The social democratic struggle on the cultural axis : a quantitative study of the Dutch social democratic party struggling with cultural issues being salient as never before

(1)

Name: Rob Janssen Student number: 10800778

Email address: rob.janssen@student.uva.nl Date: June 21st, 2019

Program: MA Political Science (Political Economy) Research project: Global Crisis of Democracies

Supervisor: Dr. Wouter van der Brug Second reader: Dr. Armèn Hakhverdian Assignment: MA thesis

Word count: 10.998

The social democratic

struggle on the cultural axis

A quantitative study of the Dutch social democratic party

struggling with cultural issues being salient as never before

(2)

1

1. Abstract

In this thesis, the cultural dimension as a potential cause for the decline of the Dutch social democratic party PvdA will be analysed. Since immigration and integration became heavily debated topics, social democratic parties struggle with the newly formed cultural dimension. Within the potential electorate of the PvdA, there are groups favouring more open and liberal policies and groups favouring more restrictive and nationalist policies. Economically left-wing voters appeared not to be necessarily left-left-wing concerning cultural issues. The largest voter quadrant in many European countries is the left-nationalist quadrant. Due to this increasingly salient cultural dimension, social democratic parties lost appeal to left-nationalist voters. These voters increasingly became alienated from the PvdA. Partially because of this, the potential electorate of the PvdA declined.

Through the years, the PvdA gradually lost potential voters in general, but disproportionally on the left-nationalist side of the spectrum. Before 2000, the probability level of left-nationalist voters was almost as high as of left-libertarian voters. Immigration and integration caused divergence in the entire electorate, and especially in the potential social democratic electorate. Left-nationalist voters are less likely to vote PvdA. Meanwhile, the left-libertarian share among potential PvdA voters is increasing. This quadrant is, however, smaller than the nationalist one, and there are more parties competing for the left-libertarian vote. It remains difficult for the PvdA, and probably for social democratic parties across Western Europe, to address these cultural nationalist sentiments. Whatever may be the solution, it requires serious electoral and political consideration to make social democratic parties revive again.

(3)

2

2. Introduction

Turbulent times require in-depth research. Mainstream parties seem to struggle harder than ever to preserve their voters. One election is watched even more closely than the other. A party family that has been watched for many years now, is the social democratic family, as social democracy is traditionally one of the core ideologies in Western Europe. Socioeconomic systems in European nations are often characterised as a beacon of social democracy. After 1945, many Western European countries got social democratic governments or at least governments with a lot of social democratic input. The largest merit of social democracy is the welfare state. The multiplicity of benefits, facilities, and obligations welfare states provide can be qualified as social stronghold against the excesses of capitalism. But when the welfare state ideal started to crumble from the 1970s onwards, the social democratic parties were forced to adjust their principles. Unlimited access was no longer feasible. At the same time, the economic boost of the 1950s and 1960s forced countries to attract labour migrants from Southern Europe, Turkey, and North Africa. Complemented by the family reunification that followed afterwards, and more asylum inflows from the Middle East and Central Asia later on in the 1970s, 1980, and 1990s, these developments caused a considerable change in the demographic composition of Western European countries.

There exists already a lot of literature on welfare state decline and rising economic liberalism, in relation to the shift to the centre of social democratic parties. Socioeconomics has great explanatory power, and have also a direct influence on the welfare and well-being of individuals. However, economics is not always determinant. Cultural change and preservation also have become a source of political polarisation. Cultural issues, such as immigration and integration, have become heavily politicised since the start of this century. As mentioned, Western European countries saw their populations change with the inflow of migrants of all kinds from various places. This led to tensions among the public. Social democratic parties, as one of the mainstream party families, began to struggle with it. This new cultural dimension forced parties to take stances. New parties formed around these new issues pose a challenge. If parties remain ambivalent, they risk electoral reckoning. This forms the starting point of this thesis, in which the Dutch case of social democracy will be studied. The Dutch Labour Party or Partij van de Arbeid (PvdA) is the main social democratic party and will function as the case in this research.

The aim of this thesis is to find a link between the rising salience of the cultural dimension and the decline of social democratic parties. As mentioned, there seem to be more

(5)

4 than economics to it, but it has not been clarified yet what that may be. Cultural issues started to play a role on the left side of the spectrum as well, but this does not mean the left-wing voters all moved in the direction of cultural libertarianism, as may be expected. Several pieces of research show that many voters combine economically left-wing views with cultural protectionist views (Van der Brug & Van Spanje 2009, Lefkofridi et al 2014, Van der Meer et al 2017a), which also has been referred to as left-nationalism. It seems just as plausible that with social democratic parties being the mainstream centre-left political family, many of these voters once upon a time supported these social democratic parties. And somehow, along the way, these parties lost their appeal. The guiding hypothesis will, therefore, be: Due to the

increasingly salient cultural dimension, social democratic parties increasingly lost appeal towards voters with a nationalist profile, resulting in increased alienation of these nationalists, and therefore a declining potential electorate, that exists largely out of left-libertarian voters.

The thesis is structured as follows: first off, a theoretical framework will be set forth. Herein, the importance of issue salience will be explained, as well as the rise of the cultural dimension in general, and the voting motives of left-nationalists. Second, the already existing insights on how social democratic parties could be able to find an electoral way up will be briefly explained. In the third chapter, the methodologies and means of data analysis will be described. The analysis itself is divided into three parts. In the first chapter, the ideological profile of potential PvdA voters will be sketched. First, only on the cultural dimension, later, the economic left-right dimension will be added to create a two-dimensional perspective. In the second chapter, this two-dimensional quadrant division will also be used to show how the national electorate develops on both the economic and cultural dimension. To go further on the voting probability of the PvdA, the high-probability scores will be shown of each quadrant in each year. In the third chapter, the relation between all the quadrants will be statistically tested through a multiple regression. The last part of this chapter will clarify to what extent left-nationalist voters voted PvdA disproportionally, and if that disproportionality has disappeared. Finally, a conclusion will be drawn based on the results in order to see if the guiding hypothesis will be confirmed. The outcome will also be discussed and recommendations will be given.

(6)

5

3. Theoretical framework

3.1 Issue salience

In order to understand the struggle of social democratic parties on immigration and integration, it is important to acknowledge the importance of issue salience. These issues were not very salient in the political debate before the 1990s. It is perhaps a key factor in understanding why these topics became so highly controversial and divisive. Dennison and Giddes (2019) discuss this matter in relation to the rise of the anti-immigration populist radical right. It is, however, also useful to apply it to the social democratic parties. While the increased salience of these cultural issues fuelled populist radical right parties massively, they often even base their existence on it, it also likely enhanced the ambivalence among social democratic parties.

