Assessing the Assumptions Behind the IAT
Posted: 31 Mar 2008
Date Written: March 2008
When researchers use the Implicit Association Test (IAT), they tacitly assume that attitudes towards the two target categories are strongly negatively correlated. We show that this assumption is likely to be violated in marketing contexts. As a consequence, IAT-scores are potentially invalid measures of implicit consumer attitudes. We demonstrate this by simulating possible correlation matrices of the underlying implicit attitudes and the IAT measurements, and then calculating the resulting IAT validity. We find that when the tacit assumption is satisfied the IAT validity is reasonably high. However, when the tacit assumption is violated (i.e., attitudes towards the two target categories are positively correlated), IAT validities are so dispersed that it is practically impossible to meaningfully interpret the IAT. We conclude by giving guidelines for how researchers may try to avoid violations of the assumption of the IAT.
Keywords: Implicit Association Test, measurement validity, simulation
Suggested Citation: Suggested Citation