There's an extremely interesting backstory to the current claim by intelligence researchers that men lie on the extreme ends of intelligence distribution. It used to be that intelligence research simply postulated higher intelligence in men. This made a lot of sense to intelligence researchers. 1/ - ThreadSky

jensfoell.de • 18 days ago

There's an extremely interesting backstory to the current claim by intelligence researchers that men lie on the extreme ends of intelligence distribution.

It used to be that intelligence research simply postulated higher intelligence in men. This made a lot of sense to intelligence researchers.

1/

Reposted from Moira Donegan

Matt Yglesias endorses the theory that men are naturally more inclined toward extreme intelligence over on the other site.

Comments

justinesherry.bsky.social•18 days ago

There's also a really fun statistical fallacy that often pops up in these arguments -- basically, if you have observed more samples, you're more likely to have samples from the tails of the distribution.

justinesherry.bsky.social•18 days ago

So when you look at distributions of say, chess scores, for men vs women, since men are much more likely to play chess, you see longer tails for men, but you see longer tails because there are so many more men than women in the dataset.

strangecurrencies.org•18 days ago

That this isn't obvious to people is so strange to me. If there's a 1 in 1,000,000 outlier position, and you look at 5,000,000 people in one set and 500,000 people in another, you're statistically going to see multiple such outliers in the former sample and very likely none in the latter.

yminsky.bsky.social•16 days ago

I would have thought it was more complicated than that. You can compute an unbiased estimator of a distribution from a sample, if you know the family of distributions you're sampling...

yminsky.bsky.social•16 days ago

But if you're wrong, i.e., you assumed it was normal, but I'm fact it's heavy-tailed, then you will end up biased low in the way you say.

But I would have naively thought that things like chess ability were normally distributed...

yminsky.bsky.social•16 days ago

Is there a write-up of the details of this anywhere? It would be fun to read!

justinesherry.bsky.social•16 days ago

I don't think it's impossible to statistics your way out of the problem, just most people don't do it.

gojomo.bsky.social•17 days ago

but don't tests where women are oversampled, like the SAT, also show more males at the extremes?

jensfoell.de•18 days ago

After all, that's what the world looked like to them 100+ years ago, with basically all positions of societal responsibility given to men, almost never to women. What weird world would do such a thing if it wasn't based on merit? Plus, the researchers were usually men themselves. It sounded good.
2/

jensfoell.de•18 days ago

At some point, the evidence against this idea became way too strong. It was very easily apparent that women, if they had a chance to show it, were at least as competent as men in matters of the mind, if not more. This should have been seen as a refutation of much of intelligence science.

3/

jensfoell.de•18 days ago

Instead, the new generation of intelligence researchers interpreted their data around this new perspective. Women are about as intelligent as men, they said, probably even a bit higher on average, BUT men occupy the extreme ends. The consequence of this is that all geniuses can still be male.

4/

irredeemablebunny.bsky.social•18 days ago

Ah yes, that good old trick: the difference is always in the first statistical moment we haven't yet done research on.

jensfoell.de•18 days ago

Postulating this is very important to many people (especially men) who otherwise would have to acknowledge that many of the geniuses they've come to know were/are actually just random fools who have been handed money and/or influence. Acknowledging this can be hard or even impossible for some.

5/

susannevogel.bsky.social•18 days ago

Regarding the greater variance in men hypothesis, I came accross this preprint https://osf.io/preprints/osf/83wma_v1, which is (afaik) still not published. It states that greater variance may be artifial as by those tests were designed and optimized to test intelligence in men. Do other people have a take on this?

jensfoell.de•18 days ago

"But surely you're not saying that the findings of intelligence science can just be read in whatever way suits the current cultural climate"
Weeeell for this we need to talk a tiny bit about data vs. interpretation.

6/

jensfoell.de•18 days ago

It's been clear for decades that whatever intelligence research measures correlates with many important things. Meaning: High scores on intelligence predict later educational or financial success, successful relationships and all that. Those data are out there and have been replicated plenty.

7/

bertywooster.bsky.social•18 days ago

The question is simple, can you extrapolate any score on a test beyond the context of that test? Part of the difficulty is that IQ tests definitionally return results that look like normal distributions centered on 100. The test questions are tuned specifically to deliver that result.

bertywooster.bsky.social•18 days ago

I recall being taught that Terman (?) removed questions that women were more likely to get right in order to ensure both men and women had the same average. Given the statistical goal that makes sense. However, does it obscure a material fact?

katennp.bsky.social•18 days ago

this is appalling misuse of data

Comments

Posting Rules

Reply