How Stereotypes Impair Women’s Careers in Science

While there are no sex differences in performance, both male and female employers are much more likely to hire a man than a woman to perform an arithmetic task.

• Without any information other than a candidate’s appearance, both male and female employers are twice more likely to hire a man than a woman to perform an arithmetic task. Both men and women also expect women to have a worse performance than men.
• The fraction of female candidates chosen when employers had no additional information and when they were subject to self-promotion by employees was almost identical. While receiving objective information about how well subjects had previously performed on the task attenuated the hiring bias by about 9 percentage points it did not eliminate it.  
• When candidates were allowed to self-promote, men tended to boast about their performance much more than women. Men overestimated their performance by 3.33 correct answers, while women only overestimated it by 0.44 correct answers.  
• Employers who had more biased beliefs (measured through an implicit association test) were less likely to account for the fact that men, on average, boast more than women about their future performance.

Both male and female employers are less likely to hire women for arithmetic tasks even when provided with more information about their aptitudes.

Subjects were hired to perform an arithmetic task in which they summed as many sets of four two-digit numbers as possible over a period of 4 minutes. First, all subjects performed the task and were informed of the number of problems they solved correctly. Next, two subjects were randomly selected to be candidates, while the remaining ones were to act as employers. Employers chose one of the candidates from the pair to perform the same arithmetic task. Although the employers chose candidates from pairs representing any combination of genders (to avoid making gender overly salient), the researchers only analyzed data from pairs that were one man and one woman. Each employer was randomly assigned to one of four treatments: 1) each candidate in the pair communicated to the employer about their expected future performance before the employer chose a candidate; 2) employers were told the actual performance of each candidate in the first arithmetic task before the employer chose a candidate; 3) employers first chose a candidate to hire without information other than the candidate’s appearance, and after making the choice, saw the candidate’s self-reported expected performance and were asked to update their choice; and last, 4)employers made an initial decision based only on candidate’s appearance and were then informed of their actual performance on the original arithmetic task, and asked to update their choice.  In order to measure implicit biases, all subjects were also asked to complete an implicit association test (IAT), in which they associated sex with science related abilities.

In total, 191 undergraduate students (83 men and 108 women) participated in 14 sessions. There were 94 pairs of candidates, of which 76 were mixed-gender pairs (subjects observed an average of 4.88 mixed-gender pairs).