Resumé

In a democracy, political decisions ought to be based on public opinion. In practice, however, the mechanisms connecting voter preferences and public policy are complex, and it appears that public opinion may be partly policy-shaping and partly shaped by policy. In this article, some of these mechanisms are discussed. The article presents, first, three models of public reactions to policy decisions. These models are then applied in an analysis of the liberal trends in Norwegian alcohol policy as well as attitudes towards this policy. The results are consistent with a "consumer model", where citizens evaluate public policy according to their preferences, as well as a "support model", where they tend to follow and support decisions made by political leaders. A "discontent model", where implementation brings about less acceptance of a policy, is not supported by the data.

Introduction

According to Stein Rokkan (1966, 78), "foreign observers have often found it difficult to understand the importance of alcohol as an issue in Nordic politics". Alcohol policy has been the subject of one of the main controversies in the history of Norwegian mass politics. As one of the "counter-cultures", the teetotallist movement of the 19th century defended rural values against urban standards and practices (Rokkan 1966, 1967; Valen & Rokkan 1974; Valen 1981). From the 1870s up to World War I, this movement grew rapidly. The alcohol conflict reached its zenith in the period of prohibition (1916—27). In 1919, prohibition was supported by a majority in a referendum, but seven years later the majority voted against it in a new referendum. Since then, the alcohol policy debate seems to have been more pragmatic (Nordlund 1993). Still, the teetotallist movement exercises considerable influence on public policy. Heavy taxation and a number of local and national restrictions on the sale of alcohol have made Norway's alcohol policy one of the most restrictive in Europe. However, a trend towards liberalization in attitudes towards alcohol has taken place during recent decades, and support for teetotalism has declined (Valen 1992, 157). Moreover, the common conception of "alcohol abuse" has changed;

drinking habits which were regarded as "abuse" in the 1960s are now, to an increasing extent, accepted (Arner 1993). The trend towards liberalization is also reflected in public policy. Since 1945, several local restrictions on the sale and serving of alcohol have been lifted.

Public opinion is certainly affected by several factors, ranging from longterm changes in social structure and deep-rooted values to the activities of interest groups or media coverage of current events. However, a complete discussion of all factors influencing attitudes to alcohol policy is beyond the scope of this article. The aim is narrower: to explore how the preferences of citizens can be conditioned by the political process itself. With this perspective, the dynamics of political decisions and public opinion is brought into focus.

In the first section, I discuss possible mechanisms linking policy decisions and attitude changes. The aim of this section is to describe models of public reactions to policy decisions, based on the literature on public opinion and values, which may be useful tools for untangling the links between policy and opinion. Secondly, I apply these models in an empirical analysis of trends in Norwegian alcohol policy and opinion since the 19605.

Connecting Opinion and Policy

The concept of "public opinion" is somewhat problematic: "To speak with precision of public opinion is a task not unlike coming to grips with the Holy Ghost", according to V. O. Key (1961, 8). Citizens may lack knowledge of, or interest in, policy issues. Moreover, the wording of questions in public opinion surveys affects the answers. Therefore, statements claiming that a specific policy is supported or rejected by public opinion, or "the will of the people", will seldom be meaningful. Attitudes, towards specific issues may rather be regarded as indicators reflecting changes in the ideological climate, or "policy mood" (Stimson 1991).

If public opinion moves in the same direction as public policy, the
connection may be explained in three ways (Weissberg 1976; Papadakis
1992):

1. Public opinion affects policy decisions. When there is disagreement between elected policy-makers and voters, the politicians may be replaced at the next election - or they may change their policies in order to get re-elected (Stimson et al. 1995).

2. Voters and policy-makers respond to social changes in the same way.
Particular events may move both policy and public opinion, without any
causal connection between opinion and policy.

3. Policy decisions influence public opinion.

If public opinion directs public policy, opinion and policy will move in the same direction. If policy influences opinion, the direction of this impact may be ambiguous. Citizens may respond to policy changes with protest as well as support. When implementation of a policy causes protest, the trends of opinion and policy may diverge. In public opinion research, different reactions to policy decisions are described. These perspectives will be summarized by means of three models, which I here call the "consumer model", the "discontent model" and the "support model". Subsequently, conditions which may affect the relevance of the models are discussed. Finally, the measurement of these processes by different kinds of survey questions is considered.

The Consumer Model

The logic of the consumer model is, as the name implies, borrowed from the economic theory of consumption behaviour. In this theory, preferences are exogenously determined: they are taken for granted. Total utility, gained from the consumption of all units of some commodity, is distinguished from the marginal utility resulting from the consumption of one more unit of the commodity (see e.g. Lipsey et al. 1990, 151-155). The total utility of a consumer will usually rise when his consumption increases. However, according to "the law of diminishing marginal utility", the utility increase derived from consuming an additional unit of a commodity will fall as the consumption of that commodity increases.

If demand for consumer goods is replaced with demand for political decisions, this logic can be transferred to research on attitude or value change.¹ Political preferences reflect a comparison of the total utility attached to each alternative policy output or end-state. These preferences are exogenous: they are not supposed to change as a result of shifting policies. When a political programme is implemented, the marginal utility of this policy - and thus the demand for even more of it - will diminish. While attitudes towards different policy outputs are stable, attitudes towards the direction of public policy will change.

