Background

The Norwegian political scene changed rather dramatically in the 70's and early 80's. There was a trend towards increased electoral mobility and a pronounced swing to the right which benefited the Conservative Party. Henry Valen's recent Valg og Politikk (NKS-Forlaget 1981) may be seen as an explicit attempt to account for the changing Norwegian voter, but there is considerably more to the book.

It draws heavily on Stein Rokkan's famous model of the Norwegian party
system, where the parties are seen to compete along a number of political
dimensions or cleavages (Rokkan 1967):

— a territorial cleavage (center vs periphery)
— a linguistic cleavage (riksmål vs nynorsk)
— a moral cleavage (teetotallers vs others)

— a religious cleavage (non-conformists vs conformists)

— a rural/urban cleavage (the primary vs the secondary and tertiary sectors)
— an economic cleavage (labor vs capital)

Socioeconomic, attitudinal and individual level behavioral data all provide
clues about the underlying conflict structure in a society. If a party is consistently

favored by members of the non-conformist churches, who attach great importance to moral issues like the struggle against free abortions and who rarely express discontent with their first choice by defection to one of the socialist parties, we have identified a religious cleavage which correlates strongly with the übiquitous economic left/right dimension (Berglund & Lindstrom 1978). If the advocates of some cause, say ecology, are attracted to parties or groups of parties at opposite extremes on the left/right axis, we have identified a salient secondary dimension which is totally independent of left/right. And, if membership in, say, temperance organizations fails to predict partisan affiliation, though affecting the individual priorities, we are up against an independent secondary dimension of such little import that it may be left out of the analysis.

Survey data allow the researcher to move freely among the three different sets of indicators, and Valen, who draws on an impressive series of Norwegian election studies from the mid fifties and onward, takes full advantage thereof. The book is divided into three major sections on social structure, on attitudes or ideology and on floating; it thereby also complies with the traditional concerns of the well-established Anglo-Saxon electoral research tradition (Borre 1983).

In Quest of a Theory?

The book is rich, in some cases overwhelmingly rich, on observations, hypotheses and generalizations, but has considerably less to offer by way of theory. The swing in favor of the Conservative Party is cautiously seen as a direct function of two, possibly three, related phenomena (Valen 1981, 369-370):

— the conservative mood (as brought out by the opinion polis)

— the disappearance of the Liberal Party which had served as a buffer
between left and right

— the shrinking perceived distances between the Labor and Conservative
parties.

Technically, this is a theory in the sense that we have two sets of generalizations cast in a causal framework (Meehan 1965). There is a certain circularity to it, however. There is usually a mood to match electoral gains and losses, but the direction of the causality must not be taken for granted (Key 1966). The remaining factors are equally ambiguous. In the zero-sum world of electoral politics, gains are always made at somebody else's expense. The fewer parties there are, the fewer options there are; and the fewer the options, the shorter are the party distances as given by rank order scores (Valen 1981, 213-240).

The major theoretical drawback, however, is Valen's failure to link the

three separate sections in a meaningful way. The discrepancy between the cleavage structure as given by the socioeconomic data and the one intimated by the attitudinal data deserves particularly close attention. The voter s' perceptions of the parties lend themselves to no more than two- or threedimensional solutions (Valen 1981, 213-240). The voter attitudes to current political issues also fail to comply with the full complexity of the traditional model (Valen 1981, 243-250), and — what is more — they cannot be accounted for with the aid of the standard battery of socioeconomic indicators like sex, occupation, and education (Valen 1981, 259-280). This makes Valen conclude that there is no linkage between social structure on the one hånd and ideology on the other. But is it reaJly that simple?

Taken at face value, Valen's data strongly support the notion of left/right as the primary structuring criterion in Norwegian politics. Class, particularly class identification, correlates strongly with party identification. The voters apparently take cues from the parties' relative position on the left/right axis which consistently comes out as the primary dimension in the multi-dimensional analyses; and, if nothing else, left/right clearly helps structure the voter attitudes better than anything else. Now, if class does not do the trick, why not try an alternative linkage variable — party identification. It helps determine how the voters see themselves, how they approach the parties and how they judge the issues at stake, and should make for considerable political stability even in times of rapid social change.

