Bryan Caplan  

Endogenous Sexism

Good News on Australia's Econo... Carlos Ball, RIP...
Suppose men and women are equally praiseworthy in every way.  Both genders are equally honest, fair, peaceful, hard-working, fun-loving, and so on.  With one key exception, both genders share the same trait preferences: The average man places as much weight on honesty, fairness, and so on as the average woman.  The key exception: everyone's heterosexual. 

Suppose further that no one's evaluations are gender-biased.  Men have no tendency to see women as worse than they really are, and women have no tendency to see men as worse than than really are.  People do however tend to get to know more people of the same sex because, say, men hunt together and women gather together.

Claim: Even in this rarefied setting, sexism will endogenously arise.  Men will find the men they personally know to be better than the women they personally know.  Women will find the women they personally know to be better than the men they personally know.  And if people extrapolate from the people they personally know to all humanity, men will think that men are better than women, and women will think that women are better than men.

Hint: Compare the selection filter you apply to potential friends to the selection filter to apply to the romantic partners of your friends.

Please show your work.

COMMENTS (16 to date)
Tom West writes:

Personally, I think much sexism (and various other -isms) is simply the inability of the human brain to handle easily correlations of less than 1.

As soon as we see even a moderately significant correlation in some group of humans, it becomes instinctive to reject anything that doesn't support that correlation. Even when we logically accept this exception, some part of our brain "knows" that it's wrong.

And then, of course, confirmation bias provides all the "proof" that our instinct are correct - the correlation *is* 1.

Emily writes:

Because of that "People do however tend to get to know more people of the same sex because, say, men hunt together and women gather together," and the hint, I assume the idea is that more of the same-sex people you know are people you've chosen (and therefore you like), and more of the opposite-sex people are people you haven't chosen (your same-sex contacts' romantic partners.)

But I don't think the conclusion is a given, even from the assumptions. Like, let's think of an extreme example where you know all of the same-sex people. Your impression of that distribution, therefore, is totally accurate. But the only opposite-sex person you know is your spouse. So long as your spouse matches your preferences in a more-than-average way, the average opposite-sex person you know is better than the average same-sex person.

Brandon Berg writes:

As soon as we see even a moderately significant correlation in some group of humans, it becomes instinctive to reject anything that doesn't support that correlation.

Of course, the mirror image of this is those who reject the correlation because they think that such a correlation would imply a justification for sexism or racism.

Thus, for example, both racism and "anti-racist" IQ denialism spring from the same fallacy: Failure to grasp the significance of correlations bring real-valued rather than integral. The only difference is which side of the perceived contradiction they choose to reject.

Jesse writes:

For your claim to be true you need to assume two conditions:
A- on average people's friends groups have higher than mean 'value'.
B - there is a low correlation between the 'value' of a person and the their romantic partner is low.
Both friendships and romances require that each person passes the others selection filter, hence both conditions are false.
Consider the observations of someone with below average 'value' if either of these are false...

I need to know your definition of "sexism".

If I assert that men are taller on average than women, is that an example of sexism?

How about if I assert that men score higher on average on tests of three-dimensional spatial reasoning, is that an example of sexism?

MG writes:

Tom West describes the mechanism that afflicts the "bigot" -- to force a correlation of, say 0.7, to rationalize a variance that only a correlation of 1.0 or -1.0 could explain.

But consider what the "anti-bigot at all costs", the "PC", does -- to force us to deny/conceal/hide the ability of any real correlation to explain a variance -- coor^2 can be significantly different from zero.

In other words, both the bigot and the PC subvert judgement. I suspect that either minimizes this cognitive bias the more it affects his/her pocketsbook.

Dee W writes:

It sounds like you're saying if people segregate themselves by gender, then sexism will arise. Even if the segregation isn't complete--as long as people of the same gender tend to group with others of the same gender.

It's true for other ways of categorizing people as well. If people segregate themselves by race then racism will arise. If people segregate themselves by class then classism will arise.

I don't understand the point of the word "endogenous" though. I think all sexism is endogenous since it comes from the society in which it is found. What would be the case of exogenous sexism?

