1. Our Problem

For more than a century the Danish press has in some sense of the word been a party press, the majority of newspapers wearing a party label and often being directly owned or supported by parties. Today political journalism is on the whole confined to the most important mass media: newspapers, radio and television. Periodicals are either entirely apolitical in outlook (the weekly magazines) or carry very limited circulations - apart from the trade union papers.¹ Though sometimes reproached of leftism, the Danish state radio and TV in principle administers scrupulously equal treatment of different parties in all programmes on domestic politics, it has actually been warned to do so by a court decision. All major parties are represented in the controlling board of the Danish Radio.

This leaves the newspapers as the most important channels for directed party propaganda. This article will try to answer a question which may be derived from this situation: To which extent is the party press a reality today? We shall divide it into two related problems: 1. How far is party journalism prevalent in the newspapers during election campaigns? and 2. Is there any reasonable correlation between the pattern of audience support for newspapers and the comparable support given by voters to the parties?

The relation between expressed opinions by voters in elections and the opinions expressed in channels of mass communication has been the subject of much speculation in this century.² The problem is a basic one because it concerns the nature of opinion formation and hence is a key to the understanding of the performance of political systems in the modern world. Revolutionary and authoritarian philosophies often maintain that government against the real interests of the people in systems with universal suffrage is only possible when education and the mass media is controlled by the Jewish, White or capitalist minority rulers.3

To all other theories of non-coercive stability in democratic societies, it remains
a puzzle that a complete gap can exist between voting support given to parties and
the circulation which newspapers in support of these parties have managed to win

---

* I want to thank professor Alex S. Edelstem, University of Washington, for inspiration and help rendered in preparing this manuscript.

* I want to thank professor Alex S. Edelstem, University of Washington, for inspiration and help rendered in preparing this manuscript.

from the public. A large proportion of voters choose newspapers out of accord with
their party preferences, although the daily press still contain the most important
reservoir of political information and comments which is available to them.4

A more palpable solution to this problem, would of course be that the actual bias excerted in political journalism does not correspond very much to the party label in newspapers. Recent findings at least point to a possible feed-back mechanism in the newspaper market, implying that editorial boards adapt to the changing nature of their audience: 'overspread' newspapers are inoffensive in their reporting while 'innerspread' newspapers are more aggressively supporting their parties5 It is tempting to conclude that the gap between comparable circulation and election figures to a great extent is compensated for by this difference in treatment of political events by newspapers. The 'innerspread' newspapers will reach a relatively larger amount of opinion leaders and shall thus in the ultimate process of opinion formation regain much of what is lost by a limited circulation. However, regarding the almost complete and instrumental selectivity in exposure by the audience, this comes dangerously near to dissolving the impact of a given intent in the communication.6 6

Clearly, the answers to these questions shall mainly be found in audience and group research. We shall draw on a different kind of data, namely content analysis and election and media statistics. By this procedure some of the many doors can be closed while others must be left open. This article is a summary of the author's book "Partipressen" (The Party Press), but it will also contain some further attempts to judge the validity of the results presented there.⁷

2. The Present Structure of the Press

The daily press in Denmark seems eminently fit for studies in the interdependence of politics and journalism. Its traditional structure was formed in the period between 1866 and 1914, essentially by the ambitions of four major political parties to acquire each their own press organ in all important cities throughout the country. This so-called classical "political four paper system" has been broken down by a severe process of concentration during the years after World War 11. Organs of all four parties have survived, however, and have preserved a rather close contact with their parties. The bonds between party and newspaper today are mainly to be found in the political affiliations of editors, shareholders and members of the boards of the publishing companies, and in the political press agencies distributing syndicated copy to provincial papers. Subventions are rare today outside the Social Democratic press, which is supported by its owners, the trade unions. More voluntary allegiances have kept the non-socialist papers in clearly definable positions vis-a-vis the parties. Most papers claim a definite political tendency in all official registers, and they also recommend the same party at elections.⁸ The only serious problem in registering the newspapers along political lines are two big Copenhagen noon tabloids, B.T. published by the Conservative Berlingske Tidende and Ekstrabladet published by the Radical Politiken. They formally adhere to the same parties as their mother morning papers, but do not actually show any party preference in their editorials.

They will consequently be registered separately here. Two other papers, Jyllands (Aarhus) and the financial daily Børsen (Copenhagen), do not support any definite party. But they have consistently taken a stand in the political struggle against the Social Democrats and their allies, and they will be grouped as "Antisocialist" papers here. Six papers are classified as non-partisan, namely Information and Kristeligt Dagblad (Copenhagen) and four very small local papers. There are two daily organs of small parties, in Copenhagen Land og Folk (Communist), and in the borderland Der Nordschleswiger (representing the German minority).

Adding that there are no borderline cases between dailies and other periodicals, and that "newspaper" here means daily papers with fully independent editing of total content, we can draw Table 1 to give a survey of the Danish press development since 1945:

During the period observed the Conservatives have lost nearly all their small papers, and even one big paper in Copenhagen; but their big provincials (plus the tabloid 8.T.) have expanded more than any others and kept the total circulation up pretty well. The Liberals have lost relatively few papers and the rest have grown steadily, but the party has remained unrepresented by newspapers in the Capital save for an abortive attempt 1946-49. The Social Democrats and the Radicals have seen their newspaper number halved and have been losers in circulation.

