Hostname: page-component-8448b6f56d-wq2xx Total loading time: 0 Render date: 2024-04-19T18:30:27.084Z Has data issue: false hasContentIssue false
  • English
  • Français

Reassessing Local Candidate Effects

Published online by Cambridge University Press:  28 March 2022

Semra Sevi Open the ORCID record for Semra Sevi [Opens in a new window] ,
Marco Mendoza Aviña Open the ORCID record for Marco Mendoza Aviña [Opens in a new window]  and
André Blais Open the ORCID record for André Blais [Opens in a new window]
Semra Sevi*
Affiliation:
Department of Political Science, Columbia University, Mail Code 3320, New York, NY 10027, USA
Marco Mendoza Aviña
Affiliation:
Department of Political Science, Université de Montréal, 3150 rue Jean-Brillant, Montréal, QC H3T 1N8, Canada
André Blais
Affiliation:
Department of Political Science, Université de Montréal, 3150 rue Jean-Brillant, Montréal, QC H3T 1N8, Canada
*
*Corresponding author. E-mail: ss6479@columbia.edu
Rights & Permissions [Opens in a new window]

Abstract

In a seminal article published in 2003, Blais et al. demonstrated that local candidates mattered for about 5 per cent of voters in the 2000 Canadian federal election. This study's reliance on a single election raises external validity concerns. We replicate Blais et al.'s original analyses on four elections from 2000 to 2008 using a decade's worth of data from the Canadian Election Study. The local candidate effect first uncovered by Blais et al. is not specific to a single election. Local candidates are a decisive consideration for about 5 to 8 per cent of voters outside Quebec and for about 2 to 5 per cent of voters in Quebec.

Résumé

Résumé

Dans un important article, Blais et al. (2003) démontrent que les candidats locaux ont été une considération importante pour environ 5 % des électeurs lors de l'élection fédérale canadienne de 2000. Le fait que cette étude se base sur une seule élection soulève des enjeux de validité externe. Nous reproduisons les analyses originales de Blais et al. sur quatre élections entre 2000 et 2008 en utilisant une décenie de données de l'Étude électorale canadienne. L'effet du candidat local mis en évidence par Blais et al. n'est pas spécifique à une seule élection. Les candidats locaux sont un facteur décisif pour environ 5 à 8 % des électeurs hors Québec et 2 à 5 % des électeurs au Québec.

Keywords

local candidateselectionsCanadaQuebecreplicationCandidats locauxélectionsCanadaQuébecréplication
Type
Research Note/Note de recherche
Information
Canadian Journal of Political Science/Revue canadienne de science politique , Volume 55 , Issue 2 , June 2022 , pp. 480 - 485
Creative Commons
Creative Common License - CCCreative Common License - BY
This is an Open Access article, distributed under the terms of the Creative Commons Attribution licence (https://creativecommons.org/licenses/by/4.0/), which permits unrestricted re-use, distribution, and reproduction in any medium, provided the original work is properly cited.
Copyright
Copyright © The Author(s), 2022. Published by Cambridge University Press on behalf of the Canadian Political Science Association (l’Association canadienne de science politique) and/et la Société québécoise de science politique

Introduction

The study of candidates in Canadian elections focuses for the most part on aggregate voting patterns and party leaders (Bittner, Reference Bittner2011; Sevi, Reference Sevi2021). The impact of local candidates has largely been overlooked on the grounds that most constituents do not know or care about their local candidates. There have been a few studies showing that local candidates matter, as they make a small but discernible difference in electoral outcomes (Allen Stevens et al., Reference Allen Stevens, Islam, de Geus, Goldberg, McAndrews, Mierke-Zatwarnicki, Loewen and Rubenson2019; Blais et al., Reference Blais, Gidengil, Dobrzynska, Nevitte and Nadeau2003; Blais and Daoust, Reference Blais and Daoust2017; Roy and Alcantara, Reference Roy and Alcantara2015). These studies, however, tend to draw on individual elections; we thus lack a wider picture of local candidates’ influence across elections.

