Affective polarization in the Netherlands : political parties and political orientation

(1)

1 Master Thesis

Universiteit van Amsterdam

Department of Political Sciences: Public Policy and Governance

Affective Polarization in the Netherlands

Political Parties and Political Orientation

22/06/2018

Lauren Heeremans 11783710

laurenheeremans@gmail.com

Supervisor: Dhr. E. (Eelco) Harteveld MSc & Dhr. Dr. G. (Gijs) Schumacher Second reader: Dhr. Dr. M. (Matthijs) Rooduijn

(2)

2

Abstract

Since Iyengar et al. (2012) introduced the concept of affective polarization, research has tried to map whether polarization based on partisanship could also occur in multi-party European democracies (Bankert et al., 2017; Huddy et al., 2018; Reiljan, 2016). This thesis aims to measure whether there is affective polarization in the Netherlands by not only focusing on in-group sentiments but also on out-in-group sentiments, caused by identification with political parties and political orientation. Since there is no data available on out-group sentiments, the author constructed a questionnaire in order to collect to sufficient data. This data collection does not measure affective polarization over time, affect based on political identity is instead compared to other parts of the social identity (age and religion) in order to make a substantive comparison. Affective distance – the difference between in-group and out-group feelings- is greater for political parties (42.1) than for age (3.3), religion (26.1) or political orientation (10.5). Even though the impact of this study is limited due to a rather homogenous survey sample, political parties were still not expected to create such large affective distance. In a multi-party system and especially in the Netherlands, which is known for its consensus political model, this large affective distance is found remarkable and thus needs to be explored and tested with a more heterogeneous and larger survey sample over time.

(3)

3

1. Introduction

At the end of the 20th century, polarization was a rather unfamiliar concept in the

Netherlands; the infamous Dutch ‘poldermodel’ of forming political consensus allowed for opposites of the ideological spectrum to govern with each other and either agreed to disagree on the traditional lines of conflict. This changed with the national 2002 elections, where Pim Fortuyn and consorts introduced a cultural line of conflict, causing a debate on immigration on which the political parties were, and still are, polarized (Pelikaan et al, 2007; Harteveld et al, 2017). This new line of conflict created great fluctuations in political polarization in the following years, even though ideological movement was limited (Oosterwaal, 2009; Oosterwaal & Toorenvlied, 2010). This caused some to wonder, is the Dutch population polarized as well along ideological lines?

This asymmetry in political polarization between the political elite and the people is also observed in the United States (Fiorina et al., 2005) and was captured by Iyengar et al. (2012) by introducing the concept of Affective Polarization. Polarization is defined as a conflict between two opposites, whereby there is a large asymmetry between groups and solidarity within groups (Estaban & Ray, 1994). Affective polarization, however, measures polarization not by focusing on the content of the conflict but by measuring the growing distance between groups created by the growing solidarity of one’s own group. Social distance thus becomes the measure of polarization, which overshadows the content of the conflict. By using this alternative measurement of polarization, Iyengar et al. (2012) proved that over the past five decades Americans increasingly disliked each other and drifted apart due to their partisanship; being either Democrat or Republican could be a sufficient reason for the population to polarize. They became polarized without any real source of conflict other than their party affiliation.

According to Toorenwaal (2009), this same asymmetry between the political elite and the people is present in the Netherlands. Therefore it is necessary to see whether affective polarization is present in the Netherlands; can we speak of political polarization if we don’t use the right tool to measure it?

Affective polarization is measured in terms of in-group and out-group sentiments regarding a political party, also known as partisanship. In the United States partisanship is a very important and powerful part of the citizen’s social identity (Huddy et al, 2015). In the Netherlands, partisanship is considered as a different construct, due to its multi-party political

(5)

5

system. Partisanship is considered to be more instrumental and less of a salient topic that can create inclusiveness within a group and negative exclusion of others (Bankert et al., 2017). However, with the rise of populist parties and a new line of conflict at the beginning of the 21st century, voters shifted to the extremes of the political left-right spectrum not according to their socioeconomic statuses nor ideologies – which would be instrumental – but because the felt that other parties better represented them, especially on topics of immigration

(Oosterwaal & Torenvlied, 2010; Harteveld et al., 2017). This caused some to consider the Dutch elections as one of the most unstable of the Western democratic countries (Mair, 2008). Since ideology movement was limited, one wonders where this political movement to the extremes comes from. Measuring for affective polarization could provide an answer.

Of course political polarization in the Netherlands differs from the US for several reasons. As mentioned earlier, US has a two-party-system whereas the Netherlands has a multiple party system, thereby attachment only one party is much less likely simply because there are more options for political volatility. The political culture in the Netherlands is also different; whereas in the two-party system of the US conflict is common, the Dutch multi-party is based on consensus and coalitions of different parties in order to govern (Lijphart, 1969).

Moreover, where in the US the identification based on partisanship has historically grown, the Dutch political parties and citizens identify themselves much more on the

traditional political scale of left-right. Affective polarization could therefore also occur based on political orientation, ignoring political orientation would overlook vital political in and out-group sentiments. Affective polarization based on political preferences would have different consequences for multi-party Netherlands than two-party United States. The instrumental partisanship creates common ground for cooperation, which is necessary for a government that needs to be formed out of different political parties with different political orientations. If parties would have too hostile attitudes towards each other this would

undermine efficient governing. Furthermore, if partisanship would become more expressive, citizens would become less deliberative, creating less diverse opinions, which is reduces cooperation and compromise (Garret et al., 2014). Expressive partisanship would mean that the better argument would sometimes get lost in a conflict based on membership to certain political groups. However, expressive partisanship also indicates political enthusiasm – though one-sided - and political engagement, which in turn means stability of the democratic political system (Huddy et al., 2018). With declining levels of partisanship and electoral

(6)

6

volatility in Western-European Democracies, expressive partisanship could thus also lead to some political stabilization

The research question of this thesis would thus be the following:

Is there affective polarization in the Netherlands, either based on partisanship or political orientation?

This thesis starts by explaining theory that supports affective polarization, namely the Social Identity Theory. The social identity is build out of memberships of different social groups, such as a political party. Membership of a group creates positive in-group and negative group sentiments – the more positive the in-group feelings and the more negative the out-group feelings, the larger the affective distance between two out-groups. The theoretical framework then covers underlying motivations affective polarization, such as political interest, education, selective media exposure and even salience of politics during the childhood. Further the theoretical framework explores the meaning of partisanship in the Netherlands and what effect affective polarization would have on a multi-party system. The theories, difference in political systems and motivations are all captured in hypotheses.

