Elite polarization, group memberships, and sorting
Abstract: While scholars debate over the existence of mass polarization, most accept that the correlation between partisanship and ideology has increased in response to changes in elite polarization. This manuscript argues, however, that this logic is seriously flawed in two important ways. First, the type of polarization matters. Exposure to policy-based polarization has virtually no effect on the extent to which individuals’ political identities have converged; instead, symbolic elite cues are the primary antecedent of sorting. Second, because perceptions of elite cues are inherently shaped by group (partisan) memberships, I find that American citizens have not sorted because they think the political parties are diametrically opposed; rather, this sorting is almost wholly a function of perceptions of out-group extremity and dissimilarity. These results fundamentally reshape the logic of sorting, reinforcing the significance of both the in-group / out-group paradigm and the power of social identities.
Relative to their private sector brethren, do bureaucrats vote for parties that promise to expand the public sector? Prior research indicates as much in the United States, but results are highly mixed in the comparative context. No prior research has theoretically accounted for cross-national variation in this relationship. Here, we focus on actual rhetoric in parties’ platforms aimed at expanding the public sector. Given the mutability of “left” and “right” labels across countries, we argue this is a more appropriate approach than the standard left-right dichotomy. We advance a collective action explanation for why bureaucrats differ from private employees in some countries, but not others. We find that the size of public sector employment substantially mediates these individuals’ proclivity to identify as bureaucrats and, subsequently, cast ballots in line with that identity. We test our theory on more than 31,000 survey respondents from 51 elections across 25 countries.