Abstract Party switching can pose a severe threat to party unity and deepen internal party division. To date, research on party switching has either focused on the individual motivations for changing party or on the effects of macro-level settings. The role of party-level variables, however, has received surprisingly little attention in the literature. In particular the impact of ideology has rarely been assessed. This article tests whether specific aspects related to parties’ ideology (i.e. extremism, isolation, authoritarianism, programmatic clarity and stability) are linked to different levels of defection. For this purpose I rely on an original database on party switching in 12 Western European countries from 1999 to 2015, supplemented with variables from the Chapel Hill Expert Survey. The results of multilevel negative binomial analyses show that indeed ideology and its various components have a substantial impact on the scope of switching. For instance, parties promoting authoritarian values suffer from a higher number of defectors. Moreover, parties with more unstable labels seem to be more subject to switching. This article improves our understanding of how party ideology is related to party unity and, more generally, to legislative dynamics. 1. Introduction To perform its functions, a representative democracy requires united political parties (Bowler et al., 1999). Party unity is necessary not only for cabinet stability and policy bargaining, but also to ensure politicians’ accountability. The way in which parties act as unitary actors has been challenged on two grounds. On the one hand, several studies (Carty, 2004; Bolleyer, 2012) have shown that from an organisational perspective, parties are not monolithic entities, but rather a ‘complex and variegated sets of persons and structures, each of which are, or could be, independent actors within the party’ (Olson, 2003, p. 165). This has led scholars to study more in-depth intra-party organisations, that is, whether and how members organise into factions and the impact that these subgroups have on, for instance, cabinet durability (Saalfeld, 2009), policymaking (Giannetti and Laver, 2009), party splits (Ceron, 2015), party cohesion and discipline (Depauw and Martin, 2009) and electoral volatility (Gherghina, 2014). On the other hand, the literature has also shown that parties not only organise into factions, but they also act in a non-unitary way and their members in parliament (MPs) do not always stick to the party line. Research conducted, for example, by Kam (2009) and Sieberer (2006) has shown that rebellion votes are not so infrequent in Western democracies. Therefore, party unity varies across countries, parties (Close and Gherghina, 2017) and over time. In extreme circumstances, rebellious MPs change party affiliation and such a behaviour, usually called ‘party switching’, raises threats to party stability, unity or credibility (O’brien and Shomer, 2013) and to voters’ ability to hold their representatives accountable (Heller and Mershon, 2009). Additionally, defections can affect policymaking within assemblies or even endanger the stability of governments (Giannetti and Laver, 2001). Finally, party switching has also substantial theoretical implications. Indeed, it sheds light on party change and dynamics within election and it might be considered as an alternative way of measuring party unity in parliament. Previous studies on switching have either analysed the reasons that drive switchers or the institutional settings that make this behaviour more likely to occur. In other words, the focus has either been on individuals and their motivations (Heller and Mershon, 2005; Di Virgilio et al., 2012) or on countries and their vulnerability to this phenomenon (Kreuzer and Pettai, 2009; McLaughlin, 2012). This article adopts an approach that is more similar to the institutional one, but it focuses on political parties. The role of party characteristics on defection has always been recognised by the literature (Heller and Mershon, 2005; Di Virgilio et al., 2012). However, party features are usually not tested per se, but they are used as proxies to uncover switchers’ goals. For instance, O’brien and Shomer (2013) analyse whether parties in opposition are more subject to switching in order to see whether MPs are driven by career advancement concerns. As party features are not at the centre of the explanation, we still know very little about whether and how party characteristics are related to different switching rates, both theoretically and empirically. In particular, few studies have explored the connection between ideology and unity (Owens, 2003). This article aims at filling this gap, by exploring whether ideology can explain some of the variation in the level of defections across countries and parties. In order to test my hypotheses, I rely on an original data set on party switching in 12 Western European countries from 1999 to 2015 and supplement it with variables from the Chapel Hill Expert Surveys (CHES) (Bakker et al., 2015; Polk et al., 2017) and the ParlGov database (Döring and Manow, 2015). The article is structured as follows: after a discussion of the theory and related hypotheses, I present the research design and the measurement of variables; in the fourth section, I show the results of several statistical models and I conclude with some final remarks. 2. Theory and hypotheses The literature on party switching has usually tried to uncover the motivations that lead a politician to cross the floor. MPs are considered as rational actors who try to maximise their interest and change party in order to serve their goals (Heller and Mershon, 2005). This stream of literature has enhanced our knowledge of individual reasons for changing party, yet it has not been fully able to explain why the scope of switching varies across countries, parties and time. If politicians are driven by similar goals (identified in literature as policy, office and vote (Müller and Strøm, 1999)), this does not account for the fact that defections are not so common among all political parties in Europe. There must be something else. This article moves away from the individual perspective and focuses on the relationship between party characteristics and the patterns of switching. In particular I analyse one macro-feature, namely ideology. So far, the connection between ideology and switching has been researched only by Mejia Acosta (2004) in his analysis of defections in Ecuador. The literature has more often looked at whether the distance between an MP ideal point and the policy preferences of his/her party affect the decision to switch (Desposato, 2006; Pinto, 2015). Thus, the link between parties’ ideology and their vulnerability to switching has not received large attention so far. Ideology affects the probability of a party to witness switching for two reasons. First, according to Sartori (1976), ideology is highly correlated to cohesion. Similarly, Owens (2003) states that parties’ values are a key factor to explain their level of cohesion and discipline. The mechanism linking ideology and loyalty (one of the dimensions of party unity) has been analysed by Close (2016), who illustrates how ideology affects both MPs’ perception of their role and party organisations. Ideology shapes the representational style that legislators adopt: for example, liberal parties embody values as individualism and—as a consequence—their MPs will behave in a more individualistic manner. Ideology also affects parties’ organisational structures and the level of internal democracy, that—in turn—might have an influence over legislators’ behaviour (Gauja, 2013). To put it shortly, ideology potentially affects MPs’ attitudes and behaviour both directly and indirectly. As ideology affects unity, it is most likely that it may also influence switching, which represents an extreme form of disunity. Secondly, according to the categories developed by Hirschman (1970), Kato (1998) and by Pedersen et al. (this issue), for certain MPs, ‘exit’ (i.e. changing party) is sometimes a more valuable option than ‘voice’. The decision of exiting versus staying in the party comes at specific costs for legislators. According to Yoshinaka (2015), among the many factors that may determine these costs, there is also ideology. Indeed, parties’ ultimate values affect the room granted to MPs to express their discontent. Certain parties encourage and tolerate much more that their legislators and members voice disagreement. In other groups, on the contrary, dissent is not allowed, with the result that MPs—in case of conflict—have no option but to exit their party. To put it in simple terms, my argument is that since exit and voice are inversely related, switching (exit) occurs more frequently in those parties in which it is more difficult to express dissatisfaction (voice). Nevertheless, as Close and Núñez (2016) underline, ideology is an extremely complex concept, that can only be partially grasped using parties’ ideological position on the left–right scale. For this reason, this article looks at the theoretical and empirical connection between switching and different understandings of ideology: the spatial dimension, with the extremity of parties’ position (H1) and their isolation (H2), parties’ values (H3), with their clarity (H4) and stability over time (H5). Let us now turn to each of these variables and explain how they may influence defection. 2.1 Extremism and isolation Although party placement is the most used proxy for ideology, its impact over party switching has been rarely tested. Moreover, a clear theoretical argument linking ideological placement and switching is missing. If the relationship between party position and defection is still unclear, there is another factor that instead has received more attention in the literature, that is, ideological extremism. The argument is that what matters is the extremity of party position, rather than ideological placement per se. When a party is placed at the extreme of the ideological spectrum, its MPs can only switch towards the centre. In other words, they have fewer options compared to members of centrist parties, who instead face appeals from two sides (Morgenstern, 2003). Moreover, extremist parties are usually believed to have a clearer ideology (Mejia Acosta, 2004) and they are generally more faithful to their values and less sensitive to public opinion, which makes them also more cohesive (Rahat, 2007). However, for what concerns switching, these expectations might not apply. In fact, extremist parties stress the importance of principles and values, thus, the internal ideological discussion might be more acute. Additionally, the willingness to remain faithful to ideology (Rahat, 2007) might restrict the space for legislators with alternative views. Therefore, in case of conflict or excessive debate around a specific issue, the only realistic option for dissidents is leaving the party. This argument is coherent with Hirschman’s scheme: in extremist parties it might be more difficult for MPs to voice their discontent. Discipline might be so tight that—in case of conflict—there is no alternative but switching. The expectation therefore is opposite than what was posed by the literature so far: more extremist parties might be more disciplined, but this might lead to a higher number of switchers. Extremism looks at the ideology of individual parties. However, the opportunity of switching might be also affected by the presence/absence of other political parties that are ideologically close to the one of origin. If a party is isolated in the political space, it is more difficult for potential switchers to find a group able to welcome them, simply because there are no alike parties around. Not only the availability of similar parties is lower, but also the ideological transformation undertaken by defectors is deeper and more difficult to achieve and justify to voters. On the contrary, when a party is relatively close to others (a likely circumstance, especially under coalition governments) it is easier to find akin political platforms and a conversion is less tricky to defend. The ideological isolation of a party in most of the cases might be greater for extremist parties; however, this is not always the case. In fact, there could be a system with a centrist party distant from clusters of parties on its right and/or left. In this case we have a party that is not extreme, but that is isolated. To put it simply, ideological extremism and isolation do not necessarily overlap. For this reason, the two factors are analysed separately. Following these arguments, I expect that: Hypothesis 1An extreme placement of the party in the political space favours defections. Hypothesis 2Less ideological isolation of parties favours defections. 