The methods section will explore the methodology of the survey; explain the motivation for the IDPG scale to measure in-party sentiments, the thermometer ratings and the Bogardus Social Distance Scale. Also the limits and ethical issues will be raised here. The analyses and results go hand in hand, using regression models to answer the hypotheses and see whether there are significant correlations between the variables. Finally, the

conclusion and discussion are used to open up the debate on future research on affective polarization in the Netherlands.

(7)

7

2. Theoretical Framework

The theoretical framework starts by explaining the basis of affective polarization in the United States, partisanship and the Social Identity Theory, and continues by translating that into the multi-party system in the Netherlands. In does so by mapping out the differences between expressive and instrumental partisanship and the difference in the political systems. The difference in political systems and different type of data-collection will lead to the first two hypotheses. Then the theoretical framework will set out the discussed theoretical causes for affective polarization, which leads to two hypotheses on causes of affective polarization in the Netherlands.

2.1 Partisanship and the Social Identity Theory

Party identification shapes and polarizes the United States. In fact, party identification and partisanship is considered to be “one of the most significant findings of public opinions research” (Dalton, 2008). In The American Voter, Campbell et al. (1960) already considered the American individual affectively oriented, both positively and negatively, based on partisanship. Iyengar et al. (2012) found that over the past fifty years, the United States has become more and more polarized based on affect towards the political parties; positive in-party and negative out-in-party feelings were considered as the better indicators of polarization.

Political polarization in the United States is thus related to partisanship rather than ideology. Partisanship in multi-party systems is a different concept however. Partisanship in the US is considered as expressive partisanship, which remains stable regardless of the possible policy or leadership status of the party because the partisanship is part of the

citizen’s social identity. Just like other parts of the social identity, such as gender, religion or race, partisanship provokes defensive emotions when it is threatened. In multi-party system however, partisanship is usually referred to as instrumental partisanship; a responsive and informed form of partisanship, whereby partisans actively reacted to the party’s policy stances, leadership performance and their success or failure (Huddy, Bankert & Davies, 2018). Instrumental partisanship considers their partisans as ideal citizens who rationally navigate in their political environment.

Measuring expressive partisanship is relatively easy to measure compared to

instrumental partisanship; whereas expressive partisans remain loyal to one party for most of their lives, instrumental partisans are not as expressively loyal to one party. In order to

(8)

8

measure the more complex instrumental partisanship Bankert, Huddy & Rosema (2017) relied to a series of the questions of the Identification with a Psychological Group ( IDPG) scale, created by Greene (2004), in order to measure identification and preference of political parties in multi-party systems in Europe (See table 1). This new scale better helps to identify partisanship in multi-party systems to one party, which also showed that in the Netherlands 61% of the population preferring one political party over the other.

Partisanship, the preference of one political party over another, can be traced back to social identification (Green, 1999). The Social Identity Theory states that part of the social identity is constructed by group dynamics; membership to a social group is accompanies with certain characteristics and social perceptions, it creates positive in-group feelings and

negative out-group feelings. (Green, 1999; Tajfel et al., 1971). Thus, membership to a political party would create positive in-group feelings towards other members of that party and negative out-group feelings towards members of other parties or non-members. The more salient the topic of group identification is, the more positive in-group feelings and negative out-group feelings the individuals will have (Iyengar et al., 2012). Expressive partisanship, where the political identification makes part of the social identity and partisanship is a salient topic, thus usually is accompanies with strong positive in-party and negative out-party sentiments.

Recently, in-group feelings regarding political parties have been studied by measuring expressive partisanship in the Netherlands. However, the studies of Huddy et al. (2018) & Bankert et al. (2017) mainly focus solely on the in-group feelings of partisans towards their own party. Essential to measuring affective polarization in the Netherlands is also the measuring of possible negative out-group feelings towards the ‘other party’. A strong sense of solidarity with one’s own group should also correspond with social distance towards out-groups (Westwood, Iyengar, Walgrave, Leonisio, Miller & Strijbis, 2018). Partisanship is usually understood as something positive and self-evident; you feel connected to one party and therefore you automatically exclude the other parties. Positive in-party sentiments and negative out-party sentiments are thus presented as being reciprocally activated. However, both political and psychological sciences have shown that negative feelings are not simply the bipolar opposite of positive feelings (Medeiros & Noël, 2014; Baumeister, Bratslavsly, Finkenauer & Vohs, 2001). Psychologically, negative sentiments outweigh positive

sentiments, leave a longer impression and are more difficult to ware off (Baumeister et al., 2001). In political sciences we see that the vote is moved by negative sentiments rather than positive, making it a better predictor for elections (Medeiros & Noël, 2014). Since negative

(9)

9

feelings have a larger impact, negative out-group feelings are very important to take into account in order to measure affective polarization. Negative out-group feelings should be measured individually and not simply be assumed to be the opposite of the positive in-group feelings. Silva (2018) has shown that followers of Radical Right Parties in the Netherlands have both strong positive in-group and negative out-group sentiments, and that those who oppose these parties do that strongly. In order to measure affective polarization, in-group and out-group sentiments should also be measured for the other political parties in the

Netherlands.

Again, this social categorization of in-groups and out-groups on basis of politics is rather simple the US, since there are only two major parties (Democrats and Republicans). However, in most European multi-party systems, social categorization not only happens on the basis of parties but also on the basis of the ideological continuum (Left/ Right). Recently, it has been doubted whether this political orientation is not outdated and should instead be labeled as Progressive/Conservative. However, many scholars still use the left-right dimension since it holds well over time and space (Nicholson et al., 2018). Social

categorization of politics - and its perceived stereotypes of the in- and out-group - thus has two dimensions in the Netherlands; social categorization on the basis of political party and the basis of political orientation. This political orientation of left-right especially important since Nicolson (2018) found that in multi-party systems the out-group is not necessarily defined by political party by rather by their stance on the political spectrum.

Medeiros & Noël (2014) further argue that ideology also plays a significant role of in partisan identification. Whether you vote left/right or progressive/conservative has a strong influence on both positive in-party sentiments and negative out-party in Australia, Canada, New Zealand, and the United States. After all, ideology and partisanship together are the strongest indicators of political preference and ideological self-placement can tell us much about an individuals’ intergroup attitudes (Jost et al., 2009) Both in-group and out-group sentiments are thus important to measure affective polarization, related to both political parties and ideology. Then it is possible to sketch a coherent vision of how people feel and perhaps why they would vote for a certain party.

Measuring affective polarization requires two things; in-group sentiments and out-group sentiments. Although there is data on in-out-group sentiments and how affectionate the Dutch population feels towards political parties and political orientation, no such data is available regarding out-groups. Out-group sentiments are essential to affective polarization, since they create the social distance that attribute to polarization. Creating new data is

(10)

10

therefore necessary. This already creates an obstacle. Iyengar et al. (2012) draw conclusions on affective polarization by measuring it over time. Therefore the can observe and conclude that affective polarization not only is present but furthermore grows consistently over time. This thesis is not able to do that. It can however, compare the social distance that political parties and political orientations create with other parts of the Dutch social identity. By doing so, we can actually say something about the social distance that political parties and political orientations create. Iyengar et al. (2012, pp. 416) showed that party polarization is greater than racial and religious polarization. In this thesis, age and religion are used as the affective counterparts for political parties and political orientation (Pauw & Maas, 2015; Phalet et al., 2010). This leads to the first hypothesis.

