Resumé

Danish politics in the 1970s is characterized by a remarkable increase in electoral fluidity. Increasing political mobility should be seen not as a product of passing and specific events, but as a product of basic societal change. The link between political mobility and some dimensions of social change is discussed theoretically and investigated empirically. The discussion focuses on the joint effects of socioeconomic change and political-institutional change. Some reasons why

these changes should lead to increased electoral mobility are stated. The empirical analysis is carried out at the level of municipalities. A positive relationship between various indicators of socio-economic change and political mobility is demonstrated. The relationship is sustained when a control for the effects of social structure is introduced.

1. Introduction

Political scientists sometimes seem to forget that the important thing about an election is its result. Compared with this the motivation of voters or the social factors behind party choice are secondary. It is the number of votes and mandates won or lost that matters to parties and it is the size and direction of shifts between parties that are of importance to the political system as a whole.

This study focuses on one aspect of election results: that of persistence or change. An election result implies persistence if the aggregate distribution of votes among parties is more or less the same as it was for the preceding election; it implies change if the election shows great shifts in the strength of parties.l

The degree of electoral persistence or change is significant politically, because it is related to certain aspects of political stability such as government longevity.² Great electoral mobility reduces the possibility of a government 'surviving' an election. It also increases the temptation to

have frequent elections. The formation of stable coalitions becomes more
complicated if substantial gains or losses among coalition partners are
likely.

The stability of election results may be operationalized in terms of the sum of the percentage points of the votes won or lost by each party.³ Measured in this way there has been a remarkable increase in electoral fluidity in Denmark over the last ten years or so.

Compared to the 'normal' elections of the 1950s and the early 1960s the elections of 1966, 1968, 1971, 1973, 1975, and 1977 resulted in large changes in the distribution of votes among parties. These elections are of course not the first highly volatile elections in Danish history. But never before has there been anything like six such elections in an unbroken sequence. The impression of electoral fluidity in Denmark in the last decade is underlined if one takes the length of electoral periods into consideration. Other things being equal, one should expect greater shifts in voter preferences the longer the time period that has elapsed since the last election. In the last decade, inter-election periods have been fairly short. Accordingly, if one calculates the change in election results per annum the picture of increasing fluidity becomes still more remarkable.

Denmark is not the only Western European country to experience growing fluctuations in election results. Similar fluctuations can, for instance, be observed in Norway, the Netherlands, and the U. K. The phenomenon of growing electoral fluctuations thus has a certain generality.

There are two possible interpretations. First, fluctuations may represent a transitory stage in a restructuring of the party system. Eventually a new and stable party system will emerge. The other possible interpretation of recent electoral results is that they represent the beginning of a new era

of electoral volatility. According to this interpretation, future elections
may also be expected to result in great changes in the strength of the
parties.

This paper favours the second interpretation, that the Danish political system has entered an era of electoral volatility. We shall first discuss some of the factors contributing to this development. Next we shall carry out an empirical analysis of the effects of one of the factors contributing to electoral volatility: the rate of social change. This analysis will be carried out at the level of municipalities.

2. The Changing Context of Electoral Choice

During the recent decades the socio-economic and political factors determining the choice situation of individual voters have changed. The traditional social structures of Danish society have gradually been eroded, leaving the traditional party system in something of a vacuum. As to the political institutions, a new system of interest representation has emerged, based on a close cooperation between interest organizations and the bureaucracy. It offers a new channel of access to interests which once articulated their demands through political parties.⁴ Together, socio-economic and political-institutional changes have probably modified both the behaviour of parties and the ties between parties and voters, in a way conducive to high electoral mobility.

Socio-Economic Change

Traditionally the Danish party system has been explained by reference to two main social and economic cleavages, that between workers and employers, and that between urban and rural groupings (Damgaard, 1974). Around 1920, when the party system crystallized, three of the four major parties could be identified with respect to these two cleavages. The Social Democratic Party represented the workers, the Liberal Party the farmers, and the Conservative Party the urban employers. Only the Radical Liberal Party did not fit neatly within this simple matrix of social cleavages. The Radical Liberals are believed to have been supported by a number of different social groupings, among which urban intellectuals and rural smallholders can be considered the most important. On the whole this party system seems to have been well suited to satisfy the demands of the major occupational groups of society. The ties between parties and groups were reinforced by various cultural mechanisms. Each of the four parties, for example, had its own newspaper in each region of the country. As a

consequence of this four paper system, little dissonant information reached party supporters. The ties between parties and groups were also strengthened by the prevailing pattern of housing which separated the major occupational groups from each other. Finally the educational system, which provided few opportunities to the children of the underprivileged strata, also worked to cement traditional party alignments.