This kind of explanation is not unusual in research on attitudes and values. For instance, the dilemma of Socialist and Social Democrat parties has been described in this way. Their objectives with regard to equalization have to some extent been realized, while the foundation of their appeal and political attraction has weakened (Valen 1981, 144). Inglehart (1990, 248-257) has employed the law of diminishing marginal utility to account for this development: the demand for a continued equalization policy will be reduced, when economic inequality in a society is reduced and the standard of living increases. The point is not that the leftist economic policy has been a failure. Inglehart maintains that equalization policy has been a

success; and, therefore, more of it is unnecessary.² Diminishing marginal utility can, in other words, explain demands for new priorities or changes in public policy. However, this model describes no fundamental value change. Equalization may weaken the demand for more equality, but opinion on the optimal balance between equality and freedom in a society is unchanged.

In accordance with the consumer model, comparative studies of values in Europe show that support for a leftist economic policy is relatively weak in the wealthiest countries (Inglehart 1990, 253-257; Knutsen 1995). However, Listhaug (1990a) found only a weak connection between leftism and the wealth of European countries. The difference is probably caused by different question formats -a subject to which I return later.³ Still, Listhaug supports one of Inglehart's conclusions: even though the Nordic countries have been, for the most part, governed by Social Democrats in the post-war era, this has not caused a lasting support for leftist values.⁴

The diminishing marginal utility may explain development over time as well as cross-national differences. In Norway and Britain, opposition against governmental regulations of the economy increased during the 19705. New policies were implemented by Conservative or centre-right governments in the 1980s, and the demand for more of this kind of policy declined to some extent (Listhaug 1990b; Heath et al. 1991, 171-185). A similar development has taken place in American public opinion. Stimson (1991) describes a cyclical movement, where policy and opinion fluctuate around an equilibrium. For example, voters may ask for a policy based on more conservative values. They will elect new leaders who implement this kind of policy, or the incumbent representatives will shift their policies in a new direction. Eventually, the voters will think that the new policy has been carried too far, demand a more liberal policy - and so on.

The Support and Discontent Models

The assumption of exogenous preferences is useful, but unrealistic - in economics (Haavelmo 1993, 123-162) as well as political science. Endogenous preferences, influenced by policy implementation, are the foundation of the "discontent model" and the "support model". In the discontent model, implementation causes less acceptance of a policy, while increased support is the result in the support model. In addition to changes in marginal utility, political decisions will affect the evaluation of alternative policy outputs, or total utility.

According to the discontent model, a policy is more attractive in theory than in practice. Implementation may reveal problems, and people may be dissatisfied when they see the consequences. For example, Listhaug (1990a, 222) adds another interpretation to Inglehart's account of cross-national

differences in support of equalization. Implementation of egalitarian policies may not only satisfy people and weaken demand for more equality. The result may be disappointment, and people will demand a return to the previous policy. Introduction of a market economy may also create unexpected side-effects, and recent developments in Eastern Europe are probably an example of the processes described in the discontent model. On the other hand, the consequences of a decision may be more favourable than expected, and acceptance of a new policy will grow. In addition, several psychological and sociological mechanisms work in favour of increased support. These mechanisms are the basis of the support model.

According to the consumer model, individual preferences are aggregated into collective outcomes through political decisions. In reality, decisionmakers - parties and politicians - are also shaping preferences. The public is exposed to a stream of persuasive messages. Moreover, people will seldom make extensive calculations of costs and benefits when they work out their positions on political issues. The concept of "bounded rationality" describes our decision-making in a more realistic manner. It maintains that since our information and capacity to process information is limited, we must use simple cognitive procedures or "shortcuts" to make reasonable choices (Simon 1985; Sniderman et al. 1991). When political elites disagree, people tend to follow the elites sharing their ideological or partisan predisposition - if they are aware of the elite positions (Zaller 1992). Changes in public policy may consequently influence public attitudes - at least the attitudes of citizens who trust the decision-making parties.

Secondly, a shift in the reference point may affect preferences. Psychologists have found a greater sensitivity to losses than gains: people tend to prefer the status quo over alternatives with the same expected value (Quattrone & Tversky 1988). When public policy is changed, people may eventually get used to the new status quo, and evaluate the new policy more favourably. Thirdly, large reforms - as the introduction of a national social insurance system - may limit the alternative policies discussed in a society. When a return to the previous policy is no longer a feasible option, preferences may be adjusted to reality. Finally, political decisions may change the citizens' interests, and with that their preferred policy outputs. For instance, the Thatcher government tried to promote an entrepreneurial spirit through its privatization policy.⁵

Exogenous or Endogenous Preferences?

In Rokkan and Valen's approach, the perspectives of the consumer and support models are combined. Here, preferences are products of social cleavages; they are not easily changed as a result of political decisions. On the other hand, political parties will shape values and attitudes among their

adherents. With reference to the Norwegian counter cultures, Valen &
Rokkan (1974, 364-365) write:

We do not think it possible to interpret our findings in strict terms of causality: we have to recognize possibilities of interaction, of positive or negative feedback . .. identification with one or the other culture may have led to identification with a party defending that culture and the social pressures existing within the party context may in their turn have reinforced the initial identification.

