Stefano Bartolini & Peter Mair: Identity, Competition, and Electoral Availability. The Stabilisation of European Electorates 1885-1985. Cambridge and New York: Cambridge University Press, 1990, 308 + xix pp. Appendices.

For the past decade or so, numerous studies on West European mass politics have argued that the stability of party systems format and voter allegiances is deteriorating. What has been described as a major transformation process of party systems, electoral dealignment, 'decline of parties', single-issue electoral behavior, etc., is all taken as proof of this decreasing stability. Ultimately, this discussion aimed at reassessing the often-quoted 'freezing hypothesis' of the Western European party systems advanced by Seymor Lipset and Stein Rokkan: the party systems of the 19605, they argued, reflect, with a few exceptions, the party systems of the 19205. In short, their argument was that, following a period of extension of franchise, the party systems, reflecting the major social and political cleavages, 'froze'.

Stefano Bartolini and Peter Mair's recent book identity, Competition, and ElectoralAvailability. Stabilisation of European Electorates 1885-1985 challenges the current predominant perspective on West European electoral behavior. Not only do they question previous reports about electoral instability, rather their argument is that the electorates of Western Europe have indeed undergone a process of stabilization. The West European party systems, they suggest, are characterized by a 'fundamental bias towards stability' (p. 287), and factors that may function as incentives for electoral volatility appear to be weaker than those

factors that impede such volatility. Contrary to most other analyses in this field,
they find that

.. . judged in the longer-term, the exceptional period no longer appears to be the 'volatile'
1970s and 1980s, but rather the steady-state period of the 1950s (italics in original, p. 99).

Bartolini and Mair repeatedly underscore the importance of looking at changes in electoral volatility in a longer perspective than just the 1970s or 1980s. It is only in this longer perspective, they argue, that it becomes possible to assess the more recent trends, if indeed there are any.

It should be said at this point that this is an intriguing and provocative study that is likely to be widely read and cited, and rightly so. Awarded the 1990 Stein Rokkan Prize for this book, Bartolini and Mair address a key problem related to West European politics; then, drawing on a vast literature which they are in complete control of, they proceed to elaborate a sophisticated theoretical model which is subjected to careful empirical analysis using an impressive data set; and finally, they conclude their study by returning to some of the questions raised at the outset. So far, it is an exemplary piece of scholarly work.

The authors state the four overall purposes of their study. First, they seek to assess the actual extent and significance of the electoral volatility. Given the recent debate on this topic, they argue, their initial task must be to describe the actual electoral volatility across time and across different polities. Secondly, and closely connected to the first problem, they give an assessment of the actual decline of cleavages in the aggregate electoral change. Thirdly, they analyze what is meant by electoral volatility, and fourthly, they are interested in 'the relationship between electoral instability and patterns of competition' (p. 6). The first three of these purposes dominate the analysis while the fourth aspect is addressed primarily in the concluding chapters.

The key dependent variable in the study is the aggregate electoral volatility in 13 West European democracies between 1885 and 1985. All in all, the study covers 303 general elections. Although the Lipset-Rokkan model also considers other social cleavages like, for example, the center-periphery cleavage dimension, Bartolini Mair choose to concentrate on the class cleavage. Thus, electoral volatility is treated at two different levels, total volatility and class volatility.

The authors employ six different independent variables in their model: (1) the extent of closures of important cleavages, i.e., the class cleavage; (2) the policy distance between parties on opposite sides of the class cleavage; (3) the party system format, primarily the number of parties; (4) changes in electoral institutions; (5) changes in electoral participation; and (6) short-term factors, unique events, etc. (basically a residual category). In part derived from the first of these factors, the authors then elaborate on two additional independent factors, namely cultural segmentation (primarily religious and ethnolinguistic segmentation) and organizational (the organizational strength of party organizations and trade unions).

The thrust of the volume presents an analysis of the explanatory powers of these independent variables on the aggregate electoral volatility. Policy distance is the variable that seems to cause the biggest problem and it is consequently redefined to add to the explanatory power of the model. Space forbids a full account of this analysis, but suffice it to say that all of these variables add - albeit to a varying extent - to explaining electoral volatility. However, some of the relationships are positive, whereas others are negative. Thus, change in electoral institutions, change

in electoral participation and party system format all have a positive relationship with electoral volatility, while spatial distance, cultural segmentation and organizationaldensity negative relationships with the dependent variable. Put differently, changes in the institutional framework for electoral behavior seem to be conducive to electoral instability - the analysis reports positive correlations between changes in participation and volatility as well as between extensions of franchise and volatility - whereas factors related to the political and social significanceof organizations in society (parties and trade unions) impede volatility.

