Resumé

In the 1994 EU referendum a majority of Norwegian voters rejected membership in the European Union. The outcome contrasts with victories for membership in the corresponding referendums in Austria, Finland and Sweden. The article reports a preliminary investigation of factors relating to the support for "no". Analysis of aggregate data (representing 435 municipalities) and survey data demonstrates the importance of traditional cleavages as well as new cleavages for the vote. The no-vote was strongest in northern Norway, among supporters of the traditional "counter-cultures" and among those employed in the primary sector. Women were more likely to vote no as were public sector employees. The major parties took relatively clear stands on the issue, and we find that party identification shows a strong correlation with the vote in the referendum. With a turnout of 89 percent, the 1994 referendum represents an all-time high for elections in Norway. Compared with the previous referendum on the EC in 1972, we find that the increase in mobilization was particularly strong in the no-dominated periphery, but the shift in mobilization patterns was not decisive for the outcome of the referendum.

Introduction

In the referendum of 28 November 1994, 52.2 percent of the Norwegian voters rejected membership in the European Union. This result closely mirrored the referendum of 25 September 1972, where 53.5 percent voted no to membership in the European Community. The Norwegian disapproval of EU membership followed referendums in Austria, Finland, and Sweden where majorities of the voters said "yes" to membership in the EU. Norway remains the only country which has declined membership in the European Union by a popular vote, and now even for a second time. The near identical outcomes of the two referendums, although spanning a period of 22 years, raise the question of how to account for the continuity of Norwegian Euroscepticism. Moreover, as the vote in Norway was scheduled after the decisions in Finland and Sweden, maximum pressure in a pro-EU direction

was enforced. This is in contras to 1972, when the referendum in Norway
was held before the vote in Denmark.

The outcome of the referendum can be understood by pursuing explanatory factors in three different directions. First, one may put the focus on the comparative aspect of the issue. Various comparative arguments have been raised. In a recent study Huseby & Listhaug (1995) have listed some of the factors that could be used to explain the negativity towards the EU in Norway. The geographical location of Norway on the northern periphery of Europe and the dominance of seaward and westward patterns of alliances, migration, and trade have isolated Norway from continental Europe. Norway won full sovereignty as late as 1905 and the strength of nationalism could be a force to be reckoned with. The comparative assessment must also be updated to cover the recent period. Following the decision to join the EU, Finland has dramatically shifted its political allegiance to the West. This was possible through the breakdown of the Soviet Union. For a concise analysis of the Finnish referendum see the chapters in the volume edited by Pesonen (1994). Similarly, the traditional non-alliance position of Sweden is at least weakly modified through membership. Like Finland, membership for Sweden might also be facilitated by the end of the cold war. Nevertheless, in pure foreign policy terms these nations took bolder steps than Norway would have needed to join the Union.

Second, one may search for explanations in the political and social structure within Norway. With reference to the Rokkan-Valen model of socio-economic cleavages, a case could be made to support the view that the prevailing structures of cleavage conflicts - old and new - even in the 1990s remain loaded against European integration.

Third, one can explain the outcome with reference to how the contending forces mobilized the electorate in the referendum campaign. This raises questions on the development and strength of the ad hoc organizations, the impact of the mass media for attitude formation, and the role of political parties. We are unable to pursue all these questions in this paper. Our main focus will be on the role of the cleavages for the vote in the referendum and on the effect of the mobilization of the periphery for the outcome.

The Cleavage Model

The Rokkan-Valen cleavage model is the most famous and most widely applied model of electoral behaviour in Norway. Less known is the fact that the impressions from the debate on EEC membership in 1962-63 inspired Rokkan's original contributions to the model (Rokkan & Valen 1964; Valen 1995). The cleavage model has already proven its relevance and fertility in

explaining the voting patterns in the EEC referendum of 1972 (Valen 1972;
Bjørklund 1993).

The Territorial Cleavage

According to the old saga, Norwegian medieval kings were not considered legitimate without the consent of the people in all parts of the country. The kings who neglected the regional oppositions were often short lived. The tension between the political centre and the peripheries is a central theme, perhaps the theme, in the political history of modern Norway.

Before 1814, the cities, and especially the site of the Danish governor in Norway, Oslo, were spearheads of Danish culture and strongholds of colonial administration. During the Danish rule of Norway (1536-1814) and the period under Sweden (1814-1905), the efforts to modernize Norway and to build national institutions were led by the shifting political elites of the Capital. In a sparsely populated country with a tradition of local autonomy such efforts were destined to provoke resistance, not least in the western and northern peripheries.

