University of Göteborg

Stability in the partisan division of the electorate, as V. 0. Key has pointed out, is linked to a large majority of the voters not reconsidering their party choice with the regularity and frequency provided for in electoral law.¹ Indeed, a well-established finding of modern political behavior research is that party preferences are formed, to a large extent, as enduring political allegiances. In the electorate as whole, major political cleavages tend to survive from election to election. Prevailing differences in value orientation among population groups, reinforced by the activities of organized interests, contribute to sustain party allegiances and maintain structural stability in the party system.

This is not to say that election campaigns could lead merely to mechanical repetitions of predetermined responses in the mass electorate when no profound political realignment becomes manifest. Virtually every election outcome reflects some degree of change in the electorate's mood. Policyrelated events and party controversies generate short-term factors that mark each election campaign. Sometimes such short-term shifts may even create a new parliamentary majority, although the basic pattern has not been altered appreciably.² In an earlier volume of this yearbook, the author has examined the entire sequence of such short-term fluctuations of the vote during the post-war period in Swedish politics.³

In the following inquiry we will deal with major issue controversies in Swedish
politics from 1956 to 1960 which concerned fluctuations in the party division
of the vote. The present analysis, however, will not be directed primarily at

---

* This article is a revised and extended version of a paper prepared for the Fifth World Congress of the International Political Science Association in Paris, 1961. Part of the analysis appears also in: Bo Särlvik, "Skiljelinjer i valmanskåren" (Political Cleavages in the Electorate, Statsvetenskaplig Tidskrift, 1965: 2-3). The study is basedon nationwide Interview sample surveys conducted by the Institute of Political Science, University of Göteborg. The surveys were carried out with the Survey Research Institute of the Swedish Central Bureau of Statistics which has been in charge of the sampling and the fieldwork. These studies have been financed through government grants and support from the Swedish Council for Social Science Research.

the causes of partisan change. Instead, we will highlight some structural properties of the party system which bear on the interplay between the parties' policy decisions and the formation of political opinions among their constituents. Furthermore, we will attempt to illuminate the role of party allegiance as a psychological factor which contributes to equate the voter's views with the policies espoused by his chosen party.

The analysis is based on data furnished by nationwide interview sample surveys conducted by the Institute of Political Science at the University of Göteborg in the parliamentary elections in 1956, 1960, and 1964 and the referendum on the pension Question in 1957.⁴

The Swedish Multi-Party System: Social and Political Cleavages

For a multi-party system, the Swedish political scene has a remarkably one-dimensional character.⁵ Tie major parties can be arrayed along a right-left continuum which forms the ideological basis of the party system. This arraying also provides the frsimework for short-term competition among parties in the sense that partisan change tends to take the form of switching between adjacent parties on the continuum.⁶ This characterization clearly is applicable to four of the five major parties: the Conservative Party (which should be translated literally "The Right Party"), the liberal People's Party, the Social Democratic Party, anc the Communist Party. A partial exception from the prevailing unidimensionality is the Center Party since it draws major support from the farming population and traditionally has represented an urban-rural cleavage. During the period of our study, however, the party managed to enlist much additional support from outside its previously established domain. The broadening of its electoral appeal was emphasized by a change in the party name, the earlier party label being "The Countryside Party — Farmers Union".

In Swedish political verbiage, the three non-socialist parties usually are referred to as the Bourgeois; parties. The Center Party shares the position at the "political center" with the People's Party. Over the last decades, the combined electoral strength of the Bourgeois parties has nearly equalled that of the Social Democratic Party. Despite the perennial split among the nonsocialist parties, they tend to be considered as one political bloc in opposition to the Social Democratic Pa'ty which has had a tenure of more than three decades as the party in government. Nevertheless, since the 1930's the Center Party has also had political ties with the Social Democrats and from 1951 to 1957 it participated in a coalition government. Thereafter, by joining the opposition, the Center Party moved toward its "natural" position in the party system.

We have placed the parties along the predominant dimension referred to
in such a way that the differences in the parties' socio-economic centers

of gravity within the middle class and the working class coincide with a rightleft gradation in their attitudes toward welfare state politics. To completet this characterization, we can deal briefly with differentiation along regional, religious, and ethnic lines. As sources of political cleavage such factors are almost entirely absent.

The broad lines of the socio-economic differentiation in voting behavior appear clearly in Table 1. The same pattern generally has prevailed throughout the post-war period.⁷ The table shows that the Socialist parties receive a heavily predominant share of the working class vote, while a considerable non-socialist majority exists among enterprisers and salaried employees. However, the data also reveal "lower middle class" groups as well as working class groups where the Bourgeois vote and the Socialist vote are about

equibalanced in strength. In accordance with our conception of the party system as a one-dimensional continuum, it also turns out that the bulk of the Bourgeois support within the working class goes to the two centrist parties in the system. At the other end of the social hierarchy, the Conservative party has a stronghold among large enterprisers and salaried employees in higher positions.

The basic unidimensionality of the system is also evidenced from another point of view by the relationship between the voters' party preferences and their attitudes toward "wellare state politics." Although Swedish politics has a highly pragmatic flavor, the various parties show a persistent divergency in their approaches to broad areas of politics. This lends more than a merely rhetoric substance to the traditional "Right-Center-Left", or Bourgeois-Socialist classification which we rely upon in this study of the party system. We have attempted to assess the importance of the ensuing ideological coloration of party appeals to the electorate with the aid of a measure of "attitudes toward welfare state politics". The analytic tool is an attitude index based on six interview questions about social welfare legislation, taxation, government control of the economy, the proper role of private enterprise and equality among occupational groups. The questions were framed to tap the respondents' views on general policy goals rather than their opinions on specific policy measures. In this way, me have obtained a scale which stretches from a highly consistent Bourgeois outlook at the one extreme to an equally strong Socialist position at the other.⁸ From Table 2 we see that the Social Democrat and Communist proportion of the vote shows a continuous decrease from the left to the right along the scale. The data also reveal that the Conservative vote, as would be expected, is distinctly concentrated to the far right on the attitude continuum. Marginal frequencies in the bottom row of the table disclose another significant feature, namely that a substantial part of the electorate is concentrated in the midmost region of the scale where our attitude measure has recorded only weak signs of partisanship in the voters' policy preferences.

This article is not concerned with offering complete analysis of the formation of ideologically colored attitude structures in the mass electorate. For our purpose, it is sufficient to point out one important conclusion drawn from the data in Table 2. Both blocs in the party system can rely on the support of one of the two wings in the electorate which comprise voters who fit their political beliefs into a coherent partisan outlook and are especially likely to be responsive to the parties' ideological appeals. But both the Social Democrats and the Bourgeois parties also draw a sizable portion of their voting support from a part of the electorate where political attitudes are much less consistent and where ideological concepts can be presumed to be much less salient in the voters' minds.

Party Allegiance and Opinions on Policy Questions

In a sense, fluctuations in the vote from election to election prove that election outcomes are not "devoid of political content and intent." Literature on electoral behavior is replete with findings concerning linkages between shifts in the party division of the vote and opinion trends in the mass electorate which have been generated by events in the political arena.⁹ One instance which will be more fully discussed later is the political controversy over the pension question in Sweden through 1956 to 1960. There is no doubt that the pension issue caused a comparatively high rate of partisan change, and it unquestionably affected the outcomes of a sequence of electoral contests.

In this broad sense, it is theoretically meaningful to conceive of an election outcome as the citizenry's collective decision on the choice between the policy alternatives offered by the parties. However, electoral behavior research also has shown that the representational meaning of democratic elections cannot be described as a simple relationship between majority opinions in the electorate and decisions made by officeholders. A national election outcome is not only a collective but also a composite phenomenon. It emerges as the aggregate expression of a wide range of demands and expectations held by millions of individuals. Hence, it is almost unavoidable that, when the ballots have been counted, the precise relationship between the electorate's verdict and the parties' stands on policy questions generally remains unclear.

The cases to be investigated here exemplity a variety of circumstances on different levels of the political system that affect the shaping of voter opinions over national policies. The analysis, however, will be focused on the voters' responses to party politics. For this purpose we shall use a limited set of explanatory variables. One is the voters' party allegiances.

