Resumé

This study is a comparative survey analysis of similarities and differences in the
Scandinavian party systems. These are all dominated by three big class parties,
Social Democrat, Agrarian, Conservative, each representing mainly one occupational
The study investigates the electoral basis of each type of class party
in order to explain class voting, and it finds a similar pattern of social factors
which influence class voting. The article also considers the differences between
the Scandinavian party systems, the psychological variables influencing class
voting, a possible decline in class voting, and the distinctiveness of the Scandinavian
systems compared to other types of party systems.

The aim of this article is to provide a comparative analysis of the Scandinavian systems in order to clarify their similarities and differences, especially with regard to their relation with social structure. This will be done through a secondary analysis of survey data, mainly that collected for the Scandinavian Welfare Study of 1972.¹

The relationship between the party systems and social structure is summarised in Table 1, which gives two different measures of association. The first one is extremely simple, Robert Alford's index of class voting which dichotomises the parties into left and right and the voters into manual and non-manual occupations (Alford 1963).² The index illustrates the difference in left voting between the two classes, that is, it is a measure of the degree to which party cleavage represents class cleavage. The table shows that there is a considerable distance between the classes in all four countries. This tendency is stronger in Denmark and Finland than in Norway and Sweden.

However, both-party systems and social structures in the Scandinavian
countries are more complicated than these dichotomies are able to show.3
Therefore Table 1 also presents a more comprehensive measure based

upon a MNA analysis, which takes the multiparty system into account and includes a number of social background variables simultaneously. The measure is a better overall expression of the social conditioning of the Scandinavian party systems.⁴ The entries (eta²) are measures of the proportionofthe in party choice explained by each social background variable. The table demonstrates that in all Scandinavian countries occupationisby the strongest predictor of party choice, accounting for between 11 and 20 per cent of the total party variance, more in Denmark and Norway, less in Sweden.⁵ Of the seven other social background variables included in the analysis, the most influential are father's occupation,education,and all variables that are closely connected with occupational class. But these, as well as the remaining variables, have only a marginal influence on party choice. This appears from the multiple correlation coefficient (R²), a measure of the joint prediction of all variables. Taken together, they do not explain much more of the party variance than the most important of them, occupation. This single variable alone accounts for no less than 76 per cent of the total variance explained

by social background in Denmark, 70 per cent in Finland, 66 per cent in Norway, and 56 per cent in Sweden. We may conclude that occupational class is the prime determinant of electoral behaviour in all Scandinavian countries. The general similarity between them even appears in the influenceofthe variables, their rank order being nearly identical in all four countries.

Table 2 shows how much of the variance for each individual party is accounted for by occupation (eta²), as well as by the totality of the eight social background variables (R²). We again find a very similar pattern in all the Scandinavian countries. There are three parties in each country which

are strongly related to occupational class: Social Democrat, Agrarian, and Conservative. Between 12 and 25 per cent of the variance of these parties is accounted for by occupational class. The remaining parties show very little dependence on class or on social background as such. The differencesbetween class parties and the rest are summarised in the average coefficients.

The similarity between the Scandinavian party systems is due to domination the three big class parties. Table 3 examines the social structure of the four party systems more closely. The electorates are divided into three classes: workers, farmers, and middle class. The table demonstrates that each of these generally votes for its own class party. A great majority of the workers support the Social Democrat party (74 per cent in Norway, around two-thirds in Denmark and Sweden, but less than one-half in Finland). In the same way, between 52 and 83 per cent of the farmers vote for the Agrarian party. The class basis of the Conservative parties is less clear, and we shall investigate it more closely in section 3. However, they are mainly supported by middle class voters, albeit only a minor proportion per cent) of them. The remaining parties are not exclusively supported by one social class. In Table 3 they are divided into two groups, a left group and a bourgeois group. While these left parties are frequently considered as being working class parties, Table 3 indicates that only the Social Democrats are real working class parties. The other left parties gain support equally from the working and middle classes. The exception is Finland, where the Communists constitute a major workers' party.

The similarity between and disdnctiveness of the Scandinavian party systems derive from their domination by three big class parties, each of which is supported by a-massive majority of one of the three occupational classes, and which between them are supported by a large majority of the

total electorate. At the election before the survey was carried out, these
class parties collected 77 per cent of the votes in Norway and Sweden, 70
per cent in Denmark and 60 per cent in Finland.

