Local Politicians in National Legislatures:
How Do They Vote?
A case study of voting behavior in Germany’s 17
thBundestag (2009-2013)
BACHELOR THESIS POLITICAL SCIENCE
Author: Sarah Sramota (s1797034)
First reader: Dr. T. A. Mickler
Second reader: Dr. F. Meijerink
Table of Contents
1. Introduction ... 2
2. Local political experience as a threat to party unity ... 3
3. Research Design and Case Selection ... 8
4. Operationalization ... 10
5. Data-analysis and results... 15
6. Conclusion ... 21
7. Bibliography ... 23
1. Introduction
One of the main roles of political parties is organizing likeminded individuals in legislative institutions, government and society (Saalfeld & Strøm, 2014). In national legislatures, party members are organized in parliamentary party groups (PPGs), that form the crucial link between the electorate, political parties and parliament (Heidar & Koole, 2003). In order for parties to exercise the representative function in national legislatures, their members of parliament (MPs) are supposed to work, act and vote in accordance with the party’s program and goals (Kam, 2014). In other terms: PPG leaders expect their legislators to display a high level of party unity. The concept of party unity refers to ‘the degree to which members of a party are observed to work together in the pursuance of the party’s goals’ (Kam, 2014, p. 399).
Since party (voting) unity is essential for parties to fulfill their main political functions, there is a long-standing academic debate on what mechanisms generate cohesive parliamentary party groups. Four main pathways to party unity have been recognized in literature (see Andeweg & Thomassen, 2011 and below). Despite the incentives to vote in line with the party that these pathways create, PPGs are seldom fully cohesive. Therefore, many scholars have tried to solve the puzzle of what factors motivate legislators to either toe to the party line or cast a dissenting vote.
Generally, being elected into parliament is not a legislator’s first working experience. While some new MPs look back at a successful career in business, education or engineering, others are already experienced politicians when they take place in their seat.1 Each MP arrives in parliament with their own background and a ‘rucksack’ full of personal life and working experiences. Some scholars have already pointed out, that these individual attributes, practices and convictions shape the way how MPs think about decision-making and, consequently, affect their voting behavior in parliament. An example is a study by Tavits (2010), confirming that Estonian MPs with previous local level political experience are more likely to make voting decisions that deviate from their party’s program than MPs that are not experienced in local politics.
This paper further elaborates on the idea that legislators’ prior local political experience affects party voting unity in parliament. The academic relevance of this paper lies in its explicit focus on one specific attribute and the process through which it directly affects parliamentary
1 The main method of data collection is in this research is a content analysis of legislators’ biographies. This
analysis has given me insight into the extensive variety of careers that MPs may have had before being elected into parliament.
voting behavior. Most literature on personal attributes focuses on whether and how MPs mobilize their characteristics and background in election time to earn personal votes in candidate-centered electoral systems and secure a seat in parliament (e.g. Ames, 1995; Crisp
et al., 2013; Gschwend & Zittel, 2015; Moser & Scheiner, 2005; Tavits, 2010; Zittel &
Gschwend, 2008, 2019). Other literature then focuses on the impact of different electoral systems on party voting unity (e.g. Carey, 2007; Heidar, 2006; Sieberer, 2010). Only a few authors have already made a connection between personal attributes and parliamentary voting behavior (see Baumann et al., 2015; Crisp et al., 2013; Tavits, 2009, 2010;). Following Tavits’ (2010) example, I directly link MPs’ prior local political experience to deviating voting behavior in parliament, hoping to contribute to the generalizability of her results across countries with candidate-oriented as well as party-centered electoral systems. The main research question guiding this paper is:
RQ: How does legislators’ prior local political experience affect their voting behavior in parliament?
To address this question, first, some important theoretical concepts are introduced. An in-depth explanation is provided of the mechanisms that are perceived to link MPs’ personal attributes, specifically local political experience, to deviating voting behavior. Second, I proceed with empirical tests. I analyze whether MPs’ political experience on a local level can explain their individualistic voting behavior during roll-call votes and whether the electoral context affects this relationship. The findings, based on two statistical analyses of data from the 17th Bundestag
in Germany show no significant relationship between prior local political experience and voting dissent, nor do they indicate that local political experience affects party voting unity independently of the electoral context.
2. Local political experience as a threat to party unity
Students of parliamentary party unity seek to find out whether and why members of parliament act in line with their party’s program. Some scholars measure party unity as preference unity (see for example Leston-Bandeira, 2003; Zittel & Nyhuis, 2019). This approach focuses on the ideological cohesion within PPGs and measures whether legislators’ preferences and viewpoints are in line with the party program by studying vote explanations or media
statements. Yet, most scholars study party voting unity and measure whether an MP voted according to the party line of their PPG. Whereas this approach facilitates easier and more straightforward empirical measurement, it does not necessarily provide an objective reflection of legislators’ true preferences. Quantitative data on MPs’ individual voting decisions form a good starting point for research on factors that might affect deviating voting behavior in parliament.
The debate concerning why legislators vote according to the program of their parliamentary party group is one of the key puzzles for scholars studying parliamentary politics. Since PPGs in modern democracies tend to be rather cohesive (see Sieberer, 2006, p. 161), an extensive body of literature aims to provide insight into the incentives that MPs have to toe the party line. Andeweg & Thomassen (2011) summarize these incentives into four
pathways to party unity (also see Van Vonno, 2016, p. 44 for a more detailed explanation). The
first ‘pathway’ is agreement with the party’s standpoint on the given issue (Andeweg & Thomassen, 2011). Second, legislators might show party loyalty despite disagreement through an effective process of socialization (Andeweg & Thomassen, 2011; Crowe, 1986; Davidson-Schmich, 2003; Heidar, 2006; Skjæveland, 2001). Third, disciplinary measures might force MPs into voting in line with the party program (Andeweg & Thomassen, 2011; Davidson-Schmich, 2003; Krebhiel, 1993). The last pathway is ‘cue-tipping’: legislators base their voting behavior on issue-specialists in their party (Andeweg & Thomassen, 2011).
Despite the diverse ‘pathways’ that might lead MPs into voting in line with their party’s majority, voting dissent keeps on occurring. Therefore, many scholars have tried to identify factors that contribute to incohesive PPGs. They differentiate between institutional and
individual-level factors.
