The System Works Fine: The Positive Relationship Between Emphasis on Individual Explanations for Inequality and External Political Efficacy

This article addresses the largely overlooked question of whether explanations for inequality are related to appraisals of the political system. It posits a positive relationship between individual explanations for inequality and three indicators of appraisals of the political system: satisfaction with democracy, political trust, and external political efficacy. Individual explanations for inequality are a form of system justifying belief and constitute part of a wider ideological view of the status quo social order as just and defensible. This positive view of the functioning of society may flow over into appraisals of the political system, imply a positive disposition towards high-status groups including politicians, and remove the motivation to blame the political system for ongoing inequality (which is instead seen in a positive, meritocratic light). The relationships between explanations for inequality and appraisals of the political system are tested for the first time in the United States, using 2002 ANES data, and in Great Britain, using data from a survey fielded in 2014. The results in the United States show few consistent or significant relationships between explanations for inequality and any of the appraisals of the political system. However, the results in Great Britain show consistent, robust, and statistically significant positive relationships between individual explanations for inequality and external political efficacy. The inconsistency in these results may stem from the differing temporal and national contexts of the surveys. It is also likely that the ranking measures of explanations for inequality in the GB data distinguished respondents for whom individual explanations are particularly important, who have a less negative appraisal of external political efficacy. However, more work is required to investigate the effects of question format, the impact of national and temporal context, and the causal direction of the relationship between explanations for inequality and appraisals of the political system.


INTRODUCTION
This article addresses the question of whether explanations for inequality are related to appraisals of the political system. Building on the work of Osborne et al. (2015), the article investigates the relationship in the United States, using 2002 ANES data, and Great Britain, using data from a survey fielded in 2014. It extends the literature on the relationship between inequality and appraisals of the political system (Norris, 2015;Zmerli and Castillo, 2015) by shifting the focus onto what people see as the reasons for inequality. In doing so it tests the proposition that those who see unequal outcomes in society as meritocratic and just also have positive appraisals of the political system.
Appraisals of the political system are important both as a barometer of the health of democracy and an influence on public engagement with politics. Where such appraisals are negative it implies flaws or failings in the functioning of the political system and contributes to public disengagement with politics (Hooghe and Marien, 2013;de Moor, 2016). This is particularly problematic because public action is often a key driver of the change that is needed to improve the functioning of the political system. At the same time, people's explanations for inequality matter because they are deep-seated beliefs about how society works and are logically related to a range of other important opinions. For example, whether one believes that inequality stems primarily from structural or individual factors has implications for views on education policy, how the justice system should handle criminals, and what a reasonable level of pay inequality is. Yet, despite the importance of both sets of beliefs, few studies have considered the relationship between explanations for inequality and appraisals of the political system.
System justification theory provides a basis for supposing a positive relationship between the two sets of beliefs. The theory posits a general disposition in which the status quo is seen as just and defensible, which may exist amongst both low-status and high-status individuals. This disposition is embodied in a set of lower order beliefs, including explanations for inequality, that are relatively stable and shaped in early and formative years. Core beliefs affect less stable higher order beliefs about specific objects such as the political system. As such, an important area in which explanations for inequality have implications is appraisals of that system. People who explain inequality with reference to individual traits such as hard work and ambition tend to view the social system as meritocratic and inequality as justified. As such, they are also likely to view the political system, which manifests and influences societal inequality, in a more positive light. This matters because it also suggests that people with structural explanations for inequality, such as social background and group marginalization, are more negative about the political system. Ultimately, this relationship has important implications for engagement with politics and the health of democracy.

Appraisals of the Political System
Appraisals of the political system can be assessed using at least three measures: satisfaction with democracy, political trust, and external political efficacy. These appraisals relate to the whole system, or elements of it, and are distinct from appraisals of the government of the day. Although the performance of the current government can affect them (Catterberg and Moreno, 2006;Seyd, 2015;Citrin and Stoker, 2018), appraisals of the political system are broader and encompass other considerations as well. Dissatisfaction, distrust, and low efficacy regarding the political system are distinct concepts but have been characterized as features of a wider political alienation (Southwell, 2012).

Satisfaction With Democracy
Satisfaction with democracy has been described as a "cornerstone of system support" (Loveless and Binelli, 2020). It measures satisfaction with "how the democratic political system functions in practice" as distinct from support for the nation, the principles of democracy, specific institutions, or specific political actors (Linde and Ekman, 2003, p. 394;Weßels, 2015). It may be expedient to measure the concept via a battery of questions (Linde and Ekman, 2003) relating to satisfaction with a range of specific democratic institutions (Kölln and Aarts, 2021). However, it is commonly assessed via a single question that directly asks how satisfied people are with the functioning of democracy in their country (Wagner et al., 2009;Dassonneville and McAllister, 2020;Bekiaris and Daskalopoulou, 2021).
Whilst individual socio-economic factors shape baseline levels of satisfaction with democracy, its variations tend to be driven by the nature of political institutions and their performance (Kölln and Aarts, 2021). At the individual level, social capital is positively related to satisfaction (Bekiaris and Daskalopoulou, 2021) whilst people with low income weigh economic expectations more heavily when assessing democracy (Nadeau et al., 2020). Institutionally, proportional electoral systems tend to sustain higher satisfaction with democracy (Ergun et al., 2019), though a greater range of parties does not (Dassonneville and McAllister, 2020). Further, the quality of institutions embodied in the rule of law, regulatory regimes, and levels of corruption also shapes satisfaction with democracy (Wagner et al., 2009;Ergun et al., 2019).
Political polarization is negatively related to satisfaction with democracy, though less so when diverse topics are seriously considered by parties (Merkley et al., 2019;Hoerner and Hobolt, 2020). This indicates the importance of both input (procedural) and output (performance) considerations when assessing the functioning of democracy (Hobolt, 2012). Both electoral victory (Campbell, 2015;Leiter et al., 2019) and government proximity to the mean voter on a range of ideological dimensions prompt higher satisfaction with democracy (Ezrow and Xezonakis, 2011;Stecker and Tausendpfund, 2016), as do positive economic expectations and appraisals (Nadeau et al., 2019;Loveless and Binelli, 2020). Most pertinently, people who perceive economic inequality to be unfair tend to be less satisfied with democracy (Pfeifer and Schneck, 2017;Saxton, 2021). By implication, people who see inequality as fair, such as those who believe that it stems from individual effort and is based on merit, should be more satisfied with democracy. This provides initial evidence that not only levels and perceived levels of inequality but also beliefs about inequality have implications for appraisals of the political system.

