On 22 November 1991, the Danish Prime Minister, Paul Schliiter, announced that a new election was to be held on 12 December. This announcement was remarkable for several reasons. For one thing, the date selected meant that there was to be a campaign of only 20 days, which is very close to the minimum technically possible and certainly very far from what is in keeping with the spirit of the Constitution. In the Constitution, moreover, there is an expectation that elections will only be held roughly every four years. In this case the government had held power for only two years and seven months.¹

The formal reason for calling the election was a classical political disagreementand over the state budget between the governing parties (the Conservatives, the Liberals and the Radical Liberals) and the major party in opposition (the Social Democrats). Even so, the call for an election was a surprise because, as Svend Auken, Chairman of the Social Democrats noted, there was agreement between the negotiating parties on 98 percent of the budget. This led to a political game among the parties as to whom was to blame for calling the election. No clear villain was found - contrary to what was the case in the 1988 election, when Svend Auken appeared as an untrustworthy politician and had to take the blame for causing the election. In 1990, by contrast, Auken pursued a careful and successful approach during the campaign, an approach designed to make him appear as the mature and balanced statesman looking for politically solid compromises.He proposed a majority government which by necessity would have to have included non-socialist parties such as the Conservatives. This

---

* Data from the campaign used in this article are drawn from surveys conducted by AGB Gallup by telephone for Berlingske Tidende and Radio Denmark. All other surveys are faceto-face conducted by the Danish election research group and the data are available through the Danish Data Archive. All surveys are conducted nationally with representative samples of the electorate as post-election studies. In 1971 and 1987 there were supplementary pre-election studies.

change in style, which meant that he avoided confrontations and political
meetings with opponents, was widely noted and discussed.

The real reason for calling the election, however, appears in large part to have been due to internal problems within the government. The chairmen of the three governmental parties had shown considerable public disagreement a period of time, and this disagreement was especially pronounced during the preceding summer. According to the opinion polls, moreover, the Liberals had gained in popular support and appeared to want to turn this gain into electoral mandates, thus pushing for the election in order to change the balance of power within the government.

It also seems - at least according to early statements and subsequent campaign strategies - as if the parties in government believed they had a good issue to campaign upon, namely a reduction in taxes. Yet the parties made one major mistake: the tax relief measure was presented in such an awkward, bureaucratic manner that they themselves stumbled in explaining its contents. Formally named 'the 6 percent tax law', the intent of this proposal was to lower the progression in direct taxes. Deep confusion regarding this proposal was revealed as early as the night of the call for the election, as the party leaders were gathered for a televised debate. Most noticeably the chairman of the Liberals, who was also the minister of foreign affairs and an economist, showed a lack of understanding of the proposal. For illustrative purposes he referred to female supermarket cashiers and described how they would benefit from the proposed law. The fact of the matter, however, was that they would not benefit; their income bracket was too low for the proposed relief to have any effect. Even so, supermarket cashiers figured as prominent persons throughout the campaign.

The Outcome

Despite the short campaign period, major changes took place (cf. Table 1). There was a net change of party, for example, of 13.3 percent, a figure which was double the change in 1988 and generally above normal for the 1980s as a whole.² The Social Democrats in particular were the big winners of the election, enjoying an astounding 7.6 percentage points increase in support as compared with results from the 1988 election. Half of this gain took place in the last days before the election proper. This gain was in the main offset by a loss for the Socialist People's Party, an outcome that led to a change of leadership.

A new party creation, a united left (the Marxist-Leninists, the Communistsand
Left Socialists), did not achieve enough votes for parliamentaryrepresentation.
the socialist parties taken together had a

gain of 3.1 percentage points, which is a major change across the middle line in Scandinavian politics, where the relationship between the socialist and bourgeois parties has traditionally been characterized by a very close and delicate balance. Viewed from a long-term perspective, however, it might be argued that all that really happened was that the previous leftrightbalance 1987 and before was re-established.

