Party loyalty and electoral dealignment

Abstract

Representative democracy is party democracy. Parties vote together in the legislative arena; party labels act as information shortcuts in the electoral arena, tying together co-partisan legislators’ re-election prospects. But the utility of party labels is weakened by waning party identifications in the electorate. Partisan dealignment therefore risks undercutting party loyalty on the part of backbenchers. Combining district-level data on electoral volatility and new data from the PARTIREP survey of legislators in 15 advanced industrial democracies, it is demonstrated that party loyalty is lowest where partisan dealignment is strongest – even after accounting for backbenchers’ policy preferences, whether they represent the ruling party or the opposition, and their campaign strategies. Our results have important implications for the sustainability of current models of representative democracy.

Keywords

Electoral dealignment party loyalty political representation

Introduction

Party unity is near perfect in parliamentary democracies, where the cabinet can be voted out of office by a plurality in the legislature by means of an ordinary or constructive vote of no confidence (Strøm, 2000: 265). Co-partisan legislators vote in near perfect unity on recorded votes; measurements of party unity frequently exceed 99 percent (e.g. Carey, 2009; Depauw and Martin, 2009; Sieberer, 2006). Norms of party loyalty are widely shared among legislators in parliamentary democracies (Crowe, 1983): legislators feel they should vote the party line, and not harm the party – even when they disagree.¹ Such norms of party loyalty are deeply rooted in the ‘responsible party’ model (APSA, 1950; Schattschneider, 1942). That is, parties serve to solve collective action problems in both the legislative arena and the electoral arena (Bowler, 2000; Bowler et al., 1999). On the one hand, party unity enables the parties in office to command a majority of votes in the legislature and to enact the public policies they promised during the campaign. On the other, party unity enhances the value of the party cue in the electoral arena. The party label acts as a brand name, communicating to voters what the candidates running under the label will do once elected (Cox and McCubbins, 1993; Kiewiet and McCubbins, 1991).

Party loyalty, Jackson (1968: 300) noted, is invoked in the House of Commons along with assertions about winning elections. Members of Parliament should vote the party line because they were elected on a party ticket. What’s more, the party’s success or defeat in the polls critically depends on it being united (see also Andeweg and Thomassen, 2005; Crowe, 1986; Kam, 2009). Backbenchers are loyal to the party in so far as they depend on it to obtain their career goals of policy, office and votes (Cox and McCubbins, 1993; Hechter, 1987; Saalfeld, 1995). But party identifications are waning in the electorate across most advanced industrial democracies (Dalton, 2004; Dalton et al., 2000). Fewer people vote on the basis of party predispositions, reflecting the erosion of long-standing societal cleavages, growing individualization, and rising levels of education among the population (Andeweg, 2003; Dalton, 2004; see also Saward, 2010). Party membership is declining (Mair and van Biezen, 2001; Van Biezen et al., 2012), as is turnout in elections (Blais and Dobrzynska, 1998; Gray and Caul, 2000). More people switch parties between elections, increasing electoral volatility and instability (Crewe and Denver, 1985; Pedersen, 1979; but see Bartolini and Mair, 1990).

There is a long-standing debate as to whether party dealignment in the electorate will further erode parties in the legislative arena (Bowler, 2000; Thies, 2000). Combining district-level electoral volatility data and data from the PARTIREP cross-national survey in 15 advanced industrial democracies, we demonstrate that legislators who compete for re-election in conditions of high dealignment do not defer to the party as readily as legislators who compete where dealignment is low. Partisan dealignment in the electorate undercuts the party label’s utility to legislators in the electoral arena and thereby reduces their loyalty to the party in the legislative arena. The argument proceeds in four stages. The first section sets out the hypotheses regarding the effect of partisan dealignment on party loyalty among legislators, controlling for other institutional dependencies and situational resources (see Sieberer, 2006). The second section presents the data from the PARTIREP cross-national survey of national legislators in 15 advanced industrial parliamentary democracies. The third section argues the importance of party loyalty and describes the dependent variable. The fourth section then conducts a multivariate analysis assessing the validity of our main hypothesis and demonstrates that partisan dealignment does indeed undercut party loyalty among legislators. Further implications of our main findings are addressed in the concluding section.

Partisan dealignment and the electoral incentive to dissent

Party identifications in the electorate are waning. The number of strong party identifiers in nine European countries has decreased by over 19 percent since the 1970s, Schmitt and Holmberg (1998: 126 f.) noted. The number of people without party attachment, Dalton (2000: 27) calculated, has increased by over 10 percent in the same period – though there are marked differences between countries. Membership of political parties by the late 1990s has been reduced to a third of the numbers recorded in the 1960s (Mair and van Biezen, 2001: 9). Parties’ ability to mobilize voters in the electoral arena has been undermined in this manner (Dalton et al., 2000). To make a ‘reasoned’ voting decision, voters typically use a combination of party-based and candidate-based information shortcuts (Downs, 1957; Popkin, 1991). Party-based cues refer to the party’s record in office, the state of the economy, or the legislation it has proposed; candidate-based cues originate in candidates’ individual qualities, positions and actions (Cain et al., 1987: 9). Fewer people vote on the basis of party; by contrast non-partisan voters have grown in number, increasing electoral volatility between general elections (Dalton et al., 2000; Wattenberg, 1998; but see Holmberg and Oscarsson, 2013).

