roots in an individual's moral philosophy.

Contemporary conservatism, like early 20th century liberalism, has a large focus on the individual. Certainly a lot goes on in the individual, but one could say that the entire social psychology movement has been an effort to offer evidence that much of what we take as individual action is actually an emergent property of groups of people. You can't really square that easily with strong forms of either moral realism or libertarian ideals.

Matt writes:

I suspect many of the liberal psychologists would attribute the absence of conservatives to self selection; less conservatives are interested in social psychology

See what happens if you use this explanation for other minorities (women/minorities in STEM).

Ray Lopez writes:

There is another economic benefit to diversity: greater output. This is fleshed out according to the provocative book "Moral Calculations : Game Theory, Logic and Human Frailty" by Lazlo Mero (1998) towards the end of this book.

According to Mero, a game theory researcher, if an economy is mixed, you actually get greater output. Simple example given was antitrust. Without antitrust laws, monopolists would price products to get a higher P, lower Q. With antitrust laws, they are forced to sell at lower prices, higher quantity, less deadweight loss. Other examples were given, backed by game theory logic.

triclops writes:

I agree with almost everything you said, but the revelations of social psychology lead me to question the wisdom of centralized power that progressives so love even more than before.
It seems though that many view the perfectibility of government as a given, and don't see a contradiction between moral relativism, socially constructed power, and an ever larger , more intrusive, and more centralized government.

Tom writes:

There's an assumption here that diversity is a balance of liberals and conservatives. I suspect social psychologists are broken into camps over other issues relevant to their field, and the conservatives wanting to re-open some debate their colleagues consider settled are seen as time-wasters.

A more extreme parallel would be ancient history of the Middle East, where the secular scholars have no truck at all with the religious believers.

This may just be what happens to conservatives by definition: they stick to old-fashioned views until they become a fringe minority, and the old liberals become the new conservatives. 18th-century political liberalism is today in America called conservative.

Enrique writes:

But can these social psychologists pass the ideological Turing Test?

Brian writes:

"Two very fundamental findings from social psych over the past 50 years have been 'moral relativism' and 'socially constructed power'."

Kevin H,

OK, for the sake of argument, let's say that these two ideas are the two most important ideas in social psychology, and that liberals are more readily able to understand them and give them they're rightful importance. How would that explain a 14:1 liberal/conservative ratio? Surely conservatives are capable of understanding and using these ideas, even if they don't like them, right? And even if they completely ignored those ideas, wouldn't there still be plenty of areas where they could do good work? Most fields are sufficiently compartmentalized that a researcher can be an expert in one sub-field without paying much attention to what's going on in other areas.

It seems to me the importance of those two ideas in social psychology likely argues the other way--the importance is greatly overstated on account of the ideological bias of practitioners in the field.

Brian writes:

"I suspect social psychologists are broken into camps over other issues relevant to their field, and the conservatives wanting to re-open some debate their colleagues consider settled are seen as time-wasters."


Ah, yes, ideologues often prematurely close debate on issues that are far from settled. It's only natural for them to shut down debate by those who think differently. That's exactly the problem with a lack of ideological diversity. A good example in the natural sciences is climate change, where lack of ideological diversity has led to serious errors and a premature closing of debate on likely outcomes. Fortunately, the natural sciences have stronger tools for self-correction. Th poor social sciences, sadly, have little defense against ideological corruption.

Bostonian writes:

Believing that there are racial differences in IQ is now a far right-wing position, and tolerance for such ideological diversity is thought to conflict with the goal of achieving demographic diversity, since the confidence of the under-represented could be undermined.

Daublin writes:

Academia has some factors that make it converge on the same ideas over a long period of time. So any liberal bias will tend to become more extreme over time.

First, to get anywhere in academia, you have to network. You have to ingratiate yourself to the big players, and the current big players had to do the same thing back when they were getting started. We're talking about a sphere of society where the primary measure of success is whether the other people in the sphere subjectively like you. A businessman can buck the conventional wisdom and prove they were right by getting rich; there's no equivalent in academia. A graduate student who grouses about the existing field will just be shown the door, and probably wouldn't care to stick around anyway.