In the second half of the 19th century the central political elite was discredited and finally defeated during the struggle for national independence and democratization. The modern Norwegian nation is a product of the struggle for political independence and the core political institutions were formed during the campaign for political freedom. The constitution of 1814 was one of the most radical of its time, strongly inspired by the French and American Revolutions and the idea of popular sovereignty. Parliamentarism was introduced in 1884 as the result of the the effort by the Storting to control the cabinet appointed by the Swedish king. The first political party, The Liberal Party (Venstre), was formed as an alliance of radical townspeople, farmers, and other representatives from rural areas. The establishment of independent Norwegian political institutions eventually led to the collapse of the union with Sweden in 1905. Norway was declared an independent nation after a popular referendum. An overwhelming majority of 99 percent voted in favour of independence. In this period of political turbulence new social groups were politically mobilized through the extension of voting rights, making the processes of national independence and democratization strongly connected.

The Counter-culture Cleavages: Language, Lay Christendom and Teetotalism

The secularized and continental lifestyles of the urban elites created
resentment along several lines. The hostility was primarily related to three
issues: the role of Danish and later "riksmål" as the official administrative

language and the language taught in schools, the control over religious teaching and practice, and the moral evils related to the consumption of alcohol. Through mandatory schooling and a state church, the worldviews of the educated few were introduced to people in the social and geographical peripheries. Three distinct social movements countered this "penetration from the centre", using Rokkan's terminology. A new language, "nynorsk", based on the dialects from the western parts of the country, was launched as an alternative to the urban, semi-Danish language, a religious lay movement outside the state church claimed independence in spiritual matters, and, finally, teetotallers organized to guard moral purity, basic welfare and family life. The coastal areas in the south and southwest formed strongholds of the religious counter-culture. A ban on the sale of liquor and wine ratified by a referendum in 1919 was supported by large majorities of voters in the western and southern peripheries.

Rokkan and Valen portrayed the counter-culture cleavages as closely connected to the territorial cleavage. The counter-culture movements formed an important part of the Liberal Party. Tensions between the radical and urban faction and the morally conservative religious wing of the party eventually led to a split in 1888, but the party was reunited during the final battle over the union with Sweden in 1905, and the support of the "countercultures" was integrated into the party platform. However, the success was short lived. The power of the Liberal Party gradually faded as the different factions broke away to form the Christian People's Party and the Farmers' Party (later renamed the Centre Party).

The importance of the counter cultures as predictors of the vote has gradually declined in the post-war period (Valen 1981). In the previous referendum on EC membership in 1972, the counter-cultures demonstrated the continued strength of the old alliance between the peripheries, countercultures and urban radicals (Valen 1972).

The Commodity Market Cleavage

Norway became industrialized from the second half of the 19th century. The numbers in urban and industrial areas multiplied and, consequently, the importance of the market for food, increased. A shortage of manpower lead to a gradual modernization of farming and fisheries. The primary sector was integrated into the monetarized economy of the urban areas. After the severe market crisis in the 19205, both farmers and fishermen organized to control the harvesting, production and distribution of their products. Through the new Farmers' Party they also sought to secure stable prices by pushing for restrictions on agricultural imports. These efforts were opposed by the urban population, not least by the growing class of industrial workers.

The Labour Market Cleavage

In 1889 the Labour Party was founded by the political leaders of craftsmen and skilled workers, but rose to power through the support from unskilled industrial workers, smallholders and farm workers in the 1920s and 19305. The success of the Labour Party had not been possible without the support of the poorer segments of the rural population. However, Labour had difficulty in winning support in the rural areas in the southern and western parts of the country where the counter-cultures were dominant. The rural support for Labour came mainly from Trøndelag, Northern Norway, and the interior areas of eastern Norway.

New Cleavages?

Since the late 1970s various scholars have described the historical cleavages as fading in Norwegian party politics (Bjørklund & Hagtvet 1981; Valen 1981; Valen & Aardal 1983; Knutsen 1985; Jenssen 1993). Several attempts have been made to identify new emerging cleavages. We examine four new cleavages: education, gender, generation, and sector of occupation (Vogt 1993; Bjørklund 1994).

Education is a key factor in the discussion on new values and increased cognitive competence in the younger cohorts (Inglehart 1977, 1990; Dalton et al. 1984; Jenssen 1993). Inglehart claims that the younger and bettereducated generations are more cosmopolitan in their worldviews due to their postmaterialist values and socialization in a peaceful period. If this reasoning is correct, we expect the younger and better educated to support the idea of international cooperation, including the European Union.

