Behavioral Consistency And Individual Differences In Predictive Structure

Psychological research has frequently been directed at evaluating the predictive utility of personality assessments with respect to various behavioural criteria. Bem and Allen (1974) have advocated an approach to maximizing the convergences of various modes of personality assessment which is based on the identification of behaviourally consistent and predictable individuals along specific trait dimensions. The measured behavioural consistency for individuals was used as an independent moderator variable, influencing the correlations between predictor and criterion measures of personality. Individuals who identified themselves as being consistent for a particular behavioural domain were shown by Bem and Allen to be more predictable than less consistent subjects with regard to peer ratings and other trait measures.;Subsequent research by Kenrick and Stringfield (1980) and others (e.g., Kenrick & Braver, 1982) has provided some support for the idiographic formulations of Bem and Allen. Procedural differences among empirical studies, however, have produced generally unrecognized problems of interpretation related to the outcomes of hypothesis tests. These problems are described and evaluated in the present investigation.;The results of a peer rating study of personality, and an assimilation of data from other sources, suggest that (a) behavioural consistency is related to trait level or need strength such that the greatest degrees of consistency occur at the extremes of bipolar dimensions of behaviour, and that (b) the failure to consider this relationship in classifying subjects as predictable (consistent) and unpredictable (variable) can lead to the discovery of moderator effects that are spurious. The consistent subgroups being predominately the most extreme on personality dimensions will have inflated trait score variances that increase the likelihood of finding significant correlational indices of behavioural predictability. Accounting for the relationship between consistency scores and personality scores in this study resulted in little evidence for the pervasive belief that individual differences in trait consistency effectively moderate the validity of behavioural assessments.;Additional analyses presented and discussed are concerned with other issues relevant to the validity of personality assessments. These issues include rater confidence as a moderator of predictive validity, reliability and validity as a function of the aggregation of replicated measurements, and the stability of the measured organization of personality traits