Resumé

The Danish parliamentary parties have traditionally been known for their high degree of
cohesion. Recent social and political changes may, however, justify a hypothesis on a
decline of party cohesion. In this paper two conflicting hypotheses on party cohesion in
the Danish Parliament during the 1970s are formulated and empirically tested. The
application of different operationalizations of party cohesion all give the same result: The
hypothesis on a decline of party cohesion is false, whereas the hypothesis on a high degree
of party cohesion is true. Party cohesion is as high in the 1970s as in previous decades,
and the new parties gaining representation during the 1970s are as cohesive as the old, well
established parties.

Introduction

The current political situation in Danmark is to a large extent the product of a so-called 'crisis' in the party system. This system has not only undergone drastic changes recently, it has also found it difficult to gain a new functional stability. A general impression of lack of continuity and stability dominates when one looks at the larger number of parties represented in Parliament, at the declining numbers of party members in the party organizations, and at the volatile electoral behaviour. The question posed in this paper is to what extent this lack of continuity and stability penetrates the Danish party system in all its vital aspects, or whether there are important elements of continuity and stability in the system. The focus will be on the parliamentary level of the party system and the degree of cohesion in the parliamentary party groups.

In the second section of the paper two conflicting hypotheses about party
cohesion in the Danish party system are presented and discussed. With a
view to a test of these hypotheses, the dependent variable is operationalised

---

* This is a revised version of a paper presented at for the ECPR workshop on 'Party Adaption to Societal Changes', Lancaster, March 29 — April 4, 1981. I wish to thank the participants of the workshop and in particular Mogens N. Pedersen for their comments and suggestions.

in the third section, and the findings are presented in the fourth section. Finally,
the paper is concluded with a discussion of some theoretical perspectives.

Conflicting Hypotheses on Party Cohesion

The Danish parliamentary parties have been known for their high degree of cohesion (Loewerberg & Petterson 1979, 224). From 1953 to 1965 breaks of party unity occurred in about 12 per cent of the final divisions on governmental bills (Pedersen 1967, 147). For the period 1965—70 the corresponding figure was 11 per cent (Damgaard 1973, 40). If other conditions in the Danish political system had remained unchanged, a high degree of party cohesion in the 1970's could have been expected. However, as briefly indicated above, a great number of conditions have actually changed during the last decade, and it could be argued that a number of these changes; are highly relevant for party cohesion in Parliament. This suggests that a case can be made for both the hypothesis of a high degree of party cohesion and the opposite hypothesis of a decline in party cohesion.

The High Degree of Party Cohesion Hypothesis

It can be argued that the regime norms in the Danish political system as
well as the development of the work load of the Danish Parliament tend to
sustain a high degree of party cohesion in the parliamentary groups.

The democratic character of the political system in Denmark is expressed in the regime norms which transform popular sovereignty to governmental authority and legitimacy through (1) universal suffrage, (2) proportional representation, and (3) parliamentarism or cabinet responsibility to Parliament. The last two of these regime norms are highly influential on party cohesion (Rasmussen 1971, 218).

The political parties have important functions as mediating structures between voters and Parliament and Cabinet. The party goal of vote maximation generally motivates the parties to be united. When a party has taken a position on a current political issue, every deviation from this position is likely to undermine the credibility of the party in the eyes of the voters. 'If party cohesion did not exist, it would be difficult for the voter to decide the opinion direction in the party that will succeed and thereby what standpoint the party will adopt' (Sjöblom 1968, 151).

This is so especially when the electoral system is one which combines
proportional representation and supplementary mandates, i.e. an electoral
system in which the votes cast for a party in one part of the country may result

in the election of a candidate in another part of the country (Johansen 1979). Since the introduction of proportional representation in Denmark in 1915^—20, no member has been elected to Parliament without being a candidate of one of the political parties. This means that the voters first and foremost choose among the political parties, which are supposed to stand united around certain principles and policies.

The party unity required by proportional representation is sanctioned at the individual level by the nomination procedure in Danish parties. The general rule is that candidates are nominated by the local party organizations (Meyer 1965, 190^—194). If members of Parliament are disloyal to their party they run the risk that party members in their local organization will refuse to renominate them. In addition, they might not be reelected because the voters of the party may prefer a more loyal candidate.

From this line of reasoning a close connection would be expected between on the one hand proportional representation and a high degree of party cohesion and on the other hand the relative majority system in single member constituencies and lack of party cohesion. This works out fine when most European countries are compared with the United States, but not when Britain is included in the comparison. This means that the electoral system is not a sufficient explanation of party cohesion. However, this points to the importance of the third mentioned regime norm, namely parliamentarism.

The extent to which parliamentary parties act as a unity is fundamentally influenced by whether they exist in a parliamentary or a separation of powers system. When the Cabinet is responsible to Parliament, all divisions on bills — and especially on governmental bills of some political importance — involve not only a question of the passage of the bill, but also a question of confidence in the Cabinet. Lack of party cohesion might result in governmental instability, because there would be no solid basis for the Cabinet. On the other hand, in a separation of powers system, where the executive power is directly elected and legitimatized, there is not to the same extent a need for a high degree of party cohesion.

This relationship between regime norms and party cohesion is most evident in two-party parliamentary systems (Epstein 1967, 318^—331). Party cohesion has been high in Britain, whereas it is low in the United States. The importance of parliamentarism is, in particular, noticeable in a comparison between Canada and the United States, where 'nothing except the parliamentary system distinguishes Canada from the United States in such a way as to account for the much greater cohesion of parties in the Canadian House of Commons than in the American Congress' (ibid., 330).

