The trade-off that left-authoritarians make at the ballot box
A quantitative exploration of the role of left-right identification and perceived party positionin the constrained voting decision of left-authoritarians in the Netherlands
Master thesis by Veerle Weustink Student number: 10431608
Thesis supervisor: Dhr. dr. A. (Armèn) Hakhverdian Second reader: Dhr. Prof. dr. T.W.G. (Tom) van der Meer 21-06-2020
Master thesis Political Science track Political Economy, University of Amsterdam
Abstract: Left-authoritarians (voters who hold left-wing economic positions and right-wing cultural
positions) make up a sizeable share of the Dutch electorate (20,3 percent). Still they find themselves in a suboptimal voting decision: since no party in the Netherlands represents their views on both of the two dimensions of the policy space, they have to make a trade-off between their economic and cultural believes. How do they make this decision? Recent studies point towards the importance of salience they attribute to one of the two dimensions while making this trade-off (Lefkofridi et al. 2014; Spies 2013). In this research two additional interaction factors have been explored that might play a role -next to salience - in shaping their voting decision: left-right identification and party position. With the use of the NKO 2017 data set the probability for left-authoritarians to vote for two Dutch parties was examined: the left-wing party SP and the populist-right party PVV. Although there was found no significant effect of left-right identification altogether and of the perceived position of SP, the perceived position of PVV did have a sizeable and significant effect on the voting decision of this group of voters. These results show that left-authoritarians who perceive PVV to hold left-wing economic views are more likely to vote for this party then the rest of the electorate. This finding has some important real-life implications on both the supply and demand side of the Dutch political landscape.
Keywords: voting behavior, left-authoritarians, policy space, issue salience, left-right identification,
Table of content
1. Introduction – page 3
2. Theoretical framework and hypotheses – page 6
2.1 Previous research – page 6 2. 2 Hypotheses – page 9
3. Case study – The SP and PVV- page 12 4. Data and methods – page 17
4.1. Data & variables – page 17 4.2 Control variables – page 19 4.3 Methods – page 20
5. Results – page 21
5.1 Baseline models – page 21
5.2 Models with interaction effects – page 24
• Left right identification – H1 & H2 – page 24 • Perceived party position - H3 & H4 – page 29 6. Discussion and conclusion – page 34
7. Literature - page 38 8. Appendix – page 43
1. Introduction
At the start of this century Dutch publicist Paul Scheffer published the essay Het multiculturele drama (‘a multicultural tragedy’), in which he addressed a perceived unwillingness of political elites to acknowledge how the influx of migrants (mainly of Arabic descent) would change and eventually threaten the core of the Dutch ‘tolerant and liberal’ democracy (Scheffer, 2000). This essay was controversial from the start and is still often referred to as one of the defining pieces that politicized the Dutch debate about immigration and integration. In 2020 Scheffer’s narrative is widely shared not only among the populist right but among mainstream parties as well (Van Spanje, 2010). What is remarkable about Scheffer, however, is that he was a prominent member of the Labour party (PvdA). Scheffer thus combined a set of ideas that are quite prominent among citizens, yet all but absent in the political landscape: left-wing economic issues, combined with right-wing cultural issues. Or, as this hybrid ideology is often referred to: left-authoritarianism 1.
This gap between political supply and demand has been called a ‘blind corner of representation’ (Thomassen, 2012). While the largest group of voters in Western Europe (41 percent) consist of individuals who consider themselves left-authoritarian - they lean towards a ‘left-wing’ position on socio-economic issues and towards a ‘right-wing’ position on cultural issues - it is relatively quiet on the supply-side (Van der Brug et al., 2012: 65-66; Hillen & Steiner, 2019: 1; Van der Brug & Spanje, 2009: 32). There are very few true left-authoritarian parties in Western Europe and due to this left-authoritarian supply gap left-authoritarian voters face a suboptimal voting discussion: they can’t find a party that combines their “left-wing” socio-economic views and “right-wing” cultural views (Van der Brug & Spanje, 2009: 32). Simply put: these voters are attitudinally cross-pressured and have to make a trade-off (Brader et
1 Left-authoritarianism has been described by many scholars, but not always by the same term. Lipset (1959) referred to a group of working-class citizens with left-wing economic views and authoritarian (or conservative) views in the cultural domain as ‘working class authoritarians’. Some even referred to this set of policy preferences as ‘economic populism’ (Derks, 2004), arguing that in many cases these attitudes are ‘populist’ rather than ‘leftist’ on the economic dimension. In this thesis the more commonly used term of ‘left-authoritarians’ will be used (Lefkofridi et al., 2014; Van der Brug & Spanje, 2009; Van der Brug et al., 2012; Hillen & Steiner, 2019; Thomassen 2012), because it does not make judgements or assumption on the question where these policy stances could stem from, but simply states a citizen's voter preference on the cultural/authoritarian dimension and economic dimension.
al. 2013). They either a) abstain from voting (Hillen & Steiner, 2019), b) put their right-wing positions on cultural issues above their left-right-wing positions on economic issues, voting for a party in the right-authoritarian quadrant, or c) put their left-wing positions on economic issues above their right-wing positions on cultural issues, voting for a party in the left-libertarian quadrant. But how do they make this trade-off?
Currently the most influential hypothesis that answers this question is presented by Lefkofridi et al. (2014), in which they show that whether left-authoritarians vote for left-libertarian or right-authoritarian parties depends on the salience they attach to the different dimension of the policy space: do they consider economic issues or immigration issues to be the most pressing? The aim of this thesis is to contribute to this research agenda by broadening the academic scope and considering two additional factors that could be important in shaping the voting decision of these left-authoritarian voters. These complementary factors will be in line with an unpublished paper by Steiner & Hillen (2018), in which they presented two previously neglected conditions which could shape the voting decision of left-authoritarian voters: ideological left-right identification and perceived party positions. Their researched on the 2017 German election showed strong support for the idea that ideological right identification shapes the voting decisions of German left-authoritarians: the more they identify with the ideological left, the more the will be persuaded to vote for a left-libertarian parties rather than right-authoritarian parties – and vice versa. They also found that perceived party position matters for German left-authoritarian, mainly for the case of voting for a right-authoritarian party whom they perceived to hold left-economic views. This thesis will explore whether these additional factors also play a role in the voting decision of left-authoritarian voters in the Netherlands. The Netherlands has experienced increased fragmentation (Van der Meer et al., 2012) and left and right identities are notoriously vague (De Vries et al., 2013). Whether or not German results translate to the Dutch context is therefore an open question. Hence the research question for this thesis will be:
To what extent do left-right identification and perceived party position shape the voting decision of left-authoritarian voters in the Netherlands?
