Parties,Legislators,and the Origins of Proportional Representation

Abstract

A prominent line of theories holds that proportional representation (PR) was introduced in many European democracies by a fragmented bloc of conservative parties seeking to preserve their legislative seat shares after franchise extension and industrialization increased the vote base of socialist parties. In contrast to this “seat-maximization” account, we focus on how PR affected party leaders’ control over nominations, thereby enabling them to discipline their followers and build more cohesive parties. We explore this “party-building” account in the case of Norway, using roll call data from six reform proposals in 1919. We show that leaders were more likely to vote in favor of PR than rank-and-file members, even controlling for the parties’ expected seat payoffs and the district-level socialist electoral threat facing individual legislators. Moreover, using within-legislator variation, we show that the internal cohesion of parties increased significantly after the introduction of PR.

Keywords

proportional representation electoral reform party cohesion Europe Norway

Studies of why so many European countries adopted proportional representation (PR) around the turn of the 20th century have proliferated in the last generation. The most widely accepted explanation was first articulated by Braunias (1932) and Rokkan (1970), and later formalized by Boix (1999). In this “seat-maximization theory” of PR adoption, a fragmented bloc of conservative (bourgeois) parties favored PR as a strategic, defensive reform to preserve their legislative seat shares in the face of a rising socialist threat on the left and the inability to coalesce or cooperate effectively in single-member districts (SMDs) on the right.¹

Karl Braunias, of course, was not the only contemporary to study the reasons that motivated the adoption of PR in Europe. Georges Lachapelle, an early student of electoral systems, urged the adoption of PR as a way to improve the discipline of French parliamentary parties (Lachapelle, 1911). France’s two-round elections, Lachapelle argued, fostered diverse and conflicting local electoral alliances, which in turn produced factionalized and undisciplined parties incapable of advocating and pursuing coherent public policies. The adoption of PR, Lachapelle claimed, would begin to centralize control over nominations, thus enabling party leaders to curb indiscipline and empowering parliamentary parties to act as more cohesive voting blocs.

We first review several critiques of the dominant seat-maximization theory of PR adoption. We then develop the (implicit) logic of Lachapelle’s arguments to suggest a “party-building theory” of PR adoption. After considering why moving from two-round elections in SMDs to PR elections in multimember districts (MMDs) should have improved national party leaders’ control over nominations, we explain how more centralized control over nominations should have improved both voting cohesion and party leaders’ ability to negotiate coalition governments. Our line of argument implies that party leaders should have been more favorable to adopting PR than their followers, and that parties should have become more cohesive after its introduction.

Why would rank-and-file members have supported their leaders’ proposals for PR? With a few recent exceptions (Leemann & Mares, 2014; Schröder & Manow, 2014), existing accounts of why Europe adopted PR have treated parties as unitary actors, ignoring the potentially disparate preferences of individual members of Parliament (MPs) within parties. Yet, in many cases, parties were internally divided on the question of whether to adopt PR. Following Leemann and Mares (2014), we argue that intraparty variation in support for PR among backbenchers can be explained partly by members’ differing exposure to socialist electoral threats in their own local districts.

We provide empirical evidence for the party-building theory of PR adoption from the case of Norway, which figured prominently in the initial development of the seat-maximization accounts. From 1906 to 1918, members of the Norwegian Parliament (Storting) were elected through a two-round runoff system in SMDs, which produced disproportional seat allocation outcomes across the main parties. In 1917, Parliament appointed a commission to consider changes in the electoral system. We examine the roll call votes of MPs on all six relevant reform proposals that emerged in 1919, and which culminated in the adoption of PR. The socialist Labor Party was an ardent supporter of PR. Our analyses show that, within the bourgeois Conservative and Liberal Party blocs, the reform was propelled forward mostly by party leaders but was also supported by members whose local districts exhibited higher vote shares for the Labor Party. Our analyses control (via fixed effects) for each party bloc’s aggregate seat gain or loss from PR. Thus, we are better able to discern how the incentives of leaders pursuing the parties’ long-term organizational interests and backbenchers pursuing their own short-term reelection interests affected the process of reform, controlling for each party’s seat-maximization concerns.

In addition to providing a rare examination of support for PR at the individual level, we also investigate how PR affected party cohesion. If our party-building theory is correct, we should observe an increase in party cohesion, either after the adoption of PR in 1919 or after its implementation in 1921. Based on a comprehensive sample of roll call votes on either side of these junctures, we show that voting cohesion indeed increased significantly in Norway’s three main parties following the adoption of PR and continued to increase after its implementation. These findings complement a recent investigation by Schröder and Manow (2014), who use historical roll call votes from Germany to show that party cohesion increased after the adoption of PR in 1918.

All told, our analysis suggests that party leaders’ desire to improve their own control over nominations, thereby improving their ability to discipline their MPs, was an important part of their motivation to introduce PR. We show that the adoption of PR, at least for the classic case of Norway, was not just about preserving or maximizing seats; it was also about building more disciplined and cohesive parties.

Critiques of the Seat-Maximization Theory

Several scholars have previously challenged the dominant seat-maximization theory of PR adoption. We review three lines of critique here: (a) support for PR was more widespread than would be consistent with seat maximization, (b) some parties that supported PR actually lost seats, and (c) the fragmented right had already figured out how to cooperate in SMDs with two rounds of voting.

The first line of critique stems from the fact that both the old elite parties and the socialists supported PR in many European countries. Blais, Dobrzynska, and Indridason (2005) note, for example, that “the very presence of a consensus in what is necessarily a zero-sum game (what a party gains when shifting from one electoral system to another is by definition at the expense of other parties) suggests very strongly that other factors . . . were at work” beyond seat maximization (p. 184). Rodden (2009), pursuing a similar line of argument, puts it this way: “Could it possibly be true that the socialists and their bourgeois enemies both simultaneously believed that the same electoral institution would be their salvation?” (p. 3). As not every major party could have expected to gain seats from the same reform, he sensibly concludes that there must have been other considerations influencing PR adoption.

Indeed, the second criticism of the seat-maximization theory is that many of the bourgeois parties actually lost seats as a result of the adoption of PR. Calvo (2009) views Austria, Germany, Italy, the Netherlands, and Norway as cases in which PR reforms “did not help old elites to maximize their seat shares” (p. 256). In the case of Norway, we show in the analysis to follow that the Liberals were earning a substantial bonus under the old system at the time that many of them voted to dismantle it. Other scholars, such as Ahmed (2013) and Barzachka (2014), have pointed to Belgium as another case in which the incumbent party supporting PR lost seats as a result of its adoption.

The third line of attack questions the central premise that, where the nonsocialist parties were evenly divided, “failure to reduce the electoral threshold would have led to an overwhelming victory of the socialist party” (Boix, 1999, p. 609). The two-round electoral systems in place in most of Europe prior to PR allowed a divided bourgeois bloc to unite after the first round—and such coordination was common and reasonably successful (Blais et al., 2005; Calvo, 2009; Lachapelle, 1911; Schröder & Manow, 2014). Thus, it was possible for even a fragmented right to deal with a rising socialist threat within the existing electoral rules, again suggesting that seat maximization was not the only consideration in play.