Although a fundamental similarity exists between the Scandinavian party systems, there are also differences between them. The Swedish system is the least class oriented: the Social Democrats have a strong middle class support and the Centre party has, as the only one of the Agrarian parties, considerable support among non-farming groups, while the Moderates are very weak even in the middle class. In Finland class voting is weak both among workers and farmers because of the existence of competing parties inside both classes: the Communists and Smallholders Denmark exhibits the least deviation from the class party pattern. There are, however, reasons to question the representativeness the Danish part of the Welfare Study, because a number of contemporary surveys showed a considerably weaker tendency towards class voting.⁶

In this article class voting will be used in a slightly narrower sense than that employed by Alford. It will be used as an expression of the proportion of one occupational group which votes for its class party: the workers for the Social Democrats, the farmers for the Agrarians, and the middle class for the Conservatives. The article will try to measure and explain the basis for, the extent of, and the variations in class voting.

1. The Basis of Social Democracy

In their origin the Scandinavian Social Democrat parties are typical class parties. They were founded by a lower class group that wanted to exploit its numerical strength as its main political resource, and which did not feel that it was represented by any of the existing parties. The conditions were favourable for the creation of a class party. The working class was a group with a relatively homogeneous background: type of income, working conditions, social status, educational level. It constituted an integrated milieu where workers mixed only with fellow workers. Both inside and outside the place of work their relations were limited to their own class; most lived in working class districts and associated with other workers. Social Democracy in all the Scandinavian countries is closely associated with the trade union movement, as well as with a number of other organisations take care of working class interests: cooperatives, housing societies, sick-benefit associations, adult education societies.

But although the Social Democrats-are typical class parties, not all workers vote for them. Some prefer bourgeois parties, and in all four countries there exist left-wing parties which also claim to be the representative the working class. These, however, are supported only by a minority of workers, although the proportion differs from country to country: 4 per cent in Sweden, 8 per cent in Norway, 11 per cent in Denmark, and no less than 28 per cent in Finland (see Table 3). That the Social Democrat parties in Denirnark and Finland are smaller than those in Norway and Sweden is mainly due to the greater strength of other left parties. But for the sake of comparability, we have in this analysis recognised one working class party, although in the Finnish case this is less satisfactory.

In all four countries the dominant political norm among workers is to vote for the Social Democrats. But there are variations in this tendency, and Table 4 shows how class voting also depends on a number of other factors. An obvious hypothesis; would be that the differences in material conditions were decisive, that better-off workers would be less classoriented their party choice than the poorer ones. Table 4 shows three variables that divide the working class into better- and worse-off strata. The first is the distinction between skilled and unskilled workers, which constitutes the most important status cleavage between workers. But generally support for Social Democrats is strongest among the skilled workers. The only exception is Denmark, due to the fact that the other left parties here are based primarily om skilled workers. The second distinction is between tenants and homeowners, but it turns out that there are no significant differences between the two categories. Possession of property does not in itself cause any diminution in class voting. Finally, the workers were divided into high income and low income groups, but contrary to our expectations it appears that support for the Social Democrats is strongest among the wealthier workers (except in Denmark, where the difference is insignificant). We have to conclude that there is no evidence to support the hypothesis that class voting is stronger among the worst-off workers. On the contrary, the upper strata show stronger support for the class party than do the lower strata. One might then presume that this was because the lower strata supported the more radical left parties. But actually the opposite is the case: the left wing voters are generally even better-off than supporters of the Social Democrats, consists of moire skilled workers, and have a higher level of income (although Finland is an exception).

There is, however, another possible interpretation of the variation in the
Social Democrat vote. Class voting is strongest among those workers who

are most integrated within the working class, whose environment does not subject them to contradictory influences, and who thus develop the strongest sense of class solidarity. It is not improbable that skilled workers with a high income are more integrated than lowly paid unskilled workers, whose class position may be more marginal. Two variables in Table 4 support this hypothesis. They are both measures of integration, but withoutanystatus and their impact on voting is much stronger than that of the stratification variables analysed above. The first is the class of the parents. Among workers of working class origin, class voting is much stronger in all four countries than among workers of middle class origin. One's political orientations are originally formed in youth, and the family

is the dominant framework of this initial political socialisation. If the voter remains in his childhood milieu, his party identification will be confirmed and strengthened, reinforcing class voting among second generation workers. Workers of middle class origin, by contrast, are subject to cross pressures: their occupational interests and the influence of their present milieu have to overcome the mark of social origin. A second element of integration is trade union membership. In all four countries class voting is higher among union members than among unorganised workers (a differenceofbetween and 20 per cent). The trade unions apparently play an important role in bringing about class party identification.

