Annoyed by Allies

[andrewducker]

In some ways it's more annoying when people who nominally believe things like you step over the mark and say ridiculous things than when people on the other side do so. As usual it's nominally 'objective' scientists taking a theory and then using anything to hand in order to 'prove' it. Robert Anton Wilson talks about this propensity - he divides us into two people - the thinker and the prover. In a perfect world the thinker would think whatever the prover had proved. In reality, the prover sets out to prove whatever the thinker already thinks. Robert Heinlein put it another way when he said that "Man is not a rational animal; he is a rationalizing animal." We will selectively ignore whatever evidence there is that contradicts what we already believe. We invent the most tenuous of excuses in the face of evidence of a world we don't like. I'm certainly guilty of this myself, as is everyone else I know.

However, it'd be nice to think that when people are publishing a scientific work that they would stick to things which, if not actual facts, were as close to facts as possible. It's all well and good me putting together a few chance observations of the people around me and the news and culture I have access to, but I expect scientists to work a little harder and actually try to, well, prove what they're saying.

Which makes this dissapointing. The new book from Michael Baron-Cohen (apparently very respected in the world of autism) explores differences in the brains of men and women. Sadly, while it contains some decent research and interesting studies, it also contains a vast amount of hearsay and anecdotal evidence which wouldn't stand up to any scrutiny.

So on the positive side there's things like:

...a study of rats in which the males were found to be better at finding their way through mazes. Baron-Cohen contends that this marginally supports his theory that men are better than women at reading maps. "In both the human and rat studies, a male superiority has been established when geometric (systemic) cues are available. Females tend to rely on landmarks (objects) in the room," which he says is not very systematic or reliable method.

Delving further into biology, Baron-Cohen finally gets cookin'. First off, hormones appear to play a factor in gender-typical behavior. Higher levels of testosterone, particularly in early development, he maintains, citing studies on everyone from rats to men with "very small testes" to male-to-female and female-to-male transsexuals, lead to an increased aptitude for systems and a decreased aptitude for emotional relationships. What's more, fetal testosterone may also affect the rate of growth of the brain's two hemispheres: the higher the testosterone level, the faster the growth of the right hemisphere, which has been linked to spatial ability, in which systemizing plays a role. The left side of the brain, meanwhile, is linked to language abilities, which is a key component of empathizing, Baron-Cohen says.

One particularly interesting study cited in the book finds that women tend to have larger left feet than right -- and larger left ovaries and breasts -- as well as dominant left brain hemispheres. These "left-greater" people tend to score better in language tests. Men, on the other hand, are generally found to have dominant right brain hemispheres -- and larger feet and testes. "Right-greater" people have been found to do better on spatial tests.

Still, Baron-Cohen seems much more comfortable when he turns to his area of expertise: autism. (He is the director of Cambridge's Autism Research Centre and has written two books on the subject, "Autism: The Facts" and "Mindblindness.") Here, he argues that autism, which afflicts far more men than women and is characterized by "abnormalities in social development and communication" and "unusually strong obsessional interests," and Asperger Syndrome, similar to autism but found in people with higher IQs and less severe communication issues, are both examples of "the extreme male brain." In other words, these are people for whom the systemizing/empathizing split is heavily weighted toward the former. (For the record, I scored very low on Baron-Cohen's appended Autism Spectrum Quotient test, meaning I show few signs of the syndrome.)

On the negative side there's vast amounts of things along the lines of:

For instance, to support his assertion that boys have a superior mathematical ability across cultures, after allowing that their work in school might be "less neat" than girls', Baron-Cohen looks at the entrants in the International Mathematical Olympiad, "in which the world's best mathematicians compete against each other."

"You can look up the winners on the Web if you are interested," he writes. "You will notice immediately that they are nearly all male. The Olympiad winners are listed by name, not by sex, but one can have a good guess at the sex of someone called Sanjay, David, Sergei, or Adam."

As the reviewer asks: This is valuable scientific evidence?

And the answer is that it might have been, if the time and effort had been put in to research it properly and check the backgrounds of the places these children come from - I can see Heron's immediate response to this that in most cultures the males are taught far more than the females are, which would easily slant the results, especially as "Sanjay" is a name associated with India, a very gender-based culture, "David" is possibly Jewish (another one) and so on and so forth.

It's only by sticking to the basics of what can be shown statistically to be (as) true (as possible) that inroads can be made. And if you can't show it to be true, then stating it as a fact based on trivial anecdotal evidence and guesswork harms your cause, it doesn't help it.

Comments

[heron61.livejournal.com]

For instance, to support his assertion that boys have a superior mathematical ability across cultures, after allowing that their work in school might be "less neat" than girls', Baron-Cohen looks at the entrants in the International Mathematical Olympiad, "in which the world's best mathematicians compete against each other."

"You can look up the winners on the Web if you are interested," he writes. "You will notice immediately that they are nearly all male. The Olympiad winners are listed by name, not by sex, but one can have a good guess at the sex of someone called Sanjay, David, Sergei, or Adam."

In addition to the author making a specious argument, the whole idea that men are on average better at math than women was recently revealed to be junk. A 20-year study of 1,700 students from starting public school to leaving college showed that in public school (including high school) women did (on average) slightly better than men in math. The other significant difference wrt math was that even women who excelled at math are less likely to go into math-heavy careers like engineering. The authors of this article attributes difference this to the fact that men and women on average value different things about careers. I'd also suggest that workplace bigotry likely had something to do with these decisions.

The most amusing gender studies are always the ones that definitively "prove" something already known to be incorrect.

[andrewducker]

See those bees?

I hear it's been proved that they can't fly.

Oh, I'd appreciate a link to the study. I'm collecting stuff like that.

[heron61.livejournal.com]

Sadly, the study was in a magazine, but it did have a link to the group that performed the study. I'd heard a little but about this study before and it seems fairly robust. Then again, I am far more likely to trust large sample size studies of actual human behavior than any study of brain activity or genetics.

My guess is that the earlier data about women doing less well in math was solely due to prejudice and cultural pressure. While these are only somewhat better, they are clearly enough better to eliminate that sort of discrepancy in performance.

[andrewducker]

Well I'd trust large sample sizes of actual people too.

I'm not going to believe anything firmly until we have a study of at least 10,000 people showing a correlation between their genes and some kind of social/intellectual characteristic.

My gut feeling is still that there is a strong genetic component in the underlying levels, but I'm not going to claim that it's proved until it actually is.

[heron61.livejournal.com]

I'm not going to believe anything firmly until we have a study of at least 10,000 people showing a correlation between their genes and some kind of social/intellectual characteristic.

My feeling is that you will be waiting an extremely long time, both because I don't believe that such characteristics have a significant genetic component (assuming no gross genetic and physiological defects) and have not seen and useful evidence that there are any such correlations. But you already knew that :)

OTOH, 10,000 is a good number for a study and assuming a reasonable degree of rigor in methodology and sample selection (which are sadly far too rare in such studies) I'm likely to believe the results of a study with this sample size.

[andrewducker]

Well, the NHS looks likely to start recording genetic data fairly soon and we'll have records on _everything_ for people in the UK that should allow for studies on that scale. But it'll take decades.

And of course, it may show nothing.

I may need to find some other way to occupy my time :->