Politics in Multiparty Context: Multiplicative Specifications, Social Influence, and Electoral Choice

Abstract

Explanations of electoral choice most frequently focus on the isolated characteristics of voters and examine those characteristics using linear, additive formulations. We explore two alternatives. We first argue that in a multiparty electoral context, the structuring influence of ideology on the vote often will be best represented with a multiplicative specification. Second, we contend that the impact of individual-level traits such as ideology will be moderated by the political climate of the voter's social context. We test these alternative formulations using data concerning the 1989 elections in Honduras, where two major parties and two minor parties vied for support. Multinomial logistic regression results demonstrate that a multiplicative and contextual specification substantially outperforms a more conventional modeling strategy. In explaining electoral choice, political scientists have focused considerable attention on characteristics of the individual voter. Voting behavior most frequently is depicted as a consequence of individual traits such as education, occupation, income, ideology, and partisan affiliation (e.g., Campbell et al. 1960; Jackson 1975; Markus and Converse 1979; Nie, Verba, and Petrocik 1976; Page and Jones 1979). When concentrating on characteristics such as these, analysts implicitly introduce two general assumptions to their models of the vote. First, most models of electoral choice posit the existence of additive relationDAMARYS CANACHE is a Ph.D. candidate in political science at the University of Pittsburgh. JEFFERY J. MONDAK is assistant professor in the Department of Political Science, University of Pittsburgh. ANNABELLE CONROY is a Ph.D. candidate in political science at the University of Pittsburgh. They thank Mitchell Seligson, who coordinated the Central American Public Opinion Project. Public Opinion Quarterly Volume 58:509-538 ( 1994 by the American Association for Public Opinion Research All rights reserved. 0033-362X/94/5804-0004$02.50 This content downloaded from 207.46.13.176 on Mon, 20 Jun 2016 06:29:46 UTC All use subject to http://about.jstor.org/terms 510 Damarys Canache, Jeffery J. Mondak, and Annabelle Conroy ships among the variables. Second, explanatory variables typically represent only the isolated demographic characteristics and political predispositions of the individual voter, and only such individual-level traits are considered to be potential determinants of the vote. The pervasiveness of these assumptions throughout the voting literature is easily understood in light of the parsimony of additive models and the relative ease with which individual-level data can be obtained. However, these assumptions also raise significant substantive concerns that are rarely addressed by students of electoral behavior. In this article we demonstrate that multiplicative and contextual explanations of electoral choice can substantially outperform more conventional specifications both normatively and empirically. We will begin by outlining the logic of multiplicative and contextual modeling strategies. These perspectives will then be tested empirically through analysis of electoral choice in a multiparty political context. Specifically, our data are drawn from a survey conducted in Honduras in 1991. The survey includes questions concerning the 1989 Honduran elections, the third elections following a decade of military dictatorship in that country. Unique features of the survey's design will allow us to explore the electoral significance of a mix of individual-level and contextual factors. Additive and Multiplicative Explanations of Electoral