Do unfulfilled political pledges and counterfactual relative deprivation predict populism? : an attitude survey researching the supply-side of populism in the Netherlands

(1)

1

Do unfulfilled Political Pledges and Counterfactual Relative

Depriva-tion predict Populism? An attitude survey researching the

supply-side of populism in the Netherlands

Hendrik Aldenberg 5782813

Begeleider: Dr. Sjoerdje van Heerden

Tweede lezer: Prof. Dr. Wouter van der Brug

(2)

2 Summary:

‘This thesis answers the following question: Are voter’s attitudes on the ful-fillment of pledges, an explanation for populist attitudes in the Dutch elec-torate? To answer this question, the paper introduces counterfactual relative deprivation as a concept to explain the possible relationship between both attitudes. Doing a quantitative analysis using Likert-items, Principal Component Analysis, and regression analysis, the paper shows that there is explorative evidence to conclude that populism is caused by political pledges with an underlying mechanism of counterfactual relative deprivation. The main result is that the attitudes about non-fulfillment of pledges made by politics, create feelings of counterfactual relative deprivation and thus polit-ical discontent within the representative system. Both are good predictors to conclude a possible supply-side explanation of populism. The results find that there is mild evidence that this relation is stronger concerning people in the electorate that have lower educational degrees, lower income and are younger.’

Dankwoord:

Ik dank allen, die bedankt moeten worden. In het bijzonder mijn begeleider, de 2de_{lezer, mijn vader voor zijn} statistische en materiele steun, Tinka Bakker die meegeholpen heeft met het verzamelen van data en Laurens Pollmann die geholpen heeft met nakijken.

(3)

3

1. Introduction

In a television debate surrounding the Dutch parliamentary elections of 2010, somone in the audience asked a politician how he could guarantee that the things he said will actually be done when his party would be installed in government. Briefly reflecting on the question, the politician displayed the difficulty of the issue, by showing a problematic face live on television unable of answering the question at first hand. The politician’s reply -after a short silence- was very modest and nuanced, stating that he could not absolutely account for that, but was doing the best he can to ensure that his promises were realized, if he and his party would be installed in government. Quickly after, he also stated that installation in government means that policy packages had to be exchanged within a coalition and that it was not so sure that all his utterances would directly result in policy. He noted that he and his political party would, for sure, strive for the realization of most of the election program.

The problem for the politician facing the question is twofold. On the one hand, a live debate on television could generate or break down shares of the electorate at once. So making utter-ances or pledges that large parts of the electorate find attractive is of major importance, and doing the opposite could ruin a well set-up election campaign and policy program. Being attractive should be the main goal for contemporary politicians.

On the other hand, the problem of politicians making too attractive pledges during election campaigns could backfire in the long run. A short, attractive political promise in clear language, creates an expectancy that is less likely to be fulfilled in a sphere of competing ar-guments by other politicians, experts, institutions and technocratic policy-making.

Therefore, I think that the pledge is of major importance in politics and important for the understanding of emotions on trust and satisfaction with government. Especially, I think, the pledge could be significant in the research about the continued success of populism in West-ern democracies and the Netherlands in particular. This thesis will investigate, if, and how, political promises could be an explanation for the continued success of populist parties.

1.1 Formulation of the problem; academic relevance

The rise of populism in Western democracies is an extensive and well-researched topic in contemporary political science literature. Many studies try to get a sound explanation of the phenomenon of populism. Many studies try to define populism as a concept. The lit-erature generally agrees about a minimal definition. Populism, as a minimal definition, is the antagonistic relationship between two homogeneous constructs, the pure people against the corrupt elite (Mudde: 2004; Rooduijn & Pauwels: 2011; Rooduijn: 2014). In addition to this definition, the thesis uses Taggart’s notion that the will of the people should be the point of departure for all politics (Taggert: 2000). The problem with more extensive definitional forms of populism is that it is hard to give an explanation of the rise and success in contemporary democracies according to social-economical structures in society. Especially studies that try to give an explanation to the demand side of populism fail to address a workable cross country generalization of the phenomenon. In his article, “The populist Radical Right: A Pathological Normalcy”, Mudde raises the doubt if the limitations in the search for a demand-side explanation on the rise of populism will be solved (Mudde: 2010). Mudde argues that political science literature should intensify the search for a supply-side explanation of populism (Mudde: 2010). In this thesis, the call to search for a supply-side explanation of populism is researched. The supply-side of politics is a form of the political opportunity structure, party

(6)

6 ideologies and strategies (Rydgren: 2004, p.1). The attitudes regarding possible unfulfilled po-litical promises, as shown in the example, are conceived in this thesis as a popo-litical opportunity structure for right-wing populist parties. This is why, next to a minimal definition of populism, the supplement of Taggert, regarding that the will of the people should be the point of departure for all politics is highlighted as a important element of populism.

This thesis tries to do empirical research about the supply-side, in the form of an attitude survey of the functioning of political parties and politicians regarding the feeling of unfulfilled political promises. There are numerous studies that focus on the attitudes of trust and satisfaction with democracy as a breeding ground for right-wing populism (Akkerman e.a: 2014, Doyle: 2011, Ivarsflaten: 2008, Norris: 2005, Fieschi & Heywood: 2004). Akkerman, Mudde and Zaslove highlight that studies about trust and satisfaction with politics are ‘illumi-nating’, but these studies do not directly measure populist attitudes (Akkerman: 2014 p.1325). At the end of the article they write: “… having demonstrated that it is possible to measure populist attitudes in individuals, it will be interesting to see whether populist attitudes correlate with other attitudes.” (Akkerman e.a.: 2014 p.1346). So the general problem to be solved is if we can extend the measurement of populist attitudes and relate it to attitudes on trust and satisfaction, in the form of political pledges.

In this thesis populist attitudes are measured directly with the use of Likert-items, and will be linked with voter’s attitudes on political promises. The sought relationship, is not yet made in the literature and could be explanatory for the continued success of populist parties in Dutch society and Western democracies.

Next to the main relationship between attitudes about unfulfilled political promises and populist attitudes, the second goal of this thesis is to explain the underlying mechanism, or phenomenon, why this relationship could exist in a democratic society by means of generalization. The history of the sciences have a strong tendency to investigate certain cau-salities or correlations, but as David Hume famously argued, the ‘mechanism’ underneath a correlation or causation is of general importance for understanding. This thesis argues that the underlying phenomenon, of why attitudes on political promises could be explanatory for populist attitudes, has a strong tendency towards counterfactual thinking. The underlying mechanism will be ‘framed’ as counterfactual relative deprivation and will be explained in the theory section.

1.2 Main research question

Resulting from the problem formulated by Akkerman et al., the main research question is the research for a possible correlation between attitudes on political promises and populist atti-tudes:

RQ: Are voter’s attitudes on the fulfillment of pledges able to explain populist attitudes in the Dutch electorate?

