The proximity and the directional theories of issue voting: Comparative results for the USA and Germany

In several recent studies George Rabinowitz and his co–authors challenge the 'classical' spatial model of issue voting, the proximity model, by introducing a directional model. In this article we examine whether different measurement of perceived issue positions of candidates or parties leads to diverging judgments about the predictive power of the directional model (which is claimed to be empirically superior), as compared to the proximity model, using data from the USA and Germany. The results demonstrate that the measurement preferred by Rabinowitz et al. tends to bias empirical findings in favour of directional theory. If we use a more plausible operational definition of issue positions of candidates and parties the directional model in both countries fails to turn out superior.     

Jenseits von Distanz und Richtung. Die Verbundmessung von Politikpräferenzen im Vergleich mit dem Distanz- und dem Richtungsmodell des Wählens — Empirische Befunde eines Methodenexperiments

The analysis of issue voting is of central importance in empirical electoral research. In this context, policy positions of voters and parties are normally surveyed by means of bipolar policy scales developed within the spatial model of voting. Even if such policy scales are widespread, their use can be criticised for a series of theoretical and methodological reasons. Beyond this background, conjoint measurement of policy preferences is presented as an alternative mode of operationalization. Using data from a methodological experiment it is shown that (full profile) conjoint measurement is superior to the use of traditional policy scales in the proximity and the directional model of voting.     

Party policy equilibrium for alternative spatial voting models: An application to the Norwegian storting

In recent years a lively debate has emerged concerning the empirical status of the traditional proximity spatial model versus a directional model of voter choice. The central reason for this scholarly interest concerns these models' contrasting implications for parties' policy positioning, with the directional model motivating parties to present extreme policies, but the proximity model promoting centrist positions. To this point, however, there exist no studies that compute parties' optimal strategies in historical elections, for these competing models. This article addresses this issue, by examining party policy strategies in a multiparty electorate for three different vote models: (a) the proximity model, (b) a directional model (c) a mixed model which combines proximity and directional components. Each model incorporates past voting history and the random effects of unmeasured variables. Using parameter estimates derived from analyses of survey data from the 1989 Norwegian Election Study we compute — for each of these vote models — the configuration of party policy positions that maximize each party's vote share in relation to those of the other parties. We find that for each model, such a vote–maximizing configuration exists, but — for the proximity model — represents an unrealistic, tightly clustered array. A mixed proximity–directional model, however, provides by far the most convincing account of parties' actual policy strategies with regard to dispersion and vote share.     

Party policy equilibrium for alternative spatial voting models: An application to the Norwegian Storting

Abstract In recent years a lively debate has emerged concerning the empirical status of the traditional proximity spatial model versus a directional model of voter choice. The central reason for this scholarly interest concerns these models' contrasting implications for parties' policy positioning, with the directional model motivating parties to present extreme policies, but the proximity model promoting centrist positions. To this point, however, there exist no studies that compute parties' optimal strategies in historical elections, for these competing models. This article addresses this issue, by examining party policy strategies in a multiparty electorate for three different vote models: (a) the proximity model, (b) a directional model (c) a mixed model which combines proximity and directional components. Each model incorporates past voting history and the random effects of unmeasured variables. Using parameter estimates derived from analyses of survey data from the 1989 Norwegian Election Study we compute — for each of these vote models — the configuration of party policy positions that maximize each party's vote share in relation to those of the other parties. We find that for each model, such a vote–maximizing configuration exists, but — for the proximity model — represents an unrealistic, tightly clustered array. A mixed proximity–directional model, however, provides by far the most convincing account of parties' actual policy strategies with regard to dispersion and vote share.     

A directional model of candidate evaluations in the 2016 presidential election

The 2016 U.S. presidential election provides an interesting setting for testing the directional theory of issue voting. But, when modeled using the standard approach, using the seven-point issue scales in the 2016 ANES, it is not clear whether the directional model provides any additional leverage over the more traditional proximity model. In order to get around this impasse, I examine candidate evaluations rather than issue scales. Doing so enables a direct test of directional theory against proximity theory. The empirical results show that the proximity model does not outperform the directional model. Instead, the latter outperforms the former because the directional model produces not only a succinct graphical representation of the electorate's candidate evaluations but also an explanation for the relative spatial positions of the candidates.     

Ideology and candidate evaluation

How does the expressed political ideology of voters influence their evaluation of presidential candidates? The classic answer to this question is provided by the spatial theory of electoral choice in which utility for a candidate is a function of the proximity between the voter and candidate positions on the liberal-conservative continuum. We have argued elsewhere that spatial theory, while intellectually appealing, is inadequate as an empirical model of mass behavior. We have developed a directional theory of issue voting that we believe provides a more realistic accounting of how specific policy issues influence utility for a candidate. Directional theory is based on the view that for most voters issues are understood as a dichotomous choice between two alternative positions. While ideology is widely understood as a continuum of positions, the directional model can be applied to the relationship between ideology and candidate evaluation. In this paper we compare the two theories using National Election Study data from 1972 to 1988. The results tend to favor the directional model over the traditional proximity model. We conclude by briefly tracing out the implications of this finding.     