Although a substantial portion of the electorate may have changed parties
on at least some occasion if we consider a long enough time span, it is still
true that only a tiny minority of the voters switch party preferences from any

given election to the following.¹⁰ Most voters form a partisan attachment that becomes enduring in their political outlook. As a consequence, a voter's partisan orientation is most often antecedent to, and more lasting than, his opinions on the changeful issue contents of current politics. To denote the psychological implications of such an enduring party affiliation, Angus Campbell et al. have employed the concept of party identification. In its relation to social psychological theory, this concept presumes that political parties are perceived as collectivities, or reference groups, "toward which the individual may develop an identification, positive or negative, of some degree of intensity."¹¹ A positive identification with a group generally can be expected to lead to a particular readiness to adopt its standards and standpoints. The more important the group attachment is to the individual, the stronger will his identification with the group become and the more influential will group standards be for his own attitude formation.

Thus, the concept of party identification implies that positive or negative party identification has a formative influence upon the individual's evaluation of other elements of the political reality.¹² This assumption gains importance when we turn attention to the short-term contents of politics. Cognitive theory leads us to expect the individual to feel a need to attain coherency in his outlook when he is confronted with conflict. Indeed, few citizens can escape experiences that stress the element of conflict in party politics. Almost everyone encounters policy-related conditions that call for evaluation in some form, e.g., taxation rising prices, and shortage of housing. Mass media supply interpretations of such experiences and transmit familiarity with the more remote aspects of politics. Political parties and interest groups create issues by putting forward demands that may attract or repel voters. To the extent that the individual recognizes; such controversies cropping up in current party politics, he will often find it difficult to qrasp their real meanings. In most cases he will have to rely on outside sources to make up his mind. Campbell et al. have pointed out that party identification can then serve as a cognitive short cut that enables the individual to find his bearings in this flow of conflict-loaded experiences.¹³ Confidence or lack of confidence in a political party is readily available as the basis required to distinguish between credible and non-credible sources of communication or to distinguish between policy makers with good and bad intentions.

The measure of party identification employed by Campbell et al. permits a clear distinction between party identification as a psychological tie to a political party and actual voting behavior. It has not been possible to attain the same level of methodological refinement in the present study. We will, therefore, utilize electoral party choice as an indicator of party identification. However, the 1956 and 1960 surveys include a measure of strength in party allegiance with which we can classify the voters as "convinced" or "not convinced" party adherents.¹⁴ A major disadvantage of this classification compared to that in the American studies is that we cannot discriminate between

voters who have a weak but lasting partisan attachment and "independents". Despite this methodological discrepancy, we will use the terms "strong identifiers" and "weak identifiers" in our discussion when we intend to stress that the inferences pertain to the effects of standing party loyalties.

In describing the structural characteristics of the party system, we pointed out two features that facilitate the formation of stable partisan alignments in the electorate: The parties' anchorage in different social strata as well as the linkages between the party system and a highly developed network of interest organizations presumed to connect perceptions of group interests with party identifications in the electorate. The long-established roles of parties as representatives of ideological orientations, diffuse though these may appear, also will enable the individual to equate party loyalty with support of a longterm political tendency.

These conditions take on added significance because they lend a content of shared values and demands to the affective relationship between the voter and his party which is denoted by the concept of party identification.'⁵ However, the opposite aspect is also pertinent to the problem of opinion formation over political issues. Ideologically colored attitudes and perceptions of group interests held by adherents of a political party do not always guarantee conformity to party standpoints; they can also become sources of dissension. As shown in Tables 1 and 2, the Swedish parties are far from homogeneous in either of these respects even though they have distinctive centers of gravity. Therefore, party leaderships periodically, will make policy decisions that repel various groups within their electoral followings. We will attempt to show how such situations affect the division of voter opinions on national policies.

Finally, the representational function of democratic elections sometimes becomes less a question of a potential conflict between representatives and constituents than a question of representation in the absence of opinions within a large part of the electorate. In a study of the role of public opinion in the American democracy, V. O. Key has coined the term "political stratification" to account for the existence of a wide variation in political involvement and information within the enfranchised population.¹⁶ The term is apt since it draws attention to a conspicuous feature in a picture of voter opinions on public policies which often emerges from data gathered through a crosssection of the citizenry. Stratification of respondents in an interview sample with the aid of information or interest criteria often uncovers striking differences between the distributions of opinions that obtain at different levels of attention. The reason, of course, is that hosts of policy controversies eliciting engagement among party militants, or categories of citizens who have some tanguible interest at stake, may not be noticed by less attentive sections of the electorate.

In the foregoing discussion, we have designated party allegiance, attitude
toward welfare state politics, and socio-economic group affiliations as
explanatory variables. Since these factors constitute enduring components

in the individual's political orientation, we can hypothesize that they have a structuring impact upon his evaluation of public policies. In reality, this hypothesis presupposes intervening cognitive processes which cannot be investigated exhaustively in this article.¹⁷ However, it should be clear that recognition of the standpoints of the parties is a factor of crucial importance. Party loyalty, for instance, can hardly have any influence upon the voter's opinion on a policy proposal unless he is informed about the position taken by the party he has confidence in. For this reason, information about party standpoints will be introduced as an intervening variable in several stages of the analysis.

Evaluation of Government Policies: Two Cases of Opinion Change

We have mentioned that most citizens have daily experiences of conditions which are more or less directly affected by government policies. Often such conditions are not only well-known, but also are considered as unquestionably "good" or "bad" without much partisan disagreement. Even ardent supporters of a party in government would be willing to agree, for instance, that a rise in the cost of living is deplorable. And even embittered opponents will be satisfied if the standard of living shows an upward trend. The connection between such experiences and the policy being pursued by the government, on the other hand, may often be the object of different judgments. When an election approaches, the voter's interpretation of the causes of "good" and "bad" experiences will affect his appraisal of the party in government. However, our hypothesis about the . importance of party identification leads us to expect that the voters' avaluation of government policies will be colored by their partisan orientations. Thus, we would expect government party adherents to be inclined to give the government credit for "good experiences" and to attribute "bad experiences" to circumstances beyond its control. As a correlate, we will expect voters who distrust the governing party to find reasons for the converse evaluation, i.e., improvements have been brought about despite government policies, while "bad conditions" are evidences of policy failures.

In the election surveys, a set of four questions was used to elicit evaluative judgements about governmsnt responsibility for policy-related conditions. These questions referred to recent decades of full employment and continuous increase in the standard of living in Sweden as well as rising prices and housing shortage. All of the questions were similarly phrased. Thus, it may be sufficient to quote one: "By and large, we have not had any unemployment since the world war. (According to your opinion,) is this mainly due to government policies, or is it mainly due to other circumstances?"

In view of the purpose of this analysis, it is important to note that these
questions concerned conditions that remained on the whole unchanged from

1956 to 1960. Though the two nationwide samples were not identical, we have a basis for comparing the relationships between party preferences and opinions about government policies at these two points in time. In one respect, an important political change had taken place between the two elections. The coalition formed by the Social Democrats and the Center Party came to an end and was followed by a purely Social Democratic government.

Table 3 shows how government politics were assessed by the supporters of the various parties. The measure of "the balance of opinions" in the table is a simple index indicating the degree of predominance of opinions that were favorable (+) or unfavorable (—) toward the government.

The table reveals that the supporters of the opposition parties consistently show a predominance of critical opinions, while Social Democrats are as consistently positive in their appraisal of the government's achievements. In accordance with our hypothesis, it also is found that voters with a strong party identification were much more partisan in their views than those with

weak party attachments. These trends become apparent in both the 1956
and 1960 data.

Only one exception appears in the general similarity between the data from 1956 and 1960. Amoncj supporters of the Center Party, a striking reversal of the balance of opinions is recorded. A marked predominance of unfavorable judgments in 1960 was substituted for a strong approval of all aspects of government politics in 1956. The data discloses that the shift was most dramatic among the party's "convinced adherents". Obviously we have registered the consequences of a change in the political context.¹⁸ The reality (full employment, housing {shortage, etc.) was unaltered. But when their party joined the opposition, Center Party adherents became motivated to see the connection between national policies and "good" and "bad" experiences in a new perspective. As their evaluational criteria were reversed so was their picture of the political reality. Hence, these data confirm the causal ordering presumed in our hypothesis, i.e. the voters' party identification is found to have a formative influence on thoir cognitions of policy measures.