To move beyond an exclusive focus on seat maximization by parties, we consider a wider range of actors. The adoption of PR in Europe was often a deliberate choice made by the incumbent MPs in each country.² However, backbenchers and leaders faced different pressures and incentives, and it is important to consider how these sometimes competing and sometimes coincident incentives are reflected in MPs’ support for PR. Backbench MPs should have been concerned not just with their party’s seats but also with their own reelection. Party leaders should have been concerned not just with their party’s share of seats but also with their ability to manage members and negotiate and enforce deals with other parties. In the next section, we consider how PR affected the attainment of leaders’ goals.

PR, Nominations, and Party Cohesion

In European democracies prior to the adoption of PR, the bourgeois parties had separate nominating committees in each SMD. In the countries with two-round systems, such as Norway, sometimes there were multiple local committees for each party, with each local committee putting up its own candidate in the first round. Central party leaders had no formal powers over these committees, meaning that the local actors operated with considerable autonomy.³ A leader seeking to influence local nominations through informal means, moreover, would have had to spread his fixed time and resources thinly across a large number of SMDs. To make matters worse, in the two-round systems rank-and-file members in the bourgeois parties often needed to form district-level electoral coalitions involving diverse parties, with aggregate legislative cohesion suffering as a result (Lachapelle, 1911; Schröder & Manow, 2014).

With the adoption of PR, there were fewer districts and hence fewer nominating committees. Indeed, in the nine Western European democracies that switched to PR (and had modern Parliaments beforehand), the average reduction in the number of districts was 82%.⁴ Thus, the adoption of PR allowed leaders to focus their lobbying resources on a considerably smaller number of targets. Even if not augmented by specific new formal rules giving national leaders a role in local nominations, the reduction in nominating committees alone would have helped to homogenize party nominations and improve central leaders’ ability to monitor and influence them. Moreover, to run for office, candidates would need to be part of an official party list. Independent candidacies of rogue would-be nominees, which might undercut a party’s nomination strategy, would not be possible.

Some contemporary observers and authors, such as Lachapelle (1911), predicted that party leaders would enjoy greater influence over nominations under PR. Consistent with this prediction, soon after PR had in fact been adopted, a common view was that “proportional representation has strengthened the influence of parties and weakened that of personalities . . .” (Herlitz, 1925, p. 585; see also McBain & Rogers, 1922). A bit later, Duverger (1954) offered a similar view: “Proportional representation increases the influence of parties over candidates” (p. 358).

Contemporaries also drew the logical conclusion that PR should enhance the voting cohesion of parties. Lachapelle (1911) argued that closed-list PR could help cure the endemic factionalism of French politics and thereby produce more cohesive and responsible parties. Similarly, party leaders in Austria believed they could combat socialism only if they offered a powerful set of social reforms, which would require “reformed, disciplined mass bourgeois parties” (Boyer, 1986, p. 167). In Austria, as in France, some viewed PR as part of the recipe to create such parties (Boyer, 1995, p. 619).

These arguments hinged on the idea that nomination control was a potent source of influence in all electoral systems. In his classic study of the United Kingdom, Ostrogorski (1902) argued that the pressure put on the parliamentary conduct of MPs by their local nominating committees turned them into “delegates” rather than “representatives” (p. xxx). Although each MP had various political obligations, the nominating committee held the “first mortgage over his political conscience.” A generation later, Schattschneider (1942) gave the idea that control over nominations implied control over MPs its most memorable formulation: “he who can make the nominations is the owner of the party” (p. 64).

Although Schattschneider’s aphorism is famous, systematic evidence on whether and how much European leaders increased their influence over nominations after the introduction of PR is scarce.⁵ If leaders’ influence did increase, then we would expect two effects. First, party cohesion should improve. Second, leaders’ positions on the new lists should have been favorable. Schröder and Manow (2014) make a related argument, and show evidence that backbenchers in the bourgeois parties were more likely to defect from the party majority in roll call votes before the adoption of PR in Germany. We offer similar evidence here, comparing the cohesion of Norwegian parties before and after PR.⁶ We are not aware of a systematic, comparative study of leaders’ positions on early PR lists but offer evidence on how they fared in the case of Norway.

By increasing their influence over nominations, PR could also ease leaders’ task in forming coalition governments or minority governments with stable legislative support. Under the old SMD-based systems, coalition deals could potentially be resisted by any faction in any party. In handling dissidents, party leaders could not threaten expulsion, as autonomous local selectorates controlled nominations in each district. After the introduction of PR, however, national party leaders had a new tool to secure agreement on the deals they struck: the threat of a lower list position or even de-nomination for MPs who pushed their dissent too far. This additional resource in securing agreement to coalition deals probably looked valuable to party leaders pushing for PR. In the countries that switched to PR, the median effective number of electoral parties in the last election under an SMD system was four. Thus, many party leaders must have suspected that their route to power in the future would often be through negotiations with other parties to form coalition governments.

In the case of Norway, within-bloc coalitions were already common prior to the adoption of PR. Coalitions between blocs did not ultimately materialize until the postwar period. However, several of the governments formed between 1921 and 1945 were minority governments (Strøm, 1990), which would have at least required the legislative cooperation of the bourgeois parties across blocs. To sustain this cooperation, party leaders would need to ensure that their members followed suit with regard to any negotiated terms sustaining the government’s support.

In sum, our argument is that party leaders and backbenchers may have favored PR for different reasons. Whereas support among backbenchers should be closely tied to their own reelection chances under the existing SMD system, given the level of socialist threat in their districts (Leemann & Mares, 2014), party leaders should have more uniformly supported PR as a way to better control their own members and increase party cohesion. This motivation for party leaders comes close to what Benoit (2004) calls a “general interest” motivation for reform based on governability concerns and stands in stark contrast to a simple seat-maximization account applying to parties as a whole or individual members in their districts.

Explaining British Exceptionalism

Great Britain provides a contrasting pattern to the argument just made for the continental European democracies, and it is instructive to consider why. The existing account of the British exception to the seat-maximization theory of PR adoption posits that party leaders simply miscalculated in the timing of their support (e.g., Andrews & Jackman, 2005). In the early 1900s, some Liberals were sympathetic to the idea of adopting some form of PR (Bogdanor, 1981; Hart, 1992), in part perhaps because of disproportional outcomes in general elections between 1886 and 1900. However, opinion was divided and would shift over the course of the next two decades, particularly following a successful electoral pact with Labour between 1906 and 1910. By the time the Liberals had reached a consensus on favoring PR following electoral defeats in 1918, Labour had secured its status as the second dominant party and no longer supported reform.

This miscalculation may in part explain why Britain did not adopt PR. However, an alternative account of party leaders’ incentives fits the logic of our party-building theory. At the dawn of the 20th century, Britain was the only country in Europe whose monarch had been continuously limited since the 18th century. The others had only recently secured Parliament’s power over the purse (Cox, 2016), only recently secured Parliament’s right to dismiss ministers via votes of no confidence, and were just beginning to regulate the monarch’s prerogative power to choose ministers.⁷ In contrast, British party leaders had already exercised control over such matters—with the monarch mostly operating as a national figurehead—since the 1830s.