In general the change in the political composition of the daily press has been very moderate for a twenty-year period. Whatever change has occurred is due to a geographical-typological change in the press pattern. A general upswing of tabloids has taken place, but also an overall decline in the importance of the Copenhagen

papers, which have kept roughly the same circulation outside the metropolitan area, but have lost heavily inside it. A closer analysis reveals that the general newspaper structure has followed a very regular pattern of accumulation inside local areas: the stronger local newspaper has nearly invariably grown, while the weaker ones have ceased or at best stagnated. The number of provincial cities with local production of more than one newspaper has decreased from 33 to 9, whereas the number with only one paper has grown from 21 to 33. An increasing monopolization of newspaper markets in the provinces has emerged, and the loss of local competition has not been balanced by a growth for national papers: outside Copenhagen papers have kept a total sale of some 170,000 copies. The only other national newspaper Jyllands-Posten of Aarhus, has remained at 60-70,000. In Copenhagen proper the biggest ownership group - Berlingske Tidende with Berlingske Aftenavis and B.T. - has increased its share of total circulation from 33 to 52 % since 1945. The second largest group - Politiken and Ekstrabladet - has retained some 33-35 % while the share of smaller papers has decreased from 34 to 15 %.¹⁰

A process of this character leaves little room for ideological factors. A newspaper will grow or die according to its initial relative strength in its market, irrespective of its political colour. In fact it has been possible for conservative as well as liberal and radical papers to reduce their local competitors. But the liberal papers started with the biggest circulation in most districts, and the Conservatives had the upper hand in the five most populous cities, whereas the Radicals dominated only in 4—5 of the smaller areas and the Social Democrats none (probably partly owing to their ambition to cover all areas of the country). This explains the different fate of the groups as outlined in Table 1.

If political motives among readers have had only secondary effects on the recent development of the press structure, it is still possible that newspaper monopolization could have influenced the political trends in the electorate. The condition of this would, of course, be that the newspapers tried to utilize their positions for political propaganda. This condition will be examined first.

3. Content Analysis of Party Bias in Campaign Reporting

Introduction.

As the structural pattern of the press has been changed from political diversity towards near-monopolies in local communities it might be expected that the traditional party journalism would have given way to a more objective reporting and a more open attitude to contributions from other points of view. In most other types of content newspapers strive to achieve rapport with their readers by adapting to their tastes and views, and most Danish newspapers today have a majority of their readers in support of opposing parties. It should be noted also that the development in recruiting and training of journalists is unfavourable to partisanship, and that journalists frequently move from socialist to liberal papers, from radical to conservative etc. The general estimate - confirmed by the introduction of "open" columns in most newspapers since 1945 - would be, that there has been such a shift towards less one-sided journalism.11 But it would of course be of interest to see how

far this development has gone. For this purpose the Institute of Press Research and
Contemporary History, attached to the Danish School of Journalism in Aarhus,
asked me to analyse newspaper content during the election campaign of 1964.

Intending to present a picture of the Danish press, it was essential to include several newspapers and also to deal with more than one type of content. The main period examined was the last week of the campaign (16-22 September) with a control period at the start (1-7 September). As the number of newspapers and the subject matter were not the same from one period to another there cannot be a comparison between individual papers. But it should be possible to give an account of the Danish press as a whole and of different political groupings. The sample includes all the biggest newspapers except "Ekstrabladet" which proved difficult to classify according to categories of traditional journalistic treatment. It represents between 40 and 50 % of total week-day circulation.

The result of the analysis, which should be considered a first attempt to elaborate tools for further measurements, will be summarized here. The presentation will follow the analysis in treating each aspect separately, as the technique is necessarily different from field to field.

The campaign

First, however, there should be a brief note on the political situation in the campaign. Social Democrats and Radicals had been allies in a government coalition since 1957, while the Liberals and Conservatives in declared alliance formed the opposition and the parliamentary alternative. Backing the right-wing opposition, but not recognized as an organized member, was the more extreme, antisocialist party, the Independents. Likewise, the Socialist Peoples Party (S.F.) formed a Frondeur party to the left of the Social Democrats. From the extreme left the Communists (D.K.P.) made attacks on all sides, as did the Henry George-inspired Retsforbundet (Rf) from the centre. Neither of them had or gained seats in Parliament. Two still smaller parties, plus the German Minority's "Schleswigsche Partei", brought the total number of parties up to nine. But the battle was clearly between the two blocs - Social Democrats plus Radicals versus Liberals plus Conservatives; traditional, though by no means automatic groupings. Schematically the positions can

be conveniently ordered as in Table 2, which will serve as a key to some of the later
tables.

Quantitative bias in reports from political meetings

The reports from political meetings were the focus of the analysis.

At the traditional "cornered" meetings, speakers from several parties - nearly always including the four traditional ones (S, RV, V, KF) - take part. Each report of such meetings found in twelve newspapers during the last week of the campaign (16-22 Sept.) was measured, recording what amount of column space (excluding here headings and sub-headings) had been allotted to report the individual speakers. For these twelve papers 105 separate news reports were found (with a total space of 66,210 mm). For a control, reports in eight other papers were measured during the days 16-19 September, finding here 56 reports (44,270 mm). Disregarding three non-commited papers, the results for seventeen party papers are summarized in table 3.

There was, it is seen, a strong general tendency for party papers to favour their own party in the allotment of space. Further, there was certain difference between the different political newspaper groups, but not a very large one. The differences found between individual newspapers were more striking. A provincial edition of the Social Democrat Aktuelt gave eight times more space to its own party than to any other, whereas the Conservative Aarhuus Stiftstidende only gave 30 % more. None of the 17 papers, however, differed from the tendency as such.