H1: political identity creates more affective distance than other parts of the Dutch social identity, such as age and religion.

Whereas political parties come and go, the Dutch political spectrum has existed in multi-party European systems for more than 50 years. Also in the Netherlands, the left-right political spectrum is a dominant indicator of political preferences and also a spectrum on which political parties place themselves. Identification and affective distance is expected therefore to be much stronger with political coordination than with political parties. The second hypothesis would therefore be:

H2: political orientation creates more affective and social distance than political parties.

2.2 Causes for affective polarization

The study from Iyengar et al. (2012) shows that affective polarization in the United States has increased and other scholars agree with that by expressing that “Americans’ attitudes towards members of an opposed political party have shifted from mild negativity to outright hostility” (Garret et al., 2014, pp. 309). Therefore, explanations and possible ‘solutions’ for affective

polarization have also been explored and offered. The

strong or weak identification with a political party can be traced back to childhood, which in social psychology is referred to as the Michigan model: “party-identification is a deep-seated relationship which develops during childhood, is passed down from generation to generation, becomes stronger as time passes and influences voters by helping them to comprehend

(11)

11

politics, politicians, relationships between parties and social groups, party strategies, political issues and the performance of government” (Vlachová, 2001, pp. 481). Moreover, the more salient politics was in your childhood, the more successfully it was transmitted into your adult life (Jennings et al., 2009). Political orientations rarely change after young adulthood (Rekker, 2016). After all, “support for a party not only reflects but also shapes voters’ opinion” (Harteveld et al., 2017). Once you have a membership to a political orientation, you mostly will form your political opinions based on that membership.

However, affective partisanship cannot solely be considered as a fact of life, set in stone in your childhood. Partisanship and partisan ties are daily influenced by political events, media coverage and social events, and are therefore also processed by the individual and not only by the group (Richardson, 1999). And at that individual level, partisan ties are much more unstable and have a much smaller impact (Garzia, 2013). Moreover, the

‘solution’ of affective polarization based on partisanship is found in priming a shared part of the social identity. By priming patriotism, Levendusky (2018) reduced the negative

sentiments that Republicans and Democrats towards each other by emphasizing that they both are Americans. By doing so, they include each other as members of the same group, instead of excluding each other based on party. Thus even if partisanship is a salient topic of your social identity, the affective distance that it creates can be limited by priming a different part of the social identity. However, this also means it could work the other way around; affective distance could be increased by priming the differences of parts of the social identity.

What else could explain affective distance between political groups, besides political upbringing & political priming? Garret & Bankert (2018) show that those who moralize politics and are politically engaged are more likely to express negative out-party sentiments than those who are less interested in politics. Moreover, Abramowitz & Saunders (2008) also show that well educated citizens are more politically polarized than less educated citizens. Being politically interested and highly educated could increase personal salience of politics, which in turn could lead to stronger in-group identification and out-group exclusion.

However important political interest and education, political scholars especially research the use of different types of media as possible sources for polarization. As the era of limited media choices ends and the online media platforms have entered, there are less traditional mass media that can form a counterforce to political polarization by regularly exposing people to opposite viewpoints (Iyengar & Hahr, 2009). This build upon the

Selective Exposure Theory, which argues that people expose themselves to media they agree with already, which will mostly support their positive in-group sentiments and affirm their

(12)

12

negative out-group sentiments (Stroud, 2011; Slater, 2007). Garret et al. (2014) furthermore show that selective exposure to partisan media leads to polarizing views both in the US (two-party system) as Israel (multi-(two-party system). However, when considering the Dutch case for this thesis, Trilling et al. (2016) show that selective media exposure has a less polarizing effect in the Netherlands because the traditional media show a more moderate view than the different news channels in the United States. Less polarizing effect of the media is also moderated by the Dutch multi-party political system and the general less-extreme Dutch political standpoints. However, even though this is the case for the traditional media such as TV, papers and radio, social media can create a real echo chamber of in-group sentiments and exclude opposite arguments and view-points (Iyengar & Hart, 2009). It is thus still important to observe how media use and which media type is used for political information can have an effect on affective distance.

The hypotheses on the causes for affective polarization in the Netherlands would therefore be:

H3: the more politically interested, the more affective distance these respondents will show.

(13)

13

3. Methods

Where other affective polarization studies rely on already existing data, there is no sufficient data from the Netherlands in order to measure both in-group and out-group sentiments based on partisanship and political orientation. To test the proposed hypotheses it is thus necessary to set up a questionnaire in order to collect data, using the online survey software program Qualtrics. Since this study wants to investigate affective polarization regarding Dutch political parties and political orientations, the questionnaire is conducted in Dutch. In the upcoming section I will discuss the benefits of constructing my own questionnaire, as well as the limits and the ethics of it. Next I will discuss the variables will be measured, how they are coded into relevant data, the expecting outcomes and limits of the questions and motivations to phrase certain questions in a different way.

The questionnaire in this thesis is an online survey, which makes it easier to distribute and allows for the collecting of more respondents in a limited time frame. However, an electronic survey also implies certain characteristics, limits and ethical dimensions. Surveys, both online and paper, are imperfect concepts to collect data; it relies on the judgement of the respondents, which makes the data subjective and biased to begin with. However, electronic surveys are very cost effective, both in financial means and time. Therefore, electronic surveys are becoming more and more common in academic data collection (Andrews et al, 2003).

The sample of this survey will be a probability sample; all respondents can participate in it and are not specifically targeted, making sure that each unit of the population has the opportunity to be represented in the survey. However, I cannot say that the sample of this survey is entirely randomly selected; part of the respondents have been randomly selected via online groups both others have been reached through a snowball effect of my own personal contacts. This snowball effect means that the part of the sample is homogenous, which is a good thing in order to say something about a certain part of the population, but also limits what this study can say over the entire population. Moreover, this can lead to a larger margin of error. In order to say something about the population, the sample of the survey needs to be more diverse. Due to the snowballing effect I expect to have an overpopulation of a certain strata of highly educated, left-oriented and politically interested respondents.

Moreover, due to limited time and resources, the sample is expected to be relatively small. I expect to have a 90-95% confidence interval, which translate between 100 and 300

(14)

14

respondents.

