Resumé

Referendums represent a special variant of political elections. Although it must be assumed that the behaviour of voting in referendums is determined by a range of factors identical to those that are operative in other contests, it can be asked to what extent referendum voting behaviour is sensitive to active campaigning on the part of political parties and organized groups. It may add to our understanding of referendums, therefore, to analyze them in terms of three 'ideal types': partystructured, and unstructured contests. Illustrations taken from certain recent referendums support the argument that it is useful to consider the extent to which these ideal types are in evidence - alone or in combination - when explaining referendum voting behaviour.

The Referendum as an Object of Research

In comparison with studies on legislative elections, analysis of referendums has not been carried very far. However, in recent years the subject of referendum voting behaviour has received considerable attention, especially by Danish authors (Nielsen 1970, Petersen & Elklit 1973, Svensson 1979). The latest contribution, by Palle Svensson, is entitled 'Class, Party and Ideology: A Danish Case Study of Electoral Behaviour in Referendums' (Svensson 1984). As the title indicates, it is a case study of the 1978 referendum on lowering the voting age. But Svensson also considers the earlier Danish referendums on the same subject, which took place in 1953, 1961, 1969 and 1971, respectively. These constitute a series of'natural experiments', which can be usefully compared with the 1978 case. Only the 1971 contest represents an exception; unlike the others, it was held simultaneously with a parliamentary election and was overshadowed by the latter.

In addition to his thorough analysis of the 1978 referendum, Svensson offers a contribution to the search for general propositions by referring to a sketch presented by Ole Tonsgaard (Tonsgaard 1974, 1984). The latter undoubtedly deserves to be called a pioneering work. An attempt is made by the author to construct a typology of referendums on the basis of issue characteristics. Tonsgaard (ef. Svensson 1984, 191-192):

How salient is the referendum issue to the voters?
How complex is the referendum issue?

How important is the referendum issue to the everyday life of the voters?

It is suggested that referendums can be classified on the basis of these characteristics or low saliency, etc.) in such a manner that all referendums on, for instance, the voting age should be seen as belonging to one type, while referendums on the subject of membership in the European Community should be regarded as belonging to another type, and so on.

We are not convinced, however, that this classification scheme is a useful one. In particular it seems to us misleading to designate an issue 'salient' or 'not salient' as if either saliency or non-saliency were a property, intrinsic and unchanging, any given issue. Such is not the case. Almost every political issue can be made salient to voters, if parties or interest groups work sufficiently hard for that purpose. If you say, 'The political participation of young people was a salient issue in Denmark in 1978', what you mean is that it was made a salient issue by Danish politicians, whose intensive and closely coordinated campaign for that purpose has been most convincingly described in Svensson's 1984 article. In 1961 Danish politicians waged no such campaign, so the issue of the voting age was not salient that year. At the time of the 1969 referendum, without any activity being undertaken by politicians, the issue none the less became salient to the electorate, apparently for specific reasons to which we revert below.

Somewhat similar considerations apply to the notions of complexity and importance for the daily lives of voters. No doubt observers can compare the questions posed in various referendums, for example with regard to their complexity. However, to the political scientist, the views of outside observers are of less interest than the views of the voters themselves. An issue regarded as somewhat complex by the expert will probably appear as highly complex to the average voter. But voter perceptions can change very quickly if some political party or interest group undertakes the task of simplifying the issue for the benefit of the electorate. Thus the 'objectively complex' issue of entry into the EC was presented to large groups of voters in countries like Denmark and Norway - to primary producers, for example - as a simple question of economic survival. Thereby the importance of this issue in the eyes of the voters was also affected.

