Analysis of a Large Cohort of Competitive Chess Players

Only 1% of the world’s chess grandmasters are women. This underrepresentation is unlikely to be caused by discrimination, because chess ratings objectively reflect competitive results. Using data on the ratings of more than 250,000 tournament players over 13 years, we investigated several potential explanations for the male domination of elite chess. We found that (a) the ratings of men are higher on average than those of women, but no more variable; (b) matched boys and girls improve and drop out at equal rates, but boys begin chess competition in greater numbers and at higher performance levels than girls; and (c) in locales where at least 50% of the new young players are girls, their initial ratings are not lower than those of boys. We conclude that the greater number of men at the highest levels in chess can be explained by the greater number of boys who enter chess at the lowest levels. The game of chess has been studied by computer scientists and cognitive psychologists as a model arena of human intellectual performance. Research on computer chess has culminated in programs that can defeat the best human players (e.g., Hsu, 2002), and research on chess masters has yielded seminal discoveries, such as the chunk structure of short-term memory (Chase & Simon, 1973), and has contributed to debates on the importance of pattern recognition and deliberate thought in expertise (Burns, 2004; Chabris & Hearst, 2003; Gobet & Simon, 1996). But one of the most striking facts about chess competition has received little study: the dramatic lack of women among the game’s elite performers. None of the official world champions has been a woman, no champion of a major country is a woman, and as of January 2004, only 9 of the world’s 894 chess grandmasters—1%—were women (according to data in Howard, 2005). Analyzing possible explanations for the underrepresentation of women among the chess elite may help explain the underrepresentation of women at the highest levels in other fields, such as tenured professorships in mathematics, science, and engineering. It has been suggested (e.g., Pinker, 2005; Summers, 2005) that differences between men and women in the distribution of cognitive abilities required for success in these fields can partly account for the disparity (the ability-distribution hypothesis). In particular, men and women may differ in mean performance levels, variability of performance, or both; evidence suggests that in cognitive abilities, both types of differences are found (Halpern, 2000; Hedges & Nowell, 1995). However, the possibility of ‘‘old boys networks’’ of men who function as gatekeepers to high positions in these fields, coupled with the subjective nature of assessing achievement, makes it difficult to distinguish between an objective lack of achievement or credentials and discrimination by the existing social system as causes. In chess, there are neither gatekeepers nor subjective assessments; in particular, the rating system invented by Elo (1986) objectively measures individual skill solely on the basis of results of tournament games. The U.S. Chess Federation (USCF) applies this system to rate tens of thousands of players who participate in events that are open to all. Therefore, the overrepresentation of men at the highest levels in chess is, at first glance, more consistent with an ability-distribution hypothesis than with a social-system account. (Note that a difference in mean, variance, or both could explain the observed differences at the upper tail of the distribution.) However, other explanations are possible. One is that men and women may have differential dropout rates over time. Men and women may start out with equal endowments of the abilities Address correspondence to Mark E. Glickman, Center for Health Quality, Outcomes & Economics Research, Edith Nourse Rogers Memorial Hospital (152), Bldg. 70, 200 Springs Rd., Bedford, MA 01730, e-mail: [email protected]. PSYCHOLOGICAL SCIENCE 1040 Volume 17—Number 12 Copyright r 2006 Association for Psychological Science necessary for an endeavor, but women may be less likely than men to study and practice intensively or to devote obsessive amounts of time to it (the differential-dropout hypothesis). Indeed, some people argue that it is precisely the amount of deliberate practice that predicts success in fields like chess (Ericsson, Krampe, & Tesch-Römer, 1993). Thus, perhaps more potential female than male grandmasters leave organized competition, resulting in an imbalance at the top levels. Anyone who visits an open chess tournament will be struck less by the lack of women at the top of the results table than by their near absence at all levels. Only 9.7% of all USCF-rated games in 2004 were played by women. It is possible that the lack of women at the top is an artifact of their lower overall participation rate (Charness & Gerchak, 1996): Even if men and women have the same underlying ability distribution, a larger number of top-rated players will be men if the overall number of men competing is greater (the participation-rate hypothesis). That is, if fewer women than men even begin to participate in organized competition, dropout rates (and cognitive endowments) could be equal, but women would still be relatively absent at the top. In the present study, we asked whether these three hypotheses explain the enormous imbalance between men and women among the best chess players. Previous research on sex differences in chess performance (Charness & Gerchak, 1996; Howard, 2005) has considered only players at the top end of the rating spectrum. In this study, we analyzed the annual ratings of nearly all of the chess players who participated in USCF-rated games over 13 years, from 1992 through 2004. This is the broadest and largest sample of chess performance data ever analyzed, and one of the best data sets on sex differences in intellectual performance in any domain.