As a women with an advanced STEM degree, it often feels like you can’t win. You can power through a quantitative PhD and still find people look to the man in the room for all things quantitative. It seems I am not alone.
It is a fact that women are usually paid less and work in fewer higher-powered positions. Lifestyle and education choices explain some of pay differences, though even after controlling for these factors, some of the gap persists. A group of researchers from Harvard Business School and Stanford dug into what’s behind gender discrimination at work, aiming to figure out if employers discriminate because don’t want to hire women simply because they are women or if employers think being a woman signals something about ability or skills.
In one test, the subjects were administered quizzes on sports and math. On average, the women performed slightly worse on the quizes. Potential “employers” (who were not actual employers but randomly selected survey participants) were given the average quiz scores for each group. Knowledge in sports or math was relevant for the job.
One group of employers were given information on the gender of a potential employee and their test score. The researchers found that even if a male and female candidate had identical test scores, employers were more likely to hire men. There was only a 43% chance they’d pick a woman. (With no discrimination, they should pick a woman 50% of the time).
The researchers wanted to test whether the difference was just about gender or the lower-than-average test scores. The employers were again asked to choose between people with the same scores, but instead of revealing gender, they only revealed the potential employee’s birth month (all women were born in even months, men in odd) and the average scores for the even and odd group. In this experiment, women where only chosen 37% of the time, less frequently than when employers knew their gender. They concluded:
That members of the lower-performing groups are clearly discriminated against regardless of whether they are labeled as female (in the Gender treatment) or even-month (in the Birth Month treatment) suggests that historically disadvantaged groups like women are not the only groups that suffer from discrimination, pointing to a stronger role for statistical discrimination than animus. In fact, we observe that women are less likely to be discriminated against than the 12 non-historically disadvantaged group of even-month workers.
The researchers conclude that being associated with a weaker group lowered the odds of being hired, even when the workers had equal ability. But employers are slightly less prone to discriminate when they know their decision is based on gender.
The results suggest that discrimination stems from employers using gender to form inferences about ability. Even if a woman scores well on a math or sports test, being associated with a weaker-scoring group lowers the odds she’ll be hired. Projecting negative characteristics on any individual, of course, is just another form of discrimination.
Nonetheless, the information is useful because it offers some indication that reducing the perception that women are weaker in math can reduce some of the remaining gender gap. It shows why well-meaning, misguided ideas about women’s math deficiencies are harmful and why an honest conversation women and math is critical in overcoming the gender gap.