Danish political scientists seem to agree that the cleavages of 1920 were still prominent in politics in the 19605. Poul Meyer, for one, takes these cleavages as a point of departure for discussing where the new large group of public and private employees will locate themselves in the political spectrum (Meyer, 1965, Ch. 9). Erik Damgaard concludes his examination of the Danish party system by saying: 'the party system of the 1960's on the one hand reflects the cleavages of the 1920'₅, on the other hand the changes that occurred in the 1960's might portend further structural development in the party system of tomorrow' (Damgaard, 1974, p. 121). The first part of his conclusion is based - at least partly - on an examination of data from the election survey of 1971.⁵ The interpretation of this evidence is that the traditional pattern of party loyalties to a considerable degree still remained in 1971.

Danish society, however, has for a long time been undergoing farreaching transformations, especially with regard to the social and economic factors believed to underlie party choice and thus to structure the party system.

The socio-economic structure of Danish society has been changing ever since the beginning of industrialization in the second half of the 19th century.⁶ The rate of change, however, has been extraordinarily high in the period 1958-59 to 1973. In the same period Denmark, like most Western European countries, experienced an unprecedented economic boom. Increasing wealth served as the foundation on which the modem Danish welfare state was constructed.

The changes that occurred from about 1960 onwards are fairly typical of a society moving into the post-industrial era. There was a rapid decrease in the relative proportion of the labour force employed within the agricultural sector. Only a few decades ago the agricultural sector had held a dominant position within the economy. But from 1960 to 1974, its proportion of the labour force fell from 17.8% to 9.2%. During the same period, the manufacturing sector tended to stagnate or even decrease. The expanding part of the economy has been the tertiary or service sector. Between 1960 and 1974, administration and the professions more than doubled their proportion of the labour force from 12.9% to 27.1%. The most significant growth

occurred within the public welfare sector, i.e. within education, health, and social services. The percentage of the labour force employed in the welfare sector increased from 8.2% in 1960 to 18.0% in 1973. It is notable that growth in these sectors has been somewhat higher in the second half of the period than in the first. In the same period women have been mobilized into the labour force at a very high rate.

The shifts between the sectors of the economy have been accompanied
by shifts in the class structure (table 2.1.).

There has been a decrease in the number of self-employed and a large increase in the number of white collar personnel and salaried employees. The latter group now outnumbers the workers, whose number has been stagnating or declining. Furthermore, there has been an increasing differentiation within the main categories of the labour force. The categories of 'workers' or of 'white collar employees' each now covers a wide range of groups with highly differing conditions of work and pay.

The changes in the economic structure of society have more or less eroded the social basis of the traditional party system. The opposing groups of farmers, workers and employers have lost their dominant position within the electorate, and increasing occupational differentiation has made the interests of new groupings much less clear. We are thus confronted with a much more complex configuration of interests than in the early phases of industrialization.

Social change has not, of course, been restricted to the economic sphere. There have also been changes in the cultural 'mechanisms' traditionally supporting the ties between specific occupational groups and specific parties. The old four newspaper system has disappeared and the remaining newspapers have adopted a more or less non-partisan stance. Local cultural insulation has also decreased as a consequence of the greater importance of the electronic mass media, which in Denmark are controlled by a state monopoly. Patterns of housing have also altered dramatically. There has been a rapid increase in the number of individually owned houses. This change has penetrated far into the working class, and many working-class families have left inner city working class neighbourhoods

for more socially mixed suburban areas. Ties between parties and occupational groups may also have been weakened by the expansion of the educational sector. Following the elementary school law of 1958, the quality and length of compulsory primary education have been raised. At the higher levels of the educational system younger generations have received an increasing amount of theoretical education. Parents therefore become exposed to new stimuli through their children.

