Sample Range as Subjective Variability and Implications for Experience-Based Decision Making
Abstract
Variance and standard deviation are common metrics of statistical variability in the management and decision-making literature. In everyday contexts, however, information about sample range prevails. Range is arguably easier to comprehend and compute than variance, yet it often yields equivalent judgments about variability. Hence, we reason that decision makers often rely on range to judge variability. We test this hypothesis across four behavioral experiments (total n = 1,093). When comparing the variability of two number sets, decision makers predominantly based their judgments on sample range rather than sample variance when the two metrics disagree (experiment 1). Sample range tends to increase with sample size, whereas sample variance does not. Consistent with the range as variability hypothesis, decision makers perceived distributions as more variable after observing larger samples (experiment 2). This perceived variability, in turn, influenced risky decisions, as larger sample sizes led to decreased valuations of uncertain rewards for risk-averse decision makers (experiment 3). In further support of the hypothesis, when asked to mentally simulate new samples from a previously experienced distribution, participants generated more numbers within the observed range and fewer numbers outside of it than the true distribution implies (experiment 4). To model the data, we propose a “cognitive fence” idea as a potential psychological explanation for how experience shapes subjective variability. We conclude by discussing implications for research on overconfidence, distributional learning, risk perception, and communication.
This paper was accepted by Jack Soll, behavioral economics and decision analysis.
Supplemental Material: The online appendix and data files are available at https://doi.org/10.1287/mnsc.2024.05050.