To this it should be added that reports from one-party meetings were given almost exclusively from meetings held by the newspapers' own parties. Sixty-seven one-party meeting reports were found in the 17 party newspapers during the last campaign week, but only three of these reported from other parties' meetings. These millimetre lengths should of course be added to the space found in multi-party meetings, as is done in table 4.

This contains some modifications to Table 3, especially that the Conservative
papers have risen from bottom to above middle. The reason is that big papers -
Berlingske Tidende and Aarhuus Stiftstidende, both Conservative, but also the Radical
Politiken - carried more single-party meetings than others, whereas the reports
from multi-party meetings were generally more frequent in smaller papers.
As it might be expected, an analysis of passages printed in bold types plus headings
and sub-headings showed that bias present in "bare" text emerged in a reinforced
shape here. In multi-party meeting reports the seventeen papers gave 70 %
of such space to their own party during the last campaign week, and the difference
between groups was somewhat smaller — Conservative papers gave 65 %, Liberals
76 %, Radicals 65 % and Social Democrats 81 %.
Another dimension of analysis was time. Nine of the papers from Table 3 were
also measured during the control period 1-7 September (the first week of the campaign).
The results are presented in Table 5.

This contains some modifications to Table 3, especially that the Conservative papers have risen from bottom to above middle. The reason is that big papers - Berlingske Tidende and Aarhuus Stiftstidende, both Conservative, but also the Radical Politiken - carried more single-party meetings than others, whereas the reports from multi-party meetings were generally more frequent in smaller papers.

As it might be expected, an analysis of passages printed in bold types plus headings and sub-headings showed that bias present in "bare" text emerged in a reinforced shape here. In multi-party meeting reports the seventeen papers gave 70 % of such space to their own party during the last campaign week, and the difference between groups was somewhat smaller - Conservative papers gave 65 %, Liberals 76 %, Radicals 65 % and Social Democrats 81 %.

Another dimension of analysis was time. Nine of the papers from Table 3 were
also measured during the control period 1—7 September (the first week of the campaign).
The results are presented in Table 5.

Here we observe a general intensification of the partisanship during the campaign. In bare text the average paper gave twice as much space to its own party as to any other, but in the last week this proportion had risen to 3:1. In headings the bias rose even more, to a 6 : 1 relationship, though it should be noted that the figures for the first week were rather uneven.

Qualitative dimensions of bias

The use of headlines and bold type presentation for propaganda purposes supports
in a limited way the hypothesis, that imbalance in measurable quantities reflects a

qualitative imbalance of the text. To prove this in a more material way, however, is extremely difficult. Any but a purely stenographic report is of course based on selection and interpretation by the journalist - at best on his sincere judgment of what is important for the readers and how it can be translated into written form. Any attempt to judge the bias of reports must necessarily contain non-objective criteria, as no objective standard of importance or relevance is available in political questions 15

In order to measure this dimension of bias several procedures were tried out, one
of which will be reported on here:

By 1 September three economic experts, forming the presidium of the Advisory Economic Council, issued a report containing positive as well as critical judgments on the recent development and future perspectives of the Danish economy. This document was summarized by nearly all papers. It was decided to have two coders choose the eight points in the expert report which most clearly expressed satisfaction with economic development and confidence for the future and eight other points of a correspondingly critical or negative tendency. Then it was seen whether or not these 16 points were found in the newspaper reports. The scores were computed by giving each item one point, positive or negative. Underlining with bold type gave one point extra, and the use of an item in a headline one further point. On this basis a seven-point scale was constructed. The result appear in Table 6.

Despite the informality of the procedure there emerges a very clear and positive result. The report of the "wise men" was reported with a definite leaning consistent with the views of the paper. All eight left-wing papers stressed the hopeful aspects except one radical (which simply printed one half of the text, but chose the second half, which happened to be the "wrong" one). All 15 papers on the opposite side showed a preference for negative aspects, except for Jyllands-Posten whose report was very brief.

One-sidedness in opinion sections

A promising approach was the analysis of quotations from other papers. These were
classified according to the allegiance of the cited paper to the two main blocs: a
government bloc consisting of Radical, Social Democrat and non-daily trade union

papers, and an opposition bloc consisting of Conservative, Liberal and Anti-Socialist papers. Next the quotations were divided according to the viewpoint expressed relative to the party allegiance of the quoting paper: Support of the quoting paper's party or bloc, attacks on opposing parties and bloc and attacks on own wing party were counted as supporting; opposing quotations were those doing the opposite; neutrals included any treatment of opponents' wing parties and isolated recommendation of an allied party.

Ten papers were searched during the first and the second week of the campaign and another eleven during the days 16—19 September. The total number of quotations found was 586. Leaving out the figures for independent papers, relatively uninteresting here, the results are shown in table 7.

It clearly emerges that even when newspapers are relatively liberal in deciding who is going to be quoted they still have a lot to say about the actual content of that column. Everywhere the majority of controversial quotations support the editorial policy. The number of daily papers supporting the government bloc is so small, that it may be difficult to go on finding good quotes in them. The Social Democrat papers got over this by making extensive use of the trade union papers. The Radical papers actually printed more quotations from opponent parties than from their own camp; but the selection process ensured a general support for Radicalism. It may be noted also that the difference between first and last week of the campaign is insignificant here.

