Chapter 11.  Accuracy: Criteria

Abstract

            This chapter reviews the theoretical and empirical bases for identifying the criteria appropriate for assessing the accuracy of social beliefs, interpersonal expectancies, and social stereotypes.  Many famous social psychologists have concluded that identifying appropriate criteria for assessing accuracy is so difficult that, in fact, it cannot or should not be done.  This chapter contests such pessimistic conclusions.  First, it reviews broad theoretical perspectives on accuracy (probabilistic realism, functional perspectives, and social constructivism), and it concludes that both functional and constructivist perspectives are incoherent with respect to accuracy, whereas probabilistic realism is based on the same principles that scientists (including social scientists) use to reach conclusions about the validity of their theories and hypotheses.  In short, establishing the accuracy of lay social beliefs is much like the scientific enterprise of establishing construct validity in the social sciences.  Specific contenders for criteria, such as objective measures, agreement among perceivers, target behavior, expert judgments, self-reports and self-perceptions, are critically evaluated in order to highlight their strengths and weaknesses.  The chapter also highlights several concrete examples whereby social scientists are themselves logically incoherent and hypocritical when they reach conclusions about sociological and psychological phenomena and also suggest that there are no good criteria for assessing accuracy.

EXCERPT:
    A similar contradiction occurs in expert models regarding use of stereotypes.  Nearly all social psychological theories and perspectives, and much of current cultural mythology, states or implicitly assumes that people are acting rationally and appropriately only when they judge others solely and entirely on the basis of those others’ personal characteristics, rather than their group memberships.  This view argues or implies that people act in a biased, prejudiced, or irrational manner when they allow their stereotypes to influence their judgments of individuals (e.g., Brewer, 1988; Darley & Fazio, 1980; Fiske, 1998; Fiske & Neuberg, 1991; Jones, 1986, 1990; Myers, 1998).
According to widely accepted principles of both statistics (e.g., Bayes’ theorem, see, e.g., Kahneman & Tversky, 1973) and philosophy of science (e.g., Krueger & Funder, 2004; Meehl, 1990), however, base-rates, prior beliefs and expectations should influence our interpretations of new evidence (see McCauley, et al., 1980 for an analysis focusing specifically on stereotypes).  An unlikely claim (that object I saw in the sky was an alien spacecraft) requires a much higher standard of evidence than does a likely claim (that object I saw in the sky was a cloud).  Similarly, if I am not 100% sure what that object was, I am far more likely to be correct if my generally accurate expectation colors my interpretation than if it does not (i.e., I am much more likely to be right if I figure it was probably a cloud, than if I figure it was an alien spacecraft).
Similarly, these principles also suggest that if I am not sure how tall a person is, I will, on average, be right more often if I estimate any particular male to be a few inches taller than any particular female, than if I estimate them to be exactly equal (i.e., use, rather than discard, my sex stereotype regarding height).  They also suggest that if I do not know someone’s social class background, I will, on average, be right more often if I estimate that any particular Jewish American is wealthier and more highly educated than any particular Native American.  And even if I am highly unlikely to die in a fatal car accident under any circumstances, I am, on average, far more likely to do so if I enter a car with a male under 25 doing the driving than if I enter a car with a middle aged female doing the driving.
Regardless, it cannot be better to both use and ignore prior expectations.  Thus, the expert models that have emerged from the stereotype literature conflict with those that have emerged from the cognitive judgment and decision-making literature.   ( I return to this issue in Chapter 18 when I address whether using or ignoring stereotypes increases or reduces accuracy in person perception). For now I simply point out that the expert models developed in the judgment and decision making literature emphasizing the value of relying on base-rates when situations are ambiguous conflict with those developed in the stereotype literature emphasizing the supposedly biased and irrational nature of ever relying on a stereotype to judge an individual.