2.2 Values, ideological clarity and stability The limitation of using a spatial understanding of ideology is that it does not grasp parties’ values, which instead are determinant for legislators’ representational style and—as a consequence—for switching. While it can be argued that parties’ placement on the left–right scale indeed corresponds to specific set of beliefs (Hinich and Munger, 1992), it is problematic to exactly determine parties’ values by looking only at their position on the left–right spectrum. Moreover, often the left–right continuum conceives ideology only in terms of economic policy, while for my argument, the cultural dimension is more relevant. This cultural dimension, called by Inglehart (1977) ‘the postmaterialist–materialist cleavage’ and by Kitschelt (1994) ‘authoritarian–libertarian cleavage’, has at its core the concepts of hierarchy and tolerance (Stubager, 2010). Authoritarians favour the rank ordering of individuals and dot not tolerate deviations from conventional norms. On the contrary, libertarians promote parity in social interactions, and show a high degree of tolerance for non-conformity. These values are crucial also for party switching. Indeed, we can expect that parties with authoritarian values might also not be tolerant towards dissenting positions among their legislators. Conversely, parties promoting libertarian values might encourage the expression of deviating positions. In Hirschman’s terms, I expect that libertarian parties display a higher degree of ‘voice’ which makes them more immune to switching, while authoritarian parties discourage ‘voice’ and leave no other option to potential dissidents but exiting the party. Parties’ platforms are often difficult to pin down exactly. There is always a certain degree of uncertainty regarding parties’ stances on specific issues (Bräuninger and Giger, 2016). According to Rovny (2012), parties intentionally keep their platforms vague, because this is electorally helpful. Moreover, some parties have not only very ambiguous policy platforms, but they also represent a wide range of values. To put it simply, not all parties have clear-cut ideologies. Indeed, Gunther and Diamond (2001) use programmatic clarity to classify different kinds of parties. According to the authors, catch-all parties are characterised by vague platforms, while programmatic and mass parties have clear policy manifestos and are ideologically sound (Giebler et al., 2015). Gunther and Diamond (2001) state that parties with ambiguous preferences have a greater ability to support coalitions as they can accommodate partners, thanks to their policy flexibility. Similarly, these kinds of parties are also able to host legislators with very different beliefs. As there is no well-defined ideology, there is also no pressure to stick to the party line and MPs have probably more room to express their views, even when they are conflicting with each other. In Hirschman’s scheme, in these parties the level of ‘voice’ should be greater than in parties whose platform is clearly defined and less flexible. As a consequence, the expectation is that the lower the ideological clarity, the less likely that legislators will recur to switching. The previous factors look at parties’ ideology at one specific moment. However, platforms are not stable over time and they can change significantly between two elections (Schumacher et al., 2013). For instance, parties from the right might shift their policy preferences towards the centre or to more extreme positions and the same can happen to leftist groups. The literature has studied why parties would modify their platforms and it is beyond the scope of this article to discuss the several findings of this stream of research (for a summary, see Fagerholm (2016)). For the purpose of this work, what matters are the consequences that these position have on MPs’ decision to change party. According to Heller and Mershon (2005), ‘uncertainty about party policy makes it likely that MPs will at times find party dictates on legislation to be at odds with their own or their supporters’ preferences’ (p. 539). Similarly, Ames (2009) states that when parties are unstable, legislators’ voting behaviour cannot be predicted by their affiliation and they often defect with impunity. The research on party switching thus expects parties with unstable platforms to witness more defections (Desposato, 2006). Moreover, as the literature on policy shifts has shown, parties usually change their platforms after an electoral defeat (Somer-Topcu, 2009). Poor electoral performances are one of the main determinants of switching, as politicians leave their party if they fear an electoral loss (Gherghina, 2016; Klein, 2016). A policy shift might therefore be considered as a clue for electoral concerns that—in turn—might induce switching. From a party perspective, a shift in the programmatic platform might also come at a cost. Indeed, the change most likely disappoints part of the party members, no matter what triggers it (electoral defeats, leadership alternation). Some members feel the new course as a betrayal of the original parties’ values and consider the exit option in order to preserve them, and this is particularly true for parties with well-defined ideologies (Salucci, 2008). To summarise, parties with unstable platforms are more subject to switching because programmatic change can disappoint part of the membership. Moreover, a value review might be linked to an electoral defeat that represents a great concern for legislators, who then might prefer to leave the sinking ship. All these arguments lead to the following hypotheses: Hypothesis 3Authoritarian values of parties favour defections. Hypothesis 4Programmatic clarity favours defections. Hypothesis 5Unstable ideological position favours defections. 3. Data, measurement and methods The five hypotheses are tested using a self-collected data set on all the defections1 that occurred in 12 Western European democracies2 from 1999 to 2015. The unit of analysis is party–year, that is, one observation corresponds to a party in a given year. The dependent variable (Switchers) counts the number of switchers that each party witnessed within a year. As can be seen from the descriptive statistics (Table 1), the overall average number of switchers is low. Considering all countries together, the mean number of changes per year is almost 0.6. Moreover, the variable’s distribution is extremely skewed towards the left, that is, most observations take value 0. Table 1. Descriptive statistics Variable N Mean Std. Dev. Min Max Year 1223 2.006 4.82 1999 2015 Switchers 1223 0.576 3.24 0 74 Seat 1223 5.296 7.67 1 419 LRgen 1223 4.887 2.22 0.22 9.888 GAL/TAN 1223 4.748 2.28 0.63 9.75 LRgen sd 1216 0.861 0.37 0 3.420 Tenure 1223 3.475 2.10 0 70 Seat share 1223 0.154 0.15 0.002 0.636 ENPP 1223 4.434 1.96 2.119 9.054 ΔLRgen 1223 0.41 0.37 0 2.088 Extreme 1223 1.925 1.11 0 4.888 Volatility 1217 1.391 7.37 4 48.50 Govt 1223 0.389 0.49 0 1 Isolation 1223 2.783 0.86 1.115 5.855 Variable N Mean Std. Dev. Min Max Year 1223 2.006 4.82 1999 2015 Switchers 1223 0.576 3.24 0 74 Seat 1223 5.296 7.67 1 419 LRgen 1223 4.887 2.22 0.22 9.888 GAL/TAN 1223 4.748 2.28 0.63 9.75 LRgen sd 1216 0.861 0.37 0 3.420 Tenure 1223 3.475 2.10 0 70 Seat share 1223 0.154 0.15 0.002 0.636 ENPP 1223 4.434 1.96 2.119 9.054 ΔLRgen 1223 0.41 0.37 0 2.088 Extreme 1223 1.925 1.11 0 4.888 Volatility 1217 1.391 7.37 4 48.50 Govt 1223 0.389 0.49 0 1 Isolation 1223 2.783 0.86 1.115 5.855 Table 1. Descriptive statistics Variable N Mean Std. Dev. Min Max Year 1223 2.006 4.82 1999 2015 Switchers 1223 0.576 3.24 0 74 Seat 1223 5.296 7.67 1 419 LRgen 1223 4.887 2.22 0.22 9.888 GAL/TAN 1223 4.748 2.28 0.63 9.75 LRgen sd 1216 0.861 0.37 0 3.420 Tenure 1223 3.475 2.10 0 70 Seat share 1223 0.154 0.15 0.002 0.636 ENPP 1223 4.434 1.96 2.119 9.054 ΔLRgen 1223 0.41 0.37 0 2.088 Extreme 1223 1.925 