5 Dec 2002 15:34 AR AR178-PS54-12.tex AR178-PS54-12.sgm LaTeX2e(2002/01/18) P1: FHD10.1146/annurev.psych.54.101601.145225Annu. Rev. Psychol. 2003. 54:297–327doi: 10.1146/annurev.psych.54.101601.145225Copyrightc° 2003 by Annual Reviews. All rights reservedFirst published online as a Review in Advance on August 6, 2002IMPLICIT MEASURES IN SOCIAL COGNITIONRESEARCH: Their Meaning and UseRussell H. Fazio and Michael A. OlsonDepartment of Psychology, Ohio State University, Columbus, Ohio,43210-1222; e-mail: [email protected] Words attitude measurement, automatic processing, priming, ImplicitAssociation Test■ Abstract Behavioral scientists have long sought measures of important psycho-logical constructs that avoid response biases and other problems associated with di-rect reports. Recently, a large number of such indirect, or “implicit,” measures haveemerged. We review research that has utilized these measures across several domains,including attitudes, self-esteem, and stereotypes, and discuss their predictive valid-ity, their interrelations, and the mechanisms presumably underlying their operation.Special attention is devoted to various priming measures and the Implicit AssociationTest, largely due to their prevalence in the literature. We also attempt to clarify severalunresolved theoretical and empirical issues concerning implicit measures, includingthe nature of the underlying constructs they purport to measure, the conditions underwhich they are most likely to relate to explicit measures, the kinds of behavior eachmeasure is likely to predict, their sensitivity to context, and the construct’s potentialfor change.CONTENTSINTRODUCTION ..................................................... 298SOME OPENING OBSERVATIONS ...................................... 301Where’s the Theory? ................................................. 301Where’s the Implicit? ................................................. 302THE RELATION BETWEEN IMPLICITAND EXPLICIT MEASURES .......................................... 303PREDICTIVE VALIDITY OF IMPLICIT MEASURES ....................... 305Priming Measures ................................................... 305Implicit Association Test .............................................. 307Other Measures ..................................................... 310QUESTIONABLE INTERRELATIONSAMONG IMPLICIT MEASURES ....................................... 311UNDERLYING MECHANISMS ......................................... 312What Drives Priming? ................................................ 313What Drives the Implicit Association Test? ............................... 313Implications ........................................................ 3150066-4308/03/0203-0297$14.002975 Dec 2002 15:34 AR AR178-PS54-12.tex AR178-PS54-12.sgm LaTeX2e(2002/01/18) P1: FHD298 FAZIO¥OLSONEFFECTS OF CONTEXT ON IMPLICIT MEASURES ....................... 317ADDITIONAL QUESTIONS FOR FUTURE RESEARCH .................... 318Role of Awareness ................................................... 318Stereotype Versus Attitude Activation .................................... 319Changing Automatically Activated Constructs ............................. 319FINAL THOUGHTS ................................................... 320INTRODUCTIONOver the past few years, there has been a surge of interest in the use of implicitmeasurement techniques in social psychological research. If this assertion neededany verification, one merely has to call attention to the special issues of four jour-nals recently devoted to the topic: Journal of Experimental Social Psychology on“Unconscious Processes in Stereotyping and Prejudice” (Banaji 1997), Cogni-tion and Emotion on “Automatic Affective Processing” (De Houwer & Hermans2001), Zeitschrift f¨ur Experimentelle Psychologie on “Attitude Measurement Us-ing the Implicit Association Test (IAT)” (Plessner & Banse 2001), and Journalof Personality and Social Psychology on “Implicit Prejudice and Stereotyping:How Automatic Are They?” (Devine 2001). The research has involved a varietyof domains, including attitudes (e.g., Fazio et al. 1995, Greenwald et al. 1998),stereotypes (e.g., Wittenbrink et al. 1997, Nosek et al. 2002a), self-esteem (Hettset al. 1999, Bosson et al. 2000, Koole et al. 2001, Rudman et al. 2001b), closerelationships (e.g., Banse 1999), and health behavior (e.g., Stacy et al. 1997).A variety of different implicit measurement techniques have been employed.One such technique involves various priming procedures that have proven use-ful in the past as a means of assessing what is activated from memory by thepresentation of some attitude object (e.g., Gaertner & McLaughlin 1983, Fazioet al. 1986, Greenwald et al. 1989, Perdue et al. 1990). For example, Fazio et al.(1995) examined the consequences of priming participants with photos of blackversus white undergraduates. The participants’ primary task was to indicate theconnotation of an evaluative adjective (e.g., “pleasant” or “awful”) as quickly aspossible. In the context of a cover story concerning the judgment of word meaningbeing an automatic skill that should not be disrupted by the participants havingto perform an additional task simultaneously, each target adjective was precededby the brief presentation of a photo. Participants were instructed to attend to thesefaces so that they would be able to pick them out in a later phase of the experiment.The black and white faces had different consequences for the latency with whichparticipants could indicate the connotation of the subsequently presented targetadjective. Relative to what was observed for white faces, black faces facilitatedresponding to negative adjectives and interfered with responding to positive adjec-tives. The pattern suggests that, on average, negativity was automatically activatedby the black primes.Probably the most well-known implicit measurement technique is the ImplicitAssociation Test (IAT), developed by Greenwald et al. (1998). This procedure5 Dec 2002 15:34 AR AR178-PS54-12.tex AR178-PS54-12.sgm LaTeX2e(2002/01/18) P1: FHDIMPLICIT MEASURES 299assesses the strength of an association between a target concept and an attributedimension by considering the latency with which participants can employ tworesponse keys when each has been assigned a dual meaning. The participants’task is to categorize stimuli as they appear on the screen. For example, in theGreenwald et al. (1998) IAT
View Full Document