Having secured full power to govern, British party leaders quickly established whip organizations in the House of Commons, mechanisms to centrally regulate nominations, and extraparliamentary organizations large enough to secure majorities in the House of Commons (Cox, 1987).⁸ The whips kept up good relations with the local nominating committees, and party-affiliated clubs maintained lists in London of “pre-approved” candidates from which the local committees were supposed to choose. Members had to worry about being removed from the central list and also about satisfying their local committee. Thus, British party leaders needed PR neither to improve their control over nominations, nor to facilitate negotiations between parties for government formation. Elsewhere in Europe, in contrast, PR could further both of these goals.⁹

The Norwegian Case

Norway provides a useful case for exploring our party-building account of the adoption of PR in Europe. The first electoral system used for the Norwegian Storting, established by the 1814 Constitution, was based on indirect elections. Two-round plurality runoff elections in SMDs were introduced after Norway gained full independence from Sweden in 1905. The runoff system, effective from the 1906 election, came under pressure within five elections, and in 1919 the Storting decided to adopt a system of closed-list PR in MMDs. The reform transformed the existing 126 SMDs into 29 MMDs with district magnitude ranging from 3 to 8 seats.¹⁰ In this section, we describe the political context prior to the introduction of PR, and then provide details on the various reform proposals, which constitute the main ingredients of our empirical analysis. We also provide qualitative evidence from contemporary Storting debates to illustrate the divergent thought processes of the party leaders and backbenchers.

The Political Context

Elections during the two-round runoff period (1906-1918) were dominated by three party blocs: (a) the Labor Party (Arbeiderpartiet), (b) the Liberals (Venstre) and the Labor Democrats (Arbeiderdemokratene), and (c) the Conservatives (Høyre) and the Progressive Liberals (Frisinnede Venstre). The Labor Party, representing a new socialist threat to the bourgeois establishment, gradually gained support after the extension of the franchise, first to men older than 25 years of age in 1898, and then to women in 1913 (Aardal, 2002, p. 176).

Candidates were chosen in each local electoral constituency, with relatively little control exercised by national party elites (Mjeldheim, 1978). The national party organization would recommend that its members coordinate on the top-finishing candidate from the first round, but had no power to enforce this coordination. For the bourgeois parties, the first round of the two-round system often functioned like a primary election to determine which of several copartisans or coalition partners in a district should be supported in the second round (Fiva & Smith, 2017a; Helland & Saglie, 2003a), but additional candidates sometimes ran, and even candidates who did not run in the first round could enter in the second. The Labor Party, in contrast, eschewed coalitions or cooperation with the other parties and less frequently ran multiple candidates in a district.

In line with the third line of critique against the seat-maximization theory of PR adoption, the bourgeois parties were largely successful at negotiating stand-down agreements. Fiva and Smith (2017a, p. 133) find that a failure of the bourgeois parties to coordinate between rounds—thereby splitting the vote and costing the bourgeois bloc the election—occurred in only a single district in elections between 1909 and 1918 (Nordland second district in 1915). In this instance, two Liberal candidates from different parts of the district ran in the second round, splitting the bourgeois vote and handing the seat to the single Labor Party candidate.

The SMD system produced disproportional translations of votes to seats. Figure 1 shows vote shares (white) and seat shares (black) for the main political blocs, and a residual “other” category (including independents and minor parties), over the first half of the 20th century.¹¹ Because support for the Labor Party was relatively evenly spread out, the two-round SMD system resulted in the party’s consistent underrepresentation—mostly to the advantage of the Liberals, who formed a single-party government from 1913 to 1920.¹²

Figure 1.

Vote shares and seat shares across blocs.

The 1814 Norwegian Constitution stipulated that one third of MPs should be elected from towns and two thirds from rural districts. This component of the electoral system, known as the “Peasant Clause,” historically led to the overrepresentation of urban, rather than rural, districts. Although the fraction of urban citizens increased over time, in 1919 it was still below one third, and remained below one third for all of the interwar period. In the 1906 to 1915 period, 123 MPs were elected in 82 rural and 41 urban districts. In 1918, three additional districts were established, but these new districts did not alleviate the problem of Labor Party underrepresentation.¹³

The Labor Party was, therefore, naturally the strongest advocate for electoral reform. The other parties may have been concerned by the growing radicalization and alienation of the Labor Party as a result of its underrepresentation (Rokkan, 1970), and so also publicly acknowledged a need for some kind of reform. In their 1918 party manifestos, both the Liberals and the Conservatives stated that they would work for a “more just electoral system.” The Liberal Prime Minister Gunnar Knudsen elaborated in the November 28, 1919, Storting debate,

It is both our right and duty to strive toward the ideal. It should be acknowledged that the current electoral rule does not fit with these principles . . . This was already obvious in the 1915 election, not to mention the preceding one . . . Hence, all parties acknowledged that the situation was such that this could no longer continue, and a better electoral rule should be found, a more just electoral rule, as we wrote in the Liberals’ program. After the 1918 election it became clear that the situation was even worse. Rather than winning 41 seats the [Labor] party only won 18 seats.¹⁴

Thus, the “fairness” of the electoral system was, at least publicly, part of the rhetoric surrounding reform. However, Rokkan (1970) also argues that the decision to adopt PR was “clearly not just a sense of equalitarian justice but the fear of rapid decline with further Labour advances across the majority threshold” (p. 158). This is consistent with his seat-maximization account.

Consistent with our party-building explanation, on the contrary, is that party leaders also hinted at the need, in the foreseeable future, for parties to negotiate interparty deals for governing, possibly in coalitions. For example, Prime Minister Knudsen stated,

When I arrived at the conclusion that I would recommend support for Proposal A_iv [adopting PR]—like many others—I wanted to take a political perspective. We have been used to having a majority for a single party, but since the socialists arrived . . . it will only rarely happen, if ever, that one party gets a majority. This will naturally have political consequences . . . Any party in the Storting must be prepared to engage in political and parliamentary connections with other parties.¹⁵

Rokkan (1970) himself notes that, although the bourgeois parties were not willing to put aside their differences to join forces in a single list, “they were anxious to increase their representation through provisions for electoral cartels” (p. 159).¹⁶ PR would allow parties to not only remain distinct entities, but also strengthen the ability of party leaders to make deals with the other parties for mutual gain and enforce these deals among their members.

Some backbenchers in bourgeois parties pushed back on the need for reform, and particularly in opposition to the adoption of PR. As did several others in the debates, Fridtjov Otto Hegge of the Liberal Party took for granted that PR would strengthen party control over nominations:

And then comes the big and difficult question of nomination. . . I agree that it is through the party organization that political development must occur, so generally there is no harm in parties being strengthened. . . But it is also possible that it can go too far. It may be that the party organization . . . will become so great that it is at the expense of the individual voter’s freedom and perhaps more. . . The individual voter may be left out of the game . . . because he has to vote for the [candidates decided by the party], so we can understand things will be quite different with list elections using large counties as districts.¹⁷

Thus, while Hegge acknowledged that stronger parties might promote political development, as argued by Lachapelle and other PR advocates, he worried that the party organization might exert too much control. Another Liberal Party MP, Knut Otterlei, framed this worry more dramatically, arguing that the adoption of PR would “make the party organization or the nomination convention, practically speaking, almighty.”¹⁸ Otterlei faced a very low socialist electoral threat in his own district; he and other Liberal dissidents seemed to be more concerned about parties becoming too strong, to the point that they voted against the reform being advocated by their leaders.