Accordingly, political parties should be regarded as opinion leaders as well
as spokesmen (Aardal 1993, 319-333; Iversen 1994).

The perspective here will be that attitude formation generally reflects all three theoretical models. Then, however, the question is under what conditions will citizens follow their leaders or protest against them, and under what conditions is it likely that political preferences are exogenous. Firstly, support for new policies seems to be a more likely outcome than discontent. In a study of opinion change in Norway in the 19705, Valen (1981, 300) concluded that opinion tended to follow and support decisions made by political leaders. The psychological and sociological mechanisms mentioned above make the reactions described in the discontent model less likely, unless a policy causes strong and visible negative effects.

Secondly, several factors may affect the relevance of the exogenous preferences assumption. In Tonsgaard's (1992) model of referendum behaviour, conditions which may determine the possibility of influencing voters' attitudes are described. Three central elements are the complexity of the issue, the issue's importance to individual living conditions and the strength of the connection between the issue and basic values. To simplify the discussion, the last two conditions may be combined. Thus, the complexity and importance of an issue may indicate whether political preferences are stable or affected by policy decisions. However, the complexity and importance of a given issue may vary - both across time and between citizens (Nilson & Bjørklund 1986, 265-267).

In the consumer model, we assume that the citizens possess all the relevant information, compare the alternatives and choose the option with the highest utility. Therefore, the consumer model may be well-suited for simple and ordinary issues. This is the case for the main subject of this article: alcohol policy. Moreover, most people have first-hand knowledge of the object of this policy. Compared with the formation of attitudes to, for example, drug policy (Ødegård 1995) or foreign policy, attitudes to alcohol policy may be less dependent on information from the mass media and the authorities. On the other hand, alcohol policy is not especially salient on the current Norwegian political agenda - although it is more important in Norway than in most countries. In a recent survey, Norwegian party members were asked to choose the three most important issues from a list of

19 items. The alcohol issue was given the lowest priority; only 2 percent of
the party members mentioned this issue.6 This lack of priority may
strengthen the support model.

Question Format and Attitude Change

Differences concerning simplicity and importance of issues are not the only cause of varying results in the study of opinion-policy links. In addition, different kinds of questions measure different processes. In an economics textbook, Lipsey et al. (1990, 161) emphasize the difference between survey questions measuring marginal and total value, or utility. Relative questions refer to an existing situation and demands for change in current government policy (Stimson 1991, 28-29). With this question format, one may ask for an evaluation of government policy, measure priorities (e.g. questions about spending more or less on something), or refer to a process which changes current policy (e.g. privatization). These kinds of question are able to measure diminishing marginal utility, since they refer to the demand for change from the existing situation. Only opinion measured by relative questions can be expected to move in cycles around an equilibrium, as described above. In order to measure attitudes towards a specific object, absolute questions should be used. These questions measure the total utility attached to each alternative policy output or end-state. The alternatives are concrete: their content and meaning do not change when government policy changes.

Table 1 describes the relation between policy, opinion change and question format: How will a policy decision affect survey answers, dependent on question format and citizen reactions (other things being equal)? To exemplify the distinction between absolute and relative questions, reactions to a liberalization of alcohol policy are discussed. In the consumer model, political preferences are not affected by policy decisions. Attitudes towards a specific issue - e.g. selling wine in grocery shops - will not change (cell 2). However, political decisions will influence attitudes to the direction of public policy: if rules are liberalized, fewer people will think that the rules are too restrictive (cell 1). This is quite obvious, but the point is that the current policy must be considered when relative questions are used - in both cross-national and longitudinal research.

According to the other models, political preferences shift when policy changes. The discontent model assumes that every policy is more attractive in theory than in practice. The prohibition years in Norway exemplify this model. After several years of prohibition, side-effects appeared. There was an increase in home distillation and smuggling, problems occurred in foreign trade, and support for prohibition declined (Nilson 1972, 1978; Hauge 1986). In the same way, a liberalization may lead to increased consumption,

followed by more alcohol-related problems. Then, support for restrictions may increase (cells 3 and 4). This actually happened in Finland, when more liberal rules governing the sale of beer were introduced in 1969 (Mäkelä 1987; Ahlström & Österberg 1992).

If the support model is more appropriate, a liberalization will probably not affect the demand for a more liberal policy (cell 5). If, for example, the government implements a less restrictive policy, most supporters of the government parties may still be satisfied with the alcohol policy. Both libertarians and teetotalists may still demand policy changes, each in their own direction. However, less people will support specific restrictions on the sale of alcohol (cell 6).

Trends in Norwegian Alcohol Policy and Opinion

Three models of public reaction to political decisions have been presented above. In this section, these models are employed to account for changing attitudes to alcohol policy in Norway. However, as described initially, connections between policy and opinion may be explained in different ways. Through local referendums, municipal elections and changing attitudes among municipal council members, public opinion has contributed to political decisions. Moreover, previous research has pointed to social changes which may have produced shifting preferences in the population as well as changes in public policy.

