When the method makes a difference : antagonistic effects on 'automatic evaluations' as a function of task characteristics of the measure

"Researchers often employ implicit measures as dependent variables to investigate processes of attitude formation and change. In such studies, experimentally induced differences are typically interpreted as reflecting change in automatic evaluations. We argue that experimentally induced effects on implicit measures may not always reflect genuine changes in evaluative responses, but can be driven by the mechanisms underlying the measurement procedure. In line with this assumption, the present research shows that these mechanisms can produce opposite effects of the same experimental manipulation for otherwise equivalent implicit measures. These results indicate that merely observing experimental effects on implicit measures does not allow direct inferences regarding changes in automatic evaluations. Instead, psychological interpretations of such effects hinge upon the mechanics of how a given measurement procedure responds to variations in the context. Implications for research using implicit measures are discussed." [author's abstract