Parliamentary party loyalty and party family

Introduction

Party cohesion is a crucial feature in the daily lives of parliaments. Parliamentary party cohesion directly impacts government’s survival and stability (Boucek, 2012; Saalfeld, 2009), coalition behaviour and the bargaining power of the party in public office (Giannetti and Benoit, 2009; Pedersen, 2010). It shapes the electoral strategies, influences the electoral success or failure of the party in the electorate (Kam, 2009: 130–149) and may have important consequences at the intra-party level, ranging from altering intra-party organizational rules (Kohno, 1992) to shaping the party’s positions (Harmel and Tan, 2003). Given the implication of party cohesion for the functioning of the whole polity, understanding its causes is crucial.

The degree of party cohesion varies across time, across parliaments and across parties. Explanations to this variance have included macro-level institutional factors, such as the electoral rules (Carey, 2007; Carey and Shugart, 1995; Depauw and Martin, 2009; Hix, 2004; Sieberer, 2006), the type of democratic regime (Carey, 2007; Diermeier and Feddersen, 1998) and decentralization (Bowler et al., 1999; Cordero and Coller, 2015), as well as micro-level factors linked to the MPs’ socio-demographic background and political career (Gherghina and Chiru, 2014; Kam, 2009). Party-level factors have been considered, including party size (Boucek, 2012; Sieberer, 2006), intra-party rules (Sieberer, 2006), use of ‘carrots and sticks’ by the party leader or whip (Kam, 2009; Malloy, 2003) and candidate selection processes (Cordero and Coller, 2015; Rahat and Hazan, 2001). While a few studies have examined the effect of party ideology (Close and Nuñez, 2016; Damgaard, 1995) – mostly intended as a party’s positioning in the left-right scale – the effect of party family has been so far under-explored.

This article argues that party family is a crucial element in explaining legislators’ attitudes and behaviour, and that it should be brought to the forefront of the explanation of party cohesion. This research stems from previous studies in two ways. First, in its conceptualization of the dependent variable, this article adopts a more sociological approach by focusing on the ‘loyalty’ dimension of cohesion. Loyalty is defined as a behavioural norm internalized more or less by legislators across parties and measured as MPs’ subscription to the norm that, in case of disagreement on a vote, they should follow their party line rather than their own opinion. In that way, the article examines MPs’ pre-floor attitudes, which show greater variance than actual voting behaviours. Second, this article draws on the concept of party family as defined by Mair and Mudde (1998), which appeals to the notion of party origin and to an extended understanding of party ideology, beyond the policy-based approach and left-right positioning. This research contends that a party’s family matters for its level of cohesion, as it directly and indirectly affects MPs’ attitudes towards their party. On the one hand, a party’s family’s core values and principles are shared by party members and reflected in parliamentarians’ representational style and attitudes towards their party; and on the other hand, they are reflected in the party’s organizational structure and rules, which in turn shape parliamentarians’ attitudes.

This article investigates the relationship between party family and party loyalty using attitudinal data collected by the PartiRep Comparative MP Survey ¹ among 829 MPs clustered in 49 parties and elected in national parliaments across 14 European democracies. Controlling for the effect of the oft-cited factors of party cohesion, the logit regression models bring some evidence of a party family effect that is not linked to policy or left-right positioning, but reveals different conceptions of political representation and of individual participation carried by the party families’ political culture.

Explaining parliamentary party loyalty: Why would party family matter?

Although party family is a central concept in comparative politics, scant attention has been given to its relationship with party cohesion. At best, a few scholars have suggested that ideology – which is one of the main criteria used to classify parties into families (Mair and Mudde, 1998) – could affect a party’s level of cohesion. They have theorized about the potential impact of a party’s positioning in the left-right spectrum on its level of cohesion, some assuming left-wing parties would have a higher degree of cohesion or discipline (Bailer et al., 2011: 5–6; Damgaard, 1995: 322), others showing the reverse (Close and Nuñez, 2016). However, these studies were often based on a narrow, unidimensional or policy-based understanding of ideology, whereas ideology should be more broadly understood in terms of a party’s ‘identity’ (Mair and Mudde, 1998: 220).