Political Trust
In contrast with the focus of satisfaction with democracy on the general functioning of the system, "[p]olitical trust measures the extent to which individuals feel that political parties, government, or parliament fulfill their expectations, that is, their policy expectations" (Geurkink et al., 2020, p. 250). Considered in general terms: "Trust-political or otherwise-is relational and domain specific. That is, A trusts B to do X. Trust always has an object or target (B), which could be a person, group, or institution, and a domain of action (X) where trust is given or withheld." (Citrin and Stoker, 2018, p. 50).
Those with high levels of political trust (A) trust the political system and its representatives (B) to implement policy that is beneficial to the public (X). By this formulation, they trust in the "output legitimacy" of the political system, "judged in terms of the effectiveness of . . . policy outcomes for the people" (Schmidt, 2013, p. 2). The focus of political trust may be institutions at national or local level, those with protective functions, or those with orderrelated functions, and these distinctions generally hold across varied countries (Schneider, 2017). This cross-national consistency is matched by consistency in the structure of political trust between those with differing levels of education, suggesting that it is an equally rational construct across groups (van Elsas, 2015).
Contextual factors with positive implications for political trust include elections (Echeverría and Mani, 2020), good government performance (Catterberg and Moreno, 2006;Seyd, 2015;Citrin and Stoker, 2018), sociotropic economic performance and social provision (Ellinas and Lamprianou, 2014;Hooghe et al., 2017), as well as both income equality and perceived distributional fairness (Zmerli and Castillo, 2015). Such perceptions are more important in countries with lower levels of inequality, indicating the interaction of contextual and individual effects. This interaction is also reflected in the positive relationship between education and political trust in societies with low levels of corruption but the opposite relationship in societies with high levels of corruption (Hakhverdian and Mayne, 2012). Similarly, perceived unfairness of treatment by public officials is negatively related to political trust, especially in societies where such treatment is not the norm (Marien and Werner, 2019). This points to the importance of expectations, and disappointment, in driving political trust (Hooghe et al., 2017). However, although more educated people have both higher levels of political trust (Turper and Aarts, 2017) and lower disappointment stemming from lower expectations of politics (Seyd, 2016), it seems that the performance of politicians is more important than citizens' expectations in shaping their levels of political trust (Seyd, 2015).

External Political Efficacy
Political trust is distinct from external political efficacy, which is concerned with the responsiveness of institutions rather than whether they act in the public interest (Geurkink et al., 2020). Nevertheless, external political efficacy can be seen as a specific form of political trust, in which people (A) trust the political system and its representatives (B) to respond to citizens (X). In this sense, external political efficacy is trust in the "input legitimacy" of the political system, "judged in terms of . . . responsiveness to citizen concerns as a result of participation by the people" (Schmidt, 2013, p. 2). External political efficacy is distinct from internal political efficacy, which focuses on one's individual capacity to engage with the political system (Pollock, 1983;Craig et al., 1990). People with high levels of internal political efficacy believe that they, and people like them, are equipped with the knowledge and skills to navigate the political system, whether the system itself is particularly responsive or not.
External political efficacy is less responsive to changes in the political environment (Chamberlain, 2012) than political trust, and is decreasingly related to local political cultures (Chamberlain, 2013). In authoritarian regimes, the lack of electoral losers, absence of over-responsiveness to particular electoral groups, and need for leaders to establish nonelectoral bases of legitimacy may raise external political efficacy (Zhou and Ou-Yang, 2017). In democracies, neither political trust nor external political efficacy vary greatly between ethnic groups in the United States (Koch, 2019) but greater economic inequality depresses the latter (Norris, 2015). However, external political efficacy is raised by the clear signaling of responsiveness to citizens through restoring voting rights to those with past felony offences (Shineman, 2020) and, more broadly, through engagement initiatives (Wolak, 2018). Online political engagement has positive consequences for the related concept of collective efficacy (Halpern, 2017;Wagner et al., 2017), whilst awareness of citizen deliberation initiatives increases external political efficacy (Knobloch et al., 2020).
Thus, active efforts by political institutions and forums in which citizens can exercise their political muscles increase a sense of the responsiveness of the political system. In emphasizing the openness of politics to all citizens, such initiatives are important and necessary countersignals to the idea that some citizens are more influential than others (Bartels, 2018;Schlozman et al., 2020). This points towards the importance of inequality, which has negative implications for satisfaction with democracy (Wu and dChang, 2019), political trust (Zmerli and Castillo, 2015), and external political efficacy (Norris, 2015).