Stratification variables, then, do not influence class voting among workers, while the integration variables do. Finally, Table 4 includes two demographic variables, education and age, which one would perhaps not expect to be of any political importance, but which turn out to have considerable influence on class voting. Workers with only elementary education support the Social Democrats to a greater extent than those with further education; the latter therefore seems to undermine the tendency to class voting. And the support for the Social Democrats increases with age, particularly in Denmark and Finland. Age differences can be interpreted in two ways. They may be the result of a life-cycle pattern where workers are gradually socialised within the class subculture, including the class party norm. Or they may be an expression of differences between political generations who experience different socialisation processes. Both explanations contain an element of truth. In any case, both educational age differences are expressions of relationships which will gradually cause a declining inclination towards class voting. This aspect will be analysed more thoroughly in section 5.

We have established that support for the Social Democrats among working class voters depends upon a number of social background variables. voting is not only a consequence of occupation, but also of a number of other working class characteristics, most importantly social origin, trade union membership, and education. Therefore, a cumulative index is also applied in Table 4. Working class voters are divided into four groups on the basis of the number of working class characteristics they possess. The pattern is very clear and similar in all four countries: the more working class characteristics, the greater the tendency towards class voting. The range is greatest in Denmark, where support varies from 20 per cent at the least integrated level of the working class to 75 per cent at the most integrated.

The dominant position of the Scandinavian Social Democrat parties is

due not only to their working class support. They also receive many votes from middle class voters. These are not the result of a class party norm, but Table 5 shows that they nevertheless have a special social foundation. Stratification variables which were unimportant among the workers turn out to be decisive in the middle class: it is the economically worst-off voters who support the Social Democrats. There is a much greater support from lower employees than from higher employees, more support from tenants than from homeowners, and not least, a considerable difference between high and low income groups. Finally, social origin plays an even more important role. Middle class voters of working class origin give the Social Democrats more than twice as many votes as those of middle class origin. Perhaps because the middle class voters are subject to more ambiguouspolitical the influence of origin is so much stronger.

2. The Social Basis of Agrarian Parties

In every Scandinavian country there is a typical agrarian party called the Centre party or, in Denmark, the Liberals (Venstre). The farmers have the same high level of class voting as the workers. Agriculture provides, in many ways, very favourable conditions for class party formation. Farmers have a broad community of interests in their trade, income, and residence; they constitute a rural subculture more influenced by tradition and religion than the urban one; and they live and work in exclusive contact with fellow

farmers, contributing to a more closed and integrated milieu than that of any other social group. Finally, farmers have an extensive organisational apparatus, where the overlapping membership between the farmers' associations,cooperative and the branches of the Agrarian parties is considerable, and the social pressure to follow the dominant political norms is strong. All the conditions for class voting are present.

To a lesser extent than workers, farmers are subject to cross pressure and disintegration of their interest community. But then the Agrarian parties have to face the problem that their class basis is shrinking, that migration from agriculture as a consequence of economic development has decimated the original social core of the parties; for example, i:he agricultural share of the Danish population has been halved during the last twenty years, and in the same period the vote of the Liberal party has also been halved. However, the number of votes received by the Agrarian parties is in all four countries higher than the total eigricultural population, so the parties must draw on other resources. But these are all related to agriculture in one way or another, and do not therefore affect the class character of the Agrarian parlies.

