Editorial Note. When we received Professor Barton's manuscript, "Occupational Class and Vote in the 1949 Norwegian Election, " we immediately realized how truly remarkable it was. His analyses are based on an election survey carried out in connection with the 1949 national election in Norway, i.e., only five years after the publication of Lazarsfeld, Berelson & Gaudet's The People's Choice. They were part of Professor Barton's doctoral dissertation from 1957, but have not been available to a broader audience. The first published results from a Norwegian election survey is Valen & Katz's Political Parties in Norway from 1964, covering the 1957 election, but now, thanks to Professor Barton's article, the Nordic political science community can go 8 years back in time. In our opinion, this clearly justifies the unusual practice of publishing an article with a 50 year delay!

The editors

Resumé

In 1948-49 the Sociology Office of the University of Oslo, under the guidance of Prof. Paul Lazarsfeld of Columbia University, initiated a series of studies on the Norwegian economic planning system. Part of this Planning Project was a national election survey in the fall of 1949, on a modified probability sample of 2600 people. The results show the paramount influence of economic class on voting, interpreting class as a combination of employer/employee relationship, property ownership, prestige-status, and income. Indicators of the homogeneity of class environment also played a role. Given the occupational structure of Norway, the Labor party had to win over part of the non-working-class population to obtain control of the government. They did this by winning about 1/3 of the white-collar vote, and about 1.4 of the farmers and fishermen. The policies of economic stabilization - control of inflation through price controls, wage controls, food subsidies, and rationing, and maintenance of full employment under conditions of "suppressed inflation" - were crucial to winning over these white-collar and farm voters.

Historical Background of the 1949 Election Study

The research reported here was part of a larger study of the problems of
economic planning in post-war Norway, the "Planning Project," carried out at
the Sociology Office of the University of Oslo. In 1948 the University of

---

* Allen H. Barton, 4441 SW 85 Wav, Gainesville, Florida 32608-4132, USA.

Oslo, with the support of a private group interested in social science, which later became the Norwegian Institute for Social Research, invited Professor Paul F. Lazarsfeld of Columbia University to initiate a program of training in social research at the University of Oslo. Professor Lazarsfeld saw in the Norwegian economic planning system a good object for a "demonstration project," in which many different research techniques could be employed in the study of a complex social issue. Having studied unemployment in Austria and the US in the Great Depression, and lived through the disasters of the Austrian socialist movement between the wars, he was particularly interested in studying what a social democratic government could do when in power.

The Planning Project initiated several types of research: a content analysis of newspaper coverage of economic issues (carried out by Frederik Fostvedt and Per Saugstad); an interview study of decision makers in a government planning agency (the Industry Directorate) allocating resources within the overall plan laid out in the National Economic Budget (carried out by myself with assistance of Fritz Holte), a study of the enforcement of price control and rationing (by Vilhelm Aubert), detailed interviews with citizens on their beliefs about the economy and the government's planning policy (by cand. psykol Per Mentzen), and a national survey of public economic attitudes and their relation to voting in the 1949 election. The Planning Project was supported by the Rinde family, and benefited from the close involvement of Erik Rinde. I worked on the project as Lazarsfeld's teaching assistant and as an Instructor at the University of Oslo, and continued the analysis on a Social Science Research Council fellowship on returning to the US.

This paper reports on the national survey. It is a shortened version of my dissertation chapters written in 1950-54 for the Department of Sociology of Columbia University (Barton 1957). To my knowledge, a detailed analysis of the 1949 election survey has never been published in Norway.

Sample and Field Work

The 1949 election survey was made on behalf of the Planning Project by the FAKTA Bureau for Market and Opinion Research, directed by Mr. Leif Holbaek-Hanssen. The questionnaire was designed by a group of members of the Planning project, including Erik Rinde, Allen Barton, Odd Ramsøy, and Leif Holbaek-Hanssen. Preliminary tabulations were done by the FAKTA agency in Oslo, and by Odd Ramsøy of the Planning Project staff. The analysis reported here was carried out by the author and is his responsibility.