On the bourgeois side there were two major changes. First, the balance between the two major parties - the Conservatives and the Liberals - changed in favour of the most rightist party, the Liberals, creating a situation in which the two parties had close to equal support. Second, the Progress Party lost substantially, especially considering that in the autumn of 1988 they had popular support of almost 20 percent of the populace and yet gained only 6.4 percent of the vote at the time of the election. This was

mainly due to internal trouble resulting in the expelling of the founder,
Mogens Glistrup, from the party. He tried to run outside of his 'own' party,
but did not gain re-election.

There are several reasons for this outcome. For one thing it is clear that the internal trouble in the Progress Party, resulting in the party split, contributed heavily both to Glistrup's failure to gain re-election and to the decline in support for the Progress Party. In addition, however, Glistrup's attempt to form a new party was made more difficult by a change in the election law which was adopted in April 1989. According to this act, it is necessary not only to gather approximately 18,000 signatures (equal to the number of votes necessary to get a mandate at the last general election) which must subsequently be checked (as was previously the case), but these signatures must now also be reaffirmed by the signature-giver before they are accepted by the ministry of interior affairs. In spite of extended efforts, Mogens Glistrup did not succeed in meeting this requirement and instead had to run with another populist, Møller Hansen of the Seamen's Union. Most of what these two candidates shared was populism and a distrust of the political system. Otherwise, they were so different that the joint efforts were doomed to fail, as they did.

Another reason for Glistrup's lack of success, however, was that the support he had had in attacking the political system declined after he was charged with and sentenced for tax fraud, and later for his pronounced racism. He figured prominently in politics from 1973 to 1990, and he may come back again, but his strongest legacy at this point seems to be found in Norway, where Carl I. Hagen of the Norwegian Progress Party is enjoying considerable success.³

In sum, the outcome of the 1990 election can be described as a move toward the centre in the sense that both the left wing (the Socialist People's Party) and right wing (the Progress Party) were substantially cut. On the other hand, there is a rather mixed picture of changes in the middle ground. There are changes both to the right and to the left.

Even so, possibilities for government formation were hardly changed. Although 13 parties competed for seats, the same eight parties retained representation in parliament. The bourgeois parties kept their majority, however, and only five days after the election a new bourgeois government was formed. The only difference in this case was that the government consisted solely of the Conservatives and the Liberals. The Radical Liberals, the smallest party in the previous government, suffered severe losses and announced its withdrawal from the government as early as the eve of the election.

Hence, Denmark once again has a minority government. The last
majority government retreated in 1971. And the new government has been
weakened inasmuch as the big winner of the election, the Social Democrats,

together with the Socialist People's Party may now form a majority with any of the four minor bourgeois parties. This means that so-called alternative majorities (e.g. on environment, foreign policy and social policy) that are not in line with government policy have better opportunities for being formed. The political situation is consequently less stable than before the election.

Party Change

The increase and relatively high level of net change of party which occurred in the election (13 percent) was not due to an increase in volatility, at least as this may be measured by gross change. If anything, change of party and relatively late decisions regarding voting choice were slightly below the average for the last 15 years or more, as evidence presented in Table 2 demonstrates. Even so, around one-fifth of the voters changed party and around one-fifth similarly made their voting decisions during the course of the campaign. This means that the electorate really did make moves, and these were not merely compensatory moves as in most of the preceding years.⁴ The matrix of change presented in Table 3 provides an interesting picture of this movement.⁵ Among other aspects revealed by this table, the following are especially worth noting:

First, the big gainer, the Social Democrats, received new votes from all over the political spectrum, though mostly from the neighbouring parties - the Radical Liberals and especially from the Socialist People's Party. Even earlier, Progress Party voters now voted for the Social Democrats. In these latter cases, however, it is reasonable to assume that they in large measure were old Social Democratic voters returning home to the fold.

