Trying to quit [LWN.net]

Trying to quit

Posted Oct 2, 2007 17:50 UTC (Tue) by alankila (guest, #47141)
In reply to: Trying to quit by man_ls
Parent article: To Sir, with Love: How To Get More Women Involved in Open Source (O'ReillyNet)

> ... an absolutely artificial measure, which includes gender equality in
> the definition. If the dispersion is different for men and for women,
> that is an artifact of the measure, not a fundamental truth about the
> unknown g.

This looks like a contradiction to me. If measurements are corrected to produce identical result ("gender equality in the definition"), then how could the dispersion end up being different? And why wouldn't some approximation of the "fundamental truth" about g not be found in the gender bias function used to correct the results for equality?

to post comments

No contradiction

Posted Oct 2, 2007 21:18 UTC (Tue) by man_ls (guest, #15091) [Link]

IQ tests must be corrected from time to time, or the results tend to drift. For starters, among civilized populations IQ grows about 3 IQ points per decade. Since the average for a large population must be 100, by definition, the test results must be adjusted.

I'm no expert, but I can imagine that variations in dispersion between genders are possible too. Giving more weight to tests with bigger variance on one side and smaller in the other, and recalibrating to keep the same average, would eliminate the gender bias. Or maybe it is a statistical artifact from an imperfect fit to a normal. Remember, the normal distribution in IQ scores is there by design; nobody really knows if intelligence itself (whatever it is) follows a normal distribution or not.

And why wouldn't some approximation of the "fundamental truth" about g not be found in the gender bias function used to correct the results for equality?

Because g is not an intelligence profile; it is a scalar factor. Again from APA:

Some theorists (e.g., Spearman, 1927) have emphasized the importance of a general factor, g, which represents what all the tests have in common

This factor, in other words, represents "intelligence" itself, not ability at specific tasks. It is the thing that makes people solve problems better than chimpanzees or gorillas. It is what makes disabled children unable to learn beyond a certain point (as opposed to those with behavioral issues). Even its existence and utility is also a contended issue; read the article for a good introduction.