This thesis contributes to current research on left-authoritarians by building on the explanatory power of salience, while further investigating how this sizeable group of voters makes the trade-off between their economic and cultural views. This is highly relevant since the sheer fact a large segment of voters is not being represented, could be a threat to the democratic representation of voters on the long term (Thomassen, 2012; Steiner & Hillen, 2019). Additionally, since the salience of issues relating to immigration has been rising the last decades in Western Europe (Kriesi et al. 2006) one could argue that it is to be expected that the preference of left-authoritarian voters will play a vital role in upcoming elections.
Some scholars have suggested that this blind corner of representation has recently been catered to a bit more, due to radical right parties moving to the left on economic issues (Hillen & Steiner, 2019; Harteveld 2016; Lefkofridi & Michel 2017; Kriesi et al. 2012; Spies, 2013). One could also argue this has also happened to some extent on the other side of the political spectrum: former social democratic parties in Denmark and Finland have moved to the right on cultural issues such as immigration (Nedergaard, 2017; Steiner & Hillen, 2018). The focus of this thesis is on the appeal of these two options for left-authoritarian voters: (radical) right parties who are moving left on economic issues and (radical) left parties who are moving right on cultural issues. In the context of the Dutch party system, which has a become increasingly unstable and fragmentated in the last decades (Pelikaan et al., 2018; Van der Meer et al., 2012), a decision was made to focus on two specific parties in the Netherlands for the analysis of this thesis: the PVV (Partij voor de Vrijheid, Freedom Party) and the SP (Socialistische Partij, Socialist Party). The choice for PVV and SP is based on their extraordinary position in the Dutch political landscape: both parties combine a very outspoken stance on one of the two axes (PVV on immigration and SP on socio-economic issues) and a (deliberately) vague position on the other, which makes both these parties a particularly appealing choice for left-authoritarian voters. Further on in this thesis the decision for these two parties as the focus of analysis will be elaborated more thoroughly.
In the theory section of this thesis four hypotheses will be presented based on the two additional factors. It is expected that left-authoritarian voters are more likely to vote for the SP when they self-identify with the ideological left and when they perceive the SP to hold an authoritarian position on immigration. Vice versa, it is expected that left-authoritarian voters are more likely to vote for the PVV when they
self-identify with the ideological right and when they perceive the PVV to hold economically left positions. The analysis in this thesis was based on the NKO 2017 data set, which provides a rich set of questions regarding the issue position of respondents in the Netherlands and their political preferences. With the use of multivariate regression analyses with interaction variables, the four hypotheses were tested. The results showed no strong support for the explanatory power of ideological left-right identification. In regard to perceived party position, there was no strong support to suggest that left -authoritarians are more likely to vote for the SP when they perceive this party to be authoritarian on the cultural dimension. However, strong support was found for the last hypothesis: the perception of PVV as a party who presents themselves as in favor of decreasing income differences really does make a difference for left-authoritarians.
2. Theoretical framework and hypotheses
2.1 Previous research
This thesis is based on the assumption that the political ideologies of voters in Western European countries are two-dimensional: they are located on two main policy dimensions, being a socio-economic and a libertarian-authoritarian dimension (Kriesi et al., 2006; 2008; 2012; Van der Brug et al., 2012; Lefkofridi et al., 2014). Although this two-dimensional policy space has been widely used by scholars who study the Western European party system, not all scholars agree upon the exact meaning of these dimensions (Van der Brug & Spanje, 2009: 313). In this thesis the following interpretation of these two dimensions will be deployed. The first dimension is the socio-economic dimension, which displays a voter’s preference on the level of redistribution of income. Most people refer to this on a day-to-day basis with the labels of economically left and right: ‘left’ meaning being in favor of advanced levels of redistribution through taxes, social security and market regulations, while the ‘right’ will advocate for less market regulations and a smaller role of the state in the market. The second dimension is the libertarian-authoritarian dimension, which is sometimes referred to as the socio-cultural dimension or the cultural dimension and holds a number of issues. These include conflicts over cultural values, issues of globalization and international political integration (Kriesi et al., 2006; 2008; 2012; Van der Brug & Spanje, 2009; Steiner & Hillen, 2018). Its most salient issue is however immigration,
which in our current political climate means that the ‘left’ takes a more libertarian position: being in favor of a more generous refugee policy, open borders, and libertarian values. The ‘right’ on the other hand can be defined by its generally more authoritarian position on immigration: arguing against the influx of immigrants, nationalistic values and searching for regional solutions for migration (Lefkofridi et al., 2014: 67-68; Van der Brug et al., 2012: 55- 56). This two-dimensional policy space results in four quadrants, which are illustrated in the figure below.
Figure from Lefkofridi et al., 2014: 68
As voters are dispersed among all four quadrants, one would expect parties to follow their lead and also manifest in one of the four quadrants. However, Van der Brug and Spanje (2009: 309) show that although the attitudes of citizens are structured by two dimensions, political parties mainly organize along one dominant dimension. Although the salience of economic issues has declined in the last decades and the salience of immigration has steadily grown, Kriesi et al. (2006) have shown that most parties in Western-Europe have historically positioned themselves along the economic dimension. The rise of salience of cultural issues has changed the policy space of Western-European politics, but left one quadrant empty (Kriesi et al., 2006: 937 - 938). In most of Western-European countries one can find the majority of parties in the right-authoritarian, left-liberal and right-liberal quadrant, but there are hardly any in the left-authoritarian quadrant (Lefkofridi et al., 2014: 68) - with the exceptions of the True Finns in Finland and the Social Democratic Party in Denmark (Steiner & Hillen, 2019: 2; Nedergaard, 2017). This is the ‘blind corner of representation’
Thomassen (2012) was referring to: in many countries left-authoritarians are not represented by any party (Van der Brug & Spanje, 2009: 311; Thomassen, 2012; Lefkofridi et al., 2014; Kurella & Rosset, 2017). Lipset (1959) already showed the importance of this group of voters, who he called ‘working-class authoritarians’ and these voting-preferences have been prevalent in contemporary Western Europe (Kriesi et al., 2006).