Second, the three big, old parties - the Social Democrats, the Conservativesand

servativesandthe Liberals - have traditionally been the most stable parties in terms of having the highest percentage of core voters.⁶ In 1990, however, it would seem that the Conservatives lost this stable position and rather experienced a substantial in- and out-flow of voters. Instead, as has been seen at earlier elections (Elklit & Tonsgaard 1989,155-56; Sauerberg 1988, 367), the Christian People's Party joined the other two big, older parties in their pattern of a highly stable electorate. It should none the less be noted that ever since 1979 the Christian People's Party has had very stable, albeit minimal, support - just a little above the 2 percent threshold required for representation in the Danish Parliament.

Third, also following a long-established pattern, the Radical Liberals and the Centre Democrats are both subject to highly unstable voter support. In view of this, the sizeable in- and out-flows of support from these parties might well have caused larger fluctuations in overall party support than has actually been the case. In 1990, the in- and out-going streams seem to a large degree to have counter-balanced each other for the Centre Democrats, whereas the Radical Liberals experienced larger out- than in-flows.

Having established these party changes, an interesting strategic question arises: When was the decision to change party actually taken - or, phrased more broadly, when was the voting decision taken? Table 4 provides some evidence in this regard.⁷

In 1988, the short period of time which had elapsed since the previous

election was no doubt the major reason for the large number of late party changes; almost half (46 percent) did not make the decision to change until the last days before the election. Both the 1987 and 1990 elections, by comparison, show a similar pattern for the time when decisions to change were actually taken. In these cases around half of the decisions to change party were taken during the campaign itself, but a somewhat larger share of these decisions was reached earlier in the campaign. This evidence suggests that the campaign period really does matter. Whether it is the campaigns themselves or something else that matters, however, is another question. One approach to this question is to ask what happened during the time of the campaign, and whether there are any clues as to what might have influenced the outcome of the election?

Campaign Issues and Perceptions of the Best Government

There is a rich literature on voting behaviour and attempts to explain the reasons for votes and changes of vote. There seems to be a consensus that issue-voting is becoming more important, especially as class-voting wanes. This is not to say that what happens structurally is unimportant. A case in point here is the decline of leftist parties in western as well as eastern countries since 1989, a phenomenon which might very well explain a major part of the losses experienced by the Socialist People's Party in Denmark. Nor is it to say that the competence of politicians and their ability to handle the media in a persuasive and glamorous way does not matter. Neither is it to say that ideologies are dead. But it does indicate that a focus on the political issues as perceived by the electorate, especially the perception of

which government might best be able to handle the problems and the
perceived themes of the campaign, may yield insight into the voting process.
Evidence on these matters is contained in Table 5.8

With almost 300,000 unemployed - or a good 10 percent of the labour force - it is hardly surprising that 'unemployment' consistently tops the list of problems that voters feel are 'the most important that the politicians should address'. At the point when the election was called, on 22 November, almost one-third of the electorate (31 percent) mentioned 'unemployment' as the most important problem. Almost one-third also mentioned something like 'a sound economy' and somewhat fewer mentioned 'social problems' (the health system, young people, pensions, and so forth). Following this came the issue of taxes, which was mentioned by 20 percent. Another major area of political problems was 'environment', which was considered most important by 12 percent. All other problems such as industry and trade, the Common Market, foreign policy in general, a better society, etc., were each only mentioned by 5 percent or less.

During the campaign there was a remarkably clear development. Interest for the environment almost doubled, while interest in taxes fell sharply. Interest in unemployment also increased. In particular, bourgeois voters began opening their eyes to this problem. When the election was called, it

was primarily the Social Democrats and left-wingers who felt unemploymentwas matter of emergency. Close to the election day, however, this problem was mentioned by the bourgeois voters to the same degree, even though there are, no doubt, major differences amongst bourgeois and socialist voters with respect to how they deal with unemployment.

