What's Wrong with Stereotypes?

1. Opposition to Stereotyping

I keep hearing that “stereotyping” is bad, and that it’s good to undermine stereotypes. For instance, if you have a TV show with a brilliant surgeon, you should make them a woman. Or black. Or, best of all, a black woman. Because that will defy stereotypes and thereby make the world a better place.

If you make a picture of some business people in a meeting, you have to make sure that it does not reflect what most business meetings actually look like; rather, you should gender balance it and make sure to have three different races represented (see above photo).

If someone tells a joke that relies on stereotypes about a group, that is “offensive” and hence evil. I guess because it reinforces the stereotypes? Or maybe it’s just evil to rely on a stereotype for anything.

Back when James Damore was fired from Google, it was partly because he cited research to the effect that women tend to be higher than men in the traits of “agreeableness” and “neuroticism” from the 5-factor model of personality. In doing so, he was reinforcing stereotypes, which all decent people know to be evil. If a statement sounds like a stereotype, that alone is enough to categorically reject it.

Most of the people who believe this have a predictable political orientation, and so you can usually count on a certain amount of hypocrisy. Thus, certain stereotypes are fine. You can stereotype white men as privileged oppressors, you can stereotype Republicans as uneducated, etc. It’s all a matter of stereotyping the right group in the right way. As long as your stereotype reinforces your political side, it’s cool.

But I digress. My question: what exactly is supposed to be wrong with stereotypes? Why not use and reinforce them?

2. Problems with Stereotypes

A. What are stereotypes?

First, what is a stereotype? Usually, people are talking about stereotypes about groups of people (e.g., women, black people, doctors). (I guess you could also have “stereotypes” about any class of object, but we don’t care about non-human objects.) As far as I can tell, a “stereotype” is just a widely shared belief about what a certain class of people tend to be like.

Aside: Maybe there are a few other conditions, such as: it can’t be something definitional, it has to differentiate the group from other groups, and it should be a statistical generalization. Thus, it isn’t a “stereotype” that bachelors are unmarried, or that black people tend to have two legs. But let’s not worry about all the details of the definition.

What is wrong with using or reinforcing such beliefs?

B. Are they false?

Maybe the problem with stereotypes is that they tend to be false, or to lead people to make false judgments.

On the face of it, this would be surprising. In general, people tend to form beliefs about observable reality based on observations, which generally tend to reflect that reality. If most people think that the winter is colder than the summer, that’s probably because the winter is colder than the summer. If people tend to think that humans generally have two arms, that’s probably because humans generally have two arms. Etc. Could it be that, when it comes to groups of people, we have a general tendency to go wrong about everything—that when we think a group has feature F, in general, the group usually doesn’t have F? This is possible, but it would be pretty surprising.

Btw, notice that I’m talking about beliefs about observable (or otherwise easily accessible) properties of observable objects. It’s not so surprising that people get things wrong about unobservable things, like God or the origin of the universe.

Or maybe it’s not quite that the group doesn’t have F at all; maybe it’s just that stereotypes tend to exaggerate real differences, so the group that is stereotyped as having F will have less F than people tend to think.

These are common critiques of stereotypes among educated people. One might say there is a stereotype that stereotypes are inaccurate. These critiques, however, have the character of articles of faith—almost none of these educated people who are too smart to fall for crude stereotypes has ever actually checked on whether stereotypes tend to be accurate or not.

It happens that we have evidence about this. Over 50 studies have been performed on the accuracy of demographic, national, political, and other stereotypes. The results are unequivocal: Stereotypes are not generally inaccurate, nor are they generally exaggerations. The truth is the exact opposite: stereotypes are generally accurate, except that they tend to understate real differences. There is basically no evidence that stereotypes tend to lead to inaccurate judgments. Psychologist Lee Jussim describes stereotype accuracy as “one of the largest and most replicable effects in all of social psychology”. See his paper for a general review of the evidence. (See also his blog post.)

For example: A 2011 paper titled “Beliefs About Cognitive Gender Differences: Accurate for Direction, Underestimated for Size” reported results of a study in which they asked ordinary people to guess how men and women would perform on a series of cognitive tasks. The scientists then compared the ordinary people’s expectations to the reality. This is a straightforward test. It turned out that people’s guesses were correct about the direction of gender differences (which sex would perform better on which tasks) but that the actual gender differences were larger than people thought.

Notice that this is the opposite of what educated, progressive, right-thinking people would presumably predict.

But again, this really should not be at all surprising if you’re thinking non-ideologically. Human beings can observe each other. Most of us have had many interactions with men and women. It would be bizarre if, despite that, we kept having beliefs about gender differences that had no correlation with reality.

C. Are they oppressive?

Maybe the problem is that stereotypes—whether accurate or not—are oppressive. Maybe they stop individuals from attempting or succeeding in things that would defy the stereotypes, when they would otherwise have succeeded.

This could be true, but it isn’t self-evident. One way this could work is that other people would judge you based on stereotypes and would refuse to take into account your own individual characteristics. This is another one of those things that educated, progressive people assume without checking the evidence. In fact, studies find individuating information (specific to individuals) has massively greater effects on people’s judgments of others than stereotypes do.