The regime norm of parliamentarism is also a decisive factor in multiparty parliamentary systems, such as the Danish, without a majority party. In these systems all majority coalitions are results of a cooperation between at least two parties. In negotiations between party leaders a high degree of party cohesion is a necessary condition, because they negotiate from a position of strength, which to some extent equals the number of mandates in their party groups. '. . . (I)f lack of party cohesion prevails in a party, this reduces the party's possibilities of making a coalition with other parties' (Sjöblom 1968, 87).

Comparative analyses have shown that party cohesion is high in almost all political systems which combine parliamentary government and multipartism (Epstein 1967, 333—340; Ozbudun 1970, 355—363). The only exception is France of the Third and Fourth Republics. The low party cohesion and the resulting governmental instability in France in this period may be contributed to the nature of the French party system, which has been characterized as 'extreme' multi-partism, because of the existence of strong anti-regime parties. 'In an extreme multiparty system, the incentives for party unity are weaker than in two-party or moderate multi-party systems. As the center party (or parties) will always be in government, there is little risk for its parliamentary representatives in voting against the majority of their party. Nor does such an act constitute an unpardonable breach of party loyalty, because cabinet reshuffle does not mean a loss of power or prestige for the center party' (Ozbudun 1970, 361).

It could be argued that even if the number of parties increased in Denmark during the 1970'5, and the decade was characterized by governmental instability, the Danish party system has hardly become 'extreme' in the sense mentioned. The Progress Party and some left-wing parties are to some extent of an anti-regime nature, but altogether they have never controlled more than a quarter of the seats in Parliament. A large area of basic consensus exists among the remaining parties, and most of them are involved in the kind of bargaining described above. Certainly, the Social Democrats have most often held governmental power, but alternative coalitions have actually replaced one another, the Social Democrats have been out of Government, and an alternative bourgeois coalition Government has always been a real possibility.

The incentive for party cohesion, which parliamentary government provides, does not only exist for the parliamentary party as a whole, but also for each individual legislator. '. . . (A) legislator who has contributed to the fall of his party's government also loses his share in the obvious benefits of executive power for his party' (Ozbudun 1970, 356). It is also quite

possible that his own chances for reelection are likely to suffer. He is not only personally to blame for the fall of his party's Government, but his party may also lose strength because it has shown itself uncohesive and unable to maintain its leadership in office (Møller 1974, 48^—55).

Together with the regime norms in the Danish political system, the development of the workload and structural adaption of the Danish Parliament to environmental changes have contributed to a high degree of party cohesion.

During the 20th century the scope of government has expanded in Denmark as in most other Western countries. This expansion was particularly great during the 19605. For the Danish Parliament the increased scope implied both a growing number of proposed and passed bills and treatment of more complicated bills with far-reaching social consequences. The increased pressures on Parliament and its members can also be documented by the number of proposed and passed resolutions, interpellations, reviews, and questions asked (Damgaard 1977, 29—49, 286—288).

The increasing workload on members of the Danish Parliament has resulted in a number of structural adjustments. Among these the development of a structure of specialized and permanent committees — completed by a reorganization of the committee system in 1972 — has been of essential importance for party cohesion in the parliamentary party groups. The explicit aim of the reform was to improve efficiency and coordination in legislative work: 'Specialized standing committees allow for the simultaneous consideration of a large number of different items. Furthermore, they constitute an infra-structure for division of labour in such a way that issues are handled by those members of the parties who have special knowledge, experience, or interests in relevant areas' (ibid., 292_—293).

The increased workload combined with the division of labour and specialization among members of Parliament has made it impossible for each member to follow every legislative issue closely. The members have, accordingly, to rely on the information and advice from party 'experts' on the various issues, i.e. the party spokesmen who are normally members of the permanent committees dealing with the relevant policy areas. The members have, so to speak, more or less been forced to a low information strategy of decision-making and to take cue from their colleagues (Matthews & Stimson 1977, 247_—273). In sum, they follow the party line, unless special circumstances make deviant behaviour both possible and desirable.

The Decline of Party Cohesion Hypothesis

The opposite hypothesis can, however, also be argued. Recent changes in

Danish society and party system may have weakened those factors which previously caused a high degree of party cohesion in Denmark. This hypothesis has been formulated both in the public debate in Denmark and in political science literature.

Up to the start of the 19705, Danish politics was dominated by the four 'old' parties: the Social Democrats, the Radical Liberals, the Liberals, and the Conservatives. In 1960 a new fifth party, the Socialist Peoples' Party, was added to this group of parties of considerable size and influence (Damgaard 1974). After the election of September 1971 only these five parties were represented in Parliament, with a close balance between the Social Democrats and the Socialist Peoples' Party on the one hand and the three bourgeois parties on the other.

In December 1973 this party system was completely changed. The five old parties lost almost one third of their support, and in particular the two largest parties, the Social Democrats and the Conservatives, had heavy losses. Both the socialist and the bourgeois bloc declined from about 46 per cent to 32 per cent of the vote, which meant that the reaction of the electorate against the political establishment hit both the left and the right. Three entirely new parties, the Progress Party, the Center Democrats, and the Christian Peoples' Party, obtained representation. Whereas new parties in Denmark generally used to fight to pass the electoral threshold of 2 per cent, two of the new parties were at once established as rather large parties, and in sum they won almost one quarter of the votes. Together with the new parties, some smaller parties, which had earlier been represented in Parliament, regained representation (Borre 1974).