1.11 0 4.888 Volatility 1217 1.391 7.37 4 48.50 Govt 1223 0.389 0.49 0 1 Isolation 1223 2.783 0.86 1.115 5.855 Variable N Mean Std. Dev. Min Max Year 1223 2.006 4.82 1999 2015 Switchers 1223 0.576 3.24 0 74 Seat 1223 5.296 7.67 1 419 LRgen 1223 4.887 2.22 0.22 9.888 GAL/TAN 1223 4.748 2.28 0.63 9.75 LRgen sd 1216 0.861 0.37 0 3.420 Tenure 1223 3.475 2.10 0 70 Seat share 1223 0.154 0.15 0.002 0.636 ENPP 1223 4.434 1.96 2.119 9.054 ΔLRgen 1223 0.41 0.37 0 2.088 Extreme 1223 1.925 1.11 0 4.888 Volatility 1217 1.391 7.37 4 48.50 Govt 1223 0.389 0.49 0 1 Isolation 1223 2.783 0.86 1.115 5.855 Turning to the independent variables, the first hypothesis looks at ideological extremism. In order to calculate this, I retrieved parties’ position in the political space from the CHES (variable LRgen3). Then, I calculated the absolute value of the distance between each party’s position and the centre of the spectrum, which equals 5. The variable Extreme ranges from 0 (centrist party) to 5 (extremist party). I expect that when the variable increases, the number of switchers increases as well. The second hypothesis discusses the role of isolation, which is operationalised as the average mean distance of a party from all the others in the system. Distances are calculated in absolute values (variable Isolation) and when they increase, the number of switchers should become smaller. The third hypothesis analyses the connection between values and switching. In order to measure how authoritarian are parties’ values, I retrieved parties’ position on the so-called GAL/TAN dimension4 made available by the CHES. The index ranges from 0 (extreme GAL) to 10 (extreme TAN) and measures exactly how parties’ values are close/far from the authoritarian tradition. The expectation is that the greater the score of a party on the scale, the higher the number of defections. The fourth hypothesis concerns the impact of programmatic clarity. The operationalisation of this variable is challenging. I decided to use the standard deviation of each party’s position as calculated in the expert surveys (variable LRgen sd). This indicator is far from being perfect, because it measures variation in experts’ perceived positions of parties. As Bräuninger and Giger (2016) point out, this variation can either be the result of an effective programmatic vagueness, or might also reveal the difficulty faced by experts in placing parties on the ideological space. Nevertheless, despite its limitations, this index is the best option given that other measures (e.g. those proposed by Bräuninger and Giger (2016) or Giebler et al. (2015)) are not available for the full set of cases considered in this article. The expectation is that the larger the standard deviation of a party’s position, the lower the number of switchers. Finally, ideological stability is measured as shifts in parties’ position on the left–right scale between two waves of the CHES (variable ΔLRgen). As I am not interested in the direction of the change, but only in its magnitude, the index is calculated in absolute terms. As descriptive statistics reveal, the variable obtained ranges from 0, when a party’s placement has not changed, to 2 (the maximum shift recorded).5 Based on the fifth hypothesis, larger values of ΔLRgen should correspond to a higher number of defections. For what concerns controls, I add the following five variables: ○ Tenure: measures the years a party has been in parliament. Tenure is a proxy of party institutionalization that might affect the level of switching witnessed (Ceron 2015). ○ Governing status: I control for the governing status of a party because according to several authors parties in power should be more immune to defections (Desposato 2006; Di Virgilio et al. 2012). The variable takes value 0 when a party has been in opposition in a given year, and 1 otherwise. Scores were retrieved by the ParlGov database. ○ Size: calculated as seat share, it is included because the literature suggests that smaller parties might be more subject to switching (Heller and Mershon 2005; Laver and Benoit 2003). ○ Party system fragmentation: measured as the ‘effective number of parliamentray parties’ (ENPP) (Laakso and Taagepera 1979). I include this control because it has been shown that the higher the level of fragmentation, the larger the opportunities for defecting. ○ Party system institutionalization: operationalized as electoral volatility, as one of the most common index for party system instability (Chiaramonte and Emanuele 2015). I add this control because, based on previous research, defections are more numerous in weakly institutionalized settings (Kreuzer and Pettai 2009; Mainwaring 1998). Information on electoral volatility was retrieved from the dataset by Emanuele (2015). Given that my dependent variable is a count variable, data is analysed with a negative binomial model to account for the over-dispersion of the dependent variable6 (Long, 1997). Moreover, in order to account for party size, I include an exposure variable that measures it. This exposure variable allows to adjust the estimation for the amount of opportunity an event has. In other words, it treats the count variable as a ratio. The risk of omitting the exposure variable is that larger parties always result having a higher number of switchers simply because they have a greater set of potential defectors. The advantage of the exposure variable is that it can be included in the estimation as well. In order to control for the hierarchical structure of the data, I use a random-effect model. The final data set includes 1217 observations, from 111 parties nested in 12 countries. 4. Results and discussion The results of the multivariate statistical analysis are presented in Table 2. Each hypothesis is tested separately from the others, in order to avoid potential problems of multicollinearity. The first model looks at the relationship between parties’ extremism and the number of switchers. According to Hypothesis 1, the expectation is that more extremist parties should witness more switchers. The coefficient of the variable Extreme is positive and statistically significant, thus in line with the hypothesis. The effect of the variable is considerable, given that the number of defections for the most extremist parties is above 2, while for centrist group the prediction is below the overall mean (0.5) (Figure 1a). Table 2. Random-effect negative binomial models of party switching Variables Model 1 Model 2 Model 3 Model 4 Model 5 Tenure −0.017* −0.014 −0.018* −0.015† −0.020* (0.01) (0.01) (0.01) (0.01) (0.01) Govt 0.250 0.244 0.195 0.210 0.210 (0.25) (0.25) (0.25) (0.25) (0.25) ENPP −0.165 −0.141 −0.129 −0.137 −0.140 (0.10) (0.10) (0.10) (0.10) (0.11) Seat share −2.266* −1.698† −1.782† −2.018* −2.135* (0.98) (1.00) (0.98) (1.00) (1.00) Volatility 0.022† 0.021† 0.024* 0.015 0.022† (0.01) (0.01) (0.01) (0.01) (0.01) Extreme 0.396** Isolation (0.14) 