This striking deficit between the supply and demand side thus leaves left-authoritarian with a trade-off: if she will vote (which is not a given, since Hiller and Steiner (2019: 1) have showed that these voters are less likely to vote, less satisfied with the democratic process and have lower levels of political trust) how will she vote? Most scholars who study voter behavior assume that voters make rational decisions based on policy preferences (Van der Brug et al, 2012: 65; Lefkofridi et al., 2014: 65). This is mostly based on the Downsian (1957) idea of the ability to make spatial comparisons while making electoral decisions (Devine, 2015: 509; Steiner & Hillen, 2018: 4). Policy distance matters to voters and they will most likely strive to minimize the distance between their views and the party they vote for. In a one-dimensional policy space, this is relatively easy but in a two-dimensional space it becomes more complicated. In practice left-authoritarians can only minimize their policy distance on one of the two ideological dimensions, so they have to make a trade-off. So, Lefkofridi et al. argue that voters don’t necessarily weight distances on the two dimensions equally, they prioritize one of the two dimensions based. Or in other words: they base their decision on the salience of one of the dimensions (Lefkofridi et al., 2014: 69 - 70). Is a left-authoritarian voter worried about migration? Then she will most likely vote for a right-authoritarian party like the PVV. Is she however more concerned about economic issues, then she will most likely be tempted to vote for a left-liberal party like the SP (Lefkofridi et al., 2014: 65; Steiner & Hillen, 2018: 5; Kurella & Rosset, 2017).
In most studies on the demand side of the left-authoritarianism salience is thus the default explanation, but it’s a rather tricky concept to measure. It raises questions about interactions between the supply and demand side, such as: do political parties make issues salient by employing it in their party manifesto or speaking on it in the media or do voters and their growing concerns on certain topics make issues salient? Although the concept of salience is hard to measure, scholars such as Lefkofridi et al. (2014) have empirically showed that salience is likely to be a strong driver of the left-authoritarian vote. This thesis will build upon their research and will thus ascribe to
the explanatory nature of salience, while considering two other complimentary factors that could be important in shaping the voting decision of these left-authoritarian voters. Throughout this thesis the concept of salience will be reflected upon in regard to these two complementary factors: left-right identification and perceived party position.
2. 2 Hypotheses
On a very basic level, this thesis comes down to the following question: will these two additional factors (left-right identification and perceived party position) affect the probability to vote for PVV/SP for left-authoritarians compared to the rest of the electorate? Will the probability increase, decrease or does the interacting variables have no significant effect? In the following part, four hypotheses will be formulated that show the direction of these factors. In figure 1 one can see a visualization of the causal expectation of the four hypotheses that will be tested.
Figure 1: Illustration of causal expectation of hypotheses
Kriesi et al. (2006; 2008) denounces that contemporary Western European politics is two-dimensional, creating the so-called ‘winners and losers of globalization’. However, it can be argued that ideological self-identification has not simultaneously grown into a two-dimensional concept. Although some voters will of course self-identify on two of the axes (as left-liberal or right-conservative etc.) political self-identification is still mostly measured on a continuum left-right scale (Rekker et al., 2017). These ‘left’ and ‘right’ labels are not futile. They go to the core of political ideologies, making them still empirically relevant in order to understand political behavior (Devine, 2015: 510). Political identification is broader than how you vote, it is about which group you feel
like you belong to. In an effort to reconceptualize this, scholars now often see ideological identification as a symbolic attachment to ideological groups, actors and symbols that are associated with these political ideologies. The labels “left” and “right” become ‘social identities’ or ‘super identities’ which voters emotionally identify with (Devine 2015: 510 - 519; Steiner & Hillen, 2018: 8). The phenomenon of strong group identification with the political “left” or “right” is more pronounced in a two-party system such as the U.S., but remains relevant in a multi-party system (Devine, 2015: 301 - 302). Here salience of the two-dimensions is also an important factor: De Vries et al. (2013) found that between 1980 and 2006 voters’ left/right self-placement in the Netherlands was more strongly determined by anti-immigrant attitudes than by economic attitudes.
With the case of left-authoritarians, this can present an interesting dynamic: despite the fact that left-authoritarian voters agree on their preferred policy position, they do not give heterogeneous answers when asked about how they identify on their left-right position (Steiner & Hillen, 2018: 2). Even more puzzling, Steiner and Hillen found that in Germany many left-authoritarians position themselves on the opposite poles of the left-right spectrum, identifying as extremely right or extremely left (Steiner & Hillen, 2018: 2 ; Hillen & Steiner, 2017). Simply put: two people can hold very similar left-authoritarian policy preferences, but they can fundamentally differ in whether they feel connected to the political “left” or “right”. Although these results are convincing for the case of Germany, theoretically one could argue that a different situation is just as possible. Because left-authoritarians are being cross-pressured a situation might occur in which they don’t fully identify with one of the two group identities and therefor place themselves more in the middle of the left-right ideological spectrum. What the case is for the Netherlands will be analyzed in the results section of this thesis, but for now it is expected that a left-authoritarian voter who strongly identifies with the left is more likely to vote for the SP, and vice versa. Hence, the first two hypotheses of this thesis will be:
Hypothesis 1: When left-authoritarians identify with the ideological left, they are more likely to vote for the SP.
Hypothesis 2: When left-authoritarians identify with the ideological right, they are more likely to vote for the PVV.
The second complementary factor that will be discussed is perceived party positions. As stated before, the explanatory factor of the salience-argument rests on a crucial assumption: namely that voters can accurately identify a party’s position in a two-dimensional policy space (Devine, 2015: 509; Steiner & Hillen, 2018: 4;Lefkofridi et al., 2014: 69; Kurella & Rosset, 2017: 2). But what if a left-authoritarian voter perceives the position of a party in one of the two dimensions in a way that is more similar to their own stances? For such a voter, this would be an attractive proposition, because it will make their dilemma about which party to vote for effectively non-existent: if a party is perceived to be close to their policy preference on both dimensions, they suddenly don’t have to prioritize one of their stances over the other. The concept of perceived party position works on both the supply and demand -side of party politics. On the supply-side political parties strategically emphasize one of their ideological dimensions (for example their immigration policies), while being deliberately vague about the other dimension (in this case their economic policies) in order to attract or to not alienate left-authoritarian voters (Rovny, 2013; Kurella & Rosset, 2017: 3; Steiner & Hillen, 2018: 6- 7;). Rovny (2013) has shown that there is an incentive for right-wing populist parties for this kind of ‘position-blurring’. They are known for their strong stances on cultural issues such as immigration and can therefore avoid taking a clear stand on economic issues (Rovny, 2013: 1 - 3). Also, they can often ‘politically afford’ to take a more left economic stance, since it is not their core issue. This way they can bind working-class voters with left-authoritarian views to their party (Harteveld 2016; Lefkofridi & Michel 2017; Hillen & Steiner, 2019; Kurella & Rosset, 2017). Obviously, the same tactic could theoretically work for a left-liberal party who blurs its stance on immigration. On the demand side this might results in voters who will gladly assume that a party hold views similar to their own, without having the actual proof for it. One can almost see this as a form of cognitive dissonance: since the voter can’t have the policy set she would ideally vote for, she convinces herself (with some help from position-blurring of parties) that the party she is contemplating on voting on probably wants the same as she does.
