1
Scrutinizing Strategy
The effect of coalition expectations on strategic voting
in the Netherlands
Arnal Kuntze
24-03-2016
Student number: 0591939 Supervisor: Dr. Jonas Lefevere
Master Thesis Political Communication Graduate School of Communication University of Amsterdam
2 Introduction
While cycling to the voting booth, Bert Bakker (36) deliberates on which party to vote for during the Dutch general elections. He doubts if a switch from his lifelong favourite, labour party PvdA, is necessary. These last few days he has read several news articles regarding possible future coalitions. According to the latest information the PvdA has absolutely no chance of governing after the elections. Especially the fact that the polls show that his second choice, social democratic party D66, is in a close battle to win a spot in the coalition with the right wing party PVV, his least preferred party, makes the choice even harder. Fifteen minutes and counting before he makes his final call.
This case, while fictional, gives a small insight into how during the last decades the habits of the Western voter have changed significantly. As can be seen from the literature, a growing share of the electorate no longer feels connected to longstanding predispositions based on social divisions or identification with a particular party (Hansen et al., 2011; Irwin & Holsteyn, 2008; Meer et al., 2015; Meffert & Gschwend, 2011; Schmitt-Beck & Partheymüller, 2012). Whereas the classic voter had the tendency to engage in a lifelong relationship with a party, the modern-day voter seems to be more rational when choosing which party or candidate to vote for (Fischer, 2004; Meer et al., 2015). The Netherlands is no exception to this rule of changing voting habits. Whereas half a century ago Dutch voting preferences could largely be explained by social class and religion, nowadays these sociological and structured models lack explanatory power (Irwin & Holsteyn, 2008).
When a voter casts a ballot on a party other than their preferred party based on strategical considerations this is referred to as strategic voting or tactical voting (Blais & Nadeau, 1996; Cox, 1997; Fischer, 2004; Irwin & Holsteyn, 2012; Mc Cuen & Morton, 2010; Meffert & Gschwend, 2011). The importance of this behaviour is that it has a significant impact on the outcome of an election (Hansen et al., 2011).
3 Many scholars have placed this behaviour within a rational choice theory framework (Blais et al., 2006; Cox, 1997; Downs, 1957; McCuen & Morton, 2010; Meffert & Gschwend, 2009). This implies that these voters are using the information available to them to make rational decision based on the costs and benefits of voting for a particular party. This opens up the space for an inquiry into what kind of information this could be, and in what way it could be instrumental in this consideration.
One of the main tasks of a voter in a proportional representation system like the Netherlands is to determine which coalitions are likely to govern after the elections and try to increase the odds of the coalition of their preference (Downs, 1957). They do this mainly on the basis of media coverage based on political polls (Irwin & Holsteyn, 2012). These polls predicts what coalition is likely to be made up after elections. In order to investigate the influence of this media coverage on voters, this study uses an experimental design to present respondents with future coalitions via manipulated news articles (the independent variable) and investigates to what degree these possible outcomes affect the respondent’s vote intention (the dependent variable). For this study, the following research question has been maintained: To what extent does exposure to coalition expectations influence actual strategic voting?
This study will extend the literature on strategic voting in the Dutch context by not only testing the relationship between coalition expectations and voting behaviour but also by introducing the influence of an individual’s least preferred party in this matter. This new insight could help parties to understand how to position themselves in the media and what would be the most appropriate way to react to certain other parties in order to obtain more votes in future elections.
4 Furthermore, several studies have shown the influence of political knowledge on an individual’s political behaviour (Dolan & Holbrook, 2011; Gschwend, 2007; Meffert & Gschwend, 2011). For example it could be said that political knowledge influences strategic voting behaviour in a negative way, as those with the highest political sophistication are less inclined to change (Dassonneville & Dejaeghere 2014). This study will also take into account the moderating role of voter’s political knowledge, by testing the following sub question: To what extent is the influence of exposure to coalition expectations on strategic voting moderated by a voter’s political knowledge and a voter’s political orientation?
In this thesis I will first discuss the literature already written on the subject, from which I will derive five hypotheses. After this, I will elucidate my methodological framework, showing which choices have been made in order to go about answering the research question. I will then turn to the results derived from several different statistical applied to the data derived from the experimental survey that was conducted amongst 225 participants. Finally, I reflect on these results in the conclusion and discussion by reiterating the connection with the literature discussed.
THEORETICAL FRAMEWORK
The Dutch context
This study will examine strategic voting within the Dutch multiparty system. The Dutch electorate is known to drift in their voting choices (Mair, 2008), and nowadays is even considered to be the most volatile in Western Europe (Meer et al., 2015). Distinguishing characteristics of the Dutch political system are the high amount of parties and the relatively open structure of party competition (Mair, 2008). Another important point to take into account
5 when focusing on this context is that the Netherlands is characterized by opposing a ruling minority government. In the Dutch multiparty system the government is expected to possess more than half of the 150 seats in parliament. As in practice no party is able to achieve this amount by its own, this means that parties are forced to rely on post-election bargaining, in order to form a government that is based on a coalition between different parties (Mair, 2008).
Downs’ conditions
One of the first scholars to write on strategic voting is the American economist Anthony Downs. According to Downs (1957) three conditions need to be satisfied for a voter to act in such a strategic manner. First, it has to be known what possible coalitions each party would enter after the elections. Second, an estimated probability of these different coalitions should be known. Finally, it should be clear which different policies these coalitions would implement if they would be in power. According to Downs these conditions cannot be fulfilled within the context of proportional representation systems such as the Netherlands, because precarious post-election bargaining will determine which coalitions are actually formed and which policies will consequently be implemented (Downs, 1957).
This conclusion has been questioned by many authors, including Irwin and Holsteyn (2012). They suggest that adjusting these conditions slightly could make them applicable to a proportional representation system such as the Dutch political system. They rephrase the first condition by stating that voters should have expectations not on the exact coalition that will be formed but rather on what other voters will vote and on what size a party will be after the elections. Results showed that this made the first condition applicable to the Dutch context as Dutch voters in general were able to come up with a fairly accurate estimation on these points (Irwin & Holsteyn, 2012).
6 The second condition focusses on the expectation of a voter in relation to the likelihood of certain coalitions post-election. As Dutch parties will not indicate what possible coalition preferences they might have prior to the elections (Blais et al., 2006), it seems impossible for the Dutch voter to fulfil this condition. Despite this seemingly crucial lack of information, voters have still found ways to come up with fairly correct expectations to anticipate the future government compositions (Armstrong & Duch, 2010). One key source for the Dutch voter are public opinion polls, as it is shown that Dutch voters rely mostly on them to form their expectations (Irwin & Holsteyn, 2012).