The dramatic election of December 1973 marked a shift from a relatively stable to a new and rather fluid party system. New elections were called in January 1975, February 1977, and October 1979. In 1975 the Left Socialists won representation, whereas the Justice Party lost their parliamentary seats. However, in 1977, the Justice Party was represented once again, bringing the number of parties in Parliament up to 11.

Cabinets had a rather short life-time, and Cabinet formations were in some cases of an unusual or even desperate character: A Liberal minority Cabinet from December 1973 to February 1975 was backed by only 22 of the 179 members in Parliament, and in August 1978 a minority coalition was formed between the two traditional opponents in Danish politics, the Social Democrats and the Liberals. This unusual combination lasted for only one year. Political life has been dominated by almost permanent negotiations on the economic difficulties and by a large number of compromises between different combinations of parties named after the month, i.e. the

'August Compromise' of 1976, the 'May Compromise' of 1980 etc.

This unstable political context, combined with the pressure against the old parties and with a large number of unexperienced politicians in the new parties, has time and again led to allegations about leadership crisis, lack of party loyalty, and decline of party cohesion.

Among the five old parties, only the Social Democrats have avoided a change of party leader since 1973. On the other hand, there have been sharp clashes in the Labour Movement between the prime minister and chairman of the Social Democratic Party, Anker Jørgensen, and the chairman of the trade unions, Thomas Nielsen, in particular around the formation of the Social Democratic — Liberal coalition Governmerit in August 1978. Sharp internal confrontations have also occurred within the leaderships of the Conservative party and the Socialist Peoples' Party.

The new parties have witnessed internal disagreements to an even greater extent, and in a number of cases members have left their parliamentary group in order to become a 'maverick' or to join another party group in Parliament. The Progress Party has seen four members leave their parliamentary group in the 19705. Two members have left the Center Democrats in the same period.¹

It is a widely held view that a change in the behaviour of members of Parliament has occurred since the beginning of the 19705. This view is well illustrated by a leading article in the independent-conservative newspaper Ber lings fee Tidende, which in 1975 observed:

Within the old and the new parties an increased tendency is perceived towards protesting against party cohesion in Christiansborg (the seat of Parliament in Copenhagen) — a wish for greater independence from majority decisions within each of the party groups. Previously deviants were mainly met with contempt by their fellow partisans, and with suspicion by opponents who saw themselves supported by them. During the last couple of years a special sympathy is perceived for people who — with, reference to the Constitution — maintain that they have done nothing but break the 'party discipline' commonly held in contempt.²

A hypothesis on a decline of party cohesion can also be derived from current political research. At the theoretical level it has been argued that, with higher voter volatility, heterogeneity among party members and party voters could be expected to be greater. Furthermore, it is argued that with an increase in this kind of heterogeneity, party cohesion might be expected to decline (Sjöblom 1980, 24).

An increase in electoral volatility has certainly developed in Denmark
during the last decades. It has been calculated that whereas volatility3 in the
Danish party system in the period 1948—59 was 5.5 per cent, it increased in

1960—69 to 8.9 per cent and in 1970—1977 to 18.7 per cent (Pedersen 1979, 9). It has been argued that this increased electoral mobility was the product of basic social and economic changes in Danish society and not only of passing and specific political events (Jarlov & Kristensen 1978). This development has clearly been accompanied by a decline in class voting resulting in greater heterogeneity among party voters (Worre 1979).⁴ As both volatility and heterogeneity among party voters have become higher in Denmark, it follows that a decline in party cohesion could be expected in the 19705.

The aim is now to turn to an empirical test of the two conflicting hypotheses.
As a first step, the dependent variable has to be clarified and the appropriate
data described.

Data and Operationalizations

Party cohesion has a number of different aspects. The concept can be given different meanings, the phenomenon can be studied at different levels, and its importance can be discussed in different relations (Sjöblom 1968, 85_—87, 183_—184, 264_—265). In this paper the focus is on the parliamentary level of political parties, but even at this level cohesion means different things. Generally, it can be said that a parliamentary group is coherent when it functions as a unit. This involves a number of obligations for the members of the parliamentary groups. At least four obligations form the core of party unity (Worre 1970, 169—170).

First, there is an obligation to secrecy. Party unity requires that members of parliamentary groups keep silent about internal discussions. Open and free discussion is allowed within the group — and this opportunity is probably used more now than earlier (Krag 1973, 91); but it is an established norm that members may not inform the public about the discussions, in order not to give an impression of weakness and discord. This norm prevails in all parties represented in Parliament except the Progress Party, whose group meetings are open to the press.

Second, members have a general obligation to be present and to take part in the floor divisions of the Parliament, whenever demanded by the party leadership. In any case, the members must inform their group, if they are prevented from participation.

Third, members are obliged to obtain the approval of their group before

they take any political initiatives. Whenever they raise an important issue
for public discussion, make specific proposals, or negotiate with other parties,
they are supposed to act as the representatives of their party.

Finally, the most important aspect of party unity is the obligation to follow the party line in the divisions on the floor. Members of the parliamentary groups will as far as possible try to reach a unanimous standpoint on every issue, but when disagreement remains after a thorough discussion, the general rule is that the minority has to yield to the majority.

In this paper the aim is to examine party cohesion in the last respect. This is the most important aspect, because the other elements of party cohesion are only indirectly of importance for the decisions taken by Parliament, whereas the voting behaviour directly decides which proposals are adopted or rejected. Whereas it is difficult, if not impossible, to get reliable information on the degree to which the first mentioned obligations are fulfilled, it is possible to provide reasonably good data on the last aspect of party cohesion.

