Resumé

The main purpose of this article is to argue the importance of combining several
approaches in studies of communist parties: socio-economic structures,
competitive relations to other left wing parties, organizational implantation, and
the role of political tradition. The present ecological analysis will particularly
emphasize the organizational approach and the importance of political tradition. In
keeping with Allardt's and Lipset's theories, socio-economic structure seemed to a
great extent to be a necessary condition for communist strength in this period.
Both significant employment in industry and/or forestry and a strong social
democratic party turned out to be necessary conditions for communist strength,
but only when we analysed the 'historical' and organizational dimensions in
more detail were we able to determine more accurately the areas of strong
electoral support of the Norwegian Communist Party.

The, Norwegian Political Landscape 1945- 1950

The political scene in Norway in 1945 looked quite different from what it did in 1936. The left parties increased their share of the vote from 42.8 % to 52.996. This was due to the increase in the vote for the Communist party (NKP), from 0.4% in 1936 to 11.9% in 1945: by contrast, the Labour Party (DNA) declined from 42.5% to 41.0%. The NKP success in 1945 represents one of the greatest changes that has ever taken place between two elections in Norway, but it was one that the party could not sustain; its vote in 1949 dropped to 5.8%, and the party has been in decline ever since.

If we compare this situation with that in other West-European countries
during this period, we find that it was not atypical. Norway followed a
general trend common to all countries without a very strong communist

---

* This article presents some of the main results from my thesis on the Norwegian Communist Party during the period 1945 - 1950 (to be published by Universitetsforlaget 1982). I wish to thank Einar Berntsen, Stein Kuhnle, Lars Svåsand and Derek Urwin for valuable comments.

party, and with a dominating social democratic party on the left.¹ In several countries there was a strong increase in communist support in the immediate postwar election followed by a rapid decline. The most deviant feature in the Norwegian case was the success of DNA in 1949, increasing from 41.0% to 45.796. Only Denmark, the country with the greatest social democratic decline in 1945, witnessed a similar social democratic gain in 1949.

It was only in countries with strong communist parties (France, Italy, Finland, Iceland, and perhaps Luxembourg) that these parties were strong enough to face the powerful pressures against communism, both nationally and internationally, that began around 1947.²

Important national and international factors can explain the general decline at the national level, but not the structure of the decline in different local communities. Communist voting is promoted by certain common characteristics, and these will vary among communes. The main theme of this article is to pinpoint the reasons why communist strength and communist decline vary between distinct areas. In our ecological analysis we will particularly emphasize an organizational approach and the importance of political tradition.

The Theoretical Standing₃

Theories on communist voting raise a number of problems. Traditionally, most widely held theories on communist voting attribute special importance to socio-structural factors and are part of traditional modernization theories (Allardt 1964, 1967, 1970, Lipset 1968, Dogan 1967, Tarrow 1967, 1968, 1977). These authors see mass support for communist parties as a periodical phenomenon and as an intermediary stage of development in societies undergoing rapid socio-economic change on the path towards modern highly industrialized capitalist society. To vote communist means that in one way or another you are displeased with your situation not only materially, but also psychologically: a modern working class votes for social democratic parties.

Another highly socio-economically oriented theory is represented by Korpi (1971, 1978). Contrary to Lipset and Allardt, who tend to consider communist voting as less rational than voting for other parties, Korpi takes the opposite stance: voting Communist is a highly rational act. Korpi's argument is based on general behavioural theory, according to which voting is a rational act aimed at utility maximization in a purely material sense. If it is a fact that communist parties both in their political platforms and in their propaganda demand socio-economic equality with more vehemence and intransigence than other parties, it is in the

material self-interest of working-class voters to vote Communist. But to
focus on rational man is not sufficient: contextual and historical factors
must also be taken into account.

In recent years, an increasing amount of criticism has been directed also at the purely socio-economic models. While both Dogan and Tarrow partly incorporate historical and organizational dimensions into their schemata, this is especially true of Laulajainen (1979a, 1979b, 1980), who stresses that organizational strength and political tradition are the main variables explaining communist or social democratic support in Finland. According to Laulajainen, the regional variations in communist support upon which Allardt bases his typologies are primarily explained by political traditions maintained for the most by local communist organizations. Allardt's argument that communism after World War II lacked traditions in northern and eastern Finland thus seems strange to Laulajainen because, according to him, it was precisely in northern Finland that the communists as early as the beginning of the 1920s took over the control of almost all organizations of the old unified working-class movement (Laulajainen 1979, 62).

I find Laulajainen's approach of great value. One should not only place an emphasis on variables at an individual or structural level, but also look at the 'transmission belts' between the individual and society. On the other hand, Laulajainen seems to go too far when he rejects the importance of socio-economic structure. A certain socio-economic structure is a necessary condition for socialist support, but never a sufficient one. Morever, Laulajainen fails to look at the interrelationship between electoral and organizational strength, political tradition, and socio-economic variables.

Some electoral support is required to maintain an organization, and
this support is obviously related to socio-economic variables.

To sum up, we have tried to combine the Allardt - Lipset approach and the Laulajainen approach (Selle 1980). We have thus singled out four main approaches, for which separate sets of hypotheses were developed and tested.

- A sociological approach concerning the relationship between socioeconomic
structure and communist electoral strength.

- A political approach where the competitive aspects between NKP and
DNA were analyzed.

- An organizational approach consisting of an examination of the relationship
between organizational implantation and communist electoral
support.