There are two problems with the concept of perceived party position that make researching it specifically difficult. Firstly, the way perceived party position interacts with salience. Simply put, it can be expected that the more one-dimension matters to a voter, the more important it becomes that this voter perceives the party of their
choice to keep similar positions. Let’s take for example a left-authoritarian voter who cares deeply about immigration issues. For her to even consider a left-liberal party would be a far stretch, unless she misperceives their stances on immigration. However, if this voter has low salience on immigration, it would be easier for her to vote for a left-liberal party: she does not care that much about that dimension anyway (Steiner & Hillen, 2018: 6- 7). Secondly, it is very difficult to truly establish the causal processes between the supply and demand side. Take for instance the same left-authoritarian voter who is thinking about voting for the PVV and who also believes the party’s economic views are close to hers. Is the reason she is projecting economical left-wing policies on the PVV due to exogenous factors (for example effective position-blurring from the PVV) or because of endogenous factors (she herself believes the PVV to hold left-wing economic policies for an other unknown reason) (Steiner & Hillen, 2018: 23)? These two problems should be kept in mind during the analysis and will be reflected upon in the discussion. With that being said, the following hypotheses are expected in regard to perceived party positions:
Hypothesis 3: Left-authoritarians are more likely to vote for the SP, when they perceive this party to be authoritarian on the cultural dimension.
Hypothesis 4: Left-authoritarians are more likely to vote for the PVV, when they perceive this party to hold economically left positions.
3. Case study - The SP and PVV
These hypotheses will be tested with a case study of the electoral landscape of the Netherlands. The focus of this thesis will be on two parties which are perceived to be an appealing choice for left-authoritarian voters, being the socialist left party the SP and the populist right party PVV. The reason these parties are most suitable for this analysis is that they both are very outspoken on one of the axes (respectively stricter immigration laws or economical redistributional policies) while also being and (deliberately) vague on their position on the other axe. This puts them in an ideal position to appeal to left-authoritarians who invariably have to make a trade-off on one of the axes.
To start with the SP, which was founded in 1971 as a Maoist splinter party and currently occupies the most extreme left-wing position on the economic dimension in the Dutch party system (Pelikaan et al, 2007: 289) The SP has been consistent in their left-wing economic positions, with strong anti-EU stances, an anti-elitist sentiment and an appeal to popular sovereignty (Otjes & Louwerse, 2015: 66 - 67; Veul et al. 2009: 28; De Lange & Rooduijn, 2011: 324; Van der Meer et al., 2013, 10). This makes the SP a suitable candidate for the left-authoritarian voter who is mainly concerned with socio-economic policies. However, their stance on migration issues has changed over the years. Although currently immigration issues are mostly divided pretty clear over the traditional dominant left-right dimension of politics (Van der Brug et al., 2009: 5), this has not always been the case. In the 1960s, when immigration was a relative ‘new’ issue, some left-wing parties and unions opposed new waves of immigration because they perceived it to be a strategy to keep the wages of working-class people down (Van der Brug et al., 2009: 2). Even though in many political analyses the SP is seen as an outlier when it comes to their stance on immigration compared to other left-wing parties, an analysis of party manifestos from 1994 up to 2006 did not support the claim that the SP is stricter on immigration (Van der Brug et al., 2009: 18). Between 1994 and 2006 the SP only used an average of 2,7 percent of sentences in their party manifesto on immigration (compared to almost 15 percent by the PVV) and did not have an outspoken anti-immigration stance (Van der Brug et al., 2009: 16). Internally, however the SP has been reported to be divided over this issue (NRC, 2018; Trouw, 2019). Since Lilian Marijnissen took the lead of the party in 2018 a change of trajectory can be seen: the SP is now making more forceful statements on refugees and migration and is taking a semi-positive stance on so-called ‘refugee-deals’ with Northern African countries (Trouw, 2009). Internally there is much debate on this new stance, where a very sizable group openly disagrees and is actively speaking up to urge the leaders of the SP not make their party part of the ‘Fort Europe’ parties (NRC, 2018). One can see how for a left-authoritarian voter this might be an appealing party to vote for: the SP has a long and consistent track record on advocating for left-economic policies and while their views on migration are changing and perhaps not most clear, it could be precisely this vagueness that makes a left-authoritarian voter believe the SP wants the same as she wants.
The second party in this analysis will be the PVV, which is regarded to be a right-wing populist party (Veul et al., 2016: 28; Lucardie, 2007). The PVV started when the party founder Geert Wilders decide to split from the liberal party the VVD in 2004. Since then Wilders has profiled the party with a very dominant focus on being strict on immigration and criticizing the ‘progressive elite’ for their neglectance to the threat of Islam (Van der Brug et al., 2009: 16; Otjes & Louwerse, 2015: 67). Where the PVV has a very clear profile on immigration, the economic stances of the party have changed over time. In the first year of the party (2006) Wilders was a supporter of more neo-liberal economic measures such as reducing welfare state spending (Otje & Louwerse, 2015: 67; Lucardie, 2009). This changed after the parties first electoral win: they moved away from their original neo-liberal position and advocated for more public spending on elderly care, opposed raising the age of retirement and positioned themselves against labor market liberalization (Otje & Louwerse, 2015: 67). With some of these economic stances, the PVV can be considered to sometimes position themselves centrist or left of center on certain socio-economic issues. This makes the PVV an appealing party for left-authoritarian parties, who like their unchanged anti-immigration stance and are charmed by the more left socio-economic policies in for instance elderly care.
In the 2017 election - which was the most recent national election in the Netherlands and thus the focus of the NKO 2017 data set that will be used for this analysis - the party manifestos of both parties confirm the aforementioned policy sets of the PVV and SP. In 2017 PVV famously published a party manifesto of one single page, which focuses on the ‘de-islamification of the Netherlands’. Although it is tempting to disregard such a manifesto, it does show the PVV’s stark focus on migration issues combined with some very specific left-wing economic proposals such as the abolition of deductibles in health care, lowering the age of retirement and an investments in eldercare (PVV, 2017: 1). SP did produce a traditional party manifesto, in which it showed its priorities to consist of several left-wing economic policies such as a the introduction of a ‘National Healthcare Fund’, more investment in healthcare, the abolition of deductibles in health care, lowering the age of retirement and higher taxes for multinationals and millionaires (SP, 2017: 6). Immigration did not take as much of a prominent role compared to their economic policies, but it was mentioned in their top priorities while propagating regional solutions for refugees and advocating for regulations on the influx of migrants (SP, 2017: 6).