Second, the published literature is talmudic. Each new paper must be positioned as an extension of the mass of papers that already exist. A ten page paper will often have 20-30 references; those citations are part of the price of getting published, and you just have to cite them favorably. Their authors are going to be your paper reviewers.

Tom west writes:

Surely conservatives are capable of understanding and using these ideas, even if they don't like them, right?

That's not been my experience with either the right or the left. Pretty much both sides understand that there are quite a number of "facts" which, if promulgated as science, are extremely damaging to society, and thus must, indeed, be incorrect.

They may differ as to what those truths might be and what constitutes damage, but I think there's a general understanding that science is way too important a political tool to be left to science.

(Okay, that's perhaps a little too cynical. How about "there are almost no results in the social sciences that are so unambiguous that it would prevent someone from discarding "self-evidently wrong" data that challenged their priors. After all, if evidence from your experiments said gravity pulls upward, you'd require hundreds of experiments with absolute, incontrovertible data before you'd agree. It's certainly reasonable to demand the same level of proof from results that contradict self-evident truths of human behaviour.)

KevinH writes:


Oh certainly, there's nothing in social psych that necessarily precludes conservative policy ideas, but rather just their philosophical underpinnings. One can certainly be right for the wrong reasons.

For example, take individualism. The Cato institute's bit on the philosophy of libertarianism: "Libertarians see the individual as the basic unit of social analysis. Only individuals make choices and are responsible for their actions." ( )

A social psychologist might snicker at that strong assertion, but there are many effective ideas for policy that don't still rely fairly heavily on individual responsibility.

KevinH writes:


I think you are underestimating the need for philosophical alignment to devote your life to a field. I'm not sure, but I'd imagine the ratio of biochemists who don't believe in evolution exceeds 1:14.

It isn't so much about the relative strength of the theories (evolution much more solid and well defined than social constructed power of course), but rather that the 1:14 ratio is the consequence of four things.

1) dynamic beliefs. Those who come in identifying as conservative or with philosophical ideas like individualism change their views in response to evidence (or peer pressure).

2) self selection, as you point out

3) failure to contribute. Challenging dogma isn't hard just because of intellectual inertia. There's a lot of data that got us to that dogma in the first place.

4) actual discrimination above and beyond #3

Now, my original post wasn't meant to rule out #4 completely, just to point out that 1 and 3 will be fairly substantial effects as well. I bet one could come up with a combination of surveys, fictitious resumes, and citation patterns to get a decent measurement of the factors, but 3 and 4 are going to tough to differentiate.

Brian writes:

Kevin H.,

I am not underestimating the need for "philosophical alignment" on the part of researchers. What I am denying, rather, is that particular fields, by the nature of their subject matter, require a specific philosophical alignment.

Reality, as far as I can tell, is ideologically neutral. Gravity tells us nothing about whether it's better to be liberal or conservative. Neither does evolution, surprisingly enough (don't confuse specific religious beliefs with conservatism).

Of course, it may happen that certain ways of understanding reality are more easily accepted by one viewpoint than another, but there's no reason to expect that all the ways, taken together, should favor one side over another.

When there's evidence of one side being strongly favored over another, it is clear that ideological bias on the part of the practitioners is at work.

Even your proposed causes (1 - 4) are suggestive of bias at every step. Take point 1. If a practitioner changes his or her mind because of peer pressure, as opposed to the evidence, wouldn't that be a form of indirect bias at work? Practitioners are trying to bring someone to their side, as opposed to valuing the objective and independent pursuit of truth. In point 2, why would someone self select out? Because they felt unwelcome or too different? This would also be a case of indirect bias. Point 3, why would someone interested in a field "fail to contribute?" Because journal referees don't like their point of view?

Now, I don't want to claim that the entire 14:1 discrepancy is due to bias. There are good reasons to think that liberals value the scholarly academic environment more than conservatives do, but that at most could only explain the liberal/conservative divide seen throughout academia. More extreme polarization seen in certain fields, such as social psychology, is likely only explained by the presence of bias and discrimination, as noted by the paper.

Comments for this entry have been closed
Return to top