System Justification Theory
The relationship between inequality and appraisals of politics, and between explanations for inequality and tolerance of it (Roex et al., 2019), suggests that explanations for inequality are related to appraisals of politics. Explanations for inequality may be structural, including group marginalization and people's backgrounds, or individual, including hard work and ambition. Although these explanations are not mutually exclusive, people who prioritise the latter are manifesting a form of system justifying belief. They are more apt to view society as meritocratic and inequality as legitimate, with positive implications for appraisals of the political system.
System justification theory was developed to supplement social identity theory and particularly to address the perceived lack of explanation for the rarity of action by individuals and groups to challenge hierarchies that disadvantage them Rubin and Hewstone, 2004). 1 The theory has four key tenets: "1) there is a general ideological motive to justify the existing social order 2) this motive is at least partially responsible for the internalization of inferiority among members of disadvantaged groups, 3) it is observed most readily at an implicit, nonconscious level of awareness and 4) paradoxically, it is sometimes strongest among those who are most harmed by the status quo" (Jost et al., 2004, p. 881).
Despite the theory's attention to low-status and disadvantaged groups, ideological beliefs including system justification are uncorrelated with indicators of self-interest. They can be held by low-status groups, contradicting their self-interest, and highstatus groups , coinciding with their self-interest (Osborne et al., 2019). These beliefs explain "differences in the distribution of social and material goods in terms of differences in individual effort, talent, and merit and by holding people responsible for their outcomes" (O'Brien and Major, 2009). As such, they are a form of the fundamental attribution error, in which outcomes are attributed to individual dispositions rather than situational factors (Knight, 1998).
System justifying beliefs are relatively stable, lower-order, and ideological in nature (Peffley and Hurwitz, 1985;Feldman, 1988;Jost et al., 2009). Ideological beliefs are shaped by family socialization (Dinas, 2013(Dinas, , 2014 as well as the wider societal context during early and formative years (Inglehart and Norris, 2016;Grasso et al., 2019;Roex et al., 2019). Beliefs about the drivers of unequal social outcomes function on a similar level to other common ideological beliefs, for instance about whether working people get their fair share of the nation's wealth and whether children should be taught to obey authority in school (Evans et al., 1996;Heath et al., 1994). Such beliefs change slowly and underpin higher-order beliefs about specific objects such as the political system (Bem, 1970), which are less stable. This runs counter to the argument that external political efficacy drives system justifying beliefs (Osborne et al., 2015). However, the experimental evidence used to support that claim shows a specific relationship between the act of participating and beliefs about the particular decision-making process (Lind et al., 1990;Platow et al., 2006), which does not necessarily sustain the broader claim that appraisals of the political system influence system justifying beliefs.
System justifying beliefs can manifest themselves in many ways, including Right-Wing Authoritarianism or Social Dominance Orientation (Duckitt and Sibley, 2009), the Protestant Work Ethic, Belief in Status Legitimacy, Belief in a Just World (O'Brien and Major, 2005), and individual explanations for inequality. They also manifest themselves in views of society as meritocratic, which are closely related to explanations for inequality. 2 The literature on system justification and meritocratic beliefs provides three reasons to expect a positive relationship between individual explanations for inequality and appraisals of the political system.
First, those who view unequal outcomes in society at large as just and meritocratic are also likely to have a positive view of the functioning of the political system. System justification theory "posits a general tendency to defend, legitimize, and bolster the social and political systems on which people are psychologically dependent" (Thorisdottir et al., 2009, p. 8). Given that it is a general tendency, people who have a positive view of the social system, manifested in individual explanations for inequality, may also see the functioning of the political system as just, fair, and meritocratic. These views may flow into their appraisals of the political system, manifested in satisfaction with democracy, political trust, and external political efficacy.
Second, although politicians are often viewed negatively, they are also commonly considered to be a high-status group in a prestigious position. People with system justifying beliefs tend to have positive appraisals of high-status groups (Hafer and Choma, 2009;Grigoryan et al., 2020), and to see them as competent and deserving of their positions. Applying such appraisals to politicians, even if only to moderate the prevailing negative views of the group, implies a more positive assessment of the political system. This is especially so because it implies that the political system itself is meritocratic and functions in a just and defensible way.
Third, people with system justifying beliefs see inequality in society as justified and are less likely to perceive discrimination as a driver of low status (Major et al., 2002;Hafer and Choma, 2009;O'Brien and Major, 2009). In attributing low status to individual traits (Major et al., 2003), system justifying beliefs reduce both discontent with that status and action to challenge it (Hafer and 1 In recent years, a number of scholars who draw on social identity theory have argued that no distinct motive is needed to account for system justifying beliefs in disadvantaged groups. They advocate instead that such beliefs are driven by selfinterest and group-interest motivations that work through passive acceptance of the system, identification with a superordinate group such as a nationality, or hope that their group can attain higher status in future (Owuamalam et al., 2017;2019a). The critics have also highlighted a contradiction between the theory's emphasis on dissonance reduction for disadvantaged groups and its claim that this occurs in situations of low group-interest, when dissonance is already low (Owuamalam et al., 2016). Further, they have criticized a contradiction between the palliative function of such beliefs for low-status groups and its suggestion that high-status groups feel this benefit the most (Owuamalam et al., 2019b). Advocates of system justification theory have responded robustly (Jost, 2019;Jost et al., 2019), and both sides have claimed misunderstanding and confusion on the other. However, there is agreement (Owuamalam et al., 2019b, p. 372) that system justifying beliefs exist and are consequential.
2 As with the system justifying beliefs that they are a manifestation of, meritocratic beliefs reduce anger at being discriminated against (Maitner, 2015) and reduce selfesteem whilst raising perceived fairness of treatment amongst low status-groups who suffer such discrimination McCoy and Major, 2007). Such beliefs are more readily associated with upward than downward mobility (Evans, 1997) and reflect actual levels of meritocracy in income distribution within society (Kunovich and Slomczynski, 2007 Olson, 1993). They are also associated with an alignment between perceived and ideal levels of inequality in society (García-Sánchez et al., 2019). Thus, people with individual explanations have a more positive view of inequality and are less apt to attribute its existence to the social or political system. As such, they are less likely to be critical of flawed or inadequate political action to address inequality, which implies more positive appraisals of the political system.

Hypotheses
The preceding arguments suggest a positive relationship between individual explanations for inequality and appraisals of the political system. However, despite the theoretical grounds for viewing system justifying beliefs as lower-order ideological beliefs, and appraisals of the politics as higher order beliefs, the cross-sectional nature of the available data means that the causal order of the relationship cannot be tested. This is reflected in the circumspect nature of the hypotheses: -Hypothesis 1: Individual explanations for inequality in society (system justifying beliefs) are positively related to appraisals of the political system.
People also have explanations for their own statuses, which may not align with their explanations for inequality in society. For instance, it is possible to believe that structural constraints impact on inequality in general but also to believe that one's own status was earned through individual effort. The application of structural explanations for one's own status is very personal and recognizes that one's circumstances are not only the result of one's efforts but also of structural forces. Thus, the application of explanations for own status are expected to have similar implications as explanations for inequality in society. However, given that general and personal explanations do not necessarily move in lockstep, these should be specified separately: -Hypothesis 2: Individual explanations for own status (system justifying beliefs) are positively related to appraisals of the political system.

STUDY 1 Survey Details
The 2002 ANES Time Series Study (American National Election Studies, 2021) took place around the 2002 midterm elections in the United States and was conducted using computer aided telephone interviewing and random digit dialing. The pre-election survey wave obtained a sample of 1,511 respondents, of which 1,346 also completed the post-election wave. This is currently the only established survey of a representative national sample that includes questions about both explanations for inequality and appraisals of the political system.