Table 6 shows the different spheres of influence of the Scandinavian Agrarian parties. The strongest support comes from the farm population, varying from 52 per cent in Finland to 83 per cent in Sweden. The second sphere of influence is the rural population not engaged in agricultural employment. There are at least three reasons why they should support the Agrarian party. First, the fact of living in rural areas gives the population a considerable community of interest; for example, in questions of the allocation of public funds and services. It is customary to refer to a centre-periphery cleavage in political systems. The Agrarian party will usually be the spokesman for the peripheral interests. Second, the wellbeingof

beingofother social groups in the countryside is, in many ways, dependenton wealth of the farmers: when the latter's incomes rise, grocers sell more goods, and craftsmen secure more work. Finally, the dominant position of agriculture in the local community will mark it politically: the number of influences to which one is subject in regular and incidental social intercourse will necessarily become more Agrarian-marked in rural than in urban areas. Table 6 shows that the Agrarian parties are supported by between 17 and 34 per cent of the non-agrarian rural population.

But the Agrarians even win some support in the towns. Voters are not only influenced by their present milieu, but also by that of their childhood and the political preferences of their parents. And since the agricultural population is declining rapidly, there are many more voters of agrarian origin both in the countryside and in towns than there are farmers. Table 6 shows that between 10 and 27 per cent of the off-spring of farmers in the urban population vote for the Agrarian party. The table also considers the urban voters of rural, but non-agricultural origin. They do not show any particular preference for the Agrarians, except in the case of Sweden. One of the sources of the success of the Swedish Centre party has been its ability to follow this migrant group into the cities.

The comparison between the four Scandinavian countries shows that the pattern of Agrarian party support is the same everywhere. There exists a number of spheres of influence with varying distance from the original nucleus of farmers, and the Agrarian vote declines from the inner to outer circles. However, there are also differences between the four countries. The Swedish Centre party has by far the strongest support in all spheres, while the Finnish is weakest, and the Norwegian party has a steeper drop in strength from the inner to outer spheres. National differences explain some of these variations. The surprising weakness of the Finnish Agrariansshould related to the time at which the survey was held, which was at the height of the success of the Finnish Smallholders' party, a populist faction that had defected from the Centre party. The Norwegian Centre party is the purest class party. It is strongly entrenched among the farmers, but in the outer circles it faces competition from two other parties, the Liberals and Christian People's party, which traditionally have defended peripheral and rural interests. The Danish agricultural population is traditionally subject to a certain political stratification accordingto size of holding: the larger estates support the Conservatives,the of the farmers vote for the Agrarian Liberals, while many smallholders prefer the Radicals. The Swedish Centre party has no such competition within the agrarian sphere: it is the unchallenged and dominantfarmer's

antfarmer'sparty. But that is not the only reason why the Swedish Centre
is the most successful of the Agrarian parties. Of special interest is its
strong support among urban voters.

All Agrarian parties had to face the dilemma of the declining agricultural population. To survive meant winning a foothold inside the urban population. effort to break the rural barrier is reflected in the change of name from Agrarian to Centra Party around 1960 in Norway, Sweden, and Finland. The perspective of extension of the class basis was much more threatening in Sweden than elsewhere, as the agricultural population was already reduced to 5 per cent in 1972, this being one-third of the corresponding and Finnish figures. And as urbanisation had gone much further in Sweden than elsewhere, it became absolutely necessary to transcend the traditional class basis and adopt a new programme with a wider appeal. The table indicates that this was successful, although the party's support is still much stronger in the outer spheres of Agrarian influence than among the native urban population. The result is that while the three other Agrarian parties still collect between 80 and 90 per cent of their votes in rural areas, the Swedish Centre party wins a majority of its votes in urban districts.

3. The Social Basis of Conservative Parties

Unlike the Social Democrat and Agrarian parties, the Conservatives are not clearly connected to any distinct social group, and it is a controversial question whether it is possible to speak of any Conservative class basis. Frequently, scholars refer to the middle class, a rather vague concept incorporating urban self-employed, salaried employees, and public servants. is a very heterogeneous group, including employers as well as wage-earners. Moreover, within each group there is a marked stratification large differences in responsibility, income, and education. Another complication in the consideration of the Conservatives as class parties is that they are supported only by a minority of the middle class, and that a number of other parties obtain most of their votes from this group. Nevertheless, it is possible to distinguish a number of common features in the social composition of all Scandinavian Conservative parties, fact that makes it reasonable to consider them as a type of class party. The other parties are so different and so specific in their social support that it is not expedient: to consider them as cross-national types.