Contributions from readers form a related kind of newspaper content. A search of eleven papers during the last week of the campaign for all articles or letters from readers (excluding the newspaper staff and also politicians) gave a total of 254 units. According to a classification similar to the one applied above, 159 of these supported the views of the paper, 38 went against them and 57 were neutral.¹⁸

The contributions from politicians — candidates, organization leaders and secretaries
- were observed in the same eleven papers and during the same period. One
paper, the Anti-Socialist Jyllands-Posten gave completely equal space to all parties.

The same paper had proved a deviant from the general partisanship in other areas, too, which was somewhat surprising in view of the violent tone in its own editorial expressions. The ten party papers had 68 out of 105 contributions from partisans. In terms of space the papers' own parties received 73 %, the party's ally 13 % and the opposing parties 8 %.¹⁹

A number of other editorial areas, less appropriate for systematic research due to spasmodic appearance, nevertheless showed the same clear tendency (portrayals and interviews of politicians etc.). Generally the newspapers gave equal opportunities for all parties to advertise, although not necessarily equal treatment in price and facilities or even full freedom in expression (as some party officials claimed). A survey of ten papers during the last week of the campaign showed that although some big papers carried nearly as much advertising for another party as for their own - or even more (Politiken) — this was not generally the case. Of the total space recorded, 51 % specifically recommended the papers own party.²⁰

4. The Party Press and Voting Patterns

From the analysis of press performance up to the Danish 1964 election it can be concluded that party papers in Denmark still have a strong tendency to support their own party during campaigns. There are some differences in degree, though not easily registered, as no aggregate scale of behaviour in the different sections of newspaper content can be built on non-arbitrary weights. Generally there seemed to be some tendency for the Social Democratic papers to display more partisanship than the non-socialist papers, and for the Radical papers to be less biased than normal. But the dimension innerspread-outerspread as such was not clearly indentified: Newspapers in quasi-monopoly positions did not generally differ in tendency from more competitive units, whether these were strong or weak. On the whole variations from the general tendency seemed mostly to be of an individual character.

But in all cases these were only variations in degree. There are no systematic studies showing whether Danish newspapers confirm their political allegiances as clearly between the elections as during the campaigns, but in the absence of further evidence it may be assumed that the totality of political comments and news offered by a given newspaper contains a bias corresponding in tendency to the official party label of the paper.

When this is the case some general research in the role and effects of mass media
seems possible on the basis merely of official election statistics and the surveys of
newspaper sales.

The most obvious representation is the relation of circulation figures to voting for the country as a whole. In doing so it is advisable to eliminate the tabloids altogether, because their party affiliation is problematic as outlined above. To eliminate the influence of casual shiftings the share of the parties in the elections have been averaged for 1945, 1947 and 1950 and for 1960, 1964 and 1966. Correspondingly the week-day circulations have been averaged for the years 1945 (second half only), 1948 and 1950 - and for 1960, 1964 and 1966 (table 8).

The left wing as a whole has gained in votes but lost heavily in circulation, whereas the right wing parties have increased their circulation much more than their voting strength. Of the individual parties the Social Democrats have gained in votes but lost in circulation, while Liberals have had the opposite experience. Even if Radicals and Communists have suffered as well in press as in elections, it is clear then that the appreciable gap between the parties' shares in total votes and their supporting circulation, far from being closed, rather tends to be widening.

This conclusion is of limited interest, however, owing to the increasingly monopolistic
character of the newspaper structure: a party may have its circulation concentrated
in two or three areas, whereas its voters come from all over the country.

The subdivision of circulation figures for smaller areas is not too accurate. But for the 45 provincial cities with more than 10,000 inhabitants a firm basis can be established for these last years, and hence also a correlation between the relative strength of the four traditional parties and the coverage of their newspapers. A product moment correlation coefficient on (a) the household coverage of the party's newspapers - local and national, but excluding tabloids - and (b) the percentage of votes cast at the election in 1964, shows the following results:²²

The relationship may also be stated in another way. For each of the four parties the 45 cities are divided into three groups - 15 where the party got relatively more votes in the other cities, 15 where it got relatively fewer and 15 where its share was "normal". The figures dividing into these three groups of course differ from party to party. For the Social Democrats the normal group are the 15 cities in which the party received in 1964 between 50 and 52.4 % of the votes cast, whereas the

"plus" group consists of 15 cities in which the party got 52.5 % or more, and the minus group of the 15 cities where the share was under 50 %. Similarly the "normal" group was 3-4 % for the Radicals, 12-17 % for the Liberals and 17 to 21.3 % for the Conservatives. This situation is surveyed in Table 9.

It is clear then, that there exists a clear correlation between newspaper circulation and voting behaviour in Denmark as far as the four traditional parties are concerned: Where a party has strong newspaper support it probably also gets more votes than elsewhere. Between the parties, however, there are the very clear differences outlined above - that the Social Democrats enjoy quite moderate press support even though they are nearly invariably the largest party in the city areas, while the Liberals in particular are generally strongly over-represented in the daily press of the provincial cities. The explanation for this could be that newspaper coverage varied from one social stratum to another and hence from party to party. But according to the existent readership surveys this factor is generally unimportant.²⁴ Social Democrats are simply much more often subscribers to newspapers of opposite political views than the supporters of right wing parties, which have strong press support.²⁵ The regional co-variance of circulations and voting patterns for the individual parties therefore cannot be correctly interpreted as a sign of any causal relationship, because of the tendency towards structural single unit press dominance following a cumulative pattern, which has been explained earlier: local traditions, supported perhaps by social structures, may create or maintain a party stronghold and at the same time have given that party's newspaper the initial superiority inevitably leading to its dominance over local competitors. Thus the co-variance simply results from one common historical basis finding two different expressions, i.e. in voting and in newspaper structure.