Why would party family matter in explaining legislators’ subscription to the norm of loyalty? Different types of explanation can be advanced. Party families carry specific values and principles that can have a direct implication for legislators’ perception of role and internalization of behavioural norms. We can expect that legislators’ attitudes will reflect the values that are at the core of the party’s family, for two reasons. First, because legislators have probably joined the party for purposive motives – although material and solidarity incentives may have played a non-negligible part (Wilson, 1973) – : they joined the party because they agreed and shared the values of the party. Second, we can expect that legislators will embody the values and norms enshrined in their party’s identity because they have been socialized to them.

Party family can also indirectly influence MPs’ attitudes and role perceptions, by impacting on the organizational structure in which they evolve, and through which they have been selected (for a discussion on the link between a party’s family and its organization, see Enyedi and Linek, 2008). In this vein, Gauja (2013) has examined the relationship between party family and intra-party democracy (IPD). According to her argument, the degree of participation that is allowed to party members within the party organization reflects different models of participation (direct, representative, delegate and consultative), and is influenced by party ideology, culture and origins. Party families thus carry specific models of democracy, which should be reflected in MPs’ attitudes towards their party.

Based on these two types of effects, several expectations can be foreseen. Some families clearly put more emphasis on individual freedom and reject state regulation and excess of political authority – this is the case of liberal parties. Liberal parties would be less cohesive because their MPs would embody the liberal values of individualism, and hence, would adopt a more individualistic style of representation. They would conceive their role as representative as an individual acting more independently from the party’s guidelines and following more often his or her own judgment.

In a similar vein, green MPs would embody the post-materialist and libertarian values of self-affirmation and self-fulfilment (Kitschelt, 1989). Given the ‘abhorrence of hierarchy’ (Cross and Katz, 2013: 3) that is at the core of green parties’ identity, green MPs would value the self-affirmation and independence of parliamentarians against the authority of the party organization. At the organizational level, green parties have rejected the traditional formal and centralized style of organization. As a consequence, ‘their political apparatus was often weak’ (Villalba, 2005: 83). MPs’ attitudes should reflect greens’ democratic ideals as well as the fact that green parties are primarily ‘parties of individual participation’ (Poguntke, 1993). In these parties, there is often ‘no need for coherent strategic behaviour of all the top levels’ (Deschouwer, 1994: 92); besides, they have been characterized by an important rotation in office that ‘denied them both stable and experienced representation’(Cross and Katz, 2013: 3). As a consequence, socialization processes should be less effective in these parties, and party loyalty should be weaker.

On the other hand, political parties which adhere to the values of law and order – in other words, to the Tradition-Authority-Nationalism (TAN) pole of the Green-Alternative-Libertarian (GAL)-TAN cleavage (Hooghe et al., 2002) – would show higher levels of cohesion. Radical right parties – and to some extent conservative parties – would put a greater emphasis on discipline and on the MPs’ fidelity to the party. The norm of loyalty would be stronger than in other parties, especially compared to liberal and green ones. From an organizational point of view, radical right parties have also rejected the traditional partisan organizations. But by contrast to green parties, they have rather been built on a hierarchical and centralized structure (Mudde, 2007) led by a charismatic leader ‘who exert[s] nearly dictatorial control over [its] organization’ (Art, 2011: 8). With a highly centralized power concentrated in the hands of the party leadership characterized by stability and low turnover, and with a very limited say granted to party members (Gauja, 2013), such organizations leave little space for dissent. These organizational features reflecting the radical right’s conception of (intra-party) democracy should prompt parliamentarians’ loyalty.

Between these two poles, social-democratic parties and, to some extent, Christian-democratic ones, would show average degrees of loyalty. Their MPs would have quite an important sense of solidarity, and would give a prominent role to the social group binding the individuals acting in a collective action. Besides, given their initial mass-based organization, these parties should be more effective in socializing their members to the norms of party solidarity – although party organizations have dramatically changed (Katz and Mair, 1994; Katz and Mair, 1995).

