Improving the Detection of Personality-Behavior Relationships in Consumer Research

Past suggestions for improving detection of personality-consumer behavior rela-tionships concerned the personality side of the relationship. In contrast, after Epstein (1979,1980), this research calls for an examination of the reliability of the dependent variable, the behavior measure. Using data from two large-scale lifestyle studies, this research empirically demonstrates how increasing test-retest reliabilities of re-peated behavior measures results in enhanced detection of personality-behavior relations. Theoretical and practical implications are discussed. A classic area of consumer research is the prediction of product use or choice with personality or life-style traits. Based on casual observation, many mar-keting managers believe that consumers buy and use products congruent with their personalities. Aside from some exceptions (e.g., Calantone and Sawyer 1978), re-view articles have generally concluded that strong per-sonality-behavior relationships have not been empiri-cally detected (Jacoby 1971; Kassarjian 1971, 1979; Kassarjian and Sheffet 1975; Wells and Beard 1973). A meta-analysis of more recent consumer personality re-search also suggests only weak empirical relationships. 1 Given this conflicting state of affairs, one possibility is that marketing managers are correct and that consumer researchers empirical methods are not giving true re-sults. This is not to say that recent personality-oriented consumer research is methodologically flawed. Kassar-jian (1971) concluded that much early empirical per-sonality work was weak because: (1) the trait measures had questionable validity and reliability, and/or (2) the traits investigated had little conceptual relevance to the behavior studied. Since the early 1970s, personality re-search has improved in these two key areas. First, recent work pays more attention to trait measure quality; for example, studies such as those reported by Brooker (1978) or Richins (1983) examine trait measures. Sec-ond, traits studied in contemporary research are typi-cally theoretically justified; for examples of conceptually