Changes in voting and circulation patterns

The dominant problem is of course, whether the press structure as such can have exerted any influence over the trends in voting behaviour. One way to study this would be to apply techniques similar to the ones used above to the question of relative growth or decline of the parties. The reasoning here is that one may compare two elections, both giving results fairly consistent with the results of the surrounding elections, to see where the shares of the individual parties have risen or declined more than usually. For the post-war decade, September 1953 presented such a wellbalanced election, and at the election of September 1964 the four traditional parties also kept quite close to their "usual" share.

The measure of relative growth or decline of the four parties in different cities has been defined here as the percentage received by a party of the votes cast in the 1964 election divided by the equivalent percentage received in the 1953 election. As the Social Democrats received 45 % of the votes cast in Viborg city in 1953 and 46.5 % in 1964, the party's "relative rate of growth" in Viborg will then be 46.5 : 45 X 100 or 103. A measure for the total press exposure during this period is impossible to calculate exactly on available data, but because of the rather mechanical structural development of the provincial press it is reasonable enough to take the household coverages for 1964 as representative of the press situation as a whole during the years 1953-64. Tables giving figures for 1955 are quite inaccurate for smaller papers, but still they confirm the suggestion that the press coverages of the four traditional parties have not changed radically during the period - the main trend, of course, being that strong coverage in 1964 implies strong, though not equally strong, coverage in 1955.26

A product moment correlation coefficient on this material shows the following
correlation between (a) the press support of a party and (b) the relative growth
or decline in its share of votes cast in 1953-64:27

It may be noted that the same procedure, based on comparison between the relative
shares of the four parties in 1950 and 1964, gives a very similar result.28

The problem can also be attacked by dividing the 45 cities into three groups of 15 for each of the four parties, one "plus-group", where the party has grown more or declined less than usually, one "minus group", where it has fared relatively badly, and one "zero-group" registering about the normal amount of growth or decline. For the Social Democrats the "zero group" are the 15 cities in which the party's share of votes cast in 1964 was less than 104 % but more than 97 % of the 1953share. For the Radicals the "zero-group" was between 74 and 65 %, for the Conservatives between 125 and 106 %, for the Liberals between 92 and 101.7 %. The confrontation of these groupings is shown in table 10.

In addition the correlations in table 10 will show that the relationship between

the relative growth or decline in voting strength of the traditional parties and their local press support has on the whole been - if anything - of a negative character. The evidence is not of any compelling strength - much weaker than the correlation between actual party strength and circulations as outlined earlier. But as it points to the opposite conclusion to what might have been expected, it does show that other factors have been able to reduce the final effects of newspaper party propaganda to nil. And the minus in the correlation figures does call for some sort of an explanation.

This explanation is found in a certain tendency towards gradual levelling of local differences in political adherences, some tendency for party strongholds to weaken with the passage of time.3o Economic change and internal migrations have no doubt played an important role here. For example the Conservatives and the Social Democrats seem to have gained a little by industrialization in the countryside, whereas the Liberals and Radicals have lost by the depopulation in the agrarian sector and found in return more opportunities in the cities now invaded by the sons of their old electorate. Other factors might be mentioned - but the reasons for the levelling are largely irrelevant here. It can be stated that there is such a tendency, and then concluded that it will logically run counter to the structural development in the daily press. Newspaper circulations and strength of the corresponding political parties, which earlier in this century showed rather high correlations,³¹ have since been subject to quite different, indeed diverging tendencies. A newspaper, which was strongest in its own area, when the cumulative concentration process started in the 19205, would normally have survived and increased its local dominance. But the political (voting) correlate of the initial strength of the newspaper has been influenced inter alia by the levelling process just mentioned, so that a party would normally grow less or decline more where its press is strong. In a way this means

that newspapers have quite often become reflections of the political patterns of bygone ages. About 1920 the Liberals were the strongest party throughout most of the Danish provinces. Since then the party has declined markedly, but the liberal newspapers have grown and now hold monopoly positions covering some 66 % of the territory. And though the Social Democratic party continued to grow for some decades after 1920, the earlier upswing of its daily press gradually ceased from about that date and was replaced by complete stagnation. It should be added that none of the newer parties, which have been generally expanding since the twenties, have as yet succeeded in establishing any viable daily press.

That accounts for the divergence of circulations and voting. Table 10 thus contains no proof that backing from large circulations has done the political parties any harm during campaigns! But the analysis still reveals that political change has not been any compelling reason for the newspapers to change their outlook. It also reveals that newspapers have not been able to preserve outdated patterns in the elections.

But just as the press may have influenced the political developments in many other and not less important ways than by shifting votes, it is also possible that newspapers may have modified or delayed in some way the disappearance of such outdated political patterns - "outdated" referring here only to the voting inclinations of the readers. An example of such outdated patterns would be the historically conditioned conglomerations of Radicals in different parts of the country — a declining but still observable phenomenon. In most of these Radical strongholds the dominant newspaper has also been of Radical party colour. But it is possible to find a number of constituencies where this has not been the case — where the Radicals though originally numerous at the elections had only weak support in the press, i.e. today as a rule none at all. The following table tries to compare the decline in Radical votes for areas, where the press situation differed markedly (table 11). It goes as far back as 1924, because it might reasonably be expected that any influence of the press should be felt most strongly in the long run.

