Ballot structure, district magnitude and descriptive representation: the case of New Zealand local council elections

ABSTRACT

It is sometimes claimed that the use of preferential voting facilitates more effective representation of under-represented groups than that generated by the first-past-the-post (FPP) system, particularly in combination with higher district magnitude. This claim is tested using unique observational data. Local government elections in New Zealand are conducted using either the single transferable vote (STV) or first-past-the-post, providing the opportunity to compare representation of women and Indigenous Māori across the two sets of councils. When combined with higher district magnitude, STV may sometimes promote the election of women, but has no effects on the election of Māori. The advantages of high district magnitude STV in promoting descriptive representation are counteracted by lack of coordination in candidate nomination and voting because these local electoral contests are predominantly nonpartisan.

有人认为顺序 投票法较之票过即选制更有利于不被代表的群体获得代表，尤其在地区权重较高的情况下。我们使用一种独特的观察资料对这种说法进行了检测。新西兰的地方政府选举或采用单独转票法或采用票过即选法，让我们可对妇女和毛利原住民入选两套议会的情形做一番比较。当地区权重较高时，单独转票制可以促进妇女的当选，但无助于毛利人的当选。地区权重高加单独转票制有利本色代表的长处，会因候选人提名及投票等方面缺乏协调而打折扣，因为这些地方选举的竞争多是非党派性质的。

KEYWORDS:

• STV
• district magnitude
• local elections
• Māori representation
• women's representation

Disclosure statement

No potential conflict of interest was reported by the author(s).

Notes

1 Another nominal option is provided by the SNTV system, now rarely used, provides for a vote for only one candidate in a multi-member district. The ‘limited vote’ system allows nominal votes for more than one candidate, but not for as many candidates as there are elected positions to fill (Lijphart, Pintor, and Sone 2003).

2 The limited examination of descriptive representation under ordinal ballots is easily explained. Ordinal ballots are unusual outside of the three countries in which they have been ‘embedded’ over many decades – Ireland, Australia, and Malta (Bowler and Grofman 2000).

3 Party list systems are another form of nominal voting, distinguished by their proportional electoral formulae: for example, the d’Hondt or Sainte-Laguë methods of translating votes into seats. While lists are used in national politics in New Zealand, they are not an option for local government elections. While also not relevant to the discussion here, it can also be noted that there are some forms of AV with M > 1, where the quota for victory remains at 50 per cent plus one, with either a single vote or multiple votes.

4 The same incentives also apply under nominal multi-member plurality voting.

5 An STV New Zealand local election ballot can be found at Te Tari Taiwhenua/Department of Internal Affairs, 2020. The instructions provided are as follows. ‘There are three vacancies in this ward. Instead of a tick you rank candidates in the order you prefer - '1' beside your first choice, '2' beside your second choice, '3' beside your third choice and so on. You can rank as few or as many candidates as you wish.’

6 Persons of Maori descent may choose on which roll to place their name; they constitute about 19 per cent of the total population, and 15 per cent of the age-eligible voting population. Significant numbers of Māori register on the general roll; see Table 3 below and further discussion in the final section.

7 From Local Government New Zealand 2010 (with a response rate of 47.2%). For similar data from other years see Sullivan 2003, Hayward 2011.

8 Table 2, Figure 2 and later Tables and Figures report significance levels and 95 per cent confidence intervals, many of which are below traditional levels of significance and within confidence intervals. However, this data is not based on random sampling. It constitutes the entire range of cases within the elections in question. We make no claims here of robust generalisation beyond these cases and elections. In some analyses, Ns are small. STV systems are a minority of the cases. Statistical significance and confidence intervals should therefore be considered as heuristic aids, indicating the relative certainty of findings, not arbitrary cut-offs.

9 Number of candidates=stv 1.02 (.20)** + lgM 2.27 (0.02)**, constant 1.28 (0.11)**, r2=0.79, N=3598. Ward-Level Dataset: Number of candidates=stv 1.432 (.27)** + lgdM 5.97 (0.12)**, constant 1.88 (0.11)**, r2=0.12, N=1985.

10 2016: Elected=lgdm 0.19 (0.07) + stv 0.12 (0.18) + incumbent 2.16 (0.14) + stv * incumbent 0.46 (0.47) + lgelpop -0.39 (0.05), constant 2.43 (0.46), r2=0.24. 2019: Elected=lgdm 0.28 (0.07) + stv 02 (0.18) + incumbent 2.70 (0.15) + stv * incumbent 0.06 (0.35) + lgelpop -0.34 (0.05), constant (1.85 (0.45), r2=0.25.

11 Likelihood of being unopposed=stv -2.90 (1.00) + election year -0.61 (0.17) + stv*election year 1.35 (1.16), constant -2.59 (0.10), r2=0.04. There were almost no unopposed candidates under STV, with about 7 per cent unopposed in 2019, 4 per cent in 2016.

12 As confirmed by the number of Māori electorate seats allocated after the most recent Maori electoral option in 2018 (7) to the total (72), using this total population basis.

Additional information

Notes on contributors

Jack Vowles

Jack Vowles is Professor of Comparative Politics at Victoria University of Wellington. His earlier appointments were at the University of Auckland, the University of Waikato, and the University of Exeter. He has led the New Zealand Election Study since 1996.

Janine Hayward

Janine Hayward is Professor of Politics at the University of Otago. She researches and teaches various aspects of New Zealand politics, including Treaty of Waitangi politics, New Zealand's constitution, electoral politics, environmental politics, and local government politics.