A second question central in this thesis is how we can understand this possible correlation. Therefore, counterfactual relative deprivation is introduced as a main concept in the theoret-ical framework. The second question is the following, is counterfactual relative deprivation an explanation for populist attitudes within the Dutch electorate. Next to this three fundamental concepts of populist attitudes, political pledges, and counterfactual relative deprivation, the thesis tries to look at income, age and education as possible moderators for the correlation between political pledges and populist attitudes.

(7)

7 The general model for research based on a dependent and independent variable is the fol-lowing:

(Political pledges + Relative Deprivation) -> Populist attitudes

1.2.1 Outline

In the coming chapter I will give a theoretical outline to the research question, and define concepts and relationships, than the operationalizing of the positivist method for data collection is explained. A questionnaire was made based on Likert-items measuring attitudes on political promises, populist attitudes and counterfactual relative deprivation. Thereafter the results will be presented with the methods of Principal Component Analysis, Linear Re-gression Analysis and Statistical Mediation Analysis. The last chapters will be a conclusion of the results and a discussion on the findings.

1.2.2 Why the Netherlands?

The choice of doing research among Dutch citizens in the Netherlands concerning the topic at hand is reasonable, because the Netherlands has a form of Populism that is ‘here to stay’(Akkerman: 2014). The Netherlands was famous for its technocratic style of deliberative politics during the 1990s, the so-called ‘poldermodel’. After the rise and death of the Dutch populist Pim Fortuyn, the Netherlands are widely viewed as an excellent multi-party democ-racy for researching populism. The political Party for Freedom (PVV) in the Netherlands represents populist thought in parliament, with a strong position concerning anti-immigration, anti-leftish establishment and anti-European Union.

2. Theoretical framework

In this theoretical framework, the three main concepts of this thesis will be explained. First the concept of populism is explained, second the concept of political pledges will be elabo-rated, and last the underlying mechanism of counterfactual relative deprivation is genealogi-cally set-up. The chapter concludes with a proposed hypothesis based on the main concepts.

2.1 Main concepts

2.1.1 Populism

Populism is a widely contested concept with many theoretical debates. The controversy has recently cooled down to the moderate acceptance of a minimal definition of populism. Pop-ulism, as a minimal definition, is the antagonistic relationship between two homogeneous constructs, the pure people against the corrupt elite (Mudde: 2004; Rooduijn & Pauwels: 2011; Rooduijn 2014). In addition to this definition, I use Taggart’s notion that the will of the people should be the point of departure for all politics (Taggert: 2000). This definition proves work-able across spatiotemporal problems, such as the differences between populism in Latin Amerika and Europe, the possibility of being either far-left or extreme-right or the question if it is an ideology or a strategy. Important within the minimal definition is the notion that democratic politics should be an expression of the ‘will’ of the people (Mudde: 2007 p.23). Regarding the will of the people, the definition highlights an imaginary boundary between those who govern and those who have to obey. This boundary is a central theme regarding the justification of democratic politics since Thomas Hobbes famous book Leviathan. In de-mocracy theory, obedience and governing power is justified by the central arguments of jus-tice, legitimacy, obligation, and representation that settled the discussion about the possibility

(8)

8 of government ruling. With the rise of populism, however, it appears that there is, again, a gap between the pure people that have to obey and the justified rule of governing power. The above is why in this thesis, it is fruitful to position political pledges as having a central relation with populism. I propose that populism and its appeal to the ‘will’ of the people, that should be the point of departure for all politics, is a natural reaction against the current status of the procedural legitimacy of party democracies in Western Europe. This, because promises made by politicians to attract people, are less and less clearly resulting in policy, people feel that politics is not procedurally legitimate in itself and try to regain the procedural legitimacy by voting for populist parties. This results from the way promises, as a simple utterance, has a problematic relationship with the currents of mediatized landscape of politics. Pledges by contemporary parties, I propose, result in counterfactual thinking about party politics because more and more, promises are not fulfilled and are only left as ‘conditionals’.

There is much debate about whether populism is a protest vote against ruling elites, and multiple researcher justify the findings that voters are actually ideologically motivated on left-right scale, to vote for left-right-wing populism (Van der Brug, et al.: 2000). However, researching supply-side opportunities within this thesis, I will try to focus on a political opportunity struc-ture without amplifying ideological motives and base it solely on the proposed minimal defi-nition of populism. In the discussion section, I will return to this dispute according to the findings presented below.

2.1.2 Political pledges

The second central concept within this thesis are political promises. The main body of litera-ture about political pledges is the work of Robert Thompson.

The political pledge is at the same time the utterance by which politicians try to gain their share of the electorate, but also the pledge seems to be a mechanism to represent policy mak-ing. Thereby pledges are an important source of accountability and trust in politics (Thomson: 2001). The pledge according to Thomson e.a. is important in the election program to policy linkage (Thomson: 2007, 2011). It is a reflection of salience of issues in politics and the method of differentiation of other parties in election campaigns. Furthermore, the pledge has a time scope element regarding policy making vis a vis the electorate. If the pledge is fulfilled people feel comfortable in the sense of a politician keeps one’s promise, something that is an utter necessity in the complex of modern policy making. The pledge is central to the setting up of an ‘expectancy’, in the complexity of governance and policy making. The pledge is also a unifying party political concept in the media by which media centralizes the debates between political parties in the form of polarizing pledges (Thomson: 2008). The pledge is thereby an utterance of a collective preference, and a barrier of party identification, which is followed by the duty to influence policy and thus represent the electorate. Research shows that pledges are less likely to be fulfilled in policy making processes, especially in coalition governments. The coalition government argument is a major inspiration for the proposed model because it highlights that coalitions are not sacred within the topics of trust and satisfaction.

The political promise is stronger, in presence and form, in an age where media, shortage of time, and simplicity in the form of clearness are main goals for party leaders. The simple utterance is best acknowledged in the pre-election debates where politicians do their utter best to gain votes by being attractive towards a demos that is less motivated by participating in politics. Similarly, as Colin Crouch states the politicians are trying ‘to gain votes at the maxi-mum level of minimal participation’ (Crouch: 2004 p.112). The media are debit to what Ber-nard Manin argues as a ‘party democracy’ that is evolved towards an ‘audience democracy’ (Manin: 1997). The media is central to the democratic deficit in politics, Papadopoulos argues, because the media create an ever-widening gap between front-stage and backstage politics (Papadopoulos: 2013 p.10) Media are in a search for news value at the front-stage, mostly

(9)

9 partisan politics, and do not offer an adequate description of the increasingly complex policy-making, a significant part that is taking place at the back-stage of politics which is often involving non-elected actors. So media and their search for news value, amplify the political pledge in contrast to the complex of policy making. Thereby favoring personalization in the form of charismatic leadership and reinforcing trends in the form of polarizing issues instead of the real of policy formation.

Furthermore, political promises become the knowledge carrier of trust and satisfaction with democratic government. There are problems concerning demand-side models, that they are not able to highlight short-time shifts in the electorate (Koopmans & Muis: 2009) In this thesis, I propose, that political promises can also account for subtle long term change in the supply-side of politics. This is done by setting political promises as the main correlate of pop-ulist attitudes for quantitative research. This is the main motivation to oppose both attitudes for research.