There are intimations of such a perspective throughout the book, but Valen never follows up on what might be the essence of the new Scandinavian voter, whose class identification is a function of his party identification rather than the other way around (Petersson 1982).

Valen's failure to evaluate Stein Rokkan's model of Norwegian politics may be thought of as a drawback of similar magnitude. In fact the entire book serves as a healthy reminder that models may serve as blinders. After all, who but a Norwegian scholar, enamored with the complexity of Norwegian politics, would devote pages on end to describing and explaining a dimension that accounts for no more than three (sic!) per cent of the total variance, as does the celebrated religious-moral cleavage (Valen 1981, 248-249)? And who but a Norwegian scholar, firmly anchored in a traditional model with its roots in early ecological data research, would go hunting for a territorial cleavage, oblivious of Przeworski's and Teune's seminal Logic of Social Inquiry (1970) which basically states the obvious? Region will register in a multi-variate framework only to the extent that we have failed to specify the model properly. In other words, it only serves as a proxy for contextual variables like urbanization, unemployment, party organization etc., which should have been included in the model from the beginning.

In Search of a Method?

Students of electoral behavior have often been accused of using their data as a testing ground for unnecessarily sophisticated statistical techniques. If this is true, Valen's book represents a welcome change. It relies heavily on frequency distributions, contingency tables and graphs which are easily digested by readers without formal statistical training.

The multi-variate format is used sparingly, but often enough to raise some methodological questions. The MCA analyses (Valen 1981, 208, 219) raise the spectre of multi-collinearity. If the social background variables are interrelated, the regression coefficients (read: BETA weights) will not serve as reliable indicators of the respective variables' relative importance. The factor analysis (Valen 1981, 248-249) raises questions about the underlying model. If real-life cleavages are interdependent, why force an orthogonal solution (read: varimax) with zero-correlations among the factors onto the data? And the multi-dimensional scaling technique (Valen 1981, 213-239) raises the familiar question of what is a statistical artifact and what is not. With seven parties on the stage (1965 and 1969), one gets a two-dimensional solution as opposed to the three-dimensional solution obtained with nine parties competing for the voters (1973 and 1977). The question is how much of the complexity remains if the two newcomers (read: Anders Lange's Party and the New People's Party) are excluded from the analyses of the seventies.

On the whole, however, the book is methodologically sound. The data
are solid and lend themselves to all sorts of statistical manipulations.

Each model calls for its own set of statistical techniques. Descriptive statistics, however, are helpful in the search for models, which goes a long way towards explaining Valen's reliance on the bivariate format. It is a byproduct of the theoretical flux, and not just a benevolent concession to the statistically innocent. The methods are there, the theory is not.

Summary and Conclusions

Henry Valen's Valg og Politikk is indispensible basic reading for students of Norwegian politics. It is a handbook of sorts. Few, if any, of the potentially relevant break-downs are left out of the book, and what is not included in the book is likely to be found in the appendix, which may be ordered from the Institute of Social Research in Oslo.

The book has considerably less to offer by way of theory. It is poorly integrated; the swing to the right is not properly accounted for, and it is structured by perceptions of Norwegian politics which do not appear to be borne out by the data. Norwegian politics remain multi-dimensional, but to a much lesser extent than implied by the traditional model.

REFERENCES

Berglund, S. & Lindström, U., 1978. The Scandinavian Party System(s), Studentlitteratur
Lund.

Borre, 0., 1983. Review of Henry Valen 'Valg og Politikk' in Statsvetenskaplig Tidskrift
1982:4.

Key, V.O. Jr., 1966. The responsible Electorate; Rationality in Presidential Voting 1936-1960,
The Belknap Press of Harvard University Press, Cambridge, Mass.

Meehan, E.J., 1965. The Theory and Method of Political Analysis. Dorsey Press, Homewood,
Illinois.

Petersson, 0., 1982. 'Klassidentifikation' in Kent Asp et al. Väljare, Partier, Massmedia.
Liber Tryck, Stockholm.

Przeworski, A. & Teune, H. 1970. The Logic of Social Inquiry. John Wiley & Sons Inc.,
New York.

Rokkan, S., 1967. 'Geography, Religion and Social Class. Crosscutting Cleavages in Norwegian
Politics' in S.M. Lipset & S. Rokkan (eds.) Party Systems and Voter Alignments. New
York, Free Press.