Divisions in the Danish Parliament are anonymous; roll calls seldom occur. This means that it is impossible to retrieve the voting behaviour of every single member. However, for each division the Folketingsårbogen — The official Parliamentary Yearbook — records the number of members voting yes, no and abstaining from voting. It is recorded how the parliamentary groups cast their votes, and as a rule also whether one or more members deviated from the party line.⁵

The cohesion of the parliamentary groups is examined on the basis of this kind of information about the voting behaviour at the final division after the third reading of governmental and private bills passed in the period 1971^—79. The advantage of using such data is that it makes possible a diachronic analysis of party cohesion in Denmark. Originally Mogens N. Pedersen analysed party cohesion by means of this kind of data for the period 1953^—65 (Pedersen 1967), and afterwards Erik Damgaard added the 1953—70 period (Damgaard 1973).

It is readily conceded that this kind of data has certain limitations or disadvantages. The first one has already been mentioned: there is no exact information about the voting behaviour of every individual member. More important, the data reveal only part of the truth, because a large number of important parliamentary decisions are not included. Parliament votes not only on legislation. General debates and interpellations often end with a vote on a motion technically termed 'a motivated resolution to pass to the next item on the agenda'. These divisions are not included in the data analysed in this paper, even though it must be acknowledged that important

breaks of party cohesion have taken place in connection with decisions of
this kind.

In order to make a diachronic analysis possible, the first operationalization of party cohesion in this paper is identical with that of the previous studies. This means that party cohesion is measured by the relative number of divisions with uniform behaviour, i.e. the frequency of the members of parliamentary groups acting in unity by voting in the same way in the divisions after the third reading of bills. If one or more members of a parliamentary group vote differently from the majority of the group members,⁶ a break of unity is registered. Breaks of unity may occur in more than one parliamentary group at the same division, which means that the total number of breaks is not equal to the number of divisions with breaks, but somewhat higher.

It could be argued that this operationalization of the dependent variable implies too weak a test of the two hypotheses on party cohesion, because it only concentrates on the number of divisions in which breaks of the party line have taken place. It is conceivable that even if breaks did not occur in more divisions in the 19705, party cohesion has no less declined, because more members of Parliament deviated from the party line. Whereas the number of divisions with breaks of the party line tells how often party unity is broken, the number of members deviating from the party line tells to what extent party unity breaks down.

The absolute number of individual members of Parliament deviating from the party line is, however, a rather rough measure. The absolute number of breaks should be related to the number of divisions, or more precisely the number of votes cast by each single member of parliamentary party groups. The more divisions taken and the more votes cast, the more likely it is that some members of a Parliamentary group break the party line. Similarly, when the parties are compared, the absolute numbers in each parliamentary group breaking the party line should, in principle, be related to the size of the group. The same number of members breaking the party line in a large and a small parliamentary group does not represent the same degree of disunity in the group. By computing the average number of breaks of the party line for members of each parliamentary party group, it is possible to account for this factor.

However, for a number of reasons it is rather difficult to consider both of these factors simultaneously. First, the data used are not nearly exact enough to calculate an index of party cohesion as proposed by Stuart Rice (Rice 1928, 208—209).⁷ Second, the level of party cohesion in the Danish Parliament means that the few breaks of the party line disappear almost

completely among the large number of divisions with party unity. Finally, the presentation of an index might give the reader a false impression of accuracy which the data and the assumptions do not justify. It might, on the one hand, convincingly be argued that these difficulties make it impossible to use the Stuart Rice index.⁸ On the other hand, this index might give an overall indication of the level of party unity during the whole decade and of the degree of differences between the parties. Whereas the absolute figures of party breaks tend to exaggerate disunity, the index at once puts the breaks of the party line in relation to the number of divisions, and to the size of the parliamentary groups.

In the following analysis the absolute numbers of breaks with the party line, the average number of breaks in parliamentary party groups, and the index of party cohesion will all be applied as complementary measures of party unity.

Finally, it should be acknowledged that the validity of these measures is not fully satisfactory, because they all to some extent overestimate the number of breaks with the party line. The registered degree of party cohesion is not necessarily in accordance with the perception of disagreement in the political parties. In some situations the parliamentary group may not only accept but even stimulate one or a few members to deviate from the party line, for instance when they advocate local interests in opposition to a general policy of the party or when they take care of demands from interest organizations. In particular, this may be the case when the result of the division will not be affected by the deviant behaviour of one or a few members. On specific issues it also happens that a division on the floor is more or less explicitly declared 'a free division', which means that the party does not have a common standpoint. If a break of party cohesion is defined more narrowly as a deviant behaviour against the expressed wish of the party, the number of breaks counted would certainly become smaller (Rasmussen 1971, 217—218).

To apply such an operationalization would, however, require a deeper analysis of the background of all registered cases of deviant behaviour. In addition there is an apparent danger that the gain in validity could be overshadowed by a loss in reliability because a large number of doubtful interpretations would have to be made.

In the period between 1971 and 1979, 1,336 divisions which passed private and governmental bills took place in the Danish Parliament at the end of the third reading. All deviations from the majority position of parliamentary groups have been identified and coded.⁹ The results of the analysis are reported in the next section.

Findings

Number of Divisions with Breaks of the Party Line

The first test of the two hypotheses on party cohesion in the Danish Parliament during the 1970s is based on the frequency of divisions with breaks of the party line. Figure 1 shows how often the divisions in Parliament have been characterized by party unity.