One of the institutional factors that affect the level of voting dissentin parliament is the method of legislative candidate selection (e.g. Kam 2014; Rahat & Hazan, 2001; Sieberer, 2006). Democratizing candidate selection, Rahat and Hazan (2001) argue, leads to ideologically incoherent PPGs, resulting in a decrease of party voting unity. Furthermore, the country’s political system is perceived to have an impact: Carey (2007) argues that presidentialism disrupts party unity when disputes arise between PPG leadership and the chief executive. The level of party voting unity is also likely to depend on whether a given party is in government or not (e.g. Sieberer, 2006; Carey, 2007) or the country’s electoral system (e.g. Carey, 2007; Heidar, 2006; Sieberer, 2010 and below).
Examples of individual-level factors that may lead MPs into deviating from the party line are their incumbency and seniority (Bowler et al., 1999) and parliamentary career (Bailer
& Ohmura, 2018, Benedetto & Hix, 2007), their position in the party group’s hierarchy (Kam, 2014) or their ideological viewpoint on certain morally charged issues, which depends on their personal background and characteristics (Baumann et al., 2015). More scholars have already started exploring a link between MPs’ attributes, such as personal characteristics and (working) experiences, and their voting behavior in parliament (see Tavits, 2009, 2010; Crisp et al., 2013). In their studies, both Tavits and Crisp et al. operationalize MPs’ personal attributes as their ‘localness’, and more specifically their prior local political experience. Three mechanisms underlie the link between ‘localness’ and individualistic voting behavior in parliament.
First, the ‘localness’ of these legislators is perceived to be a personal vote-earning attribute. PVEA are legislative candidates’ individual characteristics and/or experiences that are useful for them to mobilize during election time as a strategy to gain personal votes (Nemoto and Shugart, 2013; Shugart et al., 2005). Having strong PVEA allows candidates to cultivate personal votes separately from their party affiliation. Mobilizing PVEA is especially useful in candidate-centered electoral systems, such as single-member district (SMD) and open list systems, that encourage the cultivation of personal votes and stimulate electoral competition between individual candidates, rather than between parties (Carey, 2007; Carey & Shugart, 1995; Moser & Scheiner, 2005).
Nemoto and Shugart (2013) and Put et al. (2018) suggest that any objectively observable personal attributes that are useful to mobilize in personalized campaigns during election time could be considered PVEA. Examples include the candidate’s gender (Smith and Fox, 2001), age (Tavits, 2010), union membership (Nemoto & Shugart, 2013) or even their physical attractiveness (Rosar et al., 2008). In addition, a lot of literature on PVEA focuses on local roots/origins of the candidate and their earlier local/lower-level political experience (such as being an ex-city major or an ex-member of the municipal council), more broadly the ‘localness’ of candidates (see for example Shugart et al., 2005; Put et al., 2018; Tavits, 2010). Local PVEA “allow [politicians] to advertise their commitment to representing local needs” (Shugart et al., 2005, p. 446) and allow voters to easily make an informed decision by “provid[ing them] with substantive cues to a politician’s local knowledge and competence to represent them” (Put et al., 2018, p. 2).
The first mechanism that might link MPs’ local political experience to voting dissent in parliament is as follows. Voters value a candidate’s ‘localness’ (a PVEA), resulting in a number of personal votes that contribute to the candidate’s election in parliament. Studies have indeed found that politicians profit electorally from their individualized campaigns in
candidate-centered electoral systems (e.g. Crisp et al., 2013; Gschwend & Zittel, 2015; Moser & Scheiner, 2005; Tavits, 2010). In such electoral contexts, legislators depend on their constituents, rather than their party, for election and re-election (Sieberer, 2010; Tavits, 2009). Office-seeking behavior of legislators creates an incentive to take into account their voters’ interests and preferences (Bauman et al., 2015; Cain et al., 1987) and seek to deliver supporters ‘local public goods’ (Crisp et al., 2004). Therefore, when the situation arises that an MP is cross-pressured between their constituents’ and their party’s interest on a certain issue, the electoral dependence on their constituents might be enough incentive to deviate from the party line when casting a vote regarding this issue (Bowler et al., 1999; Carey, 2007; Sieberer et al., 20102). If members from one party group represent multiple groups in society that have diverse ideological interests, this increases voting dissent (Morgenstern, 2004), because “legislator[s] may vote against [their] party to signal to supporters that de party’s position is not radical enough” (Hix et al., 2005).
Tavits (2010) argues that local ties could lead to more individualism in parliament, without necessarily being responsible for the electoral success of a candidate in the first place (also see Tavits, 2009; Heidar, 2006). This makes local political experience a unique attribute amongst other PVEA: studying its effect cannot be limited by a country’s electoral system.
The second reason for legislators with prior local political experience to deviate from the party line might be their individualistic attitude towards decision-making. Their personal expertise makes them less dependent on the party to come to a well-informed voting decision (Tavits, 2010). They are more reluctant to indifferently follow party orders (Cain et al., 1987), while inexperienced legislators, on the contrary, may depend on the party’s expertise to cast a well-informed vote. This dependence stands on the basis of their further socialization into following the party line (see Jensen, 2000). Party loyalty, which is the result of effective socialization, is considered one of the four pathways to party unity (Andeweg & Thomassen, 2011, see e.g. Crowe, 1986; Heidar, 2006). For MPs with local political experience, their independent and individualistic attitude might stand in the way of an effective socialization process, leaving them less loyal to the party than others.
2 In the same paper Sieberer (2010) fails to explain voting dissent amongst German district MPs by referring to
this mechanism. He concludes that “district MPs do not simply follow a competing local principal but use their higher electoral independence in more complex ways. Such reasons could include own policy preferences or demands from various outside actors like party factions or non-local interest groups” (Sieberer, 2010, p. 494). He calls for further research on these alternative causal mechanisms.
Third, their prior local political experience provides legislators with (non-partisan) social networks they can rely on in the future, should their career in national politics come to an end. Having a back-up option makes them less dependent on their party when it comes to future career possibilities. Not having to fear possible disciplinary sanctions from the party leadership – party discipline is also one of the four pathways to party unity (Andeweg & Thomassen, 2011) – is expected to decrease their motivation to stay loyal to the party line when they do not agree with it in on a certain issue (Tavits, 2010).
In sum, having had previous local level political experience makes legislators less independent on their parties, generating incentives to express individualistic voting behavior in parliament (Tavits, 2010). This leads to my first hypothesis:
Hypothesis 1: Legislators with prior political experience on the local level are more likely to
cast votes that deviate from the party line than legislators without political experience on a local level.
Among scholars, it is a widely spread assumption that a country’s electoral system and electoral institutions affect the level of voting dissent in parliament. More specifically, deviating voting behavior is expected to be higher in electoral contexts where MPs pursue a personal vote than in closed-list systems (Cain et al., 1987; Carey, 2007; Carey & Shugart, 1995; Sieberer, 2006, 2010).Being less dependent on the party for re-election creates an incentive for constituency-oriented and individualistic voting behavior in parliament.
