The Effect of Visualization on Students’ Understanding of Bayesian Inference and Corresponding Probability Concepts in an Innovative Learning Module

Bayesian reasoning - the optimal process of updating a hypothesis or belief with the acquisition of new information - is a critical aspect of both everyday decision-making and statistics education, but strategies for effectively teaching the topic in the classroom remain elusive. This study leverages the findings of prior research on facilitating Bayesian reasoning by utilizing a visualization, called the bar display, as a method for teaching Bayes theorem and its underlying probability concepts. Data were collected from a university’s statistics in psychology course, wherein students were taught and tested on Bayesian reasoning either with (visual condition) or without (nonvisual condition) the bar display. In addition to testing the immediate efficacy of the bar display, data was also collected to test long-term retention and the potential benefits for low numeracy/high anxiety students. Students were also recruited for qualitative interview sessions that allowed for further exploration of solving strategies. Results indicated engagement with the bar display as a method for visually approximating answers to Bayesian question, and students in the visual condition provided more accurate answers to Bayesian reasoning questions before training and at long-term assessment. Evidence from the interviews uncovered a theme of false equivalency between joint and conditional probabilities that the bar display was ill-equipped to address. Recommendations for future work are discussed, including the development of an interactive bar display for use in classroom curriculums that may further facilitate students’ understanding of Bayesian concepts.