Figure 1 does not support the second hypothesis on a decline in party cohesion. Divisions with breaks of the party line do not occur more often in the Danish Parliament during the 1970s than in the previous decades. In fact, the cohesion of the parliamentary groups is very high, with a complete unity in all groups in 1,159 divisions or 86.8 per cent of all divisions passing bills after the third reading. The remaining 177 divisions with deviations from the party line in one or more parliamentary groups constitute 13.2 per cent of all divisions,¹⁰ which by and large is the same as the 11 per cent for the late 1960s and the 12 per cent for the 1953—65 period. The total number of breaks of the party line in all parliamentary groups is 240 or 35.6 per cent larger than the number of divisions with breaks. For the 1953^—65 period the total number of breaks in the groups was 233 or 31.6 per cent larger than the number of divisions with breaks. When it is taken into account that the number of parties increased in the 1970s this suggests a smaller tendency towards party disunity within the parliamentary groups.

It is further seen in Figure 1 that the December election in 1973 and the doubling of political parties represented in the Folketing did not lead to a decline in party cohesion. Actually, the percentage of divisions with breaks is 18.8 per cent for the 1971_—73 period compared with 11.1 per cent for the post 1973 period.ll

Even if there has been no aggregate decline in party cohesion in the last decade, it is still possible that cohesiveness has declined in some parties or that the new parties differ from the old parties in this respect. The alleged internal disagreement within and the number of actual withdrawals from the new parties could indicate that these parties are less cohesive than the old parties. A lower cohesiveness in the new party groups could also be expected because the members of these groups were without any legislative experience.

Table 1 shows party unity and breaks of the party line for each party. Breaks have been identified in 240 cases, whereas in 119 cases it has been impossible to make out whether a break of the party line has taken place or not.

The figures in Table 1 by no means document that the new parties are less

cohesive than the old parties — if anything, they seem more cohesive, because breaks take place less often in a number of the new parties than in the old ones. Among the new parties, the Progress Party seems less cohesive than the Justice Party and the left wing parties. Among the old parties, the lowest party cohesion is found in the Radical Liberal Party. It is remarkable that cohesion in this party is lower in the 1970s (93.9 per cent) than in the 1950s and 1960s (97.4 per cent).¹²

The Case of the Radical Liberal Party

The Radical Liberal Party is a center party par excellence, and it is known to have a very volatile and heterogeneous body of voters. It is possible, therefore, that the decline in party cohesion for this party represents a partial confirmation of the general hypothesis on decline of party cohesion. On the other hand, it is also possible that special circumstances explain the relatively low degree of party cohesion in the Radical Liberal parliamentary group in the 19705. It is reasonable to examine the kind of political issues which has caused breaks of the party line, because the decline of party cohesion in the Radical Liberal Party might not be due to general social and political changes in the Danish society but rather to specific political issues in the 19705.

Mogens N. Pedersen has identified four main groups of instances where

breaks in cohesion occurred in the 1950s and 1960s (Pedersen 1967, 147^—48). Two groups concerned the type of members who broke with the party line, the other two groups the type of issues in which such breaks took place. One issue group consisted of bilis which gave rise to strong emotionally toned public debate, in particular moral (religious and sexual) and traffic matters. The second issue group consisted of bilis on conditions of delimited geographic localities. In coding the divisions with breaks of the party line in the 19705, these categories were used, and, in addition, some social and economic issue areas were defined. This coding proved to be very difficult, and the figures in Table 2 should, accordingly, be read with caution. It is evident, however, that a special issue provoked a large number of breaks in the 19705, namely bilis relating to the Danish membership of the European Community.

Danish politics in the first part of the 1970s was not only dominated by the popular reaction against increasing taxes and public expenditures — which formed an important part of the background for the advance of the Progress Party — the EC issue was, perhaps, even more controversial, at least in the heated public debate up to the referendum in October 1972. The Social Democrats and the Radical Liberals were seriously split on this issue, not only among their voters and members, but even in their parliamentary groups (Elklit & Petersen 1973, 200, 204).

In Table 3 the 240 cases of breaks of the party line are analysed by political party and political issue. Even if the absolute numbers of several parties are too small to calculate percentages, some important conclusions may be drawn.

Table 3 shows, firstly, that the moral and traffic issues still cause many breaks, particularly in the bourgeois parties. Whereas between one quarter and one third of all breaks of party cohesion — 28.3 per cent to be exact — concern these matters, almost half of the breaks in the large bourgeois parties — Conservatives, Liberals, and the Progress Party — fall in this category. A further study may uncover to what extent there was a 'party line' in these cases or whether a 'free vote' was announced. Secondly, the breaks are most evenly distributed over the different categories in the Social Democratic Party. Here almost half of the breaks have taken place in divisions about social and economic issues (labour market, taxes, social affairs; and housing, and education). Thirdly, and most important to the problem dealt with in this paper, the table shows that most of the breaks (45.7 per cent) in the Radical Liberal Party concerned the EC issue.

The divisions on the bills caused by the Danish entry in the European Community were limited in time — almost all took place in the 1972—73 session. If it was the EC-issue which mainly contributed to the lower level of unity in the Radical Liberal Party in the 19705, the influence of this issue should be concentrated to the 1972_—73 session, whereas party unity should be high in the remaining part of the decade.

Table 4 clearly shows that party cohesion was extremely low among the Radical Liberals in 1972^—73 where party unity failed in one out of every five divisions. This is the lowest degree of party cohesion found for any political party in the whole 1971^—79 period. The table also shows that since 1974 party unity has remained high, and that no trend towards a lower party unity is observed. A single year has a higher percentage of breaks, but this is not unusual among the parties. Therefore, it is reasonable to conclude that the decline of the Radical Liberal Party cohesion from the previous decades to the 1970s was due to a specific political issue and to a specific and brief period, rather than to general social or political changes in the Danish society.