As far as the radical left family is concerned, hypotheses are hard to draw. This family is less coherent than other European families and is oscillating between several traditions – orthodox communism, social democracy, left-libertarian or ‘red-green ideology’ (Escalona and Vieira, 2013). Therefore, depending on the political tradition in which the radical left parties are rooted, MPs will either display highly loyal attitudes (i.e. orthodox communism, which surely places a great emphasis on party discipline and loyalty), or, at the other extreme, a low loyalty (following a green ideology). In between, parties positioned at the ‘left of social democracy’ would have similar degrees of MPs’ loyalty to those of social-democratic parties.

To sum up, party family matters for party loyalty in two ways: first, party family directly affects parliamentarians’ attitudes, as they adhere and have been socialized to the party’s family’s core values and principles; second, the party family’s core values are reflected in the party’s organizational structure, which in turn shapes the behavioural norms of parliamentarians. The analysis does not aim to disentangle which of these mechanisms is the most prominent or effective one, but provides a systematic testing of such party family effect, by controlling for the effect of other individual-, party- and system-level factors.

Data, method and measurement

Party loyalty as one dimension of cohesion

Party cohesion is a multi-dimensional concept, and can be used to refer to very different empirical phenomena – unity of voting behaviour, shared preferences within a party, and so on. Confusions abound in the literature. A first confusion is found in the works of scholars who have used party cohesion and party discipline interchangeably to describe a situation of party voting unity in the legislative arena (Bowler et al., 1999: 4; Hazan, 2003: 2). In their study of the American Congress, Cox and McCubbins (1993) refer indistinguishably to ‘party voting’, ‘intra-party cohesion’ and ‘party discipline’. In the same way, scholars analysing legislative behaviour in the European Parliament often use both terms to refer to European party groups voting in unison despite their national diversity (Hix, 2004).

A second widespread confusion relates to the status of cohesion: does cohesion describe the outcome – that is, voting unity – or does cohesion constitute a way to achieve voting unity? (Hazan, 2003: 3). The tenants of the former conception usually use cohesion as a synonym of unity and identify party discipline as a means used by the party whip or leader to enforce cohesion (Faas, 2003; Hix et al., 2005). The proponents of the latter conception rather define cohesion as one of the mechanisms through which voting unity can be reached in the parliamentary party (Andeweg and Thomassen, 2011; Hazan, 2003; Krehbiel, 1993).

According to that view, both cohesion and discipline would constitute pathways towards voting unity. Cohesion is conceived as a voluntary pathway, as a force that binds MPs together and that comes from socialization processes, solidary incentives and ideological consensus; whereas discipline constitutes a compulsory pathway, and refers to formal or informal rules and constraints implemented within the party and the parliamentary arena in order to achieve party unity when cohesion is low (Andeweg and Thomassen, 2011; Hazan, 2003; Van Vonno et al., 2014). Some scholars further differentiate between two dimensions of party cohesion: party agreement, which is defined as the homogeneity of preferences within a party or as ‘the extent to which co-partisans agree with one another when voting on legislation’ (Van Vonno et al., 2014: 4); and party loyalty, which derives from the legislators’ internalization of the norms of party unity and solidarity, and causes the MP to stick to the party line even in case of disagreement.

This study adheres to the latter approach, and focuses on the loyalty dimension of party cohesion. Instead of focusing on directly observable behaviours (e.g. legislators voting along or against their party’s position), the research opens the ‘black box’ of the party, and examines what usually remains hidden: parliamentarians’ attitudes of (non)loyalty towards their party in case of disagreement, in other words, ‘what takes place before voting decisions are made’ (Hazan, 2003: 8).

The PartiRep questionnaire includes a question ⁵ that enables to grasp the extent to which MPs consider that they should remain loyal to their party (i.e. MP should vote according to his/her party’s opinion) or not (MP should vote according to his/her own opinion), in case of a disagreement with their party. Table 1 displays the frequency distribution for this variable. The table shows quite a significant proportion of missing data (6.4%). This percentage certainly tells something about the reluctance or unease of MPs to choose between defending their own opinion and following the party logic. This level of non-response can also result from a lack of clarity in the question (see below). Note that these missing values are not specific to one country, party or party family; they are equally distributed across the sample. Given the dichotomous nature of this dependent variable, logit models are used in the analysis.