Dennison and Giddes define issue salience as "the degree to which a person is passionately concerned about and personally invested in an attitude" (2019: 111). Voters know very well which issues they care for the most, and which ones prevail over others (Ibid). In research data, one refers to issue salience often as the variable that measures "what people see as the most important national issue or problem" (Idem: 112). Issues with great salience, prove to be strong indicators on political attitudes and voting behaviour. Their preferences regarding this topic or cluster of topics generally serve as a criterion when considering the party they vote for. They continue on issue-voting theory, which states that parties do not alter public attitudes itself, but form a catalyser of these public attitudes (Idem: 113). Parties only enlarge the salience of the issue that they ‘own’. People associate parties with specific issues, and it is then up to these parties to make their issue as salient as possible, and try to convince voters of the importance of their topic and their solution (Ibid).

Dennison and Giddes go on to test correlations between the rise of anti-immigration parties and the salience of immigration as a public issue. The essence of this theoretical insight is that many investigations have shown that culturally nationalist or conservative attitudes were widely present before that immigration was actually a salient topic. Once this salience increased, and latent attitudes were suddenly addressed by a respective party or public figure, other parties started to address it as well. It became a publicly debated topic. Thus, with regard to the topic of this research, immigration and integration became heavily discussed problems within a short period of time, with the associated consequences.

(7)

6 3.2 New dimension forming

Van der Brug and Van Spanje have incorporated the increased salience of immigration and integration in what they called ‘the new socio-cultural dimension’ (2009: 312). In Western European countries party choice has revolved around one dimension, namely left-right. This dimension was mainly structured around socio-economic issues, but this has changed in recent decades, and they mention Kriesi et al (2008) as a scholar who contributed greatly to the literature on this topic. Van der Brug and Van Spanje also sketch the struggle of voters quite strikingly, as they state that ‘voters are limited to the supply of political parties and their attitudes when voting' (Ibid). Therefore, this means implicitly that voters have to choose which issue is the most important to them in order to determine the party they will vote for. One could say that issue salience structures both the demand and supply side of politics.

Coming back to the dimensions, Kriesi’s reasoning on the rise of the new cultural dimension, with immigration as the leading issue that enforces that dimension is incorporated (Ibid). There remains, however, discussion about the full definition of the socio-cultural dimension. Several criteria are set out to classify parties on both the economic and the cultural dimension (Idem: 313). The results of Van der Brug and Van Spanje show that attitudes of voters are structured heavily around these two dimensions, but party competition has not thus far (Idem: 327). This has resulted in a gap between supply and demand, leaving voters with a left-wing profile on the economic dimension, and a right-wing profile on the cultural dimension (Ibid).

3.3 Representation gap

This leaves question marks for Western European political parties, but even more so for the voters themselves. Lefkofridi et al (2014) investigated what drives this group voters, and what determines their voting choice. Because these voters can go two ways: they can fulfil either their economic or cultural preferences. This makes the circle round again because issue salience is an important factor. It is, however, also very complicated. Because as their analysis shows, which dimension voters care for the most, also heavily influences party identification (2014: 77). And as is known by now, immigration and integration as public issues have been on the rise greatly since approximately 25 years. The authors schematically show that salience of either economic issues (for example during an economic downturn) or cultural issues (during a migration crisis for instance) has a great effect on what kind of party they will vote for (Idem: 79). Remarkable is that they note that populist radical right parties might remain

(8)

7 economically ambivalent in order to attract left-authoritarian voters, especially when economics could be decisive in an election.

Perhaps one might guess that this could also apply the other way around to social democratic parties, and their apparent ambivalence on cultural issues. Their economic profile is, even in post-Third Way times, still indisputably centre-left. Just as the anti-immigration position of populist radical right parties is undeniable. But while populist radical right parties have the tide with them as the cultural issues prevail, social democratic parties may have paid the price for their ambivalence on this salient dimension that overshadowed their traditional socio-economic issues.

3.4 Explanations for the decline

Shifting the focus to social democratic parties in general, there are different explanations for their declining support, and how they could regain their popularity once more. An influential scholar, including Karreth et al (2012), blames the shift to the economic centre as the main cause for the great losses of social democratic parties across Europe. This is the most elaborated scholar on this theme. During the 1980s and 1990s, the social democratic parties struggled to enter office across Europe. Economic moderation and adopting more liberal policies were used as an attempt to catch more voters in the centre (Karreth et al 2012: 795). Many saw it is a logical step because a lot of working-class voters moved up to the middle class (Idem: 797). During the 1990s this seemed successful, with many social democratic parties entering office once more. But during the new millennium, it seemed that many of left-wing voters felt alienated by the economically centrist course. Meanwhile, the centrist voters neither stuck loyally to the social democratic parties, as they proved to be more volatile (2012: 796). Thus, a shift back to their ‘classical’ leftist position would potentially result in recovery.

This is, however, only an economic explanation for the decline. Taking into consideration the increasing salience of cultural matters, and the emergence of a cultural dimension, the economic explanation is not comprehensive. Immigration, integration, and European unification also define the profile of voters and their behaviour. For this reason, this thesis focusses on cultural attitudes among social democratic voters and respective losses because of cultural ambivalence among social democratic parties. Broadly spoken, there are two possibilities for social democratic parties on this cultural side: either becoming more libertarian/cosmopolitan or the other way around; more nationalist/authoritarian. Libertarian and nationalist are conceptualisations used in this thesis. The emphasis lies increasingly on

(9)

8 themes such as immigration, integration, and national identity, as these issues became salient the past decades. Cultural matters remain, however, inconvenient for social democratic parties, as Schmidtke also pointed out in his article on the German SPD (2016: 405). Multiple strategies are set forth in order to deal with the issues and will be discussed in the next paragraph.

3.5 Strategies to the rescue

Bale et al (2010) distinguish three positions social democratic parties can take. The first position is to hold on to their original position: the culturally libertarian position. Propagating multiculturalism, relatively open asylums policies, and strong opposition to populist right rhetoric are the key elements to this position (2010: 413). This cultural libertarianism has its origins in the 1960s and 1970s, also called the New Left, with initial principals such as universalist rights, pacifism, and third world support (Bornschier 2010: 426). Social democratic parties committed themselves to these values convincingly. Until the late 1980s, the laissez-fair migration discourse remained fairly strong, but equally undiscussed (Nedergaard 2018: 5). So even after the breakthrough of the anti-immigration populist parties, this strategy demands social democratic parties to stick to these culturally liberal stances. Abou-Chadi and Wagner (2018) researched the struggle of social democratic parties too and concluded that it would be the most successful if social democratic parties aim for the left-libertarian voter. Economically, they speak in terms of investment- versus consumption-based policies, and social democratic parties should head towards investment policies, together with cultural libertarian views to aim for well-educated, cosmopolitan, urban voters (2018: 25).

