The 1950s was a decade of political stability in Norway. Indeed, this situation reportedly prompted Israel's Golda Meir to remark 'Election in Norway is re-election' prior to the 1961 election (Lie 1975, 197). The stability of the 1950s was not to replicate itself in the following decades however. The 1970s saw an increase in floating voters (Valen 1981; Valen et al. 1990). No longer was it as simple to forecast the result of the election as it was to forecast the year of the election - something which, in Norway, is set by law.¹ Norwegian voters were not unique in this regard. Election research in a wide range of countries has revealed rising voter volatility (Pedersen 1983). Party preferences have become more fluid, and they can shift more easily.

As the outcome of the election has become more uncertain, the election
campaign itself has grown more decisive. Several critical questions arise in
this connection. One such question is whether activity and engagement in

Resumé

From 1945 to the end of the 1980s, Norway's election campaigns have changed substantially. The mass media - particularly television - have grown in importance, partly structuring the election campaign. In this study the period in question is divided into three phases: (1) 1945— 57, a phase characterized by a loyal party press and the breakthrough of radio; (2) 1961-69, a phase characterized by a loyal party press under pressure and the advent of television; (3) 1973-89, a phase characterized by the dissolution of the party press and the beginning of the television era. In the course of these decades, the media's coverage of election campaigns has been "liberated' from the political parties. This move from party-controlled to media-driven election campaigns coincides with a more turbulent period among the voters, a period with a higher level of voter volatility. The changing role of the mass media has not led to dramatic shifts in activity, at least not as reflected by two indicators - voter turnout and level of informal discussion. Recent changes in the pattern of voting participation, however, may be associated with the role of the media. The gap between centre and periphery in voter turnout has gradually been bridged, and television is probably one of the bridge-builders. With television as the major arena for the election campaign, the stream of political impulses which serve as mobilizing forces has become more or less the same in both centre and periphery.

election campaigns have grown as well? As this question relates to voters, we shall deal with it by means of two indicators: voter turnout and the level of informal discussion during the election campaign. Another question concerns the role of political parties as the main actors in the election campaign. Here the question is whether Norway reflects the international trend towards a decline of parties?

Two aspects can be discerned regarding the decline-of-parties thesis, one corresponding to the micro-level, the other to the macro-level. These aspects concern: (i) the relation between voters and parties, and (ii) the importance of parties compared to other actors in the political system.² In this case we shall restrict our comments to the first aspect. Evidence from election studies shows that party identification has remained fairly stable in the period from 1957 to 1985. More generally, the attitudinal component of partisanship can also be characterized by stability (Listhaug 1989). Election surveys from 1969 to 1985 indicate no decline in affective evaluation the parties: on the contrary, sympathy has risen slightly.

One can, all the same, register a certain downwards trend in party membership in Norway, a decline which becomes even more pronounced when the period is extended to 1989. This is almost exclusively due to a decreasing propensity among men to become party members. Among women, the proportion of party members has remained fairly stable. Even so, there are more men than women party members.³ Evidence from sources other than the election surveys, furthermore, more clearly support the decline-of-party thesis. The final volume of the history of the Labour movement in Norway, for example, describes activity in the party organization the 1980s as 'worrisomely low and with a falling tendency¹ (Nyhamar 1990, 542).

As the parties have become somewhat weaker, it seems logical to forecast that their potential to control the election campaign has diminished at the same time as campaigns have become more decisive for the election outcome. This prospect leads to a new question: Which actors have entered the scene and challenged the political parties? In this article we will emphasize the mass media which have gained an independent and powerful position in the election campaigns. Originally highly influenced by the political parties, the mass media today mainly follow their own course in an election campaign. Clearly, then, the relation between political parties and mass media has changed.

In order to grasp the changing forces concerning election campaigns in Norway, we shall first construct a model which highlights what would seem to be the most critical variables. On this basis we shall then look into the three channels of communication between parties and the electorate. This discussion leads to a change of focus to the role of the media - in particular

television - and the media's effect on voters. Finally the article looks at
the relationship between these changes and voter behaviour.

A Model of an Election Campaign

According to a traditional approach in communication research, a message goes from a sender to a receiver. In a campaign situation, political parties are the senders. They try to disseminate their message by different techniques in order to reach the voters, the potential receivers of the message. Various channels are used. We can distinguish between direct and indirect channels, as well as a third 'secret' channel.⁴ In the direct channels the parties meet the voters face-to-face, in public or in private. Such meetings occur in public via campaign rallies, party meetings, etc. and in private via door-todoor and in talks with committed party members.

In indirect channels the message is mediated through the mass media, which may be controlled or uncontrolled. Typical controlled media are pamphlets, advertising, posters and the traditional party press - which is, however, no longer as strictly edited in accordance with the party line. Partly as a consequence of the weakening of the party press, newspaper advertising has recently become a more common feature in election campaigns. uncontrolled media, by comparison, are television, radio and independent newspapers which are edited along journalistic and marketing lines. Recent developments in the mass media make it difficult to characterize uncontrolled media as exclusively a passive channel for sending a message. The media - especially television - have also become actors in the election campaign, as we shall discuss later.

Certain trends are evident in Norwegian election campaigns from 1945 up to recent times. The importance of direct channels has decreased, while indirect channels - and more precisely the uncontrolled mass media - have grown in importance. What makes television such a powerful medium is partly that - following our terminology - an indirect channel is transformed into a direct channel. Through television a voter meets the politicians in his or her own home.

Perhaps to compensate for a weakened position of the direct channel between voters and parties, the political parties have stepped up their use of polling. Polling is an instrument to gauge opinion in general, and among members and actual/potential voters in particular. This is the party's third and 'secret¹ channel of communication. In Norway parties have been using opinion polls since 1949, but with a substantial increase during recent decades.

