Girls underperform when they play chess against boys – real-life evidence of stereotype threat?

Judit Polgár, chess grandmaster

An analysis of girls’ performances in 12 US school chess tournaments has found they tend to underperform when playing against boys. The researchers Hank Rothgerber and Katie Wolsiefer believe this is the first real-life demonstration in children of a phenomenon known as “stereotype threat”. This is when a person fears their performance will be used to bolster stereotypes about their social group. This fear then undermines their performance.

Most examples of stereotype threat have been demonstrated in social psychology labs. This has led to concerns that the phenomenon may not be so relevant in real life, especially since some studies of real exam grades have failed to reveal any evidence of the effect.

Rothgerber and Wolsiefer first surveyed 77 female school chess players and found they were familiar with the stereotype that men are better at chess than women (a stereotype reflected in the fact that there is only one woman, Judit Polgár, in the world’s top 100 chess players; see pic).

Next, the researchers analysed the outcomes of chess matches played by 219 girls (aged 5 to 15) in 12 tournaments rated by the United States Chess Federation. These official tournaments provide a pre-rating for each player based on their past performances, and a post-rating adjusted in line with their tournament performance. For comparison, the outcomes of tournament matches played by 195 boys were analysed.

The girls lost more often to boys than they should have done given their and their opponents’ prior skill ratings. Overall, they performed at 83 per cent of their expected success rate when playing boys. “Evidence of stereotype threat among young children, then, cannot be dismissed merely as an artefact of, or limited to experimental paradigms”, the researchers said.

Girls particularly underperformed (relative to their skill rating) when playing a male opponent with a higher rating than them (in this case they performed at 56 per cent of what was expected of them, on average); and when playing an older boy (managing an average of 73 per cent of their expected success). Younger girls were more susceptible than older girls to underperformance against boys. In contrast, there was no evidence of underperformance among the boys; in fact they often exceeded expectations. “This reinforces our interpretation that there is something specific to the interaction between female and male competitors that produced these performance deficits in females,” said Rothgerber and Wolsiefer.

The researchers’ interpretation was supported by their further analysis of the girls’ participation in future tournaments. Those who underperformed more against boys in the initial analysis tended to participate in fewer future tournaments during the ensuing year, consistent with the idea that stereotype threat can encourage people to disengage from an activity when they feel threatened.

Rothgerber and Wolsiefer said their results suggest interventions to combat stereotype threat are needed at an early age. In the context of girls playing chess, they said possible remedies include providing female role models and reframing the game as a problem-solving activity. “Whatever the method of intervention, the findings indicate that for females to fully experience the cognitive and emotional benefits of chess, the earlier the intervention, the better”, they concluded.


Hank Rothgerber and Katie Wolsiefer (2014). A naturalistic study of stereotype threat in young female chess players. Group Processes and Intergroup Relations DOI: 10.1177/1368430213490212

–further reading–
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.” (pdf)
Women’s true maths skills unlocked by pretending to be someone else
Female political role models have an empowering effect on women
Girlie scientist role models could do more harm than good
Women need female role models

Post written by Christian Jarrett (@psych_writer) for the BPS Research Digest.

15 thoughts on “Girls underperform when they play chess against boys – real-life evidence of stereotype threat?”

  1. What were the stats for the boys? I don't have anything to compare these arbitrary numbers against…

  2. I think it would be interesting – and the results of this study would be more relevant and academically rigorous – if the girls were tested in some kind of “blind box” game, i.e., each player is in a separate room and can't see each other, and moves are recorded by computer. How did the researchers determine the under-performance of the girl players was due to their awareness they were playing against a boy? Also, instead of going to all the trouble of recruiting “role models” (which may take decades), why not just have each tournament played in a “blind box” fashion to eliminate the bias?

  3. Many boys, especially the ones going through a chess phase in which they are engrossed in the game, are much stronger than their ratings because their rating might lag their strength by 100-200 points or more. Study the US Chess Federation rating lists for junior boys and girls and you will see boys becoming masters, IMs and GMs in their teens while their female contemporaries are advancing more slowly. Stereotype threat? No. Boys are a real threat to anyone with an established rating, including girls, because boys are generally under rated.