Whether waning party identifications in the electorate will have further implications for party unity in the legislative arena is rooted in the long-standing debate on cohesion and discipline (Bowler, 2000; see also Smith, 2000). The ‘two-arena model’ argues that cohesive parties emerge in the legislative arena largely as a consequence of parties’ utility for mobilizing voters. That is, candidates self-sort into parties and run under the label of the party they share most ideological and policy positions with (Andeweg and Thomassen, 2011; Bowler et al., 1999). Cooper and Brady (1981) further argued that party unity waxes and wanes with intra-party cohesion and inter-party polarization, mirroring party identifications in the different districts. As such, the ‘two-arena model’ posits that partisan dealignment in the electorate will undercut cohesion in the parliamentary party. While cohesion of this kind originates in the electoral arena, discipline by contrast is enforced by the monitoring and sanctioning of those in the legislature (Hazan, 2003). The ‘one-arena model’ emphasizes the procedural advantages that parties have in the legislative arena over individual legislators, thereby stating parliamentary parties’ ability to maintain unity even in the face of dealignment in the electorate (Bowler, 2000; Cox and McCubbins, 1993). The party leadership may rely on its control of the legislative agenda to avoid recorded votes on issues on which it is divided. The leadership may further reward loyalty with plump committee assignments and sanction dissent – ultimately by (the threat of) exclusion from the party (Kam, 2009).

Why will partisan dealignment in the electorate weaken party loyalty? Legislators will defer to the party, Hechter (1987) points out, to the extent that they depend on it to get what they want in terms of policy, office or votes. They are loyal to the party as long as they recognize the utility of the party label. They are elected on a party platform and without the party’s support few would be returned to the legislature. Once in power the party will enact the policies pledged in the platform and victory or defeat in the polls will depend on its ability to present a united front in doing so (Crowe, 1986). That is, legislators are prepared to trade off (some of) their individual preferences in return for greater party gains in the electoral arena (see Aldrich, 1995). But partisan dealignment in the electorate undermines any gains they might expect from the party label. As a result party loyalty will suffer. Distancing themselves from unpopular party policies may win them a personal vote (Carey, 2009). The media attention that is typically given to intra-party dissension aids a legislator’s name recognition and public approval (Kam, 2009). Dealignment, Kam (2009) observed, may explain the frequency and extent of dissent in Canada and the United Kingdom as compared to Australia and New Zealand. Dealignment also coincides with the sharp increase in dissent in the United Kingdom in the 1970s. In that period the proportion of very strong party identifiers in the electorate declined by 15 percent (Kam, 2009: 73; for alternative explanations see King, 1981; Norton, 1978; Schwarz, 1980). This personal vote won by defecting from party ranks is not restricted to electoral institutions using single-seat districts, moreover. Legislators with strong local ties break party unity more often, Tavits (2009) found, in open and closed list pr systems (i.e. Estonia, Finland, Norway, Poland and Switzerland). They are also more electorally successful in Estonia’s multi-tiered open list system (Tavits, 2010). Breaking party unity further wins preference votes in Slovakia’s flexible-list system (Crisp et al., 2013).

Main hypothesis: Partisan dealignment in the district decreases the likelihood that legislators think they should vote the party line in case of conflicting opinions.

A number of alternative explanations need to be accounted for, moreover. Previous studies have rooted party unity in shared policy preferences within the parliamentary party, in government participation, and in the electoral institutions (for a review of the literature, see Andeweg and Thomassen, 2011; Depauw and Martin, 2009; Sieberer, 2006). Government participation and the procedural advantages it brings can be tied to the one-arena model, whereas the incentives that electoral institutions generate on the part of legislators are central to the two-arena model. Both models seek to account for legislators’ policy preferences – depending on their conception of the origin of policy preferences.

Do co-partisans act in unison in the legislature, Krehbiel (1993: 238) pointedly asked, ‘because of their agreement on policies’, or ‘in spite of their disagreement’? The notion is intuitive: because co-partisans share similar policy preferences, they will vote alike (see, e.g., Andeweg and Thomassen, 2011). Candidates self-select into parties, donning the party label that best suits their policy preferences (Bowler et al., 1999). Parties further select candidates who are closest to the party in policy preferences (Hazan and Rahat, 2010). More importantly, party loyalty is strongest among those legislators whose preferences most resemble the party positions – even in the few cases when they disagree. Not only are they in agreement more often. But because they stand to gain most in terms of policy when the party enacts its platform, they will feel a sense of duty to toe the party line – even when they disagree. By contrast, legislators whose preferences are more distant from the party will be more prepared to vote their own opinion. Krehbiel (1993) is adamant that any indication of parties’ ability to constrain individual legislators must therefore first account for the latter’s policy preferences. Above some level of cohesion, discipline is redundant (Bowler et al., 1999: 5; Hazan, 2003: 6 f.). Below some level of cohesion, the leadership lacks the mandate to enforce discipline. Discipline, Aldrich and Rohde (1997) posited, is conditional upon cohesion: when parties’ positions are homogeneous and far apart, members of the parliamentary party will agree to party unity norms and seek to strengthen the partisan organization, presenting the leadership with the resources, if necessary, to keep legislators in line.

Hypothesis 2: Legislators whose preferences most resemble party positions are more likely to think they should vote with the party in case of conflicting opinions; legislators who are more distant from the party are more likely to think they should vote their own opinion in case of conflicting opinions.

‘Discipline starts’, Hazan (2003: 3) explained, ‘where cohesion falters’. In the face of widespread and persistent disagreement in the parliamentary party, the leadership must obtain unity by other means. It may rely on the promise of career advancement, and ultimately on discipline to keep dissenting legislators in line (Kam, 2009; see also Jackson, 1968). In parliamentary systems, in particular, the cabinet’s very survival critically depends on its ability to command a majority in the legislature. On the one hand, party loyalty is commonly thought to be stronger in governing parties (e.g. Saalfeld, 1995). Dissent in the opposition may undercut its public record as a viable alternative, but dissent in the ruling parties may bring down the government and result in elections. On the other hand, opposition parties have frequently displayed the closest unity in roll-call voting (Sieberer, 2006). The finding might be but an artefact of the measurement, however. The opposition, it seems, finds it easier to duck votes on which there is disagreement; the ruling parties cannot outrun their responsibility in a similar manner (Jackson, 1968). Recorded votes are more common, moreover, when there is disagreement in the government ranks (Saalfeld, 1995). Not only is party loyalty expected to be stronger in ruling parties than in the opposition, party loyalty will also grow stronger the smaller the margin of seats that the government has in the legislature over the opposition. If the margin is narrow, government backbenchers recognize the importance of presenting a united front, when a handful of dissenters can jeopardize the outcome of the division. If the margin is large legislators who disagree do not feel the same sense of duty to toe the party line and feel they may safely dissent (Crowe, 1980).

