Logo DBNL
State formation, parties and democracy. Studies in comparative European politics (2011)

State formation, parties and democracy. Studies in comparative European politics

(2011)–Hans Daalder–rechtenstatus Auteursrechtelijk beschermd

Vorige Volgende

chapter eight | in search of the center of european party systems

Introduction*

This article has its origin in a curious anomaly. Writings about European party systems abound with references to the Left. They also give considerable attention to a less easily defined Right. Yet, there is practically no systematic treatment of the center or of center parties. Admittedly, the term center is widely, if loosely, used in both journalistic comment and in academic descriptions of the functioning of party systems in individual countries. Nevertheless, European party systems are rarely viewed from a center perspective. On the contrary, the idea of a center tends to find little favor with scholars. More than a quarter of a century ago, Duverger argued in a famous paragraph of his book, Political Parties:

‘The center does not exist in politics; there may well be a Center party but there is no center tendency, no center doctrine. The term “center” is applied to the geometrical spot at which the moderates of opposed tendencies meet: moderates of the Right and moderates of the Left. Every Center is divided against itself and remains separated into two halves, Left-Center and Right-Center. For the Center is nothing more than the artificial grouping of the right wing of the Left and the left wing of the Right. The fate of the Center is to be torn asunder, buffeted and annihilated: torn asunder when one of its halves votes Right and the other Left, buffeted when it votes as a group first Right then Left, annihilated when it abstains from voting. The dream of the Center is to achieve a synthesis of contradictory aspirations; but synthesis is a power only of the mind. Action involves choice and politics involve action. [...] There are no true Centers, only superimposed dualisms.’1.

Giovanni Sartori has written of center parties in hardly more positive terms. He greatly values what he terms centripetal drives in party systems - i.e., a competition of parties for votes at the center. But for precisely that reason he takes a negative stance toward the ‘physical occupation of the center’ by one or more center parties, which he singles out as one of the distinctive features of systems of ‘polarized pluralism’. For this implies that ‘the central area of the political system is out of competition [...] In other terms, the very existence of a center party (or par-

ties) discourages centrality, i.e., the centripetal drives of the political system. And the centripetal drives are precisely the moderating drives. This is why this type is center-fleeing, or centrifugal, and thereby conducive to immoderate or extremist politics.’2.

Such critical attitudes toward center parties stand in marked contrast to the rhetoric of many participants and observers of European party systems. Clearly, the idea of center parties has been regarded as an attractive one by politicians, who in recent decades have regarded the very word center as a positive label for their parties. The term has been adopted by former agrarian parties in Scandinavian party systems, by new parties in various re-emerging democracies of Southern Europe, and among would-be reformers of party systems in a Giscardist France or a Jenkins-type Britain. N'en déplaise Duverger, terms like centripetal competition, centrism and center generally carry positive overtones. But such sentiments are rarely given specific meaning or explanation.

Academic writings in this area are neither extensive nor well-focused. Scanning the literature (e.g., on party systems, political cleavages, or coalition theories), one finds few explicit references to the concept of a center and its use in the study of party systems. In addition to Duverger and Sartori, only Downs (1957) seems to have made the idea of a center a key variable in his writings.3. Implicitly, some attention is given to the notion by authors who have studied the formation of European cabinet coalitions, based on an assumed location of parties along one or more dominant dimensions.4. The writings of formal theorists shed some indirect light on the subject, notably because they specify the conditions that must be met if one is logically to speak of a center in a multi-actor game. As a whole, however, the literature offers few cumulative lessons, possibly owing to at least two factors.

First, the term center forms part of at least three rather different kinds of metaphors. Thus, some authors reason in locational terms, holding that a particular individual or party occupies a center position in a given political space. Others reason in metaphors taken from mechanics: e.g., certain parties hold the balance in a political system, or the interaction of parties shows the operation of centripetal or centrifugal forces. Yet other writers focus in particular on specific kinds of cleavages, which may be inherently dualistic in nature (as Duverger would argue as a matter of course), or may allow more variegated positions.

A second confusion results from failure to specify the unit of analysis. For some, the center is a theoretical point, to which parties may be near or distant. For others, the center is an actual actor, whether a pivotal deputy in an actual parliamentary vote or one or more parties in a given party system. But in treating a party as an actor, new ambiguities may arise, depending upon whether or not a party can in fact be treated as a unitary actor, or whether it should be analyzed in terms of possible internal divisions (e.g., within the parliamentary group or between this group, party activists outside parliament, and voters for a party).

The importance of these distinctions should become clearer in the course of the following exploration, which will move from the spatial concept of center (Sections One to Three), to approaches reasoning in mechanistic terms (Section Four), to analyses in terms of actual social cleavages (Section Five).

The center in voting

The Median Member or the Median Party

It is clear from the passage quoted above that Duverger bases his view of the nonexistence of the center on the need for a voting body (such as a parliament) to say either yes or no to a particular proposal. Hence, he refers to the center as a mere ‘geometrical spot’ which separates the proponents and opponents of a particular decision. As the word suggests, such a point is most easily envisaged as a point on a line on which all members of a voting body are arranged. The center, then, equals the position of the ‘median member’ - a conclusion substantiated in formal theory by Black (1958).5. Certain problems emerge from this proposition, however. How does one know who the median member is? If one is to locate the median member of a voting body, members must be somehow arranged in a logical order, which is tantamount to saying that there must be a common underlying dimension, a point that will be taken up shortly. But how can one substantiate this dimension, except in a concrete vote, and only a posteriori?

If one analyzes decisions of a voting body in terms of parties as actors rather than in terms of individuals, what is the relationship between this median member and the party in which he is included? Can one describe the latter party logically as the center party, or should one instead speak of a party containing the median or central member?

If one wishes to denote the party containing the central or median member as the central party, then difficulties arise when one tries to apply the concept to a two-party system. For in such a system the majority party would become the center party by definition. In the case of alternating majority parties, this would imply alternating center parties. If parties do not alternate, one is in the presence of a predominant party. Although such a party must be of central importance to a system of this kind, does one gain any insight by also calling such a party a center party?

In systems with more than two parties, two conditions may exist. On the one hand, although there are more than two parties in the system, one party may still have the majority. Hence, it occupies the center point by definition. In this case, nothing changes from the previous situation, except that now there are more non-central parties. On the other hand, in cases where no party has a majority, one must have some way of establishing which party contains the median member, which again raises the problem of an underlying dimension, which will be treated in the next two sections.

