Cognitive Reflection Test: Whom, how, when

Abstract

The use of the Cognitive Reflection Test as a covariate to explain behavior in Economics and Psychology experiments has significantly increased in the past few years. Experiments have shown its usefulness in predicting behavior. However, little is known about if the test is gender biased, whether incentives matter or how different implementation procedures impact outcomes. Here we report the results of a meta-study of 118 Cognitive Reflection Test studies comprising of 44,558 participants across 21 countries. We find that there is a negative correlation between being female and the overall, and individual, correct answers to CRT questions. Monetary incentives do not impact performance. Regarding implementation procedures, taking the test at the end of the experiment negatively impacts performance. Students perform better compared to non-students. We obtain mixed evidence on whether the sequence of questions matters. Finally, we find that computerized tests marginally improve results.