Hypothesis 3a: Legislators in governing parties think it more important to vote with the party in case of conflicting opinions than legislators in the opposition.

Hypothesis 3b: Legislators think it more important to vote with the party in case of conflicting opinions, as the government’s margin in the legislature grows narrower.

Third, party loyalty has been related to the notion that individual legislators face ‘competing principals’ in the electoral arena (Carey, 2009). Individual legislators are accountable to the party leadership in the legislative arena; they may be individually accountable to voters in addition. That is, individual legislators then face two principals who may present them with competing demands. A distinction is frequently made between that part of their support that is tied to the party label and the part that they win on the basis of their qualities, qualifications and actions (Cain et al., 1987). Individual legislators will be loyal to the party even when they disagree, when the party vote exceeds the personal vote (Carey and Shugart, 1995). By contrast, when the personal vote is the larger, individual legislators may benefit from taking positions that run counter to the party line – in so far as backbenchers and leaders cater to different electoral audiences, Kam (2009) observed. What’s more, defying the party line will endear legislators in particular with non-partisan voters. Dissenting legislators’ name recognition is greater than loyal legislators’ in New Zealand and the United Kingdom, Kam (2009) concluded. More importantly, non-partisan voters respond more positively to dissent than voters with strong partisan identifications. Thus, individual legislators are least prepared to defer to the party when they feel their objective of re-election is best served cultivating their personal vote and/or chasing non-partisan voters rather than mobilizing the party faithful (Rohrschneider, 2002). Their campaign strategies arguably best capture their electoral incentives as they understand them (Zittel and Gschwend, 2008: 288).

Hypothesis 4a: Legislators who run a party campaign are more likely to think they should vote with the party in case of conflicting opinions; legislators who run a personalized campaign are more likely to think they should vote their own opinion.

Hypothesis 4b: Legislators who run a campaign mobilizing the party faithful are more likely to think they should vote with the party in case of conflicting opinions; legislators who run a campaign chasing non-partisan voters are more likely to think they should vote their own opinion.

The PARTIREP cross-national legislator survey

To demonstrate how partisan dealignment in the electorate might jeopardize party loyalty, comparative data are used from the PARTIREP cross-national study hosted by the Vrije Universiteit Brussel. National legislators in 15 advanced industrial democracies have been surveyed. Included are Austria, Belgium, France, Germany, Hungary, Ireland, Israel, Italy, The Netherlands, Norway, Poland, Portugal, Spain, Switzerland and the United Kingdom.² The sample of countries is balanced in geographical terms and varies in respect of the unitary or multi-level character of their state structure. In addition, the selection of legislatures displays a wide variation in electoral institutions. A diversity of electoral formulae and ballot structures governs the allocation of seats to parties and candidates. Respondents are elected in districts of various magnitudes – ranging from single-seat districts to at-large nationwide or region-wide constituencies (Table 1).

Table 1.

The PARTIREP cross-national legislator survey in 15 advanced industrial democracies.

	Response rate
	Responses	Proportion
Austria	55	0.30
Belgium	70	0.47
France	50	0.09
Germany	134	0.22
Hungary	99	0.26
Ireland	34	0.22
Israel	39	0.33
Italy	45	0.07
Netherlands	65	0.43
Norway	46	0.27
Poland	55	0.12
Portugal	76	0.33
Spain	104	0.30
Switzerland	49	0.25
United Kingdom	62	0.10

Note: The table displays the response rates calculated as the ratio between the number of respondents and the size of the population.

Members of the selected legislatures were contacted between March 2009 and January 2012, depending on election cycles. The survey was mainly web-based: legislators typically received an introduction letter and email inviting them to participate. They also received at least two online reminders (excluding ‘hard’ refusals) and a final invitation by telephone in many countries. The option was also offered to them to fill out a print-and-mail-back questionnaire or ask for a personal interview. The selection concentrates on backbenchers, moreover. In a large number of countries ministers are replaced in the legislature upon taking office; those that continue to serve in the legislature will be dismissed when they breach party unity. As such, the rules of the game are different for them compared to backbenchers. Table 1 reports the number of respondents in each country. The overall response rate was about one in four. The legislators who responded do not markedly differ from the survey population in terms of party affiliation. Responses are nonetheless weighted by parliamentary party group in each parliament to correct for possible biases. All models were subsequently tested both weighted and unweighted and the results reported are robust.

The dependent variable: Party loyalty

Cooperation and competition in the legislature are structured by norms that are readily accepted by most legislators and the norms regarding party loyalty are among the most widely shared in the legislative arena (Crowe, 1983).³ Legislators should toe the party line in roll-call votes and will vote in almost perfect unity. In the confinement of the parliamentary party, legislators are free to give voice to the concerns of those they represent and to the reservations they might have about the policies advocated by the party. In the confinement of the party, legislators may seek to influence the party position and even threaten with dissent when they are not heard (Jackson, 1968). But they are not to disclose internal party discussions to the public, to take political initiatives that have not been authorized by the party, and they are not to vote against the party line. Frontbenchers and backbenchers in the House of Commons widely agree that breaches of discipline are more important the more visible they are to the public, and when they jeopardize the government’s ability to command a parliamentary majority (Crowe, 1983).

