Subgroups of Green Voters Identifying Distinct Subgroups of Green Voters: A Latent Profile Analysis of Crux Values Relating to Support

Lucy J. Cowie, Lara M. Greaves, Chris G. Sibley ,

Abstract In the present study we aim to explore this possibility by assessing whether The Green Party experienced unprecedented support in the 2011 New there are distinct subgroups of Green Zealand General Election. However, people may vote Green for very voters who differ in terms of their core different reasons. The Green voter base is thus likely to be comprised of social values and level of environmental a number of distinct subpopulations. We employ Latent Profile Analysis concern. to uncover subgroups within the Green voter base (n = 1,663) using data from the New Zealand Attitudes and Values Study at Time Four (2012). We The Green Party of Aotearoa delineate subgroups based on variation in attitudes about the environment, A brief history and context of the equality, wealth, social justice, climate change, and biculturalism. Core Green Green Party is warranted at this point. Liberals (56% of Green voters) showed strong support across all ideological/ The Green Party of Aotearoa can be value domains except wealth, while Green Dissonants (4%) valued the traced as far back as May 1972 to the environment and believed in anthropogenic climate change, but were low formation of the New Zealand Values across other domains. Ambivalent Biculturalists (20%) expressed strong Party, which won 2% of the vote in the support for biculturalism and weak support for social justice and equality. 1972 election (Bale & Wilson, 2006). Greens in Principle (20%) supported equality and social justice, but were Although the obtained an less supportive of biculturalism. Our study identifies points of convergence increased share (5.2%) of the vote in (such as environmental values) and crux values that represent points of the 1975 election, the electoral system divergence (such as valuing social justice and Māori rights) across distinct of the time (First Past the Post) meant subpopulations of Green voters. These results highlight the diversity of the that this did not translate into any seats Green voter base and identify different crux points the Green Party must in parliament. In 1990, members of the manage in order to maintain and grow their diverse voter base. Values Party, including future co-leaders and , Keywords: The Green Party, Latent Profile Analysis, Values, Voter Behaviour. formed the Green Party (Green Party of Aotearoa, 2015). Under this new banner, the Green Party won 6.8% (and no seats) in the 1990 election. In 1993, These islands shine in the world Commission, 2014). However, since the the Greens entered into an alliance with for a tradition of ecological introduction of MMP, the Greens have a number of other left-wing parties, living, fair trade, human rights arguably undergone fundamental shifts including NewLabour, the Democratic and peace. in policy priorities, most of which relate Party (previously Social Credit) and to an increasingly diverse focus that —Green Party of Aotearoa Vision . Under the Alliance, includes a strong voice on social policy. Statement (2014) the Greens successfully campaigned We argue that the Green Party for the introduction of MMP and won must maintain a careful balance in The Green Party of Aotearoa three of the thirteen Alliance seats at the representing the interests and values benefitted greatly from the introduction introduction of MMP in 1996. However, of a potentially diverse voter base with of Mixed Member Proportional in 1997 the Greens announced that they varying levels of concern for ecological representation (MMP) in New Zealand would leave the Alliance and contest the living, social justice, and human rights. in 1996. In the elections since, the 1999 election in their own right (Bale As the party grows in appeal to a larger party has consistently attained at least & Wilson, 2006). While the Alliance voter base, there may, however, be five percent of the party vote and thus was later disestablished, the Greens difficulties in maintaining this balance. reached the threshold to gain seats in continued on to consistently win at least It is possible, for example, that the Parliament. In the recent 2014 General 5% of the vote (and therefore win seats party may risk fragmenting or losing Election, the Greens cemented their in Parliament) at each election since the part of their potential voter base as a position as the third largest political 1999 election (Wilson, 2010). result of alienating more conservative party in New Zealand with 10.7% The core policy priorities of the environmentalists through liberal social of the vote (New Zealand Electoral Greens have undergone a number policies in non-environmental domains.

New Zealand Journal of Psychology Vol. 44, No. 1, March 2015 • 45 • Lucy J. Cowie, Lara M. Greaves, Chris G. Sibley of changes since the establishment Party voters based on different patterns builds a model by creating a latent of the Party. In our reading of the of endorsement for various attitudes and variable which accounts for the hidden political landscape, the Greens have values relevant to the Green Party. LPA structure of response patterns across shifted from primarily emphasising enables us to test the premise that when manifest variables, thus uncovering environmental concerns to being it comes to attitudes and values, Green distinct sub-populations, or subgroups increasingly committed to issues which voters are not all similar. Instead there of Green voters. Furthermore, this reflect broader social values, such as may be distinct subgroups, which all method estimates the fit of the model strong opposition to neoliberal reforms, voted for the Green Party, but who differ to the given data, thereby allowing militarism, and inequality. Although across key values and attitudes, and may various models to be compared (Nylund, this can be seen as a way to potentially vote Green for quite different reasons. Asparouhov & Muthén, 2007). increase their voter base, broadening the Previous opinion on the Here, we employ LPA to determine platform also comes with a number of configuration of Green Party voters the number of subgroups which best risks. For example, this diversification has been divided. For example, in the represents the data. Importantly, risks diluting the core message of the Australian context, Manning (2002) LPA identifies subgroups (or latent Green Party, thereby undermining their suggested that Green parties may lose categories) without enforcing a priori ability to make tangible changes or support from their traditional voter- pattern of profiles, thus summarising attract voters. The incorporation of left base through liberal social values. In the variability of the data rather than wing social values also risks alienating comparison, Carroll, Casswell, Huakau, restraining the data to what has been those who may be more conservative Perry, and Howden-Chapman (2009) hypothesised (see Liu & Sibley, 2013, environmentalists, as they could be tested a similar hypothesis in the New 2015; for discussion of the application attracted by a purely environmental Zealand context by exploring social of LPA in the social sciences). The focus. Furthermore, the Green Party values as a possible reason for the application of LPA in the social sciences may be wary that this broadening of relatively weak support for the Green remains relatively novel, and has been focus may lead to a schism within the Party in the 2005 election. Carroll et al. used to examine patterns and variation Party (much like that of the Alliance) showed that support for environmental in topics such as types of paranormal whereby the Party becomes divided over policies was correlated with support belief (Wilson, Bulbulia, & Sibley, issues such as social justice, wealth or for social justice. They concluded that 2014), attitudes toward bicultural policy Māori rights. Green voters are likely to be liberal (Sibley & Liu, 2013), experiences of Uncovering the configuration of environmentalists. Thus, there seem to deprivation (Osborne, Sibley, Smith, Green voters’ values is of fundamental be two diverging perspectives on the & Huo, in press), beliefs about climate interest to both the strategists of the possible nature and composition of the change (Milfont, Milojev, Greaves & Green Party, and political psychologists, Green voter base. On the one hand, the Sibley, in press), representations of whose primary interests include the Green Party may risk being divided historical figures (Hanke et al., 2015), ideological underpinnings of vote over social justice issues. On the other, and types of sexism (Sibley & Becker, choice. Research within political previous data implies that there may be 2012). In the context of the attitudes and psychology has tended to explore this a reasonably homogenous core group of values of Green voters, LPA allows for question by investigating variables liberal environmentalists who form the the identification of subgroups ‘hidden’ which may predict voting for one party backbone of the Green voter base. Our within the data, without relying on our instead of another. Here, we investigate use of LPA allows us to explicitly test ability to necessarily hypothesise their within group differences to uncover these possibilities. existence or directly measure them a whether or not subgroups of voters priori. exist who all vote for the Greens but Latent Profile Analysis (LPA) Having identified distinct are distinct from one another in terms of To examine Green Party support, subpopulations of Green voters based key attitudes and values. For example, we apply LPA to uncover subgroups on diverging patterns of values (what just how diverse is the Green voter of Green voters who show a similar we refer to as latent subgroups), we base in terms of its values? Are there pattern of responses across a range of then explore demographic differences genuinely distinct subpopulations who theoretically relevant variables. LPA amongst the subgroups, such as voted Green for different reasons? And is a statistical method which is used differences in education level, income, what are the core points of convergence to uncover different groups or profiles age, gender and ethnicity in order to and divergence for such subpopulations? within a population (Hagenaars & identify the defining features of these Empirical research exploring such McCutcheon, 2002). LPA uses response profiles. These demographic differences questions is scant at best. Here, we patterns from multiple continuous between subgroups may have important present a statistical model identifying variables to group together participants implications for the stability and future sub-groups of Green Party voters from a into profiles, which we will refer to direction of the Green Party of New nationally representative sample of New as latent subgroups. In our case, these Zealand. Zealanders, the New Zealand Attitudes variables are a range of theoretically Our use of LPA in this context and Values Study (N = 12,182). We relevant attitudes and values, rated provides a novel contribution to the utilise Latent Profile Analysis (LPA), by the extent that each participant study of politics, as statistical modelling to create a model of subgroups of Green endorses that attitude or value. LPA of possible subpopulations is simply

