![Sample our Social Sciences journals, sign in here to start your FREE access for 14 days]({"type":"image" , "src" : "/sda/110822/TOC-Subject-Sample-2021-Social-Sciences.jpg"})
ABSTRACT
In semi-open list systems, parties present pre-selected pools of candidates to the electorate. Candidates’ assigned ranks on the lists heavily influence their election odds and may reflect party leaders’ preferences, notably a possible gender bias. To strengthen women's representation, parties’ choices are increasingly subject to legal quotas. These quotas are expected to be less binding for left-wing parties, which tend to be more women-friendly. Analyzing 854 party lists presented to Flemish voters in the 2012 local elections, this study finds that right-wing party leadership discriminate in favor of women by offering them higher positions on the party lists. For leftist parties, the study finds discrimination in favor of men. Importantly, parties offering higher positions to women tend to do so in parts of the party list certain to lead to either election (left-wing parties) or non-election (right-wing parties). For positions with critical election-odds, no gender bias is identified.
HIGHLIGHTS
• Women's representation is a major concern in today's politics.
• Gender quotas in elections are meant to ensure women's presence on ballots.
• In the 2012 Flemish elections, women candidates were on average positioned higher on party ballots than they would be if positioning were based on anticipated electoral success.
• Left-wing parties advantage women in positions with high chances of election. Right-wing parties advance women only in positions with low chances of success.
• For positions with critical election odds, rankings across all parties are based on expected electoral success in terms of preferential votes, irrespective of candidate gender.
• Assigned ballot positions may reflect underlying gender preferences or power balance within the party leadership. Gender biases may be neutralized more effectively by placement mandates than by general quota rules.
ACKNOWLEDGEMENTS
We wish to thank three anonymous referees for very detailed and constructive reviews, as well as Joshua Holm, Paola Profeta, Joanna Siwinska-Gorzelak, Kartsen Staehr, and the participants of the European Public Choice Society Annual Meeting in Rome (April 2018) and of the Research Seminar of the Institute of Economic Analysis and Finance at Tallinn University of Technology (May 2018) for helpful comments and suggestions. Colin R. Kuehnhanss is grateful to the Research Foundation Flanders (FWO) for financial support in the form of a PhD-Fellowship from 2014 to 2018 (grant nr. 11V2117N).
SUPPLEMENTAL DATA
Supplemental data for this article can be accessed at https://doi.org/10.1080/13545701.2019.1685675
Correction Statement
This article has been republished with minor changes. These changes do not impact the academic content of the article.
Notes
1 Audrey André, Bram Wauters, and Jean-Benoit Pilet (Citation2012) show that preference voting is a form of sophisticated voting behavior. Politically interested and well-resourced voters make more active use of the option to express preference votes. Closer relationships between voters and politicians, as well as the (perceived) effectiveness of preference votes increase their use.
2 The finding that leftist parliamentarians (in Poland) were more supportive of quota regulations is therefore unsurprising (Dubrow Citation2011).
3 Voting is compulsory in Belgium, and the nationwide turnout in the 2012 municipal elections was 89.7 percent (Marien, Schouteden, and Wauters Citation2017). The share of voters expressing preference votes is about 86 percent (see Wauters, Verlet, and Ackaert [Citation2012]).
4 This “verkiesbaarheidscijfer” is calculated as the number of votes (list votes + ballots with preferential votes) multiplied by (number of seats) / [(number of seats) + 1].
5 More precisely [(number of list votes)*(seats won)] / 3 are transferred. If the first candidate did not reach the eligibility threshold already, these “transferable votes” (“over te dragen stemmen”) are added to his/her preferential votes until the threshold is met. If the threshold was already met, the second ranked candidate is considered, and so on. Remaining transferable votes are distributed to the next-ranked candidate(s) until the set of transferable votes is exhausted.
6 This allows excluding atypical cases from the main analysis, the most extreme being party lists composed of a single candidate (by definition receiving a vote share of 100 percent of the party list).
7 These ties largely correspond to zipper-lists, as some party chapters self-impose the requirement that no two successive candidates on the ballot should be of the same gender. Thus, they signal a clearly gender-neutral position.