Direct Channels - The Parties Meet the Voters

One form of direct contact is door-to-door canvassing - knocking on doors.

offering pamphlets and trying to persuade the voter. In the election surveys of 1957, 1969 and 1973, respondents were asked: 'In many places the political parties send their active supporters to people's homes in order to tell them about the election. Did anyone from any of the parties come to this house to speak with you about the election?' A decline can be observed in responses to this question; positive responses dropped from 7 percent in 1957 to 5 percent in 1969 and finally 4 percent in 1973, the last year for which information is available.

In the election surveys from 1957 to 1973 another question was also posed: 'Did you during the campaign attend any political meetings with lectures, discussions, entertainment, etc.?' Here the result remained approximately 10 percent for the entire period, the only exception being the 1957 election.⁵ This figure is lower, but we should note that the election campaign was interrupted by the death of King Haakon VII. With his death came a one-week halt in the election campaign, which probably explains the lower overall figure. All in all, we may say that approximately 1 in 10 Norwegian voters attended a political meeting in the election campaigns during the second half of the 1950s and up to 1973.

Since no surveys after 1973 included questions concerning attendance at political meetings, we cannot base our judgments on the election studies. For the period after 1973, however, it seems reasonable to suggest that there has been a decline in attendance of meetings arranged by the parties.

Indirect Channels - A Three-Step Development from Channel to Arena to Actor

The media occupy a strong position in election campaigns today, with a potential to influence the agenda and suggest solutions to specific problems. Television, as the most powerful medium in Norway, structures the election campaign, with the campaign rhythm in large measure following television programming. The official start of the campaign, for example, is usually the first election programme on television, which comes approximately one month before election day. The emergence of television, in short, has transformed the election campaign.

From a developmental perspective, it is possible to identify various stages in the period from 1945 to 1989. If we include not only the electronic media but also the printed media, where the newspapers predominate, the period can be divided into three phases:

- Phase I (1945-57): A loyal party press and the breakthrough of radio.

- Phase II (1961-69): A loyal party press under pressure and the advent of
television.

- Phase 111 (1973-89): Dissolution of the party press and the beginning of
the television era.

The role of the media in the election campaign differs, with a specific role in each of the three phases. Briefly, in Phase I the media served as a channel, in Phase II as an arena and in Phase 111 as an actor. This development is cumulative, i.e. when a new role is taken on, the old one still remains. The point is that the three-step development from channel to arena to actor follows the three phases.

As used here, 'channel' refers to a situation where the parties have control, where the media are an instrument for communicating a party message. As an 'arena', by comparison, the media have an independent role in directing the confrontation between the parties, and a function as gate-keepers determining access to the arena. In the role as 'actor', finally, the media are active players in the campaign, especially as promoted by the ideal of investigative journalism.⁶ When serving both as 'arena' and 'actor' the media are a battlefield for the content of the campaign. The primary question is: Which issues shall be brought to the forefront?

Phase I (1945-57) is characterized by a loyal party press. This is an extension of a tradition that dates back to the emergence of daily newspapers Norway at the end of the 19th century. These were times of political tension characterized by the struggle for parliamentary democracy and national independence from Sweden. In this process, the newspapers constituted important links of communication between voters and the newly established political parties. The largest and most influential newspapers became spokesmen for the various political parties, and during the election campaign the newspapers served as heralds for the parties. The press had become a channel for agitation and a link between the parties and the voters.

The period from 1945 to 1957 also saw the rise and heyday of radio.⁷ This new medium gathered a large audience. A survey conducted in connection with the 1951 local elections showed that three-quarters of the voters had listened to one or more of the election programmes broadcast on the radio.⁸ The radio was identical to NRK (the state-run Norwegian Broadcasting System). It had a journalistic profile based on a non-partisan platform and a strategy of non-interference in the election campaign. The ideal was a passive role in reporting the election campaign. The leadership of NRK in large part let the political parties hammer out a profile for the election programmes - the form of presentation, the range of election programmes, etc. Before the election campaign, representatives of the parties met the leadership of NRK to discuss these matters.

This pattern of party control remained more or less fixed from 1945 to

1965, albeit with a range of variations in the programmes.⁹ A slight change occurred in 1949 however. Before 1949 there were no open debates between the parties, as these were regarded as incompatible with the non-partisan profile of NRK. Programmes were restricted to presentations of the various parties in the form of speeches - a clear-cut example of the media serving as a channel. Beginning in 1949, the practice of party debates on the radio was established, with NRK journalists in a passive role. When the focus was on one party, the normal procedure was that other parties constituted a cross-examining panel.¹⁰ In discussions of various topics, moreover, the programmes were not to have any bias towards any party. Equality of access was the norm, and the problem was solved by allotting the same amount of time to each of the parties - a practice dubbed 'stop-watch democracy' in which the clock became the ruler.

In summary, during Phase I the media were a channel for the parties.
We can, however, observe a trend towards the role as an arena: the radio
gradually came to offer a platform for discussions among the parties.

Phase II (1961-69) is the period in which the mass media served as an arena. An important event, if not the important event, was the advent of television. Television coverage of the election campaign initially followed in old NRK tracks, with 'no interference' in the election campaign. Complaints were voiced, however, that this form of journalism was old-fashioned. Professional television journalists were reduced to passive timekeepers. Criticism grew stronger after the 1965 election, which has been described as 'the first television election' (Torsvik 1967, 143). A reasonable interpretation that such discontent was a consequence of the rapid ascent of television. Television now reached large segments of the population, providing extensive coverage of the campaign with specially assigned election To explain this, a brief historical sketch of the growth and development of television is in order.

