Ideology As Motivated Cultural Cognition How Culture Translates Personality Into Policy Preferences

Home , Cultural cognition, The Political Compass

Ideology as Motivated Cultural Cognition How Culture Translates Personality into Policy Preferences

Figure 1 James C. Harman

Abstract In different cultures, the same perceptions make for different policies.

This paper summarises the results of a quantitative analysis testing the theory that culture acts as an intermediary in the relationship between individual perceptual tendencies and political orientation. Political psychologists have long observed that more “left-wing” individuals tend to be more comfortable than “right-wing” individuals with ambiguity, disorder, and uncertainty, to equivocate more readily between conflicting viewpoints, and to be more willing to change their opinions. These traits are often summarised under the blanket term of “open- mindedness”. A recent increase in cross-cultural studies, however, has indicated that these relationships are far less robust, and even reversed, in social contexts outside of North America and Western Europe. The sociological concept of culture may provide an answer to this inconsistency: emergent idea-networks, irreducible to individuals, which nonetheless condition psychological motivations, so that perceptual factors resulting in left-wing preferences in one culture may result in opposing preferences in another. The key is that open-mindedness leads individuals to attack the dominant ideas which they encounter: if prevailing orthodoxies happen to be left-wing, then open minded individuals may become right-wing in protest. Using conditional process analysis of the British Election Study, I find evidence for three specific mechanisms whereby culture interferes with perceptual influences on politics. Conformity to the locally dominant culture mediates these influences, in the sense that open-minded people in Britain are only more left-wing because they are less culturally conformal. This relationship is itself moderated both by cultural group membership and by Philip Converse’s notion of “constraint”, individual-level connectivity between ideas, such that the strength of perceptual influence differs significantly between cultural groups and between levels of constraint to the idea of the political spectrum. Overall, I find compelling evidence for the importance of culture in shaping perceptions of policy choices.

Key words: culture, perception, political orientation, motivation, conformity, constraint

Acknowledgements I am indebted to Dr. Rob Meadows of the University of Surrey, and to Dr. Paul Stoneman of Goldsmith’s, University of London, for their thorough criticism of my work. Without them, my ego would be far greater and my work far poorer. I am also extremely grateful to Professor Steve Borgatti of the University of Kentucky, for creating the free network analysis software Netdraw, and to Professor Andrew Hayes of the Ohio State University, for creating the wonderful PROCESS macro for SPSS, both of which have made the following analysis possible. I am thankful to the team behind the British Election Study, from the universities of Manchester, Oxford, and Nottingham, for making such a valuable data resource freely available to members of the public. My forbearing friends and family members deserve thanks for their proofreading, counselling, and general support. Finally, as always, my greatest gratitude rests with God, who makes all things possible.

Table of contents Abstract ...... 2 Acknowledgements ...... 2 List of tables and illustrations ...... 4 Introduction: why study policy preferences? ...... 5 The theoretical case for a cultural political psychology ...... 6 Perceptual influences on political orientation ...... 6 The “WEIRD” bias and cross-cultural contradictions ...... 8 Culture and its conceptual utility ...... 9 Motivated cultural cognition: research framework ...... 10 Methodological choices: paradigm, data, and analysis ...... 11 Research paradigm: quality or quantity? ...... 11 Data usage: primary or secondary? ...... 11 The British Election Study: data set, ethics, and limitations ...... 12 Grasping slippery ideas: operationalising concepts as variables ...... 13 Political orientation: dimensions and underlying forces ...... 13 Perception: openness, uncertainty, and ambivalence ...... 16 Culture: cluster, conformity, and constraint ...... 17 Ruling out the alternatives: control variables ...... 19 Conditional process analysis: hypotheses and how to test them ...... 20 The results ...... 23 Overview of results...... 23 The influence of active open-mindedness on leftness, regardless of culture ...... 25 The influences of active open-mindedness on conformity ...... 25 Moderation by tolerance of uncertainty and response polarisation ...... 26 The influences of conformity on leftness ...... 26 Moderation by cultural cluster and constraint ...... 27 Direct and indirect effects, at different levels of the four moderators ...... 27 Discussion: implications for the science of policy preferences ...... 31 Conclusion: limits of the study and needs for further research ...... 33 Bibliography ...... 34 Technical Appendix ...... 41 Section 1: question wording, coding and recoding of selected variables ...... 41 Section 2: annotated summary of key statistical results ...... 53 Section 3: complete PROCESS output for the mediation and moderated mediation models .. 63 Section 4: VNA file for visualising correlations between policy preferences ...... 71 Section 5: complete SPSS syntax used in the analysis ...... 76

Word count (from introduction to conclusion): 13,071

List of tables and illustrations

Tables Table 1. Unstandardised regression coefficients from the moderated parallel multiple mediation analysis used to test the thirteen hypotheses…………………………………24 Table 2. Summary of all variables from the British Election Study selected for use in the analysis, including variable names, question-wording, response options, response coding, and response recoding by the author…………………………………..45 Table 3. Cronbach’s alphas for related items grouped into scales……………………………….57 Table 4. Principal component factor analysis results…………………………………………………..59 Table 5. Rotated factor loadings for principal components analyses which saved more than one factor……………………………………………………….……………………………………..62 Table 6. Cluster analysis results for cultural cluster and level of constraint cluster……….63 Table 7. Descriptive statistics for all continuous variables in the analysis..………………..…63 Table 8. Frequencies for all nominal variables in the analysis………………………………………64 Table 9. Bivariate comparisons between the political orientation factors and other selected variables……………………………………………………………………………………….…64 Table 10. Bivariate comparisons between the perceptual factors and other selected variables……………………………………………………………………………………………………..…65 Table 11. Bivariate comparisons between the continuous cultural variables and other selected variables……………………………………………………………………………………….…65 Figures Figure 1. “What a cruel, male-dominated culture!” cartoon by Malcolm Evans……...….. 1 Figure 2. Conceptual diagram for the theory of motivated cultural cognition……………..13 Figure 3. Network diagram showing correlations between policy preference items……14 Figure 4. Conceptual and statistical diagrams illustrating hypothesised relationships…21 Figure 5. Scatterplot showing the relationship between cultural conformity and political leftness, moderated in strength by belief system constraint……………………………28 Figure 6. High-low plot showing sizes of the indirect effects of active open-mindedness on political leftness, via both measures of cultural conformity, at different levels of the four moderating variables……………………………………………………………………29 Figure 7. Conceptual diagram showing hypothesised relationships which have been confirmed, annotated with regression coefficients………………………………...………30

Introduction: why study policy preferences? My aim in this paper is to specify and test a theoretical model which attempts to explain why people are left-wing or right-wing.

This distinction, known variously and often interchangeably as ideology, the political spectrum, or political orientation, is a concept which attempts to summarise differences in political opinion between individuals (Atwan and Roberts, 1996). The definition and scope of these terms are fiercely contested, and as such a precise description of exactly what I am interested in is necessary. I regard left and right, first and foremost, as indicators of policy preferences- individual-level opinions on what a government, or governments in general, ought to do in a given situation (Budge et al, 2001). Policy preferences are predominantly, though not unanimously, arranged into a single-dimension spectrum separating the left, or “liberals”, from the right, or “conservatives” (Sidanius, 1993: 205). The question of why different people can be found at different positions along this spectrum is what I wish to investigate.

The question of why people hold certain policy preferences at the expense of others is important for at least two reasons. Firstly, policy preferences are often deeply divisive and controversial. As Aristotle once lamented, although practically all people strive for what is “good”, radical differences in specific things considered “good” or “bad” have prevented the human race thus far from agreeing on shared ideals of how to live (Aristotle, 1968: 54, 297-9).

The second, related, reason for the importance of investigating policy preferences is that they are usually highly consequential. In an established democracy, citizens holding policy preferences may use their varying degrees of resources to vote, lobby, and protest to make those preferences a reality (Parsons, 1994). Even in authoritarian dictatorships, which are nominally unresponsive to mass opinion, individuals are able to support or undermine governing authorities in the implementation of policy, up to and including involvement in public uprisings or civil wars (Bunce and Wolchik, 2010). Public policies, once implemented, affect practically every area of human life, from how crops are grown to where roads are built, to when it is acceptable to take a life, to whether nuclear weapons are detonated (Alvarez and Brehm, 2002: 41). In short, policy preferences are important because they are part of the means by which humanity decides what to do with itself.

I begin by surveying an expansive empirical literature concerned with explaining individuals’ policy preferences using perceptual measures such as open-mindedness and need for closure. These studies have been convincing in explaining why people of different motivations, and especially with differing approaches to unfamiliar information, tend to develop different political orientations. However, as we shall see, the utility of purely psychological explanations declines sharply when making comparisons across different societies, particularly when studies involve participants from non-Western countries. I propose that sociological and anthropological understandings of the concept of culture- shared belief systems irreducible to their individual adherents- could offer a solution to these apparent inconsistencies.

Therefore, the contribution my study seeks to make is to provide an empirical test of the theory of “motivated cultural cognition”, which has been proposed previously but which has hitherto lacked full specification and empirical verification. The theory holds that more “open-minded” individuals- who are more willing to accept new and unfamiliar information and to change their opinions- tend to be more critical of ideas which are predominant and widely accepted in the societies in which they live. That is, open-minded individuals usually have lower cultural conformity. Precisely which policy preferences they hold is thus a product of prevailing social values. In cultures where the balance of opinion is skewed to the right- as it is in most of the Western world- more open-minded people tend to be more left-wing. In traditionally left-wing societies, by contrast, the open-minded become more attracted to right-wing policies.

It is this basic idea which is summed up so well by the Malcolm Evans cartoon on the front cover of this paper (Evans, 2011). Both women in the picture are driven by identical perceptions, in that they are both opposed to inequality- specifically, inequality between men and women. This constitutes the concept of “motivated cognition” identified by political psychologists. Yet their embeddedness in contrasting cultural contexts, which supply them with ideas from beyond their own desires and experiences, serve to flip these motivations on their heads so that they are diametrically opposed when it comes to the concrete question of what clothes women ought to wear. This is the meaning of the “cultural cognition” of the theory: without cultural guidelines, individuals have no way of “knowing” which specific policies are best suited to their perceptual motivations. Policy preferences can only be effectively predicted, I claim, when both perception and culture are considered.

I test this theory using a secondary quantitative analysis of data from the British Election Study. Deriving measures of political orientation, open-mindedness, and cultural considerations, I specify a set of hypotheses to be tested in a “conditional process analysis”- a kind of path analysis which combines mediation and moderation to explore how, and under what cultural conditions, perception affects policy preferences. I discover that, in the contemporary United Kingdom, and in accordance with the theory under investigation, more open-minded people are only more left-wing because they are less attached to British culture, which has a moderately right-wing influence, and that the strength of this influence varies between different cultural groups and between different levels of “constraint” to the political belief system. I conclude by summarising the limitations of my analysis, and briefly propose an agenda for further research into culture as an arbiter of political and psychological motivations.

The theoretical case for a cultural political psychology Perceptual influences on political orientation A great number of studies have discovered connections between policy preferences and a wide variety of social, economic and situational factors (Citrin et al, 1997). One of the most fruitful veins of investigation, however, suggests that policy preferences are powerfully influenced by an altogether more basic and fundamental substance: the nature and characteristics of perception and consciousness, human information processing, and sensory stimulation.

A pioneering study in this field was The Authoritarian Personality, a study by Theodor Adorno and colleagues partially driven by a need to explain support for the Nazi Party in the mid-20th Century (Adorno et al, 1964). Their key finding was that individuals inclined to political authoritarianism also tend to exhibit “rigidity of thinking”, or unwillingness to change existing opinions, and are unusually prone to negative emotions when those opinions are threatened. Subsequently, a considerable number of studies have investigated similar aspects of this relationship. Their findings have been helpfully collated and compared by John Jost and his colleagues in a series of meta-analyses on political conservatism (Jost et al, 2003; Jost, 2017). They find that conservatives are generally characterised by a preference for simple, definite opinions, clearly structured world-views, and long-term stability of existing beliefs.

More recently, this concept has come to be more precisely defined and operationalised as the theoretical construct and survey device known as the need for closure scale (Webster and Kruglanski, 1994). Webster and Kruglanski describe this scale as an opinion-based metric assessing, firstly, how eager individuals are to “seize” upon a precise, certain belief about how the world is or should be, and to “freeze” that belief by refusing to change it in the face of anything less than overwhelming evidence (Roets and Van Hiel, 2011). An equivalent, opposite, tool is the active open-mindedness scale, a measure of how readily individuals claim they change their opinions in response to new information (Stanovich and West, 1997). A very similar concept, conservation, has also been uncovered in the human values literature (Schwartz, 2007; Feldman, 2013; Goren et al, 2016). Conservation is essentially a representation of differing human needs for cognitive parsimony: a preference for simple, clear, stable opinions over complex, ambiguous, ephemeral ones.

Numerous studies have uncovered statistically significant relationships between Need for Closure and right-wing political orientation (e.g. Chirumbolo, 2002; Jost et al, 2003; Golec De Zavala et al, 2010). What’s more, Need for Closure has been shown to increase among people recalling threatening situations involving the novel and unfamiliar, whilst both factors are connected with an increase in political conservatism (Matthews et al, 2009; Thórisdóttir and Jost, 2011). In more general terms, those on the conservative right tend to demonstrate greater sensitivity to threat than those of the liberal left (Vigil, 2010).

The claim made by many social psychologists and neuroscientists is that differing political orientations are, to an extent, biologically inherited and physically ingrained in the human brain (Hatemi et al, 2009). As evidence, they cite studies of brain structure and function, which indicate that rightists usually have an increased volume of grey matter, and hence greater information-processing power, in their right amygdalae (Carraro et al, 2011).

The amygdala is a brain sub-organ which regulates responses to novel, and especially threatening, sensory stimuli: because humans tend to perceive the unusual as threatening, threat sensitivity and resistance to new ideas and information go hand in hand (Tritt et al, 2013). Political leftists, by contrast, have greater information-processing power in the brain area known as the anterior cingulate cortex, which manages desire and reward anticipation (Schreiber et al, 2013). Whilst uncertainty and change mean danger for those sensitive to

7 threats, they might mean the exact opposite for those more conscious of potential future benefit (Taber and Young, 2013). This leads to very different attitudes towards the established, orthodox way of doing things. To use a pastoral metaphor: the right-wing sheep generally prefer the certainty and restriction of the fence, as it keeps wolves at bay, whereas the left- wing sheep wish to break it down, in the hope of finding greener grass on the other side. For this reason, the mainstream argument goes, political leanings reflect perceptual style: the more closed-minded prioritise some ideas over others, and thus favour social hierarchies where some groups are prioritised over others, and they resist changes in belief over time, and thus favour social relations which abide through the ages (Jost et al, 2003).

The “WEIRD” bias and cross-cultural contradictions If only things were that simple. Recent studies, however, have shown that the relationships reported above are remarkably difficult to replicate in contexts outside of the United States and Western Europe (Van Bavel et al, 2016).

Traditional social scientific studies of political orientation and many other factors have been increasingly criticised for their heavy reliance on the inhabitants of “WEIRD” (White, Educated, Industrialised, Rich, and Democratic) societies as research participants (Henrich et al, 2010). Universalistic claims are often made based merely on smatterings of convenience samples collected from the Global North (Yilmaz and Saribay, 2016). In Jost et al’s (2003) meta-analysis, to take one example, 75% of the studies included took place solely within the United States, and the majority of the remainder were conducted in Western Europe (Jost et al, 2003). To many practitioners of social science, it has seemed less than safe to assume that the inhabitants of WEIRD countries- a small and historically atypical minority of the world’s population- are representative of “human nature”, whatever that is (Ramsay and Pang, 2015).

To rectify this shortcoming, several studies have now been conducted within historically neglected social contexts. Some, such as Yilmaz and Saribay’s (2016) study of Turkish nationals, tend to reinforce the finding that those on the right are less willing to scrutinise preconceived opinions (Yilmaz and Saribay, 2016), many other studies appear to throw these findings into doubt.

Kossowska and Van Hiel (2003) compare research subjects from the Netherlands with those from Poland, finding that whilst need for closure is associated with support for free-market policies in the Netherlands, it is in fact associated with state socialism in Poland (Kossowska and Van Hiel, 2003). Similarly, need for closure is related to a confrontational approach to conflict resolution in the United States, but actually predicts a conciliatory approach among citizens of China. Harman (2017) compares general decisiveness of opinion, a measure approximating need for closure, with support for right-wing economic policy across dozens of countries. It appears that, whilst the overall relationship between the two is positive, it is non- significant within many countries, and significantly negative among the inhabitants of Uruguay, Russia and India (Harman, 2017). If policy preferences were innate or universal, as much of established theory implies, then this sort of cross-national variation in the strength, and even direction, of perceptual influences upon political thought would be very difficult to explain. 8

Culture and its conceptual utility I wish to suggest, however, that policy preferences are not innate or universal. To support this contention, a turn from psychology to sociology and anthropology will be particularly beneficial. These latter disciplines, unlike the former, possess a credible mechanism by which systematic differences in beliefs and values can coalesce, in a way that is not explicable either by individual idiosyncrasy or by universal human constants: the concept of culture (Ellen, 2003). I believe that it is culture to which we must turn to explain the complexities in the interface between individual perception and policy decisions (see Ross, 1997).

The formation of shared identities common to group members is crucial to the study of politics (Huddy, 2013). This has been recognised particularly clearly in studies of foreign policy-making “groupthink”, a herd mentality in which opinions within a confined decision-making space become increasingly uniform (Janis, 1973; Badie, 2010). Yet broader, longer-term, and more widespread collective values have often been downplayed by political psychologists.

Social anthropologists, by contrast, have made the study of culture a central aspect of their research. One influential definition of culture is that it is a pattern of beliefs, a repetition of ideas or behaviours across individuals in a way that is too systematic to be explained by individual choice alone, yet too variable to be attributed to biological “human nature” (Benedict, 2005). Anthropologists have used the theory of schemata or “impressions” (Lodge et al, 1989) to advance systematic understandings of culture. A schema is a standardised and simplified model of reality which is used to predict forthcoming events, and to indicate which eventualities are expected and which unexpected (Axelrod, 1973).

What Sperber (2005) suggests is that these schemata, impressions or “representations”, coalesce into collective belief systems when they are transmitted epidemiologically across social networks, much like the spread of a disease (Sperber, 2005: 311). Yet the precise content of exactly which ideas become predominant is irreducible to individual psychology: it is an “emergent property” which develops unpredictably out of the complexity of interactions between people (Archer, 1995). A language, for example, is an artefact which transcends individual volition; it is the work of no one individual, and whilst all may contribute in small ways to it, all are likewise required to abide by the majority of its rules if they wish to make themselves understood (Durkheim, 2008). Across repeated iterations, these representations come to be accepted as a “social structure” of interdependent opinions, none of which can be changed without altering the others (Levi-Strauss, 1974).

A similar concept already exists in the writings of the political scientist Philip Converse- that of the belief system (Converse, 1964: 3). According to Converse, political orientation consists of networks of specific policy preferences which are more or less strongly correlated with one another, such that an individual holding a certain opinion- for example, that tax cuts should be encouraged- feels obliged to also hold other, apparently unrelated opinions, such as endorsing military action against other countries, simply because those opinions are perceived to “go together”. The sources of information for these interdependencies of opinion are “less logical in the classical sense than they are psychological- and less psychological than social” (Converse,

1964: 5). In other words, people know which opinions are supposed to accompany one another because they observe others who already exhibit these patterns of ideas. As such, “constraint” between policy attitudes is potentially more of a socially emergent property of human groups than an ingrained feature of human physiology (Feldman, 2013).

Motivated cultural cognition: research framework Jeanne Ho-Ying Fu and her colleagues suggest that culture could play a central role in cross- national differences in perceptual influences upon policy preferences. They claim that qualities such as need for closure, or any other psychological characteristics, are not directly and intrinsically motivations “for” any specific policies (Fu et al, 2007). After all, the sheer volume of potential policies is vast and ever-changing, and it would be hard to see how attitudes so specific could be genetically encoded (Toren, 2003; Gross Stein, 2013). Psychological reactions are driven by brain functions, but these functions are “activated” contextually depending on whether social norms are violated (Schreiber and Iacoboni, 2012). Instead, need for closure predisposes individuals towards “cultural conformity”, a tendency to support whichever policies are most favoured in local mainstream culture. Those high in need for closure feel the need to come to quick conclusions on unfamiliar topics, and one of the most abundant sources of ready-made opinions is the surrounding culture (West et al, 2008). Conversely, the “open- minded” are motivated to seek their own, unique opinions as a source of self-expression, and thus regard conventional ideas with suspicion (Xu and Peterson, 2017).

Fu et al describe this theory as “motivated cultural cognition” (Fu et al, 2007). It combines sociological and psychological theories, both of which have been accused of contrasting reductionisms, in an interesting and useful way. Unlike cultural sociology, which often ignores individual differences, motivated cultural cognition holds that some individuals are more likely to accept cultural influence than others. Unlike individualistic psychology, which often assumes that cognitive phenomena are either unique to individuals or universal to all humanity, motivated cultural cognition allows for intermediate-level differences between social groups depending on the content of their emergent beliefs.

Fu et al elaborate the theory of motivated cultural cognition, but they do not fully test it. Their analysis simply compares differences in the influence of need for closure on conflict judgements between the United States of America and China; they do not directly measure any form of cultural conformity or assess whether it is responsible for the relationships observed. What’s more, their main focus is on approaches to conflict, not policy preferences in general, and their use of non-probability samples makes statistical inference difficult. My aim is to test their theory of motivated cultural cognition as a predictor of policy preferences, to ascertain whether culture can truly be said to fill the gap between perception and politics. As such, my research question is as follows:

“Can culture explain variations in the relationship between perception and policy preferences?”

Methodological choices: paradigm, data, and analysis The following sections will outline the rationale behind the research design for my data analysis, before outlining and testing hypotheses to shed light on the role of culture in intermediating between perception and political orientation. Owing to the large volume of statistical data from this analysis, not all results will be fully reported. Instead, they are summarised in detail in the technical appendix to this paper, which contains information on the variables chosen and recoded, selected output from statistical tests, and the code used to generate both these statistics and the various data visualisations.

Research paradigm: quality or quantity? The domain of research methods is strongly divided between the “qualitative” and “quantitative” paradigms, and perhaps my first methodological task should be to justify my use of quantitative methods (Bryman, 2012). Methodological choices can never be divorced from theory; as such, some theories may be more readily investigated using some methods than others (Gilbert, 2016). In this case, the theory of motivated cultural cognition has been developed by Fu et al (2007) and others within the quantitative paradigm. For my results to be broadly comparable with theirs, it is advisable to employ quantitative methods.

This choice brings with it both advantages and disadvantages. Properly executed quantitative research allows discovered relationships to be reliably extrapolated to a wider population of interest, whereas qualitative methods offer no guarantee that their results exist among anyone other than the research subjects in question (Stoneman and Brunton-Smith, 2016). Yet these benefits are bought at the price of a lack of individual-level depth of measurement. Surveys are largely inimical to open-ended questions- with some exceptions- and their uniform structure prohibits most forms of probing or responsive techniques which could create a dynamic conversation between interviewer and interviewee. This increases the probability that certain results may be mere artefacts of the way a survey has been designed and administered (Zaller, 1992; Alvarez and Brehm, 2002). Consequently, quantitative methods run the risk of producing impressive and sophisticated findings which do not actually mean anything.

Data usage: primary or secondary? I have chosen, in preference to collecting my own quantitative data, to utilise secondary survey data which have already been collected. My reason for making this choice is simple: high- quality survey data are freely available on topics covering almost every aspect of social life (Stoneman and Brunton-Smith, 2016: 90). Effectively executing a representative survey is an expensive and time-consuming task, requiring its own forms of expertise and a considerable collection of resources (Dillman, 2007). Such an investment is outside of the means of most individual researchers, myself included. Furthermore, reusing data that have required such effort to collect is an excellent way to recover additional value from previous work in a timely and cost-effective manner (Economic and Social Research Council, 2015). The chief disadvantage of using secondary data, of course, is lack of autonomy in the design of sampling strategy and question content, wording, and order. In utilising standardised survey questions to answer my research question, I am resigned to using measures which are less than ideal.

The British Election Study: data set, ethics, and limitations My chosen data set consists of the first nine waves of the 2014-2017 British Election Study internet panel (Fieldhouse et al, 2016), which I analyse using SPSS (IBM Corp, 2016). The British Election Study is the United Kingdom’s foremost political survey programme, managed jointly by the universities of Manchester, Oxford, and Nottingham. Its internet panel studies, collected by YouGov, follow the same representative sample of individuals over time- usually a period of two to three years- administering questionnaires repeatedly to them for the duration of the time period. These questionnaires include both unique questions, asked only once to each respondent, and repeated instances of the same question asked multiple times in different “waves” of the study. This multiple-questionnaire structure is the chief advantage of panel data and my principal reason for utilising this data set, even though my analysis is not a longitudinal one. One key aspect of constraint, as Converse understood it, is the extent to which policy preferences remain stable over time or fluctuate, potentially due to random “noise” (Converse, 1964). Asking the same people the same question several times in succession, interspersed with gaps of several months, should help to ascertain the extent to which policy preferences are random or stable in the British electorate. What’s more, the practice of dispersing single- use questions across multiple waves has allowed the data set to compile a far broader array of variables than would feasibly fit within a single cross-sectional questionnaire.

It is easy to overlook ethical issues when analysing secondary data, which another agency has already collected and anonymised (Boddy, 2016). After all, crucial decisions about harm to participants, informed consent, invasion of privacy and deception have already been made and independently approved (Bryman, 2012: 135). As a researcher using secondary data, my main ethical duty is twofold: to produce a rigorous and high-quality analysis to ensure that the reputation of social research as a profession is not harmed by association with poor-quality work (Dillman, 2007), and to interpret my results proportionally and impartially, to ensure that individuals of certain political views are not unfairly maligned by misunderstandings of evidence (Haidt, 2013), I shall endeavour to demonstrate both of these considerations by being as transparent about my work as possible, and by including supplementary results and information in the technical appendix.

A total of 30,590 people took part in the first wave of the panel study, starting in February 2014 (British Election Study, 2015). Unfortunately, due to commercial reasons, YouGov does not make details of its response rates publicly available. This makes it difficult to know how many people were originally asked to take part. However, it is possible to ascertain that 10,172 people, 18% of those who answered the first questionnaire, responded to all nine waves represented in the data. This reveals a considerable problem with “attrition”, in which panel members contact with the researchers over time (Tarling, 2009). As those respondents who drop out of the study are likely to be different in some meaningful way to those who remain, attrition makes it likely that the remaining sample will be biased and hence unrepresentative in some way. The data set includes a carefully calculated weight variable to correct some of this bias, which I have used, but this can only correct for biases which are directly measured by questions asked in the survey. As there could be any number of potential unmeasured biases, any deviation from the full response of a completely random sample is problematic. 12

Grasping slippery ideas: operationalising concepts as variables My research question refers only to the generic abstractions of culture, perception and policy preferences (see Figure 2). These terms on their own are far too vague to use for the formulation of testable hypotheses. As such, my next task was to operationalise these concepts by selecting or constructing variables in the British Election Study data set which can be used to approximate them.

Political orientation: dimensions and underlying forces If policy preferences are indeed the stuff of which left and right are made, as I have argued above, then it is crucial to understand how a vast array of preferences for specific and heterogeneous policies can result in a simple system of classifying people into broad ideological camps. The unidimensional “political spectrum” model assumes that for every policy area there is a “left” preference and a “right” preference, and none other, apart from perhaps a centrist preference. Yet many scholars have criticised such an approach, insisting that at least two separate dimensions are necessary to understand the complexities of political orientation (Giddens, 1994; Rokeach, 2000; Feldman and Johnston, 2013).

Twenty-six Likert scale variables were available in the data set as measures of respondent attitudes towards specific policy areas, which I recoded so that higher values always represent more left-wing or liberal attitudes. Figure 3 illustrates the interrelations between these policy preferences in the form of a network diagram, where points represent individual policy items and the lines joining them represent Spearman rank correlations between each pair of items (after Azevedo, 2016). The thicker the line joining two preferences, the stronger the correlation between them. Preferences are positioned so that more correlated items are closer together.

Figure 2. Conceptual diagram indicating the potential ways in which cultural considerations may contribute to perceptual tendencies in the formation of policy preferences. It is possible that cultural factors are directly influenced by individual perception and themselves exert a direct influence on policy preferences. It is also possible that cultural factors influence the strength of other relationships, such as the direct influence of perception on political orientation. Both types of arbitration will be hypothesised using different operationalised variables and tested in the analysis below.

Figure 3

As Figure 3 demonstrates, the policy preference items are closely interconnected, indicating some kind of underlying commonality, as is suggested overwhelmingly in the literature. The significant correlations used to produce the diagram varied in strength from a low of 0.02 to a high of 0.78 and were all positive, suggesting that it is conceivable, at least in the United Kingdom, to contrast left-wing liberalism with right-wing conservatism. Nevertheless, there is still an indication of a divide between “economic” and “social” policies.

To investigate how many dimensions of policy may characterise these data, I employed a technique known as principal component factor analysis. This method involves taking a set of intercorrelated variables and attempting to explain variations in them using a smaller number of hypothetical underlying “factors” (Tarling, 2009). These factors are then associated by the researcher with presumed real-life traits which, although not directly observed, can be theoretically imagined to drive observed responses.

I first ran principal components analyses separately on each collection of repeated policy preference questions (see technical appendix). Unsurprisingly, each set of repeated measures could plausibly be explained by a single underlying factor, which I concluded could be taken to represent durable dispositions toward certain specific policies. For example, I reduced the four variables on whether people are unemployed by their own fault to a single factor measuring the same attitude as an enduring disposition over time. This has helped to confirm that policy preferences do generally remain stable and intelligible in the short-term.

The next step is to find out how thematically distinct policy preferences relate to one another. To do this, I incorporated the six repeated-measures factors already saved, and an additional four policy preference items which were only administered to the respondents once, into an additional principal component analysis. This analysis revealed that most of the variations in the ten policy preferences could be explained by three distinct dimensional factors. The first, strongest factor was most closely related to the policy preferences on issues involving immigration, the second to the items on economic management, and the third to the items on social regulation. Whilst the precise number of dimensions found is likely to be an artefact of the range of policy topics in the survey (Azevedo, 2016), it is safe to conclude that policy preferences are, at least sometimes, multidimensional in structure.

One further question is whether the distinct dimensions of political orientation are nevertheless driven by a single, unidimensional left-right divide focusing on differing attitudes to equality (Bobbio, 1997; Jost et al, 2003). Most studies are unable to investigate this question because they use a version of factor analysis which assumes that factors derived from the same analysis are totally uncorrelated with one another. This assumption is, to say the least, unreasonable (Jost et al, 2009). By contrast, my factor analyses employ direct oblimin rotation, a technique which allows derived factors to correlate, and thus be included in a further, final factor analysis. On doing this I discovered that all three of the policy dimensions could be explained by a single latent factor, which I dubbed political leftness and determined to use as my dependent variable, the chief variable of interest to be explained in my analysis.

Perception: openness, uncertainty, and ambivalence The key measure of open-mindedness in the British Election Study panels is the Active open- mindedness scale, a series of seven items asking respondents to broadly indicate how willing they are to change their opinions in the face of new information or evidence. Each item measures a slightly different aspect of active open-mindedness, and the scale was only administered to the respondents on one occasion. The seven items exhibited a high Cronbach’s alpha of 0.76, indicating that they are highly intercorrelated. A principal component factor analysis produced a single factor able to explain considerably more variance in these seven items than any other factor. I saved the factor from this analysis for use as my chief independent or explanatory variable, treating it as a measure of cognitive flexibility.

An additional set of variables on levels of comfort with uncertainty and risk may tap into a related aspect of the opposing needs for novelty and security (Roets and Van Hiel, 2011). I recoded the five factors in question so that higher values refer to greater levels of comfort with uncertainty. The five factors highly intercorrelated- with a Cronbach’s alpha of 0.72- but factor analysis produced two underlying variables rather than just one. The first factor was related most strongly to tolerance of uncertainty, whereas the second related to propensity to taking risks. I chose to use both factors, treating tolerance of uncertainty as an auxiliary measure of cognitive flexibility, but regarding risk-taking propensity as simply a control variable.

The final measure of perception which I selected is response polarisation, a measure which I introduced in an earlier work (Harman, 2017). This variable was devised by myself as an answer to the absence of usable psychometric variables in most representative social surveys. It essentially measures the same basic quality as the Need for Closure scale: a desire to sort experiences swiftly and decisively into clear categories unclouded by uncertainty or ambiguity. Response polarisation measures this tendency by finding the average extent to which survey respondents provide polarised responses to Likert scale (“agree or disagree?”) survey questions. Response polarisation removes the direction- agreement or disagreement- of the response, so that those who “strongly agree” and “strongly disagree” are given the same score. Thus it measures, irrespective of the actual content of their responses, how much on average a given respondent provides strong, decisive responses rather than moderate, equivocal ones.

I selected 100 Likert scales from the variables in the data based on the following criteria: being ordinal rather than nominal, involving subjective, non-concrete distinctions between “strong” and “weak” responses, being symmetrical by having equal number of options on the “agree” and “disagree” sides, and having at least four response options, so that there is at least one choice between “strong” and “weak” responses. The scales do not necessarily need to follow a traditional “agree or disagree” format, so long as subjective differences in strength of opinion on a given topic are allowed to manifest. I duplicated each of these 100 variables and recoded them so that the centremost response, if present, became 0, the two responses either side of the centre became 1, the two responses either side of them became 2, and so on until all the responses had been recoded. These variables were highly intercorrelated, regardless of question content or number of response options (see technical appendix), and factor analysis saved one variable as a generalised indicator of cognitive classification. 16

Culture: cluster, conformity, and constraint Culture is perhaps the hardest to operationalise of all the concepts in this study. The concept is somewhat vague, has been underused in quantitative research, and historically has been understood to mean quite different things in different theoretical traditions (Soares et al, 2007). I will specify three distinct aspects of culture, each of which is likely to bear a different statistical relationship to perception and culture to the others. I use the variables available in the British Election Study data to create four variables to represent these three basic concepts.

Perhaps the simplest version of culture as a variable is cultural cluster: a categorical variable outlining the cultural group to which each individual belongs (Fearon, 2003). Examples of cultural clusters or categories are the respondent’s country of residence, ethnic group, or religious affiliation. Previous quantitative studies have all focused solely on cultural cluster, mostly comparing differences in the relationship between perception and policy preferences by country (Kossowska and Van Hiel, 2003; Fu et al, 2007; Harman, 2017).

The British Election Study takes place solely within a single sovereign state- the United Kingdom- yet clusters of cultural affiliation are nevertheless likely to exist within this context. To identify them, I used a procedure known as two-step cluster analysis. Cluster analysis uses one or more directly measured variables, categorical or continuous, to produce a single variable classifying respondents into a small number of nominal categories, which its algorithms attempt to make as homogeneous and distinct from one another as possible. For this analysis, the constituent variables were all categorical, assessing respondents’ ethnicity, country of birth, country of residence within the United Kingdom, whether they have ever lived abroad, and whether at least one of their parents had been born in another country.

The two-step cluster analysis produced three clearly defined groups, reporting a reasonably high score for the “silhouette measure of cohesion and separation” of 0.6 (1.0 is the maximum and 0.5 is considered the minimum for quality clustering). Comparing this new cluster variable with the original, directly measured variables revealed that the first cluster consisted of heterogeneous individuals, including all non-White British respondents, most of whom had lived abroad at some point and some of whom were born outside of the United Kingdom or had one or more foreign parent. The second and third clusters were entirely White British, born and living only within the United Kingdom and with entirely British parents. The second consisted mostly of those born and resident within Scotland and Wales, whereas the third was composed entirely of those native to England. The clusters comprised 30%, 13% and 58% of valid responses in the sample respectively. For convenience, I dubbed them the “External”, “Celtic” and “Anglo-Saxon” cultural clusters. In most initial, exploratory, regression models, the “External” and “Celtic” clusters did not differ significantly. For this reason, I recoded the cluster variable to have two values, “Anglo-Saxon” and “Other”, for the final analysis.

The implication of the literature- that the relationship between perception and policy preferences differs significantly between cultures- is that cultural cluster acts as a categorical moderator of this relationship. Statistical moderation is where the value of one variable- the moderator- influences the strength of the relationship between two others (Hayes, 2013). 17

Based on this theory, I would expect to find that values for political leftness depend on the multiplicative effect of active open-mindedness and cultural cluster, working together.

An alternative metric which explores levels of individual motivation instead of group membership is cultural conformity (Fu et al, 2007; Siedlecki et al, 2016). According to the theory of motivated cultural cognition, individuals in rightward-leaning societies who are more open-minded are more left-wing, but only because they are less culturally conformal. Similarly, more open-minded individuals in left-leaning societies are more right-wing, again because they attach less value to prevailing cultural norms (Taber and Young, 2013). In statistics, this form of relationship is known as a mediation, where some or all of the effect of one variable on another is carried indirectly, through mutual connections with a third, mediating variable (Hayes, 2013). I would expect that cultural conformity mediates the direct and indirect influences of active open-mindedness on political leftness.

To produce a variable for individual cultural conformity, I ran another series of principal component factor analyses, like those used to estimate political leftness. Several variables in the British Election study tap into feelings of cultural allegiance, measuring generic feeling of “Britishness”, sense of threat to traditional culture and values, and the extent to which respondents believe British culture and society to be superior to others. I recoded these factors so that higher values refer to greater cultural conformity, and used a short series of factor analyses to generate a single scale measuring this concept.

The above measure of cultural conformity is useful, but should not be exclusively relied upon. Firstly, it is a self-reported measure; respondents may sustain a self-image of being highly attached to British culture, whilst resisting cultural influences and prevailing opinions on many specific topics. Secondly, the measure has much in common with political ideals of nationalism and opposition to immigration and thus may overestimate the strength of the (likely negative) relationship between cultural conformity and political leftness. For these reasons, I devised an alternative measure of cultural conformity based on survey-answering behaviour rather than on specific question topics. For each individual, I calculated the difference between their own score on the political leftness factor and the mean score on this factor for the entire data set. This would, of course, correlate perfectly with political leftness, but I then took the squared square root of this distance, rendering all distances positive and removing any automatic association between the distances and the direction of respondents’ political orientations. I reflected these distance values by subtracting each from the highest value present so that higher values came to mean greater conformity to the ideological mean. Finally, I derived a standardised z-score for this variable. I dubbed this variable “ideological conformity”, treating it as an auxiliary measure of conformity to the average position in public opinion. The two measures of conformity were positively correlated, with a Pearson coefficient of 0.26 at a 99.9% level of confidence.

My final understanding of culture is inspired by Philip Converse’s concept of belief system constraint. Converse found that the political belief patterns of American voters were not all equally constrained; rather, only a small minority were highly constrained or “ideological”

(Converse, 1964). Whilst my principal component factor analyses have found British public opinion to be considerably more stable and systematic than Converse might have predicted, his findings at least suggest that individuals differ markedly in the extent to which their policy preferences are interdependent (Johnston et al, 2017). This is not to say that those with constrained ideologies are likely to be cultural conformists, or at least not conformists to the majority national-ethnic culture: most citizens of most countries have little interest in politics (Alvarez and Brehm, 2002). Constraint is more likely to measure acceptance of the elite cultural understanding of a cohesive and one-dimensional “political spectrum”.

I would suggest that ideological constraint functions as a continuous moderator of the relationship between open- or closed-mindedness and policy preferences. This is similar to the potential moderating role played by cultural cluster, except that the strength of the aforementioned relationship is different at every point along the constraint scale, being strongest among those with the highest constraint. This is likely to be the case because only individuals with a systematic understanding of political culture should be able to effectively connect their perceptual predispositions with their policy choices (Zaller, 1992; Arzheimer and Schoen, 2016). In this sense, constraint may function as a different and complementary belief system dynamic to cultural conformity and culture group clusters.

As with my measure of ideological conformity, I based my measure of belief system constraint on the policy preference variables, transforming them in such a way as to make the measure of constraint independent of both ideological conformity and political leftness. I recoded the policy preference items so that all followed a uniform five-point ordinal scale structure. These items were already arranged in line with the traditional political spectrum, equating liberal and pro-immigrant policies with left-wing ones, and conservative and nationalist policies with right- wing ones. As such, a person with the same numerical score for every single policy preference item would demonstrate maximal constraint between different areas of policy, and between the same areas of policy across time. I computed a new variable, the standard deviation between the twenty-six policy preference items for each individual. This effectively became an inverse measure of constraint, with higher values indicating less constraint, which I reflected and standardised this statistic to prepare for use in my analysis.

Ruling out the alternatives: control variables I selected and refined several additional variables to use as statistical controls; consideration of these variables, which have been found to relate to political orientation in studies cited below, ensures that they are not responsible for driving the relationships observed among the variables of interest. They thus represent additional guarantees of statistical validity. The additional control variables selected are as follows:

i) Individual item-level response rate, based on the percentage of questions to which respondents provided a valid answer (Couper and Kreuter, 2011). As this variable was highly negatively skewed, with a small minority of individuals having an extremely low response rate, I reflected the variable, obtained a natural logarithmic transformation

of it, which was approximately normal in distribution, then reflected and standardised the resulting variable to return its original direction and set its mean to zero. ii) Two standardised scales tallying the number of times a respondent indicated that they feel fear and anger in response to specific political parties, as measures of the role of emotions in respondents’ policy decisions (Brader and Marcus, 2013; Burkitt, 2014). iii) Repeated measures of a question asking respondents how much they feel the economic situation of their household has improved or deteriorated in the twelve months prior to answering the question, recoded so that higher values mean greater perception of personal economic improvement and saved as a single factor (Clarke et al, 2015). iv) Repeated measures assessing respondents’ general trust in Members of Parliament, likewise saved to a single factor (Balliet et al, 2016). v) Repeated measures assessing respondents’ liking of David Cameron and the Conservative Party, condensed into a single factor measuring Conservative partisanship (Allan and Scruggs, 2004). vi) A series of questions assessing political engagement, including interest in politics, exposure to political information through various media channels, and propensity to vote. Through multiple stages of factor analysis, I condensed these measures into a single variable for political engagement (Flanagan, 2003). vii) Respondent newspaper readership, recoded into broadsheet readers, tabloid readers, and those who read no newspaper or another sort of newspaper (Gentzkow et al, 2011). viii) Level of skill required in respondents’ employment, recoded into highly skilled, moderately skilled, low skilled jobs, and never worked (Evans and Dirk de Graaf, 2013). ix) Demographic dummy variables for whether the respondent has been to university (Bobo and Licari, 1989), whether the respondent is disabled (Putnam, 2005), and whether the respondent is female (Sidanius et al, 2000). x) Respondents’ age in whole years on the birthday prior to their completing Wave 7 of the study in April-May 2016, standardised (Hellevik, 2002). Conditional process analysis: hypotheses and how to test them Having derived variables to match the concepts discussed in the theoretical literature, I am now able to present a list of hypotheses to be tested in the following analysis. Each hypothesis is accompanied by a theoretical source substantiating it, and by an equation specifying mathematically what the expected significant effects are to be.

Figure 4 illustrates the model proposed by these hypotheses, first as a conceptual diagram developed as a variable-specific version of Figure 2, then as a statistical diagram specifying the actual model to be tested, with annotated relationships representing coefficients which will be produced by the following regression analyses. These are the same letter-codes which are present in the equations underlying each hypothesis. The chief difference between the two diagrams is that in a statistical diagram, arrows cannot point to other arrows, only to variables. As such, moderating relationships are represented by interaction terms, additional variables in the model which consist of the proposed moderator multiplied by the independent variable.

Figure 4. Conceptual and statistical diagrams representing the model of motivated cultural cognition. Each box is a variable, and each arrow is a relationship between variables. Where an arrow points at another arrow in the conceptual diagram, this represents a moderation and is represented in the statistical diagram as an additional variable consisting of the two influencing variables multiplied together. The influences labelled “e” are error terms, or variations in the three variables to be predicted (the dependent variable and the two mediators) which cannot be explained by any of the variables in the model. It must be noted that the direction of arrows is theoretically informed and cannot be proved by statistical analysis.

The hypotheses are as follows:

H1. That the total effects of active open-mindedness, henceforth X, on political leftness, henceforth Y, are significant and positive (Jost, 2017).

c = c1’ + c4’ + c5’ + b21 + b22 + b31 + b32 + (a11*b11) + (a12*b12) > 0 H2. That the effect of X on Y is significantly mediated by conformity to Britishness and British values, henceforth M1 (Kossowska and Van Hiel, 2013).

a11*b11 > 0 H3. That the effect of X on Y is significantly mediated by similarity to the British ideological average point, henceforth M2 (Van Hiel et al, 2006).

a12*b12 > 0 H4. That the influence of X on M1 is significantly negatively moderated by tolerance of uncertainty, henceforth W (Greenberg and Jonas, 2003).

a41 < 0 H5. That the influence of X on M2 is significantly negatively moderated by W (Haas and Cunningham, 2014).

a42 < 0 H6. That the influence of X on M1 is significantly negatively moderated by response polarisation, henceforth Z (Roets and Van Hiel, 2011).

a51 < 0 H7. That the influence of X on M2 is significantly negatively moderated by Z (Webster and Kruglanski, 1994).

a52 < 0 H8. That the influence of X on Y is significantly negatively moderated by belonging to the Anglo- Saxon as opposed to other cultural clusters, henceforth V (Harman, 2017).

c4’ < 0 H9. That the influence of M1 on Y is significantly negatively moderated by V (Fu et al, 2007).

b21 < 0 H10. That the influence of M2 on Y is significantly negatively moderated by V (Golec De Zavala et al, 2010).

b22 < 0 H11. That the influence of X on Y is significantly negatively moderated by belief system constraint, henceforth Q (Zaller, 1994).

c5’ < 0 H12. That the influence of M1 on Y is significantly negatively moderated by Q (Feldman, 2013).

b31 < 0 H13. That the influence of M2 on Y is significantly negatively moderated by Q (Converse, 1964).

b32 < 0

The statistical model used to test my hypotheses will be drawn from the methodological tradition of “conditional process analysis” (Hayes, 2013). Conditional process analysis is an enhanced application of ordinary least squares regression, a technique which intends to explain variation in a key dependent variable using variations in a series of independent variables (Gujarati and Porter, 2008).

One of the greatest criticisms of straightforward regression analysis is that it treats variables and relationships as oversimplified monoliths: effects of independent on dependent variables are assumed to be direct, and not “carried through” intermediary variables, and the strength of the relationship is assumed to apply homogeneously to all individuals in the data. Conditional process analysis is intended to redress these shortcomings: it is “conditional” in the sense that the relationship between X and Y is not universal, but differs between different people in different circumstances. This allows it to explain not only whether a relationship exists, but how it exists- through mediation analysis- and when it exists- through moderation analysis (Hayes, 2013: 5). Conditional process analysis is thus a suitable way to test the plausibility of the theory that cultural forces moderate and mediate the influence of personality on policy preferences.

Professor Andrew Hayes, the chief proponent of conditional process analysis, has developed a useful tool called PROCESS, a macro for IBM SPSS Statistics which allows the data analysis program to readily run integrated moderation, mediation and hybrid analyses (Hayes, 2013). I have used the PROCESS macro to run a moderated mediation analysis of the statistical diagram in Figure 4, the syntax and output for which are available in the technical appendix.

My approach to missing data was to insert mean values for each variable (Downey and King, 1998). This has the advantage of being a simple and readily understandable method of estimating unobserved values which, by definition, will not alter the averages of any of the statistics. I acknowledge that the method known as multiple imputation, which generates multiple estimates of each missing value, is a more authentic approach to this problem, and avoids some of the distortions caused by insertion of means (Brunton-Smith et al, 2014). However, as the PROCESS macro is presently unable to support data sets using multiple imputation, I judged that the substantive advantages of PROCESS outweigh the inferential advantages of multiple imputation for the present analysis.

The results Overview of results The results of the conditional process analysis are summarised in Table 1, which reports the results from four regression models: a “total effect model”, measuring the complete influence of active open-mindedness on political leftness without controlling for cultural conformity (for testing H1), models measuring the influence of active open-mindedness on both cultural conformity mediators (for testing H2, H3 and H4-H7), and a model measuring the impact of active open-mindedness and the two mediators together on political leftness (for testing H2, H3 and H8-H13). I will explain these results in detail below.

Total effect model Model predicting Model predicting Final model predicting predicting leftness cultural conformity by ideological conformity leftness by open- by open- open-mindedness by open-mindedness mindedness and both mindedness conformity mediators R Squared 0.39 0.38 0.21 0.58 F Statistic 293.62*** 171.77*** 72.14*** 301.45*** Constant 0.21*** (0.03) -0.05 (0.04) -0.02 (0.05) -0.02 (0.04) Cultural conformity n/a n/a n/a -0.37*** (0.01) “b11” (M1) Ideological conformity n/a n/a n/a -0.25*** (0.01) “b12” (M2) Active open- 0.19*** (0.02) “c” -0.12** (0.04) “a11” -0.08 (0.04) “a12” 0.01 (0.04) mindedness (X) Anglo-Saxon cultural -0.10*** (0.02) n/a n/a 0.09*** (0.02) cluster (V) X * V interaction n/a n/a n/a 0.08 (0.06) “c4’” M1 * V interaction n/a n/a n/a -0.11*** (0.02) “b21” M2 * V interaction n/a n/a n/a 0.15*** (0.02) “b22” Belief system 0.03** (0.01) n/a n/a 0.04*** (0.01) constraint (Q) X * Q interaction n/a n/a n/a 0.01 (0.03) “c5’” M1 * Q interaction n/a n/a n/a -0.11*** (0.01) “b31” M2 * Q interaction n/a n/a n/a -0.02 (0.01) “b32” Tolerance of 0.09*** (0.01) -0.09*** (0.01) -0.06*** (0.01) n/a uncertainty (W) X * W interaction n/a -0.06 (0.03) “a41” -0.08* (0.04) “a42” n/a Response -0.19*** (0.01) 0.16*** (0.01) -0.37*** (0.02) n/a polarisation (Z) X * Z interaction n/a -0.09** (0.03) “a51” -0.13*** (0.04) “a52” n/a Item-level response 0.04*** (0.01) 0.00 (0.02) -0.01 (0.02) -0.01 (0.01) Risk-taking -0.03*** (0.01) 0.00 (0.01) 0.03* (0.01) -0.01 (0.01) Anger at parties 0.16*** (0.01) -0.13*** (0.01) -0.09*** (0.01) 0.00 (0.01) Fear of parties 0.10*** (0.01) -0.05*** (0.01) -0.02 (0.01) 0.02 (0.01) Conservative -0.48*** (0.01) 0.52*** (0.01) 0.10*** (0.02) -0.26*** (0.01) partisanship Trust in MPs 0.19*** (0.01) -0.10*** (0.01) -0.08*** (0.01) 0.17*** (0.01) Positive household 0.00 (0.01) -0.03* (0.01) -0.05*** (0.01) 0.01 (0.01) economic evaluations Political engagement 0.09*** (0.01) -0.03 (0.01) -0.02 (0.02) 0.08*** (0.01) Female -0.11*** (0.02) -0.04 (0.02) -0.08** (0.03) -0.13*** (0.02) Age in years -0.12*** (0.01) 0.24*** (0.02) 0.10*** (0.02) 0.02 (0.01) Attended university 0.12*** (0.02) -0.24*** (0.03) -0.10*** (0.03) 0.15*** (0.02) Disabled 0.04 (0.02) 0.06* (0.03) 0.06* (0.03) 0.03 (0.02) Highly-skilled (vs -0.03 (0.03) 0.01 (0.04) 0.06 (0.05) 0.02 (0.03) never worked) Moderately skilled (vs -0.14*** (0.03) 0.02 (0.04) 0.00 (0.05) 0.00 (0.03) never worked) Lower-skilled (vs -0.18*** (0.03) 0.05 (0.05) 0.05 (0.06) 0.02 (0.03) never worked) Broadsheet reader (vs 0.26*** (0.03) -0.25*** (0.03) -0.23*** (0.03) 0.09*** (0.02) other or no paper) Tabloid reader (vs -0.24*** (0.02) 0.27*** (0.03) 0.21*** (0.03) -0.19*** (0.02) other or no paper) Table 1. Unstandardised regression coefficients and standard errors (in brackets) from the conditional process analysis summarised in the statistical diagram in Figure 4. The R Squared of each model represents the proportion of variance in the dependent variable for each column which can be explained by the independent and control variables listed in column 1. The F statistic is an arbitrary number, but a higher statistic, when significant, demonstrates that one model is a more appropriate description of a population than another, similar model. Where a coefficient is listed as “n/a“ the variable in that row was not featured in the model in that column, and hence no coefficient was calculated. Where a coefficient is accompanied by a letter and numbers in inverted commas, e.g. “a52”, this represents the arrow in the statistical diagram in Figure 4 which the coefficient represents, and which features in the hypotheses. *=95% confidence. **=99% confidence. ***=99.9% confidence.

The influence of active open-mindedness on leftness, regardless of culture The “total effect model” summarised in the second column of Table 1 estimates the total effect of active open-mindedness on political leftness (c) when cultural conformity is not controlled for. This model is not produced by PROCESS when a moderated mediation analysis is requested. As such, I ran the total effect model using SPSS’s ordinary least squares regression commands. Full diagnostics of the residuals (variations in political leftness which cannot be explained by any of the predictor variables), not reproduced here, showed that the model meets all the assumptions required for the results of linear regression analysis to be relied upon, including absence of strong correlations between predictors, normal distribution of residuals either side of a mean of zero, lack of correlation between the residual values and the predicted values for political leftness, and lack of correlation between the residuals and their potential to seriously skew the results, a quantity known as leverage (Tarling, 2009).

The model explains around 39% of the variation in political leftness- a minority, but nevertheless a reasonably high figure by the standards of the social sciences. What’s more, the high and significant F statistic demonstrates that the model’s predictor variables make a noteworthy impact in explaining political orientation. The model’s constant, 0.21***, represents the expected value for political leftness for an individual with a score of zero on all predictor variables, thus belonging to the reference category- the category which is not assigned its own row- on all the categorical predictors. As all the continuous variables, even age, have been standardised, with the mean value as zero and negative values for all cases below the mean, the constant represents the political orientation of a person who reports the mean values for all continuous variables, does not belong to the Anglo-Saxon cultural cluster, is male, has not attended university, is not disabled, has never worked and does not read a newspaper. The effects of every predictor variable are then to be added to the baseline provided by the constant.

The coefficient of 0.19*** in column 2, row 7 of Table 1 thus represents the full, positive effect of open-mindedness on political leftness, without distinguishing between components of that effect which may be direct, or which may be transmitted via mediating variables. I find, perhaps unsurprisingly given the conclusions of the literature summarised above, that more open- minded people are more left-wing. As this effect is significant and positive, I confirm H1.

The influences of active open-mindedness on conformity The models in columns 3 and 4 attempt to predict the two hypothesised mediators- cultural conformity and ideological conformity- using active open-mindedness. This is the first step in demonstrating a mediation, and thus in testing the hypotheses H2 and H3. If the independent variable has a significant effect on the mediators, and the mediators have a significant effect on the dependent variable, then part or all of the effect of independent on dependent has been transmitted indirectly via the mediators.

The models explain 38% and 21% of the variance respectively in the measures of conformity, and their lower F statistics suggest that, in general, conformity to belief systems is harder to

25 effectively predict using the chosen variables than political leftness. As hypothesised, active open-mindedness has a significantly negative effect on cultural conformity. Its negative effect on ideological conformity is not significant. Nevertheless, as will be seen in Figure 6 below, this non-significance is not universal. At many values of the moderators, those who are more open- minded are indeed more distant on average from the mean political position.

The alternative perceptual measures, response polarisation and tolerance of uncertainty, have more consistently significant impacts on the measures of cultural conformity. It should be noted, however, that the strong negative relationship between response polarisation and ideological conformity displayed in column 4, row 18 should not be interpreted, as it is an artefact of the way those two measures have been produced. Those with a tendency to give extreme responses to Likert scales will usually also give extreme, and hence nonconformal, responses to Likert scales regarding political orientation. Response polarisation does, however, have the expected positive effect on cultural conformity (0.16***). Tolerance of uncertainty, in many ways its opposite, has, as expected, a negative effect on both measures of conformity. This provides further evidence that open, flexible perceptual dispositions make individuals more willing to question cultural identities and accepted conventions.

Moderation by tolerance of uncertainty and response polarisation The models in columns 3 and 4 also test whether the negative influences of active open- mindedness on the two measures of conformity are stronger among respondents who are higher in tolerance of uncertainty and response polarisation. It should be noted that where a negative relationship is concerned, a negative value for the interaction with the moderator indicates that higher values of the moderator make the relationship stronger (Hayes, 2013). As the coefficients in row 17 demonstrate, there is no indication that tolerance of uncertainty enhances the influence of active open-mindedness on cultural conformity, but there is some evidence that it enhances open-mindedness’ effect on ideological conformity (-0.08*). The coefficients for the interaction with response polarisation in row 19 show that greater values of response polarisation increase the negative effects of active open-mindedness on both cultural conformity (-0.09**) and ideological conformity (-0.13***). These findings allow me to confirm H5, H6, and H7, although I cannot confirm H4.