2.1.3 ‘Framing’ political discontent as counterfactual relative deprivation

The third and final central concept in this thesis is counterfactual relative deprivation. Coun-terfactuals are used in philosophy to reason about possible worlds, most notably found within the work of David Lewis (Lewis: 1973, 2000). In psychology counterfactuals also draw atten-tion to social grievances, and thus can be used for political grievances (Roese: 1997).

A counterfactual relates to, or is expressing, what has not happened or is not the case. The forgoing about promises can be connected with counterfactual reasoning because people of-ten make false conditionals about political promises and reason that the exact promise did not happen. So counterfactual theory about political promises is something like: ‘If policy ‘a’ is promised and realized, then representation functions as it should’, Policy ‘a’ is not realized exactly, so politics does not function as it should. As Thomson shows, there is a very thin line between the promise to policy linkage, one that is seldomly concretely realized.What hap-pened as an election campaign does not always result in exact policy because of the complexity of modern policy making. The promise ‘ought’ to be realized but does not result as an out-come of contemporary representative politics.

This discrepancy between what ‘ought’ to be and was ‘is’ was a central construct in political theory in the prediction of political disobedience and revolution against government. The concept was coined relative deprivation (Gurr: 1970). Relative, because what ought to be the case is a subjective emotional standard. I propose that this is a fundamental construct in the supply-side of multiparty democracies and a suitable explanation for populist attitudes. The main reason to re-introduce this theory is because it makes predictions on the basis of subjective emotions instead of, for example, rational economic behavior. With the volatility of voters and other less predictable pacey shifts in the electorate, it could be useful to turn towards certain theories that make objective what people feel.

In the coming subchapter a genealogy of the concept is presented, to justify its usefulness This mainly becaise in political science, it is set aside as a social comparison theory. If you are solely interested in de operationalization of the concept for quantitative research, it is possible to skip the chapter, because thereafter the operationalization for quantitative research is given.

(10)

10 2.1.3.1 A genealogy of relative deprivation; counterfactualism as its operator

In this chapter it is tried to show a renewd importance of relative deprivation theory in politics. Counterfactual relative deprivation will be the central concept within this theoretical frame-work. First, I will explain the histories of the conceptual use in social science literature using genealogy, and then connect it to the importance of the pledge in representational politics as described by Robert Thomson in the concluding chapter. It is important to highlight the history of relative deprivation theory because it is often discredited as being a social comparison theory, though widely used as such, it has a much more interestind history as a concept about political discontent.

In a classic in political science called ‘Why men Rebel,' Ted Robert Gurr launches a theory why men turn into political discontent and revolutions. The main hypothesis of Gurr is that people go into revolution when feelings of political discontent are high due to relative deprivation. Relative deprivation according to Gurr is defined “as a perceived discrepancy between men’s value expectations and their value capabilities” (Gurr 1970: p.13). That is, according to Gurr, a “discrepancy of what ‘ought’ to be and what ‘is’” (Gurr 1970: p.23). If the expectations are rising, and the capa-bilities stay the same or decrease then the discrepancy between what ‘ought’ to be and what ‘is’ rises. This leads, according to Gurr to frustration and discontent and in particular to po-litical disobedience. Relative deprivation is thereby an actor central theory, not a rational actor central theory but an individual psychological actor theory, about desires and emotions. It is also relative because it centers on the ‘ought to be’ value of an individual according to his standards of expectancy. To contrast, real deprivation centers around being deprived of means of absolute standards, for example, food shortage, poverty etcetera.

Among the most important values of relative deprivation Gurr categorizes welfare values, power values and interpersonal values (Gurr 1970: p.25). Gurr highlights that relative depri-vation is especially the case, where individuals and collectives are within situations of complex social and economic interdependencies (Gurr 1970: p.180). Within this spectrum, Gurr highlights the importance that political discontent is not the function of the separations between what men ‘want’ and what they ‘have’, but between what they ‘want’ and what think they are capable of attaining.

Gurr uses the concept of relative deprivation in great detail to explain how an increased level of relative deprivation leads to political violence. This model on revolutions was heavily crit-icized with success in the years to follow with strong methodological implications underneath its criticism, instead of concrete political criticism. The importance of the criticism targeted the possible predictability of the concept to be fluid from case to case, thereby to doubt its absolute rigor as a sound theory on a methodological basis. The fundamental mechanism between expectancies and outcomes is still useful for politics.

The first notion of relative deprivation theory was launched by Stouffer et.al. in the research of satisfaction by African American soldiers in the US army (Stouffer et al.: 1949). Stouffer researched the psychological situation of African American soldiers that compared themselves with white Anglo-Saxon soldiers. Stouffer operationalized the comparison with the help of the expectancies of the Afro-Americans being the same as their white counterparts but showed that there value capabilities where lagging behind. The discontent within the army therefore was operationalized by Stouffer as one of relative deprivation theory. The theoretical framework of the individual comparing to others, without moving into indebt qualitative and individual subjective analysis, was the beautiful result of relative deprivation theory.

Gurr transposed this analytical rigor of individual psychology and used it for a societal analy-sis. From Tocqueville, we know that grievances patiently endured become intolerable once a brighter future seems possible. From this perspective, Gurr developed the insight that political instability only results from relative deprivation when combined with a belief that conditions

(11)

11 are worse than they could and should be. Most important, according to Gurr, are situations arrived at from a period of improvement, that is followed by a decline in the ability of the regime to meet rising demands. Situations like these create a dangerous gap between expecta-tions in situaexpecta-tions of continued improvement and the reality of a sudden decline. Essential in the investigation on revolutions is that they are most likely to occur when a prolonged period of economic and social development is followed by a short period of sharp reversal. This reversal could come from a decline in material capabilities of the state, but the fundamental building blocks of these societal dynamics lie within the realm of the social psychological. Fundamental according to Gurr and Stouffer is how people perceive their situations, which is more important than their actual situation. The perceived situation is therefore not a mate-rial situation but one on the level of emotions, hence relative or subjective. The fundamental double edginess of this is that, although emotions are foundational in the perceived short-comings, it is also objectivist in terms of operationalization. In this manner, it proved valuable for behavioral psychologists measuring true feelings of justice, equality, trust, and disad-vantage.

Gurr’s thought came at a time that the rationality of actors became an important factor within economic studies. The idea of the rational actor in economics was expanded in the 60s to the realm of the social sciences. A foundational role for rationality seemed to be the perfect way to make functional generalization and predictions about society on a structural level. Emo-tions and perceived shortcoming are not desirable as functional starting points for inference because it is not a reliable antecedent for theoretical causality. Emotions are vivid, not steady, rationality instead came with a steady antecedent model for theoretical causality for marco-economics and social science. The grand age of the rational individual and its functional the-ories was born. It appeared to be the fundamental criticism of relative deprivation thereby reducing its imporatance to social psychology.