The Conservative parties differ strongly in their electoral strength, and

this variation depends in each country on the existence of competing middle class parties. The weakest Conservative party is the Swedish Moderates, because another party, the Liberals, appeal to the same social groups. The Danish Conservatives, pressed by the Radicals, are in a similar situation. The Conservative parties of Norway and Finland are not only the largest ones, but also those with the most distinctive profile (see Table 3).

In Table 7 the middle class voters are classified according to a number of variables in order to investigate the social basis of the Conservative parties. Occupation does not provide any convincing clue, although it is evident that support for the Conservatives is much stronger among higher salaried employees and civil servants. Income and education also indicate some kind of stratification, and it turns out to be a general pattern for the Conservatives to have strongest support from the better-off: richer people

are more Conservative than poorer ones, and the highest levels of educationare more Conservative. The Welfare Study includes some informationon possession of various kinds of assets (property, securities) and it becomes evident that the inclination towards the Conservatives rises with possession of property.

The conclusion of this investigation is, then, that the Conservative parties are not based on any particular identifiable social group, but that they are primarily supported by the upper well-to-do strata of the middle class. The best predictor of Conservative party preference is not any single one of these variables, but a combined index of social status, a general stratification based on social prestige. Support for the Conservatives step by step from the bottom to the top of the status hierarchy. The connection between status and party preference is parallel in all four countries. They differ only in the proportion of the highest strata that vote for the Conservative parly, varying from one-quarter in Sweden and one-half in Denmark to about two-thirds in Norway and Finland.

4. The Psychological Basis of Class Voting

In investigating three types of class parties, it was demonstrated that voting is widely influenced by social background. But it is evident that the political influence of social phenomena depends on how they affect the voters' perceptions and attitudes. Unfortunately, the Welfare Study does not cover the psychological factors which intervene in the decision-making between social background and political behaviour. Some relevant observations, however, can be drawn from the Danish election study.

Table 8 shows the differences between Danish class parties and other parties in the 1971 election, and the observations are probably valid also for the other Scandinavian countries. A number of indicators of stability and party loyalty all proved to be dominant among the electorates of the class parties. The first is the stability of party choice: 88 per cent of the class party voters retain the same party as in the previous election, while nearly one-half of the voters of the other parties changed their allegiance. Two-thirds of the class party voters identify with their party, but only one-third of the other voters. Formal membership of class parties is three times greater than that of other parties. The decision-making process clearly differs between the two party groups: there is much less hesitation and doubt among the class party voters. And there are differences in the

motivation of party choice. The reasons given by the respondents can be distinguished into major types. Some consider party choice as an expressionof identity, economic interest, or group membership. These long-term forces create electoral stability, impelling the voter to retain the same party from election to election, and may in the long run serve as a basis for the development of a sense of party identification. Other voters consider the election as a means of solving current political problems: they are more influenced by the controversial issues, by confidence in the individual party leaders, and by the election campaign. Such short-term forces are influential among uncommitted voters. It is evident that class will be less determinant for these marginal voters, who are rather volatile. Table 8 proves that a majority of the non-class voters are motivated by such short-term considerations, but only one-sixth of the class party electorates. On the other hand, nearly one half of the latter are motivated by long-term forces, but very few of the non-class party voters (Worre 1979). All these observations prove that adherence to class parties is of a much more fundamental, committed, and permanent nature than adherenceto parties, whose support is based on more superficial or time-limited considerations.

5. Changes in the Class Party System

The Scandinavian class party system used to be very stable: when most of the electorate considered party choice as an expression of class identity, they became stable voters. Polling was like a kind of political census; changes were small and temporary. Only changes in social structure

would cause major permanent changes in party strength. An example of
this is the decline of the Agrarian parties in Denmark and Finland, which
was a consequence of the decrease in the farm population.

In recent years, however, the stability of voting behaviour has given way to a greater mobility, and the stable balance of strength between the parties has been shaken. It is possible to distinguish at least three types of changes. Traditional parties may change their style and image in order to attract new followers from outside their original social core. The best example is the Swedish Centre party which from a nadir of 9 per cent of the vote in 1956 gradually rose to 24 per cent in 1976. Second, current issues can cause unusual waves in favour of some parties at the expense of others, across social borders. For example, the Danish Liberals doubled their vote when they formed a government for a short time in 1975, but lost everything again at the following election. After the EEC referendum the Norwegian Labour party lost one-quarter of its vote, but it all returned four years later. These short-term forces cause instability, but no lasting changes. Finally, it became easier for new parties to break through in the 1970s than was earlier the case. Particularly successful were populist protest parties like the Finnish Smallholders' Party and the Danish Progress Although the causes and consequences of these changes differ, they all imply a reduction of the political importance of class. This section will consider this reduction in the influence of social background factors. Although we lack systematic comparable data of the changes over time, national election studies illustrate some general trends and even give clues as to their possible explanation.