A different view is stated by Galtung (1964). He sees the middle aged (30-60 years) to be at the peak of their occupational careers as well as other hierarchies in society. Consequently, they will tend to support policies advocated by the elites. Younger people and the very old belong to the "social periphery" in the modern society, and are more likely to hold elitechallenging views. Seeing the EEC, and later the EU, as primarily a cooperation of elites, he predicts a curvilinear relationship between age and support for European integration.

About twenty years ago Martinussen (1975) described gender as a latent cleavage in Norwegian politics. He conceived a process of political mobilization of women to improve their status in society. In the 1970s and earlier, women were slightly more conservative in their voting patterns than men (Bjørklund 1986). This balance shifted in the opposite direction in the 1980s (Listhaug, Miller & Valen 1985; Aardal & Valen 1989; Valen, Aardal & Vogt 1990; Bratterud 1994; Listhaug, Huseby & Matland 1995).

Lafferty (1988) introduced the concept of a "public sector class" to describe the fact that public employees have an interest in high levels of public spending. On some issues, like support for welfare, public employees have been described as more to the left than the working class in the private sector (Knutsen 1986). Public employees are also more likely to vote for the Socialist Left Party than unskilled workers in the private sector (Hines 1993).

Of the new cleavages, especially gender and sector were politicized in the campaign. The no-movement argued that the EU was a threat to public sector interests and the welfare state, and - partly as consequence of this - of disadvantage to women as well.

Stability and Change Between the Two Referendums: 1972 vs. 1994

The structural changes of Norwegian society 1972-94 weakened some of the main forces of the traditional cleavage model which created resistance to the EC in 1972. Norway became more urbanized. Most notably, the population increased in the Oslofjord area - the only main region with a majority of yesvoters in 1972. While 12 percent of the population was employed in the primary sector in 1972, only 6 percent had its income from agriculture or fishing twenty years later. The proportion employed by the service sector increased from 54 percent to 68 percent. Norway has traditionally been a major exporter of raw materials and industrial products. Through the development of major oilfields in the 1970s and 1980s the country became even more dependent on international trade and access to foreign markets.

In the 1970s Norway experienced a major expansion of education at all levels. Nine years of compulsory education was introduced. The student population of the universities expanded and each of the 19 counties established regional colleges to teach courses and degrees at the undergraduate level. In 1972, 31 percent of the population had education beyond the compulsory level; in 1992 this figure had increased to 69 percent. The proportion with a university degree more than doubled, from 7 percent to 18 percent.

The modernization of Norwegian economy as well as the decline of the rurally based counter-cultures readily point to the prediction of victory for yes in 1994. This prediction could be premature if some of the new cleavages were not considered. The continued expansion of the public sector, including a dominant proportion of female employment, could constitute a new structural basis for anti-EU attitudes.

Finally, the impact of the periphery obviously depends on how strongly the voters of the periphery were mobilized in the campaign. In assessing the importance of the mobilization of the periphery for the no-victory in 1994 one should look for the strength of the no-vote as well as the relative turnout

in 1994. Or, in other terms, did an extraordinary mobilization of the
periphery compensate for the decline in numerical strength?

Centre and Periphery; One Concept, Many Issues: Analysis of Aggregate Data

We begin the discussion of the role of the periphery by analysing aggregate data for all 435 communes. The commune, or municipality, is the basic administrative unit in Norway. The Commune Database of the Norwegian Social Science Data Services contains a very large number of variables with information on the communes. The database includes a number of indicators that can be used to tap the centre-periphery dimension. Here we consider the impact of municipality size, population density, geographical location, average income, and the proportion employed in the primary sector.

Turnout: A Mobilization of the Periphery in 1994?

The distribution of turnout levels at the 1972 referendum confirmed a traditional pattern for Norwegian elections. Mobilization was highest in the small but densely populated Oslofjord area and the valleys of the interior east. As we move west and north the participation rates tend to decline, and the northern region in particular had few municipalities that could match the mobilization of the southern regions.

The mobilization pattern in the municipalities changed sharply from 1972 to 1994. The participation increased from 79 percent in 1972 to 89 percent in 1994 (Table I). As can be seen from Table I the average participation level in 1994 exceeded the maximum turnout in 1972. The lowest turnout increased by 22 percentage points and the highest turnout by 7 percentage points. More significantly, a major part of the increased mobilization came in the peripheries of western and northern Norway.

