Implicit Association Test or the measure of implicit social cognition: a critical review of the validity and the theoretical basement of its scores

In order to remedy the limits of self-report measures (self-presentational biases, introspective limits...), several indirect measures have been developed. The Implicit Association Test (IAT, Greenwald, McGhee et Schwartz, 1998) is the one that has concentrated most interest and research: it is flexible enough to measure a broad range of constructs (attitude, personality, stereotypes...), and it produces reliable and encouraging criteria valid scores. But the construct validity of the IAT scores remains controversial because their interpretation is quite ambiguous. In the same vein, because of the lack of any strong theoretical background, the interpretation of the observed dissociations between direct (self-report) and indirect measures remains problematic. This leads to multiple discussions. These "youthful limits" however stimulate researchers' creativity and open new perspectives which should lead to more theoretically grounded measures.