Electing Mayors with the Supplementary Vote Method: Evidence from Norway

Abstract

Both Norway and England have used the supplementary vote (SV) method to elect mayors. SV was intended to elect mayors with a broad popular mandate, and deemed easy for voters to use. Previous research on English mayoral elections focusing on the supply side of the elections has found that voters supporting minority candidates with their first vote face difficulties using the optional second vote to choose among the remaining two top-candidates in municipalities where there are many candidates in the race. A close race, on the other hand, seems to give voters a helping hand in making this choice. Drawing on these findings, we link data on the 48 Norwegian mayoral races to survey data to further explore how the supply side of the elections and individual voter resources affect effective voting and use of the optional second vote. We use both standard logistic regression and multi-level models as statistical tools. The result reveals that 27 out the 48 elected mayors in Norway came out as majority winners (more than 50 per cent of the votes). The analysis suggests that the supply side of the elections is of minor importance for whether voters use their second vote or not. The probability of using the second vote is reduced by age, and voters who claim to understand how the electoral system works use the second vote more frequently. The analysis also reveals considerable variation in the number of first and second votes for the top two candidates (effective votes) between municipalities, and the results suggest that this variability may be explained partly by the closeness of the elections. Voters in municipalities with competitive mayoral races are more likely to vote effectively than voters in municipalities with less competitive elections. At the individual level party members are more likely to vote ineffectively.

Key Words:

• Supplementary vote
• directly elected mayors
• majority winners
• effective voting
• voter behaviour
• supply-side effects

Acknowledgments

The authors wish to thank Tor Midtbø and two anonymous reviewers for constructive comments to the revision of this article. We also thank the Ministry of Local Government and Regional Development for financial support.

Notes

1 Different types of voting paradoxes are not discussed here. However, the literature shows that SV (as many other electoral systems) cannot guarantee the election of a Condorcet winner – that is the candidate who would win a two-candidate race against each of the other candidates. In addition a Condorcet loser – a candidate defeated in two-candidate races against every other candidate – can be elected with SV. Finally, SV does not satisfy the monotonicity criterion; the idea that extra votes for a candidate shall not reduce the same candidate's chances of winning the election (see Christensen & Aars, 2008; Van der Kolk et al., 2004).

2 ‘Effective use of the vote’ means casting a vote for at least one of the top two candidates in the race with the first or second vote (Van der Kolk, 2008).

3 The Local Authorities (Mayoral Elections) (England and Wales) Regulations 2007 No. 1024, paragraphs 8 and 10.

4 Guidance for candidates and agents. Greater London Authority election, 1 May 2008 (The Electoral Commission).

5 Independent mayoral candidates won in seven out of 23 UK races, whereas this did not happen in any of the Norwegian municipalities.

6 For an overview of different electoral systems, see Gallagher and Mitchell (2005).

7 The purpose here is to single out voters with potential preferences for the two top-candidates. As pointed out by Van der Kolk et al. (2006: 98), it may be that ‘voters were indifferent between the top two candidates’ and preferred voting for losing candidates.

8 When comparing the Norwegian results to those of the UK we have to consider that in many of the UK races the battle for second place was also very close. In Stoke-on-Trent in 2005, only 432 votes (0.5 per cent) separated the second candidate and the fourth candidate (Electoral Reform Society, 2007). In Hackney (2002) the gap between second and fourth was 597 votes (2.2 per cent). In seven out of 24 mayoral races the gap between the second- and third-placed candidates was less than 1.5 per cent (ibid.). Evidently, it is harder to guess which two candidates will make it to the final round, with a tight race between three or four candidates.

9 The lower n is caused by a considerable number of voters not willing to reveal their party preferences in the council elections.

10 The dependent variables are measured at the individual level and it is therefore natural that this level accounts for most of the variance (see Steenbergen & Jones, 2002).