Number of Members Breaking the Party Line

As mentioned above, it could be argued that the test based on the frequency of breaks of the party line in divisions is too weak, because party cohesion could have declined, because more members of Parliament broke with the party line in the same or almost the same number of divisions.

For the period 1953—65, Morgens N. Pedersen found that in 164 out of
233 cases of breaks of the party line — or 70 per cent — only one or two

members deviated from their parliamentary group (Pedersen 1965, 146, Table 1). To compare party cohesion in the 1970s with this information raises a problem. Among the 240 cases of breaks of the party line in the 19705, data are not available on the exact number of members of Parliament breaking the party line in 47 cases. Table 5 shows how many members deviated from the majority of their parliamentary group in the remaining 193 cases.

The figures in Table 5 demonstrate that no remarkable decline in party cohesion has taken place during the 19705. Whenever a break of the party line occurs, it is most often one or a few members who deviate. In half of the cases it is only a single member, in 66 per cent no more than two members, and in 85 per cent no more than three members. Even if these figures are not fully comparable with the figures from the 1953_—65 period, they form the same general pattern.

To draw a line at two or three members breaking the party line is, obviously, somewhat arbitrary. All the information can be utilized if the absolute number of members of Parliament is used as a measure of the development of party cohesion in the 19705.

The development in party cohesion measured by the absolute number of
members of Parliament breaking the party line is delineated in Figure 2.

Figure 2 does not show any growth in the absolute number of members of Parliament breaking the party line. A large number of members broke with their parliamentary groups in the beginning of the decade, whereas the numbers became smaller later on in the decade. The absolute number of individual breaks of the party line was particularly large in the 1972^—73 session. This was, as mentioned above, partly due to the EC-issue. Nevertheless, the number of individual breaks was large, even if the EC-issue is eliminated from the analysis. In fact, the division with most members deviating from their party colleagues took place in the 1972^—73 session, when 35 members from four of the five parties represented on a bill on abortion voted in another way than the majority of their group. It could be said that this was a very extraordinary division, but even if we eliminate this division from the analysis it still leaves 1972^—73 as the session with the largest number of members deviating from the party line. It is worth noticing thai: this took place before the 1973 election. Apart from the first two sessions, it is seen that party cohesion was stable during the remaining of the decade. The number of individual breaks falls, quite understandably, in the short sessions — 1973—74 Ist session, 1974—75 Ist session and 1976—77 Ist session — where only a few bills were passed before the call of new elections.

Figure 2 clearly demonstrates that the 1973 election and the representation of a number of new parties did not result in a larger number of breaks of the party line and a decline in party cohesion. The question remains, however, whether a larger number of members deviated from the party line in the new parliamentary groups than in the old ones.

The old parties have participated in more legislative divisions in the 1970s than the new parties. In order to compare the old and the new parties on the basis of the same number of divisions, it is appropriate to divide the period in two sub-periods — before and after the 1973 election.¹³

Table 6 shows that party cohesion in the 1971^—73 period was lowest among the Radical Liberals, and that the EC-issue accounted for two thirds of the number of individual breaks in the Radical Liberal parliamentary group. Actually, the same three Radical Liberals breaking with the party line in 18 divisions on the EC-issue made up 54 of the 85 breaks on that issue. It is also seen that — disregarding the EC-issue — the Radical Liberal Party has the largest number of breaks of the party line, indicating a lower party cohesion. However, a more accurate measure of party cohesion level is obtained, when the size of the parliamentary party groups is taken into consideration.

Table 7 confirms that party cohesion in the 1971_—73 period was lowest in the Radical group, and that the EC-issue produced much of the party disunity. In addition, Table 7 shows that even if this issue is eliminated, party cohesion remains somewhat lower among the Radical Liberals. The question is, however, how substantial this difference between the Radical Liberals and the other old parties is. Compared with the difference between the Swedish parties in the 1960s — where average votes against the party line was between 4.8 and 16.5 and average abstentions against the party line was between 2.8 and 18.3 (Stjernequist & Bjurulf 1970, 138) — it seems to be of a marginal nature. This impression is strengthened by the figures in

Table 8 where both group size and the number of divisions are taken into
consideration.

Table 8 demonstrates for the period 1971—73 that the Radical Liberal Party
is clearly less cohesive than the other parties, when the EC-issue is included,
but only marginally less cohesive when it is excluded.

For the 1973^—79 period Table 6 shows that the absolute number of members breaking the party line varies among the new parties. In three of the new parliamentary groups a considerable number of members have deviated from the majority in their groups, whereas only a few or one have done so in the socialist parties and in the Justice Party. The absolute number of members breaking the party line is largest in the newly represented Progress Party. In this context it should be remembered, however, that the Progress Party was not only the largest of the new parties, but in most of the period also larger than the old parties — except the Social Democrats. When the size of the parliamentary groups is taken into consideration, Table 7 shows a slightly different picture.

Table 7 indicates that party cohesion is marginally lower in three of the new parliamentary party groups, and in particular in the Progress Party and the Christian Peoples' Party. But breaks of party line do not occur more often in these new party groups than in the least cohesive of the old parties, i.e. the Radical Liberals.

Finally, Table 8 indicates that even if party cohesion is marginally lower
in some of the new parliamentary party groups, party cohesion is generally
at the same high level in the new parties as in the old ones.