This article started with a consideration of the differences between the party preferences of the working and middle classes, expressed by the Alford index of class voting. Table 9 shows the changes in this class voting index during the last twenty years in Sweden and Denmark. In these two countries the left vote has swung up and down from election to election, influenced by favourable or unfavourable factors in the current political situation. These fluctuations usually run in a parallel direction inside both classes as a result of the same short-term forces. However, there is a general tendency for the losses to be greater among the working class voters and the gains to be larger among the middle class voters, so that the difference between the two groups is gradually diminishing. The result is a convergence between the classes, each of them becoming politically less homogeneous; and more like the other. The index of class voting has been reduced by one-third during the past twenty years. In the case of Denmark,two measures from Table 1 are calculated for the whole

period: they show that the association between party and occupation is
reduced by more than two-thirds, while the total multiple correlation of
party and social background is reduced by one-half.

Although these changes are not continuous, they follow an overall pattern, a trend which must reflect some fundamental transformation in society. No simple or final explanation of this waning in class voting can be given at present, but several elements could have contributed to the decline. The homogeneity of social groups is apparently dissolving, and consequently the interest community is also disappearing. For example, the working class differs internally more than it used to do in relation to income, housing, and life style, and as many workers possess symbols of affluence and status, they will perhaps cease to consider the Social Democrats representing their interests in all areas and thus become subject to incompatible pressures.

But much more important is the erosion of the closed and self-sufficient social milieu which the workers and the farmers traditionally constituted. Formation of opinion used to be dominated by class-specific influences: peer groups, trade unions, party branches, and party newspapers. Today,

non-party mass media have more influence, particularly television. These reach members of all social classes, drawing them into similar developmentsand the class-specific subcultures in favour of a common mass culture. The strength of the traditional class parties was based on superior political resources as the basis of political mobilisation: mass membership and affiliated organisations, links to interest organisations, funds, party newspapers, and the exploitation of class solidarity. Today, new movements, lacking in these organisational resources, are able to raise popular support, mainly through television appeal, as was convincinglyproved the Danish general election of 1973, where two newly founded parties without any organisation or social basis were able to collect one-quarter of the votes. Indeed, the organisational apparatus and interest group foundation may become a liability for the traditional class parties, a restriction on their freedom of manoeuvre which prevents them from exploiting the chances of the moment, while the new parties are able to make the most of the possibilities, since they are not bound by any consideration of established interests and opinions within the party ranks.

The hypothesis has been put forward that with increasing affluence, voters need not worry so much about their daily bread: they will have sufficient reserves to be less concerned about class interests and more about value questions (Inglehart 1977). There are several indicators of a certain re-evaluation among Scandinavian voters. The success of the Swedish Centre party in breaking out of its narrow agrarian base is generally interpreted as a consequence of the adoption of a 'green' image, of a concern with environment and decentralisation (Berglund and Lindström 1978, 187-188). The surprising resurrection of Christian Democrat parties in the secularised Danish and Finnish populations and the new growth of the Norwegian party is an expression of a re-evaluation of religious and moral values. The break-up of the traditional Danish party system in 1973 is clearly connected with a new ideological multidimensionality voter attitudes (Rusk and Borre 1974).