The discussion about electoral reform also took place outside of the Parliament and party organizations. The conservative newspaper Aftenposten, for example, wrote 157 articles in the 1918 to 1919 period that included the words “electoral rule” (valgordning); on average, one article every fifth day. A front-page editorial on the morning of November 29, 1919 (the day of the reform votes in Parliament), noted Prime Minister Knudsen’s expectation of future coalition governments and argued that the Liberals should seek to cooperate with the Conservatives rather than any parties on the left. It is clear that legislators and the electorate were aware of the various potential consequences of electoral system reform.

Reform Proposals

In July 1917, Parliament appointed a commission to consider changes in the electoral system. This commission, which consisted of seven members, presented several different reform proposals in April 1919.¹⁹ The majority in the electoral reform commission proposed a system of PR in MMDs (Proposal B in Table 1).²⁰ A minority of the electoral commission suggested instead to implement a system based on plurality rule and adjustment seats, similar to the system adopted by Denmark in 1915.²¹ A majority of the commission also suggested that the Peasant Clause be abolished.²²

Table 1.

Reform Proposals.

Proposal	Rule	District structure	Percent in favor
Proposal	Rule	District structure	Labor	Liberal	Conservative	Overall
Postpone	—	—	0	35	32	29
A (A_viii)	PR (no diﬀ. rural/urban)	MMDs of equal pop. size	100	0	6	17
B (A_vi)	PR (some diﬀ. rural/urban)	MMDs follow county lines Large cities → Separate MMDs	100	4	26	27
C (A_ii)	Mix of PR and plurality	Small cities → SMDs Large cities → MMDs Rural areas → MMDs	89	2	96	53
D (B_i)	Plurality with adjustment seats	126 SMDs; 24 adjustment seats Inspired by Danish reform (1915)	6	83	18	44
E (A_iv)	PR (separate rural/urban)	Small cities → Disjoint MMDs Large cities → MMDs Rural areas → MMDs	94	72	66	73

Proposal labels in parentheses correspond to the labels used in the actual Storting debates and votes. We assign alphabetical labels for ease of presentation. Proposals A and B included provisions to abolish the Peasant Clause. The descriptive statistics are based on the 1919 Storting, which consisted of 18 Labor MPs, 54 Liberal MPs, 50 Conservative MPs, 3 MPs from the Farmer’s Association, and 1 independent. MMDs = multimember districts; SMDs = single-member districts.

The membership of the Storting in 1919 consisted of 18 Labor MPs, 54 Liberal MPs (including three Labor Democrats), 50 Conservative MPs (including nine Progressive Liberals), three MPs from the Farmer’s Association, and one independent. In Norway, MPs who are appointed to cabinet resign their seats and are replaced in Parliament by deputy MPs for the duration of their service in cabinet. Cabinet ministers do not vote on legislation.

There was much debate over the reform proposals in the Storting, but votes were not cast until November 29, 1919.²³ After an initial vote on whether to postpone or move forward with electoral reform, there were five different reform proposals considered, as detailed in Table 1. For a reform to be adopted, it required a two-thirds majority. The order in which the proposals were considered was decided by the president of the Storting (the presiding officer of the Parliament), who at the time was Otto B. Halvorsen of the Conservative Party.²⁴ Halvorsen proposed a voting sequence that he argued would “allow the largest number of MPs to vote according to their principal stances.”²⁵ This suggests that the order was set so that expressive preferences could be counted on Proposals A through D, while the leadership was aware that Proposal E would ultimately win out. In the debate leading up to the votes, there was some pushback on the decided voting order by a handful of Liberal and Conservative backbenchers, who soon conceded.²⁶

The official parliamentary record of the Storting provides complete data on how MPs voted on each of these six proposals. Using these data, combined with data on election results and leadership positions (Fiva & Smith, 2017b), we can explore whether there was more support for PR among party leaders in the Conservative and Liberal blocs, as well as among those MPs who faced a stronger socialist (Labor) threat in their own locales. To set the stage for this analysis, we first describe how many MPs were part of the party elite and how much the socialist threat varied across MPs.

Conservative and Liberal Party Elites

Our party-building theory suggests that party leaders who anticipated a future necessity of forming coalition governments or negotiating other forms of interparty agreements may have favored the introduction of PR to strengthen their hands vis-à-vis the party’s rank-and-file members. In our regression analyses in the next section, we use a variable Party elite that equals 1 for any MP who was a prereform elite politician. We consider MPs who acted as party leader, parliamentary leader, cabinet minister, or national board member in the prereform period to belong to the prereform party elite. All told, 15 of 104 MPs fulfill this criterion. Because the Liberals were in government (and hence most of the party’s elite members resigned their seats to become ministers), all but three of these elite MPs were from the Conservative Party.

Table 2 shows descriptive statistics on the sample of Liberal and Conservative MPs in the Storting at the time of the votes. Both party leaders and rank-and-file MPs tended to continue their political careers after the introduction of PR. However, whereas 73% of bourgeois party elites (11 out of 15) became top-ranked in at least one postreform election, the same is true for only 28% of nonelites (25 out of 89). It is rare that prereform nonelites were promoted to postreform cabinet ministerial positions (two out of 89), but common for prereform elites (nine out of 15). Hence, the prereform elites enjoyed higher list positions and continued to enjoy their status in the party leadership after the reform.

Table 2.

Descriptive Statistics.

	Prereform nonelite		Prereform elite
	No	Yes	No	Yes
Conservative MP (freq.)	51	38	3	12
Ran postreform (freq.)	27	62	2	13
Top-ranked postreform (freq.)	64	25	4	11
Elected postreform (freq.)	44	45	4	11
Cabinet minister postreform (freq.)	87	2	6	9
Mean district socialist vote share (SD)	0.24 (0.13)		0.30 (0.08)
Mean district socialist vote share (SD)	n = 89		n = 15

Descriptive statistics for MPs belonging to the Liberal and Conservative blocs. MPs who acted as party leader, parliamentary leader, cabinet minister, or national board member in the prereform period are classified as belonging to the party elite. An MP is classified as top-ranked if he, in the postreform period, was ranked first on a list nominated by the party he represented in 1919. The district socialist vote share is the 1918 electoral support for the Labor Party candidate in each MP’s home district. MP = member of Parliament.

All Socialist Threats Are Local

In the prereform period, as noted, the Conservative and Liberal blocs often used the first round election as a de facto party (or coalition) primary. The Labor Party, however, typically coordinated on a single candidate in the first round and never formed electoral coalitions with other parties in the second round. Hence, as support for the Labor Party grew, it posed a real threat to Conservative and Liberal candidates, particularly in districts where both parties ran, splitting the nonsocialist vote. Nevertheless, this “socialist threat” was not uniform across districts.

In 1918, the Labor Party fielded a candidate in 124 out of 126 SMDs.²⁷ On average, the Labor Party received 29% of the votes cast in the first round; however, the top panel of Figure 2 illustrates how electoral support for the Labor Party varied considerably across the SMDs. The bottom panel shows Labor Party support in the 104 SMDs won by Liberal and Conservative candidates. Clearly, not all MPs from the bourgeois parties faced the same level of socialist threat in their local districts. On average, party elites faced a slightly higher threat than nonelites (see Table 2).

Figure 2.

First-round support for the Labor Party in 1918.

