Reverse “stereotype threat” – women chess players perform better against men than against other women

Stereotype threat is one of those social psychology concepts that has managed to break out of the academic world and into everyday conversation: the idea that a fear of conforming to stereotypes – for example, that girls struggle at maths – can make those stereotypes self-fulfilling, thanks to the adverse effect of anxiety and excessive self-consciousness on performance.

A recent review suggested that stereotype threat has a robust but small-to-medium sized effect on performance, but a meta-analysis suggests that publication bias may be a problem in this literature, inflating the apparent size of the effect. Also, the majority of the work has been done under laboratory conditions, which may not reflect what happens in the wider world. So when a field study comes along, it's worth paying attention to, and a paper published as a pre-print at PsyArXiv from Tom Stafford at the University of Sheffield looks at a domain involving high pressure, clear success criteria, and a presupposition that's it's more a guy thing: chess.

The handy thing about chess is that we have so much solid data: vast databases of matches, and effective ranking systems that allow accurate predictions of who is likely to win upcoming matches. To date, lab-based studies have suggested that women playing chess do suffer from stereotype threat, so Stafford's question is: what does that look like in real play?

To find out, he explored a dataset of youngish (average age 32) players ranked officially by FIDE, the World Chess Federation, including 150,000 men and 16,000 women, and the outcomes of over five million games.

The average man had a higher ranking – 2070 versus 1978 – meaning plenty of the games would put the women players in a challenging situation, which is where stereotype threat is most likely to manifest. And yet the data showed that whether playing a stronger, matched or weaker player, women performed better when playing a man than they did a woman. In other words, the data revealed exactly the opposite pattern of performance as predicted by the concept of stereotype threat.

Stafford looked closer at the data, to see whether the expected threat effect emerged under certain conditions, such as among younger and less experienced players; for older players who conceivably grew up around more sexist assumptions; or for players from national leagues where women are even more in the minority. Nothing changed the pattern: women perform better when playing men.

Nor was there evidence that stereotype threat triggered women to drop the ball on matches they were well placed to win. Looking at matches where one player had a rating 500 or more points above the other, Stafford found that female favourites were upset by men in 3.5 per cent of cases, but female underdogs beat the male favourite even more often, in 3.7 per cent of cases.

Stafford notes that another factor that increases the likelihood of stereotype effect is when the task attempted is unfamiliar, and this isn't the case here: the women players had years of experience. Still, the data didn't merely show no effect, but a reversed one, so there seems to be some sort of gender-related psychological process exerting an effect. Perhaps in domains where women are skilled and self-confident, anticipating a stereotype-related challenge actually sharpens their focus to buckle down. Or, says Stafford, maybe it's an issue of male under-performance, whether "male underestimation of female opponents, misplaced chivalry, or 'choking' due the ego-threat of being beaten by a women."

This isn't the first field investigation of stereotype threat to show different effects from laboratory studies. For example, Thomas Wei's investigation of children's maths performance found that gender primes before the task actually led girls to do better than normal. Findings like these don't prove that stereotype threat is non-existent in every situation, but they do suggest it is not as ubiquitous or straightforward as some have claimed. Just as with many phenomena in our rich, complex real world, gendered assumptions about performance may not have the same influence in every situation.