In order to understand the impact of this accelerated rate of societal change it is important to stress that it does not simply represent generational shifts. Large numbers of people have had to change their occupation and perhaps also their place of residence in middle age. The illustrative case is not just the farmer who sends his son to town to become a salaried employee. In many cases the farmer himself has had to give up his occupation and try to make a living outside the agricultural sector. In sum, the traditional social basis of party choice clearly does not exist any longer for many Danish voters.

Changes at the Political-Institutional Level

The political effects of social change depend, of course, on the structure of the political system. Over the past few decades, important alterations have taken place in the way political decisions are made. At the core of these changes one finds the strengthening of the corporate channel of influence (Dahlerup et al., 1975).

In recent decades the major interest groups in Denmark have gradually gained more ready access to the central structures of decision-making. More precisely, interest groups have themselves become part of the regular structure of decision-making. The integration of interest groups into the state apparatus may be traced back to before the turn of the century. It has not been a simple linear developmental process: rather there has been a series of relative peaks and troughs. The first relative peak occurred in connection with a regulatory economic policy during, and immediately after World War I. The next is found in the 1930s as a result of measures taken to counter the general economic crises. Relations between state and organizations were also intense during and in the years after World War 11, again as a result of attempts to regulate the economy. Finally, there has been a new intensification of contacts between state and organizations in the 1960s and the early 19705. This time, the background is a period of unparallelled economic growth and rapid expansion of the public welfare sector. Integration has been most pronounced for the major interest groups - those of the farmers, workers, and employers.

The major interest groups can no longer be described as 'pressure groups' striving for access and influence. In the contemporary Danish political system, the access of interest groups to the decision-makers has become institutionalized, and representatives of interest groups have become coopted into the decision-centers. In general, one may say that there has been a trend towards a more corporalist system of interest representation, where organizations no longer compete for either members or access to the state authorities, and where they are given de facto recognition and are licensed by the state.⁷

It seems reasonable to assume that the trend towards corporatism has had important consequences for the role or functions performed by political parties within the political system. The integration between the state and the organizations of farmers, workers, and employers means that the political parties do become somewhat redundant in terms of the protection of the major occupational interests which 50 years ago served as the basis of the party system. Today these interests are mainly looked after within the corporative structures of decision-making.

Consequences of Socio-Economic and Political-Institutional Change

We have seen that both social and institutional change threaten the traditional function of parties as the representatives of major occupational groups in society. The numerical strength of the party-forming groups has declined and a much more complex configuration of interests has arisen. At the institutional level, there is an alternative system of interest representation, which may much more adequately reflect the diverse interest of modern society. With this background, the behaviour of voters and parties is bound to change.

Due to rapid socio-economic changes, large parts of the electorate have moved from a situation where they experienced relatively homogeneous political stimuli to a new one where they are much more likely to be confronted by divergent stimuli. Cross-pressure may cause a weakening of party identifications and accordingly remove some of the brakes on shifts in voter preferences.

Due to institutional change, voters may find that political parties are no longer necessary for the protection of their basic occupational interests. These interests are now adequately protected by the corporative system of interest representation. Recent electoral research has to some degree confirmed that voters are aware of the central role of interest groups in the political process. In a 1971 national survey, 48% of the respondents agreed that 'the big trade unions and employers' association have obtained a

power which properly belongs to the Folketing'. Even more strikingly, no fewer than 74% agreed that 'whichever party is in power, it is a few big organizations that determine what will be done' (Vælgerundersøgelsen, 1974, p. 65). Furthermore, the voter may also perceive that contemporary parties are hardly able to represent the diversified configuration of occupational interests found today. More voters than previously may therefore find themselves 'free' to use their votes to protect or maximize interests other than those of their occupation.

As voting patterns change, the behaviour of parties also changes. Nowadays political parties cannot survive only by appealing to a specific occupational group. They have to give priority to issues which are only vaguely related to occupational interests, and they have to be alert to a host of latent issues which may at any time become politicized. In the past decade, Danish parties seem to have been preoccupied with a number of 'new' issue areas, e.g. the taxation of individually owned houses, moral problems such as abortion and pornography, and red tape bureaucracy within the public sector. These kinds of issues can hardly be seen as concrete manifestations of the urban/rural or worker/employer cleavages. New parties, which are not bound by prior commitments, have gained support by focusing upon issues not adequately dealt with by the existing parties. This has been done by the Progress Party (a right wing populist party led by Mogens Glistrup), the Centre-Democrats and the Christian People's Party. Similarly, small old parties may attract more votes by appealing to current grievances. This has been the case for the Communist Party and the Justice Party (a Georgist Party).