Sieberer (2010) empirically tests this idea in a study on the German Bundestag.3 The
author predicts and finds evidence for a so-called ‘mandate divide’ in voting behavior: MPs with a district mandate are more likely to deviate from the party line than MPs with a party mandate. When generalizing this argument, it indeed confirms that a country’s electoral system and electoral institutions affect the level of voting dissent in parliament. I argue, that when studying MPs with prior local political experience, these difference in levels of voting dissent between electoral systems remain the same as for MPs without this attribute.
When viewing local political experience as a PVEA, one could expect that it affects voting behavior more in candidate-centered electoral systems, such as SMD or flexible-list,
3 Germany’s mixed-member proportional (MMP) electoral system combines a single member district system (SMD) with closed-list proportional representation (PR). With this, studying one country enables comparative research between two separate electoral contexts.
PR. PVEA simply seem more relevant in systems where candidates are (or may be) elected by personal vote. As illustrated earlier, however, Tavits (2010) identifies two mechanisms that might link prior local political experience to deviating voting behavior, without referring to implications deriving from a country’s electoral system. While active in local politics, MPs might have (1) developed an individualistic attitude towards decision-making and (2) built up a local, non-partisan social network. Both make them less dependent on their party and create incentives to deviate from the party line, even in electoral systems that are not necessarily candidate-centered. Therefore, one could argue that prior local political experience may affect parliamentary party voting unity irrespective of the electoral context of the country. This leads to my second hypothesis:
Hypothesis 2: Local political experience has the same effect on voting behavior of district
legislators as it has on voting behavior of legislators in party-list legislators.
3. Research Design and Case Selection
The research design is a case study of voting behavior in one national parliament, using data on one legislative period. A case study of this type allows for an in-depth analysis of MPs’ individual characteristics and detailed study of their vote-specific behavior in parliament given the limited time space. The goal of this case study is to contribute to the generalizability of Tavits’ (2010) findings about the impact of prior local political experience on parliamentary voting behavior of national legislators.
To test the hypotheses, certain criteria apply for the case selection. First, the country has to be an established representative democracy. Second, to test the second hypothesis, the country needs to have a mixed member proportional electoral system. A mixed-member (MM) electoral system combines a single member district (SMD) system with proportional representation (PR). This will allow for comparison of voting dissent between two different electoral systems while studying only one country. Third, roll call votes4 need to be recorded on a regular basis, so that one legislative period provides for enough data.
4 During roll call-votes (RCVs), voting behavior of all MPs is recorded individually. This makes RCVs the only
Germany is an example of an established representative democracy with an MM-proportional electoral system. During parliamentary elections, voters may cast two votes. With the
Erststimme they directly vote for a candidate that runs in their district. With the Zweitstimme
they vote for a party. With this system, half of the MPs are elected directly via personal district votes, the other half is elected via closed party-lists (Roberts, 2006), allowing for comparison between MPs with different mandates.
Furthermore, parliamentary party groups in the German Bundestag are not fully cohesive: the Rice index (average party group unity) during the period of 1987-1990 is 96.33 (Van Vonno, 2016, p. 62).5 Despite the high Rice index, previous studies on party unity in Germany have shown there is statistically significant variance in parliamentary voting behavior (see Becher & Sieberer, 2008; Sieberer, 2010; Zittel & Nyhuis, 2019). These studies on party (voting) unity in Germany focused on the mandate divide in voting behavior and lead to inconclusive results.
The legislative period studied in this paper is the 17th Bundestag, in function between 2009 and 2013. During this period, 651 MPs made a total of 157.236 voting decisions (excluding absences) in 275 roll-call votes (RCVs) (Sieberer et al., 2018d). Note that RCVs are not the default voting method in the Bundestag; they only account for about 5% of all final passage votes. The parliament only votes via roll-call if at least 5% of the MPs have requested this. Therefore, one should take into account that RCV-data most likely are an unrepresentative sample, only consisting of votes with high salience (Sieberer et al., 2018a).
The 17th Bundestag provides for relatively larger set of roll-call data compared to
electoral periods. Aside from this, it may be considered a typical or exemplifying case (Bryman, 2016): results from analyses of the 17th Bundestag should be representative for other legislative
periods in. This is especially helpful for a case study that seeks to contribute to the generalizability of results from another country.
5 A post-1990 Rice index on the entire Bundestag (Eastern and Western Germany combined) is not available
(Sieberer, 2006), but Davidson-Schmich (2003) finds evidence that ‘parliamentary parties in eastern Germany have become almost as strong as those in their western partner legislatures’ (Davidson-Schmich, 2003, p. 98).
4. Operationalization
4.1 Dependent variables
The dependent variable in this study is an MP’s individual voting behavior in parliament. I use data from Sieberer et al.’s (2018c) ‘BTVote - Voting Behaviour’ dataset. It contains a variable that shows how MPs’ voting behavior relates to their party line for every RCV they could have attended: either no deviation, deviation (weak or strong) or absence (excused or unexcused). The cases in this dataset are RCVs, but the subjects of study in this research are MPs. It is necessary to recode the variable into one that represents voting behavior per individual MP.
Sieberer et al. (2018b) differentiate between two types of deviations. Weak deviation means that the legislator voted ‘abstain’, while the party line was ‘yes’ or ‘no’. Strong deviation means that the legislator voted ‘yes’ when the party line was ‘no’ or vice versa. In case of no
deviation, the legislator in line with their party group. Both strong and weak deviations are
coded as deviations. I include weak deviations in the data because by abstaining, the MP voluntarily and actively breaks with their party by placing their lack of opinion above their party’s interests. Absences during votes (either excused or unexcused) are not considered deviations, because I am studying voting behavior and not parliamentary behavior in general. Furthermore, absence during a vote does not necessarily explain how a MPs standpoint relates to the party line.
To fit my research question and hypotheses, MP voting behavior can best be operationalized as the percentage of votes deviating from the party line in the total amount of roll call votes (RCVs) attended by an individual legislator. Using percentages accounts for MPs’ absence during a vote: an MP who attended all 275 RCVs and deviated 10 times gets a lower score (4%) than someone who deviates 10 out of 100 attended RCVs (10%). This allows for even comparison between legislators.
The suggested depended variable, however, is highly skewed and shows high kurtosis (see Figure 1). Furthermore, it has several outliers, with a maximum value of 31,1%. This makes the variable unfit for regression analysis. An alternative is necessary: the original variable is recoded into two new dependent variables; the data-analysis will, therefore, consist of two parts.