Our second case involves a change toward partisan conformity which occurred in the course of an election campaign. Here we will deal with voter opinions over a specific policy measure rather than evaluative judgments on the perfomance of the party in government. As an example of short-term change, it is perhaps moro extreme than representative of opinion formation over campaign issues in general. However, it illuminates clearly the factors that obtain when the adherents of a political party are being rallied around their party's position in a campaign controversy.

The topic is a sales tax put into effect at the beginning of 1960. The new tax caused a vehement political conflict in which all of the opposition parties were united against the Social Democratic government. A parliamentary election was held in September of the same year, and the abolishment of the sales tax became a major theme in the Bourgeois parties' election campaign.

In the 1960 interview survey, a subsample comprising half of the nationwide sample was interviewed before the election. Most of the fieldwork at this stage was completed well before the last week of the campaign period. A post-election mail questionnaire then was used to obtain data on voting behavior in this subsample. The remaining part of the sample was interviewed immediately after the election. This procedure, enables us to examine the division of opinions on two different occasions and to gauge the change that took place.

Response distributions are shown in Table 4. A stable and massive opposition to the new tax among Bourgeois voters was evident through the entire period. In contrast, the balance of opinions must have been decisively redressed among Social Democrats during the campaign. While a majority of the Social Democrats who expressed any opinion were critical in the first interview, a heavy majority in favor of the party standpoint had been formed when the election day was over. We shall now introduce some additional variables to explore the nature of this shift.

The sales tax measure ws.s presented by the Social Democratic government as necessary to balance the budget and to finance the social welfare legislation program. That aspect of the tax issue received added emphasis when the Conservative Party - but not the centrist parties - proposed considerable cuts in the social welfare expenditures as an alternative to the sales tax. Thus, one might expect that the voters' general attitudes toward welfare state politics should be a factor influencing their stands on tax issue.

Table 5 shows that such a relationship appears distinctively among Social Democrat voters. Social Democrats who had a "strong" or "moderate" leftist orientation were much more likely than others to endorse the party standpoint. Most of the Bourgeois voters, of course, are spread along the right-oriented part of the attitude continuum. Here we find only a slight tendency to a relationship between attitude position and opinion on the sales tax.

Table 6 is a more penetraling analysis of the opinion development within the Social Democratic part of the electorate with the inclusion of both strength in party identification and attitudinal orientation as independent variables and shows the situation before as well as after the election. Now it becomes clear that opinions on the tax issue among Social Democratic voters were linked to both these forms of political orientation through the entire campaign period. Before the election, support of the party standpoint already was overwhelming in one particular category, that is, those who vere decidedly leftist and also had a strong emotional atlachement to their party. Positive and negative responses were more equibalanced in strength within the two categories of respondents who met only one of these criteria. Opposition to the new tax was markedly predominant among Social Democrats who showed a low degree of partisanship in both regards. After the election, a clear majority support for the sales tax existed in all of the categories. But the same rankordering

among the categories is evident in the "balance of opinions" index. Thus, our explanatory variables are capable of locating the groups of Social Democratic voters who were most (and least) responsive to their party's campaign appeals.

Table 7 completes the analysis by disclosing the factor that generated the change. Exposure to political communication during the campaign clearly was of decisive importance. The more the Social Democratic supporters were exposed to campaign propaganda, the higher became their propensity to join the party line. As would be expected from our hypothesis, the effect of increasing familiarity with party arguments was much more pronounced among strong identifiers than among weak identifiers." From the beginning, a large majority of the voters knew that the Social Democratic Party was responsible for the new tax (at the pre-election interview, 88% of the men and 75% of the women were so informed; after the election the proportion increased to 90 % for men and 79 % for women). Table 8 shows that this information was indeed crucial.

A new tax is always a disagreeable experience. Our data concerning Social Democratic voters disclose a process in which the "natural" negative responses were outweighed by other factors. Party allegiance obviously was one factor. The relationship between "attitude toward welfare state politics" and opinion

on the sales tax furthermoro indicates that consistency in the voter's political outlook was instrumental in preparing him to accept the party standpoint. Finally, we can infer that :he shift in the balance of opinions was due to two effects of campaign exposure bearing upon these factors. By heightening public attention to party politics, an election campaign will increase the salience of party identification and supply voters with partisan arguments over policy measures.²⁰ In the case of the sales tax, these campaign effects merged to generate a polarization toward party standpoints.²¹

The same three factors, i.e., a "natural" negative response to a new tax, attitudes toward welfare state politics, and party allegiance, provide a simple explanation to the pattern of stable partisan conformity appearing among the Bourgeois voters. For these voters, nothing in the campaign situation could induce them to accept the new tax. Irrespective of their attitudes toward welfare state politics, they could rely upon their parties' testimonial that the new tax was harmful and unnecessary. In addition, much of the Bourgeois electoral support held an ideological orientation that made them suspicious toward the means and ends involved in the Government's tax policy.

Social Welfare Issues as Sources of Infra-Party Dissension In 1960

The election campaign in 1960 converged, more than usual, on a few major policy controversies that brought the right-left differentiation in the party system into the foreground. At the same time, all the issues were capable of arousing widespread public attention and had visible economic consequences for large population groups.²² The 1960 election, thus, allows us to investigate the relationships between party standpoints and voter opinions over policy questions under conditions unusually favorable for a direct channeling of citizen demands through the primary mechanism of representative democracy.

The sales tax was one of these issues. Both of the other major issues in the campaign were originated by Conservative proposals. The Conservative party had experienced a period of electoral successes during the 1950's when its political course had been laid more toward the right than in the earlier post-war years. This background enables us to see that the Conservative party was determined to present the voters with a clearcut alternative to Social Democratic policies in the 1960 election. Its campaign platform comprised several cuts in the social welfare program, but two proposals became especially controversial. One of these was the Conservative Party's demand that the newly enacted "Supplementary Pensions Scheme" (the "ATP") should be abolished. The other was a proposal to limit government support to families with minor children through reduction in the "children's allowances" scheme. In effect, the Conservative strategy, promising though it may have seemed to the party leadership, led to a severe split in the Bourgeois

opposition and an electoral setback. During the campaign the People's Party and Center Party worked together, resulting in a joint condemnation of the Conservative Party's "extremism" in the field of social welfare politics. In terms of policy standpoints, the People's Party was opposed to both of these Conservative proposals. The Center Party attacked the Conservative position on the children allowances question. The Center Party, on the other hand, had a record of outright hostility to the supplementary pensions scheme and never formally reversed its position before the 1960 election. However, it sought to keep the pension question in the background in its campaign appearance to ease its new rapprochment with the People's Party.²³

Voter opinions over the two predominating social welfare issues divided largely along party lines. Nevertheless, among supporters of the Bourgeois parties, we find much intra-party dissension over party standpoints. We will now focus on this aspect of the 1960 campaign. The data about the division of voter opinions on the child allowances issue in the percent distributions below may be taken as the point of departure.

The same picture appears generally in the data on the pension question (which we shall deal with in more detail in the following section). Social Democratic voters showed high partisan cohesion in both of the social welfare controversies, while substantial minorities within the electoral support of the Bourgeois parties deviated from party lines.

Table 9 illuminates the overall relationship between opinions on these issues and the voters' general attitudes toward welfare state politics. For comparison, we also have included the corresponding data for the sales tax issue. The relationship visible in the table is strong enough to support the assumption that the 1960 campaign brought differences in ideological orientation to the surface. This becomes especially clear if we consider voters' opinions on both of the social welfare issues (see third row in table). In a sense, this should be expected because of the nature of these issues and because party standpoints coincided neatly with the parties' location along the right-left dimension in the party system. How, then, could this seemingly neat arraying result in a considerable amount of disagreement between representatives and constituents? For an explanation, we turn to intraparty variation in general attitudes and issue opinions (Table 10).

In Table 10 we have consolidated into one simple opinion index the respondents' views on the two social welfare issues. Because of the somewhat ambiguous position of the Center Party on the pension question, we have included only the three remaining major parties.

The index has been constructed so that partisan conformity would have required all Conservative voters to be included in the "negative" category, while all People's Party voters ought to be found in the "positive" category (There would still be room for some ambiguity in opinions). As is seen in Table 10, the reality was, indeed, quite different from such a perfect accordance between party leaderships and their electoral followings. Furthermore, the data bring out one significant effect of the campaign situation in 1960. To a large extent, Conservative voters were drawn to the positions taken by the other Bourgeois parties. This trend was most pronounced among those who were least right-oriented in their general attitudes. Among People's Party voters, on the other hand, the right wing was attracted to the standpoints of the Conservative Party.

