Deliberation is often presumed to affect policy attitudes through learning. But deliberators are also more demographically and attitudinally diverse than everyday discussion partners. They encounter people and perspectives they may not often (or ever) face back home. This paper weighs the extent to which deliberation-induced attitude change stems from increased diversity versus learning. Our data come from the Europolis DP (2009), based on a probability sample from all EU member states.
Focusing on immigration policy, we estimate (hierarchical) models for post-deliberation attitudes as a (nonadditive) function of post-deliberation knowledge (proxying learning) and pre-deliberation diversity, among other variables. We then use the model and estimates to simulate counterfactual distributions of post-deliberation attitudes for full information and back-home diversity, among other scenarios. In brief, learning and diversity seem to work in tandem. Surprisingly, however, full information and DP-level diversity seem to make Europeans favor more, not less restrictive immigration policies.