There seem to be notable differences in timing here. Between 1924 and 1939 the decline was markedly weaker in the areas with good or medium press support (groups A, C and D) than elsewhere - especially compared to strongholds with weak newspapers. But since then the situation has been rather the opposite, as will be seen from table 11:

One cannot then exclude the possibility that the party's powers of resistance were stiffened for many years by its newspapers, and it would presumably be legitimate to speculate on the effects of new media - television was introduced around 1960 - in this situation. But anyway the ultimate result has been much the same all over the country, irrespective of the press support.

This, of course, could only be an example. But it does show that newspaper support has no automatic "effect" on political elections in the long run ("effects" understood as measurable positive results of newspapers' propaganda effeorts to influence the distribution of votes between parties) - even in case of a true centre party with an ambiguous position in social class antagonisms.

A case study: The Referendum of 1963

Another point for observation of press influence in Denmark is given by the referendum on four land bills - containing rules for pre-emption, indemnities etc. — held in June 1963. The Social Democrats and the Radicals had passed the bills in the Folketing and of course recommended them to the voters, supported by Socialistisk Folkeparti and Communists. Whereas the other parties, Liberals, Conservatives and Independents, urged people to vote "no". The campaign was rather vehement, party papers taking sides as strongly as usual, and it was believed, afterwards, that the outcome was influenced by the superiority of the no-parties in money and circulation. The proposals suffered a crushing defeat, only 28-31 % of the electorate voted "yes" as against 41-45 % "no". At the general elections of 1960, 44 % of the electorate had voted for the yes-parties, only 39 % for the no-parties, and the following general election in 1964 showed very closely the same distribution of votes.

Clearly there had been a major defection of voters from left-wing parties at the referendum in 1963.33 This defection may be measured as the difference between the yes-parties' percentage of the electorate (average of the elections in 1960 and 1964) and the percent share of yes-votes for the "Bill of agricultural estates" in the referendum, expressed as a percentage of the said general election share. The defection from yes-parties would then be 37 % in Greater Copenhagen, 37.5 % in the rest of the Danish Islands and 45.5 % in Jutland. But there are greater and interesting differences between the individual constituences, and it would be of interest to see whether these correlate with the varying degrees of press support for the yes-parties.

Though the circulation areas used in the present Danish media statistics do not
coincide with the constituencies — and referendum results have only been published
for these - it is possible to compute approximate coverage figures for most individual

constituencies. A few areas — Greater Copenhagen, Aarhus, Odense and Aalborg - which could not be subdivided with respect to newspaper coverage, will be represented here by their respective number of constituencies (29, 3, 2 and 2) all registering the average percentage of coverage and "defections" in the area. A general survey of the 126 constituencies can then be given in a few figures (Table 13) and in graphic illustration (see Chart on page 161).

Contrary to an earlier observation by myself on more limited materials,³⁵ it must be admitted that there exists a certain correlation. The referendum result was somewhat less unsatisfactory for the yes-parties in areas, where they had superior or just moderate press support. The product moment correlation coefficient between the relative size of the defections and the household coverage shows minus 0.55.³⁶ As will be seen, especially from the accompanying Chart, there are serious deviations from this main tendency. The yes-result was bad in Rudkøbing (—45(^—45 %, despite satisfactory coverage (46 %), and it was at its best in Odense (—32(^—32 %} though the coverage here was only 12 %. But on the whole it is evident that most of the really bad results were registered where the government had very thin coverage - under 10 % or so.

Again it seems as if press conditions could possibly have played some role in the outcome, but only on the extreme margin. The plebiscite was thoroughly also lost by the government in areas where the opposition press held only 20 or 30 per cent of the circulation. And this was a situation where voters were "on their own" to an unusual degree — and took the chance to choose for themselves. But even in the face of such temporary breakdown of traditional party and class loyalties, the newspaper readers obviously went elsewhere for ultimate guidance than to the editorial columns of their local daily.

5. Conclusions

The Danish press, traditionally a party press, is moving into a new structure of semimonopolies
inside local areas.

It can be clearly observed that it still reacts as a party press during campaigns -
favouring its own party to a high degree.

For a long time the discrepancy between the political inclination of the readers

and the tendency supported in the editorial and news columns of the daily press has
been widening.

There is no sign of any change in this situation, as the influence of the press on
distribution of votes between traditional parties can be only marginal, and as newspapers
thrive or perish irrespective of their political accord with the readership.

Even in less usual situations of political decisions by the voters - such as the referendum
of 1963 — people will not always be influenced very much by the advocacies
of the daily press.

In sum, while the traditional build-up of the Danish press has been strongly influenced by political preferences in the public, press developments and voting trends of the last decades give no visible sign of consonance-seeking from people in their double roles as voters and as subscribers to political newspapers.