OPEN IN VIEWER

Table 1.

Frequency distribution – loyalty.

	N	Percentage	Valid percentage
Non-loyal MPs	324	39.1%	41.8%
Loyal MPs	451	54.5%	58.2%
Missing	53	6.4%
Total	827	100.0%	100%

Obviously, this question does not grasp the whole process of cohesion building. It offers MPs a binary choice, between exit (not voting along the party line) or loyalty (stick to the party line) (Hirschman, 1970). It provides limited – if not none – understanding of discussion and deliberation processes that occur within the parliamentary group to reach consensus before a vote is taken.

Another limitation relates to the identity of the ‘party’ whose line should or not be followed. Is the party line referring to the policy orientation discussed during extra-parliamentary party congresses, or to the position adopted by the parliamentary group, which depends more on government-opposition dynamics? Does the MP feel loyal to the party as a collective entity or to the leader of the group – either in the parliamentary or in the extra-parliamentary arena? This lack of clarity is problematic, as loyalty to one entity within the party organization can imply dissent to another entity. Indeed, an MP dissenting from the parliamentary party group may in fact show her/his loyalty towards the extra-parliamentary organization, or vice-versa. In that regard, a centralized organizational structure and overlapping leadership (the leader of the party in public office being also the leader of the extra-parliamentary organization) may increase loyalty, while a more diffused or stratarchical organization may dilute loyalty. Depending on the party’s family, MPs may feel more loyal to their leader – for example, in a radical right party – or to the organization – for example, in a social-democratic party – or to the party’s ideological project – for example, in a green party – and this relationship depends on both values and organizational structures.

Despite these limitations, the PartiRep question distinguishes between two ways parliamentarians conceive their role as representatives (Converse and Pierce, 1986: 664–696): the party delegate or partisan, and the trustee. In short, the party delegate ‘follows the party’s lead when making decisions’ (Önnudóttir, 2014: 538), whereas the trustee ‘makes his own decisions based on deliberation of the issues under question’ (Önnudóttir, 2014: 538). I argue that indeed, this measurement indicates how MPs view their role as representatives; but that this specific conception – as party delegate or as trustee – says a lot about the extent to which they adhere to the norm of loyalty, and by extension, about how they might contribute to the cohesion of their party.

Control variables ¹²

Individual-level control variables include socio-demographic characteristics (age and gender) and MPs’ previous parliamentary experience (seniority). Young MPs are expected to be more loyal to their party leaders who control advancement to higher position (Kam, 2009: 199). The effect of gender has been less discussed, but there is some evidence that women are less prone to ‘rebel’ (Cowley and Childs, 2003) or to ‘voice’ their disagreement to the party (Close, 2011). Seniority grasps MPs’ socialization within the parliamentary party, which is expected to increase their internalization of the norm of loyalty (Gherghina and Chiru, 2014; Kam, 2009). Seniority is operationalized as the number of years since the MP’s first election to the national parliament. ¹³ At the individual level, the analysis also takes into account the frequency of disagreement between the MP and her/or his party, given the sequential relationship between party agreement and loyalty (Van Vonno et al., 2014). Disagreement is measured as the MP’s self-reported frequency of disagreement with her/his party. ¹⁴ The variable has been dichotomized to indicate whether the MP disagrees more often (‘about once a month’ or ‘about every 3 months’) or whether (s)he does it less frequently (‘about once a year’ or ‘(almost) never’).

At the party level, the models control for the degree of party institutionalization, grasped through party size (in terms of percentage of seats in the parliament) and party age. ¹⁵ Party institutionalization should affect the MPs’ socialization towards the internalization of norms. In addition, any effect of party family could be related to the peculiar history, thus to the degree of institutionalization of specific parties – for example, green, radical left and radical right parties are smaller than traditional parties. ¹⁶ The models also include a dichotomous variable ‘government/opposition’ in order to control for potential contextual effect. Being in government might increase the pressure put on MPs to reach unity (Carey, 2007; Stecker, 2015); but at the same time might increase the risk of conflict, as being in a governing coalition can lead parties to adopt positions that contradict the original party manifesto. In order to control for intra-party power structure, I add a dichotomous variable that indicates whether the parliamentary party organization (PPO) is dominant over the extra-parliamentary organization (EPO) – for example, PPO in charge of selecting the leader and/or of establishing the party’s programme and policy priorities – or whether the PPO is dominated by the EPO. ¹⁷ I expect that when the EPO holds control over the PPO, conflicts are more likely to occur between parliamentarians and their party.