Given the partially positive, partially negative relationships between the independent and the dependent variable, the concluding chapters cluster the independent variables into two major groups; first, 'institutional incentives' (factors with a positive relationship with electoral volatility), and secondly socio-organizational (factors with a negative relationship). In the overall assessment, this model explains some 45 percent of the total variation in electoral volatility, which no doubt is an impressive result.

As a first overall result - and contrary to what seems to have been the common belief during the 1980s - the analysis clearly shows that mass political behavior in Western Europe is not characterized by increasing volatility but rather that the opposite is the case. Indeed, they suggest that the past hundred years has been a 'century of electoral stabilization' (p. 287):

If national electorates can be seen to be torn between those forces which sustain and reinforce the complex of political identities, on the one hand, and those which proffer a stimulus to instability, on the other, then the single most striking impression which we have gained from this analysis is that the former will continue to prevail (ibid).

Needless to say, this significant stabilization of the Western European electorates during the past century is a key finding in the analysis and something which should challenge most other scholars of electoral behavior. What has happened, Bartolini and Mair argue, is that during the time period studied, socio-institutional bonds have gradually become stronger while the institutional incentives have been weakened.

Secondly, looking at the importance of the class cleavage over time, Bartolini and Mair argue that although the significance of this cleavage has not declined, the nature of the cleavage has changed. Class has moved from being primarily a 'dimension of competition' (high salience of the class cleavage and a high block volatility) towards becoming a 'domain of identification' (low salience of class cleavage and low block volatility) (cf. pp. 45 f., 80 ff., 290). In fact, it is suggested that the 'bias towards stability' is even more apparent when looking at class cleavage volatility than at total volatility. However,

The decline of cross-cleavage mobility signifies a reduction in the salience of the class
cleavage in any given country, and, in this sense, one can speak of its declining importance
as far as electoral competition is concerned (italics in original; p. 292).

In the third and final major set of conclusions, Bartolini and Mair suggest that the above discussed 'socio-organizational bonds' and 'institutional incentives' 'appear as the two most powerful "structural" determinants of electoral availability. ..' (p. 293). Electoral 'availability' refers to the number of mobile voters and this concept is then related to the concept of the 'political market'.

As a whole, the authors conduct a sophisticated analysis and produce powerful
empirical evidence to sustain their arguments. The book elegantly summarizes the

previous research in this field and provides detailed theoretical discussions on most
of the key concepts related to electoral behavior. In addition, the empirical analysis,
as well as the interpretation of the data, is conducted very carefully and thoroughly.

Having said this, it should also be mentioned that Bartolini and Mair's book is also associated with a couple of minor problems and ambiguities. First, the introduction the volume gives the reader the impression that this book is about the relationship between stability at the level of the party system on the one hand, and stability at the level of the polity on the other. However, as should be obvious by now, this is not the problem that the authors address in this book. They point at what they believe to be a paradox: on the one hand, 'electoral stability or even electoral stalemate appears as a necessary prerequisite for democratic consolidation' (p. 2). On the other hand, 'a degree of electoral instability also appears to be necessary in order to ensure democratic responsiveness' (p. 3). They then go on to suggest that - as most would argue - these are two different aspects of democratic theory, or problems at two different levels of the political system. Finally, the argument is brought forward that they both relate to 'the dynamics of aggregate electoral change in Western Europe' (p. 3).

This is a slightly misleading introduction to the book, simply because it does not tell the reader what the book is all about. It becomes all the more unfortunate as the authors later make an excellent theoretical and empirical dissemination of their problem. To be sure, had it been the purpose of the analysis to study the relationship between electoral stability at the level of the party system and at the level of the polity, then their findings would have provided an interesting point of departure for a discussion on that problem. Since factors that seem to encourage electoral volatility are laid down at the level of the political system, they are basically a matter of policy choice, whereas factors that inhibit such volatility are closely associated with the political parties as well as other voluntaristic associations.