Party affiliation, nevertheless, made an independent impact. Along the entire attitude continuum, People's Party voters prove to be more positive toward the social welfare policies than the adherents of the Conservative Party. To make clear the impact of party allegiances, it should be noted that both issues involved a change in party positions: The cut in the childrens' allowances program was a new standpoint for the Conservative Party, and the People's Party only a year earlier had been opposed to the pension scheme it supported in 1960.

Since social welfare reforms have formed the cornerstone of Social Democratic post-war politics, the Social Democratic Party was capable of enlisting a more wholehearted allegiance than any other party when its achievements seemed to be threatened. However, even among Social Democrats, a noticeable relationship exists between consistency in political outlook and opinions on policy questions. To make this more visible, we have given the opinion index a finer gradation pertaining to Social Democrat voters.

Partisan Competition and Electoral Opinion Formation

The pattern of intra-party split over policy questions among adherents of the Bourgeois parties that appears in our data from the 1960 election has much to do with a basic precondition for electoral competition in the Swedish multi-party system. The Conservative Party and the People's Party — and to a lesser extent the Center Party — recruit major portions of their voting strength from a shared electoral base in the urban middle class and from the same right-center part of the entire right-left attitude distribution in the electorate. Even though the Conservative Party has a special stronghold among the most well-to-do and most right-oriented Bourgeois voters, a fair amount

of variation in basic political attitudes cuts across Bourgeois party lines. Thus, an opinion situation of the 1960 type is likely to emerge whenever the Bourgeois parties choose to compete with each other by taking distinctly different standpoints in major social welfare issues. In fact, the dynamics of partisan competition between one rightist and two centrist parties - all of which are attempting to work out alternatives to Social Democratic policies and striving to enlarge their own share of the Bourgeois vote - have resulted repeatedly in this situation in Swedish politics. The effects become visible in the form of a peculiar changeability in their electoral base. Due to the continuous exchange of voters between the Bourgeois parties, their voting support in any given election will comprise a much greater proportion of voters who have changed their party preferences in recent elections than will support the Social Democratic party. Closely related is that the Bourgeois voters - as proved through a decade of election surveys - are less likely than Social Democrats to have formed a strong sense of allegiance to any political party.²⁴

The Social Democrats are by no means unaffected by overlapping electoral bases comprised in the party system, or by intra-party variation in political attitudes. The reason why its position is generally less vulnerable is that the parties tend to be grouped in the form of a two-bloc system, where the Social Democrats are entirely predominating in one of the blocs while the other bloc is divided into three parties of about equal size. Because of the relative insignificance of the Communist Party during our study period, the Social Democratic Party was much less restrained by intra-bloc competition in its choice of policy standpoints than were the Bourgeois parties. The People's Party, for instance, had to be concerned with the risk of losing marginal voting support to the right as well as to the left. The Conservative Party, on the other hand, rrust attempt to preserve its identity to the right of center while also avoiding to place itself too far to the right. The Conservative's misfortune in the 1960 election is evidence that this position entails tactical hardships of ils own.

To some extent, the parties thus are constrained in their policy formation because they are located differently along a scale of ideological positions and corresponding configurations of group interests.²⁵ Since any transparent relationships seldom exist between such "roles" in the party system and short-term politics, every political party also has considerable freedom to choose its policy decisions. That is as long as a party does not place itself "on the wrong side" of a competitor to the right or to the left, there is room for tactical considerations. For obvious competitive reasons, each party will strive to balance its appearance in important policy questions in such a way that it can appeal effectively to voters with differing attitudes and interests within its current and potential electoral base. This means, first, that one can never connect all the policies pursued by a given party with any fixed point on an attitude scale; instead, they are deliberately dispersed.

Second, it means that the parties themselves act so as to attract voting support
representing a wide range of attitude variation.

When parties have determined their standpoints, a profound influence is exerted upon the division of opinions in the electorate. As we have seen, the impact of party allegiance, however, is not so pervasive as to entirely outweigh other components in the individual's political outlook. Hence, the degree of concordance between party standpoints and voter opinions will vary from question to question.

Though we will not investigate the effects of partisan change here, it should be noted that this contributes to reduce intra-party dissension. However, even issues that split a party severely will rarely cause more than a part of the dissenters to switch votes to another party. The consequences, nevertheless, can be politically important. During the period dealt with here, controversies over social welfare policies produced electoral defeats for the People's Party in 1958 and for the Conservative Party in 1960. In both cases, the parties involved decided to shift their grounds when they found that their policy positions met an unusually unfavorable response at the polls. Thus, an election reverse can also have an ex post facto influence upon party standpoints.²⁶

Since the representational function of democratic elections was our point of departure, we will finally attempt to make a summary evaluation of the relationship between the voters' party choice and their policy opinions. The opinion index in Table 11 serves this purpose. Of course, the aggregation of voter opinions that forms the basis for Table 11 provides a somewhat artificial measure. We have only added responses to our interview questions, whereas the individuals themselves also must have weighted their various opinions on party policies. On the other hand, it might be argued that votes are counted and not weighted in the normal democratic process.

Table 11 shows only a minority of the voters supported their parties in all
of the three issues which were most intensively debated in the 1960 campaign.

For reasons already discussed, partisan cohesion became especially strong among Social Democrats, as is reflected clearly in the table. We also can summarize the voters' opinions in another form. If all outright disagreements with party positions are added up - not counting cases of "no opinion" as genuine disagreement we find 49 percent of those who voted in the 1960 election were at variance with their party on at least one of these major issues. We can assume that the proportion of voters expressing views conflicting with party lines would have become even larger if we had included a wider range of issues in the analysis.²⁷

When presented in this form, the data almost gives an impression of a political process leading to "non-representation" rather than representation of voter demands. In reality, the opposite was true with regard to "the supplementary pension scheme" as well as "the childrens' allowances". If we examine the distributions of opinions in the electorate as a whole, we see the parliamentary majority, indeed, did obtain an impressive popular majority support in both cases. Among all respondents in our nationwide sample, 62 percent favored the pension scheme and 63 percent approved the current system of support to families with children. As for the sales tax, on the other hand, the Social Democratic government obviously was not able to find such a majority endorsement, although it achieved a remarkable opinion chgange among its own electoral support. Thus, no "built-in" mechanisms in representative democracy exist that could guarantee a complete correspondence between policy decisions and voter demands. In a broader perspective, however, one could interpret the 1960 election as the citizenry's verdict over the Conservative Party's attempt to launch a new and more restrictive general policy in the field of social welfare. The voters did not like it, and the voters got their way.

Pension Politics: The Course of Opinion Formation, 1956-1964

The controversy over the Supplementary Pensions Act in the 1960 election campaign was the last stage of a drawn-out decision-making process that went through a consultative referendum in 1957 and a sequence of election contests before the pension scheme finally was enacted in 1959. The decision was carried through with the barest possible majority in Parliament by the Social Democratic government over the combined opposition of the three Bourgeois parties. In 1960 the situation underwent a profound change. The People's Party shifted sides and, thereby, in practice, guaranteed the continued existence of the pensions system. A widening consensus was already being formed.

The supplementary pensions system was to complete the existing National Pension scheme with income-graded pensions for all employees and, in effect, most enterprisers, up to a level where the goal to secure a sufficient income for the retired could be considered achieved. It was a social welfare

project, but it also comprised another component that contributed to make it politically controversial. The new pension system presupposed the creation of great pension funds controlled by public boards. Indirectly, the pension reform opened the way for extensive government control over the capital market.²⁸

The course of events was unusually dramatic in the pension conflict, but the underlying gradual adjusting of party positions until a consensus could be formed is an example of a process which certainly is not unusual. As in the case of the supplementary pensions, such a process may open with a new policy scheme engendering a partisan polarization of opinions within a stratum of politically engaged voters, while the opinion situation remains inarticulate in less attentive strata. Through a heightening of the level of public attention, the question may then be transformed into a major issue that mobilizes mass interest in the electorate. For a time then, it will be the object of a tense political cleavage where ideological principles are invoked and organized group interests become engaged. Finally, a decision is reached. The new policy measure becomes an established fact and, after some time, will be met with a general "non-ideological" acceptance. Using a term introduced by Berelson eta/, in a study of political opinion formation, we could speak of such a process as the "life history" of a political issue — it dissolves as an issue when party politics turn the focus of attention to new sources of partisan conflict and new battlegrounds for electoral competition.²⁹

In the following analysis, we will attempt to highlight the broad trends of opinion formation that mark the various stages of the pension conflict. In particular, we shall be concerned with the interplay between party politics and the shaping and reshaping of opinions in the mass electorate. To include the final stage of this process, it will be necessary to go outside the time span otherwise covered in this article. We shall utilize some data drawn from an interview sample survey of the 1964 election.³⁰

As in the foregoing analysis, party allegiance and attitudes toward welfare
state politics will be employed as explanatory variables. To these we now
add socio-economic status as a conditioning factor.

