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.