Does Increasing Ethnic Diversity Reduce Electoral Turnout? The Case of New Zealand 1957–2020

Abstract

A diversifying population may affect electoral turnout when there is an increasing proportion of minorities with weaker voting habits than those of the majority. New Zealand provides an example to test this possibility, by way of growth in the indigenous Māori population and a high level of recent immigration. A large sample dataset provides better estimates of group turnouts than hitherto available. The authors estimate how much ethnic diversification may have affected turnout in New Zealand between the elections of 1957, 2014, and 2020. By adopting a theory that resists conceptualisation of not voting as deviation from ‘civic duty’, the authors avoid ascribing ‘fault’ to this behaviour.

Disclosure Statement

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

Notes

1 The concept of political efficacy looms large in the social psychological theory of voting (Campbell et al., 1954; Coleman & Davis, 1976; Craig & Maggiotto, 1982; Miller et al., 1980) but of course is equally consistent with a rational choice model. As a combination of resources and education (lowering the information costs associated with deciding how to vote) internal efficacy also fits well. Simllarly, external efficacy affects the perceived benefits of voting; if politicians are deemed responsive, people will be more likely to vote.

2 Indeed, recent research in cognitive psychology and neuroscience makes it increasingly difficult to make a clear distinction between habitual and norm-driven behaviour (Neal et al., 2011; Neal et al., 2012; Ouellette & Wood, 1998; Wood & Neal, 2007). People often act before making a conscious choice to do so (Smith, 2008). One can express agreement with a norm in a survey questionnaire simply as a rationalisation of a habit. Habit and ‘duty’ may simply be different sides of the same coin. In line with social choice theory, Franklin’s position does not go this far. The choice to vote is fundamentally instrumental, and normative only in the sense that those norms are based in learned attitudes and values.

3 New Zealand’s MMP system was adopted in 1993 by referendum and the first election held under the new rules was in 1996. There are two votes, a party or list vote, and an electorate/ constituency/ candidate vote. Lists are national and in 2020 returned 48 members. 72 members are also elected in single-member districts. List seats are allocated on a ‘top-up’ basis using the party or list vote. There is a 5 per cent party vote threshold for representation that can also be surmounted if a party wins one or more electorate seats. See Vowles (2017) for the details, although the analysis should be supplemented by more recent developments that saw the plurality-winning party consigned to Opposition in 2017 and a majority single-party Cabinet formed after the 2020 election.

4 For further information about the Māori electorates and the Māori roll, see Bargh (2021), Riambau (2018), Vowles and Gibbons (2022).

5 Despite relatively low Māori electoral turnout, Māori tend to participate at a somewhat higher level than non-Māori in various forms of non-electoral participation (Bargh, 2013; McVey & Vowles, 2005, p. 12). The loudest calls for indigenous self-determination come from those also deeply engaged in the electoral process, most notably Te Pāti Māori (Godfery, 2015).

6 The electronic electoral rolls are available in New Zealand for approved purposes of social science research. The ‘Bigger Data’ turnout project was supported by the Marsden Fund, administered by the Royal Society of New Zealand (VUW 1624). Ethical approval was granted by Victoria University’s Human Ethics Committee (applications 0000020998, 0000026647 and 0000030817) which waived the usual requirement that those sampled should give consent to their inclusion.

7 Thus the growing proportion of the elderly in many electorates may also need to be taken into account in making this assumption in future.

8 Surnames were classified by data from https://mondonomo.com/surname/. When in doubt, the country with the highest prevalance of the name was chosen. Southern Europe and Latin America were grouped together given the large overlap of names between these two areas. Estimating ethnicity by names is widely practiced and defended as providing reasonably reliable results (CFPB, 2014; Mateos, 2007; Stillwell, 2022).

9 Estimated residential population data by ethnic groups is available by year from Stats NZ (2022a). From a 2018 base, estimates are published at 5-year intervals up to 2043. Exponential extrapolation for 2020 provides an equivalent age-eligible voting population estimate to that published by the Electoral Commission. An estimate for Māori identifiers can also be calculated. Māori descent data is available from the 2018 census in age bands (Stats NZ, 2022b), but without any projection into the future. However, close correspondence to the age-eligible population in 2018 can be estimated by breaking up the 15–19 age band, assigning 2/5 of the numbers to the age-eligible total of 18 and over. Further extrapolation from the extrapolated percentage increase for Maori identifers between 2018 and 2020 on those of Māori descent gives an estimate of the age-eligible Maori descent population. Those of Māori descent on the 2020 roll formed 92.8 percent of the those eligible, a little lower than the overall reported figure of 94.1 per cent. This is likely almost entirely because the Māori population is much younger than the rest. Chapman (1963, pp. 279–280; 1986, p. B-106) reports evidence that non-enrolment among Māori was increasing from the 1950s onward, in tandem with urban migration, but possibly over-estimates the problem by assuming that all with equal Māori and Pākehā ancestry would have been on the Māori roll. One has no way of comparing Māori non-enrolment between these two elections, but it is unlikely to have much influence to the estimates made here.

10 The average of panellists in 2020 was 49, reflecting lower coverage of young people on the roll. One can then assume the same applied in 1957, making the difference comparable.

11 Various ways of estimating the slope were applied, using age in years as a factor variable or as a linear variable between 28 and 80, on both unweighted (roll) and weighted data (by age-eligible population).

12 Other changes are more difficult to assess. Although the voting age has gone down, people are staying in the education system longer, having children later, and settling into secure employment later.

13 (Māori fraction roll 1957 * 2020 Māori turnout) + (European fraction roll 1957 * 2020 European turnout).

Additional information

Funding

This work was supported by Marsden Fund [VUW1624].

Notes on contributors

Jack Vowles

Jack Vowles is Professor of Comparative Politics at Victoria University of Wellington – Te Herenga Waka. He leads the New Zealand Election Study. He has worked at several universities in New Zealand and spent five years as Professor of Politics at the University of Exeter. He has been a Fellow of the Royal Society of New Zealand Te Aparanga since 1996.

Matthew Gibbons

Dr. Matthew Gibbons is a researcher at Victoria University of Wellington Wellington – Te Herenga Waka. His PhD in politics is from the University of Waikato.

Jie Huang

Dr. Jie Huang gained his PhD at Victoria University of Wellington – Te Herenga Waka and now works at the Research Institute of Global Chinese and Area Studies (RIGCAS), Huaqiao University, Xiamen, China. His PhD thesis The power of habit and electoral participation of ethnic Chinese in New Zealand was submitted in 2023. His research interests include political participation, public opinion, and immigrants' social and political integration.