The first dependent variable is dichotomous (deviated), with value 1 for MPs who have at least once voted against their own party, and value 0 for MPs who have not cast any deviating RCV. This variable is fit for binary logistic regression, which will allow me to measure how having had prior local political experience affects an MP’s odds (p) to cast a vote that deviates from the party line.
However, when reducing legislators’ behavior during (a maximum of) 275 RCVs into a binary code, an MP who has once weakly deviated from the party line is treated the same as someone who has structurally cast strongly deviating votes. The fact that deviated does not account for the differences in the relative number of deviating RCVs of 455 MPs (69,9% of the total) does not benefit the validity of this measurement of voting behavior. In order to do justice to the variation on the dichotomous variable, I create a second dependent variable.
Variable deviation_logged is the result of a logarithmic transformation of the originally suggested dependent variable.6 A log(10)-transformation brings values in the variable closer to each other, reducing the skew and the number of outliers (Field, 2013). Furthermore, it eliminates all original cases with value ‘0’, since one cannot solve a logarithm for zero. This
6 At first, a square root transformation of the original variable was performed. Contrary to a logarithmic
transformation, this would have allowed me to keep cases with value ‘0’ (no deviations at all) in the dataset. But after the square root transformation, problems with skewness and kurtosis remained, due to some extreme outliers and the high number of MPs who never deviated. Therefore, logarithmic
Figure 1: Distribution of values on the suggested dependent variable
reduces the problem of high kurtosis. While MPs who never deviated from the party line will not be included in the analysis (N = 196), the variable does account for the differences in voting behavior between MPs who have at least once cast a deviating RCV (N = 455). The distribution of values on this recoded variable makes comes close to a normal distribution, making it suitable for OLS-regression analysis.
Table 1: Descriptive statistics of the independent variables
N Min. Max. Mean Std. Dev.
Deviated at least once 651 0 1 0,70 0,494
Proportion of deviations
(logged) 455 -2,44 -0,52 -1,83 0,420
4.2 Independent variables
The main independent variable in this research is MPs’ local political experience that they have built up prior to their election in parliament. Local political experience is coded as dummy variable local_pol; with ‘1’ for MPs who have had prior local political experience, ‘0’ for MPs
without. Data is collected from a content analysis of legislators’ biographies, as documented in Kürschners Volkshandbuch Deutscher Bundestag (Holzapfel, 2011 & 2012)7 These
biographies include information about MPs’ previous working experience as well as their political functions.
By local political experience, I refer to representative political experience on the administrative level of a Kreis or Gemeinde or lower.8 Representative experience entails
membership of communal, democratically elected, legislative organs (e.g. Kreistag or
Gemeinderat) and experience as the head of the local executive organ (e.g. being an (Ober)Bürgermeister, Landrat or 1. Beigeordnete). Other administrative functions, such as
civil servant in a local executive organs (e.g. Kreis- or Gemeindeverwaltung), are not included,
7 I use editions 122 (10 December 2011) and 126 (10 December 2012) of the series. These editions both cover the
17th legislative period of the Bundestag. Biographies of MPs that are not included in the 126th edition, because
they joined the parliament after 10 December 2012, are retrieved from the web archive of the official website of the Bundestag (http://webarchiv.bundestag.de/cgi/archive).
8 Within the Länder, different names are used for essentially equal administrative layers. For example: a Kreis in
Saxony is called a Verwaltungsverband. I am aware of such regional differences and will take them into account during the process of data collection.
for they are not always appointed by (democratic) election and therefore not representative functions. For the Stadtstaaten Hamburg and Berlin, membership of the Landtag is operationalized as legislative political experience. Contrary to Tavits (2010), I will not take into account the geographic location where an MP has gained their local political experience. Two reasons are that (1) electoral districts in Germany are different from administrative districts9 and (2) not every legislator is elected by the district mandate (which is the case in e.g. Estonia).
It takes some time for a (new) politician to internalize their new working environment (such as a city council) and build up a useful social network and reference frame. Therefore,
and in order to prevent an inflation of positive codes, a threshold of 2 years of actual local
political experience is used during the data acquisition process.
In the analyses, I control for two demographic factors, as well as three institutional-level factors that are perceived to affect party voting unity (see Table 2). As discussed in the theoretical section of this paper, there is also a vast number of other factors that are perceived to impact the level of voting dissent in national legislatures. Yet, some factors are unfit to control for in this research: testing their effects either would acquire a comparative or longitudinal research design and/or more time for data collection.
The first control variable is the MP’s sex. Women are likely to develop distinct ideological preferences on morally charged issues, such as abortion and equal rights, that might differ from their party’s viewpoint on these subjects. To represent their female identity, women are expected to deviate from the party line when such issues are at stake in during votes in parliament (Baumann et al., 2015). Smith and Fox (2001) argue that in open-seat election races, women are likely to support and vote for female candidates. In parliament, women do not only represent their own identity, but also the preferences of their female supporters.
Second, age might serve as a means for candidates to accumulate votes as well. Tavits (2010) argues that older candidates might be advantaged over younger candidates, because they may be seen as more experienced. In parliament, older candidates’ experience might translate into incentives to deviate from the party line during a vote. To make the variable fit for statistical analysis, age is operationalized as a categorical variable, dividing MPs into three
age groups (see Table 2). The source for data on MPs’ sex and age is the ‘BTVote - MP
Characteristics’ dataset by Sieberer et al. (2018c).
The third control variable is type of candidacy; district candidates are coded ‘1’, party-list candidates are coded ‘0’. Data on an MP’s mandate is derived from Sieberer et al.’s (2018) ‘BTVote - MP Characteristics’ dataset. Generally, literature on the subject predicts that district candidates are more likely to deviate from the party line than their party-list elected colleagues, due to cross-pressures between their constituency’s interests and their party’s interests (Cain
et al., 1987; Carey, 2007; Carey & Shugart, 1995; Sieberer, 2006, 2010). Becher and Sieberer’s (2008) findings, on the contrary, challenge the existence of such mandate divide in voting behavior of German MPs. Controlling for this variable is necessary in order to test my second hypothesis, that local political experience affects voting behavior of all MPs equally, irrespective of their electoral mandate.