The first regular television broadcasts in Norway started in 1960. During the 1960s a nationwide system of transmitters was constructed and television gradually spread throughout the country. The very first election in which television was involved was in 1961. But at that time the range of programmes restricted, and more importantly, only a small segment of the population owned TV sets. Large areas of the country were beyond the range of television reception.

Even in 1965 a large part of the country lacked TV reception, and according to an election survey, only about half of the households in Norway owned a television. By the election of 1969, however, this share had increased to three-quarters of all Norwegian households.¹¹ Norway was gradually linked together by a nationwide television network and united by television. By the 1973 election, television had the potential to

reach more or less the whole electorate.¹² With some modification, in fact, the spread of television corresponds with the various phases identified here - i.e. during Phase II the diffusion curve soared and in Phase 111 a state of saturation is reached. Consequently 1973 represents the beginning of a new stage, Phase 111, the television era.

Yet more important than whether people own TV sets is whether they watch election programmes aired during the campaign. To assess this we have singled out the proportion who watched one or more election programmes during the campaign, from 1965 to 1981. As shown in Table 1, the TV audience increased steadily from 1965 to 1977 and then levelled out, beginning with nearly half of the electorate and rising to almost 90 percent, a figure which includes both voters and non-voters. These general figures, of course, hide large differences in viewer habits, habits that range from watching every election programme to only a short look at one programme. The main point, however, is simply that from the 1970s on television became a medium with more or less the entire Norwegian electorate as its potential audience.

Radio also covers the election campaign, but in comparison with television, radio is a rather weak medium. Here we have one empirical indication. On election night, after the polling stations have closed down and the counting begins, the results are communicated hour-by-hour via both television and radio. Table 2 shows a time series from 1957 - an election without television - up to 1977. Two trends are observed: (i) the proportion of no exposure decreases from 48 to 16 percent; and (ii) radio has been losing ground. Only a scant minority now rely on the radio as their sole information channel.

As regards the printed media, the loyal party press as an institution came under pressure from the 19605, a development which has been related to market forces (Høyer 1964). Adaption to the reader market served to promote depolitization. Television, also, probably promoted a non-partisan press, as NRK has a journalistic profile based on a non-partisan platform.

Phase 111 (1973-89) is the period of the mass media as actor, with

reference to both the ether media and the press. In 1965 NRK election programmes were compared to a football match.¹³ The rules were set by the parties. NRK was to be an observer with an eye on the clock, and as an observer NRK could hardly have any effect on the outcome. By 1973, the start of Phase 111, the metaphor 'observer of a football-match' was no longer apt. Or, perhaps more aptly, we could say the observer jumped down from the stand and gained a position more powerful than a traditional football referee, who is firmly restricted by rules.

The changes came gradually; NRK became an actor in the campaign step-by-step. After coverage of the 1965 election, criticism against NRK triggered off changes, but not overnight. From the 1969 election onwards, for example, interviewers were no longer politicians but television journalists. journalists normally have this job, and they have become important in the campaign, just as much commented on and evaluated in the newspapers and among the public as the leading politicians. This break with the past became even clearer with the 1973 election campaign.¹⁴

The same three-step development in the ether media can be observed in the press, i.e. from a strictly edited party press {channel), via an arena where different views or parties can be confronted and a platform for discussion offered (although still biased towards one party), to the independent following its own course. A precondition for the role of an actor is the dissolution of the party press, and that is what happened in Phase 111. Some newspapers broke their formal party affiliation. But even in newspapers which retained their links with parties, political journalism became more independent. The traditionally strong connection between the Labour Party and its own party press, for instance, became much weaker.¹⁵

Compared to television, however, the press has played a more modest
role in recent election campaigns. The importance of television makes it
appropriate to mention some of its effects. Television has contributed to a

nationalization of the election campaign, offering one common arena for the campaign. Centralizing forces have been released emphasizing the nation as context, both vis-å-vis the audience and vis-å-vis the parties. This tendency has been strengthened due to the faet that Norwegian television has, up to the 1989 election campaign, been dominated by one channel, which means a centralization of the audience's attention.

Second, the image of a political party exposed through the television is focused upon the central leadership.¹⁶ The person who represents the party is at centre stage. As a result, personification has followed in the wake of television (Keeter 1987). In a television era, leadership of a party can easily become equated with the party itself. From the voter's point of view this situation tends to create an aura of intimacy regarding the leading politicians. A new trend has emerged, therefore, with politicians known by their first names. Major election debates, for example, have been ascribed as the 'Gro-Kåre dialogue' (i.e. based on Gro Harlem Brundtland and Kåre Willoch).

Television gives the politicians the ability to speak directly to the whole electorate. But what are the reactions and effects among the voters? Polling can give an answer, and this leads us to the third channel between political parties and voters, the secret channel of opinion polls.

The Secret Channel - Opinion Polls and the Parties

In Norway, public opinion polling started just after the Second World War. From 1949 on the Norwegian Gallup Institute - established in 1946 - regularly measured party preference in the electorate. During the first years, the clients who paid for these surveys were the two largest political parties, Labour and the Conservatives, and political parties controlled all use of opinion polls concerning party choice.

How seriously the opinion polls were studied among party strategists is difficult to say. But we do know that the secretary for the Labour Party, Haakon Lie, scrutinized the data carefully. From an early point, polling became an integral part of the party's campaign strategy (Bergh 1987). Over the past two decades, the political parties - especially Labour and the Conservatives - have increased their use of polling substantially. But parties are not alone in making use of opinion polls.