Map 1 illustrates that very few municipalities in the Oslofjord area and the interior east of Norway increased the turnout by a substantial margin. An increase in the magnitude of 10 percent and more is rather common in the southern, western and northern regions. The variation in turnout change

suggests that this factor might have contributed to the outcome of referendum in 1994 as the increase in mobilization was strongest in areas where the no-vote was strong. To achieve a fuller understanding we take a look at the geographical distribution of the vote in the two referendums.

The Geographical Distribution of the Vote: Nearly Identical Patterns

When we compare the distribution of strength for no, the similarity between 1972 and 1994 is impressive (see Maps 2 and 3). For both years the strength of the supporters of membership is concentrated around the Oslofjord and in the region north of Oslo, although the strength of yes in the latter area declined in 1994. Outside this region only a few municipalities with major industries dependent on international markets, or major cites, showed majorities for yes. The near identical geographical patterns of the two referendums demonstrate the profound effects of the centre-periphery conflict on the issue. We now move on to demonstrate more directly the relationship between particular aspects of the centre-periphery dimension and the vote.

Local Government: The Swiss Argument

Switzerland has in general resisted membership in supra-national organizations, the EU included. The high degree of decentralization of Swiss government contributes to the scepticism towards European integration. Decentralization is also a strong feature of the Norwegian polity. A plausible hypothesis is that the fear of losing democratic control to the EU would be stronger in the smaller municipalities. This is also what we find. An increase in population size of 1000 reduced the percentage voting no in the commune by about 0.2 percent in both years (see Table II).

This does not seem to be much of an effect, but we have to consider that
Norwegian municipalities vary in size from a few hundred inhabitants to
nearly 500,000 (Oslo).

However, properties other than size are also of importance. Table 111 shows the effect of population density assessed on a 10-point scale and degree of geographical centrality measured on a 7-point scale. For both years there is clear evidence that the degree of urbanization and centrality of the municipality has a major effect on the strength of opposition to membership in the EU. In 1994 the two scales explain as much as 61 percent of the variance of the no-votes in the communes. The impact of geographical centrality is especially strong. When we move one point on a 7-point scale from the most peripheral municipalities to those close to a major city, the proportion rejecting membership declines by more than 12.5 percent (Table III).

Economic Development

We use two indicators of economic development, average income of
taxpayers and the proportion employed in the primary sector. The interest

organizations of the primary sector argued very strongly that membership in the EU would weaken Norwegian agriculture and fisheries. The organizations argued that the small farms of Norway with a relatively short growing season would not be able to compete in an open European market. In contrast to farming, fisheries needed access to markets, here the argument against membership primarily was one to keep control of resources. Table IV shows that in both referendums the size of the primary sector was positively correlated with the strength of the no-vote, while the affluence level of the municipalities tapped by the average income of taxpayers, was negatively correlated with support for "no".

The aggregate analysis has demonstrated that resistance to Norwegian membership was strongest in the communes of the periphery. The statistical relationships are remarkably similar in 1972 and 1994 - indicating strong patterns of continuity. On balance, the associations between centre-periphery factors and the vote are even a bit more potent in 1994 than in 1972.

Continuity in Referendum Voting 1972-94

The continuity of referendum voting can also be observed directly by looking at the correlation between the votes in the two years. Table V shows that the correlation between the votes in 1972 and 1994 is 0.88. The impact of party is also evident. To capture the party effect we have constructed a variable that sums the vote for the Centre Party, the Socialist Left Party, the Christian People's Party, and minor parties that were against membership. In

1972 we find that the correlation between the strength of the no-parties at the previous parliamentary election in 1969 and the vote in the referendum was 0.46. In the 1994 referendum the correlation is 0.51, 0.57, and 0.86 with the strength of the no-parties at the elections of 1985, 1989, and 1993, respectively. The strong impact of the vote in 1993 on the outcome of the referendum indicates that much of the opposition to Norwegian membership was already in line at the Storting elections one year prior to the referendum, where the strong showing of the Centre Party in particular must be accounted for by the forceful stand of the party leadership against membership.

