How Stereotypes Shape Performance

S. Alexander Haslam, Jessica Salvatore, and Thomas Kessler
University of Exeter, UK

Every sports fan has vivid memories of key occasions on which a favorite team or player has 'choked' under pressure. And every student who has ever taken a standardized test knows what that kind of pressure feels like. What makes for high-pressure situations, and how do they influence performance? In the last decade such issues have been explored by social psychologists researching the phenomenon of stereotype threat. Their work shows not only that pressure can compromise performance, but that this dynamic is more common among members of negatively stereotyped social groups.

Why? The classic demonstration of stereotype threat, in a 1995 paper by Claude Steele and Joshua Aronson, emerged from a series of studies in which high-achieving African American students at Stanford completed the Graduate Record Exam (GRE) under conditions where they thought either that the test was measuring intelligence or that it was not a test of ability at all. Intriguingly, these bright students did much worse when they considered it an intelligence test.. This, the researchers argued, was because "in situations where [a negative] stereotype is applicable, one is at risk of confirming it as a self-characterization, both to one's self and to others who know the stereotype."

This tendency to perform worse when conscious of being in a group stereotyped as performing poorly is what is meant by stereotype threat. This pattern of findings has been replicated with many different groups on many different dimensions of stereotype content. The work of the University of Chicago's Sian Beilock and colleagues, reported in the latest issue of the Journal of Experimental Psychology (abstract or pdf download), follows that of many previous researchers in showing that if female students are made aware of a stereotype that men have greater mathematical ability than women, they tend to do worse on complex mathematical tasks than they do if they are not alerted to this stereotype.

Understanding process 
To help people combat the effects of negative stereotypes on performance, psychologists first need to understand why it occurs. Recent work has argued that one core factor is enhanced cognitive load. For example, work by Mara Cadinu and colleagues (abstract or pdf)has shown that when women perform mathematical tasks after being exposed to the stereotype that they are worse at math than men, they report entertaining more intrusive negative thoughts about their own mathematical ability, of the form "I am not good at math." Other studies have indicated that exposing people to negative stereotypes about groups to which they belong increases their anxiety and stress when performing tasks related to that stereotype.

Beilock and colleagues' most recent paper attempts to explore and integrate these ideas by delving deeply into the cognitive dynamics of stereotype threat. Working in the domain of women's performance on mathematical tasks, they ran a series of experiments that (a) replicated the standard stereotype threat effect, (b) showed that the effect is most pronounced on tasks which place demands on phonological resources (i.e., those which require verbal reasoning), (c) demonstrated that the presence of stereotype threat increases verbal reports of worry associated with either the task or the stereotype, and finally (d) suggested that the debilitating consequences of stereotype threat can be avoided if participants learn to perform tasks in such a way that they no longer consume working memory resources.

The working memory issue, they conclude, is critical. Accordingly, the authors recommend intensive training -- involving rote learning of solutions, so that answers become automatic -- as a way for women to avoid the debilitating consequences of stereotype threat.

Practical and theoretical limitations 
There are, however, a number of practical and theoretical reasons for believing that Beilock et alia's exclusively cognitive analysis is limited. On the practical side, it is clear that rote learning will not be a viable solution for all individuals or for all tasks. Such a strategy probably could (under precisely the right circumstances) reduce the harmful impact of stereotype threat. But as a general remedy the prescription seems unrealistic, particularly for complex tasks requiring intensive training. If you have trouble remembering phone numbers when put under pressure, learning the phone book by heart isn't an ideal solution. A sense that Beilock and colleagues' theoretical analysis is incomplete derives from other work inspired by Steele and Aronson's original demonstration of the effects of stereotype threat. This work shows that exposure to stereotypes can have welcome as well as unwelcome consequences.

For instance, studies by Margaret Shih and co-investigators show that exposure to positive stereotypes about one's group (relative to outgroups) can elevate performance instead of compromising it. Clearly such elevated performance cannot be explained in terms of cognitive load -- since it is hard to see how the salience of a positive ingroup stereotype (of the form "we are good") could increase the memory resources available to participants (relative to 'no stereotype' control conditions). Ideally, a parsimonious explanation of the stereotype threat effect would account for both upward and downward change. It should also be able to explain a host of other effects reported in the research literature -- including evidence that such effects are apparent in athletic domains where cognitive capacity is not critical and vary depending upon whether participants are encouraged to focus on promoting positive outcomes or on preventing negative ones. An explanation of effects arising from stereotype threat also needs to explain why they are not particularly generalized. For it is certainly not the case that all members of a given group succumb to the perils of threat.

On the contrary, effects are restricted to individuals who value the domain in question, and who have high levels of basic competence (i.e., those who, in the abstract, have less to worry about). A woman who loves math and is good at it, in short, is more likely than others to suffer from stereotype threat. How so?

Self and identity 
One answer is that stereotype threat is not so much an issue of cognition per se as one of self and identity. A number of researchers have made this point, including Steele and Aronson themselves, who argue in a 2002 review, co-authored with Steven Spencer, that stereotype threat can be understood as a phenomenon that centers on a person's social identity. That is, stereotype threat (and lift) effects eventuate to the extent that people are encouraged to think of themselves (i.e., self-stereotype) in terms of a particular group membership. (We will elaborate on this point in a full-length version of this article soon to appear in Scientific American Mind.) Thus, as the work of Beilock and others powerfully demonstrates, whether a person tests or plays well does not simply depend on high intellectual or athletic ability.

It also depends on whether the prevailing social context and the exigencies of group life validate the development and expression of these abilities -- and in particular, whether they allow them to be experienced as valid and 'natural' expressions of self. Descartes famously observed, "It is not enough to have a good mind; the main thing is to use it well." We might modify this to say "It is not enough to have a good mind; the main thing is to be encouraged to use it well." Such encouragement can obviously come from the direct actions of others -- the things they say and the way they treat us.

Less obviously, though, the stereotype threat literature shows us that such encouragement (or destructive discouragement) can also come from the social structures and associated stereotypes that we encounter and the sense of self that these engender. Where these promote self-conflict and self-doubt, performance will tend to suffer; where they promote self-belief and a sense of self-efficacy, it will generally be enhanced.

S. Alexander Haslam is professor of social psychology at the University of Exeter in England and serves on the board of advisers for Scientific American Mind. Jessica Salvatore is a post-doctoral fellow at the University of Exeter who recently completed her PhD at Princeton University. Thomas Kessler recently moved from the University of Jena in Germany to take up a position as professor of social psychology at the University of Exeter.

Elsewhere: Earlier this year, Sian Beilock wrote a Mind Matters of a simple and effective intervention that reduced the racial achievement gap in a Colorado high school. Beilock's web page offers several papers from her own lab -- including some about working memory. A 2002 study found that white athletes responded to stereotype threat about "natural athletic ability" by practicing less, as if to provide a rationale or excuse for subsequent underperformance. And this 1999 study, gorgeously simple, found that stereotype threat harmed both white and black golfers' performance depending on whether it was suggested to the players that golf performance reflected "athletic intelligence" (advantage: white golfers) or "natural athletic ability" (advantage: black golfers). ReducingStereotypeThreat.org offers a range of information on stereotype threat. And -- forgive me, I couldn't resist -- here's 8 minutes of Woody Harrelson and Wesley Snipes trading stereotype threats in "White Men Can't Jump." Decency caution: Woody and Wesley use some bad language. review-- Mind Matters editor David Dobbs -- Edited by David Dobbs at 12/05/2007 10:28 AM