Furthermore, I control for whether an MP’s party is in government (‘1’) or not (‘0’).10 Again, ‘BTVote - MP Characteristics’ dataset by Sieberer et al. (2018) is used to collect data on MPs’ party affiliation. Studies on the relationship between being in a government party and voting dissent have reached mixed conclusions. Carey (2007) expects and proves that, at least in parliamentary systems, party unity in governing parties is higher than in opposition parties. Zittel & Nyhuis (2019), Becher & Sieberer (2008) and Sieberer (2010) expect and find the same for Germany. On the other hand, Sieberer’s (2006) results from a comparative study in party unity in 11 modern democracies question the widespread argument that government parties are better at maintaining unifying party groups than the opposition.
Finally, the effect of MPs’ seniority is addressed. Kürschners Volkshandbuch
Deutscher Bundestag (Holzapfel, 2011, 2012) provides information about the number of
legislative periods MPs have served. Seniority (in legislative periods) is recoded as a categorical variable. In order to maintain equal group size, the fifth category contains all MPs serving their fifth period or more (with a maximum of 11 legislative periods). There are two contradicting assumptions considering the impact of seniority on voting behavior. First, legislators with more seniority may be less likely to break party unity (Bowler et al., 1999), for they have undergone a long process of socialization within their party group (e.g. Crowe, 1986). According to Roberts (1988), on the other hand, MPs who have been re-elected so many times might assume that their seat is parliament is secure. The feeling of electoral security could be an incentive for individualistic voting behavior.
10 This variable is only added to the OLS-regression model. Controlling for it during logistic regression analysis
resulted in a big number of outliers and heavily influenced the model outcome. Therefore, it is left out of the definitive logistic regression model.
Table 2: Descriptive statistics of the independent variables
5. Data-analysis and results
5.1 Binary logistic regression
Table 3 shows the results of the binary logistic regression analysis of an MP’s odds to vote against the party line. I have run all relevant diagnostics, as prescribed by Field (2013). Not more than 5% of the standardized residual scores (ZRE) are higher than 1,96 (or lower than -1,96) and there is no case with |ZRE| >2,58. This indicates that there is no problem regarding outliers. Furthermore, a maximum Cook’s Distance of 0,084 shows that there are no cases that extensively influence the model outcome.
Model I only includes the main predictor of parliamentary voting behavior as studied in this paper: whether the MP has had prior local political experience (‘1’) or not (‘0’). I formulated the expectation that legislators with prior political experience on a local level are more likely to break party voting unity in parliament than legislators without this attribute. The marginal and insignificant (Wald(1) = 1,097; p = 0,59) regression coefficient of 0,093 does not indicate such relationship. Whereas the model fit is significant, the extremely low R2’s shows that prior local political experience alone is not a relevant explanatory factor of dissenting voting behavior for this set of data.
N Min. Max. Mean Std. Dev.
Local political experience 651 0 1 0,58 0,494
Sex 651 0 1 0,33 0,471
Age group 651 1 3 2,33 0,688
Mandate 651 0 1 0,46 0,499
In government party 651 0 1 0,53 0,499
Table 3: Logistic regression analysis of the odds to vote against the party line
Note: binary logistic regression coefficients with standard errors between brackets
***p < 0,001, **p < 0,01, *p < 0,05
Model I Model II Model III Model IV
(Constant) 0,788*** (0,131) 1,122*** (0,155) 2,719*** (0,380) 1.378*** (0.322) Local political experience 0,093
(0,173) 0,199 (0,177) -0,062 (0,255) -0,088 (0,258)
Mandate (ref. = Party-list)
District mandate -0,794*** (0,176) -1,085*** (0,209) -0,932*** (0,280) Local political experience
* mandate
0,507 (0,355)
0,446 (0,361)
Sex (ref. = male)
Female 0,254
(0,199) Age group (ref. = 20-40)
41-55 56+ -0,045 (0,300) -0,158 (0,311) Seniority (ref. = freshman)
2nd legislative period 3rd legislative period 4th legislative period 5th + legislative period 0,080 (0,276) -0,229 (0,048) 0,048 (0,321) -0,591* (0,272) Model fit (𝜒2) 0,289*** 21,109*** 23,150*** 34,841*** -2LL 796,239 775,419 773,378 761,687 Cox & Snell’s R2 0 0,032 0,035 0,052
Nagelkerke R2 0,001 0,045 0,049 0,074
In Model II, the variable mandate is added, that measures whether the MP has been elected directly (‘1’) or via the party list (‘0’). I expectdistrict candidates to be more likely to cast a vote that deviates from the party line, since they are more inclined towards individualistic behavior in parliament. Accordingly, the regression coefficient for mandate should be positive. The model shows a highly negative coefficient: district candidates have significantly lower odds to deviate from the party line during a roll-call vote than their party list-colleagues. This contradicts the widespread belief that voting dissent is higher in electoral systems where candidates are (or may be) directly elected via a personal vote. A study of Lithuania’s mixed-member electoral system by Clark et al. (2008) also finds party-list MPs to be more prone to vote defection than district MPs. The authors suspect that due to certain dynamics in MM-electoral systems, district MPs might not be that beholden to their constituents as theory would suggest.
Controlling for mandate is an improvement of the model fit (𝜒2(1) = 20.820,
p<0.001), but the explanatory power of the model, now 4,5% of the variance in the dependent variable, still remains low. Also note, that adding mandate to the variable increases the coefficient of local political experience, which could indicate an interaction effect between these two variables.
Model III includes an interaction variable of local political experience by mandate. The expectation is that having had prior local political experience affects the voting behavior of all MPs equally, regardless of their electoral mandate. Based on that, the model should predict a coefficient close to zero, indicating no interaction-effect of mandate. Instead, the model shows a high, positive interaction-coefficient.
Please note, that both the coefficients for local political experience (Wald(1) = 0,059; p = 0,808) and the interaction-coefficient (Wald(1) = 2,039; p = 0,153) are not significant. Therefore, no conclusive remarks can be made on whether the effect of local political
experience depends on the mandate of MPs in the 17th Bundestag. Nevertheless, I further elaborate on the interaction-effect, because it is one of the central variables in this research. Appendix 8.1 shows the odds to deviate from the party line for different groups of MPs.
District MPs without local political experience have 66,2% lower odds to deviate than party list-MPs who also lack this attribute, which indicates the ‘reversed’ mandate divide in voting behavior. District MPs who do have prior local political experience have 47,3% lower odds to deviate from the party line than party list-MPs without this attribute, and therefore have 18,9% higher odds to deviate than district MPs without local political experience. Odds for
party list-MPs with local political experience to deviate from the party line, on the other hand, are 6% lower than for their party list-colleagues without such attribute.