The second strategy is the other way around. It is adopting the position of the populist competitor, one way or the other (Bale et al 2010: 413). It relies on the thought to live up to the sentiment that apparently is present in large parts of the electorate. One might interpret it as the end justifies the means, or as a strategic correction, but this will come down to social democratic parties implementing restrictive immigration policies and assimilative integration policies (Ibid). It takes into account the salience of immigration/integration as public issues, questioning the dominant culturally liberal that thus far has been the core direction (Nedergaard 2018: 5; Bale et al 2010: 422). Nedergaard argues that many former social democratic voters started to disagree with the cultural libertarian course in the 1990s when cultural issues became salient (2018: 4). Bale et al remark that this strategy would be a reaction to the cultural libertarian movement, and thus will face criticism and resistance from

(10)

9 within the party, as cultural libertarianism has been the leading framework for many decades (2010: 421-422).

The third strategy Bale et al suggest is to defuse the salience of immigration and integration (2010: 413). Practically, this means trying to make the issue at least less important and to shift it to an issue social democratic parties are traditionally strong at. So instead of reshuffling the positions on the salient issue at that very moment, it is about attacking the salience of the dominant issue itself (Ibid). While the shifts on substance naturally have their pros and cons, this strategy is likely the most complicated of all three. Because this involves issue ownership and being the leading party on the issues that are dominant. Changing that is not only up to the party but up to the whole system to take the turn with that respective party. And the dominance of a party on important topics is also driven on developments and occasional events. Accusations such as ‘silencing or ignoring problems' will lurk, and will respectively fuel counterparties (Ibid).

3.6 Hypotheses

Increased salience of issues, such as immigration and integration, helped create a new cultural dimension. Politics started to revolve around this dimension increasingly. Economically left-wing voters appeared to be not as left-left-wing on cultural matters as one expected. This seems to put social democratic parties in a difficult position. Several explanations have been provided and several strategies have been proposed. Hence, it is important to explore if, in fact, social democratic parties have alienated left-nationalist voters, because of the salient cultural dimension. If that can be confirmed, only then it is appropriate to consider alternative strategies. To examine this, five hypotheses in total are formulated to test the guiding thesis.  H1: Cultural nationalist positions were already held by at least 30% of the potential PvdA

voters before the breakthrough of the populist radical right in 2002.

 H2: The share of potential PvdA voters classified as left-nationalist has decreased, while the share of left-nationalists among the full electorate is expected to have grown.

 H3: The high probability shares among left-nationalist voters have decreased more in comparison to the other voter groups.

 H4: Left-nationalist attitudes have increasingly become a negative predictor for the probability of voting PvdA, and the effect is stronger compared to other voter groups.  H5: Left-nationalists voted PvdA disproportionally, above 3 percentage points, often

(11)

10

4. Methodology

4.1 Methods

This thesis is based on quantitative methods. Data from the Dutch Parliamentary Election Studies (NKO) were used, respectively from 1998, 2002, 2010, and 2017 (Kamp et al 1998, Irwin et al 2002, SKON & CBS 2010, Van der Meer et al 2017b). The NKO of 1994 was also used to analyse the voting behaviour of left-nationalist voters in 1989 and 1994. These studies are always executed after every respective general election. The parliamentary elections studies are under the control of the Stichting Kiezersonderzoek Nederland (SKON). Data of a representative sample of the Dutch electorate is collected and provides information on plenty of issues, as well as information about voters themselves. The data are accessible through the website of Data Archiving Networked Services. The analysis has been executed with the statistical program SPSS.

The data will firstly be used to sketch the ideological profile of the potential PvdA electorate. First, only on cultural issues, and then the economic dimension will be added to create two-dimensional plots existing out of four quadrants. The national electorate will also be classified by the help of these four quadrants. For each quadrant, the high probability shares will be calculated. This results in a percentage for each quadrant in each year that shows how big that high probability group is. Thereafter, a multiple regression will be executed to test the relation between voting probability for the PvdA and the different quadrants through the years. In addition, the average probability scores of each quadrant in each year will be presented. Finally, the voting results of left-nationalist voters will be used in order to see to which extent they voted PvdA from 1989 to 2017.

4.2 Operationalisations

For the cultural profile, three variables are used. In the dataset, they are formulated as ‘asylum seekers' (immigration), ‘ethnic minorities' of ‘foreigners' (integration), and European unification. The attitudes are measured on the scale of 1 to 7. 1 stands respectively for open, liberal, pro-EU attitudes, 7 for restrictive, assimilative, Eurosceptic stances. For the one-dimensional graphs, the scale has been brought back to 3 categories. 1, 2, and 3 are put together, as well as 5, 6, and 7. Score 4 functions as the middle category. The addition of the economic axis will create a two-dimensional plot. The economic indicator used here is ‘income differences', which is based on the same 1 to 7-scale. Score 1 stands for large

(12)

11 differences, 7 stands for minimal differences. To create a left-wing and right-wing quadrant, 1 to 4 is categorised as ‘higher differences’, and 5 to 7 as ‘lower differences’.

The cultural axis has been created through the combination of the immigration and integration variable. European unification has been left out in these plots, as this scepticism on this variable does not necessarily correlate with strict view on immigration and integration. Views on immigration and integration do in fact correlate strongly. The variable ‘ethnic minorities'/‘foreigners' is merged with the variable ‘asylum seekers'. A scale of 1 to 13 comes out, in which 1 to 7 are coded as libertarian attitudes and 8 to 13 as nationalist attitudes. The two-dimensional plot is then formed by the combination of the economic and cultural axis. The left-libertarian quadrant scores left-wing on the economic axis, namely 5 to 7 on income differences, and libertarian (1 to 7) on immigration/integration. The left-nationalist quadrant scores left-wing on the economic axis to, but nationalist on the cultural axis (8 to 13 on immigration/integration). The right-nationalist quadrant scores also nationalist on the cultural axis, but right-wing on the income differences variable. The right-libertarian quadrant scores also right-wing on the economic axis, but libertarian on the cultural axis (1 to 7 on immigration/integration). For the calculation of the left-nationalist voting behaviour in 1989 and 1994, the NKO of 1994 was also used. In order to create the left-nationalist quadrant, the variable ‘ethnic minorities’ was used in combination with a differently measured ‘asylum seeker’ variable. This variable was measured on a scale of 1 to 5, with 1 meaning there are too many asylum seekers, and 5 means more may be granted access. This is different from the way ‘asylum seekers’ variable was measured differently than in 1998 and later NKO’s. Nevertheless, it seemed useful to include 1989 and 1994 too.