Media as Actor: The Use of Polling

New instruments in political journalism, such as polling, have given the
media an important vehicle in the role as actor. Opinion polls during the
election campaign can disclose unpleasant facts for the parties - e.g. a gap

between opinion at the grassroots level and among the establishment. The media can in this fashion challenge the legitimacy of claims made by elected political bodies. This form of 'direct democracy' can easily be presented as the purest form of democracy, and the political parties can, if not careful, be accused of distorting the free expression of voxpopuli. The media's use of opinion polls can, in short, not only decide the agenda and pace of the election campaign, but also restrict the power of the political parties and their freedom to choose their own course.

The use of public opinion polling has in any event shown a clear pattern of growth. At every new election the number of polls taken appears to increase. Use of opinion polls has been especially widespread in the daily press. In the 1989 election campaign, for example, television debates between two politicians were for the first time rapidly followed-up by polls in the daily press. Respondents were asked: 'Who do you assess as the winner of the debate?'

Such uses of polling, however, have attracted some criticism. The complaint been that attention around the issues under discussion disappears in favour of the candidates' image and the question of who was the loser or winner of that particular debate. This practice was seen as a typical example of what has been called horse-race journalism (Broh 1980). Indeed, polling favours this form of journalism. Some claim, in fact, that an election campaign with a lot of polling is reduced to a sporting event with the focus on who is to be the winner. The horse-race metaphor further emphasizes that attention is on the appearance and speed of the contestants rather than the content of the message. What is stressed is the exciting element in the election campaign as well as the candidates' image and personality, while differences on substantial issues are easily neglected. Complex issues become reduced to simple pro-con assessments. Defenders of polling, on the other hand, argue that this technique enhances public interest in the political process and thus serves democracy. In addition, they maintain that the assertion that the polls distract interest from serious and important issues is undocumented.

Up until 1966, the political parties which commissioned surveys monitoringsupport the various parties agreed not to publish the distribution of party preferences. Politicians feared that revealing the party distribution could affect voter preferences and interfere with the process of establishing party loyalty.¹⁷ The media's use of opinion polls remained internal rather than public.¹⁸ In 1969 one experienced the first parliamentary election campaign involving newspaper coverage of surveys that focused on party choice. The breakthrough in the media's use of opinion polls, however, came with the 1973 parliamentary election at the start of the television era. This election saw not only one polling institute, as in the previous elections, but three institutes, linked to three different newspapers. For the first time

polls became an important feature of the election campaign.₁₉ By the 1989 parliamentary election there were four national polling institutes with a regular and formal affiliation to newspapers. In addition there was a fifth institute with short-term contracts with regional newspapers.²⁰ Both radio and television made use of opinion polls.

From Party-Controlled to Media-Driven Campaigns: The Change in Voter-Models

In terms of campaigns, we can distinguish between two ideal types, the party-controlled and the media-driven campaign. The context surrounding the party-controlled campaign was a media situation characterized by a party press and radio heavily restricted by the parties in terms of political reporting. In this situation the printed media were a channel for the parties. Most newspapers, and certainly the most influential ones, had direct party links with the parties. Based on a newspaper's party sympathies, readers could know even before opening the newspapers what the paper would write about the campaign. One party would be emphasized as the most 'trustworthy', which defended the 'reasonable' alternatives. Newspapers thus served to confirm and reinforce old beliefs and attitudes.

From the 1970s onwards, the new media situation, with a powerful television, provided a challenge for the political parties. By means of television, parties could reach nearly the whole electorate. But this also involved communicating to a vast and unknown audience. In this situation, political advertising agencies and opinion poll institutes entered the scene with the alleged capacity to answer such important and difficult questions as 'How to hammer out a profile to reach the voters with a given message?' and 'Which social groups are reached by television programmes and by the various types of advertising campaigns?' The breakthrough of television as a mass medium in political campaigns, in short, brought in its wake a new profession, i.e. marketing experts and image-makers recruited from advertising agencies and opinion poll institutes.

Political marketing and public relations are of course intertwined, so 'PR men' both inside and outside the political parties now had golden opportunities. According to Stein Rokkan, 'In every political party of some size there is a latent conflict between organizers and public relations men" (Rokkan 1970, 423). The 'organizers' emphasize the party organization. the recruitment and training of party workers, whereas 'public relations men' focus on the general public: how to present a mass-distributed message? Which image and appeal has the most impact? Today it would seem that the public relations men have gained the upper hand, partly as a consequence of recent changes in the mass media and the new prominence of television.

The model underlying the political marketing approach highlights certain voter characteristics, while leaving others in the dark. Implicit in this approach is the floating voter. The voter is equated with the consumer. Voting, i.e. choosing a special party, becomes an act in a market, the political market. The various parties are engaged in a contest, trying to capture the largest possible share of the market.

This metaphor of the voter as consumer has several implications. Consumer can easily shift. The act of voting is not seen as a serious act reflecting deeply-rooted values, values imbedded as a part of the process of socialization and a result of earlier phases in one's lifetime. Instead, it becomes more of an act of impulse, just like selecting among available consumer goods.

Media-Driven Campaigns and Volatility

The floating voter is central to the political marketing approach. The Norwegian Election Studies programme make it possible to document trends in volatility and estimate the share of the voters who decide if they shall vote or not during the election campaign. As shown in Table 3 the proportion of late deciders has doubled in recent years, increasing from one-tenth in 1957 to one-fifth in the mid-1980s. True, the large bulk of voters had made up their minds before the start of the campaign, but even small margins can be crucial for the final outcome, and 20 percent of the electorate can indeed topple a pre-campaign election forecast. In the 1989 election the proportion of late deciders soared to a much higher level (42 percent), which is probably related to specific circumstances attached to this election.

Election surveys can also shed light on inter-electoral volatility, i.e. the
proportion who change party preference from one election to the other.