Indeed, only a modest understanding of the competing pension schemes was required to see that they were to affect different population groups in unequal ways. Without question, the compulsory insurance system espoused by the Social Democrats afforded the most substantial improvement for manual workers and other wage-earners who had no prior supplementary pension arrangements. Large groups of salaried employees, on the other hand, were already included in some existing pension scheme; they were worried that their established pension benefits might be impaired if they were to be transferred to an all inclusive system. Enterprisers were naturally apprehensive since they should bear the immediate costs of the supplementary pensions for their employees.

There also were alternatives to the Social Democratic plan which were favored by enterprisers' and farmers' organizations and several white-collar unions. Though differing in construction, they had in common the principle that the supplementary pensioning question ought to be solved through voluntary arrangements for individuals or groups who were willing to participate. Besides having less to gain from the Social Democratic plan than manual workers, enterprisers and salaried employees obviously had strong reasons to be more responsive to these voluntary schemes. Thus, occupational status defined a tangible and well-understood self-interest in the pension question.

White-collar and workers categories did not fit this simple classification as well. By and large, there appeared a cleavage of opinions between middle class and working class that coincided with the alignment of socio-economic groups underlying the Bourgeois-Socialist division of the electorate.

Table 12 gives an overview of opinion development in the pension question
from 1956 to 1964 on the basis of our four nationwide surveys.

When the proposal to construct a compulsory superannuation system was first brought up in the 1956 campaign, it still belonged to the type of campaign issues which can iduce party militants and activists in interest organizations to commit themselvs but which otherwise can attract comparatively little attention. Interviews conducted immediately after the election showed that only 43 percent of the men and 20 percent of the women voters knew that the Social Democrats had been campaigning for a compulsory supplementary pension system.³¹. When asked in the same interview to point out from a list of 16 issues the three or four they considered most important, only 15 percent mentioned the new pension proposal.³² That proportion should be compared to the 48 percent who mentioned a current proposal to increase National Pension benefits (in that context not connected with the supplementary pension question). In the same range came only taxation (60 percent), shortage of housing (49 percent), and a proposed decrease in the work-week (28 percent). Other matters obtained about the same or lower rate of interest as the supplementary pension question.

At the time of the 1956 election, the Bourgeois parties already had decided
to oppose the Social Democratic proposal. It was clear that they favored,
as an alternative, a voluntary system.

The data presented in the upper part of Table 12 were obtained in an interview shortly before the 1956 election. Respondents were shown brief statements of the two main alternatives and asked to indicate which they preferred. (The phrasing of the question is given in note to Table 12.) Opinion distributions among Bourgeois and Socialist voters were distinctly different already at this stage. However, allegiance to party lines was far from complete.

By using the same question and our measure of intensity in party identification,
we can gauge the importance of these factors in the 1956 campaign

(Table 13). These factors had a crucial relationship to the voter's opinions
on the supplementary pensions questions. Among those informed that the
compulsory plan had been proposed by the Social Democrats and also had

a strong party attachment, the proportion supporting their party's view was about 80 percent. The propensity to agree with party lines was much weaker if one factors was absent. In the six table categories comprising voters who were uninformed or weak identifiers, the proportion supporting the standpoint of their party is between 49 and 59 percent. The effect of the combination of information and strong party identification was to reduce both the proportion giving ambivalent responses and the proportion of "defectors." (About 2 percent of those who were both informed and strong identifiers gave ambivalent responses as compared to about 10 percent in the other categories.)

Even if a voter is not familiar with current party standpoints, be may still
be capable of discriminating between policy measures agreeing with his
general political orientation and those which do not. That this is to some

extent true becomes clear when examine the division of opinions among the "uniformed" respondents. Even in this group, a majority of 54 percent as supporters of the pension plan favored by the party they supported in the 1956 election, while 36 percent preferred the opposite alternative.

The situation altered drastically at the time of the consultative referendum in 1957. The voters now were offered the choice between the Social Democratic scheme and two "voluntary" plans. One of the latter was supported by the Conservative Party and the People's Party, while the other was espoused by the Center Party. The scheme proposed by the Conservatives and the People's Party presupposed that a supplementary pension system should be built up through collective agreements between the parties on the labor market. The plan implied no state-controlled organization and only a minimum of legislation to facilitate such agreements. According to the Center Party plan, on the other hand, neither the state nor the labor market organizations should engage in any form of collective arrangements. Instead, individual pension insurances should be supported in various ways. The Center Party plan had certain features particularly attractive to small businessmen and farmers. These properties paid off politically in a considerable defection of former Conservative and People's Party voters to the competing Bourgeois scheme.

By the time the referendum approached, the public had become wellinformed about the relationships between pension schemes and political parties and interest organizations. Among those who had been eligible to vote on some earlier occasion, 83 percent were informed about the position taken by the party they had voted for in the last previous election in which they had participated. The turnout (72 percent) fell somewhat below the typical level for an election, but otherwise public interest had become mobilized to make the situation similar to a regular election campaign.

The most characteristic feature of the data on voting behavior in the referendum (Table 12) is perhaps the great majority of the electorate rallying around party lines. Among those who participated in the referendum and had previous party affiliation, as much as 79 percent supported the referendum proposal sponsored by the party they had voted for in the preceding election. Thus, the power of the party system to canalize popular opinion formation made itself felt even though the decision to be made was not a choice between parties but referendum alternatives.

It follows from our reasoning that to a predominating extent, interest organizations operated to strengthen the impact of party allegiance, at least in respect to the main contest between the compulsory and voluntary superannuation systems. The two components, i.e., group influence and party allegiance, are not easily separated from each other. Most "middle-and-upper" class voters were adherents of a Bourgeois party and most workers had supported a Socialist party in previous elections, and they behaved accordingly in the referendum. There was, however, a none too small minority of voters who were

exposed to a conflict between party allegiance and the concept of self-interest
that prevailed in their own occupational group.

The nature of such a conflict can be given in a nutshell: employees who did not already have satisfactory superannuation benefits were attracted to the Social Democratic scheme, while others were inclined to prefer a voluntary system that would not affect their own pension conditions. The result is shown in Table 14. (As in Tables 12 and 15, respondents have been classified according to party preferences on the basis of their most recent electoral choice.) As would expected, party allegiance and self-interest are found to be counterpoising in effect among individuals caught in a conflict situation.³³ Nonetheless, the voter's partisan orientation must have had an exceptionally important and often decisive influence. Even in categories where a psychological conflict was present, the balance of opinions became markedly weighted 'n favor of party conformity. This inference is further substantiated by the summary of long-term trends given in Table 15.

In Table 15 we use both the voters' party preferences and their occupational status as the criteria for a classification of the respondents in the 1956, 1957, and 1960 surveys." Workers already in 1956 were more favorable than any other occupational group to a comprehensive and compulsory pension scheme for all employees, while enterprisers responded negatively to such a plan. Employees in whites-collar occupations, being a more heterogeneous category than the workers, sihow an intermedite opinion distribution. A negative attitude toward the compulsory system, however, was the preponderant view among salaried employees. At the same time, the data substantiate our hypothesis that opinions in the pension question were determined by a combination of influences originating from party affiliation and socio-economic antecedents. This is clearly illustrated in the table columns where broader

social strata are sub-divided according to voters' party preferences. Although the parties met much difficulty in mobilizing support from the periphery of their electoral bases, the major part of the Social Democratic middle class voters endorsed the compulsory system just as most of the workers with Bourgeois party preferences took the opposite standpoint. This pattern appears in the 1956 data, but it is even more conspicuous in the 1957 data since opinions were much more sharply polarized at the time of referendum.