Finally, one might attempt to shift the argument from the median member to the median party, i.e., the party that finds itself hemmed in by as many parties on one side as on the other. But such a shift presupposes that it is the number of parties which is important rather than their relative sizes. This proposition clearly leads to ridiculous conclusions where a party finds two very small actors on one side and two very much larger ones on the other. The party that would find itself in such a middle would in any case not be the party containing the center, as defined by the median member.

In summary: to make the center in a voting body identical with the median member or median party and to apply these concepts to party systems raises at least two major problems. First, one must find some way in which to rank-order members or parties or both in relation to one another. Second, one must somehow take into account the different sizes of the parties in a particular system.6.

The Problem of Successive Votes

Even though one may solve the problem of locating the median member in a particular vote, no solution is yet given for locating the center in a succession of votes. Although each vote may logically have its median member, there is no reason whatsoever that the same member will be the central member for each vote, unless all votes represent only one common dimension.

This dilemma is typically solved in escapist ways. One approach focuses on one crucial vote only, e.g., on the vote necessary to confirm a new cabinet in office. All other votes are then made subordinate to the needs of the cabinet coalition thus formed. Another approach seeks to arrive at a stable positioning of individual members (and their parties) by shifting from voting to self-location, e.g., by asking members to place themselves on a left-right scale. A third approach seeks to analyze a selected number of actual votes in order to determine whether members (or parties) fail in place around so-called pivotal actors who are thought to be of central importance. In all such cases, successive votes are really subordinated to presumed dominant alignments. If in practice more than one dimension is found, that one specific dimension is thought to be the decisive one. Hence one can maintain the concept of a median member (or center), even though that member is not central in all actual votes.

The Shifting of the Center

The assumption that the median member is the center becomes particularly questionable in the following cases:

1.In a well-known study in the mid-1960s, Giovanni Sartori arranged the Italian parties along one line and noted that different ‘centers of gravity of the parliamentary majorities’ existed according to the issues faced by the Italian parliament, e.g., an economic majority, a political majority ‘committed to the defense of the system, that would unite on basic constitutional issues, and that opposes any shift in international alliances,’ and a religious majority.7. Sartori did not pursue this idea and still speaks in the same article of the center in the singular rather than in the plural. But he then adds: ‘a center is relative to its left and right wings, it will follow in the long run the general trend of the polity’ and ‘the center will move if the balance between its left and its right should shift’.
2.The location of the effective center will change if the votes of certain parties at the extremes are regarded as illegitimate, and conversely if the area of legitimate space expands, situations that have characterized the French Fourth Republic and postwar Italy.
3.One might argue in a similar fashion that the formation of a relatively permanent political coalition logically implies a shift of the pivotal member away from the median of parliament to the median of the coalition. But this argument is in turn subject to criticism - e.g., by those who see the key to power within a specific coalition to be in the hands not of its median member or party, but of those supporters of a coalition who provide the last necessary votes and who conceivably could also shift their votes to an alternative winning coalition, which again suggests that the idea of the median member (assuming such a member can be located) is of little value in the analysis of successive votes and potentially changing coalitions.

The center in traditional left-right assumptions

The Influence of Geometric Thinking

The idea of arranging parliamentary deputies or parties on a left-right dimension has an immediate and intuitive appeal. It is general practice in expositions of single-country studies, and it is widely acceptable to political elites and citizens alike, as is amply proved by the success of surveys containing questions in which respondents must place themselves or parties on linear scales.8.

At a minimum, however, one must be wary that a left-right imagery does not take an uncritical upper hand because of the sheer ease of geometric understanding. The terms left and right had their origin in the seating arrangements of the French Convention. Just as seats gained a symbolic importance, so the very terminology has tended to acquire a bias towards the left which unavoidably affects future spatial locations.

Undoubtedly, the idea of a linear left-right dimension is highly evocative, as it permits not only notions such as the extremes or the center, but also intermediate ideas like center-left or center-right. But the same notion can also be deceptive, if it is assumed that apparently similar distances at any part of the scale are also similar in political content and affect.

Finally, the left-right image is also compatible with that other intuitively attractive geometric representation: the bell-shaped curve of a normal distribution, which is easily superimposed on a left-right scale. Since in a normal distribution the large majority is clustered near the median, the idea that there is a logical force working for moderation in electoral competition may be uncritically adopted as a basis for further analysis. In that case, no attention needs to be paid to problems of real-life situations, where the mean, median, and mode are not coterminous, let alone to the critical importance of there being only one or more modes in a given distribution of political preferences.

Because the idea of a center depends on the possibility of locating parties on a left-right scale, the manner of obtaining such placements is of critical importance. There are at least three ways in which scholars have sought to solve this problem: asking respondents thought to be representative of a party to place themselves, their own party, and other parties on an imposed left-right scale; scaling or factor-analyzing responses to other survey questions so as to trace underlying dimensions; and analyzing actual behavior.9.

Individual Self-Locations and Placement of Parties on an Imposed Left-Right Scale

Subjective placements on imposed left-right scales meet with certain problems such as non-responses (generally found less frequently among elites than in mass responses); faulty responses (the authors of the Eight-Nation study found by follow-up questions that as many as 1 out of 5 voters might have a totally irrelevant notion of the scale on which they were prepared to place themselves;10. a tendency with many respondents to play safe and choose a center location or to eschew the more extreme positions; differing interpretations about the use of the scale; assumptions concerning the interval nature of the scale, necessary to execute arithmetical operations; the problem of scale length (e.g., the finding that the use of 7-, 9-, or 10-point scales makes for material differences).11.

Assuming that these problems are not of major importance, scholars have used the responses of both mass and elite surveys to obtain an ‘average’ placement of a party. Thus, Sani and Sartori (1983) take averaged responses from the mass surveys in the eight-nation study to obtain the locations of parties, which they then use in a large-scale systematic analysis that regards the ideological distance existing in party systems as a key variable.12. They then use these data for a cross-national comparison of the extent of ideological polarization of party systems. Other writers have preferred to rely on elite data as providing more authoritative responses. The problem of finding the correct respondents for the desired location of parties is an important one whenever substantial differences are found, for example between the self-location of a party leader, individual members of the parliamentary party, the leaders outside parliament, and the party voters, not to speak about the assigned locations of one party by leaders or voters of other parties.

The problem becomes particularly crucial if certain systematic biases have an influence on the data. Thus, it has been a common finding that party activists take up more extreme ideological positions than either party leaders or party voters.13. Such a crucial finding should not be hidden behind the analytical need to locate a single placement of a party. Survey responses suggest a consistent tendency in some countries for legislators to place themselves to the left of the location they attribute to both their fellow deputies and their voters.14.