If the coefficients for local political experience and the interaction variable were significant, one could conclude that whereas party list-MPs in the 17th Bundestag are
significantly more likely to deviate from the party line during RCVs than district legislators, having had prior local political experience does not positively affect this relation (the coefficient even points towards a small negative effect). District candidates, on the contrary, are far less likely to exhibit dissenting voting behavior than MPs with a party list-mandate, but for them, it does matter whether they have had prior local political experience.
The fact that their prior local political experience seems to affect only district candidates’ incentives to deviate from the party line voting indicates that electoral motives lie at heart of this effect. This is in line with Tavits’ (2010) and Crisp et al.’s (2013) claim that PVEA affect the voting behavior of MPs elected by personal vote. Furthermore, the results dispute the idea that being experienced in local politics is a ‘unique’ attribute that affects MPs voting behavior in national legislatures regardless of the electoral system they operate in. Being politically experienced on a local level does not grant German party list-MPs in the 17th Bundestag the presumed independence and individualism that they would need to deviate from the party line in cases they do not agree with it (see Tavits, 2010).
In Model IV, I further control for sex, age group and seniority. There is no significant effect of the legislator’s sex or age on their odds to cast a deviating RCV. Only seniority,and more specifically the dummy variable for MPs who have been in parliament for five legislative periods or more, gives a significant result. Legislators serving their 5th or more period in
parliament are far less likely to cast a deviating vote than legislative freshmen: their odds are 55% lower to do so compared to the reference category. This finding is in line with Bowler et
al.’s (1999) idea that legislators with more seniority may be less likely to break party unity,
despite the strong local ties they might have (Tavits, 2010).
While Model IV is a significant representation of the data, adding more control variables does not result in a significant change of its explanatory power (𝜒2(7) = 11,691;
p>0,05). Nagelkerke’s R2 indicates that the final model explains only 7,4% of the variance in
the dependent variable. Furthermore, controlling for sex, age group and seniority decreases the effect of local political experience on the odds to deviate during RCVs for district as well as party list-candidates. For MPs in the 17th Bundestag, at least, being experienced in local politics
5.2 OLS-regression
Table 6: Linear regression models of the logged percentage of deviating votes
Note: Unstandardized OLS-regression coefficients with standard errors between brackets
***p < 0,001, **p < 0,01, *p < 0,05
Model I Model II Model III Model IV
(Constant) -1,828*** (0,031) -1,777*** (0,033) -1,768*** (0,036) -1,732*** (0,065) Local political experience 0,005
(0,040) 0,033 (0,040) 0,016 (0,050) -0,030 (0,049)
Mandate (ref. = Party-list)
District mandate -0,166*** (0,040) -0,195** (0,066) -0,113 (0,068) Local political experience
* mandate
0,046 (0,083)
0,027 (0,082)
Sex (ref. = male)
Female -0,056
(0,041) Age group (ref. = 20-40)
41-55 56+ -0,036 (0,059) -0,095 (0,063) Seniority (ref. = freshman)
2nd legislative period 3rd legislative period 4th legislative period 5th + legislative period -0,035 (0,054) -0,007 (0,055) -0,039 (0,066) -0,033 (0,065) Party in gov. (ref. = no)
Yes -0,232***
(0,043)
R2 0 0,040*** 0,041*** 0,113***
Adj. R2 -0,002 0,036 0,034 0,091
Table 5 shows the results of the OLS-regression analysis of the logged percentage of deviating votes from the total votes attended. These models examine what factors affect MPs’ voting behavior, once they have deviated from the party line. Note, that the dependent variable is logged; therefore, the coefficients cannot be directly interpreted as the ‘change in percentage of deviating votes’.
Again, I have run all relevant diagnostics and tested for the assumptions of linear regression (see Field, 2013). While detecting possible outliers, one case has been found with a standardized residual (ZRE) > 3,29. The Cook’s Distance for this case is only 0,23, so it is not necessary to leave it out of the analysis. Furthermore, regression analysis assumes that errors are independent. The Durbin-Watson test measures whether residuals are (un)correlated for any two cases. The result of the Durbin-Watson test is 1,659 for Model IV, which indicates a small positive correlation between adjacent residuals, is an acceptable result. The analysis meets the assumption of homoskedasticity, meaning that residuals are equal for each level of the predictors; in a scatterplot of standardized predicted values by standardized residuals all cases are proportionally distributed. Finally, I assessed whether two or more dependent variables show multicollinearityby using the VIF-statistic (variation inflation factor). With no VIFs higher than 10 and an average of VIFs that is not substantially higher than 1, there is no cause for concern.
Model I shows no effect of an MPs local political experience on the (logged) relative number of votes they deviated from the party line. Having had local political experience cannot explain any variance in the dependent variable (R2 = 0).
In Model II, the controlling variable for mandate is added. In line with the results of the logistic regression analysis, the negative coefficient shows that district MPs deviate less than their party-list colleagues. Model III, furthermore, includes the interaction variable of
local political experience by mandate. A regression coefficient around zero (indicating no
interaction effect) would be in line with the second hypothesis. The model predicts a coefficient of 0,046, pointing towards a very small, positive, yet insignificant (t = 0,558; p = 0,577) interaction effect. Appendix 8.2 shows how b-coefficients for the different groups of MPs are calculated. Local political experience does not affect district MPs’ level of voting dissent significantly more than it affects list-MPs’ voting behavior. Controlling for the interaction effect does not change the explanatory power of the model.
Model IV additionally controls for the MPs’ sex, age group and seniority, getting no significant results for any of these variables. In the first analysis, I found that MPs in the highest
category of seniority (5 or more legislative periods in parliament) are more likely to deviate than newcomers who serve their first period. This was in line with Bowler (1999). The OLS-model does not indicate such effect of MPs’ on their relative number of deviating votes.
Finally, I controlled for whether an MP’s party is in government or not. In line with earlier evidence from Germany (see: Becher & Sieberer, 2008; Sieberer, 2010; Zittel & Nyhuis, 2019) results, legislators of governing parties (CDU/CSU and FDP) deviate relatively less than opposition members. The final model is significant, but barely explains one tenth of the variance in the dependent variable. Controlling for sex, age group, seniority and government
party decreases the effect of local political experience on the dependence variable. One may
conclude that their prior local political experience does not explain why some members of the 17th Bundestag have deviated from the party line relatively more than others.
6. Conclusion
This paper examined whether and how one specific attribute, prior local political experience, affects legislators’ voting behavior in parliament. Data on personal attributes of German MPs and their voting behavior in the 17th Bundestag (2009-2013) have been used to test two hypotheses.
Results of both the binary logistic regression and the OLS-regression analyses did not show an effect of local political experience on voting behavior of MPs in the 17th Bundestag.