Finally, I control for two system-level factors: the degree of personalization of the electoral system and the multilevel structure of the state. ¹⁸ Electoral systems that increase the incentives to cultivate a personal rather than a party vote would increase MPs’ dissenting behaviours (Carey and Shugart, 1995). The need for cultivating a personal reputation can have different origins: it can stem from intra-party competition, increased by the openness of the ballot, or from the incentives MPs have to commit themselves in constituency work, which are particularly salient in majoritarian single-member district (Pilet et al., 2012). Accordingly, I use Mitchell’s typology (2000) of electoral systems, which combines both the electoral formula and the ballot structure to classify electoral systems in three broad categories, from the most party-centred to the most candidate-centred system (see Table 2 below).

OPEN IN VIEWER

Table 2.

Distribution – party centred, intermediate and candidate-centred systems (N = 840).

	Formula	Ballot	Percentage
Party centred	Proportional	Closed list	61.0%
		Flexible list
Intermediate	Plurality/majority	Single member	25.5%
Candidate centred	Proportional	Open list STV	13.5%

Regarding the multilevel nature of the state, the models include a dichotomous variable indicating whether the country is unitary or multilevel. On the one hand, by multiplying the levels of power, decentralization processes can increase intra-party divisions (Carey, 2007); but on the other hand, these processes also multiply the levels at which MPs can be socialized to the norm of party loyalty before entering the national parliament.

Parliamentary party loyalty and party family: Results

Table 3 presents the results of five logit models. Model 1 includes party family as the sole determinant of loyalty, with social democracy as the reference category. Almost all coefficients show a negative sign, suggesting that MPs in social-democratic parties are more loyal; but few are statistically significant. A first coefficient indicates that green MPs are significantly less likely to be loyal than social-democratic ones (at 0.05 significance level). A second coefficient shows that MPs of radical right parties are significantly more likely to be loyal (at 0.01 significance level).

OPEN IN VIEWER

Table 3.

Logit models. The effect of party family on parliamentary party loyalty.

	Model 1	Model 2	Model 3	Model 4	Model 5
Party family (ref. Social-democratic)
Radical left	−0.290			−0.678	−0.042
Green	−2.057***			−2.115***	−1.378**
Liberal	−0.327			−0.255	0.091
Christian democrat	−0.133			0.247	0.752*
Conservative	−0.236			−0.021	−0.206
Radical right	1.109#			0.666#	1.146*
Left-right party position		0.015
Left-right self-placement			−0.031
Disagreement				−0.372*	−0.342#
Woman				0.093	0.079
Age				−0.018*	−0.022*
Seniority				0.010	0.010
Party size					0.023#
Party age					0.006*
Government					−0.218
PPO dominant					0.437
Electoral system (ref. Party centred)
Intermediate				−0.726*	−1.043***
Candidate centred				−1.439**	−1.623***
Multilevel				−1.066**	−1.084*
Constant	0.480*	0.155	0.436	2.542***	1.620#
N	770	770	722	769	764
Wald χ 2 square	(6)36.02	(1).12	(1)0.29	(13)138.87	(17)160.03
Prob > χ 2 square	0.0000	0.7302	0.5873	0.0000	0.0000
Mc Fadden’s Pseudo R 2	0.0223	0.0007	0.0010	0.0887	0.1049

Note: Data are weighted by party. In models 1, 2, 3 and 4, the standard errors are adjusted for 49 clusters; in model 5, they are adjusted for 48 clusters.

^# p < 0.1, *p < 0.05, **p < 0.01, ***p < 0.001.