The limits and ethical dimensions of a survey also apply. In order to respect the participant, he/she must give its consent before it fills in the survey. Furthermore, the respondent has the option to retrieve its submission and get in contact if he/she finds that necessary. Further, the respondents are guaranteed that the questionnaire is filled in enormously, guarantying the respondent’s privacy.

Figure 1.IDPG Scale measurement (Green, 2004) translated into Partisan Identity

Items by Huddy et al. (2018)

When someone criticizes this group, it feels like a personal insult

I don’t act like the typical person of this group

I’m very interested in what others think about this group

The limitations associated with this group apply to me also

When I talk about this group, I usually say “we” rather than “they”

I have a number of qualities typical of members of this group

This group’s successes are my successes

If a story in the media criticized this group, I would feel embarrassed

When someone praises this group, it feels like a personal compliment

I act like a person of this group to a great extent

(15)

15 3.1 In-Group sentiments

For measuring in-group sentiments, questions from the Identification with a Psychological Group (IDPG) scale were used, which Huddy et al. (2018) translated into Partisan Identity items (see figure 1). Since there is some overlap between some questions, the questionnaire has included the four most effective questions according to Huddy et al. (2018). Moreover, since participants have to fill in the IDPG Scale twice (both for political party and for political orientation), being efficient with questions helps to lower the change of attrition dropout rates (Groves et al., 2011). The questions measuring in-party sentiments were:

I am interested in what other people think about this party I have a lot in common with other supporters of this party

When I meet someone who supports this party, I feel connected with this person When people praise this party, it makes me feel good

The response options are based on a five point Likert scale, which is usually used to measure attitudes (Likert, 1967), varying from Strongly Disagree (1) to Strongly Agree (5), giving people a minimal in-party sentiment of 4 and a maximum of 20. The in-group sentiments are both measured for political party as well as for political orientation. The word party is for political orientation replaced with the word ‘left’ or ‘right’, depending on whether people consider themselves as either left or right in a previous question. People who consider themselves in the ‘center’ of the political center, are measured on in-group sentiments for both left and right, in order to observe whether they have equally strong/weak in-group feelings towards left/right.

Respondents are also asked to rate both political parties and the political left/right on a scale from 0 to 100, where 0 translates into ‘cold feelings’, 50 into ‘neutral feelings’ and 100 into ‘warm feelings’, similar to Iyengar et al.’s (2012) measurement of feelings of US voters towards the respective parties. By using the terms cold and warm instead of positive or negative, the chance of skewed results due to social desirability is limited since the terms warm/cold are not as value-loaded as negative/positive. These thermometer ratings are also used to measure the other parts of social identity (age, religion, ethnicity), where respondents are first asked to identify with a group and then rate the in-group and out-groups with cold-neutral-warm thermometers.

(16)

16 3.2 Out-Group: Social Distance

Affective distance is measured by the difference between in-group feelings and out-group feelings. To speak of affective polarization however, one must not only take into the

favorability, but also the social distance that is created (Garret et al., 2014). Favorability of a political party can be circumstantial, created by short-term interests, influenced by media out-put, and creates too thin ice to speak of actual polarization. Social distance, by Iyengar et al. (2012) measured by attitudes towards a child’s out-group marriage, is considered as a long-term and instinctive concept (Garret et al., 2014).

Bogardus Social Distance Scale (1924) is one of the oldest psychological attitude scales and can be used to measure feelings of individuals towards other individuals or groups (Wark & Galliher, 2007). Using different levels of personal relations (family, friends,

neighbors, colleagues) a social concept can be used in order to see how much social distance it would create between the individual and its closest social relations, descending from family to colleagues. The more upset an individual would get if his intimate relationships would feel differently about a social concept, the more social distance the concept can create. Iyengar et al. (2012) show that in 1960 US citizens were barely upset nor pleased if their son/daughter would marry someone of the opposite party. In 2008, both Republicans and Democratic showed increasing upset feelings if their child would marry someone of the opposite party. Social distance in combination with favourability creates thick enough ice to say something about polarization.

In this questionnaire social distance is captured with the question “If (family

member/best friends/neighbour/college) would vote for a different party that would make me feel…” Response options are three point scale, varying from Not Upset (1), Somewhat Upset (2) and Really Upset (3).

The social distance question is also asked regarding political orientation, by naming the opposite end of the political spectrum. A supporter of left will therefore be asked the question “If your best friend would support right wing, that would make me feel…” For someone who is politically oriented to the right, the same question will be asked regarding the left. This helps to properly visualize the social distance regarding the out-group. For neutrally oriented people, social distance questions will be asked regarding both the right and the left.

(17)

17

map out the out-group so specifically for political parties as well, therefore the out-group is simply defines as ‘another party’. Since the out-group is better visualized with political orientation, it is expected that social distance will result with a larger correlation with political orientation than with political parties.

Finally, since the out-group for political orientation can be worded specifically, the questionnaire also includes questions where respondents have to give traits to the supporters of the out-group, both in neutral (nationalistic), negative (ignorant, naïve, hypocrite, selfish, evil) and positive ( intelligent, honest, open-minded, interested in the welfare of humanity). These are based on the Almond & Verba study on social distance, also used by Iyengar et al. (2012).

The variable of political interest is a self-indicating question, where the respondent can fill in how politically interested he is on a four-point Likert scale (1= not at all, 4 = very much). Media use is both measured in how much time the respondents spends daily on the paper, social media, radio and television (0 = no time at all, 6 = more than 3 hours) and how many times he/she receives political information through that type of media (0= no time at all, 7= constantly). Further the survey checks for control variables like age, gender, education and which party they would vote for if there were elections.

(18)

18

4. Analyses & Results

4.1 Descriptive statistics

The questionnaire was filled in by 240 respondents, of which 92 men and 148 women (respectively 38.1% and 61.9%).

Table 2. Descriptive statistics control variables

Variable Mean ± StD N (%) Age 33.3 ± 16.9 Sex (Male) 92 (38.1%) Educational Level - VMBO/MAVO - MBO - HAVO/VWO - HBO - WO 3 (1.3%) 8 (3.3%) 21 (8.8%) 76 (31.7) 132 (55%) Left (0)-Right(100) Orientation 42.3 ± 21.4 Left: 144 (60%) Centre: 27 (11.3%) Right: 68 (28.5%) Political Parties VVD: 34 (14.2%) PVV: 1 (0.4%) D66: 59 (24.6%) CDA: 8 (3.3%) GroenLinks: 85 (35.4%) SP: 4 (1.7%) PvdA: 18 (7.5%) SGP: 1 (0.4%) FvD: 8 (3.3%) Other: 14 (5.8%) Abstained: 8 (3.3%)

Political Interest Not Interested: 11 (4.6%)

Barely interested: 45 (18.8%) Reasonably Interested: 117 (48.8%) Very Interested: 67 (27, 8%) N = 240

(19)

19

These statistics already show the limits of the representativeness this study mentioned earlier in the methods section. As a result from the snowballing effect, the majority of the

respondents is highly educated (55%), reasonably politically interested (48%) and left-oriented (60%). Left parties are overrepresented in the sample (GroenLinks, SP, PvdA), but centre and right parties also appear in the sample (VVD, D66, CDA, FvD). Some parties are not (DENK, 50Plus, SGP) or barely (PVV, ChristenUnie) included in the sample, thus unfortunately no analyses can be done regarding their in-group and out-group sentiments.

