Forecast Uncertainty Information Improves Multialternative Risky Decision-Making

Abstract Research suggests that nonexperts make better decisions in binary decision tasks and have higher trust in forecasts that include numeric uncertainty estimates (e.g., 30% chance) compared to single-value forecasts (e.g., 6-in. snow accumulation). However, in real-world situations in which one must choose between protective action at a cost or risking a larger probabilistic loss, there is often an intermediate option. Two experiments, reported here, tested whether the previously shown advantages for numeric uncertainty forecasts generalize to decisions with three options. In a snow-day school closure paradigm, a 2-option condition (close, open) was compared to a 3-option condition that also included a cheaper alternative (delay) with less protection than closing, but more protection than staying open when snow occurred. The advantages of probabilistic forecasts persisted regardless of the number of options. In addition, the risk-seeking tendency, typical for decisions between losses, was attenuated in the 3-option condition in experiment 1 and reversed in experiment 2.