Models 2 and 3 test whether loyalty alternatively stems from policy positions, subsumed into a left-right orientation. Model 2 tests the effect of the party’s position in the left-right scale (as given by Benoit and Laver, 2006). The positive coefficient suggests that loyalty would be greater in rightist parties, but the relationship is not significant, and the model has a poor goodness of fit. Model 3 examines the effect of MPs’ self-placement on the left-right scale and suggests that the more rightist the placement, the less likely the MP’s loyalty. However, the significance level is not reached and the goodness of fit is weak.

The results presented in models 2 and 3 confirm preliminary analyses: loyalty is poorly related to policy orientations, either at the party or individual level. Party family seems to be a better predictor of loyalty: the pseudo R ² is much higher in model 1 than in models 2 and 3. The effect of party ideology should be examined from a ‘party family’ perspective, or at least using a conception of ideology going beyond a policy-based approach. Besides, the results indicate a clear distinction between classical and new left as well as between traditional right-wing parties and the new right. Party origin therefore certainly plays a great influence, although it overlaps to great extent with ideology broadly defined. Further, researches should try to operationalize more distinctly genetic elements from ideological ones, in order to grasp their respective influence on loyalty or cohesion at large.

Model 4 tests the effect of party family by controlling for the effect of individual- and system-level factors. The pseudo R ² is meaningfully increased, and, on the whole, the effect of party family remains similar. Adding party-level control variables (model 5) also improves the goodness of fit of the model: party characteristics appear determinant in explaining legislators’ attitudes. Positive and statistically significant coefficients are found for party size (0.01 significance) and party age (0.05 significance). Controlling for these party-level characteristics confirms the effect of the green and radical right families, though it changes the effect of other families: the coefficient for the liberals becomes positive – though not statistically significant – and the positive effect of the Christian-democrats becomes statistically significant.

Regarding the impact of other control variables, age and disagreement decreases significantly the probability of loyalty. At the party level, although the relationships are not statistically significant, loyalty is more likely when the PPO is dominant over the EPO, and when the party is in opposition. In other words, MPs would feel less at odds with the party’s policy positions when their caucus is in charge of setting the party’s program and priorities. This is in line with the idea that conflicting loyalties could be lessened in organizations centralized around the parliamentary caucus and/or leader. If the party is in government, MPs seem less prone to stick to the party line, as the party has to make compromises with coalition partners – this applies in the case of coalition governments, which represent almost all the cases included in the sample. ¹⁹ Regarding system-level control variables, loyalty is significantly less likely in multilevel systems and in more personalized electoral systems.

Figure 1 draws the effect of party age on the predicted probability of loyalty on the basis of model 5; and Figure 2 proceeds in a similar way for party size. Even when controlling for individual-, party- and system-level factors, ²⁰ the graphs show an opposition between two extremes. At one end of the spectrum, the green MPs have, by far, the lowest probability of loyalty. At the other end of the spectrum, radical right MPs have the highest probability of loyalty. A party family effect is therefore at play: controlling for the degree of institutionalization of their parties, MPs in these two families radically differ in the way they perceive their role as representatives, and in the way they conceive their relationship with their party.

OPEN IN VIEWER

Figure 1.

Effect of party age on the predicted probability of loyalty (model 5).

OPEN IN VIEWER

Figure 2.

Effect of party size on the predicted probability of loyalty (model 5).

The fact that the impact is particularly salient for these two families certainly deserves further investigation, but could partly be explained by these families’ origin along the New Politics cleavage. The explanation based only on their position on GAL-TAN issues seems insufficient: preliminary analyses have shown no significant direct effect of parties’ positioning on loyalty. However, GAL-TAN values may be reflected in these parties’ organizational rules and degree of IPD to a greater extent than for the other traditional families, which are said to have converged during the past decades – both organizationally and ideologically (Katz and Mair, 1995). These contrasting representational styles may in fact constitute an essential distinguishing feature of the green and radical right’s opposite party cultures.

Interestingly, we see that the probability of loyalty for Christian-democratic MPs is quite close to that of radical right MPs. Whereas I expected that social-democratic and Christian-democratic MPs would have similar probabilities with regard to loyalty, these two groups appear quite different. This may result from a ‘contagion’ effect: social-democratic parties have been influenced by left-libertarian and green values; whereas Christian-democratic parties have been more inspired by the radical right TAN values. Another interpretation could be that the importance attached to the group, or to the ‘community’, is still strong in Christian-democratic parties, compared to social-democratic ones.

