Negative Political Emotions
How negative affect is associated with conservatism and monocultural attitudes.
Written by Kas Woudstra 10363009
MSc Student Political Economy Graduate School of Social Sciences University of Amsterdam
Supervision by dhr. dr. G. Schumacher
Assistant Professor in Political Science Social and Behavioural Sciences
University of Amsterdam
Second reader dhr. dr. B. Bakker
Assistant Professor in Communication Studies Amsterdam School of Communication Research
University of Amsterdam
Abstract: Two political forces have shaped political conflict for as long as politics exists. The need to conserve
the political status quo and protect it from novelties is often diametrically opposed to a desire to reshape the political status into an entirely new one. Within political science, the clash between conservative and progressive thinking is traditionally studied through an analysis of theories, idea’s and attitudes. It is, however, a misconception that political behaviour is solely led by discourses. To a great extent, human behaviour is led by affective systems that consciously and unconsciously guide and shape how people think and act. The desire to conserve a status quo is founded on different affective mechanisms than the passion for novelties. The psychological and physiological affective mechanisms that guide conservatives in their behaviour are different than the mechanisms that aid progressives. Because conservatives are more protective towards novelties, they are more sensitive to threats, and they will have more anxious or hostile reactions. More often experienced negative affect might be the main force that underlies the division between conservatives and progressives.
With this article, I add the first contribution in the Netherlands to this new field of political psychology. I both measure threat sensitivity and self-reported levels of negative emotions and prove that both conservatives and people with monocultural attitudes more often or more intensively experience negative emotions.
2 Contents Page number 1. Introduction 3 2. Theory 6 2.1 Subconscious politics 6
2.2 Self-reported psychological differences between conservatives and liberals 8
2.3 Physiological differences 9
2.3.1 Brain structure and activity 10
2.3.2 Behavioural immune system: disgust and threat 10
2.3.2.1 Disgust sensitivity 11
2.3.2.2 Threat sensitivity 11
2.4 On political ideology, four concepts 12
2.4.1 Issues and conservatism 12
2.4.2 Partisanship and right-left self-placement 14
2.5 Moderating factors 14
3. Research question and hypotheses 15
4. Study I Threat sensitivity 18
4. 1 Materials and Methods 18
4.1.1 Participants 18
4.1.2 Procedure 18
4.1.3 Measures 19
4.2 Results 20
5. Study II Self-reported negativity 22
5.1 Materials and Methods 22
5.1.1 Participants and procedure 22
5.1.2 Measures 22
5.1.2.1 Dependent variables 23
5.1.2.2 Independent variables 24
5.2 Results 26
5.2.1 Factor analysis 26
5.2.2 OLS Regression Models 27
6. Conclusion 31 7. Discussion 32 7.1 Study I 32 7.2 Study II 33 8. References 35 9. Appendix 41
3 1. Introduction
Political science has studied political behaviour with a vast blind spot: the blindness for emotions. In its analysis of political behaviour, political science has nearly exclusively focussed on attitudes, values, issues, charisma, political leaders, and structural factors as education, religion, and income. But what if emotions help to shape attitudes, values and identification in distinct ways? Then political scientists have missed an important analytic tool to analyse political behaviour. That seems to have happened (Hibbing et al. 2014; Hoggett and Thompson, 2012; Dalton and Klingemann, 2007; Franklin et al. 1992).
Many people will be familiar with these two distinctive feelings: temper for traditions, and a desire to change the objectionable. Whereas many people know these feelings - by heart, one could literally say - they might be the defining feature that differentiates people from one another in politics. A temper to conserve traditions that are to cherish might underlie the attitudes that conservatives have. An urge to break traditions and replace them with better ones, might be the defining sense that unites progressives.
That conservatives want to conserve customs, and progressives want to change them, is not new. It is a political debate that can be found throughout the entire western history. It is the cultural and political division between the ancient Sparta, being traditionally rule based, and the Athens, being innovative and democratic (Powell: 1998; Hibbing, 2014: 197). It is the essential difference between the French revolutionaries and its conservative counterpart, Burke (1898). And it still drives the polemic politics of contemporary America, strongly dividing the electorates of Donald J. Trump and Hillary R. Clinton (Hochschild, 2016). What is new, is that this antique dichotomy might not be driven by mere differences of attitudes; but be underpinned by differences in emotions. It might be that the attitudinal differences between conservative and progressive people are mere manifestations of something that is equally important; differences in affect.
The consensus in psychology is that affect precedes judging and behaviour (Marcus et al, 2000; Kahneman, 2003: 710; Slovic et al., 2002). Affect and emotions, however, have not been priority of research in political and social science (Hibbing et al. 2014; Hoggett and Thompson, 2012). In these sciences, positivism and behaviouralism were and are strong paradigms (Ibid: 1; Dalton and Klingemann, 2007: 10). Political behaviour is often explained by structural factors as social class, education, income, religiosity, region and gender (Ibid; Franklin et al. 1992). As it is assumed that these structural factors lose their explanatory importance because of deteriorating cleavages (or zuilen in the Netherlands) in society have lost importance, more short-term factors as issue voting and candidate images have been studied intensively (Dalton and Klingemann, 2007: 10-12, Franklin et al. 1992; Dalton 2006). When affect was studied it was merely done to explain how people emotionally identify with specific parties (Dalton and Klingemann, 2007; Lipset and Rokkan, 1967).
4
Emotions as a driving force of political behaviour have been seriously understudied. That is remarkable, since in psychology it is seen as an important facet of human behaviour. It is even more remarkable, because in public discourse and political philosophy, emotions are seen as big forces that alter politics in not always pleasant ways (Nussbaum, 2012; Nussbaum; 2016; Mishra, 2017). When Donald J. Trump won the American elections on November the 8th, the Guardian (Roberts et al., 2016)
started its main article with the following statement:
‘Donald Trump shattered expectations on Tuesday with an election night victory that revealed deep anti-establishment anger among American voters and set the world on a journey into the political unknown’ (Roberts et al., 2016).
Was it anti-establishment anger that made Donald J. Trump win? Does the western world endure an ‘age of anger’ like Mishra (2017) states? We do not know. Maybe anger and fear are just concepts that have become more fashionable to analyse political behavioural trends. Maybe they are not significantly felt more, but just said to be sensed more often or have more effect. Or maybe the Guardian and Mishra (2017) are right, and maybe fear and anger are the prime movers of political behaviour.
Anger and fear might be two emotions that drive political behaviour in significant ways. It could be felt most by people that are prone to traditions and habits, because anger and fear are mostly felt when traditions and habits are broken. Anger and fear can be experienced most by conservatives. When they are, that has serious consequences for the ways in which we explain political behaviour. Then people would be not merely conservative because of different structural factors, or short-term explanations as issues; but also since certain negative emotions are experienced more often.
