Dominance and friendliness as implicit constructs: An SEM analysis of IAT measures

Recent work has indicated that traits, such as attitudes and self-perceptions, exist at two mainly uncorrelated levels. Explicit traits are thought to be relatively conscious and are measured using self-report measures, while implicit traits are thought to be relatively unconscious and must be measured more indirectly. Often, when studying the interpersonal traits of dominance and friendliness, researchers tend to focus on self-report (or explicit) measures. However, much of interpersonal behavior takes place outside of conscious awareness (Carson, 1969; Kiesler, 1996). The current studies investigate the constructs of implicit dominance and friendliness, which we believe should have promising explanatory power when used in conjunction with explicit measures. Accordingly, we designed implicit measures of dominance and friendliness using the Implicit Association Test (IAT; Greenwald et al., 1998). In three studies, we collected data on both implicit and explicit dominance and friendliness, using the IATs and self-report measures. We then used these data to test a structural equation model evaluating the psychometric properties of our newly devised IATs. Contrary to the widespread assumption among IAT researchers, our model suggests that each IAT may be tapping two independent underlying constructs, rather than one bipolar construct. Accordingly, we advance a novel approach to scoring the IAT, in which each IAT measures two substantive constructs. Using this new method of scoring, we examined the reliability of our IAT measures, as well as their discriminant validity with regard to explicit measures of the same constructs. The findings of the current research have important implications for interpersonal theory, which are discussed.