A comparison of the data from 1956 and 1957 discloses another outstanding feature in the opinion trend. Through all the combinations of party preferences and social status in the table, we find that voter opinions in 1957 had become divided along the same lines as displayed by the informed voters in 1956. We observed earlier that information about party standpoints contributed substantially in 1956 to generate a partisan polarization of the opinions. The same picture reproduced, almost photographically, on an enlarged scale at the end of the 1957 campaign provides the basis for the inference that the same factor was operating in both cases. It, however, influenced a much greater part of the electorate in 1957.³⁵ Thus, the main effect of the referendum campaign was to make most of the public disposed to evaluate the pension schemes in light of their partisan orientations. With a concept borrowed from V. O. Key, we could describe this process as an "activation" of a "latent" division of opinions.³⁶

The result of the referendum was inconclusive. The Social Democratic scheme obtained 46 percent of the vote. The pension plan supported by the People's Party and the Conservative Party received 35 percent. The Center Party proposal received 15 percent (the remainder being blank ballots). In comparison with the election in 1956, the outcome could be considered a setback for all parties except the Center Party. Nevertheless, the Social Democrats regarded their support as sufficient to carry on with their scheme. The Lower Chamber was dissolved, and the pension question became the major theme of the campaign before the ensuing extraordinary election in the spring of 1958. In the election contest, a regrouping took place among the Bourgeois parties. The People's Party devised a compromise plan which comprised som© of the features of the Social Democratic scheme. The Conservatives chose to draw closer to the Center Party's negative position. The parliamentary election brought moderate success for the Social Democrats and heavy defeat for the People's Party. As noted before, a supplementary pensions act based on the Social Democratic plan could then be enacted by Parliament. The People's Party, which had to risk losing votes to the left or right or possibly both if the pension conflict were to go on, declared soon after the Parliament decision that it was ready to accept the pension system thus established. The party's new attitude was justified by the argument that any attempt to dismantle the supplementary pensions system, once it had been put into effect, would lead to insurmountable difficulties on the labor market.

We already have touched on the popular majority support of the continuance of the country's new superannuation system at the time of the 1960 election. The data in Table 15 disclose that from 1957 to 1960 a distinct swing in the state of opinion took place in all categories reported. In 1960 a favorable majority appears among enterprisers and salaried employees as well as workers. Farmers were, however, still predominantly negative, and the same is true for a sub-category not separately accounted for in the table, i.e., enterprisers who supported the Conservative Party.

Data in Table 12 show that a majority of Center Party voters were still opposed to the pension system, while Conservative voters were equally divided. The decisive change in the balance of opinions occurred among the People's Party voters with a heavy majority supporting the party's new position.

When comparing data from different points of time in Tables 12 and 15, we must remember that ihe opinion distributions within the parties were affected by voters who changed their party preferences. The People's Party in 1958 must have lost a considerable number of votes to the Social Democratic Party because of the pension controversy. Likewise, it can be presumed that in 1960 the Conservative Party lost voters who considered their former party's pension policy as ntransigentce rather than realistic opinions; most of them must have switched to the People's Party. Flows of votes, of course, have gone in other directions too. Unfortunately, no comparable data lexists to determine which came first: the change in the People's Party's pension standpoint or the opinion »wing among People's Party voters. However, poll data from the spring of 1958 indicate that virtually no support for the Social Democratic scheme existed among Bourgeois voters at that time.³⁷ Furthermore, our survey data show that no significant flow of voters came from the Social Democratic Party to the Bourgeois parties between 1958 and 1960. Hence, a major opinion change must have taken place among voters who had supportetd a Bourgeois part/ in 1958 and remained non-socialist voters in 1960. It is also clear that a dramatic change occurred among consistent People's Party voters.³⁸

As in the previous stages, information about the parties' views could supply the voter with reference points. In the interview survey, respondents were asked whether they knew if any party wanted to repeal the pension act.³⁹ One can presume that familiarity with the new constellation of party positions was especially important in 1960. Most voters must have known that the Social Democrats were responsible for the decision to introduce the new system, but fewer were clear about the standpoints taken by opposition parties.

We can take People's Party voters as an example of the information effect. Among those giving a correct answer to the information question, 73 percent supported their party's new standpoint, while 14 percent were opposed and 13 percent were ambivalenl or lacked any opinion. Among the "uninformed,"

on the other hand, only 46 percent wanted to maintain the pension system,
21 percent would rather have it abolished and 33 percent had no preference.
About the same pattern appears for other parties (Table 16).

Thus, our data supports the inference that the People's Party contributed in a crucial way to the change-over in the opinion situation by switching from opposition to support of the supplementary pensions system. Among the party's adherents, the implications of party allegiance were evidently shifted when the party line shifted. Yet this cannot explain why a change in the same direction, although somewhat weaker, also occurred among Conservative voters. Since intra-party variation in views on social welfare policies had its roots in the voters' general attitudes toward welfare state politics, we must now examine how the latter factor was related to the opinion trend that appeared among Conservative and People's Party voters.

In table 17, the 1960 supporters of the two parties under study have been arrayed along our right-left attitude scale. For each attitude category the table indicates the balance of opinions in 1957 and 1960 toward the compulsory pension scheme. We must rely solely on data from the 1960 survey because the 1957 study did not comprise the general attitude measure

required for this analysis. Data on the respondents' referendum voting have been obtained through a retrospective interview question. The methodological weaknesses in this technique are obvious. Since it also leads to numerically fragile sub-categories, we cannot expect high precision from this measurement.⁴⁰ It is, however, sufficiently robust to allow a consistent pattern to emerge.

Data in Table 17 reveal that a strong plurality among supporters of both parties were opposed to the Social Democratic scheme in the 1957 referendum. Furthermore, in 1957 the voluntary pension principle enlisted a preponderant support in all categories on the attitude scale, although the data indicate a tendency that the balance of opinions was more one-sidedly weighted to the right than to the left. At that stage, there was no great difference between Conservative and People's Party adherents as long as categories with the same position on the attitude scale are compared. At the time of the 1960 election, a genera.l shift in the positive direction had taken place. The relationship between attitude toward welfare state politics and pension opinions also persisted in this situation. Now, however, striking dfference appears between Conservatives and People's Party voters. Irrespective of their general political orientation, the latter were much more disposed to change their views to approve the pension system.

Particularly significant in the change from 1957 to 1960 is that the shift proves to be so strong and so even along the whole attitude continuum when we compare 1960 opinions with the 1957 baseline. (See the "Change in Balance" rows for both of the parties.) Neither party allegiance nor attitude toward welfare state politics can account for this over-all switch. This must be attributed to some "additional factor" affecting both those who were "rightist" and those who were more "leftist" in their political orientation among Conservatives as well as among People's Party voters. Although it cannot be fully substantiated by the survey data, we venture the inference that the implementing of the new pension system had in itself a powerful impact.

Through the preceding stages of the pension conflict, the choice was between different principles; and different plans which no one could have observed in practice. In 1960 an immediate adaptation to the new situation took place in the labor market. This involved large-scale collective bargaining between the employers' organization and the organizations of employees who had been included in previously existing pension systems. Well before the 1960 election, these transition questions had been settled on terms generally considered advantageous for the employees — they got a considerable wage increase to compensate for the relative improvement other groups had obtained by the introduction of general pension benefits. After that, none of the parties on the labor market showed any enthusiasm for experiencing a new round of bargaining. As a consequence, no major interest group had any desire to keep the pension conflict alive. A new constellation of group interests, thus, had become bound up with the continuance of the pension system. Individuals also were afforded a basis to re-interpret their self-interests: The same anxiety

over a change with uncertain consequences that in 1957 made large groups of white-collar employees apprehensive to the Social Democratic referendum scheme could now be an argument favoring maintenance of the new system once it had begun to operate.

For People's Party voters, this situational change and party allegiance were concurring in effect. For Conservative voters, party allegiance was a restraining factor. Table 17 shows that the voters' general attitudes toward welfare state politics did not generate the change. Rather the role of this factor was to determine the baseline for the change. Finally, it is in accordance with our interpretation of the data that the most moderate swing occurred among Center Party voters (cf. Table 11). The electoral support of the Center Party comprised a preponderant proportion of the groups least affected by agreements between organizations on the labor market, i.e., farmers, small businessmen, etc.

Four years later, in 1964, all of the parties had committed themselves to maintain the supplementary pensions as a comprehensive and compulsory scheme. A consensus was being formed over a new link in the Swedish social welfare system. The data in Table 12 illustrate the accompanying process in the electorate: The once distinctly partisan polarization of opinions had been eroded. From the Conservatives to the Social Democrats, a heavy majority now supported continuance of the new pension system. In the electorate as a whole the majority reached 80 percent, while only 7 percent were left in outright opposition.