Party loyalty is tapped into by a dilemma (Andeweg and Thomassen, 2011; see also Wahlke et al., 1962). Individual legislators are asked to indicate how they think they should vote when there is disagreement. Should they toe the party line against their own opinion, or should they vote their own opinion in defiance of the party line? It is the assumption that in the absence of controversy over the issue in his environment a legislator simply votes the consensus (Kingdon, 1977). If there is controversy, a legislator has to consider whether or not the extraordinary circumstances would excuse a breach of party unity. That is, he has to ponder his electoral gain against the damage dissent would do to the party – and to his own career advancement. Party loyalty of this kind, in other words, can be thought of as the penultimate stage in individual legislators’ sequential decision-making process that consists of party cohesion, party loyalty and party discipline (Andeweg and Thomassen, 2011). While legislators ultimately vote in near perfect party unity in roll-call votes, there is clearly more dissension in the earlier stages of the decision-making process (as is evidenced by their floor statements, for instance; see Sieberer, 2013). What’s more, by restricting our scope to roll-call voting we are missing much that is going on below the surface (Table 2).

Table 2.

Summary statistics of the dependent variable.

	Vote party's opinion (vs. own opinion)
Austria	Vote party's opinion (vs. own opinion)	49.8
Belgium	66.0
France	31.4
Germany	46.4
Hungary	62.6
Ireland	80.9
Israel	53.7
Italy	35.6
Netherlands	88.4
Norway	90.6
Poland	48.7
Portugal	54.3
Spain	82.4
Switzerland	12.8
United Kingdom	45.0
All countries	56.8

Note: The table displays the percentage of respondents who feel that a legislator should vote according to his/her party's opinion in case of conflict with his/her own opinion.

That is not to say that legislators are in doubt about what is expected of them. There is widespread consensus among legislators about these norms. Over 70 percent of legislators in the Nordic countries are adamant that party discipline should remain ‘as it is’ (Jensen, 2000). Our data confirm the consensus: 79 percent of the national legislators in the survey estimate that party discipline on roll-call votes should remain as it is. The legislators who favour more party discipline outnumber those supporting less party discipline two to one, moreover. But the conflicting options open to them might pressurize individual legislators to ignore the tenets of party unity for the moment. Table 2 reports for each country in the selection the percentage of legislators who insist that they should toe the party line even if their personal judgement tells them otherwise. About 43 percent think that they should vote their own opinion in the case of conflict with the party; 57 percent feel they should vote the party line regardless. The proportions differ markedly between the countries. Party unity norms hardly seem to allow any exception in Spain, The Netherlands, Norway and Ireland, whereas more than one in three legislators feel they should vote their own opinion against the party line in France, Italy and Switzerland.

The effect of partisan dealignment

Does partisan dealignment in the electorate weaken party loyalty as the two-arena model suggests; or, alternatively, does party loyalty remain strong in spite of dealignment? Recall that party loyalty is measured by means of a dichotomous variable: ‘1’ indicating that the legislator feels he should vote the party line and ‘0’ referring to a vote in accordance with the legislator’s own opinion. Given the binary nature of the dependent variable, the data are best analysed using logistic regression models. To take into account the hierarchical nature of the data, multi-level analysis is used, lest we risk underestimating the standard errors (see Steenbergen and Jones, 2002).⁴ To improve the parameter estimates, penalized quasi-likelihood (PQL) iterative estimation with a second-order linearization is used. There are ‘cultural expectations’ (Saward, 2010) regarding party loyalty that representatives will share, first, in a given country and, second, in a given party.⁵ The null model in particular demonstrates that the intra-class correlation is significant and remarkably high: about 23 percent of the variance in party loyalty can be attributed to cross-national differences and 8 percent to differences across political parties. Other modelling strategies, including fixed country effects, yield similar results, moreover (Table 3).

Table 3.

Summary statistics of the main independent variables.

	Electoral volatility in the district		Policy distance from the party		Personal campaign	Persuade floating voters
	Mean	s.d.	Mean	s.d.	Percentage	Percentage
Austria	13.5	1.2	2.1	1.5	10.1	46.1
Belgium	13.5	1.2	1.1	0.8	30.9	58.6
France	21.5	8.0	0.7	0.7	55.3	11.7
Germany	11.9	3.4	1.0	0.8	39.9	35.4
Hungary	15.6	3.7	1.1	0.9	28.1	47.2
Ireland	16.5	3.9	1.6	0.9	48.7	34.7
Israel	27.7	0.0	1.4	1.4	30.6	39.1
Italy	12.7	3.2	1.1	0.7	35.2	59.7
Netherlands	20.2	0.0	0.9	0.8	17.1	46.4
Norway	19.0	2.9	0.9	0.8	3.6	36.9
Poland	30.0	2.8	1.1	0.7	52.4	47.1
Portugal	11.4	1.9	1.1	0.9	11.2	47.7
Spain	8.3	2.9	1.0	0.9	2.7	47.5
Switzerland	10.2	3.2	1.0	0.7	31.6	48.9
United Kingdom	9.5	2.8	1.0	0.7	40.6	38.0
All countries	14.9	6.6	1.1	0.9	28.3	43.4

Note: The table displays the country averages and standard deviations of electoral volatility and legislators' policy distance from the party on the left–right dimension. The table further reports the percentage of respondents who think re-election is won by running a personal rather than a party campaign and who think re-election is won by persuading floating voters and supporters of other parties rather than mobilizing the party faithful.

