Friday, November 8, 2019

About the Implicit Association Tests (IATs)... Predicting Behavior With Implicit Measures: Disillusioning Findings

Predicting Behavior With Implicit Measures: Disillusioning Findings, Reasonable Explanations, and Sophisticated Solutions. Franziska Meissner, Laura Anne Grigutsch, Nicolas Koranyi, Florian Müller and Klaus Rothermund. Front. Psychol., November 8 2019. https://doi.org/10.3389/fpsyg.2019.02483

Two decades ago, the introduction of the Implicit Association Test (IAT) sparked enthusiastic reactions. With implicit measures like the IAT, researchers hoped to finally be able to bridge the gap between self-reported attitudes on one hand and behavior on the other. Twenty years of research and several meta-analyses later, however, we have to conclude that neither the IAT nor its derivatives have fulfilled these expectations. Their predictive value for behavioral criteria is weak and their incremental validity over and above self-report measures is negligible. In our review, we present an overview of explanations for these unsatisfactory findings and delineate promising ways forward. Over the years, several reasons for the IAT’s weak predictive validity have been proposed. They point to four potentially problematic features: First, the IAT is by no means a pure measure of individual differences in associations but suffers from extraneous influences like recoding. Hence, the predictive validity of IAT-scores should not be confused with the predictive validity of associations. Second, with the IAT, we usually aim to measure evaluation (“liking”) instead of motivation (“wanting”). Yet, behavior might be determined much more often by the latter than the former. Third, the IAT focuses on measuring associations instead of propositional beliefs and thus taps into a construct that might be too unspecific to account for behavior. Finally, studies on predictive validity are often characterized by a mismatch between predictor and criterion (e.g., while behavior is highly context-specific, the IAT usually takes into account neither the situation nor the domain). Recent research, however, also revealed advances addressing each of these problems, namely (1) procedural and analytical advances to control for recoding in the IAT, (2) measurement procedures to assess implicit wanting, (3) measurement procedures to assess implicit beliefs, and (4) approaches to increase the fit between implicit measures and behavioral criteria (e.g., by incorporating contextual information). Implicit measures like the IAT hold an enormous potential. In order to allow them to fulfill this potential, however, we have to refine our understanding of these measures, and we should incorporate recent conceptual and methodological advancements. This review provides specific recommendations on how to do so.

Why does he act like this? Why does she not do what she intended to do? In our everyday life, we often try to find explanations for the behavior of others, and of ourselves, respectively. Explaining and predicting behavior is also of key interest across all fields of scientific psychology, especially when it comes to deviations between individuals’ actual behavior and the attitudes, goals, or values held by these very individuals. Why do people discriminate although they report to hold egalitarian values? Why do they not quit smoking although they know that smoking is bad? Why is there a gap between people’s self-reported attitudes and actual behavior?

Dual-process or dual-system models attribute seemingly inconsistent behavior to the triumph of an impulsive system over a reflective system of behavior control (e.g., Strack and Deutsch, 2004; Hofmann et al., 2009; Kahneman, 2011). The notion that the prediction of behavior could be improved considerably if one succeeds in measuring the processes of the impulsive system (Hofmann et al., 2007; Friese et al., 2008; Hofmann and Friese, 2008) fueled research applying so-called implicit measures of attitudes. The most popular of these measures, the Implicit Association Test (IAT, Greenwald et al., 1998) evoked enthusiastic hopes regarding its predictive value. Unfortunately, however, the IAT and its derivatives have not met these expectations.

In this article, we review findings illustrating reasons for the IAT’s unsatisfying predictive value, as well as promising ways forward. We will outline that in order to improve the predictive power of implicit measures, differentiation is key. We will argue that future research should put more emphasis on the underlying processes and concepts behind these measures. We begin with sketching the discrepancy between individuals’ behaviors and their self-expressed attitudes. We then summarize the (mostly unsatisfying) attempts to close this attitude-behavior gap with the help of implicit measures. In the main part of this article, we identify features of implicit measures that are responsible for their weak predictive validity. We review findings illustrating each of these problematic aspects along with specific, sophisticated solutions providing promising directions for future research.


Closing Thoughts

In this article, we presented an overview of possible reasons for the weak relationship between implicit measures like the IAT and behavioral criteria. We outlined that the unsatisfying predictive value of the IAT is due to (1) extraneous influences like recoding, (2) the measurement of liking instead of wanting, (3) the measurement of associations instead of complex beliefs, and/or (4) a conceptual mismatch of predictor and criterion. We presented precise solutions for each of these problems. More precisely, we suggested to switch to procedural variations that minimize extraneous influences (i.e., the SB-IAT, Teige-Mocigemba et al., 2008; and the IAT-RF; Rothermund et al., 2009), and to apply sophisticated analysis tools (i.e., the ReAL model, Meissner and Rothermund, 2013) that separate relevant processes from those extraneous influences. Second, we presented an overview of different implicit measures that go beyond the measurement of evaluative associations, and instead quantify actual implicit wanting (e.g., the W-IAT, Koranyi et al., 2017). Third, we pointed to implicit measures of beliefs (e.g., the PEP, Müller and Rothermund, 2019) that allow a more nuanced view on individual attitudes and values than measures that tap into associations. Finally, we emphasized the importance of measuring behavior proper and outlined that implicit measures incorporating contextual information might be more adequate in assessing the structure of implicit attitudes or beliefs and their implications for behavior (Casper et al., 2011; Kornadt et al., 2016). Each of the recent developments presented in the current paper has the potential to increase the predictive power of implicit measures. Future research will also have to clarify whether a combination of these approaches may lead to further improvement. Inspired by the fruitful research on dual-process or dual-systems models, we further suggest to invest in theoretical considerations: Which forms or aspects of behavior should be related to which processes involved in which implicit measures? Differentiation is key, with regard to both the predictor and the criterion.
We strongly argue not to take the validity of implicit measures like the IAT for granted. Instead, we should take into account the complexity of these measures, especially when it comes to the predictive value for real-life behavior. As outlined in the current review, the past 20 years of research have provided us with a number of good reasons for why the IAT and its derivatives did not succeed in closing the attitude-behavior gap, and enriched our toolbox with promising, sophisticated improvements. Future research will benefit from harnessing the power of such a more differentiated view on implicit measures.

No comments:

Post a Comment