- A 'historical' approach focusing on the relation between communist
strength and previous political tradition.

Here I shall concentrate on the organizational and 'historical' explanations. Finally I shall present some of the results of the path analysis which links these four approaches. But let me start with a brief summary of the findings from the sociological and political approaches (Selle 1980: chap. 6 - 7).

In keeping with Allardt's and Lipset's theories, socio-economic structure seemed to a great extent to be a necessary condition for communist strength in Norway in this period, and partly also a sufficient condition for the strength of DNA. All communes with strong communist support had an important industrial and/or forestry sector, but in a great number of communes with these characteristics the party was without any significant support. DNA has always been relatively strong in such communes. The most important structural difference in electoral support patterns between the two socialist parties was that support of DNA was less negatively correlated with agriculture and the primary sector in general. The party had strong support even in more rural areas. However, we discovered important regional differences for both parties, which will not be discussed in this context.

A look at the structure of decline in communist support between 1945 and 1949 permits no clear conclusions to be drawn. However, there is a tendency that the decline was smaller in communes with an important industrial and/or forestry sector. Where the 'basic groups' were most numerous, the decline was less marked.

With regard to the relationship between the two socialist parties, we formulated two hypotheses that may be boiled down to the following 'ideal' models: the substitute model and the cumulative model. According to the substitute model, the parties constitute alternatives to each other. The cumulative model implies that the parties are strong in the same areas. Strong positive correlations would indicate that the cumulative model is most appropriate. Negative correlations would imply that the substitute model fits in best.⁴

We found a relatively strong positive correlation between the two socialist parties both in 1945 and in 1949 (.38 and .50), although regional differences exist. In the strongest areas of Norwegian communism the correlations are low, and in the strongest communist region, Finnmark, correlations were markedly negative (-.82 and -.37). Only if we consider the Norwegian polity as a whole, does the cumulative model seem to be most appropriate, and more so in 1949 than in 1945. NKP was a serious

competitor to DNA only in a small number of communes in 1949. The relationship between the two socialist parties is summarized in Figure 1. In 1949 there was a tremendous shift from cell 1 to cell 2, and also a moderate shift from cell 4 to cell 2.

One cannot draw any clear conclusions concerning the structure of communist decline. General socialist support was important in explaining communist support by itself, but not so important for explaining the decline. There was nevertheless a tendency for the party to suffer a more modest decline in communes with strong communist support in 1945, in the same areas in which DNA had its strongest support. The stronger the total socialist support, the smaller the decline. However, DNA was not able to absorb the total communist decline. In several communes in 1949 the total socialist support was lower than in 1945.

As we have seen in this brief summary of the socio-economic and political approaches, both significant employment in industry and/or forestry and a significant support of DNA were necessary conditions for communist support. But, only when we introduced the 'historical' and organizational dimensions in addition, were we able to determine more exactly the area of strong electoral support of NKP.

Organizational Support and Electoral Strength 5

Theoretical Introduction

I shall start with a simple functional model, which suggests that it is of importance that the party is visible in the local community. The stronger the organization, the stronger the electoral support. This also means that we expect communist support to be substantially higher in communes with a party organization than in communes lacking local organizations. We do not suppose, however, that the existence of an organization accounts for the whole difference. The causal relationship between these two variables is somewhat problematic. Although organization no doubt enhances electoral support, organizations are by themselves indicators of strength. The existence of organizational infrastructure must be explained by factors other than organizational ones. We believe that a certain minimum level of political strength is necessary for an organizational apparatus to be set up in the first place, and that the differences in electoral support between communities with organizations and communities lacking organizations will tend to increase over time. And, the stronger the organization the stronger the difference.

In this part we shall not only compare the results in communes, with and without organizations. We shall also try to indicate who the members of the party were and analyse the relationship between membership strength and electoral support. We are not able to determine who left the party first, the voter or the member. The data only give us the number of members in 1945, 1946 and 1949. But, generally speaking, we assume that when a party is in a process of disintegration, losing both members and voters, the members are likely to leave the party later than the voters. One will spot a snowball effect: with a decline in electoral support, it becomes less profitable to maintain 'marginal' organizations. The process prevents the mobilization of new voters (Selle andSvåsand 1981b, Svåsand 1981).

A General Review of the Organizational Development of theNKP 1945-1950

Socialist parties in general, and communist parties in particular, have always emphasized the organizational aspect. The building of effective organizational networks has historically and ideologically been one of the most important tasks of such parties. This implies not only strong party organizations, but also youth organizations, organizations for women, organizations for children, and an extensive participation in various 'front' organizations. The build-up of a strong press and an effective internal communication system have also been of great importance. At times, a strong organizational network has been seen as equally important, or even more important, than the extent of electoral support.

NKP experienced a tremendous increase in organizational strength just after World War 11. While the party in the thirties had between 2,000 and 5,000 members, the party claimed over 34,000 members in 1946. As early as in the summer of 1949, the party was back at 16,000 members. In spite of the tense national and international situation, the party still had more members in 1949 than after the split in the Labour movement in 1923.

From the end of the twenties and the beginning of the thirties most of the newspapers that had become part of the NKP press after the split ceased to appear. For a while even the main newspaper was published only every two days. After the war the situation changed dramatically. Both the main newspaper Friheten (Oslo) and Arbeidet (Bergen) appeared daily, while three other newspapers were published two or three times a week. At the end of 1947 the total circulation of these newspapers was about 40,000. Still a significant figure, this was in reality a dramatic decline from 1945/1946.

