3 ms·
This analysis ignores both the (1) base rate, and (2) practice effect. The problem with using gender-neutral percentiles is that the proportion of males and fem
by gavazzy 11y ago
This analysis ignores both the (1) base rate, and (2) practice effect. The problem with using gender-neutral percentiles is that the proportion of males and females may not be 1:1. So, even if the percentage of women in the 99th percentile is higher than that of men, men could actually be more likely to be in the 99th percentile than women.
That is, they found p(women | 99th) / p(men | 99th) whereas they're claiming they found p(99th | women) / p(99th | men).
Regarding the second, it could be that the women were studying more, and thus not represent a genetic trait. In order to settle whether differences are genetic, you would have to compare performance from the same amount of effort.
I'm interested to see how the research continues in the future. It seems there is a lot of taboo around the subject, and I hope that doesn't dissuade scientific debate.
- Trombone12 11y ago> Regarding the second, it could be that the women were studying more, and thus not represent a genetic trait. In order to settle whether differences are genetic, you would have to compare performance from the same amount of effort. I think this is an important objection, but you stop far shy of its true implication. If just studying more would skew the results, then you cannot reasonably say that the method used has any significant power to detect the influence genetic traits. What is tested in a standardised test is arguably the combined effect of societal environment and genes, and considering that the field of the interaction between gene expression and the host environment (Epigenetics) is only just starting, even the genetic component might not be some static contribution.