Impact of Uncertainty Visualizations on Multi-Criteria Decision-Making

Abstract

This study expands on previous research on visualizing uncertainty in decision support tools by exploring the effects of two factors. First, it explores how different methods of displaying uncertainty and automated recommendations affect the decision. Second, it investigates how these factors are influenced by individual differences. In this within-subjects study, 346 participants were assigned to one of six conditions, each with a different version of a multicriteria information dashboard. Participants selected the optimal product over 12 timed trials and answered questions about their decision-making process. Results showed that dashboards with detailed decision recommendations were associated with the highest trust, reliance, and decision accuracy. Including uncertainty on charts reduced decision confidence, accuracy, and speed, but also interacted with gender and recommendation style to impact how frequently participants sought more information to make decisions. Women reported lower decision confidence and more frequently sought additional information to make decisions. These results inform designers of data dashboards about how to promote more transparent and inclusive decision support tools that convey uncertainty in the input data.