Friheten experienced an explosive increase just after the war. In May 1945, circulation increased from 40,000 to 70,000, and it further increased to 130,000 in the autumn of 1945. The newspaper had 65,000 subscribers, and was distributed to most parts of the country. What is important and remarkable is that Friheten for a while became the second largest newspaper in Norway, next to the conservative Aftenposten, and was the greatest labour newspaper in the Nordic countries. This explosive progress transformed into a systematic decline as early as the autumn of 1945. The decline continued during the whole period. In January 1947, circulation was 40,000, and at the end of 1949 the number had declined to less than 20,000.

This decline was caused by the general political development, but also

to a great extent by the fact that in the autumn of 1945 DNA had by decree already changed the newspaper from a morning paper to an afternoon paper. From the beginning the newspaper was printed in 'Arbeidernes Aktietrykkeri', controlled by DNA. This change was politically motivated. It was considered a dangerous threat to DNA that Friheten was more widely distributed than the main DNA newspaper, Arbeiderbladet. The control of the main channel of communication with the working class was crucial in the battle between the two parties. The change made it difficult to distribute the newspaper all over the country and contributed to the fact that Friheten had a much earlier decline than the party in general. In my opinion the decline of the newspaper contributed heavily to the general decline of the party. Norwegian society was not overly % media exposed' in 1945. It is likely that the mass media structure had long-term consequences, and that it may have accelerated the process of decline. It may be argued that the distance to the electorate increased, making it difficult both to reach political supporters and to maintain old ones.

The theoretical review Vår Vei never became a success in spite of the fact that the new leaders around Peder Furubotn stressed theoretical insight (Selle 1981, Titlestad 1975, 1977a, 1977b). At the end of 1945 the review had 2,000 subscribers, a figure that was dramatically reduced to only 421 by mid-July 1948. The organizational review Vårt Arbeid, only distributed among members, showed a similar pattern. Of course this was more conspicuous as it reflected the decline in membership. At the most 3,500 copies of the review were distributed, which meant that 1096 of the members were subscribers.

Little information is available about the other communist organizations, except for the mother organization. The same is true concerning the various 'front organizations'. We do not know to what extent NKP dominated these organizations which were important parts of the communist organizational set-up. We are not able to say how strong communist influence was outside strictly communist organizations. What we know is that the communists had a strong position in the Trade Unions immediately after the war, but here too DNA to a great extent took over from 1946/1947.

We have, however, some information about the youth organization: NKU. By the end of the year 1945 NKU had 230 local organizations and about 15,000 members. This marked a great increase; in 1940 there were only 1200 members. The eastern areas came out very strongly, while we found only 9 organizations in the North, thus illustrating a centre-periphery dimension. It seemed to be more difficult to build and consolidate an

organization far away from the Oslofjord area. We shall see below that the same tendency held true for the party organization, although not to the same extent. The organizational structure of the youth organization also meant that there was no strong connection between its strength and the party's electoral support in general.

That the youth organization had a great increase just after the war indicates that young people played an important role in the party. A survey in 1949 showed that in addition to the fact that the party had a male electorate to a greater extent than any other party, it also had a larger proportion of young people: 48 96 were 39 years of age or younger, and no other party had a similarly small proportion of voters over 60 years of age. (Holbæk-Hansen 1951). A study of the representation at the Party Congress in 1946 yields some very interesting and telling information. Of 303 representatives from all districts as many as 148 were under 30 years, of these 45 were even under 20 years. Only 60 representatives were over 40 years, and all together only 27 representatives were women. This means of course that new and young men had to a great extent taken over the party. Of the 303 representatives 172 were registered as members in 1945, and only 39 had been members of the party since 1923. What is most interesting is that as many as 140 of the participants were former members of the DNA. We have to look at NKP as quite a new and young party in 1945.

NKP even tried to build a party school in 1945, but had to abandon the idea as early as in 1946 for lack of pupils and a shortage of funds. The school was intended to become an important and integrated part of the party's organizational network. It was planned as the ideological centre of the party, as a real training centre for cadres.

How Did Organizational Strength Influence Electoral Support?

Table 1 shows the membership development in the period. The numbers are a bit uncertain, and it is possible that some over-registration occurred in 1945. On the other hand, documents show that the membership could have been even more numerous. Many local organizations were on their guard against members with a different ideology and political practice, fearing infiltration and destruction of the organization from within.

The number of members increased from 25,000 to 34,000 from 1945 to 1946. This must not be interpreted as a real expansion, but more as a sign that it took time to register all the people who wanted to become members of the party. One has to bear in mind that all the new members

created a 'problem' for a party organization that had been more or less
declining since the end of the twenties.

In 1949, when half of the members had left the party, the number of organizations had also declined dramatically, from 691 in 1946 to 384 in 1949. The decline occurred in all districts except Finnmark. This was partly due to the German 'scorched earth policy' in this area. The people had to move back before the party could build up an organizational network.

A similar picture appears with respect to the electoral decline. Table 2 shows the electoral decline from 1945 to 1949 by county ifylke), and also the relative decline. The decline was great in all the counties, but in general smaller the stronger the support in 1945.

