The Political Attitudes and Subjective Wellbeing of the One Percent

In capitalist societies, individuals who occupy the highest positions in the economic hierarchy feature prominently in the political discourse under the moniker of the One Percent. However, little is known about how the psychology of One Percent might differ from that of the average person. Using a large, nationally representative sample in New Zealand ( N = 14,650), we aimed to fill this gap examining the political attitudes and subjective wellbeing of the top one percent of the income distribution. We found that, compared to general public, the One Percent in New Zealand more strongly legitimize the political and economic systems in society, and express lower support for redistributive taxation. They also report higher life satisfaction, self-esteem and belongingness compared to everyone else. Thus, the One Percent benefit not only economically and politically from the current system, but also psychologically. Moreover, their political beliefs serve to bolster the inequality from which they benefit.

1 3 about how the One Percent exercise their power depend upon claims about their political attitudes as a group (see Gilens 2005). We aim to fill these gaps by comparing the political attitudes of the One Percent in New Zealand to the general population, using a large, representative sample (N = 14,650; see Table 1 for a comparison between New Zealand's One Percent and the One Percent in other major economies).
Several theories in social psychology propose that the political attitudes of (particularly) members of advantaged groups align with their own group's interests-i.e., they hold beliefs about the political and economic system that help maintain their own dominance within it (Sidanius and Pratto 1999;Rubin and Hewstone 2004). These perspectives would suggest that, having benefitted from the current political and economic systems, the One Percent would be more supportive of these systems and more opposed to policies that threaten their economic power.
On the other hand, there appears to be an increasing acknowledgement of the problem of inequality among the class of economic elites (Buffet 2011;WEF 2015), which undermines the common-sense notion that the One Percent will inevitably hold political beliefs that benefit their own group at the expense of everyone else. Indeed, System Justification Theory proposes that often it is not the most advantaged, but the most disadvantaged who show the strongest support for unequal systems cf. Brandt 2013). Therefore, more data is needed about how much the political attitudes of the One Percent differ from the general population and in which direction.
The extant literature is sparse and difficult to interpret. For example, an analysis of 397 individuals falling in the top one percent of income earners in the US found that higher proportion identified as 'Republican' (57%) compared to the rest of the population (44%), while a similar proportion identified as 'Conservative' (39% versus 40%; Saad 2011a). Another study, involving 83 of Chicago's wealthiest one percent, found that a higher percentage supported economically conservative policies than the percentages of people who supported such policies in a different general population survey (Page et al. 2013). However, both these studies are problematic. The former includes rather crude measures of political attitudes, while the latter is extremely limited in both its sampling frame and sample size. We aim to improve on these findings by testing more specific political attitudes in a representative national sample in New Zealand that includes a large number of 'One Percenters'.
This large, representative sample also allows us to compare the subjective wellbeing of the One Percent to the general population-another issue which has important implications The Political Attitudes and Subjective Wellbeing of the One… 1 3 for societal inequality but suffers from a scarce data and contradictory theoretical accounts. One strand of research that falls under the rubric of "Livability Theory" (see Veenhoven 1995) suggests that income does not predict wellbeing for people who are already fairly well-off. Evidence for this idea comes from cross-national studies showing no increases in national-level wellbeing once per capita income rises beyond USD 10,000 (Frey and Stutzer 2002;Helliwell 2003). Similarly, for people living in wealthy nations, large changes in national income across time have not been found to produce corresponding increases in wellbeing (Easterlin 1995;Oswald 1997).
On the other hand, robust effects of income on wellbeing have been found at the individual level, even at the upper ends of the income distribution in wealthy nations (see Easterlin 2001 for a review). However, the national surveys used to investigate these effects usually measure income using broad categories, the uppermost of which are not extreme enough to capture subjective wellbeing among the One Percent (e.g., Kahneman and Deaton 2010). A notable exception is a study by Diener et al. (1985b) that sampled super-rich individuals from the Forbes' list of wealthiest Americans and found that they were more satisfied with their lives than a matched control group from the same geographical area.
Understanding whether the One Percent do indeed have higher wellbeing than everyone else is particularly important because there are pervasive beliefs in society that belonging to high-status groups comes with certain downsides )-e.g., the wealthy incur a psychological cost for their wealth. These beliefs can function to legitimize inequality, making it harder to reduce it (Kay and Jost 2003). Therefore, in addition to examining the political attitudes of the One Percent, the present study will also examine their subjective wellbeing.
Thus, we aim to shed light on both the top-down and bottom-up processes that maintain inequality in society by examining, in the New Zealand context, whether (a) the elites have political preferences that exacerbate inequality, and (b) the beliefs about the elites held by the general public are accurate, or merely myths that further entrench existing inequality. We are able to fulfill these aims better than previous national surveys because instead of asking people to place themselves in one of a limited number of income categories, we asked them to report their income in an open-ended format, allowing us to better capture the extremes of the income distribution.
Specifically, in the political domain, we compare levels of three types of political attitudes that are relevant to the societal discourse on inequality-belief in the legitimacy of the political system, belief in the legitimacy of the economic system, and opposition to redistributive taxation-between the "One Percent", the "Rest of the Top 5%" and "Everyone Else" (see Table 2 for sample demopgrahics for these three income categories). We included people who fall in the Top 5% of the income distribution (while not reaching the level of the One Percent) to as an additional comparison group to see whether those on extremely high incomes differed from those who were merely on very high incomes. Despite the contradictory theoretical accounts (e.g. Jost et al. 2003;cf. Sidanius and Pratto 1999), the sparse evidence that currently exists (e.g., Page et al. 2013) points to the One Percent holding political preferences that exacerbate inequality. Based on this, we predict that the One Percent will show higher levels of belief in the legitimacy of the political and economic systems, as well higher opposition to redistributive taxation than the "Rest of the Top 5%" and "Everyone Else".
In the wellbeing domain, we compare levels of three indices of subjective wellbeinglife satisfaction, self esteem and felt belongingness-between the same three income categories. Despite the fact that prior studies have not been able to isolate the extreme upper end of the income distribution, the pattern of effects of income on wellbeing (e.g., Diener