• 46 • New Zealand Journal of Psychology Vol. 44, No. 1, March 2015 Subgroups of Green Voters not possible using earlier (and more among North American undergraduates. within the Green voter base. well-known) ordinary least squares Their results demonstrate significant regression-based approaches. Previous heterogeneity in policy endorsement on Importance of Values studies, for example, have tended to each side of the political spectrum, but There are a number of reasons to assess the extent to which variation in especially on the right. This suggests test values as points of convergence attitudes and values may predict support that people across the spectrum support or divergence within political parties for different political parties using more political ideologies as a result of a vast like the Greens. Here, we have well-known regression-based models. range of attitudes and beliefs. However, chosen to utilise the widely-used For example, Cutts, Ford and Goodwin Weber and Federico grouped participants Schwartz values as our measures of (2011) explored the values of the British according to where they fell on the environmentalism, support for social National Party to ascertain the unifying political spectrum. In comparison, we justice and value for wealth (Schwartz, features of the party. They showed that use a manifest behaviour, vote choice, 1992). This framework theorises a range of different demographic and to group participants and uncover their value for social justice, equality and attitudinal variables predicted support values. Application of LPA to a group the environment as being part of the for the British National Party, including which is defined by vote choice will domain of “universalism” which has racial prejudice and anti-immigrant reveal whether or not heterogeneity been associated with left-wing parties sentiment. Although informative of exists within a distinct voting bloc. in the past (Schwartz, Caprara & the general or average extent to which In sum, LPA has proven fruitful for Vecchione, 2010). In comparison, value different attitudes predict support for a identifying different distinct categories for wealth comes from the domain of given party, analyses of this type assume of people who support different political “power” which is more likely to be a ‘one-size fits all model’. Thus, they do parties (Greaves et al., 2014) and supported by right-leaning/conservative not allow for the possibility that there different combinations of social policy parties (Schwartz, 2010). It is thus may be distinct subgroups who express (Weber & Federico, 2013). However, as possible that many Green Party voters different combinations of strong and far as we are aware, LPA has not been may value social justice, equality and weak support for a diverse range of previously used to look at the different environmentalism as they go hand- issues. categories of people who may vote in-hand as universalism values. In LPA has been previously utilised for a political party for heterogeneous comparison, Green voters (especially to investigate political behaviour in reasons. To uncover classes which the most committed Green voters) may New Zealand more generally. Greaves, are significant to the Green Party, we not value wealth to the same extent, as Osborne, and Sibley (2014), for example, seek to build a latent profile model that it comes from a domain which is not as developed a model assessing the extent identifies ‘hidden’ or latent subgroups aligned with Green Party values. to which New Zealanders could be within the population of Green voters The alignment of New Zealand reliably categorised into different voter by differentiating them on a set of core political parties with various values profiles depending upon the types of attitudinal and social values which are has also been explored (e.g., Wilson, parties they tended to support (e.g., all relevant to the Green Party’s vision 2004). Vowles, Aimer, Catt, Lamare, Sole-National supporters, National and statement and policy. and Miller (1995), for example, argued ACT supporters, Labour and Green that New Zealand’s political parties supporters, Sole-Labour supporters). Green Party Values can be understood using a model of Using LPA, Greaves and colleagues Our analysis should help to resolve “old” and “new” politics. These old (2014) also identified a distinct latent the contention within the literature and political attitudes include issues such as voting bloc representing nearly a third the media surrounding division within state ownership of resources, welfare, of the sample, which they referred to the Green Party over social issues (e.g., regulation and unions. In comparison, as a ‘Fence-Sitter’ profile. This group Manning, 2002; Carroll et al., 2009; social justice and ecological values, of voters tended to be apathetic or Edwards, 2014). If our analysis identifies along with attitudes towards Māori, neutral in terms of their support for all two large groups that are divided across are framed as new political attitudes. political parties. This work by Greaves value for social issues and value for the Vowles et al. (1995) argue that the Green and colleagues demonstrates that LPA environment, this will provide empirical Party (at that time part of the Alliance) can be successfully utilised to explore evidence for the hypothesis that social were primarily focussed on these new political party support in New Zealand. issues are crux issues separating distinct political attitudes. Thus, in our model Our study represents a novel extension camps of Green voters. In comparison, we test environmentalism, value for of this as we explore converging and if a majority group emerges which social justice and attitudes towards diverging values within a political party displays strong support for both social Māori as new values which are crucial to uncover whether or not subgroups and environmental domains, this would to understanding the Green Party. exist within that party. provide support for the Greens’ current Additionally, LPA has also been balanced approach. However, it is also Value for the Environment used to uncover a variety of latent important to test a variety of other As environmental issues are a profiles in other domains. Notably, factors, such as support for Māori rights primary concern of the Green Party, Weber and Federico (2013) examined and value for wealth, so as to uncover and what they are largely known for, the endorsement of 19 policy issues whether these are further crux issues we would expect Green voters to value

New Zealand Journal of Psychology Vol. 44, No. 1, March 2015 • 47 • Lucy J. Cowie, Lara M. Greaves, Chris G. Sibley the environment to a greater degree than that many people may have of the (at that time part of Alliance). While these the average person. This is in line with ‘prototypical Green voter.’ However, it centrist and right-wing environmental O’Brien (2012), who argues that the is also possible that there are other voter parties experienced a low level of success of the Green Party is contingent profiles within Green supporters who support and consequently dissolved, on their ability to maintain a consistent value wealth to a greater degree. This the Green Party has continued with a environmental message which provides possibility is highlighted by the Greens’ commitment to social justice issues a link to more mainstream concerns. billboard campaign during the lead up above and beyond its environmental Thus, it is possible that environmental to the 2014 General Election, which focus. However, as these alternative support is a non-negotiable value of advocates for a “smarter economy” parties are no longer options, it is Green Party supporters which links in (Green Party of Aotearoa, 2014). possible that the Green Party has gained to more diverse issues. However, the Furthermore, the continued existence some highly environmentally-focussed extent to which they are homogenous in of “eco-consumerism”, by which voters who do not necessarily view social their support is unclear and it is possible environmentally-friendly products are justice concerns as being comparatively that there are within group differences sold at a premium, suggests that some as important as other values. Moreover, in terms of the degree to which Green environmentally minded people may there are potentially a number of voters voters value the environment. hold and value wealth. who are attracted to the Green Party Relatively high value for wealth is purely for their environmentalism and Belief in Anthropogenic Climate an important possibility to explore, as are either unaware of or unconcerned Change if this is the case, then this may point to with their liberal social values. Thus, it In line with its environmentalism, a so-called ‘fracture-point’ or division is important that we test value for social the Green Party asserts that climate within Green voters that the party may justice and equality as possible divisive change is caused by humans and is a need to carefully manage. In such a case, factors within Green voters. Along with genuine threat (Green Party of Aotearoa, the Green Party may have to walk a fine wealth, we suspect that such values may 2014). Although there has been some line in satisfying their supporters who provide another crux value which may evidence that the New Zealand value economic prosperity and those differentiate profiles, although perhaps population is not thoroughly convinced that would sacrifice it (if it were a simple may not differentiate them quite as the climate change is anthropogenic trade off), if it wants to retain support strongly given that we expect the overall (Sibley & Kurz, 2013), it is unlikely that from both groups. The potential for level of support for social values and this applies to Green Party supporters disagreement on this point is highlighted equality to be fairly strong (Sibley et due to the party’s unwavering stance by Milfont and colleagues (2012), al., 2011). on the issue. who argued that fear of a reduction in Support for Māori Rights and The centrality of climate change standards of living inhibits people’s Representation as a core platform for the Green Party support for parties who are actively is further supported by Milfont, Harré, challenging climate change. Although Sibley et al. (2011) found Sibley, and Duckitt (2012) who provide social justice to be endorsed to a large evidence that support for climate change Value for Social Justice and extent by all New Zealanders, attitudes actions predicts support for the Green Equality towards Māori were more contentious. Party. If there are indeed different latent Values of social justice and Thus, it is possible that Green Party subgroups within the Green voter base, equality tend to be endorsed by all voters’ attitudes towards Māori may it is possible that all of these subgroups New Zealanders, and are seen as a core also reflect this division. Although will express strong support for polices part of New Zealand identity (Sibley, we suspect that there may be a group that address climate change and protect Hoverd, & Liu, 2011). Although there of Green voters who see support for the natural environment. In comparison, may still be variation in these values Māori culture and political power as in we might expect them to differ in key within the Green Party, it will tend to be line with their support for social justice ways when it comes to support for in the range of strong support to those and equality, we doubt that similarly Indigenous rights, value for wealth and expressing extremely strong support. strong levels of support will be shared value for social justice. In our view, it is unclear as to whether across the party. Importantly, these Green voters, in particular, will be attitudes may be heterogeneous in Value for Wealth unified in their level of support for social themselves. Sibley (2010) argued that there are two ideologies which work Although Green voters are likely justice issues (keeping in mind that this together to legitimise material and to have relatively strong belief in is a question about relative levels). symbolic inequality in post-colonial anthropogenic climate change and high Historical events inform countries like New Zealand. Symbolic regard for the environment, it remains an understanding of the position of social Projection versus Exclusion indexes the open question as to whether they will be justice within Green parties. At the degree to which Māori culture is seen as similar in the extent to which they value introduction of MMP in New Zealand central to New Zealand identity, while wealth. It may be, for example, that in the 1996 election, a number of Historical Recognition versus Negation there is one core subgroup which is pro- environmentally-focussed parties indexes the degree to which historical environmental and anti-materialist. This formed as centrist and moderate-right injustices experienced by Māori are seen is perhaps the most salient stereotype alternatives to the left-wing Green Party