---

NOTES:

1 For a review of Danish media and audience data see and Kiihl et1966.

2 Kihlberg, 1960, p. 220 f f., Tingsten 1949, pp. 253-69, and Andersen, 1947, p. 57 ff.

3 Schramm, 1956.

4 Direct evidence for Denmark is rather meager on this point. Gallup article No. 35 (1960) shows that, out of those voting for the government parlies (S, RV and Rf) 28 % were readers of government papers only, 43 % read opposition papers only, while 9 % read both newspapers in support of the government and the opposition. The remaining 13 % read "other papers". This should be taken with usual caution, but the aggregate circulation and voting figures tell roughly the same story - see Table 8 and footnote 29.

5 Janson and Westerståhl, 1958. "Innerspread" newspapers means newspapers which on the whole do only reach party followers, while "overspread" newspapers also have managed to include a large proportion of voters from other parties in their audience. See Høyer's article in this volume on differences in the performance of party commitments between these two categories of newspapers.

6 See Berelson, Lazarsfeld & McPhee, 1966, chapter 10, and Edelstein, 1966, chapter IV with recent literature.

7 Thomsen, 1965.

8 12-18. See alsoandl964.

9 Registration of "newspapers" (units with independent production) from analysis of actual content, circulation figures for units outside the Danish ABC from "Fællesindkøbsforeningen"

---

etc. - The material was prepared by the author more completely as a report to the Danish press commission working since January 1967.

10 See the author's annual reviews in Pressehistorisk Årbog, Copenhagen, 1963, 1964, 1965, 1966, 1967. See Thomsen 1960, and Furhoff 1967.

11 See 1963.cit. 11222and1956.

12 Thomsen, 1965, p. 73 cf. p. 43. Three very small parties are omitted.

13 Ibid., p. 73 f.

14 Ibid.

15 An attempt was made under this project to compare the stenographic reports of the parties' television programmes during the first week of September and the newspaper reports of these programmes. See Thomsen, op. cit., p. 75 f., cf. p. 48 ff.

16 Ibid., p. 81, p. 58 f. Note that one point (No. 8), being positive as well as negative in meaning , registers in both groups under "items recorded". The group "conservative" calculated wrongly in the text of "Partipressen" (p. 59) is corrected here.

17 ff.

18 Ibid.. D. 79. D. 54 ff.

19 Ibid., p. 80, p. 55 ff.

20 Ibid., p. 80, p. 56 f.

21 The percentage of votes received by each party has been calculated as simple average of the percentages received in the three elections 1945, 1947, 1950 resp. 1960, 1964, 1966. The same procedure has been applied for the circulations, whose source foundations are explained in note 9.

22 With x — votes received in per cent of votes cast 1964 and y = household coverage of party press in 1964 we have: Soc. Rad. Lib. Cons. All x 51 4.2 15 19 22.4 Sx 5.3 2.4 6.4 3.8 18.05 y 18 10 45 16 22.2 Sy 11.6 20.5 26.0 22.4 24.85 Sxy 33.6 40.7 105 42.6 43.52 Sx-Sy 61.5 49.2 166 85.1 448.5

23 For the figures see Partipressen, p. 71. The table of the book (p. 28) has been rearranged. The in the first column from thehas been madeof 3000 votes cast.

24 See e. g. Gallup figures for 10 provincial daylies in 1963 in "Media 63" (Copenhagen 1963) p. D 74.

25 See Gallup article No. 35, 24 Sept. 1960, which shows that 44 % of the Social Democratic voters were reading opposition (Conservative or Liberal) newspapers only, while only 30 % were readers of government (Social Democrat and Radical) newspapers; the equivalent figures for Conservative and Liberal voters were 75 to 7 and 71 to 2, for Radicals 32 to 26. This is roughly consistent with the total voting and circulation figures in table 8.

26 Eberlins bladliste 1956 later editions in₁₉₅₉ and 1960.

27 The weighting of figures for all the parties in the extreme right column is based on the same scale as in table 9 (see note 23). With z = per cent of votes cast in 1964 divided by per cent of votes in 1953 and y = household coverage of party press in 1964 we have y and sy as in note 23 and then: All All Soc. Rad. Lib. Cons. unw. w. T" 100 70 100 120 98 97 S_z 5.3 14.5 20.2 25.7 25.4 23.8 Szy -13.7 -113 -143 -121 -26.1 -26.2 S_z-S_y 60.4 297 525 576 625 616

28 The correlation between (a) per cent of votes in 1964 divided by per cent of votes in 1950 and (b) percent of household coverage of party press in 1964 - calculated exactly as in note 28 - showed Social Democrats —0.18 Radicals —0.46 Conservatives —0.09 Liberals —0.40

29 Weighting as in Table 9 (see note 24). Apart from the change in basis election (from 1950 to 1953, which comes closer to the average of immediate post-war elections) the group limits in circulations have been modified from Partipressen, p. 30. This gives clearer insight for some parties, less for others. But the averages added should clarify the total picture.

30 The product moment correlation coefficient for x (the relative local strength of the four parties in 1964) and z (the change in relative local voting strength 1950-64) has been calculated as —0.14 for the Social Democrats, —0.36 for the Radicals, —0.43 for the Conservatives and —0.38 for the Liberals. That is overnormal election support in 1964 indicates to a mild degree that this election support has been weakened over fourteen years. - For an example of longer range moves see table 12.

---

30 The product moment correlation coefficient for x (the relative local strength of the four parties in 1964) and z (the change in relative local voting strength 1950-64) has been calculated as —0.14 for the Social Democrats, —0.36 for the Radicals, —0.43 for the Conservatives and —0.38 for the Liberals. That is overnormal election support in 1964 indicates to a mild degree that this election support has been weakened over fourteen years. - For an example of longer range moves see table 12.