Procedure
The NZAVS is an ongoing 20-year longitudinal national probability study of social attitudes, personality, and health outcomes that started in 2009. Each year, participants are posted a copy of the questionnaire, with a second postal follow-up 2 months later. Participants who provided an email address were also emailed and invited to complete an online version if they preferred. Detailed information about the sampling procedures, overall retention rates, demographic characteristics, and items included in the NZAVS questionnaires are provided on the NZAVS website (Sibley 2015). The NZAVS performs well in terms of representativeness (for more information see Sibley 2014), but oversamples women by about 10%, under-samples people in their 20 s, and those of Asian ethnicity. Here, we draw data from Time 6 (2014/15) of the NZAVS, which contained responses from 15,822 participants. We chose Time 6 because it had the largest sample size of any wave for which all target measures were available.

Income
Personal income was measured using an open-ended item that asked "Please estimate your own personal earnings (before tax) for the year 2014". For the purposes of the current analysis we divided people into three groups based on their self-reported personal income: The "One Percent", the "Rest of the Top 5%" and "Everyone else" (i.e. the bottom 95% of the income distribution). The income thresholds for being categorized into each group were determined using data from Inland Revenue Department, which provides the  (IRD 2015). Using data from 2014 (which pertained to income earned between April 1, 2013 and March 31, 2014) we determined that the One Percent of income earners were those who earned NZD 200,000 or more and the Top 5% of income earners were those who earned NZD 106,000 or more in the preceding year.

Political Attitudes
Perceived legitimacy of the political system was measured using four items adapted from Kay and Jost's (2003) General System Justification scale. This 8-item scale is the most widely used scale for measuring political legitimacy (see Jost et al. 2012 for a review). However, the scale has not been evaluated for its psychometric properties, and so there was no clear empirical criterion for selecting the smaller number of items to include in our short scale. Thus, we adapted the four items from the original scale that were most applicable to the New Zealand context: "In general, the New Zealand political system operates as it should", "In general, I find New Zealand society to be fair", "Everyone has a fair shot at wealth and happiness in New Zealand" and "Most of New Zealand's policies serve the greater good". This selection of items showed good internal reliability (α = .75) and have also been used in past research demonstrating good construct validity (Osborne and Sibley 2013;Vargas-Salfate et al. 2018a, b). Perceived legitimacy of the economic system was measured one item taken from the Economic System Justification scale developed by Jost and Thompson (2000). As with political legitimacy, there is no well validated scale for measuring economic legitimacy, and so selecting an item from an existing scale was the best remaining option. Although the items on Jost and Thompson's (2000) original scale are rather heterogenous, the social psychological literature on how people appraise the legitimacy of outcomes suggests that people can either make internal or external attributions for outcomes (e.g., Hewstone 1989;Jost and Major 2001). In the economic domain, this means making internal or external attributions for the distribution of economic resources-i.e., that economic outcomes reflect either people's own merits, or instead, have structural causes (see Bullock et al. 2003). We selected the one item from the Economic System Justification scale that we felt best captured this type of attributional appraisal of economic outcomes: "Economic positions are legitimate reflections of people's achievements". Support for redistributive taxation was measured by asking for people's degree of support versus opposition to the following policy: "Redistributing money and wealth more evenly among a larger percentage of the people in New Zealand through heavy taxes on the rich". This item was adapted from a Gallup Poll question that has been administered to thousands of Americans every year from 1999 to 2011 (Saad 2011b).