Model 1 includes the primary independent variable dealignment in the electorate. Dealignment is measured at the district level by averaging Pedersen’s (1979) indices, or half the total net change in parties’ vote percentages, across the three elections prior to the Parliament that was surveyed. To account for the intuition that changes in the high range will be less influential than changes in the low range, the decimal logarithm is taken. Table 3 clearly indicates that the countries in the selection are at different stages of dealignment: electoral volatility falls below 10 percent in Spain and the United Kingdom, but approaches 30 percent in Israel and Poland. The impact of a legislator’s ideological proximity to his or her party is gauged in model 2. Included are the party’s position on the left–right dimension and a legislator’s distance to that party position in absolute terms. To estimate the party position as accurately as possible, we have calculated the party-level average of the scores respondents have assigned their party on an 11-point left–right scale. The position of an individual legislator is measured by his/her self-placement on an identical scale. Whereas some legislators do not deviate from their party’s position at all, the ideological distance from their party amounts to 5 points on the scale for others. Model 3 accounts for differences between government and opposition parties as well as for the margin that the government has in the legislature over the opposition. The margin is measured as the decimal logarithm of the number of seats the government commands in excess of a majority. Minority cabinets are set to zero. Model 4 further looks into the assumption that party loyalty is related to legislators’ re-election strategies. These strategies are measured by means of two 5-point scales: the first opposes running a party campaign (1) to running a personal campaign (5); the second contrasts mobilizing the party faithful (1) to persuading floating voters and supporters of the other parties (5). About 28 percent of all respondents pursue a personal campaign and 43 percent chase non-partisan voters – though cross-national variation is large (see Table 3). To test the robustness of our findings, the combined effect of all explanatory variables is estimated in model 5. In addition, a number of alternative explanations is controlled for in model 6: the electoral institutions (i.e. the ballot structure and district magnitude), a legislator’s seniority in the legislature, the number of seats in the parliamentary party, and the number of parties competing in the district. The estimated regression coefficients and robust standard errors are presented in Table 4 (Figure 1).

Table 4.

Dealignment and party loyalty in 15 advanced industrial democracies.

	Model 1		Model 2		Model 3		Model 4		Model 5		Model 6
	b.	s.e.	b.	s.e.	b.	s.e.	b.	s.e.	b.	s.e.	b.	s.e.
Electoral volatility in the district (logged)	–2.169	(0.703)***							–1.516	(0.663)**	–1.563	(0.675)**
Policy distance from the party			–0.200	(0.087)**					–0.210	(0.097)**	–0.210	(0.095)**
Party's left–right position			0.063	(0.047)					0.087	(0.053)*	0.083	(0.055)
Governing party					0.345	(0.192)*			0.366	(0.215)*	0.359	(0.246)
Government margin (logged)					–1.330	(0.304)***			–1.358	(.362)***	–1.370	(0.341)***
Personal campaign (vs. party campaign)							–0.254	(0.114)**	–0.262	(0.116)**	–0.263	(0.114)**
Chasing non-partisans (vs. mobilizing the party faithful)							–0.169	(0.071)**	–0.158	(0.072)**	–0.161	(0.070)**
Preferential system											–0.318	(0.520)
District magnitude (logged)											–0.083	(0.124)
Preferential system * district magnitude											0.411	(0.323)
Seniority (in years)											–0.006	(0.011)
Party's no. of seats in Parliament (logged)											0.091	(0.245)
No. of parties running in the district											–0.005	(0.027)
Constant	2.765	(0.859)***	0.134	(0.411)	1.182	(0.375)***	1.483	(0.472)***	4.003	(1.042)***	4.082	(1.229)***
σ² (country)	1.448	(0.585)***	1.162	(0.459)***	0.260	(0.166)***	0.987	(0.401)***	0.370	(0.223)***	0.374	(0.220)***
σ² (party)	0.398	(0.121)***	0.395	(0.125)***	0.356	(0.126)***	0.423	(0.155)***	0.423	(0.155)***	0.422	(0.151)***
Pseudo log-likelihood	–513.95		–520.56		–515.97		–515.25		–496.50		–494.32
LR (df)	20.38	(1)***	7.17	(2)***	16.35	(2)***	17.78	(2)***	55.28	(7)***	59.64	(13)***
Nagelkerke R²	0.03		0.01		0.03		0.03		0.09		0.10

Note: Entries are the parameter estimates and robust standard errors (in parentheses) of three-level random-intercept logistic regression models. The dependent variable is whether, in case of conflict, a legislator should vote according to his/her party's opinion (1) or his/her own opinion (0). The number of observations equals 814 at the individual level, 82 at the party level, and 15 at the country level. *p≤0.10, **p≤0.05, ***p≤0.01, using two-tailed t-values.

The results in Table 4 firmly support our main hypothesis. The effect of district-level volatility, model 1 demonstrates, is negative and highly significant. Figure 1 demonstrates a steep downwards slope in the predicted probabilities of party loyalty for various values of partisan dealignment. In other words, party loyalty grows weaker in the face of partisan dealignment in the electorate. Electoral volatility undercuts the utility of the party label to legislators, making them more open to dissent. The size of the effect is substantively large, moreover. Increasing volatility from one standard deviation below the mean to one standard deviation above reduces the probability that a legislator thinks he should vote with the party in case of conflicting opinions by 16 percent. That probability ranges from 78 percent in districts of minimum volatility to 36 percent in districts of maximum volatility. Where party identifications are waning most and more voters turn to candidate-centred cues, legislators do not benefit from deferring to the party. At the least they are wavering in their loyalty and commitment not to rock the boat. At the most they may be prepared to cultivate a personal vote on the basis of their independent voting record.

Partisan dealignment strongly differs across countries. But even alternative modelling strategies including country fixed effects demonstrate partisan dealignment’s strong and significant negative effect weakening party loyalty – further boosting confidence in the robustness of our findings. What’s more, within-country variation in partisan dealignment similarly weakens party loyalty. In 10 of 13 countries (excluding Israel and The Netherlands) the effect of partisan dealignment on party loyalty is in the expected negative direction.⁶ Lower numbers of observations within countries mean that effects frequently fail to reach standard levels of statistical significance. But where within-country variation in partisan dealignment is strongest – in France, Germany, Spain, as well as in the United Kingdom where variation across districts in dealignment is not as great – the effect is significant at the 10 percent level. The countries that run counter to our theoretical expectations – Austria, Belgium and Norway – include among the lowest numbers of constituencies and therefore the lowest numbers of independent observations of dealignment.

