¹ Lipset, Seymour Martin and Rokkan, Stein, ‘Cleavage Structures, Party Systems, and Voter Alignments’, in Lipset, Seymour M. and Rokkan, Stein, eds, Parly Systems and Voter Alignments (New York: The Free Press, 1967), pp. 1–64.Google Scholar

² The importance of the religious cleavage has declined primarily as a result of secularization, which has reduced the potential support for religious parties. However, no consistent pattern has emerged to suggest that the correlation between religious activity (for example, church attendance) and party preference has weakened in a comparative perspective. For somewhat contradictory findings, see Inglehart, , ‘Changing Paradigms in Comparative Behavior’, paper prepared for the annual meeting of the American Political Science Association, Denver, 1982Google Scholar, and Books, John W., ‘Class and Religious Voting in Three European Nations: Political Change in 1960s’, Social Science Journal, 17 (1980), 69–87.Google Scholar

³ See Inglehart, Ronald, The Silent Revolution: Changing Values and Political Styles among Western Publics (Princeton, NJ: Princeton University Press, 1977)Google Scholar; Inglehart, Ronald: ‘The Changing Structure of Political Cleavages in Western Society’, in Dalton, Russell J., Flanagan, Scott C. and Beck, Paul Allen, eds, Electoral Change in Advanced Industrial Democracies (Princeton, NJ: Princeton University Press, 1984), pp. 25–69Google Scholar; Dalton, Russell J. and Flanagan, Scott, ‘The Changing Content of Ideological Beliefs in Western Europe, The United States and Japan’, paper presented at the annual meeting of the American Political Science Association, Denver, 1982.Google Scholar

⁴ See Inglehart, , The Silent Revolution, Chaps. 3–5Google Scholar; Inglehart, , ‘Post-Materialism in an Environment of Insecurity’, American Political Science Review, 75 (1981), 880–900CrossRefGoogle Scholar for the empirical evidence from a comparative West European perspective. The Norwegian findings are documented in Knutsen, Oddbjørn, Poliliske verdier, konfliktlinjer og ideologi (Oslo: University of Oslo, 1985), Chaps. 4 and 8Google Scholar; Lafferty, William M. and Knutsen, Oddbjørn, ‘Leftist and Rightist Ideology in a Social Democratic State: An Analysis of Norway in the Midst of the Conservative Resurgence’, British Journal of Political Science, 14 (1984), 345–67.CrossRefGoogle Scholar

⁵ The impact of the structural model is in the causal model in Figure 1, indicated not only by the direct effects (path 1). but also by the indirect effect via the intervening value model (path 2/path 3).

⁶ See Inglehart, , The Silent Revolution, Chap. 7Google Scholar; Inglehart, , ‘The Changing Structure of Political Cleavages in Western Society’.Google Scholar

⁷ See especially Inglehart, , The Silent Revolution, pp. 182–4Google Scholar and ‘The Changing Structure of Political Cleavages in Western Society’. In the latter article the cleavage transformation is characterized as a change ‘from class-based on value-based political polarization’ (p. 26). For a theoretical discussion of political values and ideologies as political cleavages, see Knutsen, , Politiske verdier, konfliktlinjer og ideologi, Chap. 1Google Scholar and Knutsen, Oddbjørn, ‘Political Cleavages and Political Realignment in Norway: The New Politics Thesis Re-examined’, Scandinavian Political Studies, 9 (1986), 235–63CrossRefGoogle Scholar. Very briefly, we here consider ideologies as a set of constrained political values, so that sets of opposed political values that are empirically related constitute sets of opposed ideological orientations and consequently ideological or value-based cleavages (pp. 238–9).

⁸ This means that the correlations between political value dimensions and political attitudes and values are not spurious; i.e. according to Figure 1 they (path 2) are not dramatically reduced after controlling for the structural cleavage model.

⁹ See especially Inglehart, , ‘The Changing Structure of Political Cleavages in Western Society’, pp. 30–1Google Scholar, in which Inglehart emphasizes ‘Left-Right identity’ and first and foremost ‘attitude towards social change’ as political phenomena in the forefront of the cleavage transformation process.