Even more important than re-evaluation has been the loosening of party ties. In the class party system, most voters considered party choice as an expression of social identity: it was never changed or even considered. But more and more voters regard themselves as being free from such ties. They consider several alternative parties and let their choice be governed by the election campaign and other short-term forces. The gradual change in basic political orientation is reflected in a number of indicators of electoral mobility. Net mobility has been gradually increasing in all four Scandinavian countries during the last thirty years, from an average of

about 4 per cent in the elections of the early 1950s to about 12 per cent in the 1970s.⁷ The increased mobility is greatest in Norway and least in Sweden. Survey investigations from Sweden and Denmark show a correspondingincrease instability at the individual level. The proportion of voters who change parties has nearly trebled in Sweden since 1960, and the number of voters making their party choice during the final election campaign grew by 50 per cent in the same period (Petersson 1978, 168, 184). Danish studies show that a similar development in mobility is accompaniedby change in electoral motivation: long-term forces like class interest and party identification are becoming less important in the decision-makingprocess, short-term forces have an increasing impact. In 1971 the two kinds of motivation were mentioned by the same number (37 and 38 per cent) of voters, but six years later the short-term reasons were indicated by nearly twice as many (46 against 26 per cent) voters as were the long-term reasons (Worre 1979:79). All these changes in the decision-making process will necessarily imply a decline in the political importance of class and, if not an undermining, at least a loosening of the traditional Scandinavian class party system.

6. The Distinctiveness of the Scandinavian Party Systems

In this analysis we have been able to establish considerable similarity between the four Scandinavian countries. This is true of the macro-level where all party systems are dominated by three big class parties. It is also true of the micro-level, where electoral behaviour reveals a strong inclination class voting, the conditions of which we have analysed.

The last question to consider is whether this Scandinavian party system constitutes a unique type, and what its distinctive character is compared to other democratic party systems. Table 10 compares three groups of countries,each by a number of indicators of social conditioning used previously (Tables 1 and 2). The first type is the Scandinavian. The second consists of the Anglo-American countries analysed by Alford. These are two-party systems. They include no agrarian party, but on the other hand the conservative parties have a clearer class basis in the absence of any serious competitors for the middle class vote. Apparently we find here a simpler kind of class polarisation, a working class party versus a middle class party. But it appears from the table than even in Great Britain (the only country from which data are available for this study), which according to Alford has the strongest class voting tendency among the Anglo-American countries, it is much weaker than in the

Scandinavian countries. The class voting index and the variance explained
by occupation as well as by other social background factors are about
one-third below the Scandinavian level.

The third group of countries are the Continental European party systems ⁸ They are characterised by an extremely low class voting tendency. Occupation seems to play no role at all. Other social variables have more importance: most influential are religious and regional cleavages, which account for one-half of the explained variance. Four of the five countries are dominated by Christian Democrat parties whose attraction goes beyond class barriers: they collect about one-third of the working class vote. France and Italy have very strong Communist parties, which further split the working class: their situation is comparable to that of Finland. On the other side, there is no equivalent of the Scandinavian or Anglo-AmericanConservative and agrarian parties are totally absent. The bourgeois end of the party spectrum differs from one country to another:

usually it is dominated by a Christian Democrat and a Liberal party, a
cleavage which is not socially conditioned and which has no parallel in
Scandinavia.

Hence, the Scandinavian party systems are more strongly connected to the economic aspects of social structure than those of any other democratic ⁹ On the other hand, the tendency towards a weakening of the influence of social background on party choice, and the increasing weight of short-term factors is found in all three types of party system.¹⁰ In the long run, they may erase the peculiarity of the individual party systems and wipe out the differences between the major types.

---

NOTES

1 The Scandinavian Welfare Study is a comparative survey-investigation carried out in Denmark, Norway, Sweden, and Finland in 1972 by the Research Group for Comparative Sociology at the University of Helsinki under the direction of Erik Allardt. It consists of about 1000 interviews from each country. Its subject is not political, but it includes information on voting at the last general election as well as social background, and it was therefore possible to use the data for a comparative analysis of class voting. The data were provided through the Danish Data Archives (Study no. 81). The data have previously been used in a comparative analysis of party preference by Uusitalo (1975). He dichotomizes party choice between left and right parties along the same lines as Alford (1963), and analyses it by means of MCA. Supplementary data material, mainly from the Danish and Swedish election studies, will occasionally be used when that in the Welfare Study is insufficient.

2 Left parties include both Social Democrat and the left wing parties (Communists, People's Socialists, etc.).

3 Alfords index was constructed with the Anglo-American two-party systems in mind. Several systematic and comparative analyses of the Scandinavian party system have employed a similar left-right dichotomy, partly because they used analytical methods which required a variable of a metric nature (Uusitalo 1975; and the Scandinavian contributions to Rose 1974): For the purposes of the present study, it is essential to maintain the qualitative nature of the party variable.

