ATTITUDES AND SOCIAL COGNITIONUnderstanding and Using the Implicit Association Test: I. An ImprovedScoring AlgorithmAnthony G. GreenwaldUniversity of WashingtonBrian A. NosekUniversity of VirginiaMahzarin R. BanajiHarvard UniversityIn reporting Implicit Association Test (IAT) results, researchers have most often used scoring conven-tions described in the first publication of the IAT (A. G. Greenwald, D. E. McGhee, & J. L. K. Schwartz,1998). Demonstration IATs available on the Internet have produced large data sets that were used in thecurrent article to evaluate alternative scoring procedures. Candidate new algorithms were examined interms of their (a) correlations with parallel self-report measures, (b) resistance to an artifact associatedwith speed of responding, (c) internal consistency, (d) sensitivity to known influences on IAT measures,and (e) resistance to known procedural influences. The best-performing measure incorporates data fromthe IAT’s practice trials, uses a metric that is calibrated by each respondent’s latency variability, andincludes a latency penalty for errors. This new algorithm strongly outperforms the earlier (conventional)procedure.The Implicit Association Test (IAT) provides a measure ofstrengths of automatic associations. This measure is computedfrom performance speeds at two classification tasks in whichassociation strengths influence performance. The apparent useful-ness of the IAT may be due to its combination of apparentresistance to self-presentation artifact (Banse, Seise, & Zerbes,2001; Egloff & Schmukle, 2002; Kim & Greenwald, 1998), itslack of dependence on introspective access to the associationstrengths being measured (Greenwald et al., 2002), and its ease ofadaptation to assess a broad variety of socially significant associ-ations (see overview in Greenwald & Nosek, 2001).The IAT’s measure, often referred to as the IAT effect, is basedon latencies for two tasks that differ in instructions for using tworesponse keys to classify four categories of stimuli. Table 1 de-scribes the seven steps (blocks) of a typical IAT procedure.The first IAT publication (Greenwald, McGhee, & Schwartz,1998) introduced a scoring procedure that has been used in themajority of subsequently published studies. The features of thisconventional algorithm (see Table 4 later in the article) include (a)dropping the first two trials of test trial blocks for the IAT’s twoclassification tasks (Blocks 4 and 7 in Table 1), (b) recodinglatencies outside of lower (300 ms) and upper (3,000 ms) bound-aries to those boundary values, (c) log-transforming latenciesbefore averaging them, (d) including error-trial latencies in theanalyzed data, and (e) not using data from respondents for whomaverage latencies or error rates appear to be unusually high for thesample being investigated. The main justification for originallyusing these conventional procedures was that, compared withseveral alternative procedures often used with latency data, theconventional procedures typically yielded the largest statisticaleffect sizes.Previous theoretical and methodological analyses have providedmethods of dealing with problems that occur in latency measuresAnthony G. Greenwald, Department of Psychology, University of Wash-ington; Brian A. Nosek, Department of Psychology, University of Virginia;Mahzarin R. Banaji, Department of Psychology, Harvard University.The revised scoring procedures described in this report are hereby madefreely available for use in research investigations. SPSS syntax for com-puting Implicit Association Test measures using the improved algorithmcan be obtained at the University of Washington Web site (http://faculty.washington.edu/agg/iat_materials.htm). However, the improved scoringprocedures described in this report (patent pending) should not be used forcommercial applications nor should they or the contents of this report bedistributed for commercial purposes without written permission of theauthors.This research was supported by three grants from National Institute ofMental Health: MH-41328, MH-01533, and MH-57672. The authors aregrateful to Mary Lee Hummert, Kristin Lane, and Deborah S. Mellott forhelpful comments on an earlier version, and also to Laurie A. Rudman andEliot R. Smith, who commented as colleagues rather than as consultingeditors for this journal.Correspondence concerning this article should be addressed to AnthonyG. Greenwald, Department of Psychology, University of Washington, Box351525, Seattle, Washington 98195-1525. E-mail: [email protected] of Personality and Social Psychology, 2003, Vol. 85, No. 2, 197–216Copyright 2003 by the American Psychological Association, Inc. 0022-3514/03/$12.00 DOI: 10.1037/0022-3514.85.2.197197in the form of speed–accuracy tradeoffs (e.g., Wickelgren, 1977;Yellott, 1971), age-related slowing (e.g., Faust, Balota, Spieler, &Ferraro, 1999; Ratcliff, Spieler, & McKoon, 2000), and spuriousresponses that appear as extreme values (or outliers; Miller, 1994;Ratcliff, 1993). Remarkably, research practice in cognitive andsocial psychology has been no more than mildly influenced by thismethodological work. That limited influence may be explained bythree practical considerations: First, some of the methodologicalrecommendations are costly to use—for example, several hours ofdata collection with each subject may be needed to obtain data setsfrom which individual-subject speed–accuracy tradeoff functionscan be constructed. Second, journal editors and reviewers rarelyinsist on the more painstaking methods. Third, researchers who usethe more sophisticated (and painstaking) methods are rarely re-warded for their extra work—conclusions based on the moreeffortful methods often diverge little from those based on simplermethods.The conventional scoring procedure for the IAT has not previ-ously been subject to systematic investigations of psychometricproperties. Additionally, the conventional scoring procedure lacksany theoretical rationale that distinguishes it from other scoringmethods (Greenwald, 2001). Consequently, the authors welcomeda fortuitous opportunity to compare the conventional procedurewith alternatives. This opportunity arose through the operation ofan educational Web site (http://www.yale.edu/implicit/) at whichseveral IAT procedures had been made available for demonstrationuse by drop-in visitors.This article first describes the IAT Web site and then presents aseries of studies that were designed to