http://www.spiegel.de/wissenschaft/m...600756,00.html
german unfortunately, and the link to this "proceedings of the royal society" doesn't seem to work anymore. He compares ratings of the top 100 of men's and women's rating lists.
He says the best 3 women are slightly better than statistically expected, next 70 are slightly worse, and the rest of the top 100 is slightly better again.
Of course the lists are biased because many women play almost exclusively women's tournaments so a similar pyramid is is to be somewhat expected. But the top women have held their own in the last years in men's tournaments afaik, so the lists should be comparable.
He also doesn't claim it's the only reason, he just says it's sufficient to explain 96% of the rating difference, whatever that means. Popular newspapers tend to get stuff like this wrong.
Edit: It should be noted that they used the german chess player database as a sample, so only players with a german national rating were apparently included. At higher levels, this rating is fairly equal to ELO ratings.
Last edited by Noir_Desir; 10-23-2009 at 08:10 AM.