Finally, the third condition of Downs is rephrased as voters being able to estimate what compromises a party will make in the coalition bargaining stage of forming a governmental program. This condition for a Dutch voter is not possible to know, as the party elite does not even know what compromises will be made (Fischer, 2004; Irwin & Holsteyn, 2012). On the other side it is shown that if Dutch voters are presented with coalition expectations, they will be able to make optimal strategic voting decisions, beneficial to their own preferences (Irwin & Holsteyn, 2012). This study follows Irwin and Holsteyn’s conclusion that Down’s conditions are in some respect applicable to proportional representation systems.
Strategic voting
Now we have established that strategic voting can occur within the Dutch multiparty system, a closer look into the topic of strategic voting is useful to understand what factors may trigger the electorate to engage in this behaviour. Based on research in Austria, Germany, Israel, New Zealand and the United States the following circumstances have been known to affect and trigger voters to engage or not in strategic voting; an uncertain or expected close outcome of
7 the upcoming elections (Acevedo & Krueger, 2004; Darmofal, 2010, Meffert & Geschwend, 2011), the voter’s belief that the casted vote could play a deciding role (Acevedo & Krueger, 2004; Darmofal, 2010), whether a voter is able to choose a plausible alternative (Abramson et al., 2010), the threshold level to get a seat in parliament (Bowler et al., 2010; McCuen & Morton, 2010; Meffert & Gschwend, 2010), and the expected outcome of the future coalition (McCuen & Morton, 2010). This last circumstance of coalition outcome expectations has also been proven to affect strategic voting within the Dutch multiparty system (Blais et al., 2006; Irwin & Holsteyn, 2012).
Especially this last circumstance is of interest, as this study exposes respondents to different coalition outcome expectations and investigates to what degree these make voters change their vote intention and engage in strategic voting. In order to be able to answer this question a clear definition on strategic voting behaviour needs to be made.
According to the rational choice models, strategic voting occurs based on a voters’ rational short-term instrumental motivations in trying to maximise the utility of their vote. In order to be able to use this concept in practice, Fisher has defined three criteria to align strategic voting1 with a rational choice framework. The first criteria states that a strategic voter has to have short-term instrumental motivations for their vote (Fischer, 2004). This study fulfils this requirement by making use of an experimental design and presenting respondents with different manipulated coalition outcome expectations. Changes in vote intention will therefore be connected to a change in one’s short-term instrumental motivation. This experimental setup will be further discussed in the method section.
1
Fischer uses the more commonly used U.K. term tactical voting in his paper which is a synonym for strategic voting.
8 The second criteria according to Fisher is that a vote can only be considered strategic if it is on a party other than the voter’s first preference (Fischer, 2004; see also Blais et al., 2006; Blais & Nadeau, 1996). Other scholars have noted that a vote on one’s own party could still be considered strategic as it might be made out of tactical considerations (Meffert & Gschwend, 2009), however this study follows Fischer and Blais and Nadeau in their focus on changing voting behaviour. The reason for this is that this focus illuminates how strategic voting can affect electoral outcomes, as an actual change in voting behaviour can be assessed.
Finally, Fischer states that a vote should be consistent with the principal of utility maximization, based on the expectations of the electorate constituency results (Fischer, 2004). He has based this criteria on the UK plurality voting system, which mostly results in a two-horse race as just one party wins all the seats corresponding to that constituency. This implies that a vote can only be considered utile if a voter chooses one of the two parties that have an actual shot of winning that constituency, as that vote could in fact be pivotal. A ballot on a third party with no realistic chances of winning the constituency could be considered a waste as it will not play any part in the distribution of the seats (Fischer, 2004).
For the Dutch fully proportional electoral system this utility is different, the extra barrier of a constituency is not present for a party to enter in parliament (Mair, 2008). The only barrier present in a proportional system is the threshold2 that needs to be cleared for a party to enter
2
The Netherlands officially does not use a threshold, but a so called electoral quota is used. This implies that a party has to obtain a minimum of votes. The quota is established by dividing the total amount of valid votes by the amount of seats in parliament (150). For the last general elections the quota was 9.424.235 : 150 = 62.828 votes needed to enter the parliament.
9 and obtain seats parliament (Gschwend, 20011; Mair, 2008). This implies that only a vote on a party falling short of the threshold could be considered ‘wasted’ (Gschwend, 2011). Therefore, in the Netherlands only a small few of the votes are wasted, as votes are directly translated into seats in parliament (Irwin & Holsteyn, 2012). As for this study none of the votes can be considered wasted, as we only focus on respondents voting on parties present in the Dutch parliament, each of the votes are considered utile.
This research has now established how short-term motivations, not voting on the preferred party and utility-maximizing are essential to strategic voting in a proportional representation system like the Netherlands. We can therefore follow Fischer’s argumentation leaving us with a fairly simple definition of strategic voting: A vote (intention) cast by an individual on a party other than one’s preferred party.
The provided definition enables us to see whether a respondent vote strategically or not, by comparing which party they indicated to be their preferred party at the beginning of the survey, to the party they indicated that they would vote for after each of the provided vignettes. If a difference between the two answers is spotted the respondent has engaged in strategic voting.
Coalition outcome expectations
Many studies have shown that voters within a proportional representation system indeed do engage in strategic voting behaviour (Abramson et al, 2009; Aldrich et al., 2003; Bowler et al., 2008; Bargsted & Kedar, 2009; Blais et al., 2006; Irwin & Holsteyn, 2012; Kedar, 2005; Meffert & Gschwend, 2007, 2011). Furthermore, research has shown that if a voter is presented with sufficient information from political polls, voters will evaluate the coalition
10 options that could be formed after the election (Blais et -al., 2006; Kedar, 2005). These coalition expectations form one of the key incentives for strategic voting in a proportional representation system (Bargsted & Kedar; 2009, Irwin & Holsteyn, 2012; Bowler et. al. 2008).