One such prior cause is increased contact - through tourism and mass media - with countries where the policy is much less restrictive (Nordlund 1993). Generally, the attitude change may reflect a decline in austerity values, giving way to a consumption-oriented culture (Hellevik 1993). Other changes such as improved communications, urbanization and economic

growth have weakened the strongholds of teetotalism in southern and western Norway (Aardal & Valen 1989, 232-237). Moreover, when communication is improved, the utility of municipal alcohol control policy is more dependent on the policy of neighbouring municipalities. This issue will be further discussed.

These social changes have certainly affected public opinion. Still, there may be reason to believe that, in addition to this, policy changes have influenced existing opinion trends - as described in the three models. A reinforcement effect is predicted in the support model: if trends in public opinion influence policy, changes in policy will strengthen the shift in opinion. In the discontent model, policy implementation will reverse opinion trends. In the following section, trends in alcohol policy are described and survey data are analysed in order to explore these connections between policy and opinion.

Norwegian Alcohol Policy

Since the starting-point of the discussion above was a liberalization of alcohol policy, it is necessary to establish whether Norwegian policy has moved in that direction since the early 19605, in the period covered by survey data. Alcohol policy may be defined as public provisions, aiming at regulating and limiting the access to alcohol (Hauge 1986, 13). Norwegian alcohol control policy is quite restrictive, compared to most Western countries (Hauge 1986, 1988; Nordlund 1988). Two main groups of alcohol control measures are price regulations and availability restrictions. Considerable taxes on alcoholic beverages have resulted in high prices. Availability is reduced through several national and municipal restrictions on the sale and serving of alcohol. Survey data contain mostly questions on availability, and the analyses below will focus on attitudes towards local and national availability regulations.⁷

The responsibility for these regulations is divided between national and municipal authorities. The State Wine and Spirits Monopoly, which has the sole right to sell wine and spirits, has been a major policy instrument. Wine and spirits are sold by the Monopoly in special stores. The sale of beer, and the serving of all alcoholic beverages, takes place in shops and restaurants which have been granted a licence by the municipality. While this framework is enacted by national authorities, alcohol control measures are largely implemented by local government. Each municipal council decides whether different alcoholic beverages can be sold or served in the municipality. However, the consent of national authorities must also be obtained when a new spirits outlet is opened. Until 1990, citizens could take part in the decision-making through binding local referendums. During the survey period, changes in national regulations have not taken a general

liberal or restrictive direction. The basic features of national alcohol policy
have been stable.

Nevertheless, municipal decisions have caused a significant liberalization. An increasing number of municipalities have granted licences for the sale and serving of alcohol. The share of the Norwegian population living in municipalities where spirits are sold or served increased from 35 percent in 1962 to 89 percent in 1991. While 25 percent lived in "dry" municipalities in 1962, this applied to only 1 percent of the population in 1991 (Saglie & Nordlund 1993, 19). National authorities have, in principle, asked for a restrictive policy, but this does not seem to have reached the local politicians. Moreover, the distance between principles and practice may be large at the municipal level (Andersen & Bugge 1993). Policy signals from the Parliament and government have often been ambiguous. Thus, municipal councils have enjoyed considerable freedom of action (Denstad 1988; Denstad & Hansen 1988).

Attitude Changes 1962-91: Absolute and Relative Questions

In the discussion of the three models, I emphasized that the measurement of public reactions to policy changes is affected by the question format. In the surveys carried out by the National Institute for Alcohol and Drug Research (SIFA) since 1962, both absolute questions, where respondents choose between concrete alternatives, and relative questions, referring to demand for changes in the current situation, were included.⁸ With absolute questions, the respondents were asked about their vote if a referendum on the sale and serving of alcohol were to be held in their municipality.⁹ The results are shown in Fig. 1. Votes for the sale of spirits and wine are steadily increasing, while support for the sale of beer has grown in a less regular way. A similar pattern was found when respondents were asked about the serving of alcohol (Saglie & Nordlund 1993, 22). Generational replacement has contributed to this development: the younger generations are much more liberal than the cohorts they replace. At the same time, attitudes became more liberal within the cohorts followed from 1962 to 1991 (Saglie & Nordlund 1993, 23-25).

The discontent model predicts that a liberalization of local alcohol policy will weaken support for the sale of alcohol. This has not happened, perhaps because the harmful effects have been limited. In spite of the liberalization, the registered sale of alcohol has fallen during the 1980s (Nordlund 1993). The results are consistent with the support model. However, the effect of prior variables - i.e. social changes - cannot be separated from the mechanisms described in the support model.

With relative questions, the relation between opinion and policy becomes
more complicated. The answers will reflect both new preferences among the
citizens, and actual changes in the policy evaluated by the respondents.

Reactions in line with the consumer model may therefore influence the answers. If the policy becomes less restrictive, demand for liberalization will decline - and this will counteract the liberal trend shown in Fig. 1. SIFA's surveys contain two relative questions: one on availability policies in the country as a whole, and a second on availability policies in the respondent's municipality.¹⁰

The relative strength of the liberal and restrictive groups describes the "policy mood". In Table 2, the balances of opinion - i.e. the percentage in favour of liberalization minus the proportion supporting a more restrictive policy - are calculated as an indicator of this "mood". The trends are liberal, and strongest in the 19605. However, attitudes were more restrictive in 1991 than in 1985. The most distinct shift concerns the rules in the country as a whole. Attitudes towards local provisions have not changed much since 1966, except for the 1985 figure. Similar results are found in the Norwegian Election Studies, where the proportion demanding more liberal alcohol rules also has been quite stable since the 1970s (Valen 1992, 157). However, attitudes were not especially liberal in the election surveys from 1985."