So, despite the central introduction of the theory of Gurr in this chapter, relative deprivation showed many shortcomings that did not make it a useful concept in political theory and po-litical behavior studies. A theory has to be explanatory by almost every occasion, and relative deprivation was not possible to account for this within the political domain due to its strong individualist psychology instead of rationality. Strangly enough it proved objectively valuable for social psychology.

Despite its critique, the theory of relative deprivation has considerable success in the expla-nation of the start of the French Revolution. In particular the starting phase of the peasant revolution in France against the aristocracy was caused by relative deprivation. In the revolu-tions debate, an important examples are found in the last days of the Weimar Republic and the Russian Revolution. Relative. Relative deprivation was coined by Gurr en Skocpol as a notable factor of post-predicting the revolutionary change for these revolutions, but as Charles Tilly argued, relative deprivation, although it has some effects, was by far not the most relevant one (Tilly: 1778). What is more successful for political theory, is the organizational bias that comes from relative deprivation. Why do so many poor people that experience high levels of relative deprivation do not account for a revolutionary change in their structural condition seemed to be a legit political theoretical question. The logical inference was that those who organize themselves are the main rationale for research. Political theory did not need a psychological individual starting point, it needs mobilized actors with rational goals. Organizations with political aims stay within the realm of the rational individual, thereby being an excerpt of the logical positivist currents of social sciences that can use rationality as its modus operandi. Although relative deprivation has a notable explanatory force on the individual psychological level and its influence on society, namely that of a suffering condition and a ‘will’ to change ultimately the political situation, the notions of collective action and

(12)

12 mobilization of resources and organizational- and leader-success is far more important (Tilly 1987: p.). Those who are organized behave rational according to their political aims and goals. So what relative deprivation lacks is, that it is not accounting for the centrifugal congruence of linguistic moments of people that experience relative deprivation. If these socio-lingual moments of social mobilization and organization of social change occur, the critique aims at the relative deprivation theory, then revolutionary tendencies will occur. Due to this argument relative deprivation was side-stepped within political science, because you could do research on the mobilized organizations that are willing and, most important, capable and able of battling for the change of power configurations. Organization with political aims are sci-entifically far more pragmatic for operationalization in comparison to psychological concepts that create unreliable antecedents for behaviorist research.

Concluding from the above, the whole psychological individual level confuses more than it standardizes for the political research spectrum, while certain emotions are feasible together and create certain political behavior, others having merely the same emotions stay politically unmoved at all, hence there is no explanatory force in emotions for political science. For this particular reason relative deprivation theory was quite arguably sidestepped by political science in favor of resource mobilization theory (Gurney&Tierney: 1982: p33).

Gurr himself also highlights on this methodological change, in the third most frequent cited political science paper of Sage Journals. The paper Gurr wrote in 1993 with the title “Why Minorities Rebel: A Global Analysis of Communal Mobilization and Conflict Since 1945“, [notice the preceding argument in the title], Gurr highlights that relative deprivation theory has a competing and preferred theory of group and resources mobilization theory as the most important independent factor for political upheaval and change. Can we thereby refute relative deprivation theory, as political scientific history has done?

For this thesis, I will argue that the original concept by Gurr is still very useful in research about contemporary Western democracies and political discontent in particular. The main reason is twofold. First, the strong positivist and rationalist tendencies of political science have been weakened by ongoing ontological debates at the end of the 20th century, and

second, political resource mobilization is not per se the main mobilization factor of the demos after the so-called ‘end of history’ and the problems of 21th century democracies. I think that relative deprivation theory and emotional behaviorism can have great explanatory power for current research on electorates and the supply-side and demand-side models of politics. To explain why relative deprivation does matter for political science, I will first highlight that the concept is still widely used within social psychology. As Stouffer did, social psychology uses relative deprivation theory to make analytical, the discrepancy between groups their re-sults and their expectancies vis a vis other groups. After Stouffer, Merton and Kitt expanded the idea within reference group frameworks. From the work of Merton and Kitt onwards a whole body of theory on relative deprivation within social psychology emerged that focused mainly on social comparisons and group comparisons (Pettigrew in Walker & Smith: 2002, p.352). Especially the study of Runciman was a notable departure where he coined the term fraternal relative deprivation (Runciman: 1972). The work of Runciman was especially im-portant because it coupled relative deprivation with attribution theory. Attribution theory, is a theory in psychology that focuses on how people judge and evaluate others in comparison to themselves (Heider: 1958; Kelly: 1967; Weiner: 1986). With the help of attribution theory and the fraternalistic relative deprivation concept, there was an analytical model that could make disadvantaged comparisons similar to the work of Stouffer, but then for groups within a civil society or society as a whole.

(13)

13 Runciman shows that relative deprivation was the fundamental theory in reasoning about social comparison, because it makes analytical how comparison can be done without subjec-tive qualitasubjec-tive research on in debt feelings. This, especially in the relationship with equity theory and distributive justice theory (Runciman: 1972; Pettigrew: in Walker & Smith 2002). Much research has been done on this subject in the course of the 20th century, making relative

deprivation the central social comparison theory, with topics including racism, health care, minorities, intergroup relations in democracy, among many, many others. What is particularly useful in the use of relative deprivation theory, is that it can side step problems within the ontology of sociology. Relative deprivation links the micro-individual level of analysis with the meso- and macro group levels of scientific research purely based on measuring discrep-ancies. Hereby relative deprivation mediates the effects of broad social variables on the indi-vidual and back to marcro-perspectives hence being a meso-level concept (Pettigrew, in Walker & Smith 2002: p.360).

From the above onwards relative deprivation is not only a concept for social comparison, it also became a concept that has interesting things to say about procedural justice (Tyler & Lind: 2002; Pettigrew: 2002, p.358). For example, high relative deprivation among a certain group in society could cause political discontent on the basis of equity and justice, not on subjective textual criteria of indebt feelings, but based on legitimate claims made by group comparisons, with an appeal to overarching institutions of the state. It returned, again, to its system justificatory qualities.

Most sociologists and psychologists who have studied relative deprivation, have focused on one of the particular types of social comparison standards described above, namely, the outcomes of groups vis a vis other people and groups, but Olson and Roese argue that this use is actually not capturing a minimal definition (Olson & Roese: 2002, p.265). Other people’s outcomes are just one of many standards with which outcomes could potentially be compared. For example, outcomes could be compared with past outcomes, such as the work of Gurr showed, whereby Gurr predicted revolutionary behavior.

Olson and Roese propose a possible third standard for the expectancy theorem of relative deprivation in relation to the duality of the social comparison theory or the spatiotemporal model. Their third way is not an absolute third way they acknowledge themselves, but a way of subsuming the different types of relative deprivation theory into one minimal component. They argue that the expectancy, not on the basis of the temporal degradation or the comparison to other social groups, is the main feature, instead there are many ways of operationalizing the gap between expectancies and outcomes. Their proposal is that relative deprivation theory could be made with the use of expectancies in relation to counterfactual comparisons of any means.