• 48 • New Zealand Journal of Psychology Vol. 44, No. 1, March 2015 Subgroups of Green Voters as relevant to contemporary society. profiles. These profiles describe groups Method Thus Symbolic Projection refers to who share similar levels of support the symbolic domain of intergroup across seven domains: value for the Participant Details relations, while Historical Recognition environment, belief in anthropogenic We limited our analyses to the describes the material. Although climate change, support for equality, 1,663 participants (1,135 women and Greaves, Osborne, Sengupta, Milojev, support for social justice, support for 528 men) who completed the NZAVS and Sibley (2014) found that overall Māori culture, recognition of colonial at Time 4 (2012) and stated that they increases in Green Party support are history and material rights for Māori, had voted for the Green Party with associated with decreases in Historical and value for wealth. As strong support their party vote in the 2011 General Negation and Symbolic Projection over across all of these variables except Election. This constituted 13.6% of the time, it is unclear if Green Party voters value for wealth is best representative full NZAVS sample. It is worth noting will be unified in supporting these of Green Party policy, this allowed us here that The Green Party won 11.1% of ideologies. to determine whether those who vote the overall vote in the 2011 election, so Furthermore, Sibley and Liu Green are unified in their values, or if the NZAVS oversampled Green voters (2013) provide evidence that people they are a diverse group with different by a margin of 2.5%. fall into distinct profiles of beliefs about motivations for supporting the Green Party. The mean age of participants in our Māori culture and the relevance of past sample of Green voters was 46 years injustices. They utilised LPA to test We expected there to be some (SD = 14.10). In terms of ethnicity, 94% attitudes towards biculturalism, with heterogeneity in domain support of Green voters identified as European four classes forming; Pro-Bicultural, within the Green Party, thus several (n=1,563), 12.7% as Māori (n=211), Moderate Differentiated, Bivalent distinct profiles should emerge. We 2.2% as Pacific (n=37), 2.7% as Asian Bicultural, and Anti-Bicultural. Sibley predicted that a large primary subgroup (n=45) and 2.5% reported another and Liu (2013) reported the percentage reflecting the Green Party position ethnicity or did not answer (n=41). The of people who voted Green in the of strong support across all domains majority of Green voting participants 2008 election who fit each of these except wealth would emerge, thereby were in paid employment (77.1%, profiles. Their results suggest that giving the party its mandate for policy. n=1,283). In addition, 28.1% identified almost half of Green Party voters However, we further hypothesised as religious (n=467). adopted the Moderate Differentiated the emergence of a smaller subgroup profile with moderate support across with weak support across all domains With regards to education, 4.8% material and symbolic domains. Only except support for the environment did not report their highest level of a few were opposed across domains, or and belief in anthropogenic climate education or said they had no education Anti-Bicultural, while Pro-Bicultural change, representing a group who are (n=79), 15.6% reported finishing some made up almost 20% of Green voters. purely motivated by environmental high school (n=259), 12.6% reported The remaining 28% supported Māori concerns. This group would encompass having studied towards a diploma or symbolically but not materially, which those who are primarily focussed on certificate (n=209), 37.1% reported that suggests that there is a high level of environmental issues but not concerned they had studied at undergraduate level diversity within the Green Party voter with social issues, and those who are (n=617), and 30.0% reported studying at base. However, as this measured voters unaware of the Green Party’s position the post-graduate level (n=499). at the 2008 election, the impact of the on social issues. Lastly, we expected increase in support for the Green Party there to be a group which is defined Sampling Procedure in the 2011 election remains unclear. by their support for Māori culture The Time Four (2012) NZAVS Moreover, it is important to investigate and political rights, so as to provide contained responses from 12,182 whether these attitudes, specific to a mandate for the Greens’ supportive participants (6,805 retained from one Māori rights, align significantly with position on these issues (Green Party or more previous waves, 5,377 new other attitudes, most notably broader of Aotearoa, 2014). Although we did participants). This sample was drawn concerns for social justice and equality. not predict any further distinct groups, from two sources. Of the sample LPA allows latent subgroups to emerge analysed here, 4,051 of these participants Overview and Guiding without a priori predictions. As there were retained from the original Time Hypotheses have been no Latent Profile models done 1 (2009) NZAVS sample of 6,518 participants. These participants were The current paper uses LPA to on the diversity of values within the randomly selected from the electoral roll investigate patterns of support for Green Party or within political parties (a national registry of voters, available ideologies and values of those who generally, it was unclear as to how many for research purposes). The initial gave their party vote to the Greens in distinct profiles would emerge, and what response rate of the original sample was the 2011 election. Using the responses form they would take. 16.6%, with a retention rate of 62.2% of those who indicated they gave their over three years. 2,705 participants party vote to the Greens in the 2011 were also retained from earlier samples, General Election in the fourth wave of having entered through non-random the New Zealand Attitudes and Values recruiting through a newspaper website, Study (NZAVS), we modelled latent recruiting through Pasifika networks

