Do Indirect Measures Predict effect-Specific Automatic Behavior —Moderated by Cognitive Control Resources? An Investigation Into the Domains of Disgust, Speech Anxiety, and Liking vs. Disliking of a Minority Group

With three studies, we scrutinized the domain specificity of indirect measures in predicting behavior. We compared indirect and direct measures of disgust (Study 1, N = 121), speech anxiety (Study 2, N = 111), and evaluation of a minority group (Study 3, N = 97). In each study, data was collected at three measurement occasions (t1, t2, t3). At t1, direct measures of explicit disgust, anxiety, and attitude (liking vs. disliking) were taken. Direct measures were each implemented by a questionnaire and a picture rating. At t2, indirect counterparts using variants of the implicit association task (IAT) were employed. Automatic and controlled behavior was recorded at t3. In Study 1 (disgust), both indirect and direct measures of disgust significantly predicted behavioral indices of approaching a cockroaches, but the effect of the direct measure was larger. By contrast, both measures did not significantly predict coded facial and bodily reactions while approaching cockroaches. In Study 2 (anxiety), the indirect measure of speech anxiety significantly predicted nervous facial and bodily reactions during a speech, while the direct measure significantly predicted the choice of audience size. In Study 3 (liking vs. disliking of a minority group), the indirect measure of attitudes toward minority group members significantly predicted facial and bodily approach reactions toward a visibly minoritized confederate compared to a visibly non-minoritized confederate, whereas a picture rating as the direct measure did not significantly predict rated sympathy and potential collaboration with the comparison of the confederates. However, the questionnaire as direct measure did significantly predict the sympathy and potential collaboration rating comparison of the confederates. In addition, we tested for moderating effects of acute and chronic cognitive control resources (dispositional self-control, working memory capacity) assessed at t1, and experimentally manipulated at t3. No consistent moderation effects emerged (analyses are reported in the supplement). The findings of our studies support the specificity of the evaluations and the discriminant validity of the direct and indirect measures, and supplement previous research on indirect measures, which has either focused on single constructs and ignored mismatching evaluations, or used self-reports as behavior proxies.