Conclusion

Given the multiple implications that party cohesion has on the functioning of the representative system, it is crucial to better assess and understand its determinants. Among these determinants, scholars have highlighted the impact of institutional rules and individual-level factors. Party-level variables have also been taken into account, but were often limited to some measures of party size and crude indicators of formal procedures of candidate selection. One of the aims of this article was to bring back party-level explanations of party cohesion in the legislative arena. Indeed, individual legislators do not exercise their role of representatives in a vacuum: they are part of an organization – that is, their party – that shapes their attitudes. Party family appeals to the main constitutive elements of a political party – that is, its ideology and origin. Paradoxically, its impact on the legislators’ attitudes and behaviours has rarely been investigated. This article has attempted to fill this gap.

This article stemmed from previous studies in two ways. First, in its conceptualization of the dependent variable, the analysis has conceived party cohesion as a sequential process, and has examined the extent to which legislators have internalized the norm of party loyalty. Such a measure of cohesion merges with the traditional opposition made in the representation literature between two roles: the party delegate and the trustee. Second, this study has relied on the notion of party family, which permitted to grasp both genetic and ideological components of a party, away from simplistic categorizations based on parties’ policy positions. I have argued that a party’s family carries a set of core values and principles that can directly and indirectly affect parliamentarians’ relationship to their party.

The analysis has indeed demonstrated the existence of a ‘party family effect’ that goes beyond the mere distinction between parties of the left and parties of the right. Moreover, the analysis has highlighted a fundamental opposition between two families: the green and the radical right families. These two families have crystallized around the New Politics cleavage or new cultural conflict that emerged in the 1960s–1970s, where both families intended to respond to the failure of the modern representative system. Whereas the greens ‘wanted to promote a model of participative democracy’ (Villalba, 2005: 82), the radical right rather rejected the representative system, and promoted more authoritarian forms of government. According to the results presented in this article, it seems that these two families have developed opposite party cultures, which are reflected into radically different styles of representation. On one side, the norm of party loyalty seems to be greatly entrenched in radical right parties, in accordance with the values of order and discipline. At the opposite, green MPs seem to value much less the norm of party loyalty, and seem to operate in accordance with the values of self-fulfilment and self-affirmation promoted by the green project.

Building on different fields of research, this study has relied on perspectives from the party organization literature to shed new light on legislative dynamics, and on the process of representation. However, the use of the PartiRep database, which was not designed in that purpose, has involved limitations. Specifically, the question used to measure loyalty has not permitted to disentangle which of the party or leader, and which of the parliamentary or extra-parliamentary organization, MPs feel loyal to. Another limitation relates to the lack of data in the questionnaire related to the MPs’ long-term experience in the organization. Length of membership, membership of other organizations and indicators of MP’s attachment to the (extra-)parliamentary party organization certainly influence MPs’ loyalty. Better understanding legislators’ attitudes therefore requires digging more into party organizational structures and membership dynamics.

Nevertheless, the use of a quantitative methodology has permitted to put forward general patterns that deserve to be further investigated. Qualitative methods and case studies might bring additional understanding to the dynamics that are at play between party ideology, party origin and the legislators’ relation to their party. Many questions are still to be addressed. What does loyalty mean for parliamentary party members? Are they loyal to their party organization and ideas, or to their party leader? How is loyalty built? How do the parties’ origins, organizational structure and political culture distinctly affect the relationship between MPs and their party, but also between MPs and their voters? What is the role of values and ideology?

By exploring the relationship between party family and MPs’ loyalty, this research has demonstrated that voting for a party involves more than voting for a manifesto: it is also about voting for a specific style of representation that partly stems from the party’s original values and principles. From a normative point of view, this is crucial, as it has been shown that specific styles of representation affect several dimensions of the representative processes – for example, the degree of congruence between voters and their representatives (Önnudóttir, 2014). Additionally, party cohesion greatly influences the policy-making process, and more globally, the stability and viability of the political system.