While most Danes regarded it as being rather unimportant for their daily lives, both before the 1972 referendum campaign started and after it had ended, they did consider the issue important as long as the campaign lasted. The latter period was comparatively short, and Tonsgaard therefore classifies the question of EC membership as belonging to the 'unimportant' type of referendum (Tonsgaard 1984,329). But the campaign period was decisive for the outcome of the contest; consequently we could just as well say that the issue of membership was perceived as important. Its character changed over time. In fact the EC issue cannot be said

to have belonged durably to one specific type either in terms of perceived importance,perceived or saliency. Tonsgaard has shown clearly that these three elements are significant for voters' behaviour in a referendum. But they cannot be used as building-blocks for the construction of a general typology. They are themselves dependent on other factors, first and foremost the activity exercised by parties, by interest groups, and by the media. Here, in our opinion, is where students of electoral behaviour should focus their attention. We thereforeventure propose a classification of the Danish referendum on the voting age which is different from the one presented by Svensson.

In support of our case, we also compare the British and Norwegian EC referendums. It seems to us that in this matter some serious misunderstandings have arisen. Svensson and Tonsgaard refer to the Norwegian contest as an example showing that 'partisan attachment is almost surely the primary force in referendum voting' (Svensson 1984, 177, cf. Tonsgaard 1984, 324). The statement is quoted from an article which appeared in 1983 in the American Journal of Political Science (Pierce et al. 1983, 61). Both Tonsgaard and Svensson appear to accept it unquestioningly. As far as Norway is concerned, however, the conclusion is a doubtful one. Therefore we find that a further comparative analysis of the EC referendums in Britain and Norway is called for.

A Tentative Typology

Students of referendum voting behaviour have made widely varying assessments of the factors that influence voters' choice. This is hardly surprising in view of the different kinds of issues that have been decided by referendum, and the different circumstances under which the decision has taken place.l It seems possible, however, draw a distinction between the following three 'ideal types':

(a) party-structured,

(b) group-structured, and
(c) unstructured referendums.

In a referendum of type (a), the voters' choice between the alternatives offered to them is determined by their party preference, while in a referendum of type (b) it is determined by the influence of interest groups. Australian referendums can be said to approximate ideal type (a) because of the exceptionally high level of partisanship in that country, while in California most referendums come closer to type (b). Californian campaigns are conducted by organized groups, which can be either ad hoc volunteer groups or permanent interest organizations, or one of the professional organizations that exists in the state solely to serve this function (Aitkin 1978, 133, Lee 1978, 101).

To the extent that voters' choice is not determined by either one of these two factors, a referendum can be designated as unstructured. In this form of referendum the voter does not follow the advice either of parties or of permanent or ad hoc interest groups. His or her choice will be determined by interaction with people in the same occupation or with friends, relatives, and neighbours; briefly, by personal interaction within some latent group to which the voter happens to belong. What we intend to convey by the word 'unstructured' is the idea not of a process devoid of structure, but of a process that is structured by unorganized groups, 'latent-group structured'.

The three referendum types are not mutually exclusive. In any process of mass opinion formation, some of the result may stem from the impact of party identification some from the co-ordinated campaign of interest groups. But some of the effect may also be produced by the voters' personal ideology, their individual decisions. And a series of such decisions may well tend in the same direction under the influence of a factor that the citizens concerned have in common (whether they belong to the same sex, the same occupation, the same age group, or perhaps the same region). This is what we call an unstructured process.

In what follows we shall compare a limited number of referendums that have been conducted during the last couple of decades. The next section considers the question of unstructured referendum behaviour and then refers briefly to an example of the party-structured type. The following section illustrates the fact that different types of voting behaviour can appear side by side in one and the same referendum.

Unstructured Referendum Behaviour

In the state of California, organized groups usually play a prominent role in referendum campaigns. But such is not always the case, and since partisan attachments are weak among Californians the outcome will then be influenced by specific cultural attitudes (Wolflnger & Greenstein 1969, 82). These differ quite markedly between the more liberal Northern and the more conservative Southern part of the state. In a sense the electorate is divided into two different regional cultures, or two latent geographical groups.