The influences of conformity on leftness The final model in column 5 again turns to predicting values for political leftness, this time including both active open-mindedness and the measures of conformity as predictors. As the model shows, once cultural and ideological conformity are taken into account in predicting political leftness (at -0.37*** and -0.25*** respectively), active open-mindedness has no significant effect. This indicates that open-mindedness only positively influences leftness because it negatively influences conformity, which itself negatively influences leftness within the United Kingdom. The overall indirect effects, which are needed to test H2 and H3, are not monolithic, but vary with differing values of the moderators. These will be discussed below.

Moderation by cultural cluster and constraint The final model also enables me to assess whether the effects of open-mindedness on leftness, direct and indirect, are different in strength between cultural clusters and between different levels of constraint to the traditional political spectrum. The relevant coefficients are located in rows 9-11 and 13-15 respectively.

Neither of the cultural moderators interacts significantly with active open-mindedness directly in influencing political leftness. I thus cannot confirm H8 or H11. There is, however, an interaction between these moderators and the two conformity-based mediators. Among members of the Anglo-Saxon cultural cluster, the negative influence of cultural conformity on political leftness is stronger than among the members of other clusters. The opposite is true, however, when considering the influence of ideological conformity on political leftness: here, its influence is strongest among those who do not belong to the Anglo-Saxon cluster. Either way, this demonstrates that culture group membership is important in determining the relationships between perception, conformity, and political orientation. However, whilst I confirm H9, I am unable to confirm H10. The reason for this is that although cultural cluster membership interacts significantly with ideological conformity, the direction of this interaction is opposite to the direction hypothesised.

Belief system constraint significantly strengthens the negative effect of cultural conformity on political leftness (-0.11***). This is illustrated in Figure 5, a scatterplot comparing the cultural conformity and political leftness of individual respondents, who I grouped into clusters based on their level of constraint using a univariate two-step cluster analysis. Each of the lines represents the average relationship between cultural conformity and leftness within a cluster of individuals with similar levels of constraint. This demonstrates that the line is steepest- implying the strongest negative relationship- among the respondents with the greatest degree of constraint. I thus confirm H12. I cannot confirm H11 or H13, however, because the relevant coefficients representing interaction terms are not statistically significant.

Direct and indirect effects, at different levels of the four moderators Finally, in order to fully test H2 and H3, it is necessary to compare the direct and indirect effects of active open-mindedness on political leftness. The direct effect is simply the effect of response polarisation on political leftness, not including any effects which have passed through the mediators. The indirect effect of each mediator is the effect of active open-mindedness on the mediator in question, multiplied by the effect of that mediator on political leftness. A complication arises from the fact that, in a mediation analysis which also involves moderators, direct and indirect effects are not absolute, but conditional upon values of the moderators. As such, PROCESS did not generate a single figure each for direct and indirect effects, but a range of conditional effects. In total, six direct effects and 108 indirect effects- 54 for each mediator- were produced (these can be found in full in the technical appendix). These reveal nuances in the effects which are not shown in table 1. For example, whilst the overall direct effect of open- mindedness on leftness is not significant overall, it does become narrowly significant and positive amongst Anglo-Saxons with average (0.09*) or high (0.10*) levels or constraint.

Figure 5. Scatterplot showing the relationship between cultural conformity and political leftness, moderated in strength by belief system constraint. The R Squared statistics represent the proportion of variance in political leftness for each cluster of constraint that can be explained by cultural conformity for each respective line.

The best way to examine the multitude of indirect effects is to observe Figure 6. This diagram displays all 108 effect sizes, with 95% confidence intervals. Crucially, even though it would appear from Table 1 that active open-mindedness does not significantly influence leftness via ideological conformity, against the prediction of H3, it emerges that in many cases it does. In many cases, also, indirect effects transmitted via cultural conformity are not significant. An examination of the tables of effect sizes in the technical appendix reveals that the largest, most significant effects occur in conditions of high constraint, high tolerance of uncertainty and high response polarisation, with differing effects of cultural cluster between the mediators.

To gain a simpler understanding of whether H2 and H3 can be confirmed, I ran a separate mediation analysis in PROCESS, including the moderators as simple control variables without interaction terms. The main results of this analysis are not reported here, but can be found in the technical appendix. In this model, unlike the model in Table 1, the effect of active open- mindedness on ideological conformity is negative and significant (-0.10**). The total effect of open-mindedness on leftness was a highly significant 0.13. The direct effect was 0.05 and was narrowly non-significant. Indirect effects via cultural and ideological conformity were 0.05 and 0.03 respectively, both significant. Indirect effects thus account for 62% of the effect of open- mindedness on leftness. In sum, I found enough evidence to confirm H2 and, narrowly, H3.

Figure 6. High-low plot showing sizes of the indirect effects of active open-mindedness on political leftness, via both measures of cultural conformity, at different levels of the four moderating variables, arranged in order of size. Dots represent effect sizes, relative to the x axis. Bars represent 95% confidence intervals. Bar colour signifies the mediator through which the effect is transmitted. The vertical line represents an effect size of zero; when a bar touches or crosses this line, the indirect effect is not significant. For information on the levels of each moderator present for each indirect effect, use the reference number for each effect size which can be found on the y axis. Each number corresponds to a row in the effect size tables in the technical appendix.

As such, I confirm eight of the original thirteen hypotheses, the relationships proposed by which are illustrated in Figure 7 below, which is a replica of the conceptual diagram in Figure 4, with significant influences annotated with coefficients from table 1, and with non-significant influences removed. Openness to changing one’s beliefs in response to new information appears to weaken commitment to the prevailing local culture, which in turn, within the United Kingdom’s social context, leads to their becoming more left-wing. The strength of these relationships is not constant and universal, however, but depends on other psychological quantities concerned with attitudes toward uncertainty and ambiguity, actual cultural group membership, and constraint to the traditional political spectrum between left and right. The inclusion of these mediators and moderators adds considerable predictive power to the final regression model, as compared with the straightforward linear regression model in which culture is not considered, enabling it to explain an additional 19% of the variance in political orientation. Ultimately, I am able to explain 58% of the variance in leftness, a remarkably high proportion by the standards of survey research in the social sciences.

Figure 7. Conceptual diagram showing hypothesised relationships which have been confirmed, annotated with regression coefficients showing the strength, direction, and significance of those relationships. Because active open-mindedness has a negative influence on both measures of conformity, and these measures have a negative influence on political leftness, then active open-mindedness has an indirect positive influence on leftness. The dashed arrow for the role of cultural cluster in moderating the influence of ideological conformity on leftness represents the fact that whilst a significant moderation was found, the hypothesis regarding this moderation cannot be confirmed. This is because the hypothesised relationship was negative, whereas the observed relationship is positive. The dotted line for the influence of open-mindedness on ideological conformity is in recognition of the fact that, although this influence is not positive overall in the model displayed in Table 1, it is significant at higher values of the moderators, and is thus a valid depiction of at least some segments of society, and, additionally, is significantly negative in a mediation model involving both mediators but no moderators.

Discussion: implications for the science of policy preferences My analysis has recovered evidence to support some theories, challenge others, and add increased layers of nuance to still more.

Policy preferences are the root of my analysis: individuals exhibit measurable differences in levels of support for specific programmes of government action. The question is, where do these differences in support come from? I have demonstrated the usefulness of Converse’s (1964) conception of “belief systems”: networks of logically distinct policy preferences which are bound together by ties of socially arbitrary “constraint” (Converse, 1964). This concept appears to be, if anything, even more useful than Converse originally thought; whilst he argued that only the opinions of the most sophisticated are highly constrained, my research suggests that constraint between policy opinions is, in fact, widespread within mass publics.

The question then emerges, can these structures in policy preferences be accurately summarised using the terms “left” and “right”? To an extent, I confirm criticisms of a simple, one-dimensional political spectrum, uncovering a clear divide between economic and non- economic policies, as claimed by Rokeach and many others (Rokeach, 2000; Bryson and McDill, 1968; Mitchell, 2006). Indeed, rather than applying an arbitrary limit of two ideological axes, I find evidence that the precise number of dimensions produced by constraint between policy preferences is both dependent on measurement techniques and situationally contingent, as suggested by Alvarez and Brehm (Alvarez and Brehm, 2002). Nevertheless, the existence of distinct policy dimensions is not the same as the idea of independent policy dimensions, as implied by the idea of the political compass. I find that the dimensions of political orientation not only correlate, they can ultimately be explained by a single underlying factor which can be thought of as a generalised attitude towards social equality combining economic regulation with social freedom, as argued by Norberto Bobbio (Bobbio, 1997). This suggests that “left” and “right” are indeed meaningful as catch-all political terms summarising meaningful differences in the human population.

My research question aimed to explain why these differences exist. I find, in line with the theories of John Jost and much of mainstream political psychology, that those on the “left” do indeed score more highly on measures of open-mindedness, willingness to accept ambiguity, and tolerance of uncertainty, and less highly on measures of commitment to one’s beliefs, decisiveness, and preference for predictability (Jost et al, 2003; Jost, 2017). As anticipated by the long theoretical tradition originated by Theodor Adorno, political views appear to be anchored in psychological antecedents representing general firmness, or flexibility, of opinion (Adorno et al, 1964). Strict classification of ideas into “right” or “wrong” appears to accompany the strict classification of people into “worthy” and “unworthy”.

Yet my findings challenge any supposition that a single, monolithic “natural” or “inherent” relationship between psychology and politics exists for all people, or even for all people within a specific society. I present cultural variation as both a problem in the field of the social sciences (after Van Bavel et al, 2016) and a solution to some of its inconsistencies (Fu et al, 2007).

I find evidence to support the proposal of Kossowska and Van Hiel (2003), Harman (2017) and others, which holds that the relationship between openness or closure and public opinion differs in strength between cultural groups (Kossowska and Van Hiel, 2003; Harman, 2017). In this sense, perceptual influences on politics are moderated by, and thus conditional on, cultural identities.

I have also substantiated the claim made by Fu et al (2007) that at least part of the reason why open- or closed-mindedness can have such different political implications across societies is that closed perceptual dispositions encourage cultural conformity, whereas open ones encourage cultural cynicism (Fu et al, 2007). Depending on cultural content, these tendencies to support or attack the dominant idea-system can result in support for very different policies. In a right-wing culture, the counterculture is left-wing, and vice versa. Deriving two measures approximating conformity, I find that more open-minded individuals generally express less attachment to British culture and society, which would otherwise have had a right-wing political influence over them. I find weaker, though non-negligible, evidence to support the case that more open-minded people tend to prefer policies which are more unusual or unconventional by the standards of their host society, and that, in a generally rightward- leaning society, these policies tend to be more left-wing. When these two conformal influences are taken into account in predicting political leftness, active open-mindedness no longer has any significant direct effect upon the latter. To underscore the importance of cultural identity in determining the precise influence of cultural conformity on policy preferences, I discover that the influences of both measures of conformity on political leftness are also significantly moderated by cultural cluster membership. I thus reveal a wide selection of evidence that perceptual influences on politics are mediated by, and thus dependent on, cultural conformity and anti-conformity (Siedlecki et al, 2016).

Finally, I find evidence that Conversian belief system constraint, itself a metric of cultural origin, interacts with conformity to strengthen the indirect influence of perception on political orientation. Individuals who accept both dominant British values and the dominant assumptions of political culture regarding which policies “go together” tend to be the most right-wing. Conversely, those who are highly opposed to British culture and yet most attuned to patterns of constraint tend to be the most left-wing. This supports the theory articulated by John Zaller and others, holding that understanding of the dominant factions and dimensions within political belief systems enables individuals to more effectively connect their general values and dispositions with support for specific programmes of state action (Zaller, 1992).

In sum, I have further specified and largely substantiated the model of “motivated cultural cognition” first elaborated by Jeanne Fu and her colleagues (Fu et al, 2007). Cognition is indeed motivated by differing psychological dispositions, as John Jost and others claim (Jost et al, 2003), but these motivations do not influence attitudes towards policies directly. Instead, perceptual motivations target socially shared belief systems. The more “open-minded” become more disposed to hold their own, unique view on a given topic, and thus develop an inbuilt suspicion toward whichever ideas are most popularly accepted. These effects are dependent on individual cultural identity and alignment to the dominant political belief system.

Conclusion: limits of the study and needs for further research This study has provided some interesting empirical evidence to suggest that policy preferences originate from an interplay between individual perceptual dispositions and collective cultural systems of integrated beliefs and values. Both the promise of this theory, and the limitations of my own approach to test it, suggest a rich vein of avenues for further research.

The chief flaw of my analytical method is its inability to demonstrate causal relationships. An association between variables is necessary to prove causality, but not sufficient. To demonstrate that open-mindedness causes reduced cultural conformity, which in turn causes reduced political leftness, it is necessary to demonstrate that the causing variables precede the caused variables in time. For this purpose, a longitudinal study of a similar nature is valuable. The data which I have used, of course, have been collected on several consecutive occasions, but their two-year time span is not long enough to measure systematic changes in attitudes within individuals. A longitudinal study taking place over a significant number of years- say, at least four or five- would considerably aid causal inference.

Having used only secondary data, I have been limited in my choice of variables and have been unable to design questionnaire content to minimise error and plan appropriate question ordering. A representative survey targeted to this topic of research, where funds allow, would be highly advantageous.

As noted above, the structure and dimensionality of policy preferences are, to a great extent, a function of the number and content of preferences measured in a survey. The policy preferences selected were fewer in number than would ideally be desired. What’s more, they were skewed in content, including far more items on immigration than any other policy area. Many additional policy preference items are available in the British Election Study, but most of these could not be used because a considerable number of responses were missing. A future study aiming to analyse a fuller, more comprehensive selection of policy preferences might employ a different data set with less missing data, a more sophisticated approach to the imputation of missing values, or both.

The restrictions and limitations of the PROCESS macro partly account for some of the shortcomings in my analysis. PROCESS is not compatible with multiple imputation of missing values; will accommodate multiple mediators, but assumes that, in models involving moderators, they are not causally associated; and provides a fixed selection of models to test which, although large, cannot be customised by the user. Structural equation modelling, which integrates the testing of flexible path-based relationships with factor analysis, could be used to help build more sophisticated models of belief system dynamics.

It is easy to confuse “left” and “right” with “right” and “wrong”- whichever side one identifies with. Psychological motivations and cultural influences are fundamental to the arbitrary subjectivity inherent in the human condition. Only by understanding these forces can we hope to unwrap the age-old mystery that is human public opinion.

Bibliography Abrica-Jacinto, N L, Kurmyshev, E, and Juárez, H A 2017 ‘Effects of the Interaction Between Ideological Affinity and Psychological Reaction of Agents on the Opinion Dynamics in a Relative Agreement Model’. Journal of Artificial Societies and Social Simulation. Vol. 20, No. 3.

Adorno, T W, Frenkel-Brunswick, E, Levinson, D J and Sanford, R N 1964 The Authoritarian Personality. Hoboken: John Wiley and Sons.

Allan, J P, and Scruggs, L 2004 ‘Political Partisanship and Welfare State Reform in Advanced Industrial Societies’. American Journal of Political Science. Vol. 48. No. 3. pp.496-512.

Alvarez, R M and Brehm, J 2002 Hard Choices, Easy Answers. Princeton: Princeton University Press.

Andrews, A C, Clawson, R A, Gramig, B M, and Raymond, L 2016 ‘Finding the Right Value: Framing Effects on Domain Experts’. Political Psychology. Vol. 38. No. 2. pp.261-278.

Archer, M S 1995 Realist Social Theory: The Morphogenetic Approach. Cambridge: Cambridge University Press.

Arendt, H 1958 The Human Condition. Chicago: University of Chicago Press.

Aristotle 1968 The Politics. London: Penguin Classics.

Arzheimer, K, and Schoen, H, 2016 ‘Political interest furthers partisanship in England, Scotland, and Wales’. Journal of Elections, Public Opinion and Parties. Vol. 26. No. 3. pp.373-389.

Atwan, R, and Roberts, J 1996 Left, Right and Center: Voices from Across the Political Spectrum. Boston: Bedford Books.

Axelrod, R 1973 ‘Schema Theory: An Information Processing Model of Perception and Cognition’. The American Political Science Review. Vol. 67. No. 4. pp.1248-1266.

Azevedo, F 2016 ‘The Structural Complexity and Validity of Conservatism Scales’. Presentation delivered to the 40th Annual Scientific Meeting of the International Society of Political Psychology. Available online: http://flavioazevedo.com/current- projects/2016/5/21/the-structural-complexity-validity-of-conservatism-scales [accessed 05/09/2017].

Badie, D 2010 ‘Groupthink, Iraq, and the War on Terror: Explaining US Policy Shift toward Iraq’. Foreign Policy Analysis. Vol. 6. No. 4. pp.277-296.

Balliet, D, Tybur, J, Wu, J, and Van Lange, P A M 2016 ‘Political Ideology, Trust, and Cooperation: In-group Favoritism among Republicans and Democrats during a US National Election’. Journal of Conflict Resolution. pp.1-22.

Benedict, R 2005 ‘The Individual and the Pattern of Culture’. In Moore, H L and Sanders, T, (eds.), Anthropology in Theory: Issues in Epistemology. Hoboken: John Wiley and Sons.

Berger, P L, and Luckmann, T 1966 ‘The Social Construction of Reality’. New York City: Anchor Books.

Block, J, and Block, J H 2005 ‘Nursery school personality and political orientation two days later’. Journal of Research in Personality. Vol. 40. pp.1-16.

Bobbio, N 1997 Left and Right: The Significance of a Political Distinction. Chicago: University of Chicago Press.

Bobo, L, and Licari, F C 1989 ‘Education and Political Tolerance: Testing the Effects of Cognitive Sophistication and Target Group Affect’. Public Opinion Quarterly. Vol. 53. No. 3. pp.285-308.

Boddy, J 2016 ‘The Ethics of Social Research’. In Gilbert, N, and Stoneman, P (eds.) Researching Social Life. 4th edition. London: SAGE Publications Ltd.

Borgatti, S.P., 2002. NetDraw Software for Network Visualization. Analytic Technologies: Lexington, KY.

Borst, A, and Theunissen, F E 1999 ‘Information theory and neural coding’. Nature Neuroscience. Vol. 2. No. 11. pp.947-957.

Bourdieu, P 2010 Distinction: A Social Critique of the Judgement of Taste. Abingdon-on-Thames: Routledge.

Boutyline, A, and Willer, R 2016 ‘The Social Structure of Political Echo Chambers: Variation in Ideological Homophily in Online Networks’. Political Psychology. Vol. 38. No. 3. pp.551-569.

Brader, T, Valentino N A, and Suhay, E 2008 ‘What Triggers Public Opposition to Immigration? Anxiety, Group Cues, and Immigration Threat’. American Journal of Political Science. Vol. 52. No. 4. pp.959-978.

Brader, T, and Marcus, G 2013 ‘Emotion and Political Psychology’. In Huddy, L, Sears, D O, and Levy, J S (eds.), The Oxford Handbook of Political Psychology. Oxford: Oxford University Press.

Brenner, C J, and Inbar, Y 2014 ‘Disgust sensitivity predicts political ideology and policy attitudes in the Netherlands’. European Journal of Social Psychology.

British Election Study 2015 British Election Study Questionnaire Wave 9 V1.0. Available online: http://www.britishelectionstudy.com/data-objects/panel-study-data/ [accessed 05/09/2017].

Brousmiche, K-L, Kant, J-D, Sabouret, N, and Prenot-Guinard, F 2016 ‘From Beliefs to Attitudes: Polias, a Model of Attitude Dynamics Based on Cognitive Modeling and Field Data’. Journal of Artificial Societies and Social Simulation. Vol. 19. No. 4.

Brunton-Smith, I, Carpenter, J, Kenward, M, and Tarling, R 2014 ‘Surveying Prisoner Crime Reduction (SPCR): Adjusting for Missing Data’. Technical Report. Ministry of Justice Research Series. Ministry of Justice. This publication is available for download at http://www.justice.gov.uk/publications/researchand-analysis/moj [accessed 04/09/2017]

Bryman, A 2012 Social Research Methods. 4th edition. Oxford: Oxford University Press.

Budge, I, Klingemann, H-D, Volkens, A, Bara, J, and Tanenbaum, E 2001 Mapping Policy Preferences: Estimates for Parties, Electors and Governments 1945-1998. Oxford: Oxford University Press.

Bunce, V J, and Wolchik, S L 2010 ‘Defeating Dictators: Electoral Change and Stability in Competitive Authoritarian Regimes’. World Politics. Vol. 62. No. 1. pp.43-86.

Burkitt, I 2014 Emotions and Social Relations. Thousand Oaks: SAGE Publications.

Caprara, G V, and Vechhione, M 2013 ‘Personality Approaches to Political Behaviour’. In Huddy, L, Sears, D O, and Levy, J S (eds.), The Oxford Handbook of Political Psychology. Oxford: Oxford University Press.

Carney, D R, Jost, J T, Gosling, S D and Potter, J 2008 ‘The Secret Lives of Liberals and Conservatives: Personality Profiles, Interaction Styles, and the Things They Leave Behind’, Political Psychology, Vol. 29. No. 6. pp.807-840.

Carraro, L, Castelli, L, and Macchiella, C 2011 ‘The Automatic Conservative: Ideology-Based Attentional Asymmetries in the Processing of Valenced Information’. PLoS ONE. Vol. 6. No. 11.

Chirumbolo, A 2002 ‘The relationship between need for cognitive closure and political orientation: The mediating role of authoritarianism’. Personality and Individual Differences. Vol. 32. No. 4. pp.603-610.

Chong, D 2013 ‘Degrees of Rationality in Politics’. In Huddy, L, Sears, D O, and Levy, J S (eds.), The Oxford Handbook of Political Psychology. Oxford: Oxford University Press.

Cichocka, A, Bilewicz, M, Jost, J T, Marrouch, N, and Witkowska, M 2016 ‘On the Grammar of Politics—or Why Conservatives Prefer Nouns’. Political Psychology. Vol. 37. No. 6. pp.799-815.

Citrin, J, Green, D P, Muste, C, and Wong, C 1997 ‘Public Opinion Toward Immigration Reform: The Role of Economic Motivations’. The Journal of Politics. Vol. 59. No. 3. pp.858-881.

Clarke, H D, Sanders, D, Stewart, M C, and Whiteley, P 2004 Political Choice in Britain. Oxford: Oxford University Press.

Clarke, D C, Goodwin, M, and Whiteley, P 2017 Brexit: Why Britain Voted to Leave the European Union. Cambridge: Cambridge University Press.

Converse, P E 1964 ‘The nature of belief systems in mass publics’. Critical Review. Vol. 18. No. 1-3. pp.1-74.

Couper, M P, and Kreuter, F 2013 ‘Using paradata to explore item level response times in surveys’. Journal of the Royal Statistical Society. Vol. 176. No. 1. pp.271-286.

Critchley, H D, Elliott, R, Mathias, C J, and Dolan, R J 2000 ‘Neural Activity Relating to Generation and Representation of Galvanic Skin Conductance Responses: A Functional Magnetic Resonance Imaging Study’. The Journal of Neuroscience. Vol. 20. No. 8. pp.3033-3040.

DeBell, M 2016 ‘Polarized Opinions on Racial Progress and Inequality: Measurement and Application to Affirmative Action Preferences’. Political Psychology. Vol. 38. No. 3.

Dillman, D A 2007 Mail and Internet Surveys: The Tailored Design Method. 2nd edition. Hoboken: John Wiley and Sons.

Douglas, M 2003 Purity and Danger. Abingdon-on-Thames: Routledge.

Downey, R G, and King, C V 1998 ‘Missing Data in Likert Ratings: A Comparison of Replacement Methods’. The Journal of General Psychology. Vol. 125. No. 2. pp.175-191. 35

Downs, A 1957 An Economic Theory of Democracy. New York: Harper Collins.

Durkheim, E 2008 The Elementary Forms of Religious Life. Oxford: Oxford World’s Classics.

Dyson, S D 2006 ‘Personality and Foreign Policy: Tony Blair's Iraq Decisions’. Foreign Policy Analysis. Vol. 2. No. 3. pp.289-326.

Economic and Social Research Council 2015 New impacts from ‘old’ data. Available online: http://www.esrc.ac.uk/news- events-and-publications/news/news-items/new-impacts-from-old-data/ [accessed 05/09/2017].

Ellen, R 2003 ‘Classification’, in Barnard, A. and Spencer, J. (ed.), Encyclopedia of Social and Cultural Anthropology. Abingdon- on-Thames: Routledge World Reference.

Evans, G, and Dirk de Graaf, N 2013 Political Choice Matters: Explaining the Strength of Class and Religious Cleavages in Cross- National Perspective. Oxford: Oxford University Press.

Evans, M 2011 ‘What a cruel, male-dominated culture!’ Produced by Malcolm Evans. http://www.evanscartoons.com/index.php [accessed 05/09/2017]. Available online: https://www.democraticunderground.com/10025207425 [accessed 05/09/2017].

Fearon, J D 2003 ‘Ethnic Structure and Cultural Diversity around the World: A Cross-National Data Set on Ethnic Groups’. To be presented at the 2002 Annual Meeting of the American Political Science Association, August 29-Sept. 1. Available online: http://web.stanford.edu/group/ethnic/workingpapers/egroups.pdf [accessed 26/07/2017]

Feldman, S 2013 ‘Political Ideology’. In Huddy, L, Sears, D O, and Levy, J S (eds.), The Oxford Handbook of Political Psychology. Oxford: Oxford University Press.

Feldman, S, and Johnston, C 2013 ‘Understanding the Determinants of Political Ideology: Implications of Structural Complexity’. Political Psychology. Vol. 35. No. 3. pp.337-358.

Fieldhouse, E, Green, J, Evans, G, Schmitt, H, van der Eijk, C, Mellon, J, and Prosser, C 2016 British Election Study Internet Panel Waves 1-9. DOI: 10.15127/1.293723.

Flanagan, C 2003 ‘Developmental Roots of Political Engagement’. PS: Political Science and Politics. Vol. 36. No. 2. pp.257-261.

Fromm, E 2001 The Fear of Freedom. Abingdon-on-Thames: Routledge Classics.

Fu, J H, Morris, M W, Lee, S, Chao, M, Chieu, C, and Hong, Y 2007 ‘Epistemic Motives and Cultural Conformity: Need for Closure, Culture, and Context as Determinants of Conflict Judgments’, Journal of Personality and Social Psychology, vol. 92, no. 2, pp.191-207.

Gentzkow, M, Shapiro, J M, and Sinkinson, M 2011 ‘The Effect of Newspaper Entry and Exit on Electoral Politics’. American Economic Review. Vol. 101. No. 7. pp.2980-3018.

Giddens, A 1994 Beyond Left and Right: The Future of Radical Politics. Cambridge: Polity Press.

Gilbert, N 2016 ‘Research, Theory and Method’. In Gilbert, N, and Stoneman, P (eds.) Researching Social Life. 4th edition. London: SAGE Publications Ltd.

Goffman, E 1974 Frame Analysis: An Essay on the Organization of Experience. Cambridge: Harvard University Press.

Golec De Zavala, A G, Cislak, A, and Wesolowska, E 2010 ‘Political Conservatism, Need for Cognitive Closure, and Intergroup Hostility’. Political Psychology. Vol. 31. No. 4. pp.521-541.

Gopnik, A, Meltzoff, A N, and Kuhl, P K 2000 The Scientist in the Crib: What Early Learning Tells Us About the Mind. New York City: William Morrow Paperbacks.

Goren, P, Schoen, H, Reifler, J, Scotto, T, and Chittick, W 2016 ‘A unified theory of value-based reasoning and U.S. public opinion’. Political Behaviour. Vol. 38. No. 4. pp.977-997.

Gough, I 2001 ‘Social assistance regimes: a cluster analysis’. Research Note. Journal of European Social Policy. Vol. 11. No. 2. pp.165-170.

Graham, J, Haidt, J, and Nosek, B A 2009 ‘Liberals and Conservatives Rely on Different Sets of Moral Foundations’, Journal of Personality and Social Psychology, vol. 96, no. 5, pp.1029-1046.

Green, E G T, Fasel, N, and Sarrasin, O 2010 ‘The More the Merrier? The Effects of Type of Cultural Diversity on Exclusionary Immigration Attitudes in Switzerland’. International Journal of Conflict and Violence. Vol. 4. No. 2. pp.177-190.

Greenberg, J, and Jonas, E 2003 ‘Psychological Motives and Political Orientation—The Left, the Right, and the Rigid: Comment on Jost et al. (2003)’. Psychological Bulletin. Vol. 129. No. 3. pp.376-382. 36

Gross Stein, J, 2013 ‘Threat Perception in International Relations’. In Huddy, L, Sears, D O, and Levy, J S (eds.), The Oxford Handbook of Political Psychology. Oxford: Oxford University Press.

Gujarati, D N, and Porter, D C 2008 Basic Econometrics. 5th edition. New York City: McGraw Hill Education.

Haas, I J, and Cunningham, W A 2014 ‘The Uncertainty Paradox: Perceived Threat Moderates the Impact of Uncertainty on Political Tolerance’. Faculty Publications: Political Science. University of Nebraska-Lincoln. Available online: http://digitalcommons.unl.edu/poliscifacpub/59 [Accessed 17/08/2017].

Harding, J, and Pribram, E D 2002 ‘The power of feeling: Locating emotions in culture’. European Journal of Cultural Studies. Vol. 5. Issue 4. pp.407-426.

Harman, J 2017 ‘The Psychological Spectrum: Political orientation and its origins in perception and culture’. Undergraduate Journal of Politics and International Relations. Forthcoming.

Hatemi, P K, Funk, C L, Medland, S E, Maes, H M, Silberg, J L, Martin, N G, and Eaves, L J 2009 ‘Genetic and Environmental Transmission of Political Attitudes Over a Life Time’. The Journal of Politics. Vol. 71. No. 3. pp.1141-1156.

Hayes, A F 2013 Introduction to Mediation, Moderation and Conditional Process Analysis:

Hayes, A F, and Preacher, K J 2013 ‘Conditional Process Modeling: Using Structural Equation Modeling to Examine Contingent Causal Processes’. In Hancock, G R, and Mueller, R O (eds.) Structural Equation Modeling: A Second Course (2nd edition). Greenwich: Information Age Publishing.

Hellevik, O 2002 ‘Age Differences in Value Orientation—Life Cycle or Cohort Effects?’. International Journal of Public Opinion Research. Vol. 14. No. 3. pp.286-302.

Henrich, J, Heine, S J, and Norenzayan, A 2010 ‘The weirdest people in the world?’. Working Paper Series des Rates für Sozial- und Wirtschaftsdaten. no. 139.

Herrmann, R K 2013 ‘Perceptions and Image Theory in International Relations’. In Huddy, L, Sears, D O, and Levy, J S (eds.), The Oxford Handbook of Political Psychology. Oxford: Oxford University Press.

Heywood, A 2012 Political Ideologies: An Introduction. London: Palgrave Macmillan.

Hodson, G, Choma, B L, Boisvert, J, Hafer, C L, MacInnis, C C, and Costello, K 2013 ‘The role of intergroup disgust in predicting negative outgroup evaluations’. Journal of Experimental Social Psychology. Vol. 49. No. 2. pp.195-205.

Hoffman, D D 2000 Visual Intelligence: How We Create What We See. New York City: W. W. Norton and Company.

Huckfeldt, R, Mondak, J J, Hayes, M, Pietryka, M T, and Reilly, J 2013 ‘Networks, Interdependence, and Social Influence in Politics’. In Huddy, L, Sears, D O, and Levy, J S (eds.), The Oxford Handbook of Political Psychology. Oxford: Oxford University Press.

Huddy, L 2013 ‘From Group Identity to Political Cohesion and Commitment’. In Huddy, L, Sears, D O, and Levy, J S (eds.), The Oxford Handbook of Political Psychology. Oxford: Oxford University Press.

IBM Corp. Released 2016. IBM SPSS Statistics for Windows, Version 24.0. Armonk, NY: IBM Corp.

Inbar, Y, Pizarro, D A, and Bloom, P 2008 ‘Conservatives are more easily disgusted than liberals’. Cognition and Emotion. Vol. 23. No. 4. pp.714-725.

Jervis, L L 2001 ‘The Pollution of Incontinence and the Dirty Work of Caregiving in a U. S. Nursing Home’. Medical Anthropology Quarterly. Vol. 15. No. 1. pp.84-99.

Johnston, C D, Lavine, H G, and Federico, C M 2017 Open versus Closed: Personality, Identity, and the Politics of Redistribution. Cambridge: Cambridge University Press.

Jost, J T 2017 ‘Ideological Asymmetries and the Essence of Political Psychology’. Political Psychology. Vol. 38. No. 2. pp.167- 208.

Jost, J T, Glaser, J, Kruglanski, A W, and Sulloway, F J 2003 ‘Political Conservatism as Motivated Social Cognition’. Psychological Bulletin. Vol. 129. No. 3.

Kandler, C, Bleidorn, W, and Riemann, R 2012 ‘Left or Right? Sources of Political Orientation: The Roles of Genetic Factors, Cultural Transmission, Assortative Mating, and Personality’. Journal of Personality and Social Psychology. Vol. 102. No. 3. pp.633-645.

Keesing, R M 2005 ‘Anthropology as interpretive quest’. In Moore, H.L. and Sanders, T. (ed.), Anthropology in Theory: Issues in Epistemology. Hoboken: John Wiley and Sons.

Khare, R S 2003 ‘Pollution and purity’. In Barnard, A. and Spencer, J. (eds.), Encyclopedia of Social and Cultural Anthropology. Abingdon-on-Thames: Routledge World Reference.

Kossowska, M, and Van Hiel, A 2003 ‘The Relationship Between Need for Closure and Conservative Beliefs in Western and Eastern Europe’. Political Psychology. Vol. 24, No. 3, pp.501-518.

Kruskal, J B 1964 ‘MULTIDIMENSIONAL SCALING BY OPTIMIZING GOODNESS OF FIT TO A NONMETRIC HYPOTHESIS’. Psychometrika. Vol. 29. No. 1. pp.1-27.

Ksiazkewicz, A, Ludeke, S, and Krueger, R 2016 ‘The Role of Cognitive Style in the Link Between Genes and Political Ideology’. Political Psychology. Vol. 37. No. 6. pp.761-776.

Levi-Strauss, C 1974 Structural Anthropology. Revised edition. New York City: Basic Books.

Lieberman, M D, Hariri, A, Jarcho, J M, Eisenberger, N I, and Bookheimer, S Y 2005 ‘An fMRI investigation of race-related amygdala activity in African-American and Caucasian-American individuals’, Nature Neuroscience, vol. 8, no. 6, pp. 720-722.

Lodge, M, McGraw, K M, and Stroh, P 1989 ‘An Impression-Driven Model of Candidate Evaluation’. The American Political Science Review. Vol. 83. No. 2. pp.399-419.

Lundberg, K B, Payne, B K, Pasek, J, and Krosnick, J A 2015 ‘Racial Attitudes Predicted Changes in Ostensibly Race-Neutral Political Attitudes under the Obama Administration’. Political Psychology. Vol. 38. No. 2. pp.313-330.

Lyons, J 2016 ‘The Family and Partisan Socialization in Red and Blue America’. Political Psychology. Vol. 38. No. 2. pp.297-312.

Malka, A, and Lelkes, Y 2010 ‘More than Ideology: Conservative–Liberal Identity and Receptivity to Political Cues’. Social Justice Research. Vol. 23. No. 2-3. pp.156-188.

Matthews, M, Levin, S, and Sidanius, J 2009 ‘A Longitudinal Test of the Model of Political Conservatism as Motivated Social Cognition’. Political Psychology, vol. 30, no. 6, pp.921-936.

Meadows, R, and Arber, S 2015 ‘Marital Status, Relationship Distress, and Self-rated Health: What Role for “Sleep Problems”?’. Journal of Health and Social Behaviour. Vol. 56. No. 3. pp.341-355.

Mitchell, B P 2006 Eight Ways to Run the Country: A New and Revealing Look at Left and Right. Santa Barbara: Praeger.

Morimoto, C H, and Mimica, M R M 2004 ‘Eye gaze tracking techniques for interactive applications’. Computer Vision and Image Understanding. Vol. 98. pp.4-24.

Nisbett, R E 2004 The Geography of Thought: How Asians and Westerners Think Differently... and Why. New York City: Free Press.

Nussbaum, M C 2015 Hiding from Humanity: Disgust, Shame, and the Law. Princeton: Princeton University Press.

Olson, J M, Vernon, P A, Harris, J A, and Jang, K L 2001 ‘The Heritability of Attitudes: A Study of Twins’. Journal of Personality and Social Psychology. Vol. 80. No. 6. pp.845-860.

Parsons, W 1995 Public Policy: An Introduction to the Theory and Practice of Policy Analysis. Cheltenham: Edward Elgar Publishing Limited.

Piurko, Y, Schwartz, S H, and Davidov, E 2011 ‘Basic Personal Values and the Meaning of Left-Right Political Orientations in 20 Countries’. Political Psychology. Vol. 32. No. 4. pp.537-561.

Putnam, M 2005 ‘Conceptualizing Disability: Developing a Framework for Political Disability Identity’. Journal of Disability Policy Studies. Vol. 16. No. 3. pp.188-198.

Ramsay, J E, and Pang, J S 2015 ‘Anti-Immigrant Prejudice in Rising East Asia: A Stereotype Content and Integrated Threat Analysis’. Political Psychology

Roets, A, and Van Hiel, A 2011 ‘Item selection and validation of a brief, 15-item version of the Need for Closure Scale’. Personality and Individual Differences. Vol. 50. pp.90-94.

Rokeach, M 2000 ‘From Individual to Institutional Values’, in Rokeach, M (ed.) Understanding Human Values. New York: Simon & Schuster, Inc.

Ross, M H 1997 ‘The Relevance of Culture for the Study of Political Psychology and Ethnic Conflict’. Political Psychology. Vol. 18. No. 2. pp.299-326. 38

Schreiber, D, and Iacoboni, M 2012 ‘Huxtables on the Brain: An fMRI Study of Race and Norm Violation’. Political Psychology. Vol. 33. No. 3. pp.313-330.

Schreiber, D, Fonzo, G, Simmons, A N, Dawes, C T, Flagan, T, Fowler, J H, and Paulus, M P 2013 ‘Red Brain, Blue Brain: Evaluative Processes Differ in Democrats and Republicans’. PLoS ONE. Vol. 8. No. 2.

Schwartz, S H 2007 ‘A Theory of Cultural Value Orientations: Explication And Applications’. In Esmer, Y, and Pettersson, T (eds.) Measuring and Mapping Cultures: 25 Years of Comparative Value Surveys. International Studies in Sociology and Social Anthropology. Vol. 104. pp.33-78.

Scotto, T 2016 ‘Measuring Attitudes’. In Gilbert, N, and Stoneman, P (eds.) Researching Social Life. 4th edition. London: SAGE Publications Ltd.

Sidanius, J 1993 ‘The Psychology of Group Conflict and the Dynamics of Oppression: A Social Dominance Perspective’. In Iyengar, S, and McGuire, W J (eds.) Explorations in Political Psychology. Durham: Duke University Press.

Sidanius, J, Levin, S, Liu, J, and Pratto, F 2000 ‘Social dominance orientation, anti-egalitarianism and the political psychology of gender: an extension and cross-cultural replication’. European Journal of Social Psychology. Vol. 30. No. 1. pp.41-67.

Siedlecki, P, Szwabiński, J, and Weron, T 2016 ‘The Interplay Between Conformity and Anticonformity and its Polarizing Effect on Society’. Journal of Artificial Societies and Social Simulation. Vol. 19. No. 4.

Simon, H A 1956 ‘Rational Choice and the Structure of the Environment’. Psychological Review. Vol. 63. No. 2. Pp.129-138.

Smaje, C 2000 Natural Hierarchies: The Historical Sociology of Race and Caste. Hoboken: Wiley-Blackwell.

Soares, A M, Farhangmehr, M, and Shoham, A 2007 ‘Hofstede's dimensions of culture in international marketing studies’. Journal of Business Research. Vol. 60. pp.277-284.

Sperber, D 2005 ‘Anthropology and psychology: towards an epidemiology of representations’, in Moore, H L, and Sanders, T (eds.) Anthropology in Theory: Issues in Epistemology. Hoboken: John Wiley and Sons.

Stanovich, K E, and West, R F 1997 ‘Reasoning independently of prior belief and individual differences in actively open-minded thinking’. Journal of Educational Psychology. Vol. 89. No. 2. pp.342-357.

Stoneman, P, and Brunton-Smith, I 2016 ‘Quantitative Research’. In Gilbert, N, and Stoneman, P (eds.) Researching Social Life. 4th edition. London: SAGE Publications Ltd.

Sturgis, P 2016 ‘Designing and Collecting Survey Samples’. In Gilbert, N, and Stoneman, P (eds.) Researching Social Life. 4th edition. London: SAGE Publications Ltd.

Stürmer, S, Rohmann, A, Mazziotta, A, and Siem, B 2016 ‘Fear of Infection or Justification of Social Exclusion? The Symbolic Exploitation of the Ebola Epidemic’. Political Psychology. Vol. 3. No. 3. pp.499-513.

Taber, C S, and Young, E 2013 ‘Political Information Processing’. In Huddy, L, Sears, D O, and Levy, J S (eds.), The Oxford Handbook of Political Psychology. Oxford: Oxford University Press.

Tamimi, N 1998 ‘A second‐order factor analysis of critical TQM factors’, International Journal of Quality Science, Vol. 3. No, 1. Pp.71-79.

Tetlock, P E 1984 ‘Cognitive Style and Political Belief Systems in the British House of Commons’. Journal of Personality and Social Psychology. Vol. 46. No. 2. pp.365-375.

Thórisdóttir, H, and Jost, J T 2011 ‘Motivated Closed-Mindedness Mediates the Effect of Threat on Political Conservatism’. Political Psychology. Vol. 32. No. 5. pp.785-811.

Toren, C 2003 ‘Culture and personality’, in Barnard, A, and Spencer, J (eds.) Encyclopedia of Social and Cultural Anthropology. Abingdon-on-Thames: Routledge World Reference.

Tritt, S M, Inzlicht, M, and Peterson, J B 2013 ‘Preliminary support for a generalized arousal model of political conservatism’. PLoS ONE. Vol. 8. No. 12.

Van Bavel, J J, Mende-Siedlecki, P, Brady, W J, and Reinero, D A 2016 ‘Contextual sensitivity in scientific reproducibility’. Proceedings of the National Academy of Sciences of the United States of America. Vol. 113. No. 23. pp.6454-6459.

Van Hiel, A, Duriez, B, and Kossowska, M 2006 ‘The Presence of Left-Wing Authoritarianism in Western Europe and Its Relationship with Conservative Ideology’. Political Psychology. Vol. 27. No. 5. pp.769-793.

Victoroff, J 2005 ‘The Mind of the Terrorist: A Review and Critique of Psychological Approaches’. The Journal of Conflict Resolution. Vol. 49. No. 1. pp.3-42.

Vigil, J M 2010 ‘Political leanings vary with facial expression processing and psychosocial functioning’. Group Processes and Intergroup Relations. Vol. 13. No. 5. pp.547-558.

Wagner, M W, Deppe, K, Jacobs, C M, and Hibbing, J R 2014 ‘Beyond Survey Self-Reports: Using Physiology to Tap Political Orientations’. International Journal of Public Opinion Research. Vol. 27. No. 3. pp.303-317.

Webster, D M, and Kruglanski, A W 1994 ‘Individual Differences in Need for Cognitive Closure’. Journal of Personality and Social Psychology. Vol. 67. No. 6. pp.1049-1062.

West, R F, Toplak, M E, and Stanovich, K E 2008 ‘Heuristics and Biases as Measures of Critical Thinking: Associations with Cognitive Ability and Thinking Dispositions’.

Xu, X, and Peterson, J B 2015 ‘Differences in Media Preference Mediate the Link Between Personality and Political Orientation’. Political Psychology. Vol. 38. No. 1. pp.55-72.

Yilmaz, O, and Saribay, S A 2016 ‘An attempt to clarify the link between cognitive style and political ideology: A non-western replication and extension’. Judgment and Decision Making. Vol. 11. No. 3. pp.287-300.

Zaller, J R 1992 The Nature and Origins of Mass Opinion. Cambridge: Cambridge University Press.

Technical Appendix Section 1: question wording, coding and recoding of selected variables

Variable name(s) Question wording Original Response labels Recoding coding Country Respondent country- coded, not 1 England Reference asked category [theme: culture] 2 Scotland Scot = 1 3 Wales Wales = 1 turnoutUKGeneralW1 “Many people don't vote in 1 Very unlikely that I would vote Same turnoutUKGeneralW2 elections these days. If there were turnoutUKGeneralW3 a UK General Election tomorrow, turnoutUKGeneralW4 how likely is it that you would vote?” [theme: engagement] 2 Fairly unlikely Same 3 Neither likely nor unlikely Same 4 Fairly likely Same 5 Very likely that I would vote Same 9999 Don’t know MISSING polAttentionW1 “How much attention do you 0 Pay no attention Same polAttentionW2 generally pay to politics?” polAttentionW3 polAttentionW4 polAttentionW6 polAttentionW7 polAttentionW8 [theme: engagement] 1 1 Same 2 2 Same 3 3 Same 4 4 Same 5 5 Same 6 6 Same 7 7 Same 8 8 Same 9 9 Same 10 Pay a great deal of attention Same 9999 Don’t know MISSING trustMPsW1 “How much trust do you have in 1 No trust Same trustMPsW2 Members of Parliament in trustMPsW3 general?” trustMPsW4 trustMPsW6 trustMPsW7 trustMPsW9 [theme: general politics] 2 2 Same 3 3 Same 4 4 Same 5 5 Same 6 6 Same 7 A great deal of trust Same 9999 Don’t know MISSING likeCameronW1 “How much do you like or dislike 0 Strongly dislike Same likeCameronW2 each of the following party likeCameronW3 leaders? [David Cameron]” likeCameronW4 likeCameronW5 likeCameronW6 likeCameronW7 likeCameronW8 likeCameronW9 [theme: partisanship] 1 1 Same 2 2 Same 3 3 Same 4 4 Same 5 5 Same 6 6 Same 7 7 Same 8 8 Same 9 9 Same 10 Strongly like Same 9999 Don’t know MISSING econPersonalRetroW1 “How does the financial situation 1 Got a lot worse Same econPersonalRetroW2 of your household now compare econPersonalRetroW3 with what it was 12 months ago? econPersonalRetroW4 Has it:” econPersonalRetroW7

[theme: general politics] 2 Got a little worse Same 3 Stayed the same Same 4 Got a little better Same 5 Got a lot better Same 9999 Don’t know MISSING likeConW7 “How much do you like or dislike 0 Strongly dislike Same likeConW8 each of the following parties? likeConW9 [Conservatives]” [theme: partisanship] 1 1 Same 2 2 Same 3 3 Same 4 4 Same 5 5 Same 6 6 Same 7 7 Same 8 8 Same 9 9 Same 10 Strongly like Same

9999 Don’t know MISSING

ptvConW9 “How likely is it that you would 0 Very unlikely Same ever vote for each of the following parties? [Conservatives]” [theme: partisanship] 1 1 Same 2 2 Same 3 3 Same 4 4 Same 5 5 Same 6 6 Same 7 7 Same 8 8 Same 9 9 Same 10 Very likely Same 9999 Don’t know MISSING immigEconW1 “Do you think immigration is good 1 Bad for economy 1 immigEconW2 or bad for Britain’s economy?” immigEconW3 immigEconW4 [theme: policies] 2 2 1.67 3 3 2.33 4 4 3 5 5 3.67 6 6 4.33 7 Good for economy 5 9999 Don’t know MISSING immigCulturalW1 “And do you think that 1 Undermines cultural life 7 immigCulturalW2 immigration undermines or immigCulturalW3 enriches Britain’s cultural life?” immigCulturalW4 immigCulturalW6 immigCulturalW8 [theme: culture] 2 2 6 3 3 5 4 4 4 5 5 3 6 6 2 7 Enriches cultural life 1 9999 Don’t know MISSING britishnessW1 “Where would you place yourself 1 Not at all British Same britishnessW2 on these scales? [Britishness]” britishnessW3 britishnessW4 britishnessW7 britishnessW8 britishnessW9 [theme: culture] 2 2 Same 3 3 Same 4 4 Same 5 5 Same 6 6 Same 7 Very strongly British Same 9999 Don’t know MISSING

reasonForUnemploymentW1 “How much do you agree or 1 Strongly disagree Same reasonForUnemploymentW2 disagree with the following reasonForUnemploymentW3 statements? [When someone is reasonForUnemploymentW4 unemployed, it’s usually through reasonForUnemploymentW7 no fault of their own]” [theme: policies] 2 Disagree Same 3 Neither agree nor disagree Same 4 Agree Same 5 Strongly agree Same 9999 Don’t know MISSING immigrantsWelfareStateW1 “How much do you agree or 1 Strongly disagree 5 immigrantsWelfareStateW2 disagree with the following immigrantsWelfareStateW3 statements? [Immigrants are a immigrantsWelfareStateW4 burden on the welfare state]” immigrantsWelfareStateW7 [theme: policies] 2 Disagree 4 3 Neither agree nor disagree 3 4 Agree 2 5 Strongly agree 1ss 9999 Don’t know MISSING govtHandoutsW1 “How much do you agree or 1 Strongly disagree 5 govtHandoutsW2 disagree with the following govtHandoutsW3 statements? [Too many people govtHandoutsW4 these days like to rely on govtHandoutsW7 government handouts]”

[theme: policies] 2 Disagree 4 3 Neither agree nor disagree 3 4 Agree 2 5 Strongly agree 1 9999 Don’t know MISSING businessBonusW1 “How much do you agree or 1 Strongly disagree 5 businessBonusW2 disagree with the following businessBonusW3 statements? [In business, bonuses businessBonusW4 are a fair way to reward hard businessBonusW7 work]”

[theme: policies] 2 Disagree 4 3 Neither agree nor disagree 3 4 Agree 2 5 Strongly agree 1 9999 Don’t know MISSING countryOfBirth “Where were you born?” 1 England [theme: culture] 2 Scotland 3 Wales 4 Northern Ireland 5 Republic of Ireland 6 Commonwealth member country 7 European Union member country 8 Rest of world 998 Skipped 999 Not asked 9999 Prefer not to answer MISSING discussPolDaysW2 “During the last week, roughly on 0 0 days Same discussPolDaysW4 how many days did you talk about discussPolDaysW5 politics with other people?” discussPolDaysW6 discussPolDaysW7

[theme: engagement] 1 1 day Same 2 2 days Same 3 3 days Same 4 4 days Same 5 5 days Same 6 6 days Same 7 7 days Same 9999 Don’t know MISSING immigrationLevelW4 “Do you think the number of 1 Decreased a lot Same immigrationLevelW6 immigrants from foreign countries who are permitted to come to the

United Kingdom to live should be increased, decreased, or left the same as it is now?” [theme: policies] 2 Decreased a little Same 3 Left the same as it is now Same 4 Increased a little Same 5 Increased a lot Same 99 Don’t know MISSING 998 Skipped MISSING 999 Not asked MISSING 9999 Don’t know MISSING conAngryW4 “Now we would like to know 0 No Same conAngryW6 something about the feelings you have towards each of the parties. Which of these emotions do you feel about each of the parties? Tick all that apply [Conservatives] [Angry]” [theme: psychology] 1 Yes Same 9999 Don’t know MISSING labAngryW4 “Now we would like to know 0 No Same labAngryW6 something about the feelings you have towards each of the parties. Which of these emotions do you feel about each of the parties? Tick all that apply [Labour] [Angry]” [theme: psychology] 1 Yes Same 9999 Don’t know MISSING ldAngryW4 “Now we would like to know 0 No Same ldAngryW6 something about the feelings you have towards each of the parties. Which of these emotions do you feel about each of the parties? Tick all that apply [Liberal Democrats] [Angry]” [theme: psychology] 1 Yes Same 9999 Don’t know MISSING ukipAngryW4 “Now we would like to know 0 No Same ukipAngryW6 something about the feelings you have towards each of the parties. Which of these emotions do you feel about each of the parties? Tick all that apply [UKIP] [Angry]” [theme: psychology] 1 Yes Same 9999 Don’t know MISSING grnAngryW4 “Now we would like to know 0 No Same grnAngryW6 something about the feelings you have towards each of the parties. Which of these emotions do you feel about each of the parties? Tick all that apply [Green] [Angry]” [theme: psychology] 1 Yes Same 9999 Don’t know MISSING conFearW4 “Now we would like to know 0 No Same conFearW6 something about the feelings you have towards each of the parties. Which of these emotions do you feel about each of the parties? Tick all that apply [Conservatives] [Afraid]” [theme: psychology] 1 Yes Same 9999 Don’t know MISSING labFearW4 “Now we would like to know 0 No Same labFearW6 something about the feelings you have towards each of the parties. Which of these emotions do you feel about each of the parties? Tick all that apply [Labour] [Afraid]” [theme: psychology] 1 Yes Same 9999 Don’t know MISSING

ldFearW4 “Now we would like to know 0 No Same ldFearW6 something about the feelings you have towards each of the parties. Which of these emotions do you feel about each of the parties? Tick all that apply [Liberal Democrats] [Afraid]” [theme: psychology] 1 Yes Same 9999 Don’t know MISSING ukipFearW4 “Now we would like to know 0 No Same ukipFearW6 something about the feelings you have towards each of the parties. Which of these emotions do you feel about each of the parties? Tick all that apply [UKIP] [Afraid]” [theme: psychology] 1 Yes Same 9999 Don’t know MISSING grnFearW4 “Now we would like to know 0 No Same grnFearW6 something about the feelings you have towards each of the parties. Which of these emotions do you feel about each of the parties? Tick all that apply [Green] [Afraid]” [theme: psychology] 1 Yes Same 9999 Don’t know MISSING electionInterestW4 “How interested are you in the 1 Not at all interested Same electionInterestW5 General Election that will be held on May 7th this year?” [theme: engagement] 2 Not very interested Same 3 Somewhat interested Same 4 Very interested Same 9999 Don’t know MISSING electionInterestW6 “How interested were you in the 1 Not at all interested Same General Election that was held on May 7th this year?” [theme: engagement] 2 Not very interested Same 3 Somewhat interested Same 4 Very interested Same 9999 Don’t know MISSING infoSourceTVW4 “During the last seven days, on 1 None, no time at all Same infoSourceTVW5 average how much time (if any) infoSourceTVW6 have you spent per day following infoSourceTVW7 news about politics or current infoSourceTVW8 affairs from each of these sources? [Television]” [theme: engagement] 2 Less than half an hour Same 3 Half an hour to an hour Same 4 One to two hours Same 5 More than two hours Same 9999 Don’t know MISSING infoSourcePaperW4 “During the last seven days, on 1 None, no time at all Same infoSourcePaperW5 average how much time (if any) infoSourcePaperW6 have you spent per day following infoSourcePaperW7 news about politics or current infoSourcePaperW8 affairs from each of these sources? [Newspaper (including online)]” [theme: engagement] 2 Less than half an hour Same 3 Half an hour to an hour Same 4 One to two hours Same 5 More than two hours Same 9999 Don’t know MISSING infoSourceRadioW4 “During the last seven days, on 1 None, no time at all Same infoSourceRadioW5 average how much time (if any) infoSourceRadioW6 have you spent per day following infoSourceRadioW7 news about politics or current infoSourceRadioW8 affairs from each of these sources? [Radio]” [theme: engagement] 2 Less than half an hour Same 3 Half an hour to an hour Same

4 One to two hours Same 5 More than two hours Same 9999 Don’t know MISSING infoSourceInternetW4 “During the last seven days, on 1 None, no time at all Same infoSourceInternetW5 average how much time (if any) infoSourceInternetW6 have you spent per day following infoSourceInternetW7 news about politics or current infoSourceInternetW8 affairs from each of these sources? [Internet (not including online newspapers)]” [theme: engagement] 2 Less than half an hour Same 3 Half an hour to an hour Same 4 One to two hours Same 5 More than two hours Same 9999 Don’t know MISSING infoSourcePeopleW4 “During the last seven days, on 1 None, no time at all Same infoSourcePeopleW5 average how much time (if any) infoSourcePeopleW6 have you spent per day following infoSourcePeopleW7 news about politics or current infoSourcePeopleW8 affairs from each of these sources? [Talking to other people]” [theme: engagement] 2 Less than half an hour Same 3 Half an hour to an hour Same 4 One to two hours Same 5 More than two hours Same 9999 Don’t know MISSING conGovTrustW5 “How much would you expect 1 Would do a bad job Same each of the following political parties to do a good job or a bad job if they are in government after the General Election (either by themselves or as part of a coalition)? [Conservatives]” [theme: partisanship] 2 2 Same 3 3 Same 4 4 Same 5 5 Same 6 6 Same 7 Would do a good job Same 9999 Don’t know MISSING lr1W6 “How much do you agree or 1 Strongly disagree Same disagree with the following statements? [Government should redistribute income from the better off to those who are less well off]” [theme: policies] 2 Disagree Same 3 Neither agree nor disagree Same 4 Agree Same 5 Strongly agree Same 9999 Don’t know MISSING al1W6 “How much do you agree or 1 Strongly disagree Same disagree with the following statements? [Young people today don’t have enough respect for traditional British values]” [theme: culture] 2 Disagree Same 3 Neither agree nor disagree Same 4 Agree Same 5 Strongly agree Same 9999 Don’t know MISSING al3W6 “How much do you agree or 1 Strongly disagree 5 disagree with the following statements? [Schools should teach children to obey authority]” [theme: policies] 2 Disagree 4 3 Neither agree nor disagree 3 4 Agree 2 5 Strongly agree 1 9999 Don’t know MISSING

