A game-based intervention for the reduction of statistical cognitive biases

Abstract

Probability and statistics is perhaps the area of mathematics education most directly applicable to everyday life. Yet, the methodologies traditionally used to cover these topics in school render the material formal and difficult to apply. In this thesis, I describe a game design that develops probabilistic concepts in real-life situations. Psychologists have coined the term cognitive bias for instances in which the intuition of the average person disagrees with the formal mathematical analysis of the problem. This thesis examines if a one-hour game-based intervention can enact a change in the intuitive mental models people have for reasoning about probability and uncertainty in real-life. Two cognitive biases were selected for treatment: overconfidence effect and base rate neglect. These two biases represent instances of miscalibrated subjective probabilities and Bayesian inference, respectively. Results of user tests suggest that it is possible to alter probabilistic intuitions, but that attention to the transitions from the current mental constructs must be carefully designed. Prototyping results suggest how some elements of game design may naturally lend themselves to deep learning objectives and heuristics.