4 Multivariate Nominal Scale Analysis (MNA) is an analysis method belonging to the OSIRIS-system. It is fully described in Andrews and Messenger (1973). It is a kind of multivariate analysis of variance which accepts nominally scaled variables as dependent variables (here party choice) and includes several predictors simultaneously. The programme two measures of the explained variance, one for the predictive power of each variable (eta², similar to the ordinary analysis of variance), the other a multiple correlation coefficient (R²) measuring the joint influence of all predictors. The concept of variance usually presumes a metrically scaled variable. MNA treats each value of the dependent variable (here each party) as a dummy variable, and both eta and R are calculated for each party in the same way as in MCA analysis (cf. Table 2). After this, a joint measure of party is calculated as a kind of weighed average of all parties.

5 Occupation is the most important explanatory variable in this study, and its measurement should therefore be described more accurately. The economically active population is divided into six categories: farmers, other independents, higher and lower salaried employees (including public servants), skilled and unskilled workers. Housewives are included under the occupation of their husbands, and pensioners under their former occupation.

---

6 The high level of class voting for Denmark which was recorded in the Scandinavian Welfare Study differs considerably from two surveys made in connection with the 1971 election. Both indicate a considerably higher left percentage for the middle class, and consequently lower values for the Alford index and for the eta and R coefficients (see below). It is difficult to explain these differences. One reason may be that class polarization sharpened during the year between the surveys, for example in connection with the EEC referendum, and this could have influenced the recall of past voting behaviour. But it is more probable that the differences are due to sampling errors, to which surveys are always subject. It is, therefore, advisable to make some reservations as to the representativeness of the Danish part of the Welfare Study, especially in comparisons with the other Scandinavian countries. The Swedish figures for example, match well with the results of other investigations (Petersson 1978). According to the two 1971 surveys, class voting in Denmark should be at the same level as in Norway. The figures for the three surveys are as follows: Percentage of Left voting in: Welfare Study 72 Election Study 71 Gallup 71 manual occupations 77 77 75 non-manual occupations 21 31 32 Alford index 56 46 43 eta² 20 14 13 R2R² 26 23 20

7 The net mobility is the sum of changes in party strength at one election, that is, the sum of the gains of the winning parties, which equal the sum of the losses of the losing ones. Borre (1979:20-22) has calculated the average increase of the net mobility for all Scandinavian as 1.5 per cent per election.

8 The Continental European countries included in this analysis are: Germany, France, Italy, Belgium and the Netherlands.

9 Rose (1974:17) arrived at a similar conclusion using a very different analytical methcd.

10 Cf. e.g. Rose (1974:258-259, 532) concerning the Netherlands and Britain.

REFERENCES

Alford, R. 1963. Party and Society. Chicago: Rand McNally

Andrews, F. M. and Messenger, R. C. 1973. Multivariate Nominal Scale Analysis. Ann
Arbor: ISR.

Berglund, S. and Lindström, U. 1978. The Scandinavian Party System(s). Lund: Studentlitteratur.

Borre, O. 1979. Ustabilitet ved parlamentsvalg i de fire nordiske lande 1950-1877, in Petersen,
N. (ed.), Dansk Politik i 1970' erne. Copenhagen: Samfunds videnskabeligt Forlag.

Inglehart, R. 1977. The Silent Revolution. Changing Values and Political Style Among
Western Publics. Princeton: Princeton University Press.

Petersson, O. 1978. Väljorna och Valet 1976. Stockholm: Liber.
Rose, R. (ed). 1974. Electoral Behavior. New York: Free Press.

Rusk. J. and Borre, O. 1974. 'The Changing Party Space in Danish Voter Perceptions
1971-73.' European Journal of Political Research 1974, 329-61.

Uusitalo, H. 1975. Class Structure and Party Choice: A Scandinavian Comparison. Helsinki:
Research Group for Comparative Sociology. Research Report 10.

Worre, T. 1979. Forandringer i det danske partisystems sociale grundlag, pp. 68-82, in
Petersen, M. N., ed., Dansk Politik i 1970'erne. Copenhagen: Samfundsvidenskabeligt
Forlag.