Zooming in on the Dutch context, Irwin & Holsteyn (2012) have done research in which different coalition expectations were presented to Dutch voters. They found that voters changed their voting behaviour when they were exposed to the different coalition compositions (Irwin & Holsteyn, 2012). One possible incentive for this strategic behaviour could be that a voter is more likely to vote on a party that will join the government (Indridason, 2011). This could be seen as the so called bandwagon effect, which means that an increase in support can be expected of parties likely to be winning the elections (Faas, et al., 2008). The incentive is backed by other authors, who have found evidence that when voters perceive that their preferred party is unlikely to enter the coalition, they are likely to change their vote intention to a more viable alternative (Bowler et al., 2008; Bargsted & Kedar, 2009).
In this study I explore in which way this strategic incentive influences voting behaviour. I have done this by directly incorporating the respondents own preferred party in the coalition expectations presented. The first hypothesis focusses on the absence of the preferred party as well as the second preferred party in the future coalition. The second hypothesis focusses on the absence of the favourite party while the second choice party does have a shot at joining the future coalition. As follows:
11 their preferred party and second choice party will not have a good chance of joining the future coalition, compared to news coverage suggesting both of the parties have a good chance of joining the future coalition.
H2: More voters will engage in strategic voting when exposed to news coverage suggesting only their second choice party will have a chance of joining the future coalition, compared to being exposed to news coverage suggesting both their most preferred parties will have a chance at joining the future coalition.
Role of the least preferred party
Another interesting issue that could influence voters into changing their vote is the expected results of the least preferred party. As to my understanding only one study has focussed on a similar matter, namely the influence of disliked parties to a voters’ vote intention (Bowler et al., 2008). According to Bowler et al. (2008), who studied strategic voting within the New Zealand context, an increase of strategic voting occurs when respondents were confronted with expectations stating that a disliked party had a good chance of winning.
This study will try to integrate the findings found in New Zealand and see how respondents in the Dutch context react to similar expectations. Respondents will be presented with expectations stating that their second preferred party is in a close battle with their least preferred party on who will win and probably entering the coalition after the elections.
I argue that the probability of having the least preferred party joining the coalition will influence respondents to engage in strategic voting and could potentially play a deciding role. It could be argued that an individual could feel threatened by the fear of having a party at
12 power whose political ideals are not line with their own. This leads to the following hypothesis:
H3: More voters will engage in strategic voting when exposed to news coverage suggesting that their second preferred party is in a close race with their least preferred party for the possibility to join the future coalition, compared to being exposed to news coverage suggesting that only their second preferred party has a chance of joining the future coalition.
Political knowledge
In earlier times, Berelson already concluded that when American voters switch their vote, this should be an indication that these voters have a low interest when it comes to politics (Berelson et al., 1963 in Schudson, 2000). These voters were characterised as so called ‘floating voters’. This reasoning affected the normative illusion of these voters as rational citizens, as volatile voters were now characterized as having a low level of political sophistication (Berelson et al., 1963 in Schudson, 2000). A problematic implication though was that these uninterested volatile voters seemed to hold the key when it came to who would win the elections.
Contrastingly, when it comes to literature on the European democracies a more positive perspective on the link between political knowledge can be spotted, as the volatile voter is regarded as someone who is politically sophisticated (Schudson 2000). Switching parties is even considered to strengthen the democratic values underlying an election. Changing one’s vote shows that a person is making an educated decision, making him or her the model democratic citizen as the choice to switch comes from a rational thinking process (Schudson, 2000). Scholars have not yet decided how the floating voter should be described. The issue
13 remains whether the balance of power is in the hands of the calculative and sophisticated voter, who is known for their considered choices, or if the key player in this ever growing spectacle of unpredictable outcomes is the volatile and unsophisticated voter (Dassonnevelle & Dejaeghere, 2014)
One thing that could be argued however, is that either way a person’s political knowledge seems to in some way affect voting behaviour. If we extend this argument it could be reasoned that political knowledge could influence a person to vote in a strategic manner. There have therefore been several studies that have investigated the link between political knowledge and strategic voting. Blais (2006) found no evidence in their study, as no difference in strategic voting was found between the least informed and most informed voters. However, according to Dassonneville & Dejaeghere (2014), a voter on an intermediate level of political knowledge is the most likely to switch parties. They also conclude that voters with the highest level of political knowledge are the least likely to change their vote to another party. Following these results, this study will implement the variable of political knowledge, to assess whether this variable could influence a respondent to engage in strategic voting. The following hypothesis is set:
H4: A voter with a higher level of political knowledge is less likely to engage in strategic voting compared to a voter with a lower level of political knowledge.
METHOD
14 In order to answer the main research question – to what extent does exposure to coalition expectations influence actual strategic voting? – the right sample needed to be obtained. The research units for this study are individuals eligible to vote in Dutch general elections. In order to obtain a useable sample the snowball sampling method was applied (Bryman, 2012). This was done for practical considerations, as it was important to obtain a large amount of respondents in a relatively short amount of time. Respondents were collected via different avenues, in order to try to obtain as diverse a sample as possible. For example colleagues and customers at work were asked, as well as sport teammates, family, friends on social media, students and further connections provided by individuals in each of these specific groups.
A disadvantage of snowball sampling is that a sampling bias can occur due to the fact that the respondents were not randomly selected. Thereby a representative sample cannot be guaranteed, resulting in a low external validity for the results presented in this paper. This is problematic of course. However, the objective of this study was not generalisability per se, but rather to compare the impact of different stimuli on individual voters. In order to account for this sampling bias, this study will talk about the results as applying to this sample only. A follow up study with a representative sample will have to be done in order to say something about the Dutch population as a whole.
In order to obtain a clear view of the demographics of the used sample the first step is to clear the uncompleted surveys out of the data, leaving us with an 83,6% of the surveys (N=188). Out of these 188 finishers, only 159 can be considered as possible strategic voters as they state that they have the intention to vote in the upcoming general elections. The remaining 22 are taken out of the further analysis as they state that they will either not vote (N=9), do not know if they will vote (N=11), or will cast a blank ballot (N=2). Finally three specific
15 outliers, due to a survey flow failure, have been dropped out of the equation. Leaving us with the polished data from respondents that can be considered as possible strategic voters (N=156).