American research suggests that this correlation exists, but in neither the European, nor the Dutch political system this relation has been studied (Marcus et al., 2000; Hibbing, 2014; Arceneaux, 2017; Dodd et al. 2012; Oxley et al. 2008). It is necessary to test whether conservative people experience more anger and fear in the continental European context. It is urgent to test an assumption that is widespread in public discourse and political philosophy. And it is essential to introduce a new concept to the analytic framework that is used to study political behaviour in continental Europe.
The Netherlands serves as case to test whether conservatives experience more negative emotions. It is a country with a early rise of populism and a stable, consistent existence of populist and conservative parties (Uitermark, 2010; Sniderman and Hagendoor., 2009; Elke, 2010). It suffered the first salafi inspired terrorist attack in Europe by a member of a radicalized salafi Muslim group: the murder on film maker Theo van Gogh (Buruma, 2006). The Netherlands has known, and still has, a strong political polemic on migrants, religion, integration, tolerance, and Dutch culture and identity. In these discussions conservatives and progressives people often opposed (Ibid, Uitermark, 2010).
5
I use two measures to test whether conservative people in the Netherlands experience more negative emotions. The first measure is most innovative and unconventional: that is to measure threat sensitivity. All people are equipped with a protective automatic system that guards people for threats, which is called the surveillance system (Marcus et al. 2000; Aarøe et al., 2017: 278-279; Inbar et al, 2012; Rozin et al, 2008). When a threat is identified, this system is directly activated, resulting in an immediate protective fight-or-flight reaction (Theatero and Penney, 2015). This can happen both consciously and unconsciously (Marcus et al. 2000; Hoggett and Thompson, 2012). The reaction is accompanied with higher arousal, and a symptom of heightened arousal is an increased transpiration (Aarøe et al., 2017: 278-279; Inbar et al, 2012; Rozin et al, 2008). The intensity of this reaction differs between people, and the strength of this reaction is called threat sensitivity (Ibid). Put simply; when people observe a threat, they shy consciously or unconsciously, and that reaction can be measured though levels of sweat production.
A second measure is more conventional in political science. Using a large self-reported survey, I test whether conservative people more often consciously experience negative emotions. A fight-or-flight reaction can be (but is not necessarily) consciously experienced as anger ‘fight’ or fear ‘flight’. With these means, I test for a categorisation what is made within psychology: that between affect and emotions. Affect concerns the unconscious, physiological realm of human activity. Emotions appertain to the conscious feelings that more easily are put into words and thinking (Hoggett and Thompson, 2012; Marcus et al., 2000). In this article, I use the term affect for all the conscious and unconscious experiences subjects can undergo. I use the frame emotion when people consciously perceive negative or positive arousal, like actually felt anger, fear, happiness or satisfaction.
Furthermore, conservatism is used in a broad sense. Conservatism of political ideology supported by an electorate emanates from a quite well demarcated tradition of political thinking. Yet the political ideology of conservatism advocated by people in society is not clearly demarcated. Conservatives in the American electorate are different from conservatives in the Dutch electorate. Conservative parties within the Dutch parliamentary systems can have different ideas on numerous issues. Yet, in a broad sense, being anti-multiculturalism is seen as conservative, as are political parties as the VVD and CDA.
This research is highly explorative. I do not seek to untangle the causal mechanism between emotions and attitude formation. Whereas emotions precede thinking when a stimuli is observed, that does not mean that people are slaves to their emotions. When people are continuously angry, they could talk with their friends about it, think about it, which could make them less angry a subsequent time when something irritating happens. Contemplation on one’s past actions could alter future reactions. Hence, it is not my intention to state that conservatism leads to anger or fear, or vice versa. It is my personal belief that a far more complex, interactive mechanism is at stake, but that is not the essential focus of
6
this research. What is important is to know whether the experiences of anger and fear play any significant role in the difference between distinct ideologies. What is important to know, is whether conservatives endure more negative affect. That is what I study in this article.
2. Theory
2.1 Subconscious politics
Subconscious forces influence how people behave and how people shape their beliefs. In psychology, subconscious factors that influence attitudes and behaviour are widely studied. Kahneman (2003) was a pioneer in disclosing unconscious psychological biases in the way people make choices. People use heuristics to reduce the complexity of making choices. Using certain heuristics can lead to systematic errors in choices people make (Tversky and Kahneman, 1974: 1124; Kahneman, 2003: 706). Among these errors are that people overvalue past certainties, avoid risk taking, and are more influenced by negative incidents than positive events (Ibid: 697-705; see also Hibbing et al. 2014). An important heuristic is the affect heuristic: the fact that choices people make are influenced by their emotions; their fears, their joy, or their anger (Kahneman, 2003: 710). After Tversky and Kahneman (1974) introduced thinking about heuristics, many other scientists have sought to exemplify which heuristics systematically contribute to the ways in which people make judgements. The field of research on affective heuristics affecting political judgement starts with the work done by Zajonc (1980) (Kahneman, 2003: 710; Slovic et al., 2002: 1334). Zajonc (1980) stated that affective reactions precede and guide information processing and judgements. Subsequently, Slovic et al. (2002) introduced a framework that describes how affect guides judgments and decisions. They distinguish a negative and positive experience as a result of a stimulus. Negative or positive reactions occur automatically, being can be conscious or unconscious, and guide the ways in which people shape their beliefs, and make their choices (Ibid: 1336).
Affective heuristics and reactions play a central role in dual-process theories, which hold that external stimuli are emotionally processed in two ways. Epstein (1994: 709-710) started with the differentiation of two cognitive systems, which subsequently were labelled a System 1 and System 2 type reaction on stimuli by Stanivich and West (2000) (Kahneman, 2011: 28-38). A System 1 reaction is an affective, fast, associative, automatic and cognitively undemanding reaction to an event. A System 2 reaction is a relatively slow, rule-based, analytic and cognitive demanding reaction (Stanovich and West, 2000: 659). The conception introduced by Zajonc (1980), that affect precedes judgement and that people often judge in a System 1-type fashion, encouraged several political scientists and psychologists to analyse how affect guides political judgement. Besides the distinction between fast reactions and slow responses, generally two affective systems shape the ways people behave. The research by Marcus et al. (2000) and Cacioppo et al. (1999) theorises that roughly two affective systems form our affective behaviour. The disposition system shapes habitual behaviour whilst the surveillance system scans the
7
environment for novelties and judges whether those novelties pose a negative threat or positive possibility. When people are obstructed to follow a habitual routine the disposition system is activated and people get frustrated, depressed, or angry. When the surveillance system notices a threat, the behavioural immune system (see paragraph 2.3.2) is activated, and people will sense unease, disgust and anxiety (Schaller and Park, 2011; Marcus et al, 2000; Brader and Marcus, 2005; Aarøe et al., 2017: 278-279; Inbar et al, 2012; Rozin et al, 2008). It is important to reiterate that System 1 is not the same as the surveillance system, nor is System 2 an equivalent of the disposition system. Novelties (as in the surveillance system) can be processed through a fast-track System 1 type reaction, or can be processed through a deliberate, cognitive System 2 type reaction. Yet it is observed that both affective systems mostly function by the merit of a System 1 type reaction (Kahneman, 2011; Erisen et al. 2014).