In our regression analyses in the next section, we measure the local socialist threat (Socialist vote share) as the 1918 electoral support for the Labor Party candidate in each MP’s home district (per the bottom panel of Figure 2). For ease of interpretation, we standardize this variable to have mean of 0 and standard deviation of 1. For deputy MPs, we use the level of socialist threat that faced the candidate for whom they substituted.

Analysis of MP-Level Support for Reform

In Table 3, we present the results of a series of linear probability models of each MP’s vote on five reform proposals (coded 1 if they voted in favor). We exclude the vote on Proposal A because all party elites and all Liberal MPs voted against it.²⁸ The main regressors are whether each MP was a member of the party elite, a dummy for the Conservative bloc (with the Liberal bloc being the excluded category), and the size of the local socialist threat each MP faced. We consider each of these factors in turn.

Table 3.

Results of Linear Probability Model of Voting for Reform Proposals.

	(1)	(2)	(3)	(4)	(5)
	Postpone	B	C	D	E
Party elite	−0.304***	−0.215**	−0.057	−0.208**	0.204*
Party elite	(0.107)	(0.093)	(0.068)	(0.090)	(0.120)
Conservative MP	0.054	0.224***	0.941***	−0.595***	−0.153
Conservative MP	(0.103)	(0.071)	(0.040)	(0.089)	(0.098)
Socialist vote share	−0.068	0.088***	0.026	−0.045	0.121***
Socialist vote share	(0.051)	(0.032)	(0.029)	(0.038)	(0.044)
Constant	0.355***	0.067**	0.027	0.836***	0.736***
Constant	(0.067)	(0.032)	(0.024)	(0.053)	(0.059)
n	104	104	104	104	104
R ²	.075	.192	.892	.457	.101

Dependent variable equals 1 if the MP voted in favor of the reform proposal (given in the table header), 0 otherwise. Proposal A is excluded because all party elites and all Liberal MPs voted against it. Sample is limited to Liberal and Conservative MPs (n = 104). Party elite is an indicator variable equal to 1 if the MP served as a cabinet minister, party leader, member of national board, or parliamentary leader, in the prereform period, 0 otherwise. Conservative MP is an indicator variable equal to 1 if the MP represents the Conservative bloc, 0 otherwise. Socialist vote share is the share of votes for the best performing Labor candidate in the MP’s home district in the first round of the 1918 election; standardized to have mean 0 and standard deviation 1. Small sample robust standard errors obtained using the vce(hc3) option in Stata are in parentheses. MP = member of Parliament.

p < .10. **p < .05. ***p < .01.

Party Elites’ Party-Building Incentives

According to our party-building account of the adoption of PR, party leaders in the Conservative and Liberal blocs would have had the most to gain from centralizing control over nominations and optimizing the bourgeois super-bloc’s chances of forming the government. The results in Table 3 are consistent with the following story.

First, the leaders reached agreement on reform Proposal E, which introduced PR but limited its political effects, as the support for the Labor Party was isolated by the separation of rural and urban areas. This resulted in smaller magnitude districts than would have resulted under the other options considered. Presumably, the conservative parties would continue to perform well enough to win some seats in the cities, while Labor would have difficulty winning seats in the rural remainder of each county.

Second, the leaders arranged a voting agenda which, given reasonably accurate whip counts, would foreseeably lead to the adoption of their agreed proposal. The Conservative presiding officer of the Storting, Otto Halvorsen, ensured that the proposals were voted on in the “correct” order. Along the way, there were some sincere and some strategic differences between the partners, which may have been for electoral consumption. Halvorsen himself voted in favor of Proposal D (presumably his sincere, expressive preference), and then in favor of Proposal E.

If this account is accurate, then the key votes revealing how leaders’ interests diverged from otherwise similar rank-and-file members of their respective blocs are the first and last votes listed in Table 1, and captured in columns 1 and 5 in Table 3.²⁹ In the first vote, which allowed consideration of the various reforms in the prearranged order, the leaders of both bourgeois blocs were significantly more likely to vote against postponing consideration of reform until the next session. In the final vote, which gave the reform proposal a two-thirds majority and thus ensured its adoption, the party elite were more likely to vote for the proposal. Most of the effect in Table 3 is driven by the Conservatives, as the Liberals were in government (so had few elites who were eligible to vote). For the final proposal, 92% (11 of 12) elites in the Conservative Party voted in favor, compared with just 58% (22 of 38) of nonelites.³⁰ Two of the three Liberal elites voted in favor. Liberal leaders who were serving in cabinet (and hence did not vote) also encouraged their members to support the final reform proposal. For example, Otto Blehr, who was serving as Justice Minister and would later become Prime Minister, made a lengthy statement during the debate, concluding, “I, once again, strongly encourage you [MPs] not to take personal interests or district interests into account, but to make this giant boost for the fatherland” [and vote for Proposal E].³¹ If we code as “party elite” the four deputy MPs who were serving in place of Liberal cabinet ministers, the effect is strengthened (see the online appendix Table A2). All four voted in favor of Proposal E.

Parties’ Seat-Maximizing Incentives

Given that the Liberal bloc was consistently and substantially overrepresented under the old electoral system, the Liberals’ support for reform at the first and last votes was more surprising than the Conservatives’ support, who had a mixed experience under the old system (Figure 1). Indeed, most Liberal MPs continued to favor a system based on plurality rule in SMDs (Proposal D). It appears that Liberal leaders made a strategic choice to abandon their bloc’s advantages in votes-to-seats translation for the party-building advantages of PR—and also succeeded in securing Conservative leaders’ cooperation. The Conservatives’ favorite option was Proposal C, which would have made each small city its own SMD. Small cities were traditionally the safest bastions of the Conservatives, so it makes sense that they would favor retaining SMDs for small cities, while expanding their opportunities to gain seats elsewhere in MMDs under PR (Danielsen, 1984, p. 57). This preference, already evident in Table 1, is confirmed in the results of the multivariate analysis presented in Table 3.

Individual MPs’ Seat-Maximizing Incentives

Finally, we can consider the office payoffs of individual MPs. Figure 3 shows that MPs facing a low socialist threat tended to (a) support postponing the decision, (b) vote in favor of plurality with adjustment seats (Proposal D), and (c) vote against Proposals B, C, and E (variations of PR).

Figure 3.

Support for reform proposals plotted against district socialist vote share.

The multivariate regression results in Table 3 provide a more controlled look at how the socialist electoral threat affected Conservative and Liberal MPs. It shows that a 1 standard deviation increase in the local socialist threat that a member faced insignificantly reduced his or her probability of supporting postponement and significantly increased his or her probability of supporting reform Proposals B and E. These results are consistent with the MP-level patterns for the case of Germany investigated in earlier work by Leemann and Mares (2014).

Collectively, these MP-level results from the Norwegian case provide empirical support for the logic of our party-building account of why incumbent legislators in the bourgeois parties opted to switch to PR, as well as additional support for the well-established socialist threat hypothesis.

