It has recently been stated (Greenwald, McGhee, Schwartz 1998, Nosek, Greenwald, & Banaji 2005) that the Implicit Association Test (IAT) measures automatic cognition. One of the advantages of measuring automatic cognitions is that they are less susceptible to faking than explicit, self-report tests. The present study investigated whether and to what degree an IAT - Anxiety can be faked. We found that the IAT - Anxiety is susceptible to faking, but to a limited degree, certainly less than self-reports. Certain conditions seem to facilitate faking: (a) detailed instruction regarding the IAT assessment mechanism, and (b) subjects' level of anxiety. However, whereas some participants faked their IAT scores in the required direction, they could not completely hide their base rate performance, as shown by the significant correlations of IAT effects at base rate and faking. So, dissimulating anxiety on the IAT is not a completely conscious and controlled process. The implications of this study are discussed.
KEYWORDS: Implicit Association Test (IAT), faking, anxiety