Download CanVotersbeEqualPart2.pdf

Contents : Review of Sociology Vol. 10 (2004) 1. 47 65 CAN VOTERS BE EQUAL A CROSS-NATIONAL ANALYSIS* Part 2. G bor T KA Central European University Budapest N dor u. 9. H-1051 tokag@ceu.hu Abstract: The paper empirically tests the proposition that because of the unequal social distribution of politically relevant resources some groups of citizens may be less successful in expressing their specifically political preferences in the vote than others. Hence the electoral arena may give different people different degrees of political influence even when the formal equality of all citizens before the law is rigorously upheld in the electoral process. The first part of the paper (published in the previous issue of this journal) explored the assumptions behind the proposition itself and the further assumptions that need to be made in order to test it empirically. The second part of the paper offers an empirical test. Survey data on voting behavior in 18 democratic party systems from the Comparative Study of Electoral Systems and Larry Bartels s (1996) simulation procedure now extended to the analysis of multiparty-systems turnout effects and non-linear information effects on the vote are utilized to explore the question. The results show that social differences in both turnout and political knowledge may lead to the hypothesized political inequalities but their size is remarkably modest. Keywords: electoral behavior level of information turnout effects MODELING INFORMATION EFFECTS Obviously not all social groups are equal in their propensity to participate in elections and to be knowledgeable about relevant political facts. The first part of this paper showed what kind of assumptions and data are needed for an analysis of whether this simple fact makes the formal equality of citizens in the electoral arena illusionary. This section shows how one can model empirically the relationship between vote * The research for this paper was completed while the author held a fellowship awarded by the Training and Mobility of Researchers Programme of the European Community at the European Union Social Science Information Research Facility branch located at the British Library of Economic and Political Sciences. A previous version of this paper was published as Voter Inequality Turnout and Information Effects in a Cross-National Perspective. Helen Kellogg Institute Working Paper Series No. 297. Notre Dame IN: The Helen Kellogg Institute for International Studies at the University of Notre Dame. The helpful comments and suggestions of R bert Tardos Lor nd Ambrus-Lakatos and several anonymous reviewers are gratefully acknowledged. 1417-8648/$ 20.00 2004 Akad miai Kiad Budapest 48 G BOR T KA choices and political knowledge so that in the next section we can proceed to an analysis of turnout- and information effects on election outcomes. Political knowledge may affect vote choice in two ways: in addition to or in interaction with the impact of other variables. For instance knowledge of a financial wrongdoing by a party may make everyone who knows about it less likely to vote for the party responsible: this is an additive information effect. But it may also happen that extra information only influences the behavior of some groups but not others or that additional information moves the vote choices of different people in opposite directions depending on the dominant tendency of their political predispositions. To capture such interaction effects Bartels (1996) simulation procedure modeled vote choice as a function of interactions between sociodemographic characteristics and political knowledge measured on a scale running from 0 to 1 and henceforth abbreviated with the INFO variable name. Recall that the choice of independent variables in the vote function follows the methodological assumptions outlined in the first part of this paper and does in no way imply a belief in a sociological model of vote choice i.e. in a particularly and universally strong impact of sociodemographic variables on vote choice. The first set of the interaction terms in Bartels model consisted of the pairwise products of political information level (INFO) with politically relevant sociodemographic variables (income gender and so forth). The second consisted of the pairwise products of the same socio-demographic variables with (1 INFO). Bartels s probit analysis yielded two constants and two sets of parameter estimates showing the impact of each interaction term on vote choice. Voters who score .4 on INFO can be conceived of as the mix of a maximally informed and a maximally uninformed voter: as 40 percent of the former and 60 percent of the latter. Thus gender s impact on vote choice at INFO .4 can be obtained as .4 times the estimated effect of FEMALE*INFO plus .6 times the estimated effect of FEMALE*(1-INFO) on vote choice. The respondents are conceived as this kind of split personali

INFO HASH : ec9495ce94787b422ccd513611ba2ab727249943

Download now

File Size: 125.6 kb