To sum up, increased voter volatility, greater heterogeneity among the voters of each party and the representation of a number of new parties have not caused any substantial decline in party cohesion in the Danish Parliament. Breaks of the party line do not occur more often, and when they occur, it is only marginally to a larger extent in some of the new parliamentary party groups than in the well established parties. Generally, party cohesion is at the same high level in all parliamentary groups. A single incidence of marked decline in party cohesion in one of the old parties is better explained by the emergence of a transient political issue than by general societal changes.

Conclusions

In this paper two conflicting hypotheses on party cohesion in the Danish parliamentary groups during the 1970s have been formulated and empirically tested. The application of different operationalizations of party cohesion has all given the same result: The second hypothesis on a decline of party cohesion is false, whereas the first hypothesis on a high degreee of party cohesion is true. Party cohesion is just as high in the 1970s as in the previous decades, and the new parties are as cohesive as the old, well established parties.

This empirical result points to certain theoretical perspectives. First, the factors of continuity have proved to be stronger than the factors of change in this respect. Regime norms such as proportional representation and parliamentarism have furthered a high degree of party cohesion. Each individual member of Parliament is encouraged to follow the party line by rewards of status and influence, whereas breaking the party line implies considerable risks. In a political system based on norms of proportional representation and parliamentarism a decline of party cohesion would be dysfunctional and cause political instability. These regime norms have neither been changed nor seriously attacked in Denmark during the 19705, and party cohesion has remained high. If Denmark has been characterized by political instability it has not, at least, been caused by a decline in party cohesion. Differently stated, a high degree of party cohesion is a necessary, though not a sufficient, condition for political stability in a parliamentary system.

Second, although social and economic changes causing voter volatility
and heterogeneity may have tended to subvert party unity, some institutional
changes have drawn in the opposite direction. A high degree of party cohesion

has been furthered by the development of the workload of Parliament
and the resulting specialization of parliamentary activities.

Third, the result of the empirical analysis and the explanation proposed demonstrate the importance of political factors in the explanation of political life. Social and economic factors are, undoubtedly, of vital importance for the functioning of political institutions, but they neither explain everything, nor are they necessarily stronger than political factors. In this case, basic social and economic changes causing voter volatility have not been strong enough to change the influence of political factors such as regime norms and parliamentary workload and specialization.

Finally, the analysis has shown that even if Danish politics has been highly dramatic during the 1970s and a number of political changes have occurred, it is doubtful whether the notion of 'crisis' is well-founded. Important elements of continuity and stability do exist in the Danish party system. A high degree of party cohesion has been sustained by elements in the political system, which are both well established and relatively unaffected by ongoing political, social and economic changes.

---

NOTES

1 In August 1980 still another member left the parliamentary group of the Progress Party. Actually the short political history of the Progress Party has been described by a trained parliamentary reporter, as 'one unbroken chain of rows and internal personal showdowns', Aarhuus Stiftstidende, September 7, 1980. Among the old parties only one member left his party in the 1971—1979 period to become a 'maverick' or 'independent'. This happened in April 1976, when a member left the Conservative group. It should be remembered, in addition, that the Center Democrats were formed as a new party just before the December Election of 1973 when its leader Erhard Jakobsen — a prominent member of the Social Democratic group — broke away from his old party and thereby caused the election.

2 Exact and comparable figures for the heterogeneity among party members for each single party are not available. The 'party line' is in this paper identified by the behaviour of the majority of the group members. If it is doubtful how the majority has voted, the 'party line' has been identified by

3 Electoral volatility is defined by Mogens N. Pedersen as net change, i.e. the net gains for winning parties numerically equal to the net losses of the parties that are defeated in the election, (Pedersen 1979, 3—4).

4 Exact and comparable figures for the heterogeneity among party members for each single party are not available.

5 The 'party line' is in this paper identified by the behaviour of the majority of the group members. If it is doubtful how the majority has voted, the 'party line' has been identified by the speech of the party spokesman at the first reading of the bill and the standpoint in the committee report.

6 As Erik Damgaard has pointed out, this can happen in the following six ways: yes/no, yes/abstain, no/abstain, no/yes, abstain/no and abstain/yes, (Damgaard, 1973, nole 8, p. 63).

7 Stuart Rice's 'index of party cohesion' is obtained by dividing the number of votes cast by the majority of party members by the total number of party members who voted and con verting the percentages to a scale from 0 to 100 (by subtracting 50 from the percentage and multiplying by 2), cited here from Ozbudun 1970, 306.

---

7 Stuart Rice's 'index of party cohesion' is obtained by dividing the number of votes cast by the majority of party members by the total number of party members who voted and con verting the percentages to a scale from 0 to 100 (by subtracting 50 from the percentage and multiplying by 2), cited here from Ozbudun 1970, 306.

8 Cf. Pedersen 1967, 148:'. .. the degree of cohesion of the groups in the Folketing cannot adequately be described by quantitative methods developed in American research. And if it had been possible to apply them, they would not have been sensitive enough to uncover the differences among parties.'

9 I want to thank Kjeld Martinussen, graduate student of political science at the University of Aarhus, for his careful coding of the data.