In contrast, Denmark is a country with a homogeneous political culture, without any important regional cleavages (Svensson 1984, 193). Yet also here we are presented with a rather clear example of an unstructured referendum, one in which not a regional but a generational cleavage assumed paramount importance. The example we have in mind is the 1969 contest about the question of lowering the voting age to 18 years.

The age limit for electoral participation, which had been fixed at 23 years in
1953, was changed to 21 years in 1961 after the holding of a mandatory

referendum. Voters at that time took very little interest in the issue, the turnout
being as low as 37.2 percent. There was a slight affirmative majority.

Towards the end of the decade proposals were put forward in the Danish parliament for a further reduction of the voting age limit. The matter was considered during the 1968/69 legislative session, when a government bill and a private bill called for a voting age of 20 years and 18 years respectively. At first the government bill seemed to hold favour with parliament, but between the first and second reading local associations, intent on soliciting the support of young people, put pressure on party leaders to lower the limit to 18 years. Accordingly, when the second reading took place in parliament on June 4 1969, there turned out to be a solid majority in favour of 18 years, and this age limit was duly accepted two days later at the third and final reading. The issue was submitted to a mandatory referendum on June 24.

In their eagerness to satisfy the wishes of constituency activists, few parliamentarians to take voters' opinions into account. According to a Gallup poll dating from the latter part of 1968, only a minority of the electorate favoured a lowering of the voting age to 18 years (Nielsen 1970), but voters were expected to regard this issue as being without importance for their daily lives. Had not the experience of 1961 indicated a state of general indifference? It must have seemed definitely unrealistic to think that 30 per cent of the eligible voters should take the trouble to go to the polling stations in order to reject the new measure. Therefore, politicians paid scant attention to public opinion in this case. Practically no attempt was made to influence the electorate through the mass media.

A new Gallup poll published on June 22 indicated a negative majority, but still the result on polling day came as a great surprise. Not only was the majority against nearly 80 per cent, but also compared with 1961 there was an enormous increase in the rate of participation. This time it reached a level of 63.6 per cent. One half of the eligible voters rejected the measure, as against only 16.6 per cent in 1961. The contrast was a striking one. It is true that lowering the age limit from 21 to 18 was not quite the same thing as lowering it from 23 to 21. But the force of the reaction indicates that also another factor had been at work. Clearly Danish voters regarded the question of letting youth take part in politics as much more important in 1969 than they had done in 1961. The proposal had been endorsed by a large majority in parliament and was not opposed by any organizations or news media of importance. Observers agreed that propaganda acitivity was conducted on a very low key right up to polling day. Yet voters turned out in large numbers to register their opposition. A probable explanation is that they did so in order to express displeasure with the youth rebellion of the late 1960'5. Teenagers had started exhibiting most unusual haircuts, clothes, and manners, while certain groups of young people, especially students, took a leading part in radical political movements. This happened not only in Denmark but also in other countries, most strikingly in France during the semi-revolutionary weeks of May

1968, as well as in the United States and Germany. All this was brought home to Danish voters by the press and television. Most of them seem to have been very unfavourably impressed by what they saw. The new currents among the younger generation were disturbing, and when people's own or neighbours' children showed signs of being carried along, the participation of youth in politics no longer seemed to be an issue without significance to ordinary voters. They felt that their daily lives might be affected.

The relatively high turnout seems to have been due mostly to increased participation among the elderly and middle-aged. The outcome was decided by the behaviour of people belonging to 'the mature age group', whose actions had in no way been coordinated in advance. It can be said that the opinion typical of their age group was moulded by the many reports and pictures of the youth rebellion appearing in the mass media. We would argue, however, that the mass media did not structure opinion. There is agreement that at no time did media people consciously wage a campaign aimed at influencing the electorate.

