Bryan Caplan  

Giles and Stereotype Accuracy

Capital vs. Institutions... Redistribution vs. Paternalism...

The weakest part of Martin Giles' Why Americans Hate Welfare is his dismissive treatment of stereotypes. He cites a number of psychological experiments on the emergence of baseless stereotypes. But he at best downplays the growing literature on stereotype accuracy. For example, he does not even cite the excellent volume Stereotype Accuracy : Toward Appreciating Group Differences, edited by Yueh-Ting Lee, Lee J. Jussim, and Clark R. McCauley. I especially love the chapter by David Funder.

Curious? Check out John Ray's fun, scholarly, and web-accessible defense of stereotypes, "Do We Stereotype Stereotyping?"

One of my favorite examples of stereotype accuracy comes from the field of personality psychology. Popular stereotypes tell us that men are more logical and women are more emotional. Lie? Bigotry? Satanism? Personality tests confirm an even stronger pattern than I expected. Myers-Briggs type tests have a Thinking/Feeling measure. The breakdown for males is about 60/40. The breakdown for females is about 30/70. You may be able to create imaginary stereotypes in a psych lab, but here is one prominent stereotype backed up by a 30 percentage-point chasm.

Does this stereotype fit you? The last time I checked, you can't take an official Myers-Briggs test over the web, but the Keirsey Temperament Sorter is a close substitute - and it's free.

Comments and Sharing

COMMENTS (7 to date)
Fazal Majid writes:

An alternative explanation is that both stereotypes and the Myers-Briggs test are equally bogus. If the Myers-Briggs test and Jung's theory of types behind it is just a fancy repackaging of stereotypes with a veneer of pseudo-scientific respectability, their correlation becomes a circular argument. Given how much of psychology or psychanalysis is scientifically dubious (e.g. Freud's infalsifiable theories), one should take all this with a big grain of salt.

There probably isn't any way to assess something as subjective as "emotional", but logic abilities should correlate well to mathematics aptitude tests, and while there is a small gender difference (specially in fields related to geometry), it is nowhere near as stark what is predicted by Myers-Briggs.

dsquared writes:

Fazal has it exactly right. I am flat out amazed that anyone would be treating the Myers-Briggs test as being unproblematic and objective science. I am sure that I could design a test that would prove all manner of stereotypes about economists if I was allowed to design the questions and define what my categories meant.

dsquared writes:

And furthermore, that "30 percentage point chasm" is obviously meaningless since there is no base unit for the test scores; a different normalisation could make it into a 0.3 point gap or a 99 point gap.

Lawrance George Lux writes:

Stereotypes have one fundamental flaw: Culture evoles while they don't. Stereotypes are already obsolete by the time they are postulated. lgl

Chuck writes:

Mr. Caplan, I've read your stuff for a while and always admired your clarity and persuasiveness. Not this time. How can you use a piece of obviously specious pseudoscience like Myers-Briggs as if it were a valid data?

Malcolm Gladwell has an accessible, compelling, and thorough demolition of personality tests here.

But let me make the point more quickly here. As someone who is familiar with the concept of "revealed preference" you know that talk is cheap and, when put to the test, people often behave differently from their stated values.

But MBTI and other personality tests are surveys rather than experiments. In that way, they measure what people imagine are their own virtues, rather than anything about their behavior.

Also MBTI is full of idiotic false alternatives. Questions are posed that ask you to choose between non-mutually exclusive alternatives. Here are a few from the test that you linked to:


Do you consider yourself:

a good conversationalist
a good listener

34. In a heated discussion do you:

look for common ground
stick to your guns

12. Common sense is:

frequently questionable
usually reliable


The answer is often "both." Or "it depends" Or "that question doesn't make any fucking sense"


Ironically, I believe that stereotypes are true more often than we care to think. But the data that you marshall does nothing to strengthen the case.

anon writes:

I'm with you - my stereotype of an economist is a white male who is quasi-autistic, with little understanding of the nuances of ethics and life who likes to apply a perversely simplistic quantitative model to the world. My stereotype of an economist is shallow, egotistical and DRASTICALLY overconfident - confusing smug cleverness with wisdom.

Randy writes:


Hey, don't hold back. What are you trying to say?


Comments for this entry have been closed
Return to top