Dominik L. Schall, Dominik Doll, and Alwine Mohnen, “Caution! Warnings as a Useless Countermeasure to Reduce Overconfidence? An Experimental Evaluation in Light of Enhanced and Dynamic Warning Designs.” Journal of Behavioral Decision Making 30(2): 347-358, April 2017.
• Two experiments look at over-precision in guessing the answers to factual questions. People tend to choose 90% confidence intervals that are much, much too narrow: people are overly optimistic about their precision.
• Warning people to be wary of over-precision in itself (even when the warning takes a fairly fancy form) seems useless, but…
• …a dynamic (pop-up) warning significantly reduces over-precision. Effective warning content and the dynamic element seem to both be required.
• Effective warning content is rather involved: it consists of a highlighted signal word ("Caution"); an explanation of the hazard (overconfidence); the consequence of the hazard (the truth lying outside of the chosen confidence interval excessively); and instructions (widen your intervals) about how to overcome the hazard.
No comments:
Post a Comment