Taber and Lodge (2005; 2006; 2013; and Erisen et al. 2014) did extensive research to identify different affective heuristics people use in their emotional political behaviour. For example, contagion effects shape political judgements. A political stimulus automatically activates an affective reaction that subsequently guides how people think about political subjects. This direct activation is what is called hot cognition (Erisen et al. 2014: 189). Hot cognition leads to different kinds of affective heuristics. Accessibility of political considerations in memory is e.g. biased by affective contagion: affectively congruent information is more easily absorbed than affective incongruent information (Ibid: 190). In another experiment, Taber and Lodge (2013) projected a subliminal image of someone smiling. The people who received the stimuli were asked to list reasons to oppose immigration. People who unconsciously absorbed the experimental stimulus, could list fewer motivations.
Affective heuristics like described above, all fit in the online model of attitude updating (Lodge, et al., 1989; Lodge et al., 1995). The online model of attitude updating entails that pre-existing affect matters to how new stimulus objects are processed. People with pre-existing positive attitudes towards refugees will be more likely to process positive messages on refugees, and vice versa.
The discussed research by Zajonc (1980) and Taber and Lodge (2005; 2006; 2013; and Erisen et al. 2014) focuses on hot cognition; an immediate effect of a political stimulus. How long cognitive effects of affective heuristics last is unclear. This is partially due to an experimental setting where only direct effects can be measured significantly. For example, Kahneman uses experiments where direct physiological reactions like pupil dilation are measured (Kahneman and Beatty, 1966). Thus, the research by Kahneman et al. (1966; 1974; 2003; 2011), Zajonc (1980) and Taber and Lodge (2005; 2006; 2013; and Erisen et al. 2014) all measure direct, temporary reactions. Structural differences between people, on the contrary, are mostly measured through self-reports.
8
2.2 Self-reported psychological differences between conservatives and liberals
The interaction between a political stimulus and physiological subconscious reactions to measure is a novel field of study in political science. That stands in contrast with the extensively studied psychological differences and their correlates with political preferences measured through self-reports (Hibbing, 2014: 300). In the postlude of the WWII, Adorno et al. (1950) sought to analyse which personality traits made people more prone to authoritarian thinking. Nine dimensions like conventionalism, easy obedience to authority and anti-intellectualism were personality traits that placed people on a F-scale; it made them more susceptible to adopt fascist (hence the ‘F’) thinking. The underlying notion, that people are receptive for political ideologies that mirror their individual personality traits had become a wide held, and undisputed notion in political psychology (Napier and Jost, 2008: 579). Yet the framework used by Adorno et al. (1950) is conceptually fuzzy; the difference between psychological traits and actual behaviour is ill-defined. In political psychology, different researches attempted to refine the framework on authoritarian personality, and Altemeyer (1981; 1996; 1998) is one of the main theoreticians in this field. Altemeyer focuses on statements like ‘Women should have to promise to obey their husbands when they get married’ and ‘The established authorities generally turn out to be right about things, while the radicals and protestors are usually just “loud mouths” showing off their ignorance’ (Altemeyer, 1998: 49). The answers on these statements, and quotes in a similar vein compose a measure of Right-Wing Authoritarianism (RWA). In the same tradition, Feldman and Stenner (1997) and Stenner (2005) composed non-political statements that focus on parenting styles. Their research shows that authoritarian parenting correlates with political conservatism.
Contemporary personality psychology agrees upon the existence of five main personality traits: the Big Five. These five traits are conscientiousness, agreeableness, openness to new experiences, extraversion and emotional stability (often called neuroticism) (Mondak et al. 2010, Caprara and Zimbardo, 2004: 585; Gerber et al., 2010). Conservatives in the USA score consistently higher on conscientiousness, and liberals have a higher general level of openness to new experiences (Caprara et al. 1999; Gerber et al. 2010; Mondak and Halperin 2008; Rentfrow et al. 2009). Dutch research shows that people who vote for the populist, right wing Dutch Freedom Party1 (Bakker et al. 2016) score
consistently lower on agreeableness than other voters within the Dutch electorate.
A second psychological scale which correlates with political beliefs is Social Dominance Orientation (SDO). People who firmly belief that a society needs to be organised hierarchically and think it is suited that specific groups dominate in such a hierarchy would score high on the SDO scale. Agreement on statements such as ‘It is unjust to attempt to make all groups in society equal’ indicates
9
an orientation towards social dominance’; and conservatives consistently score higher on the SDO scale (Levin et al., 2002; Sidanius, 1985; Sidanius and Pratto, 1999).
In an influential article, Hibbing et al. (2014) theorise that many differences between liberal and conservative people are closely associated with one psychological and physiological disposition: negativity bias. Conservatives’ inclination to social dominance; their stronger bending towards authoritarianism; and their lower levels of new experiences and agreeableness, are closely related with that one concept of negativity bias. Hibbing et al. (2014) quote Rozin and Royzman (2001: 97) who state that the negativity bias is the phenomenon of ‘negative events [being] more salient, potent, dominant in combinations, and generally efficacious than positive events.’ Negativity bias widely varies between individuals (Norris et al. 2011), and Hibbing et al. (2014) theorise that conservatives, on average, are more susceptible to negative events than liberals. This bias towards negativity can have different faces. Conservatives could be more responsive towards contagion effects that include negative frames (see Taber and Lodge, 2005; 2013; Erisen et al. 2014). But, as discussed, negative experiences occur when the affective systems are disrupted. When a habitual pattern is broken, the dispositional system reacts with anger; when a threat is observed, the surveillance system activates the behavioural immune system and reacts with fear. When these patterns are broken, conservatives could react in a more intense way (Hibbing et al., 2014: 303). Self-reports can be used to test whether these hypotheses are correct, but a more reliable and valid way to measure negative experiences is to measure physiological differences.
2.3 Physiological differences
Using self-reports, two problems are prevalent. First, only the affective reactions that are experienced consciously are measured. Moods, and affective responses outside conscious awareness are not reliable and accurately measured. Second, derived from the first problem, it could very well be that people who are more self-reflective and conscious about their emotions report to be more anxious or angry. Such problems are the springhead of a relatively new field within political psychology. People might lie, exaggerate, or trivialize what they state - but a bodily reaction is hard, not to say impossible, to control. To assess (un)conscious moods and emotions directly; untransformed by a human rationale, physiological responses to (political) stimuli can be measured. There are various ways to measure which political stimuli result in which psychological reactions, and how those reactions differ among specific political groups. Measuring physiological differences also has disadvantages. Only the intensity of an emotion (arousal) and the direction of an emotion (either negative or positive) can be measured physiologically. Distinguishing different negative emotions as anger, jealousy, sadness or fear is not possible. Nonetheless, the measurement of physiological reactions is a new way of validly and reliably studying the subconscious realm of human behaviour and attitude construction.