8 Candidates who beat their parties’ expectations, that is, receive more preferential votes than their rank in the party list would suggest, not only increase their probability of being elected. They also tend to get better (that is, higher) ranks in the following elections (Crisp et al. Citation2013; Folke, Persson, and Rickne Citation2016; André et al. Citation2017).
9 We consider two alternatives. First, taking a similar approach as Gail McElroy and Michael Marsh (Citation2010), we redefine the dependent variable as ln [vi,j/ v*j ], where v*j is the average vote share for any candidate on the party list (in other words, v*j is the inverse of the list length). Second, following Górecki and Kukołowicz (Citation2014), we use the logarithm of the absolute (instead of relative) number of preferential votes and estimate by negative binomial regression. Both alternatives give highly comparable results and do not change any of the conclusions presented in the paper. Results available upon request.
10 It could be argued that being in one of the top positions mainly offers an advantage in terms of visibility. If this were the case, we would expect absolute positions to be more relevant. Being second in a list of eleven candidates or in a list of fifty-five may be similar from a perception perspective. To capture this possibility, we re-estimate by replacing the relative FIRST DECILE indicator with dummies for absolute positions: POSIT2i, POSIT3i , POSIT4i , … POSIT10i for politicians in second, third, fourth, up to 10th position on their party's ballot (Put, Maddens, and Smulders Citation2015). The results do not change any of the conclusions reported below and are available upon request.
11 We considered three alternatives. First, we define Nj as the number of seats won in the previous 2006 election (see Esteve-Volart and Bagues [Citation2012]). In this case the sample is limited: as the party landscape at the local level is highly volatile with parties entering electoral alliances or breaking them up, the number of observations drops to 12,419. Second, we split samples in a more rudimentary way by considering subsamples of the “top x” versus “bottom 1-x” positions (with x = 20, 30, and 40 percent, respectively). Finally, the sample is split into “serious” and “non-serious” contenders. The latter approach follows Put, Maddens, and Smulders (Citation2015) and defines “(non-)serious” contenders as those that were (not) actually elected. The results are broadly in line with those presented here and available upon request.
12 With the transfer of list votes being limited to one third, the possibility remains for candidates ranked lower on the ballot to get elected purely based on their own preference votes. As discussed above, the last position is one such special case which frequently leads to election despite falling within our low-chance category. While we control for the last position, other candidates on low-chance positions may also get elected. However, the chance of being elected on such a position is below 6 percent in our sample.
13 Replicating the analysis with different values for k gives, however, highly comparable results (results for k = 2 can be found in Table A3 in the Online Appendix).
14 To rule out that the effect observed for “safe” positions is a consequence of the placement mandate (imposing parties to have both a man and a woman candidate in the first two positions), we reestimate the regressions in Table leaving out candidates in those positions. The main conclusions are unaffected, and results are available upon request.
15 To test whether any of our findings in the logistic model are driven by pairs led by (fe-)male candidates, we also run the same estimations with the sample split according to whether the female, respectively the male candidate, was ranked second. The results confirm the findings reported (results available upon request).
Additional information
Funding
Notes on contributors
Bruno Heyndels
Bruno Heyndels is Professor in micro-economics at the Vrije Universiteit Brussel (VUB), Belgium. His research lies in the fields of public choice, cultural economics, sports economics, and economic psychology. His work has been published in Electoral Studies, European Journal of Political Economy, International Tax and Public Finance, Journal of Cultural Economics, Journal of Economic Psychology, Journal of Sports Economics, Kyklos, National Tax Journal, Public Administration, and Public Choice, among others.
Colin R. Kuehnhanss
Colin R. Kuehnhanss is an Affiliated Researcher at the Department of Economics at the Stockholm School of Economics in Riga, Latvia. His research interests include political economy, behavioral economics, and political decision making. Previous work was published in the European Journal of Political Economy, JCMS: Journal of Common Market Studies, Journal of Economic Psychology, Policy and Society, and Public Administration. Colin is also an official at the European Commission, however this contribution is authored in private capacity and expresses exclusively the personal opinion of its authors. It does not necessarily reflect the official opinion of the European Commission.