How are the results in Table 2 linked to political decisions?¹² The liberal trend is quite weak when local rules are considered. Counteracting processes may have caused this result. Although attitudes towards concrete policy matters have become more liberal (as shown in Fig. 1), local authorities have enacted more liberal rules and thus weakened the demand for a less restrictive policy. The "country as a whole" item may, however, be ambiguous. If people understand rules in "the country as a whole" as the sum of local rules, the local liberalization will contribute to a maintained

balance - in the same way as for the municipal rules item. Rules in the "country as a whole" may also be perceived as the national framework. The national policy, which has been mainly stable, has not been able to weaken the demand for liberalization; there is no counterweight to the liberal trend. Both interpretations have probably influenced the answers, but the latter is presumably the reason why attitude change is stronger when the "country as a whole" is considered.

Attitude Change and Municipal Policy

Since parts of Norwegian alcohol policy are decentralized, comparing municipalities with different sales conditions may clarify the relation between political decisions and attitude change.¹³ With this comparison, it may be possible to separate the partly contradictory processes affecting the liberal trends in Table 2. Classification by sales conditions is, however, problematic in 1991. Although the group living in municipalities without any sale of alcohol in shops is large enough for statistical analysis, it is drawn from a small number of municipalities. Special conditions in a single municipality, e.g. the serving conditions, may affect the results.¹⁴

As can be seen in the upper part of Table 3, the difference between the groups of municipalities has declined during the last part of the survey period. The attitude changes have not taken place simultaneously throughout the country. In municipalities where wine and spirits are sold (mainly towns), there were about as many liberals as restrictives in 1962. Since 1966, however, there has been a relatively stable liberal predominance. Gradually,

the balance has shifted in a liberal direction in the other groups, too, first
where beer is sold, later in municipalities without any sale of alcohol.

Nevertheless, it is attitudes towards local rules that may show directly how opinion and political decisions are related. These attitudes are described in the lower part of Table 3, where the decline in geographical differences has brought about another outcome. In their analysis of the 1962 data, Brun & Krogh (1966, 206) concluded:

the principle of local self-determination had functioned in accordance with the intentions, so
that the actual sales conditions balanced the citizens' attitudes quite well - maybe
remarkably well.

This balance was to some extent disturbed as early as 1966 (Krogh 1967). Later, discontent has grown in the most restrictive municipalities. While the demand for more liberal policies has increased weakly in municipalities where alcohol is sold, this increase was much stronger in communities without any sale of alcohol.

The figures in Table 3 cover a third trend, which pulls in the opposite direction. Because several municipalities have implemented a more liberal policy, the demand for more liberal rules has declined in agreement with the consumer model. In Table 4, municipalities with a stable alcohol policy are separated from municipalities where rules were changed between 1979 and 1991. Categories where the sample is drawn from only a few municipalities - and partly different municipalities in 1979 and 1991 - are still a problem.¹⁵ Nevertheless, there are distinct differences with regard to the balance of

opinion, which hardly are caused by special features of single municipalities. In 1979, the demand for a more liberal policy was strong among one category of citizens: those who lived in municipalities where no alcohol was sold at that time, and licences were granted later. Here, local democracy has functioned. In 1991, after the introduction of the sale of beer, the balance is restored. In return, the mood has shifted in favour of liberalization in municipalities where the sale of alcohol is still prohibited. However, the balance of opinion is unchanged in municipalities where the sale of spirits is introduced.

As described earlier, social and cultural changes in Norwegian society may explain the attitude shifts. However, new values cannot explain the differences between groups of municipalities in Table 4. During the past 30 years, many "dry" communities have experienced that neighbouring municipalities have granted licences for the sale of beer. When its citizens are able to buy beer across the municipal border, the utility of a local prohibition declines. Moreover, the costs increase, as local shops lose their customers. Politicians who want to reduce the consumption of alcohol may choose a liberal policy for the sake of employment, even if they had preferred a situation where all municipalities had kept their regulations. It is easy to understand that many local politicians regard alcohol policy as a badly suited area for local self-government (Denstad & Hansen 1988, 337-338). Considerations of local trade interests become important for many members of municipal councils (Denstad 1988) - and probably for citizens, too. Liberalization of the sale of beer has built up expectations of a more

liberal policy where prohibition is maintained. The pressure of expectations appears to be weaker in municipalities where beer, but not wine and spirits, is sold. The reason may be that the opening of Wine and Spirits Monopoly outlets seems unlikely in most small municipalities.