10 A minor problem of measurement should be noticed. Party unity and breaks add up to 100 per cent only, when we have full information. However, in 8 divisions it has been impossible to obtain precise information on possible breaks in party cohesion. In Figure 1 the assumption has been that no breaks have taken place, as the figure reports the number of registered breaks. If the assumption had been the opposite — that breaks had taken place — we should have found 86.2 per cent of registered unity instead of 86.8 per cent. Obviously this difference is of no importance for the conclusions drawn. Number of divisions Percentages Break of unity in all party groups 1 .1 Break of unity in one or more party groups and lack of information in the remaining party groups 7 .5 Unity in one or more party groups and break of unity in one or more party groups and lack of information in the remaining party groups 5 .4 Unity in one or more party groups and bireak of unity in one or more party groups 164 12.3 Registered breaks of unity 177 13.2 Lack of information in all party groups 5 .4 Unity in one or more party groups and lack of information in the remaining party groups 3 .2 Unity in all party groups 1,151 86.2 All 1,336 100.0

11 Figure 1 includes all final divisions passing private as well as governmental bills. In the previous studies, however, only the governmental bills have been treated. In principle, we should therefore compare party cohesion on the governmental bills before and after 1970. No difference from earlier studies can actually be expected, because almost all bills that reach the final division in the Danish Parliament are of governmental origin, for instance 96.6 per cent in the 1971—79 period. Nevertheless, if the private and the governmental bills are studied apart, a weak tendency towards a higher degree of party cohesion is found in the divisions on governmental bills than on private bills. Whereas breaks occurred in 13.1 per cent of the divisions on governmental bills, the corresponding figure for private bills was 17.8 per cent.

12 The following percentages of party unity have been computed for the 1953—65 period from Pedersen, 1967, table 1, p. 146: Social Democrats 98.8 per cent Radical Liberals 97.4 per cent Conservatives 96.3 per cent Liberals 94.9 per cent Socialist Peoples' Party (1960—65) 98.4 per cent

---

12 The following percentages of party unity have been computed for the 1953—65 period from Pedersen, 1967, table 1, p. 146: Social Democrats 98.8 per cent Radical Liberals 97.4 per cent Conservatives 96.3 per cent Liberals 94.9 per cent Socialist Peoples' Party (1960—65) 98.4 per cent

13 In the 1973—79 period almost all parties participated in 963 divisions. Only two parties, the Justice Party (N = 564) and the Left Socialists (N = 845), participated in a smaller number, cf. Table 1.

14 The index of cohesion has been computed for the 1973—79 period together, applying the average number of seats in the three election periods, cf. note 7.

REFERENCES

Borre, O. 1974. 'Denmark's Protest Election of December 1973', Scandinavian Political
Studies, Vol. 9, 1974, pp. 197—204.

Damgaard, E. 1973. 'Party Coalitions in Danish Law-Making 1953—1970', European Journal
of Political Research, Vol. 1, 1973, pp. 35—66.

Damgaard, E. 1974. 'Stability and Change in the Danish Party System over Half a Century',
Scandinavian Political Studies, Vol. 9, 1974, pp. 103—125.

Damgaard, E. 1977. Folketinget under forandring, Copenhagen: Samfundsvidenskabeligt
Forlag. (With a summary in English).

Elklit, J. & Petersen, N. 1973. 'Denmark Enters the European Communities', Scandinavian
Political Studies, Vol. 8, 1973, pp. 198—213.

Epstein, L. D. 1967. Political Parties in Western Democracies, New York: Frederik A. Praeger.

Jarlov, C. & Kristensen, O. P. 1978. 'Electoral Mobility and Social Change in Denmark',
Scandinavian Political Studies, Vol. 1, (New Series), No. 1, 1978, pp. 61—78.

Johansen, L. N. 'Denmark', in Geoffrey Hand et al. (eds.), European Electoral Systems
Handbook, London: Butterworths, 1979, pp. 29—57.

Krag, J. O. 1973. Dagbog 1971—72, Copenhagen: Gyldendal.

Loewenberg, G. & Patterson, S. C. 1979. Comparing Legislatures, Boston: Little, Brown.

Matthews, D. R. & Stimson, J. A. 1977. 'Cue-Taking by Congressmen: A Model and a Computer
Simulation' in William O. Aydelotte (ed.), The History of Parliamentary Behavior,
Princeton, New Jersey: Princeton University Press, pp. 247—273.

Meyer, P. 1965. Politiske Partier, Copenhagen: Nyt Nordisk Forlag, Arnold Busck.

Ozbudun, E. 1970. Party Cohesion in Western Democracies: A Causal Analysis, Beverly
Hills, California: Sage Publications.

Pedersen, M. N., 1967. 'Consensus and Conflict in the Danish Folketing 1945—65', Scandinavian
Political Studies, Vol. 2, 1967, pp. 143—166.

Pedersen, M. N., 1979. 'The Dynamics of European Party Systems: Changing Patterns of
Electoral Volatility', European Journal of Political Research, Vol. 7, No. 1, :1979,
pp. 1—26.

Rasmussen, E. 1971. Komparativ Politik 1, Copenhagen: Gyldendal.

Rice, Stuart A. 1928. Quantitative Methods in Politics, New York: Alfred A. Knopf.
Sjöblom, G. 1968. Party Strategies in a Multiparty System, Lund: Studentlitteratur.

Sjöblom, G. 1980. 'Political Changes and Political Accountability. A Propositional Inventory
of Causes and Effects', Institute of Political Studies, university of Copenhagen.

Stjernquist, N. & Bjurulf, B. 1970. 'Party Cohesion and Party Cooperation in the Swedish
Parliament in 1964 and 1966', Scandinavian Political Studies, Vol. 5, pp. 129—164.

Worre, T. 1970. 'Partigrupperne i Folketinget. Et magtcentrum i dansk politik', Økonomi og
Politik, Vol. 44, No. 2, 1970, pp. 143—188.

Worre, T. 1979. 'Forandringer i det danske partisystems sociale grundlag', in Mogens N.
Pedersen (ed.), Dansk politik i 1970'erne, Copenhagen: Samfundsvidenskabeligt Forlag,
pp. 68—82.