4.2 Political In- and out-group Affect in the Netherlands

The first expectation of this thesis was that political identity would create more affective distance than other part of the Social identity. For other parts of the social identity, the survey included thermometer ratings, whereby identification based on age, religion and ethnicity was measured. The survey sample however, also limited the diversity in ethnic groups and

religious groups. For religion, only sufficient non-religious and Catholic respondents were present in the sample. For ethnic groups, only sufficient autochthonous respondents were present, making it not possible say anything about affective distance since there was no other in and out-group to compare it with.

Table 3. Average thermometer ratings of age and religion

In-Group Out-Group In-Group minus Out-Group Youths 73.9 65.7 8.2 Elderly 71.5 76.5 -5 Affective distance Age 3.2 Non-religious 79.2 51.6 27.6 Catholics 76.2 71.9 3.3 Affective distance Religion 30.1

(20)

20

Table 3 thus offers us the affective distance that identification based on age and religion can create, giving leverage to say something about political affective distance. What is

remarkable is that age does not create large affective distance. Youths have warmer feelings towards their in-group than towards their out-group elderly, but the affective distance is moderate. Elderly, on the other hand, debunk the Social Identify Theory by having warmer feelings towards their out-group than towards their own group. Perhaps because they have once belonged to that group and can therefore identity with it. Or perhaps because age is not a salient topic in the social identity of these respondents.

Religion seems to create more affective distance, where both religious and non-religious groups have warmer feelings towards their own group than the out-group. However, religion creates more affective distance for those where religion is a less salient topic – the non-religious- than for those who are considered to consider a salient topic. This also seems contradictory with the theory, from which it was expected that affective distance would be greatest by those where the topic of possible polarization is salient (Hogg et all, 1990).

(21)

21

Now that is evident that different part of the social identity can create different affective distance, let’s take a look at the social distance that is created by political parties and political orientation.

Table 4. Average thermometer ratings of Political Parties and Political Orientations In-group thermometer rating Out- Group thermometer rating Difference in-group & out-group Correlation with in-group sentiments Significance of in-group sentiments VDD 77.3 70.6 6.7 0.282 0.000* D66 82.2 34.7 47.5 0.065 0.170 CDA 74 47.6 26.4 0.184 0.05 Groenlinks 85.6 31.9 53.7 -0.296 0.000* SP 84 39.1 44.9 -0.251 0.000* PvdA 82.8 24.5 58.3 -0.337 0.000* FvD 84.4 26.9 57.5 0.230 0.001* Affective distance PP 295 Mean:42.1 Left 48 39 9 0.357 0.000* Right 65 53 12 0.064 0.045* Affective distance left-right 21 Mean: 10.5 *P<0.05

What stands out immediately from table 4 is how much greater the political affective distance is compared to age and religion. Political parties all show warm in-group feelings and

cold(er) out-group feelings and this creates a mean affective distance of 42.1, political.

Political orientation creates considerably less affective distance (mean = 10.5), yet also shows warmer in-group feelings than out-group feelings.

Supporters of left-oriented political parties show the most affective distance with other political parties. What stands out, is that the supporters of the currently governing parties VVD & CDA show the less affective distance, with the supporters of the VVD even having warm feelings towards supporters of out-groups. Why the governing parties show less affective distance could be explained by the fact that they feel less threatened by the other

(22)

22

political parties. This perceived feeling of threat could as well be an explanation why a ‘losing/defensive’ party, such as the PvdA, shows so much affective distance. I will dive into this matter more in the discussion.

The effect of in-party sentiments, measured with the IDPG-scale, is negatively correlated with affective distance for the left political parties; the more in-party sentiments there are, the less affective distance is creates. This is striking, considering that Iyengar et al. (2012) showed that in the United States the opposite occurred; the more in-party sentiments, the greater the affective distance with the out-group. Moreover, in-party sentiments are not significantly correlated for the D66 & CDA. Therefore it could be said that in-party

sentiments have less of an effect on affective distance than expected from the results from the United States. This again underlines the important of measuring the out-group sentiments individually, instead of assuming that out-group sentiments are the mirror opposite of in-group sentiments,

In-group sentiments only are strongly and significantly correlated with each other on the left of the political spectrum. This can be interpreted in two ways; the left reacts more strongly to in-group sentiments; or the right is underrepresented in this sample, therefore the correlation is not as strong or significant. In order to explore this, we need to dive into the out-group sentiments of social distance and traits, as done in the next section.

In summary, it is possible to confirm the first hypotheses; political identity creates more affective distance than other parts of the Dutch social identity, such as age and religion. Table 4 also provides some insight for the second hypothesis, since political parties create more affective distance than political orientation. However, in order to deny or confirm the second hypothesis, we must also analyse the social distance created by both political parties and political orientation.

4.3 Social Distance

The first hypothesis can be confirmed. Political parties create more affective distance than other parts of Dutch social identity, but political orientation does not. This section will analyse how much social distance is created by both characteristics of the Dutch political identity. On this matter I hypothesized also that political orientation would have a greater effect than political parties, not only due to the long-standing traditional identification with

(23)

23

political orientation, but also due to the possibility of the visualization of the out-group – left-oriented filled in the Bogardus Social Distance Scale with the right as the outgroup and the right filled out the scale with the left as the out-group.

Table 5. Social distance Political Parties

Correlation in-party sentiments Significance R-Squared Family 0.209 0.001* 0.044 Friends 0.295 0.000* 0.087 Neighbour 0.015 0.408 0.000 Colleague 0.160 0.007* 0.026 *P<0.05

In table 5 we can observe that in-party sentiments are indeed positively correlated to all levels of the Bogardus Social Distance scale. Thus, in-party sentiments have more of an effect on social distance than on affective distance. What stands out is that in-party sentiments have a stronger connection with family than with friends; the survey sample would be the most upset if their friends would vote for a different party, rather than family.

However, these correlations are not very strong (<0.30), and neighbour is not significant. What contributed to this could be the fact that the questions regarding political parties were framed towards ‘another’ party. Therefore, the out-group is not as concretely visualized as politic orientation. Another party can either be at the complete opposite of the political spectrum, which is expected to create much more social and affective distance than a political party that is at the same side of the political spectrum. Since we do not know

whether the respondent filled in the question with an ally-party or enemy-party in mind, the correlation between social distance and in-party sentiments may be skewed in either

direction.