Survey Measures
The pre-election survey asked standard questions covering a range of appraisals of the political system. Specifically, respondents were asked two questions that capture levels of satisfaction with United States democracy and trust in the federal government: "On the whole, are you satisfied, fairly satisfied, not very satisfied, or not at all satisfied with the way democracy works in the United States?" and "How much of the time do you think you can trust the government in Washington to do what is right-just about always, most of the time, or only some of the time?" To measure external political efficacy respondents were read the following two statements and asked whether they agreed, neither agreed nor disagreed, or disagreed with them: "Public officials don't care much what people like me think" and "People like me don't have any say about what the government does." Details of the distributions of the answers to these questions are presented in Table 1.
Questions covering explanations for inequality were asked in both the pre-and post-election surveys, and respondents were randomly allocated to answer the questions in one of the two waves. Within each wave, respondents were also randomly allocated to one of two conditions that asked about why people have "better jobs and higher incomes" or "'worse jobs and lower incomes." Respondents were offered seven explanations for job and income inequality relating to ability to learn, discrimination, hard work, availability of good education, choice of low-paid jobs, government policy, and differences created by God. Each explanation could be rated as "Very important," "Somewhat important," "Not important," or the respondent could volunteer that the statement was untrue.
The four conditions (pre-election, better jobs; pre-election, worse jobs; post-election, better jobs; post-election, worse jobs) each covered approximately a quarter of the sample, and the answer distributions did not differ significantly between the four treatments. 3 Respondents did not receive more than one version of the questions on explanations for inequality: those who answered the questions in the pre-election wave were not asked them in the post-election wave whilst those who were asked one version of the questions in either wave were not asked the other version of that question within the wave. 4 As such, combined variables encompassing substantive answers from all four treatments are used in the subsequent 3 Ordered logit regressions of the answers relating to each explanation for inequality on the treatment binaries show only one statistically significant difference: respondents in both of the post-election treatments are less likely than respondents in the pre-election "worse jobs and lower incomes" treatment to select the access to good education explanation. Further, when logit regressions are conducted with binary answer variables ("Very important" and "Somewhat important" vs. "Not important" and untrue) used as the dependents, there is also only one significant difference: respondents in the post-election "worse jobs and lower incomes" condition were significantly more likely than the respondents in either pre-election treatment to indicate that discrimination is important. The full results of these ordered logit and logit regressions are available in Supplementary Appendix C. 4 To check that respondents had not accidentally answered more than one version of the questions on explanations for inequality, and ensure no multiple counting of respondents, binary variables were coded indicating whether each respondent had answered the questions in each treatment. Crosstabulation of these binaries indicated that no respondents had answered more than one version of the questions on explanations for inequality. This check can be reproduced using the replication files linked to at the end of the article. analyses. The distributions of the answers to these combined questions are presented in Table 2. The data also contain the following standard demographic and political variables that are used as controls in the adjusted models: age, ethnicity, gender, education level, household income, party identity, ideology, political interest, and internal political efficacy. 5

Multivariate Analysis
Each of the indicators of appraisals of the political system is categorical so, to test Hypothesis 1, ordered logit regressions with sample weights and robust standard errors (which are the default when using sample weights) were used to estimate relationships with explanations for inequality. For ease of presentation, each explanation for inequality was coded into a binary ("Very important" and "Somewhat important" vs. "Not important" and untrue) before inclusion in the models. The explanations for inequality are weakly or moderately correlated and show no signs of multicollinearity so were included in the models simultaneously. 6 The unstandardized regression coefficients from these analyses are presented in Table 3 and show very few significant results. 7 Only the government policy explanation for inequality has a statistically significant relationship with any of the indicators of appraisals of the political system. It is positively related to the view that public officials do not care what the public think (adjusted model: coef. 0.423, SE 0.170, p < 0.05) and negatively related to satisfaction with how United States democracy works (adjusted model: coef. −0.329, SE 0.154, p < 0.05). Thus, respondents who indicate that government policy contributes to inequality offer  Numbers in each category are unweighted with percentages in parentheses.
5 Internal political efficacy was measured via two questions that asked respondents whether they agree, neither agree nor disagree, or disagree with the following two statements: "I consider myself well-qualified to participate in politics;" "I think that I am better informed about politics and government than most people." 6 The correlations and multicollinearity diagnostics for the explanations for inequality, and control variables included in the adjusted models, are available in Supplementary Appendix D. 7 The ordered logit regression results can be produced using the replications files linked to at the end of the article. negative views on two of the four indicators of appraisals of the government system. Insofar as structural explanations for inequality are opposed to individual explanations, this finding is consistent with Hypothesis 1, which posits a positive relationship between individual explanations and appraisals of the political system. However, this is sparse evidence in support of the hypothesis, and none of the other results indicate that a relationship exists.

Robustness Checks
The ANES data also contain measures of feelings (on a 0-100 thermometer) towards various state and political institutions, including the Supreme Court, Congress, and the federal government. OLS regressions with sample weights and robust standard errors were run with feelings towards each of those institutions, and a factor onto which they all load, 8 as the dependent variables. The unstandardized regression coefficients from these analyses are graphed in Supplementary Appendix B and show almost no significant relationships between explanations for inequality and feelings towards government institutions. There is also a concern that the two statistically significant results presented in the multivariate analysis section may stem from the close conceptual link between the government policy explanation for inequality and appraisals of government. This may obscure the relationships between other explanations for inequality and appraisals of the political system. As such, the models were re-estimated with the government policy explanation dropped but, as the coefficient plots in Supplementary Appendix B indicate, this did not notably alter the results. Finally, all models were also reestimated using the original ordinal explanations for inequality (treated as categorical variables) in place of their binary recodes, with similarly insignificant results. 9

Discussion
The picture emerging from Study 1 is of very few explanations for inequality, whether individual or structural, having significant or consistent relationships with appraisals of the political system. These results hold whether we focus on external political efficacy, satisfaction with United States democracy, trust in the federal government, or feelings about various government institutions. Further, the results are broadly replicated whether we use unadjusted or adjusted models and binary or ordinal independent variables. Only the government policy explanation for inequality, which is a structural explanation, is significantly related to appraisals of the political system, and then only to two of the four indicators. To the extent that structural and individual Coefficients are unstandardised, all models use sampling weights (pweight), and robust standard errors are presented in parentheses. Statistical significance: ***p < 0.01, **p < 0.05, *p < 0.1. Full results including the coefficients for control variables are available in Supplementary Appendix A.
8 The confirmatory factor analysis can be produced using the replication files linked to at the end of the article. 9 The only exception to this that the ordinal indicator of the hard work explanation is positively related to trust in government. This relationship is in line with Hypothesis 1 but constitutes limited support for it. The full results tables relating to all of the robustness checks can be produced using the replications files linked to at the end of the article.
Frontiers in Political Science | www.frontiersin.org July 2021 | Volume 3 | Article 643165 explanations for inequality are opposed, the negative form of this result is consistent with Hypothesis 1. However, the results overwhelmingly provide evidence that the hypothesized relationship does not exist. These results may be particular to the national or temporal context in which the questions were fielded, or to the manner in which they were asked. In that light, it is helpful to investigate the relationships between explanations for inequality and appraisals of the political system using more recent data from a different national context, and which asks about the concepts in a different way.