Tom West writes:

Of course, the mirror image of this is those who reject the correlation because they think that such a correlation would imply a justification for sexism or racism.

One difference. The effect on the rest of the population. By (incorrectly) insisting a natural correlation of 0, they throw some sand into the cognitive gears of the masses that would otherwise be trending towards a correlation of 1. In other words, they may be incorrect, but they make the world as a whole more correct.

(Same idea as my respect for Libertarians - I don't agree with them on many issues, but I believe their push for maximal freedom counteracts a natural tendency towards discounting freedom's value).

It's also why I wasn't sorry to see Summers go after his remarks. They may have been perfectly accurate, but the public took the "correlation is 1" message (including memorably for me, a high school science teacher). Summer's correct remark made the word less correct. And as a public bigwig, he's responsible for the outcome of his remarks, not just their veracity.

JohnBinNH writes:

"..he's responsible for the outcome of his remarks..."

Well, I might grant that as a "bigwig" he should (pragmatically but not morally) consider the outcome of his remarks, knowing as he does that they will be widely circulated.

But I see several counters:

First, no one is so clever that they can predict all the outcomes of a remark or all the emotional responses. So some sort of "reasonableness" test has to be applied here. I don't think a reasonable person would be offended by his remarks (I have read the transcript).

Second, telling the truth is always OK. It's not a teller's responsibility when the truth is uncomfortable, and we want to encourage telling the truth.

Third, feigned outrage for personal advantage is the responsibility of the feign-er, not the speaker. I don't for a minute believe that the people who claimed to be suffering severe emotional distress were telling the truth.

Fourth, if he's responsible, then he has to have matching authority, as responsibility without authority is a terrible way to run any part of society. What form would that authority take?

triclops writes:

Tom West,
I think your diagnosis is right on, but your prescription seems to be making a complex situation even more complex.
To keep with the Summers example,
You view Summer's statements and the reaction to them as poles that pull the opinions of the general pop in opposing directions, leading to a kind of balance. You acknowledge that the pull is based more on perception than reality.
I agree with all of that, but then you say that the punishment of Summers is good because it improved the knowledge of the collective. Why not try to improve it with truth?
Putting aside the issue of the justice of attacking someone for being honest and reasonable, do you think so little of the general population that honesty would fail?
Wouldn't trying to get them in just the right spot by pushing and pulling them with partisans from either side be a logistical and methodological nightmare?

Greg Heslop writes:

I am not sure about the specifics in this blog post, but I would think that, since people commonly think themselves above average and also in possession of somewhat greater capabilities than those possessed by their spouses, any suspicion of assortative mating (same ranks marry) implies that the average of one's own sex comes out as "better" than the average of the other sex. Somewhat biased evaluations here, but they do not strictly pertain to gender, only to one's self.

Of course, this holds only under the view that the bell curves for both sexes are of the same shape.

Tom West writes:

Second, telling the truth is always OK.

Sorry, no. Spotting a flaming torch in a crowded theater and yelling "Fire" doesn't absolve you of the dozens that die in the panic.

Telling the mother on her death-bed that all her children are dead as well may be truth. It is not OK.

These are extreme examples, but the fact that I hope you agree with me in these cases makes it clear that we balance the good of truth with the good of outcomes. Now we two may have a different point on the balance, but it is simply a balance.

Third, feigned outrage for personal advantage is the responsibility of the feign-er, not the speaker.

Agreed. But it's not for his effects on the outraged that I felt he had to go, but the effects of his remarks on the population who "agreed" with what he didn't say..

And I'd argue that the President of Harvard is *primarily* a PR position. He's the public face of a highly visible institution. And yes, I'd expect the head of PR to know that there are simply topics that you don't get to approach until you're no longer in the spotlight.

The President of the US doesn't get to say even more things, true or not.


Sorry, but everything I've seen in life indicates that policy is a tug of war. Much of the work that enabled the policies I support was done by people far more extreme (and who were bitterly disappointed by the outcome). Middle of the road people like me don't get off our duff to do the enormously hard work to shift society.