0.640*** GAL/TAN (0.16) 0.190** LRgen sd (0.07) −0.658* ΔLRgen (0.33) 0.543* (0.27) Constant −4.628*** −4.780*** −4.139*** −5.655*** −3.164*** (0.68) (0.72) (0.64) (0.78) (0.69) Country 0.231 0.253 0.262 0.166 0.294 (0.25) (0.26) (0.26) (0.23) (0.28) Party 1.339** 1.553** 1.416** 1.651** 1.599** (0.44) (0.49) (0.46) (0.51) (0.50) Alpha (Ln) 1.031*** 0.996*** 1.032*** 0.940*** 0.990*** (0.13) (0.13) (0.13) (0.13) (0.13) Observations 1.215 1.217 1.217 1.217 1.21 Number of groups 12 12 12 12 12 Wald χ2(6) 25.20 24.73 19.99 32.95 19.50 Prob > χ2 0.001 0.001 0.003 0.001 0.003 Variables Model 1 Model 2 Model 3 Model 4 Model 5 Tenure −0.017* −0.014 −0.018* −0.015† −0.020* (0.01) (0.01) (0.01) (0.01) (0.01) Govt 0.250 0.244 0.195 0.210 0.210 (0.25) (0.25) (0.25) (0.25) (0.25) ENPP −0.165 −0.141 −0.129 −0.137 −0.140 (0.10) (0.10) (0.10) (0.10) (0.11) Seat share −2.266* −1.698† −1.782† −2.018* −2.135* (0.98) (1.00) (0.98) (1.00) (1.00) Volatility 0.022† 0.021† 0.024* 0.015 0.022† (0.01) (0.01) (0.01) (0.01) (0.01) Extreme 0.396** Isolation (0.14) 0.640*** GAL/TAN (0.16) 0.190** LRgen sd (0.07) −0.658* ΔLRgen (0.33) 0.543* (0.27) Constant −4.628*** −4.780*** −4.139*** −5.655*** −3.164*** (0.68) (0.72) (0.64) (0.78) (0.69) Country 0.231 0.253 0.262 0.166 0.294 (0.25) (0.26) (0.26) (0.23) (0.28) Party 1.339** 1.553** 1.416** 1.651** 1.599** (0.44) (0.49) (0.46) (0.51) (0.50) Alpha (Ln) 1.031*** 0.996*** 1.032*** 0.940*** 0.990*** (0.13) (0.13) (0.13) (0.13) (0.13) Observations 1.215 1.217 1.217 1.217 1.21 Number of groups 12 12 12 12 12 Wald χ2(6) 25.20 24.73 19.99 32.95 19.50 Prob > χ2 0.001 0.001 0.003 0.001 0.003 Note: Standard errors in parentheses. *** p < 0.001, ** p < 0.01, * p < 0.05, † p < 0.1. Table 2. Random-effect negative binomial models of party switching Variables Model 1 Model 2 Model 3 Model 4 Model 5 Tenure −0.017* −0.014 −0.018* −0.015† −0.020* (0.01) (0.01) (0.01) (0.01) (0.01) Govt 0.250 0.244 0.195 0.210 0.210 (0.25) (0.25) (0.25) (0.25) (0.25) ENPP −0.165 −0.141 −0.129 −0.137 −0.140 (0.10) (0.10) (0.10) (0.10) (0.11) Seat share −2.266* −1.698† −1.782† −2.018* −2.135* (0.98) (1.00) (0.98) (1.00) (1.00) Volatility 0.022† 0.021† 0.024* 0.015 0.022† (0.01) (0.01) (0.01) (0.01) (0.01) Extreme 0.396** Isolation (0.14) 0.640*** GAL/TAN (0.16) 0.190** LRgen sd (0.07) −0.658* ΔLRgen (0.33) 0.543* (0.27) Constant −4.628*** −4.780*** −4.139*** −5.655*** −3.164*** (0.68) (0.72) (0.64) (0.78) (0.69) Country 0.231 0.253 0.262 0.166 0.294 (0.25) (0.26) (0.26) (0.23) (0.28) Party 1.339** 1.553** 1.416** 1.651** 1.599** (0.44) (0.49) (0.46) (0.51) (0.50) Alpha (Ln) 1.031*** 0.996*** 1.032*** 0.940*** 0.990*** (0.13) (0.13) (0.13) (0.13) (0.13) Observations 1.215 1.217 1.217 1.217 1.21 Number of groups 12 12 12 12 12 Wald χ2(6) 25.20 24.73 19.99 32.95 19.50 Prob > χ2 0.001 0.001 0.003 0.001 0.003 Variables Model 1 Model 2 Model 3 Model 4 Model 5 Tenure −0.017* −0.014 −0.018* −0.015† −0.020* (0.01) (0.01) (0.01) (0.01) (0.01) Govt 0.250 0.244 0.195 0.210 0.210 (0.25) (0.25) (0.25) (0.25) (0.25) ENPP −0.165 −0.141 −0.129 −0.137 −0.140 (0.10) (0.10) (0.10) (0.10) (0.11) Seat share −2.266* −1.698† −1.782† −2.018* −2.135* (0.98) (1.00) (0.98) (1.00) (1.00) Volatility 0.022† 0.021† 0.024* 0.015 0.022† (0.01) (0.01) (0.01) (0.01) (0.01) Extreme 0.396** Isolation (0.14) 0.640*** GAL/TAN (0.16) 0.190** LRgen sd (0.07) −0.658* ΔLRgen (0.33) 0.543* (0.27) Constant −4.628*** −4.780*** −4.139*** −5.655*** −3.164*** (0.68) (0.72) (0.64) (0.78) (0.69) Country 0.231 0.253 0.262 0.166 0.294 (0.25) (0.26) (0.26) (0.23) (0.28) Party 1.339** 1.553** 1.416** 1.651** 1.599** (0.44) (0.49) (0.46) (0.51) (0.50) Alpha (Ln) 1.031*** 0.996*** 1.032*** 0.940*** 0.990*** (0.13) (0.13) (0.13) (0.13) (0.13) Observations 1.215 1.217 1.217 1.217 1.21 Number of groups 12 12 12 12 12 Wald χ2(6) 25.20 24.73 19.99 32.95 19.50 Prob > χ2 0.001 0.001 0.003 0.001 0.003 Note: Standard errors in parentheses. *** p < 0.001, ** p < 0.01, * p < 0.05, † p < 0.1. Figure 1 View largeDownload slide View largeDownload slide Marginal effects of explanatory variables, created with schemes by Bischof (2017): (a) Extreme, (b) Isolation, (c) GAL/TAN, (d) Clarity, (e) Stability. Figure 1 View largeDownload slide View largeDownload slide Marginal effects of explanatory variables, created with schemes by Bischof (2017): (a) Extreme, (b) Isolation, (c) GAL/TAN, (d) Clarity, (e) Stability. Hypothesis 2 states that ideological isolation should represent a protection from defections. On the contrary, the statistical analysis returns a positive and significant coefficient for the variable Isolation. This means that the farther a party is from the other groups in the system, the greater the number of switchers. The effect of the variable is the largest observed (Figure 1b), as the expected number of switchers for isolated parties is around 3. This finding might be explained by the fact that there is no point in changing affiliation when the original and receiving parties are so similar to each other. The policy differences—and most likely also their governing/office status—are so little that switching becomes pointless. If it is true that concealing a switch between two contiguous parties might be easier, the risk is that switchers do not achieve anything with their action, neither in policy nor in office terms. Additionally, a very isolated party has less chances to find potential partners in future elections and therefore it risks an electoral defeat. Thus, those MPs who are seeking re-election might be induced to change alliance. To summarise, no matter what the motivations of switchers are, an isolated party is not a pay-off option and this might be why MPs tend to leave them more frequently. This argument might also help us to explain why switching does not occur very frequently: defection is costly and legislators are willing to pay the price only if they fully reach their goals. Turning to the variables that measure parties’ values, according to Hypothesis 3, the closer a party is to the TAN pole, the greater the scope of defections. The coefficient of the variable GAL/TAN is positive and statistically significant, and hence in line with the hypothesis. Parties that promote values like law and order are also less able to keep their ranks together. Moreover, the effect of the variable is substantial. As Figure 1c reveals, a party with maximum TAN score is expected to have almost 2 switchers, whereas at the opposite end of the spectrum, the predicted number of defections is slightly below the overall mean. Based on Hypothesis 4, parties with more unclear platforms should witness less switchers. As Model 4 shows, the coefficient of the variable LRgen sd is negative and statistically significant. This result meets the expectation of Hypothesis 4 and it implies that the larger the uncertainty around a party’s placement on the left–right scale, the lower the number of defections. The effect of LRgen sd