The median age in this study is 30.4 years old, whereas the average age of the Dutch population is 41.3, making the average age of the voting population somewhat higher. Therefore, the data includes an overrepresentation of younger voters, as 76.9% of the respondents turned out to fall in the age group of 18 to 30 (N = 120) and 23.1% fall in the older age group of 31 and over (N=36). Although this does not give a perfect representation of all the age groups, it does indicate how a large amount of the future voters could react to news articles. Further, the female versus male ratio turned out to be pretty good, as 39.1% of respondents were women and 60.1% were men, giving us a total of 61 females and 95 males who took this online survey. This is pretty close to the actual division of 50.5% female to 49.5 % male in the Netherlands. When it comes to the education level of the respondents a bias could be seen. Most of them are highly educated, as over 70% of the respondents possessed an educational HBO level degree or higher (N = 114). This could again be compared to the Dutch population, where the educational MBO level degree is the average. Although, there is a clear educational bias, the results found still reflect how coalition cues could affect a large part of the electorate.
Research design
The main objective of a research design is that the end product it provides gives the researcher the right output data to help him answer his research questions and hypotheses (Bryman, 2012). For this case, the design needs to be able to test how possible coalition outcomes could affect a respondents’ voting choice. In order to tackle this main objective an experimental
16 online survey design has been chosen, as this gives us the opportunity to test the causal relationship between coalition cues and strategic voting behaviour. A great advantage of this format is that it usually provides the researcher the security of a relatively fast response by its respondents (Bryman, 2012). This online experimental survey is no exception to that rule as a total of 225 survey were taken in a twelve day period (December 31st 2015 – January 11th 2016).
This study follows the experimental design used by Irwin & Holsteyn (2012), as they also researched the effect of exposure to coalition expectations on the voting behaviour of respondents within the Dutch context. Their research design consisted of presenting respondents with four scenarios in which hypothetical poll results along with expectations on what likely coalitions would emerge after the elections.
In this study, a similar design was employed. Respondents were first asked to answer the ranking question: Out of the parties seated in parliament, could you rank your personal three preferred parties? Second, the respondent was asked: Out of the parties seated in parliament could you name your least preferred party? This study included all parties present in the Dutch general parliament as possible answer options, namely: CDA, CU, D66, GroenLinks, PvdA, PvdD, PVV, SGP, SP, VVD and 50 Plus. As for the least preferred party choice, respondents could only choose one of the eight remaining parties, as the earlier ranked most preferred parties were not included in the answer options. This ensured us to not have any conflicting data in the scenarios presented to them later on. Only individual politicians who deserted their party but kept their seat in parliament were left out of the equation. This is an improvement in comparison to the Irwin & Holsteyn (2012) design, which left four parties seated in parliament out of the equation.
17 Strengths and weakness of using the Irwin & Holsteyn design
One problem that could be identified in the research design is that the respondents are repeatedly exposed to different news articles with minor manipulations. An issue with this is that the stimuli might lose its credibility, as participants are alerted to the fact that what they are looking at is not real. In this case however, since there are no elections coming up in the Netherlands, it is unavoidable that the articles will to some extent seem fake to respondents. Moreover, if there was an election coming up soon, it is not uncommon for people to be confronted with several news articles prior to a general election in which changing coalition outcomes are presented.
A more serious issue is the effect that the earlier stimuli might have on the reactions to the later stimuli in the experiment. It is possible that, because the stimuli are presented sequentially, he or she might feel that they should also change their vote. This might result in a larger amount of strategic voters, an aspect that needs to be taken into account when viewing the results (Irwin & Holsteyn, 2012). However, as this study is interested in what might trigger a turning point for voters to engage in strategic voting, it was important that all respondents were exposed to all of the different stimuli.
The use of a laboratory experiment also has some major advantages. An experimental research design has proven to be an appropriate tool to investigate independent variables and their effects on dependent variables (Vreese, 2004). Experimental research does well when it comes to internal validity (Bryman, 2012). The use of a laboratory experiment gives the researcher control over the provided information, as it has the ability to isolate specific voters, who are assumed to engage in strategic voting (Mc Cuen & Morton, 2010), this gives
18 researchers control over confounding variables making it possible to establish a casual relation (Morton & Williams, 2010).
This results in the main advantage of the chosen method, namely that changes in individual vote intentions under different circumstances can be observed (Irwin & Holsteyn, 2012). This way we are able to observe when individual respondents engage in strategic voting behaviour, and most importantly it becomes possible to establish the casual relation between the different exposures to coalition expectations and strategic voting.
The scenarios
As in the Irwin & Holsteyn research, this study also presents respondents with four scenarios. One difference is that this study only makes use of hypothetical coalition expectations and does not provide poll results. The following reasoning is provided for this choice. First, the coalition expectations used in this study are explicit enough to provide the respondent with sufficient information to make a strategic voting decision. Secondly, and the most important reason, is that the focus of this study is only on parties indicated by respondents as being either their most preferred, their second preferred and least preferred party of choice. Therefore, it is not necessary for this study to provide specific poll information on the other parties joining the elections.
To give a better understanding of this reasoning the vignettes with the four scenarios are presented in the table below, showing how they only make use of the most preferred, second preferred and least preferred party chosen by each of the individual respondents:
19 Vignette 1
Preferred Party and second preferred party major contenders to form government
With the upcoming parliamentary elections just around the corner, the latest polls have been made public. Both the preferred party and the second preferred party have come forward as serious candidates for the ruling coalition.
Preferred partyleader responded that he was delighted with the news, and indicated that these outcomes are a positive impulse for his party. He also mentioned that his party would gladly take a governing role after elections.
Similarly, Second preferred partyleader also spoke in a positive manner about the possibilities of a governing role for his party. He considered the polls an extra boost for the political course that his party plans to take in the future
Vignette 2
Second fiddle for the preferred party and the second preferred party
The polls indicate that both the preferred party, and the second preferred party will take a bad turn in the upcoming parliamentary elections. A role on the sidelines seems only logical and a place in the opposition thereby becomes inevitable.
The party leader of the preferred party has already indicated that his party does not aspire to take place in the newly to be formed government. According to him the polls give off a clear signal to the party that they need to get their internal affairs straightened out first. Party leader of the second preferred party, replied in a similar vein that their party needs to first take a look in the mirror before they can regain their position in the political foreground.
Vignette 3
The second preferred party favourite, the preferred party drops out
Recent polls indicate that the second preferred party has a good shot at a governing role after elections. Party leader of the second preferred party hopes that after the elections the political message of the second preferred party will be more prominent. He also indicated that he looks forward to period after the elections, during which he plans to take an active role within coalition talks.
However, a less positive reaction could be heard coming from the preferred party camp. Partly leader of the preferred party said that the low amount of seats indicated in the polls for his party provides them no real leg to stand on. His party will therefore not involve themselves in the post-election coalition talks and rather opt for a place in the opposition.