¹⁰ The other three dimensions were (a) support versus opposition to the welfare state (Welfare Support); (b) the respondents' emphasis on the role of Christian values in political decisions (Christian Values); and (c) conflicting views about subsidies and prices for agriculture and other primary products (Farmers' Subsidies). These dimensions do have a significant impact on party preference and some of the attitudinal variables emphasized by new politics theory in Norway, but it is minor when compared to the two dimensions focused upon here. See Knutsen, , Politiske verdier, konfliktlinjer og Ideologi, Chaps. 12 and 13Google Scholar, for details. Indicators tapping these value dimensions are anyway absent in the cross-national data set employed here.

¹¹ This criticism is further outlined with detailed references in Knutsen, , Politiske verdier, konjliktlinjer og ideologiGoogle Scholar, Part 5 and Knutsen, , ‘Political Cleavages and Political Realignment in Norway’.Google Scholar

¹² For details about the Norwegian findings in this particular area, see Knutsen, , Politiske verdier, konjliktlinjer og ideologiGoogle Scholar, Chap. 4 and Lafferty, William M. and Knutsen, Oddbjørn, ‘Postmaterialism in a Social Democratic State: An Analysis of the Distinctiveness and Congruity of the Inglehart Value Syndrome in Norway’, Comparative Political Studies, 17 (1985), 415–16.CrossRefGoogle Scholar

¹³ The main features of the statistical method are obtained in Klecka, William R., Discriminant Analysis (Beverly Hills, Calif.: Sage Publications, 1980).CrossRefGoogle Scholar

¹⁴ Correspondingly, if the different cleavage variables separate the party alternatives in about the same way, they only define one empirical cleavage dimension. These alternative empirical results (one or several discriminant functions) indicate respectively a cross-cutting versus overlapping cleavage structure.

¹⁵ One advantage of defining the cleavage dimensions by discriminant analysis is that these dimensions are obtained by the cleavage variables themselves, not by the respondents' rankings of different parties. The space is therefore cleavage defined, not cleavage derived as in the alternative approaches (see Budge, Ian and Farlie, Dennis J., ‘The Potentiality of Dimensional Analysis for Explaining Voting and Party Competition’, European Journal of Political Research, 6 (1978), 203–31, where these approaches are discussed in detail.CrossRefGoogle Scholar

¹⁶ The religious cleavage was operationalized by frequency of church attendance in the other West European countries, while the centre-periphery indicator is based on respondents' answers to questions on whether they lived in a ‘rural area or village’, ‘small or middle-sized town’ or ‘big town’. The measure in the Norwegian survey is a more objective indicator, which, however, taps the same theoretical concept.

¹⁷ The detailed operationalizations of two dimensions are explained in Knutsen, , ‘Political Cleavages and Political Realignment in Norway’Google Scholar; Lafferty, and Knutsen, , ‘Postmaterialism in a Social Democratic State’.Google Scholar

¹⁸ See Klecka, , Discriminant Analysis, pp. 36–7Google Scholar. The eta-coefficient is identical to Pearson's r between party preference and the cleavage variable if the different categories on the party preference variable are given their mean scores on the cleavage variables.

¹⁹ Nine per cent belong to a religious organization, while 91 per cent have no organizational affiliation.

²⁰ The religious division is first of all related to support for the Christian People's party versus all others, including the bourgeois parties (not shown).

²¹ See Knutsen, , Politiske Verdier, Konfliktlinjer og Ideologi, Chaps. 12–13Google Scholar; Knutsen, , ‘Political Cleavages and Political Realignment in Norway’.Google Scholar

²² Inglehart, , The Silent Revolution, p. 258Google Scholar. Inglehart identifies at least four factors that determine the impact of political values on voting. In addition to the level of economic development, he proposes that the extent of identification with the existing political parties, the position taken by the party leaders and the number of parties in the system will affect the role of values in voting.

²³ The correlation drops to 0.41 when Greece and Ireland are omitted (see below).

²⁴ West Germany and Great Britain in particular have inconsistent rankings along the two measures, see Appendix II.