The Impact of Mobilization

We began the analysis of aggregate data by noting the much stronger mobilization in the 1994 referendum than in the referendum of 1972, and that the increase in turnout was much higher in regions that were weakly mobilized in 1972. Furthermore, the strong continuity in geographical patterns of the vote raises the possibility that the increased mobilization was crucial for the narrow victory of no in 1994. In the final part of the section we make a first attempt to demonstrate this with empirical evidence. In Table VI we take a look at the relationship between the proportion voting no in the referendums and turnout levels, and, for 1994, the change in turnout from 1972 to 1994. The bivariate correlation of vote and turnout is weak for both years, but with a negative sign in 1972 (-0.07) and a positive sign in 1994 (0.05). More interestingly, we find that a turnout increase from 1972 to 1994 yields a positive impact on the proportion voting no. The impact is tenuous, with a regression coefficient of 0.92 in the controlled model. One caveat that we must have in mind derives from the fact that the municipalities vary greatly in size, which makes it impossible to assess the direct impact on the outcome of the referendum from the aggregate units without including information on the voting population.

To evaluate the impact of the increase in turnout we have run a series of simulations where we have moved backwards. More precisely, we have estimated three hypothetical outcomes of the 1994 referendum. In the first we reduce the difference in turnout between 1972 and 1994 in the municipalities by 10 percent, in the second by 50 percent, and in the third by 100 percent. In the third simulation the turnout is identical to the actual turnout in 1972. The results are evaluated against the actual results in 1994 (see Table VII). The gradual reduction in mobilization leads to a reduction in the proportion voting no. We show the results with municipalities as units, unweighted and weighted by the number of votes in each municipality. The weighted results are the most interesting. They show a consistent - but weak - decline in the percentage voting no when participation levels are reduced. But a replay of the referendum of 1994 using the turnout of 1972 still would have kept Norway outside the EU, but with a majority of just above 51 percent.

The aggregate analysis has demonstrated patterns of continuity in the impact of the centre-periphery factors in the two EU referendums. The periphery was as strongly opposed to membership in 1994 as in 1972. While the increase in mobilization levels slightly favoured the no-side, this advantage was so tiny that the traditional resistance would not have been strong enough to secure a no-majority when we consider the structural changes that reduced the numbers in the periphery from 1972 to 1994. Other factors must be considered. For a broader assessment we turn to survey data,

which allow for an analysis of new cleavages in addition to the traditional
structures, and to an investigation of the role of the party in the campaign.

Cleavages and the EU Vote: An Analysis of Survey Data

Is the impressive macrolevel stability from 1972 to 1994 demonstrated above
simply a consequence of the tendency for microlevel "noise" to disappear in
aggregated data, or is stability at the mass level observed as well?

Respondents in the 1994 survey who were old enough to have voted in 1972 were asked to give their preference in that referendum. When we compare their votes in 1972 and 1994, the impression is one of stability (see Table VIII).

Eighty percent of the voters taking part in both referendums, stuck to the same side. We cannot know for sure if some of them have wandered back and forth during the 22-year period. More likely, many have not given the EU issue much thought in this period, so the most frequent type of attitude shift in this group is probably between attitude and non-attitude.

The stability seems to be somewhat higher on the yes-side, but this tendency may partly reflect distorted memory on the part of some voters. Compared to the outcome of the 1972 referendum (53.5 percent no), novoting in 1972 is substantially over-reported. As a consequence, the estimated shift from no to yes may be inflated.

The 1994 survey was a three-wave panel. The respondents were interviewed before the final campaign started in August 1994, during the campaign, and, finally, after the election. In the first two interviews they were asked about their voting intention. In the third interview, we registered their actual vote. As can be seen from Table IX, the attitude stability throughout this period is striking.

Among those who stated a preference on the issue before the referendum (on the average about 11 weeks before voting), 95 percent did not change from no to yes or from yes to no during the campaign. Among the remaining 5 percent, less than 0.5 percent changed their attitude twice. The exclusion of

those without an opinion in Table IX overstates attitude stability. However, these findings suggest that models that predict stability must be of relevance in explaining opinion formation in the EU issue. The cleavage model, as outlined above, clearly is directed at explaining long-term stability in political behaviour; and it is to the question of the role of cleavages for the referendum vote of 1994 that we now turn.

The respondents are divided into two groups of equal size on the dependent variable. The estimates of OLS regression in this case will not be very different from those obtained with logit or probit techniques. To check the possibility of unreliable T-tests (Aldrich & Nelson 1984), the three models of Table X were estimated by logit regression. The results did not significantly deviate from those reported in the table.

Since the dependent variable is coded as voted yes = 0, and voted no = 1, and the independent variables are all dummies, the unstandardized regression coefficients can be interpreted directly as percentage changes in no-vote. For instance, a B-coefficient of 0.09 for members of religious organizations in model 2 means that support for the no-alternative was 9 percent higher among members of religious organizations than among nonmembers, all other factors (in the model) being equal. The constant in the equations is the percentage voting no in the reference group. Note that the reference group is not identical in the different models.