Scaling Other Survey Questions

In a number of countries (e.g., the United States, Sweden, Switzerland, and the Netherlands), studies have been undertaken to compare the responses of legislators and voters to the same issue questions. The findings of the 1972 Dutch study may illustrate some of the problems that are likely to come up if one seeks to use these data to trace the existence or nonexistence of one dominant left-right dimension. It contained the following items relevant to attempts to place parties on some form of underlying dimension (in addition to locations on a self-imposed left-right scale): preference rank orderings, sympathy ratings for particular parties, and placements of self, one's own party, party voters, and the other major parties on seven issues (each issue represented by 7- or 9-point interval scales). The following conclusions were drawn.

Strong indications were indeed found for the existence of a left-right scale. However, the best result was not one straight line, but rather a horseshoe curve, with the extremes on the left and right approaching one another at either end; the major system parties showing substantial differentiations in the lower bend of the figure. The Dutch psychologist Van de Geer explains these findings in terms of a general law of perception which allows for much greater discriminatory perceptions in intermediate zones as compared to extremes. Does this finding tally with Sartori's use of the concept of space elasticity and his particular conclusion?

A short space does not allow, or does not facilitate, the perception of a center: It has, so to speak, no room for it. A short space is defined simply by its ends - left and right. A third point of reference - the central point - becomes meaningful and perceivable only as the space extends, and particularly when the ends of the space are perceived as being two poles apart.15.

However, the placement of parties that results from the scaling of voter responses showed substantial differences from the results obtained from interviews with legislators. This is true both of the general space, and more important, also on the location of parties by voters and deputies on particular issues. Hence, the question arises again: whose placement is decisive, and what issues are the significant ones?

Also, some issues clearly broke out of a left-right dimension. This was particularly true of abortion, defense, and law-and-order issues at the elite level, and of abortion and development aid at the mass level. In other words, although a left-right dimension was consistently found (if one reads along a horseshoe curve), and although the placement of parties along this dimension tended to substantiate stereotyped notions on the assumed location of parties in relation to one another on particular issue scales, the dimension did not fit important issue areas that the parties must deal with. Said differently, the left-right dimension was sufficiently robust to allow the singling out of some parties as having a center location. But on certain issues, neither the scale nor the location held. Clearly, presumed center parties are not center parties on all issues!

Finally, a separate analysis of deputies from the major party groupings showed that not all groupings saw the space obtained from scaling all responses by all deputies in a similar manner. In particular, members of parties that others assigned to a location on the right tended themselves to see the space as one in which their location was much more in the middle. In other words, some parties' centers need not be other parties' centers!

The Analysis of Actual Behavior

At least three kinds of actual behavior lend themselves to a test for left-right (and hence center) placement of parties: participation in government coalitions, parliamentary voting, and official party pronouncement such as campaign statements or party platforms.

Analyses of coalition behavior have increasingly considered the role of ideology. The most successful explanations have been what are called ‘minimal

range’ or ‘minimum connected winning’ theories which hold that coalitions are formed containing adjacent parties in sufficient number as to ensure a majority.16. However, the placement by De Swaan (1973), Dodd (1976) and others of parties in relation to one another is based on the judgment of ‘authoritative observers’ of the politics of a particular country.17. As these judgments may well be based mainly on their knowledge of actual coalition behavior, there is a danger that coalitions are explained on the basis of knowledge of coalition behavior: this clearly smacks of a tautology. Notions like range, adjacent, and connected or closed coalitions meet with the problem of whether one dimension is sufficient to account for actual coalition behavior. In constructing a policy scale defined mainly in socioeconomic terms, De Swaan must take considerable liberties in forcing parties on to such a scale. (One should note in particular his placing of religious parties, and his reinterpretation of the right part of the scale so as to accommodate authoritarian and national-socialist movements, as well as anticollectivist groupings, at one and the same end of a continuum.) Finally, the placement of parties on a scale is generally static over time. Hence, the data throw little light on dynamic movements that may affect the center as much as either ends of the scale.

Data on the way parties line up in parliamentary voting have a considerable advantage over coalition data in any attempt to place parties on a scale. As Pedersen, Damgaard, and Nannestad Olsen (1971) have convincingly shown, analysis of such data allows inspection of the movement of parties over time.18. It also makes it possible to trace the actual behavior of parties in certain issue areas as distinct from others. Parliamentary voting is, of course, heavily influenced by whether or not parties participate in a government coalition. But this problem may to some degree be overcome, particularly in cases where there is some degree of autonomy between the governmental and the parliamentary arenas.

A third indicator may be found in a systematic study of party manifestoes along the lines of the study on British manifestoes by David Robertson (1976).19. An international project in this area has been initiated by Robertson and Ian Budge. One must await publication of the results before one can conclude whether or not this type of study makes it easier to arrange parties on some more durable left-right scale.

Conclusions on the Left-Right Scale

Whenever a robust left-right scale can be established, the assumption that there is a center at some intermediate point or area offers at least some promise. However, the foregoing paragraphs have raised certain difficulties: the actual placement of parties on such a scale may differ according to the category of respondents interviewed, the kind of data and analysis techniques used, the actual issues at stake, or the time studied. Although left-right perceptions are strongly present, there is not necessarily an immediate relation between perception and behavior. More important, for certain actors, and in certain issue areas, the left-right dimension does not hold. Often political realities prove stubbornly resistant to attempts to reduce a multidimensional universe to a unidimensional ordering.

Does a center exist in a multidimensional space?

Geometric Representations

Clearly, the idea of a center in a multidimensional world is intuitively a great deal more difficult to grasp than in a linear representation. Even in a simple two-dimensional representation of a triangle (see Figure 1a), with three parties B, C, and D at the points, one could associate the idea of a center with four positions: the midpoints between BC, CD, and BD, and the point A equidistant from B, C, and D. Even easier to grasp is the idea of a circle (Figure 1b) - or in a three- dimensional representation, a sphere - on which all parties are placed. In such a representation there is indeed only one center, but this emerges at the expense of any meaningful insight into dimensions. Another possible location is the point of intersection between two dimensions: at this point one actor (A) would be placed centrally on two dimensions, unlike other actors who might be central on one dimension, but not on the other (B and C, respectively) - see Figure 1c. Such spatial locations are less suggestive, however, than locations on a simple line.20.



illustratie



illustratie



illustratie
Figure 8.1: Three possible geometric representations of parties in a multidimensional space