In the datasets, the voting probability for the PvdA is coded as ‘Probability of voting PvdA’. This is measured on a scale of 1 to 10, with 1 standing for minimal probability, and 10 for maximal probability. This full scale is used for the multiple regression. But as mentioned, the high probability shares are calculated. These shares exist out of voters who gave the PvdA a probability score of 6 or higher. Lastly, the voting results are presented for the left-nationalist quadrant after the multiple regression. The variable ‘party voted for in the x election' has been used to get the voting results of left-nationalists. In the appendix, these variables are set forth list-wise.

(13)

12

5. Results

In this chapter, the results of the analyses will be presented. It will be done in three steps. Firstly, the cultural attitudes of potential PvdA voters will be shown. In this part, but also in all other parts, it always concerns potential PvdA voters, operationalised as explained in the previous chapter. Three cultural variables were used, namely stances on immigration, integration, and European unification. All stances of potential PvdA voters will be compared to the stances of the entire electorate (these voters are included in here as well). After the analysis of the cultural dimension, the economic dimension will be added. This combination will create a two-dimensional plot, existing out of four quadrants. The distribution of these quadrants will be shown for the potential PvdA voters. In chapter 5.2, these quadrants will also be used to see how the entire electorate is distributed. Then, the PvdA probability scores of each quadrant of every year will be shown. Finally, in chapter 5.3, a multiple regression will be used to calculate how the positioning in one of the respective quadrants will increase or decrease the chance of voting PvdA probability. It will, therefore, show if left-nationalist attitudes are actually a negative predictor. In the last part of this chapter, the voting behaviour of left-nationalist voters in relation the PvdA will be analysed. It will be graphically shown to what extent left-nationalist voted disproportionally PvdA. This will prove to what extent the statistical tests are translated to the actual voting behaviour.

5.1 The ideological profile of the potential PvdA electorate

Ideological profiles of political parties change continuously over time. The shifts, however, that occurred within the electorates of social democratic parties are quite remarkable. From the 1990s onwards, with immigration and integration becoming a very salient issue, the entire electorate started to change and to adapt in line with the increased salience of that cultural dimension. This development also affected social democratic parties, and thus the PvdA. In this paragraph, the attitudes of likely PvdA voters in 1998, 2002, 2010, and 2017 will be charted and compared to the entire electorate. The cultural variables immigration, integration, and European unification are included. Comparisons with the entire Dutch electorate are drawn on these three variables by means of bar graphs, and tables with averages and standard deviations. After analysing these cultural variables individually, they will be merged with the economic variable of income differences to create four quadrants.

(14)

13 0% 10% 20% 30% 40% 50% 60% NL 1998 NL 2002 NL 2010 NL 2017 More open Same More restrictive 5.1.1 Immigration

Graph 1: Immigration attitudes potential PvdA voters Graph 2: Immigration attitudes nationally

Source: Kamp et al (1998), Irwin et al (2002), SKON & CBS (2010), Source: Kamp et al (1998), Irwin et al (2002), SKON & CBS Van der Meer et al (2017b) (2010), Van der Meer et al (2017b)

Regarding the potential PvdA voters, the three groups are from 1998 to 2017 quite balanced, all occupying more or less than one-third. The average is always around 4.2, but also the distribution is very remarkably constant. There is a slight downward tendency of the number of those favouring more restrictive policies, from 40% in 1998 to 35% in 2010, but this goes up again to 40% in 2017. The group ‘more open’ remains remarkably stable, but also grows in 2017. Thus, the average did not alter. The group ‘same’ makes a significant drop in 2017 from 35% to 26%. This might have been caused by the electoral decimation of the PvdA. Middle groups might have left the party disproportionally more compared to the other two groups.

Table 1: Distributions and averages immigration

Average Standard deviation

PvdA 1998 4.23 1.57 PvdA 2002 4.17 1.45 PvdA 2010 4.14 1.47 PvdA 2017 4.19 1.62 National 1998 4.23 1.58 National 2002 4.59 1.49 National 2010 4.56 1.47 National 2017 4.69 1.70

Source: Kamp et al (1998), Irwin et al (2002), SKON & CBS (2010), Van der Meer et al (2017b) 0% 10% 20% 30% 40% 50% 60% PvdA 1998 PvdA 2002 PvdA 2010 PvdA 2017 More open Same More restrictive

(15)

14 In 2017, the average goes up only 0.05, but the standard deviation grows with 0.15. Clearly is, however, that restrictive preferences were already present among likely PvdA voters before the breakthrough of anti-immigration parties, something Thomassen also found (2012: 20). The average score was already above 4 in 1998. Based on these outcomes, it seems difficult for a party to determine their stance. When moving in a restrictive direction, the more liberal voters will rumble, and vote differently. The other way around, the more nationalist voters will change their choice. Schmidtke found similar ambivalence within German social democratic counterpart: SPD (2016: 406).

This ambivalence is far weaker among the entire electorate. The graph shows a clear preference for more restrictive asylum policies. The group that scores 5, 6, or 7 on this variable grows from just over 40% in 1998, to over 50% in 2017. This growth is significant compared to the stagnation among potential PvdA voters, although the restrictive group there also grows to 40% once again. The average on the scale of 1 to 7 of the potential PvdA voters is equal to the national level at 4.23 in 1998 but grows with around 0.75 points to 4.69 in 2017, while those of the PvdA stay more or less the same around 4.2. Surprisingly, the distribution of voters is quite similar. The division among potential PvdA voters is apparently a national phenomenon, because also nationally the standard deviation grows with 0.23.

A possible explanation for this could be the aftermath of the refugee crisis of 2015-16 when European states were forced to take restrictive measures to hold a grip on the inflow. Another notable observation is the fact that in the pre-populist 1990s in the Netherlands, these attitudes were already widespread. Many voters, among the PvdA and the other parties, found that the inflow of asylums seekers and other migrants should be at least regulated to limited numbers. Something anti-immigration parties were able to canalise.

(16)

15 0% 10% 20% 30% 40% 50% 60% 70% NL 1998 NL 2002 NL 2010 NL 2017 Preservation of own culture Partial adjustment Far-reaching to full adjustment 5.1.2 Integration

Graph 3: Integration attitudes potential PvdA voters Graph 4: Integration attitudes nationally

Regarding integration, the picture of potential PvdA voters is considerably less ambivalent. From 1998 to 2017, there has existed a relatively large majority that likes to see foreigners adapt to Dutch culture far-reaching or even completely. However, this group has been shrinking slightly since 2002. From 1998 to 2002, this group initially grew from 51% to 54%. But from 2002 to 2017 it shrunk to 45%. The groups ‘partial adjustment' and ‘preservation of own culture both increased with around 5 percentage points. The assimilative group still stays the largest but does not form a majority among the potential PvdA voters any longer. This is also reflected in the averages. In 2002, the peak was 4.75, but this decreased to 4.49 in 2010, and 4.33 in 2017. The distribution of views, however, remains stable since 2002, so electoral performances did not affect this much.