What’s more, partisan dealignment reduces party loyalty even when accounting for legislators’ policy preferences. Legislators whose preferences resemble the party positions closest, model 2 demonstrates, feel strongest that they should toe the party line even when they disagree. Legislators who are more distant from the party are more prepared to defy the party and feel they should be able to vote their own opinion when they disagree, confirming hypothesis 2. The probability that they defer to party unity norms decreases by about 4 percent for each point they are further removed from the party position on the 11-point left–right scale. At the maximum, that probability drops by 20 percent – when compared to the party loyalty of those who perfectly match their party position. The effect is independent from the party’s position, moreover. In spite of long-standing observations that party unity is most in line with socialist ideologies (see Jensen, 2000; Oüzbudun, 1970), the position of the party in the left–right policy space does not shape legislators’ attitude regarding party unity. Most importantly, controlling for legislators’ ideological proximity to the party does not detract from our main findings with regard to partisan dealignment.

Party loyalty is strongest in the ruling parties. Model 3 demonstrates that legislators are more likely to think they should toe the party line when they support the cabinet, confirming hypothesis 3a. The probability of being loyal to the party is 7.5 percent greater in governing than in opposition parties. As such, that party unity in roll-call voting is frequently found to be highest in the parties of the opposition can be attributed to the measurement. Opposition parties are able to duck votes in the case of dissension in a manner that government parties cannot. Party unity is commonly overestimated in smaller parties, moreover (Desposato, 2005). In fact, legislators feel they should toe the party line in case of conflicting opinions in particular when the margin the cabinet has in the legislature over the opposition is narrow, supporting hypothesis 3b. Increasing the cabinet’s margin by one standard deviation over the mean decreases the probability of legislators being loyal to the party by 21 percent. The probability of being loyal ranges from 79 percent in the case of a minority cabinet to 32 percent in the case of a cabinet commanding a comfortable margin of over 50 seats more than the combined opposition. When the margin grows larger, more legislators feel they should be able to vote their own opinion in case of conflicting opinions – as doing so will not jeopardize the cabinet’s ability to command a majority in the legislature. Opposition legislators, however, display similar concerns: the narrower its margin the greater the opportunity to topple the cabinet – provided the opposition is united (Figure 2).

Finally, partisan dealignment undercuts the party label’s utility in the electoral arena, creating incentives on the part of individual legislators to win a personal vote on the basis of an independent voting record. The incentive to cultivate their personal record, rather than the party record, model 4 points out, reduces the sense of duty legislators have that they should vote the party line even when they disagree. The more legislators think that their electoral fortunes would be better served by running a personal campaign rather than a party campaign, the stronger they feel they should be able to vote their own opinion when they disagree with the party – thereby confirming hypothesis 4a. Figure 2’s left-hand panel demonstrates this downwards slope in the predicted probabilities of being loyal for different values on the scale of personal campaigning. The probability of a legislator being loyal to the party decreases by 19 percent when he ran a personal rather than a party campaign. The more legislators chase non-partisans, furthermore, the more they feel they should be able to vote their own opinion. The more legislators mobilize the party faithful, the more they feel they should vote the party line even when they disagree, supporting hypothesis 4b. Figure 2’s right-hand panel again plots this downwards slope when campaigns target non-partisans rather than the party faithful. The probability that a legislator will be loyal to the party decreases by 13 percent when he ran a campaign chasing non-partisans rather than a campaign mobilizing the party electorate. That is, through different data and different methods we find support for Kam’s (2009) observation that dissent is tied to the incentive to cultivate a personal vote – in particular among non-partisan voters.

It should be noted that legislators’ perception of the incentive to cultivate a personal vote is not linked to the electoral institutions in as straightforward a manner as could be expected (see model 6). Contrary to Carey’s (2009) argument, party loyalty does not seem to be lower in systems that allow for preferential voting (Karvonen, 2004).⁷ Nor does district magnitude have the differential effect, Carey and Shugart (1995) hypothesized, increasing party loyalty in preferential systems, but decreasing it in non-preferential systems. Preferential voting frequently does not amount to a personal vote in the sense that it is not independent from the party (Cain et al., 1987): legislators frequently seek the support of a particular subsection of the party electorate that may not welcome dissent undercutting the party record. As such, party loyalty does not markedly grow weaker as district magnitude increases in preferential systems, nor does it grow stronger with district magnitude in non-preferential systems. Rather, electoral incentives as legislators understand them shape their party loyalty.

In sum, party loyalty grows weaker in the face of partisan dealignment in the electorate – even after accounting for legislators’ policy preferences and campaign strategies. Model 5, combining all explanatory factors, corroborates our earlier findings. Partisan dealignment undercuts the value to legislators’ re-election of the party record and thereby their willingness to toe the party line. Individual legislators are less loyal to the party, in particular in order to cultivate a personal vote and to cater to the growing non-partisan electorate. Party loyalty remains conditional on the circumstances, moreover: party unity is more important in the governing parties than in the opposition (though the effect is significant only at the 90 percent level) and when the cabinet’s margin over the opposition is narrow. Model 6 further demonstrates that including alternative explanations related to legislators’ seniority in the legislature, the size of the party, or the number of parties does not affect any of our findings. Contrary to common expectations, party loyalty does not markedly grow stronger the longer backbenchers have served in the legislature (Asher, 1973), the larger the party and the more resources the leadership is provided with (Saalfeld, 1995), or the more parties compete in the district and the more likely it is there is a party that perfectly matches legislators’ policy positions (Carey, 2009).