²⁵ This theory can be derived from new politics theory according to formulations that discuss the cleavage transformation process without directly relating the value or issue cleavages to the MPM dimension. See Inglehart, , ‘The Changing Structure of Political Cleavages in Western Societies’, pp. 26–33Google Scholar and Inglehart, , ‘Changing Paradigms in Comparative Political Behavior’.Google Scholar

²⁶ These perspectives are further outlined in Knutsen, , Politiske verdier, konfliktlinjer og ideologi, Chaps. 12–13Google Scholar and Knutsen, , ‘Political Cleavages and Political Realignment in Norway’.Google Scholar

²⁷ The statistical coefficient employed here is Wilks's lambda, which is a multivariate measure of group differences over several discriminating variables. Wilks's lambda is an ‘inverse’ measure which indicates the residual discrimination in the system: values of λ which are near to zero denote high discriminatory power, while values near 1.0 indicate less discrimination for the variables involved. To avoid confusion we have transformed Wilks's lambda into more common properties by reversing its inverse character by the simple formula (1.00 – λ). For example, if λ = 0.30, our transformed measure is 0.70. For a discussion of Wilks's lambda in discriminant analysis, see Klecka, , Discriminant Analysis, pp. 38–40.Google Scholar

²⁸ It is the religious cleavage which clearly contributes most to the impact of the pre-industrial cleavage model in almost every country. Only in Denmark and Ireland does place of residence (along the urban-rural dimension) make an almost equal contribution to that of religion. Elsewhere the effect of place of residence is relatively small.

²⁹ For Norway and the Netherlands this only evidently applies for the controlled model.

³⁰ See, for example, Whyte, J. H., ‘Ireland: Politics without Social Bases’Google Scholar, Chap. 12 in Rose, Richard, ed., Electoral Behavior: A Comparative Handbook (New York: The Free Press, 1974)Google Scholar, and Carty, R. K.: Party and Parish Pump: Electoral Politics in Ireland (Ontario: Wilfrid Laurier University Press, 1981).Google Scholar

³¹ In the further analysis these countries turn out to be ‘outliers’ in different directions, so there are statistical reasons for omitting them given that we have few cases in the country-unit analysis.

³² The correlations are 0.30 for GNP per capita and 0.17 for size of service sector.

³³ The corrections for the controlled value model (see Table 5, column B) are – 0.57 for GNP per capita and – 0.29 for size of tertiary industries.

³⁴ Inglehart, , ‘Changing Structure of Political Cleavages’, p. 32.Google Scholar

³⁵ Both hypotheses are expressed in Inglehart, , The Silent Revolution, Chaps. 7–9Google Scholar and in Inglehart, , ‘The Changing Structure of Political Cleavages in Western Society’Google Scholar, where Inglehart postulates that ‘economic issues are less urgent at a high level of economic development than at a low one’ (p. 33).Google Scholar

³⁶ In the three remaining countries, Canada, Ireland and the United States, the salience of the socio-economic dimension is characterized as ‘medium’. See Lijphart, Arend, Democracies (New Haven, Conn.: Yale University Press, 1984), pp. 129–31.CrossRefGoogle Scholar

³⁷ Lijphart, , Democracies, pp. 139–41 and Table 8.1.Google Scholar

³⁸ Budge, Ian and Farlie, Dennis J., Explaining and Predicting Elections: Issue Effects and Party Strategies in Twenty-Three Democracies (London: George Allen & Unwin, 1983), Chap. 2 and especially Table 2.3, p. 36Google Scholar. In calculating these percentages we have collapsed the categories ‘socioeconomic redistribution’, ‘governmental control and planning’, ‘governmental regulation in favour of individual’ and ‘initiative and freedom’ formulated by the authors. Two categories in Budge and Farlie's issue typology (‘Candidate Reactions’ and ‘Government Record and Prospects’) are omitted in the calculations since by their nature they cannot be called political cleavages.