PR and Voting Cohesion

An additional implication of our analysis is that PR should have increased the voting cohesion of the parties and blocs. To explore this claim, we focus on roll call votes from the last three legislative sessions before the PR system took effect (1919, 1920, and 1921) and first two sessions post-PR (1922 and 1923).³²

According to our explanation, PR should have pushed party leaders to centralize control over nominations, which would in turn have discouraged rank-and-file members from deviating from the party line on legislative votes. In fact, the Nomination Act of 1920, which came into effect with the new PR system, formally set down rules for nominating candidates (Valen, Narud, & Skare, 2002). Under the new system, the composition and rank order of candidate lists was still determined through a decentralized process involving nomination conventions in each PR district.³³ National party leaders were technically not allowed to alter the decision taken at this level but could still influence the outcome through informal channels and would have had more of a say in districts near the capital where many leaders were themselves candidates (Valen, 1956). Nevertheless, with just 29 districts postreform as opposed to 126 SMDs in the prereform period, nomination decisions were relatively more centralized (at the regional level) than in the past.

To identify the consequences of electoral reform, we estimate a fixed effects model that utilizes within-legislator changes in voting behavior. For each of the three party blocs, we estimate the following:

\begin{array}{l} {Deviation}_{i j c t} = α_{i} + γ_{c} + β_{1} {PR}_{t}^{adopt} + β_{2} {PR}_{t}^{implement} + \\ θ O d e l s t i n g_{j} + ω topi c_{j} + u_{i j c t}, \end{array}

where i denotes MP, j denotes vote, c denotes cabinet and t denotes time. Our dependent variable, Deviation, equals 1 if the MP voted differently than a majority of his party bloc peers; $β_{1}$ and $β_{2}$ , the parameters of interest, capture how the adoption of PR in 1919, and its implementation in 1921, affected each MP’s probability of deviating. We include a set of cabinet fixed effects $γ_{c}$ (Halvorsen I, Blehr, Halvorsen II).³⁴ For the purposes of identification, it is convenient that the Knudsen cabinet (January 1913 to June 1920) surrounds the adoption of PR in 1919, and the Blehr cabinet (June 1921 to March 1923) surrounds the implementation of PR in 1921 (both were Liberal cabinets). Odelsting is a dummy variable for the vote being cast in the Odelsting (“lower house”). The reference category is a plenary session.³⁵ Finally, ${topic}_{j}$ is a vector of dummies indicating the topic of the vote based on a simple string search in the title given in the Voting Archive, which we include as controls.³⁶ We cluster standard errors at the vote level (j).

Table 4 shows that the probability of deviating fell an estimated 5 to 10 percentage points for all blocs following the adoption of PR. The implementation of PR appears to have further reduced the probability of deviating for MPs in the Labor and Conservative blocs (columns 1 and 3). For the Liberal bloc (column 2), the effect is positive but small and not statistically significant.³⁷ In other words, the introduction of PR appears to have made all major party blocs significantly more cohesive in legislative voting, with a large portion of the impact occurring before the new rules even went into effect.³⁸

Table 4.

PR and Voting Cohesion (Deviation From Majority of Party Bloc).

	(1)Labor	(2)Liberal	(3)Conservative
PR^adopt	−0.103***	−0.053*	−0.056*
PR^adopt	(0.029)	(0.031)	(0.029)
PR^implement	−0.057**	0.009	−0.067**
PR^implement	(0.026)	(0.023)	(0.030)
Odelsting	−0.026**	0.020	−0.007
Odelsting	(0.011)	(0.016)	(0.017)
Halvorsen I cabinet (Con.)	0.049***	0.072***	0.014
Halvorsen I cabinet (Con.)	(0.019)	(0.025)	(0.026)
Blehr cabinet (Lib.)	0.046*	0.042	0.061*
Blehr cabinet (Lib.)	(0.025)	(0.027)	(0.033)
Halvorsen II cabinet (Con.)	0.079**	0.010	0.081**
Halvorsen II cabinet (Con.)	(0.031)	(0.035)	(0.041)
n	8,063	15,170	14,309
R ²	.040	.069	.041

Linear probability model. Dependent variable, Deviation, is a dummy variable that equals 1 if the MP voted differently from a majority of his or her party bloc peers, zero otherwise. MP fixed effects and topic area dummies (see Note 36) included in all specifications. Standard errors in parentheses are clustered at the vote level (282 votes in plenary meetings, 135 votes in Odelsting meetings). PR = proportional representation.

p < .10. **p < .05. ***p < .01.

Discussion

We have argued that European party leaders should have favored PR more than their followers and have shown evidence that indeed they did in the case of Norway. But if PR increased leaders’ control over nominations and facilitated their negotiations to enter government, then two questions arise. First, why didn’t leaders push for PR earlier? Second, were there any functional equivalents (other reforms that could have achieved the same ends) and, if so, why didn’t leaders opt for these?

As to why PR emerged when it did, note that cohesive parliamentary parties are considerably more valuable to power-seeking leaders when (a) Parliament controls who gets into the cabinet and (b) the cabinet controls the national budget. Condition (a) hinges largely on whether Parliament can remove ministers from office via votes of no confidence. But confidence procedures existed only in the United Kingdom, Belgium, and Norway in the 1830s; the rest of the continent followed relatively slowly. Such procedures were not instituted in Austria, Denmark, Finland, Germany, Portugal, and Sweden until the opening decades of the 20th century. Condition (b) was also relatively new in several countries. Sometimes this was because parliamentary power over the purse was a late development (as in Austria, Denmark, and Germany). In France, the Third Republic’s unusual lottery-based budget procedures effectively deprived the cabinet of any privileged position until the reforms of 1910 (Cirone & Van Coppenolle, 2018; Cox, 2016, pp. 44-45). Thus, one reason that PR emerged when it did is that the value of forming cohesive parliamentary parties increased only when conditions (a) and (b) were both met, which did not happen until near or after 1900 in most countries.

Of course, political entrepreneurs might have reacted to Parliament’s newfound importance as a vehicle to power by (a) seeking to establish a one-party regime, (b) seeking to found a majority-seeking party (as in the United Kingdom), or (c) seeking to form a smaller party that could then join coalition governments. Several countries in Western Europe chose the first option, sometimes after adopting PR (Austria under Dollfuss, Germany under Hitler, Italy under Mussolini) and sometimes without doing so (Portugal under Salazar, Spain under Franco). In the remaining Western European nations, political leaders typically chose to form small parties, meaning that PR could help enhance their internal authority and bolster their bargaining ability.

But were there other reforms that could have achieved the same ends for the small-party leaders? We cannot consider the entire array of possible electoral systems. But we can consider the other option that was clearly on the table: the status quo. It would have been possible to centralize control over nominations in the old two-round systems by establishing some central committee that oversaw all nominations. Such committees have functioned in various countries with low-magnitude districts, including Chile, Japan, and the United Kingdom. Central committees could then have been used as fora within which to trade withdrawals, thereby ensuring a more favorable translation of votes into seats within whatever alliance might be negotiated.

European party leaders were well aware of British practices. However, the U.K. system of central nomination lists worked because there were only two major parties and candidates were not required to reside in the districts in which they ran. Two parties made it maximally difficult for a particular district nominating committee to opt out of its party, if it did not like the potential nominees its party offered on the preapproved list. The lack of residency requirements meant that the party could offer a large menu of potential nominees.

On the continent, in contrast, there were already many parties when PR was adopted (Colomer, 2005), and there were often de jure or de facto residency requirements (including in the case of Norway).³⁹ Thus, the central parties would have had to compile information on each SMD’s situation to recommend local candidates, and the local selectorates could have more easily ignored centrally compiled lists.