Results pertaining to this issue are presented in Table 4. Inasmuch as absentees are excluded, the move from voting to non-voting, or vice versa, is not taken into account. As we see, the share of shifters shows a peak in the 1973 election, which came in the wake of the referendum campaign regarding Norwegian membership in the EC. This campaign triggered considerable political turmoil and contributed to an increase in volatility. In subsequent elections, volatility decreased, but then toward the end of the 1980s it climbed to a level higher than ever before. Norway thus confirms the international trend involving increased volatility from the 1960s up to the end of the 1980s, but this increase is not a consistent one. The deviation - the peak in 1973 - reflects a specific national event, Norway's EC referendum.

The change from party-controlled to media-driven campaigns coincides with a more turbulent period among the voters. In a society in constant change, however, it is difficult to isolate the effects of the mass media. A simple causal relation between a new type of election campaign and increased voter volatility is difficult to examine empirically. But for sake of simplicity, let us sketch a relation between mass media and voter volatility and for a moment disregard all other changes which may occur concomitantly with television.

In order to underscore the societal changes implicit in a new media situation, we shall apply a generation approach. A political generation is based on common experiences in a rather early phase in life, a time when impressions can leave a stamp of lasting character (Mannheim 1968). The formative years for this are assessed as being from approximately 17 to 25 years of age, a period when the right to vote is acquired. Experiences from election campaigns during one's youth may leave behind an impression of what Karl Mannheim (1968) dubbed 'a natural view" - in this respect "a natural view' of how an election campaign is. There will certainly be a difference if the election campaigns one experiences are party-controlled or media-driven. These events can affect the establishing of voter habits.

For nearly all post-World War II cohorts, the first election campaign in
which they could vote occurred in a period with television. Norway's

postwar generations are the television generations. Since party loyalty tends to be established during a certain period of youth, and consequently the youngest voters frequently shift party preferences, we shall distinguish between two postwar generations - the first born in the period (1946-60) and the second after 1961.

The generation before the postwar generations were either children/ teenagers during the wartime occupation or were born during this war. Most of them reached voting age in the end of the 1940s and in the 19505, the heyday of the radio, a fact which substantiates the label 'the radio generation'. We may furthermore operate with two other generations, both of which reached maturity before the war, in a period without television and with only weak radio coverage. The election campaigns during the formative years of these generations were party-controlled campaigns. Thus we shall call them the party press generations.

We anticipate that the proportion of voters who decide which party to vote for during the election campaign will be especially large among the television generations and small among the generations from the heyday of the party press. This is in fact the case, as shown in Table 5. In every election to 1985, approximately 35 percent of the postwar generations have decided which party to vote for during the election campaign. For the oldest generations this proportion is around 10 percent, remaining fairly stable from 1957 to 1985 although with a peak in the turbulent 1973 election. The radio generation occupies an in-between position with a proportion oscillating around 20 percent. Yet it is important to note that the level for all generations - but most dramatically for the youngest - rose substantially in the 1989 election.

Unpredictable Election Outcomes and Increased Activity?

The increase in volatility seems to be related to the media-driven campaign, although it is impossible to substantiate any simple causal relation. What is clear, however, is that as a result of higher volatility, the election campaigns have been more decisive. Has this situation then led to higher activity among the voters?

The review of a changing pattern in election campaigns has pinpointed two contrasting features, both of which may have implications for voter activity in the election campaign. First, the flow of information has been greater. Television has penetrated all social segments, and printed, as well as electronic media are more widespread than before. Add to this a better educated electorate, and one can forecast an increase in political activity. Second, the traditionally strong grip of the political parties around the

election campaign has loosened. Control over communication channels during the election campaign has been dispersed. Relatively speaking, the parties have become weaker, but that does not necessarily mean that the position of the parties in absolute terms has been weakened. The empirical evidence here is, as previously noted, somewhat ambiguous, but overall it would appear that the decline-of-parties thesis is substantiated. The position of the parties as mobilizing agents in the election campaign appears to have been undermined. Consequently it makes sense to predict lower voter activity in the campaign.

In summary, we are left with two trends exerting countervailing forces with a possible neutralizing effect on voter activity. To investigate these relations, a test will be conducted on the basis of two indicators, one measuring a formal aspect of activity, voter turnout, the other an informal aspect, level of discussion.

Activity I - Voter Turnout

In an election campaign, a political party will have two aims: (a) to mobilize old adherents, and (b) to recruit new ones. The success of the first aim - mobilization - can be measured by voting participation, and the success of the second aim - capturing voters from other parties - by volatility.

Patterns in voting participation can be traced back to the breakthrough of political parties in Norway in the 1880s. In their study of voter turnout from the turn of the century and up to the 19505, Stein Rokkan & Henry Valen emphasize the gradual increase observed (Rokkan & Valen 1962). When newly enfranchised voters went to the polls for the first time, the probability for doing so again in the next election increased. There were clear patterns in this mobilization process. The gap between centre and periphery decreased. As recently as in the 19505, the periphery still lagged behind the urban areas in voting participation, but this was in decline.

Rokkan & Valen's analysis ends with data from before the advent of television. Does television mobilize the voter and thus function as an instrument to increase voter turnout?²¹ Apparently not in the most straightforward The spread of television does not seem to have increased voter turnout in Norway. During the introduction phase of television in the 19605, turnout did not rise in a distinct pattern. In most Western democracies, moreover, television has been widespread for many years, with the potential audience more or less identical with the whole electorate. Despite this situation, the overall trend in the voting turnout in recent decades seem to be downwards rather than upwards.