An obvious question then is who benefited and who lost from the voters' priorities and changes in them during the course of the campaign? Part of the answer to this question may be found in voters' perceptions of which type of government would be most capable of handling political problems with the highest priority. The voters were asked and, as indicated by the right-hand portion of Table 5, the result is rather clear: the Social Democrats supporters of a Social Democratic-headed government benefited most, while the bourgeois parties lost. Of the top five issues, a Social Democratic government was considered to be the best to handle three - unemployment, environment and social problems - whereas a bourgeoisheaded was perceived best in handling taxes and the economy. Already from the outset, the clumsy election issue of abolishing the socalled percent tax (which would benefit middle and larger incomes) seemed to give the Social Democrats the best strategic situation. As the priorities changed during the campaign, the situation of the party constantly improved. It is tempting to ascribe the Social Democratic success at least in part to this advantageous point of departure and later campaign developments. conclusion is supported, moreover, by the government evaluations of earlier bourgeois voters who changed to the Social Democrats.

The Electoral Campaign and the Role of the Mass Media

Another question, which will only be briefly addressed here, is what role did the campaign as such actually play? At the end of the campaign, the voters were asked what they felt had been the principal issues of the campaign.⁹ It is hardly surprising that one-fifth responded by saying, 'persons in politics/government possibilities/eventual prime minister", while one-sixth said, 'nothing/personal fights/empty election promises' and a few (6 percent) said they had not followed the campaign. These are fairly standard responses for most electoral situations. But a good half mentioned a political area of substance, and an overwhelming majority mentioned taxes as the core of the campaign. In fact, almost one-third of the electorate and more than half of the voters who mentioned a political area said taxes, while 8 percent said unemployment and 4 percent environment.

The fact that taxes went down in saliency and unemployment and environmentwent

mentwentup during the campaign constitutes something of an explanatory dilemma. This dilemma is illustrated in Table 6. First of all, voters perceived the campaign as dealing with problems that were completely out of line with their own priorities. In short, on this aggregate issue level there is no explanation for the election outcome. Taken on face value, moreover, there is no support for a thesis regarding the agenda-setting role of the media in election campaigns.¹⁰ On the contrary, it looks very much as though the processes by which the voters' changed priorities happened elsewhere. It could have happened, for example, in connection with a broader structurally determined change of attitudes and/or in various social settings during informal discussions, where the media, especially television, plays a more limited role as agenda-setter.

These possibilities, however, remain matters of speculation at this point. There is some evidence, however, that there has been a decline in support for the idea that 'it is important to follow the campaign on television in order to follow discussions'.¹¹ There seem to be several reasons for this decline. One is the multiplicity of television channels now available, which has led a larger part of the audience or voters to select alternatives to the election programmes. The audience for election broadcasting, in other words, has decreased (Nordahl Svendsen 1991, 1-2). Thus, contrary to Asp's (1986, 230) hypothesis that increased supply of political communicationwill

municationwilllead to increased political interest, it seems that the option to avoid political broadcasts is more often used than not. This has caused campaign interest to decrease as the supply of TV programmes has increased.

Not only did satellite and cable television increase heavily throughout the 1980s, the 1990 general election was also the first general election where Radio Denmark, the national broadcasting system, did not have a monopoly televising the election. A commercial television station had been started on 1 October 1988. This led to a split of the remaining audience for political broadcasting. Hence, in contrast to the 1970s and most of the 1980s, when often half of the voters had seen the same election broadcast 'the night before', in 1990 this percentage was reduced to one-quarter or less (Nordahl Svendsen 1991, 2). The socially felt need not to be an outsider -i.e. one who does not follow what is happening - has consequently been reduced. By the same token common themes from 'the night before' for discussions with colleagues at work or with friends are no longer so clear. Yet the two main channels - Radio Denmark and Television 2 - are not all that different, so some common ground for discussion topics picked from the media is still to be found.

There is, all the same, reason to believe that the agenda-setting role of the media has been reduced. It is still a valid concept, but it has become more diffuse as a societal split in subcultures has been reinforced by the new television situation which emerged during the 1980s. Much of the agenda-setting, therefore, takes place in these subcultures, with their specific forms of communication behaviour. Tendencies toward growth in the global media village, in other words, are offset by richer possibilities for forming subcultures.

