| Many democratic theorists argue that citizens who deliberate form enlightened and consensual opinions. Too often, scholars who value deliberation do not adequately explain how it affects individual-level opinions. I provide a theory of deliberative opinion updating based upon Bayesian learning whereby citizens revise their prior beliefs with new information obtained through discussion. This theory focuses analytical attention on both the overall opinions that citizens report to survey researchers as well as the distributions of considerations underlying their individual attitudes. Empirical analyses using two-stage conditional maximum likelihood models demonstrate support for three out of four research hypotheses generated from this theory: (1) significant opinion change on Social Security reform and spending preferences for deliberators with diffuse prior opinions, (2) changes in considerations mirroring the actual content of deliberation, and (3) reductions in uncertainty as deliberators gain information. Unexpectedly, ambivalence did not significantly increase opinion variability as it should when two opposing considerations become salient. These results suggest that public discussion in an organized deliberative forum or in ordinary situations can increase knowledge and alter opinions, but it does so selectively based upon the quality and diversity of the deliberation. |
