Helping girls like math is not the same as helping them to do well.
Attitudes and beliefs about math have different effects on performance.
Posted Oct 27, 2010
There is also evidence that women sometimes perform more poorly on important tests of math achievement than they should given their ability. That is, their scores on these tests do not reflect their true ability, because of a phenomenon called stereotype threat. Stereotype threat was first described by Claude Steele and Joshua Aronson in a 1995 paper in the Journal of Personality and Social Psychology. The basic idea is that if you are a member of a group that is the subject of a negative stereotype, then your concern about showing that stereotype to be valid can harm your performance. So, women who are aware of the stereotype--women are worse at math than men--do poorly on math tests.
Stereotype threat has been an important area of research, and I have written about it before in this blog. One stream of work relating to stereotype threat tries to find ways to eliminate it. An interesting set of studies in that vein was presented by Chad Forbes and Toni Schmader in the November, 2010 issue of the Journal of Personality and Social Psychology. They were particularly interested in the difference between women's attitudes about math and their beliefs in stereotypes about math performance. This issue is important, because it is possible that women are less likely than men to pursue math because they don't like it. It is also possible that they don't pursue math because they are concerned with doing poorly.
Later in the experiment, participants were given the opportunity to work on different kinds of problems, and some of them were math problems. Participants could choose which problems to work on. Women in this study were more likely to choose to work on math problems when they were given the training that math is good than when given the training that math is bad. However, this training did not influence their performance on the problems. That is, they did not solve a higher percentage of the problems when trained to have a more positive attitude about math.
In another study, participants got training related to the stereotype about math. In this case, they classified items either as math-related or language-related or things that women are good at or that men are good at. The group that was trained to discount the stereotype used one button to respond both to things that women are good at and to math words and to use the other button to respond to things that men are good at and language words. A second group did the task the opposite way (women are good at and language words; men are good at and math words).
Later, the participants did math problems. In this case, women who were given training to help them discount the stereotype actually performed better on the later math test than those who got training that was consistent with the stereotype. The researchers also gave participants a test of their working memory capacity. Working memory is the kind of memory you need to use when you are solving problems. It refers to the amount of information you can keep in mind at once. In general, having more working memory is better than having less when solving math problems. Previous research by fellow PT blogger Sian Beilock and her colleagues has found that stereotype threat situations decrease people's working memory capacity. Forbes and Schmader found that the training to discount the stereotype increased women's working memory capacity.
There was a lot of detail here, so let me try to summarize. Improving women's attitude about math made them more likely to choose to do math problems, but not more likely to do better on those problems. In contrast, changing women's belief about the stereotype affected their performance on math problems, but not their willingness to do math.
These results suggest that if we want to increase the number of women who participate in math and science careers, it will be important to influence both attitudes about math and beliefs about stereotypes. The results of this study suggest that both are possible.
Follow me on Twitter.