Duverger: The Center as a Result of Superimposed Dualisms

As we noted above, Maurice Duverger derives the existence of a center or centers from ‘superimposed dualisms.’ Figure 2 reproduces his well- known circle of overlapping cleavages. The figure indeed suggests one possible explanation of multipartism. But such an explanation does not say anything about a center, unless one projects all parties onto one dimension. Thus, the Socialists, the Progressive Christians, the Radicals, and the MRP become center parties if one takes the Communists and the RPF as the end points of the decisive dimension. This is saying little else than that a center emerges if one reduces a multidimensional space to a unidimensional ordering. Such apparently is the strength of the left-right imagery, even for an author who sets out deliberately to analyze the simultaneous existence of more cleavage lines!



illustratie
Figure 8.2: Overlapping cleavages in France


One should add that on closer inspection Duverger does not offer just one theory about the (‘non-existing’) center, but many. Thus one may distinguish at least the following three:

1.the assumption that all real political choices are inherently dualistic. Hence, a center can only be defined as a pale reflection of the real forces at the poles and is inevitably divided between the opposing attraction exercised from the two poles, of one and the same dimension;
2.the proposition that a center is the result of projecting parties on one overriding axis where others form the defining poles (e.g., Communists vis-à-vis the Right);
3.the view - still to be treated - that center parties are really the residue of past history, as once-progressive forces are being overtaken by new forces on the Left (Duverger's sinistrisme).

Sartori: A Unidimensional Space of Competition in a Multidimensional Political Universe?

No writer has grappled with the dimensionality problem more gingerly and persistently than Giovanni Sartori.21. Sartori has been aware from the outset of the multidimensionality of issues and identifications in contemporary politics, and he has been very conscious of the indeterminacy of meaning, and the strong biases, of left-right imagery. He has strongly criticized authors who rely on a socioeconomic interpretation of a left-right scale in the face of contrary evidence and has suggested instead that priority should be given to a constitutional-political interpretation. He has warned against the facile assumption that similar intervals in a spatial representation also reflect the same actual political distances. A major point in all his writings has been that there are indeed major ideological divides which separate system parties from non-system parties. He therefore uses terms like different spacings and disjointed spaces. Yet, again and again, he has attempted to reduce the multidimensionality of modern politics to a mainly unidimensional interpretation, at least as far as the dynamics of interparty competition are concerned. To do so he follows a number of roads:

1.His primary interest in Volume I of Parties and Party Systems is to push for an interpretation of the interaction of parties as independent rather than dependent variables;22. hence his strong emphasis on the role of the number of parties, and the mechanical effects of their interaction. From it he deduces a number of properties vital for his interpretation, including his notion of space elasticity (the larger the number of parties, the wider the space), the appearance of one or more parties occupying a center position, the displacement of centripetal competition in a system with bipolar mechanics by centrifugal competition in a multipolar one, and eventually as a result of a policy of out-bidding and out-flanking by irresponsible extremist parties, the likelihood of the growth of antisystem parties in a polarized system bound for destruction.
2.Following Converse, Sartori suggests that in given systems a variety of dimensions may coexist, but that these may be of unequal ‘length.’ In that case, a process of foreshortening may effectively drive a party on the shorter dimension onto the longer - more salient and dominant - dimension.23.
3.As we saw, Sani and Sartori (1983) present extensive evidence that on a self-imposed left-right scale voters place themselves much wider apart in some European countries than in others.24. The resulting degree of polarization (for which they provide a variety of measures) shows a gratifying correspondence with Sartori's distinction between ‘polarized’ and ‘moderate’ multiparty systems. In the same article, these two authors also relate
self-locations on a left-right scale to responses to a number of important issue and attitude questions and often find a substantial relationship.
4.Sartori also continues to grapple with the problems, most clearly posed by the so-called segmented societies, in which there clearly is a world of multidimensional identifications. He submits that these identifications cannot by definition enter strongly into the world of electoral competition because that would effectively suppose the presence of at least two parties competing along one and the same ideological dimension for such voters as are available to either contestant. Instead, Sartori submits, the major dimension for electoral competition in most European countries today is left-right competition. Parties not on that dimension can only engage in defensive competition, not in expansive strategies. They also do not really affect the decisive interplay of the party system, unless they engage in the struggle along the general and dominant dimension. In this way, Sartori explains why in certain party systems (e.g., the consociational democracies characterized by segmented societies), there are five parties or more, which yet do not show a tendency toward increasing polarization. There may be numerous parties with various identifications, but for the effective dimension of competition, such systems represent cases of party crowding rather than of strong polarization.

I have summarized Sartori's arguments rather extensively because his distinction between ‘domains and identification’ (which are often multidimensional) and a ‘space of electoral competition’ (which is much more likely to be unidimensional) could solve the problem of whether a center can exist in a multidimensional party landscape. If the space of competition is effectively unidimensional, a center emerges as a result of the interplay of numerical factors in the way Sartori describes.

I offer the following comments:

At first reading, Sartori's reduction of a multidimensional world of identifications to a unidimensional space of competition shows some resemblance to Duverger's reasoning. In Duverger's case center parties emerge because they are forced to locate themselves between the end points of a continuum which is apparently thought to be the dominant one. Duverger does not make clear, however, why parties not on the dominant dimension should end up near ‘the’ center of a party system, or even at some point in between the extreme poles. Is this because parties formed on the basis of another dimension are likely to harbor rather heterogeneous elements, if looked at from the standpoint of the overriding dimension, which in the analogy of the ‘crosscutting’ proposition forces such parties to choose a somewhat indeterminate and hence ‘centrist’ position? Whenever different dimensions exist, there is no logical reason to hold that they will intersect anywhere near the center of any other relevant dimension. The manner of intersection is a matter for empirical enquiry and should take into account that a particular point of intersection need not be a stable one. Much will depend on such varying factors as the angle of crosscutting, the compatibility of the dimensions, the bargainability of

issues, and ad hoc or lasting coalitions (which may lessen the salience of particular dimensions).

The actual location of parties on a dimension (which is not the dimension on which they themselves are based) proves in practice to be a task of considerable complexity. As we saw, Abram de Swaan went through the exercise of trying to fit all European parties in the countries he covers on one overriding dimension, which, following Downs, but unlike Sartori, he defines in socioeconomic terms. He inevitably met with great difficulties which he could only solve by a combination of arbitrary choices and a redefinition of the right end of the scale, so as to accommodate clerical, extreme liberal, nationalist, authoritarian, and fascist parties indiscriminately, rather a case of saving the hypothesis of one unidimensional analysis!