New Zealand Journal of Psychology Vol. 44, No. 1, March 2015 • 49 • Lucy J. Cowie, Lara M. Greaves, Chris G. Sibley and self-selecting in at previous years. a median of roughly 81 residents (M = a moderate level of deprivation Participants were posted a copy of the 95.95, SD = 73.49, range = 0-1899). In questionnaire, followed by a second 2013, at the time of the latest census, Questionnaire Measures copy two months later. Those who there were a total of 44,211 meshblocks Participants were asked “did you provided an email address were also for which data was available. vote in the last (2011) New Zealand invited to complete the questionnaire Statistics New Zealand (2013) general election?” Those who answered online instead. defined a meshblock as “a defined yes (89.7% of the sample) also then Secondly, participants in Time Four geographic area, varying in size from answered the open-ended question “If (2012) of the NZAVS were drawn from part of a city block to large areas of yes, to which party did you give your five independent booster samples aimed rural land. Each meshblock abuts electorate vote?” We selected only at increasing sample diversity. The against another to form a network participants who had indicated that they first of these consisted of a randomly covering all of New Zealand including had voted for the Green Party. selected sample of 20,000 people from coasts and inlets, and extending out to Social values were assessed using the 2012 New Zealand Electoral Roll, the two hundred mile economic zone. specific items from the Schwartz Values of which 2,431 responded (representing Meshblocks are added together to ‘build Scale (Schwartz, 1992) on a nine point a response rate of 12.34%). The second up’ larger geographic areas such as area scale, with -1 representing opposition frame consisted of a regional booster units and urban areas.” to values and all other values falling of 10,000 people who lived in the The New Zealand Deprivation between 0 (not important) to 7 (of Auckland region, randomly selected Index (Atkinson et al., 2013; Salmond supreme importance). Value for the from the New Zealand Electoral Roll. et al., 2007) uses aggregate census environment was assessed by the 890 participants responded to this information about the residents of each statement “Protecting the environment regional booster, representing a response meshblock to assign a decile-rank index (preserving nature)”, value for equality rate of 9.04%. The third frame was also from 1 (most affluent) to 10 (most was assessed by the statement “Equality a regional sample. 3,000 people living impoverished) to each meshblock unit. (equal opportunity for all)”, social in Christchurch who were randomly Because it is a decile-ranked index, justice was phrased as “Social justice selected from the Electoral Roll, of the 10% of meshblocks that are most (correcting injustice, care for the weak)” whom 333 responded (adjusted response affluent are given a score of 1, the next and wealth was assessed by “Wealth rate = 13.52%). The fourth frame 10% a score of 2, and so on. The index (material possessions, money)”. Ratings randomly selected 9,000 people who is based on a Principal Components were rescaled so that they ranged from lived in mesh block units which rated Analysis of the following nine variables 1 (low value) to 7 (high value) so that moderate to high in deprivation on the (in weighted order): proportion of they had the same range as other item New Zealand Deprivation Scale on the adults who received a means-tested responses in our model. Electoral Roll. Of these, 767 responded, benefit, household income, proportion Belief in anthropogenic climate representing an adjusted response not owning own home, proportion change was assessed using the item rate of 9.73%. The fifth sample frame single-parent families, proportion from Sibley and Kurz (2013): “Climate represented a random sample of those unemployed, proportion lacking change is caused by humans” on a scale who identified as Māori on the 2012 qualifications, proportion household of 1 (strongly disagree) to 7 (strongly Electoral Roll. Of the 9,000 randomly crowding, proportion no telephone agree). selected, 690 responded (adjusted access, and proportion no car access. response rate = 7.79%). Beliefs about the role of Māori The New Zealand Deprivation culture in contemporary society were Index thus reflects the average level of measured using three items from Deprivation at the Local Area deprivation for small neighborhood- Unit Symbolic Projection versus Exclusion type units (or small community areas scale developed by Sibley (2010). The We measured the affluence of of about 80-90 people each) across items were: “New Zealand would be a participants’ immediate (small area) the entire country. The index is a better place to live if we forgot about neighborhood using the 2013 New well-validated index of the level of trying to promote Māori culture to Zealand Deprivation Index (Atkinson, deprivation of small area units, and has everyone”, “I think that Māori culture Salmond, Crampton, 2013; see also been widely used in health and social helps to define New Zealand in positive Salmond, Crampton & Atkinson, 2007). policy research examining numerous ways” (reverse coded) and “I reckon New Zealand is unusual in having rich health outcomes, including mortality, that Māori culture should stay where it census information about each area unit/ rates of hospitalization, smoking, cot belongs – with Māori. It doesn’t concern neighborhood of the country available death, and access to health care, to other New Zealanders”. These were for research purposes. The smallest of name just a few examples (e.g., HURA rated on a scale of 1 (strongly disagree) these area units are meshblocks. The Research Alliance et al., 2006; Mitchell, to 7 (strongly agree) and averaged to NZAVS includes the meshblock code Stewart, Crampton, & Salmond, 2000; give an overall mean scale score. for each participant. The geographic Salmond & Crampton, 2000; Crampton, Beliefs about the relevance of size of these meshblock units differs Salmond, Woodward & Reid, 2000). In injustices experienced by Māori in depending on population density. Each our sample, Green voters’ mean score on today’s society were assessed by the unit tends to cover a region containing this scale was 4.67 (SD=2.66), indicating

• 50 • New Zealand Journal of Psychology Vol. 44, No. 1, March 2015 Subgroups of Green Voters

Historical Recognition versus Negation “How many children have you given model run on the sample to one another scale (Sibley, Liu, Duckitt, & Khan, birth to, fathered, or adopted?” and in terms of which model best explains 2008). Items included “We should those who answered 1 or above were the data (Nylund, Asparouhov, & all move on as one nation and forget coded as 1, and those who answered Muthén, 2007). As more parameters are about past differences and conflicts none were coded as 0. Participants included, more variance is explained. between ethnic groups”, “We should were asked “what is your relationship The BIC is known to penalise these not have to pay for the mistakes of our status?” and those who answered that increases more harshly, while AIC is ancestors” and “People who weren’t they were married, civil union, de facto, more liberal. Thus, a focus on the BIC around in previous centuries should not living together, or engaged were coded tends to lead to the selection of simpler feel accountable for the actions of their as 1, with single, dating, separated, and models, while a focus on the AIC in ancestors”. Items were rated on a scale widowed being coded as 0. isolation is more likely to lead to the ranging from 1 (strongly disagree) to 7 selection of more complex models (strongly agree) and averaged to give an Analytic Approach (Posada & Buckley, 2004). overall mean scale score. Latent Profile Analysis (LPA) was In interpreting the BIC and AIC, Support for the Green Party was used to examine the number of distinct smaller values suggest that more measured with the instruction to “Please subgroups emerging from the analysis variance has been accounted for. Results rate how strongly you oppose or support of values, beliefs about climate change indicated that a four profile solution each of the following political parties.” and attitudes toward biculturalism and provided a good fit to the data. The The Green Party was listed as one Māori. In essence, this analysis allowed final fit index we used to assess model of these parties and responses were us to test the core premise that when it performance was entropy. Entropy measured on a scale of 1 (strongly comes to these attitudes and values, not values range from 0 - 1.0, with higher oppose) to 7 (strongly support). all Green voters are similar. Instead, values representing improvements in In terms of demographic features, there may be distinct subgroups, which prediction (see Vermunt & Magidson, gender was asked with the question “Are all voted for the Green party, but who 2004, for discussion). Entropy indices you male or female?” participants could differ statistically in their values and for our analysis suggested that a four tick a box indicating that they identify attitudes, and may vote Green for quite profile model was as effective as a six as female (coded as 0) or male (coded different reasons. Having identified profile model and performed better as 1). Education level was measured whether there are distinct subgroups, than a five profile model. Thus on the by “what is your highest level of we extended our analysis to explore how basis of parsimony we opted for a four qualification?” and coded as an ordinal Table 1. Model fit for the different class solutions of the LPA. variable, with -2 as no qualification, 0 Class Solution BIC AIC aBIC Entropy as having obtained or studied toward a diploma/certificate, and 2 indicating Two 33372.391 33253.257 33302.500 .778 the participant had obtained or studied Three 32877.191 32714.735 32781.885 .790 towards a post graduate qualification. Four 32623.789 32418.013 32503.069 .814 Age was measured with the question “What is your date of birth?” and was Five 32490.401 32241.303 32344.266 .791 then coded in years. Six 32405.338 32112.918 32233.788 .814

Religion was gathered by the Seven 32313.994 31978.253 32117.029 .825 question “Do you identify with a religion and/or spiritual group?” and employment was measured with the these different subgroups differ in core profile solution. A four profile solution question “Are you currently employed?”, demographics, such as the proportion also provided a clear and interpretable with both being coded with no as 0 and of men and women, levels of socio- solution, with the identification of yes as 1. Birthplace was asked with the economic deprivation, age, ethnicity, additional profiles simply extracting question “Where were you born?”, with birthplace, parenthood, and education. more fine grained distinctions in the all answers that were “New Zealand” relative level of all indicators (with one being coded as a 1 and all others as profile splitting into two, both following a 0. Ethnicity was measured with the Results the same pattern, but in which one standard NZ Census question “Which profile reflected people with slightly ethnic group(s) do you belong to?” Model Estimation higher scores than the other), rather with those who answered New Zealand Latent Profile solutions ranging than qualitatively distinct patterns of European being coded as a 1 and all from two to seven profiles were combinations of high/low belief. others being coded as a 0 for the New specified in Mplus 7.2. Fit statistics for Finally, we estimated the probability Zealand European demographic item. these models are presented in Table 1. that each participant belonged to each For the Māori ethnicity question, those Bayesian Information Criterion (BIC), of the four classes. The probabilities who answered Māori were coded as the Akaike Information Criterion (AIC), (averaged across participants) that a one and all others were coded as a 0. and the sample size adjusted BIC (aBIC) given participant belonging to a given For parenthood, participants were asked are relative fit indices that compare each

New Zealand Journal of Psychology Vol. 44, No. 1, March 2015 • 51 • Lucy J. Cowie, Lara M. Greaves, Chris G. Sibley class would be correctly categorized domains of value for social justice, social issues when voting, and those are presented on the diagonal in Table value for equality, Symbolic Projection with no awareness of the Green Party 2. This provides an intuitive way to assess the reliability of the latent class model. As shown, these probabilities are Figure 1. Mean levels of support across attitude and value variables for the all well above .85, indicating excellent four subgroups and the overall NZAVS sample. classification likelihood and only a small average likelihood of incorrect classification.

