Providing information for decision making: Contrasting description and simulation

Abstract

Providing information for decision making should be like telling a story. You need to know, first, what you want to say; second, whom you are addressing; and third, how to match the message and audience. However, data presentations frequently fail to follow these simple principles. To illustrate, we focus on presentations of probabilistic information that accompany forecasts. We emphasize that the providers of such information often fail to realize that their audiences lack the statistical intuitions necessary to understand the implications of probabilistic reasoning. We therefore characterize some of these failings prior to conceptualizing different ways of informing people about the uncertainties of forecasts. We discuss and compare three types of methods: description, simulation, and mixtures of description and simulation. We conclude by identifying gaps in our knowledge on how best to communicate probabilistic information for decision making and suggest directions for future research.