Presentation
Utility Visualizations in Decision Aiding Do Not Facilitate Intuitive Data Estimation
SessionWednesday Poster Session
DescriptionWe report an experiment evaluating people's ability to accurately extract data from a utility transformed geospatial visualization. The study employed a 2 (Map Type: heat vs. utility map) × 2 (Color Scheme: rainbow vs. blackbody radiation) × 2 (Subjectivity: given vs. chosen location) mixed design. Participants (n = 199) were presented single-attribute map displays and asked to estimate the attribute value at a marked location. They completed 64 trials. Brier scores were computed to measure estimation uncertainty. We conducted Type II Analysis of Deviance (Wald χ² test) tests to assess the effects of each factor mentioned in the design. We found clear evidence that participants consistently provided more accurate estimations of environmental properties when using heat maps compared to utility maps. Proponents of utility-based visualizations may argue that this point is moot as such displays could be toggled on and off to grant access to the original data. However, how utility information is extracted from displays and integrated into human decision-making processes is of theoretical importance if only to anticipate when use of such systems is unwise. Moreover, adding features to decision aids to mitigate the limitations of utility-based visualizations undermines the goal of reducing load on operators.
Event Type
Poster
TimeWednesday, October 21st5:30pm - 6:30pm PDT
Location