In this research note, we replicate Blais et al.'s (Reference Blais, Gidengil, Dobrzynska, Nevitte and Nadeau2003) article, which found that a preference for a local candidate was a decisive consideration for about 5 per cent of voters in the 2000 Canadian federal election. In what follows, we reproduce the same analysis used by Blais et al. (Reference Blais, Gidengil, Dobrzynska, Nevitte and Nadeau2003) across four Canadian federal elections: 2000, 2004, 2006 and 2008.Footnote 1

Data

Our replication effort is twofold: we replicate Blais et al.'s (Reference Blais, Gidengil, Dobrzynska, Nevitte and Nadeau2003) original analyses for the 2000 Canadian election and we extend them to encompass three additional elections. Like Blais et al. (Reference Blais, Gidengil, Dobrzynska, Nevitte and Nadeau2003), we use the Canadian Election Study (CES) to examine how local candidates affect vote choice, and we run separate models for respondents in Quebec and outside Quebec, as the latter cannot vote for the Bloc Québécois. Our analyses focus on respondents who voted in each election.

Two survey items pertaining to local candidates were included in the CES between 2000 and 2008. Respondents were first asked the following question: “Now the local candidates in your riding. Was there a candidate in your riding you particularly liked?” Those who answered in the affirmative were then asked, “Which party was the candidate you liked from?”

Table 1 shows the percentage of respondents who said they liked a candidate in their riding. The distribution for 2000 is very similar to that reported by Blais et al. (Reference Blais, Gidengil, Dobrzynska, Nevitte and Nadeau2003), but there is more variation in preferences for local candidates in subsequent elections.Footnote 2 In all four elections, this percentage is higher outside Quebec than in Quebec. Among those who voted, the pattern is similar, with slightly more citizens indicating a preference for a local candidate. This suggests that in each election, about half of voters care about their local candidates while the other half do not.

Table 1. Percentage of Respondents and Voters Who Liked a Candidate in Their Riding

Results

Following Blais et al. (Reference Blais, Gidengil, Dobrzynska, Nevitte and Nadeau2003), we first ascertain the independent effect of having a preference for a local candidate on vote choice.Footnote 3 We control for party identification (“Which party do you feel closest to?”) and feeling thermometer evaluations for all parties and leaders (rescaled to range from 0 to 1).Footnote 4 These are the same controls used in the original study. It is crucial to account for these confounders when ascertaining the independent impact of local candidates on vote choice, as each of these variables has its own impact on the outcome of interest. We run two models for each election: one for respondents outside Quebec and the other for those in Quebec. For the first set of models, we control for the region (Ontario, Atlantic, West), as Blais et al. (Reference Blais, Gidengil, Dobrzynska, Nevitte and Nadeau2003) do.

The results are presented in Table 2 (outside Quebec) and Table 3 (Quebec).Footnote 5 The findings are easy to summarize. In both tables, the local candidate coefficient is positive and significant in all four elections, meaning that even after controlling for party identification and feelings toward leaders and parties, local candidates have an independent impact on vote choice. However, the local candidate coefficient is smaller than those for party identification and party and leader evaluations. This is consistent with Blais et al.'s (Reference Blais, Gidengil, Dobrzynska, Nevitte and Nadeau2003) results for the 2000 election. While having a preference for a local candidate does matter, it matters less than partisan predispositions and attitudes toward parties and leaders. All in all, these findings confirm previous work showing that local candidates have a noticeable effect on vote choice.

Table 2. The Impact of Liking a Local Candidate on Vote Choice outside Quebec

Note: Multinomial logit regression coefficients with standard errors in parentheses. The reference category for vote choice is Liberal. The reference category for region is Ontario.

* p < .05; **p < .01; ***p < .001

Table 3. The Impact of Liking a Local Candidate on Vote Choice in Quebec

Note: Multinomial logit regression coefficients with standard errors in parentheses. The reference category for vote choice is Liberal.