Voters reacted on their own, immediately, 'spontaneously'. The referendum was unstructured, that is to say, latent-group-structured. The Gallup poll showed a sharply decreasing rate of support for the 18-year limit with increasing age (Nielsen 1970,304):

In 1978 a proposal to lower the voting age to 18 years was once more put before the Danish electorate, and this time it was approved. This is the case that Svensson has studied in detail. He shows how Danish politicians took care not to repeat their mistake of underrating the reaction of the electorate. The new proposal was put forward in November 1976.

The Gallup Institute found ... that only 30 per cent would support a voting age of 18 years, (and) there was hardly any change in public opinion during the next year ... By the time the Folketing passed the bill in the early summer of 1978 there was still a clear majority against the proposed lowering of the voting age. The next three months, however, led to a total shift in public opinion (Svensson 1984, 182).

The reason for this change is quite clear:

The public debate in 1978 on the voting age issue was rather special in at least two respects. First, only the supporters of lowering the voting age to 18 years were organized. Second, the campaign was by and large an efforton the part of the political elite to convince a reluctant electorate. Most of the political parties used their organization to support the proposed lowering of the voting age ... During the first half of September, all political parties were given time to present their views on the issue on both radio and television. These broadcasts were predominantly and almost tediously in favour of lowering the voting age to 18 years (Svensson 1979, 66; 1984, 183-184).

The measure was approved by 53.8 per cent of the voters on September 19, the
turnout being 63.2 per cent.

The decrease in the national percentage of opponents, by one third in the course of three months - from 55 per cent in June to 37 per cent in September - is an eloquent testimony to the persuasive power exerted by political parties when they undertake the task of structuring public opinion. In contrast to the 1969 consultation on lowering the voting age to 18 years, the 1978 referendum on the same issue deserves the designation party-structured rather than unstructured.

None the less it must be admitted that a certain percentage of the electorate
proved impervious to the appeals of the parties and voted No, apparently following
personal ideology.

Party-Structured, Group-Structured, Unstructured Response

The question of partisan influence can also be studied in connection with referendums held in different countries on entry into the European Community. British and Norwegian voters were asked to take a stand on that issue in the 1970'5. The former replied with a positive, the latter with a negative majority.

These referendums can be studied from different angles, depending on whether the purpose of research is to find how far party leaders succeeded in making their influence felt, or, on the contrary, to what extent they failed in attempts to make their point of view prevail. The former approach has been chosen in a recent article in which the British and Norwegian EC referendums are compared. Although it mainly focuses on left-right attitudes, the article also contains a section on the relation between party allegiance and the EC vote on the basis of survey data (Pierce et al. 1983, 46 fif.) The British referendum is found to be clearly party-structured, and it is concluded that also the Norwegian case can be characterized as party-structured. As suggested in our introductory section, we disagree with the latter conclusion. We will also question the way the concept of partisanship is operationalized.

We consider that in the case of some Norwegian citizens, referendum voting behaviour was party-structured, in the case of some others it was group-structured, and in the case of still others unstructured, i.e. latent-group-structured. Not only was the influence of parties on the result a limited one; the referendum also clearly weakened the loyalty of many a partisan.

The leaders of two Norwegian parties, Agrarians and Left Socialists, opposed entry into the EC, while it was advocated by the leadership of two other parties, the Labour Party and the Conservatives.² The latter two were the country's largest parties; at the latest parliamentary election, held in 1969, they had received together two thirds of the total vote. At the next election, however, which took place one year after the referendum, their strength was considerably reduced.

Ecological data show that their loss was particularly large in those areas where
opposition to the EC was strong. This appears most strikingly in the main fishery
districts of the North, as can be seen from Table 1.

The vote shares obtained by the various parties had been very stable all through
the nineteen-fifties and sixties, both in the North and elsewhere.