(24)

24

Table 6. Social Distance Political Orientations

Correlation in-group sentiments Significance R-Squared Left_Family 0.456 0.000* 0.217 Left_Friends 0.446 0.000* 0.199 Left_Neighbor 0.180 0.012* 0.032 Left_Colleague 0.276 0.000* 0.76 Right_Family -0.410 0.000* 0.168 Right_Friends -0.483 0.000* 0.234 Right_Neighbors -0.361 0.012* 0.130 Right_Collegue 0.035 0.374 0.01 *P<0.05

Both for left and right political orientations we can see a correlation on affective distance. For left and right-oriented, both family and friends show a strong and significant correlation. Interestingly, whereas political party preference created no affective distance with neighbours, correlation on the basis of political orientation is both strong and significant. Moreover, the correlation between in-group sentiments and social distance regarding political orientation is stronger for almost all parts of the Bogardus Social Distance Scale. This could be attributed to the better visualization of the out-group for political orientation.

Table 6 also shows us again a difference between the left-oriented and the right-oriented, which was also observed at affective distance. Where at the left in-group and social distance are positively correlated, the right shows negative correlations between in-group sentiments and affective distance; the more in-group sentiments the right has, the less social distance it creates. Again the right disconfirms the polarizing argument of the Social Identity theory that in-group sentiments leads to negative out-group sentiments. Could this mean that in-group sentiments have a less polarizing effect on the right than on the left?

(25)

25

Figure 1. Bar charts of proportion of traits attributed by political orientations

This difference between left and right is also observed when we look at the traits that left, neutral and right-oriented groups attribute to each other’s followers in figure 1. While the highest scoring trait from the right to left is negative (naïve), the other scoring treats are positive (Kind and Open-Minded). Left mostly attributes negative (Selfish, Ignorant) or neutral (nationalistic) traits to the right. The left barely attributes positive traits to the right, whilst the right does so. This asymmetry between left and right is very remarkable, since one would expect that there would be a significant overlap in negative out-group feelings from both groups towards each other.

Nonetheless, we see that political orientation scores on all three aspects on affective polarization; affective distance, social distance and traits. However, political parties create more affective distance than political orientation. The second hypotheses can thus not be confirmed, since it expected that political orientation would create both large affective and social distance.

What must be pointed out here is that the measuring of traits is essential to observe the difference between left and right negative out-group sentiments. Nonetheless, the fact that both the political groups attribute such negative traits towards each other is alarming.

Considering someone as naïve, selfish, ignorant and even evil considers the ‘other’ as not a worthy opponent in the political conversation, which is undesirable in a multi-party system that is based on consensus and political coalitions. The possible consequences of this will be addressed in the final part of this section.

0.00% 5.00% 10.00% 15.00% 20.00% 25.00% 30.00% Left Neutral Right

(26)

26 4.4 Political Interest & Media

For hypothesis 3 and 4, a multiple regression analysis was executed, using affective distance as the dependent variable.

Table 7. Regression table affective distance and relevant variables

Independent Variables R-Squared B(Intercept) (SE B) Gender (Male) 0.068 (0.137) 0.006 44.27 (.078) Age 0.42 (0.023)** 0.001 47.78 (.035) Education 0.089 (0.151) 0.000 46.95 (.016) Left (0) – Right (100) -0.268 (0.018)** 0.077 58.54 (-.231) Political interest 0.265 (0.003)*** 0.047 33.24 (.218) Time Paper 0.009 (0.09)* 0.002 50.17 (-.049) Time SocMedia 0.038 (0.471) 0.006 45-.34 (.075) Time TV -0.152 (0.698) 0.01 53.21 (-.101) Time Radio -0.88 (0.819) 0.006 50.79 (-.076) PolInfo Paper 0.126 (0.530) 0.003 47.12 (.052) PolInfo SocMedia 0.216 (0.039)** 0.041 41.46 ((.203) Pol Info TV -0.073 (0.036)** 0.010 52.89 (-.099) PolInfoRadio -0.049 (0.058)* 0.007 R2 full model = .216 50.91 (-.082) N= 240

(27)

27

We can observe from table 7 that age, left-right orientation, political interest, daily time spent on the paper and political information retrieved from social media, TV and Radio are

significant. Let’s first take a look at political interest, since that applies to hypotheses 3: the more politically interested, the more affective distance the respondents will show. We can indeed confirm this hypothesis, by having a positive correlation of 0.265 and a significance at the 99% level. By confirming this hypothesis, it contributes to Iyengar et Al.’s statement that politically interested people are usually interested in political because they favour one

political party. Their political interest will this therefore lead them to arguments and discussions that are in favour of their political conviction, causing them to draw upon the subjectively available pool of pro and con arguments that is in their favour (Hogg et al. 1990). However, even though this biased interest is politics is true for some, it is not the case for all. We can observe from this table as well that the correlation between affective distance and political interest is present but is rather small (<0.30). The effect of political interest may thus be different with a different survey sample

In terms of media use, only daily time spent on the paper is positively correlated with affective distance; therefore we cannot say anything whether time spent on social or

traditional media causes more affective distance. However, the variables on political

information retrieved on media do show sufficient significant correlations to say something about their effect on affective distance. Politics information received via traditional forms of media are negatively correlated with affective distance (TV, -0.073; Radio, -0.049), whilst political information received through social media is positively correlated (0.216).

Hypotheses 4 - media use, specifically social media, increases affective distance – can therefore also be accepted.

4.5 Summary results

This thesis aimed to measure whether there is affective polarization in the Netherlands by not only focusing on in-group sentiments but also out-group sentiments, motivated by

identification with political parties and political orientation. The results of this thesis show that two indicators of affective polarization – affective distance and social distance – do occur both based on partisanship and political orientation. The third indicator – traits – was also

(28)

28

observed on political orientation; both left and right oriented people attribute negative traits to their out-group.

The first finding of this thesis is that the political identity creates more affective distance than other parts of the social identity, compared to age and religion. However, it must be said that the trustworthiness of this finding is limited; the other parts of the social identity were not measured as detailed as the political identity, for which social distance and traits have been measured too. Nonetheless, by comparing how much affective distance political identity creates to other parts of the social identity, this thesis can confirm that politics is a salient topic that creates strong in-group and out-group sentiments.

The second finding of this thesis is that political parties create greater affective distance than political orientation, which rejects the hypothesis that political orientation, creates greater affective distance than political parties. This implicates something fundamental for the political partisanship in the Netherlands. Before, the prevailing assumption was that partisanship in European multi-party systems was instrumental; a responsive and informed form of partisanship, whereby partisans actively reacted to the party’s policy stances, leadership performance and their success or failure (Huddy, Bankert & Davies, 2018). However, the fact that political parties create much more affective distance than political orientation indicates that partisanship in the Netherlands is more expressive than expected. People feel a strong connection to the party that they vote for, which also results in negative feelings towards other parties.

