Postwar issues in 23 democracies

Postwar Issues in 23 Democracies Ian BUDGE Essex University Working Paper n.44 Barcelona 1992 Summary The paper starts by considering problems in the definition and measurement of political issues -in particular, whether issue-creation involves all parties and both leadership and electors; or whether issues can be "created" by only one of these. Confrontational approaches to the handling of issues are contrasted with saliency theories. The choice of a conceptual approach affects the way issues and change are measured and represented, whether this is done through typologies or by spatial analysis of party movements. Spatial analysis presents two possibilities in regard to issue change: parties may change positions over time within the space; but the space itself (defined in terms of issues relevant at each time-point) may change. Both are considered on the basis of codings of sentences in party election programmes across a number of democracies, since the war. This is supplemented by trends in newspaper reports of campaign issues over the same period, which seem to indicate that, in the long term, issues are generated by the objective problems facing governments at the time. The discussion ends by briefly considering consequences of issues for government formation and action, and how far commitments to take action on issues are put into effect, as party mandate theory would imply. The question of issue change is so vast, extending from party competition and government change through media and communication studies to election studies and voting behaviour, that it is narrowed down here to comparative research done over the last 15 years on party interrelationships in elections and governments, in roughly the same group of-democracies -Western Europe, North America, Australasia, Japan and Israel. Within these constants a variety of data -surveys, documents, reported eventshave been used and analysed by various techniques, including spatial analysis. The distinction between party leaders and electors has always been central and where possible their mutual influence on each other has been examined. Use of different sources, techniques and subjects for analysis throws into relief questions usually neglected at a purely abstract level -such as: how do we identify issues in the first place?- and how do we put them into the same frame of analysis for comparative purposes? These are the first questions the paper takes up, before going on to more theoretical points, (which still affect measurement and operationalisation however). 1. DEFINING ISSUES While it is possible to think of issues being defined and discussed independently of political parties (in the media for example) they are for purposes of political analysis usually related to what the parties say and do. What do we mean by parties however? -leadership or electoral base or both? Bearing in mind this potential ambiguity we could define an issue variously as: a) a point "at issue" or anyway discussed by all parties, so that in terms of a spatial analysis the concurrence of all is necessary to define the space, b) a point or topic emphasised by only one of the parties (each can contribute independently to defining any space, so that a dimension of concern to one may not interest another -this has consequences for the kind of metric one might want to use in measuring distances within a space) c) a point or topic or "position", whether or not stressed by leaders, important in: i) defining party support among electors (in which case it could presumably include the "permanent issues" of class, religion, ethnicity or urban-rural divisions stressed by Lipset and Rokkan (1967) or ii) moving support between the parties -a narrower and more dynamic concept, within which issues are more likely to be short term. These are very different conceptions with strong bearings on what one might identify as issues and therefore how one might identify change in relation to them. There is little basis for deciding in the abstract which definition is better: presumably it all depends on one's research purposes. As my interest has been primarily in the interaction of issues as defined by the party leadership with those important in consolidating party support among electors (and sometimes in moving electors between parties) I would want to be as inclusive as possible and define issues as topics raised by one or more party leaderships and/or important among electors. Part of the research interest then lies in estimating the extent to which issues defined in each of these ways coincide, not simply assuming that they do as in most theories of the Downsian tradition (Downs, 1957). In discussing issues in this sense it is important, obviously, to specify their provenance and source (in which party leadership and/or electorate do they originate or exert influence for example?). 2. DOCUMENTING ISSUES AND TRACKING CHANGE In tracking issues one may draw on various data-sources. Again what one discovers is fairly heavily dependent on the source used, so it is worthwhile to comment briefly on these in passing: a) By far the commonest source for identifying issues are election surveys -indeed so common are they that many analysts use surveys with the implicit assumption that their questions define the universe of issues that existed for a given election. While these questions usually represent the best guesses of the investigators about what would be issues at a point 2-3 months before the election, they are informed guesses by experts using polls and similar sources as evidence. Surveys have two other features. One is that certain questions get carried on from previous years because of a long term interest in electors' reactions to them, not because they were necessarily prominent in that particular election. The other is that electors are often invited to talk about politics generally, in response to "open" questions, though so far as I know no one has ever used this issue material to check on the "closed" questions included by the designers of the questionnaire. b) An alternative source for tracing issues are