Both Finnmark and Hedmark, crucial provinces in the history of
Norwegian communism, came to appear as even more dominant communist
areas in Norway in 1949. We observe that the decline was

generally strongest both electorally and organizatonally in the weakest communist provinces - the South-West area, the most backward area of Norwegian socialism. But even in these provinces important communist enclaves existed in industrial communes. In Nordland and Troms, both communist strongholds in 1945, the party suffered a dramatic setback. After 1949 one cannot say that these provinces are communist strongholds. It is DNA's tremendous support that makes Nordland and Troms two of the most important socialist areas in Norway.

As a brief and preliminary conclusion we can state that there was a strong correspondence between organizational and electoral support. We found, however, a kind of centre periphery effect. In the North, the membership was not as high as could be expected compared to the electoral support. In the Inner East region the party had the strongest organizations. Here Hedmark was quite special, with over 4,000 members in 114 organizations in 1946. Trøndelag and parts of the Oslofjord area also posessed a strong organizational network.

Who were the members of NKP? We have already mentioned that the
average member or voter tended to be young, but we do not know the

age composition exactly. What we do know, however, is the distribution by sex and social class at district level. The data are provided in table 3. We note the great differences between districts. People in the primary sector made up an important group in some districts. Their share was greatest in the Inner East region, and especially in Hedmark. This was also the area with a strong positive correlation between employment in forestry and communist support (.4). Also in Nordland and Troms, the only areas in which we did not find negative correlations between employment in the primary sector and communist support, people in the primary sector made up a significant part of the members.

In the South and the West and in the Oslofjord area the party seemed almost to lack support in the primary sector. Here the secondary and tertiary sectors dominated. It was also in these areas that the negative correlations between employment in the primary sector and communist electoral support were highest. At the same time the correlations between employment in the secondary sector and communist strength were highest in these regions (together with Trøndelag).

In Nordland we discover a high degree of membership from people employed in the public sector, but we are not able to offer a satisfactory explanation. Perhaps it was related to the large military sector in this area. The intellectuals were of some importance only in Oslo-Akershus and in the western region (Bergen, Hordaland, Sogn og Fjordane). Intellectuals have never played an important role in Norwegian communism in contrast to, for example, the pattern in Denmark. The low percentage of women reinforces the picture of a party dominated by men. But it is not at all certain that this was a characteristic of the communist party alone. In 1938/1939 only 14.8 96 of the members of DNA were women, and Lars Alldén (1980) points out that before 1972/ 1973 only 13 -15 % of the members in SF (Socialist People's Party) were women. The organizational participation of women in the Norwegian Labour movement has always been low. NKP also had the lowest percentage of women in its electorate in 1949, 2496 compared to 40% for the DNA, while it was as high as 65 96 for KRF (Christian People's Party) (Holbæk-Hansen 1951).

In sum, taking into consideration both the composition of the members and the ecological correlations, I think we can talk about two kinds of communism. In the Inner East region, especially in Hedmark, and partly also in Nordland and Troms, we can talk about a primary sector communism strongly related to forestry. Elsewhere communism is related to the secondary sector. This does not necessarily imply the usefulness of concepts of 'industrial communism' and 'backwood

communism' in the sense Allardt uses these terms. Finnmark was a quite densely populated area in the sense that people were grouped in small towns and villages, not isolated as in Allardt's characterization of vbackwood communism'. At the same time Finnmark had the lowest electoral participation and a strong positive correlation between employment in the tertiary sector and communist strength (.70). These facts also run contrary to Allardt's theories. In Hedmark the electoral participation was low too, and here NKP had a strong organizational network, both contrary to the Allardt assumption (Allardt 1964, 1967, 1970).

We shall now turn to a strict commune level of analysis. We have data that show where NKP had organizations in 1946, the name of the organizations, and the number of members of each organization. The amount of 'missing' membership data precludes a detailed analysis of that relationship at the commune level. But we have sufficient data to carry out studies in limited areas.

We start by looking at the difference in communist support between communes with and without organizations. Table 4 shows that even in its heyday NKP was not a nationwide party. The party participated in only about 37% of the communes in the local elections of 1945 and 1947, and we discern significant regional differences. The party participated in most towns and was in general much stronger in urban areas than in the countryside. In 1945 NKP got as much as 16.9% of the votes in the towns as compared to 9.6 % elsewhere. In 1945 the party had less than 3% support in 353 out of 731 communes, and in 1949 in as many as 474 communes. Altogether this indicates a strong skewness in the distribution of communist votes. There were at both elections some large communes that provided strong communist support, thus having a decisive effect for the relatively high support at the national level.

In the Inner East region and in the Oslofjord area, more than 50 % of the communes carried a communist party list. In the South-West, the percentage was less than 15. We also see that party organizations were not so widely distributed in Nordland-Troms. Perhaps this faet can explain to some extent the great communist decline in this region in 1949. It is also interesting that in the strongest NKP region, Finnmark, the participation was relatively low.

As a conclusion we can say that it is not unambiguously true that the
party had its best electoral results in regions with the best distributed
organizational network.

But it is certainly true that the poorest results were located in regions where the party had the least comprehensive organizational network. Even here we observe a kind of centre-periphery pattern. Generally, it was the people nearest to the organizational centre (Oslo) who had the best opportunity to vote for the communists.

In table 5 we see the exact differences in communist support between communes with and without organizations. Both in 1945 and in 1949 the difference was conspicuous for the country as a whole. In 1945 the difference was also great in all regions. However, the differences were smaller in the North, and particularly small in Finnmark in 1949. Here the relative decline in 1949 was greatest in communes that had communist organizations. In 1949 we find that NKP was almost without any support at all in communes without organizations, except in the North.