is smaller compared to other explanatory variables, yet not negligible. As Figure 1d shows, very cohesive groups are expected to witness 1.5 switchers, but this number falls below average for parties with more unclear positions. The fifth hypothesis finds also supports from the analysis. According to the theory presented, the more unstable the party positions are, the larger the scope of switching. The related variable (ΔLRgen) has a positive sign and significant effect. As shown by Figure 1e, parties that did not undergo a substantial revision of their platforms are expected to have 0.5 switchers, while parties with unstable programmes witness almost two defections per year. Finally, for what concerns control variables, more experienced parties are—in line with what predicted by the literature—also more stable (significant and negative coefficient). Parties in government are usually considered to be more united than parties in opposition, but my analysis returns a different result: variable Govt is positive, yet it does not reach statistical significance. Interestingly enough, larger parties are also less subject to switching, in line with what was found by research on other dimensions of party unity (Close, 2016). At the party system level, the effect of fragmentation is not confirmed by the analysis, as the coefficient of the ENPP is not significant (and it is negative, i.e. opposite to expectations). On the contrary, volatility seems to boost parliamentary disunity. Very volatile elections are followed by a greater level of switching during the legislative term. This finding supports the idea that defections are also the product of low party system institutionalisation. 5. Conclusion The aim of this article was to shed light on the relationship between party ideology and unity, looking at a very specific form of party dis-unity, that is, defection. The importance of ideology has been largely neglected by both the literature on unity and the one on switching. This article tried to fill this gap, bringing parties and their ideological features at the centre of the explanation. Political parties profoundly influence the behaviour and attitudes of their MPs. As Kam (2001) reminds us, party affiliation is a better predictor of a legislator’s behaviour than his/her preferences. This argument suggests that party characteristics might be linked to different levels of unity and switching. Among all parties’ features, ideology has a determinant role in shaping MPs’ behaviour. Indeed, as clarified by Close (2016), ideology affects both the representational style of MPs, who act in a more or less independent way from their group, and the level of intra-party democracy, that is, the internal tolerance for dissenting views. I tested whether this argument holds also when we look at switching as another dimension of (dis)unity. Given that ideology is a multifaceted concept, I looked at whether various understandings of ideology are associated with different levels of defections. The results of my analysis confirmed this intuition, as all the ideological variables tested have a significant and (in most of the cases) substantial effect on the number of defections experienced by the parties analysed. These results are particularly solid as they hold across 12 polities and for more than 15 years. To put it simply, my analysis shows that ideology, in its various meanings, is indeed linked with different levels of party switching. My theoretical expectations and results are partially conflicting with previous literature on unity. In particular, ideological extremism, clarity and isolation have proven to induce switching, rather than reducing it. Also the result that parties with authoritarian values are more unstable contradicts the finding that extreme right groups are usually more united (Mejia Acosta, 2004). Nevertheless, the results of my analysis are coherent with each other. In particular, extremism, isolation, authoritarianism and programmatic clarity, are characteristics that can be theoretically associated with a smaller room for dissent. Therefore, the fact that these features are all related to higher number of defections suggests that switching occurs in those parties that do not allow multiple views, but rather impose one strict line. In other words, switching is trigged by an absence of ‘voice’. On the contrary, parties that grant their members more freedom of opinion are more immune to ‘exit’. Clearly, this is only one plausible explanation to make sense of the findings of my analysis. More theoretical work is needed to shed light on what makes more authoritarian parties so exposed to defection. What it is specific of authoritarian ideology that makes parties less tolerant towards different opinions and views? Moreover, the results of my analysis would benefit from any index that could capture and measure the level of voice granted to party members and MPs. To put it differently, my results would be more solid if I could show that the features that are linked with more switching are also predictors for lower levels of intra-party democracy. The challenge is to find a good indicator for all the countries and the time frame covered by my dataset. It is for this reason that, for instance, I could not use for this work the index developed by the Political Party Database Project (Poguntke et al., 2016). Alternatively, I could try to look at different kinds of defections. Under the label ‘switching’ fall indeed different types of behaviour. In particular we can distinguish between individual and collective forms of switching. Individual defectors are those who change party without coordinating with other fellows. Collective switches, instead, are rather the results of party merges or splitting. In this case, individual MPs switch in order to stay faithful to their faction. The underlying logic of collective and individual switching are different and the mechanisms leading to these two forms of defections might also not be the same. Looking at these two types of switching separately might help to test whether parties from different ideological traditions experience different kinds of defections. For instance, switchers from authoritarian parties might change affiliation only collectively, while—on the contrary—libertarian parties might be more subject to individual movements, as an effect of the representational style promoted by these groups. Moreover, it is plausible that some of the factors analysed in this article explain better one form of switching than the other. For example, ideological stability and isolation might be at the origin of collective changes. A programmatic shift might trigger the reaction of an entire faction that does not approve the ideological revision. Similarly, very isolated parties might arrive at a stage in their political life in which they have to decide whether they want to make compromise and get fully involved