Tempering Extremism: A Longitudinal Analysis of Polarizing Language Surrounding November 2012

In a two-party system, the median voter model explains how individuals seeking election maximize their potential vote share by presenting themselves to the electorate as moderates. It does not explain, though, why we have observed a rightward shift in the American median voter over the past couple of decades. There are a number of possible reasons for this shift, and we add to this list by tapping the median voter model. We predict that during lulls in the election cycle, politicians are more extreme in their behavior meaning that when elections come around and officials exhibit more moderate behavior, the median voter has already shifted. To test whether politicians’ behavior between elections is more extreme than their behavior during campaigns, we use data from their social media activity. We analyze nearly 290,000 Twitter postings by 522 politicians’ accounts from November 2011 to July 2013 using automated classifiers to identify their polarizing behaviors and compare their behaviors longitudinally. This paper largely confirms our expectations. There is evidence of extreme behavior decreasing from November 2011 to November 2012 and increasing afterwards. We also see that Republicans are more extreme in their rhetoric than Democrats, offering an explanation for why the median voter shifts right and not left.