Uncertainty Information & Non-Expert Decisions

This talk will describe an experimental research demonstrating the benefit of including numeric uncertainty information in weather forecasts intended for non-experts. Our results suggest that numeric uncertainty estimates (e.g. 30% chance) allow users to better differentiate situations that do and do not require precautionary action while also increasing understanding and trust in the forecast. People appear to understand that all forecasts involve uncertainty and find forecasts that acknowledge it explicitly more plausible.  Moreover, these benefits are not dependent on higher education or special abilities—they extend to a broad range of users. However, this work also suggests that it is important to present numeric uncertainty estimates in a manner that is compatible with the way in which people process information and with their decision goal.