Visually Communicating Bayesian Statistics to Laypersons

Abstract

Effectively communicating Bayesian statistics to laypersons has been an open challenge for many years. Recent research in psychology proposed that there is a direct correlation between comprehension and representation. Specifically, a series of studies suggests that pictorial representations with icon arrays may be better suited for communicating Bayesian statistics than Euler diagrams. Though these results are compelling, the experiments were conducted in controlled lab settings and with limited samples. In this paper, we extend the previous research by expanding the sample to a more diverse population through crowdsourcing. We conducted a user study that compares three different pictorial representations of Bayesian statistics icon arrays, Euler diagrams and discretized Euler diagrams. Our findings fail to replicate previous results and demonstrate no significant difference between the three representations. We discuss possible explanations for these findings and propose directions for future investigations.

Extracted Key Phrases

4 Figures and Tables

Cite this paper

@inproceedings{Ottley2012VisuallyCB, title={Visually Communicating Bayesian Statistics to Laypersons}, author={Alvitta Ottley and Barbara Biesecker and Remco Chang}, year={2012} }