What stands out here is that supporters of certain parties show much more affective distance than other parties; the supporters of the VVD show both warm feelings towards their own party as well as to other parties, whereas supporters for the PvdA shows warm in-group feelings and cold out-group feelings. A possible explanation for this could be the threat could be the concept of threat, which Mason (2016) describes as the possibility of losing support for the next elections. According to Mason threat to a party’s status tends to drive cold feelings amongst supporters towards other political parties. The supporters of the PvdA have lost the two previous national elections, creating a sense of threat towards their party. Supporters of the VVD – a party that has been in government for more than ten years – consider other parties less as a threat to their party, causing warmer feelings. Whether threat has an effect on affective distance could be examined in a further research.

However, even though political parties create greater affective distance, political orientation creates larges social distance. As mentioned before, this could be due to the fact that the out-group for political orientation is better visualized in the questionnaire – with

(29)

29

political orientation they were questioned specifically about their out-group (left/right), whereas with political parties the questionnaire referred to ‘the other party’.

Nonetheless, the results from political orientation show us a surprising out-come: left-oriented people show more affective distance, social distance and ascribe more negative traits to right-oriented people. Right-oriented people on the other hand, have neutral feelings towards the left-group, have a negative social distance correlation and attribute more positive traits towards the left-oriented. This, again, could be attributed to the threat-theory, since the right-wing parties have been winning elections recently and that threatens the left-oriented people, causing more negative out-group feelings. However, left-wing parties such as GroenLinks have also had electoral success over the past two elections. The threat theory would therefore not be sufficient to explain these negative feelings of the left towards the right. It could be an ideological spillover, but that does not explain why the right does not show the same negative feelings towards the left. It would be very interesting for future research to explore this.

Finally the last hypothesis on the causes for affective polarization was confirmed. Those who are politically interested and retrieve their political information through social media show more affective distance than those who are less interested in politics or retrieve their political information via traditional (Radio & TV) forms of media. Again, this was only measured through basic five-point Likert items. More research on why politically interested people show more affective distance or how social media has an influence on affective distance would be very interesting.

(30)

30

5. Conclusion & Discussion

This thesis aimed to answer the same question in the Netherlands as the Iyengar et al. (2012) try to do in the context of the United States: Is the public affectively polarized? By measuring affective polarization both on the basis of political parties and political orientation and

comparing it to different parts of the Dutch social identity, I can conclude that affective distance indeed is present in the Dutch society. Moreover, what this thesis reveals is that political parties create a larger affective distance than political orientation, which is remarkable in a political system where multiple party-identifications are observed

(Thomassen, 1993) and partisanship is deemed as instrumental rather than expressive (Huddy et al., 2018). One would expect that the steady left-right political orientations would create more affective distance towards the other than political parties.

Parties thus create more affective distance than expected. What does this mean for the Dutch democratic & political system? Large affective distance created by political parties implies strong partisanship, which brings along its own problems such as “motivated

reasoning, ignore well-grounded arguments, exhibit hostility and intolerance of out-partisans, and focus on winning or losing elections at the expense of pursuing a well-thought-out-policy agenda” (Huddy et al., 2018, pp. 195). Especially in the Netherlands, where consensus and coalitions form the basis of governance, such partisanship would form would hinder the political system; the conversation needs to remain open, but if supporters of parties consider each other either as too naïve or selfish, the window for cooperation will slowly but steadily close. However, such strong partisanship cannot be concluded from this study; in-party sentiments correlated with affective distance with some parties, not all. Nonetheless, even an indication of this partisanship could have consequences in the Netherlands. We do not know what the results will be for a larger survey sample. Either way, great affective distance is in contrast with the consensus character of Dutch politics.

However, the most revealing founding from this study is not necessarily whether political parties or political orientation create greater affective distance, but rather the asymmetry in affective distance between left-oriented respondents and right-oriented respondents. Since both left and right represented opposites of the political spectrum, I presumed equal affective distance, social distance and attribution of traits. However, being right-oriented does not only create more social distance for the left, the left also attributes more negative traits to the right, considering them selfish, nationalistic and ignorant. The right, on the other hand, does consider the left as naive, but also as kind and open-minded.

(31)

31

This asymmetry leads to important follow-up questions on the Dutch political landscape. Does the left, as right-wing politicians like to say, indeed demonize the right? Or is the left-right dimension simply outdated in the Netherlands? Should it be replaced with a

conservative/progressive dimension instead? For further research it would be interesting to measure affective distance using this dimension instead, in order to see whether it has different results.

The limits of this research go hand in hand with further future research it wishes to inspire; the sample of the survey is too small and not representative for the whole of the population. This survey sample was a homogenous group of mostly highly educated,

politically interested respondents, with a majority of supporters for left-oriented parties. Not all political parties were properly represented or represented at all. Especially PVV-voters would be necessary to include in a national questionnaire, since populist partisans are

expected to show the most positive in-group and negative out-group sentiments (Silva, 2018). Those who wish to say something about affective polarization in the Netherlands on a

national scale need thus a survey sample that is more representative of its population.

Moreover, a study over time would reveal how much of this affective distance is made up out of ‘principled’ dislike of the other party or whether it is influenced by external events, such as elections, campaigns or political scandals.

Nevertheless, I believe that this study and results have showed some surprising and revealing conclusions. Politics is more salient than other parts of the social identity; political parties create more positive in-group and negative out-group sentiments than political orientation; and the left expresses more cold feelings towards the right, whereas the right deems supporters of the left as naïve but kind. Currently, the Dutch media keeps writing about a polarized political landscape; affective polarization also gives substance on the polarization of its citizens.

(32)

32

6. Bibliography

Abramowitz, A. I., & Saunders, K. L. (2008). Is polarization a myth?. The Journal of Politics, 70(2), 542-555.

Almond, G. A., & Verba, S. (2015). The civic culture: Political attitudes and democracy in five nations. Princeton University Press.

Andrews, D., Nonnecke, B., & Preece, J. (2003). Electronic survey methodology: A case study in reaching hard-to-involve Internet users. International journal of human-computer interaction, 16(2), 185-210.

Bankert, A., Huddy, L., & Rosema, M. (2017). Measuring partisanship as a social identity in multi- party systems. Political behavior, 39(1), 103-132.

Baumeister, R. F., Bratslavsky, E., Finkenauer, C., & Vohs, K. D. (2001). Bad is stronger than good. Review of general psychology, 5(4), 323.

Campbell, A., Converse, P. E., Miller, W. E., & Stokes, D. E. (1960). The American voter. University of Chicago Press.