A sample survey including 2,320 respondents conducted by the Norwegian Bureau of Statistics in connection with the 1972 referendum throws light on the change in the country as a whole. The data make it possible to ascertain how many voters followed the cues given by party leaders, how many did not, and what were the main motives that impelled them to make one or the other choice. The sample was divided into two; one half was interviewed shortly before the 1972 referendum, the other half shortly after. The referendum vote of the first half was subsequently ascertained by mail. Respondents were also asked how they voted at the last parliamentary election, in 1969, and what parties they would have voted for had an election taken place at the time of the referendum.

The answers reveal significant movements both among supporters and opponents of the idea of Norway joining the Common Market. Table 2 shows how the parties taking a clear stand retained the allegiance of those adherents who shared their point of view (pro-EC Labourites and Conservatives, anti-EC Agrarian and Left Socialist voters). But a considerable number of those who disagred with their party would no longer vote for it. Only 54 per cent of anti-EC Labour voters and 35 per cent of anti-EC Cortservative voters declared their intention to vote for the same party as they did in 1969 had there been an election in 1972. Such was also the case with very few pro-EC Agrarians and Left Socialists.

This seems to corroborate a view expressed by many observers to the effect that the EC referendum in Norway was party-structured only to a limited degree. It belonged largely to the group-structured type. The campaign was no doubt influenced to a considerable extent by the cross-party, ad hoc 'People's Movement Against Norwegian EC Membership'. Founded in the autumn of 1970, two years before the referendum, and with links to the earlier anti-EC campaign of 1961 -62, it became a well led and most effective organization. This

was different from what happened in Britain, where only loosely structured groups of an improvised nature took part in the referendum campaign. The Norwegian cross-party organizational network covered no less than 404 out of the country's 445 communes (its smallest administrative units of some 10,000 inhabitant each, on the average). In 319 of these there was a local branch of the People's Movement; in the remaining 85 there was a local contact person. Strong centralization prevailed, propaganda material and slogans being produced and formulated in Oslo for distribution through the national network. Some ten per cent of all who voted against the EC were registered as members of the People's Movement (Bjørklund 1982).

Multivariate analysis reveals the presence of an unstructured type of referendum the anti-EC organization had a stronger impact on the outcome the region around the capital city, where the largest proportion of positive votes was found, than in the area registering the highest proportion of negative votes, North Norway. This is hardly surprising. Where a large number of people were inclined to look favourably on the idea of entry, there was need for an organization that could provide opponents with counter-arguments and moral support. Where opposition to the EC had great strength in itself, organized efforts were less necessary. They were even quite superfluous in certain places, where the reaction of voters seems to have been almost wholly unstructured. In one commune, the island of Røst in Lofoten, the People's Movement had no registered members. But that same commune in the country's main fishing district recorded the highest of all negative votes at the referendum, 93.5 per cent against entry.

The Norwegian referendum campaign influenced the relative strength of the different Norwegian parties, but there is no denying that some of the parties on their side were also active in the campaign and did exert a certain influence. The causal weight to be attributed to the parties is dependent, however, on the operationalization the concept of partisanship. The measure of party allegiance at the termination of the Norwegian referendum campaign in September 1972 employed in Pierce et al. (1983) is derived from a question about voting intentions at a hypothetical election. This makes a direct comparison with Britain possible, but it is a misleading comparison. In the Norwegian case the party preference of voters at the time of the referendum was markedly different from their preference at the preceding election.³ Party identification in September 1972 was influenced by the voters' stand on the EC issue (cf. Table 2, page 273). Such was not the case in Britain. The party preference indicated by respondents at the time of the June 1975 referendum was almost identical to the reported partisan vote at the British general election of October 1974 (Pierce et al. 1983, 46, 60). Very few British voters seem to have deserted their party because of its stand on the EC issue.

In this connection it seems of interest to compare, in an ecological study of

Norway, three different operationalizations. The first has reference to partisanship at the 1969 election three years prior to the referendum, the second to partisanship at the 1973 election one year after it, and the third to partisan choice at the nationwidelocal of 1975.

