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. (C) 2012 Elsevier Ltd. All rights reserved.