Belief change: A bayesian perspective

The question of whether citizens are open- or closed-minded is as old as political thought. A crucial aspect of this question is the degree to which citizens are able and willing to revise their prior beliefs. To address this question, we invoke a Bayesian framework at the individual level, deriving an expected posterior belief for each respondent, which then serves as the benchmark against which actual posterior beliefs can be compared. Focusing on general patterns of belief change, we observe two biases: a confirmation bias, in which decision makers interpret evidence that contradicts prior beliefs as if it were neutral or even consistent with those beliefs, and a conservatism bias, in which decision makers adjust their prior beliefs insufficiently in the light of new evidence. These results are of both theoretical and normative significance. Our findings shed new light on the processes that mediate the impact of belief resistance and change.