CAREER: HCC: Designing Visualizations to Support Critical Thinking and Calibrated Trust in Data

University Of Massachusetts Amherst, Amherst MA

Investigators

Abstract

People often use data visualizations to understand, communicate, and make decisions about important questions in science, education, healthcare, and other domains. However, data visualization is not neutral; choices about what to visualize and how to visualize it can impact how a viewer interprets the visualization, especially when they don’t have access to the underlying data. This fact raises important questions about how and whether people trust their interactions with visualized data, and how visualization designers can help viewers develop appropriately calibrated trust (not overtrusting unreliable inferences nor undertrusting meaningful relationships) in the visualization. This project’s goal is to advance research around assessing and designing for trust calibration in visualizations: creating appropriate measures of trust in human-data interaction, building models of how visualization design choices can impact trust, and developing design guidelines for both creators and consumers of visualizations to help them identify signs of trust and mistrust in visualized data. The project team will also create visualization tools, educational materials, and empirical data to help researchers, policymakers, and members of the general public think about trustworthy computing and visualization ethics. The project is structured around three research objectives. First, referencing theories and research methodologies from social sciences, the project team will produce a collection of methods to reliably measure user trust in data visualizations. Second, using these methods, the team will conduct a systematic series of empirical studies that identify which visualization design factors can drive critical thinking and calibrated trust. These factors will include manipulations of the perceived clarity, complexity, accuracy, and amount of data, using real-world datasets across multiple domains related to public policy and finance. The use of such diverse data will allow the team to identify domain-specific versus generalizable visualization elements that promote critical thinking and calibrated trust. The third objective is a formal evaluation of these guidelines through testing prototype visualizations and visualization systems that facilitate calibrated trust. This will involve field studies and interviews with researchers, practitioners, and members of the general public to examine the effectiveness of the trust guidelines and identify ethical practices for trustworthy visual data communication. The team will also develop interdisciplinary coursework and initiatives that bring together computer science, public policy, psychology, and behavioral economics to advance practice and education around visual data communication. This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.