Interactive graphics for communicating health risks.

摘要

Materials for consumer informatics, patient decision support, and health promotion frequently incorporate quantitative risks such as percentages, rates, or proportions. These risks are frequently illustrated with stick figures, bar charts, or other graphics. However, risk communication models (such as the extended parallel process model) and decision models (such as prospect theory) generally focus on features of the verbal message, while failing to explore the effects of number format and illustration design. That these factors are important is shown by growing literatures that describe consumers' difficulty comprehending numbers, as well as strong effects of numeracy and risk graphic design on perceived risk and choices. This dissertation proposes an integrated model of risk communication that draws from cognitive psychology and health behavior models to consider contributions of the verbal message, features of the graphic illustration, and the audience's numeracy. Next, the dissertation presents an ontology of features of risk graphics (including animation and interactivity) and their cognitive/perceptual effects, developed through a systematic literature review. Third, a qualitative formative study was conducted that resulted in the design of several animated and interactive risk graphics with applications in Web-based communication. Finally, a questionnaire study was conducted to assess the effect of the interactive graphics on risk estimates, risk feelings, and decisions, and interactions with numeracy. Numeracy strongly affected risk estimates, risk feelings, and decisions, with lower numeracy correlated with higher perceived risks. Interaction with one of the interactive graphics affected risk perceptions and narrowed differences between high- and low-numeracy respondents. Computer-based graphical displays such as the ones developed in this project have the potential to be applied in informatics interventions for health education, tailored health and risk communication, shared medical decision-making, and patient decision support. The methods used are also promising for assessing effects of other scientific data graphics.