Steven J. Stroessner
Barnard College, Columbia University
Jason E. Plaks
Illusory Correlation and Stereotype Formation: Tracing the Arc of Research
Over a Quarter Century
Research on stereotype formation has developed substantially since the advent of the social cognition perspective nearly a quarter century ago. To appreciate the degree of change, one can compare how two authoritative reviews of the literature—one authored by Brigham (1971) and a more recent review by Hamilton and Sherman (1994)—differ in their discussion of the formation of stereotypes. Brigham bemoaned the lack of research on the processes underlying stereotype formation, noting that existing evidence indicated only that White children from an early age associated negative attributes with African Americans. The specific origin of these stereotypes was unclear, but Brigham raised the possibility that stereotypes reflect an overgeneralized “kernel of truth” that might be conveyed to children through societal or social learning variables. Given the prevailing view that stereotypes reflected the exaggeration of small group differences, he briefly speculated about how parents, peers, teachers, and the media help perpetuate stereotypes that have some basis in reality.
By Hamilton and Sherman's review, a sizeable literature had amassed that recognized how basic cognitive processes, in and of themselves, contributed to the formation and maintenance of stereotypes. Although earlier theorists (e.g., Allport, 1954; Lippmann, 1992) alluded to the possibility that stereotypes might arise from the manner in which people process information about their social environment, a body of research supporting that view did not emerge until the advent of the “cognitive revolution” in social psychology. By utilizing methods largely developed by cognitive psychologists, social psychologists were able to identify how basic cognitive processes could produce perceptions of groups that were factually unwarranted. The mere categorization of social targets as members of groups, for example, led to the perception of group differences (e.g., Tajfel, 1969) and to the perception of homogeneity within groups (Rothbart, Davis-Stitt, & Hill, 1997; Wilder, 1978). Biases in attention were also implicated, such as when a contextually unusual or salient individual received inordinate notice, yielding exaggerated trait judgments and evaluations (e.g., Lord & Saenz, 1985; Taylor, Fiske, Etcoff, & Ruderman, 1978) that might be generalized to their group.
In contrast to earlier approaches that viewed stereotypes as exaggerations of real group differences, the social cognition perspective suggested that the ordinary cognitive processes involved in social perception were sufficient for the formation of stereotypes. Therefore, many stereotypes were likely to have no basis in fact. Stereotypes might form simply because of characteristics of individuals' information processing systems and not because stereotypes reflected an overgeneralized “kernel of truth.”
The line of research best reflecting this view of stereotype formation focused on the formation of distinctiveness-based illusory correlations, beginning with a seminal article published by Hamilton and Gifford in 1976. This research was actually inspired by studies originally performed by Chapman (1967) on learning. Chapman found that experimental participants consistently overestimated the frequency of the co-occurrence of the two longest words presented in a series of word pairs. He suggested that the co-occurrence of the two stimuli might have been noticed because both words shared the unusual feature of length. If these word pairings did differentially attract attention, they would have been differentially accessible in memory, leading to the erroneous conclusion that the infrequent pair of stimuli appeared more often than they actually did.
Hamilton and Gifford (1976) reasoned that similar processes might occur when individuals encounter information describing social groups. All cultures contain groups that vary in size, and members of infrequently encountered groups might be particularly likely to draw perceivers' attention. In addition, the frequency of desirable and undesirable behavior is also likely to vary, with negative behavior typically rarer than positive