Specification Issues in Proximity Models of Candidate Evaluation (with Issue Importance)

Abstract

The use of the proximity model to represent the relationship between citizens' policy attitudes and the positions of candidates on the issues of the day has considerable appeal because it offers a bridge between theoretical models of political behavior and empirical work. However, there is little consensus among applied researchers about the appropriate representation of voter behavior with respect to the measurement of issue distance, candidate location, or whether to allow heterogeneity in the weight that each individual places on particular issues. Each of these choices suggests a different, and reasonably complicated, nonlinear relationship between voter utility and candidate and voter issue positions which may have a meaningful influence on the substantive conclusions drawn by the researcher. Yet, little attention has been given to the best way to represent the proximity model in applied work. The purpose of this paper is to identify which forms of the proximity model work best, with particular consideration given to the identification of functional forms that are invariant to the choice of scale for the independent variables.

Benchmarking across Borders: Electoral Accountability and the Necessity of Comparison

“…We used the respondent placements to validate the expert placements and found that mean respondent placements and expert placements were highly correlated. 22 Although quadratic distance has often been used in the literature, Grynaviski and Corrigan (2006) report that the best fit for policy distance is provided by a linear distance measure. between these variables and individual-level characteristics may uncover important demographic effects in the patterns of support for certain types of parties.…”

Political knowledge and campaign effects in the 2008 Irish referendum on the Lisbon Treaty

“…We make a clear distinction between the perception that the Lisbon Treaty does indeed imply future increases in tax rates and the evaluating attitude the respondent has towards the idea of such increase. 2 There is an extensive literature in the US context outlining the conditional and nonlinear nature of issue voting (e.g., Grynaviski and Corrigan, 2006) and the impact of political knowledge therein (Campbell et al, 1960;Nie et al, 1979;Palfrey and Poole, 1987;Krosnick, 1988;Zaller, 1992;Alvarez, 1997;Basinger and Lavine, 2005;De Vries et al, 2011).…”

Evaluating Voter–Candidate Proximity in a Non-Euclidean Space

“…By contrast, the Euclidean distance minimizes the impact of small disagreements across a range of issues. But the Euclidean distance -particularly when squared -imposes a large penalty for substantial disagreement on a single dimension (Grynaviski & Corrigan, 2006). Although the squared Euclidean distance will always produce the same preference ordering as the Euclidean distance, when used in models that have both additive and multiplicative terms (e.g.…”

“…Finally, it should be noted that while some proximity models test the additive impact of multiple distances along individual policy dimensions (e.g. Macdonald et al, 1998;Grynaviski & Corrigan, 2006), our model tests the distance in a real spatial context. Consider the utility functions specified in Equations [1] and [2].…”

“…In a multidimensional space, however, there are potentially infinite notions of distance and the meaning of spatial distance is not as self-evident as we usually assume. Except for a few brief discussions (Ordeshook, 1986;Westholm, 1997;Grynaviski & Corrigan, 2006), the choice of distance metric has received scant attention. In this study, we contend that various distance metrics reflect different ways whereby voters "penalize" candidates for their policy disagreements.…”