We shall consider the weight of partisan allegiance, as defined by these three operationalizations, in relation to two particular factors which many observers regarded as important in the Norwegian EC referendum. (1) Special weight was attributed to the occupational background of voters in the primary branch of the: economy and also to (2) ad hoc organizational activity.

We have operationalized anti-EC activity by considering the percentage of eligible voters registered as members of the ad hoc People's Movement in the various communes, while occupation is operationalized by the percentage of the population engaged in fishing, fish-processing and agriculture - occupational groups most strongly opposed to the idea of entry into the Common Market.

The independent variables are the distribution of the vote for the pro-EC parties (1969, 1973, 1975) along with anti-EC activity and occupation in fisheries and agriculture. The dependent variable is represented by the distribution the Yes vote in the referendum.

It will be seen from Table 3 that the 1973 results differ sharply from those of
1969, while in 1975 a movement in the direction of the pattern of the 1960's is
visible.

As expected, voting for the two pro-EC parties correlates with voting for entry. However, the predictive force of voters' preference for Labour and the Conservatives on referendum voting is not very significant if we define partisanship terms of votes cast at the last pre-referendum parliamentary election. A rather different picture appears when party allegiance is defined by votes cast four years later at the parliamentary election. This took place a year after the referendum in an atmosphere still strongly influenced by the 1972 struggle over the Common Market. While 'party 1969' is characterized by a low beta coefficient, the regression shows a higher coefficient in the case of'party 1973'. It is our contention that preference should be given to the former of the two definitions, 'party 1969', since an independent variable should precede the dependent in time.

In the regression analysis at the ecological level we are also in the happy situation of having all variables measured on the same scale. They represent the percentage of people in a commune who voted Yes, the percentage engaged in fisheries etc. It is therefore to some extent permissible to compare the relative importance of the variables by comparing the unstandardized regression coefficients. The picture that emerges is particularly revealing. In the regression where party reference is measured with votes cast in 1969, party is the least important variable, while the percentage of the population in fisheries is by far the most important, with ad hoc activity and percentage in agriculture as the second and third most important items.

Summing up the Norwegian experience, we would say that party influence was comparatively weak, whereas a not inconsiderable impact was made by the anti-EC People's Movement, which could boast more than one hundred thousand registered members in an electorate of barely two million. Besides this ad hoc group, a permanent interest organization also contributed to the structuring of the referendum, viz. the Farmers' League. However, the League was so closely associated the anti-EC Agrarian Party that farmers can be said to have reacted at one and the same time as group members and party identifiers.

The position of the fishing population was different. There never has been a national fishermen's party, and their permanent interest organization possessed no ties to a particular party comparable to the tie between farmers and the Agrarian Party. The fishermen did, however, have sufficiently close relations with the Labour Party to make it difficult for the leaders of their organization to oppose the party's stand in favour of entry into the EC. Only when at last the concessions demanded by the Community became known, did a break occur. The organization protested, and the Minister of Fisheries resigned from the Labour cabinet. But this was a development following upon a strong reaction among the rank and file. Organization leaders did not take the initiative. With regard to agriculture, the development was different. Elements within the organizational leadership displayed great activity and showed firmness of

purpose throughout. The farmers' associations represented the main financial
source of the People's Movement Against Membership.

With regard to the fishermen of North Norway, it can be said that the referendum belonged to the unstructured type. They formed an element in the population which needed no cues, either from parties or from ad hoc or permanent organizations. The announcement of certain 'new principles' by the ECprovoked immediate reaction among the fishing population. Fearing the depletion of the stock offish that mig;ht ensue if trawler fleets from EC countries got access to Norwegian territorial waters, people connected with the fishing and fish-processing industries turned decisively against membership. As noted above, there were overwhelming negative majorities in the area where fisheries constitute the most important source of livelihood. Media reports about foreign demands regarding fishing rights and the partial acceptance of such demands by Norwegian negotiators determined the attitude of these voters. News reporting sufficed. This kind of reaction seems best characterized by the designation unstructured'.