Olson and Roese argue that the body of theory on counterfactual thinking is very comparable to the operationalization of relative deprivation by Gurr. That is the relative deprivation de-fined “as a perceived discrepancy between men’s value expectations and their value capabilities” (Gurr 1970: p.13). That is according to Gurr a “discrepancy of what ‘ought’ to be and what ‘is’” (Gurr 1970: p.23). Counterfactual thinking, according to Olson and Roese, can subsume different uses of relative deprivation such as; normative standards as for example social justice (Runciman: 1966) or gender comparison (Crosby et al. in Walker & Smith: 2002) race comparisons (Duckitt in Walker & Smith: 2002), economical comparisons, or temporal comparison (Gurr: 1970). These uses of relative deprivation can be subsumed because they all use relative deprivation theory according to the way of the discrepancy about what ‘ought’ and what ‘is’ (Walker in Walker & Smith: 2002, p.6)

Olson and Roese state that this is the scientifically necessary and the main critical use of rela-tive deprivation theory (Olson & Roese: 2002 p.266). Counterfactual comparison is mainly

(14)

14 the usage of evaluating, with negative emotions, how things are turning out not according to their expectancies. Hereby I think the theoretical path for a renewed usage of relative depri-vation is possible for political theory.

2.1.3.2 Counterfactual relative deprivation as a operationalized concept

In this thesis, I define counterfactual relative deprivation as the emotion out of a counterfac-tual comparison of what ‘ought’ to and what ‘is’. So counterfaccounterfac-tual theory about political promises is something like: ‘If policy ‘a’ is promised and realized, then representation functions as it should’, Policy ‘a’ is not realized exactly, so politics does not function as it should.

As Thomson shows, there is a very thin line between the promise to policy linkage, one that is seldomly concretely realized. Accordingly people experience political behaviorism according to emotions regarding perceived shortcomings. What ‘ought’ to be in representa-tional politics is the promise, or promises someone voted for. The political pledge in Western multi-party democracies represent feelings of concrete expectancies, whereby people ‘ought’ to expect their fulfillment’ because that seems, for some, the sole requirement of procedural legitimacy of contemporary representative democracy.

This could be an explanation for the success of populism because it can merge attitudes on trust and satisfaction with the minimal notion of the ‘will of the people’ versus government. Thereby it connects populism with the notions of justice, equality, trust, and feelings of dis-advantage. However it does not apply to the standards of objectivity that these concepts entail in political theory. To measure equality, for example, could be done objectively, but perceived equality could be of a more fundamental importance. Counterfactual relative deprivation is introduced because it could be able to make certain predictions about that what is relative or subjective.

Especially issue-salience, the mediatization of politics, and the discrepancy between front-stage and back-front-stage politics, amplify this model of the political pledge as a source for feelings of counterfactual relative deprivation in representational politics. This because promises dis-tort the relation between what ‘ought’ to be and what ‘is’, because there is no direct promise to policy linkage as Thompson argues. In the model I propose, people feel relatively deprived in the expectation they have on procedural democracy, based on pledges made by politicians as a possible counterfactual way of reasoning.

2.2 Main hypothesis for quantitative research

Returning to the main question of this thesis, ‘Are voter’s attitudes on the fulfillment of pledges able to explain populist attitudes in the Dutch electorate?’, I would like to propose the following main hypothesis for quantitative analysis:

H1: If voters have an attitude that politicians make unfulfilled pledges, then people score high on populist attitudes

As a way of understanding this explanatory model of populist attitudes, I propose that this made understandable by the phenomenon of counterfactual relative deprivation. So the sec-ond main hypothesis in this thesis is:

H2: If voters have feelings of counterfactual relative deprivation, then voters favor populist attitudes With these hypotheses proposed I will operationalize the three main concept in a question-naire using Liker-items to measure attitudes.

(15)

15 The general model for research based on a dependent and independent variable is the fol-lowing:

(16)

16

3. Method and Data Collection

I hope foregoing raised interest for an empirical study. In this chapter, I will describe the setup of a questionnaire that was made to collect data for answering my main research question. Are voter’s attitudes on the fulfillment of pledges able to explain populist attitudes in the Dutch electorate? The proposed theoretical answer in the theory section leads to the following hypothesis. The hypothesis results in the following main model for investigation with an in-dependent and in-dependent variable for correlation:

(Political pledges + Relative Deprivation) -> Populist attitudes

The questionnaire is inspired by the article of Agnes Akkerman, Cas Mudde and Andrej Zaslove “How Populist Are the People? Measuring Populist Attitudes in Voters”. Recalling the formulated problem: “… having demonstrated that it is possible to measure populist attitudes in individuals, it will be interesting to see whether populist attitudes correlate with other attitudes.” (Akkerman e.a.: 2013 p.47) The questionnaire will measure if people’s attitudes concerning the fulfillment of pledges by politicians and attitudes concerning relative deprivation will correlate with pop-ulist attitudes. First I will operationalize the pledges variable, second the counterfactual rela-tive deprivation variable, and last the populist variable.

3.1 Method

The attitudes will be measured with Likert-items. Likert-items are item-response questions with ordinal answers on a 5 point scale that are widely used in quantitative psychological and sociological research. The scale goes from strongly disagree, Disagree, Neutral, Agree, Strongly agree. Likert-items are generally an understandable and accessible survey method, with good possibilities for statistical analysis. The items will be used to compute Likert-scales using the statistical method of Principal Component Analysis (PCA). All variables will be formulated by at least 4 items so we have enough possibilities to build valid Likert-scales in terms of construct validity (Jolliffe: 2014). With the PCA analysis, variables will be computed for researching the possibilities of a correlation. I will use linear regression analysis based on PCA computed variables to show variance (Jolliffe: 1982, Draper & Smith: 1998). Regression is able to show possible anomalies in the data, and thus preferable to Pearson correlation. Pearson correlation is used to check for possible problems with parametric sta-tistics.

3.1.1 Explorative Principal Component Analysis

The exact questionnaire (in Dutch) can be found in the appendices no.2. I used an extensive number of Likert-items on the concept of counterfactual relative deprivation because partly this is an explorative research. Therefore, it will result in the fact that some items do not correlate according to the proposed concept and in exploratorive research this should not be a problem. These will be removed for the analysis. Normally when working with established concepts it is possible to apply prove their existence. Now it is also meant to prove its own existence.

3.2 The questionnaire, Operationalization of the theoretical framework

3.2.1 General questions

The questionnaire starts with general questions concerning age, education, income, and the respondents vote at the last national elections. The respondents were given the opportunity

(17)

17 to fill in either ‘Blanco’ or ‘did not vote’. Both income and education are multiple choice questions. Income was categorized according to minimal wage, modal wage and a high wage, with two in between possibilities. Education was measured according to obtained degrees in the Dutch educational system.

Table 1 General questions

• 1.1 What is your year of birth?

• 1.2 What is your highest educational degree • 1.3 What is your net income?

• 1.4 Did you vote at the last national elections o What party did you vote for?

3.2.2 Political promises

The questions in part two of the questionnaire are concerned with measuring respondents’ attitudes on political pledges. All questions where fitted in a giant table as Likert-items. I choose to disentangle political pledges into four questions about political pledges. Hereby I hope to grasp different formats of the political pledge, such as party programs and television debates. They also have a hierarchy in complexity what could be used for Mokken-scale anal-ysis.