Subjective Wellbeing
Life satisfaction was measured using the following two items: "I am satisfied with my life" and "In most ways my life is close to ideal" (r = .70). These items were selected by taking the two items that showed the highest factor loadings on the latent construct from the 5-item Satisfaction with Life Scale as reported in the scale-validation study by Diener et al. (1985a). This short version of the scale has been used extensively in prior research, demonstrating excellent construct validity (Girme et al. 2016;Greaves et al. 2017;Osborne et al. 2016;Sibley and Bulbulia 2015;Stronge et al. 2016a, b).
Self-esteem was measured using the following three items: "On the whole am satisfied with myself", "Take a positive attitude toward myself" and "Am inclined to feel that I am a failure". These items were selected by taking the two highest-loading pro-trait items and the highest-loading con-trait item from Rosenberg's (1965) self-esteem scale as reported in a large-scale principal components analysis conducted on a New Zealand sample (in addition to samples from 52 other countries; Schmitt and Allik 2005). This three-item version of the scale showed excellent internal reliability (α = .80). Moreover, the construct validity of this scale can be established by the fact that it has been used extensively in prior research, yielding the expected correlations with other theoretically relevant constructs (e.g., Greaves et al. 2017;Lee and Sibley 2017;Manuela and Sibley 2014;Matika et al. 2017;Osborne et al. 2015;Stronge et al. 2016a;Waddell et al. in press).
Belongingness was measured using the following three items: "Know that people in my life accept and value me", "Feel like an outsider" and "Know that people around me share my attitudes and beliefs". Selecting items for the belongingness scale was less straightforward than the other two wellbeing constructs because the study in which the scale was first presented did not provide factor loadings for the individual items on the scale (Cutrona and Russell 1987). However, a subsequent validation study by Mancini and Blieszner (1992) did provide this information. Based on this analysis, we chose the highest-loading item from the 'social integration' subscale and the highest-loading item from 'reassurance of worth' subscale. In addition we developed a con-trait item that we thought best captured the lack of belongingness connoted by the con-trait items on the original scale (i.e., "I feel like an outsider"). This item was preferred because it captured a more general sense of social exclusion than the con-trait items on the original scale, which were more specific in content (e.g., "There is no one I can turn to for guidance in times of stress"; "Other people do not view me as competent"). This three-item version of the scale showed good internal reliability (α = .71). This scale has also been used in prior research, showing good construct validity (Stronge et al. 2015;Greaves et al. 2017). All dependent variables were rated on scales that ranged from 1 (strongly disagree/strongly oppose) to 7 (strongly agree/ strongly support).

Participants
The participants in this analysis were the approximately 14,600 individuals from the NZAVS Time 6 (2014/15) who had provided information on the relevant measures. This number is approximate because the exact sample size varied for each analysis depending on the number of people who had responded to the items that constituted the dependent variable in each analysis. In all, personal income data was available for 14,648 individuals and therefore it was these individuals who were subsequently categorized into the "One Percent", "Rest of the Top 5%" and "Everyone Else" groups. Of these, 326 participants (i.e., 2.2%) fell into the One Percent category, 1170 participants (i.e., 8.0%) fell into the Rest of the Top 5% category and the remaining 13,152 participants (i.e., 89.8%) fell into the Everyone Else category (based on the income thresholds described above). Thus, the NZAVS Time 6 (2014/15) oversampled the Top 1% and Top 5% of the income distribution.
As shown in Table 2, in the One Percent category, 77% were male, 41.6% had a University degree, and 90% were New Zealand European (with 6.6% being Maori, .6% being Pacific Islanders and 2.8% being Asian). The mean age of this group 51.37 years The Political Attitudes and Subjective Wellbeing of the One… 1 3 (SD = 8.87). Of the Rest of the Top 5%, 65% were male, 34% had a University degree, and 88.4% were New Zealand European (with 8.5% being Maori, 1.0% being Pacific Islanders and 2.1% being Asian). The mean age of this group was 50.00 years (SD = 9.35). Of Everyone Else, 34.1% were male, 41.1% had a University degree, and 80.5% were New Zealand European (with 12.6% being Maori, 2.7% being Pacific Islanders and 4.1% being Asian). The mean age of this group was 48.93 years (SD = 14.21).

