Distinguishing stereotype threat from priming effects: On the role of the social self and threat-based concerns.

Abstract

It has been argued that priming negative stereotypic traits is sufficient to cause stereotype threat. The present research challenges this assumption by highlighting the role of the social self and targets' concerns about confirming a negative group-based stereotype. Specifically, in 3 experiments the authors demonstrate that stereotype threat adversely affects the test performance and threat-based concerns of targets (but not nontargets) because only targets' social self is linked to the negative group stereotype. Trait priming, however, harms the test performance of both targets and nontargets but has no effect on their threat-based concerns because trait priming does not require such a link between the social self and the group stereotype. Moreover, the authors show that merely increasing the accessibility of the social self in nonthreatening situations leads to the underperformance of targets but has no meaningful effect on nontargets' test performance.

Understanding Stereotype Lift: On the Role of the Social Self

“…Our results further demonstrated that as the social self manipulation became more direct participants saw greater overlap between themselves and their gender group. The present findings also complement our previous work examining activation of the social self in stereotype threat testing situations (see Marx & Stapel, 2006;Marx et al, 2005).…”

“…We suggest that stereotype lift is more likely when the social self is activated in a diagnostic testing situation, because when the social self is activated the associated stereotypes are also likely to be activated; hence, non-targets should perform better due to the positive stereotype linked to their group. Support for this notion can be found in our past work on the preconditions for stereotype threat (Marx & Stapel, 2006;Marx, Stapel, & Muller, 2005). This research showed that activation of the social self led to poor test performance for stereotyped targets because their social self is linked to the negative stereotype about their group's ability.…”

“…For female participants, however, the manner in which the social self was activated (be it indirect or direct) had much less of an impact. That is, female participants performed worse in both social self conditions (i.e., We Prime and Gender) and Diagnostic condition relative to the Control condition: "extra" activation of the social self was not necessary because the links between the test description, targets' social self, and the negative group stereotype were already strong enough to harm their test performance (Major & O'Brien, 2005;Marx et al, 1999;Marx & Stapel, 2006). Considerable research has investigated the adverse effects of negative stereotypes on stereotyped targets' test performance, but far less research has been conducted on the enhanced performance of non-targets in those same testing situations.…”

“…For this task, participants were told that as part of a proofreading and word search exercise they would read a short paragraph detailing a trip to the city (see Brewer & Gardner, 1996;Marx & Stapel, 2006;Stapel & Koomen, 2001). As a way to manipulate participants' mindsets we asked those in the I Prime condition to circle all the personal pronouns (e.g., I, my, mine, me), those in the We Prime condition to circle all the group-based pronouns (e.g., we, our, ourselves, us), and those in the remaining conditions to circle the letters "x" "y," or "z," while they read the story.…”

“…As a way to manipulate participants' mindsets we asked those in the I Prime condition to circle all the personal pronouns (e.g., I, my, mine, me), those in the We Prime condition to circle all the group-based pronouns (e.g., we, our, ourselves, us), and those in the remaining conditions to circle the letters "x" "y," or "z," while they read the story. Of course for participants in the We Prime condition this "general" manipulation of the social self ("we-ness") may make a variety of social identities accessible, but once the context and situational cues are factored in (I am a person about to take a math test) then the social identity (gender) that is most relevant to the situation (taking a math test) may "win out" in the end (e.g., Davies, Spencer, & Steele, 2005;Major & O'Brien, 2005;Marx & Stapel, 2006, Onorato & Turner, 2004Turner, Hogg, Oakes, Reicher, & Wetherell, 1987). If participants' social identity is linked to a stereotype (be it positive or negative), then it seems reasonable to suggest that their test performance should correspond to that stereotype (e.g., Marx et al, 1999).…”