or to stay away from any possible coalition. The tension generated by these two contrary strategies might as well result in a split between opposing factions. To summarise, when we take into account the collective or individual nature of defections, the five ideological factors analysed in this article might not have the same explanatory power. Acknowledgements The author is very grateful to Caroline Close, Helene Helboe Pedersen, Sergiu Ghergina and all the other participants in the ECPR Joint Session of Workshops on Intra-party cohesion (Nottingham, April 2017) for their comments and insights. Conflict of interest The author has not reported any conflict of interests. Footnotes 1 I adopt a more restricted definition of switching, compared to the most used one by Heller and Mershon (2009). While the two authors count as switches also label changes, I discarded them. I considered as switching the following situations: switching to a party already existing, becoming independent, party merging and splits and establishment of a new group. Moreover, my data have been collected through an analysis of parliamentary archives; therefore, only in case the switch has been recorded, I counted it. 2 Austria, Belgium, Denmark, Finland, France, Germany, Greece, Ireland, Italy, the Netherlands, Spain, UK. Norway and Switzerland are not included because they are not in the CHES’ trend file. Sweden and Portugal are not part of the data set because of considerable problems in collecting the data on defectors. 3 It ranges from 0 (extreme left) to 10 (extreme right). 4 Green-Alternative-Libertarian versus Traditional-Authoritarian-Nationalist continuum (Hooghe et al., 2002). 5 Theoretically the maximum is 10, hence stability is the rule. 6 As most of the observations take value 0, the variance of dependent variable exceeds the mean. References Ames B. ( 2009) The Deadlock of Democracy in Brazil , Ann Arbor, University of Michigan Press. Bakker R., de Vries C., Edwards E., Hooghe L., Jolly S., Marks G., Polk J., Rovny J., Steenbergen M., Vachudova M. A. ( 2015) ‘Measuring Party Positions in Europe: The Chapel Hill Expert Survey Trend File, 1999–2010’, Party Politics , 21, pp. 143– 152. Bischof D. ( 2017) ‘New Graphic Schemes for Stata: plotplain & plottig’, Stata Journal 17, 748– 759. Bolleyer N. ( 2012) ‘ New Party Organization in Western Europe: Of Party Hierarchies, Stratarchies and Federations’, Party Politics , 18, 315– 336. Google Scholar CrossRef Search ADS Bowler S., Farrell D. M., Katz R. S. (eds) ( 1999) Party Discipline and Parliamentary Government , Columbus, Ohio State University Press. Bräuninger T., Giger N. ( 2016) ‘ Strategic Ambiguity of Party Positions in Multi-Party Competition’, Political Science Research and Methods , DOI: 10.1017/psrm.2016.18. Carty R. K. ( 2004) ‘ Parties as Franchise Systems: The Stratarchical Organizational Imperative’, Party Politics , 10, 5– 24. Google Scholar CrossRef Search ADS Ceron A. ( 2015) ‘ The Politics of Fission: An Analysis of Faction Breakaways among Italian Parties (1946-2011)’, British Journal of Political Science , 45, 121– 139. Google Scholar CrossRef Search ADS Chiaramonte A., Emanuele V. ( 2015) ‘ Party System Volatility, Regeneration and De-Institutionalization in Western Europe (1945–2015)’, Party Politics , 23, 376– 388. Google Scholar CrossRef Search ADS Close C. ( 2016) ‘ Parliamentary Party Loyalty and Party Family: The Missing Link?’, Party Politics , DOI: https://doi.org/10.1177/1354068816655562. Close C., Núñez L. ( 2016) ‘At the Root of Parliamentary Party Cohesion: The Role of Intraparty Heterogeneity and Party Ideology’, CEVIPOL Working Papers/Cahiers du CEVIPOL 1. Close C., Gherghina S., ( 2017) ‘Rethinking Intra-party Cohesion: Towards a Conceptual and Analytical Framework’, ECPR Joint Session of Workshops, Nottingham. Depauw S., Martin S. ( 2009) ‘Legislative Party Discipline and Cohesion in Comparative Perspective’. In Giannetti D., Benoit K. (eds) Intra-Party Politics and Coalition Governments , London, Routledge, pp. 103– 120. Desposato S. W. ( 2006) ‘ Parties for Rent? Ambition, Ideology, and Party Switching in Brazil’s Chamber of Deputies’, American Journal of Political Science , 50, 62– 80. Google Scholar CrossRef Search ADS Di Virgilio A., Giannetti D., Pinto L. ( 2012) ‘ Patterns of Party Switching in the Italian Chamber of Deputies, 2008-2011’, Rivista Italiana Di Scienza Politica , 42: 29– 58. Döring H., Manow P. ( 2015) Parliaments and Governments Database (ParlGov): Information on Parties, Elections and Cabinets in Modern Democracies, Development version. Emanuele V. ( 2015) Dataset of Electoral Volatility and its internal components in Western Europe (1945-2015), Rome, Italian Center for Electoral Studies, DOI: http://dx.doi.org/10.7802/1112. Accessed at http://cise.luiss.it/cise/dataset-of-electoralvolatility-and-its-internal-components-in-western-europe-1945-2015/ on July 2017. Fagerholm A. ( 2016) ‘ Why Do Political Parties Change Their Policy Positions? A Review’, Political Studies Review , 14, 501– 511. Google Scholar CrossRef Search ADS Gauja A. ( 2013) ‘Policy Development and Intra-party Democracy’. In Cross W. P., Katz R. S. (eds) The Challenges of Intra-Party Democracy , Oxford, Oxford University Press, pp. 116– 135. Google Scholar CrossRef Search ADS Gherghina S. ( 2014) Party Organization and Electoral Volatility in Central and Eastern Europe: Enhancing Voter Loyalty , London, Routledge. Gherghina S. ( 2016) ‘ Rewarding the “Traitors”? Legislative Defection and Re-Election in Romania’, Party Politics , 22, 490– 500. Google Scholar CrossRef Search ADS Giannetti D., Laver M. ( 2001) ‘ Party System Dynamics and the Making and Breaking of Italian Governments’, Electoral Studies , 20, 529– 553 Google Scholar CrossRef Search ADS Giannetti D., Laver M. ( 2009) ‘Party Cohesion, Party Discipline and Party Factions in Italy’. In Giannetti D., Benoit K. (eds) Intra-Party Politics and Coalition Governments , London, Routledge, pp. 146– 168. Giebler H., Lacewell O., Regel S., Werner A. ( 2015) ‘Niedergang oder Wandel?’ In Merkel W. (ed.) Demokratie Und Krise , Wiesbaden, Springer, pp. 181– 219. Gunther R., Diamond L. ( 2001) ‘Types and Functions of Parties’. In Diamond L., Gunther R. (eds) Political Parties and Democracy , Baltimora, JHU Press, pp. 3– 39. Heller W. B., Mershon C. ( 2005) ‘ Party Switching in the Italian Chamber of Deputies, 1996–2001’, Journal of Politics , 67, 536– 559. Google Scholar CrossRef Search ADS Heller W. B., Mershon C. ( 2009) ‘Introduction: Legislative Party Switching, Parties, and Party Systems’, In Heller W. B., Mershon C. (eds) Political Parties and Legislative Party Switching , Palgrave Macmillan, pp. 3– 28. Google Scholar CrossRef Search ADS Hinich M. J., Munger M. C. ( 1992) ‘ A Spatial Theory of Ideology’, Journal of Theoretical Politics , 4, 5– 30 Google Scholar CrossRef Search ADS Hirschman A. O. ( 1970) Exit, Voice, and Loyalty; Responses to Decline in Firms, Organizations, and States , Cambridge, MA, Harvard University Press. Hooghe L., Marks G., Wilson C. J. ( 2002) ‘ Does Left/Right Structure Party Positions on European Integration?’ Comparative Political Studies , 35, 965– 989. Google Scholar CrossRef Search ADS Inglehart R. ( 1977) The Silent Revolution: Changing Values and Political Styles among Western Publics , Princeton, Princeton University Press. Kam C. ( 2001) ‘ Do Ideological Preferences Explain Parliamentary Behaviour? Evidence from Great Britain and Canada’, The Journal of Legislative Studies , 7, 89– 126 Google Scholar CrossRef Search ADS Kam C. ( 2009) Party Discipline and Parliamentary Politics , Cambridge, Cambridge University Press. Google Scholar CrossRef Search ADS Kato J. ( 1998) ‘ When the Party Breaks Up: Exit and Voice among Japanese Legislators’, American Political Science Review , 92, 857– 870. Google Scholar CrossRef Search ADS Kitschelt H. ( 1994) The Transformation of European Social Democracy , Cambridge, Cambridge University Press. Google Scholar CrossRef Search ADS Klein E. ( 2016) ‘The Personal Vote and Legislative Party Switching’, Party Politics , DOI: https://doi.org/10.1177/1354068816678886. Kreuzer M., Pettai V. ( 2009) ‘Party Switching, Party Systems, and Political Representation’. In Heller W. B., Mershon C. (eds) Political Parties and Legislative Party Switching , Palgrave Macmillan, pp. 265– 285. Google Scholar CrossRef Search ADS Laakso M., Taagepera R. ( 1979) ‘ Effective Number of Parties a Measure with Application to West Europe’, Comparative Political Studies , 12, 3– 27. Google Scholar CrossRef Search ADS Laver M., Benoit K. ( 2003) ‘ The Evolution of Party Systems between Elections’, American Journal of Political Science , 47, 215– 233. Google Scholar CrossRef Search ADS Long J. S. ( 1997) Regression Models for Categorical and Limited Dependent Variables , Thousand Oaks, Sage. Mainwaring S. ( 1998) ‘ Party Systems in the Third Wave’, Journal of Democracy , 9, 67– 81. Google Scholar CrossRef Search ADS McLaughlin E. ( 2012) ‘ Electoral Regimes and Party-Switching: Floor-Crossing in South Africa’s Local Legislatures’, Party Politics , 18, 563– 579. Google Scholar CrossRef Search ADS Mejia Acosta J. A. ( 2004) ‘Ghost Coalitions: Economic Reforms. Fragmented Legislatures and Informal Institutions in Ecuador (1979-2002)’, PhD Thesis, University Of Notre Dame. Morgenstern S. ( 2003) Patterns of Legislative Politics: Roll-Call Voting in Latin America and the United States , Cambridge, Cambridge University Press. Google Scholar CrossRef Search ADS Müller W. C., Strøm K. ( 1999) Policy, Office, or Votes?: How Political Parties in Western Europe Make Hard Decisions , Cambridge, Cambridge University Press. Google Scholar CrossRef Search ADS O’brien D. Z., Shomer Y. ( 2013) ‘ A Cross-National Analysis of Party Switching’, Legislative Studies Quarterly , 38, 111– 141. Google Scholar CrossRef Search ADS Olson D. M. ( 2003) ‘ Conclusion’, The Journal of Legislative Studies , 9, 164– 178. Google Scholar CrossRef Search ADS Owens J. ( 2003) ‘ Explaining Party Cohesion and Discipline in Democratic Legislatures: Purposiveness and Contexts’, The Journal of Legislative Studies , 9, 12– 40. Google Scholar CrossRef Search ADS Pedersen H. H., Kaldahl M., Møgeltoft Andersen A. ( this issue) ‘ Balancing Costs of Legislative Party Switching in the Danish Parliament 1953-2015’, Parliamentary Affairs . Pinto L. ( 2015) ‘ The Time Path of Legislative Party Switching and the Dynamics of Political Competition: The Italian Case (1996-2011)’, The Journal of Legislative Studies , 21, 323– 341 Google Scholar CrossRef Search ADS Poguntke T., Scarrow S. E., Webb P. D., Allern E. H., Aylott N., van Biezen I., Calossi E., Lobo M. C., Cross W. P., Deschouwer K., Enyedi Z., Fabre E., Farrell D. M., Gauja A., Pizzimenti E., Kopecký P., Koole R., Müller W. C., Kosiara-Pedersen K., Rahat G., Szczerbiak A., van Haute E., Verge T. ( 2016) ‘Party Rules, Party Resources and the Politics of Parliamentary Democracies How Parties Organize in the 21st Century’, Party Politics , 22, 661– 678. Polk J., Rovny J., Bakker R., Edwards E., Hooghe L., Jolly S., Koedam J., Kostelka F., Marks G., Schumacher G., Steenbergen M., Vachudova M. A., Zilovic M. ( 2017) ‘ Explaining the Salience of anti-Elitism and Reducing Political Corruption for Political Parties in Europe with the 2014 Chapel Hill Expert Survey Data’, Research & Politics , 4, 1. Rahat G. ( 2007) ‘ Determinants of Party Cohesion: Evidence from the Case of the Israeli Parliament’, Parliamentary Affairs , 60, 279– 296. Google Scholar CrossRef Search ADS Rovny J. ( 2012) ‘ Who Emphasizes and Who Blurs? Party Strategies in Multidimensional Competition’, European Union Politics , 13, 269– 292. Google Scholar CrossRef Search ADS Saalfeld T. ( 2009) ‘Intra-party Conflict and Cabinet Survival in 17 Western European Democracies, 1945-1999’, In Giannetti D., Benoit K. (eds) Intra-Party Politics and Coalition Governments , Routledge, pp. 169– 186. Salucci L. ( 2008) ‘Left No More: Exit, Voice and Loyalty in the Dissolution of a Party’, Annual Meeting of the Midwest Political Science Association , Chicago. Sartori G. ( 1976) Parties and Party Systems: A Framework for Analysis , Cambridge, Cambridge University Press. Schumacher G., de Vries C., Vis B. ( 2013) ‘ Why Do Parties Change Position? Party Organization and Environmental Incentives’, The Journal of Politics , 75, 464– 477. Google Scholar CrossRef Search ADS Sieberer U. ( 2006) ‘ Party Unity in Parliamentary Democracies: A Comparative Analysis’, The Journal of Legislative Studies , 12, 150– 178. Google Scholar CrossRef Search ADS Somer-Topcu Z. ( 2009) ‘ Timely Decisions: The Effects of past National Elections on Party Policy Change’, The Journal of Politics , 71, 238– 248. Google Scholar CrossRef Search ADS Stubager R. ( 2010) ‘ The Development of the Education Cleavage: Denmark as a Critical Case’, West European Politics , 33, 505– 533. Google Scholar CrossRef Search ADS Yoshinaka A. ( 2015) Crossing the Aisle: Party Switching by US Legislators in the Postwar Era , Cambridge, Cambridge University Press. © The Author(s) 2018. Published by Oxford University Press on behalf of the Hansard Society; all rights reserved. For Permissions, please e-mail: firstname.lastname@example.org
Parliamentary Affairs – Oxford University Press
Published: Feb 22, 2018
It’s your single place to instantly
discover and read the research
that matters to you.
Enjoy affordable access to
over 18 million articles from more than
15,000 peer-reviewed journals.
All for just $49/month
Query the DeepDyve database, plus search all of PubMed and Google Scholar seamlessly
Save any article or search result from DeepDyve, PubMed, and Google Scholar... all in one place.
All the latest content is available, no embargo periods.
“Whoa! It’s like Spotify but for academic articles.”@Phil_Robichaud