The explanatory power of the first model of each analysis is marginal, if not negligible. Therefore, I reject Hypothesis 1, that legislators with prior political experience on the local level are more likely to cast votes that deviate from the party line in parliament than legislators without political experience on a local level.
This research paper does not support Tavits’ (2010) idea that their prior local political experience is one of the factors that explain why certain MPs do not follow their party line during votes in national legislatures. While my results suggest that Tavits’ findings are not straightforwardly generalizable, follow-up research in other countries is necessary before any conclusive remarks can be made about the effect of local political experience on voting dissent in national parliaments. I further suggest taking into account the length of one’s career in local politics during the coding process, instead of creating a dichotomous variable. A categorical variable is more representative of the diversity within MPs and allows one to study whether more local political experience leads to a higher number of deviating votes.
The binary logistic regression analysis indicated that the (very small) effect of local political
experience might depend on an MP’s mandate. Yet, due to very high standard errors and
insignificance of the coefficients for both local political experience and the interaction variable, I cannot make any conclusive remarks about this effect. The linear regression analysis, on the contrary, did not point towards the existence of a clear interaction effect of
mandate on the relation between local political experience and deviating voting behavior. In
sum, the data and model did not provide clear evidence either in favor or against Hypothesis 2, that local political experience has the same effect on voting behavior of district legislators as it has on voting behavior of legislators in party-list legislators.
It remains unclear whether MPs with local political experience are mainly guided by electoral motives or simply have a more individualistic and independent attitude when deviating from the party line. In order to find decisive results, I suggest comparative follow-up research between two countries with respectively a candidate-centered and a closed-list electoral system, where earlier studies already found evidence for a significant effect of local political experience on voting behavior.
Both analyses showed a mandate divide in voting behavior that is opposite to the theoretical expectation; party-list legislators are significantly more likely to deviate from the party line than district legislators, instead of vice versa. This is a surprising discovery that requires further attention. Since literature that explains why party-list MP would be more prone to deviate than district MPs is very limited, I call for follow-up research to elaborate on this phenomenon.
This thesis was introduced with the observation that newly elected legislators enter parliament with a figurative ‘rucksack’ full of life and working experiences, expertise and convictions. Whereas PPGs are expected to act and vote as unitary actors, these personal attributes might create incentives for individual MPs deviate from their party line during votes.
This thesis showed, that when casting an RCV that deviates from the party line, German MPs are not guided by incentives created by the experience they have acquired during a previous career in local politics. Their voting decisions are guided by other factors, either personal or institutional. Finding out what these factors are, requires further scholarly attention. No personal attribute or other factor is expected to fully explain legislators’ dissenting voting behavior in parliament, but each small contribution to the literature on this subject might add a piece to the complex puzzle that scholars are facing.
7. Bibliography
Ames, B. (1995). Electoral Strategy under Open-List Proportional Representation. American
Journal of Political Science, 39(2), 406-433.
Baumann, M., Debus, M., & Müller, J. (2015). Personal Characteristics of MPs and Legislative Behavior in Moral Policymaking. Legislative Studies Quarterly, 40(2), 179-210.
Bailer, S., & Ohmura, T. (2018). Exploring, Maintaining, and Disengaging—The Three Phases of a Legislator's Life. Legislative Studies Quarterly, 43(3), 493-520. Becher, M., & Sieberer, U. (2008). Discipline, Electoral Rules and Defection in the
Bundestag, 1983–94. German Politics, 17(3), 293-304.
Benedetto, G., & Hix, S. (2007). The Rejected, the Ejected, and the Dejected: Explaining Government Rebels in the 2001-2005 British House of Commons. Comparative
Political Studies, 40(7), 755-781.
Bowler, S., Farrel, D. M., & Katz, R. S. (1999). Party Discipline and Parliamentary Government. Columbus, OH: Ohio State University Press.
Bryman, A. (2016). Social research methods (5th ed.). New York: Oxford University Press. Cain, B., Ferejohn, J., & Fiorina, M. (1987). The Personal Vote: Constituency Service and
Electoral Independence. Cambridge, MA: Harvard University Press.
Carey, J. (2007). Competing Principals, Political Institutions, and Party Unity in Legislative Voting. American Journal of Political Science, 51(1), 92-107.
Carey, J., & Shugart, M. S. (1995). Incentives to cultivate a personal vote: A rank ordering of electoral formulas. Electoral Studies, 14(4), 417-439.
Clark, T. D., Martinaitis, Ž., & Dilba, R. (2008). Electoral Mandate and Party Cohesion: Does It Matter in Lithuania? Journal of Communist Studies and Transition Politics, 24(3), 317-337.
Crisp, B. F., Olivella, S., Malecki, M., & Sher, M. (2013). Vote-earning strategies in flexible list systems: Seats at the price of unity. Electoral Studies, 32(4), 658-669.
Crisp, B. F., Escobar-Lemmon, M. C., Jones, B. S., Jones, M. P., & Taylor-Robinson, M. M. (2004). Vote-Seeking Incentives and Legislative Representation in Six Presidential Democracies. The Journal of Politics, 66(3), 823-846.
Crowe, E. (1986). The web of authority: Party loyalty and social control in the British House of Commons. Legislative Studies Quarterly, 11(2), 161-185.
Eastern German state legislatures 1990-2000. Journal of Legislative Studies, 9(4), 88-101.
Field, A. (2013). Discovering statistics using IBM SPSS statistics (4th ed.). Los Angeles, CA:
Sage Publications.
Gschwend, T., & Zittel, T. (2015). Do constituency candidates matter in German Federal Elections? The personal vote as an interactive process. Electoral Studies, 39, 338-349. Heidar, K. (2006). Parliamentary Party Group Unity: Does the Electoral System
Matter?. Acta Politica, 41(3), 249-266.
Heidar, K., & Koole, R. (2000). Parliamentary Party Groups in European Democracies:
Political Parties Behind Closed Doors (Routledge/ECPR studies in European political
science; vol. 13). London: Routledge.
Hix, S., Noury, A., & Roland, G. (2005). Power to the Parties: Cohesion and Competition in the European Parliament, 1979-2001. British Journal of Political Science, 35(2), 209-234.
Holzapfel, K. J. (2011). Kürschners Volkshandbuch Deutscher Bundestag 17. Wahlperiode (122th ed.). Rheinbreitbach: Neue Darmstädter Verlagsanstalt.
Holzapfel, K. J. (2012). Kürschners Volkshandbuch Deutscher Bundestag 17. Wahlperiode (126th ed.). Rheinbreitbach: Neue Darmstädter Verlagsanstalt.
