Abstract
District magnitude is a central aspect of the institutional context in PR elections, and it influences parties’ and voters’ strategies. The incentives for strategic behaviour are stronger in smaller districts, as only large parties are likely to be viable. This article investigates how much the vote is affected by this characteristic of the electoral context, focusing on the 2005 and 2009 Portuguese elections. Portugal is one of the countries with the largest degree of variation in district magnitude and represents thus an ideal case for analysing district magnitude effects. Relying on data from the Comparative Study of Electoral Systems, this study shows a strong mechanical effect of district magnitude and a limited psychological effect.
Acknowledgements
A previous version of this manuscript was presented at the 2012 conference on "The Effects of District Magnitude" at the University of Lisbon. We would like to thank the participants in this workshop, the editors of this journal, and the anonymous reviewers, for their helpful comments. All remaining errors are ours.
Supplemental Material
Supplemental data for this article can be accessed at http://dx.doi.org/10.1080/17457289.2014.1002791.
Notes
1. In 2009, from 2 to 47.
2. The appendix is available online at http://dx.doi.org/10.1080/17457289.2014.1002791.
3. The corresponding results for 2009 are in Table A2 in the appendix.
4. The absence of a relation between district magnitude and votes for the PCP-PEV is largely due to its strong support in the small districts of Beja (29% for PCP-PEV, magnitude 3) and Évora (22%, magnitude 3), from the agricultural region of Alentejo in the South of the country, which is a traditional stronghold of the communist party. We thank one of the anonymous reviewers for emphasizing this point.
5. If districts were even more proportional than in Lisbon, it could be that additional parties would enter the race and try to gather some votes. But when making a comparison between the existing system and one in which all districts would be as proportional as the Lisbon district, we can consider that additional minor parties represent a negligible issue.
6. We cannot test for this relation in 2005: respondents were asked if they were contacted by parties during the campaign, but not by which parties.
7. Not voting for a non-viable party is one form of strategic voting, but not the only one (Cox Citation1997). In this paper, however, we only focus on that type of strategic behaviour.
8. In 2005, this is the only available information on citizens’ evaluations of party leaders. In 2009, respondents were also asked to rate party leaders on a like–dislike scale. For the 2009 replication, we rely on that second battery of questions, as it offers a more detailed assessment of leader sympathies and as it improves the predictive power of the voting choice model.
9. For respondents who do not identify with any party, this dummy variable takes the value 0 for all choice alternatives.
10. We have performed a robustness test by replicating this analysis with a more fine-grained measure of viability, based on the additional share of votes a party would have needed in the previous election in order to win the last allocated seat. The corresponding results, presented in Table A1 in the appendix, are consistent with those of the main model specification.
11. For each district, the point estimate is the median value from the 1000 simulations, while the lower and upper bounds of the confidence interval are the 2.5th and 97.5th percentiles of this distribution.