Descriptive findings of the NPO 2017 data set confirm these theoretical indications of SP and PVV being a good case study for this analysis. Figure 2 shows a set of scatterplots that display how the whole electorate and left-authoritarians in particular perceive the positions of SP and PVV on the two-axes. This figure shows that for the electorate as a whole the position of SP on economic stances is relatively clear, while their position on immigration is perceived quite diverse. With the PVV one can see the same pattern, but even more distinctively: their position on immigration is very clear while their perceived position on economic issues is quite diverse. When focusing only on the group of left-authoritarian, the same pattern is visible. This underline the theoretical expectation that due to (deliberate) vagueness on either one of the two ideological axes, a voter can perceive a PVV and SP to hold vastly different views. These descriptive findings also underline that this ‘wiggle room’ in perception will only take place on one of the axes: PVV it is most likely to be ambiguously perceived by left-authoritarians on economic issues, while SP is most likely to be perceived ambiguously by left-authoritarians on immigration issues.
2: P erc ei ved p o si ti o n o f S P ( in red ) a n d P V V ( in b lu e) i n t h e tw o -di m ens ional pol ic y spac e. e le ct o ra te i s to p t w o f ig u re s, l ef t-au th or it ar ian s is b ot tom t w o fi gu res
4. Data and methods
4.1. Data & variables
In this section the data and methods that are used to test the four hypotheses outlined in the theoretical framework will be discussed. In this thesis use is made of data of the LISS (Longitudinal Internet Studies for the Social sciences) panel administered by CentERdata (Tilburg University, The Netherlands). To be more precise: the data set of the Dutch Parliamentary Study (NKO, 2017) that has been running since 1968. Its most recent wave of 2017 will be used. The NKO 2017 data set provides a rich set of questions regarding the issue position of respondents and their political preferences. The main independent variable of this thesis is ideology (the four quadrants of the political spectrum, which left-authoritarian is one of) and the main dependent variable is the probability of future vote for PVV or SP. For measuring a respondents’ probability to vote for respectively SP or PVV, questions S092 and S093 were used from the NKO 2017 data set. These questions asked: “Would you please indicate on a scale from 1 to 10 how probable it is that you will ever vote for each party? On this scale, 1 means that you will never vote for this party and 10 means that you will certainly vote for this party sometime.” In regard to the decision of using probability to vote as dependent variable, it is important to note that there is of course a difference between respondents’ answers to the question if they would potentially vote for a party and how they actually vote. In this thesis ptv scores will be used as the main dependent variable instead of actual vote in the 2017 election because of two reasons. Firstly, the use ptv scores allow for a multivariate analysis, in which the change of each increase on the interaction variables on the ptv scores can be observed. In contrast with the logistic model, which would be necessary for measuring actual voting behavior. Secondly, while using ptv scores allows for incorporating the whole electorate in the analysis, the binary option of actual voting behavior (voted for SP/ did not vote for SP) will give a much smaller subset which could skew the results. However, the actual vote in the 2017 election has been controlled for as a robustness check.2
2In appendix 3 one can find descriptive findings on the voting behavior of left-authoritarians compared with the
other three quadrants and the rest of the electorate as a whole. These findings confirm the popularity of PVV, SP and CDA for left-authoritarians: with 20,5% PVV is the most common choice for left-authoritarians, while the SP and CDA share the second place with 17,5%
The main independent variable of this thesis is ideology: where do respondents place themselves in the two-dimensional policy space? For measuring a respondents’ views on socio-economic issues, question S132 has been used, which states: “The government should take measures to reduce differences in income levels”. For measuring respondents’ placement on migration issues, question S156 was used, which states: “The Dutch culture is threatened by immigrants”. Both questions are measured on a 5-point scale, ranging from 1) Fully agree, 2) Agree, 3) Neither agree nor disagree, 4) Disagree, 5) Fully disagree. These two questions were selected because these are used in similar research (Van der Meer et al., 2018; Van der Brug & Spanje, 2009; Steiner & Hillen, 2018) and are thus believed to represent the two dimensions quite well 3. With the use of these two variables five groups can be made: the four quadrants and a neutral category (respondents who answered neutral to either one or both of the two questions). Positive responses are considered “Fully agree” and “Agree”, neutral responses are considered “Neither agree nor disagree” and negative responses are “Disagree” and “Fully disagree”. This results in the distribution of respondents as illustrated in figure 3, which shows that the left-authoritarian group in the Netherlands consists of 21,3% of the electorate.
Figure 3: Economic and cultural stances of Dutch electorate
In order to test the main hypotheses, left-authoritarianism is interacted with left-right self-placement (H1 & H2) or party perceptions (H3 & H4). For H1 and H2 an additional independent interaction variable will be added for left-right self-rating for which question V133 will be used, which states: “When you think of your own political
3Question S132 addresses both the level of redistribution and the level of active government involvement, while
S156 does not only shows a respondent’s attitudes against immigration but also addresses feelings of loss of one's own culture.
beliefs, where would you place yourself on this line? Please tell the number that applies to you.” This is measured on an 11-point scale, with 0 being “Left” and 10 being “Right”. The interaction variable is constructed as: dummy left-authoritarian * left-right self-rating. For H3 & H4 additional independent interaction variables will be added respectively for the perceived position that SP holds regarding foreigners or the perceived position of PVV on income difference. For the first question V114 was used: “Where would you place SP on a line from 1 to 7, where 1 means preservation of own culture for foreigners and 7 means that they should fully adapt?”. For the latter question V092 was used, stating: “Where would you place PVV on a line from 1 to 7, where 1 means differences in income should be increased and 7 means that differences in income should be decreased?”. The interaction variables are constructed as: dummy left-authoritarian * perceived position SP on immigration/ perceived position PVV on income difference.
4.2. Control variables
A standard set of socio-demographic control variables for the multivariate regression analyses was used in this thesis, which are based on similar research (Steiner & Hillen, 2018: 14; Lefkofridi et al., 2013: 74; Hillen & Steiner, 2019: 18). They include age, gender, current employment status (worker vs. non-worker), education level (divided into three categories: low, middle, high), self-rating of social class (high to low), religion (religious vs. non-religious) and level of urbanization. For some of these control variables, dummy variables were created. Gender (question V010) already was a two-category variable, but for clarity it was recoded into female= 1, male=0. Current employment status (question V355) was recoded into 1= employed, 0= not-employed. For highest education of respondent (question V368) three dummy variables were created for low, middle and high. The first category was not included in the analyses and thus functioned as a control group. Religion (question V350) got recoded into 1= religious, 0= not religious 4.