Jensen, T. K. (2000). Party Cohesion. In P. Esaiasson and K. Heidar (Eds.), Beyond
Westminster and Congress: The Nordic Experience. Columbus, OH: Ohio
State University Press.
Kam, Ch. (2014). Party Discipline. In S. Martin, T. Saalfeld, & K. W. Strøm (Eds.), The
Oxford Handbook of Legislative Studies. Oxford: Oxford University Press.
Krehbiel, K. (1993). Where's the Party?. British Journal of Political Science, 23(2), 235-266. Leston-Bandeira, C. (2009). Dissent in a Party-Based Parliament: The Portuguese
Case. Party Politics, 15(6), 695-713.
Morgenstern, S. (2004). Patterns of legislative politics roll-call voting in Latin America and
the United States. Cambridge, U.K.; New York: Cambridge University Press.
Moser, R., & Scheiner, E. (2005). Strategic Ticket Splitting and the Personal Vote in Mixed Member Electoral Systems. Legislative Studies Quarterly, 30(2), 259-276.
Nemoto, K., & Shugart, M. S. (2013). Localism and coordination under three different
electoral systems: The national district of the Japanese House of Councillors. Electoral
Studies, 32(1), 1.
affect the aggregate party vote share: Evidence from the Belgian flexible-list PR system (2003–2014). Politics. https://doi.org/10.1177/0263395718811969. Rahat, G., & Hazan, R. (2001). Candidate Selection Methods: An Analytical
Framework. Party Politics, 7(3), 297-322.
Rosar, U., Klein, M., & Beckers, T. (2008). The frog pond beauty contest: Physical
attractiveness and electoral success of the constituency candidates at the North Rhine‐ Westphalia state election of 2005. European Journal of Political Research, 47(1), 64-79.
Roberts, G. K. (1988). German Federal Republic: Two Lane Route to Bonn. In M. Gallagher, & Michael Marsh (Eds.), Candidate Selection in Comparative Perspective: The Secret
Garden of Politics. London [etc.]: Sage Publications.
Roberts, G. K. (2006). German Electoral Politics. Manchester: Manchester University Press. Saalfeld, T., & Strøm, K. W. (2014). Political Parties and Legislators. In S. Martin, T.
Saalfeld, & K. W. Strøm (Eds.), The Oxford Handbook of Legislative Studies. Oxford: Oxford University Press.
Shugart, M. S., Valdini, M. E., & Suominen, K. (2005). Looking for locals: Voter
information demands and personal vote-earning attributes of legislators under proportional representation. American Journal of Political Science, 49(2), 437-449. Sieberer, U. (2006). Party unity in parliamentary democracies: A comparative analysis. The
Journal of Legislative Studies, 12(2), 150-178.
Sieberer, U. (2010). Behavioral consequences of mixed electoral systems: Deviating voting behavior of district and list MPs in the German Bundestag. Electoral Studies,29(3), 484-496.
Sieberer, U., Saalfeld, T., Ohmura, T., Bergmann, H., & Bailer, S. (2018a). Roll-Call Votes in the German Bundestag: A New Dataset, 1949–2013. British Journal of Political
Science, 1-9.
Sieberer, U., Saalfeld, T., Ohmura, T., Bergmann, H., & Bailer, S. (2018b). The Bundestag
Roll Call Vote Data (BTVote) [Code book]. https://doi.org/10.7910/DVN/RESB5L Sieberer, U., Saalfeld, T., Ohmura, T., Bergmann, H., & Bailer, S. (2018c). BTVote MP
Characteristics [Dataset, V1]. https://doi.org/10.7910/DVN/QSFXLQ
Sieberer, U., Saalfeld, T., Ohmura, T., Bergmann, H., & Bailer, S. (2018d). BTVote Voting
Behavior [Dataset, V1]. https://doi.org/10.7910/DVN/24U1FR
Skjæveland, A. (2001). Party Cohesion in the Danish Parliament. The Journal of Legislative
Smith, E. R. A. N., & Fox, R. L. (2001). The Electoral Fortunes of Women Candidates for Congress. Political Research Quarterly, 54(1), 205-221.
Tavits, M. (2009). The Making of Mavericks: Local Loyalties and Party Defection. Comparative Political Studies, 42(6), 793–815.
Tavits, M. (2010). Effect of Local Ties On Electoral Success and Parliamentary Behaviour: The Case of Estonia. Party Politics, 16(2), 215-235.
Vonno, C. (2016). Achieving party unity: A sequential approach to why MPs act in concert. Political Science Institute, Faculty of Social and Behavioural Sciences, Leiden University.
Zittel, T., & Gschwend. T. (2008). Individualised Constituency Campaigns in Mixed Member Electoral Systems: Candidates in the 2005 German Elections. West
European Politics, 31(5), 978-1003.
Zittel, T., & Nyhuis, D. (2019). Two Faces of Party Unity: Roll-Call Behavior and Vote Explanations in the German Bundestag. Parliamentary Affairs, 72(2), 406-424.
8. Appendix
8.1 Calculation of odds (p) to cast a deviating RCV per mandate-group
Constant (party list-MPs [mandate = 0], no local political experience [local_pol = 0]) = 2,719
𝑂𝑑𝑑𝑠 𝑟𝑎𝑡𝑖𝑜 for party list MPs with local political experience (in relation to 𝑐𝑜𝑛𝑠𝑡𝑎𝑛𝑡) = 𝑒−0,062 = 0,940.
𝑂𝑑𝑑𝑠 (𝑝) = −(1 − 0,940) = −0,06 = −𝟔%
𝑂𝑑𝑑𝑠 𝑟𝑎𝑡𝑖𝑜 for district MPs, no local political experience = 𝑒−1,085 = 0,338. 𝑝 = −(1 − 0,338) = −0,662 = −𝟔𝟔, 𝟐%
𝑂𝑑𝑑𝑠 𝑟𝑎𝑡𝑖𝑜 for district MPs with local political experience = 𝑒(−0,062−1,085+0,507) = 𝑒−0,64 = 0,527.
𝑝 = −(1 − 0,527) = −0,473 = −𝟒𝟕, 𝟑%
8.2 Calculation of OLS regression coefficients per mandate-group
𝐶𝑜𝑛𝑠𝑡𝑎𝑛𝑡 (party list-MPs, no local political experience) = −1,768 Effect (𝑏) for party list-MPs, with local political experience = 𝟎, 𝟎𝟏𝟔 𝑏 for district MPs, no local political experience = −𝟎, 𝟏𝟗𝟓