The 1952 presidential election in the USA and the 1965 parliamentary election in Norway have been characterized as the first television election in their respective countries even though television networks were not complete at those elections, either in the USA or in Norway. Density of television coverage varied from region to region. This situation provides possibilities for a research design to test the effect of television. Did people living in the television area go to the polls more - in relation to earlier participation level - than those outside the range of television reception? In the US case, Simon & Stern (1955) studied the potential effect of television in the 1952 election campaign in lowa, but found that no clear connection was discernible. In the Norwegian case, by comparison, Valen & Torsvik (1967) concluded that television apparently functioned as a mobilizing agent in the voting process.²² The effect was clearest and strongest in the periphery. In regions with dispersed population and normally political communication (e.g. little activity from the parties), television seemed to stimulate voter turnout.

To substantiate the mobilizing effect of television in the periphery we
may take a brief look at voter turnout in sparsely populated regions

compared to urban ones. In Figure 1, voting participation in local and parliamentary elections is shown during the period from 1945 to the latest elections. We distinguish between municipalities with less than 2000 inhabitantsand with over 50 000 inhabitants.

We see that in local elections, voting participation in sparsely populated areas in fact exceeded, albeit with small margins, the level in the cities in the two first decades of nationwide television reception. With respect to parliamentary elections during the same period, by contrast, urban voter turnout was still ahead of turnout in peripheral municipalities. But the gap was narrowing: in 1965 the difference was 8.7 percentage points, and by 1989 only 4.0 percentage points.

According to one interpretation, television is the great equalizer, levelling differences between centre and periphery. Television is a mobilizing in the election campaign, and it does not discriminate between the urban areas and the remote and sparsely populated periphery. If we go back to the days before the electronic media, differences between the centre and periphery were indeed clear regarding the stream of political impulses. Yet as television has become the major arena for election campaigns, process has been triggered which levels out centre/periphery differences in voter turnout. But this trend cannot be related exclusively to the introduction of television. Concomitant events are also of importance for example the development and expansion of the welfare state. The underlying philosophy has been that the welfare state shall not discriminate between centre and periphery. The welfare state has thus contributed to an equalization of living conditions along the centre-periphery dimension, and possibly political behaviour as well.

Activity II - Informal Political Communication

Increased media coverage makes the flow of information accessible to nearly everybody. How does the penetrating power of the mass media affect informal communication during an election campaign? According to one version, the mass media make people more passive, placing them in an isolated position as receivers, while the informal communication network is undermined. In this view, before the mass media corrupted the voters with 'direct information', we had a Golden Age of democratic politics. This 'nostalgia for the past' interpretation has been attacked by Converse (1962) however. Converse points out that a weaker information flow from the mass media may undercut informal political communication. Opinion formation presupposes information about issues, and this is often provided by the mass media. In addition, according to the two-step flow of communication, opinion leaders use mass media as an important means of conveying opinions. It is also well-documented that a high level of informal communicationcoincides

municationcoincideswith high consumption of mass media. Consequently
we could hypothesize that mass media stimulate informal communication.

From the 1950s up to recent years, exposure to the ether as well as the printed media has widened. Has this trend been accompanied by more intense informal communication? The following question has been posed in every election study from 1957 to 1989: 'We would like to know if you participated in political discussion or talks before the election this year. How often did you talk about the election with your family or with friends and acquaintances outside the family - would you say daily, a couple of times a week, seldom or never?' As shown in Table 6, no clear trend emerges from 1957 to 1985: the catchword is stability. The level of discussions in 1989, but inasmuch as this does not happen in tandem with a change in the media situation, it seems reasonable to attribute the increase to specific conditions in the 1989 election.

The stability in informal communication runs contrary to what one might expect in view of the explosive growth of higher education. As Converse (1962) has underscored, two trends have converged: (i) the heightened level of education has given a larger proportion of voters access to the printed media; and (ii) the breakthrough of the electronic media - which in fact are spoken media - makes electoral news accessible also for those not motivated to read. Taken together, it is therefore reasonable to expect a higher level of informal communication during the election campaign. And indeed, we may note a positive correlation between frequency of discussion and level of education. Other things being equal, in fact, discussion the election campaign would have shown a downwards trend if the education level from 1957 had remained frozen up to 1985.

Overall stability in the level of discussion does not imply stability in the
form of discussion however. Perhaps as a consequence of a larger proportionof

portionofundecided voters, the tone of discussions seems to have grown somewhat more aggressive. Attempts to persuade voters to vote for a particular party have increased (see Table 5). This increase is not a monotonousone a gradual upwards trend from 1965 to 1985. A deviation is observed in the 1965 parliamentary election, which is perhaps not so strange since the 1965 election was a 'mobilization election' which resulted in a governmental shift.

If this change in the form of discussion is related to fundamental societal shifts, it would be reasonable to suggest a generation profile. In Table 5 we have shown the proportion in the different generations who have been subjected to persuasion attempts concerning voting. A relatively clear generation profile is indeed revealed, with parallels to what was documented the late deciders. These two aspects are obviously interconnected. late decider is an attractive object for an agitator, but the late decider can also call on a true believer, a person who will provide unambiguous advice.

Conclusions

We have emphasized changing trends in the election campaigns during the postwar period in Norway. The relative strength of the political parties has declined, whereas an opposite trend can be observed for the media. The media have moved from being party-controlled to being more or less independent actors in the election campaign. The rise of consumer oriented media with extensive use of public opinion polling parallels the rise of political marketing - equating voters with consumers. The choice of a political party is seen more or less the same as choosing a consumer good. Consequently the media as an actor in the election campaign have the potential to nurture volatility among the voters.²³

Election campaigns are no longer party-controlled, but media-driven. However, this has not resulted in dramatic shifts in activity - at least not according to our two indicators. Voter turnout as well as the level of discussions have remained fairly stable. But changes which we have related to the new media setting are revealed in the pattern of voting participation and in the form of discussions. In voting participation there has been a narrowing or bridging of the gap between centre and periphery, in keeping with a notion of television as the great equalizer between cities and remote areas. The form of discussions also seems to have changed as the frequency of persuasion attempts among voters has increased substantially. A disclosure a generation profile in the frequency of persuasion attempts among voters, and also in the distribution of late deciders, substantiates fundamental changes.