10 2.3.1 Brain structure and activity
The analysis of a brain structure is commonly done by magnetic resonance imaging (MRI). Using this technique, Kanai et al. (2011) found that political liberalism and conservatism correlate with brain structure. Conservatives generally have a bigger righter amygdale, and liberals on average have a bigger grey matter volume in the anterior congulate cortex (ACC). These differences in brain structure reflect the personality differences between conservatives and liberals (see Mondak et al. 2010). ACC activity and volume is associated with the ability to alter habitual responses in conflict situations (Kanai et al. 2011: 677; Amodio et al. 2007: 1246). More grey matter in the ACC could therefore be the material aspect of more openness to new experiences by liberals. Increased amygdala size and activity is associated with raised arousal. This could be the physiological manifestation of the higher threat and disgust sensitivity experienced by conservatives. However, the research by Kanai et al. (2011) was conducted on a small scale (N = 90) and the participants were college students. Ever since, the study has not been replicated. A promising line of research focuses on neurocognitive activity instead of brain structure. Neurocognitive activity is measured through functional magnetic resonance imaging (fMRI). Amodio et al. (2007) tentatively conclude that liberals have an increased activity in the ACC, measuring the amplitude of event-related potentials. Using the same technique, Rule et al. (2010) showed pictures of actual politicians and analysed the consequential amygdala activity. Participants (N = 28) elicited stronger amygdala responses when they viewed politicians who they appreciated, indicating more positive arousal. Additionally, participants were shown in- and outgroup faces, and amygdala activity was when outgroup faces were depicted. Altogether, studies that analyse the relation between brain activity and political preferences are sparse. The number of participants is consistently small, and it is too soon to derive stringent conclusions from the research that has been conducted. Still, the research regarding neuroimaging and -activity is suggesting that liberals are in general more able to break habitual patterns, thus being more susceptible to new experiences; and that conservatives are more likely to sense disgust and threat.
2.3.2 Behavioural immune system: disgust and threat
Every living organism has an immune system; an automatic biological reaction that protects the body from pathogens that cause diseases. Besides the well known biological immune system, humans have evolutionary created a behavioural immune system: a term coined in social sciences to cover all the automatic protective activities people can perform to protect them from social pathogens (Schaller and Park, 2011; Aarøe et al., 2017: 278-279; Inbar et al, 2012; Rozin et al, 2008). The behavioural immune system not the same as the surveillance system. The surveillance system still judges whether something disrupting is a positive novelty or a negative threat. When the surveillance system judges that a negative threat exists in the form of a social pathogens the behavioural immune system automatically generates a system-1 type reaction to those pathogens. The incorporation of the
11
behavioural immune system in political theory is a novelty. In their meta-analysis, Aarøe et al. (2017) found 14 articles that integrate disgust sensitivity - a component of the behavioural immune system - into the realm of political science, of which 66 percent found a positive correlation. Two forms of automatic reactions towards pathogens that correlate with political preferences have been studied; the automatic reaction of disgust, and of fear (Ibid).
2.3.2.1 Disgust sensitivity
A reaction of disgust is a discouragement to be in contact with dangerous substances (Inbar et al, 2012; Rozin et al, 2008; Smith et al., 2011). Besides experienced disgust to foul objects, the sense of disgust plays a role in making social and moral judgments (Inbar et al, 2012: Chapman et al, 2009). People that are easily disgusted dirty by substances, are also more likely to experience disgust to people who violate presumptive sexual and spiritual rules (Inbar et al, 2012; Horberg et al, 2009: 173). Therefore, experiencing a relative great amount of disgust could shift people to more conservative beliefs, since it is associated with the appraisal of more stringent social norms and less openness to new forms of behaviour (Graham et al, 2009; Haidt and Hersch, 2001; Inbar et al, 2012; Smith et al, 2011). Following this argument, several researchers have sought to proof that disgust sensitivity (DS) is associated with the support of conservative values and parties. Using a large scale (N = 31.045) Inbar et al. (2012) show that there is a positive correlation between conservatism and DS. Especially social conservatism (opposed to economic conservatism, see method section below) is strongly correlated with DS. DS also correlated with a conservative voting intention in the 2008 U.S. presidential election; red states scored significantly higher on DS than blue states. An earlier study (Terrizzi et al. 2010) confirms those findings, noting that DS correlates with conservative values that regard intergroup relations. The researchers find that aversion to contact with homosexual people (r = 0.32) highly correlates with DS (Ibid: 589-590). In a follow-up study, Crawford et al. (2014) show that besides homosexual people, outgroups that advocate the same transformation of traditional rules on sexuality (such as pro-choice activists) are also object of disgust. Brenner and Inbar (2015) found that in the Netherlands, people with higher DS are more likely to vote for the Freedom party, dislike sexual minorities, be anti-immigration and dislike foreign outgroups. In a more recent study Aarøe et al. (2017) examine the relation between DS and opposition to immigration. In four representative samples collected in the U.S and Denmark, the researchers confirm that DS for immigrant outgroups is an important factor that underlies the conscious opposition to immigrants.
2.3.2.2 Threat sensitivity
A second reaction of the behavioural immune system is an automatic physiological response as result of a perceived threat. In psychophysiology, this is better known as the fight-or-flight response; a immediate, rapid physiological set of movements is set in motion to either confront the perceived threat (fight) or to escape it (flight) (Theatero and Penney, 2015). When a threat is perceived and is
12
pressing, this defensive set of movements is made to protect vital organs; this can be an eye blink or an retraction of the head (Oxley et al., 2008; Carver, 2009). All people are equipped with these rapid response movements, but the intensity of a reaction to a threat varies widely between people. The intensity of a defensive reaction is called threat sensitivity (Ibid: 1667). Recent research on the intersection between psychology, biology and political science has questioned whether threat sensitivity is associated with political preferences. Oxley et al. (2008) were first to study whether conservatives are more sensitive to threats. With a small sample (N = 46), and participants who declared that they were either strongly socially liberal or conservative, they found that conservatives had more intense reactions to threatening stimuli. This is in line with the theoretical assumption that conservatives are on average less open to new experiences than liberals, and seek to conserve the status quo, guarding it from exterior threats (Hibbing et al, 2014: Lilienfeld and Latzman, 2014: 318-319). Dodd et al. (2012; 641-645) successfully replicated the study, using a mutually small sample (N = 48). Some years later, Knoll et al. (2015) replicated the studies conducted by Oxley et al. (2008) and Dodd et al. (2012), and using a larger body of participants (N = 200) they failed to replicate the significant results by the former studies. Two suggestions can be derived from this failure to replicate the studies. First, the number of studies is yet too small to deduce clear conclusions. More research has to be conducted, and that notion is the stem of the research done in this paper. Second, threat sensitivity cannot clear differences between broad ideological concepts as liberal and conservative; but can clarify which particular themes in politics draw people to specific ways of thinking. Like the studies on DS that, inter alia, zoom in on aversion towards homosexual people and groups who want to stretch sexual norms, research on threat sensitivity might delineate why people oppose migration influxes (as in the study by Aarøe et al. 2017). A sophisticated example of different research as such, is the study by Arceneaux (2017). Arceneaux (2017) seeks to elucidate why certain conservatives oppose government intervention in the economy, and why other conservatives advocate such government interventions. His study confirms that threat sensitivity is the underlying mechanism; people who are sensitive to threats generally seek to confirm traditional norms, and thus seek to protect the current state of the economy.
