Beyond an associative conception of automatic self-evaluations: applying the relational responding task to measure self-esteem

According to dual-process models, implicit self-esteem (SE) is based on automatic self-associations that can be measured with indirect techniques based on an associative conception of implicit cognition (e.g., Implicit Association Test; IAT). However, alternative theoretical proposals (e.g., relational frame theory; RFT) propose that implicit SE might not be based on automatic self-associations, but on implicit propositional self-evaluations that can be captured only with nonassociative implicit measures (e.g., Implicit Relational Assessment Procedure; IRAP). In the present study, both reliability and validity of a new propositional measure of implicit self-esteem (relational responding task; RRT) were assessed, and compared with the SE-IAT and with two self-report scales of self-esteem. In the first study, two alternative self-esteem RRTs (SE-RRT and RSE-RRT) were administered along with a SE-IAT and other scales, to assess reliability and validity issues. The results showed: 1) acceptable, though not optimal, reliability for both RRTs; 2) an adequate support for convergent validity, with significant correlations between implicit and explicit measures of SE; 3) the criterion validity was supported for the RSE-RRT (with significant correlations with all theoretically linked scales), although only partially supported for the SE-RRT (with a significant correlation only with depression; 4) RRTs were not significantly correlated with impression management and self-deception; and 5) incremental validity of implicit propositional SE on depression, controlling for automatic SE associations and explicit self-esteem. In a second study, it was experimentally demonstrated that SE-RRT showed levels of “fakeability” similar to a classical implicit self-esteem measure like the SE-IAT, and considerably lower than SE scales.