Given a society of millions of people, I’m sure there are some people who are deterred from attempting something due to stereotypes, and who are thereby worse off. So that is a cost. On the other hand, there can also be costs to not having stereotypes—e.g., perhaps some people would be encouraged to do things that they would not be suited to, and they would thereby be made worse off.

As a general rule, truth is good. Knowing the truth about things that are relevant to your interests can make you worse off sometimes, but in general, it makes you better off. So, given the accuracy of stereotypes, stereotypes are probably beneficial. But whether they are or not, trying to suppress them is a fool’s errand. Normal people won’t stop noticing group differences just because elites try to hide them; normal people will just conclude that the elites are dishonest propagandists.

D. Stereotype threat

Stereotype threat is a theory in psychology that says that the existence of stereotypes tends to cause people to act in ways conforming to those stereotypes. Particularly negative stereotypes. E.g., it is said that if you remind people of their race before giving them a scholastic aptitude test, then the black students will tend to perform worse, due to stereotypes about their race. If you remind people of their sex, then the women will supposedly perform worse on the math questions than they otherwise would.

Some people claim that this effect actually explains the entire gap in test scores between blacks and whites. Of course that’s false (it derives from a simple misunderstanding of a graph in the original study—progressives then just kept passing on this misunderstanding to each other). The effect only (at most) slightly increases already-existing gaps.

In Progressive Myths, I discuss several problems with the stereotype threat literature. Here is just one interesting point: There have been many studies of the stereotype threat effect. Most of them are done in the lab. Some of them use more realistic conditions than others, and four of them have been done using actual administrations of standardized tests. The more realistic the test is, the smaller the effect is, and the effect vanishes for the actual standardized tests. So whatever is going on in the lab, stereotype threat appears to have no relevance for the important cases.

E. Do we overgeneralize?

The anti-stereotype people often seem to think that to hold a stereotype is to believe that every single member of a given group conforms to the stereotype. Of course that’s ridiculous. For example, there is a stereotype that corporate CEO’s tend to be men. Many people are aware of this fact, but no one thinks that there are no female CEO’s, and there is no reason to ascribe such a stupid belief to people. Left-wing elites tend to think people are too stupid to handle the idea that “most but not all” A’s are B. But it is the left-wing elites who can’t handle that idea.

3. The Inescapability of Stereotypes

A. Stereotyping as learning

Why do people harbor stereotypes about groups?

Here’s a hypothesis: it’s just normal learning. You go out in the world, you see things, you notice correlations, and you learn from them. What you saw in the past affects your expectations for future cases. This is crucial for any normal life. “Stereotypes” are just a product of that normal process, applied to types of people that you observe.

Example: You see a chair-shaped object. You immediately think that it’s a chair, and since it is a chair, it will be suitable for sitting on. That is relying on “stereotypes” about chairs. It’s just that no one gets upset about this, because chairs aren’t people.

Here’s an example involving people: You enter a coffee shop. You see someone standing behind the counter wearing an apron. Even though you’ve never been to this specific shop before or seen that specific person before, you have certain expectations. You think you can go to that person and ask for some coffee drink, give them money, and they will get you the drink. That’s all based on “stereotypes” that you have about baristas. It isn’t guaranteed to be true; there may be some baristas who will not give you the coffee drink you ask for, and there may be some people who stand behind coffee shop counters wearing aprons even though they are not baristas. Nevertheless, if you’re going to make it through the world, you have to make assumptions like these, while remaining ready, of course, to give up your assumptions if the evidence turns against them.

In that case, nobody cares that you’re stereotyping baristas, because there is no political history involving people oppressing baristas. But people are oversensitive about certain other groups because there has historically been political conflict and struggle surrounding the oppression and liberation of those groups. So people start to make up rationalizations for why stereotypes about these specific, historically oppressed groups are immoral, inaccurate, offensive, etc., when really all that is happening is that some people are getting emotionally upset because they were reminded of something in the past.

B. Statistical evidence

Sometimes, people worry that it’s wrong to form beliefs about individuals based merely on statistical generalizations about their group. But this might rule out all or nearly all predictions. Example: Progressives think that it’s wrong to think that a person is more likely to commit a crime just because they are black and blacks are statistically more likely to commit crimes.

But these same progressives think (I assume) that it’s okay to think a person is more likely to commit a crime because that person has committed crimes in the past. But this is also based on a statistical generalization and a “stereotype”—in this case, the group being stereotyped is people who have committed crimes in the past, and the generalization is that those people are more likely to commit crimes in the future than the general population.

Or suppose someone actually tells you that they plan on robbing a bank, so you believe that they are more likely to rob a bank than a random person is. This, too, can be viewed as a kind of stereotyping: here, the stereotyped group is people who say they are going to commit crimes, and you’re assuming they are more likely to commit crimes than the general population.

Conclusion

Opposition to stereotyping in general is incoherent, unless you want to oppose learning. The idea that it is bad in general to form generalizations about groups is on its face crazy. Nor is there anything objectively offensive about recognizing group differences. Being “offended” by the recognition of patterns in the world is not really something that a serious, adult thinker does.