³⁹ These issue domains constitute about 50 per cent of all issues in the period 1945–54: the same holds true for the period 1975–81. In the period 1960–75 their dominance was not so pronounced (35–40 per cent); see Budge, and Farlie, , Explaining and Predicting Elections, Table 2.4, p. 39.Google Scholar

⁴⁰ Budge and Farlie only include aspects of the authoritarian-libertarian dimension in their categories (issues related to ‘Civil Order’ and ‘Defence’) while the growth-environmental protection dimension is neglected. The included aspects constitute about 10–20 per cent of the total issues, and there is no clear developmental trend from the first postwar period to the 1970s and 1980s. See Budge, and Farlie, , Explaining and Predicting Elections, Tables 2.3 and 2.4.Google Scholar

⁴¹ See especially Lafferty, William, ‘Political Participation in the Social Democratic State: A Normative Empirical Framework for the Analysis of Decision-Making Involvement in Norway’, Scandinavian Political Studies, 6 (1983), 281–308.CrossRefGoogle Scholar

⁴² Of the single items included in the two indices, ‘public ownership of private industry’ and ‘reduction of inequality of income’ are on average most strongly correlated with party preference among Left-Right Materialist items, respectively 0.30 and 0.28. Among the MPM items ‘stronger defence’ (0.29) and ‘nuclear energy’ (0.27) are highest correlated, while ‘environmental protection’ is clearly more weakly correlated (0.18). Among all the issues ‘public ownership’ has the largest impact on party preference in the European Community countries, and this especially applies to three: France, Denmark and Greece. The ‘nuclear energy’ item correlates most strongly with party preference in the Netherlands and Ireland, while ‘stronger defence’, ‘terrorism’ and ‘regional autonomy’ are the most important items in West Germany, Italy and Great Britain, respectively.

⁴³ The exclusion of Ireland and Greece does not change these figures substantially: they are respectively 0.41 and 0.38, based on only eight countries.

⁴⁴ While the correlations between the impact of the MPM dimension (Table 3A) and the two measures for post-industrialism are negative, but less than 0.10, the corresponding correlations between the impact of Left–Right Materialism and post-industrialism are 0.52 (size of the tertiary sector) and 0.57 (GNP per capita).

⁴⁵ See especially Inglehart, , ‘Value Priorities and Socioeconomic Change’, in Barnes, Samuel H., Kaase, Max et al. , Political Action (Beverly Hills, Calif.: Sage Publications, 1979), pp. 329–33.Google Scholar

⁴⁶ For the MPM dimension this contradicts the Inglehart thesis that the ‘level of value change’ in a given country should be important as regards impact on voting; see Inglehart, , The Silent Revolution, p. 258.Google Scholar

⁴⁷ See Inglehart, , The Silent Revolution, p. 258.Google Scholar

⁴⁸ Knutsen, , ‘Political Cleavages and Political Realignment in Norway’, p. 256.Google Scholar

⁴⁹ Knutsen, , ‘Political Cleavages and Political Realignment in Norway’, p. 256.Google Scholar

⁵⁰ ‘Left-Right Materialism defines one of the most important cleavage dimensions’ means that Left-Right Materialism has the highest standardized discriminant function coefficient along that dimension.

⁵¹ That the status variables have a larger impact than Left-Right Materialism, while the latter cleavage is the most important variable in defining a dimension which we have interpreted as a Left-Right dimension in many countries, may be considered contradictory. The explanation is related to the fact that the so-called social status variables appear to have an impact along several cleavage dimensions, and apart from the Left-Right dimension these cannot be considered to be tapping the classical industrial cleavage pattern. For example, in almost all countries education and occupation go together with the MPM values in defining the new politics dimension, and in some countries (the Netherlands, West Germany, Italy, Ireland and Norway) the status variables are related to the centre-periphery dimension. These patterns are revealed from the dimensional solutions which we have not displayed in the tables here.

⁵² The figures (r) are 0.46 (size of service sector) and 0.26 (GNP per capita).

⁵³ The average impact of the ideological model is about 80 per cent of the total impact of the structural (pre-industrial and industrial) cleavages in the uncontrolled model, and drops to about 40 per cent in the controlled model (see Table 2).