Differences in opinions still existed below the surface. In the 1964 survey, respondents were not only queried about their preferences in the maintenance of the pension system but also were asked to indicate how strongly they felt. Table 18 employs the right-left attitude scale to show the underlying differentiation of opinions that becomes visible when accounting also for intensity in opinions. In 1964 a majority of "favorable" opinions existed along the entire attitude continuum. Even in the "right-most" category, this majority amounted to 69 percent. However, the degree of concern about the supplementary pensions system shows a gradual decline from the left to the right. The voters' views on the supplementary pensions were not completely disentangled from the ideological context which had been so salient in party propaganda through the earlier stages of the pension conflict.⁴¹

The bottom row of Table 18 pertains, in a sense, to an new issue. In the 1964 campaign, the Conservative Party launched a proposal directed at the component in the pension system that from the beginning had been disliked most intensively by the Bourgeois parties, that is, the government-controlled pension funds. It was no longer questioned that the supplementary pensions should be retained for all employees. Indeed, the Conservatives proposed improved pensions for old-age groups which were not entitled to full pension benefits under the original plan. The aim was to reduce the growth in the pension funds to limit government influence on the capital market. Since the Conservative plan did not refer primarily to the supplementary pension

system as such, we will nol deal in detail with this new political issue here. It is sufficent to note (cf. Table 18) that the right-left differentiation in the electorate appeared again, as distinctly as ever, in the ensuing division of opinions.

In retrospect the struggle over the pension system may appear difficult to account for. A policy measure which becomes so widely accepted after a few years cannot, it may be asserted, have had the profound importance originally attributed to it. Such an interpretation would be based on a misjudgement of the true character of the pension conflict, however. If the parties are to exert any influence on the broad trends of social development, they must select some questions; and place them before the citizens as crossroads where they have to decide kind of society they want to live in. On the other had no strong reason to seek a mutual understanding when a change in can be broken or initiated at several points of time. The parties must decide when and where they can anticipate to win electoral approval of the policies they want to pursue. Considerations about voting support in "marginal" population groups may, in some situations, lead one or more of the parties to favor a compromise rather than to evoke a clashing of interests or principles even when potentially controversial measures are to be decided. On other occasions, the political context may induce the parties to avoid seeking a compromise. This obviously was the case after the election setback for the coalition government in 195(3. The People's Party and the Social Democrats had no strong reason to seek a mutual understanding when a change in

government seemed to be approaching. The Center Party felt a strong need
to regain a political profile of its own. For too many parties, tactical needs
were incompatible with a compromise on the pension question at that stage.

That the Bourgeois parties, one after the other, preferred to shift their ground of opposition rather than to fix electoral competition in an area where they had been unsuccessful is not evidence that the principles involved in the pension conflict were insignificant. Nor does it prove that social and political cleavages underlying the party system had vanished. They survived but they became related to new issues when the political scene was changed. As this process continues, political parties will often serve as points of orientation for electoral opinion formation. As we have seen through the "life history" of the pension question, however, other factors also will enter into the process and influence the popular division of opinions.

---

NOTES

1 V. O. Key Jr, "A Theory of Critical Elections", The Journal of Politics, vol. 17, February 1955, pp. 3-18.

2 A. Campbell, P. E. Converse, W. E. Miller, D. E. Stokes, The American Voter, John Wiley, 1960, p. 521 ff. See also: Campbell, Converse, Miller, Stokes, Elections and the Political Order, New York—London—Sidney, John Wiley, 1966, Part I, pp. 7-157.

3 80 Särlvik, "Political Stability and Change in the Swedish Electorate", Scandinavian Political Studies, vol. 1, 1966, p. 188 ff.

4 A description of the design and sample of the 1960 survey is found in: S.O.S. Allmänna val. Riksdagsmannavalen åren 1959—1960, Part 11, Stockholm 1961. The 1957 survey is described in: Bo Särlvik, Opinionsbildningen vid folkomröstningen 1957, 5.0.U., Stockholm 1959. See also: Jörgen Westerståhl—Bo Särlvik, Svensk valrörelse 1956, (Swedish Election Campaign 1956), Research Report from the Institute of Political Science, University of Göteborg, 1957 (mimeo).

5 For a discussion of the characteristics of multi-party systems, see: M. Duverger, Political Parties. Their Organization and Acitivity in the Modern State, London, Methuen, 1955, p. 228 ff. See also: A. Downs, An Economic Theory of Democracy, New York, Harper, 1957.

6 Särlvik, "Political Stability and Change", Loc. cit. p. 193 ff.

7 See table 17, p. 217 in: Särlvik, "Political Stability and Change". For the socioeconomic classification, an office memorandum prepared by the Swedish Central Bureau of Statistics was utilized. For the purpose of the analysis of the survey data, we have made several modifications in the Central Bureau's classification, e.g. the division of the working class used in Table 1 in this article. The categories in the table have been obtained by combining several sub-groups. See also comments to the socio-economic classification in the article cited above. A somewhat different classification is presented in the author's reports on the elections surveys in 1960 and 1964 in the official election statistics publications; see: S.O.S. Allmänna val Riksdagsmannavalen åren 1959—1960, 11, Stockholm 1961, and Riksdansmannavalen åren 1961—1964, II Stockholm 1965.

8 A technical description of the attitude index is given in: Bo Särlvik, "Skillelinier i valmanskåren", Statsvetenskaplig Tidskrift, 1965: 2—3. In Table 2, Social Democratic and Communist voters have been put in a common "Socialist" category. In much of the following analysis, Communist voters will be excluded because of the numerical smallness of this category.

9 V. O. Key Jr, Public Opinion and American Democracy, New York, Knopf, 1961 (p. 458 ff, esp. p. 460), and V. O. Key Jr, with the assistance of Milton C. Cummines, The Responsible Electorate. Rationality in Presidential Votinn 1936—1960. Cambridge, Mass., Harvard Univ. Pr., 1966. In the latter work, Key has brought together a massive collection of survey data to support his conclusion that there has, indeed, been a strong element of "rationality" in the trends of partisan

---

change during the period he has investigated. Cf. also: A. Campbell et. al., opera cit.: these studies also demonstrate how the electorate's responses to various events and issues have affected the outcomes of recent presidential elections in the U.S.A.

10 Data on changeability in voting behavior are presented in: Särlvik, "Political Stability and Change", pp. 193—197. See also the transition tables for the elections in 1958—1960 and 1962—1964 in the election statistics reports; Riksdagsmannavalen åren 1959—1960, 11, and Riksdag smannavalen åren 1961—1964, 11.

11 For a full treatment of the concept of party identification, see: A. Campbell G. Gurin, W. E. Miller, The Voter Decides, Evanston, 1954, p. 88 ff. and Campbell, Converse, Miller, Stokes, The American Voter, p. 121 ff. See also: R. E. Lane, Political Life. Why People Get Involved in Politics, Glencoe, 111., Free Press, 1959, pp. 299—302. The theory of reference groups is treated in: R. K. Merton, Social Theory and Social Structure, New York, Free Press, rev. ed. 1966, pp. 300 ff.

12 The 1956 and 1960 surveys do not comprise any measure of negative party identification. As this author will demonstrate in another study, based on data from a survey of the 1964 election, negative identification toward political parties is an important factor in the formation of partisan attitudes.

13 Campbell et al., The American Voter, pp. 185—187.

14 The same measure was used in: Särlvik, "Political Stability and Change", pp. 196—197, p. 206 ff. See also: Särlvik, "Skiljelinjer i valmanskåren", p. 165. The interview question was framtid as follows: "Some people are strongly convinced supporters of their party. Others are not so strongly convinced. Do you consider yourself as a strongly convinced supporter of some political party?"

15 See: H. Eulau, Class and Party in the Eisenhower Years, Glencoe 111., Free Press 1962, esp. p. 59.

16 Key, Public Opinion and American Democracy, p. 182 ff.

17 Relevant cognitive theories are presented in: L. Festinger, A Theory of Cognitive Dissonance, Stanford Cal., Stanford Univ. Press, 1957, and in: M. J. Rosenberg et al., Attitude Organization and Change. An Analysis of Consistency Among Attitude Components, New York and New Haven, Yale Univ. Pr., 1960.