A declining demand for more liberal rules when the sale of beer is introduced is no surprise. The relation between liberalization and support for the sale of alcohol - the absolute questions - may be more interesting. Two of these items are included in Table 4. The geographical differences are decreasing, especially with regard to the sale of beer. Local self-government appears to have functioned well. In municipalities without any sale of alcohol in 1979, attitudes towards sale of beer were more liberal where this sale was actually permitted during the 1980s. In 1979, support for the sale of spirits was also stronger where this sale later was introduced. In other words, opinion seems to have influenced policy.

Discerning any effects of policy on opinion is more difficult. Support for the sale of beer and spirits has increased where this sale is introduced, but also where local rules are unchanged. Case studies may provide a more accurate picture of such effects. Results from surveys carried out by SIFA, in municipalities before and after changes in sales conditions, are in line with the support model: acceptance of the sale of wine and spirits has increased where Monopoly outlets were opened. Moreover, a restrictive measure - the introduction of local beer monopolies - was also met with increased support (Nordlund 1978, 122-127; Hauge & Amundsen 1994, 43^7).

The Dynamics of Opinion and Policy

The point of departure of this article is the complex connection between opinion and policy. Public opinion is partly policy-shaping, and partly shaped by policy. Citizens respond to policy changes with a mixture of protest and support - and the balance between protest and support may direct future policies. The liberal shift in alcohol policy as well as opinion in Norway may exemplify these mechanisms. A restrictive policy has lost support. The balance between demands for liberalization and stricter regulations has shifted, although a majority is satisfied with the present rules. Municipal authorities have - as opposed to national authorities - responded to the demand for liberalization. Therefore, the shift in the balance of opinion is not dramatic.

Three models of public reactions to political decisions were presented to account for trends in public opinion. The discontent model predicted that policy implementation will reverse existing opinion trends. Such counteracting effects were not found. This result confirms previous research, where support for new policies is described as a more likely outcome than protest.

It is likely that a liberalization must result in much larger - and more visible - harmful effects, if the public reaction is likely to be discontent. The results are, on the other hand, consistent with a combination of the support and consumer models. When relative questions are employed, reactions described by the consumer model are found: even though the main trend is liberal, the demand for liberalization has declined in municipalities where the sale of beer is introduced. While the evidence in favour of the support model is inconclusive, it seems likely that local policy decisions have contributed to increased support for liberal rules. However, conclusions about the causal relationship are still uncertain. In future research on the effects of policy on opinion, panel surveys should be used. In addition, case studies of decision-making processes could tell us more about the impact of opinion on policy.

Several conditions affect both opinion shifts and policy decisions. Social and cultural changes shape new attitudes among local politicians and among citizens, and with that policy changes. New values result in more positive attitudes towards alcohol, and a growing scepticism towards public prohibitions and regulations. Moreover, doubts about the utility of alcohol policy restrictions may be increasing. Liberalization in a municipality may bring about changes among its neighbouring municipalities. No corresponding liberalization has taken place on the national level. At present, national restrictions are also being put under pressure, e.g. through import from countries where alcohol is less expensive. How Norwegian alcohol policy - and opinion - will be affected by this pressure, remains to be seen.

---

NOTES

1. In this context, the utility concept is used in a broad sense. "Utility" may include both economic gain and ideological or political objectives.

2. See Flanagan (1987) and Aardal (1993) for discussions of Inglehart's use of the marginal utility concept.

3. Inglehart's questions are relative, Listhaug's are mainly absolute. Knutsen has employed both question formats. The contrast between absolutes and relatives is seen in Knutsen's (1995, 178) table 6.5, when, for example, the distributions of answers on the "business management" (absolute) and "individual/public responsibility" (relative) items are compared.

4. Similar results were reported in Svallfors's (1993a, 1993b) comparisons of attitudes in Sweden, Germany and Britain. For a review of the literature, see Listhaug (1995, 598-599).

---

5. However, most new home- and shareowners did not swing to the right (Heath et al. 1991, 120-130).

6. Data from the Norwegian Party Member Survey 1991 (not published). N = 1789, "not answered" are excluded. For the question wording of the items, see Saglie (1992, 153-159).

7. In the SIFA surveys, one question on alcohol prices is also included. Most people think, not surprisingly, that prices are too high (Saglie & Nordlund 1993, 27).

8. The Norsk Gallup Institutt and Norsk Opinionsinstitutt were responsible for sampling and fieldwork. "Not answered" is excluded from all computations. The 1966 and 1991 figures are weighted. See Nordlund (1977, 1981, 1987) for further information about the surveys. See also Brun-Gulbrandsen & Krogh (1966) and Krogh (1967) for an analysis of attitudes in 1962 and 1966, and Nordlund (1993) and Saglie & Nordlund (1993) for further analyses of the whole period.

9. "If it were decided to hold a referendum on the issue of whether the sale or serving of alcohol should be permitted in this municipality, would you vote for or against each of the following options: the sale of spirits from outlets; the sale of wine from outlets; the sale of strong beer in shops; the sale of ordinary beer, such as lager, in shops?"

10. "In this country many legal provisions regulate the sale of alcohol, as is well known. Some people think that these rules, broadly speaking, are too restrictive, so that it is too difficult or troublesome to buy alcohol. Other people think that these rules, on the contrary, are too liberal, so that it is too easy for people to buy alcohol. What is your opinion? When you consider the country as a whole, do you think the rules, broadly speaking, are too liberal, acceptable or too restrictive? And if you consider the conditions here in this municipality, do you think the rules, broadly speaking, are too liberal, acceptable or too restrictive?"