* p < .05; **p < .01; ***p < .001

We now turn to the question of how much local candidates matter. We first estimate each individual's predicted likelihood of voting for each party with all the variables at their observed values. We then estimate new predicted likelihoods in a counterfactual scenario in which we set the coefficient for the local candidate at zero while keeping every other variable in the model constant. We finally compare the predicted vote (that is, the party that has the highest predicted probability of being supported by each individual) under both scenarios. When the predicted vote remains the same for a given individual, it means that local candidates do not matter, as our model predicts that the individual in question would vote the same regardless of her local candidates. When the prediction differs, we infer that the local candidate is a decisive consideration, since the person is predicted to have voted differently if she had not cared about her local candidates.

Table 4 reports these predictions for each election, both in Quebec and outside Quebec, as well as in Canada as a whole. Outside Quebec, the local candidate was decisive for 5 per cent of voters in 2000, 6 per cent in 2004, 8 per cent in 2006, and 7 per cent in 2008. In Quebec, only 2 per cent of voters are swayed by their local candidate in the 2000, 2004 and 2008 elections, with 2006 emerging as an outlier with 5 per cent. These results illustrate the relative stability of local candidate effects across the four elections under study. There might be an upward trend for voters outside Quebec, but overall the magnitude of these effects remains very similar across all elections when we examine the country as a whole. These results confirm that local candidates play a real but limited role in federal elections.

Table 4. Percentage of Voters for Whom the Local Candidate Was a Decisive Consideration

Conclusion

We ascertain the influence that local candidates play in Canadian federal elections. Using the CES from 2000 to 2008, we replicate Blais et al.'s (Reference Blais, Gidengil, Dobrzynska, Nevitte and Nadeau2003) original findings and extend their analysis to include four elections in total. We find that local candidates matter consistently for about 5 per cent of Canadian voters. We also confirm one of Blais et al.'s (Reference Blais, Gidengil, Dobrzynska, Nevitte and Nadeau2003) key findings, namely that local candidates matter mostly outside Quebec. We note that local candidates’ significant effect on vote choice is consistent with the argument that first-past-the-post electoral systems allow for the representation of local and regional interests, especially in comparison to systems with proportional representation (Blais, Reference Blais2008).

While this replication focused on elections held during the 2000s, we have no reason to believe that our findings are specific to that period. More recent work on local candidates (Allen Stevens et al., Reference Allen Stevens, Islam, de Geus, Goldberg, McAndrews, Mierke-Zatwarnicki, Loewen and Rubenson2019; Blais and Daoust, Reference Blais and Daoust2017; Roy and Alcantara, Reference Roy and Alcantara2015) reports effects of similar magnitude as ours. Future work should aim to explain what type of local candidates voters prefer.

Supplementary Material

To view supplementary material for this article, please visit https://doi.org/10.1017/S000842392200004X.

Footnotes

1 We include all the elections for which the Canadian Election Study (CES) asked the question used in the study we aim to replicate. We acknowledge that in 2015 and 2019, the CES asked another question related to local candidate preferences: “How do you feel about the candidates in your local riding? Set the slider to a number from 0 to 100, where 0 means you really dislike the candidate and 100 means you really like the candidate.” This question is conceptually distinct from the original question; furthermore, it was asked in the Campaign Period Survey whereas the original question was asked in the Post-Election Survey. Given these data constraints, we decided not to include these elections.

2 Unfortunately, the Blais et al. (Reference Blais, Gidengil, Dobrzynska, Nevitte and Nadeau2003) article does not have replication files available. We did our best to reproduce its results by following the authors’ explanations about their empirical design. It is plausible that the small difference we find for the 2000 election is due to our sample size being slightly different from that of the original study.

3 The unit of observation is the respondent-party pair. In Tables 2 and 3, we report the number of observations (respondent-party pairs) and the number of unique cases (respondents). Following Blais et al. (Reference Blais, Gidengil, Dobrzynska, Nevitte and Nadeau2003), we use multinomial probit regressions with alternative-specific controls because the key variables of interest (party identification, leader and party evaluations, and preference for a local candidate) are all alternative specific.