4Although level of religiosity could provide a good additional control variable, the question which should be used
for this (question V352: “How often do you generally attend church, the mosque or religious services?”) did not have that many respondents and made the n of the regression analyses drop from n= 1842 to n= 693. Hence it was decided not to include level of religiosity.
Since the premise of this thesis is that these two factors (left-right identification, perceived party position) are an addition to the most prevalent theory that left-authoritarians base their constrained voting decision mostly on salience (Lefrokidi et al., 2014), control variables for both salience of economic issues and salience of immigration will be used in these analyses as well. For constructing these control variables, question V061 was used, which asks respondents to name the most important national problem from a set of standardized answers. For salience of migration, a dummy variable was created which is 1 for respondents who placed “minorities” as the most important national problem and 0 for the rest. For salience of economic issues, a dummy variable was created which is 1 for respondents who identified either “economic/ financial situation”, “social security”, “income/price level/taxes” or “employment” as the most important national problem and 0 for the rest. This resulted of groups sizeable enough to use in the analyses: 20 percent of voters placed immigration as the most important national problem and 12,2 percent did so with economic issues. It is important to stress that salience in this thesis is only implemented as a control variable and drawing conclusions from the coefficients of these control variables is not advised. However, as a check, an interaction variable has been created for salience on both axes and a multivariate regression analysis with interaction effects has been conducted for salience. The results can be found in appendix 1 and were not significant.
4.3. Methods
First, whether left-authoritarians are more likely to vote for respectively SP and PVV will be tested with a multivariate regression analyses with socio-demographic control variables. For these analyses a choice was made to begin with comparing the “pure” quadrants with each other (left-authoritarian, left-liberal, authoritarian, right-liberal), allowing for a first exploratory analysis of how left-authoritarians ptv scores compare to the other political dimensions. In these analyses dummy variables have been made for all quadrants and the neutral group is left out and thus functions as a control group. Since these regression analyses showed no large differences is strength and direction, the analyses will from thereon out switch to comparing left-authoritarians with the whole of the rest of electorate (including the neutral group). This simplifies testing the four hypotheses with interaction effects. A robustness check has also been conducted for running a logistic analysis for actual vote in the 2017
election instead of ptv score. This will be discussed more thoroughly in the results section and one can find the results in appendix 2. The results section will continue with testing H1, H2, H3 and H4 for which models with interaction effects were created. This was done with multivariate regression analyses with control variables a set of socio-demographic control variables and salience.
5. Results
In the first part of this analyses, the focus will be on taking a closer look at the relationship between ideology and the probability to vote for SP or PVV. Subsequently, the second part of this results section will focus on the question whether this relation changes once two interaction variables (ideological left-right placement and perceived party position) are added. Throughout this results section descriptive findings will be presented, followed by multivariate regression analyses in order to give the most complete picture of the relationship between ideology and the probability to vote for respectively SP or PVV, taking in mind the two interaction variables.
5.1. Baseline models
Firstly, some exploratory descriptive analyses have been conducted on probability to vote for respectively SP and PVV. In figure 4 one can see simple bar charts of the distribution of probability to vote for SP/PVV for left-authoritarians compared with the rest of the electorate. In line with theoretical expectations one can see that PVV and SP are seen as more of a probable future voting choice for left-authoritarians compared to the rest of the electorate 5. For left-authoritarians’ ptv scores for SP the M = 4,86 and SD = 3,128, while their ptv scores for PVV the M =4,24 and SD = 3,408. For the rest of the electorate’s ptv scores for SP the M = 3,99 and SD = 2,775, while their ptv scores for PVV the M = 2,50 and SD = 2,623.
5Noticeable is that CDA is also very much favored by left-authoritarians. This makes theoretical sense from a policy
Figure 4: Probability to vote for PVV and SP for left-authoritarians (red) and rest of the electorate (blue)
These figures also show that PVV is a very divisive party were a large segment of the electorate say they will “never” vote for them. With left-authoritarians, this percentage is still high, but compared to the rest of the electorate one can also see more positive attitudes towards the party and even quite some respondents saying the will “certainly one day” vote for the party. With the SP a similar trend can be observed: although they are generally perceived more favorable than PVV, this party als0 gets more potential votes from left-authoritarians compared to the rest of the electorate. However, the differences are not that stark for both parties.
Subsequently, multivariate analyses on the relation between ideology and ptv scores were conducted. Table 1 is in essence the baseline model for this whole thesis. For model 1 for the ptv scores of SP the M = 4,17 and SD = 2,880, while for the ptv scores of PVV the M = 2,87 and SD = 2,894. When looking at model 1, which includes all four quadrants, one can observe a certain symmetry in the relationship between ideology and ptv scores.
Table 1: Probability to vote for SP and PVV for the four political quadrants and left-authoritarians
Standard errors in parentheses, * p <0,10, ** p < 0,05, ***p < 0,01
For probability to vote for SP one can see that identifying as left on the political spectrum increases the likelihood to vote for this party (respectively 0,688 points for left-authoritarians and 0,884 points for left-liberals) and identifying as right on the political spectrum decreases the likelihood to vote for this party (respectively - 1,014 for right-authoritarians and -0,634 for right-liberals). With the ptv scores of PVV the direction follows the opposite axe: the groups that are authoritarian have an increased probability to vote for PVV (respectively 1,402 higher for left-authoritarians and 2,272 for right-authoritarians) while the groups that are liberal have a decreased probability to vote for PVV (respectively -0,999 lower for left-liberals and -0,027 for liberals). The coefficients are significant (p < 0,010) for all groups, except for the
right-liberals. This can be explained by the fact that this is the smallest group (n= 94) and this likely would have affected the significance.
Model 2 is a simplified version of model 1: here left-authoritarians are compared with the whole of the rest of the electorate (the three quadrants and neutral categories). However, as one can see the strength and direction of the relationship between the main independent variable and main dependent variable are quite similar. The model 2 shows that left-authoritarians score 0,621 points higher on the 10-point scale of probability to vote for SP and 1.389 points higher on the 10-point scale of probability to vote for PVV compared to the rest of the electorate. The mean and standard deviation of model 2 also resemble model 1 closely: for ptv scores for SP the M = 4,18 and SD = 2,878, while for ptv scores of PVV M = 2,87 and SD = 2,894. Both models are better at explaining the PVV vote (adjusted R-squared of 0,169 for model 1 & 0,092 for model 2) then the SP vote (adjusted R-squared of 0,089 for model 1 & 0,059 for model 2). Thus, for simplification of the multivariate regression analyses with interaction variables, a choice has been made to from now on compare left-authoritarians with the rest of the electorate as a whole.