Dalton, R. J. (2008). Citizen politics: Public opinion and political parties in advanced industrial democracies (5th ed.). Washington, DC: CQ Press.

Esteban, J. M., & Ray, D. (1994). On the measurement of polarization. Econometrica: Journal of the Econometric Society, 819-851.

Fiorina, M. P., Abrams, S. J., & Pope, J. (2005). Culture war?: The myth of a polarized America. New York: Pearson Longman.

Garrett, K. N., & Bankert, A. (2018). The moral roots of partisan division: How moral conviction heightens affective polarization. British Journal of Political Science, 1-20.

Garrett, R. K., Gvirsman, S. D., Johnson, B. K., Tsfati, Y., Neo, R., & Dal, A. (2014). Implications of pro-and counter attitudinal information exposure for affective polarization. Human Communication Research, 40(3), 309-332.

Garry, J. (2007). Making ‘party identification’ more versatile: Operationalizing the concept for the multiparty setting. Electoral Studies, 26(2), 346-361

(33)

33

Greene, S. (1999). Understanding party identification: A social identity approach. Political Psychology, 20(2), 393-403.

Greene, S. (2002). The social-psychological measurement of partisanship. Political Behavior, 24(3), 171-197.

Groves, R. M., Fowler Jr, F. J., Couper, M. P., Lepkowski, J. M., Singer, E., & Tourangeau, R. (2011). Survey methodology (Vol. 561). John Wiley & Sons.

Harteveld, E., Kokkonen, A., & Dahlberg, S. (2017). Adapting to party lines: the effect of party affiliation on attitudes to immigration. West European Politics, 40(6), 1177-1197.

Hogg, M. A., Turner, J. C., & Davidson, B. (1990). Polarized norms and social frames of reference: A test of the self-categorization theory of group polarization. Basic and Applied Social Psychology, 11(1), 77-100.

Huddy, L., Mason, L., & Aarøe, L. (2015). Expressive partisanship: Campaign involvement, political emotion, and partisan identity. American Political Science Review, 109(1), 1-17.

Huddy, L., Bankert, A., & Davies, C. L. (2018) Expressive vs. Instrumental Partisanship in Multi- Party European Systems.

Iyengar, S., & Hahn, K. S. (2009). Red media, blue media: Evidence of ideological selectivity in media use. Journal of Communication, 59(1), 19-39.

Iyengar, S., Sood, G., & Lelkes, Y. (2012). Affect, Not Ideology: A Social Identity Perspective on Polarization. Public opinion quarterly, 76(3), 405-431.

Jansen, G., Evans, G., & De Graaf, N. D. (2013). Class voting and Left–Right party positions: A comparative study of 15 Western democracies, 1960–2005. Social science research, 42(2), 376-400.

Jennings, M. K., Stoker, L., & Bowers, J. (2009). Politics across generations: Family transmission reexamined. The Journal of Politics, 71(3), 782-799.

Jost, J. T., Federico, C. M., & Napier, J. L. (2009). Political ideology: Its structure, functions, and elective affinities. Annual review of psychology, 60, 307-337.

(34)

34

Levendusky, M. S. (2018). Americans, Not Partisans: Can Priming American National Identity Reduce Affective Polarization?. The Journal of Politics, 80(1), 59-70.

Likert, R. (1967). The method of constructing and attitude scale. Methods and Techniques in Business Research, 54.

Lijphart, A. (1969). Consociational democracy. World politics, 21(2), 207-225.

Mair, P. (2008). Electoral volatility and the Dutch party system: A comparative perspective. Acta Politica, 43(2-3), 235-253.

Mason, L. (2016). A cross-cutting calm: How social sorting drives affective polarization. Public Opinion Quarterly, 80(S1), 351-377.

Medeiros, M., & Noël, A. (2014). The forgotten side of partisanship: negative party identification in four Anglo-American democracies. Comparative Political Studies, 47(7), 1022-1046.

Oosterwaal, A. (2009). Polarisatie in de Nederlandse samenleving en politiek: het integratiebeleid. Mens en maatschappij, 84(4), 369-392.

Oosterwaal, A., & Torenvlied, R. (2010). Politics divided from society? Three explanations for trends in societal and political polarisation in the Netherlands. West European Politics, 33(2), 258-279.

Pauw, R., & Maas, I. (2015). Ziet men een tegenstelling tussen jong en oud?. Mens en maatschappij, 90(3), 245-274.

Pellikaan, H., De Lange, S. L., & Van der Meer, T. (2007). Fortuyn's legacy: Party system change in the Netherlands. Comparative European Politics, 5(3), 282-302.

Phalet, K., Baysu, G., & Verkuyten, M. (2010). Political mobilization of Dutch Muslims: Religious identity salience, goal framing, and normative constraints. Journal of Social Issues, 66(4), 759-779.

Rekker, R. (2016). The lasting impact of adolescence on left-right identification: Cohort replacement and intracohort change in associations with issue attitudes. Electoral Studies, 44, 120-131.

(35)

35

Richardson, B. M. (1991). European party loyalties revisited. American Political Science Review, 85(3), 751-775.

Silva, B. C. (2018). Populist radical right parties and mass polarization in the Netherlands. European Political Science Review, 10(2), 219-244.

Slater, M. D. (2007). Reinforcing spirals: The mutual influence of media selectivity and media effects and their impact on individual behavior and social

identity. Communication theory, 17(3), 281- 303.

Stroud, N. J. (2010). Polarization and partisan selective exposure. Journal of communication, 60(3), 556-576.

Tajfel, H., Billig, M. G., Bundy, R. P., & Flament, C. (1971). Social categorization and intergroup behaviour. European journal of social psychology, 1(2), 149-178.

Thomassen, J. J. (1993). Party identification as a cross-national concept: its meaning in the Netherlands. In Richard G. Niemi en Herbert F. Weisberg (red.) Classics in voting behavior. Congressional Quarterly Inc.

Trilling, D., van Klingeren, M., & Tsfati, Y. (2016). Selective exposure, political

polarization, and possible mediators: Evidence from the Netherlands. International Journal of Public Opinion Research, 29(2), 189-213

Van der Eijk, C., & Niemöller, B. (1983). Electoral change in the Netherlands: empirical results and methods of measurement. CT-press.

Vlachová, K. (2001). Party identification in the Czech Republic: inter-party hostility and party preference. Communist and Post-Communist Studies, 34(4), 479-499.

Wark, C., & Galliher, J. F. (2007). Emory Bogardus and the origins of the social distance scale. The American Sociologist, 38(4), 383-395.

Westwood, S. J., Iyengar, S., Walgrave, S., Leonisio, R., Miller, L., & Strijbis, O. (2018). The tie that divides: Cross‐national evidence of the primacy of partyism. European Journal of Political Research, 57(2), 333-354.