Parties are not totally free to take whatever position they may find gratifying in terms of attracting votes. This is especially true for some of the older ideological parties, which must take great pains to integrate new positions into their worldview and to make the party's position acceptable to its members. Nevertheless, all parties have to intensify their efforts to contrive new issues or to design new approaches to already politicized issues in order to retain or expand their vote.

Changes in the positions and behaviour of voters and parties do not automatically cause increasing electoral fluidity. Little electoral change will occur so long as psychological party identifications are widespread and strong. A party system may thus survive for a long period after the disappearance of its original raison d'etre. This is probably the reason why electoral instability in Denmark did not materialize until late in the 19605, even though social and institutional change started much earlier.

A party system that is no longer sustained by socio-economic and

political-institutional underpinnings, however, will not last forever. Sooner or later political events will provide the necessary catalyst. For Denmark, two political events may be considered the immediate causes of instability. One was the policy of the center-right government from 1968-71, which frustrated many voters who had expected the government to reverse the welfare policy of the Social Democrats. The other was the EEC referendum of 1972, which left a great many voters with the experience of having voted against the position taken by their respective parties.

Once initiated, the process of increasing electoral mobility in the Danish case was eased by the 2% threshold in Danish election law, which is the lowest threshold requirement in Western Europe except for the Netherlands. The low barrier makes it rather easy for new parties to gain sufficient support in the opinion polls to prevent the old parties from invoking 'the wasted vote argument'.⁸

3. The Rate of Socio-Economic Change and Electoral Fluidity

An Ecological Conception of the Relationship

The relationship between socio-economic and electoral instability may operate at the individual as well as the societal level. Social change may cause electoral instability because individuals who change their position in the social structure also react politically. A change in, say, occupation or neighbourhood may produce a loosening of party identification and hence less stability in party choice, The importance of the relationship between social change and electoral stability at the individual level should not, however, be exaggerated. In spite of the great changes in Danish society it is only a minority of the electorate who have personally experienced profound social change in the midst of their adult lives.

The relationship also works at the level of the 'social context'. Almost everybody living in a rapidly changing society is bound to perceive the occurrence of social and economic change, no matter whether their individual situation is directly affected or not. One need not personally change occupation or neighbourhood to perceive that things are not what they used to be: the mere perception of profound change in the social context may be sufficient to weaken former party identification and thus in itself be a cause of electoral fluidity.

It is the 'contextual' conception which will be the focus of our analysis.
When we pose the problem this way we have a genuine ecological problem,

and ecological data may serve as indicators of the social context, as it can be presumed that the social context may at least partly be geografically defined. It is a genuine ecological problem in the sense that the ecological units are given a direct sociological interpretation.⁹ Thus the purpose of the ecological analysis is not to infer from one level of analysis to another.

Municipalities have been chosen as units of analysis. There are 277 municipalities in Denmark, 95% of which range in population from 4,000 to 75,000.¹⁰ It is problematical whether this unit is the most relevant for a discussion of the effect of the perception of social change. At one extreme, it might be argued that changes in the social context are the same for all Danes, regardless of where they live in the country. According to this view, the perception of social change is not a function of where in the country a person lives, and thus the relevant contextual unit should be the whole country. At the other extreme it might be argued that it is necessary to go below the level of the municipality to find the relevant contextual unit. Measuring change at the level of the municipality may, however, serve as a preliminary indicator of change in the social context of voters.

Based on our general hypothesis concerning the relationship between social change and aggregate electoral instability at the national level one might expect a high rate of social change to be associated with high aggregate electoral fluidity at the level of the municipality. This hypothesis will be tested with data from the 1971 and 1973 elections.

The Relationship Between Socio-Economic Change and Electoral Mobility

The dependent variable of the analysis is aggregate electoral change at the
municipal level as measured in terms of the sum of the percentage points
won or lost by each party.