2.4 On political ideology, four concepts 2.4.1 Issues and conservatism
In the U.S. and the U.K., political colour is arranged along quite simple, clear dividing lines between liberal and conservative ideology. Much of the research on psychological and physiological differences between groups of people with different ideology is done in the U.S., which has a highly polarised, two-party system (Levendusky et. al., 2008: 750-752). It is therefore not possible to directly copy the semantics of ‘conservatives’ with higher TS to the Netherlands. The Netherlands has a multi-party parliamentary system, which - at the time of writing - holds fifteen different parties in the House
13
of Representatives and the Senate (Parlement en Politiek, 2017). With that many parties, it is more helpful to use two dimensions of political ideology instead of the common distinction between conservatives and liberals (Stoetzer and Zittlau, 2015; Krouwel et al. 2014). The first continuum on which parties can be arranged reaches from economically right to economically left. The second axis runs from social conservatism or nationalism to social progressive or cosmopolite.
Right, left; progressive, conservative; these are categories that serve as umbrella terms (Walcak and Van der Brug, 2017: 247-249). Being economically right-wing necessarily means that you agree on issues which are economically right. Of course, personal attitudes can be mixed. Right-wing economist Friedman (2013) was one of the main advocates of a guaranteed basic income, an idea that one would typically identify as left-wing. Yet different attitudes on multiple issues combined, construct people’s overall ideology, reaching from right to left, and from progressive to conservative. Due to his many other fairly right-wing attitudes, Friedman (2013) still was ideologically right-wing, despite his favouritism of guaranteed income. Combing different attitudes in an equation gives an overall level of right-wing and conservatism.
Yet which attitudes belong to what ideology? Being economically right wing is associated with low levels of government interference, trust in market allocations, low tax levels, low levels of wealth and income distribution and high levels of austerity measures as a fiscal policy to reshape the economy. On the contrary, being economically left wing is associated with government interference in the market system, higher levels of tax levying, wealth and income distribution, and dependence on economic growth as a fiscal policy (Stoetzer and Zittlau, 2015; Davies and Ashford, 1991; 56-59, 80-90, 257-260, 275-280; Krouwel et al, 2014).
The second axis, between social conservatism or nationalism and social progressive or cosmopolite, relates to other issues. Besides nationalistic or patriotic attitudes, social conservatism is associated with anti-immigrant and migration stances, favouring strict law-and-order safety measures, and spending on the military apparatus. Besides cosmopolite ideas, progressive ideology is associated with welcoming migration policies, softer approaches to criminality and investments in sustainability (Stoetzer and Zittlau, 2015; Davies and Ashford, 1991; 56-59, 80-90, 257-260, 275-280; Krouwel et al, 2014).
One issue that has been most defining in the last dozen of years in the Netherlands, is that of migration, integration and multiculturalism (Uitermark, 2010; Sniderman and Hagendoorn, 2009; Elke, 2010). This topic clearly divides parties socially conservative parties from socially progressive parties. The former group generally wants to restrict the influx of foreigners and wants them to adopt Dutch culture. The second group generally wants to accept foreigners and beliefs in a multicultural society where different cultures can exist synchronous. The topic matches the supposed psychological difference between conservative and progressive people quite well. Said exaggeratedly: foreigners and
14
other cultures can threaten the status quo (according to the former group) or enrich and develop culture (according to the latter group).
2.4.2 Partisanship and right-left self-placement
Identifying with a certain ideology, and having certain political attitudes, leads to a vote on, or an identification with, a political party. Party identification is usually called partisanship (Holmberg, 2009). Thus partisanship is different from party membership; partisanship is an identification with a political party that advocates the same values and articulates the same policy preferences as the individual (Ibid).
The placement of people on a conservative - progressive scale is based on attitudes on issues like euthanasia. It is something different than asking people themselves to identify their own ideology. An easy, yet controversial way, is to ask people how ‘right wing’ they are (Piurko et al, 2011: 540-541). The distinction is controversial because the meaning of left and right has historically shifted. During the existence of communist countries, left-wing was closely associated with strict egalitarianism and anti-aristocratism (Ibid, 541). The collapse of communism and the rise of third-way politics in social democracy, altered the way left and right are perceived (Ibid, 541; Green-Pedersen and Van Kersbergen, 2002). As noted above left and right can also be interpreted on two scales (Stoetzer and Zittlau, 2015). However, with a broad brush, people interpret right and left in comparable ways. For example, it will be likely that the PVV, VVD and CDA are commonly considered to be right from the mean, and parties like GL, SP and PvdA left from the mean. Taken altogether, self-placement can be a rough but useful estimation of self-perceived ideology.
2.5 Moderating factors
Many different factors can influence how people react to threats.. Like Arceneaux (2017) states, conservatives that are relatively sensitive to threats are likely to favour government reforms. Federico et al. (2014) make a likewise argument, pointing out that political engagement can be a moderating factor. Political engagement ensures that people absorb political information that is in line with their ideological thinking. Therewith, conservatives equate their ideology with party ideologies. Consequentially, conservatives with a high level of political sophistication will be most likely to favour government retrenchment, and conservatives with low levels of political sophistication will favour government intervention in the economy. This would make politically sophisticated conservatives less prone to threats, and vice versa. This theoretical assumption outlined by Federico et al. (2014) is not yet empirically tested. It is however important to note that levels of political sophistication and education can deeply alter the ways in which people perceive threats.
In this study, another assumption is tested. In public discourse, the suggestion that people with low education working in the lower echelons of the labour market would be more anti-immigrant, and fear
15
them relatively more, is common (In press: Van de Grient and Vries, 2015; De Witt Wijnen and Kas, 2017). This assumption can already be found in Lipsets’(1960: 160) Political Man, where he assumed that people with low education, living in rural areas, doing hard work, and having authoritarian family structures, are more prone to authoritarian, thus highly socially conservative strict politics (Napier and Jost , 2008: 599). Arzheimer (2009) makes a likewise argument, stating that men, with low levels of education, doing manual work, are more likely to vote on extreme right parties, motivated by ‘xenophobic feelings’ (Ibid: 2009; Van der Brug and Fennema, 2003). Hainmueller and Hiscox (2007) likewise argue that anti-migrant attitudes are mainly the result of low levels of education. Scheepers et al. (2002) argue that people in a competitive position in the labour market will adopt anti-migration opinions. In a meta-analysis, Ceobanu and Escandell (2010: 319) state that the relation between high education and welcoming attitudes towards refugees is consistent, but varies between different countries. On the contrary, Citrin et al. (1997) argue that anti-migrant attitudes are mainly associated with the perception of the socio-economic state of the country, and that individual differences like low education do not lead to anti-immigrant attitudes.
