Stimuli, data, and analysis script for Witt & Dhami (in press @ JARMAC). Abstract: Icon arrays can help patients to understand statistical information such as their risk of having a disease given a diagnostic test result. Bayesian problems such as these are computationally challenging but can be solved via the simple ratio of Ʃ(disease|positive)/Ʃ(positive cases). We tested whether organizing an icon array to place the two sides of this ratio nearer each other (proximal condition) enhanced risk assessments compared with an organization separating them (distal condition). Organizing icons proximally improved Bayesian reasoning by 31-43% across two experiments. Participants were more likely to use the simple ratio in the proximal condition, but they used a different, less accurate ratio to determine risk in the distal condition. Current guidelines for risk communication endorse icon arrays but do not provide detailed design instructions. The present research demonstrates the importance of array organization and, in particular, proximity for communicating risk and facilitating Bayesian reasoning.
Stored only when both papers are in the corpus — 6 citations exist upstream.