It needs the be reiterated that choosing to measure the probability of future vote for these two parties instead of the actual vote for these parties comes with its own challenges. While a respondent can rate a probability to vote for all political parties on a 10-point scale, an actual vote during an election can only be given to one party. To control for this, a robustness check has been conducted. In appendix 2 binomial logistic analyses can be found for actual vote for respectively SP and PVV, which showed a similar pattern in terms of coefficient strength and direction.
5.2. Models with interaction effects
After examining the relationship between ideology and probability to vote for respectively SP and PVV, it is time to look at the four hypotheses which will be tested with the use of multivariate regression analyses with interaction effects.
Left-right identification - H1 & H2
To start with some descriptive findings, figure 5 shows the left-right self-placement of left-authoritarians compared with the rest of the electorate. The descriptive findings
as illustrated by figure 5 indicate that left-authoritarians seem to ideologically identify strongly with the most middle segment (22 percent) as is also underlined by the smaller standard deviation of left-authoritarians. For left-authoritarians the M= 5,38 and SD = 2,503, while for the rest of the electorate M =5,10 and SD = 3,43.
Figure 5: Left-right self-rating of left-authoritarians (left) and the rest of the electorate (right)
When looking at the two opposites of the spectrum, one can see that there is not a striking difference between left-authoritarians and the rest of the electorate. On the right-side of the spectrum, it seems left-authoritarians feel more comfortable placing themselves on the more extreme right, but the difference is not that stark. This differs from the findings of Steiner & Hillen (2018) in Germany, where left-authoritarians can be found on the more extremes of the political spectrum. Why this is the case will be discussed more thoroughly in the discussion section, but for now the important takeaway from figure 5 is that there is variance in the group of left-authoritarians when it comes to ideological left-right self-placement. Left-authoritarians self-identify all over the political spectrum - from extreme-left to extreme-right - indicating that there are respondents all over the spectrum of the interaction variable of left-right self-identification.
Subsequently, multivariate regression analyses were conducted to test H1 and H2, which one can find in table 2, which shows the difference in the effect of left-right self-rating on the probability to vote for respectively SP and PVV between left-authoritarians and the rest of the electorate. Since coefficients and standard errors are conditional (Brambor et al., 2005) a visualization of the interaction effect of left-right self-identification has been created as can be observed in figure 6.
Table 2: Multivariate regression models with interaction effects for left-right identification
Standard errors in parentheses, * p <0,10, ** p < 0,05, ***p < 0,01
Table 2 shows that, although most of the main effects are significant (p < 0,01), the interaction effect for left-right self-identification were not significant for ptv scores of SP and for ptv scores of PVV. Figure 6 visualizes these findings and there are two main things that stand out from these visualizations.
Figure 6: Visualization of interaction effects of left-right self-rating for probability to vote for SP and PVV
Firstly, note that the two figures go in opposite directions. As was expected, the more a respondent identifies with the right the more likely they will be to vote for PVV, and vice versa: the more a respondent identifies with the right the less likely they will be to vote for SP. Secondly, one can see that the lines of left-authoritarians and the rest of
the electorate mostly go parallel. Indicating that there is a difference between left-authoritarians and the rest of the electorate, but that it is rather small and not significant. This mainly shows that, indeed left-authoritarians are more prone to vote for SP and PVV than the rest of the electorate in a general sense. But this is true despite how they identify on a left-right spectrum. When adding left-right identification as an interacting variable one can see that the difference between left-authoritarians and the rest of the electorate is futile. There is one small exception, which is that the lines of left-authoritarians and the rest of the electorate are overlapping at the most extreme left for the ptv scores of SP. This indicates that if one identifies as extreme left, one is always likely to vote for SP no matter what policy preferences one subscribes to.
These findings indicate that there is no support for H1 and H2 suggesting that right self-placement does not significantly affect the likelihood of left-authoritarians to vote for respectively SP or PVV more than it does to the rest of the electorate. But that leaves us with the question why this is the case in the Netherlands, since Steiner and Hillen (2018) did find an effect of left-right self-placement in Germany. Firstly, this could indicate that the terms of “left” and “right” are more slippery and vague in the Netherlands then in Germany. This is in line with literature on how left and right identities are notoriously vague in the Netherlands (De Vries et al., 2013) and could explain this difference in outcome. Additional to this general statement, it could be argued that it makes sense that left-authoritarians place themselves more in the middle. Since they always have to make a trade-off in their voting decision they might feel especially less at home in the traditional “left” and “right” group identities. Besides this, it should also be kept in mind that with the way left-right self-placement is measured in this thesis extremity (placing oneself at the far left or far right) does not necessarily reflect the intensity of how one feels connected to these identities. This brings us to the issue of salience, which keeps being a tricky concept in this matter 6. It could be possible that salience plays an additional interacting factor in the relationship between ideology and probability to vote and that the causal relationship is actually a three-way interaction: if you find economic policies most important, you would always identify more with the left and vice versa. More reflection on the limitations of the use of salience can be found in the discussion, but
6 In table 2 the control variables for salience were only significant for ptv scores for PVV with salience of
immigration, suggesting that even checked for ideology, left-right and interaction effect there is an almost 1-point increase of ptv scores for PVV when immigration is regarded to be the most salient issue.
to better understand this the relationship between salience and left-right self-placement a simple bivariate correlation analysis has been conducted as can be seen in table 3.
Table 3: Bivariate correlation on the effect of salience of immigration/economics on left-right self-placement
The coefficients in table 3 provide a quick preliminary glance at the relationship between salience of the two dimensions and right self-rating for both left-authoritarians and the rest of the electorate. These show that left-right self- placement seems to be affected by how salient people find immigration (0,139/0,212 points towards the right side of the political spectrum when immigration is perceived as the most urgent problem) but not by salience of economics which is not significant. These findings seem to indicate that left-right placement is more influenced by salience of immigration then salience of economics. However, this is just a quick preliminary test to get some grip on how left-right ideological placement works in the Netherlands. More in-depth research should be done to truly confirm this, this is however beyond the scope of this thesis.
Perceived party position - H3 & H4
The second set of hypotheses that will be tested are H3 and H4, which focus on perceived party position. Figure 2 already presented descriptive findings that showed that PVV it is most likely to be ambiguously perceived by left-authoritarians on economic issues, while SP is most likely to be perceived ambiguously by left-authoritarians on immigration issues. Moving forward on this, a multivariate regression analyses with interaction effects and a set of control variables has been
conducted to test the effect of perceived party position as an interacting variable which the results can be seen in table 4.