Alternative models of issue voting: The case of the 1991 and 1995 elections in Belgium

Abstract. Previous research has indicated that the success of the directional model of issue voting depends on levels of political sophistication and how party position is measured. Using 1991 and 1995 Belgian Election Surveys, the predictive power of proximity and directional measures are compared controlling for both variables. It is shown that when one uses overall mean placements, instead of mean placements by level of political sophistication, the proximity effect declines most among the highly sophisticated voters. The article also compares the performance of the proximity and directional measures across party systems. Contrary to theoretical expectations, party–system differences between Flanders and Wallonia do not affect the explanatory power of either of these measures. It is only in the cases of the liberal, socialist and extreme right parties that the directional measure is clearly superior. A closer analysis of this result indicates that the relative success of the directional measure is due to the limited number of issues from which those parties draw support.     

Conceptualizing voter choice for directional and discounting models of two-candidate spatial competition in terms of shadow candidates

In contrast to the traditional modeling of voter choice based on proximity, under directional models, selection of candidates is based on the direction and/or intensity of change from a status quo or neutral point. Voter choice can also be modeled as representing both approaches, e.g., as a directional model with proximity restraint, or alternatively, in terms of proximity to discounted positions. We provide a unified perspective for these seemingly disparate models in terms of what we call “shadow” positions. We demonstrate that voter choice in a variety of spatial models including directional components can be viewed as proximity-based choices. Voters choose the candidate whose shadow is nearer, where shadow locations are defined by a simple transformation. We apply this approach to equilibrium analysis, showing that results for a discounted proximity model can be carried over – via shadows – to a variety of directional models.     

Direction Versus Proximity in the Social Influence Process

Rabinowitz and Macdonald (Am Polit Sci Rev 83(1):93–121, 1989) have advanced a directional theory of electoral choice that stands as an alternative to the proximity models that have dominated thinking in this area for a number of years. In this paper, we assess the utility of directional theory in another area of political behavior: the evaluation and influence of politically significant others in an individual’s social environment. Using two datasets collected during presidential election campaigns in 1984 and 1996, we find that respondents are more likely to evaluate their political discussants highly and be influenced by discussant vote choice if they agree in a directional rather than proximity manner. In looking at agreement on party identification, ideology, and issue positions, the directional model prevailed in 11 of 17 estimations, with neither explanation acquiring empirical support in the other six. In no instance did the proximity model prevail as an explanation of how political discussants relate to each other. We conclude by discussing the consequences of these results for political behavior and practical electoral politics.

Issues,the Spatial Theory of Voting,and British General Elections: A Comparison of Proximity and Directional Models

Competing spatial models of voter choiceare compared in the context ofparliamentary representatives selectedthrough single-member district, pluralityelections where party platforms areemphasized over individual candidates.Respondents of the 1987, 1992, and 1997British general election surveys ratepolitical parties on a series of issuescales. Ordered logistic regressions ofparty evaluations under proximity,directional, and mixed models reveal thatthe classic spatial model and thedirectional model perform equally well.Differences center on perceptions of thestatus quo, as voters appear to evaluatethe incumbent party (here, theConservatives) slightly differently thanminority parties (Labour and the LiberalDemocrats). The proximity model worksbetter for voter evaluations of governingparties while the directional model workswell for opposition parties.     

Ideology and party support in comparative perspective

Abstract. From the original Downs (1957) conceptualization to the Enelow and Hinich (1984) reformulation, ideology has been the key organizing dimension for electoral competition in the spatial model. The unidimensional spatial model is best suited to responsible multiparty systems where left-right position is known to be an important determinant of individual political behaviour. We contrast the traditional spatial model with the directional model recently proposed by Rabinowitz and Macdonald (1989) and test the models using data from six northern European democracies. Based on 57 different parties, the results suggest that even in this unidimensional ideological context the cleavage-oriented directional model provides a better explanation of mass attachments to political parties.     

Congruence between voters and parties: The role of party-level issue salience

The level of congruence between parties and their voters can vary greatly from one policy issue to another, which raises questions regarding the effectiveness of political representation. We seek to explain variation in party–voter congruence across issues and parties. We focus on the hypotheses that (1) average proximity between the positions of voters and the party they vote for will be highest on the issues that the party emphasises in the election campaign and that (2) this relationship will be stronger for niche parties. We test these hypotheses using data on the policy preferences of voters, party positions, party attention profiles and salience on concrete policy issues in four countries: The Netherlands, Ireland, Germany and Sweden. Overall, we find that voter–party proximity tends to be higher on issues that the party emphasises. As these are the issues where parties typically have the greatest policy impact, this implies that the quality of representation is highest where it matters most. There is some limited evidence that the positive relationship between issue salience and proximity is stronger for niche parties. In sum, the quality of policy representation varies strongly with party-level issue salience and to a lesser extent with the type of political party.     

