Simon Ellerbrock, Richard Traunmüller, Christopher Claassen
Estimating the Opinion of Religious Minority Groups Using Bayesian Multilevel Models with Poststratification

9th Annual Conference of the European Political Science Association, Belfast, June 20th to June 22nd, 2019

Europe is becoming more religiously diverse and the visibility of religious minorities has increased significantly in recent decades. However, our knowledge about religious minorities and their attitudes is very limited. We propose to apply Multilevel Regression with Poststratification, or MRP, to accurately measure attitudes of religious minorities over time and across countries. We validate this method by using MRP to predict political trust of Muslims in the UK based on cross-national survey data and subsequently comparing the estimates to large-sample UK datasets. The predictions from the most accurate models on average differed from the 'true' values by as little as 2.5 percentage points. We further demonstrate that even with very little information on religious minorities, MRP enables us to approximate their attitudes. The ability to estimate attitudes of small religious groups from openly available cross-national survey datasets can lay the ground for a better informed discussion on integration of religious minorities into European societies.