pedant [12.19.07 05:19 PM]

The distributions *are* the important bits, so it's important to get them right. Boys do *not* have a bimodal distribution of mathematical ability. Both boys and girls have roughly normal (bell-curve shaped) distributions. The distribution for males is broader and flatter than for females. The average (i.e., center of the curve) is almost the same for males and females, but the tails of the distribution are vastly different.

And the tails are really, really important. If you look at the population of people with mathematical ability above average, it will be almost equally males and females. However, if you look at the population of people with mathematical ability 3 standard deviations above the mean, it will be hugely dominated by males. (If you look at the population of people with mathematical ability 3 standard deviations below the mean, it will also be dominated by males.) The further out on the tails you go, the more the disparity. For example, if you look at the population of the 100 most mathematically talented people in history, it is likely to be *all* males.

  Dale Dougherty [12.19.07 06:05 PM]

Thanks for bringing up this important topic on Radar and pointing to the Scientific American article. I agree that we need to explore more ways to integrate verbal and analytical skills across all disciplines. It seems so wrong to begin sorting students into exclusive categories, artists in one corner and scientists in the other, to use one obvious example. This reflects our cultural bias, and our own oversimplification of the idea of intelligence. We justify it somehow on its predictive ability for determining future success.

A New Yorker article I just read touches on a similar hot topic -- race and IQ. "None of the Above" by Malcolm Gladwell in the December 17th issue analyzes the cultural context (or bias) of IQ tests and how they seem to measure those who have learned to think abstractly -- what Gladwell humorously calls putting on "scientific spectacles." (The key is that they have learned to think this way; and therefore IQ is not measuring an innate ability.) Children from different economic backgrounds start out equal but disadvantaged kids begin to fall steadily behind on IQ tests. A poor child is more likely to live in a less complex, less stimulating world. His article ends with the statement: "I.Q. measures not just the quality of a person’s mind but the quality of the world that person lives in."

I'll add one other point about science. I recently received a press release for a science competition for youth. It proudly announced that among the winners, several of them were girls. It also added that the winners came from parents who were themselves scientists. So, here's what I thought: good news that you have more girls in science but bad news if it takes a scientist to produce a scientist.

Family matters. It matters even more if the educational system is dysfunctional. It's another way the rich get richer, and the poor get left out.

  gnat [12.19.07 07:08 PM]

@pedant: thanks for the correction re:bimodality. As I said to Adam, the point of studying about this is to be able to think and talk accurately about it. All such "pedantry" gratefully accepted!

An interesting question is whether gender-specific education can make a difference. I know that there's a movement towards single-sex classes to help boys with reading and writing. The converse would be to have a math class just for girls, where they are pushed to improve their number skills. The reasoning goes that school caters to girls who are, in the classes I've taught, far more likely to be able to sit down and listen and act than the pointy-bum boys.

I would argue that separating kids by gender is pretty crude. I'd rather target education on performance and learning style than whether they have a willy or not. That way everyone learns to read, write, and count.

And next I'll tackle world peace ... :-)

  gnat [12.19.07 07:16 PM]

@Dale: "Family matters. It matters even more if the educational system is dysfunctional." I've been volunteering in my local school and I can't agree enough. Both of these are important: the kid is fighting an uphill battle if the home life isn't promoting learning, just as much as if the school life is miserable, uninspiring, or just plain failing.

  Nicholas [12.20.07 12:19 AM]

>Larry Wall, the creator of Perl, is a linguist by
>training, and there's a similar elegance in great
>code as in great poetry.

>Perl ... elegance in great code as in great poetry.

>Perl ... elegance

  jerng [12.20.07 03:16 AM]

Can't help you with the CS departments... but I thought I'd share my story. I'm a guy who was decent in math (790/800 for SAT 1 math SAT 2 physics) who scored low in verbal (700 SAT 1). The funny thing is, I think I'm better at languages... and later while studying introductory calculus and abstract math, I found that I spent more time thinking about the psychology behind how people process mathematical notation, than in the business of mathematics itself. Later, I found that I should just call myself a phenomenology :-/ Later I worked out how the semantics of math work, and nowadays, I'm interested in machine intelligence. Life never ceases to get weirder.

