Reappraising Volatility and Nationalization: A Voter-centric, Multi-level Theory

• In Event: Political Preferences and Public Opinion in Comparative Perspective

Abstract

What accounts for vote choices and party affinities in inchoate party systems? We approach this question with a theory rooted in voters as the unit of analysis, their party choices as the outcome, and their spatial and temporal locations as explanations for their preferences. The theory--a voter-centric, multi-level theory of party performance--is about the forces that structure voting behavior and party affiliation. It is a "thin" theory in the sense that a voter's location in a given place, at a given time, is only a proxy for thicker accounts of their choices. For example, spatial location may stand-in for a social identity (e.g. ethnic or racial group), and temporal location may stand-in for a particular feature during a given election (e.g. the incumbency status of the candidate). We make efforts to enrich these explanations when the data allow by considering a variety of observable implications of the theory.

The theory is novel because it reorients the theoretical foundations of many works in the comparative parties literature. As we will explain below, it addresses two fundamental and ubiquitous errors of attribution: one about the unit of analysis, and the other about the level of analysis. The first error arises when research using aggregate electoral data--especially studies of electoral volatility and party nationalization--ascribes attributes to parties that are instead explanations of voting behavior. For example, projects sometimes declare that parties are poorly nationalized, when in fact it is voters and their preferences that are heterogeneous. The second error of attribution arises when scholars ascribe explanatory power to forces acting at one level of analysis when they are instead operating at a different one. This error can operate between spatial levels, as for example by confusing regional influences with district influences, or between spatial and temporal levels, as when it appears that spatial differences shape vote choice when in fact it is temporally-varying forces that cause voters in different physical locations to behave differently from each other. In sum, our theory argues against the party-centric and single-level treatments of party affiliation or vote choice.

The errors of attribution undermine theoretical and empirical research across many settings, but they are most problematic in so-called inchoate party systems. In the most stable party systems, voting behavior is tolerant of a wide variety of analytic approaches because it is regular and easy to account for. Where politics is placid, quick and dirty measures of volatility and nationalization perform just fine. However, to the extent that voting behavior is unstable, the errors of attribution cut ever more deeply into the validity of the conclusions. At the extreme, in inchoate party systems, party-centric and single-level approaches to politics are most flawed, and our theory offers the largest payoff.

For the sake of argument, we use one of the most classic cases of instability as the setting to examine the theory and its implications. Political behavior in Ecuador has been subjected to a wide and incompatible set of characterizations. It has been designated one of the most volatile, populist, regionalized, localized, and inchoate party systems in the region and the world. These are logically contradictory conclusions. If behavior is hyper volatile, then region can not play much of a role. If behavior is dominated by regional difference, then it can not also be dominated by local forces. We examine actors in Ecuador because they are difficult to account for. Overall, while our goal is to develop a theory with range, we also bring a degree of order to the empirical confusion about the case.

After developing our theory, we identify a set of observable implications and adopt two empirical strategies. First, we seek leverage to understand vote choice using aggregate election results. Second, we trace the patterns of party affiliation of citizens who run for political office. As a general matter, we will show how the forces operating on the electorate can be decomposed to their appropriate level of analysis. Moreover, for the case of Ecuador, we will show 1) that voters are overwhelmingly shaped by short-term forces (i.e. arising from the election cycle); 2) that region-level forces barely operate; and 3) that province-level (i.e. district-level) forces are much weaker than many projects have claimed. The big story in Ecuador is that the forces causing volatility are stronger, and spatial identities are weaker, than the conventional wisdom would have us believe.

Authors

• Thomas Mustillo, University of Notre Dame

• John Polga-Hecimovich, U.S. Naval Academy