STUDY 2 Survey Details
Study 2 uses data that were gathered in an original two-wave survey focusing on political participation, levels of economic, social, and cultural capital, and perceptions of privilege in Great Britain. Amongst the items included in the surveys were questions on explanations for inequality in society and explanations for own status, which allow both Hypothesis 1 and Hypothesis 2 to be tested. 10 Further, the surveys included items covering the responsiveness of the political system, attention paid to public complaints by elected representatives, and the point at which such representatives lose touch with the people. All the measures relate to external political efficacy so the hypotheses cannot be tested in relation to satisfaction with democracy or political trust, which is a limitation of Study 2. However, the largest of the two significant results in Study 1 relates to a measure of external political efficacy (the view that public officials don't care what the public think) and the data used in Study 2 allow us to further investigate the relationship between explanations for inequality and external political efficacy. The survey that produced the data was fielded just over one year before the Conservative victory in the 2015 United Kingdom general election, and slightly more than 2 years before the 2016 Brexit referendum. Due to its length, the survey was split into two waves to maintain respondent engagement and separate the perceptual measures from the behavioral ones. Each wave was fielded to a sample of GB adults drawn from YouGov's online panel of respondents, the first wave between 17 March and April 1, 2014, and the second wave between 4 April and April 17, 2014. Overall completion time for the two waves was between 35 and 40 min, and 1,405 respondents (73.8% of first wave starters) completed both waves without displaying signs of satisficing (Krosnick, 1999). The respondents each have unique identities on the YouGov system and could only see the questions in each wave once, meaning that it is not possible for respondents to be counted multiple times.
YouGov uses quotas based on census and electoral data to recruit samples from amongst its online panel of voluntary respondents. The sample recruited for this study is broadly representative of the GB population in terms of gender and region of residence but is less so in terms of age. Those in their late teens, 20s, 40s, and 80s or older are underrepresented, whilst those in their 30s, 50s, 60s, and 70s are overrepresented. In terms of ethnicity, the sample overrepresents White British respondents whilst underrepresenting those in other ethnic groups (Ford et al., 2015). Further, the sample overrepresents those with higher level qualifications (A-level and above) and concomitantly underrepresents those with lower level (GCSE or below) or no formal qualifications. Finally, although the sample is largely representative in terms of choice in the 2010 general election amongst those who turned out, it underrepresents those who did not vote. The application of weights (Lynn, 1996) provided as standard by YouGov improves the representativeness of the sample to a limited extent for some of these demographic and political measures. A full comparison of the weighted and unweighted sample with the GB population, is available in Supplementary Appendix F.

Survey Measures
The main focus of the survey was on political participation and its material and psychological correlates so a large number of its questions drew on the Citizen Participation Study (Verba et al., 1995, p. 556). This includes measures of external political efficacy that ask respondents how much influence they have over decisions at local and national level, how much attention a local councilor and a local MP would pay to a complaint from them (all four questions have the same answer options: "None at all," "Hardly any," "A little bit," "Quite a lot," or "A great deal"), and when MPs lose touch with the public (answer options: "They stay in touch with the public even once they're elected," "They lose touch with the public once they're elected," or "They were never in touch with the public in the first place").
The survey items clearly relate to common measures of external political efficacy. The first two questions prompt an appraisal of the idea that members of the public have influence over politics. This is similar to asking whether respondents agree with the common external political efficacy statement: "People like me don't have any say about what the government does" (Craig et al., 1990, pp. 307-310). The third and fourth questions invoke opinions about how much attention politicians would pay to complaints, which is similar to asking whether they agree or disagree with the statement: "I don't think public officials care much what people like me think." (Craig et al., 1990, pp. 307-310). Finally, the question about MPs losing touch with the public is similar to asking whether respondents agree that: "Generally speaking, those we elect to public office lose touch with the public pretty quickly" (Craig et al., 1990, pp. 307-309). Thus, the variables in the GB data are similar to external political efficacy items that have been included in surveys fielded by the ANES (Norris, 2015, p. 799), the British Election Study (BES), and British Social Attitudes (BSA) (UK Data Service, 2021). As such, although they were only included in one previous study (Verba et al., 1995) and have not been validated, the variables used in Study 2 at least have face validity as indicators of external political efficacy.
The measures of explanations for inequality in society are unique to the GB data but similar to questions included in surveys fielded by the ANES (Bartels, 2018, p. 123) and BSA (UK Data Service, 2021). The measures used here focus on status difference in society, rather than jobs and incomes or "getting ahead in life," and use ranking rather than rating of explanations. Respondents were asked to select from amongst seven possible "reasons for some people achieving 10 Full question wordings for the key variables used in the analysis of the GB data are available in Supplementary Appendix E.
Frontiers in Political Science | www.frontiersin.org July 2021 | Volume 3 | Article 643165 higher status than others": luck, hard work, inevitability, backgrounds, ambition, cleavage-based inequalities, and other factors. They then ranked the top three most important explanations that they selected. Rankings are inverse coded so that explanations that are ranked first, second, or third have scores of four, three, or two, respectively, whilst an answer that is selected but not ranked has a score of one, and an answer that is not selected has a score of zero. Respondents were also asked to rank the same seven explanations for inequality in relation to their own status. There was no selection step before the ranking, but respondents could opt not to rank items that they considered unimportant. Again, the rankings are inverse coded so that ranking an item first results in a score of seven, ranking an item seventh results in a score of one, and not ranking an item results in a score of zero. The data also contain control variables including measures of internal political efficacy. These indicate how much influence respondents have at local and national level, when compared to most people ("Much less," "Slightly less," "About the same," "Slightly more," or "Much more"), and their level of agreement that it is hard to understand politics ("Strongly agree," "Agree," "Tend to agree," "Tend to disagree," "Disagree," or "Strongly disagree"). Further, the data include controls covering political engagement, left-right and liberal-authoritarian ideology, knowledge of national politics, and party identity. They also include measures of perceived privilege of activists and politicians, social capital, a single-item measure of selfesteem, and multiple measures of economic capital, which are important concepts to control for when dealing with explanations for status and about relations with others. Finally, in addition to measures of self-perceived class and any experience of private schooling, the data contain standard demographic controls covering highest educational qualification, social grade, sex, and age. All of these variables are included in the adjusted models.