Sartori's view on the existence of one overriding ‘space of electoral competition’ is argued in terms of logic rather than based on strong empirical validation. To my reading, he is in danger of paying too little regard to the dynamic nature of interparty conflict. Is interparty conflict not exactly about the relative salience of different dimensions? Both old and new parties seek to exploit new issues, possibly resulting in new dimensions of conflict. Each party tries to exploit to the maximum that dimension in which it finds itself most comfortable, while attempting simultaneously to drive a wedge into the ranks of its adversaries whose main advantage may lie on another dimension. Sani and Sartori offer interesting data on the congruence of attitudes on diverse questions and self-locations on a left-right scale. But even their own graphs show such a relationship to be far from absolute, which must imply that there is ample opportunity for parties to compete selectively with one another.25. Sani and Sartori single out Belgium as a special case of two-dimensional electoral competition. Is Belgium really unique or only one rather clear example of multidimensional competition, not unknown elsewhere?

Sartori's propositions need more empirical enquiry, and his theoretical scheme needs further consideration from the perspective of center and center parties as well as in other ways. For Sartori uses not only spatial metaphors and insights derived from properties of perception (the center becoming salient as an ideological space stretches), but also views about the mechanical interaction among parties as the number of relevant actors increases - changing a centripetal into centrifugal competition - and criteria of relative size (e.g., the development of more than two poles in a system, and the occupation or nonoccupation of the center by one or more center parties).

Can a Center Exist in a Nonreduceable Multidimensional Space?

Here we tread on a terrain that is best covered by formal theory. Judging by one study that specifically investigates our problem26. it seems solvable only if we redefine the notion of the center as some form of equilibrium point in a game-theoretical setting. It appears that such a point can be obtained only if certain specific conditions are set, e.g., considerations of minimum size, familiarity or inertia considerations which focus on established links between actors and which restrict the range of possible coalitions ruling out unfamiliar combinations; and consid-

erations concerning the relative dependence of different preferences of actors on one another, in order to compel certain combinations of stands because they are logically connected, and to rule out others. I fear that at this point I must leave the argument to specialists in formal theories who move in a mathematical world well beyond my horizon.

Is the Spatial Image (Whether Unidimensional or Multidimensional in Nature) Applicable to All European Party Systems?

One could think of two situations in which the spatial representation becomes meaningless:

A situation of complete Allgemeinkoalitionsfähigkeit of all parties: In such a situation only office counts, rather than with whom one gets office.27. Neither past coalition experiences, nor ideological dimensions, nor future policies offer any restraints. This situation is nearest to the reasoning of the coalition theory of the minimum-size variety. That this theory performed most poorly in empirical tests of European cabinet coalitions suggests that the assumptions are not very realistic.28. The need to reintroduce ideological dimensions or other restraining conditions or both may therefore yet save the notion of a centre in European party systems.

The second situation might be one of complete segmentation, dividing society into closed subcultures between which no logical relations exist. Typically, in such a situation neither normal majority assumptions nor considerations of ideological affinity hold. Instead, government proceeds by Proporz, a deliberate depoliticization of issues, a far-reaching autonomy of subcultures.29.

The center in terms of mechanics

If one moves from spatial reasonings to conceptions drawn from the world of mechanics, one meets with two rather different metaphors. First, one can imagine a set of scales that a particular force tips toward one side or the other. Second, one may conceive of centripetal against centrifugal forces working in a given party system. In either case, one can reason in terms of voters or of parties. In the case of voters, the emphasis is on the outcome of elections. In the case of parties the accent is rather on the formation or breakup of government coalitions.

The Image of the Scales: Floating Voters

A large body of writings - whether on formal models of party systems30. or more descriptive surveys of particular countries31. - gives a crucial place to voters in the middle. By moving their weight from one (alternative) coalition of parties to another, voters can determine the composition of governments. One often assumes that voters in the middle also represent a force for moderation. But many analysts question why this must be so. Are the floating voters the best informed or the most responsible? Or, on the contrary, are they the least interested and most indifferent? Or should one regard them as those most likely to be subject to cross-pressures, which makes them reject extremist demands on any particular issue? The debate on the floating voters is now at least two decades old32. and it is clear from empirical study that there is not one simple answer. A reification of the floating vote does

insufficient justice to the fact that it is in practice composed of varied and changing groups of voters. The very image of one ‘floating’ vote pays too little attention to crosscurrents, to changes in the makeup of the electorate through the entry of new voters and the deaths of older ones, and to the additional effects of voting or nonvoting. Voters in the assumed ‘floating vote’ represent very different opinions, regard rather different issues as salient, and need not be in the center at all.

Such problems are evaded in another mechanistic metaphor - e.g., the assumption that there is a natural swing of the pendulum. A party or parties in power - so it is argued cannot escape growing criticism that eventually will lead to a loss of voters shifting to opposition parties. That mechanism holds governments at least indirectly accountable, so that it cannot afford to move far out from the middle ground between it and the opposition. Empirical data do not offer substantial proof for that mechanical conception.33.

The Impact of the Scales: Key Parties

In a similar analogy, particular parties of various sizes are thought to occupy a position in which they can determine the formation (or ending) of a particular government coalition. Their role in this respect has been characterized also with a variety of other mechanistic metaphors such as pivotal parties, key parties, hinge parties, or partis charnières. Logically, their special position depends on the circumstance that their cooperation is necessary to form or maintain a government, given a situation that parties on either side of them lack a majority of their own and prove unable to enter into a (larger) coalition between themselves, bypassing parties in between. The argument has been strongly influenced by the traditional finding that before the Fifth Republic France was governed by what Duverger called the éternel marais of the Center, and by the postwar position of the Liberal FDP in West Germany. Yet here again questions must be raised. Is it justified to lump together - all and sundry - small parties that can balance two very much larger parties, the size of which makes the inclusion of any one of them in government a foregone conclusion? Should one argue that in an evenly balanced position of rival governing teams any party can develop into a hinge party, irrespective of its own political positioning in relation to the other parties? There are some empirical pointers in that direction (as the wooing of the nationalist parties in Britain in the 1970s. or the rather odd position of the Austrian FÖP in relation to Socialists and Catholics has shown). Yet in that case the mechanical metaphor is pushed so far that what would otherwise remain an out-party is made into a center party by definition. Compared to that type of argument, the earlier assumption about a relatively persistent patterning of parties in relation to one another in known dimensions of politics seems rather more persuasive.