If we look at the patterns of decline, we see that the decline occurred to a similar and great extent in all regions, and that it was generally greatest in communes lacking organizations. But the difference was not very marked.

All in all, these findings support our hypothesis. The difference was marked between communes with and without organizations, and there was a tendency for this difference to increase over time. We have also found a great covariation between organizational strength and electoral support both at the county level and the commune level of analysis. All the same our data are not good enough to decide exactly how organizational strength affects levels of electoral support. What we need is more accurate information about when the members left the party and in addition more qualitative internal information about different local organizations. However, our results emphasize the extensive change in the communist party. In our terminology the party changed from a mobilization party to a party in disintegration during a short period of time.

The Relationship between Communist Strength in 1924 and Communist Strength in the Immediate Postwar Period

We shall now turn to the impact of political tradition on communist voting strength. Did the renaissance of NKP in the immediate postwar years take place in the same areas that had been communist strongholds in 1924 - this in spite of the fact that NKP almost disappeared from the Norwegian political scene in the 19305? And, did the decline in 1949 occur to a greater extent in communes without communist or socialist traditions? In 1945 the general support of NKP was much higher than in 1924, which means that the party had spread to new geographical areas and even to new social groups. We believe the decline will be strongest in these new areas.

This approach implies some serious theoretical and practical problems. It is difficult to explain the voting behaviour in 1924, because it was problematic for the electorate to see the real difference between the two socialist parties, DNA and NKP, just after the split. During the election campaign, DNA was arguing that the party was at least as radical as the NKP (Lorenz 1978, Maurseth 1972). To a great extent questions other than purely political-ideological ones are believed to have influenced the voter's choice. Questions like: which party did the old leaders follow, which party took over the local newspaper, which party had the strongest

local organization, the dominant position in central and local unions,
were obviously all factors of importance. This is indicated by the great
regional differences in their electoral support.

To vote DNA in 1924 may have appeared just as 'radical' or 'revolutionary' from the voter's point of view, in spite of the fact that there existed real political differences between the parties. To a certain degree the same difficulties arise in 1945, but to a lesser extent. Just after the war too the platforms of the two socialist parties had much in common.

These problems are also theoretically important. How a party understands itself is one thing, how the voter understands the same party is quite a different matter. (How we as researchers perceive parties maybe a third problem.) Very often we know far too little about the voter. We really need knowledge about how information is spread over time.

The split in 1923 led in the short run to a dramatic decline for DNA. The new party, NKP, got 14,000 members and DNA had about 40,000 members left. The communist party took over 11 newspapers and DNA kept the remaining 28 newspapers. In addition 11 out of the 28 socialist members in the Storting also became members of NKP.

Most interesting are the striking geographical differences in electoral support. Table 6 shows how the distribution of communist votes differs from those of other socialist parties. The Norwegian Social Democratic Labour Party (NSA), formed as a splinter party from DNA in 1921 when DNA became part of the Third International, was particularly strong in Østfold, but the party also had strong support in Akershus, Bergen, Møre (towns), Troms and Finnmark.

NKP had its strongest support in 1924 in Bergen, with 30.3 96 of the votes, but was also particularly strong in Hedmark and in the city of Trondheim. In Finnmark, Hordaland, and Oppland the party gained about 10 % or more of the vote.

DNA, the dominating party on the left, was more evenly spread geographically. Its position was weak in Østfold, the main NSA area, and in the South and the West. The situation was quite special in Hordaland. In this weak socialist region, NKP was the dominating socialist party. In general the communist party became stronger the stronger the total socialist support. Both in Bergen and Trondheim , core cities of Norwegian socialism, NKP became the dominant party. On the other hand, the communist party was almost without any support at all in the Capital; DNA has always had the upper hand of the two in Oslo.

If we compare the election results of 1924 with those of 1945 and
1949 (see table 2), we see that the main communist areas remain the

same. At both junctures the party was strongest in typically industrial areas, and, as in the immediate post-World War II period, the party was strong in forestry communes, particularly in Hedmark. In 1924 Finnmark was also a communist stronghold, but not to the same extent as in 1945 and 1949. But important differences emerge. In 1924 NKP gained support far below its own national average in most parts of the Oslofjord

area and even in Nordland and Troms. This was not the situation in the
immediate postwar period.

We can safely conclude that the strongest and weakest communist areas were to a great extent overlapping. The really new aspect in 1945 was that NKP became a much more nationwide party. However, the party was, as we have seen, still concentrated in towns and densely populated areas.

A look at the total socialist support reveals a tremendous increase during the period under study. In 1924 the total socialist vote was 33.396, in 1945 52.9%. It was during the thirties that DNA experienced 'its great leap forward' and became able to participate in local elections in almost every commune. This means that the structural position of NKP in 1945 was quite different from what it was in 1924. In 1945 DNA was the dominant socialist party all over the country, and we have found significant positive correlations between the strengths of DNA and NKP in all regions, except in the North. In table 7 we see that this is not at all the case in 1924.

In 1924 the substitute model fits in best. Although NKP was much stronger in 1945 than in 1924, DNA had an even more dominant position. In 1945 DNA was stronger than NKP in NKP's main areas. In 1924, on the other hand, NKP was the strongest socialist party in crucial areas of Norwegian socialism. As compared to the situation in 1945, the party was not able to turn these structural advantages to account. The party had a continuous decline. As early as in 1930, the party was back at 1.7%.

