Monday, March 07, 2005
Raise Your Hand If You're A Woman in Science
The finding that emerges from the research, in experiment after experiment, is that bias is a problem not because it is deliberate, but because it is the outcome of assumptions of which we are not consciously aware. Take, for example, a study published last year by New York University professor Madeline Heilman and her colleagues. The researchers asked people to rate individual men and women who were described as holding the position of assistant vice president in an aircraft company. The evaluators' job was to rate how competent and likable the employees were. They were given background information about the person, the job and the company.
In half the cases, the employee was described as about to have a performance review (his or her competence was thus unknown); in the other half, the person was described as having been a stellar performer.
When the evaluators had received no information about how well the assistant VP was doing in the job, they rated the man as more competent than the woman, and rated them as equally likable. When the background information made clear that the person was extremely competent, evaluators rated the man and woman as equally competent. But both men and women rated the highly competent woman as much less likable than her male counterpart, and considerably more hostile.
Thus, in evaluating a woman in a male-dominated field, both male and female observers see her as less competent than a similarly described man unless there is clear information that she is a top performer. And in that case, they see her as less likable than a comparable man.
We also know that the differences within each sex are far larger than the average difference between the sexes. And we know that sex differences in math are smaller than cross-national differences. One study, comparing the United States, Taiwan and Japan, found that Japanese girls in grammar school scored almost twice as high on certain tests as American boys and almost always scored distinctly higher.
Maybe Asians are innately better at math. If so, following Summers's reasoning, Harvard should be preferentially hiring Asian women over American men. (We don't know what's behind the large cross-national differences -- although education is key -- and, as Americans, we're a little reluctant to think we're inferior.)
There is one cognitive ability that appears to be linked to sex differences in hormones. It's called mental rotation: the ability to look at a picture of a three-dimensional block figure and imagine it rotated in space. Males are much better than females at this task (although, with practice, someone of either sex can improve), and that result appears to be related to testosterone level. Girls who have experienced excess androgen in utero show higher mental rotation scores than normal girls. That's the kind of evidence we need to demonstrate a hormonal connection. We don't have that evidence for math or other cognitive differences. Does mental rotation ability matter? Maybe for a couple of scientific fields, but on balance, differences in math abilities seem better accounted for by differences in what we expect of women and how we treat them.
In half the cases, the employee was described as about to have a performance review (his or her competence was thus unknown); in the other half, the person was described as having been a stellar performer.
When the evaluators had received no information about how well the assistant VP was doing in the job, they rated the man as more competent than the woman, and rated them as equally likable. When the background information made clear that the person was extremely competent, evaluators rated the man and woman as equally competent. But both men and women rated the highly competent woman as much less likable than her male counterpart, and considerably more hostile.
Thus, in evaluating a woman in a male-dominated field, both male and female observers see her as less competent than a similarly described man unless there is clear information that she is a top performer. And in that case, they see her as less likable than a comparable man.
We also know that the differences within each sex are far larger than the average difference between the sexes. And we know that sex differences in math are smaller than cross-national differences. One study, comparing the United States, Taiwan and Japan, found that Japanese girls in grammar school scored almost twice as high on certain tests as American boys and almost always scored distinctly higher.
Maybe Asians are innately better at math. If so, following Summers's reasoning, Harvard should be preferentially hiring Asian women over American men. (We don't know what's behind the large cross-national differences -- although education is key -- and, as Americans, we're a little reluctant to think we're inferior.)
There is one cognitive ability that appears to be linked to sex differences in hormones. It's called mental rotation: the ability to look at a picture of a three-dimensional block figure and imagine it rotated in space. Males are much better than females at this task (although, with practice, someone of either sex can improve), and that result appears to be related to testosterone level. Girls who have experienced excess androgen in utero show higher mental rotation scores than normal girls. That's the kind of evidence we need to demonstrate a hormonal connection. We don't have that evidence for math or other cognitive differences. Does mental rotation ability matter? Maybe for a couple of scientific fields, but on balance, differences in math abilities seem better accounted for by differences in what we expect of women and how we treat them.