Latent Subgroups Means for the levels of support across the seven attitudes and values domains for each of the subgroups and the overall NZAVS sample are shown in Figure 1. All latent subgroups were relatively similar in three of the seven domains; with high value for the environment, strong belief in anthropogenic climate change and low value for wealth. However, value placed on social justice and equality, along with support for Māori culture and recognition of the importance of past injustices against Māori proved to be issues which differentiated the groups The largest of the subgroups made up 56% of the sample. This group had relatively strong support across all domains except value for wealth and thus they were labelled as Core Green

Table 2. Latent Class probabilities for most likely classification by latent class membership (row) and latent class (column).

1. 2. 3. 4. 1. Green Dissonants .898 .066 .000 .035 2. Ambivalent Biculturalists .021 .860 .075 .044 3. Core Green Liberals .000 .053 .918 .029 4. Greens in Principle .018 .052 .066 .864 Values along the diagonal (shown in bold) represent the average probability that a person in a given latent class was correctly categorized as belonging to that class.

Liberals, as the values of this subgroup and Historical Negation. They were position on these social issues. best reflected current Green Party policy. strong in their value for the environment This subgroup was higher across all The two remaining latent subgroups and belief in human caused climate domains except value for wealth (on were of equal size and in direct contrast change, but were weaker in all other which they were lower) in comparison to each other across the four disputed domains. The small size of this subgroup to all other subgroups. In comparison, domains of value for social justice, suggests that very few people who vote the smallest subgroup, making up just value for equality, Symbolic Projection, Green do so purely for environmental 4% of the sample, was labelled Green and Historical Negation. The first of concerns, as we expect this small group Dissonants. This group had relatively these groups were labelled Greens to encompass those who knowingly weak support across the four contentious in Principle, due to their value for prioritise environmental concerns above

• 52 • New Zealand Journal of Psychology Vol. 44, No. 1, March 2015 Subgroups of Green Voters social justice and equality being Having identified a solution that LPA and is presented in Table 3. We relatively strong but their support for performed well in terms of model fit assigned the majority group, the Core Symbolic Representation and Historical while also making theoretical sense, Green Liberals, as the reference class Recognition being relatively weak. This we then examined differences in the due to their size and alignment with pattern represented the values of 20% demographic features of each group. Green Party policy. A significant effect of the sample. In comparison, the final We conducted a multinomial logistic of gender was found for each of the group we detected made up the final regression predicting the likelihood groups, with more men in the Greens 20% and were labelled Ambivalent of differential profile membership in Principle (b = .394, p = .024), Green Biculturalists. This group valued social based on participants’ demographic Dissonants (b = .696, p = .044) and justice and equality to a weaker extent, features of age, deprivation, gender, Ambivalent Biculturalist (b = .402, p = but had strong support for symbolic ethnicity, religious status, birthplace, .032) groups relative to the Core Green Māori culture, and relatively strong employment, relationship status, and Liberals. This means that the Core support for material support for Māori education levels. This regression Green Liberal group had a significantly when compared to the other subgroups. analysis used a three-step approach higher proportion of women than any Demographic Differences which weighted the parameters based other group. on classification likelihoods from our

Table 3. Multinomial logistic regression predicting likelihood of differential profile membership (reference category is Core Green Liberals)

b se t p Green Dissonants Intercept/Threshold -2.065 1.503 -1.367 .172 Gender (0 female, 1 male) .696 .346 2.011 .044 Age (years) -.014 .015 -.953 .340 NZ Deprivation Index 2013 (1 – 10) -.042 .061 -.688 .491 New Zealand European (0 no, 1 yes) 1.155 1.208 .956 .339 Māori (0 no, 1 yes) -.887 .672 -1.320 .187 Born in New Zealand (0 no, 1 yes) -.618 .361 -1.710 .087 Religion (0 no, 1 yes) .045 .350 .127 .899 Parent (0 no, 1 yes) -.397 .399 -.995 .320 Relationship (0 no, 1 yes) .218 .394 .554 .580 Employment (0 no, 1 yes) -.306 .366 -.836 .403 Education (ordinal -2 to 2) -.535 .135 -3.957 .000 Greens in Principle Intercept/Threshold -.954 .583 -1.637 .102 Gender (0 female, 1 male) .394 .175 2.249 .024 Age (years) .021 .007 3.003 .003 NZ Deprivation Index 2013 (1 – 10) -.009 .032 -.286 .775 New Zealand European (0 no, 1 yes) -.390 .372 -1.048 .294 Māori (0 no, 1 yes) -1.067 .363 -2.937 .003 Born in New Zealand (0 no, 1 yes) -.503 .191 -2.638 .008 Religion (0 no, 1 yes) .119 .177 .673 .501 Parent (0 no, 1 yes) -.333 .191 -1.740 .082 Relationship (0 no, 1 yes) .210 .195 1.075 .282 Employment (0 no, 1 yes) -.072 .206 -.352 .725

New Zealand Journal of Psychology Vol. 44, No. 1, March 2015 • 53 • Lucy J. Cowie, Lara M. Greaves, Chris G. Sibley

b se t p Education (ordinal -2 to 2) -.350 .076 -4.622 .000 Ambivalent Biculturalists Intercept/Threshold -.368 .566 -.651 .515 Gender (0 female, 1 male) .402 .188 2.140 .032 Age (years) -.012 .007 -1.599 .110 NZ Deprivation Index 2013 (1 – 10) -.043 .035 -1.238 .216 New Zealand European (0 no, 1 yes) -.484 .393 -1.231 .218 Māori (0 no, 1 yes) -.710 .313 -2.270 .023 Born in New Zealand (0 no, 1 yes) .579 .250 2.313 .021 Religion (0 no, 1 yes) -.515 .209 -2.464 .014 Parent (0 no, 1 yes) .175 .216 .812 .417 Relationship (0 no, 1 yes) -.118 .198 -.595 .552 Employment (0 no, 1 yes) .029 .224 .131 .896 Education (ordinal -2 to 2) .090 .079 1.129 .259

Note. These results employed a three-step approach which weighted the estimates based on the likelihood of profile membership.

Education levels also significantly membership. Those in the Greens in demographic variables were entered predicted differential group membership. Principle group were significantly into a regression analysis predicting Lower levels of education predicted less likely to be born in New Zealand the classification probabilities of each membership in the Greens in Principle (b (b = -.503, p = .008), while those in of the four classes in turn (which = -.350, p < .001) and Green Dissonants the Ambivalent Biculturalists were were estimated and saved as scores (b = -.535, p < .001). There was no significantly more likely to be born in in the primary analysis). We then significant difference between the New Zealand (b = .579, p = .021) when used the values from this regression Ambivalent Biculturalist group and compared to the Core Green Liberals. equation to estimate the proportion of the Core Green Liberals (b = .090, p There was no significant effect for the each demographic group classified as = .259). Thus, those in the Core Green Green Dissonants. belonging to each class, following the Liberals are more educated than those Religious affiliation significantly approach implemented in Milojev et in the Greens In Principle and Green predicted group membership for the al. (2014). Note that these proportions Dissonants profiles, but not those in the Ambivalent Biculturalist group (b = are based on the unique associations Ambivalent Biculturalist subgroup. -.515, p = .014). This means that the of each demographic, adjusting for A significant effect of age was only Core Green Liberals had a significantly all other demographics listed in Table found for one profile, with older people higher proportion of religious members 3. These results provide a descriptive being more likely to fall in the Greens than the Ambivalent Biculturalist group. supplement to the inferential statistics in Principle group (b = .021, p = .003) This was not the case when compared presented in the formal multinomial than the Core Green Liberals. Age did to any other profile. logistic regression model presented in Table 3. The overall proportion of not significantly predict membership The remaining demographic each demographic who voted for The into any other profile. variables included in the model were Green Party or another party in the Māori ethnicity predicted group unassociated with class membership. NZAVS sample is also included in membership for the Greens in Principle Thus, socioeconomic status (as Table 4. As shown, and consistent with (b = -1.067, p = .003) and Ambivalent assessed by the NZDep 2013 measure), logits reported in Table 3, women were Biculturalist (b = -.710, p = .023) employment, parenthood, relationship more likely to be Core Green Liberals, profiles. Therefore, the Core Green status, and European ethnicity did not whereas men tended to be more highly Liberals has a significantly higher uniquely predict profile membership. concentrated in the other subgroups. percentage of people identifying as The proportion of men, women, Similarly, those of Māori ethnicity were Māori than the Greens in Principle and European and non-European, and so more highly concentrated in the Core Ambivalent Biculturalist groups, but not forth, in each of the latent profiles are Green Liberal subgroup than any other the Green Dissonants. presented in Table 4. Analyses of the subgroup. Country of birth also had a proportion of people in each profile were significant effect in predicting profile based on a secondary analysis in which