In addition to “imitating the British” being expensive and unlikely to succeed, centralizing nominations while keeping the SMDs would have exposed parties to greater electoral risks than PR (per Calvo, 2009) and would have been less flexible (as central nominating committees under majoritarian electoral systems typically tie a party into a single alliance). Thus, there was not a close substitute to PR available to the small-party leaders based on the existing electoral rules.

Conclusion

The Braunias–Rokkan–Boix theory holds that Europeans adopted PR in the early 20th century because the old elite parties, fragmented on the right and facing a rising socialist vote on the left, foresaw electoral disaster if they kept the existing SMD systems. But the two-round system used in many countries prior to PR already allowed reasonably successful coordination (Blais et al., 2005; Calvo, 2009); and some bourgeois parties supported PR even though their seat shares seemed unlikely to increase ex ante and declined ex post (Calvo, 2009; Rodden, 2009). These empirical discrepancies undermine the credibility of the simple “seat-maximization” theory for the origins of PR.

Contemporary advocates (and opponents) of reform sometimes highlighted PR’s effect on internal party cohesion, rather than its effect on party seat shares. In Lachapelle’s analysis, for example, electoral coordination in two-round systems came at the cost of parliamentary discipline. The members of any given party would divide into factions, depending on which other parties’ support each MP needed to secure election in his or her own district. For the case of Germany, Schröder and Manow (2014) similarly conjecture that backbench MPs in the bourgeois parties might have been less likely to toe the party line because they were beholden to the diverse interests of the coalitions of parties that supported their district-level election, not just their own party leadership. Closed-list PR would enable parties to commit to national policy platforms as MPs would be beholden for their nominations to regional and national party leaders.

Although the idea that PR might promote party discipline was common in the decades before and after its adoption, systematic evidence of such an effect was not marshaled by contemporaries. In this article, we have revisited the idea that PR adoption was motivated by a desire to increase the discipline of bourgeois parties, rather than to preserve their seat shares. Theoretically, we argue that increased discipline and cohesion should have been valuable to party leaders because it would enable them to commit to policy platforms and facilitate their efforts to negotiate coalition governments.

Empirically, we have focused on the case of Norway—which originally inspired Rokkan’s account. Our analyses show that (a) the leaders of the bourgeois parties were more likely to support PR than their backbenchers, holding constant their parties’ expected seat gains/losses and their own personal electoral situations; (b) the voting cohesion of the bourgeois parties increased after PR; and (c) leaders tended to receive favorable list positions after PR was implemented. This pattern is consistent with the idea that party leaders acquired significantly greater control over nominations under PR than they had enjoyed under the old SMD system.

Looking beyond the Norwegian case, is there similar evidence that PR was favored because of its expected effects on party discipline and, hence, on the formation of coalition governments? In some cases, such as France (Lachapelle, 1911), advocates explicitly argued for PR on the basis of its disciplinary effects. In the case of Germany, although PR was ultimately introduced by decree in 1918, Schröder and Manow (2014) have shown that voting cohesion increased after its introduction, as in Norway. We also know that party leaders throughout post–World War II Europe routinely received favorable list positions although we do not know whether this pattern emerged immediately after the adoption of PR.

Cumulatively, the evidence suggests that European party leaders in the early 20th century were familiar with the claim that PR would improve their control over nominations and hence improve party discipline. To the extent they credited such claims, leaders should have favored PR. Thus, we argue that the decision to adopt PR was influenced not only by calculations of party seat gains (or losses), but also by calculations of party portfolio gains (or losses), mediated by leaders’ enhanced control over nominations.

Supplemental Material

Cox-Fiva-Smith-CPS-Replication – Supplemental material for Parties, Legislators, and the Origins of Proportional Representation

Supplemental material, Cox-Fiva-Smith-CPS-Replication for Parties, Legislators, and the Origins of Proportional Representation by Gary W. Cox, Jon H. Fiva and Daniel M. Smith in Comparative Political Studies

Supplemental Material

WhyPR-R2-appendix – Supplemental material for Parties, Legislators, and the Origins of Proportional Representation

Supplemental material, WhyPR-R2-appendix for Parties, Legislators, and the Origins of Proportional Representation by Gary W. Cox, Jon H. Fiva and Daniel M. Smith in Comparative Political Studies

Footnotes

Acknowledgements

We thank Stefanie Bailer, Brenda van Coppenolle, Masataka Harada, Mark Kayser, Orit Kedar, Daniel Kselman, Carissa Tudor, anonymous referees, and workshop or panel participants at the American Political Science Association (APSA), Council for European Studies (CES), European Consortium for Political Research (ECPR), European Political Science Association (EPSA), Fukuoka University, London School of Economics, Stanford University, and Toulouse School of Economics for helpful comments.

Authors’ Note

Replication materials are available at Dataverse:

Declaration of Conflicting Interests

The authors declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The authors received no financial support for the research, authorship, and/or publication of this article.

ORCID iD

Daniel M. Smith

Supplemental Material

Supplementary material for this article is available online at the CPS website

Notes

Author Biographies

Gary W. Cox is the William Bennett Munro Professor of Political Science at Stanford University. He is the author of numerous articles and books in the areas of legislative and electoral politics. His most recent book is Marketing Sovereign Promises: Monopoly Brokerage and the Growth of the English State (2016).

Jon H. Fiva is professor of Economics at the Norwegian Business School (BI), and is also affiliated with the Centre for the Study of Equality, Social Organisation, and Performance (ESOP) at the Department of Economics, University of Oslo. His most recent research has been published in the American Political Science Review, British Journal of Political Science, and the Journal of Politics.

Daniel M. Smith is associate professor of comparative politics in the Department of Government at Harvard University. He is the author of Dynasties and Democracy: The Inherited Incumbency Advantage in Japan (2018). His most recent research has been published in the American Political Science Review, Journal of Politics, Political Analysis, Electoral Studies, and Legislative Studies Quarterly.

References

Aardal

(2002). Electoral systems in Norway. In Grofman

Lijphart

(Eds.), The evolution of electoral and party systems in the Nordic countries (pp. 167-224). New York, NY: Agathon Press.

Ahmed

(2013). Democracy and the politics of electoral system choice: Engineering electoral dominance. Cambridge, UK: Cambridge University Press.

Andrews

J. T.

Jackman

R. W.

(2005). Strategic fools: Electoral rule choice under extreme uncertainty. Electoral Studies, 24, 65-84.

Barzachka

N. S.

(2014). When winning seats is not everything: Tactical seat-loss during democratization. Comparative Politics, 46, 209-229.

Benoit

(2004). Models of electoral system change. Electoral Studies, 23, 363-389.

Blais

Dobrzynska

Indridason

I. H.

(2005). To adopt or not to adopt proportional representation: The politics of institutional choice. British Journal of Political Science, 35, 182-190.

Bogdanor

(1981). The people and the party system: The referendum and reform in British politics. Cambridge, UK: Cambridge University Press.

Boix

(1999). Setting the rules of the game: The choice of electoral systems in advanced democracies. The American Political Science Review, 93, 609-624.

Boix

(2010). Electoral markets, party strategies, and proportional representation. The American Political Science Review, 104, 404-413.

10.

Boyer

J. W.

(1986). The end of an old regime: Visions of political reform in late imperial Austria. The Journal of Modern History, 58, 159-193.