The independent variables should give a broad description of changes in the social structure of municipalities. From the discussion in the previous section of the paper, it follows that changes in at least four dimensions of socio-economic change might be important for the explanation of electoral instability. The dimensions are the distribution of the labour force among different economic sectors (farming, manufacturing etc.), the distribution of the labour force among different occupational categories (self-employed, salaried employees, workers, and so on), the number of women mobilized into the labour force and, finally, the distribution of dwellings of different types, i.e. the relative numbers of individually owned houses, apartments and so on.

To represent changes in the distribution of the labour force among economic sectors and among occupational categories, we have chosen the decline of agriculture and the growth of white collar personnel. Changes in housing structure are represented by an index measuring changes in the relative number of apartments and houses. The rate of mobilization of women in the labour force is represented by the growth of the relative number of married women in the labour force.

The analysis is based on a dichotomization of municipalities along each of the four dimensions of social change. For each dimension, the median has been used as the dividing point between 'large' and 'small' changes. Dichotomization permits a simple and easily interpretable analysis with a sufficient number of cases in each category. The use of dichotomized variables will also enable us to construct some simple indices of social change.

When the average on the index of electoral mobility is calculated within
each of the eight categories resulting from the procedure of dichotomization,
the patterns of Table 3.1. emerge.

For three of the four dimensions of social change the expected relationships are found. For changes in the distribution of dwellings and in the proportion of married women and white collar personnel in the labour force, electoral mobility is higher in municipalities with large changes than in municipalities with small changes. The differences between the two kinds of municipalities range from 2.3 to 5.3. The larger figure is found in connection with the growth of white collar employees.

As for the fourth independent variable, decline of agriculture, electoral
mobility is somewhat higher in the municipalities with little social change
than in the municipalities with considerable social change. In this respect,

stable municipalities seem to have higher electoral mobility than unstable ones. Accordingly, a partial contradiction to our major hypothesis seems to have been found. In a little while, however, we shall demonstrate that this apparent contradiction is due to a spurious correlation.

To round off the preliminary analysis of the relationship between social change and electoral fluidity, we shall concentrate briefly on the three dimensions where large changes are associated with large electoral fluidity. We shall examine the hypothesis that the amount of electoral mobility increases with the number of social dimensions that are simultaneously undergoing change. A simple index of overall change has been constructed by counting for each municipality the number of social dimensions where it lies in the upper half of the dichotomized change.

The relationship between overall socio-economic change and electoral mobility is presented in Table 3.2. As expected, the table shows that electoral mobility increases with the number of dimensions with large social change. Consequently there is a clear-cut difference between the socially most stable and the socially most unstable municipalities.

This preliminary investigation of the relationship between social change and electoral mobility justifies the tentative conclusion that three of the dimensions of social change are positively related to the level of electoral mobility, and that the overall number of dimensions undergoing change is also positively related to mobility.

Elaborating the Relationship Between Social Change and Electoral Mobility

The argument we wish to test empirically is that changes in the social context contribute to increasing electoral mobility. An obvious alternative explanation of variation in the aggregate level of electoral mobility is, of course, that it is the social structure of municipalities per se, and not changes in the social structure, which are important. Such an explanation might run as follows: the more a municipality can be characterized as

urban, and the more it is characterized by individually owned houses, white collar employees, and women in the labour force, the more its voters will be electorally mobile. If so, the increase in electoral mobility at the national level might be explained by the greater number of urban onefamily house dominated municipalities. According to this alternative explanation it is not the dynamics of social change but the distribution of the population among fixed social contexts, which is central.

So far, we have only demonstrated a statistical association between social change and electoral mobility. The importance of existing social structure might, however, be so great that the independent effect of social change has no significance. We shall therefore investigate first the relationship between the social structure itself and electoral mobility, and then the relationship between social change and electoral mobility while controlling for social structure.

The socio-economic structure of municipalities may be described by
twelve indicators from the public census corresponding to the above
mentioned four dimensions of socio-economic change.

The twelve indicators of social structure have been factor analysed, solely to reduce the number of indicators without losing too much variance. The factor analysis resulted in three factors, which explained some 80% of the total variance. For further analysis we picked out the indicators loading most heavily on each factor. These indicators are the proportion of the labour force in agriculture, the proportion of private houses to total dwellings, and the proportion of workers in the labour force. Their loadings were .97, .97, .93 on the respective factors.

