Table of Contents
The concept of stereotype threat and stereotype boost
The role of self-relevance
The role of manner of stereotype activation
The concept of stereotype threat (Steele & Aronson, 1995) has received considerable attention in the past few years. In several studies, Steele and his colleagues took a close look at the influence that negative stereotypes can have on individuals in performance-related situations. As a reaction to the initial concept, the research also extended to other phenomena related to stereotype threat, such as the influence of positive stereotypes in performance-related situations. However, this investigation of the other side of stereotype threat and further studies that have been done regarding stereotype threat in general resulted in contradicting findings.
My thesis presents a focused review of the available literature first. This is done to provide a basis for the conceptual framework Shih and colleagues proposed (Shih, Ambady, Richeson, Fujita & Gray, 2002). Their work integrates the conflicting findings and suggests two possible factors that might regulate the effects that positive and negative stereotypes have on people: self-relevance and the manner of stereotype activation. In my study, I tested this framework in replicating and critically evaluating the study Shih et al. (2002) have conducted. The results and implications for future research are presented.
The concept of stereotype threat and stereotype boost
Stereotype threat refers to“the threat of being viewed through the lens of a negative stereotype or the fear of doing something that would inadvertently confirm that stereotype” (Steele, 1999, p.46). Steele and Aronson introduced the concept of stereotype threat in 1995 when they conducted a study with White and Black participants taking a test of verbal abilities. They found that merely asking Black participants to record their race on a demographic questionnaire before taking the exam significantly decreased their test scores compared to scores of White participants. They had less items correct, fewer items answered and spent more time on questions. When not asked to record their race, all participants scored equally well (Steele & Aronson, 1995). The researchers controlled for prior ability differences by taking into account the SAT scores of the participants.This seemed to indicate that simply making the group membership salient for which there is a negative stereotype worsens performance in test situations that refer to that stereotype. The same results were found when women’s math performance scores were compared to those of men and the women had been told in advance that men usually did better on that test, again with appropriate controls for ability differences (Spencer, Steele & Quinn, 1997). Steele extended the findings in showing stereotype threat and resulting performance deficits in White men being compared to Asians on math tests when they were told that Asians usually scored better on the test (Steele, 1999). These studies showed how heavily people can be temporarily affected by salient negative stereotypes, even when they do not belong to a group that is usually stigmatized, as in the latter example. Further studies indicating the same pattern of results concerned Caucasians from a low socioeconomic background and their test scores (Croizet & Claire, 1998) or Latino students and test scores (Gonzales et al., 2002; Aronson, Quinn & Spencer, 1998). Steele (1997) proposed five main features of stereotype threat:
1. Stereotype threat affects members of any group about whom there exists some negative stereotype. Stereotype threat has the potential to affect members of any group, if a situation makes them believe they will be viewed in light of a negative stereotype.
2.A stereotype must be relevant to one’s self for it to be threatening; one must care about the domain or behavior that the stereotype describes.
3.Stereotype threat is variable across different groups and situations.Different groups experience different degrees of threat depending on the content of the stereotype and the situation.
4.One need not believe in the stereotype for it to be threatening.
5. Trying to disprove a stereotype – for example by outperforming - has detrimental effects and paradoxically leads to a decrease in performance.
Most impaired by stereotype threat in one of Steele’s studies were “the most achievement oriented students, who were also the most skilled, motivated, and confident” (Steele, 1999, p.48).This surprising finding underlines points four and five. By trying to disconfirm the negative stereotype about their group - and the more talented they are, the harder they might try- people are plagued with distraction, self-consciousness, evaluation apprehension, test anxiety, and loss of motivation - leading to a decrease in test performance (Croizet et al., 2001). One condition that is important to note is that the performance task needs to be challenging. On easy or well-learned tasks, additional effort can actually boost performance under stereotype threat (O'Brien & Crandall, 2003).
Now, if stereotype threat worsens performance, should positive stereotypes not boost performance then? Following this notion, researchers subsequently conducted studies where the stereotype activated was a positive one. Levy (1996) found that stereotype activation through subtle priming can lead both to a decrease and an increase in performance, depending on the valence of the stereotype. Positive stereotypes of the elderly (e.g., wise, experienced) improved the memory performance of elderly participants whereas negative stereotypes (e.g., senile, dementia) worsened memory performance. Similarly, Shih et al. (1999) found that Asian American women performed better on a math test when their Asian identity was made salient, however they worsened when their gender was made salient. Those studies seem to support the stereotype boost assumption and extend the influence that stereotypes can have on behavior from negative ones to positive ones, indicating that it is not only a threat making people sensitive to stereotype information, but rather a general susceptibility to stereotype information in relevant situations (Shih et al., 2002).
However, there are also studies showing that when exposed to positive stereotypes, participants reacted with decreased performance like those under a stereotype threat condition. For example, in a study where Asian women were told that Asians usually did better in math than non-Asians, they performed worse on a math test than the control group (Cheryan & Bodenhausen, 2000) In addition, findings of stereotype-consistent performance with participants to whom the stereotype was self-irrelevant (e.g. Bargh et al., 1996, Dijksterhuis & van Knippenberg, 1998) increased confusion about what actually happens in settings where positive or negative stereotypes are made salient. Two possible moderators will be presented here.
