“…Comparative studies, both country case studies and large-N multicountry analyses, have identified a number of variables, including public opinion (Ward, Ezrow, and Dorussen, 2011;Adams et al, 2004), parties' organizational attributes (Schumacher, De Vries, and Vis, 2013), rival parties' policy positions (Adams and Somer-Topcu, 2009;Han, 2015;Spoon, 2011;Williams, 2015), the preferences of the affluent (Gilens and Page, 2014), the preferences of the party's core supporters , and past election results (Laver, 2005;Somer-Topcu, 2009). Investigations of party and candidate polarization in U.S. states yield still more predictive variables, including legislative professionalism (Squire, 2017), population size, divided government, unemployment, population size, term limits (Olson and Rogowski, 2018), economic inequality, and the proportion of the population that is foreign born (McCarty, Poole, and Rosenthal, 2006). While the subnational method does not totally eliminate variation in these other sources of polarization, it significantly mitigates them, which makes identifying the relationship between institutional variation and candidate behavior more feasible.…”