Table 2 Items measuring Political promises

• 2.1 Politicians do what they promise

• 2.2 The political party which I voted for do what they promise • 2.3 Promises in party programs will be fulfilled

• 2.4 Promises in pre-election television debates will be realized

3.2.3 Counterfactual relative deprivation

Part 3 of the questionnaire is intended to measure attitudes of relative deprivation. I choose to ask specific positive and negative questions concerning relative deprivation to check whether respondents are honest and to remove them if possible. Also a few more specific concepts were used that are highly prevalent in relative deprivation literature, such as trust, justice, fairness, feeling of disadvantage and in-group out-group disadvantage. The reason for the abundance of questions within part three is that measuring relative deprivation concerning representational politics is not done before. By asking a high number of questions I can be explorative in the possibilities for measuring an overarching concept with PCA analysis. Off course this could be a problem concerning construct validity. Construct validity is mostly solved by embeddedness in other literature. This also adds to the motivation to add specific concepts that are highly prevalent in the relative deprivation literature, such as trust, equality, justice, and disadvantage.

Table 3 Items measuring Counterfactual relative deprivation

• 3.1 My vote gave me what I expected and that is positive for me and my country • 3.2 If I see or hear a politician, I know what I can expect

• 3.3 If I read a party program I know what I can expect

• 3.4 By following the election campaign on television, national politics resulted in what I expected

(18)

18 • 3.5 If a political party joins the government, I know what I can expect from them • 3.6 If a political party joins the opposition, I know what I can expect from them • 3.7 Politics always decide differently then they promise, which is frustrating • 3.8 If political decisions are different from their promises, then I feel

disadvan-taged

• 3.9 If political decisions are different from their promises, then politics does not function as it should

• 3.10 Political pledges for other are more frequently realized then for me • 3.11 If policy is different than they promise, then politician cannot be trusted • 3.12 If I do not get what I expect from my vote, then politics is not just • 3.13 If I do not get what I expect from my vote, then politics cannot be trusted • 3.14 If everybody gets what he wants, then everybody is equal

• 3.15 If policy is different than the promise I feel disadvantaged

• 3.16 Politics is complicated, therefore decisions sometimes do not result as I expected and that is no problem

It must be stressed that there is no intention that all questions must be used to form a Likert-scale. I was not sure how to operationalize relative deprivation so I choose to ask a little more and to do an explorative PCA analysis to build insight in conducting future research. With the use of the PCA analysis I paid attention to the textual content of the clustered Likert-items.

3.2.4 Populism

In part 4 of the questionnaire populist attitudes are operationalized. I decided to split the operationalization into two different forms of populism. First, I operationalize according to the minimal definition of populism with the addition from Paul Taggert described above, these are questions 4.1, 4.2, 4.3, 4.7, 4.8, 4.11. Second I operationalized populism according to the themes of the Dutch right-wing Populist Party, this is represented in questions 4.4, 4.5, 4.6 and 4.9.

Table 4 Items measuring Populism

• 4.1 Politicians and governors have greater advantages out of politics than ordi-nary citizens

• 4.2 If citizens could determine politics, then politics would be more just • 4.3 Politics should make decisions

• 4.4 Left has more advantages from politics then me

• 4.5 Dutch citizens receive less from national politics than they deserve because of the European Union

• 4.6 Immigration upsets national politics

• 4.7 If Dutch citizens could determine politics, politics would be real and decisive • 4.8 The elite has more advantages from politics then me

• 4.9 Political pledges are in favor of immigrants

• 4.10 If politicians do not do what the promise, there must be a form of immedi-ate accountability

• 4.11 If politicians formulate policy that is different from the promises, citizens must be able to immediately change the policy

(19)

19 A full version of the questionnaire is included in the appendices.1

3.3 Data collection

The data was collected in two different ways; on the internet and with participants asked on the streets of various cities throughout the Netherlands.

3.3.1 Web based survey

The data was collected on the internet, using Dutch citizens, in two ways. The document was distributed using www.thesistools.com, the questionnaire is available at the following link http://www.thesistools.com/web/?id=470931.

The questionnaire was distributed on the internet in three different ways. First, on a large Dutch internet forum, www.fok.nl, which has 10.000 or more unique visitors per day. The forum concerns a wide variety of topics and should not be considered to generate a bias. The thesis was placed in ‘general chat’ and the topic remained active for approximately two weeks. Second, the questionnaire was spread in comments on political news items. Especially news items concerning populist topics were used. This because after the first week of data collec-tion, there was an underrepresentation of PVV voters, and thus populist voters.

3.3.2 Face to face street interviews

As a second distribution method, I choose Almere, Helmond, Tilburg, Venlo, Amsterdam, and Utrecht to interview people on the streets. The cities represent some variance in voting preference. I asked people on the streets and occasionally asked people at their houses, in trains, and on train stations. The paper versions of the questionnaire are preserved and stored and are available, as is the thesistools.com account and password.

3.3.3 Problems with data collection

The biggest problem collecting data was that most of the questionnaires were not finished by respondents. The final datasheet includes 103 respondents of which 35 came from street in-terviews and 68 where internet respondents. The datasheet is accessible in .sav format, and can be used for recalculations. From the 103 respondents, none has missing data. I decided to remove all the respondents from the datasheets that had missing data within them, this where about 35 respondents. The datasheet was created using SPSS version 22.

In the following chapter I will analyze the results and test the hypothesis formulated in this chapter.

1_{In the original researchdesign I also included technocracy and social capital in the questionnaire. My}

hypothesis was that people with low social capital are more cynical towards technocratic politics. Social capital and technocracy appeared non-valuable for the data-analysis and was not needed for further investigation either. For this reason there is a part 2 & 5 in the questionnaire that is not described here.

(20)

20

4. Results

In this Chapter I will look at the data resulting from the questionnaire. The data will be analyzed with the use of Principal Component Analysis, Regression, and Mediation. The main hypothesis will be put to the test.

H1: If voters have an attitude that politicians make unfulfilled pledges, then people score high on populist attitudes

H2: If voters have feelings of counterfactual relative deprivation, then voters favor populist attitudes Since this questionnaire has an abundance of Likert-item questions, the first topic is creating suitable Likert-scales from the Likert-items. The process of developing Likert-scales for fur-ther research will be done by examining them with the use of, first, Cronbach’s α, and second Principal Component Analysis. Principal Components Analysis gives the option to save the components generated by the Likert items as variables for each respondent. The first part of this chapter will be explorative, analyzing the Likert-items and computing components for further research. The second part of this chapter will be about analyzing the components in relation to each other with linear regression.