No doubt there was a certain number of Norwegian citizens whose referendum vote was determined not by organized or latent groups but by their party allegiance. It is no less interesting to observe, however, that in many cases it must have been voters' stand on the EC issue which determined, at least for a time, their choice of party.

Conclusion

A typology, it seems to us, can be useful even in the analysis of a concrete case that does not come close to any one specific 'ideal type'. The Danish referendums of 1969 and 1978 about the question of lowering the voting age can be said to exemplify quite clearly two different types. Latent population groups; played the most important role in 1969, while the impact of political parties was preponderant nine years later. So was the impact of parties in the British EC referendum of 1975. In contrast, the Norwegian 1972 consultation on entry into the EC showed the simultaneous effect of three distinct types of influence, to all of which must be ascribed some weight when the final outcome is to be: explained: partisan influence, ad hoc organizational influence, and latent group» influence. Party allegiance was not a negligible factor. But the advice given by a majority of Norwegian parliamentarians was taken only by a minority of voters:; consequently the partisan attachment of the latter cannot be regarded as the: primary force in their referendum voting. Partisanship should be operationalized in a way which elucidates this fact.

---

NOTES

1. According to A. Lijphart, with regard to the referendum no valid formulation of general propositions proved possible (Lijphart 1984,206). However, certain tentative generalizations have been proposed, for example by G. Smith (Smith 1976).

2. In addition to these four, two other parties were represented in the Norwegian parliament, the Liberals and the Christian Democrats. Their leadership, particularly that of the Liberals, was split on the EC issue.

3. A comparative analysis of the voters' attitudes in 1969 and 1972 has been performed by H. Valen (Valen 1976).

REFERENCES

Aitkin,D. 1978. 'Australia', pp. 123-137 in David Butler and Austin Ranney, eds., Referendums: A
Comparative Study of Practice and Theory. Washington, D.C.: American Enterprise Institute.

Björklund, T. 1982. Mot strømmen: Kampen mot EF 1961-1972. Oslo: Universitetsforlaget.

Lee, C. 1978. 'California', pp. 87-122 in Butler D. & Ranney, A., eds., Referendums, Washington,
D.C.: American Enterprise Institute.

Lijphart, A. 1984. Democracies. New Haven: Yale University Press.

Nielsen, H. 1970. 'Voting Age of 18 Years', Scandinavian Political Studies 5, 301-305.

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

Pierce, R., Valen, H. & Listhaug, O. 1983. 'Referendum Voting Behaviour: The Norwegian and
British Referenda on Membership in the European Community', A merican Journal of Political
Science 27, 43-63.

Smith, G. 1976. 'The Functional Properties of the Referendum'', European Journal ofPolitical Research
1-23.

Svensson, P. 1979. 'The Lowering of the Voting Age in Denmark: The Referendum of September
1978', Scandinavian Political Studies 2 (New Series), 65-72.

Svensson, P. 1984. 'Class, Party and Ideology: A Danish Case Study of Electoral Behaviour in
Referendums', Scandinavian Political Studies 7 (New Series), 175-196.

Tonsgaard, O. 1974. 'Party Voting and Referendum Voting', in Elklit, J. et al., The Danish EC
Referendum: Some Preliminary Results, Institute of Political Science, University of Aarhus.

Tonsgaard, O. 1984. 'En skitse til en teori om folkeafstemningsadfærd', in Jørgen Elklit & Ole Tonsgaard,
Valg og vælgeradfærd, 322-337. Aarhus: Forlaget Politica.

Valen, H. 1976. 'National Conflict Structure and Foreign Politics: The Impact of the EEC Issue
on Perceived Cleavages in Norwegian Polities', European Journal of Political Research 4, 47-82.

Wolfinger, R. E. & Greenstein, F. I. 1969. 'Comparing Political Regions: The Case of California'.
American Political Science Review 63, 74-85.