The chief problem with both the voter and the party metaphor based on the image of the scales is the underlying assumption that a party system provides potential rival teams that must be in some form of rough balance with one another. In many actual party systems, that is not a realistic assumption: e.g., in systems with a predominant (let alone a hegemonic) party, or in complex party systems that do not spontaneously divide into two alternative blocs. In the latter situation, the like-

lihood that movement by a specific group of voters, or by one particular party, can force a drastic readjustment of forces is not very great. How useful is it, anyway, to reason in a metaphor of a set of scales, if a rather large party, rather than a small one, actually straddles the equilibrium point?

Centripetal and Centrifugal Mechanics

We must now return to Sartori's analysis of centripetal competition for voters at the center, which he regards as characteristic for systems with two to four relevant parties, competing along one and the same dimension with a low degree of ideological polarization, and of centrifugal competition, which to him is the mechanical disposition of systems of five or more relevant parties - not being segmented polities - with a high degree of ideological polarization.34. A major element in his analysis is the view that in the latter case the occupation of the center by one or more center parties encourages centrifugal drives. I offer the following comments.

Sartori has generally taken a rather negative stand toward center parties, although he grudgingly gave these parties a somewhat more positive evaluation in later years. I quote from two different publications:

In a situation of centrifugal pluralism the center party (or parties) appears to be more than anything else a feedback, or a retroaction, of the centrifugal drives [...] According to this interpretation, then, the center is more a negative convergence, a sum of exclusions, than a positive agency of instigation. And this is why it is likely to be a passive, rather inert, and, all in all, immobile kind of aggregation.35.

and I still believe in this diagnosis, but the recent Chilean experience, which was characterized by a chronic fickleness of the in-between parties, vindicates a more positive interpretation. I would put it thus: Even though the center parties tend to be immobilistic, they remain an equilibrating force that performs a ‘mediating role,’ and mediation, or brokerage, is not the same as immobilism. This having been conceded, I hasten to add that a center positioning seemingly condemns to a policy of mediation, in the sense that a different role backfires on the party's positioning without paying off in performance or accomplishment. A center party that attempts to outdo the parties located on its left or right will contribute, more than to anything else, to a crescendo of escalation and extremization.36.

Sartori posits that the very occupancy of the center area by one or more center parties puts that area ‘out of competition,’ and hence encourages centrifugal rather than centripetal drives. It is difficult to concur with this argument: as long as there are votes in the center, why should parties not in the center refrain from going after these votes in what must by definition be centripetal electoral tactics? Rather than speaking of centripetal competition in systems of moderate pluralism (generally having three to five parties) and of centrifugal competitions in systems of polarized pluralism (having five or more parties effectively competing with one another), is it not more reasonable to argue that in any system of three or more parties (in which at least one party finds itself in an intermediate position between other parties), there will always be both centripetal and centrifugal drives? In a

visual representation of competition among five parties,37. large extreme parties on the Left and Right overlap only with the intermediate parties, not with the large party at the center. But why should large out-parties not attempt to move toward the center at the expense of the intermediate and center parties both? Are not both the PCI and the MSI competing with the DC directly for electoral votes in contemporary Italy?

Sartori's argument is possibly influenced by the fact that the DC represents a center that is really based on a criterion other than the formal left- right dimension. Yet if so, he should be led to different conclusions by his own view that it is typically the left-right space of competition rather than domains of identification which governs the decisive relations among parties. A comparison of the Italian and the Dutch case would seem to confirm that conclusion. In both systems, the center is occupied and has been so occupied for many years. In the Dutch case, this fact has been a major element in forcing both Socialists and Liberals to tone down whatever anticlerical proclivities they may once have had, and indeed in forcing them to compete both in policies and electorally in above all a centripetal manner.38. Sartori will argue that the Dutch and Italian cases are not really comparable, given the presence of sizeable antisystem oppositions and a much wider ideological space in Italy as compared to the Netherlands. This is of course true, but is not one major reason for the lower polarization in the Netherlands that parties once distant from one another eventually learnt to compete in a centripetal manner, and hence in the end contributed toward depolarization?

On a more theoretical plane, Sartori finds one distinguishing feature of ‘moderate’ as contrasted to ‘polarized’ pluralism in the existence of ‘bilateral oppositions’ in the latter against ‘bipolar mechanics’ in the earlier case. In a self-locating left-right scale, the Netherlands comes out as a country that generally shows a number of distinct modes.39. One should note, however, that this finding comes out much more clearly with the use of 9- or 10- than with 7-point scales.40. Although the weight of the extremes in the Dutch system is very much smaller than that in Italy, it is far from evident that the Netherlands could really be regarded as an example of a ‘bi-polar coalitional configuration.’41.

One is struck by the fact that Sartori gives relatively little attention to size factors, in that the proportionate size of a center party, of adjacent parties, and of more extreme parties including whenever relevant antisystem parties, may be more important for the actual functioning of party systems than the absolute number of relevant parties. One should attempt to specify the importance of such size factors.

Analyses in terms of cleavage

Yet another manner to treat the presence or absence of a center in party systems is through analysis in terms of social and political cleavages. Here we again encounter Duverger, with his reference to the center as ‘the moderates of opposed tendencies’ and also in his derivation of the center from a ‘superimposition of dualisms.’ The following comments seem in order.42.

The Dangers of the Cleavages Terminology

The term cleavage has the unavoidable connotation that there is a sharp divide that separates social or political groups. It follows that the middle ground is a dangerous place to be, or even one that is logically nonexistent as indeed Duverger would have us believe.43. However, if one inspects the nature of actual political and social cleavages, it becomes apparent that the idea of a clear divide is much more applicable to certain categories of cleavages than to others. There is indeed little between blacks and whites, Catholics and Calvinists, native speakers of one tongue or another, or members of different tribes. But such a statement is not true for many other social bases on which parties have been formed, e.g., the criteria that to Hume were the three alternative bases of party: interest, affection, and principle; social class (except in a vulgarized Marxist sense); regionalism, or even some of the new social cleavages arising from a reorientation of values in a postindustrial society toward the environment, women's rights, issues of organizational scale, international peace, or special concerns with the Third World. A realistic analysis in terms of cleavages should lead one to conclude therefore that centrist and intermediate positions are a much more realistic proposition for some cleavages than for others, without succumbing immediately to what one might call the tyranny of polar types.

The Importance of Numbers

Even if political life were organized solely around exclusive cleavages, it makes considerable difference in practice whether it leads to the formation of two groups or more. Compromise and adjustment would seem much easier to reach when more than two ascriptive groups are involved, and especially when no single group has a definite majority position. This finding is indeed one common in both anthropological literature and in the writing on modern segmented polities.44. When one group has an unassailable majority, there is much less scope for brokerage than when numerical relations make some form of bargaining inevitable even to reach a majority point.