Analysis Plan
We propose to conduct six one-way ANOVAs to test for mean differences in the three indices of well-being (Life Satisfaction, Self-Esteem and Belongingness) and the three kinds of political attitudes (Legitimacy of the Political System, Legitimacy of the Economic System and Support for Redistributive Taxation) between the three income categories (i.e. 'the One Percent,' 'the Rest of the Top 5%' and 'Everyone Else').

Results
Results of the six one-way ANOVAs are summarized in Fig. 1. Descriptive statistics and correlations between all variables used in these ANOVAs are presented in Table 3. Effects sizes for each mean difference tested are presented in Table 4. We selected a more conservative critical alpha of p = .01 due to the large sample size.

Legitimacy of the Political System
The one-way ANOVA testing for differences in the perceived legitimacy of the political system between the three income categories was significant (F (2, 14,599) = 124.26, p < .001). Bonferonni post hoc tests revealed that the One Percent (M = 5.11, SD = .98, 95% CI [5.00, 5.22]) perceived the political system to be significantly more legitimate on average than the Rest of the

Legitimacy of the Economic System
The one-way ANOVA testing for differences in the perceived legitimacy of the economic system between the three income categories was significant (F (2, 14,146) = 22.90, p < .001). Bonferonni post hoc tests revealed that the One Percent perceived the economic system to be significantly more legitimate on average (M = 3.89, SD = 1.72, 95% CI [3.70, 4.08]) than the Rest of the Top 5% (M = 3.54, SD = 1.67, 95% CI [3.44, 3.63], p = .002,

Opposition to Redistributive Taxation
The one-way ANOVA testing for differences in opposition to redistributive taxation between the three income categories was significant (F (2, 14,370) = 171.13, p < .001). Bonferonni post hoc tests revealed that the One Percent were significantly more opposed

Life Satisfaction
The one-way ANOVA testing for differences in life satisfaction between the three income categories was significant (F (2, 14216)

Self-Esteem
The one-way ANOVA testing for differences in self-esteem between the three income categories was significant (F (2, 14598) = 72.13, p < .001). Bonferonni post hoc tests revealed that the One Percent had significantly higher self-esteem on average (

Discussion
The very nature of the group-i.e., small and affluent-makes the One Percent extremely difficult to survey. Huge representative samples are required to capture large enough numbers of One Percenters to enable meaningful analyses. Moreover, the types of national surveys that do have the requisite sample size often do not assess income with enough precision to isolate the One Percent (e.g., Kahneman and Deaton 2010). As a result, we know very little about this group, despite the fact that it features prominently in the political discourse of many Western nations. We were able to take a step towards filling this gap in the present study. We found that One Percent in New Zealand believe more strongly in the legitimacy of the political and economic systems that govern society and express lower support for the inequality-reducing policy of redistributive taxation. These findings are consistent with self-and group-interest theories in psychology, which suggest that people hold political beliefs that align with their desire to advance their own position (e.g., Sidanius and Pratto 1999;Rubin and Hewstone 2004).
These results also provide evidence for a core assumption underlying much of the political discourse about the One Percent. This discourse holds that the concentration of wealth in the hands of the One Percent is corrosive for democracy because they use their political influence to maintain the inequality from which their group benefits (e.g., Stilgitz 2011). This argument appears to be supported by the finding that the types of policies that are implemented in democratic societies correlate more strongly with the policy preferences of economic elites than with those of the general public, especially when the preferences of these groups diverge (Gilens 2005). However, the missing piece in this narrative was whether the political attitudes of the elites do, in fact, diverge from the general public in a direction that would maintain existing inequality. By showing that this is indeed the case, our study helps explain part of the top-down process through which inequality is maintained in free-market democracies.
We are also able to speak to the bottom-up process that maintains inequality-i.e., the legitimizing myths held by the general public that prevent them from challenging inequality (Jost and Hunyady 2005). One of these myths is that no group in society 'has it all'-some people might be poor, but they are happy, and those that are rich are probably unhappy (Kay and Jost 2003). Indeed, beliefs about the unhappy rich are not confined merely to lay opinions. Much of the psychological literature on wellbeing also advances the notion that not only does money not buy happiness (e.g., Diener and Seligman 2004;Csikszentmihalyi 1999), it positively impedes the pursuit of happiness-inculcating a materialistic ethic that increases stress and relationship conflict, while lowering self-esteem, empathy, intrinsic motivation and sociality (Kasser et al. 2004;Vohs et al. 2008).
Whether or not high-earners face these kinds of tribulations, our findings suggest that they are still able to maintain levels of life-satisfaction, self-esteem and belongingness that, on average, outstrip not just the general population but even the fairly well-off (i.e., the 'Rest of the Top 5%'). Being in the One Percent seems to benefit people not only economically and politically, but also psychologically. One group can have it all.