Conclusion

The norms regarding party loyalty are among the most widely shared because parties help solve collective action problems in the electoral arena and in the legislative arena. Political parties advocate distinct policy platforms and provide the electorate with coherent electoral alternatives. In the legislative arena parties will then enact the pledges made. As such, disagreement may be voiced behind the closed doors of the parliamentary party group meetings, whereas in the legislative arena individual legislators are to be loyal to the party even when they disagree. But recent processes of partisan dealignment in the electorate have decreased the utility of the party label as a meaningful voter cue. Party predispositions and identifications are waning in the electorate across most advanced industrial democracies, increasing electoral volatility in turn. Combining district-level data on electoral volatility and new attitudinal data from the PARTIREP survey of legislators in 15 advanced industrial democracies, the article provides strong support for the ‘two-arena’ contention that the decreased utility of the party label in the electoral arena reduces party loyalty on the part of individual legislators. Party loyalty, we find, suffers in districts with volatile electorates – even when we control for individual legislators’ policy preferences, government participation or personal vote-seeking campaign strategies.

Our findings have further implications for party discipline and ultimately for the sustainability of current models of representative democracy. Partisan dealignment has reached unprecedented levels in the electorates of many advanced industrial democracies. What’s more, dealignment, we demonstrated, tends to weaken party loyalty. Whenever there are conflicting opinions in the party, individual legislators will be less likely to vote the party line out of a sense of duty to the party. That is not to say they will publicly speak out or even vote against the party line – there has been little evidence that party unity is waning on recorded votes (but see Kam, 2009). Weakening party loyalty thus raises concerns about the lengths the party leadership will have to go to in order to obtain party unity by other means. Increasingly, the party leadership will have to rely on the promise of career advancement and ultimately on discipline to prevent disagreement from spilling over into the media or recorded votes. While some dog-whistle position-taking may earn support among non-partisans, ultimately people will grow confused as to the meaning of the party label (further reducing its utility to individual legislators). Future research should therefore continue to look beyond roll-call voting and into the earlier stages of the sequence. Future research should delve deeper in disciplinary measures and at the same time include more subtle instances of disagreement drawing on survey data, in-depth interviews or position statements in the traditional and new (social) media.

Footnotes

Acknowledgements

The research for this paper was supported by the partirep research project, funded by the Belgian Federal Science Policy Office (BELSPO). The authors further acknowledge the great debt they bear to the project’s partners in each of the participating countries () and all persons who have at one time or another lent a hand to the data collection.

Notes

References

Aldrich

(1995) Why Parties? The Origin and Transformation of Political Parties in America. Chicago, IL: University of Chicago Press.

Aldrich

Rohde

(1997) The transition to Republican rule in the House: Implications for theories of congressional politics. Political Science Quarterly 112: 541–567.

Andeweg

(2003) Beyond representativeness? Trends in political representation. European Review 11: 147–161.

Andeweg

Thomassen

(2005) Modes of political representation: Toward a new typology. Legislative Studies Quarterly 30: 507–528.

Andeweg

Thomassen

(2011) Pathways to party unity: Sanctions, loyalty, homogeneity and division of labour in the Dutch Parliament. Party Politics 17: 655–672.

APSA (1950) Toward a more responsible two-party system. American Political Science Review 44: supplement.

Asher

(1973) The learning of legislative norms. American Political Science Review 67: 499–513.

Bartolini

Mair

(1990) Identity, Competition and Electoral Availability: The Stabilisation of European Electorates, 1885–1985. Cambridge: Cambridge University Press.

Blais

Dobrzynska

(1998) Turnout in electoral democracies. European Journal of Political Research 33: 239–261.

10.

Bowler

(2000) Parties in legislature: Two competing explanations. In: Dalton

Wattenberg

(eds) Parties Without Partisans: Political Change in Advanced Industrial Democracies, 157–179. Oxford: Oxford University Press.

11.

Bowler

Farrell

Katz

(1999) Party cohesion, party discipline, and parliaments. In: Bowler

Farrell

Katz

(eds) Party Discipline and Parliamentary Government, 3–22. Columbus, OH: Ohio State University Press.

12.

Cain

Ferejohn

Fiorina

(1987) The Personal Vote: Constituency Service and Electoral Independence. Cambridge, MA: Harvard University Press.

13.

Carey

(2009) Legislative Voting and Accountability. Cambridge: Cambridge University Press.

14.

Carey

Shugart

(1995) Incentives to cultivate a personal vote: A rank ordering of electoral formulas. Electoral Studies 14: 417–439.

15.

Cooper

Brady

(1981) Institutional context and leadership style: The House from Cannon to Rayburn. American Political Science Review 75: 411–425.

16.

Cox

McCubbins

(1993) Legislative Leviathan: Party Government in the House. Berkeley, CA:University of California Press.

17.

Crewe

Denver

(1985) Electoral Change in Western Democracies: Patterns and Sources of Electoral Volatility. London: Croom Helm.

18.

Crisp

Olivella

Malecki

Sher

(2013) Vote-earning strategies in flexible list systems: Seats at the price of unity. Electoral Studies. doi:10.1016/j.electstud.2013.02.007.

19.

Crowe

(1980) Cross-voting in the British House of Commons: 1945–1974. Journal of Politics 42: 487–510.

20.

Crowe

(1983) Consensus and structure in legislative norms: Party discipline in the House of Commons. Journal of Politics 45: 907–931.

21.

Crowe

(1986) The web of authority: Party loyalty and social control in the British House of Commons. Legislative Studies Quarterly 11: 161–185.

22.

Dalton

(2000) The decline of party identifications. In: Dalton

Wattenberg

(eds) Parties Without Partisans: Political Change in Advanced Industrial Democracies, 19–36. Oxford: Oxford University Press.

23.

Dalton

(2004) Democratic Challenges, Democratic Choices: The Erosion of Political Support in Advanced Industrial Democracies. Oxford: Oxford University Press.

24.

Dalton

McAllister

Wattenberg

(2000) The consequences of partisan dealignment. In: Dalton

Wattenberg

(eds) Parties Without Partisans: Political Change in Advanced Industrial Democracies, 37–63. Oxford: Oxford University Press.

