My Sept. 11 post mentioned studies of how construals can influence performance. Another example comes now from Dar-Nimrod and Heine, who manipulated participants' beliefs regarding the source of gender differences in math and measured their subsequent math performance (see figure). In study 1 (7), women undertook a Graduate Record Exam–like test in which they completed two math sections separated by a verbal section. The verbal section contained the manipulation in the form of reading comprehension essays. Each test condition used a different essay. Two of the essays argued that math-related sex differences were due to either genetic (G) or experiential causes (E). Both essays claimed that there are sex differences in math performance of the same magnitude. Two additional essays served as a traditional test of stereotype threat. One essay, designed to eliminate underperformance, argued that there are no math-related gender differences (ND). The other essay, designed as a standard stereotype-threat manipulation (S), primed sex without addressing the math stereotype. Controlling for performance on the first math section, they used analyses of covariance to demonstrate that women in the G and the S conditions exhibited similar performances on the second math test. Women in the E and the ND conditions, although not different from each other, significantly outperformed women in G and S conditions).
Click to enlarge figure. Legend: (Left) Study 1 results (133 women). Scores on second math test (controlling for scores on first test) after reading essays. (Right) Study 2 results (92 women), used a slightly different protocol (see methods)
These studies demonstrate that stereotype threat in women's math performance can be reduced, if not eliminated, when women are presented with experiential accounts of the origins of stereotypes. People appear to habitually think of some sex differences in genetic terms unless they are explicitly provided with alternative experiential arguments.