Advances and Challenges in the Indirect Measurement of Individual Differences at Age 10 of the Implicit Association Test

Abstract

Ever since Sigmund Freud proposed that many core psychological processes cannot be self-observed because they occur outside of the conscious mind, psychologists have searched for methods that can be used to make diagnostic inferences about a person’s dispositions without having to ask the person directly. On the one hand, the development of such indirect methods was driven by the assumption that persons may have only limited access to the underlying (automatic) processes that may be responsible for their thoughts, feelings, and actions (Nisbett & Wilson, 1977). On the other hand, the search for indirect measures was also motivated by the need to overcome a major weakness of direct (i.e., self-report) measures: their sensitivity to the distorting effects of the need for approval and other self-presentational concerns. Despite many proposals of indirect measures, none received unanimous approval from the scientific community and none obtained the trust that researchers put in self-report measures despite their well-known shortcomings. The limited success of indirect measures can be accounted for by two major reasons: First, little agreement was achieved regarding the kinds of behavior that would be suitable indicators of unconscious processes and the inferences that could be drawn from them. For example, there is little consensus on what a specific comment on a TAT picture tells us about the personality profile of a person. Second, the reliability and validity of indirect measures tends to be disappointingly low when compared to direct measures. With the fruitful exchange between cognitive and social psychology and the advance of computer-based assessment during the last 20 years, a new wave of mostly reaction-timebased indirect measures has been proposed (also referred to as implicit measures). Ten years ago, a general and flexible indirect assessment procedure appeared on the stage that seems to have considerable potential for overcoming the problems of previous indirect measures: the Implicit Association Test (IAT) as proposed by Greenwald, McGhee, and Schwartz (1998; see Schnabel, Asendorpf & Greenwald, 2008, for an up-to-date review). The IAT employs patterns in reaction time that people need for the classification of stimuli (words, pictures, symbols) according to their category membership. Classification is performed via simple motor responses such as pressing a key on a computer keyboard. The IAT is based on the assumption that classification can be performed faster if the classification of stimuli from associated concepts requires the same motor response. In the case of attitude measurement, the IAT presumes that attitudes are stored in memory as associations between an attitude category (nuclear energy) and an evaluative category (bad, evil). Accordingly, personality-IATs assume that traits are stored in memory as associations between the self and attribute concepts (talkative, intelligent, etc). The IAT is a very flexible procedure because it allows for the combination of a large number of concepts whose associations define the constructs to be measured. The publication of the IAT had tremendous impact first on social cognition and attitude research from where it originated, and then soon after on several other research areas such as personality assessment, self-esteem research, special areas of clinical psychology, and consumer behavior. It instigated a renewed interest in indirect measures and promoted the development of other reaction-time-based indirect measures. It also directed attention toward the interplay of automatic versus controlled processes and helped to refine dual-process and dual-system conceptions of the mind. Finally, it prompted a new discussion of the consistency issue that had been a focus of debate in the attitude and personality literature three decades earlier. As is the case with most new diagnostic paradigms, the IAT and other new indirect measures were not incorporated uncritically into the body of psychological assessment methods. Rather, many researchers submitted them to rigorous theoretical, methodological, and empirical analyses. These analyses identified a number of problems and open questions. Many of these ongoing issues will be addressed in the cut-