Simplified mental representations as a cause of overprecision

Abstract

All materials related to this study (z-tree programs, raw data, Stata-scripts, and Mathematica programs) are publicly available on the Open Science Framework (OSF) under the link https://osf.io/wq5nm/?view_only=2581875617c942f9bdf1c4db02ade4fd.

An Experimental Test of Some Economic Theories of Optimism

“…On the other hand, most underprecise individuals observed larger proportions of female names. We stress that similar findings have been also obtained in the study by López-Pérez et al (2020), where the elicited intervals were incentivized. This suggests that our results here are not an artifact of our experimental design.…”

“…In other words, they indicated a lower and an upper bound for θ, such that they believed that the correct θ was 'almost surely' in the interval determined by those limits. Confidence intervals were not incentivized; as we discuss later, however, our results do not differ much from those in López-Pérez et al (2020), where subjects were paid for accuracy. After this interval estimation, additionally, we included an incentivized 'recall task' and two questions so as to check whether they expected to recall better female than male extractions, i.e., good than bad news; this data is irrelevant for the test of the theories considered here, but see Caballero and López-Pérez (2020b) for a full description and analysis.…”

Constraints on Thinking Cause Overprecision

“…If the agent neglects to think through (X 2 , Y 2 ) when estimating P r(X 1 , Y 1 |E) she gets: 20) which is higher than P r(X 1 |E) = .1, i.e., the conjunction fallacy emerges.…”

“…They focus on how the representativeness heuristic guides consideration of possibilities in a multivariate environment, and do not address overprecision. [20] discusses how not considering all possibilities can produce narrow confidence intervals. We provide a general approach, which enables us to account for a wider variety of empirical results.…”