18 Since respondents have been classified according to their party preferences at the time of the election survey, all three party categories include persons who changed parties between 1956 and 1960. A detailed examination of the 1960 data however, shows that the trends discussed in the comments to the table are not affected by party changers to any appreciable extent. — In the case of "housing shortage" the change among Center Party adherents is relatively modest. This is in itself an interesting observation. As the shortage of housing is, primarily, an urban phenomenon, the countryside voters had less reasons than others to be concerned about that question.

19 The measure of Campaign Exposure is presented in: Särlvik, "Political Stability and Change", p. 206, see also note 29, p. 220.

20 For a discussion of "polarization" trends during an election campaign, see: B. R. Berelson, P. F. Lazarsfeld, W. N. McPhee, Voting. A Study of Opinion Formation in a Presidential Campaian, Chicago, Univ. of Chicago Press, 1954. A theoretical model of the effect of increasing public attention to a political issue is propounded in: W. N. McPhee and W. A. Glaiser, Public Opinion and Congressional Elections, 1962, p. 102 ff. Concerning the assumption of a "natural" negative response toward a tax increase, cf. Campbell et al., The American Voter, p. 195 ff.

21 It should be noted that the Social Democratic Party's election victory may have contributed to convince its adherents that their party had taken the right position in the tax issue. Such an effect would not contradict our hypothesis about the influence of party allegiance, but it would mean that the comparison between the two interview occasions overestimates the change that took place during the campaign itself. The classification according to party preferences in Tables 5—7 is based on actual voting behavior as recorded in post-election data. Change in voting intentions, of course, must have contributed to an increase in partisan conformity with regard to issue opinions. Examination of the data obtained from the subsample interviewed before the election proves, however, that those who changed there voting intentions were too few to influence the trends discussed in this analysis.

22 In the pre-election interview, 80 % of the men and 57 % of the women were informed that the Conservative Party had proposed a cut in the child allowances program; in the post-election interview these proportions had increased to 89 % and 73 %, respectively. Data on the information level in the pensions question are discussed in the following section of this paper.

23 The leaders of the People's Party and the Center Party announced their

---

standpoint through a joint declaration released to the press during the election campaign. Their disagreement with the Conservative standpoint was furthermore stressed through statements in radio and television appearances (cf. statements by the Liberal leader, September 9 and 16, and by the Center leader September 15 and 16, 1960; verbatim records released by the Sveriges Radio.)

24 Cf. Särlvik, "Political Stability and Change", p. 197. The same pattern appeared in a consecutive study of the 1964 election.

25 A theoretical model for partisan competition in a multi-party system which stresses the restrictions imposed by the system upon the parties' policy formation is presented in: Downs, An Economic Theory of Democracy, see pp. 125—127 et passim. See also: B. Molin, Tjänstepensionsfrågan. En studie i svensk partipolitik. (The Supplementary Pensions Question. A Study in Swedish Party Politics.) Göteborg, Akademiförlaget, 1965, esp. p. 141 ff.

26 The 1960 election was held September 18. On October 10—11, the Parliamentary Conservative Party and the National Committee of the party organization came together to review the situation. The meeting announced that the Conservatives had decided to abandon the demand for a repeal of the supplementary pensions act.

27 When it comes to policy questions where the attention level is much lower than in the the cases dealt with in this analysis, it is obvious that the relationships between voters' views and policy decisions must become even more attenuated. In addition, we encounter the methodological problem of ascertaining whether the survey data actually record "genuine opinions". For a discussion of the applicability of the majority rule on opinion formation in the mass electorate, see: R. A. Dahl, A Preface to Democratic Theory, Chicago, Univ. of Chicago Pr., 1956, esp. p. 124 ff.

28 As the development of the pension conflict on the level of parliamentary party politics has been trated in another article in this yearbook, we shall deal very summarily with that aspect here. See: B. Molin, Swedish Party Politics: A Case Study. Scandinavian Political Studies, Vol. I, 1966. Changes in the distribution of votes among the parties through 1956—1962 are analyzed in: Särlvik, "Political Stability and Change". A detailed analysis of the referendum campaign in presented in: B. Särlvik, Opinionsbildningen vid folkomröstningen 1957 (Opinion Formation in the 1957 Referendum), S. O. U., Stockholm 1959.

29 Berelson, Lazarsfeld, McPhee, Voting, p. 207 ff.

30 The sample and the survey design of the 1964 study are presented in a report prepared by this author for the official election statistics see: Sveriges Officiella Statistik. Allmänna val. Riksdagsmannavalen 1961—1964, 11, Stockholm, 1965.

31 The interview question was worded as follows: "Do you remember which of the parties it was that wanted to introduce a compulsory superannuation system, basedon legislation, for all employees?" Cf. J. Westerståhl and B. Särlvik, Svensk valrörelse 1956. Arbetsrapport I (Swedish Election Campaign. Research Report I), Statsvetenskapliga institutionen, Göteborgs universitet, 1957 (mimeo.), p. 20.

32 See: Westerståhl and Särlvik, öp. cit. p. 36 ff.

33 For a detailed account of the interview questions used to ascertain the respondents' appreciation of their pension conditions, see: Särlvik, Opinionsbildningen vid folkomröstningen 1957, p. 28—32 and 61. Actually the relationship between pension situation and referendum voting was a little more complicated than the table may suggest. Among enterprises with Conservative or People's Party preferences, there was a tendency to switch to the Center Party standpoint rather than to the Social Democratic if they were dissatisfied with their pension conditions. In a sense, this trend is quite consistent with our interpretation of the role of self-interest in the pension question. Small businessmen with low income could have reason to fear that a pension system basedon collective agreements would solve their employees' pension problem without affording any improvement to themselves.

34 The socio-economic classification appearing in Table 15 is based on the same set of sub-categories as Table 1. However, we have transferred two "marginal" middle class groups ("formen' and "shop assistants"), to the working class in Table 15. These categories are predominantly Social Democratic and they belong to trade unions which supported the Social Democratic pension scheme. Columns for occupational groups include also respondents who did not vote for the parties accounted for separately in the table as well as non-voters. The "Balance of Opinions" entries have been calculated by subtracting the percent porportion of respondents opposed to the Social Democratic proposal from the proportion supporting it; ambivalent responses are included in the percent distributions. In computing the balance values for the referendum, non-voting has been regarded as an ambivalent response.

---

35 The importance of information about party standpoints can be shown also with data pertaining to the referendum campaign itself. Among those who voted for a referendum proposal supported by their own party, 92 % also were informed about the position taken by that party. The corresponding proportion for those who voted for another proposal was 73 %. (The information question was asked in an interview before the referendum.)

36 Key, Public Opinion and American Democracy, p. 263 ff.

37 Press release from SIFO (The Swedish Institute for Opinion Research), June 1958. Respondents were given short descriptions of the parties' pension plans in the 1958 election campaign. They were then asked which plan they liked best. Less than 1 per cent of the Baurgeois voters preferred the Social Democratic plan. About 5 per cent of the Social Democratic voters preferred one of the Bourgeois plans (Data used with the kind permission of Director Sten Hultgren.)

38 For an overview of partisan change from 1958 to 1960, see the transition tables in: Riksdagsmannavalen åren 1959-1960, 11, pp. 58—59.

39 Interview question: "Are there any parties that want to abolish the supplementary pension system now when a decision has been made?" Responses referring to one or both of both the Conservative and Center parties have been considered as correct in the table where respondents have been classified as "informed" or "uninformed". In the sub-sample interview before the election, 53 % of the men and 30 % of the women gave correct answers; the corresponding proportions in the post-election sample was 66 % and 42 %. The opinion distributions in the two sub-samples differ somewhat. In the pre-election interviews 58 % wanted to maintain the pension system; in the post-election sub-sample that proportion had increased to 66 %. The switch was most marked among Conservative voters; (from 36 % to 45 %).

40 We have to assume that the scale of attitudes towards welfare state politics measures a resonably stable political outlook.

41 With some modifications the interview questions employed in the right-left scale in the 1960 survey were included also in the 1964 study. The scale version used in Table 17 has been constructed with the aid of a regression analysis where voting behavior was included as the dependent variable. The scale categories are basedon the predicted values obtained through the resulting regression equations. The technical procedure is not decribed in this article, since the author intends to investigate the right-left dimension in the party system more exhaustively in another study.