Measurement of subtyping in stereotype change

Researchers in the area of stereotype change have accumulated strong eviden ce for the subtyping phenomenon, whereby disconfirming members of a group a re "fenced-off' or excluded from group perceptions. A variety of measures h ave been used in past work as indicators of subtyping with no consensus on what is the best measure. This study compared a large number of potential m easures of subtyping by first manipulating through study instructions wheth er participants subtyped and then examining what measures covaried with the resulting group perceptions in the subtyping condition. Three measures sho wed statistically significant relationships with group perceptions. These i ncluded perceived typicality of disconfirming relative to confirming group members, memory confusions among confirming and disconfirming individuals, and rated similarity of confirming and disconfirming instances. Conceptuall y, the latter two measures are somewhat preferable to ratings of typicality because they are less invasive. Importantly atypicality did not appear to mediate stereotype change, only to covary with it. (C) 2001 Academic Press.