Table 2: Distributions and averages integration

Source: Kamp et al (1998), Irwin et al (2002), SKON & CBS (2010), Van der Meer et al (2017b)

0% 10% 20% 30% 40% 50% 60% 70% PvdA 1998 PvdA 2002 PvdA 2010 PvdA 2017 Preservation of own culture Partial adjustment Far-reaching to full adjustment

(17)

16 But as with the immigration attitudes, the picture appears to be similar. Assimilative attitudes regarding foreigners were also present among potential PvdA voters before the populist radical right broke through. This majority could have been a basis for stricter integration policies, but the Dutch context in this is probably relevant for the policies of all parties that have governed the past 60 years. In short, Dutch integration policies were heavily based on the concept of multiculturalism, in which cultural and religious organisations were subsidised in order to foster the participation of minorities (Koopmans 2002: 87). These have been the policies for decades, so stepping away from them takes an effort. Perhaps that more assimilative attitudes, although not salient yet in the 1990s, were not enough to change direction.

The picture nationally is also similar, although the group that scores 5, 6, or 7 is significantly larger than among potential PvdA voters. The average scores lie structurally higher as table 2 shows. The group grows from 58% in 1998 to its peak at almost 65% in 2002. In 2010, it is still 62%, but eventually, in 2017, it shrinks back to the level of 1998. The averages also decreased, but not as fast as among potential PvdA voters. The other two groups remain relatively balanced, staying quite constant at or just below 20%, but indeed increasing in 2017 as the other group shrunk. But just as among potential PvdA voters, the assimilative group remains the largest. This also explains the relatively stable standard deviation from 2002 onwards. The same sentiment that already existed regarding immigration, exists here to too. The overall decrease of the assimilative group. From 2002 onwards could be explained by new, stricter integration policies, introduced in the 2000s, in which the emphasis shifted from cultural organisations and preservation of own identities, towards adjustment to language requirements, integration exams and civic Dutch values (Fischler 2014: 11). This possibly contributed to tempering voters’ opinions to what extent foreigners should adjust to Dutch society.

(18)

17 0% 5% 10% 15% 20% 25% 30% 35% 40% 45% 50% NL 1998 NL 2002 NL 2010 NL 2017 More Same Less 5.1.3 European unification

Graph 5: EU unification attitudes potential PvdA voters Graph 6: EU unification attitudes nationally

European unification is a more complicated issue that immigration and integration, as international co-operation and integration is not an issue in itself, but incorporates many issues, financial, social, cultural and so forth, on which the international institution is going to have a bigger or smaller say in. Criticism can rise from the idea that the national state is eroding, but also from disagreement on one or several policy issues the European institutions are acting on. That is also why this variable is not included in the two-dimensional schemes. The next chapter will discuss this in more detail.

Back to the attitudes. There is similar ambivalence as on immigration, but the polarisation is bigger. This confirmed by the standard deviations. The average grows slightly from 1998 to 2017, but the division is the most eye-catching here. In 1998, there is a slight tendency towards more positive opinions in the European Union, but in the years thereafter the group ‘same', that seems content with the current pace of integration, shrinks in favour of the group favouring ‘less’. The ‘more’ group stays the largest at just below 40%, even exceeding that 40% in 2017. The group ‘less’ also grows from 30% to 38%, almost equalling the group ‘more’ in 2010. Perhaps the European debt crisis fuelled critical sentiments towards the EU and the euro, but there remains, nevertheless, a group of one-third that stands quite negative towards further integration. In 2017, with the big electoral loss, these groups grow further apart again. Just as with immigration, division among the electorate makes taking stances on it more difficult. And the division is even bigger here. Promoting more European

0% 5% 10% 15% 20% 25% 30% 35% 40% 45% 50% PvdA 1998 PvdA 2002 PvdA 2010 PvdA 2017 More Same Less

(19)

18 integration will alienate critical voters, but being too critical and negative will push away pro-EU voters. While the popularity of debating the pro-EU seems to be fluctuating, reluctance of PvdA on this issue would be understandable.

Table 3: Distributions and averages EU unification

The entire electorate is more negatively outspoken on European unification. The averages lie structurally higher and are growing too, which indicate more and growing Eurosceptical attitudes. The EU stances of potential PvdA voters are more stable than those of the entire electorate. This is reflected in the growth of the Eurosceptical group from 30% in 1998 to 46% in 2017. The middle group ‘same’ shrunk from 31% to 22%. The pro-integration group decreased a little less from 38% to 32%. The standard deviations show that indeed the division is bigger than among potential PvdA voters, as they lie, with exception of 1998, 0.05 to 0.1 higher. The lack of salience on European unification in the 1990s is similar to immigration and integration but voiced differently. The first Eurosceptical outing in the Netherlands was in the referendum on the European constitution in 2005 when 61% voted against. But Euroscepticism as an issue was initially not ‘owned’ by the populist radical right, in contrast to immigration and integration. But as mentioned, criticism on the EU can be based on multiple perspectives.

5.1.4 Two-dimensional profile

In plot 1, the schematic quadrants and their distributions of potential PvdA voters are displayed, with one for every researched year. These quadrants are formed with the economic dimension on the horizontal axis, with lower and higher income differences as classification. The cultural dimension is on the vertical axis, which exists out of immigration and integration attitudes combined. European unification has been left out in these plots. On this axis, the options are a more restrictive stance or a more open stance towards these cultural indicators.

(20)

19

34,3%

20,2%

The quadrant left above is the left-libertarian quadrant, the right above the right-libertarian quadrant. Below on the left, the left-nationalist quadrant is located, and the right-nationalist is on the right.

Plot 1: quadrant division of the potential PvdA electorate

1998 2002

2010 2017

In 1998, the left-libertarian is the largest with 34,1%, followed by the left-nationalist quadrant with 32,8%. The two left-wing quadrants together account for two-thirds of the entire potential PvdA electorate. This is not surprising for a centre-left social democratic party, and it is a ratio that remains stable throughout the decades. The libertarian quadrants are 46,7% against 53,3%, for the nationalist quadrants. Thus, at this point, one could not yet speak of a loss of cultural nationalist voters, as they were, in fact, a majority within the potential PvdA electorate.

In 2002, the shifts are not great, despite major electoral losses. The left-nationalist quadrant has become the largest with 36%, while the left-libertarian quadrant loses a little. The nationalist quadrants expand their majority over the libertarian quadrants to 56,2% against 43,8%. The left-wing majority remains overwhelming. However, with the politicisation of the immigration/integration issue, the division on the cultural axis remains.

32,8%

34,1%

20,5%

13,5%

R es tr ic ti v e O pe n Lower Higher

36%

32,2%

36,2%

16,7%

12,8%

30%

39,4%

19,2%

11,3%

11,6%

(21)

20 The nationalist groups are even bigger than the libertarian ones. In the next chapter, will reveal if this is a national trend or a PvdA-specific trend.