Vignette 4
Two horse race between the second preferred party and the least preferred party
As can be seen from recent polls, a two horse race between the second preferred party and the least preferred party is currently taking shape for the upcoming parliamentary elections. According to the latest count, both parties are heading towards the same amount of seats. Seeing as party leader of the second preferred party and party leader of the least preferred party indicated to be diametrically opposed to each other, which political course will be taken in the Netherlands in the next 4 years will remain unpredictable until after the elections.
From the side line, party leader of the preferred party indicated that the preferred party will in any case not be engaging in this battle for a place in the post-election coalition. With the internal struggles still fresh in everyone's memory, he sees it as highly inappropriate to take a governing role, as he believes it is necessary to first regain the confidence of the voters.
Another difference in contrast to the Irwin & Holsteyn design, is that their design uses prefixed coalition expectations, regardless of the answers provided by the respondents (Irwin
20 & Holsteyn, 2012). As can be seen in the vignettes presented in table 1, this study provides the respondents with personalized scenarios, as earlier choices of respondents are integrated in the scenarios. A disadvantage of this method is that some respondents could be presented with less realistic scenarios, as they may have chosen parties that usually do not join coalition governments in the Netherlands.
The reason why this design does not completely follow Irwin & Hoslteyn, is that the chosen setup brings us another major advantage. The method used ensures us that each respondent – apart from the inserted parties – is presented with the same stimuli, regardless of their party preferences. Therefore a change in voting behaviour seen in a respondent can be related directly to the scenarios a respondent has been exposed to.
The independent manipulations
In the first manipulation the participant was presented with a possible coalition expectation in which both their preferred as well as their second preferred party had a high chance of joining the coalition. This stimuli provides no distinction between the preferred and second preferred party, as both parties have the incentive of an increase of votes by possibly joining the government (Indridason, 2011) and an equal bandwagon effect (Faas et al., 2008) is expected. Therefore, this stimuli could be seen as the control condition, as a vote for their preferred party seems the most logic choice and is expected.
In the second manipulation the opposite information was provided as neither of the two most preferred parties had a chance of joining the future coalition. This presents the participant with an opportunity to change their vote, expecting that respondents would prefer to vote on a party joining the government (Indridason, 2011), as they are likely to change their vote
21 intention to a viable alternative (Bowler et al., 2008; Bargsted & Kedar, 2009). This could be linked to the first hypothesis of this study, which posed that the situation described above could be seen as a situation in which strategic voting might occur.
The third manipulation respondents faced was one in which their preferred party had no possible chance of joining the future coalition. On the other side a positive election outcome was presented for their second choice party, explaining this party would likely join the future coalition. This incentive is perfect for respondents willing to engage in strategic voting, as it fulfils the requirement of Bowler et al., (2008) and Bargsted & Kedar (2009) presenting respondents with their most viable alternative, as well as complying to the incentive of Indridason (2011) of more voters more likely to vote on a party joining the government. From this it was possible to test the second hypothesis derived from the literature, which posed that when the second party had a chance of joining the coalition, strategic voting would be more likely to occur.
In the fourth and last manipulation, just as in the third scenario, the preferred party had no possible chance of joining the future coalition. But an extra incentive to strategic voting behaviour was integrated by entering the least preferred party of each participant into the equation, being in a hypothetical close race for a spot in the coalition with their second choice party. Following Bowler et al. (2008), who reason that the influence of a disliked party could increase strategic voting, this provided respondents with an even bigger incentive to engage in strategic voting. This all was done to test the third hypothesis, which linked the presence of the least preferred party to a voter to engage in strategic voting behaviour.
22 As in the previous section it has been established which independent manipulations are presented to the respondents, I will now shortly discuss the operationalisation and coding of the dependent variable and the moderating variable.
The dependent variable in this study is strategic voting. Following our definition on strategic voting – a vote (intention) cast by an individual on a party other than one’s preferred – the main issue that needs to be established is whether a respondent changes their vote intention when presented with the independent manipulations. Using the ranking question we could establish the most preferred and second choice party of each respondent. By then comparing the most preferred party choice with the vote intention after the manipulations, we could establish if strategic voting occurred. As an example, if respondent A indicated that the CDA was their most preferred party, and chose the CDA as answer after the first scenario, this was coded as the value 0. But if the same respondent, after being exposed to the third scenario, did change their vote to another party, for example the PvdA, this was seen as strategic voting and coded with the value 1. Only one exception was made, as a few respondents chose their least preferred party as vote intention. This answer got the value 9 and the survey was not incorporated in the results, due to the fact that the given answer is contradictory and not realistic with a respondents’ previous answers.
Lastly, the moderating variable political knowledge is presented. Political knowledge should be seen as a relatively unidimensional concept that measures a citizens’ level of factual political knowledge. This factual knowledge can be obtained by using a short series of survey questions (Delli Carpini & Keeter, 1993). Therefore, five political knowledge questions and answer possibilities on Dutch politics were presented (Table 2). Respondents either scored the value 0 if they answered the question incorrectly and the value 1 was given if they answered
23 correctly. Finally, a sum scale was created based on these items, providing us with the total political knowledge for each of the respondents.
Table 2
Political knowledge questions Answer options
(bold answer(s) is/are correct 1. How many members does the first Dutch
chamber possess?
a. 50 b. 70 c. 75 d. 85 e. 90 f. 100 g. 125 h. 140 i. 150 j. 175 k. 200 l. 250 z. Don’t know
2. How many members does the second Dutch chamber possess?
a. 50 b. 70 c. 75 d. 85 e. 90 f. 100 g. 125 h. 140 i. 150 j. 175 k. 200 l. 250 z. Don’t know
3. The odd one out?
Recent and former prime-ministers, except one
a. Mark Rutte b. Jan-Peter Balkenende c. Wim Kok d. Alexander Pechtold e. Ruud Lubbers z. Don’t know
4. Which party proclaims herself against women’s’ voting rights?
a. RPF b. CU c. CDA d. SGP e. SP f. 50 Plus g. VVD
z. Don’t know 5. Which parties form the Dutch parliament at the
moment?
a. CDA b. CU c. D66 d. GroenLinks e. PvdA f. PvdD g. PVV h. SP
i. VVD z. Don’t know
RESULTS
Now, the method has been established, we can work towards answering the main research question – to what extent does exposure to coalition expectations influence actual strategic voting? – In the next paragraphs the hypotheses set will be tested by applying several analytical tests to the four stimuli. This way we can assess to what degree strategic voting occurs when confronted with different stimuli.
