A quantitative and individual measure of stereotyping is proposed, based on denning stereotypes as probabilistic predictions that distinguish the stereotyped group from others. Data indicate that the proposed measure, though related to the familiar Katz and Braly checklist, is a substantially new measure of stereotyping rather than simply a quantitative version of the checklist. It is argued that the theoretical and empirical value of the proposed measure is justification for abandoning the Katz and Braly measure. Theoretically, the new measure relates stereotype research to attribution theory as part of a Bayesian approach to the psychology of prediction. Empirically, the new measure opens interesting questions about stereotypes, especially about stereotype validity and the "kernel of truth" hypothesis. A study using the new measure reveals that diverse groups of subjects have some similar stereotypes of black Americans, that these stereotypes are relatively accurate, and that contrary to the "kernel of truth" hypothesis, these stereotypes are seldom exaggerated.Empirical study of stereotypes began with Katz and Braly (1933) and goes on today with the same measure they first used with Princeton students. Subjects are given a list of traits and asked to check the five traits "most typical" of a particular group. A stereotype is said to exist to the extent subjects agree on the choice of traits. This stereotype is properly a social stereotype because it is a group measure, defined by agreement across subjects.The authors wish to acknowledge the participation of Gerry Ritrovato and Terryll Schasse in carrying out the study of stereotypes of black Americans (Study 3 in this article).

A brief review of stereotype research indicates that little has been discovered about stereotypes that are defined as bad generalizations and measured with the Katz and Braly checklist. McCauley and Stitt have proposed that stereotypes can be better understood as probabilistic predictions that distinguish one group from another. The practicality of the proposed diagnostic-ratio measure of stereotyping is argued, and the relation of the new measure to the Katz and Braly measure is discussed. Lippmann's arguments against stereotypes-that they are illogical in origin, resistant to new information, and obviously invalid -are shown to be inapplicable to probabilistic stereotypes. Other common arguments against stereotypes-their ethnocentrism, genetic implications, projected hostility, and exaggeration of real group differences-are shown to be arguments against particular stereotypes rather than arguments against stereotyping. Recent research indicating various cognitive biases in the formation and use of stereotypes is reviewed, and it is suggested that what is wrong with stereotyping is no more and no less than what is wrong with human conceptual behavior generally.

No abstract

A series of experiments related inhibition and facilitation of the startle response, elicited by an intense auditory signal, to a change of the frequency characteristic of a 70-db. continuous acoustic signal. The data indicated that if a frequency change occurs in the acoustic environment 64 msec, before the startle-eliciting stimulus, the amplitude of the startle response is reduced; and if frequency change occurs 4 msec, prior to the startle-eliciting stimulus, the response latency is reduced. These results extend the generality of previous research employing weak antecedent acoustic signal onset and offset. The results indicate that neural mechanisms mediating the startle reflex may be activated by any change in the acoustic environment and that these mechanisms may be a component of the orienting reflex arc.

No abstract