Table 8 shows the correlations between communist strength in 1924 and in 1945 and 1949. In both instances we find strong correlations. In general the correlations were highest between communist strengths in 1924 and 1949, but the difference was negligible. Nevertheless, this indicates a stronger decline in 'new' communist areas without a former communist tradition. As we have seen, Finnmark was the only region where this tendency was absent.

As mentioned earlier, there are problems connected to the interpretation of the 1924 election. Therefore we shall also look at the relationship between the support for DNA and NSA in 1924 and communist strength after World War 11. In addition it is of interest to establish the relationship between what we may call 'radical tradition' (operationalized as DNA plus NKP support in 1924) and communist strength in the immediate postwar period.

Table 9 shows strong correlations between DNA support in 1924 and communist strength after World War 11, but these correlations were not as high as those of communist support at the same junctures. This means however, that in general there existed a strong connection between 'radical tradition' and NKP support in the immediate postwar period. In the South-West and in Finnmark we found no relationship. In the South-West this was presumably due to the fact that NKP was the dominant left party in 1924. In Finnmark this situation was caused by the fact that the NKP was to a great extent strong in the same areas as in 1924 and at the same time there were strong negative correlations between communist support and DNA support in the immediate postwar period.

The correlations between NSA support in 1924 and communist support after World War II showed quite a different picture. Only in the Oslofjord area, the main NSA area, in the South-West and in the towns in 1945 were there significant correlations. In the Oslofjord area: NKP had expanded strongly in the previous core area of NSA. The relationship in the South-West shows that in most parts of this region NKJ? and NSA were the main competitors on the left in 1924.

To sum up, we may say that there was a strong connection between total socialist support in 1924 and communist support after World War 11. These correlations were weaker in the South-West and in the North. This was partly so because it was in these regions that Norwegian socialism had expanded most during the interwar period. If we look at the correlations between total socialist support in 1924 and in the

immediate postwar period, we also spot these regional differences. The
correlations were as high as .80 and .81 for the country as a whole, but
far less in the above-mentioned regions.

In this paragraph we shall analyse in more detail whether the decline in communist support in 1949 was smaller, the stronger the support in 1924. In table 10 we see that this was the case. However, the correlations are not particularly high. Nevertheless it was only in the South-West that this tendency was absent, so the results lend important support to our hypothesis. The decline became smaller in areas where the party had a political tradition.

If we look at DNA, we see that it was only in the Oslofjord region that NKP had a smaller decline, the stronger the support for DNA in 1924. Even if there existed strong correlations between DNA support in 1924 and communist strength in 1945 and in 1949, this fact did not generally influence the structure of communist decline. In my opinion these differences between the effects of the NKP and the DNA strength in 1924 are very important and strengthen the significance of the organizational and political tradition approach. It was in areas with a former political and organizational tradition that the party experienced only moderate decline in 1949, but a strong socialist tradition was by itself not a.ble to prevent exposure to the general pressure on the communist vote.

Only in Finnmark do we find a significant correlation in the relationship between communist decline in 1949 and NSA support in 1924. Here there is a clear tendency for the communist decline to become greater the stronger the 'right socialism' in 1924, an interesting discovery.

On the whole the relationships seem to corroborate our expectations. There were high correlations between communist strength in 1924 and in the post World War II period. The NKP had its renaissance mainly in areas where the party had been strong in its earlier heyday, and in areas with a historically strong socialist tradition. There is continuity, but separated by a wide gap of almost non-existence of the communist party in the thirties. This means that communist political traditions survived in the 1930s and stayed latent in spite of the absence of organizations to maintain them, and then displayed an almost incredible resurrection in the years after World War 11. At the same time communist political tradition acted as a brake on the overall decline in communist support. The communists had expanded to new areas in 1945, but these were former socialist strongholds. It was DNA that during the thirties, and also in the immediate postwar period, was the vehicle of socialist ideas in areas unfamiliar with socialist thoughts.

A 'Path' Analysis of Factors Influencing Communist Support

In part two we started out with four main approaches, and so far we
have through bivariate analysis emphasized two of them: organizational
strength and political tradition.

The main approaches of the correlational analysis were as a next step
incorporated into a complex model for communist support, which was
tested by means of regression and path analysis. In Figure 3 we see the

results for the country as a whole. The model explains .57 and .65 of the total variance in 1945 and 1949. This must be characterized as a fairly good result. Our hypothesis that the 'model' should fit in better in 1949 was only weakly supported. The basis for our hypothesis was our contention that the election in 1945 was quite special because of the war, and that the NKP therefore would also appeal to groups outside the traditional socialist ones. In 1949 we expected a turn to a more 'normal' relationship.

Through residual analysis we discovered that the model underestimated the support in a lot of communes, especially in the Inner East and in the North. Our model overestimated the communist support in almost all communes in the South-West, and also in parts of the inner East, especially in Oppland and Telemark, and in parts of the Oslofjord region.

The model fit was quite good in Trøndelag. Altogether this means that in every region, and also in every county, there were important internal variations. Even in the strongest communist areas, Hedmark and Finnmark, we found communes with extreme negative residuals.⁶ Such phenomena highlight the difficulties related to regionalization in ecological analysis and should be kept in mind when interpreting the results of the analysis.