47 al4W6 “How much do you agree or 1 Strongly disagree 5 disagree with the following statements? [Censorship of films and magazines is necessary to uphold moral standards]” [theme: policies] 2 Disagree 4 3 Neither agree nor disagree 3 4 Agree 2 5 Strongly agree 1 9999 Don’t know MISSING al5W6 “How much do you agree or 1 Strongly disagree 5 disagree with the following statements? [People who break the law should be given stiffer sentences]” [theme: policies] 2 Disagree 4 3 Neither agree nor disagree 3 4 Agree 2 5 Strongly agree 1 9999 Don’t know MISSING disabilityW6 “Do you have any long-term 0 No Same illness, health problem or disability which limits your daily activities or the work you can do? Include problems which are due to old age” [theme: demographics] 1 Yes Same 9999 Don’t know MISSING euRefTurnoutW7 “Many people don’t vote in 1 Very unlikely that I would vote Same elections these days. How likely is it that you will vote in the referendum on Britain’s membership of the European Union on June 23rd?” [theme: engagement] 2 Fairly unlikely Same 3 Neither likely nor unlikely Same 4 Fairly likely Same 5 Very likely that I would vote Same 9999 Don’t know MISSING euRefInterestW7 “How interested are you in the EU 1 Not at all interested Same referendum that will be held on June 23rd?” [theme: engagement] 2 Not very interested Same 3 Somewhat interested Same 4 Very interested Same 9999 Don’t know MISSING ethno1W7 “How much do you agree or 1 Strongly disagree 5 disagree with the following statements? [Britain has a lot to learn from other countries in running its affairs]” [theme: culture] 2 Disagree 4 3 Neither agree nor disagree 3 4 Agree 2 5 Strongly agree 1 9999 Don’t know MISSING ethno2W7 “How much do you agree or 1 Strongly disagree Same disagree with the following statements? [I would rather be a citizen of Britain than of any other country in the world]” [theme: culture] 2 Disagree Same 3 Neither agree nor disagree Same 4 Agree Same 5 Strongly agree Same 9999 Don’t know MISSING ethno3W7 “How much do you agree or 1 Strongly disagree 5 disagree with the following statements? [There are some things about Britain today that make me ashamed to be British]”

[theme: culture] 2 Disagree 4 3 Neither agree nor disagree 3 4 Agree 2 5 Strongly agree 1 9999 Don’t know MISSING ethno4W7 “How much do you agree or 1 Strongly disagree Same disagree with the following statements? [People in Britain are too ready to criticise their country]” [theme: culture] 2 Disagree Same 3 Neither agree nor disagree Same 4 Agree Same 5 Strongly agree Same 9999 Don’t know MISSING ethno5W7 “How much do you agree or 1 Strongly disagree Same disagree with the following statements? [The world would be a better place if people from other countries were more like the British]” [theme: culture] 2 Disagree Same 3 Neither agree nor disagree Same 4 Agree Same 5 Strongly agree Same 9999 Don’t know MISSING ethno6W7 “How much do you agree or 1 Strongly disagree 5 disagree with the following statements? [I am often less proud of Britain than I would like to be]” [theme: culture] 2 Disagree 4 3 Neither agree nor disagree 3 4 Agree 2 5 Strongly agree 1 9999 Don’t know MISSING radicalW7 “How much do you agree or 1 Strongly disagree 5 disagree with the following statements? [We need to fundamentally change the way society works in Britain]” [theme: culture] 2 Disagree 4 3 Neither agree nor disagree 3 4 Agree 2 5 Strongly agree 1 9999 Don’t know MISSING riskTakingW7 “Generally speaking, how willing 1 Very unwilling to take risks Same risktakingW8 are you to take risks?” [theme: psychology] 2 Somewhat unwilling to take risks Same 3 Somewhat willing to take risks Same 4 Very willing to take risks Same 9999 Don’t know MISSING aom1W7 “Do you agree or disagree with 1 Strongly disagree Same the following statements? [Allowing oneself to be convinced by an opposing argument is a sign of good character]” [theme: psychology] 2 Disagree Same 3 Neither agree nor disagree Same 4 Agree Same 5 Strongly agree Same 9999 Don’t know MISSING aom2W7 “Do you agree or disagree with 1 Strongly disagree Same the following statements? [People should take into consideration evidence that goes against their beliefs]” [theme: psychology] 2 Disagree Same 3 Neither agree nor disagree Same 4 Agree Same 5 Strongly agree Same

9999 Don’t know MISSING aom3W7 “Do you agree or disagree with 1 Strongly disagree Same the following statements? [People should revise their beliefs in response to new information or evidence]” [theme: psychology] 2 Disagree Same 3 Neither agree nor disagree Same 4 Agree Same 5 Strongly agree Same 9999 Don’t know MISSING aom4W7 “Do you agree or disagree with 1 Strongly disagree 5 the following statements? [Changing your mind is a sign of weakness]” [theme: psychology] 2 Disagree 4 3 Neither agree nor disagree 3 4 Agree 2 5 Strongly agree 1 9999 Don’t know MISSING aom5W7 “Do you agree or disagree with 1 Strongly disagree 5 the following statements? [Changing your mind is a sign of weakness]” [theme: psychology] 2 Disagree 4 3 Neither agree nor disagree 3 4 Agree 2 5 Strongly agree 1 9999 Don’t know MISSING aom6W7 “Do you agree or disagree with 1 Strongly disagree 5 the following statements? [It is important to persevere in your beliefs even when evidence is brought to bear against them]” [theme: psychology] 2 Disagree 4 3 Neither agree nor disagree 3 4 Agree 2 5 Strongly agree 1 9999 Don’t know MISSING aom7W7 “Do you agree or disagree with 1 Strongly disagree 5 the following statements? [One should disregard evidence that conflicts with one's established beliefs]” [theme: psychology] 2 Disagree 4 3 Neither agree nor disagree 3 4 Agree 2 5 Strongly agree 1 9999 Don’t know MISSING anyUniW7 “Have you ever attended a 0 No, I have never attended higher 0 University or other higher education education institution?” [theme: demographics] 1 Yes, I am currently enrolled in higher 1 education 2 Yes, but I didn’t complete higher 1 education 3 Yes, I graduated from higher 1 education 9999 Don’t know MISSING ageW7 “What is your age?” [asked April- Any whole no. Age last birthday shown by whole Same May 2016] no. [theme: demographics] -9 Not asked MISSING -8 Skipped MISSING profile_work_typeW7 [work type as recorded 1 Professional or higher technical Upskilled independently by YouGov] work / higher managerial - work that dummy = 1 requires at least degree-level qualifications [theme: demographics] 2 Manager or Senior Administrator / Upskilled intermediate managerial / dummy = 1 professional

3 Clerical/junior managerial/ Midskilled professional/ administrator dummy = 1 4 Sales or Services Midskilled dummy = 1 5 Foreman or Supervisor of Other Midskilled Workers dummy = 1 6 Skilled Manual Work Midskilled dummy = 1 7 Semi-Skilled or Unskilled Manual Unskilled Work dummy = 1 8 Other MISSING 9 Have never worked Reference category 98 Skipped MISSING 99 Not asked MISSING tolUncertain1W8 “To what extent do you agree or 1 Strongly disagree 5 disagree with the following statement? [I hate not knowing what the future holds]“ [theme: psychology] 2 Disagree 4 3 Neither agree nor disagree 3 4 Agree 2 5 Strongly agree 1 9999 Don’t know MISSING tolUncertain2W8 “To what extent do you agree or 1 Strongly disagree Same disagree with the following statement? [I strongly prefer to be certain about the outcome before making a decision]“ [theme: psychology] 2 Disagree Same 3 Neither agree nor disagree Same 4 Agree Same 5 Strongly agree Same 9999 Don’t know MISSING tolUncertain3W8 “To what extent do you agree or 1 Strongly disagree Same disagree with the following statement? [I hate uncertainty]“ [theme: psychology] 2 Disagree Same 3 Neither agree nor disagree Same 4 Agree Same 5 Strongly agree Same 9999 Don’t know MISSING livedAbroadW8 “Have you ever lived in a country 0 No Same other than the UK?” [theme: culture] 1 Yes Same 9999 Don’t know MISSING parentsForeignW8 “Were either of your parents born 0 No Same outside the United Kingdom?” [theme: culture] 1 Yes Same 9999 Don’t know MISSING gender [gender as recorded 1 Male 0 independently by YouGov] [theme: demographics] 2 Female 1 profile_ethnicity [Ethnicity as recorded 1 White British 1 independently by YouGov] [theme: culture] 2 Any other white background 0 3 White and Black Caribbean 0 4 White and Black African 0 5 White and Asian 0 6 Any other mixed background 0 7 Indian 0 8 Pakistani 0 9 Bangladeshi 0 10 Any other Asian background 0 11 Black Caribbean 0 12 Black African 0 13 Any other black background 0 14 Chinese 0 15 Other ethnic group 0 16 Prefer not to say MISSING

51 profile_newspaper_readership_201 [daily newspaper readership as 1 The Express Tabloid dummy recorded independently by = 1 YouGov] [theme: demographics] 2 The Daily Mail / The Scottish Daily Tabloid dummy Mail = 1 3 The Mirror / Daily Record Tabloid dummy = 1 4 The Daily Star / The Daily Star of Tabloid dummy Scotland = 1 5 The Sun Tabloid dummy = 1 6 The Daily Telegraph Broadsheet dummy = 1 7 The Financial Times Broadsheet dummy = 1 8 The Guardian Broadsheet dummy = 1 9 The Independent Broadsheet dummy = 1 10 The Times Broadsheet dummy = 1 11 The Scotsman Broadsheet dummy = 1 12 The Herald (Glasgow) Broadsheet dummy = 1 13 The Western Mail Broadsheet dummy = 1 14 Other local daily morning newspaper Reference category 15 Other Newspaper Reference category 16 None Reference category Table 2. Summary of all variables from the British Election Study selected for use in the analysis, including variable names, question wording, response options, response coding, and response recoding by the author for analytical purposes.

Section 2: annotated summary of key statistical results

Cronbach’s alphas for related items grouped into scales Cronbach’s alpha is a measure of how closely intercorrelated several separate variables are when a researcher wishes to treat them as multiple measures of the same underlying concept, and thus to group them into a scale and reduce them to one variable. The statistic varies between zero and one, with higher values indicating greater similarity between the selected variables, and thus a greater probability that they measure the same thing. Unsurprisingly, those variables which consist of the same question asked at multiple points in time are the most highly intercorrelated, and can be collated most validly into single measures.

List of variables in each proposed scale Cronbach’s alpha for scale britishnessW1

britishnessW2 0.94 britishnessW3 britishnessW4 britishnessW7 britishnessW8 britishnessW9 immigCulturalW1 immigCulturalW2 0.95 immigCulturalW3 immigCulturalW4 immigCulturalW6 immigCulturalW8 ethno1W7 ethno2W7 0.65 ethno3W7 ethno4W7 ethno5W7 ethno6W7 radicalW7 polAttentionW1

polAttentionW2 0.97 polAttentionW3 polAttentionW4 polAttentionW6 polAttentionW7 polAttentionW8 discussPolDaysW2 discussPolDaysW4 0.87 discussPolDaysW5 discussPolDaysW6 discussPolDaysW7 infoSourcePeopleW4 infoSourcePeopleW5 0.80 infoSourcePeopleW6 infoSourcePeopleW7 infoSourcePeopleW8 infoSourceInternetW4 infoSourceInternetW5 0.84 infoSourceInternetW6 infoSourceInternetW7 infoSourceInternetW8 infoSourcePaperW4 infoSourcePaperW5 0.88 infoSourcePaperW6 infoSourcePaperW7 infoSourcePaperW8 infoSourceRadioW4 infoSourceRadioW5 0.90 infoSourceRadioW6 infoSourceRadioW7 infoSourceRadioW8 infoSourceTVW4 infoSourceTVW5 0.84 infoSourceTVW6 infoSourceTVW7 infoSourceTVW8 electionInterestW4 electionInterestW5 0.88 electionInterestW6 turnoutUKGeneralW1 turnoutUKGeneralW2 0.93 turnoutUKGeneralW3 turnoutUKGeneralW4 trustMPsW1 trustMPsW2 0.93 trustMPsW3 trustMPsW4 trustMPsW6 trustMPsW7 trustMPsW9

54 econPersonalRetroW1 econPersonalRetroW2 0.85 econPersonalRetroW3 econPersonalRetroW4 econPersonalRetroW7 likeCameronW1 likeCameronW2 0.98 likeCameronW3 likeCameronW4 likeCameronW5 likeCameronW6 likeCameronW7 likeCameronW8 likeCameronW9 likeConW7 likeConW8 0.97 likeConW9 aom1W7 aom2W7 0.76 aom3W7 aom4W7 aom5W7 aom6W7 aom7W7 tolUncertain1W8 tolUncertain2W8 0.72 tolUncertain3W8 riskTakingW7 risktakingW8 conAngryW4 conAngryW6 0.59 labAngryW4 labAngryW6 ldAngryW4 ldAngryW6 ukipAngryW4 ukipAngryW6 grnAngryW4 grnAngryW6 conFearW4 conFearW6 0.49 labFearW4 labFearW6 ldFearW4 ldFearW6 ukipFearW4 ukipFearW6 grnFearW4 grnFearW6 businessBonusW1 businessBonusW2 0.84 businessBonusW3 businessBonusW4 businessBonusW7 govtHandoutsW1 govtHandoutsW2 0.91 govtHandoutsW3 govtHandoutsW4 govtHandoutsW7 immigEconW1 immigEconW2 0.94 immigEconW3 immigEconW4 immigrationLevelW4 immigrationLevelW6 0.75 immigrantsWelfareStateW1 immigrantsWelfareStateW2 0.93 immigrantsWelfareStateW3 immigrantsWelfareStateW4 immigrantsWelfareStateW7 reasonForUnemploymentW1 reasonForUnemploymentW2 0.74 reasonForUnemploymentW3 reasonForUnemploymentW4 reasonForUnemploymentW7 [6 four-point Likert scales with directions removed, as indicators of response polarisation] 0.66 [51 five-point Likert scales with directions removed, as indicators of response polarisation] 0.86

[23 seven-point Likert scales with directions removed, as indicators of response polarisation] 0.86 [20 eleven-point Likert scales with directions removed, as indicators of response polarisation] 0.93 [20 cultural conformity-related Likert scales with directions removed, as indicators of response polarisation] 0.83 [6 political, non-policy-related Likert scales with directions removed, as indicators of response polarisation] 0.86 [21 political engagement-related Likert scales with directions removed, as indicators of response polarisation] 0.86 [14 political partisanship-related Likert scales with directions removed, as indicators of response polarisation] 0.95 [27 policy preference-related Likert scales with directions removed, as indicators of response polarisation] 0.86 [12 psychology-related Likert scales with directions removed, as indicators of response polarisation] 0.75 [the complete 100 Likert scales with directions removed, as indicators of response polarisation] 0.93 Table 3. Cronbach’s alphas for related items grouped into scales.

Factor analyses for reducing numbers of variables Factor analysis takes a set of intercorrelated variables, such as any of those identified immediately above, and attempts to explain variations in those variables using a smaller number of hypothetical “factors” assumed to cause the intercorrelations between different items identified by Cronbach’s alpha. These factors can then be saved as continuous scale variables, have substantive meanings attributed to them, and used in analyses, thus condensing several different highly correlated variables into a single measure, or a smaller number of measures. Factors with an “eigenvalue” of greater than 1 are considered to represent something real which can be reliably measured. The greater the percentage of variance in the scale items which can be explained by a factor, the more that factor accurately represents the content of the items in question. In second and other higher-order factor analyses, several of which are below, one or more factors which have been generated by a previous factor analysis (for example, FAC1_1, which represents feeling of “Britishness” throughout the panel study) are themselves included as items in a subsequent factor analyses, to reduce the number of variables still further. This is useful as a technique for both acknowledging the particular validity of certain dimensions of variables- for example, the highly intercorrelated repeat questions, or economic policy items as opposed to social or immigration ones- whilst at the same time situating these dimensions as aspects of still deeper factors- for example, general cultural conformity, political engagement, or political leftness.

List of items in Eigenvalue(s) Percentage of Name of Interpretation each analysis of saved variance in items saved and label of saved factor(s) explained by saved factor(s) factor(s) factor(s) britishnessW1 britishnessW2 5.12 73.17% FAC1_1 Feeling of britishnessW3 Britishness britishnessW4 britishnessW7 britishnessW8 britishnessW9 immigCulturalW1 immigCulturalW2 4.01 80.19% FAC1_2 Immigration immigCulturalW3 undermines immigCulturalW4 immigCulturalW6 cultural life immigCulturalW8 Fac1_2 al1W6 (young people have 1.48 74.20% FAC1_14 Cultural threat no respect for traditional perception British values) ethno1W7 ethno2W7 2.24 32.04% FAC1_22 Intolerance of ethno3W7 criticism of ethno4W7 ethno5W7 British society ethno6W7 radicalW7 1.40 19.94% FAC2_22 Feelings of shame at being British FAC1_1 FAC1_14 1.77 58.94 Conform General cultural FAC1_22 conformity polAttentionW1 polAttentionW2 5.70 81.47% FAC1_3 Attention to polAttentionW3 politics polAttentionW4 polAttentionW6 polAttentionW7 polAttentionW8

56 discussPolDaysW2 discussPolDaysW4 2.77 69.30% FAC1_4 Frequency discussPolDaysW5 discussing politics discussPolDaysW6 discussPolDaysW7 infoSourcePeopleW4 infoSourcePeopleW5 2.67 53.34% FAC1_5 Time talking infoSourcePeopleW6 about politics to infoSourcePeopleW7 infoSourcePeopleW8 other people infoSourceInternetW4 infoSourceInternetW5 2.98 59.58% FAC1_6 Time following infoSourceInternetW6 politics on the infoSourceInternetW7 infoSourceInternetW8 internet infoSourcePaperW4 infoSourcePaperW5 3.26 65.11% FAC1_7 Time following infoSourcePaperW6 politics in infoSourcePaperW7 infoSourcePaperW8 newspapers infoSourceRadioW4 infoSourceRadioW5 3.52 70.35% FAC1_8 Time following infoSourceRadioW6 politics on the infoSourceRadioW7 infoSourceRadioW8 radio infoSourceTVW4 infoSourceTVW5 2.95 58.91% FAC1_9 Time following infoSourceTVW6 politics on TV infoSourceTVW7 infoSourceTVW8 electionInterestW4 electionInterestW5 2.36 78.73% FAC1_10 Interest in 2015 electionInterestW6 general election turnoutUKGeneralW1 turnoutUKGeneralW2 3.16 79.06% FAC1_12 Likelihood to turnoutUKGeneralW3 vote in 2015 GE turnoutUKGeneralW4 FAC1_3 FAC1_4 4.87 44.22% FAC1_11 Interest in FAC1_5 politics FAC1_6 FAC1_7 FAC1_8 FAC1_9 1.69 15.36% FAC2_11 Likelihood to FAC1_10 FAC1_12 vote in elections euRefInterestW7 (interest in EU referendum) euRefTurnoutW7 (likelihood to vote in EU referendum) trustMPsW1 trustMPsW2 4.30 71.73% TrustMPs Trusting MPs in trustMPsW3 general trustMPsW4 trustMPsW6 trustMPsW7 trustMPsW9 econPersonalRetroW1 econPersonalRetroW2 3.10 62.02% Econom Thinking personal econPersonalRetroW3 household econPersonalRetroW4 econPersonalRetroW7 situation has improved in year likeCameronW1 likeCameronW2 7.46 82.94% FAC1_17 Liking David likeCameronW3 Cameron likeCameronW4 likeCameronW5 likeCameronW6 likeCameronW7 likeCameronW8 likeCameronW9 likeConW7 likeConW8 2.78 92.65% FAC1_18 Liking the likeConW9 Conservatives FAC1_17 FAC1_18 2.73 90.83% ConPart Conservative Partisanship

57 ptvConW9 (probability of voting for the Conservatives) conGovTrustW5 (trusting the Conservatives to do a good job in government) aom1W7 aom2W7 2.85 40.65% Aom Active open- aom3W7 mindedness aom4W7 aom5W7 aom6W7 aom7W7 tolUncertain1W8 tolUncertain2W8 2.34 46.73% TolUncer Tolerance of tolUncertain3W8 uncertainty riskTakingW7 risktakingW8 1.33 73.26% Willrisk Risk willingness businessBonusW1 businessBonusW2 2.68 67.04% FAC1_25 Hostility toward businessBonusW3 business bonuses businessBonusW4 businessBonusW7 govtHandoutsW1 govtHandoutsW2 3.07 76.77% FAC1_26 Defensiveness of govtHandoutsW3 welfare claimants govtHandoutsW4 govtHandoutsW7 immigEconW1 immigEconW2 3.29 82.19% FAC1_27 Immigration immigEconW3 helps economy immigEconW4 immigrationLevelW4 immigrationLevelW6 1.58 79.16% FAC1_28 Would increase immigration immigrantsWelfareStateW1 immigrantsWelfareStateW2 3.82 76.35% FAC1_29 Immigration not immigrantsWelfareStateW3 a burden on the immigrantsWelfareStateW4 immigrantsWelfareStateW7 welfare state reasonForUnemploymentW1 reasonForUnemploymentW2 2.22 55.48% FAC1_30 Sympathy for the reasonForUnemploymentW3 unemployed reasonForUnemploymentW4 reasonForUnemploymentW7 FAC1_25 FAC1_26 3.63 36.31% Immig Support for FAC1_27 immigration FAC1_28 FAC1_29 FAC1_30 1.55 15.53% Interv Government lr1W6 (favours economic redistribution) al3W6 (disfavours authority interventionism in schools) al4W6 (disfavours 1.14 11.39% Social Social liberalism censorship) al5W6 (disfavours stiffer sentences for law breakers) Immig Interv 1.55 51.65% Leftness General political Social leftness [the 100 Likert scales with directions removed, as 11.95 11.95% ResPol Polarisation of indicators of response Likert scale polarisation; factor analysis set to save factors with responses eigenvalues of 10 or above, rather than 1 or above, due (cognitive to the extremely large classification) number of items included] Table 4. Principal component factor analysis results including details of factors saved

Constituent variables Factor 1 Loadings Factor 2 Loadings Factor 3 Loadings FAC1_22 (intolerance of criticism of FAC2_22 (feelings of shame at --- British society) being British) ethno1W7 (Britain does not have a 0.51 0.21 --- lot to learn from other countries) ethno2W7 (often as proud of 0.66 -0.46 --- Britain as would like to be) ethno3W7 (would rather be citizen 0.62 0.42 --- of Britain than of any other country) ethno4W7 (people in Britain are too 0.42 0.49 --- ready to criticise their country) ethno5W7 (other countries should 0.56 0.50 --- be more like Britain) ethno6W7 (some things make me 0.65 -0.47 --- ashamed to be British) radicalW7 (do not need to 0.51 -0.49 --- fundamentally change the way society works in Britain)

FAC1_11 (interest in politics) FAC2_11 (likelihood to vote in --- elections) FAC1_3 (attention to politics) 0.72 0.71 --- FAC1_4 (days in a week discussing 0.78 0.41 --- politics) FAC1_5 (talking about politics to 0.83 0.23 --- other people) FAC1_6 (following politics on the 0.76 0.15 --- internet) FAC1_7 (following politics in the 0.71 0.30 --- papers) FAC1_8 (following politics on the 0.58 0.14 --- radio) FAC1_9 (following politics on TV) 0.70 0.37 --- FAC1_10 (interest in the 2015 0.56 0.81 --- general election) FAC1_12 (likelihood to vote in the 0.22 0.82 --- 2015 general election) euRefInterestW7 (interest in EU 0.37 0.70 --- referendum) euRefTurnoutW7 (likelihood to vote 0.13 0.78 --- in EU referendum)

TolUncer (tolerance of uncertainty) Willrisk (risk willingness) --- tolUncertain1W8 (comfortable not 0.82 0.12 --- knowing what the future holds) tolUncertain2W8 (comfortable with 0.83 0.29 --- uncertainty) tolUncertain3W8 (does not feel the 0.75 0.24 --- need to be certain about the outcome before making a decision) riskTakingW7 (willingness to take 0.23 0.93 --- risks in wave 7) risktakingW8 (willingness to take 0.27 0.93 --- risks in wave 8)

Immig (support for immigration) Interv (government economic Social (social liberalism) interventionism) FAC1_25 (anti-bonuses) 0.07 0.60 0.03 FAC1_26 (pro-handouts) 0.56 0.70 0.38 FAC1_27 (immigration good for 0.91 0.17 0.35 economy) FAC1_28 (wants to increase 0.81 0.16 0.33 immigration) FAC1_29 (immigrants not burden 0.92 0.27 0.46 on the welfare state) FAC1_30 (unemployed not at fault) 0.19 0.76 0.21 lr1W6 (favours redistribution) 0.12 0.71 0.02 al3W6 (disfavours authority in 0.37 0.15 0.81 schools) al4W6 (disfavours censorship) 0.22 0.04 0.74 al5W6 (disfavours stiffer sentences 0.41 0.18 0.77 for law breakers) Table 5. Rotated factor loadings for principal components analyses which saved more than one factor. The closer the factor loading number is to 1, the more strongly associated the relevant item is to the relevant factor.

Cluster analysis for grouping individuals into categories Two-step cluster analysis classifies a set of individual cases into a discrete number of categories, or “clusters”, based on similar responses to a certain number of selected variables, which can be either categorical or continuous. The clustering algorithm attempts to makes differences in these variables within a category as small as possible, whilst making differences in the same variables between categories as large as possible. These metrics are jointly assessed in the form of the “silhouette measure of cohesion and separation, a statistic which ranges from -1 to 1. The larger the value of the statistic, the greater are the between-group differences and the lesser the within-group differences on the selected variables. A silhouette measure of 0.5 or greater is generally considered necessary to conclude that the cluster analysis has identified groups which are sufficiently distinguished from one another. Cluster analysis can be used to save a new nominal variable signifying which of the identified clusters each individual case belongs to.

Input Silhouette Clustering Number Populations Composition of Names of Variables measure variable of of clusters clusters clusters clusters country Mixed ethnicity; live countryOfBirth 0.60 TSC_4171 3 2850 across UK; born across “External” livedAbroadW8 the world; many lived parentsForeignW8 abroad; many foreign profile_ethnicity parents; White British; live in 1241 Scotland and Wales; born “Celtic” in UK; few lived abroad; no foreign parents; White British; live in 5530 England; born in UK; “Anglo- none lived abroad; no Saxon” foreign parents; Cnstrnt Low average level of 0.65 TSC_5725 3 1677 constraint “Low” Moderate average level 4375 of constraint “Medium” High average level of 3894 constraint “High” Table 6. Cluster analysis results for cultural cluster and level of constraint cluster.

Descriptive statistics for continuous variables

Variable Mean Standard deviation Skewness Political leftness 0.00 1.00 0.71 Cultural conformity 0.00 1.00 -0.80 Ideological conformity 0.00 1.00 -1.35 Active open-mindedness 0.00 0.48 0.64 Anglo-Saxon cultural cluster membership * cultural conformity 0.11 0.65 -0.35 interaction term Anglo-Saxon cultural cluster membership * ideological conformity 0.03 0.69 -1.39 interaction term Anglo-Saxon cultural cluster membership * active open- -0.01 0.35 -0.25 mindedness interaction term Belief system constraint 0.00 1.00 -0.40 Constraint * cultural conformity interaction term -0.13 1.05 -0.64 Constraint * ideological conformity interaction term -0.14 1.06 -2.48 Constraint * active open-mindedness interaction term 0.01 0.50 -0.50 Tolerance of uncertainty 0.00 1.00 0.31 Active open-mindedness * tolerance of uncertainty interaction term 0.00 0.48 0.47 Response polarisation 0.00 1.00 0.16 Active open-mindedness * response polarisation interaction term 0.00 0.53 -0.45 Table 7. Descriptive statistics for continuous variables used in the analysis. The interaction terms are simply pairs of other variables multiplied together, used to assess whether they are required to work together to produce an outcome.

Frequency tables for categorical variables

Variable Response categories Frequencies Valid percentages Cultural cluster membership Anglo-Saxon 4807 48.4% Not Anglo-Saxon 5128 51.6%

Level of education Has not attended higher education 4176 42.0% Is enrolled at, has been enrolled at, 5775 58.0% or has completed higher education

Disability Is not disabled 7114 71.2% Is disabled 2872 28.8%

Gender Male 5583 54.9% Female 4589 45.1%

Newspaper readership Broadsheet 2301 22.6% Tabloid 3532 34.7% Other or none 4339 42.7%

Level of skill in work Highly-skilled 4014 39.5% Moderately skilled 4098 40.3% Lower-skilled 1005 9.9% Never worked 1055 10.3% Table 8. Frequency tables for all categorical variables used in the analysis. “Valid percentages” refers to the percentage of non-missing responses which fall within each response category.

Bivariate comparisons with key variables Bivariate analysis involves comparing pairs of variables to see whether and how variations in one variable accompany variations in the other. The following tables summarise a series of comparisons between each of the key political, cultural and psychological variables (all of which are continuous) and each of the other variables selected or created for the analysis. Where two continuous variables are compared, the comparison consists of a Pearson correlation, a regression coefficient which varies between -1 (a perfect negative correlation) and 1 (a perfect positive correlation). The stars following these are indicative of the statistical significance of the relationship: * indicates that the likelihood of there being no true relationship between the two variables is 5% or less but greater than 1%; ** indicates that this likelihood is 1% or less but greater than 0.1%; *** indicates that the likelihood is 0.1% or less. As such, the more stars a correlation coefficient has, the more confident one can be that the coefficient represents a meaningful real-world relationship. Where a continuous variable is compared with a categorical one, a t-test to assess differences of means has been carried out. This compares mean values of the continuous variable between categories of the categorical variable, with a difference in means being considered significant if the probability of the mean values being the same in the real-world population is 5% or less. The full t-tests have not been reported here; they can be replicated using the syntax below and the essential results have been described in the following tables. Only statistically significant differences in means have been reported. For example, the results of the first t-test, listed as External = Celtic > Anglo- Saxon, can be translated thus: members of the “External” cultural cluster do not have a significantly different mean level of immigration support to members of the “Celtic” cluster, but both have significantly greater mean immigration support than members of the “Anglo-Saxon” cluster. When a mean difference is only marginal- that is, one group has a significantly higher or lower mean than another but the probably of it being spurious is very nearly 5%- a ≥ or ≤ symbol is used.

Pro-immigration Pro-economic Pro-social General interventionism liberalism leftness Pro-immigration --- 0.23*** 0.42*** 0.81*** Pro-economic ------0.15*** 0.55*** interventionism Pro-social liberalism ------0.77*** Cultural cluster External = Celtic > Anglo- Celtic > Anglo-Saxon = External = Celtic > Anglo- External = Celtic > Anglo- Saxon External Saxon Saxon Self-reported cultural conformity -0.51*** -0.24*** -0.56*** -0.63*** Computed ideological conformity -0.21*** -0.17*** -0.25*** -0.30*** Belief system constraint 0.18*** -0.28*** 0.15*** 0.07*** Newspaper readership Broadsheet > other or none No significant differences by Broadsheet > other or none Broadsheet > other or > tabloid newspaper readership > tabloid none > tabloid Work type Highly skilled > never Low skilled > never worked = Highly skilled > never Highly skilled > never worked > middling skilled > middling skilled > highly worked > middling skilled > worked > middling = low low skilled skilled low skilled skilled Age -0.21*** 0.10*** -0.25*** -0.20*** Education Graduates > non-graduates Non-graduates ≥ graduates Graduates > non-graduates Graduates > non- graduates Disability Not disabled > disabled Disabled > not disabled Not disabled > disabled No significant differences by disability Gender Male > female No significant differences by Male > female Male > female gender Political engagement 0.16*** 0.12*** 0.11*** 0.18*** Positive household economic evaluations 0.08*** -0.33*** 0.09*** -0.03**

Trust in MPs 0.18*** -0.16** 0.06*** 0.07*** Conservative partisanship -0.14*** -0.53*** -0.20*** -0.36*** Active open-mindedness 0.13*** 0.02* 0.20*** 0.17*** Anger at parties 0.12*** 0.29*** 0.18*** 0.25*** Fear of parties 0.12*** 0.04*** 0.09*** 0.12*** Tolerance of uncertainty 0.09*** -0.04*** 0.18*** 0.13*** Risk-taking 0.06*** -0.08*** 0.08*** 0.04*** Item-level response rate -0.01 0.06*** -0.02 0.01 Response polarisation -0.20*** 0.28*** -0.10*** -0.06*** Table 9. Bivariate comparisons between the political orientation factors and other selected variables.

Active open- Tolerance of Response polarisation mindedness uncertainty Cultural cluster External = Celtic > Anglo-Saxon External > Anglo-Saxon (both = Celtic > Anglo-Saxon = External Celtic) Self-reported cultural conformity -0.18*** -0.09*** 0.09*** Computed ideological conformity -0.10*** -0.06*** -0.33*** Belief system constraint 0.01 0.09*** -0.48*** Newspaper readership Broadsheet > Other or none > Broadsheet > Other or none > Tabloid > Other or none > Tabloid Tabloid Broadsheet Work type Highly skilled > never worked = Highly skilled > never worked > Low skilled > never worked = middling skilled > low skilled middling skilled = low skilled middling skilled > highly skilled Age -0.08*** 0.07*** 0.13*** Education Graduates > non-graduates No significant differences Non-graduates > Graduates Disability Not disabled ≥ disabled Not disabled > disabled Disabled > not disabled Gender Male > female Male > female No significant differences Political engagement 0.08*** 0.06*** 0.18*** Positive household economic evaluations 0.04*** 0.07*** -0.26*** Trust in MPs 0.00 0.09*** -0.28*** Conservative partisanship -0.04*** 0.05*** -0.37*** Table 10. Bivariate comparisons between the perceptual factors and other selected variables.

Self-reported cultural Computed Belief system conformity ideological constraint conformity Cultural cluster Anglo-Saxon > Celtic = External Anglo-Saxon = External > Celtic Celtic = Anglo-Saxon > External Newspaper readership Tabloid > other or none > Tabloid > other or none > Broadsheet > other or none > broadsheet broadsheet tabloid Work type Low skilled = middling skilled > Low skilled > middling skilled > All others > low skilled, other never worked > highly skilled never worked > highly skilled differences not significant Age 0.35*** 0.03* -0.11*** Education Non-graduates > graduates Non-graduates > graduates Graduates > non-graduates Disability Disabled > not disabled No significant differences Not disabled > disabled Gender No significant differences No significant differences Female > male Political engagement -0.03** -0.16*** -0.09*** Positive household economic evaluations -0.03** 0.00 0.18*** Trust in MPs 0.04*** 0.05*** 0.14*** Conservative partisanship 0.32*** 0.14*** 0.17***- Table 11. Bivariate comparisons between the continuous cultural variables and other selected variables.

Section 3: complete PROCESS output for the mediation and moderated mediation models

Note: the indirect effect sizes for the moderated mediation analysis have been assigned numbers from 1-108 to correspond to the effect sizes displayed in Figure 6.

Mediation analysis Run MATRIX procedure:

************** PROCESS Procedure for SPSS Release 2.16.3 *****************

Written by Andrew F. Hayes, Ph.D. www.afhayes.com

************************************************************************** Model = 4 Y = Leftness X = Aom M1 = Conform M2 = Compform

Statistical Controls: Columns 1 - 14 CONTROL= TolUncer ResPol Cnstrnt Anglo Celtic ResRate Willrisk AngerPty FearPty ConPart TrustMPs Econom Engage Columns 15 - 22 CONTROL= Zage Graduate Disabled Upskill Midskill Unskill Bdsheet Tbloid

Sample size 6108

************************************************************************** Outcome: Conform

Model Summary R R-sq MSE F df1 df2 p .6387 .4080 .7458 182.2824 23.0000 6084.0000 .0000

Model coeff se t p LLCI ULCI constant -.2050 .0471 -4.3536 .0000 -.2974 -.1127 Aom -.1326 .0349 -3.7952 .0001 -.2011 -.0641 TolUncer -.0911 .0116 -7.8578 .0000 -.1139 -.0684 ResPol .1043 .0153 6.8086 .0000 .0743 .1343 Cnstrnt -.1242 .0136 -9.1348 .0000 -.1508 -.0975 Anglo .3046 .0265 11.4856 .0000 .2526 .3566 Celtic .0111 .0309 .3584 .7201 -.0495 .0717 ResRate -.0084 .0174 -.4827 .6293 -.0425 .0257 Willrisk .0070 .0128 .5459 .5852 -.0181 .0321 AngerPty -.1140 .0128 -8.8870 .0000 -.1391 -.0888 FearPty -.0478 .0113 -4.2450 .0000 -.0699 -.0257 ConPart .5031 .0142 35.5275 .0000 .4753 .5308 TrustMPs -.0985 .0133 -7.4153 .0000 -.1245 -.0725 Econom -.0234 .0127 -1.8386 .0660 -.0483 .0015 Engage -.0202 .0141 -1.4307 .1526 -.0480 .0075 Gender -.0255 .0233 -1.0921 .2748 -.0712 .0203 Zage .2348 .0147 15.9177 .0000 .2059 .2637 Graduate -.2200 .0263 -8.3567 .0000 -.2717 -.1684 Disabled .0629 .0260 2.4198 .0156 .0119 .1138 Upskill .0017 .0411 .0424 .9661 -.0788 .0823 Midskill .0007 .0399 .0170 .9864 -.0776 .0789 Unskill .0324 .0516 .6269 .5308 -.0688 .1336 Bdsheet -.2105 .0298 -7.0609 .0000 -.2689 -.1521 Tbloid .2540 .0270 9.4245 .0000 .2012 .3069

************************************************************************** Outcome: Compform

Model Summary R R-sq MSE F df1 df2 p .5738 .3292 .8013 129.8236 23.0000 6084.0000 .0000

Model coeff se t p LLCI ULCI constant -.0197 .0488 -.4043 .6860 -.1154 .0760 Aom -.1042 .0362 -2.8785 .0040 -.1752 -.0332 TolUncer -.0555 .0120 -4.6181 .0000 -.0791 -.0320 ResPol -.5822 .0159 -36.6710 .0000 -.6133 -.5511 Cnstrnt -.4706 .0141 -33.3933 .0000 -.4982 -.4430 Anglo .0359 .0275 1.3055 .1918 -.0180 .0898 Celtic -.0314 .0320 -.9801 .3271 -.0942 .0314 ResRate -.0237 .0180 -1.3126 .1894 -.0590 .0117 Willrisk .0218 .0133 1.6417 .1007 -.0042 .0478 AngerPty -.0564 .0133 -4.2394 .0000 -.0824 -.0303

FearPty -.0157 .0117 -1.3480 .1777 -.0387 .0072 ConPart .0718 .0147 4.8883 .0000 .0430 .1005 TrustMPs -.0581 .0138 -4.2204 .0000 -.0851 -.0311 Econom -.0249 .0132 -1.8941 .0583 -.0508 .0009 Engage -.0100 .0147 -.6853 .4932 -.0388 .0187 Gender -.0231 .0242 -.9539 .3402 -.0705 .0243 Zage .0856 .0153 5.5969 .0000 .0556 .1155 Graduate -.1062 .0273 -3.8908 .0001 -.1597 -.0527 Disabled .0383 .0269 1.4213 .1553 -.0145 .0910 Upskill -.0613 .0426 -1.4384 .1504 -.1447 .0222 Midskill -.0287 .0414 -.6933 .4881 -.1098 .0524 Unskill .0072 .0535 .1355 .8922 -.0977 .1121 Bdsheet -.1799 .0309 -5.8222 .0000 -.2405 -.1193 Tbloid .1175 .0279 4.2055 .0000 .0627 .1723

************************************************************************** Outcome: Leftness

Model Summary R R-sq MSE F df1 df2 p .7659 .5866 .5025 345.1409 25.0000 6082.0000 .0000

Model coeff se t p LLCI ULCI constant .0210 .0387 .5411 .5884 -.0550 .0969 Conform -.3491 .0109 -32.0183 .0000 -.3704 -.3277 Compform -.3005 .0105 -28.5715 .0000 -.3211 -.2799 Aom .0522 .0287 1.8188 .0690 -.0041 .1085 TolUncer .0375 .0096 3.9172 .0001 .0187 .0563 ResPol -.2658 .0142 -18.7743 .0000 -.2936 -.2381 Cnstrnt -.0584 .0121 -4.8133 .0000 -.0822 -.0346 Anglo .0547 .0220 2.4849 .0130 .0115 .0978 Celtic -.0064 .0254 -.2539 .7996 -.0562 .0433 ResRate -.0121 .0143 -.8490 .3959 -.0401 .0159 Willrisk -.0175 .0105 -1.6636 .0962 -.0381 .0031 AngerPty .0477 .0106 4.5028 .0000 .0269 .0685 FearPty .0236 .0093 2.5485 .0108 .0054 .0418 ConPart -.3472 .0128 -27.1437 .0000 -.3722 -.3221 TrustMPs .1418 .0110 12.9432 .0000 .1203 .1633 Econom -.0041 .0104 -.3901 .6965 -.0245 .0164 Engage .1236 .0116 10.6477 .0000 .1009 .1464 Gender -.1283 .0192 -6.6970 .0000 -.1658 -.0907 Zage .0053 .0124 .4302 .6670 -.0189 .0295 Graduate .1201 .0217 5.5221 .0000 .0774 .1627 Disabled .0577 .0213 2.7066 .0068 .0159 .0996 Upskill -.0038 .0337 -.1138 .9094 -.0700 .0623 Midskill -.0190 .0328 -.5784 .5630 -.0832 .0453 Unskill .0177 .0424 .4172 .6765 -.0654 .1008 Bdsheet .0871 .0246 3.5409 .0004 .0389 .1354 Tbloid -.1717 .0223 -7.7043 .0000 -.2154 -.1280

************************** TOTAL EFFECT MODEL **************************** Outcome: Leftness

Model Summary R R-sq MSE F df1 df2 p .6460 .4174 .7079 189.4910 23.0000 6084.0000 .0000

Model coeff se t p LLCI ULCI constant .0985 .0459 2.1457 .0319 .0085 .1884 Aom .1298 .0340 3.8147 .0001 .0631 .1965 TolUncer .0860 .0113 7.6106 .0000 .0638 .1081 ResPol -.1273 .0149 -8.5298 .0000 -.1565 -.0980 Cnstrnt .1263 .0132 9.5374 .0000 .1004 .1523 Anglo -.0624 .0258 -2.4157 .0157 -.1131 -.0118 Celtic -.0009 .0301 -.0290 .9769 -.0599 .0582 ResRate -.0021 .0169 -.1228 .9022 -.0353 .0311 Willrisk -.0265 .0125 -2.1224 .0338 -.0510 -.0020 AngerPty .1044 .0125 8.3589 .0000 .0799 .1289 FearPty .0450 .0110 4.1022 .0000 .0235 .0666 ConPart -.5443 .0138 -39.4566 .0000 -.5714 -.5173 TrustMPs .1937 .0129 14.9652 .0000 .1683 .2190 Econom .0116 .0124 .9355 .3496 -.0127 .0359 Engage .1337 .0138 9.7042 .0000 .1067 .1607 Gender -.1125 .0227 -4.9470 .0000 -.1570 -.0679 Zage -.1023 .0144 -7.1223 .0000 -.1305 -.0742 Graduate .2288 .0257 8.9182 .0000 .1785 .2791 Disabled .0243 .0253 .9602 .3370 -.0253 .0739 Upskill .0140 .0400 .3487 .7273 -.0645 .0924 Midskill -.0106 .0389 -.2718 .7858 -.0868 .0657

Unskill .0042 .0503 .0836 .9334 -.0944 .1028 Bdsheet .2147 .0290 7.3908 .0000 .1577 .2716 Tbloid -.2957 .0263 -11.2611 .0000 -.3472 -.2442

***************** TOTAL, DIRECT, AND INDIRECT EFFECTS ******************** Total effect of X on Y Effect SE t p LLCI ULCI .1298 .0340 3.8147 .0001 .0631 .1965 Direct effect of X on Y Effect SE t p LLCI ULCI .0522 .0287 1.8188 .0690 -.0041 .1085 Indirect effect of X on Y Effect Boot SE BootLLCI BootULCI TOTAL .0776 .0194 .0412 .1170 Conform .0463 .0133 .0214 .0738 Compform .0313 .0111 .0102 .0539 Partially standardized indirect effect of X on Y Effect Boot SE BootLLCI BootULCI TOTAL .0921 .0231 .0487 .1390 Conform .0549 .0158 .0253 .0874 Compform .0372 .0132 .0120 .0641 Completely standardized indirect effect of X on Y Effect Boot SE BootLLCI BootULCI TOTAL .0292 .0073 .0155 .0442 Conform .0174 .0050 .0080 .0276 Compform .0118 .0042 .0038 .0202 Ratio of indirect to total effect of X on Y Effect Boot SE BootLLCI BootULCI TOTAL .5977 .2644 .3210 1.1304 Conform .3564 .1713 .1616 .7312 Compform .2412 .1268 .0859 .4907 Ratio of indirect to direct effect of X on Y Effect Boot SE BootLLCI BootULCI TOTAL 1.4857 59.9040 -2.9861 21.6596 Conform .8860 35.6705 -1.9728 12.8463 Compform .5997 24.6272 -1.1596 8.1679

******************** ANALYSIS NOTES AND WARNINGS *************************

Number of bootstrap samples for bias corrected bootstrap confidence intervals: 10000

Level of confidence for all confidence intervals in output: 95.00

NOTE: Some cases were deleted due to missing data. The number of such cases was: 4064

------END MATRIX -----

Moderated mediation analysis

Run MATRIX procedure:

************** PROCESS Procedure for SPSS Release 2.16.3 *****************

Written by Andrew F. Hayes, Ph.D. www.afhayes.com

************************************************************************** Model = 50 Y = Leftness X = Aom M1 = Conform M2 = Compform W = ResPol Z = TolUncer V = AngloCul Q = Cnstrnt

Statistical Controls: Columns 1 - 14 CONTROL= ResRate Willrisk AngerPty FearPty ConPart TrustMPs Econom Engage Gender Zage Graduate Disabled Upskill Columns 15 - 17 CONTROL= Unskill Bdsheet Tbloid

Sample size 6108

************************************************************************** Outcome: Conform

Model Summary R R-sq MSE F df1 df2 p .6190 .3831 .7769 171.7722 22.0000 6085.0000 .0000

Model coeff se t p LLCI ULCI constant -.0509 .0446 -1.1405 .2541 -.1383 .0366 Aom -.1154 .0369 -3.1278 .0018 -.1877 -.0431 ResPol .1585 .0143 11.0685 .0000 .1304 .1866 int_1 -.0937 .0332 -2.8202 .0048 -.1588 -.0286 TolUncer -.0923 .0119 -7.7641 .0000 -.1156 -.0690 int_2 -.0638 .0335 -1.9066 .0566 -.1294 .0018 ResRate .0049 .0178 .2743 .7839 -.0301 .0398 Willrisk -.0004 .0130 -.0320 .9745 -.0260 .0251 AngerPty -.1263 .0131 -9.6771 .0000 -.1519 -.1007 FearPty -.0452 .0115 -3.9328 .0001 -.0677 -.0227 ConPart .5158 .0144 35.8536 .0000 .4876 .5440 TrustMPs -.1026 .0135 -7.5836 .0000 -.1292 -.0761 Econom -.0313 .0130 -2.4153 .0158 -.0567 -.0059 Engage -.0282 .0144 -1.9522 .0510 -.0564 .0001 Gender -.0377 .0237 -1.5886 .1122 -.0842 .0088 Zage .2389 .0150 15.8740 .0000 .2094 .2684 Graduate -.2434 .0268 -9.0775 .0000 -.2959 -.1908 Disabled .0635 .0265 2.3971 .0166 .0116 .1155 Upskill .0061 .0419 .1451 .8846 -.0761 .0883 Midskill .0166 .0407 .4076 .6836 -.0632 .0964 Unskill .0476 .0527 .9038 .3662 -.0556 .1508 Bdsheet -.2503 .0303 -8.2499 .0000 -.3097 -.1908 Tbloid .2682 .0275 9.7564 .0000 .2143 .3221

Product terms key: int_1 Aom X ResPol int_2 Aom X TolUncer

************************************************************************** Outcome: Compform

Model Summary R R-sq MSE F df1 df2 p .4548 .2069 .9473 72.1439 22.0000 6085.0000 .0000

Model coeff se t p LLCI ULCI constant -.0175 .0492 -.3557 .7221 -.1140 .0790 Aom -.0774 .0407 -1.8989 .0576 -.1572 .0025 ResPol -.3696 .0158 -23.3754 .0000 -.4006 -.3386 int_1 -.1326 .0367 -3.6142 .0003 -.2045 -.0607 TolUncer -.0641 .0131 -4.8862 .0000 -.0898 -.0384 int_2 -.0824 .0370 -2.2294 .0258 -.1549 -.0099 ResRate -.0079 .0197 -.4020 .6877 -.0465 .0307 Willrisk .0294 .0144 2.0459 .0408 .0012 .0576 AngerPty -.0876 .0144 -6.0752 .0000 -.1158 -.0593 FearPty -.0226 .0127 -1.7837 .0745 -.0475 .0022 ConPart .1043 .0159 6.5653 .0000 .0732 .1354 TrustMPs -.0824 .0149 -5.5164 .0000 -.1117 -.0531 Econom -.0498 .0143 -3.4813 .0005 -.0778 -.0218 Engage -.0212 .0159 -1.3280 .1842 -.0524 .0101 Gender -.0828 .0262 -3.1607 .0016 -.1342 -.0315 Zage .0999 .0166 6.0141 .0000 .0674 .1325 Graduate -.1046 .0296 -3.5330 .0004 -.1626 -.0466 Disabled .0636 .0293 2.1728 .0298 .0062 .1210 Upskill -.0611 .0463 -1.3197 .1870 -.1519 .0297 Midskill -.0040 .0450 -.0884 .9296 -.0921 .0842 Unskill .0525 .0581 .9025 .3668 -.0615 .1664 Bdsheet -.2309 .0335 -6.8920 .0000 -.2965 -.1652 Tbloid .1175 .0304 3.8721 .0001 .0580 .1770

Product terms key: int_1 Aom X ResPol int_2 Aom X TolUncer

************************************************************************** Outcome: Leftness

Model Summary R R-sq MSE F df1 df2 p .7624 .5813 .5091 301.4505 28.0000 6079.0000 .0000

Model coeff se t p LLCI ULCI constant -.0197 .0374 -.5271 .5981 -.0931 .0536 Conform -.3716 .0127 -29.1841 .0000 -.3966 -.3467 Compform -.2489 .0127 -19.6582 .0000 -.2737 -.2241 Aom .0107 .0412 .2598 .7950 -.0701 .0916 AngloCul .0914 .0191 4.7972 .0000 .0541 .1288 Cnstrnt .0367 .0107 3.4258 .0006 .0157 .0577 int_3 -.1114 .0187 -5.9456 .0000 -.1481 -.0747 int_4 -.1141 .0091 -12.4690 .0000 -.1320 -.0962 int_5 .1498 .0184 8.1316 .0000 .1137 .1859 int_6 -.0171 .0098 -1.7453 .0810 -.0363 .0021 int_7 .0799 .0575 1.3892 .1648 -.0328 .1926 int_8 .0096 .0296 .3238 .7461 -.0485 .0676 ResRate -.0111 .0144 -.7687 .4421 -.0393 .0172 Willrisk -.0065 .0101 -.6466 .5179 -.0264 .0133 AngerPty .0026 .0104 .2491 .8033 -.0178 .0230 FearPty .0179 .0093 1.9250 .0543 -.0003 .0362 ConPart -.2635 .0126 -20.9697 .0000 -.2881 -.2388 TrustMPs .1670 .0110 15.2481 .0000 .1455 .1885 Econom .0121 .0105 1.1565 .2475 -.0084 .0326 Engage .0813 .0115 7.0655 .0000 .0588 .1039 Gender -.1282 .0193 -6.6404 .0000 -.1660 -.0903 Zage .0151 .0123 1.2212 .2220 -.0091 .0393 Graduate .1482 .0218 6.7852 .0000 .1054 .1910 Disabled .0304 .0214 1.4189 .1560 -.0116 .0724 Upskill .0211 .0340 .6213 .5344 -.0455 .0877 Midskill -.0027 .0330 -.0831 .9338 -.0674 .0619 Unskill .0142 .0426 .3339 .7385 -.0694 .0978 Bdsheet .0905 .0248 3.6538 .0003 .0420 .1391 Tbloid -.1914 .0224 -8.5436 .0000 -.2353 -.1475

Product terms key: int_3 Conform X AngloCul int_4 Conform X Cnstrnt int_5 Compform X AngloCul int_6 Compform X Cnstrnt int_7 Aom X AngloCul int_8 Aom X Cnstrnt

******************** DIRECT AND INDIRECT EFFECTS *************************

Conditional direct effect(s) of X on Y at values of the moderator(s): AngloCul Cnstrnt Effect SE t p LLCI ULCI .0000 -.9426 .0017 .0488 .0344 .9726 -.0939 .0973 .0000 -.0072 .0106 .0412 .2582 .7963 -.0702 .0915 .0000 .9282 .0196 .0506 .3880 .6981 -.0795 .1187 1.0000 -.9426 .0816 .0474 1.7226 .0850 -.0113 .1744 1.0000 -.0072 .0905 .0404 2.2388 .0252 .0113 .1698 1.0000 .9282 .0995 .0506 1.9657 .0494 .0003 .1987

Conditional indirect effect(s) of X on Y at values of the moderator(s):

Mediator ResPol TolUncer AngloCul Cnstrnt Effect Boot SE BootLLCI BootULCI Effect size number featured in Figure 6 Conform -.9006 -.8453 .0000 -.9426 -.0061 .0151 -.0369 .0221 1 Conform -.9006 -.8453 .0000 -.0072 -.0085 .0211 -.0515 .0311 2 Conform -.9006 -.8453 .0000 .9282 -.0109 .0272 -.0662 .0400 3 Conform -.9006 -.8453 1.0000 -.9426 -.0086 .0214 -.0520 .0317 4 Conform -.9006 -.8453 1.0000 -.0072 -.0110 .0274 -.0666 .0407 5 Conform -.9006 -.8453 1.0000 .9282 -.0135 .0335 -.0816 .0497 6 Conform -.9006 .1755 .0000 -.9426 .0112 .0118 -.0119 .0347 7 Conform -.9006 .1755 .0000 -.0072 .0157 .0166 -.0167 .0490 8 Conform -.9006 .1755 .0000 .9282 .0202 .0214 -.0216 .0631 9 Conform -.9006 .1755 1.0000 -.9426 .0159 .0168 -.0169 .0499 10 Conform -.9006 .1755 1.0000 -.0072 .0204 .0216 -.0216 .0641 11 Conform -.9006 .1755 1.0000 .9282 .0249 .0264 -.0265 .0779 12 Conform -.9006 1.1962 .0000 -.9426 .0284 .0156 -.0013 .0603 13