The territorial cleavage is represented by six dummy variables in model 1 in Table X. The first variable differentiates between voters living in cities (municipalities with official status as cities) and voters living in rural areas. The other five dummies represent the regions. The Oslofjord area serves as the reference category. The results in Model 1 (Table X) mirror the geographical pattern displayed in Map 3. The opposition to EU membership is strongest in northern and western parts of the country and in the rural areas. The opposition is weaker in towns and close to the political centre. All

coefficients in the model are statistically significant, but the model explains
only 8 percent of the variance.

Since the counter-cultures have their strongholds in the western and southern peripheries, there are substantial correlations between the three variables representing the counter-cultures and at least two of the variables representing the territorial cleavage. To avoid underestimating any of the theoretical variables, the effects of the counter-culture cleavages and the two economic cleavages are represented separately in Model 2 in Table X. The results can be interpreted straightforwardly: the opposition to EU membership is stronger among people belonging to the counter-cultures than in the rest of the population. The effect of being attached to the primary sector either through occupation, former occupation (if retired) or through spouse's occupation (if the respondent had never been employed) is very strong. There is no significant effect of being employed as a skilled or unskilled worker. The traditional working class did not constitute an important part of the resistance to EU membership in 1994. This is surprising since there is a significant correlation between low occupational prestige and no-vote in the referendum and between low income and no-vote.

The full cleavage structure is represented by Model 3. Compared with Model 1, the effects of living in the southern and western peripheries are somewhat reduced when they are controlled for involvement in the countercultures. Not surprisingly, we observe that the effect of attachment to the primary sector is slightly reduced compared with the effect in Model 2. 14 percent of the variance in voting is explained by this model.

The inclusion of gender, generation, education and sector of occupation improves the explanatory power of the model somewhat (R² = 0.17). 11 percent more women than men voted no, 10 percent more no-votes were cast by voters employed in the public sector compared with voters in the private sector (farmers and fishermen excluded), people in the youngest generation were significantly more likely to vote no than older people, and the yes-vote increased with level of education.

The Arguments

We have shown that the location of the individual in the cleavage structure is important for the vote in the referendum. Do the arguments that the voters give for their choice also reflect these cleavages? The two sides sought to mobilize quite a large number of issues to support their views. Hence, the arguments that the respondents in the survey report for their choice cover a lot of ground. On the basis of the answers to the open-ended questions in the survey we have coded the responses into the categories that we report in Figs. 1 and 2.

Figure 1 shows that national independence and democracy were the most
frequent arguments on the no-side. The argument sounds almost like an echo
from 1905 and 1972.

Concern for the interests of the fisheries, farmers and the rural population in general is also among the most prevalent arguments for those who voted no to membership. Some voters also mentioned the classical counter-culture concerns: alcohol, the future of the nynorsk language and fear of secularization or Catholicism. However, arguments related to the classical cleavages are present in the other categories as well. For instance, the arguments labelled "Criticism of the European Union" are often based on the postulate that the nation state is threatened by the EU. The economic arguments often reflect a wish to maintain political control (meaning control by Norwegian political authorities) over big business.

Remains of historical cleavages are harder to detect in the yes-arguments, but they are even present here. First of all, with a few, but interesting exceptions, all politicians on the yes-side argued that EU membership would be economically favourable. This position, which is also reflected at the mass level as the most prevalent reason given for voting yes, is related to the basic free-market nature of the EU. Only a small minority of the yes-voters favour the federal idea of "The United States of Europe". The majority takes Norway's national interests as its point of reference. In other words, the

presence of a definite national identity is not confined to the no-side. In general, the EU is seen as an arena for international cooperation, not an institution which is in conflict with Norwegian institutions. Again, the views of the elite are congruent with mass opinion.

Political Parties as Opinion Leaders

A recurrent theme in the study of referendums is the conflict between direct and representative democracy (Butler & Ranney 1994). In representative democracy, political parties have a major function. Some scholars argue that parties are sidelined in referendums as voters will not follow normal partisan loyalties (Kobach 1994, 132-134). Other scholars tend to give parties a more decisive role in shaping the referendum vote decisions of citizens (Pierce, Valen & Listhaug 1983). The main political parties in 1994 were the Labour Party and the Conservative Party on the yes-side, and the (Agrarian) Centre Party and the Left Socialist Party on the no-side. A vocal minority in the governing Labour Party went against the party line and organized a campaign, "Social democrats against EU", that played an important role to move Labour identifiers to vote no (Jenssen & Valen (eds.) 1995, ch. 1).