Data Reduction
The data contain multiple measures of external political efficacy, explanations for inequality in society, and explanations for own status. As such, a confirmatory factor analysis was conducted in Mplus using the WLSMV estimator to accommodate the ordinal nature of many of the indicators. All five measures of external political efficacy load onto a single factor (coefs. range from −0.970 to 1.319, standard errors range from 0.077 to 0.083, all p values < 0.01). The distributions of answers to the external political efficacy questions, and their loadings onto the factor, are presented in Table 4. The distributions indicate an overwhelmingly negative view of the responsiveness of the political system and the attention paid to the public by representatives. As such, it is apt, subsequently, to refer to people with system justifying beliefs being less negative (rather than more positive) in their appraisals of the political system.
Turning to explanations for inequality in society, the ranking variables prompt respondents to compare options, which removes the possibility of indicating that explanations are equally important. This promotes cognitive engagement but also means that answer options are dependent on each other (Hino and Imai, 2019). When one option is ranked as most important it means that no other option can be, and the ranking of each option reduces the slots available for others. Nevertheless, the options that respondents choose to rank high or low remain at their discretion. To retain some of this ranking information whilst reducing dependence, the variables were recoded into binaries indicating whether each explanation was in a respondent's top three.
The confirmatory factor analysis focused on the four indicators that most clearly embody individual and structural explanations for inequality in society and own status: hard work (individual), background (structural), ambition (individual), and group-based inequality (structural). These four indicators load onto a single societal inequality factor and the structural explanations (background: coef.  and structural explanations for inequality in society. The indicator distributions, and their loadings on the factor, are shown in Table 5. The table also includes the distributions of the luck and inevitability explanations, which are included in some of the subsequent analyses but were excluded from the confirmatory factor analysis because they are neither individual nor structural explanations for inequality. As with explanations for inequality in society, the hard work (coef. 1 (constrained)), background (coef. 0.287, SE 0.081, p < 0.01), ambition (coef. 0.723, SE 0.108, p < 0.01), and group-based inequality (coef. −0.790 SE 0.126, p < 0.01) binary indicators load onto a single factor encompassing explanations for own status. There is a key difference between the two factors: the background explanation for inequality in society indicator loads negatively onto its factor whilst  the background explanation for own status indicator loads positively onto its factor. Thus, people who emphasise hard work and ambition as explanations for inequality in society are less likely to emphasise the background explanation. By contrast, those who emphasise hard work and ambition as explanations for their own status are also likely to emphasise the background explanation. Thus, the explanations for own status factor does not capture a neat opposition between individual and structural explanations. Table 6 presents the distributions of the indicators and their loading on the factor, and again includes the luck and inevitability indicators that were excluded from the confirmatory factor analysis. The difference between the two factors may stem from people applying individual explanations to their own status more readily than to inequality in society. More than four fifths of respondents (81.3%) pick hard work as a top three explanation for their own status, compared to less than three fifths (57.2%) who do so for inequality in society. On the other hand, two fifths of respondents (41.4%) indicate that background is an important explanation for their own status, compared to two thirds (65.2%) who do so in relation to society. This means that a quarter of respondents chose the option in relation to inequality in society but not their own status. Approximately the same proportion of respondents ranked ambition as a top three explanation in both cases (society: 55.8%; own status: 57.4%), but more people rank group-based marginalisation as an important driver of societal inequality (15.6%) than of their own status (7.3%). Thus, people are more willing to emphasise individual drivers when explaining their own status than when explaining inequality in society, and this is matched by a reduced willingness to emphasise structural explanations. This underlines the importance of measuring the two sets of explanations separately.
To reduce the number of independent variables included in the subsequent models, the confirmatory factor analysis also included seven factors covering internal political efficacy, perceived privilege of activists and politicians, political engagement, size of social network, activeness of social network, acquaintance with people in intermediate and lower occupations, and acquaintance with people in senior positions and professions. 11 The model fit statistics for the full ten-factor model indicate good (RMSEA 0.027) or adequate (CFI 0.931; TLI 0.923) fit (Byrne, 2012, pp. 70-71) so the factor scores were saved for use in the subsequent analysis.

Multivariate Analysis
Ordered logit regression models were estimated in relation to each of the five indicators of external political efficacy, and OLS regression models were estimated in relation to the external political efficacy factor. 12 All models used multiple imputation  All models are estimated using multiple imputation with chained equations, for which the R 2 statistic is not available.
Coefficients are unstandardised, all models use sampling weights (pweight), and robust standard errors are presented in parentheses.

Full results including the coefficients for control variables are available in
Supplementary Appendix I. Tables 4-6 are available in Supplementary Appendix G. 12 A series of regressions diagnostics were run for the OLS regressions and indicated a number of particularly influential cases in the unadjusted models. However, the results of regressions were replicated with the influential cases removed, and these analyses can be produced using the replications files linked to at the end of the article. Further, the diagnostics indicated some signs of non-normality of residuals and heteroscedasticity in the unadjusted models, particularly at the tails of the distributions. This supports the use of robust standard errors in the regression models. The diagnostic statistics and figures are in Supplementary Appendix H. with chained equations to estimate missing values for some of the demographic and political control variables, and all models use sample weights and robust standard errors. In the first instance, the factors representing explanations for inequality in society and explanations for own status were included as the key independent variables, and the unstandardized regression coefficients are presented in Table 7.

Details of the factors and indicator loadings not presented in
There is a consistent positive relationship between individual explanations for inequality in society (which load positively onto the factor) and the indicators of external political efficacy. Focusing on the adjusted model using the external political efficacy factor as the dependent variable (column 12 of Table 7), a one-point increase in the individual explanations for inequality in society factor is associated with a 0.444-point increase in the external political efficacy factor (SE 0.030, p < 0.01). The adjusted models relating to representatives paying attention to complaints, when MPs lose touch, and the external political efficacy factor also show a negative relationship with individual explanations for own status. In the latter case (again, column 12 of Table 7), a one-point increase in individual explanation for own status is associated with a 0.149-point decrease in external political efficacy (SE 0.028, p < 0.01).
To check that these results are not an artefact of the factors used as independent variables, the models were re-estimated using each of the original ranking variables in turn. Each model contains the ranking of the given explanation both for inequality in society and own status, and the rankings are treated as interval variables. Models were also run on the luck and inevitability explanations for reference. Again, multiple imputation and robust standard errors were used to estimate all models, ordered logit regressions were used for the five indicators of external political efficacy, and OLS regressions were used for the factor. 13 The results are consistent with the analyses using the factors representing explanations for inequality as the independent variables. Hard work and ambition, which are individual explanations for inequality in society, have consistently positive and usually statistically significant relationships with the indicators of external political efficacy. The results relating to structural explanations for inequality in society, in the form of background and group-based inequality, are less consistent. However, in five of the models, either background or groupbased inequality have a negative relationship with external political efficacy. These results provide evidence in support of Hypothesis 1. By contrast, counter to Hypothesis 2, ranked explanations for own status show no sign of a significant relationship with external political efficacy. Thus, whilst explanations for own status are distinct from explanations for inequality in society, they do not appear to have implications for appraisals of the political system.