The interviewing was carried out starting seven weeks before the election of October 10, 1949, and continuing about two weeks after it. The sample was planned as a probability sample, based on random selection from official lists of households. For practical reasons the interviewing was done in clusters of

five, one randomly selected household plus the four nearest neighbors. Almost 10 percent of the original sample had to be substituted after being repeatedly unavailable. The number of usable interviews was about 2600. In breakdowns of the opinion tabulations these are supplemented by 156 additional interviews from cities in the Eastern part of Norway, to raise the otherwise small numbers of business and white-collar people. This oversampling is not included in tables showing the structure of the electorate and overall responses. The selection of a respondent within the household was made on a mechanical, random basis.¹

Occupational Class and Vote

Party Structure and Class Structure

Twentieth century Norway, like most industrialized democracies (with the conspicuous exception of the United States), developed a strong working class party. The Labor Party was based in the blue-collar trade union movement, and emphasized governmental economic action for full employment, social security, and a more equal distribution of income. The "bourgeois parties" (Conservatives, Liberals, Farmers Party, Christian People's Party) were based on different groups of business, professional. and farming interests. These parties had somewhat different historical ideologies but were basically committed to less government economic intervention and less egalitarian goals.

The starting point of this analysis of the 1949 election lies in the occupational structure of the population. Based on our sample, the electorate consisted of the following main occupational groups: 43 percent manual workers, including farm workers and self-employed artisans; 24 percent selfemployed farmers and fishermen; 23 percent white collar employees; 10 percent businessmen. If one assumes that all employed persons, blue collar and white collar, are "working class, and all self-employed businessmen. artisans, fishermen and farmers are "property owners."" then there is a working class majority of about 2/3, and a "property-owning" class of about 1/3. However this simple "two-class" structure has never corresponded to voting behavior in Norway or anywhere else (Lipset et al. 1954.) If we maintain a distinction between the blue-collar and white-collar employees. and between business and farming/fishing property owners, as "occupational classes," the electorate does not have any class in the majority.

The basic question for this analysis is how the Labor party put together a coalition of "occupational classes" to win a majority. To refine the analysis we will distinguish further within each of the four main occupational groups. Within the blue-collar category we will distinguish those employed in farming, fishing and forestry from industrial and other workers, and those

who are self-employed manual workers. Within the white-collar category we will distinguish by income, which roughly corresponds to amount of skill and authority on the job. Differences in interest within the property owning groups based on their wealth or income will be examined by distinguishing small and large farmers and small and large businessmen by income. This gives eleven detailed occupational groups.

How the Occupational Classes Voted

The great importance of occupation for voting is clear (Table 1). Industrial workers and self-employed fishermen voted heavily for Labor; the whitecollar employees gave about 2/3 of their votes to "bourgeois parties"; farmers and small business gave 3/4 or more of their votes to "bourgeois parties'", and larger businessmen did so almost exclusively.

It will be noted that a good many who intend to vote refuse to reveal their intention, or are undecided, especially among white collar employees, small farmers, and business people. If we simply allocate the unknowns in proportion to those who reveal their intentions, the totals underestimate the vote for the Conservative and Christian People's parties as well as the Communist party. (See last two lines of Table 1 for comparison of the sample vote with the actual vote.)

An examination of the newspaper reading habits of those not reporting their vote intention within broad occupational groups permitted allocations of the unknown groups which suggested that the Labor party won 67 percent of the vote of all manual workers, 34 percent of all white collar employees, 27 percent of all farmers and fishermen, and 17 percent of all businessmen (Barton 1957. 314-16).