Next, the municipalities were divided into two categories on each of the indicators of social structure. The median was once more used as the dividing point. The relationship between social structure and electoral mobility is presented in Table 3.3.

Aggregate electoral mobility is clearly related to social structure. Municipalities

with some agriculture are more stable than those with relatively little agriculture. Municipalities with relatively many workers are more stable than those with relatively few workers, and municipalities with relatively few private houses are more stable than those with relatively many private houses.

The next and essential problem is whether social structure alone can explain differences in electoral mobility or whether changes in the social structure do independently contribute to electoral mobility when we control for social structure.

When controlling for social structure, one cannot expect substantial differences in aggregate electoral mobility between municipalities with large and small social changes. There are two reasons for this. First, the relationship between social and electoral change is not a direct one. It is mediated through a complicated set of intervening variables and social processes, which are not adequately covered by our data and thus cannot be explored further here. Secondly, the variables of socio-economic structure and of socio-economic change are heavily interrelated, giving rise to a problem of multicolinearity or 'block-booking' among the independent and control variables.ⁿ Controlling for the social structure means that only a residual variance of the dependent variable is left when we investigate the effect of social change upon electoral mobility. Furthermore, because of block-booking a greater or lesser proportion of the variance explained by the control variables may in fact be due to the independent variables. The data do not make it possible precisely to determine how much variance is due to the control variables and how much to the independent variables.

As we have four variables of social change and three dichotomized variables representing social structure the control for social structure gives us 24 possible relationships (Table 3.4.). The figures represent the effect of social change on electoral mobility within each category of the three variables of social structure. Thus the figure in the upper right corner of the table indicates that there is a difference of 1.3 in the average values on the index of electoral mobility between municipalities with a large and municipalities with a small decline of agriculture, which have a high proportion of the population in agriculture. In this instance, we find that social change does contribute independently - though modestly - to the explanation of electoral mobility. The same result is found in 20 out of the 24 controlled relationships in the table. In general, the table shows that the relationship between social change and electoral mobility is not substantially weakened when the control for social structure is introduced. Furthermore, it demonstrates once again that changes in the proportion of white collar personnel constitute the most important dimension of social change in terms of voter mobility.

The last column of the table deserves a separate commentary. It shows the effect of decline in agriculture when controlling for social structure. Earlier, when we examined the bivariate relationship between decline in agriculture and electoral mobility, the results seemed to contradict our hypothesis. Electoral mobility turned out to be somewhat higher in municipalities with small social changes than in municipalities with significant changes. We now obtain a reversal of this negative finding.

When the effect of decline of agriculture is analysed under control for

the relative proportion of agriculture, the original hypothesis is confirmed. Under this control, municipalities with a substantial decline in agriculture turn out to be less electorally stable than municipalities with a small decline. The 'negative' finding of the bivariate analysis seems to have been caused by a distortion effect of the level of agriculture. The reason for this is that there is a high correlation between the level of agriculture and the rate of decline of agriculture. It should also be noted that three of the four cells in Table 3.4. which contradict our hypothesis, are found in connection with a decline in agriculture. These exceptions to the general hypothesis are probably due to the distorting effects of the level of agriculture, which is not controlled for in these instances.

So far we have controlled for the three variables representing social structure one by one. The proportion of the labour force in agriculture, the proportion of houses, and the proportion of workers in the labour force may, however, all be used to construct a typology of social structure. If we take all combinations of these three dichotomized variables we obtain a typology of eight types, each indicating a unique socio-economic structure. The definition of each type and the level of electoral mobility associated with each type can be seen from Table 3.5.

Generally, the types differ with regard to the level of electoral mobility. The first three types and the next two, however, are quite similar with respect to electoral mobility, and will therefore be grouped together when we investigate the relationship between social change and electoral mobility within types of social structures.

For this final test, social change is defined by an index constructed by counting the number of dimensions on which municipalities show large changes. As a consequence of the reversion of the 'negative' finding with respect to the effects of decline of agriculture all four dimensions of social change have been taken into consideration in this index.

The relationships between socio-economic change and electoral mobility
within types of socio-economic structure are set out in Table 3.6.

There is not a linear or even a strictly monotonic relationship between the overall index of social change and electoral mobility. Changes in three or four dimensions do not as a rule have a greater effect than changes in two dimensions. On the other hand the basic hypothesis that large social change is associated with significant electoral mobility is once more generally supported.