The role of self-relevance
According to Steele, a stereotype must be relevant to one’s self in order for it to be threatening; one must care about the domain or behavior that the stereotype describes. The mentioned study by Levy (1996) confirmed that notion by comparing the effects that priming of elderly-related words had on the elderly versus young participants’ performance. Indeed the younger participants were not affected by the prime. The stereotypes did not relate to them. In contrast, there is also evidence in the following studies that people sometimes behave in a stereotype-consistent way even though the stereotype that is activated has no self-relevance for them. Bargh et al. (1996) conducted a study similar to Levy’s. After priming young college students with stereotypes of the elderly, researchers found that the participants walked down a hallway more slowly than the control group, consistent with the stereotype information they just had received. Similarly, Dijksterhuis & van Knippenberg (1998) found that asking students to think of traits that either characterized professors or soccer hooligans influenced their performance on a general knowledge test. According to the stereotype they had thought of, the people in the professor condition performed significantly better than the people in the soccer hooligan condition.
The question to be answered is whether there is a difference between people to whom the stereotype is self-relevant and those to whom it is irrelevant. Wheeler and Petty (2001) suggest that stereotypes influence people through two different processes: one affecting them when the self is implicated, implying additional motivational concerns because of the self-relevance, and one affecting them without self-relevance, where motivational concerns do not play a role. The authors refer to the two processes as “hot” when the self is concerned, and “cold” when it is not.
It is crucial to mention that there is also research investigating if and how motives and task-related goals can be activated automatically by aspects of the environment. People then engage in goal-directed cognition and action without awareness, i.e., also if the motives are activated very subtly (Bargh & Chartrand, 1999; Dijksterhuis, Smith, Baaren, Wigboldus, 2005).
However, the hot versus cold process idea can account for the conflicting findings mentioned. A possibility of how nontargets, i.e., people whose self is not implicated in a given situation (cold process), could be affected at all by stereotype information is through the “perception-behavior-expressway” model of social perception (Bargh et al., 1996). It states that an observed behavior or a perception influences behavior automatically. Priming traits (e.g., rude) leads to trait-consistent behavioral tendencies, like interrupting an ongoing conversation faster (Bargh et al., 1996). Since stereotypes are also a set of traits, priming a nontarget positively could lead to consistent behavior through the “cold” process where no self-relevance is needed. This would result in stereotype boost and therefore improved performance if the person is primed with a positive stereotype. If a negative stereotype is activated, however, performance should decrease. This perception-behavior link should be effortless and automatic and should not need awareness of the prime. In a study by Wheeler et al. (2001) non-African Americans who were instructed to write a story about a day in a certain Black person’s life later performed significantly worse on a math test, in congruence with the stereotype about academic abilities of African Americans.
The reason why stereotypes could affect targets differently from nontargets is because they are influenced through the hot process, where, when the self is implicated, additional motivational concerns come into play. This extra burden presumably interferes with the ability to perform as usual and the typical decrease in performance occurs (Schmader & Johns, 2003). A related study was able to show that under classical stereotype threat conditions, targets’ working memory capacity was reduced (Schmader & Johns, 2003). Several other studies indicate that under stereotype threat, people also struggle with their emotions, as is evident in increased blood pressure (Blascovich, Spencer, Quinn & Steele, 2001) or higher levels of anxiety (Aronson et al, 1999; Spencer et al., 1999; Stone, Lynch, Sjomeling, & Darley, 1999).
The differentiation between hot and cold processes and the behavior-perception link could explain why self-irrelevant stereotypes also have an influence on people. When the perception-behavior link is activated, group membership does not matter. But still it seems that stereotype information does not always influence nontargets in this way as was the case in Levy’s study mentioned earlier (Levy, 1996) Also the question remains as to why among targets positive stereotypes can create both performance boost like in the study about Asian women performing differently depending on the identity made salient (Shih et al, 1999) and performance decrease like in the study mentioned about Asian women suffering from a positive stereotype (Cheryan & Bodenhausen, 2000).
The role of manner of stereotype activation
Shih et al. (2002) argue that these remaining questions can be answered by a hypothesized relationship between the self-relevance and the manner of stereotype activation. Stereotypes can either be activated quite blatantly, so that the participant is aware of the stereotype and maybe even the link between the stereotype information and subsequent behavior (Bargh, 1994), or rather subtly as to avoid the participants’ awareness of the stereotype and its relation to behavioral tendencies. A very blatant activation could be to simply tell participants about the stereotype that refers to the situation, e.g., that men usually perform better on a math test than women (Spencer, Steele & Quinn, 1997). The most subtle way of activation would be to present stereotype information below the person’s conscious awareness threshold (Banaji & Hardin, 1996, Bargh & Chartrand, 1999). This can be done for example by presenting words on a computer screen for a very short time so that the participants will notice a flash but cannot tell which word just appeared (Bargh & Pietromonaco, 1982). Shih et al. (2002) suggest that people have a different sensitivity regarding stereotype information depending on whether the self is implicated or not, and that this sensitivity moderates their behavioral response. Specifically, they propose that targets should be more attentive and sensitive to self-relevant information, i.e., in this case, the stereotype. That self-relevant information is actually recognized faster, remembered better and processed in a favored fashion is an accepted phenomenon (Cherry, 1953; Wegener & Bargh, 1998).