Conform -.9006 1.1962 .0000 -.0072 .0398 .0219 -.0021 .0842 14 Conform -.9006 1.1962 .0000 .9282 .0513 .0282 -.0023 .1086 15 Conform -.9006 1.1962 1.0000 -.9426 .0403 .0221 -.0017 .0859 16 Conform -.9006 1.1962 1.0000 -.0072 .0518 .0284 -.0023 .1096 17 Conform -.9006 1.1962 1.0000 .9282 .0632 .0348 -.0026 .1341 18 Conform .0982 -.8453 .0000 -.9426 .0187 .0132 -.0085 .0436 19 Conform .0982 -.8453 .0000 -.0072 .0262 .0185 -.0116 .0610 20 Conform .0982 -.8453 .0000 .9282 .0337 .0238 -.0147 .0789 21 Conform .0982 -.8453 1.0000 -.9426 .0265 .0189 -.0117 .0623 22 Conform .0982 -.8453 1.0000 -.0072 .0341 .0242 -.0151 .0797 23 Conform .0982 -.8453 1.0000 .9282 .0416 .0295 -.0184 .0972 24 Conform .0982 .1755 .0000 -.9426 .0359 .0100 .0174 .0563 25 Conform .0982 .1755 .0000 -.0072 .0504 .0139 .0244 .0785 26 Conform .0982 .1755 .0000 .9282 .0648 .0179 .0314 .1011 27 Conform .0982 .1755 1.0000 -.9426 .0510 .0143 .0244 .0806 28 Conform .0982 .1755 1.0000 -.0072 .0655 .0182 .0315 .1028 29 Conform .0982 .1755 1.0000 .9282 .0800 .0222 .0381 .1250 30 Conform .0982 1.1962 .0000 -.9426 .0531 .0147 .0256 .0839 31 Conform .0982 1.1962 .0000 -.0072 .0745 .0204 .0356 .1166 32 Conform .0982 1.1962 .0000 .9282 .0960 .0264 .0460 .1512 33 Conform .0982 1.1962 1.0000 -.9426 .0754 .0208 .0365 .1189 34 Conform .0982 1.1962 1.0000 -.0072 .0969 .0265 .0469 .1521 35 Conform .0982 1.1962 1.0000 .9282 .1183 .0325 .0573 .1860 36 Conform 1.0970 -.8453 .0000 -.9426 .0434 .0176 .0085 .0774 37 Conform 1.0970 -.8453 .0000 -.0072 .0609 .0245 .0118 .1082 38 Conform 1.0970 -.8453 .0000 .9282 .0784 .0316 .0149 .1393 39 Conform 1.0970 -.8453 1.0000 -.9426 .0617 .0252 .0120 .1106 40 Conform 1.0970 -.8453 1.0000 -.0072 .0792 .0321 .0149 .1409 41 Conform 1.0970 -.8453 1.0000 .9282 .0967 .0392 .0180 .1716 42 Conform 1.0970 .1755 .0000 -.9426 .0606 .0156 .0308 .0923 43 Conform 1.0970 .1755 .0000 -.0072 .0850 .0218 .0430 .1280 44 Conform 1.0970 .1755 .0000 .9282 .1095 .0281 .0554 .1649 45 Conform 1.0970 .1755 1.0000 -.9426 .0861 .0224 .0437 .1311 46 Conform 1.0970 .1755 1.0000 -.0072 .1106 .0285 .0560 .1670 47 Conform 1.0970 .1755 1.0000 .9282 .1351 .0348 .0682 .2047 48 Conform 1.0970 1.1962 .0000 -.9426 .0778 .0193 .0412 .1169 49 Conform 1.0970 1.1962 .0000 -.0072 .1092 .0268 .0576 .1631 50 Conform 1.0970 1.1962 .0000 .9282 .1406 .0347 .0743 .2105 51 Conform 1.0970 1.1962 1.0000 -.9426 .1106 .0274 .0580 .1652 52 Conform 1.0970 1.1962 1.0000 -.0072 .1420 .0349 .0749 .2115 53 Conform 1.0970 1.1962 1.0000 .9282 .1734 .0427 .0915 .2593 54

Mediator ResPol TolUncer AngloCul Cnstrnt Effect Boot SE BootLLCI BootULCI Effect size number featured in Figure 7 Compform -.9006 -.8453 .0000 -.9426 -.0260 .0153 -.0571 .0026 55 Compform -.9006 -.8453 .0000 -.0072 -.0278 .0163 -.0607 .0030 56 Compform -.9006 -.8453 .0000 .9282 -.0296 .0174 -.0647 .0034 57 Compform -.9006 -.8453 1.0000 -.9426 -.0093 .0061 -.0247 .0000 58 Compform -.9006 -.8453 1.0000 -.0072 -.0111 .0069 -.0264 .0007 59 Compform -.9006 -.8453 1.0000 .9282 -.0128 .0079 -.0308 .0009 60 Compform -.9006 .1755 .0000 -.9426 -.0064 .0112 -.0296 .0146 61 Compform -.9006 .1755 .0000 -.0072 -.0069 .0120 -.0318 .0156 62 Compform -.9006 .1755 .0000 .9282 -.0073 .0128 -.0342 .0166 63 Compform -.9006 .1755 1.0000 -.9426 -.0023 .0042 -.0120 .0050 64 Compform -.9006 .1755 1.0000 -.0072 -.0027 .0049 -.0135 .0062 65 Compform -.9006 .1755 1.0000 .9282 -.0032 .0057 -.0156 .0071 66 Compform -.9006 1.1962 .0000 -.9426 .0132 .0133 -.0125 .0394 67 Compform -.9006 1.1962 .0000 -.0072 .0141 .0142 -.0134 .0413 68 Compform -.9006 1.1962 .0000 .9282 .0150 .0151 -.0143 .0443 69 Compform -.9006 1.1962 1.0000 -.9426 .0047 .0050 -.0037 .0165 70 Compform -.9006 1.1962 1.0000 -.0072 .0056 .0058 -.0050 .0178 71 Compform -.9006 1.1962 1.0000 .9282 .0065 .0067 -.0058 .0205 72 Compform .0982 -.8453 .0000 -.9426 .0048 .0114 -.0172 .0280 73 Compform .0982 -.8453 .0000 -.0072 .0052 .0122 -.0183 .0296 74 Compform .0982 -.8453 .0000 .9282 .0055 .0130 -.0193 .0316 75 Compform .0982 -.8453 1.0000 -.9426 .0017 .0043 -.0060 .0113 76 Compform .0982 -.8453 1.0000 -.0072 .0021 .0050 -.0073 .0126 77 Compform .0982 -.8453 1.0000 .9282 .0024 .0057 -.0084 .0144 78 Compform .0982 .1755 .0000 -.9426 .0244 .0093 .0069 .0434 79 Compform .0982 .1755 .0000 -.0072 .0261 .0098 .0073 .0460 80 Compform .0982 .1755 .0000 .9282 .0278 .0105 .0077 .0488 81 Compform .0982 .1755 1.0000 -.9426 .0087 .0041 .0025 .0192 82 Compform .0982 .1755 1.0000 -.0072 .0104 .0044 .0032 .0206 83 Compform .0982 .1755 1.0000 .9282 .0121 .0051 .0035 .0234 84 Compform .0982 1.1962 .0000 -.9426 .0440 .0142 .0182 .0737 85

Compform .0982 1.1962 .0000 -.0072 .0470 .0149 .0194 .0773 86 Compform .0982 1.1962 .0000 .9282 .0500 .0160 .0202 .0823 87 Compform .0982 1.1962 1.0000 -.9426 .0157 .0066 .0053 .0317 88 Compform .0982 1.1962 1.0000 -.0072 .0187 .0068 .0074 .0344 89 Compform .0982 1.1962 1.0000 .9282 .0217 .0079 .0086 .0396 90 Compform 1.0970 -.8453 .0000 -.9426 .0356 .0161 .0056 .0684 91 Compform 1.0970 -.8453 .0000 -.0072 .0381 .0171 .0056 .0724 92 Compform 1.0970 -.8453 .0000 .9282 .0405 .0182 .0059 .0772 93 Compform 1.0970 -.8453 1.0000 -.9426 .0127 .0068 .0025 .0302 94 Compform 1.0970 -.8453 1.0000 -.0072 .0152 .0075 .0029 .0324 95 Compform 1.0970 -.8453 1.0000 .9282 .0176 .0087 .0031 .0371 96 Compform 1.0970 .1755 .0000 -.9426 .0552 .0167 .0236 .0887 97 Compform 1.0970 .1755 .0000 -.0072 .0590 .0175 .0250 .0938 98 Compform 1.0970 .1755 .0000 .9282 .0628 .0189 .0262 .1003 99 Compform 1.0970 .1755 1.0000 -.9426 .0197 .0080 .0073 .0391 100 Compform 1.0970 .1755 1.0000 -.0072 .0235 .0083 .0098 .0423 101 Compform 1.0970 .1755 1.0000 .9282 .0273 .0096 .0109 .0480 102 Compform 1.0970 1.1962 .0000 -.9426 .0748 .0214 .0350 .1185 103 Compform 1.0970 1.1962 .0000 -.0072 .0799 .0224 .0364 .1236 104 Compform 1.0970 1.1962 .0000 .9282 .0851 .0241 .0386 .1326 105 Compform 1.0970 1.1962 1.0000 -.9426 .0267 .0104 .0100 .0513 106 Compform 1.0970 1.1962 1.0000 -.0072 .0318 .0107 .0140 .0559 107 Compform 1.0970 1.1962 1.0000 .9282 .0369 .0123 .0161 .0638 108

Values for quantitative moderators are the mean and plus/minus one SD from mean. Values for dichotomous moderators are the two values of the moderator.

******************** ANALYSIS NOTES AND WARNINGS ************************* Number of bootstrap samples for bias corrected bootstrap confidence intervals: 10000 Level of confidence for all confidence intervals in output: 95.00

NOTE: Some cases were deleted due to missing data. The number of such cases was: 4064

------END MATRIX -----

Section 4: VNA file for visualising correlations between policy preferences

The following data represent Spearman rank correlations between every policy preference variable and every other policy preference variable. Spearman correlations are similar to Pearson correlations, varying between -1 and 1, except that they are more appropriate when comparing ordinal variables, which all of the policy preference items are. These data were used to create the network diagram shown in Figure 2, where points represent policy preference variables and lines represent correlations between them. Only significant correlations are featured; non-significant correlations are valued as zero. To be reproduced, the following text should be copied into Notepad and saved as a .vna file, then opened in Netdraw. ode data unempw4 unempw1 0 20 2 redist bonusw1 0.19 ID name theme median iqr unemployednotatfaultwave1 UnemployedNotAtFault1 bonusw1 sentence 0.05 bonusw1 bonusnotfairwave1 economy 3 1 unempw2 0 20 2 sentence bonusw1 0.05 economy 2 2 censor UnemployedNotAtFault2 bonusw1 schoolauth 0.06 bonusw2 bonusnotfairwave2 censorshipnotnecessary social unempw3 0 20 2 schoolauth bonusw1 0.06 economy 2 2 3 2 UnemployedNotAtFault3 bonusw2 bonusw3 0.58 bonusw3 bonusnotfairwave3 redist unempw4 0 20 2 bonusw3 bonusw2 0.58 economy 2 2 govshouldredistributeincomes UnemployedNotAtFault4 bonusw2 bonusw4 0.55 bonusw4 bonusnotfairwave4 economy 4 1 censor 255 20 1 bonusw4 bonusw2 0.55 economy 2 2 sentence CensorshipBad bonusw2 handoutsw1 0.23 handoutsw1 lawbreakersnotgetstiffersente redist 0 30 2 handoutsw1 bonusw2 0.23 handoutsjustifiedwave1 ncces social 2 2 RedistributeIncomesGood bonusw2 handoutsw2 0.21 economy 2 2 schoolauth sentence 255 10 1 handoutsw2 bonusw2 0.21 handoutsw2 schoolsnotteachobeyauthorit StifferSentencesBad bonusw2 handoutsw3 0.19 handoutsjustifiedwave2 y social 2 2 schoolauth 255 10 1 handoutsw3 bonusw2 0.19 economy 2 2 *node properties SchoolAuthorityBad bonusw2 handoutsw4 0.21 handoutsw3 ID color size shape shortlabel *tie data handoutsw4 bonusw2 0.21 handoutsjustifiedwave3 bonusw1 0 10 1 from to strength bonusw2 immigeconw1 0.02 economy 2 2 BonusesAreBad1 bonusw1 bonusw2 0.54 immigeconw1 bonusw2 0.02 handoutsw4 bonusw2 0 10 1 bonusw2 bonusw1 0.54 bonusw2 immigeconw2 0.05 handoutsjustifiedwave4 BonusesAreBad2 bonusw1 bonusw3 0.55 immigeconw2 bonusw2 0.05 economy 2 2 bonusw3 0 10 1 bonusw3 bonusw1 0.55 bonusw2 immigeconw3 0.02 immigeconw1 BonusesAreBad3 bonusw1 bonusw4 0.54 immigeconw3 bonusw2 0.02 immigrationgoodeconomywav bonusw4 0 10 1 bonusw4 bonusw1 0.54 bonusw2 immigeconw4 0.05 e1 immigration 3 2 BonusesAreBad4 bonusw1 handoutsw1 0.25 immigeconw4 bonusw2 0.05 immigeconw2 handoutsw1 0 10 1 handoutsw1 bonusw1 0.25 bonusw2 immlevelw4 0.03 immigrationgoodeconomywav HandoutsAreGood1 bonusw1 handoutsw2 0.21 immlevelw4 bonusw2 0.03 e2 immigration 3 2 handoutsw2 0 10 1 handoutsw2 bonusw1 0.21 bonusw2 immlevelw6 0.05 immigeconw3 HandoutsAreGood2 bonusw1 handoutsw3 0.21 immlevelw6 bonusw2 0.05 immigrationgoodeconomywav handoutsw3 0 10 1 handoutsw3 bonusw1 0.21 bonusw2 immwelfarew1 0.05 e3 immigration 3 2 HandoutsAreGood3 bonusw1 handoutsw4 0.20 immwelfarew1 bonusw2 0.05 immigeconw4 handoutsw4 0 10 1 handoutsw4 bonusw1 0.20 bonusw2 immwelfarew2 0.07 immigrationgoodeconomywav HandoutsAreGood4 bonusw1 immigeconw1 0.05 immwelfarew2 bonusw2 0.07 e4 immigration 3 2 immigeconw1 127 20 1 immigeconw1 bonusw1 0.05 bonusw2 immwelfarew3 0.06 immlevelw4 ImmigrationGoodForEconomy bonusw1 immigeconw2 0.05 immwelfarew3 bonusw2 0.06 increaseimmigrationW4 1 immigeconw2 bonusw1 0.05 bonusw2 immwelfarew4 0.09 immigration 2 2 immigeconw2 127 20 1 bonusw1 immigeconw3 0.05 immwelfarew4 bonusw2 0.09 immlevelw6 ImmigrationGoodForEconomy immigeconw3 bonusw1 0.05 bonusw2 immwelfarew8 0.08 increaseimmigrationW6 2 bonusw1 immigeconw4 0.05 immwelfarew8 bonusw2 0.08 immigration 2 2 immigeconw3 127 20 1 immigeconw4 bonusw1 0.05 bonusw2 unempw1 0.15 immwelfarew1 ImmigrationGoodForEconomy bonusw1 immlevelw4 0.05 unempw1 bonusw2 0.15 immigrantsnotwelfareburden 3 immlevelw4 bonusw1 0.05 bonusw2 unempw2 0.09 wave1 immigration 2 1 immigeconw4 127 20 1 bonusw1 immlevelw6 0.05 unempw2 bonusw2 0.09 immwelfarew2 ImmigrationGoodForEconomy immlevelw6 bonusw1 0.05 bonusw2 unempw3 0.10 immigrantsnotwelfareburden 4 bonusw1 immwelfarew1 0.10 unempw3 bonusw2 0.10 wave2 immigration 2 2 immlevelw4 127 10 1 immwelfarew1 bonusw1 0.10 bonusw2 unempw4 0.13 immwelfarew3 IncreaseImmigration1 bonusw1 immwelfarew2 0.10 unempw4 bonusw2 0.13 immigrantsnotwelfareburden immlevelw6 127 10 1 immwelfarew2 bonusw1 0.10 bonusw2 censor 0 wave3 immigration 2 2 IncreaseImmigration2 bonusw1 immwelfarew3 0.10 censor bonusw2 0 immwelfarew4 immwelfarew1 127 10 2 immwelfarew3 bonusw1 0.10 bonusw2 redist 0.18 immigrantsnotwelfareburden ImmigrantsNotWelfareBurden bonusw1 immwelfarew4 0.11 redist bonusw2 0.18 wave4 immigration 2 2 1 immwelfarew4 bonusw1 0.11 bonusw2 sentence 0.03 immwelfarew8 immwelfarew2 127 10 1 bonusw1 immwelfarew8 0.11 sentence bonusw2 0.03 immigrantsnotwelfareburden ImmigrantsNotWelfareBurden immwelfarew8 bonusw1 0.11 bonusw2 schoolauth 0.04 wave8 immigration 2 2 2 bonusw1 unempw1 0.15 schoolauth bonusw2 0.04 unempw1 immwelfarew3 127 10 1 unempw1 bonusw1 0.15 bonusw3 bonusw4 0.59 unemployednotatfaultwave1 ImmigrantsNotWelfareBurden bonusw1 unempw2 0.10 bonusw4 bonusw3 0.59 economy 3 1 3 unempw2 bonusw1 0.10 bonusw3 handoutsw1 0.20 unempw2 immwelfarew4 127 10 1 bonusw1 unempw3 0.11 handoutsw1 bonusw3 0.20 unemployednotatfaultwave2 ImmigrantsNotWelfareBurden unempw3 bonusw1 0.11 bonusw3 handoutsw2 0.22 economy 3 1 4 bonusw1 unempw4 0.13 handoutsw2 bonusw3 0.22 unempw3 immwelfarew8 127 10 1 unempw4 bonusw1 0.13 bonusw3 handoutsw3 0.22 unemployednotatfaultwave3 ImmigrantsNotWelfareBurden bonusw1 censor 0 handoutsw3 bonusw3 0.22 economy 3 1 5 censor bonusw1 0 bonusw3 handoutsw4 0.22 bonusw1 redist 0.19 handoutsw4 bonusw3 0.22 72 bonusw3 immigeconw1 0 sentence bonusw4 0.06 handoutsw2 immigeconw4 immwelfarew4 handoutsw3 immigeconw1 bonusw3 0 bonusw4 schoolauth 0.04 0.34 0.39 bonusw3 immigeconw2 0.05 schoolauth bonusw4 0.04 immigeconw4 handoutsw2 handoutsw3 immwelfarew8 immigeconw2 bonusw3 0.05 handoutsw1 handoutsw2 0.69 0.34 0.41 bonusw3 immigeconw3 0 handoutsw2 handoutsw1 0.69 handoutsw2 immlevelw4 0.33 immwelfarew8 handoutsw3 immigeconw3 bonusw3 0 handoutsw1 handoutsw3 0.69 immlevelw4 handoutsw2 0.33 0.41 bonusw3 immigeconw4 0.05 handoutsw3 handoutsw1 0.69 handoutsw2 immlevelw6 0.34 handoutsw3 unempw1 0.36 immigeconw4 bonusw3 0.05 handoutsw1 handoutsw4 0.66 immlevelw6 handoutsw2 0.34 unempw1 handoutsw3 0.36 bonusw3 immlevelw4 0.03 handoutsw4 handoutsw1 0.66 handoutsw2 immwelfarew1 handoutsw3 unempw2 0.35 immlevelw4 bonusw3 0.03 handoutsw1 immigeconw1 0.41 unempw2 handoutsw3 0.35 bonusw3 immlevelw6 0.04 0.33 immwelfarew1 handoutsw2 handoutsw3 unempw3 0.30 immlevelw6 bonusw3 0.04 immigeconw1 handoutsw1 0.41 unempw3 handoutsw3 0.30 bonusw3 immwelfarew1 0.06 0.33 handoutsw2 immwelfarew2 handoutsw3 unempw4 0.28 immwelfarew1 bonusw3 0.06 handoutsw1 immigeconw2 0.46 unempw4 handoutsw3 0.28 bonusw3 immwelfarew2 0.09 0.34 immwelfarew2 handoutsw2 handoutsw3 censor 0.15 immwelfarew2 bonusw3 0.09 immigeconw2 handoutsw1 0.46 censor handoutsw3 0.15 bonusw3 immwelfarew3 0.08 0.34 handoutsw2 immwelfarew3 handoutsw3 redist 0.26 immwelfarew3 bonusw3 0.08 handoutsw1 immigeconw3 0.41 redist handoutsw3 0.26 bonusw3 immwelfarew4 0.07 0.33 immwelfarew3 handoutsw2 handoutsw3 sentence 0.28 immwelfarew4 bonusw3 0.07 immigeconw3 handoutsw1 0.41 sentence handoutsw3 0.28 bonusw3 immwelfarew8 0.10 0.33 handoutsw2 immwelfarew4 handoutsw3 schoolauth 0.25 immwelfarew8 bonusw3 0.10 handoutsw1 immigeconw4 0.39 schoolauth handoutsw3 0.25 bonusw3 unempw1 0.14 0.31 immwelfarew4 handoutsw2 handoutsw4 immigeconw1 unempw1 bonusw3 0.14 immigeconw4 handoutsw1 0.39 0.36 bonusw3 unempw2 0.10 0.31 handoutsw2 immwelfarew8 immigeconw1 handoutsw4 unempw2 bonusw3 0.10 handoutsw1 immlevelw4 0.29 0.42 0.36 bonusw3 unempw3 0.12 immlevelw4 handoutsw1 0.29 immwelfarew8 handoutsw2 handoutsw4 immigeconw2 unempw3 bonusw3 0.12 handoutsw1 immlevelw6 0.30 0.42 0.37 bonusw3 unempw4 0.13 immlevelw6 handoutsw1 0.30 handoutsw2 unempw1 0.39 immigeconw2 handoutsw4 unempw4 bonusw3 0.13 handoutsw1 immwelfarew1 unempw1 handoutsw2 0.39 0.37 bonusw3 censor 0 0.42 handoutsw2 unempw2 0.36 handoutsw4 immigeconw3 censor bonusw3 0 immwelfarew1 handoutsw1 unempw2 handoutsw2 0.36 0.37 bonusw3 redist 0.19 0.42 handoutsw2 unempw3 0.26 immigeconw3 handoutsw4 redist bonusw3 0.19 handoutsw1 immwelfarew2 unempw3 handoutsw2 0.26 0.37 bonusw3 sentence 0.04 0.39 handoutsw2 unempw4 0.28 handoutsw4 immigeconw4 sentence bonusw3 0.04 immwelfarew2 handoutsw1 unempw4 handoutsw2 0.28 0.36 bonusw3 schoolauth 0.07 0.39 handoutsw2 censor 0.14 immigeconw4 handoutsw4 schoolauth bonusw3 0.07 handoutsw1 immwelfarew3 censor handoutsw2 0.14 0.36 bonusw4 handoutsw1 0.21 0.37 handoutsw2 redist 0.28 handoutsw4 immlevelw4 0.35 handoutsw1 bonusw4 0.21 immwelfarew3 handoutsw1 redist handoutsw2 0.28 immlevelw4 handoutsw4 0.35 bonusw4 handoutsw2 0.21 0.37 handoutsw2 sentence 0.28 handoutsw4 immlevelw6 0.36 handoutsw2 bonusw4 0.21 handoutsw1 immwelfarew4 sentence handoutsw2 0.28 immlevelw6 handoutsw4 0.36 bonusw4 handoutsw3 0.20 0.38 handoutsw2 schoolauth 0.26 handoutsw4 immwelfarew1 handoutsw3 bonusw4 0.20 immwelfarew4 handoutsw1 schoolauth handoutsw2 0.26 0.41 bonusw4 handoutsw4 0.22 0.38 handoutsw3 handoutsw4 0.69 immwelfarew1 handoutsw4 handoutsw4 bonusw4 0.22 handoutsw1 immwelfarew8 handoutsw4 handoutsw3 0.69 0.41 bonusw4 immigeconw1 0.05 0.40 handoutsw3 immigeconw1 handoutsw4 immwelfarew2 immigeconw1 bonusw4 0.05 immwelfarew8 handoutsw1 0.34 0.39 bonusw4 immigeconw2 0.05 0.40 immigeconw1 handoutsw3 immwelfarew2 handoutsw4 immigeconw2 bonusw4 0.05 handoutsw1 unempw1 0.43 0.34 0.39 bonusw4 immigeconw3 0 unempw1 handoutsw1 0.43 handoutsw3 immigeconw2 handoutsw4 immwelfarew3 immigeconw3 bonusw4 0 handoutsw1 unempw2 0.39 0.35 0.41 bonusw4 immigeconw4 0.06 unempw2 handoutsw1 0.39 immigeconw2 handoutsw3 immwelfarew3 handoutsw4 immigeconw4 bonusw4 0.06 handoutsw1 unempw3 0.32 0.35 0.41 bonusw4 immlevelw4 0.05 unempw3 handoutsw1 0.32 handoutsw3 immigeconw3 handoutsw4 immwelfarew4 immlevelw4 bonusw4 0.05 handoutsw1 unempw4 0.33 0.35 0.44 bonusw4 immlevelw6 0.06 unempw4 handoutsw1 0.33 immigeconw3 handoutsw3 immwelfarew4 handoutsw4 immlevelw6 bonusw4 0.06 handoutsw1 censor 0.16 0.35 0.44 bonusw4 immwelfarew1 0.05 censor handoutsw1 0.16 handoutsw3 immigeconw4 handoutsw4 immwelfarew8 immwelfarew1 bonusw4 0.05 handoutsw1 redist 0.29 0.32 0.42 bonusw4 immwelfarew2 0.08 redist handoutsw1 0.29 immigeconw4 handoutsw3 immwelfarew8 handoutsw4 immwelfarew2 bonusw4 0.08 handoutsw1 sentence 0.30 0.32 0.42 bonusw4 immwelfarew3 0.07 sentence handoutsw1 0.30 handoutsw3 immlevelw4 0.31 handoutsw4 unempw1 0.36 immwelfarew3 bonusw4 0.07 handoutsw1 schoolauth 0.25 immlevelw4 handoutsw3 0.31 unempw1 handoutsw4 0.36 bonusw4 immwelfarew4 0.11 schoolauth handoutsw1 0.25 handoutsw3 immlevelw6 0.31 handoutsw4 unempw2 0.35 immwelfarew4 bonusw4 0.11 handoutsw2 handoutsw3 0.69 immlevelw6 handoutsw3 0.31 unempw2 handoutsw4 0.35 bonusw4 immwelfarew8 0.09 handoutsw3 handoutsw2 0.69 handoutsw3 immwelfarew1 handoutsw4 unempw3 0.29 immwelfarew8 bonusw4 0.09 handoutsw2 handoutsw4 0.68 0.40 unempw3 handoutsw4 0.29 bonusw4 unempw1 0.12 handoutsw4 handoutsw2 0.68 immwelfarew1 handoutsw3 handoutsw4 unempw4 0.30 unempw1 bonusw4 0.12 handoutsw2 immigeconw1 0.40 unempw4 handoutsw4 0.30 bonusw4 unempw2 0.11 0.35 handoutsw3 immwelfarew2 handoutsw4 censor 0.17 unempw2 bonusw4 0.11 immigeconw1 handoutsw2 0.39 censor handoutsw4 0.17 bonusw4 unempw3 0.11 0.35 immwelfarew2 handoutsw3 handoutsw4 redist 0.27 unempw3 bonusw4 0.11 handoutsw2 immigeconw2 0.39 redist handoutsw4 0.27 bonusw4 unempw4 0.08 0.38 handoutsw3 immwelfarew3 handoutsw4 sentence 0.30 unempw4 bonusw4 0.08 immigeconw2 handoutsw2 0.43 sentence handoutsw4 0.30 bonusw4 censor 0 0.38 immwelfarew3 handoutsw3 handoutsw4 schoolauth 0.27 censor bonusw4 0 handoutsw2 immigeconw3 0.43 schoolauth handoutsw4 0.27 bonusw4 redist 0.20 0.36 handoutsw3 immwelfarew4 immigeconw1 immigeconw2 redist bonusw4 0.20 immigeconw3 handoutsw2 0.39 0.78 bonusw4 sentence 0.06 0.36

73 immigeconw2 immigeconw1 immigeconw2 immwelfarew3 immigeconw4 immlevelw4 sentence immlevelw4 0.33 0.78 0.69 0.59 immlevelw4 schoolauth 0.28 immigeconw1 immigeconw3 immwelfarew3 immigeconw2 immlevelw4 immigeconw4 schoolauth immlevelw4 0.28 0.76 0.69 0.59 immlevelw6 immwelfarew1 immigeconw3 immigeconw1 immigeconw2 immwelfarew4 immigeconw4 immlevelw6 0.53 0.76 0.67 0.57 immwelfarew1 immlevelw6 immigeconw1 immigeconw4 immwelfarew4 immigeconw2 immlevelw6 immigeconw4 0.53 0.75 0.67 0.57 immlevelw6 immwelfarew2 immigeconw4 immigeconw1 immigeconw2 immwelfarew8 immigeconw4 immwelfarew1 0.53 0.75 0.64 0.65 immwelfarew2 immlevelw6 immigeconw1 immlevelw4 immwelfarew8 immigeconw2 immwelfarew1 immigeconw4 0.53 0.55 0.64 0.65 immlevelw6 immwelfarew3 immlevelw4 immigeconw1 immigeconw2 unempw1 0.12 immigeconw4 immwelfarew2 0.55 0.55 unempw1 immigeconw2 0.12 0.66 immwelfarew3 immlevelw6 immigeconw1 immlevelw6 immigeconw2 unempw2 0.11 immwelfarew2 immigeconw4 0.55 0.54 unempw2 immigeconw2 0.11 0.66 immlevelw6 immwelfarew4 immlevelw6 immigeconw1 immigeconw2 unempw3 0.10 immigeconw4 immwelfarew3 0.57 0.54 unempw3 immigeconw2 0.10 0.66 immwelfarew4 immlevelw6 immigeconw1 immwelfarew1 immigeconw2 unempw4 0.12 immwelfarew3 immigeconw4 0.57 0.71 unempw4 immigeconw2 0.12 0.66 immlevelw6 immwelfarew8 immwelfarew1 immigeconw1 immigeconw2 censor 0.19 immigeconw4 immwelfarew4 0.54 0.71 censor immigeconw2 0.19 0.72 immwelfarew8 immlevelw6 immigeconw1 immwelfarew2 immigeconw2 redist 0.10 immwelfarew4 immigeconw4 0.54 0.65 redist immigeconw2 0.10 0.72 immlevelw6 unempw1 0.13 immwelfarew2 immigeconw1 immigeconw2 sentence 0.28 immigeconw4 immwelfarew8 unempw1 immlevelw6 0.13 0.65 sentence immigeconw2 0.28 0.64 immlevelw6 unempw2 0.14 immigeconw1 immwelfarew3 immigeconw2 schoolauth immwelfarew8 immigeconw4 unempw2 immlevelw6 0.14 0.65 0.28 0.64 immlevelw6 unempw3 0.11 immwelfarew3 immigeconw1 schoolauth immigeconw2 immigeconw4 unempw1 0.10 unempw3 immlevelw6 0.11 0.65 0.28 unempw1 immigeconw4 0.10 immlevelw6 unempw4 0.14 immigeconw1 immwelfarew4 immigeconw3 immigeconw4 immigeconw4 unempw2 0.11 unempw4 immlevelw6 0.14 0.64 0.77 unempw2 immigeconw4 0.11 immlevelw6 censor 0.17 immwelfarew4 immigeconw1 immigeconw4 immigeconw3 immigeconw4 unempw3 0.10 censor immlevelw6 0.17 0.64 0.77 unempw3 immigeconw4 0.10 immlevelw6 redist 0.12 immigeconw1 immwelfarew8 immigeconw3 immlevelw4 immigeconw4 unempw4 0.12 redist immlevelw6 0.12 0.60 0.57 unempw4 immigeconw4 0.12 immlevelw6 sentence 0.31 immwelfarew8 immigeconw1 immlevelw4 immigeconw3 immigeconw4 censor 0.19 sentence immlevelw6 0.31 0.60 0.57 censor immigeconw4 0.19 immlevelw6 schoolauth 0.29 immigeconw1 unempw1 0.12 immigeconw3 immlevelw6 immigeconw4 redist 0.09 schoolauth immlevelw6 0.29 unempw1 immigeconw1 0.12 0.56 redist immigeconw4 0.09 immwelfarew1 immigeconw1 unempw2 0.12 immlevelw6 immigeconw3 immigeconw4 sentence 0.29 immwelfarew2 0.73 unempw2 immigeconw1 0.12 0.56 sentence immigeconw4 0.29 immwelfarew2 immigeconw1 unempw3 0.09 immigeconw3 immwelfarew1 immigeconw4 schoolauth immwelfarew1 0.73 unempw3 immigeconw1 0.09 0.68 0.27 immwelfarew1 immigeconw1 unempw4 0.12 immwelfarew1 immigeconw3 schoolauth immigeconw4 immwelfarew3 0.72 unempw4 immigeconw1 0.12 0.68 0.27 immwelfarew3 immigeconw1 censor 0.19 immigeconw3 immwelfarew2 immlevelw4 immlevelw6 0.69 immwelfarew1 0.72 censor immigeconw1 0.19 0.67 immlevelw6 immlevelw4 0.69 immwelfarew1 immigeconw1 redist 0.09 immwelfarew2 immigeconw3 immlevelw4 immwelfarew1 immwelfarew4 0.71 redist immigeconw1 0.09 0.67 0.54 immwelfarew4 immigeconw1 sentence 0.30 immigeconw3 immwelfarew3 immwelfarew1 immlevelw4 immwelfarew1 0.71 sentence immigeconw1 0.30 0.72 0.54 immwelfarew1 immigeconw1 schoolauth immwelfarew3 immigeconw3 immlevelw4 immwelfarew2 immwelfarew8 0.67 0.27 0.72 0.56 immwelfarew8 schoolauth immigeconw1 immigeconw3 immwelfarew4 immwelfarew2 immlevelw4 immwelfarew1 0.67 0.27 0.68 0.56 immwelfarew1 unempw1 immigeconw2 immigeconw3 immwelfarew4 immigeconw3 immlevelw4 immwelfarew3 0.13 0.81 0.68 0.58 unempw1 immwelfarew1 immigeconw3 immigeconw2 immigeconw3 immwelfarew8 immwelfarew3 immlevelw4 0.13 0.81 0.64 0.58 immwelfarew1 unempw2 immigeconw2 immigeconw4 immwelfarew8 immigeconw3 immlevelw4 immwelfarew4 0.15 0.78 0.64 0.60 unempw2 immwelfarew1 immigeconw4 immigeconw2 immigeconw3 unempw1 0.15 immwelfarew4 immlevelw4 0.15 0.78 unempw1 immigeconw3 0.15 0.60 immwelfarew1 unempw3 immigeconw2 immlevelw4 immigeconw3 unempw2 0.13 immlevelw4 immwelfarew8 0.11 0.56 unempw2 immigeconw3 0.13 0.54 unempw3 immwelfarew1 immlevelw4 immigeconw2 immigeconw3 unempw3 0.12 immwelfarew8 immlevelw4 0.11 0.56 unempw3 immigeconw3 0.12 0.54 immwelfarew1 unempw4 immigeconw2 immlevelw6 immigeconw3 unempw4 0.13 immlevelw4 unempw1 0.08 0.14 0.55 unempw4 immigeconw3 0.13 unempw1 immlevelw4 0.08 unempw4 immwelfarew1 immlevelw6 immigeconw2 immigeconw3 censor 0.18 immlevelw4 unempw2 0.10 0.14 0.55 censor immigeconw3 0.18 unempw2 immlevelw4 0.10 immwelfarew1 censor 0.24 immigeconw2 immwelfarew1 immigeconw3 redist 0.09 immlevelw4 unempw3 0.08 censor immwelfarew1 0.24 0.68 redist immigeconw3 0.09 unempw3 immlevelw4 0.08 immwelfarew1 redist 0.10 immwelfarew1 immigeconw2 immigeconw3 sentence 0.31 immlevelw4 unempw4 0.09 redist immwelfarew1 0.10 0.68 sentence immigeconw3 0.31 unempw4 immlevelw4 0.09 immwelfarew1 sentence 0.34 immigeconw2 immwelfarew2 immigeconw3 schoolauth immlevelw4 censor 0.20 sentence immwelfarew1 0.34 0.72 0.29 censor immlevelw4 0.20 immwelfarew1 schoolauth immwelfarew2 immigeconw2 schoolauth immigeconw3 immlevelw4 redist 0.09 0.30 0.72 0.29 redist immlevelw4 0.09 schoolauth immwelfarew1 immlevelw4 sentence 0.33 0.30

74 immwelfarew2 immwelfarew3 unempw1 immwelfarew4 censor 0.26 unempw1 sentence 0.10 immwelfarew3 0.74 0.12 censor immwelfarew4 0.26 sentence unempw1 0.10 immwelfarew3 unempw1 immwelfarew3 immwelfarew4 redist 0.09 unempw1 schoolauth 0.09 immwelfarew2 0.74 0.12 redist immwelfarew4 0.09 schoolauth unempw1 0.09 immwelfarew2 immwelfarew3 unempw2 immwelfarew4 sentence 0.37 unempw2 unempw3 0.40 immwelfarew4 0.73 0.13 sentence immwelfarew4 0.37 unempw3 unempw2 0.40 immwelfarew4 unempw2 immwelfarew3 immwelfarew4 schoolauth unempw2 unempw4 0.37 immwelfarew2 0.73 0.13 0.33 unempw4 unempw2 0.37 immwelfarew2 immwelfarew3 unempw3 schoolauth immwelfarew4 unempw2 censor 0.06 immwelfarew8 0.66 0.08 0.33 censor unempw2 0.06 immwelfarew8 unempw3 immwelfarew3 immwelfarew8 unempw1 unempw2 redist 0.25 immwelfarew2 0.66 0.08 0.16 redist unempw2 0.25 immwelfarew2 unempw1 immwelfarew3 unempw4 unempw1 immwelfarew8 unempw2 sentence 0.08 0.12 0.11 0.16 sentence unempw2 0.08 unempw1 immwelfarew2 unempw4 immwelfarew3 immwelfarew8 unempw2 unempw2 schoolauth 0.07 0.12 0.11 0.16 schoolauth unempw2 0.07 immwelfarew2 unempw2 immwelfarew3 censor 0.24 unempw2 immwelfarew8 unempw3 unempw4 0.48 0.13 censor immwelfarew3 0.24 0.16 unempw4 unempw3 0.48 unempw2 immwelfarew2 immwelfarew3 redist 0.11 immwelfarew8 unempw3 unempw3 censor 0.06 0.13 redist immwelfarew3 0.11 0.13 censor unempw3 0.06 immwelfarew2 unempw3 immwelfarew3 sentence 0.35 unempw3 immwelfarew8 unempw3 redist 0.35 0.11 sentence immwelfarew3 0.35 0.13 redist unempw3 0.35 unempw3 immwelfarew2 immwelfarew3 schoolauth immwelfarew8 unempw4 unempw3 sentence 0.13 0.11 0.32 0.15 sentence unempw3 0.13 immwelfarew2 unempw4 schoolauth immwelfarew3 unempw4 immwelfarew8 unempw3 schoolauth 0.11 0.13 0.32 0.15 schoolauth unempw3 0.11 unempw4 immwelfarew2 immwelfarew4 immwelfarew8 censor 0.20 unempw4 censor 0.09 0.13 immwelfarew8 0.70 censor immwelfarew8 0.20 censor unempw4 0.09 immwelfarew2 censor 0.24 immwelfarew8 immwelfarew8 redist 0.12 unempw4 redist 0.29 censor immwelfarew2 0.24 immwelfarew4 0.70 redist immwelfarew8 0.12 redist unempw4 0.29 immwelfarew2 redist 0.11 immwelfarew4 unempw1 immwelfarew8 sentence 0.37 unempw4 sentence 0.15 redist immwelfarew2 0.11 0.14 sentence immwelfarew8 0.37 sentence unempw4 0.15 immwelfarew2 sentence 0.35 unempw1 immwelfarew4 immwelfarew8 schoolauth unempw4 schoolauth 0.15 sentence immwelfarew2 0.35 0.14 0.33 schoolauth unempw4 0.15 immwelfarew2 schoolauth immwelfarew4 unempw2 schoolauth immwelfarew8 censor redist 0 0.34 0.13 0.33 redist censor 0 schoolauth immwelfarew2 unempw2 immwelfarew4 unempw1 unempw2 0.50 censor sentence 0.32 0.34 0.13 unempw2 unempw1 0.50 sentence censor 0.32 immwelfarew3 immwelfarew4 unempw3 unempw1 unempw3 0.40 censor schoolauth 0.36 immwelfarew4 0.73 0.10 unempw3 unempw1 0.40 schoolauth censor 0.36 immwelfarew4 unempw3 immwelfarew4 unempw1 unempw4 0.40 censor sentence 0.05 immwelfarew3 0.73 0.10 unempw4 unempw1 0.40 sentence censor 0.05 immwelfarew3 immwelfarew4 unempw4 unempw1 censor 0.05 censor schoolauth 0.37 immwelfarew8 0.68 0.12 censor unempw1 0.05 schoolauth censor 0.37 immwelfarew8 unempw4 immwelfarew4 unempw1 redist 0.26 sentence schoolauth 0.53 immwelfarew3 0.68 0.12 redist unempw1 0.26 schoolauth sentence 0.53

Section 5: complete SPSS syntax used in the analysis

* Encoding: UTF-8. infoSourcePeopleW6 reasonForUnemploymentW3 RECODE infoSourcePeopleW7 reasonForUnemploymentW4 profile_work_typeW7 DATA SET ACTIVATE Data infoSourcePeopleW8 al4W6 lr1W6 al5W6 al3W6 (MISSING=SYSMIS) (7=1) set3. infoSourceInternetW4 aom1W7 aom4W7 (ELSE=0) INTO Unskilled. FREQUENCIES infoSourceInternetW5 aom5W7 aom6W7 aom7W7 VARIABLE LABELS Unskilled VARIABLES=wt_full_W1W2W infoSourceInternetW6 aom3W7 aom2W7 '[theme: demographics] Work 3W4W5W6W7W8W9 infoSourceInternetW7 conAngryW4 conAngryW6 type low skilled or unskilled as /FORMAT=NOTABLE infoSourceInternetW8 grnAngryW4 grnAngryW6 opposed '+ /STATISTICS=STDDEV infoSourcePaperW4 labAngryW4 labAngryW6 'to never worked'. MINIMUM MAXIMUM MEAN infoSourcePaperW5 ldAngryW4 ldAngryW6 EXECUTE. /ORDER=ANALYSIS. infoSourcePaperW6 ukipAngryW4 ukipAngryW6 infoSourcePaperW7 conFearW4 conFearW6 COMPUTE GRAPH infoSourcePaperW8 grnFearW4 policy1=businessBonusW1f. /HISTOGRAM(NORMAL)=wt_f infoSourceRadioW4 grnFearW6 labFearW4 VARIABLE LABELS policy1 ull_W1W2W3W4W5W6W7W infoSourceRadioW5 labFearW6 ldFearW4 '[theme: policies] [bonus] In 8W9. infoSourceRadioW6 ldFearW6 ukipFearW4 business, bonuses are a fair infoSourceRadioW7 ukipFearW6 way to '+ RECODE infoSourceRadioW8 personality_agreeableness 'reward hard work W1'. wt_full_W1W2W3W4W5W6 infoSourceTVW4 EXECUTE. W7W8W9 (20 thru infoSourceTVW5 personality_conscientiousness Highest=20). infoSourceTVW6 personality_extraversion RECODE ageW7 EXECUTE. infoSourceTVW7 personality_neuroticism (MISSING=SYSMIS) (Lowest infoSourceTVW8 personality_openness thru 49=0) (50 thru GRAPH electionInterestW4 riskTakingW7 riskTakingW8 Highest=1) INTO ageGroup. /HISTOGRAM(NORMAL)=wt_f electionInterestW5 tolUncertain1W8 VARIABLE LABELS ageGroup ull_W1W2W3W4W5W6W7W electionInterestW6 tolUncertain3W8 'Under 50 or 50 or over?'. 8W9. euRefInterestW7 tolUncertain2W8 . EXECUTE. registeredW3 registeredW4 FILTER OFF. registeredW6 registeredW7 RECODE gender (1=0) (2=1). COMPUTE demogCleave=0. USE ALL. registeredW8 EXECUTE. VARIABLE LABELS SELECT IF euRefTurnoutW7 demogCleave '[theme: (MISSING(wt_full_W1W2W3 turnoutUKGeneralW1 RECODE anyUniW7 (0=0) demographics] eight-way W4W5W6W7W8W9)=0). turnoutUKGeneralW2 (1=1) (2=1) (3=1). demographic cleavage EXECUTE. turnoutUKGeneralW3 EXECUTE. between age, '+ turnoutUKGeneralW4 'gender and education'. DATA SET ACTIVATE Data genElecTurnoutRetroW6 RECODE country (2=1) EXECUTE. set1. genElecTurnoutRetroW7 (MISSING=SYSMIS) (ELSE=0) SORT VARIABLES BY NAME euroTurnoutRetroW2 INTO Scot. DO IF (ageGroup = 1 & gender (A). partyMemberW6 VARIABLE LABELS Scot = 1 & anyUniW7 = 1). econPersonalRetroW1 '[theme: culture] [cluster] RECODE demogCleave (0=1). SORT VARIABLES BY LABEL (A). econPersonalRetroW2 Country is Scotland as END IF. econPersonalRetroW3 opposed to England'. EXECUTE. MVA VARIABLES=id econPersonalRetroW4 EXECUTE. wt_full_W1W2W3W4W5W6 econPersonalRetroW7 DO IF (ageGroup = 1 & gender W7W8W9 country trustMPsW1 trustMPsW2 RECODE country = 1 & anyUniW7 = 0). countryOfBirth trustMPsW3 trustMPsW4 (MISSING=SYSMIS) (3=1) RECODE demogCleave (0=2). profile_ethnicity gor trustMPsW6 (ELSE=0) INTO Wales. END IF. livedAbroadW8 trustMPsW7 VARIABLE LABELS Wales EXECUTE. profile_religion generalElectionVoteW5 '[theme: culture] [cluster] parentsForeignW8 likeCameronW1 Country is Wales as opposed DO IF (ageGroup = 1 & gender britishnessW1 britishnessW2 likeCameronW2 to England'. = 0 & anyUniW7 = 0). britishnessW3 likeCameronW3 EXECUTE. RECODE demogCleave (0=3). britishnessW4 likeCameronW4 END IF. britishnessW7 britishnessW8 likeCameronW5 RECODE profile_ethnicity EXECUTE. britishnessW9 ethno1W7 likeCameronW6 (1=1) (MISSING=SYSMIS) ethno6W7 ethno2W7 likeCameronW7 (ELSE=0). DO IF (ageGroup = 1 & gender ethno4W7 likeCameronW8 EXECUTE. = 0 & anyUniW7 = 1). ethno5W7 ethno3W7 likeCameronW9 likeConW7 RECODE demogCleave (0=4). radicalW7 immigCulturalW1 likeConW8 likeConW9 RECODE END IF. immigCulturalW2 ptvConW9 conGovTrustW5 profile_work_typeW7 EXECUTE. immigCulturalW3 businessBonusW1 (MISSING=SYSMIS) (1=1) (2=1) immigCulturalW4 businessBonusW2 (ELSE=0) INTO Upskilled. DO IF (ageGroup = 0 & gender immigCulturalW8 al1W6 businessBonusW3 VARIABLE LABELS Upskilled = 1 & anyUniW7 = 1). ageW7 businessBonusW4 '[theme: demographics] Work RECODE demogCleave (0=5). profile_newspaper_readershi govtHandoutsW1 type highly skilled as opposed END IF. p_201 disabilityW6 gender govtHandoutsW2 to never '+ EXECUTE. anyUniW7 govtHandoutsW3 'worked'. profile_work_typeW7 govtHandoutsW4 EXECUTE. DO IF (ageGroup = 0 & gender polAttentionW1 immigEconW1 immigEconW2 = 1 & anyUniW7 = 0). polAttentionW2 immigEconW3 RECODE RECODE demogCleave (0=6). polAttentionW3 immigEconW4 profile_work_typeW7 END IF. polAttentionW4 immigrationLevelW4 (MISSING=SYSMIS) (3=1) (4=1) EXECUTE. polAttentionW6 immigrationLevelW6 (5=1) (6=1) (ELSE=0) INTO polAttentionW7 immigrantsWelfareStateW1 Midskilled. DO IF (ageGroup = 0 & gender polAttentionW8 immigrantsWelfareStateW2 VARIABLE LABELS Midskilled = 0 & anyUniW7 = 0). discussPolDaysW2 immigrantsWelfareStateW3 '[theme: demographics] Work RECODE demogCleave (0=7). discussPolDaysW4 immigrantsWelfareStateW4 type middling skilled as END IF. discussPolDaysW5 immigrantsWelfareStateW8 opposed to '+ EXECUTE. discussPolDaysW6 reasonForUnemploymentW1 'never worked'. infoSourcePeopleW4 EXECUTE. DO IF (ageGroup = 0 & gender infoSourcePeopleW5 reasonForUnemploymentW2 = 0 & anyUniW7 = 1).