4 Consistent with the original study, outside Quebec, we include the Conservatives, Liberals, the New Democratic Party (NDP) and (for the 2000 election only) the Canadian Alliance. In Quebec, we include those parties that obtained at least 4 per cent of the province vote in a given election. This is the case for the Alliance (but not the NDP) in 2000. Starting in 2004, the NDP gained more than 4 per cent of the vote in Quebec, so we include it in our analyses for this period. Finally, we note that the Green Party met this threshold in 2006, but the CES did not include party and leader evaluations for the Greens that year; we thus cannot include this party in the analysis.

5 Whereas Blais et al. (Reference Blais, Gidengil, Dobrzynska, Nevitte and Nadeau2003) use multinomial probit regression, we opt for multinomial logit regression instead, as these are more robust than their probit alternative (Dow and Endersby, Reference Dow and Endersby2004) and more commonly used in voting behaviour research (Alvarez and Nagler, Reference Alvarez and Nagler1998). We nevertheless present the probit regression estimates in the online appendix. Results are very similar regardless of model specification.

References

Allen Stevens, Benjamin, Islam, Md Mujahedul, de Geus, Roosmarijn, Goldberg, Jonah, McAndrews, John R., Mierke-Zatwarnicki, Alex, Loewen, Peter John and Rubenson, Daniel. 2019. “Local Candidate Effects in Canadian Elections.” Canadian Journal of Political Science 52 (1): 83–96.CrossRefGoogle Scholar
Alvarez, Michael and Nagler, Jonathan. 1998. “When Politics and Models Collide: Estimating Models of Multiparty Elections.” American Journal of Political Science 42 (1): 55–96.CrossRefGoogle Scholar
Bittner, Amanda. 2011. Platform or Personality? The Role of Party Leaders in Elections. London: Oxford University Press.CrossRefGoogle Scholar
Blais, André. 2008. To Keep or to Change First Past the Post? The Politics of Electoral Reform. Oxford: Oxford University Press.CrossRefGoogle Scholar
Blais, André and Daoust, Jean François. 2017. “What Do Voters Do When They Like a Local Candidate from Another Party?” Canadian Journal of Political Science 50 (4): 1103–9.CrossRefGoogle Scholar
Blais, André, Gidengil, Elisabeth, Dobrzynska, Agnieszka, Nevitte, Neil and Nadeau, Richard. 2003. “Does the Local Candidate Matter? Candidate Effects in the Canadian Election of 2000.” Canadian Journal of Political Science 36 (3): 657–64.CrossRefGoogle Scholar
Dow, Jay K. and Endersby, James W.. 2004. “Multinomial Probit and Multinomial Logit: A Comparison of Choice Models for Voting Research.” Electoral Studies 23 (1): 107–22.CrossRefGoogle Scholar
Roy, Jason and Alcantara, Christopher. 2015. “The Candidate Effect: Does the Local Candidate Matter?” Journal of Elections, Public Opinion & Parties 25 (2): 195–214.CrossRefGoogle Scholar
Sevi, Semra. 2021. “Who Runs? Canadian Federal and Ontario Provincial Candidates from 1867 to 2019.” Canadian Journal of Political Science 54 (2): 471–76.CrossRefGoogle Scholar
Figure 0

Table 1. Percentage of Respondents and Voters Who Liked a Candidate in Their Riding

Figure 1

Table 2. The Impact of Liking a Local Candidate on Vote Choice outside Quebec

Figure 2

Table 3. The Impact of Liking a Local Candidate on Vote Choice in Quebec

Figure 3

Table 4. Percentage of Voters for Whom the Local Candidate Was a Decisive Consideration

Supplementary material: File

Sevi et al. supplementary material

Online Appendix

Download Sevi et al. supplementary material(File)
File 30 KB