This is the author's version of the work. Please cite as:

Political Geography

Abstract Previous research has demonstrated a significant relationship between the geographical distance from a voter to a candidate and the likelihood of the voter choosing that candidate. However, models of this relationship may be mis- or under-specified, by not taking into account voters' perceptions of distance or not controlling for other possible factors related to a candidate's 'localness' which may influence vote choice. Using a two-wave panel survey carried out during the 2015 \{UK\} General Election, this article tests a more fully specified alternative-specific multinomial probit model of candidate-voter distance. We show that, although the effect size is smaller than in previous tests, candidate-voter distance mattered in the 2015 General Election, an effect that is robust to controls not only for party support and incumbency, as previous research had demonstrated, but also to measures of voter information, candidate presence and marginality. We also find that contiguity mattered: candidates living in non-neighbouring constituencies have a lower likelihood of vote than those living in neighbouring constituencies or in the constituency itself.

@Article{evans-et-al-2017,
author = {Jocelyn Evans and Kai Arzheimer and Rosie Campbell and Philip
Cowley},
title = {Candidate Localness and Voter Choice in the 2015 General Election
in England},
journal = {Political Geography},
year = 2017,
volume = 59,
html = {https://www.kai-arzheimer.com/paper/candidate-localness-voting-england/},
pages = {61-71},
abstract = {Abstract Previous research has demonstrated a significant
relationship between the geographical distance from a voter to a
candidate and the likelihood of the voter choosing that candidate.
However, models of this relationship may be mis- or
under-specified, by not taking into account voters' perceptions of
distance or not controlling for other possible factors related to a
candidate's 'localness' which may influence vote choice. Using a
two-wave panel survey carried out during the 2015 \{UK\} General
Election, this article tests a more fully specified
alternative-specific multinomial probit model of candidate-voter
distance. We show that, although the effect size is smaller than in
previous tests, candidate-voter distance mattered in the 2015
General Election, an effect that is robust to controls not only for
party support and incumbency, as previous research had
demonstrated, but also to measures of voter information, candidate
presence and marginality. We also find that contiguity mattered:
candidates living in non-neighbouring constituencies have a lower
likelihood of vote than those living in neighbouring constituencies
or in the constituency itself.},
doi = {10.1016/j.polgeo.2017.02.009},
}

Candidate localness and voting in England

How important is ‘localness’ – the fact that a candidate lives close to their prospective voters – in General Elections in England? Building on our previous research on General Elections and English County Council elections, we look at the 2015 General Election to answer that question. The final article is published in Political Geography. Here is the abstract:

Previous research has demonstrated a significant relationship between the geographical distance from a voter to a candidate and the likelihood of the voter choosing that candidate. However, models of this relationship may be mis- or under-specified, by not taking into account voters’ perceptions of distance or not controlling for other possible factors related to a candidate’s ‘localness’ which may influence vote choice. Using a two-wave panel survey carried out during the 2015 UK General Election, this article tests a more fully specified alternative-specific multinomial probit model of candidate-voter distance. We show that, although the effect size is smaller than in previous tests, candidate-voter distance mattered in the 2015 General Election, an effect that is robust to controls not only for party support and incumbency, as previous research had demonstrated, but also to measures of voter information, candidate presence and marginality. We also find that contiguity mattered: candidates living in non-neighbouring constituencies have a lower likelihood of vote than those living in neighbouring constituencies or in the constituency itself.

Photo by Matt From London