What your test scores don’t say about you

A new study finds negative stereotypes can mask people’s academic abilities


Consider the following scenario: university admissions officers have narrowed applications for the final place in an engineering program down to two. The candidates have similar credentials and identical test scores; the only difference is that one is a woman and the other is a man. Who should they choose?

The answer may come as a surprise. According to a paper slated for publication in Psychological Science, the fear of confirming negative stereotypes about the intellectual capacity of women in math and sciences likely led the female applicant to underperform. Though her test scores may be the same as those of her male counterpart, the woman has a “significant untapped potential,” says University of Waterloo professor Steven Spencer, who co-authored the study with Stanford University’s Greg Walton. Put simply, she’s the better choice.

Test scores and grades have long shown an academic achievement gap between genders and ethnicities. In the past, this discrepancy has been explained by factors like poverty and poor schooling, which, it has been believed, lead to real differences in ability. Latent Ability: Grades and Tests Systematically Underestimate the Intellectual Ability of Negatively Stereotyped Students makes a new and very different case for affirmative action. While Spencer and Walton don’t deny that socio-economic factors play a role in academic performance, their research, gleaned from a compendium of studies that include 19,000 students in Canada, France, Germany, Sweden and the U.S., has identified stereotype threat as another cause.

090318_testsAccording to Spencer, stereotype threat comes into play whenever “you feel you can be judged based on a negative stereotype about your group.” As he explains, for non-Asian minorities and women (in quantitative fields), the belief that they don’t belong, or that the odds are stacked against their success, causes these students to “become excessively careful” when answering questions, a strategy that’s particularly ill advised on standardized tests. At the same time, he says efforts to “tamp down thinking about a stereotype … actually eats up a lot of their cognitive resources,” reducing the capacity of short-term memory. The feeling of belonging may be an abstract concept, but the implications are very real. On the SAT, for example, the professors say black and Hispanic students score about 40 points below their true ability, and on the math portion, women score about 20 points under where they should. And it’s not just on college entrance exams. Spencer says that stereotype threat in high schools and even junior highs mean grades and test scores could underestimate the ability of these students for the majority of their academic careers.

But there is hope. Compared to factors like poverty and poor schooling, says Spencer, reducing stereotype threat is relatively easy. Simple interventions, such as telling college students that there is no group differences on a particular test or getting junior high kids to write about values that are important to them can “make a big difference in performance,” he says. To eliminate it completely, he says academic institutions must work to identify at-risk groups and develop long-term strategies to make them feel accepted. “That feeling of belonging is really the antidote to this [belief] that you’re going to be judged based on stereotypes about your group,” says Spencer. In the meantime however, he says admissions officers should take stereotype threat into account when making decisions-not because women and minorities need a boost to succeed, but because tests hide the fact that they already have.


What your test scores don’t say about you

  1. So, given that boys are more likely to be disciplined at school, suspended, or expelled, are more likely to be arrested or commit suicide, and less likely to take advanced classes, participate in extracurricular activities, or go to college, shouldn’t their scores be incredibly deflated also? Sure, the female applicant may have “significant untapped potential,” but who’s to say the male applicant doesn’t also?

  2. It must be so difficult for girls to know that the majority of university students are female, and have been for the past 20 years. That must be a massive emotional hurdle to overcome. They’re all heros as far as I’m concerned.

    • “Heroines”.

    • Although I was taken to task a little while ago for talking about “actresses”. They are all “actors”, apparently.

  3. Negative stereotypes don’t disappear just because you’re not writing a test. Ok, so a woman’s test scores are lower than they could be because of stereotypes, but wouldn’t her performance at the university program she’s admitted too also be lower, again because of those stereotypes? Later on, will her performance at a potential job be equally lower because of that same stereotype?

    No one cares if people are more intelligent than their test scores indicate if those test scores are indicative of future performance. Taking a woman over a man just because women do worse than they should on math tests is downright sexist if woman also do worse than they should on other math-related tasks – the tests indicate that they’ll be equally effective in the program, so they’re equal.

    Besides, this “groundbreaking” research only goes to show that confidence matters when taking tests. If you believe you’ll do well, you will. If you don’t, then you won’t. Stereotyping obviously has a notable effect on confidence, and we should be making efforts to combat this dangerous effect in all situations (not just test-taking), but come on – this is just putting two and two together and claiming you’ve discovered four.

    • Agreed.

      If the issue is that so-called “identifiable groups” underperform because they’ve internalized some batty stereotype and aren’t confident enough to overcome it, then would they not continue to underperform even if they were hired?

      I don’t want to work with underachievers; I don’t care where they come from or what they look like. I can’t depend on a chronic underachiever any more than I can a cheap pair of nylons, no matter how much “potential” they may have.

      Oh, and Bill – apparently every actor, male or female, is an actor – but I suspect, like all PC-isms, this choice of phrasing blows in and out like the wind, and preference varies person-to-person.

  4. What a load of tripe.

    • No one said the women were underachievers, they just do not test ACCURATELY on certain types of tests. If the tests can be made more gender-free and accurate, I think that would be better. Women engineers actually perform well, and in some areas outperform men–in communicating with clients, for instance. Engineering workplaces have been crying out that engineers need more communication skills for years.

      Gender issues are no where near as simplistic as some of you seem to think. There are many layers of biases, including the way we tend to think about these problems. We should all want everyone to achieve their potential. It would make for a much better world for both men and women..

      • Nowhere in the article (or, to my knowledge, the study) is the type of test mentioned as a factor in performance. Only the inherent bias about the test is mentioned and the implication is that this inherent bias is present in all tests in a mathematics-oriented field. Thus, without corresponding evidence that, on average, if a man and a woman perform equally on typical test, the woman will do better in a math-related job or in a math-related university program, the findings of this study are essentially meaningless when interpreting test results. If underperformance on tests indicates underperformance in future performance as well, then the test is accurate, because potential is intangible and ultimately worthless if it doesn’t express itself in a person’s work.

        Look, no one is claiming that stereotypes don’t exist or have an impact on performance. Of course they do, and it doesn’t take a study to figure that out. The issue is trying to fix a problem of implicit or societal discrimination with explicit discrimination in the opposite direction. In the situation in the article, not only is it sexist against men, but it also reinforces the notion that women underperform on math tests, which would only further the impact of stereotyping.

  5. This article and the underlying study is garbage.

Sign in to comment.