media descriptions and analyses, especially as summarized in compendia like "Keesing's Contemporary Archives" or in retrospective accounts of campaigns like the Nuffield General Election Series in Britain; the Making of the President or the Enterprise Research Institute series in the USA; dealing with elections in a variety of countries. Media accounts are clearly biased by the position of the observers and are unlikely to be based on any very systematic sampling of opinions. Nevertheless they do comprehend the whole campaign and reflect opinions and events both at elite and electoral levels in a way no other source does. c) One can also base oneself on party documents, particularly the election programmes (in the US the "platforms") of the parties. Though these are directed at electors they clearly originate with the leadership and reflect its opinions or strategies. They are also written at varying times before the campaign opens and in many countries are read by few electors. As against this they are read by commentators and media people, who use them as a basis from which to begin their discussions. In this indirect way they reach the general public and shape many of the themes of the campaign. They thus constitute a major source for the analysis of issues change (Sections 6-8 below). d) One should also mention legislative sources and proceedings, such as roll calls, bills and other outputs and debates. These differ from the sources mentioned above in that they are non electoral, falling generally between campaigns. They tell you a lot about the questions which preoccupy party leaderships at a particular point in time: these often of course precipitate the fall and rise of governments and could provide issue-information highly relevant to coalitionformation. In theory such information could also be related to polling information about electoral opinions. Clearly the different source-materials may give rather different pictures of what issues are relevant at any one point in time. An attempt at comparing judgements based on surveys, media and party documents for the 1960 Presidential election in the US and the 1964 General Elections in Britain is made in Tables 1 and 2. Obviously themes culled from each of the sources often coincide: when they do they are listed on the same line, parallel to the corresponding issue among electors as ascertained from survey sources (column 4). Some themes stressed by politicians were not asked in surveys (presumably because their resonance among electors was not apparent). Some survey questions which divided electors very effectively into partisan groups were not stressed at all by politicians according to our sources, e.g. for Britain in 1964, the question of entry to the Common Market. These are listed in column 5 of the tables. It will be noted that only one trivial issue appears in this column for the U.S. in 1960, as compared to three issues of some weight for Britain in 1964. This contrast holds generally for British and American elections. Generally in the United States for all elections examined, all issues could be seen as originating with leaders. For Britain most do, but there are usually two or three questions, important and divisive among electors, about which leaders seem to have said nothing. Two additional aspects of leaders' influence on electors need to be noted. First politicians must be credited with putting up particular Presidential and Prime Ministerial candidates, who make some impact on electors. Candidate reactions are therefore listed in both tables at the top of column 4 as an effect attributable to leaders' cues. Secondly, leaders can at times give considerable weight to long-standing issues (such as nationalisation in Britain) which are not very prominent among electors. In spite of discrepancies between leaders and electors and between sources, there is perhaps more convergence on characterisations of issues in these Tables than one might have expected. So far as I know no one has done a systematic comparison of this kind over an extended series of elections. This simple validation is obviously a first step to be undertaken in a more comprehensive study of issue change. The encouraging results of the limited comparisons for 1960 and 1964 encourage some substitution of one source for another in the general tracking of issues, though it also appears from the Tables that leaders are likely to nominate many more issues than electors actually take up. 3. HOW DO ELECTORS AND PARTIES REACT TO ISSUES-CONFRONTATION VERSUS SALIENCY The way in which parties and electors react to issues has considerable implications for the way we measure them. The conventional picture of party competition is one in which the parties endorse opposing positions on the same set of issues; and electors estimate the agreement between their own position and those of the parties and vote for the alternative which is closest. This kind of picture is very familiar from Downs' classic treatment (Downs, 1957) and from the many spatial analyses in the Downsian tradition. However when one actually studies electoral programmes it is hard to find much evidence for confrontation, at least at the leadership level. If one examines the number of references to other parties it is on average about 10 per cent of all sentences in the documents and party-related issue references are even lower – about 5 per cent. Clearly this minority of sentences does not contain the main thrust of what is being said -which we must take as the party leaderships' attempt to attract electors to vote for them (Budge, Robertson, Hearl, eds., 1987, pp. 389-391). What do these documents contain? For the most part assertions of the importance of various problems (unemployment, social services, strong defence, law and order, the erosion of freedom, the environment, etc. etc.)