District Magnitude and Home Styles of Representation in European Democracies

Abstract

How political representatives interact with constituents is contingent on the conditions under which they compete for re-election. It has been the assumption that intra-party competition shapes legislators’ incentive to cultivate a personal reputation and thereby the incentive to engage in any constituency-oriented action that reputation may be built on. Using data from the PARTIREP cross-national survey among regional and national legislators in 12 European democracies, the assumption is proven to be inaccurate: focusing on more than one action at a time, it becomes clear that district magnitude increases the time representatives spend on some constituency-oriented actions and decreases the time they spend on others (even when the electoral incentive to nurture a personal reputation is strong). The article demonstrates that representatives’ choice to engage in four constituency-oriented actions is related to the nature of these home styles and their electoral utility in a given context.

Acknowledgements

The authors acknowledge BELSPO for funding the PARTIREP ‘Participation and Representation in Modern Democracies’ international study. Audrey André also acknowledges FWO-Vlaanderen for financial support. The authors would further like to thank David Farrell, Thomas Zittel, Kris Deschouwer, Rudy Andeweg, Jean-Benoit Pilet, Peter Esaiasson and the two anonymous reviewers for their comments on earlier versions of this article.

Notes

1. Where there are penalties for overnomination, under the single non-transferable vote for instance, parties will tend to field as many candidates in the district as there are seats they expect to win. Fielding more candidates risks splitting their vote and winning fewer seats in the face of concentrated opposition (Cox 2008). As such, district magnitude can no longer be assumed to be the key determinant of that ratio.

2. Political parties are absent from the state legislature of Appenzell-Innerrhoden.

3. Dutch voters indicate a preference for a candidate, but the impact of preference votes on the order of intra-party seat allocation is negligible. In the 2003 general elections, Andeweg (2008) observed, no more than two candidates were elected in defiance of the predetermined list order. Leapfrogging past higher ranked co-partisans is much more common in Belgium – in particular since the 2000 reform cutting the number of list ballots to be allotted to the top candidates in half. In recent elections, the number of legislators owing their seat to their personal performance exceeded 10 per cent (André et al. 2012).

4. The ballot structure is closed for the Italian Camera dei Deputati and for the Tuscan regional council but open for the regional councils in Campania and Valle d’Aosta. The regions of Calabria, Lazio, and Lombardy use a two-tiered PR system. Preference votes alone determine which candidates are elected at the lowest level. The second tier allocates a majority bonus to the coalition or party list of the candidate elected for presidency. Candidates are elected in the order of the regional party list.

5. There is some discussion regarding the correct coding of single-seat districts: it is our opinion that the practice of coding single-seat districts as candidate-centred is inspired more by its observed consequences than by its mechanics: the preference for a candidate cannot be separated from the party preference (Karvonen 2004). Voters are not offered intra-party choice and cannot hold an individual candidate to account without changing party affiliation (Mitchell 2000).

6. The Austrian Nationalrat employs a three-tiered system with preferential voting at the lowest levels, but nominal voting is quite uncommon in the second tier and as such is ineffective. The ease of supporting an individual candidate differs markedly at both levels. While candidates are fully listed on the ballot and voters simply tick the box next to the name of their preferred candidate in the first tier, voters have to write in a candidate’s name in the second (Müller 2008). Both the second and third tiers are therefore coded as closed lists. Though reduced to two levels, the regional parliaments largely mirror the electoral system used at the national level.

7. The overdispersion parameter, alpha equals 0.58 and a likelihood ratio test indicates that it differs significantly from zero (G² = 50,212; p < 0.000). The likelihood ratio test for alpha is computed as follows: G² = 2(lnL _nbrm – lnL _prm ) = 2(–6,794 – (–31,900)).

8. The district’s geographical location is defined by the distance from the seat of parliament to the largest city in the district as the crow flies – using Google maps. In mixed-member systems, the distance was measured to the first tier district the legislator stood in. Only for legislators who stood only on the party list was the distance set to zero. In the Netherlands reference was made to the 19 administrative electoral regions and in Israel to the legislator’s place of residence.

9. The controls do not detract from the effect of the electoral rules. Reported findings are robust to excluding the control variables from the regression equations.

10. District magnitude was increased from 1 to 100 (i.e. the maximum in open-list systems) to compute the expected counts and the 90 per cent confidence intervals of the discrete change. Dichotomous variables were set to zero and continuous variables were fixed at their mean values.

11. To ascertain that the negative effect in closed-list systems is not an artefact of the question wording in Israel and the Netherlands, the model was rerun excluding both cases. The direction, significance, and even magnitude of district magnitude’s effect remain practically unaltered.

12. Analogous to the previous analysis, district magnitude was increased from 1 to 100. Dichotomous variables were set to zero and continuous variables were fixed at their mean values.

13. Due to word limit constraints, the multivariate probit model and post-estimations are not reported – but they can be obtained upon simple request.