Evaluative priming as an implicit measure of evaluation: An examination of outlier-treatments for evaluative priming scores

Abstract

Evaluative priming is based on the notion that evaluative classifications of target stimuli are faster (vs. slower) when they are preceded by a prime stimulus of the same (vs. opposite) valence. Although evaluative priming is widely used as an implicit measure of evaluation, there is no common procedure for the treatment of response-latency outliers. To address this limitation, four studies examined common outlier-treatments in terms of (1) the overall size of evaluative priming effects, (2) their internal consistency, and (3) their relation to corresponding explicit measures in the domains of conditioned attitudes (Study 1), political attitudes (Study 2), racial attitudes (Study 3), and ethnic attitudes (Study 4). The algorithm with the best performance uses a priori cutoffs of 300 ms and 1000 ms, treating response times beyond these cutoffs as missing values. Internal consistency was low for all algorithms, indicating limits in the usefulness of evaluative priming for correlational studies.