Robustness Checks
The analyses using each ranked explanation for inequality in society and own status were repeated with the ranking variables treated as ordinal rather than interval, with similar results. 14 Further, to test whether the preceding results stem from the ranked nature of the variables, the models were also re-estimated using simple binaries indicating whether each explanation was selected. Again, models were estimated using each indicator of external political efficacy, and the overarching factor, as the dependent variables. All the binaries indicating explanations for inequality in society and own status were included in each model. Ordered logit regressions were used for the indicators of external political efficacy, and OLS regression was used for the factor. Again, the models used multiple imputation with chained equations, sample weights and robust standard errors. The unstandardized coefficients from these models are graphed in Supplementary Appendix J, and the results do not obtain statistical significance. 15

Discussion
In contrast to Study 1, Study 2 provides evidence in support of Hypothesis 1 and the idea that individual explanations for inequality in society are positively related to appraisals of the political system. Specifically, they are positively related to multiple indicators of external political efficacy and the factor encompassing those indicators. Concomitantly, in many cases, structural explanations for inequality in society are negatively related to indicators of external political efficacy. These results hold whether we focus on each indicator of external political efficacy or the factor that they load onto, and whether we treat explanations for inequality in society as a factor or as separate interval or ordinal indicators. However, most of the results lose significance when binary indicators of selected explanations for inequality are used in place of ranked variables or the factor. The coding of explanations for inequality as binaries brings them closer to the three-category ratings used in the ANES data, and it is notable that analysis of the binaries in the GB data replicate the nonsignificant results observed in the ANES data. This suggests the importance of how explanations for inequality are asked about. The act of ranking the options seems to distinguish people with different explanations in a way that simply asking them to select explanations does not.
There is little evidence in support of Hypothesis 2. The negative relationship observed between individual explanations for own status and external political efficacy does not hold when the ranked indicators of explanations for own status are used in place of the factor. Thus, although they are distinct from explanations for inequality in society, and notably more orientated towards individual explanations, they do not appear to relate to appraisals 13 The unstandardised coefficients for the rankings of explanations for inequality in society and own status when treated as interval variables are available in Supplemenatary Appendix J.
14 The ordered logit and OLS regression results using rankings of explanations for inequality in society and own status, treated as ordinal variables, can be produced using the replications files linked to at the end of the article. 15 The ordered logit and OLS regression results using binary indicators of explanations for inequality in society and own status can be produced using the replications files linked to at the end of the article. of the political system. This lack of relationship may reflect the more personal nature of explanations for own status, which could have implications for other personal beliefs or feelings, such as well-being and self-esteem.

GENERAL DISCUSSION
The first key finding of this research is that there are almost no consistent and significant relationships between explanations for inequality and appraisals of the political system in the 2002 ANES data. Only the government policy explanation for inequality, which is a structural explanation, has any relationship with such appraisals. To the extent that structural and individual explanations for inequality are opposed, the negative form of this relationship is consistent with the Hypothesis 1. However, the government policy explanation is an unusual one that is conceptually close to views of the political system, which may drive the relationship between the two. Indeed, none of the other explanations for inequality relate to appraisals of the political system, even when the government policy explanation is removed from the models. Overall, Study 1 provides evidence that there is no relationship between explanations for inequality and appraisals of the political system.
Turning to Study 2, the second key finding of this research is the confirmation of a distinction between explanations for status as applied to society and the self. People tend to prioritise individual explanations for their own status and, concomitantly, to deprioritise structural explanations. By contrast, they are more willing to endorse structural explanations for inequality in society. This means that many people take personal credit and responsibility for their own status whilst thinking that structural factors are important in society at large. The results of Study 2 also show that explanations for own status do not have implications for appraisals of the political system. However, the personal nature of such explanations suggests that they may relate to other individual beliefs and psychological traits such as well-being and self-esteem. Further research is needed to corroborate the value of measuring these beliefs separately from beliefs about the reasons for inequality in society, and to investigate their relationships with other beliefs and dispositions.
The third key finding of this research also stems from Study 2. When asked to rank the most important reasons for inequality in society, people who prioritise individual explanations are less negative about the political system than are people who prioritise structural explanations. Thus, people who emphasise system justifying explanations for inequality in society are more apt to view the political system as open to influence by the public, elected representatives as more attentive to public complaints, and MPs as in touch with the public. This finding holds when controlling for a range of demographic and political variables, whether models are estimated using a factor capturing explanations for inequality in society or the ranking variables that load onto it, and whether the external political efficacy factor or indicators are used as the dependent variable. Thus, the GB data shows a consistent and positive relationship between individual explanations for inequality in society and appraisals of the external political efficacy. This corroborates the findings of the only other study related to this topic (Osborne et al., 2015), which observed a positive relationship between system justifying beliefs and external political efficacy amongst M aori people in New Zealand.
The inconsistency of the findings emerging from Study 1 and Study 2 has at least three possible explanations; temporal context, national context, and question format. The 2002 ANES survey was fielded prior to the 2008 financial crash and subsequent recession, and only a year after the September the 11th terrorist attacks on the United States. The rally round the flag effect of the attacks may have influenced appraisals of the United States system of government, and even of the strengths and weaknesses of United States society more broadly. By contrast, the GB survey was fielded after the financial crash and recession, as well as after the MPs expenses scandal. Again, these events may have influenced appraisal of government and views of the social system more generally. In both cases the temporal context also relates to events that were specific to the country in question. More broadly, the concept of the American Dream in the United States and the prominence of class and its relationship with politics in Great Britain may create different national contexts that shape how people explain inequality, and how those explanations relate to appraisals of the government.
It is also possible that the differing relationships observed in Study 1 and Study 2 stem from the different ways in which the key questions were asked in each survey. The process of ranking explanations for inequality used in the GB survey is cognitively demanding and forces respondents to choose between competing answers. This distinguishes people for whom a given explanation is particularly important from a wider group that simply select the explanation. In other words, many people may be willing to acknowledge that, for instance, ambition has some part to play explaining inequality in society. However, that group may have quite different views of the political system than the sub-set of them who indicate that ambition is one of the most important explanations for inequality in society. The idea that question format is a key factor in the difference between the results in Study 1 and Study 2 is supported by the robustness checks in the latter. These show that when the explanations for inequality in the GB data are recoded into binaries, indicating whether each explanation was selected, they cease to have a relationship with external political efficacy. These binaries are more akin to the three-category rating indicators used in the United States data in Study 1, which also showed no relationship with appraisals of the political system. This is, however, only suggestive evidence, and work is required to directly test the effects of question format on answers regarding explanations for inequality and their relationships with appraisals of the political system.
The need for evidence regarding the effects of asking about explanations for inequality in different ways indicates the first limitation of this research: the unusual nature of the questions included in the GB survey. This applies particularly to the measures of explanations for inequality, which are not replicated in other data. It would be useful to investigate the relationships between explanations for inequality in society and standard measures of external political efficacy beyond the constrained categorical measures in the 2002 ANES data, and the unusual measures in the GB data that are based on the Citizens Participation Study (Verba et al., 1995). Unfortunately, such investigations seem unlikely given that the 2019 ISSP survey, which asked about reasons for some people "getting ahead in life", did not ask about appraisals of the political system. Further, explanations for inequality in society and own status represent a particular form of system justifying beliefs. It remains to be seen whether assessments of the political system are related to other forms of system justifying beliefs, such as the Protestant Work Ethic, Belief in Status Legitimacy, or Belief in a Just World (O'Brien and Major, 2005).
The second limitation of the research is its minimally comparative nature given its restriction to two countries, the different temporal contexts noted above, and the differing question formats. This means that the results emerging from the two studies are not comparable. Further, both surveys were fielded in wealthy English-speaking capitalist countries so findings cannot be read across to national contexts with differing economic and cultural circumstances. As above, the absence of measures covering appraisals of government from the 2019 ISSP data precludes the possibility of a more comparative study. Thus, we cannot currently investigate the extent and ways in which relationships between explanations for inequality and appraisals of the political system vary between national contexts.
Finally, the third limitation of this research is its non-causal nature, given that the data analysed are cross-sectional. This article argues that explanations for inequality are lower-order beliefs (Bem, 1970) akin to components of ideology (Peffley and Hurwitz, 1985;Feldman, 1988;O'Brien and Major, 2009). By contrast, appraisals of the political system are higher order beliefs, which are causally posterior to explanations for inequality. However, it may be the case that views of politics, and observing the performance of the political system, shape or alter broader beliefs about society, including ideas about the drivers of inequality. This causal order cannot be disentangled using the data analysed here or the 2019 ISSP data, given its crosssectional nature and lack of measures relating to appraisals of the political system. As such, further investigation of the causal nature of the relationships observed in this article is required via either time-series or experimental analysis.

