Here’s an interesting study (paywall) by a team of psychologists from the University of Wisconsin-Whitewater and the University of British Columbia that speaks to just how strong is the link between our personal identity and the way we perform on academic tests, especially mathematics tests. In the study, a group of 110 female and 72 male undergraduates were given a 30-question multiple choice math test. At the beginning of the test, all participants were told that men usually outperform women on math performance. (Never mind whether this is true for the moment.) Then, one group of participants completed the test using their own names on the test papers, while another group used one of four fake names – two of which were male names and the other two females.
The males who took the test did equally well regardless of whether they used an alias or not – even if they used a female alias. However, the females who took the test did significantly better if they used an alias instead of their own names – even if the alias they used was female.
The authors of the study refer to the phenomenon driving this result as “stereotype threat”, in which a stereotyped group (women, in this case) fear doing poorly either because they fear that a poor performance on the test will confirm a negative stereotype as being true of themselves personally. Their identity gets so wrapped up in the performance on the test that it becomes a distraction and ultimately a cause of poor performance.
This result makes a lot of sense, although I wish it were counterintuitive. Negative stereotypes about women in the STEM disciplines start early and persist for a very long time in the lives of many women. I’ve seen this negative stereotyping happen even with my two daughters, who just finished the first and third grades respectively, in the form of peer pressure to think that girls aren’t good at math – or worse, that they aren’t supposed to be good at math and so if you are good at it, you’re some kind of a freak. For a girl coming up fast on her teenage years, there are few threats as grave as being different. And as for my students, anecdotally I can say that I’m much more likely to hear a female say “I’m no good at math” when I am working with her than I am a male. Girls get it into their heads that math deficiency is normative; and then deficiency becomes identity.
As a parent, two things I realize from this study are (1) this negative stereotyping is an active, invasive force in the lives of young girls, and it takes active positive stereotyping to counteract it, and (2) it’s primarily my job, as my daughters’ dad, to provide it. There are some great resources for bringing positive views of math and science to my kids; for example, they’ve become addicted to Mythbusters and I take every opportunity to talk up the show and particular Kari Byron’s presence on it. There’s also Danica McKellar’s books, which admittedly I haven’t read, but they might be appropriate for my oldest in a couple of years. But the primary thing to realize is that this is not somebody else’s problem (read: my daughters’ teachers’ problem).
From the standpoint of being a teacher, the article references alternative ways of giving tests that unlink students’ identities from the test itself, like putting numerical code on tests as an identifier rather than a name. (This has the added benefit of providing a somewhat blind review when grading the test.) Such methods are worth considering, but I wonder if they go far enough. I think part of the problem is the mode of assessment itself – specifically, the high stakes, timed math test. I won’t suggest that we should get rid of these, because I think they serve an important purpose. But if these are the only kind of assessments in a course, or even just the primary means of determining a grade, then I think we’re asking for the stereotype threat effect to rear its head. A mix of assessment types, including timed tests but also including group work, untimed written work, portfolio assessment, oral exams, etc. can be just as rigorous and produce authentic results that are less tainted by these kinds of psychological confounds.
(hat tip: Smithsonian.com)