Differences in coalition make-up
The first hypothesis that needs to be answered is whether a difference in strategic voting can be seen when the preferred party and second choice are likely to join the future coalition, versus when they are both unlikely to join the future coalition. This resulted in the following hypothesis:
24 H1: More voters will engage in strategic voting when exposed to news coverage suggesting their preferred party and second choice party will not have a good chance of joining the future coalition, compared to news coverage suggesting both of the parties have a good chance of joining the future coalition.
In this hypothesis a comparison is made between the first scenario, in which the expectation is made that individuals’ two most preferred parties will join the future coalition versus the scenario in which these two parties will not join the future coalition. The difference between the two parties in obtaining a seat in parliament stays equal, as either both parties will enter or both will not enter, the relative difference does not change. Therefore, the only influence on respondents to engage in strategic voting is the expectation whether or not their main parties of choice join the future coalition. Following the incentive by Indridason (2011), stating that a voter is more likely to vote on a party that will join the government, we expect to find a relative difference in respondents engaging in strategic voting. Especially in the second scenario, our expectation is that more voters will vote on neither of the two most preferred parties, and cast a ballot on a third party.
In order to address this hypothesis the results of the first two stimuli the respondents faced need to be compared. A paired-samples t-test is conducted to compare the amount of strategic voting between the two conditions are compared (N=153). A significant difference in scores is found between the first (M=0.14, SD=0.35) and second (M=0.25, SD=0.43) conditions; t(152) = -3.38; p = 0.001, 95%CI [-0,17, -0,04]. In the condition in which both join the coalition 14.4% of the respondents (N=22) engage in strategic voting, whereas 24.8% of the respondents vote strategically in the condition where both will not join the coalition (N=38).
25 These result suggest that voters have a higher tendency to cast a strategic vote if their preferred party does not have a chance to join the future coalition in comparison to when it is likely to join the future coalition. This confirms our first expectation, that more voters will cast a ballot on a third party in the second scenario.
When taking a closer look at the amount of strategic voting between the two exposures, another interesting issue can be spotted. When faced with the scenario in which both parties have a chance of joining the coalition, only 14.4% of the respondents will cast a strategic vote on their second choice party and no vote was casted to a third party. Whereas in reaction to the scenario in which both parties will not enter the coalition after elections 11.1% of the respondents vote on their second choice party and another 13.7% of the respondents cast a vote for a third party, combined as a total of 24.8% that can be considered as strategic votes. An increase of 10.4% in strategic votes when compared to the reaction to the earlier stimuli. This result suggest that some voters are not keen on ‘wasting’ their vote, and only consider voting on parties which may have a shot at ruling after the elections, and confirming our second expectation.
In sum, the results given by the t-test analysis provide sufficient proof to accept the first hypothesis. Therefore, it can be concluded that voters are more likely to engage in strategic voting when their preferred party and second choice party are presented as having no chance to join coalition after elections, compared to when they are presented as both having the possibility to join the future coalition.
26 The third scenario the respondents faced was set in order to create a situation in which a higher amount of strategic voting could be expected. Respondents were provided with a news article in which was stated that their second preferred party was likely to join the future coalition and their favourite party would definitely not be part of this coalition. The following expectation was stated in the second hypothesis:
H2: More voters will engage in strategic voting when exposed to news coverage suggesting only their second choice party will have a chance of joining the future coalition, compared to being exposed to news coverage suggesting both their most preferred parties will have a chance at joining the future coalition.
To assess whether an increase in strategic voting would occur, a second paired-samples t-test was conducted, this time applied to the difference between the condition in which both parties were set to join the post-election coalition and the condition in which only the second preferred party had this opportunity (N=147). The results show that a significant difference in scores is found for the first (M=0.14, SD=0.34) and the second (M=0.50, SD=.50) conditions; t(146) = -8.71, p = 0.000, 95% CI [-0.45, -0.28]. These results suggest that whether the preferred party joins or does not join the future coalition has an effect on strategic voting. Specifically, the results suggest that the amount of strategic voting will increase when one’s preferred party has no possibility to join the future coalition and one’s second choice does have that opportunity.
Furthermore, a total of 50.3% of the respondents turned out to vote strategically after stimuli in which only the second preferred party was said to have a chance at joining the coalition (N=74). An increase of 37.7% compared to the earlier stimuli in which both parties joined the
27 coalition, in which only 13.6% of the respondents voted strategically (N=20). According to the results in this sample more than half of the respondents would vote for a second choice party that is likely to join the future coalition, showing that joining a future coalition is an important trigger when it comes to voting strategically. Therefore, the second hypothesis formulated in this paper is also accepted.
The influence of the least preferred party
For the final hypothesis the results obtained from the third and fourth scenario were compared. The last scenario respondents faced was incorporated to assess the influence of the least preferred party into the equation of strategic voting. This effectively introduced an added incentive for respondents to engage in strategic voting, and with this the possibility to test the third hypothesis:
H3: More voters will engage in strategic voting when exposed to news coverage suggesting that their second preferred party is in a close race with their least preferred party for the possibility to join the future coalition, compared to being exposed to news coverage suggesting that only their second preferred party has a chance of joining the future coalition.
Again a paired-samples t-test was conducted to evaluate if the least preferred party can play a deciding factor when it comes to strategic voting. The difference between the strategic voting scores of the respondents after the condition in which only their second choice had a chance of entering the coalition and the condition in which second preferred party was in a close race with the least preferred party are assessed (N=145). According to the results a significant difference in scores is found for the first (M=.50, SD=.50) and the second (M=.61, SD=.49) conditions; t(144) = -3.39, p = 0.001, 95% CI [-0.19, -0.05]. These results suggest that the
28 possibility that the least preferred party could join a future coalition has an increasing effect on the strategic voting of the respondents. According to the data 61.4% of the respondents would vote strategically after being exposed to the final scenario including the least preferred party (N=89), while 49.7% respondents vote strategically after they are exposed to the earlier mentioned scenario in which only the second preferred party has a chance of entering the coalition (N=72). In sum, the incorporation of the least preferred party into the scenarios increases the amount of strategic voters with 11.7%.
These results suggest that having a party that you least prefer possibly entering the future coalition can be seen as the extra incentive for people to engage in strategic voting. As a consequence the third hypothesis formulated needs to be accepted; we accept that more voters will vote strategically if a close race is expected between their second preferred party and their least preferred party.