In 2010, the libertarians gain ground. The left-libertarian quadrant gains 4 points and the wing equivalent 1 point. The left-nationalist quadrant also loses a little, the right-nationalist quadrant even more. The left-right-nationalist quadrant still remains second at 34%, but together with its right-wing quadrant, the nationalist quadrants are down to 51%. The libertarian quadrants together grow significantly, to 49%, thanks to that increase of the left-libertarian quadrant that is now the largest quadrant. Overall this means the left-libertarian and nationalist potential voters are now almost at an even level.

In 2017, the electoral decimation to 5,7% adds an extra dimension to this analysis. What does exist the most loyal core (potential) electorate of the PvdA out of? The result goes further in the same direction. Unsurprisingly, the left side is still the largest with almost 70%. The left-libertarian voters gain again and are by far the largest at almost 40%. The libertarian quadrants together are still 51%. The left-nationalist quadrant goes down to 30%, and for some reason the right-nationalist quadrant gains a few points, making the nationalist side account for 49%. It is indeed true that the libertarian course is supported now by a majority of voters. However, the left-nationalist quadrant, with 30%, remains the second most important factor. It is perhaps best reflected in, on one hand, more restrictive asylum policies and activating integration policies of the second Rutte-cabinet. On the other hand, the party continued to distance itself rhetorically from anti-immigration parties, in line with the hold-strategy of Bale et al (2014). The quadrant analysis, however, does not take into account the overall electoral performance. So indeed, the libertarians are at 40% and the left-nationalists at 30%, but that does not say anything about the overall a potential of the PvdA among these groups of voters. This will be covered in the next chapter.

(22)

21

5.2 National quadrant distributions and PvdA probability scores

As discussed in the theoretical chapter, political dimensions are subjected to changing cleavages. With the rise and fall of new, salient issues, definitions of left and right have changed quite rapidly, but the supply side of political systems, namely political parties, have not. It resulted in a discrepancy between that system and voters with an economically left-wing, and culturally nationalist profile (Van der Brug & Van Spanje 2009; Lefkofridi et al 2014). As explained, it is relevant for social democratic parties to take this into consideration. Socio-economic matters still have explanatory strength, but the cultural profile of voters also increasingly determine their voting behaviour. Therefore, it is important to not only look at one or another but to combine them, and to see how these distributions are actually shaped.

In this chapter, the entire electorate will be mapped in the same quadrants as in 5.1.4. However, the perspective will now shift away from potential PvdA voters towards the national electorate. The charts will say something about the entire electorate. Firstly, the overall distribution will be shown for all the years of analysis. These will be interpreted, also in comparison to the distributions of the potential PvdA voters. Hereafter, the probability of voting PvdA of all respective quadrants will be shown. So this will show us in which quadrants, in which year, have a higher probability of voting PvdA. During the analysis, both the left-wing and right-wing quadrants, and the libertarian and nationalist quadrants respectively will also be added to sketch a broader picture.

5.2.1 National electorate

The quadrant charts that have been made for the potential PvdA voters, will also be used here. The rising salience of immigration and integration did not only affect the potential PvdA voter base, but it affected the entire electorate. Therefore, it is important to look at how voters are distributed over these quadrants throughout the years. This is illustrated in plot 2. In 1998, the left-nationalist quadrant is the biggest at 31,7%. The left-wing quadrants together account for over 60%. The left-nationalist quadrant is almost equal to the left-nationalist quadrant of the potential PvdA electorate. The right-nationalist quadrant is also already at 26%. The nationalist quadrants together are almost at 58%. The left-libertarian quadrant is almost 29%, and the right-libertarian quadrant is by far the smallest at 13,5%. This remains stable from 1998 until 2017. These outcomes the same expectation that restrictive and assimilative attitudes were already widely held before populist radical right parties broke through.

(23)

22 Plot 2: quadrant division of the entire electorate

1998 2002

2010 2017

In 2002, the left-nationalist quadrant grows to almost 36% and the right-nationalist quadrant to over 32%. On the contrary, the left- and the right-libertarian quadrant shrink 7 and 3 points respectively. The right division becomes slightly more balanced, with the left-wing quadrants going down to almost 58%. The nationalist quadrants together expand to over two-thirds of the entire electorate. So not only among potential PvdA voters but among all voters left-nationalist attitudes grew. The majority of the nationalist quadrants is also a national phenomenon, instead of merely a specific PvdA phenomenon. Although among potential PvdA voters the left-libertarian quadrant lost a little in 2002, the PvdA is neither resistant to nationalist tendencies in the public opinion.

In 2010, the electorate shifts a little left, as both the libertarian and the left-nationalist quadrant grow. Possibly due to the aftermath of the economic crisis that struck the US and Europe in 2008-2009. The left-wing quadrants even surpass the level of 1998. The left-nationalist quadrant stands now at almost 38%. The libertarian and the right-nationalist score a little below and a little above one-quarter of the electorate respectively. The nationalist quadrants together account for 64%, a little lower than in 2002. On the contrary, among the potential PvdA electorate, the left-nationalists do not gain. The left-libertarians do.

31,7%

28,8%

26%

13,5%

35,9%

21,9%

32,1%

37,8%

24,1%

26,2%

11,9%

34%

25,9%

28,4%

11,7%

10,1%

(24)

23 In 2017, the tide turns slightly. The left-nationalist quadrant seems to spread out a little, with a loss of 3,8 points. This benefits the left-libertarian and right-nationalist quadrants, which grow to 25,9% and 28,4% respectively. In total, the left-wing quadrants make up 60%, against 40% for the right. The nationalist quadrants cover 62,4%, against 37,6% for the libertarian quadrants. Compared to the developments among potential PvdA voters, is that there is on both groups a tendency towards the economic left. Even other voters than those probably voting PvdA would like to see the gap between rich and poor to shrink. But on the cultural axis, there is divergence. The nationalist quadrants among the potential PvdA voters are losing ground, while among the entire electorate these are growing. The share of left-libertarians started to form the largest group among the potential electorate. But the appeal to left-nationalist voters has been declining according to the structure of the potential voter base.

5.2.2 Probability PvdA among quadrants

It may be clear that the PvdA loses and gains different potential voters than the ideological distributions among the national electorate would suggest. That does not immediately say much about how popular the PvdA used to be throughout these years. Therefore, it is also very clarifying to see it from the opposite perspective. What are the probability scores of the PvdA among the different quadrants in each year of research? Graph 7 shows these scores.