11.

Boyer

J. W.

(1995). Culture and political crisis in Vienna: Christian Socialism in power, 1897-1918. Chicago, IL: University of Chicago Press.

12.

Braunias

(1932). Das parlamentarische Wahlrecht: Ein Handbuch über die Bildung der gesetzgebenden Körperschaften in Europa [Parliamentary suffrage: A handbook on the formation of legislatures in Europe] (Vol. 2). Berlin, Germany: W. de Gruyter.

13.

Calvo

(2009). The competitive road to proportional representation. World Politics, 61, 254-295.

14.

Carey

J. M.

(2007). Competing principals, political institutions, and party unity in legislative voting. American Journal of Political Science, 51, 92-107.

15.

Cirone

Van Coppenolle

(2018). Cabinets, committees and careers: Estimating the causal effect of committee service. The Journal of Politics, forthcoming.

16.

Colomer

J. M.

(2005). It’s parties that choose electoral systems (or, Duverger’s laws upside down). Political Studies, 53, 1-21.

17.

Cox

G. W.

(1987). The efficient secret: The cabinet and the development of political parties in Victorian England. Cambridge, UK: Cambridge University Press.

18.

Cox

G. W.

(2016). Marketing sovereign promises: Monopoly brokerage and the growth of the English state. Cambridge, UK: Cambridge University Press.

19.

Cox

G. W.

Fiva

J. H.

Smith

D. M.

(2016). The contraction effect: How proportional representation affects mobilization and turnout. The Journal of Politics, 78, 1249-1263.

20.

Cusack

T. R.

Iversen

Soskice

(2007). Economic interests and the origins of electoral systems. The American Political Science Review, 101, 373-391.

21.

Cusack

T. R.

Iversen

Soskice

(2010). Coevolution of capitalism and political representation: The choice of electoral systems. The American Political Science Review, 101, 373-391.

22.

Danielsen

(1984). Borgerlig oppdemningspolitikk, Høyres historie Bd. 2 [Nonsocialist containment politics: The history of the Conservative Party Vol. 2]. Oslo, Norway: Gyldendal.

23.

Duverger

(1954). Political parties: Their organization and activity in the modern state. London, UK: Methuen.

24.

Eggers

A. C.

Spirling

(2016). Party cohesion in Westminster systems: Inducements, replacement and discipline in the house of commons, 1836–1910. British Journal of Political Science, 46, 567-589.

25.

Elklit

(2002). The politics of electoral system development and change: The Danish case. In Grofman

Lijphart

(Eds.), The Evolution of electoral and party systems in the Nordic countries (pp. 15-66). New York, NY: Agathon Press.

26.

Firth

(1993). Bias reduction of maximum likelihood estimates. Biometrika, 80, 27-38.

27.

Fiva

J. H.

Smith

D. M.

(2017a). Local candidates and voter mobilization: Evidence from historical two-round elections in Norway. Electoral Studies 45, 130-140.

28.

Fiva

J. H.

Smith

D. M.

(2017b). Norwegian parliamentary elections, 1906-2013: Representation and turnout across four electoral systems. West European Politics, 40, 1373-1391.

29.

Gerber

A. S.

Green

D. P.

(2012). Field experiments: Design, analysis, and interpretation. New York, NY: W.W. Norton.

30.

Hallerberg

von Hagen

(1999). Electoral institutions, cabinet negotiations, and budget deficits in the European Union. In Poterba

von Hagen

(Eds.), Fiscal institutions and fiscal performance (pp. 209-232). Chicago, IL: University of Chicago Press.

31.

Hart

(1992). Proportional representation: Critics of the British electoral system 1820-1945. Oxford, UK: Clarendon Press.

32.

Helland

Saglie

(2003). Candidate competition and strategic coordination: Evidence from four early Norwegian elections. Electoral Studies, 22, 581-602.

33.

Herlitz

(1925). Proportional representation in Sweden. The American Political Science Review, 19, 582-592.

34.

Hix

(2004). Electoral institutions and legislative behavior: Explaining voting defection in the European Parliament. World Politics, 56, 194-223.

35.

Kreuzer

(2010). Historical knowledge and quantitative analysis: The case of the origins of proportional representation. The American Political Science Review, 104, 369-392.

36.

Lachapelle

(1911). La représentation proportionnelle en France et en Belgique [Proportional representation in France and Belgium]. Paris, France: Librarie Félix Alcan.

37.

Leemann

Mares

(2014). The adoption of proportional representation. The Journal of Politics, 76, 461-478.

38.

McBain

H. L.

Rogers

(1922). The new constitutions of Europe. New York, NY: Doubleday.

39.

Mjeldheim

(1978). Parti og rørsle: Ein studie av Venstre i landkrinsane 1906-1918 [Party and movement: A study of Venstre in rural districts, 1906-1918]. Oslo, Norway: Universitetsforlaget.

40.

Nohlen

Stöver

(Eds.). (2010). Elections in Europe: A data handbook. Baden-Baden, Germany: Nomos.

41.

Ostrogorski

(1902). Democracy and the organization of political parties (Translated from the French by Clark

). London, UK: Macmillan.

42.

Rodden

(2009). Why did Western Europe adopt proportional representation? A political geography explanation. Unpublished manuscript.

43.

Rokkan

(1970). Citizens, elections, parties: Approaches to the comparative study of the processes of development. Oslo, Norway: Universitetsforlaget.

44.

Schattschneider

E. E.

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

45.

Schröder

Manow

(2014). Elektorale koordination, legislative kohäsion und der aufstieg der modernen massenpartei: Die grenzen des mehrheitswahlrechts im Deutschen Kaiserreich, 1890-1918 [Electoral coordination, legislative cohesion, and the rise of the modern mass party: The limits of majoritarian representation in Germany, 1890–1918]. Politische Vierteljahresschrift, 55, 518–554.

46.

Seymour

Frary

D. P.

(1918). How the world votes: The story of democratic development in elections. Springfield, MA: C.A. Nichols.

47.

Sieberer

(2010). Behavioral consequences of mixed electoral systems: Deviating voting behavior of district and list MPs in the German Bundestag. Electoral Studies, 29, 484-496.

48.

Strøm

(1990). Minority government and majority rule. Cambridge, UK: Cambridge University Press.

49.

Strøm

(1995). Parliamentary government and legislative organisation. In Döring

(Ed.), Parliaments and majority rule in Western Europe (pp. 51-82). Mannheim, Germany: Mannheim Centre for European Social Research (MZES) at University of Mannheim.

50.

Valen

(1956). Nominasjon ved stortingsvalg - En demokratisk prosess eller et eksempel på partidiktatur? [Nomination at Storting elections—A democratic process or an example of party dictatorship?]. Statsøkonomisk Tidsskrift, 70, 115-152.

51.

Valen

Narud

H. M.

Skare

(2002). Norway: Party dominance and decentralized decision-making. In Narud

H. M.

Pedersen

M. N.

Valen

(Eds.), Party Sovereignty and citizen control: Selecting candidates for parliamentary elections in Denmark, Finland, Iceland and Norway (pp. 169-215). Odense, Denmark: University Press of Southern Denmark.

Supplementary Material

Please find the following supplemental material available below.

For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.

For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.

2.80 MB

0.00 MB

0.42 MB