RECODE demogCleave (0=8). /ANALYSIS britishnessW1 /MISSING MEANSUB /EXTRACTION PC END IF. britishnessW2 britishnessW3 /ANALYSIS polAttentionW1 /CRITERIA ITERATE(25) EXECUTE. britishnessW4 britishnessW7 polAttentionW2 DELTA(0) britishnessW8 polAttentionW3 /ROTATION OBLIMIN SORT CASES BY demogCleave. britishnessW9 polAttentionW4 /SAVE REG(ALL) SPLIT FILE LAYERED BY /PRINT INITIAL EXTRACTION polAttentionW6 /METHOD=CORRELATION. demogCleave. ROTATION polAttentionW7 /PLOT EIGEN ROTATION polAttentionW8 RELIABILITY FREQUENCIES /CRITERIA MINEIGEN(1) /PRINT INITIAL EXTRACTION /VARIABLES=infoSourceIntern VARIABLES=businessBonusW1 ITERATE(25) ROTATION etW4 infoSourceInternetW5 f businessBonusW2f /EXTRACTION PC /PLOT EIGEN ROTATION infoSourceInternetW6 businessBonusW3f /CRITERIA ITERATE(25) /CRITERIA MINEIGEN(1) infoSourceInternetW7 businessBonusW4f DELTA(0) ITERATE(25) infoSourceInternetW8 govtHandoutsW1f /ROTATION OBLIMIN /EXTRACTION PC /SCALE('ALL VARIABLES') ALL govtHandoutsW2f /SAVE REG(ALL) /CRITERIA ITERATE(25) /MODEL=ALPHA govtHandoutsW3f /METHOD=CORRELATION. DELTA(0) /SUMMARY=TOTAL. govtHandoutsW4f /ROTATION OBLIMIN immigEconW1f RECODE immigCulturalW1 /SAVE REG(ALL) FACTOR immigEconW2f immigCulturalW2 /METHOD=CORRELATION. /VARIABLES immigEconW3f immigCulturalW3 infoSourceInternetW4 immigEconW4f immigCulturalW4 RELIABILITY infoSourceInternetW5 immigrationLevelW4f immigCulturalW8 (1=7) (2=6) /VARIABLES=discussPolDaysW infoSourceInternetW6 immigrationLevelW6f (3=5) (4=4) (5=3) (6=2) 2 discussPolDaysW4 infoSourceInternetW7 immigrantsWelfareStateW1f (7=1). discussPolDaysW5 infoSourceInternetW8 EXECUTE. discussPolDaysW6 /MISSING MEANSUB immigrantsWelfareStateW2f /SCALE('ALL VARIABLES') ALL /ANALYSIS immigrantsWelfareStateW3f RELIABILITY /MODEL=ALPHA infoSourceInternetW4 immigrantsWelfareStateW4f /VARIABLES=immigCulturalW1 /SUMMARY=TOTAL. infoSourceInternetW5 immigCulturalW2 infoSourceInternetW6 immigrantsWelfareStateW8f immigCulturalW3 FACTOR infoSourceInternetW7 reasonForUnemploymentW1f immigCulturalW4 /VARIABLES infoSourceInternetW8 reasonForUnemploymentW2f immigCulturalW8 discussPolDaysW2 /PRINT INITIAL EXTRACTION /SCALE('ALL VARIABLES') ALL discussPolDaysW4 ROTATION reasonForUnemploymentW3f /MODEL=ALPHA discussPolDaysW5 /PLOT EIGEN ROTATION reasonForUnemploymentW4f /SUMMARY=TOTAL. discussPolDaysW6 /CRITERIA MINEIGEN(1) al4W6f lr1W6f al5W6f al3W6f /MISSING MEANSUB ITERATE(25) /FORMAT=NOTABLE FACTOR /ANALYSIS discussPolDaysW2 /EXTRACTION PC /STATISTICS=MEDIAN /VARIABLES discussPolDaysW4 /CRITERIA ITERATE(25) /ORDER=ANALYSIS. immigCulturalW1 discussPolDaysW5 DELTA(0) immigCulturalW2 discussPolDaysW6 /ROTATION OBLIMIN DO IF (ageGroup = 0 & gender immigCulturalW3 /PRINT INITIAL EXTRACTION /SAVE REG(ALL) = 0 & anyUniW7 = 1). immigCulturalW4 ROTATION /METHOD=CORRELATION. RECODE gor immigCulturalW8 /PLOT EIGEN ROTATION (MISSING=SYSMIS) (7=3) /MISSING MEANSUB /CRITERIA MINEIGEN(1) RELIABILITY (10=5) (11=6) (12=7) (1 thru /ANALYSIS immigCulturalW1 ITERATE(25) /VARIABLES=infoSourcePaper 3=1) (4 thru 6=2) (8 thru 9=4). immigCulturalW2 /EXTRACTION PC W4 infoSourcePaperW5 END IF. immigCulturalW3 /CRITERIA ITERATE(25) infoSourcePaperW6 EXECUTE. immigCulturalW4 DELTA(0) infoSourcePaperW7 immigCulturalW8 /ROTATION OBLIMIN infoSourcePaperW8 SPLIT FILE OFF. /PRINT INITIAL EXTRACTION /SAVE REG(ALL) /SCALE('ALL VARIABLES') ALL ROTATION /METHOD=CORRELATION. /MODEL=ALPHA RECODE ethno1W7 (1=5) /PLOT EIGEN ROTATION /SUMMARY=TOTAL. (2=4) (3=3) (4=2) (5=1). /CRITERIA MINEIGEN(1) RELIABILITY EXECUTE. ITERATE(25) /VARIABLES=infoSourcePeopl FACTOR /EXTRACTION PC eW4 infoSourcePeopleW5 /VARIABLES RECODE ethno6W7 (1=5) /CRITERIA ITERATE(25) infoSourcePeopleW6 infoSourcePaperW4 (2=4) (3=3) (4=2) (5=1). DELTA(0) infoSourcePeopleW7 infoSourcePaperW5 EXECUTE. /ROTATION OBLIMIN infoSourcePeopleW8 infoSourcePaperW6 /SAVE REG(ALL) /SCALE('ALL VARIABLES') ALL infoSourcePaperW7 RECODE ethno3W7 radicalW7 /METHOD=CORRELATION. /MODEL=ALPHA infoSourcePaperW8 (1=5) (2=4) (3=3) (4=2) (5=1). /SUMMARY=TOTAL. /MISSING MEANSUB EXECUTE. RELIABILITY /ANALYSIS /VARIABLES=polAttentionW1 FACTOR infoSourcePaperW4 RELIABILITY polAttentionW2 /VARIABLES infoSourcePaperW5 /VARIABLES=britishnessW1 polAttentionW3 infoSourcePeopleW4 infoSourcePaperW6 britishnessW2 britishnessW3 polAttentionW4 infoSourcePeopleW5 infoSourcePaperW7 britishnessW4 britishnessW7 polAttentionW6 infoSourcePeopleW6 infoSourcePaperW8 britishnessW8 polAttentionW7 infoSourcePeopleW7 /PRINT INITIAL EXTRACTION britishnessW9 polAttentionW8 infoSourcePeopleW8 ROTATION /SCALE('ALL VARIABLES') ALL /SCALE('ALL VARIABLES') ALL /MISSING MEANSUB /PLOT EIGEN ROTATION /MODEL=ALPHA /MODEL=ALPHA /ANALYSIS /CRITERIA MINEIGEN(1) /SUMMARY=TOTAL. /SUMMARY=TOTAL. infoSourcePeopleW4 ITERATE(25) infoSourcePeopleW5 /EXTRACTION PC FACTOR FACTOR infoSourcePeopleW6 /CRITERIA ITERATE(25) /VARIABLES britishnessW1 /VARIABLES polAttentionW1 infoSourcePeopleW7 DELTA(0) britishnessW2 britishnessW3 polAttentionW2 infoSourcePeopleW8 /ROTATION OBLIMIN britishnessW4 britishnessW7 polAttentionW3 /PRINT INITIAL EXTRACTION /SAVE REG(ALL) britishnessW8 polAttentionW4 ROTATION /METHOD=CORRELATION. britishnessW9 polAttentionW6 /PLOT EIGEN ROTATION /MISSING MEANSUB polAttentionW7 /CRITERIA MINEIGEN(1) RELIABILITY polAttentionW8 ITERATE(25)

/VARIABLES=infoSourceRadio electionInterestW5 /CRITERIA ITERATE(25) econPersonalRetroW3 W4 infoSourceRadioW5 electionInterestW6 DELTA(0) econPersonalRetroW4 infoSourceRadioW6 /MISSING MEANSUB /ROTATION OBLIMIN econPersonalRetroW7 infoSourceRadioW7 /ANALYSIS /SAVE REG(ALL) /MISSING MEANSUB infoSourceRadioW8 electionInterestW4 /METHOD=CORRELATION. /ANALYSIS /SCALE('ALL VARIABLES') ALL electionInterestW5 econPersonalRetroW1 /MODEL=ALPHA electionInterestW6 CORRELATIONS econPersonalRetroW2 /SUMMARY=TOTAL. /PRINT INITIAL EXTRACTION /VARIABLES=FAC1_11 econPersonalRetroW3 ROTATION FAC2_11 econPersonalRetroW4 FACTOR /PLOT EIGEN ROTATION /PRINT=TWOTAIL NOSIG econPersonalRetroW7 /VARIABLES /CRITERIA MINEIGEN(1) /MISSING=PAIRWISE. /PRINT INITIAL EXTRACTION infoSourceRadioW4 ITERATE(25) ROTATION infoSourceRadioW5 /EXTRACTION PC FACTOR /PLOT EIGEN ROTATION infoSourceRadioW6 /CRITERIA ITERATE(25) /VARIABLES FAC1_11 /CRITERIA MINEIGEN(1) infoSourceRadioW7 DELTA(0) FAC2_11 ITERATE(25) infoSourceRadioW8 /ROTATION OBLIMIN /MISSING MEANSUB /EXTRACTION PC /MISSING MEANSUB /SAVE REG(ALL) /ANALYSIS FAC1_11 FAC2_11 /CRITERIA ITERATE(25) /ANALYSIS /METHOD=CORRELATION. /PRINT INITIAL EXTRACTION DELTA(0) infoSourceRadioW4 ROTATION /ROTATION OBLIMIN infoSourceRadioW5 RELIABILITY /PLOT EIGEN ROTATION /SAVE REG(ALL) infoSourceRadioW6 /VARIABLES=turnoutUKGener /CRITERIA MINEIGEN(1) /METHOD=CORRELATION. infoSourceRadioW7 alW1 turnoutUKGeneralW2 ITERATE(25) infoSourceRadioW8 turnoutUKGeneralW3 /EXTRACTION PC RELIABILITY /PRINT INITIAL EXTRACTION turnoutUKGeneralW4 /CRITERIA ITERATE(25) /VARIABLES=trustMPsW1 ROTATION /SCALE('ALL VARIABLES') ALL DELTA(0) trustMPsW2 trustMPsW3 /PLOT EIGEN ROTATION /MODEL=ALPHA /ROTATION OBLIMIN trustMPsW4 trustMPsW6 /CRITERIA MINEIGEN(1) /SUMMARY=TOTAL. /SAVE REG(ALL) trustMPsW7 ITERATE(25) /METHOD=CORRELATION. /SCALE('ALL VARIABLES') ALL /EXTRACTION PC FACTOR /MODEL=ALPHA /CRITERIA ITERATE(25) /VARIABLES GRAPH /SUMMARY=TOTAL. DELTA(0) turnoutUKGeneralW1 /HISTOGRAM(NORMAL)=FAC1 /ROTATION OBLIMIN turnoutUKGeneralW2 _13. FACTOR /SAVE REG(ALL) turnoutUKGeneralW3 /VARIABLES trustMPsW1 /METHOD=CORRELATION. turnoutUKGeneralW4 CORRELATIONS trustMPsW2 trustMPsW3 /MISSING MEANSUB /VARIABLES=FAC1_1 FAC1_2 trustMPsW4 trustMPsW6 RELIABILITY /ANALYSIS al1W6 trustMPsW7 /VARIABLES=infoSourceTVW4 turnoutUKGeneralW1 /PRINT=TWOTAIL NOSIG /MISSING MEANSUB infoSourceTVW5 turnoutUKGeneralW2 /MISSING=PAIRWISE. /ANALYSIS trustMPsW1 infoSourceTVW6 turnoutUKGeneralW3 trustMPsW2 trustMPsW3 infoSourceTVW7 turnoutUKGeneralW4 CORRELATIONS trustMPsW4 trustMPsW6 infoSourceTVW8 /PRINT INITIAL EXTRACTION /VARIABLES=FAC1_1 al1W6 trustMPsW7 /SCALE('ALL VARIABLES') ALL ROTATION FAC1_2 /PRINT INITIAL EXTRACTION /MODEL=ALPHA /PLOT EIGEN ROTATION /PRINT=TWOTAIL NOSIG ROTATION /SUMMARY=TOTAL. /CRITERIA MINEIGEN(1) /MISSING=PAIRWISE. /PLOT EIGEN ROTATION ITERATE(25) /CRITERIA MINEIGEN(1) FACTOR /EXTRACTION PC FACTOR ITERATE(25) /VARIABLES infoSourceTVW4 /CRITERIA ITERATE(25) /VARIABLES FAC1_2 al1W6 /EXTRACTION PC infoSourceTVW5 DELTA(0) /MISSING MEANSUB /CRITERIA ITERATE(25) infoSourceTVW6 /ROTATION OBLIMIN /ANALYSIS FAC1_2 al1W6 DELTA(0) infoSourceTVW7 /SAVE REG(ALL) /PRINT INITIAL EXTRACTION /ROTATION OBLIMIN infoSourceTVW8 /METHOD=CORRELATION. ROTATION /SAVE REG(ALL) /MISSING MEANSUB /PLOT EIGEN ROTATION /METHOD=CORRELATION. /ANALYSIS infoSourceTVW4 RELIABILITY /CRITERIA MINEIGEN(1) infoSourceTVW5 /VARIABLES=FAC1_3 FAC1_4 ITERATE(25) RELIABILITY infoSourceTVW6 FAC1_5 FAC1_6 FAC1_7 /EXTRACTION PC /VARIABLES=likeCameronW1 infoSourceTVW7 FAC1_8 FAC1_9 FAC1_10 /CRITERIA ITERATE(25) likeCameronW2 infoSourceTVW8 FAC1_12 euRefInterestW7 DELTA(0) likeCameronW3 /PRINT INITIAL EXTRACTION euRefTurnoutW7 /ROTATION OBLIMIN likeCameronW4 ROTATION /SCALE('ALL VARIABLES') ALL /SAVE REG(ALL) likeCameronW5 /PLOT EIGEN ROTATION /MODEL=ALPHA /METHOD=CORRELATION. likeCameronW6 /CRITERIA MINEIGEN(1) /SUMMARY=TOTAL. likeCameronW7 ITERATE(25) CORRELATIONS likeCameronW8 /EXTRACTION PC FACTOR /VARIABLES=FAC1_1 likeCameronW9 /CRITERIA ITERATE(25) /VARIABLES FAC1_3 FAC1_4 FAC1_14 /SCALE('ALL VARIABLES') ALL DELTA(0) FAC1_5 FAC1_6 FAC1_7 /PRINT=TWOTAIL NOSIG /MODEL=ALPHA /ROTATION OBLIMIN FAC1_8 FAC1_9 FAC1_10 /MISSING=PAIRWISE. /SUMMARY=TOTAL. /SAVE REG(ALL) FAC1_12 euRefInterestW7 /METHOD=CORRELATION. euRefTurnoutW7 RELIABILIT FACTOR /MISSING MEANSUB /VARIABLES=econPersonalRet /VARIABLES likeCameronW1 RELIABILITY /ANALYSIS FAC1_3 FAC1_4 roW1 econPersonalRetroW2 likeCameronW2 /VARIABLES=electionInterest FAC1_5 FAC1_6 FAC1_7 econPersonalRetroW3 likeCameronW3 W4 electionInterestW5 FAC1_8 FAC1_9 FAC1_10 econPersonalRetroW4 likeCameronW4 electionInterestW6 FAC1_12 euRefInterestW7 econPersonalRetroW7 likeCameronW5 /SCALE('ALL VARIABLES') ALL euRefTurnoutW7 /SCALE('ALL VARIABLES') ALL likeCameronW6 /MODEL=ALPHA /PRINT INITIAL EXTRACTION /MODEL=ALPHA likeCameronW7 /SUMMARY=TOTAL. ROTATION /SUMMARY=TOTAL. likeCameronW8 /PLOT EIGEN ROTATION likeCameronW9 FACTOR /CRITERIA MINEIGEN(1) FACTOR /MISSING MEANSUB /VARIABLES ITERATE(25) /VARIABLES /ANALYSIS likeCameronW1 electionInterestW4 /EXTRACTION PC econPersonalRetroW1 likeCameronW2 econPersonalRetroW2 likeCameronW3

79 likeCameronW4 grnFearW6 labFearW4 /VARIABLES ethno1W7 likeCameronW5 FACTOR labFearW6 ldFearW4 ethno6W7 ethno2W7 likeCameronW6 /VARIABLES aom1W7 ldFearW6 ethno4W7 ethno5W7 likeCameronW7 aom4W7 aom5W7 aom6W7 ukipFearW4 ukipFearW6 ethno3W7 radicalW7 likeCameronW8 aom7W7 aom3W7 aom2W7 /SCALE('ALL VARIABLES') ALL /MISSING MEANSUB likeCameronW9 /MISSING MEANSUB /MODEL=ALPHA /ANALYSIS ethno1W7 /PRINT INITIAL EXTRACTION /ANALYSIS aom1W7 /SUMMARY=TOTAL. ethno6W7 ethno2W7 ROTATION aom4W7 aom5W7 aom6W7 ethno4W7 ethno5W7 /PLOT EIGEN ROTATION aom7W7 aom3W7 aom2W7 RELIABILITY ethno3W7 radicalW7 /CRITERIA MINEIGEN(1) /PRINT INITIAL EXTRACTION /VARIABLES=conAngryW4 /PRINT INITIAL EXTRACTION ITERATE(25) ROTATION conAngryW6 grnAngryW4 ROTATION /EXTRACTION PC /PLOT EIGEN ROTATION grnAngryW6 labAngryW4 /PLOT EIGEN ROTATION /CRITERIA ITERATE(25) /CRITERIA MINEIGEN(1) labAngryW6 ldAngryW4 /CRITERIA MINEIGEN(1) DELTA(0) ITERATE(25) ldAngryW6 ITERATE(25) /ROTATION OBLIMIN /EXTRACTION PC ukipAngryW4 ukipAngryW6 /EXTRACTION PC /SAVE REG(ALL) /CRITERIA ITERATE(25) /SCALE('ALL VARIABLES') ALL /CRITERIA ITERATE(25) /METHOD=CORRELATION. DELTA(0) /MODEL=ALPHA DELTA(0) /ROTATION OBLIMIN /SUMMARY=TOTAL. /ROTATION OBLIMIN RELIABILITY /SAVE REG(ALL) /SAVE REG(ALL) /VARIABLES=likeConW7 /METHOD=CORRELATION. COMPUTE /METHOD=CORRELATION. likeConW8 likeConW9 angerScale=(conAngryW4 + /SCALE('ALL VARIABLES') ALL RECODE tolUncertain1W8 conAngryW6 + grnAngryW4 + CORRELATIONS /MODEL=ALPHA tolUncertain3W8 grnAngryW6 + labAngryW4 + /VARIABLES=ethno1W7 /SUMMARY=TOTAL. tolUncertain2W8 (1=5) (2=4) labAngryW6 + ethno6W7 ethno2W7 (3=3) (4=2) (5=1). ldAngryW4 + ldAngryW6 + ethno4W7 ethno5W7 FACTOR EXECUTE. ukipAngryW4 + ethno3W7 radicalW7 /VARIABLES likeConW7 ukipAngryW6)/10. /PRINT=TWOTAIL NOSIG likeConW8 likeConW9 RELIABILITY VARIABLE LABELS angerScale /MISSING=PAIRWISE. /MISSING MEANSUB /VARIABLES=riskTakingW7 '[theme: psychology] /ANALYSIS likeConW7 riskTakingW8 [emotion] Proportional CORRELATIONS likeConW8 likeConW9 tolUncertain1W8 additive scale for '+ /VARIABLES=FAC1_22 /PRINT INITIAL EXTRACTION tolUncertain3W8 'feeling anger when FAC2_22 ROTATION tolUncertain2W8 thinking about political /PRINT=TWOTAIL NOSIG /PLOT EIGEN ROTATION /SCALE('ALL VARIABLES') ALL parties'. /MISSING=PAIRWISE. /CRITERIA MINEIGEN(1) /MODEL=ALPHA EXECUTE. ITERATE(25) /SUMMARY=TOTAL. NONPAR CORR /EXTRACTION PC FREQUENCIES /VARIABLES=FAC2_22 /CRITERIA ITERATE(25) FACTOR VARIABLES=angerScale ethno3W7 DELTA(0) /VARIABLES riskTakingW7 /FORMAT=NOTABLE /PRINT=SPEARMAN /ROTATION OBLIMIN riskTakingW8 /NTILES=4 TWOTAIL NOSIG /SAVE REG(ALL) tolUncertain1W8 /STATISTICS=STDDEV MEAN /MISSING=PAIRWISE. /METHOD=CORRELATION. tolUncertain3W8 MEDIAN SKEWNESS SESKEW tolUncertain2W8 KURTOSIS SEKURT COMPUTE RELIABILITY /MISSING MEANSUB /HISTOGRAM NORMAL EthnoProud=FAC2_22 * - 1. /VARIABLES=FAC1_17 /ANALYSIS riskTakingW7 /ORDER=ANALYSIS. VARIABLE LABELS EthnoProud FAC1_18 ptvConW9 riskTakingW8 '[theme: culture] [ethno] conGovTrustW5 tolUncertain1W8 COMPUTE Ethno factor for pride in /SCALE('ALL VARIABLES') ALL tolUncertain3W8 fearScale=(conFearW4 + British identity '+ /MODEL=ALPHA tolUncertain2W8 conFearW6 + grnFearW4 + 'created by inverting /SUMMARY=TOTAL. /PRINT INITIAL EXTRACTION grnFearW6 + labFearW4 + FAC2_22 by multiplying it by - ROTATION labFearW6 + ldFearW4 1'. FACTOR /PLOT EIGEN ROTATION + ldFearW6 + ukipFearW4 + EXECUTE. /VARIABLES FAC1_17 /CRITERIA MINEIGEN(1) ukipFearW6)/10. FAC1_18 conGovTrustW5 ITERATE(25) VARIABLE LABELS fearScale RELIABILITY /MISSING MEANSUB /EXTRACTION PC '[theme: psychology] /VARIABLES=FAC1_1 /ANALYSIS FAC1_17 FAC1_18 /CRITERIA ITERATE(25) [emotion] Proportional FAC1_22 EthnoProud conGovTrustW5 DELTA(0) additive scale for '+ FAC1_14 /PRINT INITIAL EXTRACTION /ROTATION OBLIMIN 'feeling fear when thinking /SCALE('ALL VARIABLES') ALL ROTATION /SAVE REG(ALL) about political parties'. /MODEL=ALPHA /PLOT EIGEN ROTATION /METHOD=CORRELATION. EXECUTE. /SUMMARY=TOTAL. /CRITERIA MINEIGEN(1) ITERATE(25) CORRELATIONS FREQUENCIES FACTOR /EXTRACTION PC /VARIABLES=FAC1_21 VARIABLES=fearScale /VARIABLES FAC1_1 /CRITERIA ITERATE(25) FAC2_21 /FORMAT=NOTABLE FAC1_22 FAC1_14 DELTA(0) /PRINT=TWOTAIL NOSIG /NTILES=4 /MISSING MEANSUB /ROTATION OBLIMIN /MISSING=PAIRWISE. /STATISTICS=STDDEV MEAN /ANALYSIS FAC1_1 FAC1_22 /SAVE REG(ALL) MEDIAN SKEWNESS SESKEW FAC1_14 /METHOD=CORRELATION. CORRELATIONS KURTOSIS SEKURT /PRINT INITIAL EXTRACTION /VARIABLES=FAC1_20 /HISTOGRAM NORMAL ROTATION RECODE aom4W7 aom5W7 FAC1_21 FAC2_21 /ORDER=ANALYSIS. /PLOT EIGEN ROTATION aom6W7 aom7W7 (1=5) (2=4) personality_agreeableness /CRITERIA MINEIGEN(1) (3=3) (4=2) (5=1). personality_conscientiousness RELIABILITY ITERATE(25) EXECUTE. personality_extraversion /VARIABLES=ethno1W7 /EXTRACTION PC personality_neuroticism ethno6W7 ethno2W7 /CRITERIA ITERATE(25) RELIABILITY personality_openness ethno4W7 ethno5W7 DELTA(0) /VARIABLES=aom1W7 /PRINT=TWOTAIL NOSIG ethno3W7 radicalW7 /ROTATION OBLIMIN aom4W7 aom5W7 aom6W7 /MISSING=PAIRWISE. /SCALE('ALL VARIABLES') ALL /SAVE REG(ALL) aom7W7 aom3W7 aom2W7 /MODEL=ALPHA /METHOD=CORRELATION. /SCALE('ALL VARIABLES') ALL RELIABILITY /SUMMARY=TOTAL. /MODEL=ALPHA /VARIABLES=conFearW4 COMPUTE /SUMMARY=TOTAL. conFearW6 grnFearW4 FACTOR ItemResponseNumber=NVALI

D(profile_ethnicity, gor, likeCameronW4, FREQUENCIES RECODE britishnessW1 livedAbroadW8, likeCameronW5, VARIABLES=ItemResponseRat britishnessW2 britishnessW3 profile_religion, likeCameronW6, e britishnessW4 britishnessW7 parentsForeignW8, likeCameronW7, /FORMAT=NOTABLE britishnessW8 britishnessW1, britishnessW2, likeCameronW8, /STATISTICS=MAXIMUM britishnessW9 britishnessW3, britishnessW4, likeCameronW9, likeConW7, /HISTOGRAM NORMAL immigCulturalW1 britishnessW7, likeConW8, likeConW9, /ORDER=ANALYSIS. immigCulturalW2 britishnessW8, ptvConW9, immigCulturalW3 britishnessW9, ethno1W7, conGovTrustW5, COMPUTE immigCulturalW4 ethno6W7, ethno2W7, businessBonusW1, ReflectItemResponseRate=10 immigCulturalW8 ethno4W7, ethno5W7, businessBonusW2, 0 - ItemResponseRate. trustMPsW1 trustMPsW2 ethno3W7, businessBonusW3, VARIABLE LABELS trustMPsW3 trustMPsW4 radicalW7, businessBonusW4, ReflectItemResponseRate trustMPsW6 trustMPsW7 immigCulturalW1, govtHandoutsW1, '[theme: psychology] conGovTrustW5 immigCulturalW2, govtHandoutsW2, Reflected item-level response immigEconW1 immigCulturalW3, govtHandoutsW3, rate'. immigEconW2 immigCulturalW4, govtHandoutsW4, EXECUTE. immigEconW3 immigEconW4 immigCulturalW8, immigEconW1, (4=0) (3=1) (5=1) (2=2) (6=2) al1W6, ageW7, immigEconW2, FREQUENCIES (1=3) (7=3) (MISSING=SYSMIS) profile_newspaper_readershi immigEconW3, VARIABLES=ReflectItemRespo INTO p_201, disabilityW6, immigEconW4, nseRate Likert1 Likert2 Likert3 anyUniW7, immigrationLevelW4, /FORMAT=NOTABLE Likert4 Likert5 Likert7 Likert8 profile_work_typeW7, immigrationLevelW6, /STATISTICS=MAXIMUM Likert9 Likert10 Likert11 polAttentionW1, immigrantsWelfareStateW1, /HISTOGRAM NORMAL Likert12 Likert13 polAttentionW2, /ORDER=ANALYSIS. Likert14 Likert15 Likert16 polAttentionW3, immigrantsWelfareStateW2, Likert17 Likert18 Likert19 polAttentionW4, immigrantsWelfareStateW3, COMPUTE Likert20 Likert21 Likert22 polAttentionW6, immigrantsWelfareStateW4, LnReflectedItemResponseRat Likert23 Likert24. polAttentionW7, e=LN(ReflectItemResponseRat VARIABLE LABELS Likert9 '7- polAttentionW8, immigrantsWelfareStateW8, e). point, culture' /Likert10 '7- discussPolDaysW2, reasonForUnemploymentW1, VARIABLE LABELS point, culture' /Likert11 '7- discussPolDaysW4, reasonForUnemploymentW2, LnReflectedItemResponseRat point, '+ discussPolDaysW5, e '[theme: psychology] 'culture' /Likert12 '7-point, discussPolDaysW6, reasonForUnemploymentW3, Natural logarithm of reflected culture' /Likert13 '7-point, infoSourcePeopleW4, reasonForUnemploymentW4, '+ culture' /Likert14 '7-point, '+ infoSourcePeopleW5, al4W6, lr1W6, al5W6, al3W6, 'item-level response rate'. 'general politics' /Likert15 infoSourcePeopleW6, aom1W7, aom4W7, EXECUTE. '7-point, general politics' infoSourcePeopleW7, aom5W7, aom6W7, /Likert16 '7-point, general infoSourcePeopleW8, aom7W7, aom3W7, aom2W7, FREQUENCIES politics' infoSourceInternetW4, conAngryW4, conAngryW6, VARIABLES=LnReflectedItemR /Likert17 '7-point, general infoSourceInternetW5, grnAngryW4, grnAngryW6, esponseRate politics' /Likert18 '7-point, infoSourceInternetW6, labAngryW4, /FORMAT=NOTABLE general politics' /Likert19 '7- infoSourceInternetW7, labAngryW6, ldAngryW4, /STATISTICS=MAXIMUM point, '+ infoSourceInternetW8, ldAngryW6, ukipAngryW4, SKEWNESS SESKEW KURTOSIS 'general politics' /Likert20 infoSourcePaperW4, ukipAngryW6, conFearW4, SEKURT '7-point, partisanship' infoSourcePaperW5, conFearW6, grnFearW4, /HISTOGRAM NORMAL /Likert21 '7-point, policy' infoSourcePaperW6, grnFearW6, labFearW4, /ORDER=ANALYSIS. /Likert22 infoSourcePaperW7, labFearW6, ldFearW4, '7-point, policy' /Likert23 '7- infoSourcePaperW8, ldFearW6, ukipFearW4, NONPAR CORR point, policy' /Likert24 '7- infoSourceRadioW4, ukipFearW6, riskTakingW7, /VARIABLES=LnReflectedItem point, policy'. infoSourceRadioW5, riskTakingW8, ResponseRate EXECUTE. infoSourceRadioW6, tolUncertain1W8, ItemResponseRate infoSourceRadioW7, tolUncertain3W8, /PRINT=SPEARMAN RECODE ethno1W7 infoSourceRadioW8, tolUncertain2W8). TWOTAIL NOSIG ethno6W7 ethno2W7 infoSourceTVW4, VARIABLE LABELS /MISSING=PAIRWISE. ethno4W7 ethno5W7 infoSourceTVW5, ItemResponseNumber ethno3W7 radicalW7 al1W6 infoSourceTVW6, 'Number of items with valid COMPUTE euRefTurnoutW7 infoSourceTVW7, responses, out of 160'. ReflectedLnReflectedItemRes turnoutUKGeneralW1 infoSourceTVW8, EXECUTE. ponseRate=4.56 - turnoutUKGeneralW2 electionInterestW4, LnReflectedItemResponseRat turnoutUKGeneralW3 electionInterestW5, COMPUTE e. turnoutUKGeneralW4 electionInterestW6, ItemResponseRate=(ItemResp VARIABLE LABELS econPersonalRetroW1 euRefInterestW7, onseNumber / 160)*100. ReflectedLnReflectedItemRes econPersonalRetroW2 euRefTurnoutW7, VARIABLE LABELS ponseRate '[theme: econPersonalRetroW3 turnoutUKGeneralW1, ItemResponseRate '[theme: psychology] Reflected natural econPersonalRetroW4 turnoutUKGeneralW2, psychology] Item-level '+ econPersonalRetroW7 turnoutUKGeneralW3, response rate'. 'logarithm of reflected businessBonusW1 turnoutUKGeneralW4, EXECUTE. item-level response rate'. businessBonusW2 econPersonalRetroW1, EXECUTE. businessBonusW3 econPersonalRetroW2, FREQUENCIES businessBonusW4 econPersonalRetroW3, VARIABLES=ItemResponseRat FREQUENCIES govtHandoutsW1 econPersonalRetroW4, e VARIABLES=ReflectedLnReflec govtHandoutsW2 econPersonalRetroW7, /FORMAT=NOTABLE tedItemResponseRate govtHandoutsW3 trustMPsW1, trustMPsW2, /NTILES=4 /FORMAT=NOTABLE govtHandoutsW4 trustMPsW3, trustMPsW4, /STATISTICS=STDDEV MEAN /STATISTICS=MAXIMUM immigrationLevelW4 trustMPsW6, trustMPsW7, MEDIAN SKEWNESS SESKEW SKEWNESS SESKEW KURTOSIS immigrationLevelW6 generalElectionVoteW5, KURTOSIS SEKURT SEKURT immigrantsWelfareStateW1 likeCameronW1, /HISTOGRAM NORMAL /HISTOGRAM NORMAL immigrantsWelfareStateW2 likeCameronW2, /ORDER=ANALYSIS. /ORDER=ANALYSIS. immigrantsWelfareStateW3 likeCameronW3, immigrantsWelfareStateW4 immigrantsWelfareStateW8

/Likert65 '5-point, policy' euRefInterestW7 Likert8 Likert9 Likert10 reasonForUnemploymentW1 /Likert66 '5-point, psychology' riskTakingW7 Likert11 reasonForUnemploymentW2 /Likert67 '5-point, psychology' riskTakingW8 (2=1) (3=1) Likert12 Likert13 Likert25 reasonForUnemploymentW3 /Likert68 '5-point, (1=2) (4=1) (MISSING=SYSMIS) Likert26 Likert27 Likert28 reasonForUnemploymentW4 psychology' /Likert69 '5-point, INTO Likert96 Likert97 Likert29 Likert30 Likert31 al4W6 lr1W6 al5W6 al3W6 psychology' /Likert70 '5-point, Likert98 Likert99 Likert32 aom1W7 aom4W7 aom5W7 psychology' Likert100 Likert101. /SCALE('ALL VARIABLES') ALL aom6W7 aom7W7 aom3W7 /Likert71 '5-point, VARIABLE LABELS Likert96 '4- /MODEL=ALPHA aom2W7 tolUncertain1W8 psychology' /Likert72 '5-point, point, engagement' /Likert97 /SUMMARY=TOTAL. tolUncertain3W8 psychology' /Likert73 '5-point, '4-point, engagement' tolUncertain2W8 (3=0) (2=1) psychology' /Likert98 RELIABILITY (4=1) (1=2) (5=2) /Likert74 '5-point, '4-point, engagement' /VARIABLES=Likert14 (MISSING=SYSMIS) INTO psychology' /Likert75 '5-point, /Likert99 '4-point, Likert15 Likert16 Likert17 Likert25 psychology'. engagement' /Likert100 '4- Likert18 Likert19 Likert26 Likert27 Likert28 EXECUTE. point, psychology' /SCALE('ALL VARIABLES') ALL Likert29 Likert30 Likert31 /Likert101 '4-point, /MODEL=ALPHA Likert32 Likert33 Likert34 RECODE polAttentionW1 psychology'. /SUMMARY=TOTAL. Likert35 Likert36 polAttentionW2 EXECUTE. Likert37 Likert38 Likert39 polAttentionW3 RELIABILITY Likert40 Likert41 Likert42 polAttentionW4 RELIABILITY /VARIABLES=Likert33 Likert43 Likert44 Likert45 polAttentionW6 /VARIABLES=Likert96 Likert34 Likert35 Likert36 Likert46 Likert47 polAttentionW7 Likert97 Likert98 Likert99 Likert37 Likert38 Likert39 Likert48 Likert49 Likert50 polAttentionW8 Likert100 Likert101 Likert40 Likert41 Likert51 Likert52 Likert53 likeCameronW1 /SCALE('ALL VARIABLES') ALL Likert42 Likert76 Likert77 Likert54 Likert55 Likert56 likeCameronW2 /MODEL=ALPHA Likert78 Likert79 Likert80 Likert57 Likert58 likeCameronW3 /SUMMARY=TOTAL. Likert81 Likert82 Likert96 Likert59 Likert60 Likert61 likeCameronW4 Likert97 Likert98 Likert62 Likert63 Likert64 likeCameronW5 RELIABILITY Likert99 Likert65 Likert66 Likert67 likeCameronW6 /VARIABLES=Likert25 /SCALE('ALL VARIABLES') ALL Likert68 Likert69 likeCameronW7 Likert26 Likert27 Likert28 /MODEL=ALPHA Likert70 Likert71 Likert72 likeCameronW8 Likert29 Likert30 Likert31 /SUMMARY=TOTAL. Likert73 Likert74 Likert75. likeCameronW9 likeConW7 Likert32 Likert33 VARIABLE LABELS Likert25 '5- likeConW8 likeConW9 Likert34 Likert35 Likert36 RELIABILITY point, culture' /Likert26 '5- ptvConW9 (5=0) (4=1) (6=1) Likert37 Likert38 Likert39 /VARIABLES=Likert20 point, culture' /Likert27 '5- (3=2) (7=2) (2=3) (8=3) Likert40 Likert41 Likert42 Likert83 Likert84 Likert85 point, '+ (1=4) (9=4) (0=5) (10=5) Likert43 Likert44 Likert86 Likert87 Likert88 'culture' /Likert28 '5-point, (MISSING=SYSMIS) INTO Likert45 Likert46 Likert47 Likert89 Likert90 culture' /Likert31 '5-point, Likert76 Likert77 Likert78 Likert48 Likert49 Likert50 Likert91 Likert92 Likert93 culture' /Likert32 '5-point, Likert79 Likert80 Likert81 Likert51 Likert52 Likert53 Likert94 Likert95 culture' Likert82 Likert83 Likert84 Likert54 Likert55 /SCALE('ALL VARIABLES') ALL /Likert33 '5-point, Likert85 Likert86 Likert87 Likert56 Likert57 Likert58 /MODEL=ALPHA engagement' /Likert34 '5- Likert88 Likert89 Likert59 Likert60 Likert61 /SUMMARY=TOTAL. point, engagement' /Likert35 Likert90 Likert91 Likert92 Likert62 Likert63 Likert64 '5-point, engagement' Likert93 Likert94 Likert95. Likert65 Likert66 RELIABILITY /Likert36 '5-point, VARIABLE LABELS Likert76 Likert67 Likert68 Likert69 /VARIABLES=Likert21 engagement' /Likert37 '5- '11-point, engagement' Likert70 Likert71 Likert72 Likert22 Likert23 Likert24 point, engagement' /Likert38 /Likert77 '11-point, Likert73 Likert74 Likert75 Likert43 Likert44 Likert45 '5-point, engagement' engagement' /Likert78 /SCALE('ALL VARIABLES') ALL Likert46 Likert47 /Likert39 '5-point, '11-point, engagement' /MODEL=ALPHA Likert48 Likert49 Likert50 engagement' /Likert40 '5- /Likert79 '11-point, /SUMMARY=TOTAL. Likert51 Likert52 Likert53 point, engagement' /Likert41 engagement' /Likert80 '11- Likert54 Likert55 Likert56 '5-point, engagement' point, engagement' RELIABILITY Likert57 Likert58 /Likert42 '5-point, /Likert81 '11-point, /VARIABLES=Likert1 Likert2 Likert59 Likert60 Likert61 engagement' /Likert43 '5- engagement' /Likert82 '11- Likert3 Likert4 Likert5 Likert7 Likert62 Likert63 Likert64 point, policy' /Likert44 '5- point, engagement' /Likert83 Likert8 Likert9 Likert10 Likert65 point, policy' '11-point, '+ Likert11 /SCALE('ALL VARIABLES') ALL /Likert45 '5-point, policy' 'partisanship' /Likert84 '11- Likert12 Likert13 Likert14 /MODEL=ALPHA /Likert46 '5-point, policy' point, partisanship' /Likert85 Likert15 Likert16 Likert17 /SUMMARY=TOTAL. /Likert47 '5-point, policy' '11-point, partisanship' Likert18 Likert19 Likert20 /Likert48 /Likert86 Likert21 Likert22 RELIABILITY '5-point, policy' /Likert49 '5- '11-point, partisanship' Likert23 Likert24 /VARIABLES=Likert66 point, policy' /Likert50 '5- /Likert87 '11-point, /SCALE('ALL VARIABLES') ALL Likert67 Likert68 Likert69 point, policy' /Likert51 '5- partisanship' /Likert88 '11- /MODEL=ALPHA Likert70 Likert71 Likert72 point, '+ point, partisanship' /SUMMARY=TOTAL. Likert73 Likert74 'policy' /Likert52 '5-point, /Likert89 '11-point, Likert75 Likert100 Likert101 policy' /Likert53 '5-point, partisanship' /Likert90 '11- RELIABILITY /SCALE('ALL VARIABLES') ALL policy' /Likert54 '5-point, point, partisanship' /Likert91 /VARIABLES=Likert76 /MODEL=ALPHA policy' '11-point, '+ Likert77 Likert78 Likert79 /SUMMARY=TOTAL. /Likert55 '5-point, policy' 'partisanship' /Likert92 '11- Likert80 Likert81 Likert82 /Likert56 '5-point, policy' point, partisanship' /Likert93 Likert83 Likert84 RELIABILITY /Likert57 '5-point, policy' '11-point, partisanship' Likert85 Likert86 Likert87 /VARIABLES=Likert1 Likert2 /Likert58 /Likert94 Likert88 Likert89 Likert90 Likert3 Likert4 Likert5 Likert7 '5-point, policy' /Likert59 '5- '11-point, partisanship' Likert91 Likert92 Likert93 Likert8 Likert9 Likert10 point, policy' /Likert60 '5- /Likert95 '11-point, Likert94 Likert95 Likert11 point, policy' /Likert61 '5- partisanship'. /SCALE('ALL VARIABLES') ALL Likert12 Likert13 Likert14 point, '+ EXECUTE. /MODEL=ALPHA Likert15 Likert16 Likert17 'policy' /Likert62 '5-point, /SUMMARY=TOTAL. Likert18 Likert19 Likert20 policy' /Likert63 '5-point, RECODE electionInterestW4 Likert21 Likert22 policy' /Likert64 '5-point, electionInterestW5 RELIABILITY Likert23 Likert24 Likert25 policy' electionInterestW6 /VARIABLES=Likert1 Likert2 Likert26 Likert27 Likert28 Likert3 Likert4 Likert5 Likert7

Likert29 Likert30 Likert31 Likert78 Likert79 Likert80 /VARIABLES=FAC1_24 RECODE Likert96 Likert97 Likert32 Likert33 Likert81 Likert82 Likert83 ReflectedLnReflectedItemRes Likert98 Likert99 Likert100 Likert34 Likert35 Likert36 Likert84 Likert85 Likert86 ponseRate Likert101 (1=50) (2=100). Likert37 Likert38 Likert39 Likert87 Likert88 /PRINT=SPEARMAN EXECUTE. Likert40 Likert41 Likert42 Likert89 Likert90 Likert91 TWOTAIL NOSIG Likert43 Likert44 Likert92 Likert93 Likert94 /MISSING=PAIRWISE. RECODE Likert25 Likert26 Likert45 Likert46 Likert47 Likert95 Likert96 Likert97 Likert27 Likert28 Likert29 Likert48 Likert49 Likert50 Likert98 Likert99 NONPAR CORR Likert30 Likert31 Likert32 Likert51 Likert52 Likert53 Likert100 Likert101 /VARIABLES=ReflectedLnRefle Likert33 Likert34 Likert54 Likert55 /MISSING MEANSUB ctedItemResponseRate Likert35 Likert36 Likert37 Likert56 Likert57 Likert58 /ANALYSIS Likert1 Likert2 FAC1_20 Likert38 Likert39 Likert40 Likert59 Likert60 Likert61 Likert3 Likert4 Likert5 Likert7 /PRINT=SPEARMAN Likert41 Likert42 Likert43 Likert62 Likert63 Likert64 Likert8 Likert9 Likert10 TWOTAIL NOSIG Likert44 Likert45 Likert65 Likert66 Likert11 /MISSING=PAIRWISE. Likert46 Likert47 Likert48 Likert67 Likert68 Likert69 Likert12 Likert13 Likert14 Likert49 Likert50 Likert51 Likert70 Likert71 Likert72 Likert15 Likert16 Likert17 NONPAR CORR Likert52 Likert53 Likert54 Likert73 Likert74 Likert75 Likert18 Likert19 Likert20 /VARIABLES=ReflectedLnRefle Likert55 Likert56 Likert76 Likert77 Likert21 Likert22 ctedItemResponseRate Likert57 Likert58 Likert59 Likert78 Likert79 Likert80 Likert23 Likert24 Likert25 angerScale fearScale Likert60 Likert61 Likert62 Likert81 Likert82 Likert83 Likert26 Likert27 Likert28 /PRINT=SPEARMAN Likert63 Likert64 Likert65 Likert84 Likert85 Likert86 Likert29 Likert30 Likert31 TWOTAIL NOSIG Likert66 Likert67 Likert87 Likert88 Likert32 Likert33 /MISSING=PAIRWISE. Likert68 Likert69 Likert70 Likert89 Likert90 Likert91 Likert34 Likert35 Likert36 Likert71 Likert72 Likert73 Likert92 Likert93 Likert94 Likert37 Likert38 Likert39 NONPAR CORR Likert74 Likert75 (0=0) (1=50) Likert95 Likert96 Likert97 Likert40 Likert41 Likert42 /VARIABLES=ReflectedLnRefle (2=100). Likert98 Likert99 Likert43 Likert44 ctedItemResponseRate EXECUTE. Likert100 Likert101 Likert45 Likert46 Likert47 FAC1_21 FAC2_21 /SCALE('ALL VARIABLES') ALL Likert48 Likert49 Likert50 /PRINT=SPEARMAN RECODE Likert1 Likert2 Likert3 /MODEL=ALPHA Likert51 Likert52 Likert53 TWOTAIL NOSIG Likert4 Likert5 Likert7 Likert8 /SUMMARY=TOTAL. Likert54 Likert55 /MISSING=PAIRWISE. Likert9 Likert10 Likert11 Likert56 Likert57 Likert58 Likert12 FACTOR Likert59 Likert60 Likert61 CORRELATIONS Likert13 Likert14 Likert15 /VARIABLES Likert96 Likert97 Likert62 Likert63 Likert64 /VARIABLES=FAC1_20 Likert16 Likert17 Likert18 Likert98 Likert99 Likert100 Likert65 Likert66 angerScale fearScale FAC1_21 Likert19 Likert20 Likert21 Likert101 Likert67 Likert68 Likert69 FAC2_21 Likert22 Likert23 /MISSING MEANSUB Likert70 Likert71 Likert72 ReflectedLnReflectedItemRes Likert24 (0=0) (1=33) (2=67) /ANALYSIS Likert96 Likert97 Likert73 Likert74 Likert75 ponseRate (3=100). Likert98 Likert99 Likert100 Likert76 Likert77 FAC1_24 EXECUTE. Likert101 Likert78 Likert79 Likert80 /PRINT=TWOTAIL NOSIG /PRINT INITIAL EXTRACTION Likert81 Likert82 Likert83 /MISSING=PAIRWISE. COMPUTE ROTATION Likert84 Likert85 Likert86 ResponsePolarisation=MEAN( /PLOT EIGEN ROTATION Likert87 Likert88 REGRESSION Likert76, Likert77, Likert78, /CRITERIA MINEIGEN(1) Likert89 Likert90 Likert91 /MISSING LISTWISE Likert79, Likert80, Likert81, ITERATE(25) Likert92 Likert93 Likert94 /STATISTICS COEFF OUTS R Likert82, Likert83, Likert84, /EXTRACTION PC Likert95 Likert96 Likert97 ANOVA Likert85, Likert86, Likert87, /CRITERIA ITERATE(25) Likert98 Likert99 /CRITERIA=PIN(.05) Likert88, Likert89, Likert90, DELTA(0) Likert100 Likert101 POUT(.10) Likert91, /ROTATION OBLIMIN /PRINT INITIAL EXTRACTION /NOORIGIN Likert92, Likert93, Likert94, /SAVE REG(ALL) ROTATION /DEPENDENT FAC1_24 Likert95, Likert96, Likert97, /METHOD=CORRELATION. /PLOT EIGEN ROTATION /METHOD=ENTER FAC1_20 Likert98, Likert99, Likert100, /CRITERIA FACTORS(1) angerScale fearScale FAC1_21 Likert101, Likert25, FACTOR ITERATE(25) FAC2_21 Likert26, Likert27, Likert28, /VARIABLES Likert1 Likert2 /EXTRACTION PC ReflectedLnReflectedItemRes Likert29, Likert30, Likert31, Likert3 Likert4 Likert5 Likert7 /CRITERIA ITERATE(25) ponseRate. Likert32, Likert8 Likert9 Likert10 DELTA(0) Likert33, Likert34, Likert35, Likert11 /ROTATION OBLIMIN CORRELATIONS Likert36, Likert37, Likert38, Likert12 Likert13 Likert14 /SAVE REG(ALL) /VARIABLES=ReflectedLnRefle Likert39, Likert40, Likert41, Likert15 Likert16 Likert17 /METHOD=CORRELATION. ctedItemResponseRate Likert42, Likert18 Likert19 Likert20 FAC1_24 Likert43, Likert44, Likert45, Likert21 Likert22 NONPAR CORR personality_agreeableness Likert46, Likert47, Likert48, Likert23 Likert24 Likert25 /VARIABLES=FAC1_24 Likert49, Likert50, Likert51, Likert26 Likert27 Likert28 FAC1_20 personality_conscientiousness Likert52, Likert29 Likert30 Likert31 /PRINT=SPEARMAN personality_extraversion Likert53, Likert54, Likert55, Likert32 Likert33 TWOTAIL NOSIG personality_neuroticism Likert56, Likert57, Likert58, Likert34 Likert35 Likert36 /MISSING=PAIRWISE. personality_openness Likert59, Likert60, Likert61, Likert37 Likert38 Likert39 /PRINT=TWOTAIL NOSIG Likert62, Likert40 Likert41 Likert42 NONPAR CORR /MISSING=PAIRWISE. Likert63, Likert64, Likert65, Likert43 Likert44 /VARIABLES=FAC1_24 Likert66, Likert67, Likert68, Likert45 Likert46 Likert47 angerScale fearScale RECODE Likert76 Likert77 Likert69, Likert70, Likert71, Likert48 Likert49 Likert50 /PRINT=SPEARMAN Likert78 Likert79 Likert80 Likert72, Likert51 Likert52 Likert53 TWOTAIL NOSIG Likert81 Likert82 Likert83 Likert73, Likert74, Likert75, Likert54 Likert55 /MISSING=PAIRWISE. Likert84 Likert85 Likert1, Likert2, Likert3, Likert56 Likert57 Likert58 Likert86 Likert87 Likert88 Likert4, Likert5, Likert7, Likert59 Likert60 Likert61 NONPAR CORR Likert89 Likert90 Likert91 Likert8, Likert62 Likert63 Likert64 /VARIABLES=FAC1_24 Likert92 Likert93 Likert94 Likert9, Likert10, Likert11, Likert65 Likert66 FAC1_21 FAC2_21 Likert95 (0=0) Likert12, Likert13, Likert14, Likert67 Likert68 Likert69 /PRINT=SPEARMAN (1=20) (2=40) (3=60) (4=80) Likert15, Likert16, Likert17, Likert70 Likert71 Likert72 TWOTAIL NOSIG (5=100). Likert18, Likert73 Likert74 Likert75 /MISSING=PAIRWISE. EXECUTE. Likert19, Likert20, Likert21, Likert76 Likert77 Likert22, Likert23, Likert24). NONPAR CORR

VARIABLE LABELS /PLOT EIGEN ROTATION /SCALE('ALL VARIABLES') ALL /PRINT INITIAL EXTRACTION ResponsePolarisation '[theme: /CRITERIA FACTORS(1) /MODEL=ALPHA ROTATION psychology] Percentage ITERATE(25) /SUMMARY=TOTAL. /PLOT EIGEN ROTATION polarisation of Likert '+ /EXTRACTION PC /CRITERIA MINEIGEN(1) 'scale survey responses'. /CRITERIA ITERATE(25) FACTOR ITERATE(25) EXECUTE. DELTA(0) /VARIABLES /EXTRACTION PC /ROTATION OBLIMIN immigrantsWelfareStateW1 /CRITERIA ITERATE(25) CORRELATIONS /SAVE REG(ALL) immigrantsWelfareStateW2 DELTA(0) /VARIABLES=ResponsePolarisa /METHOD=CORRELATION. immigrantsWelfareStateW3 /ROTATION OBLIMIN tion FAC1_24 immigrantsWelfareStateW4 /SAVE REG(ALL) /PRINT=TWOTAIL NOSIG RELIABILITY immigrantsWelfareStateW8 /METHOD=CORRELATION. /MISSING=PAIRWISE. /VARIABLES=immigEconW1 /MISSING MEANSUB immigEconW2 immigEconW3 /ANALYSIS RELIABILITY RECODE businessBonusW1 immigEconW4 immigrantsWelfareStateW1 /VARIABLES=FAC1_25 businessBonusW2 /SCALE('ALL VARIABLES') ALL immigrantsWelfareStateW2 FAC1_26 FAC1_27 FAC1_28 businessBonusW3 /MODEL=ALPHA immigrantsWelfareStateW3 FAC1_29 FAC1_30 al4W6 businessBonusW4 (1=5) (2=4) /SUMMARY=TOTAL. immigrantsWelfareStateW4 lr1W6 al5W6 al3W6 (3=3) (4=2) immigrantsWelfareStateW8 /SCALE('ALL VARIABLES') ALL (5=1). FACTOR /PRINT INITIAL EXTRACTION /MODEL=ALPHA EXECUTE. /VARIABLES immigEconW1 ROTATION /SUMMARY=TOTAL. immigEconW2 immigEconW3 /PLOT EIGEN ROTATION RELIABILITY immigEconW4 /CRITERIA FACTORS(1) CORRELATIONS /VARIABLES=businessBonusW /MISSING MEANSUB ITERATE(25) /VARIABLES=FAC1_32 1 businessBonusW2 /ANALYSIS immigEconW1 /EXTRACTION PC FAC2_32 FAC3_32 businessBonusW3 immigEconW2 immigEconW3 /CRITERIA ITERATE(25) /PRINT=TWOTAIL NOSIG businessBonusW4 immigEconW4 DELTA(0) /MISSING=PAIRWISE. /SCALE('ALL VARIABLES') ALL /PRINT INITIAL EXTRACTION /ROTATION OBLIMIN /MODEL=ALPHA ROTATION /SAVE REG(ALL) RELIABILITY /SUMMARY=TOTAL. /PLOT EIGEN ROTATION /METHOD=CORRELATION. /VARIABLES=FAC1_32 /CRITERIA FACTORS(1) FAC2_32 FAC3_32 FACTOR ITERATE(25) RELIABILITY /SCALE('ALL VARIABLES') ALL /VARIABLES /EXTRACTION PC /VARIABLES=reasonForUnemp /MODEL=ALPHA businessBonusW1 /CRITERIA ITERATE(25) loymentW1 /SUMMARY=TOTAL. businessBonusW2 DELTA(0) reasonForUnemploymentW2 businessBonusW3 /ROTATION OBLIMIN reasonForUnemploymentW3 FACTOR businessBonusW4 /SAVE REG(ALL) /VARIABLES FAC1_32 /MISSING MEANSUB /METHOD=CORRELATION. reasonForUnemploymentW4 FAC2_32 FAC3_32 /ANALYSIS businessBonusW1 /SCALE('ALL VARIABLES') ALL /MISSING MEANSUB businessBonusW2 RELIABILITY /MODEL=ALPHA /ANALYSIS FAC1_32 FAC2_32 businessBonusW3 /VARIABLES=immigrationLevel /SUMMARY=TOTAL. FAC3_32 businessBonusW4 W4 immigrationLevelW6 /PRINT INITIAL EXTRACTION /PRINT INITIAL EXTRACTION /SCALE('ALL VARIABLES') ALL FACTOR ROTATION ROTATION /MODEL=ALPHA /VARIABLES /PLOT EIGEN ROTATION /PLOT EIGEN ROTATION /SUMMARY=TOTAL. reasonForUnemploymentW1 /CRITERIA MINEIGEN(1) /CRITERIA FACTORS(1) reasonForUnemploymentW2 ITERATE(25) ITERATE(25) FACTOR reasonForUnemploymentW3 /EXTRACTION PC /EXTRACTION PC /VARIABLES /CRITERIA ITERATE(25) /CRITERIA ITERATE(25) immigrationLevelW4 reasonForUnemploymentW4 DELTA(0) DELTA(0) immigrationLevelW6 /MISSING MEANSUB /ROTATION OBLIMIN /ROTATION OBLIMIN /MISSING MEANSUB /ANALYSIS /SAVE REG(ALL) /SAVE REG(ALL) /ANALYSIS reasonForUnemploymentW1 /METHOD=CORRELATION. /METHOD=CORRELATION. immigrationLevelW4 reasonForUnemploymentW2 immigrationLevelW6 reasonForUnemploymentW3 CORRELATIONS RECODE govtHandoutsW1 /PRINT INITIAL EXTRACTION /VARIABLES=FAC1_31 govtHandoutsW2 ROTATION reasonForUnemploymentW4 FAC1_32 FAC2_32 FAC3_32 govtHandoutsW3 /PLOT EIGEN ROTATION /PRINT INITIAL EXTRACTION FAC1_25 FAC1_26 FAC1_27 govtHandoutsW4 (1=5) (2=4) /CRITERIA FACTORS(1) ROTATION FAC1_28 FAC1_29 FAC1_30 (3=3) (4=2) (5=1). ITERATE(25) /PLOT EIGEN ROTATION al4W6 EXECUTE. /EXTRACTION PC /CRITERIA FACTORS(1) lr1W6 al5W6 al3W6 /CRITERIA ITERATE(25) ITERATE(25) /PRINT=TWOTAIL NOSIG RELIABILITY DELTA(0) /EXTRACTION PC /MISSING=PAIRWISE. /VARIABLES=govtHandoutsW1 /ROTATION OBLIMIN /CRITERIA ITERATE(25) govtHandoutsW2 /SAVE REG(ALL) DELTA(0) RECODE govtHandoutsW3 /METHOD=CORRELATION. /ROTATION OBLIMIN profile_newspaper_readershi govtHandoutsW4 /SAVE REG(ALL) p_201 (MISSING=SYSMIS) (1 /SCALE('ALL VARIABLES') ALL RECODE /METHOD=CORRELATION. thru 5=1) (ELSE=0) INTO /MODEL=ALPHA immigrantsWelfareStateW1 Tabloid. /SUMMARY=TOTAL. immigrantsWelfareStateW2 RECODE al4W6 al5W6 al3W6 VARIABLE LABELS Tabloid immigrantsWelfareStateW3 (1=5) (2=4) (3=3) (4=2) (5=1). '[ANALYSE] Tabloid FACTOR immigrantsWelfareStateW4 EXECUTE. newspaper readership as /VARIABLES govtHandoutsW1 immigrantsWelfareStateW8 opposed to other or none'. govtHandoutsW2 (1=5) (2=4) (3=3) (4=2) (5=1). FACTOR EXECUTE. govtHandoutsW3 EXECUTE. /VARIABLES FAC1_25 govtHandoutsW4 FAC1_26 FAC1_27 FAC1_28 RECODE /MISSING MEANSUB RELIABILITY FAC1_29 FAC1_30 al4W6 profile_newspaper_readershi /ANALYSIS govtHandoutsW1 /VARIABLES=immigrantsWelfa lr1W6 al5W6 al3W6 p_201 (MISSING=SYSMIS) (6 govtHandoutsW2 reStateW1 /MISSING MEANSUB thru 13=1) (ELSE=0) INTO govtHandoutsW3 immigrantsWelfareStateW2 /ANALYSIS FAC1_25 FAC1_26 Broadsheet. govtHandoutsW4 immigrantsWelfareStateW3 FAC1_27 FAC1_28 FAC1_29 VARIABLE LABELS Broadsheet /PRINT INITIAL EXTRACTION immigrantsWelfareStateW4 FAC1_30 al4W6 lr1W6 al5W6 '[ANALYSE] Broadsheet ROTATION immigrantsWelfareStateW8 al3W6

84 newspaper readership as Overall=SQRT(FAC1_31 * END IF. NoDirDistanceFromMeanOfCu opposed to other or none'. FAC1_31). EXECUTE. lturalCluster. EXECUTE. VARIABLE LABELS VARIABLE LABELS PoliticalOrientationDifference DO IF (TSC_5887 = 1). ClusterCulturalPoliticalConfor RECODE immigEconW1 Overall '[theme: culture] RECODE mity '[theme: culture] immigEconW2 immigEconW3 Individual difference '+ MeanOrientationofCulturalClu Individual conformity to '+ immigEconW4 (1=1) (2=1.67) 'in general political leftness ster (0=0.57). 'mean cluster political (3=2.33) (4=3) (5=3.67) from mean political leftness END IF. orientation'. (6=4.33) (7=5). for the overall data'. EXECUTE. EXECUTE. EXECUTE. EXECUTE. DO IF (TSC_5887 = 2). FREQUENCIES COMPUTE FREQUENCIES RECODE VARIABLES=ClusterCulturalPol SDPolicyPreferences=SD(busin VARIABLES=PoliticalOrientatio MeanOrientationofCulturalClu iticalConformity essBonusW1, nDifferenceOverall ster (0=-0.20). /FORMAT=NOTABLE businessBonusW2, /FORMAT=NOTABLE END IF. /NTILES=4 businessBonusW3, /NTILES=4 EXECUTE. /STATISTICS=STDDEV businessBonusW4, /STATISTICS=STDDEV MINIMUM MAXIMUM MEAN govtHandoutsW1, MINIMUM MAXIMUM MEAN DO IF (TSC_5887 = 3). MEDIAN SKEWNESS SESKEW govtHandoutsW2, MEDIAN SKEWNESS SESKEW RECODE KURTOSIS SEKURT govtHandoutsW3, KURTOSIS SEKURT MeanOrientationofCulturalClu /HISTOGRAM NORMAL govtHandoutsW4, /HISTOGRAM NORMAL ster (0=-0.32). /ORDER=ANALYSIS. immigEconW1, /ORDER=ANALYSIS. END IF. immigEconW2, EXECUTE. CORRELATIONS immigEconW3, COMPUTE immigEconW4, OverallCulturalPoliticalConfor DO IF (TSC_5887 = 4). /VARIABLES=OverallCulturalPo immigrationLevelW4, mity=4.07- RECODE liticalConformity immigrationLevelW6, PoliticalOrientationDifference MeanOrientationofCulturalClu ClusterCulturalPoliticalConfor immigrantsWelfareStateW1, Overall. ster (0=0.10). mity VARIABLE LABELS END IF. /PRINT=TWOTAIL NOSIG immigrantsWelfareStateW2, OverallCulturalPoliticalConfor EXECUTE. /MISSING=PAIRWISE. immigrantsWelfareStateW3, mity '[theme: culture] immigrantsWelfareStateW4, individual conformity to '+ FREQUENCIES GRAPH 'overall mean political VARIABLES=MeanOrientation immigrantsWelfareStateW8, orientation'. ofCulturalCluster /SCATTERPLOT(BIVAR)=Overal reasonForUnemploymentW1, EXECUTE. /ORDER=ANALYSIS. lCulturalPoliticalConformity reasonForUnemploymentW2, WITH FREQUENCIES COMPUTE ClusterCulturalPoliticalConfor reasonForUnemploymentW3, VARIABLES=OverallCulturalPol DistanceFromMeanOfCultural mity reasonForUnemploymentW4, iticalConformity Cluster=MeanOrientationofCu /MISSING=LISTWISE. al4W6, lr1W6, al5W6, al3W6). /FORMAT=NOTABLE lturalCluster - FAC1_31. VARIABLE LABELS /NTILES=4 VARIABLE LABELS SORT VARIABLES BY LABEL (A). SDPolicyPreferences '[theme: /STATISTICS=STDDEV DistanceFromMeanOfCultural culture] Individual-level MINIMUM MAXIMUM MEAN Cluster "[theme: culture] CORRELATIONS standard deviation of '+ MEDIAN SKEWNESS SESKEW distance between individual /VARIABLES=FAC1_31 'policy preference items, as KURTOSIS SEKURT "+ FAC1_23 an inverse measure of /HISTOGRAM NORMAL "political leftness and mean OverallCulturalPoliticalConfor constraint'. /ORDER=ANALYSIS. political leftness for mity BeliefSystemConstraint EXECUTE. individual's culcluster". ageW7 FAC1_13 FAC1_15 SORT CASES BY TSC_5887. EXECUTE. FAC1_16 FAC1_19 FAC1_20 FREQUENCIES SPLIT FILE LAYERED BY angerScale fearScale FAC1_21 VARIABLES=SDPolicyPreferenc TSC_5887. COMPUTE FAC2_21 es NoDirDistanceFromMeanOfCu ReflectedLnReflectedItemRes /FORMAT=NOTABLE DESCRIPTIVES lturalCluster=SQRT(DistanceFr ponseRate FAC1_24 /NTILES=4 VARIABLES=FAC1_31 omMeanOfCulturalCluster* /PRINT=TWOTAIL NOSIG /STATISTICS=STDDEV /STATISTICS=MEAN. /MISSING=PAIRWISE. MINIMUM MAXIMUM MEAN DistanceFromMeanOfCultural MEDIAN SKEWNESS SESKEW SPLIT FILE OFF. Cluster). CORRELATIONS KURTOSIS SEKURT VARIABLE LABELS /VARIABLES=FAC3_32 /HISTOGRAM NORMAL COMPUTE NoDirDistanceFromMeanOfCu FAC1_23 /ORDER=ANALYSIS. MeanOrientationofCulturalClu lturalCluster "[theme: culture] OverallCulturalPoliticalConfor ster=0. distance between "+ mity BeliefSystemConstraint COMPUTE VARIABLE LABELS "political leftness and mean ageW7 BeliefSystemConstraint=2.12- MeanOrientationofCulturalClu political leftness for FAC1_13 FAC1_15 FAC1_16 SDPolicyPreferences. ster "[theme: culture] mean individual's culcluster no +-". FAC1_19 FAC1_20 angerScale VARIABLE LABELS general political "+ EXECUTE. fearScale FAC1_21 FAC2_21 BeliefSystemConstraint "leftness of individual's '[ANALYSE] [theme: culture] cultural cluster". FREQUENCIES ReflectedLnReflectedItemRes Reflected standard deviation EXECUTE. VARIABLES=NoDirDistanceFro ponseRate FAC1_24 '+ mMeanOfCulturalCluster /PRINT=TWOTAIL NOSIG 'of policy preferences, as a DO IF /FORMAT=NOTABLE /MISSING=PAIRWISE. measure of belief system (MISSING(TSC_5887)=1). /NTILES=4 constraint'. RECODE /STATISTICS=STDDEV CORRELATIONS EXECUTE. MeanOrientationofCulturalClu MINIMUM MAXIMUM MEAN /VARIABLES=FAC2_32 ster (0=SYSMIS). MEDIAN SKEWNESS SESKEW FAC1_23 DESCRIPTIVES END IF. KURTOSIS SEKURT OverallCulturalPoliticalConfor VARIABLES=FAC1_31 EXECUTE. /HISTOGRAM NORMAL mity BeliefSystemConstraint /STATISTICS=MEAN. /ORDER=ANALYSIS. ageW7 DO IF (TSC_5887 = 0). FAC1_13 FAC1_15 FAC1_16 COMPUTE RECODE COMPUTE FAC1_19 FAC1_20 angerScale PoliticalOrientationDifference MeanOrientationofCulturalClu ClusterCulturalPoliticalConfor fearScale FAC1_21 FAC2_21 ster (0=0.13). mity=3.94 -