The influence of political knowledge on strategic voting
In this paragraph the sub-research question of this study is investigated. The focus lies on whether a person’s political knowledge could be a moderator to the behaviour of strategic voting. The following expectation focusses on a person’s political knowledge as it predicts that a person with a higher political knowledge is less inclined to engage in the behaviour of strategic voting. Bringing us to the subsequent hypotheses:
H4: A voter with a higher level of political knowledge is less likely to engage in strategic voting compared to a voter with a lower level of political knowledge.
29 In order to be able to accept or reject this hypothesis a linear regression analysis needs to be conducted. In advance the moderating variable political knowledge, which is used as the independent variable in the equation, as well as the dependent variable strategic voting need to be formed out of the collected data. First, the total political knowledge score for each respondent is set, by adding each of the five knowledge scores answers, giving a total knowledge score on a zero to five scale. Finally, the dependent variable strategic voting in this particular analysis was formed by checking whether a respondent had voted strategically in either one of the four stimuli provided. A positive answer to this question makes the respondent to be labelled as a strategic voter, whereas a negative answer leads a respondent to be labelled as non-strategic voter.
Finally, in order to determine the relationship between a respondents’ political knowledge and strategic voting a simple linear regression was performed. An insignificant result was found between the moderating variable political knowledge and the dependent of strategic voting (F (1,145) =.542, p =.463). This result indicates that no relation is found between a respondent’s political knowledge and their strategic voting behaviour. This implies that an individual with a higher level of political knowledge is not more probable to engage in strategic voting compared to a person with a lower level of political knowledge. Therefore, the study has to reject the fourth hypothesis, concluding that having the level of political knowledge is no indicator for whether an individual will engage in strategic voting.
CONCLUSION AND DISCUSSION
This study adds to the emerging literature on strategic voting within a proportional representation system, as it examines how different coalition expectations influence this particular behaviour within the Dutch multiparty system. In line with many scholars, the
30 results give a clear indication that voters within a proportional representation system are concerned with the role parties will have after the elections, and therefore engage in strategic voting (Abramson et al., 2010; Aldrich et al., 2003; Bargsted & Keller, 2009; Bowler et al, 2008; Blais et al, 2006; Irwin & Holsteyn, 2012; Kedar, 2005; Meffert & Gschwend, 2007, 2011). As for the Dutch context, this study adds to Irwin & Holsteyn’s conclusion that the vote intentions of Dutch voters are likely to change based on which coalition government they expect will be formed after the elections (Irwin & Holsteyn, 2012).
This research follows part of the methodological design introduced by Irwin & Holsteyn (2012). A limitation to the used design is that it confronts respondents with sequential scenarios, making it possible to overestimate the amount of strategic voting as respondents might feel the need to chance their vote intention under different circumstances. On the other hand an advantage is that the design makes it possible to view the vote intentions on an individual level establishing the causal relation between coalition expectations and strategic voting. This study introduces an innovative aspect to the methodological design by integrating the respondents least preferred party in the design. One limitation to this innovation is that certain respondents are presented with coalition expectations which are not very realistic, as their personally chosen preferred, second and least preferred party are integrated into the given scenarios. If these personally chosen parties are small opposition parties they would not have had a chance of joining a future coalition under normal circumstances.
The reason this innovation was added however, because it provides us with a major advantage. By having respondents faced with personal manipulations each respondent is confronted with the same coalition expectations, as the scenarios stay intact and only the name of the parties change according to their pre-established answers. This ensures us that the
31 causal relation between the independent variable coalition expectations and the dependent variable is even stronger, as the independent variable is the only factor changing in the equation for each individual.
Focussing on the empirical data provided by this study, some conclusions can be drawn. The first and second hypothesis set in this study were confirmed, this outcome is in line with conclusions made by several scholars (Bargsted & Kedar, 2009; Bowler et al., 2008; Blais et al., 2006; Irwin & Holsteyn), as they also show that voters are willing to desert their initial vote intention in favour of a more viable option. The conclusion that voters seek to vote for a party that will take a seat in the future government, is highlighted even more, as the first hypothesis showed that 13.7% of the respondents were not only willing to desert their most preferred party for a second choice party, but would even vote on a third choice party. This study therefore agrees with the incentive formulated by Indridason (2011), that voters are motivated to engage in strategic voting in order to vote for a party that is likely to join the government. Apparently the ‘winning’ factor of a party is seen as great incentive for a respondent to change their vote, this confirms the bandwagon effect in which voters support only winning parties. On the other side we also found evidence of voters engaging in strategic voting when their preferred party was winning. This could be linked to studies on pre-electoral-coalitions, which shows that voters supporting larger parties would find it easier to vote for these pre-electoral-coalitions (Meffert & Gschwend, 2011). The fact that these respondents voted strategically could well be due to a personal motivation to guide the larger party into the appropriate coalition direction.
The novel contribution in this study has been the incorporation of the least preferred party in the relation between strategic voting and coalition expectations. To our knowledge the only
32 study focussing on a similar issue are Bowler et al., (2008). They focus on the influence of disliked parties within the New-Zealand setting. The third hypothesis set for this study was accepted; the increase of a further 11.7% in the total strategic voting rate of the respondents could be accounted to the presence of this least preferred party. We conclude that the inclusion of the least preferred party in the coalition expectation could be the factor that could win over a voter to engage in strategic voting. Bowler et al., (2008) also found evidence of an increase in strategic voting influenced by expectations of the disliked party gaining a governing role after elections. We can therefore conclude that having an unwanted party in a future coalition does in fact play a role in the issue of strategic voting within a proportional representative system. The fear of having a party in command you intrinsically do not like, can outweigh the will to vote on your preferred party and change your vote to another option.
When focussing on the sub-question – To what extent is the influence of exposure to coalition expectations on strategic voting moderated by a voter’s political knowledge? – no statistical evidence was found that this variable had a moderating influence on the relationship between exposure to coalition expectations and strategic voting. Therefore, the fourth hypothesis had to be rejected. This study is in this respect in line with the conclusion stated by Blais et al. (2006), which also does not find evidence that a more knowledgeable voter is more likely to use coalition expectations to engage in strategic voting. However, other studies did find evidence that knowledge could play a factor for a voter to switch parties (Dassonneville & Dejaeghere, 2014). Therefore, this study adds to the theory provided on this subject and reasons that more research is needed to exactly conclude if and in what way the influence of an individual’s political knowledge is related to exposure to coalition expectations and strategic voting.
