Improving Implicit Attitude Measurement

University Of North Carolina At Chapel Hill, Chapel Hill NC

Investigators

Abstract

One of the most important recent insights in the study of attitudes and preferences is that they can be measured implicitly, without asking a person to introspect and self-report. Implicit measurements allow researchers to study attitudes that respondents may be unwilling or unable to report explicitly. They therefore are valuable for studying socially sensitive topics, where respondents may be less than candid. Yet despite this progress, current methods have several psychometric limitations to overcome. Reliability and predictive validity for implicit methods are often lower than accepted for explicit measurement methods. And explicit and implicit measurements of attitudes toward the same topic often differ in multiple ways at the same time, allowing for confounds and alternative explanations. This project develops a novel method for implicitly measuring attitudes, the Affect Misattribution Procedure. It measures attitudes by using the tendency to unintentionally misattribute affective responses from one source to another. Three sets of studies, focusing on racial attitudes and alcohol abuse, compare the method to currently available methods. The first set of studies tests the ability of this method to predict racial discrimination and alcohol abuse behaviors, in comparison to two commonly used implicit methods. The second set tests for resistance to social desirability. The third set uses the method to equate implicit and explicit measures on all dimensions except the crucial difference of intent to express one's attitude. Unlike other implicit methods, the metric for this method is an evaluation rather than response time. Therefore, self-reports and implicit responses can be directly compared on the same scales. By eliminating extraneous differences between implicit and explicit methods, the relationship between implicit and explicit measures can be tested more directly than previously has been possible. The project has potentially important broader impacts, increasing the fidelity of measurement in important domains such as racial discrimination and substance abuse. The project will help advance research infrastructure for social and behavioral science by making publicly available a new validated method for implicit attitude measurement. More accurate measurement will facilitate theory testing and future efforts to identify causes of, and remedies for, destructive behavior.