The formation of party preferences: Testing the proximity and directional models

We review the methodological debate between defenders of the proximity and directional models. We propose what we believe to be a rigorous and fair test of the two models, using the 1997 Canadian Election Study. The analysis is based on responses to questions in which the various issue positions are explicitly spelled out. We rely on individual perceptions of party positions because it is individual perceptions that matter in the formation of party preferences but we control for projection effects through a multivariate model that incorporates, in addition to indicators of distance and direction, socio–demographic characteristics, party identification, and leader ratings. We also take into account whether a party is perceived to be extreme. The empirical evidence vindicates the proximity model.     

The formation of party preferences: Testing the proximity and directional models

Abstract. We review the methodological debate between defenders of the proximity and directional models. We propose what we believe to be a rigorous and fair test of the two models, using the 1997 Canadian Election Study. The analysis is based on responses to questions in which the various issue positions are explicitly spelled out. We rely on individual perceptions of party positions because it is individual perceptions that matter in the formation of party preferences but we control for projection effects through a multivariate model that incorporates, in addition to indicators of distance and direction, socio–demographic characteristics, party identification, and leader ratings. We also take into account whether a party is perceived to be extreme. The empirical evidence vindicates the proximity model.     

Testing proximity versus directional voting using experiments

A long-running debate about how voters use issues to evaluate candidates pits the proximity theory of voting against directional theory. Using surveys, both sides of the debate have found support for their preferred theory, but disagreement remains because of differing ways of analyzing the data. Lewis and King (2000) point out that these researchers make assumptions that bias results in favor of their theory. To avoid these difficulties, our approach creates fictitious candidates with controlled positions, presents these candidates to randomly-assigned subjects, and examines the relationship between subjects’ evaluations of these candidates and their ideological beliefs as a neutral test of proximity and directional theory. Our results provide reasonably strong support for proximity theory but little for directional theory.     

Estimating relative prevalence of competing issue voting rules in multiparty elections using finite mixture modeling

How do citizens of democratic polities translate their policy preferences into voting choices? Proximity and directional theories of issue voting offer different answers to this question that have strong implications for parties’ strategies. Controlled scenarios in imaginary two-candidate contests have gained popularity as a method to estimate proportions of proximity and directional voters in the population. However, this method is not always applicable in comparative research where scholars often have to study multiparty elections with observational data. In the present paper, I demonstrate how relative prevalence of different issue voting rules can be estimated using finite mixture modeling. Using both a paradigmatic and a recent case, I demonstrate that the mixture model predicts observed voting choices better than the alternatives. I also show how finite mixture modeling can be used to study individual-level characteristics of proximity and directional voters using education as an example.     

Polarization and ideological congruence between parties and supporters in Europe

The relationship between parties and their supporters is central to democracy and ideological representation is among the most important of these linkages. We conduct an investigation of party-supporter congruence in Europe with emphasis on the measurement of ideology and focusing on the role of party system polarization, both as a direct factor in explaining congruence and in modifying the effects of voter sophistication. Understanding this relationship depends in part on how the ideology of parties and supporters is measured. We use Poole’s Blackbox scaling to derive a measure of latent ideology from voter and expert responses to issue scale questions and compare this to a measure based on left–right perceptions. We then examine how variation in the proximity between parties ideological positions and those of their supporters is affected by the polarization of the party system and how this relationship interacts with political sophistication. With the latent ideology measure, we find that polarization decreases party-supporter congruence but increases the effects of respondent education level on congruence. However, we do not find these relationships using the left–right perceptual measure. Our findings underscore important differences between perceptions of left–right labels and the ideological constraint underlying issue positions.     

An assessment of voting systems under the proximity and directional models of the vote

I evaluate five single-winner voting systems according to their tendency to elect Condorcet candidates under alternative models of issue voting derived from behavioral research. These behavioral models posit that voters have both issue and nonissue motivations; within this framework, I study the effects of both the directional and proximity voting models, with varying degrees of issue voting. Under the proximity metric, all voting systems are most efficient when voters attach little importance to issues, while the opposite is generally the case under directional voting. In contrast to previous results, voting systems tend to be more efficient for large than for small electorates. All voting systems – including the widely-criticized plurality method – are extremely efficient when voters in mass elections are inattentive to issues.     

Assimilation and contrast effects in voter projections of party locations: Evidence from Norway, France, and the USA

In the standard Downsian model, voters are assumed to choose parties based on the extent of ideological proximity between the voter's own position and that of the party. Yet it is also well known that there are rationalization and projection effects such that voters tend to misestimate the policy platforms of candidates or parties to which they are sympathetic by overstating the correspondence between those positions and the voter's own preferences (see, e.g., Markus & Converse 1979; Granberg & Brent 1980; Granberg & Holmberg 1988; Merrill & Grofman 1999). Here we follow insights in the psychological literature on persuasion (Sherif & Hovland 1961; Parducci & Marshall 1962) by distinguishing between assimilation and contrast effects. Assimilation refers to shortening the perceived ideological distance between oneself and parties one favors; contrast refers to exaggerating the distance to parties for which one does not intend to vote. Using survey data on voter self–placements and party placements on ideological scales for the seven major Norwegian parties, five major French parties, and two major American parties we show that both assimilation and contrast effects are present in each country to a considerable degree.We also investigate the possible effects of randomness in party placement and scale interpretation – effects that can easily be confounded with assimilation but not so easily with contrast.