A generation approach underscores that in every social process there are elements of both change and stability. And as the shifts often occur gradually or incrementally, they may easily appear untraceable if the timespan is not sufficiently long. In our case the empirical evidence is based on the Norwegian Election Studies which cover a period stretching from 1957 to 1989. The generation profile revealed in these data indicates that the current trend, with an election campaign more decisive than before (as a consequence of high volatility and a large proportion of late deciders) will grow even more pronounced in the years to come. The explanation is simple enough: feelings of belonging to a political party, with a loyalty anchored in traditions and unshaken by the conjunctures of an election campaign are attributes most often attached to the oldest generation. As the years pass, generations moulded by the period of party-controlled election campaigns are gradually disappearing from the scene.

---

NOTES

1. Elections follow a regular pattern, with elections being held every second year, either to the Storting (the Norwegian Parliament) or to local government.

2. This distinction is drawn by Selle & Svåsand (1991) who question the decline-of-parties thesis.

3. Figures from the Norwegian Election Studies illustrate the trend in the propensity to be a member of a political party. In 1957, 23 percent of all men were party members compared to 11 percent among women. (Collective trade union membership in Labour is excluded here.) In the 1985 election survey, corresponding figures were 18 percent for men versus 12 percent for women.

4. The main features in this model were originally formulated by Peter Esaiasson (1990). One revision is made, however, with the addition of a third channel.

5. Figures from the Norwegian Election Studies were 7 percent in 1957. 10 percent in 1965, 9 percent in 1969, and 10 percent in 1973.

6. For elaboration of these terms, see Østbye (1991).

7. During the Nazi occupation, radios were forbidden. Immediately after the war. the number of radio licences increased rapidly, but with some difficulties owing to import restrictions during reconstruction years. From the beginning of the 1950s it was possible to receive radio signals all over the country. It is estimated that by the mid-1950s nearly the entire population were radio listeners. The radio as a mass medium had penetrated all social segments. (NOU 1982: 30, p. 286, The Norwegian Power Study. Report on the Mass Media.)

8. Aftenposten, 15 December 1951.

9. As an illustration of party control we shall quote from a discussion in the Broadcasting Council (Kringkastingsrådet) 14 September 1961. A member with a Labour affiliation. Andreas Andersen, said: 'The leadership in NRK has in fact not been involved in the planning of the election programmes. The task has been transferred to the parties, and our views about the programmes must be directed to the parties'. In a meeting of the Kringkastingsråd just after the 1965 election-September 30-this form of party-control was criticized. The chairman Ivar Eskeland complained that "the NRK-expertise was reduced to timekeepers'. Head of the NRK, Hans Jacob Ustvedt. said that the political debates had been covered by NRK since the 1949 election, and that from that time there has been an agreement concerning election programmes that the parties and the leadership of NRK must come to terms*.

10. This was the case tor radio election coverage in 1949 and in most of the elections from then on. This form was also used in the television programmes in 1961 and 1965, with the different parties being cross-examined by representatives from other parties.

---

10. This was the case tor radio election coverage in 1949 and in most of the elections from then on. This form was also used in the television programmes in 1961 and 1965, with the different parties being cross-examined by representatives from other parties.

11. The exact percentages were 49 percent for 1965 and 75 percent for 1969.

12. The election surveys in 1973 did not ask about television ownership. The figure can be estimated, however, if we assume that the same proportion of voters who lived in a household with television watched one or more election programmes in 1973 as in 1965 and 1969. Following this procedure, the proportion of television ownership in 1973 was 89 percent. Data from the diffusion curve of television licenses supports other data sources. This curve was exponential during the 19605. By the 1970s the curve was still rising, but there was also a tendency to level out.

13. The metaphor was used by a member of the Broadcasting Council, Kjølv Egeland, in a meeting on 9 September 1965.

14. The new style was appreciated in newspaper comments. For instance, the Oslo daily Dagbladet wrote in an editorial article 'Honours to NRK' (8 September 1973).

15. At present the Labour press can openly criticize the Labour Party and carry news which the leadership may prefer to hide. The party which traditionally has been loyally defended no longer has a guarantee of a positive assessment. This independent style was clearly manifested in January 1981 in a turbulent phase in the Labour Party. Prime Minister Odvar Nordli had secretly declared that he wanted to resign. The party-linked Labour Press News Agency, (Arbeidernes Pressekontor) disclosed and distributed this news, which the party definitely wanted to hide (Nyhammar 1990, 345).

16. Peter Esaisson (1990), in a study of election campaigns in Sweden, has empirically shown that as television has become the most important arena in the election campaign, the focus on party leadership has become stronger. Party leaders more often than before externally represent and defend the parties. They provide their parties with a profile, and it may even seem that party leaders can be equated with the party.

17. The results of opinion polis were rapidly disseminated in political circles, but remained a secret there. By the mid-19605, however, it also happened that newspapers occasionally quoted the distribution of party choice. (Interview with the head of 'Norsk Gallup Institutt', Bjørn Balstad, 15 January 1991.)

18. In the 1953 election campaign, for example, surveys with a focus on party preference were sent from the Labour Party organization to editors in the Labour press. The statistical distributions were stamped confidential and were not to be published. This material was not meant for mass circulation, but was offered as an instrument in the editors' hånds in order to refine the campaign strategy. Source: The Labour Party Archives, Arbeiderbevegeisens arkiv.

