Applying the latent state-trait analysis to decompose state, trait, and error components of the self-esteem implicit association test
In the literature, self-report scales of Self-Esteem (SE) often showed a higher test-retest correlation and a lower situational variability compared to implicit measures. Moreover, several studies showed a close to zero implicit-explicit correlation. Applying a latent state-trait (LST) model on a sample of 95 participants (80 females, mean age: 22.49 ± 6.77 years) assessed at five measurement occasions, the present study aims at decomposing latent trait, latent state residual, and measurement error of the SE Implicit Association Test (SE-IAT). Moreover, in order to compare implicit and explicit variance components, a multi-construct LST was analyzed across two occasions, including both the SE-IAT and the Rosenberg Self-Esteem Scale (RSES). Results revealed that: (1) the amounts of state and trait variance in the SE-IAT were rather similar; (2) explicit SE showed a higher consistency, a lower occasion-specificity, and a lower proportion of error variance than SE-IAT; (3) latent traits of explicit and implicit SE showed a positive and significant correlation of moderate size. Theoretical implications for the implicit measurement of self-esteem were discussed.