* Chart Builder. DATA: GUIDE: ReflectedLnReflectedItemRes GGRAPH MEAN_FAC3_32=col(source(s) text.footnote(label("Error ponseRate FAC1_24 /GRAPHDATA SET , name("MEAN_FAC3_32")) Bars: 95% CI")) /PRINT=TWOTAIL NOSIG NAME="graphdata set" DATA: LOW=col(source(s), SCALE: linear(dim(2), /MISSING=PAIRWISE. VARIABLES=Tabloid name("MEAN_FAC3_32_LOW include(0)) MEANCI(FAC1_31, ")) ELEMENT: CORRELATIONS 95)[name="MEAN_FAC1_31" DATA: HIGH=col(source(s), interval(position(Broadsheet* /VARIABLES=FAC1_32 name("MEAN_FAC3_32_HIGH MEAN_FAC3_32), FAC1_23 LOW="MEAN_FAC1_31_LOW ")) shape.interior(shape.square)) OverallCulturalPoliticalConfor " GUIDE: axis(dim(1), ELEMENT: mity BeliefSystemConstraint HIGH="MEAN_FAC1_31_HIGH label("[ANALYSE] [theme: interval(position(region.sprea ageW7 "] MISSING=LISTWISE demographics] [readership] d.range(Broadsheet*(LOW+HI FAC1_13 FAC1_15 FAC1_16 REPORTMISSING=NO Tabloid newspaper ", GH))), FAC1_19 FAC1_20 angerScale /GRAPHSPEC "readership as opposed to fearScale FAC1_21 FAC2_21 SOURCE=INLINE. other or none")) shape.interior(shape.ibeam)) BEGIN GPL GUIDE: axis(dim(2), END GPL. ReflectedLnReflectedItemRes SOURCE: label("Mean [ANALYSE] ponseRate FAC1_24 s=userSource(id("graphdata [theme: policy] 3 pro-social * Chart Builder. /PRINT=TWOTAIL NOSIG set")) freedom regression ", GGRAPH /MISSING=PAIRWISE. DATA: Tabloid=col(source(s), "factor")) /GRAPHDATA SET name("Tabloid"), GUIDE: NAME="graphdata set" * Chart Builder. unit.category()) text.footnote(label("Error VARIABLES=Broadsheet GGRAPH DATA: Bars: 95% CI")) MEANCI(FAC1_32, /GRAPHDATA SET MEAN_FAC1_31=col(source(s) SCALE: linear(dim(2), 95)[name="MEAN_FAC1_32" NAME="graphdata set" , name("MEAN_FAC1_31")) include(0)) VARIABLES=Broadsheet DATA: LOW=col(source(s), ELEMENT: LOW="MEAN_FAC1_32_LOW MEANCI(FAC1_31, name("MEAN_FAC1_31_LOW interval(position(Tabloid*MEA " 95)[name="MEAN_FAC1_31" ")) N_FAC3_32), HIGH="MEAN_FAC1_32_HIGH DATA: HIGH=col(source(s), shape.interior(shape.square)) "] MISSING=LISTWISE LOW="MEAN_FAC1_31_LOW name("MEAN_FAC1_31_HIGH ELEMENT: REPORTMISSING=NO " ")) interval(position(region.sprea /GRAPHSPEC HIGH="MEAN_FAC1_31_HIGH GUIDE: axis(dim(1), d.range(Tabloid*(LOW+HIGH) SOURCE=INLINE. "] MISSING=LISTWISE label("[ANALYSE] [theme: )), BEGIN GPL REPORTMISSING=NO demographics] [readership] shape.interior(shape.ibeam)) SOURCE: /GRAPHSPEC Tabloid newspaper ", END GPL. s=userSource(id("graphdata SOURCE=INLINE. "readership as opposed to set")) BEGIN GPL other or none")) * Chart Builder. DATA: SOURCE: GUIDE: axis(dim(2), GGRAPH Broadsheet=col(source(s), s=userSource(id("graphdata label("Mean [ANALYSE] /GRAPHDATA SET name("Broadsheet"), set")) [theme: policy] General NAME="graphdata set" unit.category()) DATA: Political Leftness ", VARIABLES=Broadsheet DATA: Broadsheet=col(source(s), "regression factor")) MEANCI(FAC3_32, MEAN_FAC1_32=col(source(s) name("Broadsheet"), GUIDE: 95)[name="MEAN_FAC3_32" , name("MEAN_FAC1_32")) unit.category()) text.footnote(label("Error DATA: LOW=col(source(s), DATA: Bars: 95% CI")) LOW="MEAN_FAC3_32_LOW name("MEAN_FAC1_32_LOW MEAN_FAC1_31=col(source(s) SCALE: linear(dim(2), " ")) , name("MEAN_FAC1_31")) include(0)) HIGH="MEAN_FAC3_32_HIGH DATA: HIGH=col(source(s), DATA: LOW=col(source(s), ELEMENT: "] MISSING=LISTWISE name("MEAN_FAC1_32_HIGH name("MEAN_FAC1_31_LOW interval(position(Tabloid*MEA REPORTMISSING=NO ")) ")) N_FAC1_31), /GRAPHSPEC GUIDE: axis(dim(1), DATA: HIGH=col(source(s), shape.interior(shape.square)) SOURCE=INLINE. label("[ANALYSE] [theme: name("MEAN_FAC1_31_HIGH ELEMENT: BEGIN GPL demographics] [readership] ")) interval(position(region.sprea SOURCE: Broadsheet newspaper ", GUIDE: axis(dim(1), d.range(Tabloid*(LOW+HIGH) s=userSource(id("graphdata "readership as opposed to label("[ANALYSE] [theme: )), set")) other or none")) demographics] [readership] shape.interior(shape.ibeam)) DATA: GUIDE: axis(dim(2), Broadsheet newspaper ", END GPL. Broadsheet=col(source(s), label("Mean [ANALYSE] "readership as opposed to name("Broadsheet"), [theme: policy] 1 pro- other or none")) * Chart Builder. unit.category()) immigration regression GUIDE: axis(dim(2), GGRAPH DATA: factor")) label("Mean [ANALYSE] /GRAPHDATA SET MEAN_FAC3_32=col(source(s) GUIDE: [theme: policy] General NAME="graphdata set" , name("MEAN_FAC3_32")) text.footnote(label("Error Political Leftness ", VARIABLES=Tabloid DATA: LOW=col(source(s), Bars: 95% CI")) "regression factor")) MEANCI(FAC3_32, name("MEAN_FAC3_32_LOW SCALE: linear(dim(2), GUIDE: 95)[name="MEAN_FAC3_32" ")) include(0)) text.footnote(label("Error DATA: HIGH=col(source(s), ELEMENT: Bars: 95% CI")) LOW="MEAN_FAC3_32_LOW name("MEAN_FAC3_32_HIGH interval(position(Broadsheet* SCALE: linear(dim(2), " ")) MEAN_FAC1_32), include(0)) HIGH="MEAN_FAC3_32_HIGH GUIDE: axis(dim(1), shape.interior(shape.square)) ELEMENT: "] MISSING=LISTWISE label("[ANALYSE] [theme: ELEMENT: interval(position(Broadsheet* REPORTMISSING=NO demographics] [readership] interval(position(region.sprea MEAN_FAC1_31), /GRAPHSPEC Broadsheet newspaper ", d.range(Broadsheet*(LOW+HI shape.interior(shape.square)) SOURCE=INLINE. "readership as opposed to GH))), ELEMENT: BEGIN GPL other or none")) interval(position(region.sprea SOURCE: GUIDE: axis(dim(2), shape.interior(shape.ibeam)) d.range(Broadsheet*(LOW+HI s=userSource(id("graphdata label("Mean [ANALYSE] END GPL. GH))), set")) [theme: policy] 3 pro-social DATA: Tabloid=col(source(s), freedom regression ", * Chart Builder. shape.interior(shape.ibeam)) name("Tabloid"), "factor")) GGRAPH END GPL. unit.category()) /GRAPHDATA SET NAME="graphdata set"

VARIABLES=Tabloid DATA: LOW=col(source(s), SCALE: linear(dim(2), MEANCI(FAC1_32, name("MEAN_FAC2_32_LOW include(0)) LOW="MEAN_FAC1_31_LOW 95)[name="MEAN_FAC1_32" ")) ELEMENT: " DATA: HIGH=col(source(s), interval(position(Broadsheet* HIGH="MEAN_FAC1_31_HIGH LOW="MEAN_FAC1_32_LOW name("MEAN_FAC2_32_HIGH MEAN_FAC2_32), "] MISSING=LISTWISE " ")) shape.interior(shape.square)) REPORTMISSING=NO HIGH="MEAN_FAC1_32_HIGH GUIDE: axis(dim(1), ELEMENT: /GRAPHSPEC "] MISSING=LISTWISE label("[ANALYSE] [theme: interval(position(region.sprea SOURCE=INLINE. REPORTMISSING=NO demographics] [readership] d.range(Broadsheet*(LOW+HI BEGIN GPL /GRAPHSPEC Tabloid newspaper ", GH))), SOURCE: SOURCE=INLINE. "readership as opposed to s=userSource(id("graphdata BEGIN GPL other or none")) shape.interior(shape.ibeam)) set")) SOURCE: GUIDE: axis(dim(2), END GPL. DATA: s=userSource(id("graphdata label("Mean [ANALYSE] Midskilled=col(source(s), set")) [theme: policy] 2 pro- * Chart Builder. name("Midskilled"), DATA: Tabloid=col(source(s), government involvement in ", GGRAPH unit.category()) name("Tabloid"), "economy regression /GRAPHDATA SET DATA: unit.category()) factor")) NAME="graphdata set" MEAN_FAC1_31=col(source(s) DATA: GUIDE: VARIABLES=Upskilled , name("MEAN_FAC1_31")) MEAN_FAC1_32=col(source(s) text.footnote(label("Error MEANCI(FAC1_31, DATA: LOW=col(source(s), , name("MEAN_FAC1_32")) Bars: 95% CI")) 95)[name="MEAN_FAC1_31" name("MEAN_FAC1_31_LOW DATA: LOW=col(source(s), SCALE: linear(dim(2), ")) name("MEAN_FAC1_32_LOW include(0)) LOW="MEAN_FAC1_31_LOW DATA: HIGH=col(source(s), ")) ELEMENT: " name("MEAN_FAC1_31_HIGH DATA: HIGH=col(source(s), interval(position(Tabloid*MEA HIGH="MEAN_FAC1_31_HIGH ")) name("MEAN_FAC1_32_HIGH N_FAC2_32), "] MISSING=LISTWISE GUIDE: axis(dim(1), ")) shape.interior(shape.square)) REPORTMISSING=NO label("[ANALYSE] [theme: GUIDE: axis(dim(1), ELEMENT: /GRAPHSPEC demographics] [work] Work label("[ANALYSE] [theme: interval(position(region.sprea SOURCE=INLINE. type middling skilled ", demographics] [readership] d.range(Tabloid*(LOW+HIGH) BEGIN GPL "as opposed to never Tabloid newspaper ", )), SOURCE: worked")) "readership as opposed to shape.interior(shape.ibeam)) s=userSource(id("graphdata GUIDE: axis(dim(2), other or none")) END GPL. set")) label("Mean [ANALYSE] GUIDE: axis(dim(2), DATA: [theme: policy] General label("Mean [ANALYSE] * Chart Builder. Upskilled=col(source(s), Political Leftness ", [theme: policy] 1 pro- GGRAPH name("Upskilled"), "regression factor")) immigration regression /GRAPHDATA SET unit.category()) GUIDE: factor")) NAME="graphdata set" DATA: text.footnote(label("Error GUIDE: VARIABLES=Broadsheet MEAN_FAC1_31=col(source(s) Bars: 95% CI")) text.footnote(label("Error MEANCI(FAC2_32, , name("MEAN_FAC1_31")) SCALE: linear(dim(2), Bars: 95% CI")) 95)[name="MEAN_FAC2_32" DATA: LOW=col(source(s), include(0)) SCALE: linear(dim(2), name("MEAN_FAC1_31_LOW ELEMENT: include(0)) LOW="MEAN_FAC2_32_LOW ")) interval(position(Midskilled* ELEMENT: " DATA: HIGH=col(source(s), MEAN_FAC1_31), interval(position(Tabloid*MEA HIGH="MEAN_FAC2_32_HIGH name("MEAN_FAC1_31_HIGH shape.interior(shape.square)) N_FAC1_32), "] MISSING=LISTWISE ")) ELEMENT: shape.interior(shape.square)) REPORTMISSING=NO GUIDE: axis(dim(1), interval(position(region.sprea ELEMENT: /GRAPHSPEC label("[ANALYSE] [theme: d.range(Midskilled*(LOW+HIG interval(position(region.sprea SOURCE=INLINE. demographics] [work] Work H))), d.range(Tabloid*(LOW+HIGH) BEGIN GPL type highly skilled as ", )), SOURCE: "opposed to never shape.interior(shape.ibeam)) shape.interior(shape.ibeam)) s=userSource(id("graphdata worked")) END GPL. END GPL. set")) GUIDE: axis(dim(2), DATA: label("Mean [ANALYSE] * Chart Builder. * Chart Builder. Broadsheet=col(source(s), [theme: policy] General GGRAPH GGRAPH name("Broadsheet"), Political Leftness ", /GRAPHDATA SET /GRAPHDATA SET unit.category()) "regression factor")) NAME="graphdata set" NAME="graphdata set" DATA: GUIDE: VARIABLES=Unskilled VARIABLES=Tabloid MEAN_FAC2_32=col(source(s) text.footnote(label("Error MEANCI(FAC1_31, MEANCI(FAC2_32, , name("MEAN_FAC2_32")) Bars: 95% CI")) 95)[name="MEAN_FAC1_31" 95)[name="MEAN_FAC2_32" DATA: LOW=col(source(s), SCALE: linear(dim(2), name("MEAN_FAC2_32_LOW include(0)) LOW="MEAN_FAC1_31_LOW LOW="MEAN_FAC2_32_LOW ")) ELEMENT: " " DATA: HIGH=col(source(s), interval(position(Upskilled*M HIGH="MEAN_FAC1_31_HIGH HIGH="MEAN_FAC2_32_HIGH name("MEAN_FAC2_32_HIGH EAN_FAC1_31), "] MISSING=LISTWISE "] MISSING=LISTWISE ")) shape.interior(shape.square)) REPORTMISSING=NO REPORTMISSING=NO GUIDE: axis(dim(1), ELEMENT: /GRAPHSPEC /GRAPHSPEC label("[ANALYSE] [theme: interval(position(region.sprea SOURCE=INLINE. SOURCE=INLINE. demographics] [readership] d.range(Upskilled*(LOW+HIG BEGIN GPL BEGIN GPL Broadsheet newspaper ", H))), SOURCE: SOURCE: "readership as opposed to s=userSource(id("graphdata s=userSource(id("graphdata other or none")) shape.interior(shape.ibeam)) set")) set")) GUIDE: axis(dim(2), END GPL. DATA: DATA: Tabloid=col(source(s), label("Mean [ANALYSE] Unskilled=col(source(s), name("Tabloid"), [theme: policy] 2 pro- * Chart Builder. name("Unskilled"), unit.category()) government involvement in ", GGRAPH unit.category()) DATA: "economy regression /GRAPHDATA SET DATA: MEAN_FAC2_32=col(source(s) factor")) NAME="graphdata set" MEAN_FAC1_31=col(source(s) , name("MEAN_FAC2_32")) GUIDE: VARIABLES=Midskilled , name("MEAN_FAC1_31")) text.footnote(label("Error MEANCI(FAC1_31, Bars: 95% CI")) 95)[name="MEAN_FAC1_31"

DATA: LOW=col(source(s), SCALE: linear(dim(2), DATA: LOW=col(source(s), name("MEAN_FAC1_31_LOW include(0)) LOW="MEAN_FAC3_32_LOW name("MEAN_FAC1_32_LOW ")) ELEMENT: " ")) DATA: HIGH=col(source(s), interval(position(Upskilled*M HIGH="MEAN_FAC3_32_HIGH DATA: HIGH=col(source(s), name("MEAN_FAC1_31_HIGH EAN_FAC3_32), "] MISSING=LISTWISE name("MEAN_FAC1_32_HIGH ")) shape.interior(shape.square)) REPORTMISSING=NO ")) GUIDE: axis(dim(1), ELEMENT: /GRAPHSPEC GUIDE: axis(dim(1), label("[ANALYSE] [theme: interval(position(region.sprea SOURCE=INLINE. label("[ANALYSE] [theme: demographics] [work] Work d.range(Upskilled*(LOW+HIG BEGIN GPL demographics] [work] Work type low skilled or ", H))), SOURCE: type highly skilled as ", "unskilled as opposed to s=userSource(id("graphdata "opposed to never never worked")) shape.interior(shape.ibeam)) set")) worked")) GUIDE: axis(dim(2), END GPL. DATA: GUIDE: axis(dim(2), label("Mean [ANALYSE] Midskilled=col(source(s), label("Mean [ANALYSE] [theme: policy] General * Chart Builder. name("Midskilled"), [theme: policy] 1 pro- Political Leftness ", GGRAPH unit.category()) immigration regression "regression factor")) /GRAPHDATA SET DATA: factor")) GUIDE: NAME="graphdata set" MEAN_FAC3_32=col(source(s) GUIDE: text.footnote(label("Error VARIABLES=Unskilled , name("MEAN_FAC3_32")) text.footnote(label("Error Bars: 95% CI")) MEANCI(FAC3_32, DATA: LOW=col(source(s), Bars: 95% CI")) SCALE: linear(dim(2), 95)[name="MEAN_FAC3_32" name("MEAN_FAC3_32_LOW SCALE: linear(dim(2), include(0)) ")) include(0)) ELEMENT: LOW="MEAN_FAC3_32_LOW DATA: HIGH=col(source(s), ELEMENT: interval(position(Unskilled*M " name("MEAN_FAC3_32_HIGH interval(position(Upskilled*M EAN_FAC1_31), HIGH="MEAN_FAC3_32_HIGH ")) EAN_FAC1_32), shape.interior(shape.square)) "] MISSING=LISTWISE GUIDE: axis(dim(1), shape.interior(shape.square)) ELEMENT: REPORTMISSING=NO label("[ANALYSE] [theme: ELEMENT: interval(position(region.sprea /GRAPHSPEC demographics] [work] Work interval(position(region.sprea d.range(Unskilled*(LOW+HIG SOURCE=INLINE. type middling skilled ", d.range(Upskilled*(LOW+HIG H))), BEGIN GPL "as opposed to never H))), SOURCE: worked")) shape.interior(shape.ibeam)) s=userSource(id("graphdata GUIDE: axis(dim(2), shape.interior(shape.ibeam)) END GPL. set")) label("Mean [ANALYSE] END GPL. DATA: [theme: policy] 3 pro-social * Chart Builder. Unskilled=col(source(s), freedom regression ", * Chart Builder. GGRAPH name("Unskilled"), "factor")) GGRAPH /GRAPHDATA SET unit.category()) GUIDE: /GRAPHDATA SET NAME="graphdata set" DATA: text.footnote(label("Error NAME="graphdata set" VARIABLES=Upskilled MEAN_FAC3_32=col(source(s) Bars: 95% CI")) VARIABLES=Midskilled MEANCI(FAC3_32, , name("MEAN_FAC3_32")) SCALE: linear(dim(2), MEANCI(FAC1_32, 95)[name="MEAN_FAC3_32" DATA: LOW=col(source(s), include(0)) 95)[name="MEAN_FAC1_32" name("MEAN_FAC3_32_LOW ELEMENT: LOW="MEAN_FAC3_32_LOW ")) interval(position(Midskilled* LOW="MEAN_FAC1_32_LOW " DATA: HIGH=col(source(s), MEAN_FAC3_32), " HIGH="MEAN_FAC3_32_HIGH name("MEAN_FAC3_32_HIGH shape.interior(shape.square)) HIGH="MEAN_FAC1_32_HIGH "] MISSING=LISTWISE ")) ELEMENT: "] MISSING=LISTWISE REPORTMISSING=NO GUIDE: axis(dim(1), interval(position(region.sprea REPORTMISSING=NO /GRAPHSPEC label("[ANALYSE] [theme: d.range(Midskilled*(LOW+HIG /GRAPHSPEC SOURCE=INLINE. demographics] [work] Work H))), SOURCE=INLINE. BEGIN GPL type low skilled or ", BEGIN GPL SOURCE: "unskilled as opposed to shape.interior(shape.ibeam)) SOURCE: s=userSource(id("graphdata never worked")) END GPL. s=userSource(id("graphdata set")) GUIDE: axis(dim(2), set")) DATA: label("Mean [ANALYSE] * Chart Builder. DATA: Upskilled=col(source(s), [theme: policy] 3 pro-social GGRAPH Midskilled=col(source(s), name("Upskilled"), freedom regression ", /GRAPHDATA SET name("Midskilled"), unit.category()) "factor")) NAME="graphdata set" unit.category()) DATA: GUIDE: VARIABLES=Upskilled DATA: MEAN_FAC3_32=col(source(s) text.footnote(label("Error MEANCI(FAC1_32, MEAN_FAC1_32=col(source(s) , name("MEAN_FAC3_32")) Bars: 95% CI")) 95)[name="MEAN_FAC1_32" , name("MEAN_FAC1_32")) DATA: LOW=col(source(s), SCALE: linear(dim(2), DATA: LOW=col(source(s), name("MEAN_FAC3_32_LOW include(0)) LOW="MEAN_FAC1_32_LOW name("MEAN_FAC1_32_LOW ")) ELEMENT: " ")) DATA: HIGH=col(source(s), interval(position(Unskilled*M HIGH="MEAN_FAC1_32_HIGH DATA: HIGH=col(source(s), name("MEAN_FAC3_32_HIGH EAN_FAC3_32), "] MISSING=LISTWISE name("MEAN_FAC1_32_HIGH ")) shape.interior(shape.square)) REPORTMISSING=NO ")) GUIDE: axis(dim(1), ELEMENT: /GRAPHSPEC GUIDE: axis(dim(1), label("[ANALYSE] [theme: interval(position(region.sprea SOURCE=INLINE. label("[ANALYSE] [theme: demographics] [work] Work d.range(Unskilled*(LOW+HIG BEGIN GPL demographics] [work] Work type highly skilled as ", H))), SOURCE: type middling skilled ", "opposed to never s=userSource(id("graphdata "as opposed to never worked")) shape.interior(shape.ibeam)) set")) worked")) GUIDE: axis(dim(2), END GPL. DATA: GUIDE: axis(dim(2), label("Mean [ANALYSE] Upskilled=col(source(s), label("Mean [ANALYSE] [theme: policy] 3 pro-social * Chart Builder. name("Upskilled"), [theme: policy] 1 pro- freedom regression ", GGRAPH unit.category()) immigration regression "factor")) /GRAPHDATA SET DATA: factor")) GUIDE: NAME="graphdata set" MEAN_FAC1_32=col(source(s) GUIDE: text.footnote(label("Error VARIABLES=Midskilled , name("MEAN_FAC1_32")) text.footnote(label("Error Bars: 95% CI")) MEANCI(FAC3_32, Bars: 95% CI")) 95)[name="MEAN_FAC3_32"

SCALE: linear(dim(2), DATA: LOW=col(source(s), GUIDE: include(0)) LOW="MEAN_FAC2_32_LOW name("MEAN_FAC2_32_LOW text.footnote(label("Error ELEMENT: " ")) Bars: 95% CI")) interval(position(Midskilled* HIGH="MEAN_FAC2_32_HIGH DATA: HIGH=col(source(s), SCALE: linear(dim(2), MEAN_FAC1_32), "] MISSING=LISTWISE name("MEAN_FAC2_32_HIGH include(0)) shape.interior(shape.square)) REPORTMISSING=NO ")) ELEMENT: ELEMENT: /GRAPHSPEC GUIDE: axis(dim(1), interval(position(Unskilled*M interval(position(region.sprea SOURCE=INLINE. label("[ANALYSE] [theme: EAN_FAC2_32), d.range(Midskilled*(LOW+HIG BEGIN GPL demographics] [work] Work shape.interior(shape.square)) H))), SOURCE: type middling skilled ", ELEMENT: s=userSource(id("graphdata "as opposed to never interval(position(region.sprea shape.interior(shape.ibeam)) set")) worked")) d.range(Unskilled*(LOW+HIG END GPL. DATA: GUIDE: axis(dim(2), H))), Upskilled=col(source(s), label("Mean [ANALYSE] * Chart Builder. name("Upskilled"), [theme: policy] 2 pro- shape.interior(shape.ibeam)) GGRAPH unit.category()) government involvement in ", END GPL. /GRAPHDATA SET DATA: "economy regression NAME="graphdata set" MEAN_FAC2_32=col(source(s) factor")) * Chart Builder. VARIABLES=Unskilled , name("MEAN_FAC2_32")) GUIDE: GGRAPH MEANCI(FAC1_32, DATA: LOW=col(source(s), text.footnote(label("Error /GRAPHDATA SET 95)[name="MEAN_FAC1_32" name("MEAN_FAC2_32_LOW Bars: 95% CI")) NAME="graphdata set" ")) SCALE: linear(dim(2), VARIABLES=anyUniW7 LOW="MEAN_FAC1_32_LOW DATA: HIGH=col(source(s), include(0)) MEANCI(FAC1_31, " name("MEAN_FAC2_32_HIGH ELEMENT: 95)[name="MEAN_FAC1_31" HIGH="MEAN_FAC1_32_HIGH ")) interval(position(Midskilled* "] MISSING=LISTWISE GUIDE: axis(dim(1), MEAN_FAC2_32), LOW="MEAN_FAC1_31_LOW REPORTMISSING=NO label("[ANALYSE] [theme: shape.interior(shape.square)) " /GRAPHSPEC demographics] [work] Work ELEMENT: HIGH="MEAN_FAC1_31_HIGH SOURCE=INLINE. type highly skilled as ", interval(position(region.sprea "] MISSING=LISTWISE BEGIN GPL "opposed to never d.range(Midskilled*(LOW+HIG REPORTMISSING=NO SOURCE: worked")) H))), /GRAPHSPEC s=userSource(id("graphdata GUIDE: axis(dim(2), SOURCE=INLINE. set")) label("Mean [ANALYSE] shape.interior(shape.ibeam)) BEGIN GPL DATA: [theme: policy] 2 pro- END GPL. SOURCE: Unskilled=col(source(s), government involvement in ", s=userSource(id("graphdata name("Unskilled"), "economy regression * Chart Builder. set")) unit.category()) factor")) GGRAPH DATA: DATA: GUIDE: /GRAPHDATA SET anyUniW7=col(source(s), MEAN_FAC1_32=col(source(s) text.footnote(label("Error NAME="graphdata set" name("anyUniW7"), , name("MEAN_FAC1_32")) Bars: 95% CI")) VARIABLES=Unskilled unit.category()) DATA: LOW=col(source(s), SCALE: linear(dim(2), MEANCI(FAC2_32, DATA: name("MEAN_FAC1_32_LOW include(0)) 95)[name="MEAN_FAC2_32" MEAN_FAC1_31=col(source(s) ")) ELEMENT: , name("MEAN_FAC1_31")) DATA: HIGH=col(source(s), interval(position(Upskilled*M LOW="MEAN_FAC2_32_LOW DATA: LOW=col(source(s), name("MEAN_FAC1_32_HIGH EAN_FAC2_32), " name("MEAN_FAC1_31_LOW ")) shape.interior(shape.square)) HIGH="MEAN_FAC2_32_HIGH ")) GUIDE: axis(dim(1), ELEMENT: "] MISSING=LISTWISE DATA: HIGH=col(source(s), label("[ANALYSE] [theme: interval(position(region.sprea REPORTMISSING=NO name("MEAN_FAC1_31_HIGH demographics] [work] Work d.range(Upskilled*(LOW+HIG /GRAPHSPEC ")) type low skilled or ", H))), SOURCE=INLINE. GUIDE: axis(dim(1), "unskilled as opposed to BEGIN GPL label("[ANALYSE] [theme: never worked")) shape.interior(shape.ibeam)) SOURCE: demographics] Attended GUIDE: axis(dim(2), END GPL. s=userSource(id("graphdata university as opposed to ", label("Mean [ANALYSE] set")) "not attended university")) [theme: policy] 1 pro- * Chart Builder. DATA: GUIDE: axis(dim(2), immigration regression GGRAPH Unskilled=col(source(s), label("Mean [ANALYSE] factor")) /GRAPHDATA SET name("Unskilled"), [theme: policy] General GUIDE: NAME="graphdata set" unit.category()) Political Leftness ", text.footnote(label("Error VARIABLES=Midskilled DATA: "regression factor")) Bars: 95% CI")) MEANCI(FAC2_32, MEAN_FAC2_32=col(source(s) GUIDE: SCALE: linear(dim(2), 95)[name="MEAN_FAC2_32" , name("MEAN_FAC2_32")) text.footnote(label("Error include(0)) DATA: LOW=col(source(s), Bars: 95% CI")) ELEMENT: LOW="MEAN_FAC2_32_LOW name("MEAN_FAC2_32_LOW SCALE: cat(dim(1), interval(position(Unskilled*M " ")) include("0", "1")) EAN_FAC1_32), HIGH="MEAN_FAC2_32_HIGH DATA: HIGH=col(source(s), SCALE: linear(dim(2), shape.interior(shape.square)) "] MISSING=LISTWISE name("MEAN_FAC2_32_HIGH include(0)) ELEMENT: REPORTMISSING=NO ")) ELEMENT: interval(position(region.sprea /GRAPHSPEC GUIDE: axis(dim(1), interval(position(anyUniW7* d.range(Unskilled*(LOW+HIG SOURCE=INLINE. label("[ANALYSE] [theme: MEAN_FAC1_31), H))), BEGIN GPL demographics] [work] Work shape.interior(shape.square)) SOURCE: type low skilled or ", ELEMENT: shape.interior(shape.ibeam)) s=userSource(id("graphdata "unskilled as opposed to interval(position(region.sprea END GPL. set")) never worked")) d.range(anyUniW7*(LOW+HI DATA: GUIDE: axis(dim(2), GH))), * Chart Builder. Midskilled=col(source(s), label("Mean [ANALYSE] shape.interior(shape.ibeam)) GGRAPH name("Midskilled"), [theme: policy] 2 pro- END GPL. /GRAPHDATA SET unit.category()) government involvement in ", NAME="graphdata set" DATA: "economy regression * Chart Builder. VARIABLES=Upskilled MEAN_FAC2_32=col(source(s) factor")) GGRAPH MEANCI(FAC2_32, , name("MEAN_FAC2_32")) /GRAPHDATA SET 95)[name="MEAN_FAC2_32" NAME="graphdata set"

VARIABLES=anyUniW7 DATA: GUIDE: MEANCI(FAC1_31, MEANCI(FAC3_32, MEAN_FAC1_32=col(source(s) text.footnote(label("Error 95)[name="MEAN_FAC1_31" 95)[name="MEAN_FAC3_32" , name("MEAN_FAC1_32")) Bars: 95% CI")) DATA: LOW=col(source(s), SCALE: cat(dim(1), LOW="MEAN_FAC1_31_LOW LOW="MEAN_FAC3_32_LOW name("MEAN_FAC1_32_LOW include("0", "1")) " " ")) SCALE: linear(dim(2), HIGH="MEAN_FAC1_31_HIGH HIGH="MEAN_FAC3_32_HIGH DATA: HIGH=col(source(s), include(0)) "] MISSING=LISTWISE "] MISSING=LISTWISE name("MEAN_FAC1_32_HIGH ELEMENT: REPORTMISSING=NO REPORTMISSING=NO ")) interval(position(anyUniW7* /GRAPHSPEC /GRAPHSPEC GUIDE: axis(dim(1), MEAN_FAC2_32), SOURCE=INLINE. SOURCE=INLINE. label("[ANALYSE] [theme: shape.interior(shape.square)) BEGIN GPL BEGIN GPL demographics] Attended ELEMENT: SOURCE: SOURCE: university as opposed to ", interval(position(region.sprea s=userSource(id("graphdata s=userSource(id("graphdata "not attended university")) d.range(anyUniW7*(LOW+HI set")) set")) GUIDE: axis(dim(2), GH))), DATA: DATA: label("Mean [ANALYSE] shape.interior(shape.ibeam)) disabilityW6=col(source(s), anyUniW7=col(source(s), [theme: policy] 1 pro- END GPL. name("disabilityW6"), name("anyUniW7"), immigration regression unit.category()) unit.category()) factor")) * Chart Builder. DATA: DATA: GUIDE: GGRAPH MEAN_FAC1_31=col(source(s) MEAN_FAC3_32=col(source(s) text.footnote(label("Error /GRAPHDATA SET , name("MEAN_FAC1_31")) , name("MEAN_FAC3_32")) Bars: 95% CI")) NAME="graphdata set" DATA: LOW=col(source(s), DATA: LOW=col(source(s), SCALE: cat(dim(1), VARIABLES=gender name("MEAN_FAC1_31_LOW name("MEAN_FAC3_32_LOW include("0", "1")) MEANCI(FAC1_31, ")) ")) SCALE: linear(dim(2), 95)[name="MEAN_FAC1_31" DATA: HIGH=col(source(s), DATA: HIGH=col(source(s), include(0)) name("MEAN_FAC1_31_HIGH name("MEAN_FAC3_32_HIGH ELEMENT: LOW="MEAN_FAC1_31_LOW ")) ")) interval(position(anyUniW7* " GUIDE: axis(dim(1), GUIDE: axis(dim(1), MEAN_FAC1_32), HIGH="MEAN_FAC1_31_HIGH label("[ANALYSE] [theme: label("[ANALYSE] [theme: shape.interior(shape.square)) "] MISSING=LISTWISE demographics] Disabled as demographics] Attended ELEMENT: REPORTMISSING=NO opposed to not disabled")) university as opposed to ", interval(position(region.sprea /GRAPHSPEC GUIDE: axis(dim(2), "not attended university")) d.range(anyUniW7*(LOW+HI SOURCE=INLINE. label("Mean [ANALYSE] GUIDE: axis(dim(2), GH))), BEGIN GPL [theme: policy] General label("Mean [ANALYSE] shape.interior(shape.ibeam)) SOURCE: Political Leftness ", [theme: policy] 3 pro-social END GPL. s=userSource(id("graphdata "regression factor")) freedom regression ", set")) GUIDE: "factor")) * Chart Builder. DATA: gender=col(source(s), text.footnote(label("Error GUIDE: GGRAPH name("gender"), Bars: 95% CI")) text.footnote(label("Error /GRAPHDATA SET unit.category()) SCALE: cat(dim(1), Bars: 95% CI")) NAME="graphdata set" DATA: include("0", "1")) SCALE: cat(dim(1), VARIABLES=anyUniW7 MEAN_FAC1_31=col(source(s) SCALE: linear(dim(2), include("0", "1")) MEANCI(FAC2_32, , name("MEAN_FAC1_31")) include(0)) SCALE: linear(dim(2), 95)[name="MEAN_FAC2_32" DATA: LOW=col(source(s), ELEMENT: include(0)) name("MEAN_FAC1_31_LOW interval(position(disabilityW6 ELEMENT: LOW="MEAN_FAC2_32_LOW ")) *MEAN_FAC1_31), interval(position(anyUniW7* " DATA: HIGH=col(source(s), shape.interior(shape.square)) MEAN_FAC3_32), HIGH="MEAN_FAC2_32_HIGH name("MEAN_FAC1_31_HIGH ELEMENT: shape.interior(shape.square)) "] MISSING=LISTWISE ")) interval(position(region.sprea ELEMENT: REPORTMISSING=NO GUIDE: axis(dim(1), d.range(disabilityW6*(LOW+H interval(position(region.sprea /GRAPHSPEC label("[ANALYSE] [theme: IGH))), d.range(anyUniW7*(LOW+HI SOURCE=INLINE. demographics] Gender female GH))), BEGIN GPL as opposed to male")) shape.interior(shape.ibeam)) shape.interior(shape.ibeam)) SOURCE: GUIDE: axis(dim(2), END GPL. END GPL. s=userSource(id("graphdata label("Mean [ANALYSE] set")) [theme: policy] General * Chart Builder. * Chart Builder. DATA: Political Leftness ", GGRAPH GGRAPH anyUniW7=col(source(s), "regression factor")) /GRAPHDATA SET /GRAPHDATA SET name("anyUniW7"), GUIDE: NAME="graphdata set" NAME="graphdata set" unit.category()) text.footnote(label("Error VARIABLES=gender VARIABLES=anyUniW7 DATA: Bars: 95% CI")) MEANCI(FAC3_32, MEANCI(FAC1_32, MEAN_FAC2_32=col(source(s) SCALE: cat(dim(1), 95)[name="MEAN_FAC3_32" 95)[name="MEAN_FAC1_32" , name("MEAN_FAC2_32")) include("0", "1")) DATA: LOW=col(source(s), SCALE: linear(dim(2), LOW="MEAN_FAC3_32_LOW LOW="MEAN_FAC1_32_LOW name("MEAN_FAC2_32_LOW include(0)) " " ")) ELEMENT: HIGH="MEAN_FAC3_32_HIGH HIGH="MEAN_FAC1_32_HIGH DATA: HIGH=col(source(s), interval(position(gender*MEA "] MISSING=LISTWISE "] MISSING=LISTWISE name("MEAN_FAC2_32_HIGH N_FAC1_31), REPORTMISSING=NO REPORTMISSING=NO ")) shape.interior(shape.square)) /GRAPHSPEC /GRAPHSPEC GUIDE: axis(dim(1), ELEMENT: SOURCE=INLINE. SOURCE=INLINE. label("[ANALYSE] [theme: interval(position(region.sprea BEGIN GPL BEGIN GPL demographics] Attended d.range(gender*(LOW+HIGH)) SOURCE: SOURCE: university as opposed to ", ), s=userSource(id("graphdata s=userSource(id("graphdata "not attended university")) shape.interior(shape.ibeam)) set")) set")) GUIDE: axis(dim(2), END GPL. DATA: gender=col(source(s), DATA: label("Mean [ANALYSE] name("gender"), anyUniW7=col(source(s), [theme: policy] 2 pro- * Chart Builder. unit.category()) name("anyUniW7"), government involvement in ", GGRAPH DATA: unit.category()) "economy regression /GRAPHDATA SET MEAN_FAC3_32=col(source(s) factor")) NAME="graphdata set" , name("MEAN_FAC3_32")) VARIABLES=disabilityW6

DATA: LOW=col(source(s), ELEMENT: " GUIDE: axis(dim(1), name("MEAN_FAC3_32_LOW interval(position(disabilityW6 HIGH="MEAN_FAC1_32_HIGH label("[ANALYSE] [theme: ")) *MEAN_FAC3_32), "] MISSING=LISTWISE demographics] Gender female DATA: HIGH=col(source(s), shape.interior(shape.square)) REPORTMISSING=NO as opposed to male")) name("MEAN_FAC3_32_HIGH ELEMENT: /GRAPHSPEC GUIDE: axis(dim(2), ")) interval(position(region.sprea SOURCE=INLINE. label("Mean [ANALYSE] GUIDE: axis(dim(1), d.range(disabilityW6*(LOW+H BEGIN GPL [theme: policy] 2 pro- label("[ANALYSE] [theme: IGH))), SOURCE: government involvement in ", demographics] Gender female s=userSource(id("graphdata "economy regression as opposed to male")) shape.interior(shape.ibeam)) set")) factor")) GUIDE: axis(dim(2), END GPL. DATA: gender=col(source(s), GUIDE: label("Mean [ANALYSE] name("gender"), text.footnote(label("Error [theme: policy] 3 pro-social * Chart Builder. unit.category()) Bars: 95% CI")) freedom regression ", GGRAPH DATA: SCALE: cat(dim(1), "factor")) /GRAPHDATA SET MEAN_FAC1_32=col(source(s) include("0", "1")) GUIDE: NAME="graphdata set" , name("MEAN_FAC1_32")) SCALE: linear(dim(2), text.footnote(label("Error VARIABLES=disabilityW6 DATA: LOW=col(source(s), include(0)) Bars: 95% CI")) MEANCI(FAC1_32, name("MEAN_FAC1_32_LOW ELEMENT: SCALE: cat(dim(1), 95)[name="MEAN_FAC1_32" ")) interval(position(gender*MEA include("0", "1")) DATA: HIGH=col(source(s), N_FAC2_32), SCALE: linear(dim(2), LOW="MEAN_FAC1_32_LOW name("MEAN_FAC1_32_HIGH shape.interior(shape.square)) include(0)) " ")) ELEMENT: ELEMENT: HIGH="MEAN_FAC1_32_HIGH GUIDE: axis(dim(1), interval(position(region.sprea interval(position(gender*MEA "] MISSING=LISTWISE label("[ANALYSE] [theme: d.range(gender*(LOW+HIGH)) N_FAC3_32), REPORTMISSING=NO demographics] Gender female ), shape.interior(shape.square)) /GRAPHSPEC as opposed to male")) shape.interior(shape.ibeam)) ELEMENT: SOURCE=INLINE. GUIDE: axis(dim(2), END GPL. interval(position(region.sprea BEGIN GPL label("Mean [ANALYSE] d.range(gender*(LOW+HIGH)) SOURCE: [theme: policy] 1 pro- * Chart Builder. ), s=userSource(id("graphdata immigration regression GGRAPH shape.interior(shape.ibeam)) set")) factor")) /GRAPHDATA SET END GPL. DATA: GUIDE: NAME="graphdata set" disabilityW6=col(source(s), text.footnote(label("Error VARIABLES=disabilityW6 * Chart Builder. name("disabilityW6"), Bars: 95% CI")) MEANCI(FAC2_32, GGRAPH unit.category()) SCALE: cat(dim(1), 95)[name="MEAN_FAC2_32" /GRAPHDATA SET DATA: include("0", "1")) NAME="graphdata set" MEAN_FAC1_32=col(source(s) SCALE: linear(dim(2), LOW="MEAN_FAC2_32_LOW VARIABLES=disabilityW6 , name("MEAN_FAC1_32")) include(0)) " MEANCI(FAC3_32, DATA: LOW=col(source(s), ELEMENT: HIGH="MEAN_FAC2_32_HIGH 95)[name="MEAN_FAC3_32" name("MEAN_FAC1_32_LOW interval(position(gender*MEA "] MISSING=LISTWISE ")) N_FAC1_32), REPORTMISSING=NO LOW="MEAN_FAC3_32_LOW DATA: HIGH=col(source(s), shape.interior(shape.square)) /GRAPHSPEC " name("MEAN_FAC1_32_HIGH ELEMENT: SOURCE=INLINE. HIGH="MEAN_FAC3_32_HIGH ")) interval(position(region.sprea BEGIN GPL "] MISSING=LISTWISE GUIDE: axis(dim(1), d.range(gender*(LOW+HIGH)) SOURCE: REPORTMISSING=NO label("[ANALYSE] [theme: ), s=userSource(id("graphdata /GRAPHSPEC demographics] Disabled as shape.interior(shape.ibeam)) set")) SOURCE=INLINE. opposed to not disabled")) END GPL. DATA: BEGIN GPL GUIDE: axis(dim(2), disabilityW6=col(source(s), SOURCE: label("Mean [ANALYSE] * Chart Builder. name("disabilityW6"), s=userSource(id("graphdata [theme: policy] 1 pro- GGRAPH unit.category()) set")) immigration regression /GRAPHDATA SET DATA: DATA: factor")) NAME="graphdata set" MEAN_FAC2_32=col(source(s) disabilityW6=col(source(s), GUIDE: VARIABLES=gender , name("MEAN_FAC2_32")) name("disabilityW6"), text.footnote(label("Error MEANCI(FAC2_32, DATA: LOW=col(source(s), unit.category()) Bars: 95% CI")) 95)[name="MEAN_FAC2_32" name("MEAN_FAC2_32_LOW DATA: SCALE: cat(dim(1), ")) MEAN_FAC3_32=col(source(s) include("0", "1")) LOW="MEAN_FAC2_32_LOW DATA: HIGH=col(source(s), , name("MEAN_FAC3_32")) SCALE: linear(dim(2), " name("MEAN_FAC2_32_HIGH DATA: LOW=col(source(s), include(0)) HIGH="MEAN_FAC2_32_HIGH ")) name("MEAN_FAC3_32_LOW ELEMENT: "] MISSING=LISTWISE GUIDE: axis(dim(1), ")) interval(position(disabilityW6 REPORTMISSING=NO label("[ANALYSE] [theme: DATA: HIGH=col(source(s), *MEAN_FAC1_32), /GRAPHSPEC demographics] Disabled as name("MEAN_FAC3_32_HIGH shape.interior(shape.square)) SOURCE=INLINE. opposed to not disabled")) ")) ELEMENT: BEGIN GPL GUIDE: axis(dim(2), GUIDE: axis(dim(1), interval(position(region.sprea SOURCE: label("Mean [ANALYSE] label("[ANALYSE] [theme: d.range(disabilityW6*(LOW+H s=userSource(id("graphdata [theme: policy] 2 pro- demographics] Disabled as IGH))), set")) government involvement in ", opposed to not disabled")) DATA: gender=col(source(s), "economy regression GUIDE: axis(dim(2), shape.interior(shape.ibeam)) name("gender"), factor")) label("Mean [ANALYSE] END GPL. unit.category()) GUIDE: [theme: policy] 3 pro-social DATA: text.footnote(label("Error freedom regression ", * Chart Builder. MEAN_FAC2_32=col(source(s) Bars: 95% CI")) "factor")) GGRAPH , name("MEAN_FAC2_32")) SCALE: cat(dim(1), GUIDE: /GRAPHDATA SET DATA: LOW=col(source(s), include("0", "1")) text.footnote(label("Error NAME="graphdata set" name("MEAN_FAC2_32_LOW SCALE: linear(dim(2), Bars: 95% CI")) VARIABLES=gender ")) include(0)) SCALE: cat(dim(1), MEANCI(FAC1_32, DATA: HIGH=col(source(s), ELEMENT: include("0", "1")) 95)[name="MEAN_FAC1_32" name("MEAN_FAC2_32_HIGH interval(position(disabilityW6 SCALE: linear(dim(2), ")) *MEAN_FAC2_32), include(0)) LOW="MEAN_FAC1_32_LOW shape.interior(shape.square))

ELEMENT: lCulturalPoliticalConformity interval(position(region.sprea WITH FAC1_31 /SCATTERPLOT(BIVAR)=FAC2_ LOW="MEAN_FAC1_24_LOW d.range(disabilityW6*(LOW+H /MISSING=LISTWISE. 21 WITH FAC1_31 " IGH))), /MISSING=LISTWISE. HIGH="MEAN_FAC1_24_HIGH GRAPH "] MISSING=LISTWISE shape.interior(shape.ibeam)) GRAPH REPORTMISSING=NO END GPL. /SCATTERPLOT(BIVAR)=BeliefS /GRAPHSPEC ystemConstraint WITH /SCATTERPLOT(BIVAR)=Reflec SOURCE=INLINE. FREQUENCIES FAC1_31 tedLnReflectedItemResponse BEGIN GPL VARIABLES=FAC1_23 /MISSING=LISTWISE. Rate WITH FAC2_32 SOURCE: /FORMAT=NOTABLE /MISSING=LISTWISE. s=userSource(id("graphdata /NTILES=4 GRAPH set")) /STATISTICS=STDDEV GRAPH DATA: MINIMUM MAXIMUM MEAN /SCATTERPLOT(BIVAR)=BeliefS Broadsheet=col(source(s), MEDIAN SKEWNESS SESKEW ystemConstraint WITH /SCATTERPLOT(BIVAR)=FAC1_ name("Broadsheet"), KURTOSIS SEKURT FAC1_32 24 WITH FAC2_32 unit.category()) /HISTOGRAM NORMAL /MISSING=LISTWISE. /MISSING=LISTWISE. DATA: /ORDER=ANALYSIS. MEAN_FAC1_24=col(source(s) GRAPH CORRELATIONS , name("MEAN_FAC1_24")) FREQUENCIES /VARIABLES=FAC1_20 DATA: LOW=col(source(s), VARIABLES=OverallCulturalPol /SCATTERPLOT(BIVAR)=BeliefS angerScale fearScale FAC1_21 name("MEAN_FAC1_24_LOW iticalConformity ystemConstraint WITH FAC2_21 ")) /FORMAT=NOTABLE FAC2_32 ReflectedLnReflectedItemRes DATA: HIGH=col(source(s), /NTILES=4 /MISSING=LISTWISE. ponseRate name("MEAN_FAC1_24_HIGH /STATISTICS=STDDEV FAC1_24 ")) MINIMUM MAXIMUM MEAN GRAPH /PRINT=TWOTAIL NOSIG GUIDE: axis(dim(1), MEDIAN SKEWNESS SESKEW /MISSING=PAIRWISE. label("[ANALYSE] [theme: KURTOSIS SEKURT /SCATTERPLOT(BIVAR)=BeliefS demographics] [readership] /HISTOGRAM NORMAL ystemConstraint WITH CORRELATIONS Broadsheet newspaper ", /ORDER=ANALYSIS. FAC3_32 /VARIABLES=FAC1_23 "readership as opposed to /MISSING=LISTWISE. OverallCulturalPoliticalConfor other or none")) FREQUENCIES mity BeliefSystemConstraint GUIDE: axis(dim(2), VARIABLES=BeliefSystemCons GRAPH /PRINT=TWOTAIL NOSIG label("Mean [ANALYSE] traint /MISSING=PAIRWISE. [theme: psychology] Response /FORMAT=NOTABLE /SCATTERPLOT(BIVAR)=ageW polarisation chief ", /NTILES=4 7 WITH FAC1_31 GRAPH "regression factor")) /STATISTICS=STDDEV /MISSING=LISTWISE. GUIDE: MINIMUM MAXIMUM MEAN /BAR(SIMPLE)=MEAN(FAC1_2 text.footnote(label("Error MEDIAN SKEWNESS SESKEW GRAPH 3) BY TSC_5887 Bars: 95% CI")) KURTOSIS SEKURT /INTERVAL CI(95.0). SCALE: linear(dim(2), /HISTOGRAM NORMAL /SCATTERPLOT(BIVAR)=FAC1_ include(0)) /ORDER=ANALYSIS. 13 WITH FAC1_31 GRAPH ELEMENT: /MISSING=LISTWISE. interval(position(Broadsheet* FREQUENCIES /BAR(SIMPLE)=MEAN(Overall MEAN_FAC1_24), VARIABLES=ageW7 GRAPH CulturalPoliticalConformity) shape.interior(shape.square)) /FORMAT=NOTABLE BY TSC_5887 ELEMENT: /NTILES=4 /SCATTERPLOT(BIVAR)=FAC1_ /INTERVAL CI(95.0). interval(position(region.sprea /STATISTICS=STDDEV 15 WITH FAC1_31 d.range(Broadsheet*(LOW+HI MINIMUM MAXIMUM MEAN /MISSING=LISTWISE. GRAPH GH))), MEDIAN SKEWNESS SESKEW KURTOSIS SEKURT GRAPH /BAR(SIMPLE)=MEAN(BeliefSy shape.interior(shape.ibeam)) /HISTOGRAM NORMAL stemConstraint) BY TSC_5887 END GPL. /ORDER=ANALYSIS. /SCATTERPLOT(BIVAR)=FAC1_ /INTERVAL CI(95.0). 16 WITH FAC1_31 * Chart Builder. FREQUENCIES /MISSING=LISTWISE. CORRELATIONS GGRAPH VARIABLES=FAC1_13 FAC1_15 /VARIABLES=FAC1_20 /GRAPHDATA SET FAC1_16 FAC1_19 FAC1_32 GRAPH FAC1_21 FAC1_24 FAC1_23 NAME="graphdata set" FAC2_32 FAC3_32 FAC1_31 OverallCulturalPoliticalConfor VARIABLES=Tabloid FAC1_20 /SCATTERPLOT(BIVAR)=FAC1_ mity MEANCI(FAC1_24, angerScale fearScale 19 WITH FAC1_31 BeliefSystemConstraint 95)[name="MEAN_FAC1_24" FAC1_21 FAC2_21 /MISSING=LISTWISE. ageW7 FAC1_13 FAC1_15 ReflectedLnReflectedItemRes FAC1_16 FAC1_19 LOW="MEAN_FAC1_24_LOW ponseRate FAC1_24 GRAPH /PRINT=TWOTAIL NOSIG " /FORMAT=NOTABLE /MISSING=PAIRWISE. HIGH="MEAN_FAC1_24_HIGH /NTILES=4 /SCATTERPLOT(BIVAR)=FAC1_ "] MISSING=LISTWISE /STATISTICS=STDDEV 20 WITH FAC1_31 CORRELATIONS REPORTMISSING=NO MINIMUM MAXIMUM MEAN /MISSING=LISTWISE. /VARIABLES=FAC1_23 /GRAPHSPEC MEDIAN SKEWNESS SESKEW OverallCulturalPoliticalConfor SOURCE=INLINE. KURTOSIS SEKURT GRAPH mity BeliefSystemConstraint BEGIN GPL /HISTOGRAM NORMAL ageW7 FAC1_13 SOURCE: /ORDER=ANALYSIS. /SCATTERPLOT(BIVAR)=angerS FAC1_15 FAC1_16 FAC1_19 s=userSource(id("graphdata cale WITH FAC1_31 /PRINT=TWOTAIL NOSIG set")) GRAPH /MISSING=LISTWISE. /MISSING=PAIRWISE. DATA: Tabloid=col(source(s), name("Tabloid"), /SCATTERPLOT(BIVAR)=FAC1_ GRAPH * Chart Builder. unit.category()) 23 WITH FAC1_31 GGRAPH DATA: /MISSING=LISTWISE. /SCATTERPLOT(BIVAR)=FAC1_ /GRAPHDATA SET MEAN_FAC1_24=col(source(s) 21 WITH FAC1_31 NAME="graphdata set" , name("MEAN_FAC1_24")) GRAPH /MISSING=LISTWISE. VARIABLES=Broadsheet DATA: LOW=col(source(s), MEANCI(FAC1_24, name("MEAN_FAC1_24_LOW /SCATTERPLOT(BIVAR)=Overal GRAPH 95)[name="MEAN_FAC1_24" "))

