Candidate turnover in EP- versus national elections: a new perspective on the second-order logic? A comparative analysis

ABSTRACT

This article investigates the differences between candidate turnover at national legislative and European Parliament elections. Multilevel analysis is applied on original data containing 2754 electoral lists, clustered within 79 political parties, within 48 national and European elections. These comparative data allow (1) to test hypotheses on second-order EU elections within the context of electoral recruitment and (2) to measure the effect of systemic turnover drivers, which in turn may differ for EP elections. As such, this paper builds further on the emerging literature on candidate renewal by offering a comparative perspective and understanding the particular dynamic for EP elections. Results suggest that candidate turnover is lower at EP elections. Moreover, the effects of turnover drivers differ at EP elections: candidate turnover is lower on lists of fringe parties at the European level and elections with higher electoral volatility are characterised by higher candidate turnover, particularly at European elections.

KEYWORDS:

• Candidate turnover
• candidate selection
• EU
• second-order elections
• list-PR systemist

Acknowledgments

This work was supported by [FWO] under Grant [G070418N].

Disclosure statement

No potential conflict of interest was reported by the authors.

Supplemental data

Supplemental data for this article can be accessed online at https://doi.org/10.1080/07036337.2023.2220878

Notes

1. All nine countries under investigation in this article apply a nation-wide electoral district, apart from Belgium, which has two districts across regional borders. However, as our dataset only includes electoral lists of parties within the region of Flanders, no electoral district subdivisions are to be taken into account.

2. No lists were stored by the Swedish ministry of interior for the Swedish national elections of 2006 in the Jönköpings district. To collect this information we contacted the national library archive and individual political parties.

3. As the presented models are fractional logit regressions, we predict the logit at candidate renewal and the beta coefficients follow an additive arithmetic.

4. After we have included all relevant EU interaction terms (model 2), the EU main effect is no longer significant. It can be seen that the strongest interaction effects are with ‘fringe party position’, ‘EU saliency’ and ‘electoral volatility’. Both a forward and a backward selection procedure (not shown here) suggest that the EU main effect is no longer significant after introducing interactions with the variables ‘fringe party position’, ‘EU saliency’, ‘electoral volatility’ and ‘ENEP’ in the model.

5. We control for district magnitude and list magnitude, but differences could also occur depending on the variation in the ratio between the number of candidates on a list and the number of incumbents. However, the robustness check presented in Annex D in the supplemental material does not show any substantial impact.

Additional information

Funding

The work was supported by the Fonds Wetenschappelijk Onderzoek [G070418N]

Notes on contributors

Gertjan Muyters

Gertjan Muyters is postdoctoral researcher at the KU Leuven Public Governance Institute. His research focuses on candidate turnover and political careers. He has published in West European Politics, Party Politics, and European Political Science Review, amongst others.

Bart Maddens

Bart Maddens is Professor of Political Science at the KU Leuven Public Governance Institute. His research interests include political finance, elections and multi-level systems. His articles have appeared in journals such as West European Politics, Party Politics and Electoral Studies, amongst others.