Reducing bias in judgment and decision making; improving intergroup and interpersonal negotiation

This line of research examines how people’s efforts to maintain identity increase closed-mindedness, bias in social judgment, and inflexibility in negotiation. We also have been successful at developing theory-informed interventions to increase openness to threatening ideas and information (e.g., in political and health contexts) and to reduce inflexibility in negotiation.

Resistance to change. We have developed a theoretical framework positing that long-held beliefs often serve as bases of identity. Beliefs relevant to socially divisive and value-laden issues, we posit, arise less from a dispassionate assessment of facts and arguments than from their social meaning—the message they convey about one’s moral identity (see Cohen, 2003).

In one series of studies, liberals’ and conservatives’ evaluations of a new social policy depended more on the stated position of peers who shared their political allegiances than on the objective content of the policy in question or even their own prior ideological beliefs.

Beliefs linked to an important identity persist in the face of strong logical and empirical challenges. Our studies show, however, that interventions that decrease the personal significance of the identity challenged by a communication make people less biased and more open to counter-attitudinal information. In one line of studies, partisans of various social issues (such as capital punishment and abortion) displayed the standard information assimilation biases (see Cohen, Aronson, & Steele, 2000; Cohen et al., in press; Sherman & Cohen, 2002; Sherman & Cohen, in press). They accepted belief-congruent information uncritically, resisted belief-incongruent evidence, and became more polarized in their views when exposed to mixed evidence. By contrast, partisans who had received an affirmation of an alternative source of self-integrity (e.g., who reflected on an important personal value unrelated to their political beliefs) proved relatively open to identity-threatening information and relatively unbiased in their assessment of new evidence. Indeed, in one study, such affirmation eliminated the difference between patriotic and non-patriotic Americans in their evaluations of a report critical of U.S. terrorism policy (see Cohen et al., in press).

We have documented similar examples of “de-biasing” through self-affirmation in a number of different contexts, including resistance to threatening health-risk information, defensive reactions to failure, and unwillingness to compromise in negotiation (see Cohen et al., in press; Sherman & Cohen, 2002). We have recently completed a review of this research ( see Sherman & Cohen, in press), and are currently developing a general theoretical account that links resistance to persuasion and intransigence in negotiation to concerns of identity maintenance (see Cohen et al., in press). One distinctive feature of our research is that it goes beyond documenting biases to test theory-driven de-biasing interventions. In another study, we found that affirmation paired with a situational cue highlighting participants’ partisan identity increased abortion advocates’ willingness to compromise on abortion legislation (see Cohen et al., in press).

Prejudice and discrimination. In research on a phenomenon we have termed constructed criteria, we have identified a particularly subtle form of discrimination—one that enables decision-makers to engage in gender and racial bias while maintaining a personal identity as “objective” and “fair.” Principles of meritocracy assert that the criteria used to assess merit should be constant across individuals and groups. However, we find that people spontaneously construct (and re-construct) criteria of merit to the advantage of members of socially privileged groups.

In one study, for instance, participants constructed criteria of merit for the job of “police chief” congenial to the idiosyncratic credentials of male applicants (see Uhlmann & Cohen, 2005). Regardless of whether a male applicant had “book smarts” or “street smarts,” or had children versus not, whatever qualities that he happened to have were viewed as relatively important to the job in question (whereas the qualities he lacked were viewed as relatively unimportant). By contrast, the qualities of female applicants had no impact on their perceived importance. When participants subsequently hired the male candidate over the female one, they could thus entertain the illusion that they had “picked the right man for the job,” when in reality they had “picked the right criteria for the man.” The phenomenon of constructed criteria plays an important role in discrimination and points to an intervention strategy: We found that having decision-makers commit to criteria prior to reviewing job candidates and learning their gender eliminated discrimination in hiring choices.

Through the mechanism of constructed criteria, people can rest assured of the objectivity of their decisions, having failed to realize that their definition of merit has been determined, in large part, by a candidate’s group membership. Indeed, in our study, the people who displayed the most anti-female bias also rated their decisions as the most objective and free of bias. Current research tests whether the causal arrow also runs in the opposite direction—whether feeling objective can, in turn, lead to greater sexism. Our recent findings suggest that people who feel objective have an “I think it, therefore it’s true” mentality (Uhlmann & Cohen, 2007). They assume that any sexist thoughts and beliefs they hold are, by virtue of being theirs, valid and therefore worthy of being acted upon. Indeed, we found that increasing men’s self-perceived objectivity via an experimental manipulation led them to discriminate against women more, and made them more influenced by nonconsciously primed gender stereotypes.

Other Problems our Research is Addressing. Further examples of current and future directions include (a) zeroing in on the mechanisms mediating the effects of self-affirmation documented in our previous research; (b) using our theoretical framework to understand the contagion of anti-social and health-risk behavior among adolescent youth (in the context of work undertaken with a colleague in clinical psychology; see Cohen & Prinstein, 2006 [link]); and (c) examining how assessments of societal risks (e.g., about industrial pollution and private gun ownership) arise from people’s cultural identities as individualists versus communitarians (in the context of work undertaken with a colleague in the law school). Additionally, we have (d) begun to investigate defensive denial of social inequality. Some initial research suggests that people who base their social identity on a cherished social value (such as equality) may, under some circumstances, assume that because a desirable societal condition should be true, that it is true . As an example, we have found that priming Americans on long-held national values (such as equality and colorblindness) subsequently blinds them to real violations of those values (Hahn & Cohen, 2007; Cohen, Scrivener, & Steele, 2007).

We are also looking at how negative stereotypes benefit the non-stereotyped, such as white men in science and math. We find that non-stereotyped individuals show a performance boost in the presence of negatively stereotyped individuals. We call this phenomenon stereotype lift (see Walton & Cohen, 2003).