DATA: HIGH=col(source(s), N_FAC1_21), "] MISSING=LISTWISE GUIDE: axis(dim(1), name("MEAN_FAC1_24_HIGH shape.interior(shape.square)) REPORTMISSING=NO label("[ANALYSE] [theme: ")) ELEMENT: /GRAPHSPEC demographics] [readership] GUIDE: axis(dim(1), interval(position(region.sprea SOURCE=INLINE. Tabloid newspaper ", label("[ANALYSE] [theme: d.range(Tabloid*(LOW+HIGH) BEGIN GPL "readership as opposed to demographics] [readership] )), SOURCE: other or none")) Tabloid newspaper ", shape.interior(shape.ibeam)) s=userSource(id("graphdata GUIDE: axis(dim(2), "readership as opposed to END GPL. set")) label("Mean [ANALYSE] other or none")) DATA: [theme: psychology] [aom] GUIDE: axis(dim(2), * Chart Builder. Broadsheet=col(source(s), Active open-mindedness ", label("Mean [ANALYSE] GGRAPH name("Broadsheet"), "regression factor")) [theme: psychology] Response /GRAPHDATA SET unit.category()) GUIDE: polarisation chief ", NAME="graphdata set" DATA: text.footnote(label("Error "regression factor")) VARIABLES=Broadsheet MEAN_FAC1_20=col(source(s) Bars: 95% CI")) GUIDE: MEANCI(FAC1_21, , name("MEAN_FAC1_20")) SCALE: linear(dim(2), text.footnote(label("Error 95)[name="MEAN_FAC1_21" DATA: LOW=col(source(s), include(0)) Bars: 95% CI")) name("MEAN_FAC1_20_LOW ELEMENT: SCALE: linear(dim(2), LOW="MEAN_FAC1_21_LOW ")) interval(position(Tabloid*MEA include(0)) " DATA: HIGH=col(source(s), N_FAC1_20), ELEMENT: HIGH="MEAN_FAC1_21_HIGH name("MEAN_FAC1_20_HIGH shape.interior(shape.square)) interval(position(Tabloid*MEA "] MISSING=LISTWISE ")) ELEMENT: N_FAC1_24), REPORTMISSING=NO GUIDE: axis(dim(1), interval(position(region.sprea shape.interior(shape.square)) /GRAPHSPEC label("[ANALYSE] [theme: d.range(Tabloid*(LOW+HIGH) ELEMENT: SOURCE=INLINE. demographics] [readership] )), interval(position(region.sprea BEGIN GPL Broadsheet newspaper ", shape.interior(shape.ibeam)) d.range(Tabloid*(LOW+HIGH) SOURCE: "readership as opposed to END GPL. )), s=userSource(id("graphdata other or none")) shape.interior(shape.ibeam)) set")) GUIDE: axis(dim(2), * Chart Builder. END GPL. DATA: label("Mean [ANALYSE] GGRAPH Broadsheet=col(source(s), [theme: psychology] [aom] /GRAPHDATA SET * Chart Builder. name("Broadsheet"), Active open-mindedness ", NAME="graphdata set" GGRAPH unit.category()) "regression factor")) VARIABLES=Upskilled /GRAPHDATA SET DATA: GUIDE: MEANCI(FAC1_20, NAME="graphdata set" MEAN_FAC1_21=col(source(s) text.footnote(label("Error 95)[name="MEAN_FAC1_20" VARIABLES=Tabloid , name("MEAN_FAC1_21")) Bars: 95% CI")) MEANCI(FAC1_21, DATA: LOW=col(source(s), SCALE: linear(dim(2), LOW="MEAN_FAC1_20_LOW 95)[name="MEAN_FAC1_21" name("MEAN_FAC1_21_LOW include(0)) " ")) ELEMENT: HIGH="MEAN_FAC1_20_HIGH LOW="MEAN_FAC1_21_LOW DATA: HIGH=col(source(s), interval(position(Broadsheet* "] MISSING=LISTWISE " name("MEAN_FAC1_21_HIGH MEAN_FAC1_20), REPORTMISSING=NO HIGH="MEAN_FAC1_21_HIGH ")) shape.interior(shape.square)) /GRAPHSPEC "] MISSING=LISTWISE GUIDE: axis(dim(1), ELEMENT: SOURCE=INLINE. REPORTMISSING=NO label("[ANALYSE] [theme: interval(position(region.sprea BEGIN GPL /GRAPHSPEC demographics] [readership] d.range(Broadsheet*(LOW+HI SOURCE: SOURCE=INLINE. Broadsheet newspaper ", GH))), s=userSource(id("graphdata BEGIN GPL "readership as opposed to set")) SOURCE: other or none")) shape.interior(shape.ibeam)) DATA: s=userSource(id("graphdata GUIDE: axis(dim(2), END GPL. Upskilled=col(source(s), set")) label("Mean [ANALYSE] name("Upskilled"), DATA: Tabloid=col(source(s), [theme: psychology] [risk] 1 * Chart Builder. unit.category()) name("Tabloid"), Tolerance of ", GGRAPH DATA: unit.category()) "uncertainty regression /GRAPHDATA SET MEAN_FAC1_20=col(source(s) DATA: factor")) NAME="graphdata set" , name("MEAN_FAC1_20")) MEAN_FAC1_21=col(source(s) GUIDE: VARIABLES=Tabloid DATA: LOW=col(source(s), , name("MEAN_FAC1_21")) text.footnote(label("Error MEANCI(FAC1_20, name("MEAN_FAC1_20_LOW DATA: LOW=col(source(s), Bars: 95% CI")) 95)[name="MEAN_FAC1_20" ")) name("MEAN_FAC1_21_LOW SCALE: linear(dim(2), DATA: HIGH=col(source(s), ")) include(0)) LOW="MEAN_FAC1_20_LOW name("MEAN_FAC1_20_HIGH DATA: HIGH=col(source(s), ELEMENT: " ")) name("MEAN_FAC1_21_HIGH interval(position(Broadsheet* HIGH="MEAN_FAC1_20_HIGH GUIDE: axis(dim(1), ")) MEAN_FAC1_21), "] MISSING=LISTWISE label("[ANALYSE] [theme: GUIDE: axis(dim(1), shape.interior(shape.square)) REPORTMISSING=NO demographics] [work] Work label("[ANALYSE] [theme: ELEMENT: /GRAPHSPEC type highly skilled as ", demographics] [readership] interval(position(region.sprea SOURCE=INLINE. "opposed to never Tabloid newspaper ", d.range(Broadsheet*(LOW+HI BEGIN GPL worked")) "readership as opposed to GH))), SOURCE: GUIDE: axis(dim(2), other or none")) s=userSource(id("graphdata label("Mean [ANALYSE] GUIDE: axis(dim(2), shape.interior(shape.ibeam)) set")) [theme: psychology] [aom] label("Mean [ANALYSE] END GPL. DATA: Tabloid=col(source(s), Active open-mindedness ", [theme: psychology] [risk] 1 name("Tabloid"), "regression factor")) Tolerance of ", * Chart Builder. unit.category()) GUIDE: "uncertainty regression GGRAPH DATA: text.footnote(label("Error factor")) /GRAPHDATA SET MEAN_FAC1_20=col(source(s) Bars: 95% CI")) GUIDE: NAME="graphdata set" , name("MEAN_FAC1_20")) SCALE: linear(dim(2), text.footnote(label("Error VARIABLES=Broadsheet DATA: LOW=col(source(s), include(0)) Bars: 95% CI")) MEANCI(FAC1_20, name("MEAN_FAC1_20_LOW ELEMENT: SCALE: linear(dim(2), 95)[name="MEAN_FAC1_20" ")) interval(position(Upskilled*M include(0)) DATA: HIGH=col(source(s), EAN_FAC1_20), ELEMENT: LOW="MEAN_FAC1_20_LOW name("MEAN_FAC1_20_HIGH shape.interior(shape.square)) interval(position(Tabloid*MEA " ")) ELEMENT: HIGH="MEAN_FAC1_20_HIGH interval(position(region.sprea

93 d.range(Upskilled*(LOW+HIG BEGIN GPL "unskilled as opposed to d.range(Midskilled*(LOW+HIG H))), SOURCE: never worked")) H))), s=userSource(id("graphdata GUIDE: axis(dim(2), shape.interior(shape.ibeam)) set")) label("Mean [ANALYSE] shape.interior(shape.ibeam)) END GPL. DATA: [theme: psychology] [risk] 1 END GPL. Unskilled=col(source(s), Tolerance of ", * Chart Builder. name("Unskilled"), "uncertainty regression * Chart Builder. GGRAPH unit.category()) factor")) GGRAPH /GRAPHDATA SET DATA: GUIDE: /GRAPHDATA SET NAME="graphdata set" MEAN_FAC1_20=col(source(s) text.footnote(label("Error NAME="graphdata set" VARIABLES=Midskilled , name("MEAN_FAC1_20")) Bars: 95% CI")) VARIABLES=Upskilled MEANCI(FAC1_20, DATA: LOW=col(source(s), SCALE: linear(dim(2), MEANCI(FAC1_21, 95)[name="MEAN_FAC1_20" name("MEAN_FAC1_20_LOW include(0)) 95)[name="MEAN_FAC1_21" ")) ELEMENT: LOW="MEAN_FAC1_20_LOW DATA: HIGH=col(source(s), interval(position(Unskilled*M LOW="MEAN_FAC1_21_LOW " name("MEAN_FAC1_20_HIGH EAN_FAC1_21), " HIGH="MEAN_FAC1_20_HIGH ")) shape.interior(shape.square)) HIGH="MEAN_FAC1_21_HIGH "] MISSING=LISTWISE GUIDE: axis(dim(1), ELEMENT: "] MISSING=LISTWISE REPORTMISSING=NO label("[ANALYSE] [theme: interval(position(region.sprea REPORTMISSING=NO /GRAPHSPEC demographics] [work] Work d.range(Unskilled*(LOW+HIG /GRAPHSPEC SOURCE=INLINE. type low skilled or ", H))), SOURCE=INLINE. BEGIN GPL "unskilled as opposed to BEGIN GPL SOURCE: never worked")) shape.interior(shape.ibeam)) SOURCE: s=userSource(id("graphdata GUIDE: axis(dim(2), END GPL. s=userSource(id("graphdata set")) label("Mean [ANALYSE] set")) DATA: [theme: psychology] [aom] * Chart Builder. DATA: Midskilled=col(source(s), Active open-mindedness ", GGRAPH Upskilled=col(source(s), name("Midskilled"), "regression factor")) /GRAPHDATA SET name("Upskilled"), unit.category()) GUIDE: NAME="graphdata set" unit.category()) DATA: text.footnote(label("Error VARIABLES=Midskilled DATA: MEAN_FAC1_20=col(source(s) Bars: 95% CI")) MEANCI(FAC1_21, MEAN_FAC1_21=col(source(s) , name("MEAN_FAC1_20")) SCALE: linear(dim(2), 95)[name="MEAN_FAC1_21" , name("MEAN_FAC1_21")) DATA: LOW=col(source(s), include(0)) DATA: LOW=col(source(s), name("MEAN_FAC1_20_LOW ELEMENT: LOW="MEAN_FAC1_21_LOW name("MEAN_FAC1_21_LOW ")) interval(position(Unskilled*M " ")) DATA: HIGH=col(source(s), EAN_FAC1_20), HIGH="MEAN_FAC1_21_HIGH DATA: HIGH=col(source(s), name("MEAN_FAC1_20_HIGH shape.interior(shape.square)) "] MISSING=LISTWISE name("MEAN_FAC1_21_HIGH ")) ELEMENT: REPORTMISSING=NO ")) GUIDE: axis(dim(1), interval(position(region.sprea /GRAPHSPEC GUIDE: axis(dim(1), label("[ANALYSE] [theme: d.range(Unskilled*(LOW+HIG SOURCE=INLINE. label("[ANALYSE] [theme: demographics] [work] Work H))), BEGIN GPL demographics] [work] Work type middling skilled ", SOURCE: type highly skilled as ", "as opposed to never shape.interior(shape.ibeam)) s=userSource(id("graphdata "opposed to never worked")) END GPL. set")) worked")) GUIDE: axis(dim(2), DATA: GUIDE: axis(dim(2), label("Mean [ANALYSE] * Chart Builder. Midskilled=col(source(s), label("Mean [ANALYSE] [theme: psychology] [aom] GGRAPH name("Midskilled"), [theme: psychology] [risk] 1 Active open-mindedness ", /GRAPHDATA SET unit.category()) Tolerance of ", "regression factor")) NAME="graphdata set" DATA: "uncertainty regression GUIDE: VARIABLES=Unskilled MEAN_FAC1_21=col(source(s) factor")) text.footnote(label("Error MEANCI(FAC1_21, , name("MEAN_FAC1_21")) GUIDE: Bars: 95% CI")) 95)[name="MEAN_FAC1_21" DATA: LOW=col(source(s), text.footnote(label("Error SCALE: linear(dim(2), name("MEAN_FAC1_21_LOW Bars: 95% CI")) include(0)) LOW="MEAN_FAC1_21_LOW ")) SCALE: linear(dim(2), ELEMENT: " DATA: HIGH=col(source(s), include(0)) interval(position(Midskilled* HIGH="MEAN_FAC1_21_HIGH name("MEAN_FAC1_21_HIGH ELEMENT: MEAN_FAC1_20), "] MISSING=LISTWISE ")) interval(position(Upskilled*M shape.interior(shape.square)) REPORTMISSING=NO GUIDE: axis(dim(1), EAN_FAC1_21), ELEMENT: /GRAPHSPEC label("[ANALYSE] [theme: shape.interior(shape.square)) interval(position(region.sprea SOURCE=INLINE. demographics] [work] Work ELEMENT: d.range(Midskilled*(LOW+HIG BEGIN GPL type middling skilled ", interval(position(region.sprea H))), SOURCE: "as opposed to never d.range(Upskilled*(LOW+HIG s=userSource(id("graphdata worked")) H))), shape.interior(shape.ibeam)) set")) GUIDE: axis(dim(2), END GPL. DATA: label("Mean [ANALYSE] shape.interior(shape.ibeam)) Unskilled=col(source(s), [theme: psychology] [risk] 1 END GPL. * Chart Builder. name("Unskilled"), Tolerance of ", GGRAPH unit.category()) "uncertainty regression * Chart Builder. /GRAPHDATA SET DATA: factor")) GGRAPH NAME="graphdata set" MEAN_FAC1_21=col(source(s) GUIDE: /GRAPHDATA SET VARIABLES=Unskilled , name("MEAN_FAC1_21")) text.footnote(label("Error NAME="graphdata set" MEANCI(FAC1_20, DATA: LOW=col(source(s), Bars: 95% CI")) VARIABLES=Upskilled 95)[name="MEAN_FAC1_20" name("MEAN_FAC1_21_LOW SCALE: linear(dim(2), MEANCI(FAC1_24, ")) include(0)) 95)[name="MEAN_FAC1_24" LOW="MEAN_FAC1_20_LOW DATA: HIGH=col(source(s), ELEMENT: " name("MEAN_FAC1_21_HIGH interval(position(Midskilled* LOW="MEAN_FAC1_24_LOW HIGH="MEAN_FAC1_20_HIGH ")) MEAN_FAC1_21), " "] MISSING=LISTWISE GUIDE: axis(dim(1), shape.interior(shape.square)) HIGH="MEAN_FAC1_24_HIGH REPORTMISSING=NO label("[ANALYSE] [theme: ELEMENT: "] MISSING=LISTWISE /GRAPHSPEC demographics] [work] Work interval(position(region.sprea REPORTMISSING=NO SOURCE=INLINE. type low skilled or ",

/GRAPHSPEC demographics] [work] Work d.range(Unskilled*(LOW+HIG BEGIN GPL SOURCE=INLINE. type middling skilled ", H))), SOURCE: BEGIN GPL "as opposed to never s=userSource(id("graphdata SOURCE: worked")) shape.interior(shape.ibeam)) set")) s=userSource(id("graphdata GUIDE: axis(dim(2), END GPL. DATA: set")) label("Mean [ANALYSE] anyUniW7=col(source(s), DATA: [theme: psychology] Response * Chart Builder. name("anyUniW7"), Upskilled=col(source(s), polarisation chief ", GGRAPH unit.category()) name("Upskilled"), "regression factor")) /GRAPHDATA SET DATA: unit.category()) GUIDE: NAME="graphdata set" MEAN_FAC1_21=col(source(s) DATA: text.footnote(label("Error VARIABLES=anyUniW7 , name("MEAN_FAC1_21")) MEAN_FAC1_24=col(source(s) Bars: 95% CI")) MEANCI(FAC1_24, DATA: LOW=col(source(s), , name("MEAN_FAC1_24")) SCALE: linear(dim(2), 95)[name="MEAN_FAC1_24" name("MEAN_FAC1_21_LOW DATA: LOW=col(source(s), include(0)) ")) name("MEAN_FAC1_24_LOW ELEMENT: LOW="MEAN_FAC1_24_LOW DATA: HIGH=col(source(s), ")) interval(position(Midskilled* " name("MEAN_FAC1_21_HIGH DATA: HIGH=col(source(s), MEAN_FAC1_24), HIGH="MEAN_FAC1_24_HIGH ")) name("MEAN_FAC1_24_HIGH shape.interior(shape.square)) "] MISSING=LISTWISE GUIDE: axis(dim(1), ")) ELEMENT: REPORTMISSING=NO label("[ANALYSE] [theme: GUIDE: axis(dim(1), interval(position(region.sprea /GRAPHSPEC demographics] Attended label("[ANALYSE] [theme: d.range(Midskilled*(LOW+HIG SOURCE=INLINE. university as opposed to ", demographics] [work] Work H))), BEGIN GPL "not attended university")) type highly skilled as ", SOURCE: GUIDE: axis(dim(2), "opposed to never shape.interior(shape.ibeam)) s=userSource(id("graphdata label("Mean [ANALYSE] worked")) END GPL. set")) [theme: psychology] [risk] 1 GUIDE: axis(dim(2), DATA: Tolerance of ", label("Mean [ANALYSE] * Chart Builder. anyUniW7=col(source(s), "uncertainty regression [theme: psychology] Response GGRAPH name("anyUniW7"), factor")) polarisation chief ", /GRAPHDATA SET unit.category()) GUIDE: "regression factor")) NAME="graphdata set" DATA: text.footnote(label("Error GUIDE: VARIABLES=Unskilled MEAN_FAC1_24=col(source(s) Bars: 95% CI")) text.footnote(label("Error MEANCI(FAC1_24, , name("MEAN_FAC1_24")) SCALE: cat(dim(1), Bars: 95% CI")) 95)[name="MEAN_FAC1_24" DATA: LOW=col(source(s), include("0", "1")) SCALE: linear(dim(2), name("MEAN_FAC1_24_LOW SCALE: linear(dim(2), include(0)) LOW="MEAN_FAC1_24_LOW ")) include(0)) ELEMENT: " DATA: HIGH=col(source(s), ELEMENT: interval(position(Upskilled*M HIGH="MEAN_FAC1_24_HIGH name("MEAN_FAC1_24_HIGH interval(position(anyUniW7* EAN_FAC1_24), "] MISSING=LISTWISE ")) MEAN_FAC1_21), shape.interior(shape.square)) REPORTMISSING=NO GUIDE: axis(dim(1), shape.interior(shape.square)) ELEMENT: /GRAPHSPEC label("[ANALYSE] [theme: ELEMENT: interval(position(region.sprea SOURCE=INLINE. demographics] Attended interval(position(region.sprea d.range(Upskilled*(LOW+HIG BEGIN GPL university as opposed to ", d.range(anyUniW7*(LOW+HI H))), SOURCE: "not attended university")) GH))), s=userSource(id("graphdata GUIDE: axis(dim(2), shape.interior(shape.ibeam)) shape.interior(shape.ibeam)) set")) label("Mean [ANALYSE] END GPL. END GPL. DATA: [theme: psychology] Response Unskilled=col(source(s), polarisation chief ", * Chart Builder. * Chart Builder. name("Unskilled"), "regression factor")) GGRAPH GGRAPH unit.category()) GUIDE: /GRAPHDATA SET /GRAPHDATA SET DATA: text.footnote(label("Error NAME="graphdata set" NAME="graphdata set" MEAN_FAC1_24=col(source(s) Bars: 95% CI")) VARIABLES=anyUniW7 VARIABLES=Midskilled , name("MEAN_FAC1_24")) SCALE: cat(dim(1), MEANCI(FAC1_20, MEANCI(FAC1_24, DATA: LOW=col(source(s), include("0", "1")) 95)[name="MEAN_FAC1_20" 95)[name="MEAN_FAC1_24" name("MEAN_FAC1_24_LOW SCALE: linear(dim(2), ")) include(0)) LOW="MEAN_FAC1_20_LOW LOW="MEAN_FAC1_24_LOW DATA: HIGH=col(source(s), ELEMENT: " " name("MEAN_FAC1_24_HIGH interval(position(anyUniW7* HIGH="MEAN_FAC1_20_HIGH HIGH="MEAN_FAC1_24_HIGH ")) MEAN_FAC1_24), "] MISSING=LISTWISE "] MISSING=LISTWISE GUIDE: axis(dim(1), shape.interior(shape.square)) REPORTMISSING=NO REPORTMISSING=NO label("[ANALYSE] [theme: ELEMENT: /GRAPHSPEC /GRAPHSPEC demographics] [work] Work interval(position(region.sprea SOURCE=INLINE. SOURCE=INLINE. type low skilled or ", d.range(anyUniW7*(LOW+HI BEGIN GPL BEGIN GPL "unskilled as opposed to GH))), SOURCE: SOURCE: never worked")) shape.interior(shape.ibeam)) s=userSource(id("graphdata s=userSource(id("graphdata GUIDE: axis(dim(2), END GPL. set")) set")) label("Mean [ANALYSE] DATA: DATA: [theme: psychology] Response * Chart Builder. anyUniW7=col(source(s), Midskilled=col(source(s), polarisation chief ", GGRAPH name("anyUniW7"), name("Midskilled"), "regression factor")) /GRAPHDATA SET unit.category()) unit.category()) GUIDE: NAME="graphdata set" DATA: DATA: text.footnote(label("Error VARIABLES=anyUniW7 MEAN_FAC1_20=col(source(s) MEAN_FAC1_24=col(source(s) Bars: 95% CI")) MEANCI(FAC1_21, , name("MEAN_FAC1_20")) , name("MEAN_FAC1_24")) SCALE: linear(dim(2), 95)[name="MEAN_FAC1_21" DATA: LOW=col(source(s), DATA: LOW=col(source(s), include(0)) name("MEAN_FAC1_20_LOW name("MEAN_FAC1_24_LOW ELEMENT: LOW="MEAN_FAC1_21_LOW ")) ")) interval(position(Unskilled*M " DATA: HIGH=col(source(s), DATA: HIGH=col(source(s), EAN_FAC1_24), HIGH="MEAN_FAC1_21_HIGH name("MEAN_FAC1_20_HIGH name("MEAN_FAC1_24_HIGH shape.interior(shape.square)) "] MISSING=LISTWISE ")) ")) ELEMENT: REPORTMISSING=NO GUIDE: axis(dim(1), GUIDE: axis(dim(1), interval(position(region.sprea /GRAPHSPEC label("[ANALYSE] [theme: label("[ANALYSE] [theme: SOURCE=INLINE.

95 demographics] Attended d.range(disabilityW6*(LOW+H /GRAPHSPEC demographics] Gender female university as opposed to ", IGH))), SOURCE=INLINE. as opposed to male")) "not attended university")) BEGIN GPL GUIDE: axis(dim(2), GUIDE: axis(dim(2), shape.interior(shape.ibeam)) SOURCE: label("Mean [ANALYSE] label("Mean [ANALYSE] END GPL. s=userSource(id("graphdata [theme: psychology] Response [theme: psychology] [aom] set")) polarisation chief ", Active open-mindedness ", * Chart Builder. DATA: "regression factor")) "regression factor")) GGRAPH disabilityW6=col(source(s), GUIDE: GUIDE: /GRAPHDATA SET name("disabilityW6"), text.footnote(label("Error text.footnote(label("Error NAME="graphdata set" unit.category()) Bars: 95% CI")) Bars: 95% CI")) VARIABLES=disabilityW6 DATA: SCALE: cat(dim(1), SCALE: cat(dim(1), MEANCI(FAC1_21, MEAN_FAC1_24=col(source(s) include("0", "1")) include("0", "1")) 95)[name="MEAN_FAC1_21" , name("MEAN_FAC1_24")) SCALE: linear(dim(2), SCALE: linear(dim(2), DATA: LOW=col(source(s), include(0)) include(0)) LOW="MEAN_FAC1_21_LOW name("MEAN_FAC1_24_LOW ELEMENT: ELEMENT: " ")) interval(position(gender*MEA interval(position(anyUniW7* HIGH="MEAN_FAC1_21_HIGH DATA: HIGH=col(source(s), N_FAC1_24), MEAN_FAC1_20), "] MISSING=LISTWISE name("MEAN_FAC1_24_HIGH shape.interior(shape.square)) shape.interior(shape.square)) REPORTMISSING=NO ")) ELEMENT: ELEMENT: /GRAPHSPEC GUIDE: axis(dim(1), interval(position(region.sprea interval(position(region.sprea SOURCE=INLINE. label("[ANALYSE] [theme: d.range(gender*(LOW+HIGH)) d.range(anyUniW7*(LOW+HI BEGIN GPL demographics] Disabled as ), GH))), SOURCE: opposed to not disabled")) shape.interior(shape.ibeam)) shape.interior(shape.ibeam)) s=userSource(id("graphdata GUIDE: axis(dim(2), END GPL. END GPL. set")) label("Mean [ANALYSE] DATA: [theme: psychology] Response * Chart Builder. * Chart Builder. disabilityW6=col(source(s), polarisation chief ", GGRAPH GGRAPH name("disabilityW6"), "regression factor")) /GRAPHDATA SET /GRAPHDATA SET unit.category()) GUIDE: NAME="graphdata set" NAME="graphdata set" DATA: text.footnote(label("Error VARIABLES=gender VARIABLES=disabilityW6 MEAN_FAC1_21=col(source(s) Bars: 95% CI")) MEANCI(FAC1_21, MEANCI(FAC1_20, , name("MEAN_FAC1_21")) SCALE: cat(dim(1), 95)[name="MEAN_FAC1_21" 95)[name="MEAN_FAC1_20" DATA: LOW=col(source(s), include("0", "1")) name("MEAN_FAC1_21_LOW SCALE: linear(dim(2), LOW="MEAN_FAC1_21_LOW LOW="MEAN_FAC1_20_LOW ")) include(0)) " " DATA: HIGH=col(source(s), ELEMENT: HIGH="MEAN_FAC1_21_HIGH HIGH="MEAN_FAC1_20_HIGH name("MEAN_FAC1_21_HIGH interval(position(disabilityW6 "] MISSING=LISTWISE "] MISSING=LISTWISE ")) *MEAN_FAC1_24), REPORTMISSING=NO REPORTMISSING=NO GUIDE: axis(dim(1), shape.interior(shape.square)) /GRAPHSPEC /GRAPHSPEC label("[ANALYSE] [theme: ELEMENT: SOURCE=INLINE. SOURCE=INLINE. demographics] Disabled as interval(position(region.sprea BEGIN GPL BEGIN GPL opposed to not disabled")) d.range(disabilityW6*(LOW+H SOURCE: SOURCE: GUIDE: axis(dim(2), IGH))), s=userSource(id("graphdata s=userSource(id("graphdata label("Mean [ANALYSE] set")) set")) [theme: psychology] [risk] 1 shape.interior(shape.ibeam)) DATA: gender=col(source(s), DATA: Tolerance of ", END GPL. name("gender"), disabilityW6=col(source(s), "uncertainty regression unit.category()) name("disabilityW6"), factor")) * Chart Builder. DATA: unit.category()) GUIDE: GGRAPH MEAN_FAC1_21=col(source(s) DATA: text.footnote(label("Error /GRAPHDATA SET , name("MEAN_FAC1_21")) MEAN_FAC1_20=col(source(s) Bars: 95% CI")) NAME="graphdata set" DATA: LOW=col(source(s), , name("MEAN_FAC1_20")) SCALE: cat(dim(1), VARIABLES=gender name("MEAN_FAC1_21_LOW DATA: LOW=col(source(s), include("0", "1")) MEANCI(FAC1_24, ")) name("MEAN_FAC1_20_LOW SCALE: linear(dim(2), 95)[name="MEAN_FAC1_24" DATA: HIGH=col(source(s), ")) include(0)) name("MEAN_FAC1_21_HIGH DATA: HIGH=col(source(s), ELEMENT: LOW="MEAN_FAC1_24_LOW ")) name("MEAN_FAC1_20_HIGH interval(position(disabilityW6 " GUIDE: axis(dim(1), ")) *MEAN_FAC1_21), HIGH="MEAN_FAC1_24_HIGH label("[ANALYSE] [theme: GUIDE: axis(dim(1), shape.interior(shape.square)) "] MISSING=LISTWISE demographics] Gender female label("[ANALYSE] [theme: ELEMENT: REPORTMISSING=NO as opposed to male")) demographics] Disabled as interval(position(region.sprea /GRAPHSPEC GUIDE: axis(dim(2), opposed to not disabled")) d.range(disabilityW6*(LOW+H SOURCE=INLINE. label("Mean [ANALYSE] GUIDE: axis(dim(2), IGH))), BEGIN GPL [theme: psychology] [risk] 1 label("Mean [ANALYSE] SOURCE: Tolerance of ", [theme: psychology] [aom] shape.interior(shape.ibeam)) s=userSource(id("graphdata "uncertainty regression Active open-mindedness ", END GPL. set")) factor")) "regression factor")) DATA: gender=col(source(s), GUIDE: GUIDE: * Chart Builder. name("gender"), text.footnote(label("Error text.footnote(label("Error GGRAPH unit.category()) Bars: 95% CI")) Bars: 95% CI")) /GRAPHDATA SET DATA: SCALE: cat(dim(1), SCALE: cat(dim(1), NAME="graphdata set" MEAN_FAC1_24=col(source(s) include("0", "1")) include("0", "1")) VARIABLES=disabilityW6 , name("MEAN_FAC1_24")) SCALE: linear(dim(2), SCALE: linear(dim(2), MEANCI(FAC1_24, DATA: LOW=col(source(s), include(0)) include(0)) 95)[name="MEAN_FAC1_24" name("MEAN_FAC1_24_LOW ELEMENT: ELEMENT: ")) interval(position(gender*MEA interval(position(disabilityW6 LOW="MEAN_FAC1_24_LOW DATA: HIGH=col(source(s), N_FAC1_21), *MEAN_FAC1_20), " name("MEAN_FAC1_24_HIGH shape.interior(shape.square)) shape.interior(shape.square)) HIGH="MEAN_FAC1_24_HIGH ")) ELEMENT: ELEMENT: "] MISSING=LISTWISE GUIDE: axis(dim(1), interval(position(region.sprea interval(position(region.sprea REPORTMISSING=NO label("[ANALYSE] [theme: d.range(gender*(LOW+HIGH))

), DATA: [theme: culture] [conformity] shape.interior(shape.ibeam)) Broadsheet=col(source(s), Generalised (British) ", shape.interior(shape.square)) END GPL. name("Broadsheet"), "Cultural Conformity ELEMENT: unit.category()) Regression Factor")) interval(position(region.sprea * Chart Builder. DATA: GUIDE: d.range(Tabloid*(LOW+HIGH) GGRAPH MEAN_FAC1_23=col(source(s) text.footnote(label("Error )), /GRAPHDATA SET , name("MEAN_FAC1_23")) Bars: 95% CI")) shape.interior(shape.ibeam)) NAME="graphdata set" DATA: LOW=col(source(s), SCALE: linear(dim(2), END GPL. VARIABLES=gender name("MEAN_FAC1_23_LOW include(0)) MEANCI(FAC1_20, ")) ELEMENT: * Chart Builder. 95)[name="MEAN_FAC1_20" DATA: HIGH=col(source(s), interval(position(Tabloid*MEA GGRAPH name("MEAN_FAC1_23_HIGH N_FAC1_23), /GRAPHDATA SET LOW="MEAN_FAC1_20_LOW ")) shape.interior(shape.square)) NAME="graphdata set" " GUIDE: axis(dim(1), ELEMENT: VARIABLES=Broadsheet HIGH="MEAN_FAC1_20_HIGH label("[ANALYSE] [theme: interval(position(region.sprea MEANCI(OverallCulturalPolitic "] MISSING=LISTWISE demographics] [readership] d.range(Tabloid*(LOW+HIGH) alConformity, REPORTMISSING=NO Broadsheet newspaper ", )), /GRAPHSPEC "readership as opposed to shape.interior(shape.ibeam)) 95)[name="MEAN_OverallCult SOURCE=INLINE. other or none")) END GPL. uralPoliticalConformity" BEGIN GPL GUIDE: axis(dim(2), SOURCE: label("Mean [ANALYSE] * Chart Builder. LOW="MEAN_OverallCultural s=userSource(id("graphdata [theme: culture] [conformity] GGRAPH PoliticalConformity_LOW" set")) Generalised (British) ", /GRAPHDATA SET DATA: gender=col(source(s), "Cultural Conformity NAME="graphdata set" HIGH="MEAN_OverallCultural name("gender"), Regression Factor")) VARIABLES=Tabloid PoliticalConformity_HIGH"] unit.category()) GUIDE: MEANCI(OverallCulturalPolitic MISSING=LISTWISE DATA: text.footnote(label("Error alConformity, REPORTMISSING=NO MEAN_FAC1_20=col(source(s) Bars: 95% CI")) /GRAPHSPEC , name("MEAN_FAC1_20")) SCALE: linear(dim(2), 95)[name="MEAN_OverallCult SOURCE=INLINE. DATA: LOW=col(source(s), include(0)) uralPoliticalConformity" BEGIN GPL name("MEAN_FAC1_20_LOW ELEMENT: SOURCE: ")) interval(position(Broadsheet* LOW="MEAN_OverallCultural s=userSource(id("graphdata DATA: HIGH=col(source(s), MEAN_FAC1_23), PoliticalConformity_LOW" set")) name("MEAN_FAC1_20_HIGH shape.interior(shape.square)) DATA: ")) ELEMENT: HIGH="MEAN_OverallCultural Broadsheet=col(source(s), GUIDE: axis(dim(1), interval(position(region.sprea PoliticalConformity_HIGH"] name("Broadsheet"), label("[ANALYSE] [theme: d.range(Broadsheet*(LOW+HI MISSING=LISTWISE unit.category()) demographics] Gender female GH))), REPORTMISSING=NO DATA: as opposed to male")) /GRAPHSPEC MEAN_OverallCulturalPolitical GUIDE: axis(dim(2), shape.interior(shape.ibeam)) SOURCE=INLINE. Conformity=col(source(s), label("Mean [ANALYSE] END GPL. BEGIN GPL [theme: psychology] [aom] SOURCE: name("MEAN_OverallCultural Active open-mindedness ", * Chart Builder. s=userSource(id("graphdata PoliticalConformity")) "regression factor")) GGRAPH set")) DATA: LOW=col(source(s), GUIDE: /GRAPHDATA SET DATA: Tabloid=col(source(s), name("MEAN_OverallCultural text.footnote(label("Error NAME="graphdata set" name("Tabloid"), PoliticalConformity_LOW")) Bars: 95% CI")) VARIABLES=Tabloid unit.category()) DATA: HIGH=col(source(s), SCALE: cat(dim(1), MEANCI(FAC1_23, DATA: name("MEAN_OverallCultural include("0", "1")) 95)[name="MEAN_FAC1_23" MEAN_OverallCulturalPolitical PoliticalConformity_HIGH")) SCALE: linear(dim(2), Conformity=col(source(s), GUIDE: axis(dim(1), include(0)) LOW="MEAN_FAC1_23_LOW label("[ANALYSE] [theme: ELEMENT: " name("MEAN_OverallCultural demographics] [readership] interval(position(gender*MEA HIGH="MEAN_FAC1_23_HIGH PoliticalConformity")) Broadsheet newspaper ", N_FAC1_20), "] MISSING=LISTWISE DATA: LOW=col(source(s), "readership as opposed to shape.interior(shape.square)) REPORTMISSING=NO name("MEAN_OverallCultural other or none")) ELEMENT: /GRAPHSPEC PoliticalConformity_LOW")) GUIDE: axis(dim(2), interval(position(region.sprea SOURCE=INLINE. DATA: HIGH=col(source(s), label("Mean [ANALYSE] d.range(gender*(LOW+HIGH)) BEGIN GPL name("MEAN_OverallCultural [theme: culture] individual ), SOURCE: PoliticalConformity_HIGH")) conformity to overall ", shape.interior(shape.ibeam)) s=userSource(id("graphdata GUIDE: axis(dim(1), "mean political END GPL. set")) label("[ANALYSE] [theme: orientation")) DATA: Tabloid=col(source(s), demographics] [readership] GUIDE: * Chart Builder. name("Tabloid"), Tabloid newspaper ", text.footnote(label("Error GGRAPH unit.category()) "readership as opposed to Bars: 95% CI")) /GRAPHDATA SET DATA: other or none")) SCALE: linear(dim(2), NAME="graphdata set" MEAN_FAC1_23=col(source(s) GUIDE: axis(dim(2), include(0)) VARIABLES=Broadsheet , name("MEAN_FAC1_23")) label("Mean [ANALYSE] ELEMENT: MEANCI(FAC1_23, DATA: LOW=col(source(s), [theme: culture] individual interval(position(Broadsheet* 95)[name="MEAN_FAC1_23" name("MEAN_FAC1_23_LOW conformity to overall ", MEAN_OverallCulturalPolitical ")) "mean political Conformity), LOW="MEAN_FAC1_23_LOW DATA: HIGH=col(source(s), orientation")) " name("MEAN_FAC1_23_HIGH GUIDE: shape.interior(shape.square)) HIGH="MEAN_FAC1_23_HIGH ")) text.footnote(label("Error ELEMENT: "] MISSING=LISTWISE GUIDE: axis(dim(1), Bars: 95% CI")) interval(position(region.sprea REPORTMISSING=NO label("[ANALYSE] [theme: SCALE: linear(dim(2), d.range(Broadsheet*(LOW+HI /GRAPHSPEC demographics] [readership] include(0)) GH))), SOURCE=INLINE. Tabloid newspaper ", ELEMENT: BEGIN GPL "readership as opposed to interval(position(Tabloid*MEA shape.interior(shape.ibeam)) SOURCE: other or none")) N_OverallCulturalPoliticalConf END GPL. s=userSource(id("graphdata GUIDE: axis(dim(2), ormity), set")) label("Mean [ANALYSE] * Chart Builder.

GGRAPH DATA: demographics] [work] Work /GRAPHDATA SET HIGH="MEAN_BeliefSystemCo MEAN_BeliefSystemConstrain type middling skilled ", NAME="graphdata set" nstraint_HIGH"] t=col(source(s), "as opposed to never VARIABLES=Broadsheet MISSING=LISTWISE name("MEAN_BeliefSystemCo worked")) MEANCI(BeliefSystemConstrai REPORTMISSING=NO nstraint")) GUIDE: axis(dim(2), nt, /GRAPHSPEC DATA: LOW=col(source(s), label("Mean [ANALYSE] SOURCE=INLINE. name("MEAN_BeliefSystemCo [theme: culture] Reflected 95)[name="MEAN_BeliefSyste BEGIN GPL nstraint_LOW")) standard deviation of ", mConstraint" SOURCE: DATA: HIGH=col(source(s), "policy preferences, as a LOW="MEAN_BeliefSystemCo s=userSource(id("graphdata name("MEAN_BeliefSystemCo measure of belief system nstraint_LOW" set")) nstraint_HIGH")) constraint")) DATA: Tabloid=col(source(s), GUIDE: axis(dim(1), GUIDE: HIGH="MEAN_BeliefSystemCo name("Tabloid"), label("[ANALYSE] [theme: text.footnote(label("Error nstraint_HIGH"] unit.category()) demographics] [work] Work Bars: 95% CI")) MISSING=LISTWISE DATA: type highly skilled as ", SCALE: linear(dim(2), REPORTMISSING=NO MEAN_BeliefSystemConstrain "opposed to never include(0)) /GRAPHSPEC t=col(source(s), worked")) ELEMENT: SOURCE=INLINE. name("MEAN_BeliefSystemCo GUIDE: axis(dim(2), interval(position(Midskilled* BEGIN GPL nstraint")) label("Mean [ANALYSE] MEAN_BeliefSystemConstrain SOURCE: DATA: LOW=col(source(s), [theme: culture] Reflected t), s=userSource(id("graphdata name("MEAN_BeliefSystemCo standard deviation of ", shape.interior(shape.square)) set")) nstraint_LOW")) "policy preferences, as a ELEMENT: DATA: DATA: HIGH=col(source(s), measure of belief system interval(position(region.sprea Broadsheet=col(source(s), name("MEAN_BeliefSystemCo constraint")) d.range(Midskilled*(LOW+HIG name("Broadsheet"), nstraint_HIGH")) GUIDE: H))), unit.category()) GUIDE: axis(dim(1), text.footnote(label("Error DATA: label("[ANALYSE] [theme: Bars: 95% CI")) shape.interior(shape.ibeam)) MEAN_BeliefSystemConstrain demographics] [readership] SCALE: linear(dim(2), END GPL. t=col(source(s), Tabloid newspaper ", include(0)) name("MEAN_BeliefSystemCo "readership as opposed to ELEMENT: * Chart Builder. nstraint")) other or none")) interval(position(Upskilled*M GGRAPH DATA: LOW=col(source(s), GUIDE: axis(dim(2), EAN_BeliefSystemConstraint), /GRAPHDATA SET name("MEAN_BeliefSystemCo label("Mean [ANALYSE] shape.interior(shape.square)) NAME="graphdata set" nstraint_LOW")) [theme: culture] Reflected ELEMENT: VARIABLES=Unskilled DATA: HIGH=col(source(s), standard deviation of ", interval(position(region.sprea MEANCI(BeliefSystemConstrai name("MEAN_BeliefSystemCo "policy preferences, as a d.range(Upskilled*(LOW+HIG nt, nstraint_HIGH")) measure of belief system H))), GUIDE: axis(dim(1), constraint")) 95)[name="MEAN_BeliefSyste label("[ANALYSE] [theme: GUIDE: shape.interior(shape.ibeam)) mConstraint" demographics] [readership] text.footnote(label("Error END GPL. LOW="MEAN_BeliefSystemCo Broadsheet newspaper ", Bars: 95% CI")) nstraint_LOW" "readership as opposed to SCALE: linear(dim(2), * Chart Builder. other or none")) include(0)) GGRAPH HIGH="MEAN_BeliefSystemCo GUIDE: axis(dim(2), ELEMENT: /GRAPHDATA SET nstraint_HIGH"] label("Mean [ANALYSE] interval(position(Tabloid*MEA NAME="graphdata set" MISSING=LISTWISE [theme: culture] Reflected N_BeliefSystemConstraint), VARIABLES=Midskilled REPORTMISSING=NO standard deviation of ", shape.interior(shape.square)) MEANCI(BeliefSystemConstrai /GRAPHSPEC "policy preferences, as a ELEMENT: nt, SOURCE=INLINE. measure of belief system interval(position(region.sprea BEGIN GPL constraint")) d.range(Tabloid*(LOW+HIGH) 95)[name="MEAN_BeliefSyste SOURCE: GUIDE: )), mConstraint" s=userSource(id("graphdata text.footnote(label("Error shape.interior(shape.ibeam)) LOW="MEAN_BeliefSystemCo set")) Bars: 95% CI")) END GPL. nstraint_LOW" DATA: SCALE: linear(dim(2), Unskilled=col(source(s), include(0)) * Chart Builder. HIGH="MEAN_BeliefSystemCo name("Unskilled"), ELEMENT: GGRAPH nstraint_HIGH"] unit.category()) interval(position(Broadsheet* /GRAPHDATA SET MISSING=LISTWISE DATA: MEAN_BeliefSystemConstrain NAME="graphdata set" REPORTMISSING=NO MEAN_BeliefSystemConstrain t), VARIABLES=Upskilled /GRAPHSPEC t=col(source(s), shape.interior(shape.square)) MEANCI(BeliefSystemConstrai SOURCE=INLINE. name("MEAN_BeliefSystemCo ELEMENT: nt, BEGIN GPL nstraint")) interval(position(region.sprea SOURCE: DATA: LOW=col(source(s), d.range(Broadsheet*(LOW+HI 95)[name="MEAN_BeliefSyste s=userSource(id("graphdata name("MEAN_BeliefSystemCo GH))), mConstraint" set")) nstraint_LOW")) LOW="MEAN_BeliefSystemCo DATA: DATA: HIGH=col(source(s), shape.interior(shape.ibeam)) nstraint_LOW" Midskilled=col(source(s), name("MEAN_BeliefSystemCo END GPL. name("Midskilled"), nstraint_HIGH")) HIGH="MEAN_BeliefSystemCo unit.category()) GUIDE: axis(dim(1), * Chart Builder. nstraint_HIGH"] DATA: label("[ANALYSE] [theme: GGRAPH MISSING=LISTWISE MEAN_BeliefSystemConstrain demographics] [work] Work /GRAPHDATA SET REPORTMISSING=NO t=col(source(s), type low skilled or ", NAME="graphdata set" /GRAPHSPEC name("MEAN_BeliefSystemCo "unskilled as opposed to VARIABLES=Tabloid SOURCE=INLINE. nstraint")) never worked")) MEANCI(BeliefSystemConstrai BEGIN GPL DATA: LOW=col(source(s), GUIDE: axis(dim(2), nt, SOURCE: name("MEAN_BeliefSystemCo label("Mean [ANALYSE] s=userSource(id("graphdata nstraint_LOW")) [theme: culture] Reflected 95)[name="MEAN_BeliefSyste set")) DATA: HIGH=col(source(s), standard deviation of ", mConstraint" DATA: name("MEAN_BeliefSystemCo "policy preferences, as a LOW="MEAN_BeliefSystemCo Upskilled=col(source(s), nstraint_HIGH")) measure of belief system nstraint_LOW" name("Upskilled"), GUIDE: axis(dim(1), constraint")) unit.category()) label("[ANALYSE] [theme:

GUIDE: ELEMENT: /GRAPHDATA SET "] MISSING=LISTWISE text.footnote(label("Error interval(position(region.sprea NAME="graphdata set" REPORTMISSING=NO Bars: 95% CI")) d.range(Unskilled*(LOW+HIG VARIABLES=Upskilled /GRAPHSPEC SCALE: linear(dim(2), H))), MEANCI(OverallCulturalPolitic SOURCE=INLINE. include(0)) alConformity, BEGIN GPL ELEMENT: shape.interior(shape.ibeam)) SOURCE: interval(position(Unskilled*M END GPL. 95)[name="MEAN_OverallCult s=userSource(id("graphdata EAN_BeliefSystemConstraint), uralPoliticalConformity" set")) shape.interior(shape.square)) * Chart Builder. DATA: ELEMENT: GGRAPH LOW="MEAN_OverallCultural Upskilled=col(source(s), interval(position(region.sprea /GRAPHDATA SET PoliticalConformity_LOW" name("Upskilled"), d.range(Unskilled*(LOW+HIG NAME="graphdata set" unit.category()) H))), VARIABLES=Midskilled HIGH="MEAN_OverallCultural DATA: MEANCI(OverallCulturalPolitic PoliticalConformity_HIGH"] MEAN_FAC1_23=col(source(s) shape.interior(shape.ibeam)) alConformity, MISSING=LISTWISE , name("MEAN_FAC1_23")) END GPL. REPORTMISSING=NO DATA: LOW=col(source(s), 95)[name="MEAN_OverallCult /GRAPHSPEC name("MEAN_FAC1_23_LOW * Chart Builder. uralPoliticalConformity" SOURCE=INLINE. ")) GGRAPH BEGIN GPL DATA: HIGH=col(source(s), /GRAPHDATA SET LOW="MEAN_OverallCultural SOURCE: name("MEAN_FAC1_23_HIGH NAME="graphdata set" PoliticalConformity_LOW" s=userSource(id("graphdata ")) VARIABLES=Unskilled set")) GUIDE: axis(dim(1), MEANCI(OverallCulturalPolitic HIGH="MEAN_OverallCultural DATA: label("[ANALYSE] [theme: alConformity, PoliticalConformity_HIGH"] Upskilled=col(source(s), demographics] [work] Work MISSING=LISTWISE name("Upskilled"), type highly skilled as ", 95)[name="MEAN_OverallCult REPORTMISSING=NO unit.category()) "opposed to never uralPoliticalConformity" /GRAPHSPEC DATA: worked")) SOURCE=INLINE. MEAN_OverallCulturalPolitical GUIDE: axis(dim(2), LOW="MEAN_OverallCultural BEGIN GPL Conformity=col(source(s), label("Mean [ANALYSE] PoliticalConformity_LOW" SOURCE: [theme: culture] [conformity] s=userSource(id("graphdata name("MEAN_OverallCultural Generalised (British) ", HIGH="MEAN_OverallCultural set")) PoliticalConformity")) "Cultural Conformity PoliticalConformity_HIGH"] DATA: DATA: LOW=col(source(s), Regression Factor")) MISSING=LISTWISE Midskilled=col(source(s), name("MEAN_OverallCultural GUIDE: REPORTMISSING=NO name("Midskilled"), PoliticalConformity_LOW")) text.footnote(label("Error /GRAPHSPEC unit.category()) DATA: HIGH=col(source(s), Bars: 95% CI")) SOURCE=INLINE. DATA: name("MEAN_OverallCultural SCALE: linear(dim(2), BEGIN GPL MEAN_OverallCulturalPolitical PoliticalConformity_HIGH")) include(0)) SOURCE: Conformity=col(source(s), GUIDE: axis(dim(1), ELEMENT: s=userSource(id("graphdata label("[ANALYSE] [theme: interval(position(Upskilled*M set")) name("MEAN_OverallCultural demographics] [work] Work EAN_FAC1_23), DATA: PoliticalConformity")) type highly skilled as ", shape.interior(shape.square)) Unskilled=col(source(s), DATA: LOW=col(source(s), "opposed to never ELEMENT: name("Unskilled"), name("MEAN_OverallCultural worked")) interval(position(region.sprea unit.category()) PoliticalConformity_LOW")) GUIDE: axis(dim(2), d.range(Upskilled*(LOW+HIG DATA: DATA: HIGH=col(source(s), label("Mean [ANALYSE] H))), MEAN_OverallCulturalPolitical name("MEAN_OverallCultural [theme: culture] individual Conformity=col(source(s), PoliticalConformity_HIGH")) conformity to overall ", shape.interior(shape.ibeam)) GUIDE: axis(dim(1), "mean political END GPL. name("MEAN_OverallCultural label("[ANALYSE] [theme: orientation")) PoliticalConformity")) demographics] [work] Work GUIDE: * Chart Builder. DATA: LOW=col(source(s), type middling skilled ", text.footnote(label("Error GGRAPH name("MEAN_OverallCultural "as opposed to never Bars: 95% CI")) /GRAPHDATA SET PoliticalConformity_LOW")) worked")) SCALE: linear(dim(2), NAME="graphdata set" DATA: HIGH=col(source(s), GUIDE: axis(dim(2), include(0)) VARIABLES=Midskilled name("MEAN_OverallCultural label("Mean [ANALYSE] ELEMENT: MEANCI(FAC1_23, PoliticalConformity_HIGH")) [theme: culture] individual interval(position(Upskilled*M 95)[name="MEAN_FAC1_23" GUIDE: axis(dim(1), conformity to overall ", EAN_OverallCulturalPoliticalC label("[ANALYSE] [theme: "mean political onformity), LOW="MEAN_FAC1_23_LOW demographics] [work] Work orientation")) " type low skilled or ", GUIDE: shape.interior(shape.square)) HIGH="MEAN_FAC1_23_HIGH "unskilled as opposed to text.footnote(label("Error ELEMENT: "] MISSING=LISTWISE never worked")) Bars: 95% CI")) interval(position(region.sprea REPORTMISSING=NO GUIDE: axis(dim(2), SCALE: linear(dim(2), d.range(Upskilled*(LOW+HIG /GRAPHSPEC label("Mean [ANALYSE] include(0)) H))), SOURCE=INLINE. [theme: culture] individual ELEMENT: BEGIN GPL conformity to overall ", interval(position(Midskilled* shape.interior(shape.ibeam)) SOURCE: "mean political MEAN_OverallCulturalPolitical END GPL. s=userSource(id("graphdata orientation")) Conformity), set")) GUIDE: * Chart Builder. DATA: text.footnote(label("Error shape.interior(shape.square)) GGRAPH Midskilled=col(source(s), Bars: 95% CI")) ELEMENT: /GRAPHDATA SET name("Midskilled"), SCALE: linear(dim(2), interval(position(region.sprea NAME="graphdata set" unit.category()) include(0)) d.range(Midskilled*(LOW+HIG VARIABLES=Upskilled DATA: ELEMENT: H))), MEANCI(FAC1_23, MEAN_FAC1_23=col(source(s) interval(position(Unskilled*M 95)[name="MEAN_FAC1_23" , name("MEAN_FAC1_23")) EAN_OverallCulturalPoliticalC shape.interior(shape.ibeam)) DATA: LOW=col(source(s), onformity), END GPL. LOW="MEAN_FAC1_23_LOW name("MEAN_FAC1_23_LOW " ")) shape.interior(shape.square)) * Chart Builder. HIGH="MEAN_FAC1_23_HIGH GGRAPH