4.1 Questionnaire response and distribution over age, income, education,

and political parties

The first important step in analyzing the data is to check whether it is normally distributed according to the Dutch electorate. The dataset generated by the questionnaire has a number of 103 respondents. This is slightly low for statistical analysis but not too low that it could cause problems for statistical analysis. The most important implication is that the data should not but biased, but should be normally distributed similar to society as a whole. For this, I looked at political party preference, income and education. Income and education seem more or less normally distributed as well, see Fig. 1 and Fig. 2. An important bias could be the frequency of representation of political parties see Fig. 3. Here the data also shows a more or less normal distributed representation of the current demos in the Netherlands. There is a slight overrepresentation of D66 and non-voters, what is generally the reasonable result of the small overrepresentation of people between the ages of 20 and 30.

(21)

21

Figure 2 Education distribution dataset

(22)

22

4.2 Exploring the Likert-item data for further research, from dimensions to

scales through Cronbach’s alpha

The exploratory phase of new conceptual research is always a bit puzzling, with combining and removing Likert-items to check and compare their variance and correlations and statistical suitability within Principal Component Analysis. It is mandatory in research to be open about this process as much as possible, although most amplification will be upon those items that do form components. Normally, when testing already widely used and narrow down or defined scientific concepts, it could be argued to leave this developmental phase out of the thesis, but because this research is partly exploratory in the use of counterfactual relative deprivation theory, some phases in the process will be argued about. It is important that the items together create a valid concept as described in the hypothesis, that is, measuring the concept that it is supposed to measure.

An important problem for Likert-items and Likert-scales, is the problem surrounding the use of nonparametric or parametric statistics. There are many hurdles with Likert-item analysis especially when using parametric statistical methods. Therefore many researcher choose the statistical reasoning that ordinal data can only be analyzed by non-parametric methods. The biggest reason is that using parametric statistics with ordinal data, as are Likert-items, is sta-tistically unsound, because most generalization could be erroneous. This debate will not be disputed here, to be sure all regressions are checked with non-parametric two tailed correla-tions.

As a first step for building Likert-scales from the Likert-items, the focus will be on internal consistency and in particular reliability using Cronbach α. The first scale to look at is political pledges. Using the Likert-items no. 2.1, 2.2, 2.3 and 2.4 the Likert-scale of political pledges will be made. Usually a Cronbach’s α below 0.5 is unacceptable, and above 0.7 is good for analysis. The pledges items show good internal reliability. Cronbach α = 0.80 See Table 5.

• 2.1 Politicians do what they promise

• 2.2 The political party which I voted for do what they promise • 2.3 Promises in party programs will be fulfilled

• 2.4 Promises in pre-election television debates will be realized

Relative deprivation had, as expected, a view problems with internal reliability. Therefore, the decision was made to combine the explicit negative items for relative deprivation theory, namely 3.7, 3.8, 3.10, 3.11, 3.12, 3.13, 3.14 and 3.15. Leaving out the positive items 3.1, 3.2, 3.3, 3.4 and 3.16.

It seems to be the case that phrases such as “If I see or hear a politician, then I know what to expect from this politician’, was a little sloppy in terms of formulation. For example, some people with negative attitudes interpreted the question as if the answer was ‘yes, namely noth-ing’ while others answered within the scope of ‘yes and I am happy with this politician’. This makes question 3.1, 3.2, 3.3, 3.4 and 3.16 to ambiguous for analysis what is also noticeable comparing the data per respondent in relation to explicit negative or positive political posi-tions per respondent.

The Cronbach α of items 3.7, 3.8, 3.10, 3.11, 3.12, 3.13, 3.14 is computed and it shows a good internal reliability with Cronbach α = 0.89 See Table 5. To be precise the Likert-scale from the Likert-items consist of the following:

• 3.7 Politics always decide differently then they promise, which is frustrating • 3.8 If political decisions are different from their promises, then I feel disadvantaged

(23)

23 • 3.10 Political pledges for others are more frequently realized then for me

• 3.11 If policy is different than they promise, then politician cannot be trusted • 3.12 If I do not get what I expect from my vote, then politics is not just • 3.13 If I do not get what I expect from my vote, then politics cannot be trusted • 3.14 If everybody gets what he wants, then everybody is equal

The last items are the items concerning populism. After Inspection of the data the decision was made to combine the items concerning the minimal definition of populism and the items that represent Dutch populism. The reason is that Dutch populism has a concerning anti-immigrant spectrum that is mostly not applicable to SP voters that are generally culturally more open but also show populist tendencies. For evidence see 4.4.4. For populism as an abstract definition the items of 4.1, 4.2, 4.3, 4.7, 4.8 and 4.11 are combined. These show good results for internal reliability Cronbach α = 0.86 See Table 5..

• 4.1 Politicians and governors have greater advantages out of politics than ordinary citizens • 4.2 If citizens could determine politics, then politics would be more just

• 4.3 Politics should make decisions

• 4.7 If Dutch citizens could determine politics, politics would be real and decisive • 4.8 The elite has more advantages from politics then me

• 4.11 If politicians formulate policy that is different from the promises, citizens must be able to immediately change the policy

For the Dutch populist scale the items 4.4, 4.5, 4.6, 4.9 are combined. These items also show a good internal reliability Cronbach α = 0.885. It must be noted that an internal reliability of all the combined items concerning populism also scores a good on internal validity.

• 4.4 Left has more advantages from politics then me

• 4.5 Dutch citizens receive less from national politics then they deserve because of the European Un-ion

• 4.6 Immigration upsets national politics • 4.9 Political pledges are in favor of immigrants

Table 5 Reliability of possible components

4.3 Principal Component Analysis

The next step for the possibility of regression with ordinal Likert-items is to do a Principal Component Analysis. This statistical method has two important features for this research. First, it shows if the items are generally measuring the same underlying construct. Construct validity is an important issue concerning Likert-items because on a five-point scale there could be a chance of underlying correlations resulting in components that are measuring different

Items N of Items Cronbach's Alpha

Pledges 4 0.80

Relative deprivation 8 0.89

Populism Minimal 6 0.86

(24)

24 concepts. The second important feature of Principal Components Analysis is that it can gen-erate an output as a scale that is continuous rather than ordinal. The PCA will be used to generate variables for each respondent lying between -2 and 2. With these components, linear regression should not be a problem even though the data is ordinal in essence.

Concerning the suitability of the data for PCA there could be a problem with the number of respondents. The sample size should ideally be at least 300 cases or more, otherwise the PCA is less reliable. The reliability decreases when more items form a component. For the political pledges component, generated out of four different items, with a good score on internal reli-ability, this should not be a problem. To be sure, a Kaiser-Meyer-Olkin Measure of sampling Adequacy is run. Generally this should be above 0.6, if below, then there is a reliability prob-lem concerning the number of respondent vis-a-vis the number of items. The Bartlett’s test should be less than 0.05 for PCA to be appropriate. For the pledge component the KMO test = 0.756 and the Bartlett’s Test is significant at 0.000 See Table 7.. Therefore the generated output of the PCA for further analysis is reliable. The pledge scale will be plotted after the rest of the components are extracted.

4.3.1 Political Promises PCA

The first component that will be constructed is the pledge component. Running a PCA anal-ysis shows that the four items, 2.1, 2.2, 2.3 and 2.4 indeed measure one and the same component. With a good internal consistency, it explains 63% of the total variance and an eigenvalue of above 1. All items score above the lower bound of 0.5 See Table 6.