• 54 • New Zealand Journal of Psychology Vol. 44, No. 1, March 2015 Subgroups of Green Voters

Table 4. Proportion of men, women, European and non-European, and so forth, in each of the latent profiles, and that voted for The Green Party or not in the total NZAVS sample (analysis of demographic proportions in each latent profile are based on a secondary analysis of classification likelihoods from the primary logistic regression model).

Green Subtype Classification Likelihoods Total NZAVS Green Ambivalent Core Greens In N o t V o t e d Voted Dissonants Biculturalists Green Principle Green Green Liberals Gender Men .052 .222 .496 .231 .867 .133 Women 033 .192 .589 .186 .822 .178 European Yes .040 .199 .563 .197 .926 .074 No .012 .239 .493 .255 .826 .174 Māori Yes .024 .192 .674 .111 .871 .129 No .041 .203 .543 .213 .833 .167 Born in NZ Yes .036 .212 .566 .186 .840 .160 No .052 .158 .535 .255 .835 .165 Religion Yes .037 .162 .581 .219 .891 .109 No .039 .217 .551 .193 .802 .198 Parent Yes .036 .214 .567 .183 .868 .132 No .044 .180 .546 .230 .741 .259 Partner Yes .038 .197 .554 .210 .847 .153 No .040 .210 .571 .178 .821 .179 Employed Yes .037 .202 .562 .199 .826 .174 No .046 .200 .550 .203 .872 .128 N 60 309 859 307 8683 1663

Note. Analyses of the proportion of people in each profile were based on a secondary analysis in which demographic variables were used to simultaneously predict each classification probability (CPROB in Mplus, saved as part of the output of the primary model). We then estimated predicted values (proportions) for each classification probability at (centered) conditional values for each categorical covariate. The predicted values thus adjusted for all other demographics included in the model. Note that we did not estimate conditional values for continuous covariates included in the primary model. 126 cases were excluded from this secondary analysis due to missing data on one or more demographic variable, yielding N = 1,535 for model predicting classification likelihoods.

New Zealand Journal of Psychology Vol. 44, No. 1, March 2015 • 55 • Lucy J. Cowie, Lara M. Greaves, Chris G. Sibley

Differences in Support for the levels of support for the party overall. Consistent with our predictions, a Green Party Put another way, strong support for the small group formed which did not align Greens was not homogenous across with Green Party values except in their We conducted additional analyses different profiles of Green voters. support for the environment and belief examining whether the different Green in anthropogenic climate change. This voter subgroups varied in their mean group was labelled Green Dissonants level of support for the Green Party. Discussion and made up 4% of the sample. Although To examine this issue, we conducted The current study investigated the small, this group remained stable across a LPA which again estimated the same attitudes and values of Green voters to different profile solutions, suggesting solution as our primary analysis, but uncover whether or not subgroups of that it exists and is distinct from the modelled ratings of support for the Green voters exist. Over the past ten years other groups within the voter base. Green Party as a continuous outcome the Green Party has arguably shifted to The Green Dissonants profile would variable using the three-step procedure represent broader social values. This encompass those who are unaware of the developed by Lanza, Tan, and Bray shift could in turn risk alienating more Greens’ social policy, along with those (2013) to estimate distal outcome scores hardline environmentalists and thereby who are aware but purely motivated by in LPA. At step one, this approach lose the Greens an important voter base. environmental concerns. This group was allowed us to estimate a standard latent At the same time, the broadening of their significantly less educated than the Core profile model independent of covariates. policy platform away from a primarily Green Liberals. This suggests that there Step two then estimated the most likely environmental focus may, we suggest, are distinct groups of voters who have class variable, or the likelihood of each put the Greens at risk of creating a diverse motivations for voting Green. person’s classification in a profile. schism within their voter base, should The last two groups each made up In the third step, when using a distal their voter base be divided on social 20% of the sample and were divided approach, profile membership was used issues. It is thus important to explore across support for Māori culture and to predict covariates (here, demographic the values of Green voters to uncover recognition of past injustice and value factors) that were weighted to adjust for whether or not distinct subgroups of for social justice and equality. The misclassification in profile membership. Green voters exist, all of which vote for first of these profiles, the Ambivalent The extent to which people in one profile the Greens for different reasons. differed from those in other profiles Biculturalists, consisted of those who Our analysis identifies four distinct was then assessed using equality tests were strongly supportive of Māori subpopulations of people who all vote of the means and probabilities (for culture and recognised past injustice Green but who differ from one another continuous and categorical covariates) against Māori. Interestingly, they had in core social values and attitudes in across profiles. relatively low value for social justice key ways. Although all four of the and equality, suggesting that they were The overall test for mean differences subgroups that we identified valued the largely motivated by their concern for in support for the Green Party between environment, believed in anthropogenic Māori issues. This is in contrast to the the four Green voter subtypes was climate change valued social justice Green Party position, which situates significant (χ²(3)=221.38, p<.001). Of and equality, and placed relatively little broader social justice principles as the four classes, Green Dissonants (M importance on wealth, support for Māori underlying support for Māori (Bale & = 5.52) reported weakest support for culture and recognition of past injustice Wilson, 2006). Members of this group the Green Party, being significantly against Māori proved to be issues that were significantly more likely to have lower than the Core Green Liberals, distinguished the subgroups from one been born in New Zealand but less who had the strongest support (M = another. Thus, our first hypothesis that likely to be religious. This subgroup 6.41; χ²(1)=37.19, p<.001). The Greens there would be heterogeneity in attitudes were significantly less likely to be of in Principle also expressed weaker and values within Green Party voters Māori ethnicity when compared to the support for the Green Party (M = 5.63), was supported. Core Green Liberals, suggesting that and this group did not differ in mean As we predicted, the largest of the this subgroup is not motivated by self- support from the Green Dissonant groups reflected Green Party policy, interest as could be suggested. class (χ²(1)=.49, p = .483). Finally, with strong support across all domains the Ambivalent Biculturalists (M = Although we hypothesised the except wealth. This group was labelled 5.82) sat in the middle of the range, existence of these first three subgroups, the Core Green Liberals and constituted expressing significantly weaker support we did not predict a specific profile 56% of the sample. The formation of this for the Green Party relative to the Core evidenced by the Greens in Principle majority subgroup demonstrates that a Green Liberals (χ²(1)=90.37, p < .001), subgroup. The uncovering of unforeseen large proportion of Green voters have but stronger support than the Greens subgroups is made possible by LPA, values which align with current Green in Principle (χ²(1)=5.32, p =.021) and which creates models which best explain Party Policy. This group tended to be also (marginally) the Green Dissonants the heterogeneity of manifest variable more educated when compared to the (χ²(1)=3.77, p =.052). and how they interact with each other, Greens in Principle and Green Dissonant rather than enforcing an a priori pattern In sum, while the people in each subgroups, and had a significantly on the data. The Greens in Principle of these profiles all voted for the Green higher proportion of women when subgroup represented those who had Party in the 2011 General Election, they compared to any other group. strong support for social justice and expressed significantly different mean

