Practice Improves Performance of a 2D Uncertainty Integration Task Within and Across Visualizations

IEEE Trans Vis Comput Graph. 2023 Sep;29(9):3949-3960. doi: 10.1109/TVCG.2022.3173889. Epub 2023 Aug 1.

Abstract

Information uncertainty is ubiquitous in everyday life, including in domains as diverse as weather forecasts, investments, and health risks. Knowing how to interpret and integrate this uncertain information is vital for making good decisions, but this can be difficult for experts and novices alike. In this study, we examine whether brief, focused practice can improve people's ability to understand and integrate bivariate Gaussian uncertainty visualized via ensemble displays, summary displays, and distributional displays, and we examine whether this is influenced by the complexity of the displayed information. In two experiments (N=118 and 56), decision making was faster and more accurate after practice relative to before practice. Furthermore, the performance improvements transferred to use of display types that were not practiced. This suggests that practice with feedback may improve underlying skills in probabilistic reasoning and provides a promising approach to improve people's decision making under uncertainty.