4. Conclusion

The thesis of this article has been that the Danish political system has entered an era of high electoral mobility. It has been demonstrated empirically that at the level of municipalities electoral mobility is related to social change; that is, the erosion of electoral stability is at least partly due to recent rapid socio-economic change. Hence increasing electoral mobility is also a product of basic societal factors and not only of passing and specific political events.

This background makes it hard to imagine a basis on which a new stable party-system may materialize. At least it is hard to the authors. If our interpretation is right, the prognosis must be one of continued electoral mobility.

We have only dealt with the size of aggregate electoral mobility, not with the direction of electoral changes. In addition to the extent of electoral mobility it also of course matters whether shifts occur mainly among old established parties or whether shifts involve the sudden rise of new and/or extremist parties. In the latter case, the consequences for the functioning of the political system are probably aggravated. For Denmark, increasing electoral mobility has been accompanied by the rise of large new parties, some of which may be viewed as extremist ones. The prospect of continuing high electoral mobility may thus be quite a gloomy one.

---

NOTES

1 Under the complex provisions of Danish electoral law, shifts in the popular vote are always reflected directly in the number of mandates assigned to each party.

2 The by no means well defined concept of political stability is discussed in Hurwitz (1973).

3 That is, the range of the index is from 0 (total stability) to 200 (total instability). A similar index is used by Rose and Urwin (1970).

4 This new channel is well-known from discussions of other European countries, see e.g. Rokkan, 1966.

5 This survey was made by a group from the universities of Copenhagen and Aarhus. See Vælgerundersøgelsen, 1974.

---

6 For an overall description of this development, see Hansen, 1974.

7 For a general discussion of this type of development see Heisler and Kvavik, 1974. The elements of the corporatism concept used are from Schmitter, 1974.

8 The importance of the wasted vote argument is demonstrated in Fisher, 1973.

9 Ecological analysis may be of two types: one doing it in its own right and one where the purpose is to infer from the ecological level to the individual. See Valkonen, 1969.

10 The municipality makes up the smallest unit of self-government in Denmark. More important for our purpose is the fact that the municipality in most cases is the lowest unit in the public census data.

11 The term 'block-booking' is from Rosenberg, 1968, pp. 26-27

REFERENCES

Dahlerup, D., Jarlov, C, Johansen, L. N., and Kristensen, O. P. 1975. Korporatismebegrebet
og studiet af samspillet mellem politiske institutioner', Økonomi og Politik 49, 317—
344.

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

Fisher, S. L. 1973. 'The Wasted Vote Thesis', Comparative Politics 5, 293-299.

Hansen, S. Å. 1974. Økonomisk vækst i Danmark. København: Akademisk Forlag.

Heisler, M. O. and Kvavik, R. B. 1974. 'Patterns of European Politics: The 'European Polity'
Model', pp. 27-28 in Heisler M. 0., ED., Politics in Europe. New York: David McKay.

Hurwitz, L. 1973. 'Contemporary Approaches to Political Stability', Comparative Political
Studies 3, 449-463.

Meyer, P. 1965. Politiske Partier. København: Munksgård.

Rokkan, S. 1966. 'Norway: Numerical Democracy and Corporate Pluralism', pp. 70-115 in
Dahl, R., ed., Political Oppositions in Western Democracies. New Haven and London:
Yale University Press.

Rose, R. and Urwin, D. W. 1970. 'Persistence and Change in Western Party Systems Since
1945', Political Studies 18, 287-319.

Rosenberg, M. 1968. The Logic of Survey Analysis. New York: Basic Books.

Schmitter, P. C. 1974. 'Still the Century of Corporatism?', pp. 85-131 in Pike, F. B. and
Stritch, eds., The New Corporatism. Notre Dame and London: University of Notre Dame
Press.

Vælgerundersøgelsen. 1974. Vælgerundersøgelsen August-Oktober 1971. Århus: Institut for
Statskundskab.

Valkonen, T. 1969. 'Individual and Structural Effects in Ecological Research', pp. 53-68 in
Dogan, M. and Rokkan, S., eds., Quantitative Ecological Analysis in the Social Sciences.
Cambridge: The M.I.T. Press.