Secondly, and more importantly, there is the obvious methodological problem: to what extent can indeed electoral volatility by analyzed at the aggregate level? Bartolini and Mair (rightfully) address this question at some length. Drawing on studies comparing survey analyses with data reporting the aggregate electoral change that report a correlation value between individual and aggregate volatility of some 0.5, they argue that aggregate data do indeed offer a good measure of electoral volatility. In an effort to give this statement further support, they suggest that by excluding "deviant" cases, the correlation may in fact rise to a figure as high as 0.92 (p. 29). Furthermore, a simulation analysis produced a correlation between individual and aggregate volatility of 0.60 or higher, depending on the number of parties in the simulation (p. 31).

Just as no-one would argue that individual electoral volatility and aggregate electoral volatility are two different phenomena, not everyone would agree that they are identical. Indeed, after investigating both previous analyses of aggregate data and conducting their simulation, Bartolini and Mair conclude that in both analyses there is a correlation of never below 0.5 and

(S)uch a degree of concurrence confirms the basic validity - given a sufficient number of cases - of considering aggregate volatility as one approximate indicator of individual volatility. And while the relationship should be interpreted with some caution, it can be justified not only as a postulated heuristic device, but also as one which is both empirically and statistically likely (p. 34).

While this appears to be a reasonable argument, it occasionally appears as if the
authors seem to pay only a fairly modest attention to the implications of this

introductory methodological caveat. Indeed, in several instances averages are calculated on other averages in order to categorize and summarize the vast amounts of data and in this process the perspective on aggregate volatility as 'one approximate indicator of individual volatility' is lost. It might have made the reader feel more comfortable if some of the most aggregated findings and conclusions had been framed by the initial note of methodological caution.

This problem also becomes apparent in some of the correlation analyses. In a number of such analyses, the authors choose to exclude deviant cases or outliers in order to strengthen the correlation under study (cf. e.g. pp. 136, 167 f., 183, 186, 208, 236 f., 241 f., 256, 257). Certainly, there are often good arguments for this exclusion - cases are said to be deviant because of unique events or representing periods of extraordinarily high political turmoil, etc. - but sometimes it is not quite clear why deviant cases are excluded, apart from the fact that they are deviant. If the only purpose of excluding deviant cases is to strengthen the correlation, then a more heuristic strategy might have been to accept a low overall correlation coefficient in a residual analysis assess why the deviant cases do indeed deviate.

The slight confusion over the exact focus of the analysis caused by the introductory pages and the choice of method and type of data are two closely associated issues. Had it indeed been the authors' purpose to investigate the relationship between polity level stability and party system stability, then building the empirical analysis on aggregate data on electoral volatility is probably the best choice. At the level of the polity, the democratic theoretical concern for stability does not have to address the question of the total exchange of votes between the political parties. Instead, it is the long-term stability of the most salient cleavages in society and the political accommodation of these cleavages which become the most important issues. Conversely, looking at the stability of the party system, one would assume that - next to changes in the format of the party system - the total electoral volatility is the key analytical problem, while the net or aggregate changes are only an approximation (albeit a good one) of these total figures.

Bartolini and Mair's chief interest is in how aggregate electoral volatility can be explained. They find that variables related to the level of the polity (institutional incentives) as well as factors related to the party system (socio-organizational bonds) explain this volatility, albeit in opposite directions: polity-level factors seem to encourage electoral volatility, whereas party system factors seem to impede it. Although this is a strong simplification of a complicated relationship, it nevertheless suggests that Bartolini and Mair with their results in hand could have returned to the paradox outlined in the introduction and made an interesting argument about the relationship between the polity's need for democratic stability and the party system's need for electoral flexibility to be responsive. Indeed, as it happens they turn their paradox upside-down: polity-level factors do not impede volatility but instead encourage it, and party systems factors which, according to the theory of democratic responsiveness, should encourage volatility, in fact turn out to do just the opposite.

However, in the overall picture these are minor comments and thoughts triggered by the analysis advanced in this book. They should by no means conceal the fact that this is an excellent book and one of the rare volumes in this field that cover theory and discourse as well as sophisticated empirical tests of the theoretical propositions. No doubt, this is a highly worthy Stein Rokkan Prize-winner.