Measuring automatic associations: Validation of algorithms for the Implicit Association Test (IAT) in a laboratory setting

Background and objectives: In their paper, "Understanding and using the Implicit Association Test: I. An improved scoring algorithm", Greenwald, Nosek, and Banaji (2003) investigated different ways to calculate the IAT-effect. However, up to now, it remained unclear whether these findings based on internet data also generalize to laboratory settings. Therefore, the main goal of the present study was to cross-validate scoring algorithms for the IAT in a laboratory setting, specifically in the domain of psychopathology.

Methods: Four known IAT algorithms and seven alternative IAT algorithms were evaluated on several performance criteria in the large-scale laboratory sample of the Netherlands Study of Depression and Anxiety (N = 2981) in which two IATs were included to obtain measurements of automatic self-anxious and automatic self-depressed associations.

Results and conclusions: Results clearly demonstrated that the D-2SD-measure and the D-600-measure as well as an alternative algorithm based on the correct trials only (D-noEP-measure) are suitable to be used in a laboratory setting for IATs with a fixed order of category combinations. It remains important to further replicate these findings, especially in studies that include outcome measures of more spontaneous kinds of behaviors.