Statistical Generalizations, Stereotypes, and Rational Behavior

Stereotypes are none other than a statistical generalization; and a statistical generalization may be warranted or it may be unwarranted.  The following is a statistical generalization that is very much warranted: Most American-born and raised individuals are not bilingual.  Whenever an American-born person shows up in a non-English speaking country, it is assumed that the person does not speak the language of that country.  The stereotype, then, of American-born individuals is that they only speak English.  And guess what: It is true.  The stereotype is true although there are in fact American-born individuals who are bilingual.

So it is just plain nonsense to insist that a stereotype is false simply in virtue of being a stereotype.

Of course, a statistical generalization can be downright false.  Here is one: It is false that particularly frumpy and/or very plain looking people cannot sing.  Needless to say, Susan Boyle and Paul Potts give the lie to that claim.  There is in fact no correlation between singing talent and physical appearances.  What is true, to be sure, is that the music industry does much to beautify its stars—so much so, in fact, that it seems plausible to suppose that some people would not enjoy the success they enjoy in the music industry were it not for their physical attractiveness.

So the automatic assumption that homely people cannot sing is a negative stereotype.  Moreover, unlike the stereotype about American-born individuals not being bilingual, the stereotype that homely people are particularly lacking in singing talent is entirely without warrant.

A negative stereotype can be very much warranted.  A negative stereotype can be entirely unwarranted.

One thing is for certain, though, is that statistical generalizations are here to say.  This is because making such generalizations is a defining feature of being rational.  Suppose for example, that the University of Western Canada has a student body population of 1200 students; and we determine that 60 out of 70 U of WC students quite randomly chosen sing classical music, it becomes pretty reasonable to suppose that singing classical music.  It would certainly be reasonable to suspect that the next U of WC student whom one encounters will also sing classical music.  Indeed, it would be nonsense to insist that it is as likely that the next student whom one encounters does not sing classical music as it is that she or he does sing classical music.

There can be no objection to making warranted statistical generalizations.  The problem is twofold: One problem is that of reacting foolishly when the generalization does not apply.  The other is failing to see the limits of a generalization.

Here is a simple example of the first.  It is overwhelmingly the case that men who braid their hair take themselves to be black or to have black ancestry.  So it is regardless of how light-skin the person is.  Yet, there is nothing that prevents a white male from adopting precisely that hairstyle.  Perhaps he is dating a black woman and wants to impress her.  If, given his head full of braids, we assume that the male is black and he informs us that in fact he is a white male who decided to try out the style, then my view is that we should simply accept the white male at his word and leave the matter at that.  There is no need for any form of hyperbolic reaction.

We could tell an analogous kind of story with a woman who always dresses in male attire.  From my experience, this is a very, very good indication that the woman is a lesbian.  Certainly, the woman could not be offended if anyone thought that.  She need not be a lesbian, however; and if she walks in one day with a man on her arm who she announces is her husband and the father of their 3-children, then precisely what I should do is accept this reality and move on with.  A hyperbolic reaction is out of order.

It is a simple truth that a warranted generalization with regard to a given population is not thereby a universal law.  And it is a sign of maturity on our part to be responsive to this reality.

Let me now turn to the point of seeing the limitations of statistical generalization; and I shall do so in somewhat controversial way.

Let us suppose for the sake of argument that it is reasonable for 60-something white women to fear having their purse snatched by black male youth who run away with it.  Well, even if we grant this fear is reasonable, what does not thereby follow is that it is also reasonable for 60-something white women to fear that a 50-something and older black males will snatch their purse and run away with it.  Why?  Because running is not something that men who are 50-something and older do very well, be they black or white or whatever.

In other words, even a warranted fear about black male youth does not thereby extend to all blacks, especially when the fear is based upon the other having a capacity that obviously diminishes sharply with age.

We can offer an analogous case from the other direction.  I do not know many 70-something white women who love rap music.  I do not know any, in fact.  Surely, it is a warranted stereotype that 70-something white women do not like rap music.  In fact, it is a warranted stereotype that 70-something women of any ethnic configuration do not like rap music.  Alas, the assumption of white plays into the stereotype ever so well.  So, if I offered to make a music CD for such an individual and she remarked “Be sure to include a few rap song or two on the CD”, the appropriate response is: “With pleasure”—and not “Oh my God, I can’t believe you like rap music!”

The example of a making a music CD for a 70-something person, be the individual male or female, is a very nice way to end this blog-entry.  In the absence of any clear indication to the contrary, surely the warranted generalization is that rap music is not what I should put on the CD.  This assumption is not at all defeated by the rare exception.

It is a quite warranted stereotype of the 70-something person that her or his taste in music does not include rap music.  Interestingly enough, though, when the 20-something of the present become the 70-something of the future, a very different stereotype may be in place for the 70-something.  It is more than a little amusing to imagine 70-something folks in nursing homes with their head bobbing to rap music.  But that reality may very well be what the present serves up.

About Laurence Thomas

Laurence Thomas is Professor in the Department of Political Science and the Department of Philosophy at Syracuse University. His most recent book is The Family and the Political Self and his most recent article in French is "Juifs et Noirs: Au-delà du Mal" in Trigano (ed.) Juifs et Noirs: du Mythe à la Réalité. Thomas has published numerous essays on the topic of friendship. The essay "The Character of Friendship" has appeared in volume on friendship, entitled Thinking About Friendship, edited by Damian Caluori and the essay "Friendship in the Shadow of Technology" has appeared in the anthology Moral and Moral Controversies edited by Steven Scalet and John Arthur. His most recent essay--entitled "Being Moral and Handling the Truth"--is about circumstances under which it is morally permissible to lie. Indeed, an example is given in Section IV of a lie being morally virtuous.
This entry was posted in Articles. Bookmark the permalink.

One Response to Statistical Generalizations, Stereotypes, and Rational Behavior

  1. fleeb says:

    What you are discussing in a sense is reasoning using probability ala’ Bayes theorem. You start with a belief based upon the most basic generalization based upon the data you have about someone. Then you update it when more information comes in. Let’s say with the purse snatching, we have reliable data that 2% of black people in a town snatch women’s purse while only 1$ of white people do so. So, the conditional probability of being a purse snatcher given you are white is 0.01 in this town. There are other conditions that further influence the probability like age, gender, style of dress, area you live, etc. Let’s assume all this for the sake of argument. Now, a woman with a purse (who is only concerned with keeping her purse at this moment) can choose to walk 1 of 2 paths. On the first is a white person, the second a black person. Based on the information she has, is she rational or racist to choose the first path. This is something that all rational, non-PC, non-racist people think about but seldom discuss.

Leave a Reply

Your email address will not be published. Required fields are marked *

You may use these HTML tags and attributes: <a href="" title=""> <abbr title=""> <acronym title=""> <b> <blockquote cite=""> <cite> <code> <del datetime=""> <em> <i> <q cite=""> <s> <strike> <strong>