While occupation was a major factor in the vote. the extent to which the Labor party penetrated the "middle groups" is also striking - they won about one-third of the white-collar class. and a quarter of the farmer fisherman group. Even among businessmen Labor seemed to win between one-eighth and one-sixth of the votes - almost entirely among small businessmen. Only among larger businessmen was the Labor vote entirely negligible. The Labor party with some truth could claim to be what it set out to become in the L93os - a "people's party" rather than a class party.

The Class Composition of the Parties

The significance of the voting figures is further clarified by presenting the same data in another way: to show the occupational composition of the vote received by each party. (Table 2) This form of presentation reveals clearly the importance of the "middle groups" to Labor: over one-third of Labor's vote came from white-collar employees, farmers, fishermen, and small businessmen.

Farm workers and self-employed workers supplied another 14 percent.
Industrial and other non-farm employed manual workers made up only half of
the Labor vote.

The various bourgeois parties present interesting variations in their class basis. The two center parties, the Christian People's party and the Liberals, drew support from all groups. The Christian People's party especially, appealing on a largely moral-religious platform, came close to the occupational makeup of the population as a whole. The two right-wing parties, Farmers and Conservative, specialized in attracting the rural and the urban middle-class respectively. The Farmers" party ran candidates only in rural districts and was supported almost entirely by farmers and farm workers. The Conservative party got its largest block of votes from white-collar employees, with most of the rest coming from business. Since independent business owners were such a tiny minority of the population, even a "businessman's party" had to depend mainly on hired employees for its votes.

Taking the "bourgeois parties" as a whole, the two pillars of their voting strength were the farmers and the white collar workers, which together provided about two-thirds of their votes. The rest came about equally from businessmen and manual workers, each providing about one sixth.

These figures all come from voters in households where the "head" was in the labor force. About 9 percent of the sample were in households headed by someone who was retired, pensioned, widowed, or living on income from wealth. There was very little involuntary unemployment in Norway in 1949. The respondents from "not employed" households were mostly older people. They formed a particularly large proportion of two parties total vote: the Christian People's party (18 percent) and the Conservative Party (21 percent) compared with 8 percent of the total voters in the sample.

Stability of Party Attachment

The survey included several measures of stability of party attachment. Respondents were asked how they had voted in 1945, and those who supported a given party in that year could be classified as "defectors" if they shifted to another party, to undecided, or non-voting. Those supporting a party in 1949 could be classified as to whether they had voted for a different party in 1945 or had not voted then. Those who voted for a different party in 1945 could be classified as "cross-class" changers if they had switched from a bourgeois party to one of the left parties or vice versa. Respondents were also asked whether they were "a supporter of a particular party, or rather in doubt," and whether they had made up their mind how to vote within the election year.

The results of these measures show that the two clearly "occupational"
parties. Labor and the Farmers party, were the most stable, and the "middle

parties,"' Christian and Liberal, were much more unstable. (Table 3) The Conservatives were not as unstable as the "middle parties" but more so than the two "occupational" parties. The defections from the 1945 Communist party voters are severely underreported; the Communists got 12 percent of the vote that year, but only 5 percent of the sample admits to voting Communist then.

While there was a good deal of "defecting" from 1945 vote, most changes were from one bourgeois party to another, or from one left party to another. "Cross-class change" was quite low, and likely to be to one of the "center parties" if it occurred.

Factors Related to Deviation from Class Voting

Of the 13 to 20 percent of manual workers (depending on how one allocates refusals) who voted for bourgeois parties and provided them with one-sixth of their votes, about half were farm workers and self-employed workers, rather than hired industrial labor. The farm workers had some tendency to support the Farmers' party which their employers strongly supported; the selfemployed workers who did not vote Labor tended to be Liberals.

The voting behavior of industrial and other non-farm hired workers was examined to see what might account for their voting either Communist or for "bourgeois" parties. (Table 4) It is notable that the Communists got a higher vote among the better-paid workers than among the lowest-paid, and that the highest rate of voting for bourgeois parties was among the poorest workers.