Graph 7: High probability shares PvdA among quadrants

In 1998, the PvdA scored generally high among all quadrants. With a general result of almost 29% (see appendix) is not surprising. Still, the scores of the quadrants differ from each other. Voters in the left-libertarian quadrant generally give the PvdA a probability score of 6

0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 1998 2002 2010 2017 Left-libertarians Left-nationalists Right-libertarians Right-nationalists

(25)

24 or higher the most often: 84,7%. Of the left-nationalist quadrant, almost as many voters potentially vote PvdA, with a score of 77,7%. The right-wing quadrants also score quite high with 67,5% for the right-libertarian block, and 58,3% of the right-nationalist block. The overall averages lie quite high. It may be obvious that immigration and integration were not very salient issues yet. Cultural nationalism did not yet alienate a voter from the PvdA.

This changes in 2002, when the populist radical right broke through. The high probability share within the cultural nationalist quadrants drops considerably. This share drops with 32 points among the left-nationalist voters, and 30 points among right-nationalist voters. The share within the left-libertarian quadrant drops too, but only with 18 points. The right-libertarian quadrant, in 1998 less favourable regarding potentially voting PvdA than left-nationalists, decreases only 15 points. The high probability share of right-libertarians has become larger than that same share of left-nationalists. This applies also for the averages. All averages drop, but those of the left- and the right-nationalist quadrants drop around 2 points, while the libertarian quadrants around or a bit above 1 point. In 2010, the high probability shares recover a little. The high probability share exceeds 80% again, close to the score of 1998. This recovery occurs only little at the left-nationalist quadrant. The high probability share increases a little to just above 50%, not nearly close to the 1998 scores. The right-libertarian quadrant still scores better. Finally, among the right-nationalists, the PvdA is still the least popular, with a high probability share of just over one-third.

In 2017, with the poor electoral performance, the high probability shares drop heavily. Nevertheless, clear distinctions are also noticeable. Now, only among left-libertarians, the PvdA receives significant high probability scores. The high probability share drops to 60%. The scores among the other quadrants are considerably less favourable. The left-nationalist and the right-libertarian quadrants drop to a high probability share of just over one-third. The right-nationalist quadrant reaches a low with a high probability share of just over one quarter. Relatively seen, high probability values dropped the most within the left-nationalist quadrant over the course of 1998 to 2017. This quadrant lost 42 percentage points. The left-libertarian quadrant lost ‘only’ 25 points in the same period. The share within the right-libertarian quadrant decreased with 28 points. The right-nationalist, always the weakest quadrant in terms of high probability of voting PvdA, lost 31 points. Thus, the PvdA lost potential voters across the spectrum, but relatively the most on the left-nationalist quadrant.

(26)

25

5.3 A correlation between left-nationalism and voting PvdA?

In this final chapter, the focus lies at that expected lack of appeal is statistically tested. The four quadrants as formulated in the previous chapter will function as independent variables in a multiple linear regression model. They have been re-coded into dummy variables. In this model, each profile will be connected to the probability of voting PvdA, the dependent variable. The multiple regression will show to what extent are the four respective profile quadrants predictors for the probability of voting PvdA. In the regression model, the coefficients are standardised to the scale of the probability variable. This one goes from 1 to 10.

After the interpretation of these models, the actual electoral outcome will be discussed. How often did left-nationalists vote PvdA from 1998 to 2017 compared to the general results? This will show how the regression model is reflected in the real electoral behaviour, and how the appeal of the PvdA towards left-nationalist voters evolved. These results might also provide an clarity on where left-nationalist part of the PvdA voters went, as the electoral decline has been a fact for the social democratic parties since the start of this millennium.

5.3.1 Regression models

In order to determine if left-nationalism is negative predictor, a multiple regression has been executed. The results are presented in table 4. What important is to keep in mind, is that the coefficients of the intercept (B) contain the value of the remaining quadrant of left-libertarianism. The choice to use left-libertarianism as the starting point is based on the fact that this thesis investigates to what extent left-nationalist attitudes used to be as present among potential PvdA voter as left-libertarian attitudes, and if the PvdA lost these potential left-nationalist voters. As shown in the previous chapter, voting probability of all remained the highest within the libertarian quadrant. Therefore, it is important to see how the left-nationalist coefficients will develop over the years, also in comparison with the two right-wing quadrants. In order to test the statistical significance, two controlling variables, namely education (low) and gender (male), were added in all years. Table 5 shows the averages of each quadrant in each year. The averages reveal the outcomes if the regression models would be executed without the controlling variables.

(27)

26 Over time, the classification of the four quadrants increasingly explains the covariance of the probability score of the PvdA. In 1998, R-square is at 7,3%, but increases to 12,7% in 2010. In 2017, it decreases again to 9,5%. This is likely due to the poor electoral in March of that year. The coefficients nevertheless develop over the years, as do the differences between the coefficients of the quadrants. In 1998, the intercept stands at 7.752. This is the value of the left-libertarian quadrant since this is the basic category of the regression model. Left-nationalism does not deviate much from left-libertarianism, as the coefficient is only -0.565. This means voters from both left-wing quadrants have a high probability of voting PvdA. The coefficients of the libertarian and nationalist quadrant are lower, with right-nationalism reaching almost -2. The addition of the two control variables ‘low education’ and ‘male’ do not change this image.

In 2002, major shifts occur. The intercept decreases, pointing at lower probability among left-libertarian voters. But the coefficients of left-nationalism and right-libertarianism swap positions. The right-libertarian coefficient increases to -1.02, while the left-nationalist coefficient drops to 1.565. The right-nationalist coefficient also drops further. This development corresponds with the outcome of the previous paragraph on high probability shares. The decline among the two cultural nationalist quadrants is larger than among the

Table 4: Multiple linear regression regarding PvdA voting probability

1998 2002 2010 2017

Model 1 Model 2 Model 3 Model 4 Model 5 Model 6 Model 7 Model 8 Left- nationalism -0.565*** (0.161) -0.591*** (0.162) -1.565*** (0.204) -1.486*** (0.206) -1.856*** (0.153) -1.837*** (0.16) -1.605*** (0.17) -1.526*** (0.174) Right-libertarianism -1.265*** (0.204) -1.211*** (0.205) -1.02*** (0.287) -0.985** (0.288) -1.56*** (0.208) -1.567*** (0.212) -1.324*** (0.23) -1.349*** (0.231) Right- nationalism -1.903*** (0.168) -1.873*** (0.17) -2.692*** (0.208) -2.623*** (0.211) -2.861*** (0.166) -2.854*** (0.169) -2.334*** (0.177) -2.32*** (0.178) Low educated -0.256* (0.127) -0.416* (0.161) 0.087 (0.155) -0.332* (0.166) Male -0.246 (0.146) -0.146 (0.153) -0.16 (0.122) -0.275* (0.132) Intercept (B) 7.752*** (0.115) 7.919*** (0.137) 6.572*** (0.161) 6.729*** (0.178) 7.469*** (0.118) 7.529*** (0.133) 6.077*** (0.128) 6.257*** (0.144) Adjusted R2 _0.073 _0.073 _0.10 _0.103 _0.127 _0.131 _0.095 _0.10 Observations ₁₇₅₀ ₁₇₃₁ ₁₅₁₇ ₁₅₁₇ ₂₁₀₃ ₁₉₇₀ ₁₇₀₂ ₁₆₇₆ Significance values: *** p <0.001, ** p <0.01, * <0.05

(28)

27 libertarian quadrants. The variables ‘low education’ and ‘male’ do not change this image, although the right-libertarian quadrant loses a little of its significance. Thus, education might explain some variance of the right-libertarian quadrant, but this effect is relatively small.