… in the form of rather rambling analyses of past history and of the current state of affairs, with a view to stressing the urgency of action in the particular area. These analyses do not give great prominence to other parties or what they have done. But they do talk a lot about the originating party's record and attitudes. The importance of party documents of this kind is that, unlike survey questionnaires or media reports, they come directly and unmediated from participants in the struggle for votes. They must therefore be taken as direct evidence for the way parties handle issues and for the way in which the leadership at any rate thinks electors will react. How do they think of issues in this context? Clearly not primarily as direct confrontation, since so few of their references are dedicated to contrasting party policies. Rather they seem to be engaged in stressing the priority of certain topics, which they discuss at great length, in comparison with others, which they discuss in passing or hardly at all. Why should parties present themselves in this way? It seems that party leaders have a picture of the way electors react to parties which differs somewhat from the one prevalent among survey analysts. Party leaders stress certain issues more than others because they think they would benefit if those they favour become prominent in the campaign. This in turn seems to assume an almost automatic linkage between the prominence of a certain type of issue and party advantage (in terms of votes). The most plausible explanation of why politicians present themselves in this way, is that they endorse a "saliency" rather than a confrontational theory of voting. They think that the majority of electors see different parties as having the obviously "correct" policy for particular areas. Thus "left" or reformist parties will expand social services; right wing parties will be tough on offenders in the field of law and order; left or reformist parties will generally intervene more where intervention is needed; while right wing parties are more likely to reduce government activity and implement tax cuts. There seems little doubt in politicians' minds about most electors endorsing one "obvious" course of action in each area and clearly identifying the party most likely to carry it through. In view of this, they seem to think that elections will be won by the party most of whose proprietary issues assume prominence in the election. So, much party effort seems to be devoted to making "their" issues prominent and down playing those of the opposition. (For a similar interpretation of issues effects, which he terms the dominance principle, see W.H. Rikers's discussion in Riker, ed. 1922, Chaps. 1 and 2). These tendencies are reinforced by the difficulty parties have in dissociating themselves from their own history and actions in government, which give them an indelible association with certain policy stances. This implies in turn that they are able to shift their policy positions only gradually, so they are far from enjoying unlimited mobility over the entire policy field. Clearly again they must find it difficult to leapfrog as it would be quite impossible for a Labour party to pretend it was more conservative than the Conservatives. They would lose many longstanding supporters to no avail as nobody would believe they had changed radically in any case. All this has considerable implications for the way one operationalises and codes issues, pointing to the importance of the space or attention devoted to a topic, rather than the nuances of what is said about it. And in fact one striking empirical finding in coding party documents is that remarks about any area are usually of one kind. One does not discuss defence for example to attack military expenditure but only to support it. The same goes for social services. Proof for the assertions of saliency theory comes above all from the very plausible results, including spatial maps, obtained when one applies it to documentary data (Robertson, 1976; Budge and Farlie, 1977, Chap. 12; Budge, Robertson, Hearl, eds. 1987). This supports the general impression you have from reading party documents or following political discussions in the media, that party spokesmen rather than discussing a problem (say crime) on some common ground, will switch: the more right-wing emphasises conventional law and order, the more reformist the fact that crime breeds in poor social conditions, which call for more intervention and services. One further piece of specific evidence, apart from the low rate of programmatic references to other parties and their issue-positions, is the fact that specific commitments and pledges are so limited in party documents and confined almost exclusively to marginal areas (Rallings, 1987). One simply cannot code and election platform in terms of specific stands without throwing away most of the material: but surely everything that is written has some purpose? If so, saliency theory is the only explanation that makes sense of 90% of the sentences in the election programmes. 4. CHARACTERISING ISSUE POSITIONS; SCORES AND TYPOLOGIES It is possible to characterise party and electoral positions by simply recording stances on every conceivable issues. For most purposes however one needs to summarise and estimate the overall issue position. This is particularly necessary when one wants to estimate distances between parties, and between parties and electors, to operationalise voting and coalition theories. Again, there are a number of ways of doing this. It is salutary to reflect that many -perhaps mostsummary characterisations are not spatial in nature. I review these here, going on in Section 5 to consider the spatial characterisations that may be applied. a) A first, heroic solution is to concentrate only on the most important issue and characterise parties and electors in regard to that. (In terms of a saliency characterisation this would be the contrast between each party's leading topics). A specific example of this is where Budge and Keman (1990, p. 83) -partly from paucity of data as well as from a desire to simplify-hypothesise that parties agreeing on the most important issue would be more likely to enter into governmental coalition with each other than those that did not, where no obvious structural impetus to coalition existed. Given the difficulty of getting data from newspaper accounts, the hypothesis had