DATA: HIGH=col(source(s), ELEMENT: /GRAPHSPEC name("MEAN_FAC1_23_HIGH interval(position(Unskilled*M 95)[name="MEAN_OverallCult SOURCE=INLINE. ")) EAN_FAC1_23), uralPoliticalConformity" BEGIN GPL GUIDE: axis(dim(1), shape.interior(shape.square)) SOURCE: label("[ANALYSE] [theme: ELEMENT: LOW="MEAN_OverallCultural s=userSource(id("graphdata demographics] [work] Work interval(position(region.sprea PoliticalConformity_LOW" set")) type middling skilled ", d.range(Unskilled*(LOW+HIG DATA: "as opposed to never H))), HIGH="MEAN_OverallCultural anyUniW7=col(source(s), worked")) PoliticalConformity_HIGH"] name("anyUniW7"), GUIDE: axis(dim(2), shape.interior(shape.ibeam)) MISSING=LISTWISE unit.category()) label("Mean [ANALYSE] END GPL. REPORTMISSING=NO DATA: [theme: culture] [conformity] /GRAPHSPEC MEAN_BeliefSystemConstrain Generalised (British) ", * Chart Builder. SOURCE=INLINE. t=col(source(s), "Cultural Conformity GGRAPH BEGIN GPL name("MEAN_BeliefSystemCo Regression Factor")) /GRAPHDATA SET SOURCE: nstraint")) GUIDE: NAME="graphdata set" s=userSource(id("graphdata DATA: LOW=col(source(s), text.footnote(label("Error VARIABLES=anyUniW7 set")) name("MEAN_BeliefSystemCo Bars: 95% CI")) MEANCI(FAC1_23, DATA: nstraint_LOW")) SCALE: linear(dim(2), 95)[name="MEAN_FAC1_23" anyUniW7=col(source(s), DATA: HIGH=col(source(s), include(0)) name("anyUniW7"), name("MEAN_BeliefSystemCo ELEMENT: LOW="MEAN_FAC1_23_LOW unit.category()) nstraint_HIGH")) interval(position(Midskilled* " DATA: GUIDE: axis(dim(1), MEAN_FAC1_23), HIGH="MEAN_FAC1_23_HIGH MEAN_OverallCulturalPolitical label("[ANALYSE] [theme: shape.interior(shape.square)) "] MISSING=LISTWISE Conformity=col(source(s), demographics] Attended ELEMENT: REPORTMISSING=NO university as opposed to ", interval(position(region.sprea /GRAPHSPEC name("MEAN_OverallCultural "not attended university")) d.range(Midskilled*(LOW+HIG SOURCE=INLINE. PoliticalConformity")) GUIDE: axis(dim(2), H))), BEGIN GPL DATA: LOW=col(source(s), label("Mean [ANALYSE] SOURCE: name("MEAN_OverallCultural [theme: culture] Reflected shape.interior(shape.ibeam)) s=userSource(id("graphdata PoliticalConformity_LOW")) standard deviation of ", END GPL. set")) DATA: HIGH=col(source(s), "policy preferences, as a DATA: name("MEAN_OverallCultural measure of belief system * Chart Builder. anyUniW7=col(source(s), PoliticalConformity_HIGH")) constraint")) GGRAPH name("anyUniW7"), GUIDE: axis(dim(1), GUIDE: /GRAPHDATA SET unit.category()) label("[ANALYSE] [theme: text.footnote(label("Error NAME="graphdata set" DATA: demographics] Attended Bars: 95% CI")) VARIABLES=Unskilled MEAN_FAC1_23=col(source(s) university as opposed to ", SCALE: cat(dim(1), MEANCI(FAC1_23, , name("MEAN_FAC1_23")) "not attended university")) include("0", "1")) 95)[name="MEAN_FAC1_23" DATA: LOW=col(source(s), GUIDE: axis(dim(2), SCALE: linear(dim(2), name("MEAN_FAC1_23_LOW label("Mean [ANALYSE] include(0)) LOW="MEAN_FAC1_23_LOW ")) [theme: culture] individual ELEMENT: " DATA: HIGH=col(source(s), conformity to overall ", interval(position(anyUniW7* HIGH="MEAN_FAC1_23_HIGH name("MEAN_FAC1_23_HIGH "mean political MEAN_BeliefSystemConstrain "] MISSING=LISTWISE ")) orientation")) t), REPORTMISSING=NO GUIDE: axis(dim(1), GUIDE: shape.interior(shape.square)) /GRAPHSPEC label("[ANALYSE] [theme: text.footnote(label("Error ELEMENT: SOURCE=INLINE. demographics] Attended Bars: 95% CI")) interval(position(region.sprea BEGIN GPL university as opposed to ", SCALE: cat(dim(1), d.range(anyUniW7*(LOW+HI SOURCE: "not attended university")) include("0", "1")) GH))), s=userSource(id("graphdata GUIDE: axis(dim(2), SCALE: linear(dim(2), shape.interior(shape.ibeam)) set")) label("Mean [ANALYSE] include(0)) END GPL. DATA: [theme: culture] [conformity] ELEMENT: Unskilled=col(source(s), Generalised (British) ", interval(position(anyUniW7* * Chart Builder. name("Unskilled"), "Cultural Conformity MEAN_OverallCulturalPolitical GGRAPH unit.category()) Regression Factor")) Conformity), /GRAPHDATA SET DATA: GUIDE: NAME="graphdata set" MEAN_FAC1_23=col(source(s) text.footnote(label("Error shape.interior(shape.square)) VARIABLES=disabilityW6 , name("MEAN_FAC1_23")) Bars: 95% CI")) ELEMENT: MEANCI(BeliefSystemConstrai DATA: LOW=col(source(s), SCALE: cat(dim(1), interval(position(region.sprea nt, name("MEAN_FAC1_23_LOW include("0", "1")) d.range(anyUniW7*(LOW+HI ")) SCALE: linear(dim(2), GH))), 95)[name="MEAN_BeliefSyste DATA: HIGH=col(source(s), include(0)) shape.interior(shape.ibeam)) mConstraint" name("MEAN_FAC1_23_HIGH ELEMENT: END GPL. LOW="MEAN_BeliefSystemCo ")) interval(position(anyUniW7* nstraint_LOW" GUIDE: axis(dim(1), MEAN_FAC1_23), * Chart Builder. label("[ANALYSE] [theme: shape.interior(shape.square)) GGRAPH HIGH="MEAN_BeliefSystemCo demographics] [work] Work ELEMENT: /GRAPHDATA SET nstraint_HIGH"] type low skilled or ", interval(position(region.sprea NAME="graphdata set" MISSING=LISTWISE "unskilled as opposed to d.range(anyUniW7*(LOW+HI VARIABLES=anyUniW7 REPORTMISSING=NO never worked")) GH))), MEANCI(BeliefSystemConstrai /GRAPHSPEC GUIDE: axis(dim(2), shape.interior(shape.ibeam)) nt, SOURCE=INLINE. label("Mean [ANALYSE] END GPL. BEGIN GPL [theme: culture] [conformity] 95)[name="MEAN_BeliefSyste SOURCE: Generalised (British) ", * Chart Builder. mConstraint" s=userSource(id("graphdata "Cultural Conformity GGRAPH LOW="MEAN_BeliefSystemCo set")) Regression Factor")) /GRAPHDATA SET nstraint_LOW" DATA: GUIDE: NAME="graphdata set" disabilityW6=col(source(s), text.footnote(label("Error VARIABLES=anyUniW7 HIGH="MEAN_BeliefSystemCo name("disabilityW6"), Bars: 95% CI")) MEANCI(OverallCulturalPolitic nstraint_HIGH"] unit.category()) SCALE: linear(dim(2), alConformity, MISSING=LISTWISE DATA: include(0)) REPORTMISSING=NO MEAN_BeliefSystemConstrain

100 t=col(source(s), GUIDE: axis(dim(1), ELEMENT: name("MEAN_BeliefSystemCo label("[ANALYSE] [theme: interval(position(disabilityW6 LOW="MEAN_OverallCultural nstraint")) demographics] Disabled as *MEAN_FAC1_23), PoliticalConformity_LOW" DATA: LOW=col(source(s), opposed to not disabled")) shape.interior(shape.square)) name("MEAN_BeliefSystemCo GUIDE: axis(dim(2), ELEMENT: HIGH="MEAN_OverallCultural nstraint_LOW")) label("Mean [ANALYSE] interval(position(region.sprea PoliticalConformity_HIGH"] DATA: HIGH=col(source(s), [theme: culture] individual d.range(disabilityW6*(LOW+H MISSING=LISTWISE name("MEAN_BeliefSystemCo conformity to overall ", IGH))), REPORTMISSING=NO nstraint_HIGH")) "mean political /GRAPHSPEC GUIDE: axis(dim(1), orientation")) shape.interior(shape.ibeam)) SOURCE=INLINE. label("[ANALYSE] [theme: GUIDE: END GPL. BEGIN GPL demographics] Disabled as text.footnote(label("Error SOURCE: opposed to not disabled")) Bars: 95% CI")) * Chart Builder. s=userSource(id("graphdata GUIDE: axis(dim(2), SCALE: cat(dim(1), GGRAPH set")) label("Mean [ANALYSE] include("0", "1")) /GRAPHDATA SET DATA: gender=col(source(s), [theme: culture] Reflected SCALE: linear(dim(2), NAME="graphdata set" name("gender"), standard deviation of ", include(0)) VARIABLES=gender unit.category()) "policy preferences, as a ELEMENT: MEANCI(FAC1_23, DATA: measure of belief system interval(position(disabilityW6 95)[name="MEAN_FAC1_23" MEAN_OverallCulturalPolitical constraint")) *MEAN_OverallCulturalPolitic Conformity=col(source(s), GUIDE: alConformity), LOW="MEAN_FAC1_23_LOW text.footnote(label("Error " name("MEAN_OverallCultural Bars: 95% CI")) shape.interior(shape.square)) HIGH="MEAN_FAC1_23_HIGH PoliticalConformity")) SCALE: cat(dim(1), ELEMENT: "] MISSING=LISTWISE DATA: LOW=col(source(s), include("0", "1")) interval(position(region.sprea REPORTMISSING=NO name("MEAN_OverallCultural SCALE: linear(dim(2), d.range(disabilityW6*(LOW+H /GRAPHSPEC PoliticalConformity_LOW")) include(0)) IGH))), SOURCE=INLINE. DATA: HIGH=col(source(s), ELEMENT: BEGIN GPL name("MEAN_OverallCultural interval(position(disabilityW6 shape.interior(shape.ibeam)) SOURCE: PoliticalConformity_HIGH")) *MEAN_BeliefSystemConstrai END GPL. s=userSource(id("graphdata GUIDE: axis(dim(1), nt), set")) label("[ANALYSE] [theme: * Chart Builder. DATA: gender=col(source(s), demographics] Gender female shape.interior(shape.square)) GGRAPH name("gender"), as opposed to male")) ELEMENT: /GRAPHDATA SET unit.category()) GUIDE: axis(dim(2), interval(position(region.sprea NAME="graphdata set" DATA: label("Mean [ANALYSE] d.range(disabilityW6*(LOW+H VARIABLES=disabilityW6 MEAN_FAC1_23=col(source(s) [theme: culture] individual IGH))), MEANCI(FAC1_23, , name("MEAN_FAC1_23")) conformity to overall ", 95)[name="MEAN_FAC1_23" DATA: LOW=col(source(s), "mean political shape.interior(shape.ibeam)) name("MEAN_FAC1_23_LOW orientation")) END GPL. LOW="MEAN_FAC1_23_LOW ")) GUIDE: " DATA: HIGH=col(source(s), text.footnote(label("Error * Chart Builder. HIGH="MEAN_FAC1_23_HIGH name("MEAN_FAC1_23_HIGH Bars: 95% CI")) GGRAPH "] MISSING=LISTWISE ")) SCALE: cat(dim(1), /GRAPHDATA SET REPORTMISSING=NO GUIDE: axis(dim(1), include("0", "1")) NAME="graphdata set" /GRAPHSPEC label("[ANALYSE] [theme: SCALE: linear(dim(2), VARIABLES=disabilityW6 SOURCE=INLINE. demographics] Gender female include(0)) BEGIN GPL as opposed to male")) ELEMENT: MEANCI(OverallCulturalPolitic SOURCE: GUIDE: axis(dim(2), interval(position(gender*MEA alConformity, s=userSource(id("graphdata label("Mean [ANALYSE] N_OverallCulturalPoliticalConf 95)[name="MEAN_OverallCult set")) [theme: culture] [conformity] ormity), uralPoliticalConformity" DATA: Generalised (British) ", disabilityW6=col(source(s), "Cultural Conformity shape.interior(shape.square)) LOW="MEAN_OverallCultural name("disabilityW6"), Regression Factor")) ELEMENT: PoliticalConformity_LOW" unit.category()) GUIDE: interval(position(region.sprea DATA: text.footnote(label("Error d.range(gender*(LOW+HIGH)) HIGH="MEAN_OverallCultural MEAN_FAC1_23=col(source(s) Bars: 95% CI")) ), PoliticalConformity_HIGH"] , name("MEAN_FAC1_23")) SCALE: cat(dim(1), shape.interior(shape.ibeam)) MISSING=LISTWISE DATA: LOW=col(source(s), include("0", "1")) END GPL. REPORTMISSING=NO name("MEAN_FAC1_23_LOW SCALE: linear(dim(2), /GRAPHSPEC ")) include(0)) * Chart Builder. SOURCE=INLINE. DATA: HIGH=col(source(s), ELEMENT: GGRAPH BEGIN GPL name("MEAN_FAC1_23_HIGH interval(position(gender*MEA /GRAPHDATA SET SOURCE: ")) N_FAC1_23), NAME="graphdata set" s=userSource(id("graphdata GUIDE: axis(dim(1), shape.interior(shape.square)) VARIABLES=gender set")) label("[ANALYSE] [theme: ELEMENT: MEANCI(BeliefSystemConstrai DATA: demographics] Disabled as interval(position(region.sprea nt, disabilityW6=col(source(s), opposed to not disabled")) d.range(gender*(LOW+HIGH)) name("disabilityW6"), GUIDE: axis(dim(2), ), 95)[name="MEAN_BeliefSyste unit.category()) label("Mean [ANALYSE] shape.interior(shape.ibeam)) mConstraint" DATA: [theme: culture] [conformity] END GPL. LOW="MEAN_BeliefSystemCo MEAN_OverallCulturalPolitical Generalised (British) ", nstraint_LOW" Conformity=col(source(s), "Cultural Conformity * Chart Builder. Regression Factor")) GGRAPH HIGH="MEAN_BeliefSystemCo name("MEAN_OverallCultural GUIDE: /GRAPHDATA SET nstraint_HIGH"] PoliticalConformity")) text.footnote(label("Error NAME="graphdata set" MISSING=LISTWISE DATA: LOW=col(source(s), Bars: 95% CI")) VARIABLES=gender REPORTMISSING=NO name("MEAN_OverallCultural SCALE: cat(dim(1), MEANCI(OverallCulturalPolitic /GRAPHSPEC PoliticalConformity_LOW")) include("0", "1")) alConformity, SOURCE=INLINE. DATA: HIGH=col(source(s), SCALE: linear(dim(2), BEGIN GPL name("MEAN_OverallCultural include(0)) 95)[name="MEAN_OverallCult PoliticalConformity_HIGH")) uralPoliticalConformity"

101

SOURCE: profile_ethnicity GRAPH /MODEL=LINEAR s=userSource(id("graphdata livedAbroadW8 /BAR(SIMPLE)=MEAN(FAC1_2 QUADRATIC CUBIC set")) parentsForeignW8 3) BY TSC_4171 /PLOT FIT. DATA: gender=col(source(s), /DISTANCE LIKELIHOOD /INTERVAL CI(95.0). name("gender"), /NUMCLUSTERS AUTO 15 REGRESSION unit.category()) BIC GRAPH /MISSING DATA: /HANDLENOISE 0 /BAR(SIMPLE)=MEAN(Overall MEANSUBSTITUTION MEAN_BeliefSystemConstrain /MEMALLOCATE 64 CulturalPoliticalConformity) /STATISTICS COEFF OUTS R t=col(source(s), /CRITERIA INITHRESHOLD(0) BY TSC_4171 ANOVA COLLIN TOL name("MEAN_BeliefSystemCo MXBRANCH(8) MXLEVEL(3) /INTERVAL CI(95.0). /CRITERIA=PIN(.05) nstraint")) /VIEWMODEL DISPLAY=YES. POUT(.10) DATA: LOW=col(source(s), GRAPH /NOORIGIN name("MEAN_BeliefSystemCo TWOSTEP CLUSTER /BAR(SIMPLE)=MEAN(BeliefSy /DEPENDENT FAC1_31 nstraint_LOW")) /CATEGORICAL stemConstraint) BY TSC_4171 /METHOD=ENTER FAC1_20 DATA: HIGH=col(source(s), VARIABLES=country /INTERVAL CI(95.0). FAC1_21 FAC1_24 FAC1_23 name("MEAN_BeliefSystemCo countryOfBirth ZOverallCulturalPoliticalConfo nstraint_HIGH")) profile_ethnicity RECODE TSC_4171 (2=1) rmity GUIDE: axis(dim(1), livedAbroadW8 (MISSING=SYSMIS) (ELSE=0) ZBeliefSystemConstraint label("[ANALYSE] [theme: parentsForeignW8 INTO Celtic. AngloSaxon Celtic demographics] Gender female /DISTANCE LIKELIHOOD VARIABLE LABELS Celtic ZReflectedLnReflectedItemRe as opposed to male")) /NUMCLUSTERS AUTO 15 'Celtic cultural cluster as sponseRate FAC2_21 GUIDE: axis(dim(2), BIC opposed to External cluster'. ZangerScale label("Mean [ANALYSE] /HANDLENOISE 0 EXECUTE. ZfearScale FAC1_19 [theme: culture] Reflected /MEMALLOCATE 64 FAC1_16 FAC1_15 FAC1_13 standard deviation of ", /CRITERIA INITHRESHOLD(0) RECODE TSC_4171 (3=1) gender ZageW7 anyUniW7 "policy preferences, as a MXBRANCH(8) MXLEVEL(3) (MISSING=SYSMIS) (ELSE=0) disabilityW6 Upskilled measure of belief system /VIEWMODEL DISPLAY=YES INTO AngloSaxon. Midskilled constraint")) /SAVE VARIABLE=TSC_3694. VARIABLE LABELS Celtic Unskilled Broadsheet GUIDE: 'Anglo-Saxon cultural cluster Tabloid text.footnote(label("Error FREQUENCIES as opposed to External /SCATTERPLOT=(*ZRESID Bars: 95% CI")) VARIABLES=TSC_4171 cluster'. ,*ZPRED) SCALE: cat(dim(1), /ORDER=ANALYSIS. EXECUTE. /RESIDUALS include("0", "1")) DESCRIPTIVES HISTOGRAM(ZRESID) SCALE: linear(dim(2), CROSSTABS VARIABLES=FAC1_23 NORMPROB(ZRESID). include(0)) /TABLES=TSC_4171 BY OverallCulturalPoliticalConfor ELEMENT: profile_ethnicity mity BeliefSystemConstraint REGRESSION interval(position(gender*MEA countryOfBirth country ageW7 /MISSING N_BeliefSystemConstraint), livedAbroadW8 FAC1_13 FAC1_15 FAC1_16 MEANSUBSTITUTION shape.interior(shape.square)) parentsForeignW8 FAC1_19 FAC1_32 FAC2_32 /STATISTICS COEFF OUTS R ELEMENT: /FORMAT=AVALUE TABLES FAC3_32 FAC1_31 FAC1_20 ANOVA interval(position(region.sprea /STATISTICS=CHISQ angerScale fearScale /CRITERIA=PIN(.05) d.range(gender*(LOW+HIGH)) /CELLS=COUNT ROW FAC1_21 FAC2_21 POUT(.10) ), COLUMN ReflectedLnReflectedItemRes /NOORIGIN shape.interior(shape.ibeam)) /COUNT ROUND CELL ponseRate FAC1_24 /DEPENDENT FAC1_32 END GPL. /BARCHART. /STATISTICS=MEAN STDDEV /METHOD=ENTER FAC1_20 MIN MAX. FAC1_21 FAC1_24 FAC1_23 TWOSTEP CLUSTER GRAPH ZOverallCulturalPoliticalConfo /CATEGORICAL /BAR(SIMPLE)=MEAN(FAC1_3 DESCRIPTIVES rmity VARIABLES=country 2) BY TSC_4171 VARIABLES=OverallCulturalPol ZBeliefSystemConstraint countryOfBirth /INTERVAL CI(95.0). iticalConformity AngloSaxon Celtic profile_ethnicity gor BeliefSystemConstraint ZReflectedLnReflectedItemRe livedAbroadW8 GRAPH ageW7 angerScale sponseRate FAC2_21 parentsForeignW8 /BAR(SIMPLE)=MEAN(FAC2_3 fearScale ZangerScale /DISTANCE LIKELIHOOD 2) BY TSC_4171 ReflectedLnReflectedItemRes ZfearScale FAC1_19 /NUMCLUSTERS AUTO 15 /INTERVAL CI(95.0). ponseRate FAC1_16 FAC1_15 FAC1_13 BIC /SAVE gender ZageW7 anyUniW7 /HANDLENOISE 0 GRAPH /STATISTICS=MEAN STDDEV disabilityW6 Upskilled /MEMALLOCATE 64 /BAR(SIMPLE)=MEAN(FAC3_3 MIN MAX. Midskilled /CRITERIA INITHRESHOLD(0) 2) BY TSC_4171 Unskilled Broadsheet MXBRANCH(8) MXLEVEL(3) /INTERVAL CI(95.0). DESCRIPTIVES Tabloid /VIEWMODEL DISPLAY=YES. VARIABLES=ZOverallCulturalP /SCATTERPLOT=(*ZRESID GRAPH oliticalConformity ,*ZPRED) TWOSTEP CLUSTER /BAR(SIMPLE)=MEAN(FAC1_3 ZBeliefSystemConstraint /RESIDUALS /CATEGORICAL 1) BY TSC_4171 ZageW7 HISTOGRAM(ZRESID) VARIABLES=country /INTERVAL CI(95.0). ZangerScale ZfearScale NORMPROB(ZRESID). countryOfBirth ZReflectedLnReflectedItemRe profile_ethnicity GRAPH sponseRate REGRESSION livedAbroadW8 /BAR(SIMPLE)=MEAN(FAC1_2 /STATISTICS=MEAN STDDEV /MISSING parentsForeignW8 0) BY TSC_4171 MIN MAX. MEANSUBSTITUTION /DISTANCE LIKELIHOOD /INTERVAL CI(95.0). /STATISTICS COEFF OUTS R /NUMCLUSTERS AUTO 15 * Curve Estimation. ANOVA BIC GRAPH TSET NEWVAR=NONE. /CRITERIA=PIN(.05) /HANDLENOISE 0 /BAR(SIMPLE)=MEAN(FAC1_2 CURVEFIT POUT(.10) /MEMALLOCATE 64 1) BY TSC_4171 /VARIABLES=FAC1_31 /NOORIGIN /CRITERIA INITHRESHOLD(0) /INTERVAL CI(95.0). FAC1_32 FAC2_32 FAC3_32 /DEPENDENT FAC2_32 MXBRANCH(8) MXLEVEL(3) FAC1_23 /METHOD=ENTER FAC1_20 /VIEWMODEL DISPLAY=YES. GRAPH ZOverallCulturalPoliticalConfo FAC1_21 FAC1_24 FAC1_23 /BAR(SIMPLE)=MEAN(FAC1_2 rmity ZOverallCulturalPoliticalConfo TWOSTEP CLUSTER 4) BY TSC_4171 ZBeliefSystemConstraint rmity /CATEGORICAL /INTERVAL CI(95.0). WITH FAC1_20 ZBeliefSystemConstraint VARIABLES=country /CONSTANT AngloSaxon Celtic

102

ZReflectedLnReflectedItemRe /CRITERIA=PIN(.05) /MISSING /MISSING sponseRate FAC2_21 POUT(.10) MEANSUBSTITUTION MEANSUBSTITUTION ZangerScale /NOORIGIN /STATISTICS COEFF OUTS R /STATISTICS COEFF OUTS R ZfearScale FAC1_19 /DEPENDENT ANOVA COLLIN TOL ANOVA COLLIN TOL FAC1_16 FAC1_15 FAC1_13 ZOverallCulturalPoliticalConfo /CRITERIA=PIN(.05) /CRITERIA=PIN(.05) gender ZageW7 anyUniW7 rmity POUT(.10) POUT(.10) disabilityW6 Upskilled /METHOD=ENTER FAC1_20 /NOORIGIN /NOORIGIN Midskilled FAC1_21 FAC1_24 /DEPENDENT Leftness /DEPENDENT Leftness Unskilled Broadsheet AngloSaxon Celtic /METHOD=ENTER Aom /METHOD=ENTER Aom Tabloid ZReflectedLnReflectedItemRe TolUncer Conform Compform TolUncer Conform Compform /SCATTERPLOT=(*ZRESID sponseRate Cnstrnt ResPol Anglo Celtic Cnstrnt ResPol Anglo Celtic ,*ZPRED) FAC2_21 ZangerScale AngloAom CeltAom ResRate AngloRP CeltRP ResRate /RESIDUALS ZfearScale FAC1_19 FAC1_16 Willrisk AngerPty FearPty Willrisk AngerPty FearPty HISTOGRAM(ZRESID) FAC1_15 FAC1_13 gender ConPart TrustMPs Econom ConPart TrustMPs Econom NORMPROB(ZRESID). ZageW7 anyUniW7 Engage Gender Zage Engage Gender Zage disabilityW6 Graduate Disabled Upskill Graduate Disabled Upskill REGRESSION Upskilled Midskilled Midskill Unskill Bdsheet Midskill Unskill Bdsheet /MISSING Unskilled Broadsheet Tabloid Tbloid Tbloid MEANSUBSTITUTION /SCATTERPLOT=(*ZRESID /SCATTERPLOT=(*ZRESID /SCATTERPLOT=(*ZRESID /STATISTICS COEFF OUTS R ,*ZPRED) ,*ZPRED) ,*ZPRED) ANOVA /RESIDUALS /RESIDUALS /RESIDUALS /CRITERIA=PIN(.05) HISTOGRAM(ZRESID) HISTOGRAM(ZRESID) HISTOGRAM(ZRESID) POUT(.10) NORMPROB(ZRESID). NORMPROB(ZRESID) NORMPROB(ZRESID) /NOORIGIN /SAVE PRED. /SAVE PRED. /DEPENDENT FAC3_32 REGRESSION /METHOD=ENTER FAC1_20 /MISSING * Chart Builder. * Chart Builder. FAC1_21 FAC1_24 FAC1_23 MEANSUBSTITUTION GGRAPH GGRAPH ZOverallCulturalPoliticalConfo /STATISTICS COEFF OUTS R /GRAPHDATA SET /GRAPHDATA SET rmity ANOVA COLLIN TOL NAME="graphdata set" NAME="graphdata set" ZBeliefSystemConstraint /CRITERIA=PIN(.05) VARIABLES=Aom PRE_1 VARIABLES=Aom AngloSaxon Celtic POUT(.10) TSC_4171 MISSING=LISTWISE PRE_1RpCluster TSC_4171 ZReflectedLnReflectedItemRe /NOORIGIN REPORTMISSING=NO MISSING=LISTWISE sponseRate FAC2_21 /DEPENDENT /GRAPHSPEC REPORTMISSING=NO ZangerScale ZBeliefSystemConstraint SOURCE=INLINE. /GRAPHSPEC ZfearScale FAC1_19 /METHOD=ENTER FAC1_20 BEGIN GPL SOURCE=INLINE. FAC1_16 FAC1_15 FAC1_13 FAC1_21 FAC1_24 SOURCE: BEGIN GPL gender ZageW7 anyUniW7 AngloSaxon Celtic s=userSource(id("graphdata SOURCE: disabilityW6 Upskilled ZReflectedLnReflectedItemRe set")) s=userSource(id("graphdata Midskilled sponseRate DATA: Aom=col(source(s), set")) Unskilled Broadsheet FAC2_21 ZangerScale name("Aom")) DATA: Aom=col(source(s), Tabloid ZfearScale FAC1_19 FAC1_16 DATA: PRE_1=col(source(s), name("Aom")) /SCATTERPLOT=(*ZRESID FAC1_15 FAC1_13 gender name("PRE_1")) DATA: ,*ZPRED) ZageW7 anyUniW7 DATA: PRE_1RpCluster=col(source(s) /RESIDUALS disabilityW6 TSC_4171=col(source(s), , name("PRE_1RpCluster")) HISTOGRAM(ZRESID) Upskilled Midskilled name("TSC_4171"), DATA: NORMPROB(ZRESID). Unskilled Broadsheet Tabloid unit.category()) TSC_4171=col(source(s), /SCATTERPLOT=(*ZRESID GUIDE: axis(dim(1), name("TSC_4171"), DATA SET ACTIVATE Data ,*ZPRED) label("[ANALYSE] [theme: unit.category()) set1. /RESIDUALS psychology] [aom] Active GUIDE: axis(dim(1), REGRESSION HISTOGRAM(ZRESID) open-mindedness ", label("[ANALYSE] [theme: /MISSING NORMPROB(ZRESID). "regression factor")) psychology] [aom] Active MEANSUBSTITUTION GUIDE: axis(dim(2), open-mindedness ", /STATISTICS COEFF OUTS R COMPUTE label("Unstandardized "regression factor")) ANOVA AngloAom=AngloSaxon * Predicted Value")) GUIDE: axis(dim(2), /CRITERIA=PIN(.05) Aom. GUIDE: label("Unstandardized POUT(.10) VARIABLE LABELS AngloAom legend(aesthetic(aesthetic.col Predicted Value for Rp*cluster /NOORIGIN 'Anglo*Aom interaction'. or.exterior), label("TwoStep regression")) /DEPENDENT FAC1_23 EXECUTE. Cluster Number [CULTURAL ", GUIDE: /METHOD=ENTER FAC1_20 "CLUSTER!]")) legend(aesthetic(aesthetic.col FAC1_21 FAC1_24 COMPUTE CeltAom=Celtic * SCALE: or.exterior), label("TwoStep AngloSaxon Celtic Aom. cat(aesthetic(aesthetic.color.e Cluster Number [CULTURAL ", ZReflectedLnReflectedItemRe VARIABLE LABELS CeltAom xterior), include("1", "2", "3")) "CLUSTER!]")) sponseRate 'Celt*Aom interaction'. ELEMENT: SCALE: FAC2_21 ZangerScale EXECUTE. point(position(Aom*PRE_1), cat(aesthetic(aesthetic.color.e ZfearScale FAC1_19 FAC1_16 color.exterior(TSC_4171)) xterior), include("1", "2", "3")) FAC1_15 FAC1_13 gender COMPUTE END GPL. ELEMENT: ZageW7 anyUniW7 AngloRP=AngloSaxon * point(position(Aom*PRE_1Rp disabilityW6 ResPol. COMPUTE CstAom=Cnstrnt * Cluster), Upskilled Midskilled VARIABLE LABELS AngloRP Aom. color.exterior(TSC_4171)) Unskilled Broadsheet Tabloid 'Anglo*Response Polarisation VARIABLE LABELS CstAom END GPL. /SCATTERPLOT=(*ZRESID interaction'. 'Constraint*aom interaction'. ,*ZPRED) EXECUTE. EXECUTE. REGRESSION /RESIDUALS /MISSING HISTOGRAM(ZRESID) COMPUTE CeltRP=Celtic * COMPUTE CstRP=Cnstrnt * MEANSUBSTITUTION NORMPROB(ZRESID). ResPol. ResPol. /STATISTICS COEFF OUTS R VARIABLE LABELS CeltRP VARIABLE LABELS CstRP ANOVA COLLIN TOL REGRESSION 'Celt*Response Polarisation 'Constraint*response /CRITERIA=PIN(.05) /MISSING interaction'. polarisation interaction'. POUT(.10) MEANSUBSTITUTION EXECUTE. EXECUTE. /NOORIGIN /STATISTICS COEFF OUTS R /DEPENDENT Leftness ANOVA REGRESSION REGRESSION

103

/METHOD=ENTER Aom mCst), /GRAPHDATA SET VARIABLE LABELS ConstCmp TolUncer Conform Compform color.exterior(TSC_5725Const NAME="graphdata set" 'Constraint*computed Cnstrnt CstAom ResPol Anglo raint)) VARIABLES=ResPol conformity interaction'. Celtic ResRate Willrisk END GPL. PRE_1RpCluster TSC_4171 EXECUTE. AngerPty FearPty ConPart MISSING=LISTWISE TrustMPs Econom Engage REGRESSION REPORTMISSING=NO REGRESSION Gender Zage Graduate /MISSING /GRAPHSPEC /MISSING Disabled Upskill Midskill MEANSUBSTITUTION SOURCE=INLINE. MEANSUBSTITUTION Unskill Bdsheet Tbloid /STATISTICS COEFF OUTS R BEGIN GPL /STATISTICS COEFF OUTS R /SCATTERPLOT=(*ZRESID ANOVA COLLIN TOL SOURCE: ANOVA COLLIN TOL ,*ZPRED) /CRITERIA=PIN(.05) s=userSource(id("graphdata /CRITERIA=PIN(.05) /RESIDUALS POUT(.10) set")) POUT(.10) HISTOGRAM(ZRESID) /NOORIGIN DATA: ResPol=col(source(s), /NOORIGIN NORMPROB(ZRESID) /DEPENDENT Leftness name("ResPol")) /DEPENDENT Leftness /SAVE PRED. /METHOD=ENTER Aom DATA: /METHOD=ENTER Aom TolUncer Conform Compform PRE_1RpCluster=col(source(s) TolUncer Conform Compform TWOSTEP CLUSTER Cnstrnt ResPol CstRP Anglo , name("PRE_1RpCluster")) Cnstrnt ResPol Anglo Celtic /CONTINUOUS Celtic ResRate Willrisk DATA: AngloCnf CeltCnf ResRate VARIABLES=Cnstrnt AngerPty FearPty ConPart TSC_4171=col(source(s), Willrisk AngerPty FearPty /DISTANCE LIKELIHOOD TrustMPs Econom Engage name("TSC_4171"), ConPart TrustMPs Econom /NUMCLUSTERS AUTO 15 Gender Zage Graduate unit.category()) Engage Gender Zage BIC Disabled Upskill Midskill GUIDE: axis(dim(1), Graduate Disabled Upskill /HANDLENOISE 0 Unskill Bdsheet Tbloid label("[ANALYSE] [theme: Midskill Unskill Bdsheet /MEMALLOCATE 64 /SCATTERPLOT=(*ZRESID psychology] Response Tbloid /CRITERIA INITHRESHOLD(0) ,*ZPRED) polarisation chief ", /SCATTERPLOT=(*ZRESID MXBRANCH(8) MXLEVEL(3) /RESIDUALS "regression factor")) ,*ZPRED) /VIEWMODEL DISPLAY=YES HISTOGRAM(ZRESID) GUIDE: axis(dim(2), /RESIDUALS /SAVE VARIABLE=TSC_2997. NORMPROB(ZRESID) label("Unstandardized HISTOGRAM(ZRESID) /SAVE PRED. Predicted Value for Rp*cluster NORMPROB(ZRESID) GRAPH regression")) /SAVE PRED. /BAR(SIMPLE)=MEAN(Cnstrnt) * Chart Builder. GUIDE: BY TSC_5725. GGRAPH legend(aesthetic(aesthetic.col REGRESSION /GRAPHDATA SET or.exterior), label("TwoStep /MISSING FREQUENCIES NAME="graphdata set" Cluster Number [CULTURAL ", MEANSUBSTITUTION VARIABLES=TSC_5725 VARIABLES=Aom PRE_1RpCst "CLUSTER!]")) /STATISTICS COEFF OUTS R /ORDER=ANALYSIS. TSC_5725Constraint SCALE: ANOVA COLLIN TOL MISSING=LISTWISE cat(aesthetic(aesthetic.color.e /CRITERIA=PIN(.05) * Chart Builder. REPORTMISSING=NO xterior), include("1", "2", "3")) POUT(.10) GGRAPH /GRAPHSPEC ELEMENT: /NOORIGIN /GRAPHDATA SET SOURCE=INLINE. point(position(ResPol*PRE_1R /DEPENDENT Leftness NAME="graphdata set" BEGIN GPL pCluster), /METHOD=ENTER Aom VARIABLES=Aom SOURCE: color.exterior(TSC_4171)) TolUncer Conform Compform PRE_1AomCst s=userSource(id("graphdata END GPL. Cnstrnt ConstCnf ResPol Anglo TSC_5725Constraint set")) Celtic ResRate Willrisk MISSING=LISTWISE DATA: Aom=col(source(s), COMPUTE AngloCnf=Anglo * AngerPty FearPty ConPart REPORTMISSING=NO name("Aom")) Conform. TrustMPs Econom Engage /GRAPHSPEC DATA: VARIABLE LABELS AngloCnf Gender Zage Graduate SOURCE=INLINE. PRE_1RpCst=col(source(s), 'Anglo*conformity Disabled Upskill Midskill BEGIN GPL name("PRE_1RpCst")) interaction'. Unskill Bdsheet Tbloid SOURCE: DATA: EXECUTE. /SCATTERPLOT=(*ZRESID s=userSource(id("graphdata TSC_5725Constraint=col(sour ,*ZPRED) set")) ce(s), COMPUTE CeltCnf=Celtic * /RESIDUALS DATA: Aom=col(source(s), name("TSC_5725Constraint"), Conform. HISTOGRAM(ZRESID) name("Aom")) unit.category()) VARIABLE LABELS CeltCnf NORMPROB(ZRESID) DATA: GUIDE: axis(dim(1), 'Celt*conformity interaction'. /SAVE PRED. PRE_1AomCst=col(source(s), label("[ANALYSE] [theme: EXECUTE. name("PRE_1AomCst")) psychology] [aom] Active * Chart Builder. DATA: open-mindedness ", COMPUTE ConstCnf=Cnstrnt * GGRAPH TSC_5725Constraint=col(sour "regression factor")) Conform. /GRAPHDATA SET ce(s), GUIDE: axis(dim(2), VARIABLE LABELS ConstCnf NAME="graphdata set" name("TSC_5725Constraint"), label("Unstandardized 'Constraint*conformity VARIABLES=Conform unit.category()) Predicted Value for interaction'. PRE_2CnfCst GUIDE: axis(dim(1), Rp*constraint regression")) EXECUTE. TSC_5725Constraint label("[ANALYSE] [theme: GUIDE: MISSING=LISTWISE psychology] [aom] Active legend(aesthetic(aesthetic.col COMPUTE AngloCmp=Anglo * REPORTMISSING=NO open-mindedness ", or.exterior), label("TwoStep Compform. /GRAPHSPEC "regression factor")) Cluster Number for VARIABLE LABELS AngloCmp SOURCE=INLINE. GUIDE: axis(dim(2), Constraint")) 'Anglo*computed conformity BEGIN GPL label("Unstandardized SCALE: interaction'. SOURCE: Predicted Value for cat(aesthetic(aesthetic.color.e EXECUTE. s=userSource(id("graphdata Aom*constraint regression")) xterior), include("1", "2", "3")) set")) GUIDE: ELEMENT: COMPUTE CeltCmp=Celtic * DATA: legend(aesthetic(aesthetic.col point(position(Aom*PRE_1Rp Compform. Conform=col(source(s), or.exterior), label("TwoStep Cst), VARIABLE LABELS CeltCmp name("Conform")) Cluster Number for color.exterior(TSC_5725Const 'Celt*computed conformity DATA: Constraint")) raint)) interaction'. PRE_2CnfCst=col(source(s), SCALE: END GPL. EXECUTE. name("PRE_2CnfCst")) cat(aesthetic(aesthetic.color.e DATA: xterior), include("1", "2", "3")) * Chart Builder. COMPUTE ConstCmp=Cnstrnt TSC_5725Constraint=col(sour ELEMENT: GGRAPH * Compform. ce(s), point(position(Aom*PRE_1Ao

104 name("TSC_5725Constraint"), culture] Z individual Predicted Value for computed /ORDER=ANALYSIS. unit.category()) conformity to overall mean ", conformity*constraint ", GUIDE: axis(dim(1), "political orientation")) "regression")) GRAPH label("[ANALYSE] [theme: GUIDE: axis(dim(2), GUIDE: /BAR(SIMPLE)=MEAN(Aom) culture] [conformity] label("Unstandardized legend(aesthetic(aesthetic.col BY TSC_5075. Generalised (British) ", Predicted Value for computed or.exterior), label("TwoStep "Cultural Conformity conformity*cluster ", Cluster Number for RECODE TSC_5075 (3=1) (1=2) Regression Factor")) "regression")) Constraint")) (2=3). GUIDE: axis(dim(2), GUIDE: SCALE: EXECUTE. label("Unstandardized legend(aesthetic(aesthetic.col cat(aesthetic(aesthetic.color.e Predicted Value for or.exterior), label("TwoStep xterior), include("1", "2", "3")) * Chart Builder. conformity*constraint Cluster Number [CULTURAL ", ELEMENT: GGRAPH regression")) "CLUSTER!]")) point(position(Compform*PR /GRAPHDATA SET GUIDE: SCALE: E_1CmpCst), NAME="graphdata set" legend(aesthetic(aesthetic.col cat(aesthetic(aesthetic.color.e color.exterior(TSC_5725Const VARIABLES=ResPol or.exterior), label("TwoStep xterior), include("1", "2", "3")) raint)) PRE_1AomRP TSC_5075 Cluster Number for ELEMENT: END GPL. MISSING=LISTWISE Constraint")) point(position(Compform*PR REPORTMISSING=NO SCALE: E_1CmpCluster), COMPUTE AomRP=Aom * /GRAPHSPEC cat(aesthetic(aesthetic.color.e color.exterior(TSC_4171)) ResPol. SOURCE=INLINE. xterior), include("1", "2", "3")) END GPL. VARIABLE LABELS AomRP BEGIN GPL ELEMENT: 'Aom*response polarisation SOURCE: point(position(Conform*PRE_ REGRESSION interaction'. s=userSource(id("graphdata 2CnfCst), /MISSING EXECUTE. set")) color.exterior(TSC_5725Const MEANSUBSTITUTION DATA: ResPol=col(source(s), raint)) /STATISTICS COEFF OUTS R COMPUTE AomTlunc=Aom * name("ResPol")) END GPL. ANOVA COLLIN TOL TolUncer. DATA: /CRITERIA=PIN(.05) VARIABLE LABELS AomTlunc PRE_1AomRP=col(source(s), REGRESSION POUT(.10) 'Aom*tolerance of name("PRE_1AomRP")) /MISSING /NOORIGIN uncertainty interaction'. DATA: MEANSUBSTITUTION /DEPENDENT Leftness EXECUTE. TSC_5075=col(source(s), /STATISTICS COEFF OUTS R /METHOD=ENTER Aom name("TSC_5075"), ANOVA COLLIN TOL TolUncer ResPol Conform COMPUTE RPTlunc=ResPol * unit.category()) /CRITERIA=PIN(.05) Compform Cnstrnt ConstCmp TolUncer. GUIDE: axis(dim(1), POUT(.10) Anglo Celtic ResRate Willrisk VARIABLE LABELS RPTlunc label("[ANALYSE] [theme: /NOORIGIN AngerPty FearPty ConPart 'Response psychology] Response /DEPENDENT Leftness TrustMPs Econom Engage polarisation*tolerance of polarisation chief ", /METHOD=ENTER Aom Gender Zage Graduate uncertainty interaction'. "regression factor")) TolUncer Conform Compform Disabled Upskill Midskill EXECUTE. GUIDE: axis(dim(2), Cnstrnt ResPol Anglo Celtic Unskill Bdsheet Tbloid label("Unstandardized AngloCmp CeltCmp ResRate /SCATTERPLOT=(*ZRESID REGRESSION Predicted Value for Willrisk AngerPty FearPty ,*ZPRED) /MISSING aom*response polarisation ", ConPart TrustMPs Econom /RESIDUALS MEANSUBSTITUTION "regression")) Engage Gender Zage HISTOGRAM(ZRESID) /STATISTICS COEFF OUTS R GUIDE: Graduate Disabled Upskill NORMPROB(ZRESID) ANOVA COLLIN TOL legend(aesthetic(aesthetic.col Midskill Unskill Bdsheet /SAVE PRED. /CRITERIA=PIN(.05) or.exterior), label("TwoStep Tbloid POUT(.10) Cluster Number for active ", /SCATTERPLOT=(*ZRESID * Chart Builder. /NOORIGIN "open-mindedness")) ,*ZPRED) GGRAPH /DEPENDENT Conform SCALE: /RESIDUALS /GRAPHDATA SET /METHOD=ENTER Aom cat(aesthetic(aesthetic.color.e HISTOGRAM(ZRESID) NAME="graphdata set" TolUncer ResPol AomRP xterior), include("1", "2", "3")) NORMPROB(ZRESID) VARIABLES=Compform Compform Cnstrnt Anglo ELEMENT: /SAVE PRED. PRE_1CmpCst Celtic ResRate Willrisk point(position(ResPol*PRE_1A TSC_5725Constraint AngerPty omRP), * Chart Builder. MISSING=LISTWISE FearPty ConPart TrustMPs color.exterior(TSC_5075)) GGRAPH REPORTMISSING=NO Econom Engage Gender Zage END GPL. /GRAPHDATA SET /GRAPHSPEC Graduate Disabled Upskill NAME="graphdata set" SOURCE=INLINE. Midskill Unskill REGRESSION VARIABLES=Compform BEGIN GPL Bdsheet Tbloid /MISSING PRE_1CmpCluster TSC_4171 SOURCE: /SCATTERPLOT=(*ZRESID MEANSUBSTITUTION MISSING=LISTWISE s=userSource(id("graphdata ,*ZPRED) /STATISTICS COEFF OUTS R REPORTMISSING=NO set")) /RESIDUALS ANOVA COLLIN TOL /GRAPHSPEC DATA: HISTOGRAM(ZRESID) /CRITERIA=PIN(.05) SOURCE=INLINE. Compform=col(source(s), NORMPROB(ZRESID) POUT(.10) BEGIN GPL name("Compform")) /SAVE PRED. /NOORIGIN SOURCE: DATA: /DEPENDENT Compform s=userSource(id("graphdata PRE_1CmpCst=col(source(s), TWOSTEP CLUSTER /METHOD=ENTER Aom set")) name("PRE_1CmpCst")) /CONTINUOUS TolUncer ResPol Conform DATA: DATA: VARIABLES=Aom AomRP Cnstrnt Anglo Celtic Compform=col(source(s), TSC_5725Constraint=col(sour /DISTANCE LIKELIHOOD ResRate Willrisk AngerPty name("Compform")) ce(s), /NUMCLUSTERS AUTO 15 FearPty ConPart TrustMPs DATA: name("TSC_5725Constraint"), BIC Econom Engage Gender Zage PRE_1CmpCluster=col(source( unit.category()) /HANDLENOISE 0 Graduate Disabled Upskill s), GUIDE: axis(dim(1), /MEMALLOCATE 64 Midskill Unskill name("PRE_1CmpCluster")) label("[ANALYSE] [theme: /CRITERIA INITHRESHOLD(0) Bdsheet Tbloid DATA: culture] Z individual MXBRANCH(8) MXLEVEL(3) /SCATTERPLOT=(*ZRESID TSC_4171=col(source(s), conformity to overall mean ", /VIEWMODEL DISPLAY=YES ,*ZPRED) name("TSC_4171"), "political orientation")) /SAVE VARIABLE=TSC_7109. /RESIDUALS unit.category()) GUIDE: axis(dim(2), HISTOGRAM(ZRESID) GUIDE: axis(dim(1), label("Unstandardized FREQUENCIES NORMPROB(ZRESID) label("[ANALYSE] [theme: VARIABLES=TSC_5075 /SAVE PRED.

105

/CRITERIA=PIN(.05) /METHOD=ENTER Aom ELEMENT: * Chart Builder. POUT(.10) TolUncer ResPol RPTlunc point(position(ResPol*PRE_1R GGRAPH /NOORIGIN Compform Cnstrnt Anglo PTluncCnf), /GRAPHDATA SET /DEPENDENT Compform Celtic ResRate Willrisk color.exterior(TSC_8103)) NAME="graphdata set" /METHOD=ENTER Aom AngerPty END GPL. VARIABLES=ResPol TolUncer ResPol AomTlunc FearPty ConPart TrustMPs PRE_1AomRPCmp TSC_5075 Conform Cnstrnt Anglo Celtic Econom Engage Gender Zage REGRESSION MISSING=LISTWISE ResRate Willrisk AngerPty Graduate Disabled Upskill /MISSING REPORTMISSING=NO FearPty ConPart TrustMPs Midskill Unskill MEANSUBSTITUTION /GRAPHSPEC Econom Engage Gender Zage Bdsheet Tbloid /STATISTICS COEFF OUTS R SOURCE=INLINE. Graduate Disabled Upskill /SCATTERPLOT=(*ZRESID ANOVA COLLIN TOL BEGIN GPL Midskill Unskill ,*ZPRED) /CRITERIA=PIN(.05) SOURCE: Bdsheet Tbloid /RESIDUALS POUT(.10) s=userSource(id("graphdata /SCATTERPLOT=(*ZRESID HISTOGRAM(ZRESID) /NOORIGIN set")) ,*ZPRED) NORMPROB(ZRESID) /DEPENDENT Compform DATA: ResPol=col(source(s), /RESIDUALS /SAVE PRED. /METHOD=ENTER Aom name("ResPol")) HISTOGRAM(ZRESID) TolUncer ResPol RPTlunc DATA: NORMPROB(ZRESID) TWOSTEP CLUSTER Conform Cnstrnt Anglo Celtic PRE_1AomRPCmp=col(source( /SAVE PRED. /CONTINUOUS ResRate Willrisk AngerPty s), VARIABLES=TolUncer FearPty ConPart TrustMPs name("PRE_1AomRPCmp")) * Chart Builder. /DISTANCE LIKELIHOOD Econom Engage Gender Zage DATA: GGRAPH /NUMCLUSTERS AUTO 15 Graduate Disabled Upskill TSC_5075=col(source(s), /GRAPHDATA SET BIC Midskill Unskill name("TSC_5075"), NAME="graphdata set" /HANDLENOISE 0 Bdsheet Tbloid unit.category()) VARIABLES=TolUncer /MEMALLOCATE 64 /SCATTERPLOT=(*ZRESID GUIDE: axis(dim(1), PRE_1AomTluncCmp /CRITERIA INITHRESHOLD(0) ,*ZPRED) label("[ANALYSE] [theme: TSC_5075 MISSING=LISTWISE MXBRANCH(8) MXLEVEL(3) /RESIDUALS psychology] Response REPORTMISSING=NO /VIEWMODEL DISPLAY=YES HISTOGRAM(ZRESID) polarisation chief ", /GRAPHSPEC /SAVE VARIABLE=TSC_1139. NORMPROB(ZRESID) "regression factor")) SOURCE=INLINE. /SAVE PRED. GUIDE: axis(dim(2), BEGIN GPL FREQUENCIES label("Unstandardized SOURCE: VARIABLES=TSC_8103 process vars=Leftness Predicted Value for s=userSource(id("graphdata /ORDER=ANALYSIS. Conform Aom TolUncer aom*response polarisation on set")) ResPol Compform Cnstrnt ", DATA: GRAPH Anglo Celtic ResRate "compformity regression")) TolUncer=col(source(s), /BAR(SIMPLE)=MEAN(TolUnce Willrisk AngerPty FearPty GUIDE: name("TolUncer")) r) BY TSC_8103. ConPart TrustMPs legend(aesthetic(aesthetic.col DATA: Econom Engage Gender or.exterior), label("TwoStep PRE_1AomTluncCmp=col(sour * Chart Builder. Zage Graduate Disabled Cluster Number for active ", ce(s), GGRAPH Upskill Midskill Unskill "open-mindedness")) name("PRE_1AomTluncCmp") /GRAPHDATA SET Bdsheet SCALE: ) NAME="graphdata set" Tbloid/y=Leftness/x=Aom/m= cat(aesthetic(aesthetic.color.e DATA: VARIABLES=ResPol Conform xterior), include("1", "2", "3")) TSC_5075=col(source(s), PRE_1RPTluncCnf TSC_8103 Compform/model=4/total=1/ ELEMENT: name("TSC_5075"), MISSING=LISTWISE effsize=1/boot=10000. point(position(ResPol*PRE_1A unit.category()) REPORTMISSING=NO omRPCmp), GUIDE: axis(dim(1), /GRAPHSPEC RECODE TSC_4171 (3=1) color.exterior(TSC_5075)) label("[ANALYSE] [theme: SOURCE=INLINE. (MISSING=SYSMIS) (ELSE=0) END GPL. psychology] [risk] 1 Tolerance BEGIN GPL INTO AngloCulture. of uncertainty ", SOURCE: VARIABLE LABELS REGRESSION "regression factor")) s=userSource(id("graphdata AngloCulture 'Belongs to the /MISSING GUIDE: axis(dim(2), set")) Anglo-Saxon cultural cluster?'. MEANSUBSTITUTION label("Unstandardized DATA: ResPol=col(source(s), EXECUTE. /STATISTICS COEFF OUTS R Predicted Value for name("ResPol")) ANOVA COLLIN TOL aom*tolerance of uncertainty DATA: FREQUENCIES /CRITERIA=PIN(.05) on ", PRE_1RPTluncCnf=col(source( VARIABLES=AngloCul POUT(.10) "compformity regression")) s), name("PRE_1RPTluncCnf")) /ORDER=ANALYSIS. /NOORIGIN GUIDE: DATA: /DEPENDENT Conform legend(aesthetic(aesthetic.col TSC_8103=col(source(s), SORT VARIABLES BY LABEL (A). /METHOD=ENTER Aom or.exterior), label("TwoStep name("TSC_8103"), TolUncer ResPol AomTlunc Cluster Number for active ", unit.category()) process vars=Leftness Compform Cnstrnt Anglo "open-mindedness")) GUIDE: axis(dim(1), Conform Aom TolUncer Celtic ResRate Willrisk SCALE: label("[ANALYSE] [theme: ResPol Compform Cnstrnt AngerPty FearPty ConPart cat(aesthetic(aesthetic.color.e psychology] Response AngloCul ResRate Willrisk TrustMPs Econom Engage xterior), include("1", "2", "3")) polarisation chief ", AngerPty FearPty ConPart Gender Zage Graduate ELEMENT: "regression factor")) TrustMPs Disabled Upskill Midskill point(position(TolUncer*PRE_ GUIDE: axis(dim(2), Econom Engage Gender Unskill Bdsheet Tbloid 1AomTluncCmp), label("Unstandardized Zage Graduate Disabled /SCATTERPLOT=(*ZRESID color.exterior(TSC_5075)) Predicted Value for response Upskill Midskill Unskill ,*ZPRED) END GPL. polarisation*tolerance ", Bdsheet /RESIDUALS "of uncertainty on Tbloid/y=Leftness/x=Aom/m= HISTOGRAM(ZRESID) REGRESSION conformity regression")) Conform NORMPROB(ZRESID) /MISSING GUIDE: Compform/w=ResPol/z=TolUn /SAVE PRED. MEANSUBSTITUTION legend(aesthetic(aesthetic.col cer/v=AngloCul/q=Cnstrnt/mo /STATISTICS COEFF OUTS R or.exterior), label("TwoStep del=50/boot=10000. REGRESSION ANOVA COLLIN TOL Cluster Number for tolerance /MISSING /CRITERIA=PIN(.05) ", REGRESSION MEANSUBSTITUTION POUT(.10) "of uncertainty")) /MISSING /STATISTICS COEFF OUTS R /NOORIGIN SCALE: MEANSUBSTITUTION ANOVA COLLIN TOL /DEPENDENT Conform cat(aesthetic(aesthetic.color.e /STATISTICS COEFF OUTS R xterior), include("1", "2")) ANOVA COLLIN TOL

106

/CRITERIA=PIN(.05) SCALE: cat(dim(1), businessBonusW4 POUT(.10) include("1.00", "2.00")) govtHandoutsW1 /NOORIGIN ELEMENT: govtHandoutsW2 /DEPENDENT Leftness interval(position(region.sprea govtHandoutsW3 /METHOD=ENTER Aom d.range(whichmed*(MINIMU govtHandoutsW4 AngloCul Cnstrnt TolUncer M_lowCI+MAXIMUM_highCI) immigEconW1 immigEconW2 ResPol ResRate Willrisk *effno)), immigEconW3 immigEconW4 AngerPty FearPty ConPart shape(shape.ibeam), immigrationLevelW4 TrustMPs Econom Engage color.interior(whichmed)) immigrationLevelW6 Gender Zage Graduate ELEMENT: immigrantsWelfareStateW1 Disabled Upskill Midskill point(position(whichmed*ME immigrantsWelfareStateW2 Unskill Bdsheet Tbloid AN_indeffect*effno), immigrantsWelfareStateW3 /SCATTERPLOT=(*ZRESID shape(shape.circle)) immigrantsWelfareStateW4 ,*ZPRED) END GPL. immigrantsWelfareStateW8 /RESIDUALS reasonForUnemploymentW1 HISTOGRAM(ZRESID) GRAPH reasonForUnemploymentW2 NORMPROB(ZRESID). /HILO(SIMPLE)=MEAN(highCI) reasonForUnemploymentW3 MEAN(lowCI) reasonForUnemploymentW4 *The following commands MEAN(indeffect) BY rp al4W6 lr1W6 al5W6 al3W6 were performed on a data set /MISSING=LISTWISE. /PRINT=SPEARMAN of indirect effect sizes for TWOTAIL NOSIG both mediators at a series of GRAPH /MISSING=PAIRWISE values of the four moderator. /HILO(SIMPLE)=MEAN(highCI) N = 108 MEAN(lowCI) MEAN(indeffect) BY tlunc * Chart Builder. /MISSING=LISTWISE. GGRAPH /GRAPHDATA SET GRAPH NAME="graphdata set" /HILO(SIMPLE)=MEAN(highCI) VARIABLES=effno MEAN(lowCI) MAXIMUM(highCI)[name="M MEAN(indeffect) BY constraint AXIMUM_highCI"] /MISSING=LISTWISE.

MINIMUM(lowCI)[name="MI GRAPH NIMUM_lowCI"] /HILO(SIMPLE)=MEAN(highCI) MEAN(indeffect)[name="MEA MEAN(lowCI) N_indeffect"] whichmed MEAN(indeffect) BY culclust MISSING=LISTWISE /MISSING=LISTWISE. REPORTMISSING=NO /GRAPHSPEC DATA SET ACTIVATE Data SOURCE=INLINE. set1. BEGIN GPL RELIABILITY SOURCE: /VARIABLES=FAC1_2 al1W6 s=userSource(id("graphdata /SCALE('ALL VARIABLES') ALL set")) /MODEL=ALPHA. DATA: effno=col(source(s), name("effno"), DATA SET ACTIVATE Data unit.category()) set1. DATA: FREQUENCIES MAXIMUM_highCI=col(source VARIABLES=businessBonusW1 (s), businessBonusW2 name("MAXIMUM_highCI")) businessBonusW3 DATA: businessBonusW4 MINIMUM_lowCI=col(source( govtHandoutsW1 s), name("MINIMUM_lowCI")) govtHandoutsW2 DATA: govtHandoutsW3 MEAN_indeffect=col(source(s govtHandoutsW4 ), name("MEAN_indeffect")) immigEconW1 immigEconW2 DATA: immigEconW3 whichmed=col(source(s), immigEconW4 name("whichmed"), immigrationLevelW4 unit.category()) immigrationLevelW6 COORD: rect(dim(1,2), immigrantsWelfareStateW1 cluster(3,0)) immigrantsWelfareStateW2 GUIDE: axis(dim(3), immigrantsWelfareStateW3 label("effno")) immigrantsWelfareStateW4 GUIDE: axis(dim(2), immigrantsWelfareStateW8 label("Maximum(highCI), reasonForUnemploymentW1 Minimum(lowCI), Mean(indeffect)")) reasonForUnemploymentW2 GUIDE: reasonForUnemploymentW3 legend(aesthetic(aesthetic.col reasonForUnemploymentW4 or.interior), al4W6 lr1W6 al5W6 al3W6 label("whichmed")) /FORMAT=NOTABLE SCALE: linear(dim(2), /NTILES=4 include(0)) /ORDER=ANALYSIS. SCALE: cat(aesthetic(aesthetic.color.i NONPAR CORR nterior), include("1.00", /VARIABLES=businessBonusW "2.00")) 1 businessBonusW2 businessBonusW3

107