The Degree of Political Affect

One further disadvantage of a reasoning in terms of cleavages is that one concludes too rapidly that particular cleavages are unavoidably loaded with political content. It may be true that survey data demonstrate highly skewed patterns which suggest that different groups are opposed on matters like religion, class, or language. Yet the political relevance of such a finding depends on the actual degree to which such cleavage divisions are politicized, a point underlined also by Sani and Sartori when they argue that conflicting identifications need not result in effective political combat.45.

The Superimposition Hypothesis Once More

Duverger's derivation of a center through a superimposition of dualistic cleavages logically presupposes that particular cleavages have different salience for different groups of political actors. By assuming one cleavage to be the dominant one from

a systemic point of view, groups on another cleavage end up having a heterogeneous, hence by definition an ambiguous, position. This is indeed Duverger's biased view, but tenable only if one accepts the a priori assumption of the existence of one overriding dimension. It is a telling illustration of the powerful normative sentiments that accompany discussions of a center. What is detested by Duverger as preventing ‘real’ choices is hailed elsewhere under the perspective of ‘cross-cutting cleavages,’ regarded as the force par excellence which ensures moderation in a homogeneous political culture!

The Center as a Historical Residue

In yet another generalization, Duverger explains parties in a center position as the result of new parties springing up to the left of existing ones, which force the previously existing parties from their original left position to the center or even the right of the system. One is probably not far amiss in thinking that Duverger sees the new parties as representing the future, relegating older parties to eventual oblivion as no longer representative of living forces. The argument underplays the possibility of new parties coming in, not from the Left but from the Right, or even into the middle of the existing political spectrum (e.g., in Duverger's France the Gaullists and reforming Centrists, D'66 in the Netherlands, or the not easily placeable Progress Party in Denmark).

Even if one were to accept Duverger's law of a gradual displacement of older parties by newer parties on their left (i.e., in the light of the historical rise of new working-class parties), one should recognize that older cleavage lines are not necessarily doomed to extinction. One famous example is the course of events in Norway, where the original Left of peripheral protest parties was indeed shifted away from its original position to such an extent that one wing (the Agrarians) was eventually to re-name itself Center Party. Yet as Rokkan and Valen had correctly forecast,46. older cleavages could indeed become salient again once their values were threatened. In the 1972 battle over the EEC Referendum, many of the presumed Centrists joined modern New Left groups in polar opposition to a modern pro-European front of conservatives and establishment socialists alike. Clearly, one must reason not in terms of a static superposition of dualisms, but in terms of a dynamic interplay of cleavages, historically and in the contemporary world. History does not freeze parties in immovable positions, nor does it force them into an irreversible momentum in one direction only.

Labels and positions: the concrete experience of European party systems

So far, our search for the center has mainly followed conceptual lines. We have explored spatial analogies, ranging from a concentration on the median voter or median party, moving on to the assumption of the existence of a left-right dimension (which easily suggests the notion of a center location as well), and ending with the intractable problem of a multidimensional political universe. We have turned toward mechanistic metaphors in terms of a balancing of the scales and of centripetal versus centrifugal forces. Finally, we have concentrated on analyses

of cleavages which can be more or less dualistic, and hence allow or not allow for intermediate or brokerage positions. Our search has been far from conclusive. Our most difficult problem has been that of the inevitable multiplicity of issues, dimensions, and cleavages in political life which stubbornly resist attempts to reduce them to one dominant underlying dimension. (If that analysis is true, then one should query the facile use of the terms left and right as much as that of the term center, as all such terms become inherently ambiguous.) A further problem is the mixing of metaphors. Finally, there are clear signs of bias. Whereas some like a center, others do not. And with Sartori, one may even frown on center parties because one does like competition for the center.

Can one do better by inspecting the actual record of European party systems? One could do so by investigating to what degree the use of the term center by particular parties in particular party systems throws light on our problem, and by exploring the extent to which European party systems are in practice characterized by having parties at the center or not.

Checking the Labels

There is little doubt that the label center has gained in popularity among European parties in recent years. In Nordic countries a number of parties formally took on the label center as former agrarian parties renamed themselves Center Party in Sweden (1957), Norway (1959), and Finland (1965). In Denmark a splinter party from the Socialist Party called itself the Center Democrats (1973). In the re-emerging democracies in Southern Europe at least one party adopted the label center, although this label covered rather different parties in different countries, e.g., Suarez's Union of Democratic Center in Spain, the Democratic Social Center (now Christian Democratic Party) in Portugal, and the Center Union of Mavros in Greece (Duverger would have noted with satisfaction that neither parties nor labels proved durable, however).47.

Appeals to a center designation are also made in other countries, even though parties did not take on the actual label. Thus, in Italy the DC, the PRI, and the PSDI have claimed the presumed center space in the political system, and even the Liberal PLI has sought to out-center the DC from the right. In French politics a center tendency has been a consistent phenomenon, and the label has constantly reappeared for one or other group within it.48. One finds a general usage of the term in Dutch political dialogue. There was widespread discussion in Britain of the need for a Center Party before the new Social Democratic Party was formed in 1981 (e.g., The Times, January 17, 1980, following Roy Jenkins' Richard Dimbleby Lecture of November 22, 1979).

The concept left has been so strong, however, as practically to preclude the use of the term center by parties that find themselves hemmed in by a large party on their left (whether Communists or Socialists or both), and large parties on their right. Non-socialist reform movements have generally preferred terms like radicals, democrats, or progressives to that of center. In some cases parties that have willy-nilly found themselves in what might easily be regarded as objectively a center position have deliberately disclaimed the label that they found uncongenial

for ideological reasons (e.g., the Italian PSI). Even academic writers unblushingly dichotomize European parties into left and center-right parties, clearly regarding the label center as being more right than left.49.

Checking the Positions

One can also explore when and where parties have in fact acted as center parties in European party systems, irrespective of their labels. The following attempt at a simple classification is based mainly on the mechanistic assumption of the presence or absence of one or more parties which can effectively tip the balance to one government combination rather than another.

In a pure two-party system (such as the stereotyped British one), the idea of a center party would not logically fit. At most, with Downs and Sartori one should argue in such a situation in terms of the center as a point of gravitation to which the two competing parties would be irresistibly drawn. The traditional interpretation of British politics in terms of a Front Benchers' Constitution where the ‘Butskell’ rule, whatever the more militants in either party say, illustrates - and probably inspired - this way of reasoning.50. But at least two developments have since then changed this classical conception: the loss by Labour of a majority position in the Commons in the 1970s, and growing concern about the Downsian movement of the two main system parties away from the center. The first event gave life to the idea that other parties might occupy a balancing position (e.g., the Liberals, but also the various nationalist parties and, since 1981, the new SDP, i.e. the Alliance). Resistance against a growing adversary politics, on the other hand, has done much to upgrade the view that what Britain needs are not majority parties, but center parties.51.