To further localize the "left and right deviationists," these non-farm
workers were classified by sex, rural-urban residence and income. This leaves
a rather small number of cases in some categories. The following results seem

well-founded: The strongest Labor support came from the better-paid workers outside the cities. Large industrial plants in Norway were scattered over the countryside, each surrounded by a workers' community. In these the workingclass voter found a homogeneous social environment, highly organized by the unions and the Labor party (Lipset et al., 1954). The low-paid workers outside the cities, on the other hand, probably worked in small plants in communities dominated by farmers, small merchants, and small industrial employers; they were in a less homogeneously working-class environment. In the cities, the best-paid workers may have been in frequent contact with white-collar people and artisans. The lowest-paid city workers, almost exclusively women, tended to be service workers who would have close contacts with members of other classes. These groups too would have a relatively heterogeneous class environment, and tended toward "right" deviant voting. The especially high bourgeois vote among women of the best paid urban working-class households may reflect the middle-class aspirations which have been found in corresponding women in England and America. Unfortunately we have no direct data on subjective class identification.

The Communist vote was highest where the bourgeois vote was lowest - among workers in the higher pay groups outside the cities, and among urban male workers. It has been suggested that the Communists gained strength among the highest-paid workers because of their discontent with the wage

freeze and the Labor party-trade union policy of giving pay raises to the
lower-paid while making the better-paid wait.

The importance of status of occupation, as distinct from income, is shown by the great difference between manual and white-collar employees on the same income levels. (Table 5) Labor had indeed made important inroads into the white-collar group; nevertheless the majority of white-collar employees at all but the lower income level remained loyal to the "bourgeois parties." Various explanations have been advanced for this: their exposure to businessclass ideology in the course of their work, their status identification with the upper classes, a reaction against the traditional Marxian glorification of the manual worker at the expense of non-manual ("unproductive") occupations.

It has been suggested that the Labor party offered the manual worker not only economic betterment, but a collective increase in status. By organizing in unions and in the Labor party, the workers were able to place their leaders in positions of great power and prestige - bargaining equally with the leaders of business, occupying the highest governmental posts, serving as national leaders during the great wartime and postwar crises. The white-collar worker already had a feeling of relatively high status, and would not be attracted on this basis.

As has been found in most countries during this period, women were more conservative than men within each occupational stratum, and especially in the white-collar group. (Table 6) Women were notably more likely to vote for the Christian People's party. Since there is some reason to believe that women were not sufficiently represented in the sample, especially in rural areas, sex ratio figures for the various parties must be interpreted with caution. They do however clearly confirm facts well-known from other countries: that extreme

leftist parties are overwhelmingly male, that religious parties are mainly
female in their voting support, and that conservative parties also tend to draw
more female support.

Economic Policy and Occupational Group Interests

Economic issues were the most prominent topic in the mass media and in public perceptions of problems facing the country.² Economic policies were also the means by which the coalition of occupational groups supporting the Labor party was constructed. This is evidenced by questions asking whether the voters supported or opposed the government's economic policy, and what they saw as the advantages and disadvantages of that policy.

General attitude toward economic policy

The opposition parties were divided on their responses to the Labor government's economic policies. (Table 7) In neither of the center parties is there a majority of definite opponents of Labor's policies. Over 40 percent of both the Christian and Liberal voters were unable to say whether they were supporters or opponents, and a significant fraction actually supported the government's policies. The only parties in which an overwhelming majority was explicitly opposed were the Conservatives on the right and the Communists on the left.

Class Differences in Perceived Advantages and Disadvantages

To see which aspects of the government's economic policy were significant to the voters and how this influenced Labor support within various classes, the approval/disapproval question was followed by two open-ended questions: "Can you mention any advantages of this policy?" "Can you mention any disadvantages brought about by this policy?" The answers have been located

under the main headings of effects on production, effects on distribution, and objections to the "means of implementation" (e.g. "controls."" "bureaucracy"). The results are reported for male voters separated into Labor and bourgeois party supporters. (Table 8). The restriction of the analysis to men results from the extraordinarily high proportion of women - 50 percent overall - who answered "don't know" to the basic question of support or

opposition to the government's economic policy, and who had no answer to
its advantages and disadvantages.