In Table 11 we see the explained variance and the most significant
beta-weights in each region. Coefficients below .20 are excluded. We find
some striking differences in explained variance between the regions. In

general the model fits in best in the strongest NKP regions. It was, however, only in the weakest NKP region, the South-West and in the relatively strong communist region of Nordland-Troms, that the 'model' fitted considerably better in 1949. It seems to be in areas where the party had few but important strongholds in 1924, and also made a successful election in 1945, that the party actually reached new groups in 1945. We can speculate whether these groups also were the first to leave the paity.

If we look at the internal paths, some interesting results come to the fore. For the country as a whole both employment in the secondary sector and in forestry had a limited direct effect on communist voting. The secondary sector had, however, a relatively strong direct effecting several regions, and forestry had a strong direct effect in the strongest forestry area, the Inner East.

All in all, socio-economic structure is not entirely without importance, but as in the Finnish context, organization and political tradition no doubt had the strongest direct effect on communist voting. NKP strength in 1924 had a significant effect in every region in 1949, and only lacked importance in the North in 1945. The weak effect in the North was caused by the great communist progress in this area in the meantime. That the effect was also strong in the South-West was due to relatively few but important strongholds which overlapped at these two junctures.

Organization also had a relatively strong effect in all regions except in Finnmark. This was not suprising, since we have seen that it was only in this area that the difference between communes with and without organizations was small, relatively speaking.

What is most interesting is to observe to what extent socio-structural variables are related to organization. We see that the beta-weights between the secondary sector and orgaization were significant in all regions except in Finnmark. The coefficients were strongest in the Oslofjord region and in the Inner East. Even in the Inner East the beta-weights between the secondary sector and organization were higher than between employment in forestry and organization. The secondary sector seems to be a stronger necessary condition for building an organization. These results indeed stress the importance of socio-structural variables for communist voting in this period.

We have also tried to explain the relative change in communist voting between 1945 and 1949 by a multiple stepwise regression, but the results were not convincing. The amount of explained variance is now much lower and great multicollinearity problems arise. Nevertheless, as a cautious conclusion we can say that the stronger NKP was in 1924 the smaller the decline, and the stronger the increase in suport for DNA in 1949, the stronger the communist decline.

The fact that the explained variance becomes lower when we try to explain change is not surprising. In general, change is very difficult to explain as an isolated factor. It is not logically evident that the factors which are important for explaining the general strength of a phenomenon, are also the most important ones in explaining change in the same phenomenon during a certain period.

Conclusion

In this article we have argued that a combination of the classic AUardt-Linset approach, focusing on socio-economic structure, and an organizational approach also taking into account the importance of political tradition, is fruitful in studies of communist voting behaviour. The findings show that organizational strength and political tradition are the most important factors explaining communist strength, a fact that underlines the importance of the Laulajainen approach. But even if the organizational pattern is most important statistically, the organizational structure is in turn to a great extent explained by socio-structural factors. This means that a working-class milieu is an important conditioning factor for the existence of communist organizations. This is the fundamental lacuna of Laulajainen's approach. He fails to put the existence of organization in the Finnish case into a wider causal context. In the Norwegian context a socio-economic structure dominated by industry and/or forestry was a necessary condition for communist strength and even a sufficient one with regard to the strength of DNA. But in several communes with this socio-economic structure and a strong DNA, NKP was almost without any support at all. Only when we took into account the factors of organizational strength and political tradition were we able to predict communist strength more precisely. Only in very few extreme communes was communist strength significant without a former radical tradition and an organizational network.

However, I should like to stress that I do not regard my model as a strictly causal one, but rather as a simple methodological way of summarizing important factors that influenced communist voting in this period. I have also tried out a lot of linear and non-linear variants of the 'model' in each region, but the present version has provided the best results when one takes into account the requirement of a high explained variance and model simplicity. This means that we did not find any important thresholds in the material, as Tingsten's law of'social gravity' predicts (Tingsten 1937, 230 - 230). Nowhere did we find, for instance, that communist strength became relatively stronger when employment in industry and/or forestry increased.

To conclude, a digression-. To what extent is our approach relevant for studies of other political parties in the same, period (subject)? Would the approach also be relevant for studies of communist voting behaviour in quite different types of society (space)? And last, but not least, would our approach be as fruitful in different periods of time? These are important methodological and theoretical problems not to be elaborated on here.

---

NOTES

1. For a comparative analysis of the political changes in Western Europe from the late interwar period to the early fifties, see Selle 1980: chap. 2. This chapter provides a review of relevant literature and extensive statistical information.

2. This process is analysed in Selle 1980: chap. 5. Here I start out with a rejection of the old 'monolithic' approach so common in studies of the communist movement. It is not sufficient to analyse the intentions and the interests of the Soviet Union in order to understand the policies of the different national parties. Specific national factors always play an important role. In the classical tradition the work of Borkenau 1971 is crucial. Also Claudin 1977/1978 is still part of this tradition (his important work on the Komintern). Important names in the 'new' tradition (excluding here the literature on 'Eurocommunism') are Blackmer 1975, Mclnnes 1975, Tannahill 1978, and Middlemas 1980.

3. Selle 1980: chap. 3 gives a broad and critical analysis of different theories on communist voting.

4. These theorical statements are explained in greater detail in Svåsand 1981 and in Selle and Svåsand 1981a and b.

5. Most of the political and socio-structural data are taken from the Kommunedatabank of NSD (Norwegian Social Data Services). Selle 1980: chap. 4:19-26 and chap. 7:52-55 evaluate the data. The author has also had access to organizational data not often available for Communist studies. Through T. Titlestad, I have had access to both his private archive and the archive of Peder Furubotn, the leader of NKP during the war ami the immediate postwar period. The data are evaluated in detail in Selle 1980: chap. 4:20 - 22, chap. 7:52 - 55 and in appendix B: 1 - 3. For a presentation of the regions, see Selle 1980: chaps. 4 and 7. For a discussion of the choice of regions and problems connected with regionalization in general, see Selle 1980: chap. 7:1 -4. In Selle 1980: chap. 8 and chap. 9 one will find a discussion of methodological and statistical problems.