33 Theoretically, this study shows that following a rational choice theory framework can explain why voters are willing to engage in strategic voting based on analysis of the costs and benefits of different votes. This can be made up out of the fact that many of the respondents did not seem to want to waste their vote on a party that would not play a deciding role within the post-election government. However, the findings on strategic voting in relation to the influence of the least preferred party could suggest that voters are willing to change their vote, perhaps based on their ‘fear’ in relation to that party. This should be further investigated as it remains unclear if the choice to engage in strategic voting can totally be subscribed to a rational thinking process or whether an emotional thinking process could be at the basis of this strategic voting behaviour. This question has not yet been investigated within the current literature and is therefore an interesting avenue for future study.
This brings us to the end of this study, but not before knowing what choice Bert Bakker made when casting his ballot. After long and emotional deliberation and after all these years of being a loyal voter to his favourite party PvdA, he decided to try to counter the possible right wing win of the PVV and switched his vote to social democratic party D66. Now Bert Bakker can proudly say that he can be labelled as a strategic voter. Whether this is due to a completely rational choice or this is due to an emotional outcry, only Bert knows.
34 LITERATURE
Abramson, P. R., Aldrich, J. H., Blais, A., Diamond, M., Diskin, A., Indridason, I. H., ... & Levine, R. (2009). Comparing strategic voting under FPTP and PR. Comparative
Political Studies.
Acevedo, M., & Krueger, J. I. (2004). Two egocentric sources of the decision to vote: The voter's illusion and the belief in personal relevance. Political Psychology, 25(1), 115-
134.
Aldrich, J. H., Blais, A., Indridason, I. H., & Levine, R. (2003). Coalition considerations and the vote. The elections in Israel, 180-211.
Armstrong, D. A., & Duch, R. M. (2010). Why can voters anticipate post-election coalition formation likelihoods?. Electoral Studies, 29(3), 308-315.
Bargsted, M. A., & Kedar, O. (2009). Coalition‐targeted Duvergerian voting: how expectations affect voter choice under proportional representation.American Journal of Political Science, 53(2), 307-323.
Blais, A. (2006). What affects voter turnout?. Annu. Rev. Polit. Sci., 9, 111-125.
Blais, A., & Nadeau, R. (1996). Measuring strategic voting: A two-step procedure. Electoral Studies, 15(1), 39-52.
Bowler, S., Karp, J. A., & Donovan, T. (2010). Strategic coalition voting: evidence from New Zealand. Electoral Studies, 29(3), 350-357.
Bryman, A. (2008). Social research methods. Oxford university press.
Carpini, M. X. D., & Keeter, S. (1993). Measuring political knowledge: Putting first things first. American Journal of Political Science, 1179-1206.
Cox, G. W. (1997). Making votes count: Strategic coordination in the world's electoral systems (Vol. 7). Cambridge: Cambridge University Press.
35 Dassonneville, R., & Dejaeghere, Y. (2014). Bridging the ideological space: A cross‐national analysis of the distance of party switching. European Journal of Political Research, 53(3), 580-599.
Darmofal, D. (2010). Reexamining the calculus of voting. Political Psychology, 31(2), 149-174.
Dolan, K. A., & Holbrook, T. M. (2001). Knowing versus caring: The role of affect and cognition in political perceptions. Political Psychology, 22(1), 27-44.
Downs, A. (1957). An economic theory of political action in a democracy.The journal of political economy, 135-150.
Faas, T., Mackenrodt, C., & Schmitt-Beck, R. (2008). Polls that mattered: effects of media polls on voters’ coalition expectations and party preferences in the 2005 German parliamentary election. International Journal of Public Opinion Research, 20(3), 299-325.
Fisher, S. D. (2004). Definition and measurement of tactical voting: the role of rational choice. British Journal of Political Science, 34(1), 152-166.
Gschwend, T. (2007). Ticket‐splitting and strategic voting under mixed electoral rules: Evidence from Germany. European Journal of Political Research, 46(1), 1-23.
Hansen, K. M., Slothuus, R., & Stubager, R. (2011). Late Deciders: Changing Patterns in Which Voters Make Up Their Mind during Campaigns?.
Indridason, I. H. (2011). Proportional representation, majoritarian legislatures, and coalitional voting. American Journal of Political Science, 55(4), 955-971.
Irwin, G. A., & Van Holsteyn, J. J. (2008). What are they waiting for? Strategic information for late deciding voters. International Journal of Public Opinion Research, 20(4), 483-493.
36 Irwin, G. A., & Van Holsteyn, J. J. (2012). Strategic electoral considerations under
proportional representation. Electoral Studies, 31(1), 184-191.
Kedar, O. (2005). When moderate voters prefer extreme parties: Policy balancingin parliamentary elections. American Political Science Review,99(02), 185-199.
Mair P (2008) Electoral volatility and the Dutch party system: A comparative perspective. Acta Politic 43: 235–253.
McCuen, B., & Morton, R. B. (2010). Tactical coalition voting and information in the laboratory. Electoral Studies, 29(3), 316-328.
Meer, van der, T. W., van Elsas, E., Lubbe, R., & van der Brug, W. (2015). Are volatile voters erratic, whimsical or seriously picky? A panel study of 58 waves into the nature of electoral volatility (The Netherlands 2006–2010). Party Politics, 21(1), 100-114. Meffert, M. F., & Gschwend, T. (2011). Polls, coalition signals and strategic voting: An
experimental investigation of perceptions and effects. European Journal of Political Research, 50(5), 636-667.
Meffert, M. F., & Gschwend, T. (2010). Strategic coalition voting: Evidence from Austria. Electoral Studies, 29(3), 339-349.
Meffert, M. F., & Gschwend, T. (2009). Strategic voting under proportional representation and coalition governments: A simulation and laboratory experiment.
Morton, R. B., & Williams, K. C. (2010). Experimental political science and the study of causality: From nature to the lab. Cambridge University Press.
Schmitt-Beck, R., & Partheymüller, J. (2012). Why voters decide late: A simultaneous test of old and new hypotheses at the 2005 and 2009 German federal elections. German Politics, 21(3), 299-316.
Schudson, M. (2000). Good citizens and bad history: Today’s political ideals in historical perspective.Communication Review 4(1): 1–19.
37 Vreese, de C. (2004). The effects of strategic news on political cynicism, issue evaluations,
and policy support: A two-wave experiment. Mass Communication & Society, 7(2), 191-214.