11. Thus, there may be reason to be somewhat sceptical of SIFA's liberal 1985 figures. A direct comparison between SIFA's figures and the election surveys is impossible, since the question wording is different. In addition, the proportion thinking that the rules are too restrictive is generally lower in the election studies than in SIFA's surveys. The context has probably affected the distribution of answers. In the SIFA surveys, the attitude items are located below a large number of questions on drinking habits, home distillation, smuggling, etc., while the alcohol policy item follows a question on teetotalism in the electoral surveys.

12. See Saglie & Nordlund (1993, 31-37) for analyses of the connection between social and demographic variables and attitudes to alcohol policy.

13. Classification by a combination of sales and serving conditions might provide a more complete measure of municipal policy. However, the number of respondents living in municipalities with neither sale nor serving of alcohol becomes too small for analysis towards the end of the period. Besides, the sale of alcohol in shops constitutes a large part of the registered consumption in Norway (Nordlund 1985, 66).

14. The group living in municipalities without any sale of alcohol contains 63 persons in 1991. A two-stage sampling procedure was employed (see Nordlund 1981, 3-4), so that these 63 are drawn from only five different municipalities.

15. This applies mainly to municipalities without any sale of alcohol at both points of time, where the sample was drawn from 8 municipalities in 1979 and 5 in 1991, and municipalities where sales conditions were changed from sale of beer to sale of all alcoholic beverages, where the sample was drawn from 8 municipalities in 1979 and 9 in 1991. In both groups, 3 municipalities were represented in 1979 as well as 1991.

ACKNOWLEDGEMENTS

This article is based on a research report (Saglie & Nordlund 1993) from the National Institute
for Alcohol and Drug Research (SIFA). lam grateful to colleagues, especially those at SIFA
and the Department of Political Science, who have read and commented on earlier drafts.

REFERENCES

Aardal, B. 1993. Energi og miljø: Nye stridsspørsmål i møte med gamle strukturer. Rapport
93:15, Oslo: Institute for Social Research.

Aardal, B. & Valen, H. 1989. Velgere, partier og politisk avstånd. Oslo: Central Bureau of
Statistics.

Ahlström, S. & Österberg, E. 1992. "Changes in Climate of Opinion Concerning Alcohol
Policy in Finland in the 1980s", Contemporary Drug Problems 19, 431-457.

Andersen, J. & Bugge, L. S. 1993. "Planlegging og politikk - foreløpige resultater fra en studie
av arbeidet med alkoholpolitiske handlingsprogrammer i to kommuner," ØF-notat 32/93,
Lillehammer: Østlandsforskning.

Arner, O. 1993. Hva er alkoholmisbruk? SIFA-rapport 1/93, Oslo: National Institute for
Alcohol and Drug Research.

Brun-Gulbrandsen, S. & Krogh, P. 1966. "Folkemening om alkoholomsetning", Norsk
Tidsskrift om Alkoholspørsmålet 18, 165-206.

Denstad, B. 1988. "Kommunal alkoholpolitikk: Politisk ideologi og alkoholpolitiske
strategier", Alkoholpolitik 5, 73-82.

Denstad, B. & Hansen, T. 1988. "Kommunal håndheving av statlig alkoholpolitikk:
partikonflikter og representantroller", Tidsskrift for samfunnsforskning 29, 333-356.

Flanagan, S. C. 1987. "Value Change in Industrial Societies", American Political Science
Review 81, 1303-1319.

Haavelmo, T. 1993. Økonomi, individ og samfunn. Oslo: Universitetsforlaget.

Hauge, R. 1986. Alkoholpolitikken i Norge. Oslo: National Directorate for the Prevention of
Alcohol and Drug Problems.

Hauge, R. 1988. "Alcohol Control Policies", in Skog, 0.-J. & Waahlberg, R., eds., Alcohol
and Drugs: The Norwegian Experience. Oslo: National Directorate for the Prevention of
Alcohol and Drug Problems.

Hauge, R. & Amundsen, A. 1994. Virkninger av økt tilgjengelighet på alkohol. SIFA-rapport
2/94. Oslo: National Institute for Alcohol and Drug Research.

Heath, A. et al. 1991. Understanding Political Change. Oxford: Pergamon Press.

Hellevik, O. 1993. "Postmaterialism as a Dimension of Cultural Change", International
Journal of Public Opinion Research 5, 211—233.

Inglehart, R. 1990. Culture Shift in Advanced Industrial Society. Princeton: Princeton
University Press.

Iversen, T. 1994. "The Logic of Electoral Politics: Spatial, Directional and Mobilizational
Effects", Comparative Political Studies 27, 155-189.

Key, V. O. 1961. Public Opinion and American Democracy. New York: Knopf.

Knutsen, O. 1995. "Left-Right Materialist Value Orientations", in van Deth, J. & Scarbrough,
E., eds., The Impact of Values. Oxford: Oxford University Press.

Krogh, P. 1967. "Folkemening om alkoholomsetning 1962-1966". Norsk Tidsskrift om
Alkoholspørsmålet 19, 1-33.

