Implicit Attitudes and Racism: Effects of Word Familiarity and Frequency on the Implicit Association Test

Abstract

Greenwald, McGhee and Schwartz (1998) described a new method-the Implicit Association Test (IAT)-for unobtrusively measuring racial attitudes. This article assesses the validity of the IAT by investigating whether Greenwald et al.'s implicit racism findings resulted from two confounds present in their studies: differential familiarity and frequency of the words that comprised their target concepts. Experiment 1 produced large IAT effects when both low and high familiarity words comprised nonsocial target categories (insects and flowers) and demonstrated that the IAT is more sensitive when high familiarity exemplars form the target concepts. In Experiment 2, we obtained large implicit racism effects for both African American and Hispanic racial groups even when the familiarity and frequency of the names that comprised the racial categories were controlled and even though participants described themselves as unprejudiced. Additionally, explicit self-reports of racial attitudes were only weakly related to the IAT measures. These experiments indicate that (a) although familiarity clearly exerts an important influence on the IAT, the use of low familiarity stimuli does not eliminate the sensitivity of the IAT, (b) stimulus familiarity and frequency can not account for the implicit racism effect and (c) stimulus familiarity is an important moderating variable that can influence the sensitivity of implicit attitude measures. We discuss the results in relation to the validity of the Implicit Association Test and theories of implicit social cognition.