Raw Thought

by Aaron Swartz

Discrimination and Causation

First, let’s imagine that tomorrow scientists announced the discovery of rock-solid, unimpeachable, 100% convincing evidence of differences in mental function between men and women. Let’s say, for example, they notice that there’s a tiny hole where the “math center” of the brain should be. No wonder they do worse at math!

No doubt, The Times would respond with a handwringing article about the important scientific implications and David Brooks would throw a party and denounce closed-minded liberals. George Bush would cancel programs aimed at helping girls learn math and Harvard University would shut down their task force on getting women tenure.

But are these really appropriate responses? Showing genetic differences is only the first in a long line of things that need to be shown to prove that gender-based disparities in tenure are unavoidable. As Jeremy Freese has pointed out, it’s a long line from genes to social outcomes. To make the case, you need to go a lot further.

Second, you have to show the genetic differences are relevant. It’s possible the hole in the math center could be completely insignificant, that women do just as well at math irrespective. So you need to show that the hole causes differences in functioning. One way to do this is to find different people with differing sizes of holes, control for as many other factors as possible, and see if the size of hole is correlated with some test of math functioning.

Third, you need to prove that the differences are unavoidable. The brain has amazing levels of neuroplasticity. Perhaps with the right environment, women can be taught to do math with another part of their brain. Perhaps, as a result, they might even do better than men at math. Again, Freese has pointed out that the same genetic differences (or genetic similarities) can go all sorts of different places in different environments. If there’s an easy environmental change that makes even genetically different women equally good at math, we ought to make it.

Fourth, you need to show a causal link from the genetic difference to the tenure disparity. Why is it that doing worse at math causes you to do worse at tenure? Are speed-math-tests used as a relevant factor in tenure decisions? If so, maybe you guys should really cut that out, because that’s a pretty stupid test.

Fifth, you need to show that it’s the only cause of discrimination. Even if genetic differences cause some of the disparity, it’s still morally required for us to remove the rest. Do guys with holes in their math center do just as bad as women at getting tenure? Do women with no holes do just as well as men?

Right now, there’s only even arguable evidence for the very first of these. Those of us who want to shove discrimination under the rug need to do a lot more work on the other four.

You should follow me on twitter here.

May 23, 2007


“But are these really appropriate responses?”

No. We all know plenty of women who are good at math.

Given a mental task X that a man is good at, you’ll likely find a woman who is good at it as well. This we simply know from experience.

posted by itistoday on May 23, 2007 #

No. We all know plenty of women who are good at math. Given a mental task X that a man is good at, you’ll likely find a woman who is good at it as well. This we simply know from experience.

Yes. And given a tall person, you can almost certainly find a short person who could beat that tall person at basketball. A correlation doesn’t have to be 100% to be valid — and as far as I know, there isn’t even a correlation between mathematical talent and gender. There is a correlation between variance in talent and gender: men are (as demonstrated by standardized test scores and subsequent careers) more likely to be really good or really bad, and similarly less likely to be average. Could be cultural, could be genetic, could be random — but that’s only really important if you want to fix it, rather than just wanting to know it. And if you want to fix it, I’d like to know why.

The higher standard deviation means that for accomplishments that require the top fraction of a percentile, men will be overwhelmingly dominant even though the average man isn’t better at math than the average woman.

posted by Byrne on May 23, 2007 #

Aaron, as you see from Byrne’s response - not to mention The Bell Curve - you are dealing with rationalization, not rationality.

Not that I mean to discourage fighting the good fight, especially for higher-readership people such as you. But myself, I find the limited power of rational argument to be very frustrating.

posted by Seth Finkelstein on May 26, 2007 #

Before someone get huffy, I should note the refution of the above statistical argument, which granted is a bit subtle: The problem is that it relies on a simplistic statistical model in exactly the range where that model is least reliable - which, rationally, should at least give pause regarding how correct an explation it is.

The deep problem is that I’ve never seen anyone who advanced that argument then say, “Y’know, you have a really good point there. All models are provisional, and I never realized there was a subtle flaw in my statistics there”. It’s always something like “No! The model says! It’s just natural, men get to be in overwhelmingly in the top spots, science proves it!”. This is why a certain type of liberal comes to hate science - I strongly disagree with them on their anti-science views, but I sure understand where they are coming from.

posted by Seth Finkelstein on May 26, 2007 #

There is a correlation between variance in talent and gender: men are (as demonstrated by standardized test scores and subsequent careers) more likely to be really good or really bad, and similarly less likely to be average. Could be cultural, could be genetic, could be random

It could be that standardized tests measure the ability to perform well on standardized tests, not apply abstraction and logic to real-world problem solving.

The career observation is completely bound in gender-role norms/expectations and, most likely, indicative of only that.

Such makes ‘mathematical talent’ extremely difficult to measure.

posted by Carty on May 28, 2007 #

The main reason genetic differences matter is that we might be able to enhance humans by knowing them.

If there is a “smarts” gene ( i know, i know — a whole set of em with tradeoffs … ), we should find it.

The problem is that it is completely taboo to even suggest people study the issue earnestly.

Think about the opportunity cost of that decision. Specifically, think about the unenhanced researchers whose development is relatively impaired compared to where they could have been without the taboo.

Clearly there are huge genetic and sociological issues leading to the trends in disparity of outcome between groups.

Hoping to fight the unjust causes, such as blatant discrimination, should not come at the expense of hindering research.

posted by Ivan Kirigin on May 29, 2007 #

I’d be delighted to respond to Seth’s counterarguments about the correlation between genotype and phenotype, but so far, Seth has only addressed the correlation between being me and being Bad and Wrong and Dumb.

Of course models are provisional! Of course they could be wrong! I’d be delighted to have some sort of laboratory to test them — perhaps some kind of educational system in which people of given attributes take tests, and these scores are recorded and later compared to the average success rates in those groups. For example, if our tests revealed that men and women had roughly equal ability, but different variances in skill, and if their later success demonstrated approximately equal average ability, but substantial variance in skill, we might decide that the correlation between gender and variance in mathematical ability implied some sort of correlation between gender and variance in mathematical ability.

Until that happy day, Seth and I will just have to use ad hominem arguments to emphasize how awful it is to disagree with us. Which is too bad, y’know?

posted by Byrne on May 29, 2007 #

Men are notably bad at designing civilizations. Case in point: human history.

posted by Connelly Barnes on June 10, 2007 #

Actually, the most likely response to this would be a simultaneous outcry from those who say, “See, we always knew this, now we have proof” and those who say, “Studies like these could lead to discrimination and the scientists were irresponsible for even studying the question.” Its largely the exact same response when studying race, IQ, poverty, religion, sexuality, etc. Aggressive egalitarianism has many sacred-cows when it comes to refusing to acknowledge differences, for fear they can be used for further discrimination.

My own feeling is that the largest reason for the appearence of a gap in male/female math abilities, reflects preferences related to the types of employment math is used in, and how it is used. Perhaps the job of university physics prefessor isn’t, for completely different reasons, as attractive to women as other employment.

On the other hand, people who firmly deny sex-based differences in ability, have no problems embracing the idea that men are more aggressive and violent than women.

posted by William Crim on June 14, 2007 #

You can also send comments by email.

Email (only used for direct replies)
Comments may be edited for length and content.

Powered by theinfo.org.