Table 4: Multivariate regression models with interaction effects for perceived party position of SP and PVV
Standard errors in parentheses, * p <0,10, ** p < 0,05, ***p < 0,01
Table 4 shows that for ptv scores for SP both the main effects and the interaction effect are not significant and rather small. The coefficient of the interaction is 0,066, meaning that left-authoritarians with every step to the right on how they perceive SP’s party position on immigration score 0,018 points (-0,048 + 0,066) higher on ptv scores of SP. Simply put: not a lot is happening. For ptv scores for PVV the two main effects are both not significant, indicating that for the rest of the electorate the perception of PVV’s economic stances seems to make little difference in their
probability to vote for this party. However, the interaction effect of perceived party position of PVV is significant (p < 0,01) and rather large. The coefficient is 0,455, meaning that with every step towards a perception that PVV is in favor of decreasing of income differences, left-authoritarians would score 0,439 (0,016 - 0,455) points higher on the 10-point scale of probability to vote for PVV. This confirms H4, that indeed left-authoritarians who perceive PVV to hold left-economic views are more likely to vote for PVV compared to the rest of the electorate 78.
Figure 7 visualizes these findings: the first figure shows how the probability to vote for SP is affected by perceiving the party to hold right-wing immigration stances. Since both lines are basically flat, the results could be bluntly captured as: left-authoritarians (and the rest of the electorate) are generally more prone to vote for SP, but their cultural stances don’t seem to matter at all. This indicates that there is no support found for H3. This is quite an interesting finding, considering that SP is internally divided over the question where to place themselves on cultural issues such as immigration (NRC, 2018; Trouw, 2019). In the light of these results, this debate would seem rather futile. In regard to the question why this the cultural dimension does not seem to have an effect for the ptv scores of SP of left-authoritarians, I can for now only speculate that most of these voters find economic policies the most salient and therefor immigration stances don’t have much effect on their voting decision.
The part of figure 7 which shows how the probability to vote for PVV is affected by perceiving the party to hold left-wing economic stances, is perhaps the most impactful of this thesis. One can clearly see that the effect of perceiving PVV to be left-wing on economic stances is significant and rather large.
7 For both table 2 and 4 it was checked if the interaction effect was significant without the control variables of
salience and the socio-demographic control variables. The results stayed roughly the same, meaning the interaction effect was still insignificant for ptv scores SP without the control variables and still significant for ptv scores PVV.
8The control variables for salience were only significant with salience of immigration, suggesting that even checked
for ideology, perceived party position and interaction effect there is roughly an 1,5 point increase of ptv scores for PVV when immigration is regard to be the most salient issue and roughly half a point decrease of ptv scores for SP when immigration is regard to be the most salient issue. Gender and age are still significant for ptv scores for PVV and self-identification of social class remains a significant control variable for both ptv score
With every step on the 7-point scale towards a perceived more left-wing policy of PVV on economic issues, left-authoritarians’ probability to vote for PVV goes up with almost half a point. Meaning that the difference between the two extremes (perceiving PVV to be in favor of making income differences bigger versus perceiving PVV to be in favor of making income differences bigger smaller) is more than 3 points on the probability to vote for PVV. The figures show that for the rest of the electorate the position of PVV on economic issues seems to have no big effect, while for left-authoritarians this really influences their decision to vote for this party. These findings show support for H4 and indicate that the perception of PVV as a party who presents themselves as in favor of decreasing income differences really does make a difference for left-authoritarians in their constrained voting decision.
The implications of this finding are interesting to elaborate upon. Because, while this suggest that left-authoritarians who perceive PVV to hold left-wing economic issues will be more likely to vote for this party, it is not clear why left-authoritarians perceive PVV to hold these left-wing economic issues. Does it come from the voters themselves? Meaning they already have positive feelings towards the PVV and a) don’t care that much about economic views, and/or b) decide as a sort of cognitive dissonance to perceive the economic views of this party to be in line with their own left-wing believes. Or is this confusion on the economic stance of PVV due to an active strategy from the PVV itself who know left-authoritarians are susceptible to perceived left-wing economic policies? With the use of position-blurring (Rovny, 2013) a party such as PVV can easily appear to be left-wing on economic issues such as better elder care and with this give left-authoritarians who are already receptive the last little nudge, without alienating their right-wing economic voters too much. Or is it a combination of both? The literature around cue-taking shows how voters can make complicated voting trade-offs with limited information with the help of cues from – among other things – political parties (Hobolt, 2007; Sniderman, 1985; Sniderman, 2000). When there is a lack of knowledge or interest from the side of the voter, a party’s endorsement can offer a sort of shortcut for voters to establish their own position without too much detailed information (Hobolt, 2007, Sniderman, 2000). In the case of the PVV this could mean that voters who either don’t know that much about economic policies or don’t care that much for them, would gladly take cues from a party such as PVV that the way this party frames economic issues is the right way and thus will vote for that
party. Perhaps they start to perceive the PVV’s left-wing policies on for example elderly care as part of the ‘normal’ right-wing agenda. This does not only have to be the case for left-authoritarians: it could potentially work on all PVV-voters.
6. Discussion and conclusion
This thesis is an attempt to build on and contribute to the previous research on the trade-off that left-authoritarians have to make in Western-Europe: although their policy stances are widely shared, there are hardly any parties offering this specific set of policy preferences (Van der Brug et al., 2012; Van der Brug & Spanje, 2009; Thomassen, 2012; Hillen & Steiner, 2019). While previous research indicated that salience is the default explanation in regards to how these voters make a suboptimal decision for which party they will vote (Lefkofridi et al., 2014), the goal of this thesis was to investigate the effect of two additional variables which could contribute to better understanding the voting decision that left-authoritarians make. These two additional variables are left-right self-identification and perceived party position, and previous research in the case study of Germany found that these factors influence the trade-off left-authoritarians have to make when they go to the voting booth (Steiner & Hillen, 2018). The analysis in this thesis was based on the NKO 2017 data set, which provides a rich set of questions regarding the issue position of respondents in the Netherlands and their political preferences. Based on their unique position in the Dutch political landscape and the appeal they pose to left-authoritarians, a choice was made to focus on two parties in particularly: SP and PVV.
First, the general relationship between ideology and ptv scores of SP and PVV was tested with multivariate regression analyses, which functioned as baseline model for this thesis. This model showed that indeed, left-authoritarians are more likely to vote for SP/PVV than respectively respondents from the three other quadrants and the rest of the electorate as a whole. Subsequently the relationship between ideology and ptv scores of SP/PVV was tested with the addition of two interaction variables: left-right identification and perceived party position. These multivariate regression analyses with interaction variables, showed that no strong support was found to confirm the hypotheses on ideological left-right-placement. Indicating that left-right self-placement does not significantly affect the likelihood of left-authoritarians to vote for