In 2010, this image is further confirmed. The intercept increases again, as the electoral performance during the election was better. The coefficient is almost back at the 1998 level. But the coefficients of the left- and right-nationalist quadrants drop further, and so does the coefficient of the right-libertarian quadrant. The variables for education and gender do not alter this picture. The education variable even loses significance compared to the models of 1998 and 2002 for an unknown reason. This might be an indication that the PvdA structurally lost potential among voters of other quadrants. This is no evidence that these voters directly walked away, but this does head towards the distancing of voters outside the left-libertarian quadrant.

Table 5: Average probability scores PvdA of voters from the four quadrants

1998 2002 2010 2017

Left-libertarians 7.75 6.57 7.47 6.08

Left-nationalists 7.18 5.01 5.61 4.47

Right-libertarians 6.49 5.55 5.91 4.75

Right-nationalists 5.85 3.88 4.57 3.74

In 2017, the polarisation of the coefficients seems to come to some sort of standstill. The intercept decreases to just above 6. All the other coefficients increase somewhat. But overall seen, these are the lowest statistics for the PvdA so far. Even with the increased coefficients of the left-nationalist, right-libertarian, and right-nationalist quadrants, the probability scores are at a low point. The addition of education and gender increases the intercept, the right-nationalist, and the left-nationalist quadrant, but no considerable shifts occur. Nationalist attitudes have apparently become a negative predictor of voting PvdA structurally. This performance is also reflected in table 5, in which the average probability scores of each quadrant in each year are shown. These averages are factually the result of the executed regression models without the controlling variables.

The left-libertarian averages match with the intercepts of model 1, 3, 5, and 7 of the regression table. The averages of the other quadrant also match with the coefficients. These averages confirm the development as sketched with the multiple regression models, but also with the graph on high probability shares in paragraph 5.2.2. Most striking are the different decreases in 2002. The high probability shares, the regression models, and the averages show

(29)

28 that voting probability of the PvdA has decreased more among right- and left-nationalist voters. For the left-nationalist quadrant, the result is the most outstanding, as it was in 1998 almost on the same level as the left-libertarian quadrant, but tumbling down the years thereafter. Thus, it is worth looking at the left-nationalist quadrant itself, and see how these voters actually behaved through years.

5.3.2 Voting behaviour left-nationalists

In this last paragraph, the voting behaviour of left-nationalist voters will be analysed. While most of the thesis revolved around voting probability, this incorporates real voting behaviour. It will help to confirm or reject the guiding hypothesis of this thesis because it will show if left-nationalist voters actually voted PvdA at first instance, and to what extent these voters actually stopped voting PvdA. For this analysis, a graph has been created with the voting behaviour of left-nationalists regarding the PvdA. A full table with the general voting behaviour of left-nationalist can be found in the appendix.

Graph 8: Left-nationalists voting PvdA

Source: Thomassen (1994), Kamp et al (1998), Irwin et al (2002), SKON & CBS (2010), Van der Meer et al (2017b), www.kiesraad.nl

Graph 8 shows the voting behaviour of left-nationalist voters. The first result is that, indeed, left-nationalist voted PvdA more than average in 1989 and 1998. During both elections, the left-nationalist voters are 11 points above the general result of the PvdA. Mark that this is indeed before the cultural dimension became salient. The left-right socio-economic scale determined voting behaviour first and foremost, as Van der Brug and Van Spanje already demonstrated (2009: 317). But this picture already changes in 1998. While the PvdA

0% 5% 10% 15% 20% 25% 30% 35% 40% 45% 50% 1989 1994 1998 2002 2003 2006 2010 2012 2017 Left-nationalists General results

(30)

29 generally recovers, among left-nationalist voters the PvdA loses, ending up with 31,4%. Still 2,5 points above average, but the margin is decreasing.

In 2002, the party face a great loss, but recovers in 2003, reaching almost the same level as in 1998. The advantage among left-nationalist is basically disappeared. In 2006, the score is slightly above average again but also loses some percentage points. In 2010, the score is just below 20%, equal to the general result. In 2012, the PvdA wins again, but not among left-nationalist voters. This is perhaps the most striking result, in which left-nationalist voters disproportionally abstained from voting PvdA. In 2017, an electoral low point is reached at 5,7%. This also applies to the left-nationalists (7,7%). Regarding the left-nationalist voters, it has been a decline from 2003 onwards. But the decline already started in the late 1990s, when the advantage of the PvdA disappeared.

One might wonder to which parties these voters turned. Cultural attitudes in general became a voting predictor as the data results prove, in line with the rising salience of immigration and integration (Dennison & Geddes 2019). After the 1990s, left-nationalist voters turned disproportionally to the Christian democratic CDA in 2002 and 2003 (Kamp et al 1998, Irwin et al 2002). From 2010 onwards, they started voting disproportionally for the democratic socialist SP, and the populist radical right PVV. From then, they either vote left-wing to fulfil their economic preferences or right-left-wing to fulfil their cultural views (Lefkofridi 2014: 79). Before the cultural dimension became salient, this dilemma did not exist. Hence, the disproportional vote for the PvdA in the 1990s and before makes sense.

The social democratic struggle on the cultural axis : a quantitative study of the Dutch social democratic party struggling with cultural issues being salient as never before

The social democratic

struggle on the cultural axis

A quantitative study of the Dutch social democratic party

struggling with cultural issues being salient as never before

1. Abstract

Table of contents

2. Introduction

3. Theoretical framework

4. Methodology

5. Results

5.1 The ideological profile of the potential PvdA electorate

34,3%

20,2%

32,8%

34,1%

20,5%

13,5%

36%

32,2%

36,2%

16,7%

12,8%

30%

39,4%

19,2%

11,3%

11,6%

5.2 National quadrant distributions and PvdA probability scores

31,7%

28,8%

26%

13,5%

35,9%

21,9%

32,1%

37,8%

24,1%

26,2%

11,9%

34%

25,9%

28,4%

11,7%

10,1%

5.3 A correlation between left-nationalism and voting PvdA?