Beyond the two-state setting: Do individuals generally underinfer from high-weight information?

Abstract

Individuals have been shown to systematically deviate from Bayes' law when updating probabilistic beliefs. Experimental studies indicate that individuals' underinference with respect to new signal sets is more pronounced if the weight or reliability of the signal set is high. We challenge the generality of this prominent finding and argue that the two-state setting, which dominates experimental research, is an extreme and peculiar case with respect to the normative importance of signal set characteristics. Thus, previously identified judgment biases might not extend to settings with more than two states. Our experimental analyses support this conjecture as they show that individuals' weight-dependent underinference is no general phenomenon but specific to the two-state setting. Given that many real-world information environments do not resemble such a simplified setting, our results caution against an indiscriminate transfer of popular biases in belief updating to a broad set of real-world applications.

Keywords: Information Weight, Over- and Underinference, Judgment Biases

JEL Classification: C91, D91