31 This tendency emerges clearly from materials collected by the author for a larger manuscript examining structural developments in the Danish press 1870-1965.

32 The eight constituencies in group A are Roskilde, Lejre, Holbæk, Nykøbing, Middelfart, Rudkøbing, Skive and Vinderup. The eight in group B are Kalundborg, Ringsted, Bogense, Kjerteminde, Otterup, Højrup, Viborg and Kjellerup. In all these the percentage of Radical votes was over 16 % in 1924, and except Roskilde and Viborg over 22 %. Group G includes only the three "storkredse" of København-Frederiksberg, not the suburbs. I have consulted approximative circulation figures for newspapers in the areas (data collected for another study), so even if coverage figuresexist only from 1955—63 there is every reason to suppose that the division into areas with strong and weak support holds true even for the 19205. In some areas of group E (Østsjælland and Møn, etc.) the press support was at times very strong, but never reliable, while in another (Helsingør) it was always lacking in intensity.

33 A Gallup article, 1963 No. 32 (21.9) illustrates this as follows (cf. No. 31). S.F. Soc. Rad. Lib. Cons. Of active voters voted "yes" 62 87 52 3 0 Of active voters voted "no" 38 • 13 48 97 100 Percent of "electorate" voting 84 81 74 90 91 The distribution suggest a somewhat higher participation than actually registered, especially on the "yes"-side, but seems reasonable in toto.

34 Defection is computed as where x is the votes received by yes-parties (S, RV, SF, and DKP) in per cent of the electorate taken as average of the general elections in 1960 and 1964 — while y is the per cent of the electorate voting at the referendum in 1963 (specifically on the "Bill of Agricultural Estates", which got the least favourable response of the four technically independent bills, though the difference was very small).

35 Thomsen, op. cit., p. 31 ff, cf. p. 72.

36 For defections we have an average of 45.5 % and a standard deviation of 6.81. For coverage of households by yes-press we have an average of 20.1 and a standard deviation of 13.9. Data sources: 1. Media statistics a. Oplagsbulletinen 1946. Copenhagen, 1946 - the annual bulletin for net circulations of the Danish A.B.G. - and subsequent editions up to 1967. b. Eberlins bladliste 1956. Copenhagen, 1956 and subsequent editions to 1960. c. Oplagstal og markedstal 1964 (mimeo.) Copenhagen, 1964 (b. and c. contain available coverage data). d. "Media 63". Copenhagen, 1963) and subsequent editions - annual bulletins of advertising and readership surveys. e. Gallups artikler 1959-64 (mimeographed sheets). f. Private materials (see notes 9 and 32). 2. Official statistics a. Statistisk Årbog 1967. Copenhagen, 1967, and earlier editions back to 1926. b. Statistiske meddelelser 1965 and 1954. Copenhagen. 3. Newspapers 23 daily newspapers for the days 1 to 22 Sept. 1964. (For composition of sample, differing from field to field, see Thomsen, op. cit., pp. 75—82).

REFERENCES

Andersen, Martin: Avisen i dag (The Newspaper Today). Erichsens Forlag, Copenhagen
1947.

Andersen, Martin: Avisen (The Newspaper). Berlingske Forlag, Copenhagen 1964.

Berelson, Bernhard, Lazarsfeld, Paul F. & William N. McPhee: Voting. Phoenix edn.,
The University of Chicago Press, Chicago 1966.

Edelstein, Alex: Perspectives in Mass Communication. Harck, Copenhagen 1966.

Furhoff, Lars: Up plage s pir alen (The Circulation Spiral). Aldus/Bonniers, Stockholm 1967.

Janson, Carl-Gunnar &Westerståhl, Jørgen: Politisk press (Political Press). Tidningsstatistik
A/8., Gothenburg 1958.

Jansen, Torben Bo: Massekommunikationsmidler i Danmark (Mass Media in Denmark).
Paludan, Copenhagen 1967.

Jensen, H.: Journalistik (Journalism). Institut for Presseforskning og Samtidshistorie,
Århus 1963.

Kihlberg, Leif: Dagens Nyheter och demokratins genombrott II - Seglar sin egen sjø 1911-1921
(Dagens Nyheter and the Breakthrough of Democracy). Bonnier, Stockholm 1960.

Kiihl, P. H. et al.: Fritidsvaner i Danmark (Leisure Time Activities in Denmark). Social-

forskningsinstituttet, 1966.

Ree, Knud in: "Avisen idag" (The Newspaper Today). Copenhagen 1956.

Schramm, Wilbur: Four Theories of the Press. University of Illinois Press, Urbana 1956.

Slangerup, H. & Poulsen, E.: Newspapers in Denmark. Det danske Selskab, Copenhagen
1953.

Thomsen, Niels: "Bladdød og koncentration" (Newspaper mortality and concentration). Copenhagen,
Pressemuseets årsberetning 1960.

Thomsen, Niels: Partipressen. En undersøgelse af danske dagblades politiske bindingei
(The Party Press. An investigation of political allegiances in Danish daily newspapers).
Institut for Presseforskning og Samtidshistorie, Århus 1965.

Thorsen, Svend: Den danske Dagspresse I-11. Det danske Selskab, Gad, Copenhagen
1947-51.

Tingsten, Herbert: in Hal Koch, ed.: Nordisk Demokrati (Nordic Democracy.) Westermann,
Copenhagen 1949.