Table 6 Component extraction Political Pledges

Table 7 KMO and Bartlett's test Political Pledges

4.3.2 Counterfactual Relative Deprivation PCA

The second component to be extracted is the Relative Deprivation component consisting of the items 3.7, 3.8, 3.10, 3.11, 3.12, 3.13, 3.14 and 3.15. This component was a little harder to analyze and compute from the data. Especially before the shifting of items in the data analysis, removing the positive formulated items. With 3.7, 3.8, 3.10, 3.11, 3.12, 3.13, 3.14 and 3.15, two components are generated that show a little difficulty to explain the variation in the data-set. With other mixtures of items the exploration of different relative deprivation components was less optimal, therefore this choice has been made. I do not want to elaborate on the trial and error phase of removing and adding items to much, because this diffuses more than it

Component Political Pledges

Eigenvalue 2.5 Var. 63.2 Pledge item 2.1 0.85 Pledge item 2.2 0.63 Pledge item 2.3 0.89 Pledge item 2.4 0.79 Kaiser-Meyer-Olkin Measure of Sampling Adequacy. 0.76

Bartlett's Test of Sphericity 143.8

(25)

25 explains. Mostly the removed items are concerned with the ambiguity of language. Especially 3.1, 3.2, 3.3, 3.4, show this ambiguity, and therefore they much more and different compo-nents than is desirable. Therefore it could do more damage to validity problems than the other way around. This is for I hope, partly a mistake in the design of the questionnaire and the items in particular. Though it must be highlighted that arguments against this can be made on the basis of construct validity. For now I will leave this for further discussion and continue the analysis with items 3.7, 3.8, 3.10, 3.11, 3.12, 3.13, 3.14 and 3.15. The reason for highlighting this is to be most genuine about the first data exploration. Mistakes in language should not happen, and mostly should be solved before collecting the data.

Looking at the components generated by the PCA analysis we see two components that are not very explanatory at first hand. On the basis of this component extraction it could be useful for our analysis to continue only with component 1. But statistics offer a method to look at the components from a different perspective. Removing component 2 just out of the analysis would surely add full force to those who would play the construct validity card.

Table 8 Component extraction Relative deprivation

Table 9 KMO and Bartlett's Test ReLdep

Component ReLdep 1 ReLdep 2

Eigenvalue 3.0 2.5 Var 37.9 31.7 ReLdep3.7 0.85 0.21 ReLdep3.8 0.81 0.13 ReLdep3.10 0.07 0.83 ReLdep3.11 0.77 0.34 ReLdep3.12 0.37 0.80 ReLdep3.13 0.56 0.66 ReLdep3.14 0.28 0.69 ReLdep3.15 0.72 0.33

Kaiser-Meyer-Olkin Measure of Sampling

Adequacy. 0.83

Bartlett's Test of Sphericity 445.90

(26)

26 4.3.2.1 Component rotation

Therefor a Varimax-component rotation is used to see which components consists strongly of what items and how to centralize them. Looking at the rotated analysis we see two com-ponents that are strongly influenced by one 3.7 and 3.8, and two, 3.10 and 3.12. This is a very interesting finding analogue to the discussion within the operationalization of relative depri-vation in the research model. Component 1 is structured around relative depridepri-vation based on frustration out of expectancies, and the feeling disadvantage or counter action out of po-litical promises. Hence component 1 could be a reflection of the possibility of relative depri-vation as a concept of counterfactual thinking.

Figure 4 Rotated Component Plot ReLdep

Looking at the graph we see the second component centering around 3.10 that ‘promises for others are more realized by politics’, and 3.12 about feelings of justice.

Hereby we are in some ways finding the same shift that the whole body of literature on relative deprivation has been undertaken in its history. Recalling the theoretic outline that relative deprivation for most of the 20th century is about the feelings that are caused by intergroup

relations and especially feelings of justice that are related to intergroup differences. Construct validity is mostly solved by the embeddedness in other literature, and with this result we could maybe speak of a possibility that this rotation is useful, and more importantly does not stand on its own. The exact nature of this rotation is beyond the scope of this thesis. It could be possible that certain possibilities such as Oblimin rotation could create more clarity on this subject. Otherwise it could be argued that renewed understanding of the language that create the items could generate more clarity on how the items are measuring different components. For this research the decision is made to use both components for analysis and extraction for regression analysis. Both components show a cumulative total variance of 70%, and both have an eigenvalue greater than 1 See Table 8.. The difference in explainability sure does add to the explorative nature of this research. Looking at the KMO and Bartlett’s test we see good scores for the Relative Deprivation components extraction. Therefore the generated variable can be used for linear regression analysis. The KMO-measure of Sampling Adequacy = 0.831, and Bartlett is significant See Table 9..

(27)

27 With relative deprivation as a variable the there are two important components left to extract. First the minimal definition of populism will be extracted, thereafter the variable for Dutch populism will be extracted.

4.3.3 Populism PCA

For the minimal definition of populism, the Likert-items 4.1, 4.2, 4.3, 4.7, 4.8, and 4.11 are used to run a Principal Component Analysis. These items show good results for composing a populism scale, because neither items scores below 0.5. The Likert-items result in one com-ponent that is significant and usable for further research with eigenvalue above 1 and explain-ing nearly 60% of the total variance See Table 10. Notice that 4.3 the Likert-item about decisionism is the least scoring item in the list. This is normal regarding the nature of popu-lism. The empowerment of the demos against the political elite is the has the highest explan-atory value in this component with items 4.2, 4.11, and 4.7. I think thereby an ‘us against them’ element can be reasoned without explicitly having a Likert-item about this dubious concept.

For Dutch populism the last component will be based on the items 4.4, 4.5, 4.6 and 4.9. These all have a very different topics, anti-leftish feeling, anti-immigrant, and anti-EU. A few re-spondents complained about these questions in the way they were proposed. For example, an immigrant asked how he himself as immigrant should answer the question, and somebody with very progressive leftish attitude asked how he should answer the question about the anti-left politics. These are very legit questions, but in the data, there are now signs of ambiguity in answering these question, for example in comparison with the positive statements about relative deprivation. The PCA analysis show just one component with good internal con-sistency, and with an eigenvalue larger than 1, and 74% of the variance explained See Table 11. A variable was extracted for further research. Both Populism minimal and Dutch populism score positive on the KMO and Barlett’s test.

Table 10 Component extraction PopMin

Table 11 Component extraction PopNL

Component PopMin Eigenvalue 3.5 Var. 59.7 Pop4.1 0.76 Pop4.2 0.83 Pop4.3 0.63 Pop4.7 0.80 Pop4.8 0.78 Pop4.11 0.82 Component PopNL Eigenvalue 2.9 Var. 74.3 Pop4.4 0.82 Pop4.5 0.88 Pop4.6 0.87 Pop4.9 0.88

Do unfulfilled political pledges and counterfactual relative deprivation predict populism? : an attitude survey researching the supply-side of populism in the Netherlands