What is more, available evidence suggests it is still important to discuss politics in smaller groups before a decision to change party is taken.¹² As has been the case for the last twenty years, the more voters discuss politics, the more likely a change of party is, or vice versa. Social radar, in short, still works, i.e. traditional networks still exert some influence, but they do so in a new communication setting, the impact of which we have just the first indications.

---

NOTES

1. It should be noted, however, that a period of 2 years and 7 months was a little above normal for political practices in Denmark during the last 20 years. Regarding the frequency and other features of Danish election campaigns, see Bille (1991b).

2. Net change as used here is equal to the sum of the losses by parties suffering decline. which is also equal to the sum of the gains enjoyed by parties experiencing growth. Gross change, by comparison, is the percentage of the voters that voted at both elections, but for different parties. Volatility is often used synonymously with gross change as it is the actual change of party which occurs at an election. Occasionally. however, 'volatility' is also used with reference to voters contemplation of changing party, thereby including a subjective component. This more encompassing definition is not used here.

---

2. Net change as used here is equal to the sum of the losses by parties suffering decline. which is also equal to the sum of the gains enjoyed by parties experiencing growth. Gross change, by comparison, is the percentage of the voters that voted at both elections, but for different parties. Volatility is often used synonymously with gross change as it is the actual change of party which occurs at an election. Occasionally. however, 'volatility' is also used with reference to voters contemplation of changing party, thereby including a subjective component. This more encompassing definition is not used here.

3. For a long time it looked as if the strong barriers to change in Sweden would prevent a similar phenomenon in that country. But in 1988 the Greens were the first new party to enter parliament since 1920 and in 1990 a party kindred to Mogens Glistrup's old party was founded. Through the spring of 1991 this party, named the New Democracy, has averaged around 10 percent in the opinion polis, representation with 7.2 percent in the September 1991 election. The Finns also seem to have kept the Progress Party spirit alive ever since Vennamo's success in the 19705, even though it has been dwindling in recent years.

4. In order to get a full overview of the changes which took place one should also consider the party choice of the 170,000 voters who voted for the first time. These voters constitute 4.3 percent of the electorate, but they have a considerably lower turnout than the rest of the electorate, which means that they account for only 3 percent of actual voters. In addition, a correction should be made for the behaviour of the deceased - around 140,000 voters. Finally around 8 percent of the electorate 'were in and out of the sofa', meaning that they only voted at one of the two elections.

5. In this table only percentages based on a sufficient number of cases to provide relatively stable estimates of transition rates have been included. Most of the cells which are empty in the table would in any event involve relatively small percentage figures (i.e. roughly 1 percent or less).

6. Core voters are defined as the percentage of voters that voted for a given party in 1988 and voted for the same party again in 1990. Core voters, in other words, are found in the matrix on the diagonal and are represented with bold figures. This definition means that a gaining party will have a relatively high proportion of core voters, as the 'old' voters are prone to stay with the party, whereas a losing party by definition will lose what otherwise could have been core voters. Or, to put it the other way around, a losing party will often be left with only core voters, namely the voters that voted for the party at the last election. In Table 3, for example, it can be seen that only 72 percent of the Socialist People's Party voters in 1988 voted for them again in 1990. But calculated the other way around, 94 percent of their 1990 voters were voters who also voted for them in 1988. The party, in other words, gained few new voters from other parties and instead lost voters to other parties. The Socialist People's Party was cut to the bone, but some real bone may be expected to remain.

7. The exact wording of the question upon which Table 4 is based is as follows: 'When did you decide to vote for the party which you voted for? Was it in the last days before the election, was it earlier in the election campaign, or did you know before the election campaign which party you would vote for?'