Put short, low education or low income could have a significant effect on your experiences fear and negative emotions. Low education or low income can also have a significant effect on people’s attitudes towards foreigners, which would make you more anxious too. The level of negative emotions of conservatives that have low incomes, in conclusion, should be the highest.
3. Research question and hypotheses
Conservative people want to conserve the things they cherish. They value the status quo and its traditions. Therefore, they need to protect the state of things from alteration by new influences and impulses. This, overall, makes conservatives more on guard. They are prepared for threats, and will react with power.
Progressive people attempt to alter the world for the better. The state of things could better be a status quo ante: upholding traditions is preventing improvement. Therefore, liberals seek new influences and impulses that can help lead a country to progress. They are seeking new experiences and welcome them with open arms.
These are, in short and quite embellished, the assumptions that underlie much of the literature that seeks to explain how conservatives differ from liberals or progressives and why they do so. This protective state of being of conservatives should make that conservatives have a more active behavioural immune system (Aarøe et al., 2017: 278-279; Inbar et al, 2012; Rozin et al, 2008). Therefore, conservatives should experience a fight-or-flight reaction more often, which results in the emergence of fear, or of hostility (Theatero and Penney, 2015; Hibbing et al. 2014). That emergence can move people unconsciously, or be experienced consciously (Marcus et al. 2000). As Hibbing et al.
16
(2014) suggest, socially conservatives might not only be more prone to negative impulses, but experience more negative emotions as well (Ibid: 303). An activation of the behavioural immune system, and the physiological symptomatic set of movements, initially happens unconsciously. That is why study I measures people’s threat sensitivity through physiological reactions. Study II subsequently measures emotions in the realm of conscious awareness. Both studies, however, measure two manifestations of the same phenomenon: negative affect. Therefore, the main research question is:
Do conservatives have more negative affect?
Conservatism is measured on four levels. First, one essential ideological dividing issue is chosen; concern about multiculturalism. For decades, migration, the integration of migrants, and multiculturalism are some of the most precarious topics in Dutch politics (Uitermark, 2010; Sniderman and Hagendoorn, 2009; Elke, 2010). It clearly divides parties socially conservative parties from socially progressive parties. The former group generally wants to restrict the influx of foreigners and wants them to adopt Dutch culture. The second group generally wants to accept foreigners and beliefs in a multicultural society where different cultures can exist synchronous. The topic matches the supposed psychological difference between conservative and progressive people quite well. Put exaggeratedly: foreigners and other cultures can threaten the status quo (according to the former group) or enrich and develop culture (according to the latter group). As Arceneaux (2017) and Federico et al. (2014) suggest; it can very well be that individual topics correlate with TS, instead of the broad, encompassing categories of ideology. Thus, the first hypothesis is:
Hypothesis 1: Opposing multiculturalism and the acceptance of foreigners is associated with having more negative affect.
Second, a composition of different issues is made, to measure one’s overall conservatism.. Fourth, conservative partisanship is used.
Hypothesis 2: Conservatism is associated with having more negative affect.
Third, one’s estimation of the personal ideology serves as a measure.
Hypothesis 3: Identifying as right-wing is associated with having more negative affect. The fourth measure is partisanship.
Hypothesis 4: Identifying with right-wing parties is associated with having more negative affect. As suggested by several authors (Arceneaux, 2017; Federico et al, 2014; Arzheimer, 2009; Van der Brug and Fennema, 2003; Hainmueller and Hiscox; 2007; Scheepers et al. 2002; Ceobanu and Escandell, 2010) your personal situation might strongly affect how anxious you are, and what your
17
attitude towards migrants is. People with lower levels of education or lower levels of income might experience more negative emotions due to e.g. employment uncertainty or lower wages. They might oppose the influx of foreigners more often because they pose a direct threat on the labour market. Thus, level of education can have a considerable effect on both anxiety levels and attitudes towards foreigners. This intersection of causes of fear could make that people with lower education, that oppose foreigners, experience the highest levels of negative emotions. Thus, the fifth hypothesis is:
Hypothesis 5: Conservative people with low income have more negative affect.
18 4. Study I Threat sensitivity
4. 1 Materials and Methods 4.1.1 Participants.
In the first half of the academic year of 2016, students at the University of Amsterdam - mainly from the communications department - were asked to participate in a lab experiment. The participants were told that they would participate in a study to analyse the ‘political gut feeling’ (onderbuik) of the Netherlands. They were aware of the purpose of the experiment. The study was approved by the University of Amsterdam ASCOR committee and participants filled in informed consent contracts before they
could partake in the study. Participants cooperated in exchange for ECTS, or 10 euros per hour.
The sample (N = 120) is collected amongst students of the University of Amsterdam, which has a relatively homogeneous public. Consequentially, the sample is not highly diverse nor representative for the Dutch population as a whole. The average age is slightly higher than 25 years old (SD = 7.71, ranging from 19 - 49). 34.4 percent of the participants is highly educated, and another 60.8 percent finished the two highest variants of secondary education (VWO and Havo). 86.1 percent of the respondents is student. In political colour, the sample is socially progressive, and economically diverse (see figure 1).
4.1.2 Procedure
Participants first completed a self-reported survey. Amongst the questions were the usual demographics, along with political preferences, opinions on specific topics like migration reforms and climate change, SDO attitudes, and agreeableness. Subsequently, participants took part in the experimental test phase. The experimental phase included tests for several other studies including physiological responses to political video’s and pictures of politicians. The stimuli that were used to measure threat sensitivity were pictures of a snake, a gun, and two dogs. These were combined with two neutral stimuli. There pictures were randomly in displayed on a laptop in full screen (see figure 2). Each test began with a central fixation cross for 20 seconds, followed by a picture for 20 seconds. Participants were asked to sit as silent as possible during the trial, to minimize the volume of unusable
22% 23% 9% 6% 4% 12% 3% 9% 12% D66 GroenLinks PvdA SP PvdD VVD CDA No vote Other Figure 1. Party preference sample study I
19
data. The experimental phase lasted about 20 minutes, and it took participants on average 40 minutes to complete the whole procedure.