The Emphasis on Distribution and Security

A general feature of the answers was the emphasis on distribution and security, rather than the effects of policies on production. The largest single category of advantages mentioned in each social group was "stabilization policy, equal distribution;" the second and third categories in all groups but one were either "full employment, social security" or "helps workers, low income groups." The exception was businessmen, who not surprisingly put "no strikes" ahead of "helps the workers" as an advantage. The "stabilization policy" refers to the government's strong policy of holding the line on inflation through price and wage controls, rationing, and food subsidies. The "full employment" policy refers to their refusal to use orthodox deflationary, unemployment-producing policies to get rid of inflation, but rather to tolerate "suppressed inflation" while rebuilding the economy's infrastructure with a high rate of investment.

The answers were not entirely clear-cut as between production versus distribution and security. The production emphasis was mainly found in the category, "better living conditions, reconstruction," but perhaps the references to "full employment" also imply "full production" as compared with the economic slump which followed the First World War. As to "disadvantages" we have also the question to what extent those who complained about taxes and controls were worried about their effects on production as distinct from their effects on the incomes of the upper classes or their interference with traditional ownership prerogatives.

A White-Collar Class Issue

On most items the white-collar group just took a middle position between the workers on one side and business on the other. "Stabilization policy, equal distribution" however was an unusual kind of class issue - one with its peak in the white-collar group. This supports the notion that the white-collar group had essentially a "consumer-interest" orientation rather than a producerinterest outlook. The issue of stabilization and equitable distribution must naturally appeal to a fixed-income group with weak economic bargaining power as compared to the other three classes. Lacking their own bargaining organizations or a naturally strong economic power position, they relied on the national government to protect them from inflation. This suggests the white-collar class playing the role of a "balancing group," checking the tendency toward a mutually ruinous struggle for larger shares of the national income among powerful producer-groups.

That this issue may have had an important causal role in bringing whitecollar support to Labor is suggested by the considerable percentage of whitecollar workers even among bourgeois-party supporters who mentioned it as an advantage of the government's economic policy. This was an issue which could penetrate even the screen of party prejudice and selective exposure to one's own party propaganda, to reach the minds of an opposition group. This is an essential condition for an issue to cause change in voting, rather than merely to generate appropriate rationalizations of an already existing party affiliation. "Full employment, social security" may also have been such an issue, since it was the second most frequently mentioned advantage among bourgeois-voting white-collar and business people.

---

NOTES

1. The use of one interview per household under-represented members of large households. However, checks showed that on most questions family size made no difference in opinions within occupational strata. Therefore no weighting appeared necessary, since figures are normally are given separately by occupational strata. In the table showing the occupational composition of each party's vote, where a difference would be made, each occupational group is weighted by its average number of adults per household.

2. A content analysis of editorials in the five party newspapers in Oslo during the four years before the election showed that economic topics were the topic of about half of all editorials (Barton 1957. 355-56).

REFERENCES

Barton. Allen H. 1957. Sociological and Psychological Problems of Economic Planning in
Norway. Doctoral dissertation. Columbia University. University Microfilms International.
Ann Arbor. Michigan. Mic 58-1331.

Lipset. S. M.. Lazarsfeld. P. F.. Barton, A. H. & Linz, J. 1954. "The Psychology of Voting." pp. 1124-75 in Lindzey. G., ed.. Handbook of Social Psychology, v.2. Cambridge. MA: Addison-Wesley. Revised version (1960) published as ch. VII. "Elections: the Expression of the Democratic Class Struggle." in Lipset. S. M.. Political Man. Garden Cit\. NY: Doubledav.