19. The use of polling as an element in political journalism in the 1973 election campaign was particularly evident in the two newspapers KG and Dagbladet. The referendum campaign in 1972 concerning Norwegian membership in the EC was a catalyst for the breakthrough of polling by mass media in Norway. A referendum campaign with two contesting camps of relatively equal size is an ideal situation for such opinion polis. Two clear alternatives exist - yes or no - and there is no intervening chain between votes and result as is the case in an ordinary election where an electoral system translates votes into candidates. In 1972 there were two public opinion polling institutes which regularly reported the distribution of pro- and anti-EC sentiments.

20. These institutes and their affiliations were as follows: Scanfact for VG, Opinion A/S for Aftenposten, MMI (Markeds- og Mediainstituttet) for Dagbladet, Norges Markedsdata for Arbeidernes Pressekontor, and Norsk Gallup Institutt for short-term contracts.

21. Angus Campbell (1962) has posed the question as to whether television has raised voter turnout. He frankly admits that a scientific answer 'would require surgery of a delicacy we cannot at present perform' (Campbell 1962, 10). But after studying turnout and the diffusion of radio and television at the aggregate level, Campbell concludes: 'The advent of radio was followed by a general and significant increase of turnout in national elections; the arrival of television was not' (Campbell 1962, 12).

22. In the 1965 election, more than 40 percent of households had access to television, and television did follow the election campaign, with a total of about 25 hours of special election programmes.

---

22. In the 1965 election, more than 40 percent of households had access to television, and television did follow the election campaign, with a total of about 25 hours of special election programmes.

23. Some have questioned if there is any relationship at all between the mass media and volatility. A study in the US basedon the 1976 election campaign concludes: 'Contrary to expectations, the dominant direction of television exposure's relationship was toward lower levels of volatility' (Bybee et al. 1981, 69). However, such a result from one election campaign does not necessarily contradict the overall tendency that a mediadriven normally contributes to voter volatility.

Data sets from the Norwegian Election Studies 1957, 1969, 1977, 1981, 1985 were made available by the Norwegian Social Science Data Service (NSD). The data set from the 1989 election study is not yet available. However, the Norwegian Elections Study programme at the Institute for Social Research has kindly given me access to the particular distributions in question.

Professor Henry Valen has been in charge of the election studies. As regards data, the 1957, 1969 and 1973 surveys were conducted by the Norwegian Gallup Institute and the 1977. 1981, 1985 and 1989 surveys by the Central Bureau of Statistics. Neither the Norwegian Gallup Institute, the Central Bureau of Statistics nor NSD are responsible for my analyses or interpretations of the data. That responsibility rests entirely with the author.

REFERENCES

Bergh, T. 1987. Storhetstid (1945-1965). Vol. 5 in Arbeiderbevegelsens historie i Norge. Oslo:
Tiden.

Broh, C. A. 1980. 'Horse-Race Journalism: Reporting the Polis in the 1976 Presidential
Election', Public Opinion Quarterly 44, 514-529.

Bybee, C. R., McLeod, J. M., Luetscher. W. D. & Garramone. G. 1981. 'Mass Communication
Voter Volatility", Public Opinion Quarterly 45, 69-90.

Campbell. A. 1962. 'Has Television Reshaped Polities'. Columbia Journalism Review 1. 10-13.

Converse, P. E. 1962. 'Information Flow and the Stability of Partisan Attitudes', Public
Opinion Quarterly 26, 578-599.

Crewe, I. & Denver, D. eds., 1985. Electoral Change in Western Democracies. Patterns and
Sources of Electoral Volatility. London: Croom Helm.

Esaiasson, P. 1990. Svenska valkampanjer 1866-1988. Stockholm: Allmånna Forlaget.

Keder, S. 1987. 'The Illusion of Intimacy. Television and the Role of Candidate Personal
Qualities in Voter Choice", Public Opinion Quarterlv 51. 344-358.

Lie, H. 1975. . . . slik jeg ser det, Oslo: Tiden.

Listhaug. O. 1989. Citizens, Parties and Norwegian Electoral Politics 1957-1985, An Empirical
Study. Trondheim: Tapir

Mannheim, K. 1968. 'The Problem of Generation'. Essays on the Sociology of Knowledge.
London: Routledge & Kegan Paul Ltd.

Nyhammar, J. 1990. Nve utfordringer (1965-1990). Vol. 6 in Arbeiderbevegelsens historie i
Norge. Oslo: Tiden.

Pedersen, M. N. 1983. 'Changing Patterns of Electoral Volatility in European Party Systems.
1948-77', in Daalder, H. & Mair, P. eds.. Western European Party Systems. Beverly Hills:
Sage Publications.

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

Rokkan, S. & Valen, H. 1962. 'The Mobilization of the Periphery: Data on Turnout. Party
Membership and Candidate Recruitment in Norway", Acta Sociologien, reprinted in Rokkan
(1970).

Selle. P. & Svåsand, L. 1991. 'Membership in Party Organizations and the Problems of
Decline of Parties'. Comparative Political Studies 23. 459-477.

Simon, H. A. & Stern, F. 1955. "The Effect of Television upon Voting Behavior in the 1952
Presidential Election', American Political Science Review 49, 47(M77.

Torsvik, P. 1967. 'Valgkampen i fjernsyn og radio', Tidsskrift for samfunnsforskning 8, 143—
162.

Valen, H. 1981. Valg og politikk. Oslo: NKS-forlaget.

Valen, H. & Torsvik, P. 1967. 'Økningen i valgdeltakelsen ved kommunevalget i 1963 og
stortingsvalget 1965', Tidsskrift for samfunnsforskning 8, 187-218.

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

Østbye, H. 1991. 'Media in Politics: Channels, Arenas, Actors, Themes', in Strøm, K. &
Svåsand, L. eds., Challenges to Political Parties (in preparation).