Lipsey, R., Steiner, P., Purvis, D. & Courant, P. 1990. Economics (9th ed.). New York: Harper
& Row.

Listhaug, O. 1990a. "Macrovalues: The Nordic Countries Compared", Acta Sociologica 33,
219-234.

Listhaug, O. 1990b. "Velferdsstat og veljarar: Komparative funn", in Hovdum, A., Kuhnle, S.
& Stokke, L., eds., Visjoner om velferdssamfunnet. Bergen: Alma Mater.

Listhaug, O. 1995. "Komparativ offentlig opinion i Europa: En översikt over nyere
forskning", Tidsskrift for samfunnsforskning 36, 583-615.

Mäkelä, K. 1987. "Attitudes and Opinions", in Simpura, J., ed., Finnish Drinking Habits.
Helsinki: The Finnish Foundation for Alcohol Studies.

Nilson, S. S. 1972. Politisk avstånd ved norske folkeavstemninger. Oslo: Gyldendal.

Nilson, S. S. 1978. "Scandinavia", in Butler, D. & Ranney, A., eds., Referendums.
Washington: American Enterprise Institute.

Nilson, S. S. & Bjørklund, T. 1986. ' "Ideal Types' of Referendum Behaviour", Scandinavian
Political Studies 9, 265-278.

Nordlund, S. 1977. "Data om alkoholspørsmål", SIFA-stensilserie nr. 6, Oslo: National
Institute for Alcohol and Drug Research.

Nordlund, S. 1978. Lokale ølmonopol - virkninger på drikkevanene. Oslo: Universitetsforlaget.

Nordlund, S. 1981. "Alkoholdata 1979", SIFA-stensilserie nr. 50, Oslo: National Institute for
Alcohol and Drug Research.

Nordlund, S. 1985. "Norske drikkevaner. En beskrivelse på grunnlag av offentlig statistikk og

Nordlund, S. 1987. "Data om alkohol og andre stoffer 1985", SIFA-stensil nr. 1/87, Oslo:
National Institute for Alcohol and Drug Research.

Nordlund, S. 1988. "Alcohol Control Policies in Norway and Their Effects", in Skog, 0.-J. &
Waahlberg, R., eds., Alcohol and Drugs: The Norwegian Experience. Oslo: National
Directorate for the Prevention of Alcohol and Drug Problems.

Nordlund, S. 1993. "Alcohol Policy and Public Opinion". Paper presented at the Kettil Bruun
Society's Alcohol Epidemiology Symposium, Krakow.

Papadakis, E. 1992. "Public Opinion, Public Policy and the Welfare State", Political Studies
40, 21-37.

Quattrone, G. A. & Tversky, A. 1988. "Contrasting Rational and Psychological Analyses of
Political Choice", American Political Science Review 82, 719-736.

Rokkan, S. 1966. "Norway: Numerical Democracy and Corporate Pluralism", in Dahl, R., ed.,
Political Oppositions in Western Democracies. New Haven: Yale University Press.

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

Saglie, J. 1992. Partier, medlemmer og ideologi. Thesis. Oslo: Department of Political
Science, University of Oslo.

Saglie, J. & Nordlund, S. 1993. Alkoholpolitikken og opinionen. SIFA-rapport 3/93, Oslo:
National Institute for Alcohol and Drug Research.

Simon, H. A. 1985. "Human Nature in Politics: The Dialogue of Psychology with Political
Science", American Political Science Review 79, 293-304.

Sniderman, P., Brody, R., Tetlock, P. et al. 1991. Reasoning and Choice: Explorations in
Political Psychology. Cambridge: Cambridge University Press.

Stimson, J. A. 1991. Public Opinion in America. Boulder: Westview Press.

Stimson, J., MacKuen, M. & Erikson, R. 1995. "Dynamic Representation", American Political
Science Review 89, 543-565.

Svallfors, S. 1993a. "Dimensions of Inequality: A Comparison of Attitudes in Sweden and
Britain", European Sociological Review 9, 267-287'.

Svallfors, S. 1993b. "Policy Regimes and Attitudes to Inequality: A Comparison of Three
European Nations", in Boje, T. & Olsson Hört, S., eds., Scandinavia in a New Europe. Oslo:
Scandinavian University Press.

Tonsgaard, O. 1992. "A Theoretical Model of Referendum Behaviour", in Gundelach, P. &
Siune, K., eds., From Voters to Participants. Aarhus: Politica.

Valen, H. 1981. Valg og politikk - et samfunn i endring. Oslo: NKS-Forlaget.

Valen, H. & Rokkan, S. 1974. "Norway: Conflict Structure and Mass Politics in a European
Periphery", in Rose, R., ed., Electoral Behavior. New York: The Free Press.

Weissberg, R. 1976. Public Opinion and Popular Government. Englewood Cliffs, NJ:
Prentice-Hall.

Zaller, J. R. 1992. The Nature and Origins of Mass Opinion. Cambridge: Cambridge
University Press.

Ødegård, E. 1995. "Legality and Legitimacy: On Attitudes to Drugs and Social Sanctions",
British Journal of Criminology 35, 525-542.