  Greg Wilson [12.20.07 03:34 AM]

The best review of recent research is Ceci and Williams' "Why Aren't More Women in Science"? (American Psychological Association, 2006, 159147485X, reviewed at http://pyre.third-bit.com/blog/archives/847.html). The most interesting finding comes from Dweck et al: if people (male or female) are told that something is an innate ability, they will *all* do less well when they first try it (because early failures signal "I don't have the ability"). On the other hand, if they're told that it's learned, they all do better on average, regardless of gender. Couple that with social stereotypes, and "math is harder for girls" becomes a self-fulfilling prophecy.

  Syndaryl [12.20.07 07:00 AM]

Posting as a woman in computer science:

Abstract reasoning and mathmatical logic are of course very important in programming, but I've always said I approach programming as much as a linguistic exercise as a logical one.

Good language skills don't just make you a good communicator to others (reporting on your findings to your peers, explaining to your boss why your department neeeeeds a new server, etc) but they also help you receive information from others.
If your professor isn't a great communicator, strong language skills might help you figure out what he was trying to say despite the problems, for instance.

  Meredith L. Patterson [12.20.07 07:58 AM]

The University of Iowa (where I did my MA in linguistics and am still ABD on my PhD in CS) offers three versions of its CS 101 class: the standard large-lecture-hall version with TA-led sections, an accelerated curriculum which covers more material, and the accelerated curriculum with only girls allowed in the class. (Girls can, of course, enroll in the mixed-gender section if they want.)

In practice, this ended up being similar to the remedial courses you suggest, since the few really sharp kids (of both genders) ended up in the accelerated sections and the TAs for the regular sections had more time to spend on the kids who needed the help (and meanwhile the TAs for the accelerated sections could actually field the questions that the sharper kids had, instead of waving them off till after class).

Aptitude testing for mathematical and verbal skills could really take it to the next level, though. I was recruited into my PhD program by a CS professor who taught my first computational linguistics course and poached me away from the linguistics department back when I didn't know much beyond Hello World in Python; I ended up being given a waiver on the "design of programming languages" class because it duplicated a lot of the material I'd already learned in linguistics, and spent my extra time becoming a better coder and tutoring other grad students for whom the Chomsky hierarchy and the like weren't especially intuitive.

  Alex Tolley [12.20.07 10:44 AM]

"What does this mean? I think it shows we need to do a better job of emphasizing that science and technology can be verbal as well as numerical"

Maybe. But the article clearly shows that the existing academic structure is very sexist, so change needs to happen at this level.

While I support the argument that career choices should be at the individual level and not based on society's biases, the very premise that the US is short of scientists and that women should be encouraged to fill that gap is a societal choice, not an individual one. Given the poor rewards for science careers, it may make far better sense for women to choose jobs that reward them more highly.

  Jo [12.20.07 12:02 PM]

Oh, I like Alex' comment. Indeed the differences are not sufficient to warrant the discrimination. Follow the money!

And PS you need a better spam catcher. This is close to being another "too hard basket"!

6th try - a right pain in a... - can't read this at all - could be anything

  EOM [12.20.07 01:27 PM]

This article is very timely for me. I've encountered the idea of the different bell-shaped distribution for men and women, but had never thoroughly thought about it in terms of "boys and girls look at their abilities and if they're better at numbers go into sciences and if they're better at words go into arts." Funny, because that seems to be exactly what I did. (Oh, and I'm a woman.) I was eerily good at foreign languages and literary analysis in high school, and picked my college major accordingly--even though I've loved science since I was a small child and first found my father's old George Gamow books, Nova, etc.

Well, circuitously I've found my way back in that direction and am planning to apply for a CS grad program next year. You can apply to the program in question conditionally, pending your completion of a yearlong prep program for people without a professional or undergrad CS background. I don't think program tests for different aptitudes in incoming students, but I believe the goal is indeed to give people with a different background a chance.

  Twiss Butler [12.31.07 09:07 AM]

It is refreshing to to see a group of people approaching this politicized subject with a clear focus on pedagogical strategies and values and with an evident will to avoid sex stereotyping. Since the underlying lesson taught by sex-segregated classes, whatever the pretext, is sex discrimination, I agree with gnat's common sense observation: "I would argue that separating kids by gender is pretty crude. I'd rather target education on performance and learning style than [on] whether they have a willy or not. That way everyone learns to read, write, and count."

  HG [01.03.08 09:09 AM]

There is another way to consider this for traditional Computer Science (not Physics, Math, or hard sciences). The question is whether software development even requires traditional Computer Science skills like studying algorithms or compiler theory.

Instead, are art and design, communication skills, and psychology (especially cognitive and behavioral for usability) more important?

The focus on math in software development is possibly the wrong approach given the current type of software we produce.