4.1.3 Measures
Participants self-reported their opinion on themes regarding foreigners and income distribution on a seven-point-scale. Left-right self-placement was reported on a ten-point scale. Political knowledge was measured through three questions and the answers to those three questions were combined to create one scale of political knowledge. Using the two-dimensional spatial measurement of political party ideology (Stoetzer and Zittlau, 2015; Krouwel et al. 2014), party preference was recoded on two separate scales. The first scale is economical conservative party support, and this includes support for parties that are regarded right wing on the economic left-right wing scale (D66, CDA, VVD). The second scale reaches from cosmopolite parties (GL, D66, PvdA, PvdD, SP) to more nationalistic, socially conservative parties (PVV, CDA, VVD) (Krouwel et al. 2014). Support for economical and socially conservative parties recoded as a dummy variable. All other dependent variables were additionally recoded, so that they range from 0 to 1.
In order to measure threat sensitivity, participant’s electrodermal responses to threatening photos were recorded. Using electrocariographic (ECG) monitoring, three electrodes were attached to the body to measure heart rate intensity (Soroka and McAdams, 2012; Mulder and Mulder, 1981). Second, electrodes were attached to participants’ index- and ring fingers to measure electrodermal activity (EDA) through skin conductance levels: the extend of sweat production (Arceneaux, 2017; Dodd et al. 2012; Oxley et al. 2008; Knoll et al. 2015; Soroka and McAdams, 2012). Third, electrodes were attached to the frown muscles (corrugators supercilii) to measure negative affect through electromyograpic (EMG) activity (Fodor et al. 2008; Cacioppo et al 1986).
To measure threat sensitivity, the measurement of SCL through EDA is a common approach (Arceneaux, 2017; Knoll et al. 2015). In the present study, the other two measurements are not used. Increased sweating is a symptom of increased arousal, and a sign that the behavioural immune system is active. Because SCL were measured while participants were watching the stimuli, direct affective reactions could be measured.
Participants received four threatening stimuli - two containing a picture of a barking dog, one of a snake, and one of a pointed gun. These images were shown randomly after the participants saw neutral images, a basket and a spoon. The images were on display for 20 seconds. Between the images, a central fixation cross was displayed for 20 seconds (the interstimulus interval: ISI). Each SCL during the threat stimulus is compared with the SCL during the ISI. The average of the four differences in SCL between the threat stimulus and the ISI is the general TS. In the sample, TS ranged from -0.05 to 0.05. The sample mean was -0.0021. This variable was recomputed so that it ranged from 0-10, to
20
make the effects of the independent variables on TS comparable with the other dependent variables in the second study.
4.2 Results
None of the measurements of political conservatism are associated with threat sensitivity in the study sample. Opposition to the acceptance of foreigners in the Netherlands, self-declared right wing ideology, or support for either economical or socially conservative parties do, in the sample, not lead to heightened threat sensitivity. Therewith, the first four hypotheses are not confirmed, nor debunked. The second and third models drop in observations, because not everyone answered the questions sufficiently to include them in these models. The third regression model controls for level of education, gender and age. Solely gender has a significant effect on threat sensitivity; one a 10-point scale, women are significantly less sensitive to threat (p = 0.007, two tailed). These findings conflict with former research with likewise small samples that found a consistent relationship between socially conservative attitudes and threat sensitivity (Arceneaux, 2017; Dodd et al. 2012; Oxley et al. 2008; Smith et al. 2011). On the contrary, the findings are in line with the reproductions performed by Knoll et al. (2015), who fail to find a correlation between socially conservative attitudes and TS. It is important to reiterate that the failed reproduction could also be the result of the low level of representativeness, and especially the small amount of conservative and right wing people in the sample. The consequences of the composition of the sample will be discussed below, in the discussion section of the present study.
21 Table 1. Regression Threat Electrodermal Response
(1) (2) (3)
Opposition acceptance foreigners -0.028 (0.084) -0.012 (0.109) -0.015 (0.105) Conservatism -1.836 (1.194) -1.218 (1.190) Right-wing self-placement -0.053 (1.253) 294 (1.264) Conservative party support -0.600
(0.470) -0.803# (0.460) Political Knowledge 0.020 (0.662) Education -1.474 (0.985) Gender: Female -1.186* (0.439) Age 0.244 (1.490) Constant -0.141 3.261 5.521 N 119 93 93 R2 0.001 0.043 0.152
The dependent variable was recoded so that it ranges from 1-10. All independent variables were recoded so that they range from 0 to 1. Standard errors in parentheses.
22 5. Study II Self-reported negativity
5.1 Materials and Methods 5.1.1 Participants and procedure
Since October 2007, CentERdata, the Institute for Data Collection and Research, yearly collects data in the Netherlands (CentERdata, 2017). Its LISS panel is composed from 4500 households, consisting of 7000 individuals. These respondents are drawn from the population register by Statistics Netherlands, and form a true probability sample. The sample is representative for the Netherlands. Participants monthly answer the survey questions and in return they get a payment. It takes them about 15 to 30 minutes to complete the questions. An essential part of those surveys is the LISS Core Study. This longitudinal study is carried out yearly, and provides the same set of variables since 2007. The core study is comprised of ten sets of questions, sorted by theme, ranging from questions on health, to information about income. For the present study, three of those ten sets are used: the set on personality, background variables and political values. These sets are merged so that differences between individuals could be measured on both emotions, political values and background variables like education. The most complete and recent dataset on personality was taken from November 3, 2014 to December 31, 2014. From the participants (N = 7739), 83.2 percent answered the questions completely (N = 6442). The dataset on political values was chosen to be as near as possible in time. This happened to be the dataset that was comprised end 2015 (from December 7 to December 29). The number of completed surveys was 5962 (a rate of 86.2 percent). The third dataset is on background variables. This set was chosen to match the date of the personality dataset, and was conducted in December 2014. The number of valid responses is 11736. The number of respondents that answered all the questions that were needed to compose the independent variables and dependent variables for the regression models, is 5149 for the first three models, and slightly more (respectively N = 5177, N = 5173 and N = 5201) for the last three models. The number of observations is reduced, because not anyone answered all the questions in the three datasets.
5.1.2 Measures
Hypotheses one, two, three and four state that there is a positive linear relation between levels of negativity and, opposition to an influx of foreigner conservatism, wing self-placement and right-wing partisanship. First, I test for a common latent dimension. All six dependent variables could measure the same concept; or measure negativity in specific, different ways. To test for a possible common latent dimension, a CPA factor analysis is performed. Subsequently the hypotheses are tested. To do so, I employed an ordinary least squares (OLS) regression. The OLS regression is a relatively simple, straightforward test to measure the predictive strength of independent, explanatory variables on the dependent variable, calculating a model with minimised variance. This allows for a
23
clear comparison of the explanatory strength of ideology. An OLS model tests for causal relations, yet theoretically it is hard to unwind how the relation between emotions and ideology is causally constructed. The main purpose of the test is to assess which associations are prevalent, and the OLS regression should give clear insight in that.
