Stereotypes and Their Uses

Steve Sailer directs people to some posts by Bryan Caplan on the Econolog blog.

Most interesting are these two posts that deal with stereotypes:

Crashing into Stereotypes and
How I Fight Statistical Discrimination.

Looking around on the web, I also found this article by Chris Naud of the University of Chicago.

The general argument here is that stereotypes are usually based on statistical truth, that is, groups that are stereotyped to have certain characteristics probably do have those characteristics in greater proportions than the general population. Also, people use stereotypes for the simple fact that trying to "treat people as individuals," i.e. deal with each person on a case-by-case basis, is not always cost-effective, and may be cost-prohibitive. Or put as Mr. Caplan put it, people tend to assume that a person is an average member of whatever group he or she can be seen to belong to until evidence suggests otherwise (obviously, this applies not just to race but to sex, sexual orientation, lodge membership, etc.).

Of course, one may retort that judging based on stereotypes is a sign of ignorance. Well, technically that is true; one pre-judges situations based on partial knowledge. But on the other hand, one might argue that prejudices are how we cope with ignorance; when we don't have all the information, we make the guess that is the most likely to get us the best outcome. We can never totally overcome ignorance unless we become omniscient; moreover, as stated before, information is not cost-free, so getting more information before making a decision is not always practical.

None of this is to say, of course, that government policy should be based on stereotypes per se. Obviously, at the levels of corporations and governments, there are a lot more resources to use to get information, and the consequences of making wrong decisions are a lot greater than at the individual level. Of course, on some level decisions will still be made based on a statistical truth (for example, an insurance actuarial table can't predict who will and won't have accidents, just the likelihood based on a number of factors), but they are based on a huge number of factors including ones that are not immediately obvious.

But on a personal level, decisions often have to be made with very little information, and it isn't always possible to find out someone's age, marital status, their criminal record (or lack thereof), their family history, etc.

In any case, two thoughts occur to me, looking at this.

First, I think that the best way to look at racism is to consider what Niger Innis of the Congress of Racial Equality (CORE) said during an "audience discussion" after an episode of Any Day Now that dealt particularly heavily with the issue of prejudices.

I can't find the exact quote, but what he said was in essence that the problem isn't people assuming things as much as it is when people refuse to alter their assumptions in the face of contravening evidence.

This, I think, is a good rule of thumb for when prejudice becomes destructive; when it is based not just on ignorance, but on willful ignorance and not subjected to evidence.

Second, any stereotype that is based on statistical truth will get more accurate the larger the population sample you have to deal with, assuming that the sample is representative. For example, if a stereotype about Hispanics is based on statistical truths, it might not apply when you are thinking of hiring a Hispanic for a job, but would likely be very important if you are thinking of moving into a predominantly Hispanic neighborhood (unless, of course, the neighborhood is an unrepresentative one; I would assume that if, for example, the neighborhood consists mainly of people making over 100,000 a year, stereotypes about the general population would be much less accurate as predictors).

Statistics is not as dry a subject as most people think, now, is it?

btw, I also have to see Crash.

That is all.