Why Keep Arguing: A Bayesian Model for Predicting Engagement in Political Conversations Online

• In Event: JSS SESSION: Ideal Point Estimation

Brief Overview

We develop a Bayesian model of argument engagement using a new dataset of conversations about President Trump. The model incorporates user, tweet, and thread-level features to infer user ideology and predict user participation with over 98% accuracy.

Authors

• Sarah Shugars, Northeastern University

• Nick Beauchamp, Northeastern University