  4. If they aren't willing to do the test on one of the dozens of computer chess games that exist, with opponents not knowing their opponent's gender, then I would doubt the validity of these findings.

  5. I just read the research article this is based on. The results seem to me to show pretty clearly that girls underperform in chess tournaments and boys perform better than expected. This might be evidence of stereotype threat, though unfortunately the authors just seem to assume that it is rather than presenting the data then discussing what it may mean. I am very surprised that as academic researchers they don't make any attempt to consider alternative explanations for their results. Essentially they took measures of chess performance prior to a tournament and from these generated 'expected performance' in the tournament. Girls performed worse than expected and boys better. The authors phrase this as 'girls perform worse against male opponents', without acknowledging that this is the same as saying girls perform worse overall. In other words, any gender-based effect will only appear when a girl plays a boy, but it doesn't mean the effect is only there in those cases. If a girl plays against a girl and they are both performing below normal – i.e. the effect is down to playing in a tournament rather than playing against males – no effect will be observed but it is still there.
    So a fairly obvious alternative explanation for the results is that the girls do not cope with the stress of a tournament as well as boys, and this affects their performance. Unless I have missed something about the analysis, this would seem to explain all the findings in the report – e.g. the drop-out rate (girls who feel more stress stop playing chess) and the stronger effect when playing higher-rated players (more stress). However, the authors do not consider this. Despite their assumption that stereotype threat is the only explanation here, in reality further research is needed to assess whether different responses to stress, or the presence of stereotype threat, explain these findings.

  6. That women of lose more than expected when they play against men of equivalent rating is entirely consistent with regression towards the mean.

    Some facts:

    1. Elo ratings has error in it

    2. Men are on average better at chess [1]


    The average true value that elo ratings are trying to measure will be lower for women than men at equivalent elo ratings.


    [1] About 1 SD,

  7. This already exists in a sense.

    Most chess players also practice online, so you could try to find the correlations with all chess players inperson and online ratings.

    But there's also the chance of other thoughts getting in the way. One being the assumption of “maleness” online or in most cases where parties are anonymous. Another could be how players read their opponents, emotion and soft reads tend to not transfer well online(those being stereotypical feminine traits).

  8. The reason why girls underperform in chess isn't because of “the threat of stereotypes”, it's maybe because they just aren't as good since they a) don't practice as much OR b) just aren't good at strategic type games.

  9. When girls play in females-only tournaments -yes, there is such a thing, and they are fairly common- they gain or lose rating points to/from other girls, and that has the usual effect of overate or under rate them; the over rated ones are the most likely ones to play in gender mixed tourneys, and their ratings do not get to “normalize” till they keep playing solely in this kind of events. Hence is not surprising altogether that they appear to “underperform”. Get rid of that nonsense that “female only tourneys” are!

  10. One must take into serious consideration that intelligence and well played chess have a very weak correlation, if any at all. There are other elements at play (pun unintended). These may be the result of left-brain right-brain capabilities, which have been shown to differ between sexes. This does not mean that individual girls can't be good at chess, or that most good chess players are very smart. Chess is a game where achievement can be seen across all levels of IQ and education. And some very smart people basically admit that when it comes to chess, no matter how hard they study and struggle, they suck. In conclusion, it appears there really is no conclusion with regard to chess.

  11. Why is this framed in a way to make women seem like victims? You could use the exact same data to argue that boys try harder and exceed expectation when challenged by a girl, and even more so when challenged by a younger girl.

    An interesting follow-up would be to hold tournaments where only one player knows the gender of the opponent.

  12. Thank you for sharing valuable information. Nice post. I enjoyed reading this post. The whole blog is very nice found some good stuff and good information here Thanks..Also visit my page chess training win at chess, win chess, chess, winning chess, chess beginners, chess tips, chess training, chess strategy

Comments are closed.