Partisan polarization is one of the principal features of contemporary US politics. This paper is designed to address the extent to which the partisan homogeneity of networks is related to, and may even contribute, to it. Drawing upon a national post-election survey of the US electorate in 2012 and its discussion network battery of questions developed in the Comparative National Election Project (CNEP), the paper will examine the relationship between network partisan homogeneity and the partisanship of voters. It measures network partisanship from respondent reports in two ways: as the candidate favored by generic types of discussants (family, friends, neighbors, and co-workers) and by more “intimate” networks including spouse/partner (in the two-thirds of cases where they exist) and two additional personal discussants selected using a “most important matters” name generator. It measures voter partisanship by the strength of party identification, the presidential vote, and comparative evaluations of each presidential candidate on a 10-point scale. The network batteries include reports of discussant gender, knowledge, frequency of interaction and relationship to the respondent, which will be analyzed as potential moderators. Similar questions to these were asked in 1992 and 2004 US election studies and in 17 other CNEP countries (from Africa, Asia, Europe, and Latin America) across two decades. This enables us to view the 2012 US analysis from both longitudinal and cross-national perspectives. Our principal hypotheses to be examined from these comparative perspectives are that network homogeneity has become greater in the US and that American networks are among the most homogeneously partisan in the democratic world.