CONCLUSION
When asked to rank explanations for inequality, people who emphasise individual causes tend to be less negative in their assessments of external political efficacy than are people who emphasise structural explanations. Specifically, they are apt to think of the responsiveness of the political system and those who inhabit it in less negative terms. External political efficacy's focus on the responsiveness of the system links to input legitimacy, whereas political trust is concerned with the output legitimacy of the political system. However, external political efficacy and political trust are frequently considered alongside each other (Catterberg and Moreno, 2006;Southwell, 2012;Boulianne, 2019;Geurkink et al., 2020) and satisfaction with democracy, are often affected by the same factors (Anderson, 2010), and are positively related to each other (Craig, 1979;Chao et al., 2017). As such, the results emerging from GB data indicate that explanations for inequality in society may be positively related to political trust and satisfaction with democracy, though these propositions require further testing.
Explanations for status difference in society are distinct from explanations for own status, and the latter tend to be more individual and less structural than the former. However, the relationship between external political efficacy and explanations for inequality in society does not extend to explanations for own status. This may reflect the focus of external political efficacy on actors beyond the self, for which beliefs about society are more relevant. Further, the relationships between explanations for inequality in society and appraisals of the political system are not consistent across studies 1 and 2, and the ANES data indicate almost no relationships between the concepts. This inconsistency in the results of the two studies may stem from their differing temporal or national contexts, or the format of the questions. The latter possibility is supported by the lack of relationship observed in the GB data when explanations for inequality are recoded into a similar format as the indicators in the ANES data. In addition to work investigating the effects of question format, additional work is needed on the impact of national context, and on the causal direction of the relationship between explanations for inequality and appraisals of the political system.
The GB results suggest that believing status is unearned has negative implications for appraisals of the political system. People react negatively to inequality when it is not seen as the consequence of individual effort, and this reflects badly on systems that sustain such outcomes. These findings are, perhaps, less surprising when politicians are becoming visibly more distinctive from the people they represent in terms of education and occupation (Heath, 2015;Audickas, 2016;Heath, 2018). The impact of explanations for inequality has implications for how we talk about politics. It may be worth placing more emphasis on the hard work and dedication of many people in the political system, both in reaching their positions and once they are in them. However, this is not mutually exclusive with recognizing that many of those same people have been the beneficiaries of structural advantage, and that it is incumbent on those in politics to make the system more open and representative. The very principle of democratic equality, of government amongst equals rather than by superiors, implies that the work of politics should not be the preserve of those with advantage. As long as people have grounds for viewing inequality in general, and political inequality specifically, as unearned, they may be reasonable in moderating their appraisals of the political system.

DATA AVAILABILITY STATEMENT
The dataset analysed in Study 2, and the analysis files for both Study 1 and Study 2, can be found in the following online repository: Pure, the University of Strathclyde Research Management System. DOI: https://doi.org/10.15129/ca1eaa95-09c2-44e5-8674-e94f41e50a1a.

ETHICS STATEMENT
The survey that produced the data in Study 2 was reviewed and approved by the University of Essex Ethics Committee. The participants provided their written informed consent to YouGov for their anonymised survey answers to be analysed. The anonymised survey data used in Study 1 was gathered by the American National Election Study and made publicly available by them.

AUTHOR CONTRIBUTIONS
The author of this article (JG-H) conducted the literature review outlined in the opening sections, formulated the hypotheses, conducted the fieldwork that produced the data in Study 2, and processed that data as well as conducting the analyses in both studies. As such JG-H is the sole author of the article.

FUNDING
The data analysed in Study 2 was gathered during the author's PhD in the Department of Government at the University of Essex, which was funded by an ESRC +3 collaborative studentship linked with YouGov. The article was written during the author's employment as British Academy Postdoctoral Fellow (Award number: PF20/ 100039) in the School of Government and Public Policy at the University of Strathclyde.

ACKNOWLEDGMENTS
Thanks are due to the British Academy for providing the funding that allowed this research to be completed, and to the School of Government and Public Policy at the University of Strathclyde for providing such a supportive work environment. Thanks are also due to the Ben Seyd, Raynee Gutting, and Rob Johns for their helpful comments on an earlier draft of the article presented at the 2020 United Kingdom Political Psychology Conference. Further, the author is very grateful to the Joseph Hamm, the three reviewers, and the other Frontiers editors for their detailed and constructive comments and suggestions during the review process. These prompted changes that undoubtedly improved the article. Finally, special thanks are due to Sonja Zmerli for her help and support in relation to the research underpinning this article.