5.1.2.1 Dependent variables
To be able to confirm or reject hypothesis one, two, three and four, dependent variables are chosen to give an insight in the emotional state of participants. To give a direct, neutral assessment of perceived emotions, the most direct questions in the datasets were used. Indicate to what extent you feel, right now, that is, at the present moment scared?, was the question to measure one’s level of anxiety (model 1). For model 2 and 3, the same question was posed, using the words afraid and hostile. The fourth model establishes the relation between ideological factors and sadness. In contrast to the first three models, this question along with the fourth and fifth model, does not measure a perceived feeling on the very moment of answering the survey, but a general feeling: I often feel blue. The fifth model estimates ones overall self-confidence: All in all, I am inclined to feel that I am a failure. The sixth model, finally, assesses ones level of happiness: On the whole, how happy would you say you are?
The first three models are chosen to mirror TS in the most direct way. The experience of a threat results in a rapid defensive responsive movement, and the immediate experience of arousal. The flight-or-fight reaction generated by the behavioural immune system, can result in generally two acts that are accompanied with higher arousal: the retreat or the confrontation (Theatero and Penney, 2015). The flight reaction is accompanied with fear (Ibid; Blanchard et al. 2008: 2-3). Semantic equivalents of fear are ‘being scared’ (model 1), or ‘afraid’ (model 2). These two models therefore measure the same theoretical concept. The fight reaction is accompanied with the experience of aggression (model 3) (Theatero and Penney, 2015). Therefore, the first three models estimate the two conscious emotional manifestations of the reaction to a threat; to flight or to fight; to feel scared or to feel hostile.
The last three models measure a more general state of wellbeing. This might not cover the concept of threat sensitivity that precise, but alludes to the notion that a more often active behavioural immune system leads to a bias to negativity in a general sense (see Hibbing et al. 2014). Second, it is likely that higher TS protracts to a general feeling of discomfort. People have limited energy, and energy dedicated to threats, fear, and aggression, might detract energy from comfort, rest, and satisfaction. Third, a different way of reporting emotions (not a present in the moment felt assessment of emotions, but a general evaluation of one’s emotional state) makes the general measurement of negativity more reliable. Therefore, sadness is measures to be felt ‘often’, the feeling to be a failure ‘all in all’, and happiness ‘on the whole’.
24
Taken together, the six dependent variables should give insight in the average emotional state of the respondents. Directly present sensations and overall felt emotions are measured separately. Experienced fear, aggression and happiness can be analysed independently. Yet the variables are also composed to give a general assessment of negativity.
5.1.2.2 Independent variables
The Dutch parliamentary party system does not hold a binary ideological distinction between conservatives and liberals. This stands in contrast with first-past-the-post electoral systems in the U.S. and U.K. which mostly result in one party with a clear conservative, and one party with a clear liberal identity. Therefore several measures of ideology have been used. First, one dividing issue is chosen: that of the acceptance of foreigners in the Netherlands. Second, various issues are together combined into a conservatism scale. Third, a left-right self-assessment is used. Finally, the PVV is used to measure right-wing partisanship.
Opposition to the acceptance of foreigners is chosen for multiple reasons. In Dutch politics, it can be seen as one of the structural, core themes of campaigning in the last decade. Nearly all parties have clear standpoints on migration and multiculturalism. Parties that favour multiculturalism are often opposing patriotism, and vice versa (Uitermark, 2010; Sniderman and Hagendoor., 2009; Elke, 2010). Second, an opinion on multiculturalism is associated with group thinking. People that oppose a multicultural society, can be supposed to have a clear demarcated, heterogeneous idea about Dutch identity; and outsiders might pose a threat to this idea. People that favour multiculturalism will often have a more open, homogeneous idea about ‘being Dutch’, and might experience the influx of outsiders as a welcome novelty. The question measuring this explanatory variable is: ‘People of foreign origin or descent are not accepted in the Netherlands.’
Conservatism is a scale that combines four variables: attitudes on euthanasia, immigrants, and on the role of women in firms and in families. Unfortunately, the LISS panel does not answer questions in climate change attitudes, as study I does. Therefore, different variables are used to measure the concept of conservatism. The first used variable measures attitudes on euthanasia: Where would you place yourself on a scale from 1 to 5, where 1 means that euthanasia should be forbidden and 5 means that euthanasia should be permitted? The second variable measures attitudes on immigrants: Where would you place yourself on a scale of 1 to 5, where 1 means that immigrants can retain their own culture and 5 means that they should adapt entirely? The third and fourth variables measure attitudes towards the role of women in society. It is unnatural for women in firms to have control over men, and The father should earn money, while the mother takes care of the household and the family. All variables were answered on a five-point-scale. The answers on the question on euthanasia were reversed. All the variables were combined into a conservatism measure, and subsequently recoded so that it ranges from 0-1.
25
Thirdly, PVV party vote serves as the second explanatory variable. The PVV is the most right-wing, nationalistic, socially conservative party in the Dutch parliamentary party system (Stoetzer and Zittlau, 2015; Krouwel et al. 2014). Monoculturalism and anti-migration measures belong to its main ideological points (PVV, 2016). It therefore makes a good case, to analyse whether it’s electorate experiences more negativity than other electorates.
Left-right placement is the broadest assessment of ideology. It is asked broadly: ‘Where would you place yourself on the scale below, where 0 means left and 10 means right?’, and does not elaborate on what left or right means to people. It is a form of self-placement. When people would be placed on a left-right scale by experts, placed on a scale based on their opinions on specific topics, the result could be different. Yet simply asking where people place themselves, gives a proper insight in the self-perceived ideology individuals have. Therefore it is a useful explanatory factor for the negative emotions people might perceive. It is however not clear which axis on the two-dimensional spectrum the respondents have in mind (Stoetzer and Zittlau, 2015).
Education, income, age and gender serve as control variables. How well educated you are, how much money you make, how old you are, and the gender you have, could all have independent effects on the level of fear, hostility and happiness.
To test the final hypothesis, the moderation effect of education on levels of negativity ordered by opinion on multiculturalism is analysed. Using independent sample t-tests, the chart in figure three shows whether the average levels of negativity of lower educated people are higher than that of higher educated people, and what effect an anti-multiculturalism stance has on levels of negativity. In those models, opinion on multiculturalism was recoded to be a dummy variable. Respondents who answered neutrally were excluded, people below the mean were recoded to be pro-multiculturalism, and respondents above the mean were recoded to be anti-multiculturalism.
All dependent variables were recoded, so that they range from 1 to 10. All independent variables were recoded, so that they range from 0 to 1. The recoding is the same as in study I, which makes the coefficients and results in the second study directly comparable with the results in the first study.
To perform the independent sample t-test, two final measures were used. First, the same four questions to measure conservatism were used to create another measurement of conservatism. In total, people answered four questions, all on a five-point-scale. Being fully conservative, would mean that people agreed with all the questions, what would result in a total of 20 points. The contrary being true for people that were fully not conservative. These four variables were merged and subsequently recoded. People who scored 10 points were neither conservative, nor not-conservative, and excluded from the test. All people that scored more than 10 points were recoded as being conservative, and vice versa. Income was also recoded to be a dummy variable. What is a high income can of course be object of