I think of society like an elephant. It lumbers on, and it's probably not quite on the path I would like, but it's "close enough" to make it not worth the huge effort to change it. The extremists on either side are the ones who are enraged enough that they spend their life pushing the elephant, and yes, it then moves a few inches closer to the path I'd like.

I disagree with their policies and often their methods, but they'd done more to make the path closer to what I'd like than me and my "more rational" peers.

JohnBinNH writes:

Tom, if there's a actual torch with actual fire, then yelling "Fire" is not only the truth, it's the right thing to do. Or did you mean it was on stage and clearly part of the play rather than a present danger? In that case, that kind of fire is not the kind of fire for which a yell of "Fire" is true.

I disagree with the mother issue as well, if the scenario is she asks directly about her children. If she doesn't ask, I don't think there's a duty to tell her. If she doesn't ask and you do tell her, I agree that she might be hurt (depending on her beliefs about an afterlife). But I'm not willing to say telling her is a bad act, so I don't agree with you yet that there's a need to consider outcomes of the hurt feelings sort. I will agree that not telling the whole truth to children, the insane and the like is OK (e.g., the evil henchmen ask you which way the innocent victim went; you don't owe them the truth). But all of this was not Summer's position. We're not talking about circumstances in which one may lie, or should lie, but about whether you can say something true, with no immediate consequence and so not an emergency situation, which other people who are not in distress don't like.

In the transcript, he was very clear that he was speaking as himself, not as the President of Harvard; he was very clear that he had been asked to speculate and was about to; he framed the talk quite clearly. I would accept similar framing from the President of the US, and did so when one of the Bushes said he didn't like broccoli. That was clearly personal, not a statement of Federal policy.

What do you mean by 'the population who "agreed" with what he didn't say'?

Tom West writes:

What do you mean by 'the population who "agreed" with what he didn't say'?

That a lot of people agreed with him that biology proves that women can't really do science.

Of course, that's *not* what he said. But that's what just about everyone heard.

And that was a pretty much a given if he said *anything* that wasn't an out-and-out denunciation. Since he wasn't prepared to do that, his responsibility was to keep his speculations to himself.

And, as I'm certain the Broccoli industry can attest, Bush should have kept his opinions of the broccoli to himself until he wasn't president. (Although it's pretty small potatoes among his mistakes :-))

Shane L writes:

Even without dividing men and women into your example of hunter and gatherer groups, I can imagine that sexism would emerge in a mostly heterosexual society. This is because most people would have sexual and romantic feelings only for the other sex, and such feelings really make us vulnerable. Certainly most of the emotional pain I've suffered as an adult has come from women over whom I had powerful feelings of some kind. This isn't because they were cruel or hurtful, but sometimes things don't work out and, while it is nobody's fault, such failed romantic relationships are very painful. Likewise I fear I have hurt women much more than I have hurt men, again without malice but simply because these romantic relationships make us vulnerable to pain. Heterosexuals do not suffer such vulnerability towards members of their own sex.

I'm trying to be thoughtful about it, but I guess many people do not. Instead they recognise that most of the emotional pain they endure comes from the other sex. It's understand, if unfortunate, that they would then feel a general bitterness towards the other sex, and make negative judgements about that entire half of the human population.

Dave Hamilton writes:

Telling the truth in an academic setting or a public forum should always be encouraged. Like others I can think of plenty of times when telling a falsehood or neglecting to tell the whole story might be the most appropriate thing. However, the case where we are intellectually dishonest because some Adolph Hitler or Margaret Sanger may take it the wrong way and do some outrageous things with the information is not a reason for someone to be less than honest. In fact it is the less than honest intellectual who is responsible for much of the problems we face today. Just like the distrust in the media is fostered by their lack of truth telling and covering up of facts, once people see they have been lied to and misled they then fail to believe them even when they are telling the truth creating sometimes more serious problems. Using very personal examples of when it is ok to lie gives no insight into whether it is ok to lie in a public forum when we are talking about ideas and outcomes.

Comments for this entry have been closed
Return to top