• 56 • New Zealand Journal of Psychology Vol. 44, No. 1, March 2015 Subgroups of Green Voters equality, but weaker support for Māori high value for the environment, high people weigh various attitudes and culture and lower recognition of past belief in anthropogenic climate change values when voting, especially in a injustice against Māori. This subgroup and relatively low value for wealth. multi-party system, so as to explore this had significantly lower proportions of These were supported universally by relationship further. those born in New Zealand and people the four latent subgroups. Thus, it Our findings speak to this question of Māori ethnicity. They also had appears there is not a group within the to some extent. Our model indicates significantly lower education levels in Green Party who vote for only non- that only 4% of Green Voters (those comparison to the Core Green Liberals. environmental factors. The lack of a belonging to the Green Dissonant These last two groups provide further non-environmental group suggests profile) are motivated purely by their evidence for our first hypothesis, that that the Greens are able to attract concern for the environment. On the there are distinctly different groups supporters by linking environmental one hand, this could be seen to support within the Green Party who vote for the concern with more diverse issues, such Manning’s (2002) position that the Greens for diverse reasons. as social policy and support for Māori. Greens have alienated a voter base of Our findings would seem to Attracting supporters in this way has centrist environmentalists, as this group contradict the argument that the Green some interesting implications for the of solely environmentally concerned Party would benefit from maintaining future positioning of the Green Party, Green voters is small. However, it is a primarily environmental stance as attracting new voter bases may rely unclear as to whether there is a group (Manning, 2002; Edwards, 2014). The on the Greens’ ability to link their of people in the voter base who would Core Green Liberals embody the vision environmental position with economic vote for the Greens should they shift statement quoted in our epigraph, with or geopolitical concerns. to a primarily environmental focus. attitudes and values consistent with These findings highlight the nuances Instead, it appears likely that most a desire for New Zealand to embrace of the relationship between values and people are mobilised by a number of “a tradition of ecological living, fair voting behaviour. For example, we values, all acting in unison to provide trade, human rights and peace” (Green have found that the vast majority of the foundation for vote choice. Party of Aotearoa, 2014). This subgroup Green supporters do not vote purely as This position is in line with the constituted 56% of the Green voter a result of their environmental values, findings of Carroll et al. (2009) that base. As show in Figure 1, the Greens but in unison with other concerns, environmentalism was correlated with in Principle (20%) were similar to the such as concern for Indigenous rights left wing social values. Vowles and Core Green Liberals in their expressions and culture, or value for social justice colleagues’ (1995) work provides of strong support for social justice and and equality. Furthermore, no group further support for this position, as they equality. Put another way, roughly three had weak support for the environment, provide evidence that certain values quarters of Green voters thus expressed suggesting that people do not vote for cluster together, into groups of old strong support for social justice and the Greens if they do not value the and new political attitudes. Thus, the equality. The Ambivalent Biculturalist environment. However, it is unclear as Greens’ environmentalism can be seen (20%) subgroup were similar to the Core to whether or not this is their primary as working in unison with a number of Green Liberals in a different domain, motivation for voting Green. Although other new political attitudes (such as in their expressions of strong support we determined the extent to which support for Māori and social justice) to for Māori rights. Thus, three quarters voters endorsed various attitudes and motivate people to vote Green. Thus, it of Green voters also expressed strong values, we were unable to explore appears risky for the Greens to become support for Māori rights, although how these attitudes and values were a primarily environmentalist party, as it the people doing so only partially weighted in relation to their vote choice. is likely that very few people would be overlapped with those expressing strong For example, those in the Ambivalent motivated purely by environmentalism. support for social justice and equality Biculturalist group may choose to Rather, it appears that people vote as (i.e., those in the Core Green Liberal vote Green primarily as a result of a result of a number of attitudes and subtype who were high on both). This their attitudes towards Māori, with values, all acting in unison. implies that if the Green Party were to the environmentalism as a secondary The Green Dissonants profile change their position in any of these concern. includes voters who choose to vote for domains, it may risk departing from the Whether or not environmental a party which does not represent their values held by roughly three quarters concerns are primary motivation for social values and attitudes towards of their voter base, just a different three vote choice, or a secondary concern, Māori, and also those who perhaps quarters depending on the position in has important implications for both have no knowledge of the Greens’ question. the Greens and larger parties. For policies of strong support for social We also provide empirical support example, if future research uncovers justice, equality and Māori rights for O’Brien’s (2012) position that the that the environment is not a crucial and are only aware of the Greens’ success of the Green Party is contingent motivation for voting it may shift how environmental position. As the fourth on their ability to maintain a consistent environmental issues are treated in the wave of the NZAVS did not test political environmental message which provides lead up to future elections and during sophistication or knowledge, we are a link to more mainstream concerns. governmental terms. Researchers should unable to differentiate between these two The unifying factors of the party were seek to uncover the extent to which groups. Future research could answer

New Zealand Journal of Psychology Vol. 44, No. 1, March 2015 • 57 • Lucy J. Cowie, Lara M. Greaves, Chris G. Sibley this theoretically interesting question Green Liberals (56%), showed strong Greaves, L. M., Osborne, D., Sengupta, by testing the political sophistication support across all ideological/value N. K., Milojev, P., & Sibley, C. G. of Green voters to see whether this domains except wealth. By comparison, (2014). Politics and post-colonial group is itself heterogeneous in non- the smallest, Green Dissonants (4%), ideology: historical negation and symbolic exclusion predict political environmental values. valued the environment but expressed party preference. New Zealand Journal Our secondary analysis of party less support for the other ideological/ of Psychology, 43, 39-54. value domains we examined. Ambivalent support can provide some insight into Greaves, L. M., Osborne, D., & Sibley, C. these issues of political knowledge and Biculturalists (20%) valued equality and G. (2014). Profiling the fence-sitters in identification. We found that those in social justice to a lesser extent, while New Zealand elections: a latent class the Core Green Liberal group expressed they had relatively strong support for model identifying distinct voting blocs. stronger support for the Green Party than Māori culture and reparation for past Manuscript submitted for publication. any of the other three subgroups. This injustice. Greens in Principle (20%) Green Party of Aotearoa (2014). Climate finding is not surprising, as this group expressed strong support for equality Change Policy - Kicking the Carbon is voting for a party which fits with and social justice, but weaker support Habit. Retrieved from www.Greens.org their position across a range of attitudes for the rights of Māori. Our study Green Party of Aotearoa (2014). Election and values. However, it is interesting identifies points of convergence (such as 2014: Downloads. Retrieved from www. to note that the Greens in Principle environmental values) and crux values Greens.org subgroup expressed a similarly weak that represent points of divergence (such Green Party of Aotearoa (2014). Māori level of support for the Greens as that as valuing social justice and Māori Issues Policy. Retrieved from www. of the Green Dissonants class, while the rights) across distinct subpopulations Greens.org Ambivalent Biculturalist class was in of Green voters. As political parties Green Party of Aotearoa (2014). Vision the middle of the range. This suggests generally seek to increase their market Statement. Retrieved from www.Greens. that the Greens in Principle and Green share (see Lees-Marshment, 2001), a org Dissonants subgroups do not identify key challenge for the Green Party will be Green Party of Aotearoa (2015). The History with the Greens as much as those in the continuing to grow its voter base while of the Green Party. Retrieved from www. Greens.org Ambivalent Biculturalist, or to a greater representing the interests of the diverse degree, the Core Green Liberals. This subpopulations of Green voters. Hagenaars, J. A., & McCutcheon, A. L. is of interest to strategists of the Green (2002). Applied latent class analysis. New York: Cambridge University Press. Party and rival parties, as it suggests References that those in the Green Dissonants and Atkinson, J., Salmond, C., & Crampton, Hanke, K., Liu, J. H., Sibley, C. G., Paez, D., Gaines, S. O., Moloney, G., … Greens in Principle profile are not as P. (2014). NZDep2013 Index of & Cabecinhas, R. (2015). “Heroes” supportive of the Green Party and thus Deprivation. Department of Public Health, Wellington. and “Villains” of world history across more likely to change their vote in future cultures. PLoS ONE, 10, e0115641. elections. Bale, T. & Wilson, J. (2006). The Greens. In R. Miller (Ed.), New Zealand HURA Research Alliance, McLeod, D., Government and Politics Fourth Edition Cormack, D., Love, T., Salmond, C., Conclusion (pp. 392-404). Victoria, Australia: Oxford Robson, B., et al. (2006). Ethnicity, University Press. socioeconomic deprivation and Our analysis indicates that there consultation rates in New Zealand general may be distinct Green policies that Carroll, P., Casswell, S., Huakau, J., Perry, practice. Journal of Health Service attract quite distinct types of voters. P., & Howden Chapman, P. (2009). Research Policy, 11, 141-49. The Green voter base, in other words, Environmental attitudes, beliefs about social justice and intention to vote Green: Lanza, S. T., Tan, X., & Bray, B. C. is composed of a number of distinct lessons for the New Zealand Green Party? (2013). Latent class analysis with distal subpopulations who differ across a Environmental Politics, 18, 257-278. outcomes: A flexible model-based number of crux values. We employed approach. Structural Equation Modelling: Crampton, P., Salmond, C., Woodward, A Multidisciplinary Journal, 20, 1-26. Latent Profile Analysis to model these A., & Reid, P. (2000). Socioeconomic different subpopulations and examine deprivation and ethnicity are both Lees-Marshment, J. (2001). The marriage of their values, attitudes and demographic important for anti-tobacco health politics and marketing. Political studies, characteristics using data from the Green promotion. Health Education and 49, 692-713. voter base sampled from the broader Behavior, 27, 317-27. Liu, J. H., & Sibley, C. G. (2013). From and nationally representative New Cutts, D., Ford, R. & Goodwin, M.J. (2011). ordinal representations to representational Zealand Attitudes and Values Study. Anti-immigrant, politically disaffected profiles: A primer for describing and We uncovered four distinct profiles that or still racist after all? Examining the modelling social representations of attitudinal drivers of extreme right history. Papers on Social Representations, differed in their pattern of support across 22, 5.1-5.30. seven attitudinal domains; value for the support in Britain in the 2009 European Liu, J. H., & Sibley, C. G. (2015). environment, equality, social justice, elections. European Journal of Political Research, 50, 418–440. Representations of world history. In wealth, belief in anthropogenic climate G. Sammut, E. Andreouli, G. Gaskell, change, views about historical injustice Edwards, B. (2014). Political roundup: Will the gutted Greens go blue? New Zealand & J. Valsiner (Eds.), The Cambridge and reparations for Māori, and value for Herald, Retrieved from http://www. Handbook of Social Representations. Māori culture. nzherald.co.nz/opinion/news/article. Cambridge University Press. The largest of these profiles, the Core cfm?c_id=466&objectid=11336457 Manning, H. (2002). The Australian Greens

