Our goal was to initiate a series of explorations of the techniques used to investigate person perception and stereotypes. More specifically, do different techniques uncover the same, and assumedly correct, underlying cognitive structure of the perceivers or is our current understanding of social cognition merely a reflection of our data collection and analytic techniques? We also hoped to draw some conclusions concerning White participants' perceptions of Blacks. Participants were given two tasks. One task involved sorting fourteen racial type labels and rating the sorted piles on four scales (e.g., respectable-not respectable). The second task involved generating attributes that described a randomly selected racial type. Both tasks provided data that could be used as input for multidimensional-scaling and hierarchical -clustering analyses. Further, the data from the adjective-generation task was used as input for a discriminant-function analysis. It was predicted that the different data collection and analysis tasks would produce results that emphasized the importance of evaluation in racialtype perception but that the sorting task would reveal more prejudice against Black targets. The results supported the hypotheses.
The extant literature on person perception and stereotyping is immense and impressive (Hilton & Von Hippel, 1996) and there is a wide consensus on a number of important issues. For example, many would agree that stereotypes are pragmatic in that they help guide perceivers through their daily social interactions (Fiske, 1993). An implicit acceptance of this belief has led social cognition researchers to move from an emphasis on the global social-stereotype categories toward investigations of the seemingly more pragmatic subcategories. For example, research indicates that perceivers utilize subcategories when thinking about gender (e.g., Green & Ashmore, 1998), the elderly (Brewer, Dull, & Lui, 1981), Latinos (Huddy & Virtanen, 1995), and Blacks (Devine & Baker, 1991). A wide range of techniques and statistical analyses has been used to investigate these subcategories and little discussion has arisen concerning what, if any, techniques and analyses are most useful. That is, can different methods of investigating person perception lead to different results and interpretations? One noteworthy exception involves the investigation by Devine and Baker of racial subgroups, which included an inquiry into the utility of discriminant function analysis for understanding stereotype organization. We decided to follow their precedent and used the study of racial-subgroup organization as a testing ground for two different methodologies. More specifically, the structure of racial subgroups was investigated using discriminant-function analysis, more conventional multidimensional-scaling and hierarchical-clustering analyses. The results of these techniques were compared and contrasted.
A REVIEW OF DEVINE AND BAKER (1991)
Devine and Baker (1991) examined White' perceivers' perceptions of Blacks to empirically document whether Whites understand and use well-differentiated subgroups of Blacks. They gave participants one of nine labels (Black, Black athlete, businessman Black, ghetto Black, militant Black, Oreo cookie, streetwise Black, Uncle Tom, and welfare Black) and asked them to generate a list of characteristics that best captured the cultural conception of the group. Using discriminant-function analyses, they found two distinct subgroups under the general heading "Blacks" (i.e., "athlete" and "businessman"), but no other distinctions were found among the remaining subgroups. A caveat offered by the authors is that their subject sample (White Midwestern college students) had no real concept of some of the subgroups (e.g., "Oreo cookie," "militant Black").
The research of Devine and Baker (1991) was valuable for two reasons. First, in terms of social cognition, it provided evidence that perceivers can differentiate between the superordinate category "Black" and some subgroup labels. …