6. The intervals are defined like this: Approximately predicted cases are cases where the difference between the predicted and the real value are less than ± 2.5%. Moderate residuals are defined as lying between ± 2.5 96 and ± 7.0%. Extreme residuals are residuals greater than ± 7.0%.

REFERENCES

Allardt, E. 1964. 'Patterns of Class Conflicts and Working Class Consciousness in
Finland', in Allardt, E. and Littunen, Y., ed., Cleavages, Ideology and Party Systems.
Helsinki: The Westmark Society.

Allardt, E. and Pesonen, P. 1967. 'Cleavages in Finnish Polities', in Upset, S. M. and
Rokkan S. ed., Party Systems and Voter Alignments. N. Y. Free Press.

Allardt, E. 1970. 'Types of Protest and Alienation', in Allardt, E. and Rokkan, S., ed., Mass
Politics, N. Y.: Free Press.

Alldén, L. 1980. Medlemmene i Sosialistisk Folkeparti. Magisteravhandling i Sosiologi,
Universitetet i Oslo.

Blackmer, D. L. M. 1975. 'Continuity and Change in Postwar Italian Communism' in
Blackmer, D. L. M. and Tarrow S., ed., Communism in Italy and France. New Jersey:
Princeton Univ. Press.

Borkenau, F. 1971. World Communism. Ann Arbor: Univ. of Michigan Press.
Claudin, F. 1977. Krisen i den Kommunistiske Verdensbevægelse I. Viborg: Pax.
Claudin, F. 1978. Krisen i den Kommunistiske Verdensbevægelse 11. Viborg: Pax.

Dogan, M. 1967. 'Political Cleavage and Social Stratification in Italy and France', in
Lipset, S. M. and Rokkan. S., ed., Party Systems and Voter Aligment. N. Y.: Free Press.

Holbæk-Hanssen, L. 1951. 'Litt om velgerne ved Stortingsvalget i 1949', in Sak og
Samfunn'. Bergen: Chr. Michelsens Institutt

Korpi, W. 1971. 'Working Class Communism in Western Europe', ASR, American
Sociological Review, Vol. 36.

Korpi, W. 1978. The Working Class in Welfare Capitalism, Work, Unions and Politics in
Sweden. London: Routledge & Kegan.

Laulajainen, P. 1979a. 'Some Aspects of the Division of the Finnish Working Class Movement
after the Civil War: Research Note', Scandinavian Political Studies Vol. 2, New Series
no. 1.

Laulajainen, P. 1979b. Sosialdemokratti vai Kommunisti (English Summary) Mikkeli:
ITA-Suomen Instituutti.

Laulajainen, P. 1980. 'The electoral support of the working class movement in Finnish
Lappland'. Mikkeli: Mimio.

Lipset, S. M. 1968. Det Politiske Menneske. Oslo: Gyldendal.

Lorentz, E. 1978. Norwegische Arbeiterbewegung und Kommunistische Internationale
1919- 1930. Oslo: Pax.

Maurseth, P. 1972. Fra Moskva teser til Kristiania forslag, Oslo: Pax.

McCauley, M. ed., Communist Power in Europe 1944-1949. London: MacMillan
Press.

Mclnnes, N. 1975. The Communist Parties of Western Europe. London: Oxford Univ.
Press.

Middlemas, K. 1980. Power and the Party. London: André Deutsch.

Rydenfelt, S. 1954. Kommunismen i Sverige. Lund: Gleerupska Univ. Bokhandel.

Selle, P. 1980. NKPvedStortingsvalga i 1945 og 1949. Thesis in Comparative Politics, Univ.
of Bergen. To be published at Universitetsforlaget, Bergen 1982.

Selle, P. 1981. "Peder Furubotn. Teoretisk perspektiv og politisk praksis.' Institute of
Comparative Politics, Bergen, mimeo.

Selle, P. and Svåsand L. 1981a. 'Radikalismens økologi i Nord-Norge', to appear in
Lindstrøm, U., ed., Politikkpå Nordkalotten.

Selle, P. and Svåsand L. 1981b. 'Organisasjonsstyrke og valgoppsutning', paper,
Nordiska Statsvetenskapliga kongress, Åbo.

Svåsand, L. 1981 'Membership trends in Norwegian political parties', paper, ECPR Joint
Session, Lancaster.

Tannahill, R. N. 1978. The Communist Parties of Western Europe. London: Greenwood
Press.

Tarrow, S. 1967. Peasant Communism in Southern Italy. New Haven: Yale Univ. Press.

Tarrow, S. 1968, 'Political Dualism and Italian Communism', American Political
Science Review, vol. XXXI, nr. 1.

Tarrow, S. 1977. Between Centre and Periphery. New Haven: Yale Univ. Press.

Titlestad, T. 1975. Peder Furubotn 1890 - 1938., Oslo: Gyldendal.

Titlestad, T. 1977a. Peder Furubotn. Stalin midt i mot 1938 - 1941. Oslo: Gyldendal.