The Christian People's Party opposed EU membership but supported the
EEA agreement. Hence, the party leadership kept at some distance from the

Centre Party and the Socialist Left Party, as these parties voted against the EEA as well as the EU. The two minor parties, the Liberals and the rightwing Progressive Party, were divided on the issue, and the party leadership chose a rather low profile throughout most of the campaign. However, in the final stage of the campaign, the Progressives came out for yes, and the Liberals for no.

Some voters obviously came to experience cross-pressure between party loyalty and cleavage group loyalty (Jenssen & Valen 1995 (eds.), ch. 5). A detailed examination of cross-pressure conflicts goes well beyond the task that we have set out to accomplish in the present paper. By adding party identification to the cleavage model (see Table XI) we hope to grasp some of

the importance of parties for referendum vote. Since the party structure partly reflects the cleavage structure already in the model, we expect cleavages to loose some impact. The inclusion of party identification increases the proportion of variance explained from 17 to 29 percent. Some of the cleavage variables drop below significance levels while other effects are weakened.

According to the estimates of Model 5 (Table XI), the no-vote is 36 percent lower among those who identify with the Conservatives than among independents. The effect of identification with Labour is -25 percent. The effect of identification with the Progressive Party is -15 percent, but the coefficient is not statistically significant. Among those who identify with the Centre Party and the Socialist Left Party a strong effect in favour of no is observed. In sum, party identification works as expected and proves to be a strong and consistent predictor of the referendum vote. The specific dynamics of partisan factors in the referendum campaign remains to be analysed, hence firm conclusions should be postponed until subsequent publications.

Conclusion

Our analysis of aggregate data (representing 435 municipalities) and survey data demonstrates the importance of traditional cleavages as well as new cleavages for the vote in the referendum. The no-vote was strongest in northern Norway, among supporters of the traditional "counter-cultures" and among those employed in the primary sector. Women were more likely to vote no as were public sector employees. The major parties took relatively clear stands on the issue, and we find that party identification shows a strong correlation with the vote in the referendum. With a turnout of 89 percent, the 1994 referendum represents an all-time high for elections in Norway. Compared with the previous referendum on the EC in 1972 we find that the increase in mobilization was especially strong in the no-dominated periphery. However, the overall finding of this preliminary analysis is one of stability, both on the macrolevel and among individuals. In order fully to evaluate this impressive stability, we must bear in mind that close on one million new voters entered into the electorate after 1972. Our empirical results indicate that the cleavages that emerged in the 19th century still make a difference in Norwegian politics. The EU issue in 1994 - as in 1972 - revitalized these otherwise declining cleavages.

ACKNOWLEDGEMENTS

An earlier version was presented as a paper at the annual meeting of the Midwest Political
Science Association, The Palmer House Hotel, Chicago, April 6-8 1995. The interview data
were collected by the survey division of Statistics Norway. The survey was funded by a

generous grant from the Ministry of Local Government and Labour. Data on Norwegian
municipalities are from the Commune Database of the Norwegian Social Science Data
Services. We appreciate the comments and help of Arne Kolstad and Kathrine L. Moen.

REFERENCES

Aardal & Valen, H. 1989.

Aldrich, J. H. & Nelson, F. D. 1984. Linear Probability, Logit and Probit Models. Beverly
Hills: Sage.

Bjørklund, T. 1986. "Kvinners og menns partipreferanse", Tidsskrift for samfunnsforskning
27, 417-443.

Bjørklund, T. 1993. "Sentrum mot periferi: Norsk EU-strid belyst med ulike sentrum-periferi modeller", in Asbjørnsen, N. & Vogt, G., eds., EF, Norge og 1990-årene. Seks artikler om EF-sakens betydning i norsk politikk. Research report 2193. Oslo, Department of Political Science, University of Oslo.

Bjørklund, T. 1994. "Norsk EU-opinion: Velgere og ledere i utakt", in Bingen, J. & Lindahl,
R., eds., Nordisk skjebnevalg. Oslo: Europaprogrammet.

Bjørklund, T. & Hagtvet, B. 1981. "De bleknede skillelinjer - Arbeiderkultur og motkultur
som bølgebryter mot Høyre", in Bjørklund, T. & Hagtvet, B. eds., Høyrebølgen-epokeskifte
i norsk pplitikk. Oslo: Aschehoug.

Bratterud, Ä. 1994. "Blir kvinnene velferdsstatens viktigste støttespillere?", in Jenssen, A. T.
& Martinussen, W., eds., Velferdsstaten i vare hjerter. Oslo: Ad Notam Gyldendal.

