Dissertation

Graphic Communication about Scientific Uncertainties: Examples from Influenza and COVID-19 Forecasts

Yang, Yanran

2021

Overview

The Epidemic Prediction Initiative of U.S. Centers for Disease Control and Prevention holds annual forecasting challenges for various diseases, including influenza, Aedes, and the current forecasting challenge for the ongoing COVID-19 pandemic. Universities, research institutes, independent researchers, corporations, consultancies, and government agencies make such forecasts on many other public health events as well. The value of these forecasts depends on how well people trust and understand them, when making decisions regarding public policy, preparedness, and personal actions. This thesis applies research on risk and science communication, visualization, and decision making to communicating the scientific uncertainties of influenza and COVID-19 forecasts. The thesis (1) designs multiple graphic displays and communication aids for communicating influenza and COVID-19 forecasts, (2) creates hypothetical yet realistic decision scenarios and tasks to examine non-expert users’ performance of interpreting and using such forecasts and risk information in related decision making, and (3) examines how decision makers accomplish these tasks, as a function of both how the forecast information is displayed and their own personal properties, including their perceived and actual ability to interpret the displays and how motivated reasoning may influence their response to risk communication messages. Study 1 (Chapter 2) compares performances in making four influenza-related decisions based on four graphic presentations of influenza forecasts uncertainty, extending Barbara Tversky’s research on conceptual-spatial congruence to probabilistic forecasts. I find that participants have better comprehension with the most familiar display (bar chart), for all four decisions, but do not perceive it as more helpful or have more confidence in their responses. I find that participants who report greater familiarity with a display actually perform more poorly, express greater decision confidence, and rate the display as more helpful.Study 2 (Chapter 3) compares judgments and decisions based on a suite of uncertain forecasts, depending on whether they are presented individually, in aggregate (an ensemble), or both, and in what order. I find that participants who see just the suite or just the ensemble respond differently. However, they respond similarly when they update their judgments and decisions after seeing the other display as well.Study 3 (Chapter 4) compares risk judgments and decisions regarding school attendance and vaccinations, for individuals with different political beliefs. It examines the effects of asking participants, in one experimental group to self-affirm their commitment to good deeds, and intervention that had reduced the impact of beliefs in other settings. I find clear disparities between vaccinated liberals and unvaccinated conservatives in their risk estimates and decision making. However, the self-affirmation intervention had no effect on participants’ risk estimates and decisions here.Overall, this thesis extends risk communication and visualization literature to develop and test graphic displays for uncertainties in epidemiological forecasts and show the value of pretesting the communication aids with potential users. It extends theories of visual information processing, self-identity, and judgment under uncertainty.

• Publisher: ProQuest Dissertations & Theses
• Subject: Behavioral Sciences / Public policy
• ISBN: 9798759996743