A second case has been dubbed a two-and-a-half party system, i.e., a system with two large rival parties, each of which needs the support of a third party to obtain a majority in parliament. The archetype is of course the German Federal Republic since the early 1960s, where the FDP has used its leverage to obtain representation in national and Linder-governments well beyond its proportionate share. As compared to the German FDP, the Austrian Freedom Party seemed to be too far removed from what would be regarded as a legitimate center position to play a similar role. Yet with the fall of the Kreisky Cabinet in 1983, the strength of mechanical factors was such as indeed to propel that party into government with the Socialists. Could Irish Labor eventually develop similarly into a third force between Fianna Fail and Fine Gael?

In a number of European systems a situation has occurred where party conflict consisted of one large party facing a number of smaller parties. The clearest examples of such a situation have been Ireland and Sweden, where Fianna Fail and the Socialists face a number of much smaller rivals. Somewhat similar situations occurred for a time in Norway and in the German Federal Republic (when the CDU/CSU crossed or came near to the majority point). Often the out-parties had little choice but to remain opposition parties in such a situation, ruling out the possibility of regarding them as center parties. Yet such a situation presents a standing temptation to one or other of the smaller opposition parties to move over, as indeed the Agrarians did repeatedly in Scandinavia even before they called themselves center parties.

In a fourth situation conflict has been mainly between two rival blocs, as in the traditional two-bloc parliamentarism of Denmark, or in the more recent period of the French Fifth Republic. In such a situation, the movement of a single party can be of decisive importance. Witness, for example, the change in the balance of Danish politics when the Danish radicals went over decisively to a bourgeois bloc, or the flirtation of some socialists and majority centrists with the idea of a new center coalition, before the victory of Mitterrand in 1981 put an end to such speculations for the time being. What clearly distinguishes this type of political situation from the third situation mentioned above is the absence of a realistic expectation of an independent majority for any one party. Hence, there is room for tactical maneuvering for what is intuitively described as the Center-Left and the Center-Right to move either toward each other, or each to a possible ally in a centrifugal direction, which underlines the importance of quantitative proportions in addition to locational ones. It also suggests that terms such as centripetal and centrifugal may have a different explanation, depending on electoral or coalitional politics.

In other countries or at different times in countries mentioned earlier the typical situation has been one of continuous (or at least very frequent) rule by one or more parties that have changed allies from time to time.52. The typical cases are those of the Netherlands, Italy, and to a degree also Finland. The hold on the system by the Catholics (and to a large extent their Protestant allies, now all amalgamated in a new Christian Democratic Appeal) in Holland, of the DC in Italy, and of the Finnish Agrarians who were later to call themselves explicitly the Center Party, has been such as to make them of pivotal importance to the system. Yet within this category certain differences exist, depending on the extent to which these parties can actually enter alternating coalitions (the case of semi-turnover, typically found in the Netherlands), or are forced to look mainly in one direction (Italy, ever since parties on the right of the DC were regarded as clearly illegitimate, thus leaving only the choice between minority government and an apertura a sinistra).

In this category one should also place most of the French Fourth Republic and Belgium and Luxemburg, even though the role of the Christian Democrats has been somewhat less dominant in these countries than that of their Dutch counterparts.

Some countries resort frequently to the practice of minority governments,53. Parties called to exercise the role of government in such countries could conceivably be regarded as center parties to the extent that their presence in government, paradoxically, is resented least by all other relevant parties. However, for an understanding of the nature of actual interparty relations, the operation of parliamentary coalitions may be just as telling as the presence of a particular party in (minority) government.

Finally, there is a residual class which for want of a better term one might call that of changing coalitions. Relations of parties are such that neither clear alternation nor semi-turnover is present sufficiently to constitute a regular pattern. The chief candidates for this category would seem to be Iceland, but also Finnish and maybe Belgian politics offer some illustrations. Typically, parties may ally themselves along frequently changing dimensions and opt for grand coalitions as much as for caretaker governments of experts or minority parties.

The relevance of the notion of center parties for different European party systems

If the short inventory given above is correct, European party systems seem to fall in at least three different groups.

In some systems the notion of a center party hardly applies. This is logically the case under any of three conditions: countries with stable single-party majority governments (e.g., Britain, France under De Gaulle and Pompidou, Austria under Kreisky, Sweden until at least 1976), countries with clear alternating governments even when often not of the single-party majority variety (Denmark until 1973, Norway, Ireland), and in the rather different case of Proporz systems, in which all major parties enter government by definition (e.g., Switzerland).

In a second group of countries, the idea of center party or center parties is a key variable. This was the case in the French Fourth Republic and has been true of Italy, Finland, and also the Netherlands ever since the religious system parties forced Liberals and Socialists into the position of alternate suitors. It is also true for the German Federal Republic as long as the FDP can effectively play off Christian Democrats and Socialists against one another.

Finally, there is a third group of countries where the relevance of the notion of center parties is contingent on both size and positioning of major parties in relation to one another, which is dependent on the outcome of elections and the issues to be handled. In this intermediate category one might place Belgium, Luxemburg, Denmark since 1973, and possibly Iceland.

As this grouping suggests, the major determining variables are the relative size of parties (above all, majority or minority) and how stable the parties' positions are in relation to one another. Countries differ markedly from one another in both respects, and a particular country may also change from one category to another over time.54.

Power and Prejudice

The preceding pages leave two possible approaches still untreated.

First, one might pay special attention to parties continuously in power. Continuous exercise of power is clearly not the same as being a dominant party, since the first notion is compatible with the continuous presence of a small party in successive government coalition. Such a categorization points to a group of parties which provide important questions for detailed study, e.g., of the forces that make such a party indispensable for governments, and of the consequences that result from long government tenure for the electoral position of the party, for its role in relation to other political actors (notably the bureaucracy), as well as the consequences that result from its relationship to parties not in the government. But equating center parties with perennial governing parties otherwise confuses rather than clarifies understanding.

Second, we clearly need an explicit study of normative assessments of center parties. We have frequently drawn attention to the many unspecified value judgments associated with the treatment of notions like center and center parties. A care-

ful review of such value judgments might help us to obtain clearer insights into alternative theories of good government, which after all laid a heavy hand on our discussions and comparative research on the empirical functioning of party systems.


Vorige Volgende