Butler, D. & Ranney, A., eds. 1994. Referendums Around the World. Washington, DC: The
AEI Press.

Dalton, R. J. et al., eds. 1984. Electoral Change in Advanced Industrial Democracies:
Realignment or Dealignment? Princeton: Princeton University Press.

Galtung, J. 1964. "Foreign Policy Opinion as a Function of Social Position", Journal of Peace
Research 1, 206-231.

Hines, K. 1993. "SVs velgeroppslutning - sektor, generasjons- eller kjønnsbestemt?", Norsk
Statsvitenskapelig Tidsskrift 9, 171-188.

Huseby, B. & Listhaug, O. 1995. "Identifications of Norwegians with Europe: The Impact of
Values and Centre-Periphery Factors", Values in Western Societies. Tilburg: Tilburg
University Press.

Inglehart, R. 1977. The Silent Revolution. Princeton: Princeton University Press.

Jenssen, A. T. 1993. Verdivalg: Ny massepolitikk i Norge. Oslo: Ad Notam Gyldedal.

Jenssen, A. T. & Valen, H., eds. 1995. Brussel midt imot: Folkeavstemningen om EU. Oslo: Ad
Notam Gyldedal.

Knutsen, O. 1985. Politiske verdier, konfliktlinjer og ideologi: Den norske politiske kulturen i
komparativt perspektiv. Oslo, Department of Political Science, University of Oslo.

Knutsen, O. 1986. "Offentlige ansatte: Mulige årsaker til partipolitisk særpreg", Norsk
Statsvitenskapelig Tidsskrift 2, 21^44.

Kobach, K. W. 1994. "Switzerland", in Butler, D. & Ranney, A., eds., Referendums Around
the World. Washington, DC: The AEI Press.

Lafferty, W. M. 1988. "Den offentlige-sektorklassen: Syndebukk for privatisme eller arvtaker
til sosialisme?", in Bogen, H. & Langeland, 0., eds, Offentlig eller privat? FAFO-report
078. Oslo: FAFO.

Listhaug, 0., Huseby, B. & Matland, R. 1995. "Valgadferd blant kvinner og menn
1957-1993", in Raaum, N., ed., Kvinner i norsk politikk. Oslo: TANO.

Listhaug, 0., Miller, A. & Valen, H. 1985. "The Gender Gap in Norwegian Voting
Behaviour", Scandinavian Political Studies 8, 187-206.

Martinussen, W. 1975. "Politiske skillelinjer og politisk deltakelse", in Ramsøy, N. R. & Vaa,
M. eds., Det norske samfunnet, Vol. 2. Oslo: Gyldendal.

Pesonen, P., ed. 1994. Suomen EU-kansaäanestys 1994. Helsinki, Painatuskeskus Oy.

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

Rokkan, S. 1967. "Geography, Religion and Social Class. Crosscutting Cleavages in
Norwegian Politics", in Seymour M., Lipset, S. M. & Rokkan, S., eds., Party Systems and
Voter Alignments. New York: Free Press.

Rokkan, S. 1970. Citizens, Elections, Parties: Approaches to the Comparative Study of the
Processes of Development. Oslo: Universitetsforlaget.

Rokkan, S. & Valen, H. 1964. "Regional Contrasts in Norwegian Politics", in Allardt, E. &
Littunen, Y., eds., Cleavages, Ideologies and Party Systems. Helsinki, Westermarck Society.

Valen, H. 1972. "Norway: 'No' to EEC", candinavian Political Studies 8, 214-226.

Valen, H. 1995. "EU-saken i norsk politikk", in Aardal, B. & Valen, H., Konflikt og opinion.
Oslo: NKS-forlaget.

Valen, H. & Aardal, B. 1983. Et valg i perspektiv. En studie av stortingsvalget 1981. Sosiale
og økonomiske studier 54. Oslo: Statistisk sentralbyrå.

Valen H., Aardal, B. & Vogt, G. 1990. Endring og kontinuitet: Stortingsvalget 1989. Sosiale
og økonomiske studier 74. Oslo: Statistisk sentralbyrå.

Vogt, G. 1993. "Norsk EU-opinion 1972-1993: Gamle og nye mønstre", in Asbjørnsen, N. &
Vogt, G., eds., EF, Norge og 1990-årene: Seks artikler om EF-sakens betydning i norsk
politikk. Research report 2/93. Oslo, Department of Political Science, University of Oslo.