8. The exact wording of the questions upon which Table 5 is based is as follows: Saliency: 'Which problem or problems do you feel are the most important for the politicians to address.' Best government: 'I shall now mention some area of society, which the politicians must see to that they are functioning in the best way. For each area I would like you to tell me whom you find most able to solve the problems, a Social Democratic headed government or a bourgeois headed government?' For the item on 'best government' ten areas were mentioned, whereas only the top five on the saliency list are reported. The number of respondents is only 235 for the 11 December survey, but 964 answered the same question between 23 and 28 November. The results are very alike. If there is any difference, there seems to be a slight decrease for the Social Democrats. They did not do quite as well the day before the election as they did around a fortnight before. They did a little worse on unemployment and social problems, but still much better than a bourgeois-headed government. The bourgeois government seems instead to have strengthened its reputation during the campaign for being best able to handle taxes.

9. The exact wording of the question was: 'When you think about the campaign as it has developed until now, what do you think it especially dealt with? The distribution of answers was as follows: Answers Percent Various policy areas (see Table 6) 57 Forming a government, prime minister, politicians per se 20 Nothing, empty promises, quarrels 16 Other answers 3 Did not follow the campaign 6 DK/NA y Sum 111 N (1662) NB: Responses total more than 100 because respondents could indicate more than one answer.

---

9. The exact wording of the question was: 'When you think about the campaign as it has developed until now, what do you think it especially dealt with? The distribution of answers was as follows: Answers Percent Various policy areas (see Table 6) 57 Forming a government, prime minister, politicians per se 20 Nothing, empty promises, quarrels 16 Other answers 3 Did not follow the campaign 6 DK/NA y Sum 111 N (1662) NB: Responses total more than 100 because respondents could indicate more than one answer.

10. In all likelihood a content analysis of the campaign will not show such a dominance of taxes as a focal point. It can none the less be argued that what is most important in the agenda-setting process is the voters' perception of the campaign, not more objective evidence.

11. This situation is evident from responses to the following question asked at various points in time: 'Do you find it important to watch the election campaign on TV in order to be able to take part in discussions?' Answers to this question have been as follows: 1973 1975 1977 1979 1990 'Yes' 79 65 55 54 48 N (533) (1047) (921) (1694) (968) Note: DK varies from 2 to 16 percent. Between 10 and 14 percent of the sample is excluded. furthermore, because they did not watch television. This means that the N's should be reduced with approximately 25 percent for statistical tests.

12. The following question was posed in connection with the 1990 campaign: 'Have you in the last three weeks before the election discussed politics with somebody in your nearest family, with some of your friends or with some of your colleagues?' Positive responses among those who changed party between 1988 and 1990 are as follows: Number of discussion groups reported 0/1 2 3 Total Voters who changed party 1988-90 14% 19% 25% 19% N (281) (215) (240) (736) Note: 24 percent of the sample is not represented. as the universe consists of voters who voted for a party in both 1988 and 1990.

REFERENCES

Asp, K. 1986. Mäktiga mass medier. Stockholm: Akademilitteratur.

Bille, L. 1991b. 'The 1988 Election Campaign in Denmark'. Scandinavian Political Studies
14, 205-218.

Danmarks Statistik: Befolkning og valg. 1990:19.

Elklit, J. & Tonsgaard, 0., eds., 1989. To folketingsvalg. Århus: Politica.
Folketingets Håndbog 1988 og 1990. Copenhagen: Schultz.

Nordahl Svendsen, E. 1991. 'Lighedsprincip i konkurrence: Folketingsvalget 1990 i Danmarks
Radio'. Paper from Radio Denmark, Department of Media Research.

Sauerberg, S. 1988. "The General Election in Denmark 1988', Scandinavian Political Studies
11, 361-371.

Siune, K. 1990. 'Massemedier mellem partierne og vælgerne' in Carlsson, U., ed., Medier
Människor Samhälle. Göteborg: Nordicom-nytt/Sverige 3-4 1990.

Siune, K. 1991. EF pä dagsordenen. Århus: Politica.