idealstan: A Package for Bayesian Ideal-Point Modeling with Stan
Abstract
At StanCon 2018 I will present my new R package for ideal-point modeling with the Stan Markov Chain Monte Carlo sampler for Bayesian inference. This package allows for clustering of data around polarized axes to understand how and why people polarize themselves, whether in terms of political opinions or market products. I illustrate the package with applications drawn from the U.S. Senate and from Amazon product reviews of coffee products. I show how ideal point models can help us understand who the most polarized Senators and coffee products are, and also how the Bayesian version of the model adds powerful new features, including censoring for missing data.
Robert Kubinec
Assistant Professor of Political Science
My research centers on political-economic issues such as corruption, economic development, and business-state relations in developing countries, and in particular the Middle East and North Africa. I am also involved in the development of Bayesian statistical models with Stan for hard-to-study subjects like corruption, polarization, and other latent social constructs.