Limitations
Here, we operationalized the One Percent category using the distribution of personal income rather than household income or wealth. Wealth inequality in Western nations is even more severe that income inequality, and that therefore, the 'true' elites are not necessarily those on high incomes, but the very wealthy (Piketty 2014). However, as difficult as it is to survey the upper echelons of the income distribution, it is even more difficult to survey those with large fortunes, or even to get them to report their wealth in questionnaires (see Page et al. 2013 for a discussion). Therefore, until data on the wealthy are available we must rely on data from high-earners to understand elite opinion on our societies. We were also unable to use household income because no data exists in New Zealand that is fine-grained enough to isolate the Top 1% of the household-income distribution. However, detailed information on the personal income of every New Zealand taxpayer is available (IRD 2015).
Our findings are also limited by the fact that we have not utilized fully validated measurement instruments for any of the constructs in our analysis. There are two reasons for this. First, the empirical literature on some constructs we sought to measure (e.g., political legitimacy) does not include the of use consistent, validated measurement instruments (see Sengupta 2016;p. 24, 27). Second, constructs for which well-validated scales do exist (e.g., self-esteem) are measured using scales that are too long to be viable for a large-scale, longitudinal, survey designed to measure hundreds of socially-important constructs.
As a result there is considerable variation in the empirical and conceptual strength of the criteria we could implement for each selection being made. However, the fact that we obtained the same pattern of effects across measures for which we had more robust item-selection criteria, and measures for which we had less robust criteria, gives us greater confidence that our findings reflect the true relationship between income and the two second-order constructs in our study-i.e., subjective wellbeing and political attitudes. Moreover, all but one of the measures (i.e., economic legitimacy) have already been used extensively in their current forms demonstrating excellent construct validity (e.g., Manuela and Sibley 2014;Osborne et al. 2016;Osborne and Sibley 2013;Stronge et al. 2015;Saad 2011b).
Nevertheless, there is a need for future research that uses more comprehensive measures to assess the relationships tested here. The need for more research is highlighted by the fact that, despite the limitations of our measures, our analysis provides a more comprehensive picture of the political attitudes and wellbeing of the upper end of the income distribution than existing studies. For example, coarse-grained income measures and a disproportionate focus on one particular dimension of wellbeing (i.e., life-satisfaction; see Diener and Oishi 2000) have resulted in a very limited current understanding of the wellbeing of the One Percent.
The breadth and validity of measurement in the political domain is even more constrained. For example, the only analyses of the political attitudes of the extremely well-off, to date, have used either a single, dichotomous measure of conservative versus liberal self-identification (Saad 2011a), or nominal scales measuring support for specific policy proposals (Page et al. 2013). Therefore, the present study represents an important incremental step in our understanding of the political psychology of an important but under-researched group.
Finally, the fact that our data come from one country-New Zealand-might limit the broader relevance of our findings. However, New Zealand is extremely similar to other Western nations in which debates about inequality and the One Percent are currently raging. For example, it has a highly developed free-market economy (Miller and Kim 2015), robust democratic institutions (EIU 2016), and a recent history of neoliberal economic policies that have been blamed for rising inequality since the 1980s, in a similar way to Thatcherism in the UK and Reagonomics in the US (Marcetic 2017).
Moreover, as one of the most unequal countries in the OECD (OECD 2017), New Zealand's One Percent represent the same class of economic elites that are referred to in the political discourse of other capitalist societies (and accrue very similar sociopolitical advantages as a result of this elite status; see Table 1). Therefore, to the extent that the differences in political attitudes and wellbeing reported here are a function of group status, there is no reason to expect a substantively different pattern of results in other Western countries.
The Political Attitudes and Subjective Wellbeing of the One… 1 3

Conclusion
For the first time, our study has documented how the political attitudes and subjective wellbeing of the One Percent differ from the rest of the population. We found that, in New Zealand, this group shows higher inequality-enhancing political preferences and higher subjective wellbeing than the general public. Thus, we have taken the first step towards understanding the psychology of societal elites, and consequently, the processes by which inequality is maintained in free-market democracies.