How Much Disagreement is Good for Democratic Deliberation? The CaliforniaSpeaks Health Care Reform Experiment

Abstract

Are we the kind of creatures who are suited to govern ourselves through deliberation? We seek to answer one important component of this question: how do individuals respond to deliberation in groups with varying levels of disagreement? We use a natural experiment in which approximately 3000 individuals were divided into small groups composed of about 8-10 persons. These groups deliberated for one day about health care reform in California. We demonstrate that there is a non-monotonic effect of disagreement upon deliberative quality. Elements of deliberative quality include mutual respect, understanding, proferring of reasons and arguments, equal opportunity for discursive engagement, and neutrality. Deliberative quality is maximized at moderate levels of disagreement and lower at high levels of ideological agreement or disagreement. Furthermore, individuals exhibit higher levels of persuasion in deliberative contexts of moderate disagreement. These findings support the view that many individuals have elements of a political psychology that is well suited for deliberation. They do not recoil when encountering disagreement nor do they especially value deliberating with those who see the world in very similar ways. Instead, they regard as most successful deliberations with moderate levels of difference -- perhaps those in which they were acquire new information, perspectives, or reasons. Beyond our substantive finding, this paper offers a methodological template for experimental studies of deliberation.

Keywords: Deliberative Experiment, Homophily, Stochastic Treatment, Ideal Points, Latent Variable Models

JEL Classification: C11, C93