• 58 • New Zealand Journal of Psychology Vol. 44, No. 1, March 2015 Subgroups of Green Voters

and the handicap of left legacies. AQ: (2007). NZDep2006 Index of Deprivation. University Press. Journal of Contemporary Analysis, 74, Wellington. Department of Public Health, Weber, C.R. & Federico, C.M. (2013). 17-40. University of Otago Wellington. Moral foundations and heterogeneity Milfont, T.L., Harrѐ, N., Sibley, C.G. & Schwartz, S. H. (1992). Universals in in ideological -preferences. Political Duckitt, J. (2012). The climate-change the content and structure of values: Psychology, 34, 107-126. dilemma: Examining the association Theoretical advances and empirical tests Wilson, J. (2010). Greens. In R. Miller (Ed.), between parental status and political in 20 countries. Advances in experimental New Zealand Government and Politics party support. Journal of Applied Social social psychology, 25, 1-65. Fifth Edition (pp. 497-508.) Victoria, Psychology, 42, 2386–2410. Schwartz, S. H., Caprara, G. V., & Australia: Oxford University Press. Milfont, T. L., Milojev, P., Greaves, L. M., & Vecchione, M. (2010). Basic personal Wilson, M. S. (2004). Values and political Sibley, C. G. (in press). Socio-structural values, core political values, and voting: ideology: Rokeach’s two-value model and psychological foundations of climate A longitudinal analysis. Political in a proportional representational change beliefs. New Zealand Journal of Psychology, 31, 421-452. environment. New Zealand Journal of Psychology. Sibley, C.G. (2010). The dark duo of post- Psychology, 33, 155-162. Milojev, P., Osborne, D., Greaves, L. M., colonial ideology: a model of symbolic Wilson, M. S., Bulbulia, J., & Sibley, C. Bulbulia, J., Wilson, M. S., Davies, C. exclusion and historical negation. G. (2014). Differences and similarities L., Liu, J. H., & Sibley, C. G. (2014). International Journal of Conflict and in religious and paranormal beliefs: a Right-Wing Authoritarianism and Social Violence, 4, 106-123. typology of distinct faith signatures. Dominance Orientation predict different Sibley, C. G., & Becker, J. C. (2012). Religion, Brain and Behavior, 4, 104-126 moral signatures. Social Justice Research, On the nature of sexist ambivalence: . 27, 149-174. profiling ambivalent and univalent Mitchell, E., Stewart, A., Crampton, P., & sexists. European Journal of Social Acknowledgements: This research Salmond, C. (2000). Deprivation and Psychology, 42, 589-601. was supported by a Templeton sudden infant death syndrome. Social Sibley, C. G., Hoverd, W. J., & Liu, J. H. World Charity Foundation Grant Science and Medicine, 51, 147-50. (2011). Pluralistic and monocultural (ID: 0077). The funders had no New Zealand Electoral Commission (2014). facets of New Zealand national character role in study design, data collection Official Count Results -- Overall Status. and identity. New Zealand Journal of and analysis, decision to publish, Retrieved from www.electionresults. Psychology, 40, 19-29. or preparation of the manuscript. govt.nz Sibley, C. G. & Kurz, T. (2013). A model of As per the NZAVS data access Nylund, K. L., Asparouhov, T., & Muthén, climate change belief profiles: How much statement, a copy of the B. O. (2007). Deciding on the number does it matter if people question human anonymous data reported in each of classes in latent class analysis and causation? Analyses of Social Issues and NZAVS publication is available growth mixture modeling: A Monte Carlo Public Policy, 13, 245-261. from CS upon request from simulation study. Structural Equation Sibley, C.G. & Liu, J.H. (2013). appropriately qualified researchers. Modeling, 14, 535-569. Relocating attitudes as components Such data will be provided with O’Brien, T. (2012). Leaving the minors: of representational profiles: Mapping the explicit understanding that it The Green Party of Aotearoa New the epidemiology of bicultural policy is used solely for the purposes of Zealand and the 2011 General Election. attitudes using latent class analysis. Representation, 49, 69-81. European Journal of Social Psychology, replicating or otherwise checking the validity of analyses reported in Osborne, D., & Sibley, C. G., Smith, H. 43, 160–174. J., & Huo, Y. J. (in press). Doubling- Sibley, C. G., Liu, J. H., Duckitt, J., & Khan, scientific papers analysing NZAVS down on deprivation: Using Latent S. S. (2008). Social representations of data. Mplus syntax for the models Profile Analysis to evaluate an enduring history and the legitimation of social reported here is available on the assumption in relative deprivation theory. inequality: The form and function of NZAVS website (http://www.psych. European Journal of Social Psychology. historical negation. European Journal of auckland.ac.nz/uoa/NZAVS). Posada, D., & Buckley, T. R. (2004). Social Psychology, 38, 542-565. We thank Nikhil Sengupta, Sam Model selection and model averaging Statistics New Zealand (2013). Geographic Stronge and Yanshu Huang for in phylogenetics: advantages of Akaike Definitions. Wellington, NZ: Statistics helpful comments on an earlier information criterion and Bayesian New Zealand. http://www.stats.govt.nz/ draft of this manuscript. approaches over likelihood ratio tests. Census/2013-census/info-about-2013- Systematic biology, 53, 793-808. census-data/2013-census-definitions- The authors declare no competing interests. Salmond, C., & Crampton, P. (2000). forms/definitions/geographic.aspx Deprivation and health. In P. Howden- Vermunt, J. K., & Magidson, J. (2004). Chapman & M. Tobias (Eds.), Social Latent class models. In Kaplan, D. Corresponding Author Inequalities in Health: New Zealand (Ed.), The Sage handbook of quantitative Lucy Cowie 1999 (pp. 9-63). Wellington: Ministry methodology for the social sciences (pp. School of Psychology of Health. 175-198). United States of America: Sage University of Auckland Salmond, C., & Crampton, P, (2012). Publications. Private Bag 92019 Development of New Zealand’s Vowles, J., Aimer, P., Catt, H., Lamare, J., Auckland, New Zealand deprivation index (NZDep) and its & Miller, R. (1995). Towards Consensus? [email protected] uptake as a national policy tool. Canadian The 1993 Election in New Zealand Journal of Public Health, 103, 7-11. and the Transition to Proportional Salmond, C., Crampton, P., & Atkinson, J. Representation. Auckland: Auckland

New Zealand Journal of Psychology Vol. 44, No. 1, March 2015 • 59 •