25.

Depauw

Martin

(2009) Legislative party discipline and cohesion in comparative perspective. In: Giannetti

Benoit

(eds) Intra-party Politics and Coalition Governments, 103–120. London: Routledge.

26.

Desposato

(2005) Correcting for small group inflation of roll-call cohesion scores. British Journal of Political Science 35: 731–744.

27.

Downs

(1957) An Economic Theory of Democracy. New York: Harper and Row.

28.

Gray

Caul

(2000) Declining voter turnout in advanced industrial democracies, 1950 to 1997: The effects of declining group mobilization. Comparative Political Studies 33: 1091–1122.

29.

Hazan

(2003) Introduction. Journal of Legislative Studies 9: 1–11.

30.

Hazan

Rahat

(2010) Democracy within Parties: Candidate Selection Methods and Their Political Consequences. Oxford: Oxford University Press.

31.

Hechter

(1987) Principles of Group Solidarity. Berkeley, CA: University of California Press.

32.

Holmberg

Oscarsson

(2013) Party leader effects on the vote. In: Aarts

Blais

Schmitt

(eds) Political Leaders and Democratic Elections, 35–51. Oxford: Oxford University Press.

33.

Jackson

(1968) Rebels and Whips. An Analysis of Dissension, Discipline and Cohesion in British Political Parties. London: Macmillan.

34.

Jensen

(2000) Party cohesion. In: Esaiasson

Heidar

(eds) Beyond Westminster and Congress: The Nordic Experience, 210–236. Columbus, OH: Ohio State University Press.

35.

Kam

(2009) Party Discipline and Parliamentary Politics. Cambridge: Cambridge University Press.

36.

Karvonen

(2004) Preferential voting: Incidence and effects. International Political Science Review 25: 203–226.

37.

Kiewiet

McCubbins

(1991) The Logic of Delegation: Congressional Parties and the Appropriations Process. Chicago, IL: University of Chicago Press.

38.

King

(1981) The rise of the career politician in Britain — and its consequences. British Journal of Political Science 11: 249–285.

39.

Kingdon

(1977) Models of legislative voting. Journal of Politics 39: 563–595.

40.

Krehbiel

(1993) Where’s the party? British Journal of Political Science 23: 235–266.

41.

Loewenberg

Mans

(1988) Individual and structural influences on the perception of legislative norms in three European parliaments. American Journal of Political Science 32: 155–177.

42.

Mair

van Biezen

(2001) Party membership in twenty European democracies, 1980–2000. Party Politics 7: 5–21.

43.

Matthews

(1973) U.S. Senators and their World. New York, NY: Norton.

44.

Norton

(1978) Conservation Dissidents: Dissent Within the Parliamentary Conservative Party, 1970–74. London: Temple Smith.

45.

Özbudun

(1970) Party Cohesion in Western Democracies: A Causal Analysis. Beverly Hills, CA: Sage.

46.

Pedersen

(1979) The dynamics of European party systems: Changing patterns of electoral volatility. European Journal of Political Research 7: 1–26.

47.

Popkin

(1991) The Reasoning Voter: Communication and Persuasion in Presidential Campaigns. Chicago, IL: University of Chicago Press.

48.

Rohrschneider

(2002) Mobilizing versus chasing: How do parties target voters in election campaigns? Electoral Studies 21: 367–382.

49.

Saalfeld

(1995) Parteisoldaten und Rebellen: Eine Untersuchung zur Geschlossenheit der Fraktionen im Deutschen Bundestag (1949–1990). Opladen: Leske + Budrich.

50.

Saward

(2010) The Representative Claim. Oxford: Oxford University Press.

51.

Schattschneider

(1942) Party Government. New York, NY: Rinehart.

52.

Schmitt

Holmberg

(1998) Political parties in decline? In: Fuchs

Kaase

Klingemann

(eds) Citizens and the State, 95–133. Oxford: Oxford University Press.

53.

Schwarz

(1980) Exploring a new role in policy making: The British House of Commons in the 1970s. American Political Science Review 74: 23–37.

54.

Sieberer

(2006) Party unity in parliamentary democracies: A comparative analysis. Journal of Legislative Studies 12: 150–178.

55.

Sieberer

(2013) Using MP statements to explain voting behaviour in the German Bundestag: An individual level test of the competing principals theory. Party Politics. doi:10.1177/1354068812472584.

56.

Smith

(2000) Positive theories of congressional parties. Legislative Studies Quarterly 25: 193–215.

57.

Steenbergen

Jones

(2002) Modeling multilevel data structures. American Journal of Political Science 46: 218–237.

58.

Strøm

(2000) Delegation and accountability in parliamentary democracies. European Journal of Political Research 37: 261–289.

59.

Tavits

(2009) The making of mavericks: Local loyalties and party defection. Comparative Political Studies 42: 793–815.

60.

Tavits

(2010) Effect of local ties on electoral success and parliamentary behavior: The case of Estonia. Party Politics 16: 215–235.

61.

Thies

(2000) On the primacy of party in government: Why legislative parties can survive party decline in the electorate. In: Dalton

Wattenberg

(eds) Parties Without Partisans : Political Change in Advanced Industrial Democracies, 238–257. Oxford: Oxford University Press.

62.

van Biezen

Mair

Poguntke

(2012) Going, going, … gone? The decline of party membership in contemporary Europe. European Journal of Political Research 51: 24–56.

63.

Wahlke

Eulau

Buchanan

Ferguson

(1962) The Legislative System: Explorations in Legislative Behavior. New York, NY: Wiley.

64.

Wattenberg

(1998) The Decline of American Political Parties, 1952–1996. Cambridge, MA: Harvard University Press.

65.

Zittel

Gschwend

(2008) Individualised constituency campaigns in mixed-member electoral systems: Candidates in the 2005 German elections. West European Politics 31: 978–1003.