a fair degree of success: 58 per cent of relevant cases. b) From the point of view of trying to predict the direction of vote in elections, summary scores for each party may do very well. The simplest solution was suggested by Kelley and Mirer (1974). For elegance and parsimony it has never been bettered. They simply took advantage of the "open" questions about issues and candidates included in practically all national election surveys, and added up positive and negative references to each candidate algebraically (each counting one). Electors were then postdicted to vote for the party with the highest score -a characterisation which seemed to fit the data for the US very well. c) More directly related to the discussion of issues as such, rather than vote-prediction is the idea of typing specific issues into long term "clusters" or "areas" as in Table 3. The advantage of doing so is that the "types" are relevant over all elections whereas specific issues appear and disappear. In line with saliency ideas one can assign most of these types to one party or another (in terms of their prominence in an election causing a net inflow of votes to one of the parties). Election failure or success can then be explained in terms of more of one parties' issues than another's predominating at an election. The advantage of such a characterisation is that it can be applied over all elections for which newspaper accounts of prominent issues are available i.e. in practice, over all elections in democracies since the second world war. Not only that, it can be used to predict election results in advance on the basis of such accounts. Predictions on this basis have had a fair degree of success (Budge and Farlie, 1983, pp. 84-114). The advantage of this approach in the present context is that it can be used to survey the incidence of issues of the various types in elections since the war: Tables drawn from this study form in fact the centrepiece of Section 8 (Table 10 below). 5. SUMMARIZING ISSUE POSITIONS: DIMENSIONS AND SPATIAL ANALYSES The power and flexibility of non-spatial analyses are much greater than is commonly credited. But spatial analyses have well-known advantages of direct, intuitive representation and openness, to mathematical analyses and formal modelling, which render them very appealing. This is even truer of one-dimensional representations which have the advantage, usually, of avoiding cyclical voting problems. A pervasive problem with many analyses, however, has been that in their eagerness to utilize the power of a spatial approach they have been unselfcritical -even unconsciousof the operational assumptions necessary to produce it and on which their results depend just as much as on the highly refined mathematical postulates which they use explicitly (cf Budge and Keman, 1990, pp. 2130). This is even more true when theoretical analysts have availed themselves of empirically derived spaces, for there it is even clearer that what goes in determines what comes our -and that is not always an issue-space! I illustrate these points below: a) The commonest type of space used to explain and predict government coalitions is a judgmentally derived Left-Right space, where country specialists have been asked to order and space out parties along a continuum between (usually undefined) Left and Right ends. Usually this puts communist and radical left parties as well as extreme Right parties way out at each end of the line, so it is not hard to predict that the coalition will form from those parties of the Centre which are closest to each other. This kind of judgmental placement often leads to an implicit tautology. It is difficult to specialists making this kind of judgement to confine themselves to however, may be a misleading analogy for issues, as one gets the impression both from party try-ons in their programmes, and from the way in which issues bob up and down in elections, that they are never fixed enough to be on a definite agenda: it is rather a process of constant change, and party competition to push others' issues out and your's in). It is difficult to generalize very much on the basis of Table 8. The same broad topics clearly appear among leading issues in all countries, for example, "social issues", "economic orthodoxy", etc. However, they do not show the same pattern of rise and decline. In the Netherlands and Britain the attention paid to social services (up to the early 'eighties) was stable: in Belgium it was declining, as also in Sweden (but from a very high base there). Economic orthodoxy and free enterprise, which in retrospect one might have expected to be consistently rising, were instead declining in Denmark, Belgium and the Netherlands, but certainly rising in Australia and Denmark. The few general tendencies which emerge are: a) questions connected with traditional morality (which in party programmes is more the virtues of the family and Christian values rather than the minefields of abortion and divorce) are in general decline. This may reflect deconfessionalization, particularly in the Netherlands b) attention to agriculture is generally declining, reflecting the migration from country into town which has taken place in practically all these countries c) attention to the military seems to be declining to the early 'eighties (apart from the United States) -running in parallel with the reduction of international tensions. The difference between the United States, where the attention the topic gets is stable, and other countries, is interesting because of course the U.S. was more heavily involved in military ventures during the later period than its allies d) along with the military and defence, there seems a general decline in attention to international affairs in the later period, again perhaps reflecting the more reassuring atmosphere as detente proceeded. Issues to which more attention was paid in the later period are: a) technology -rising in the US, UK and Italy and stable elsewhere b) a cluster of "new politics issues" -democracy, decentralization, social justice and environmentnot always the same in every country but recognizably related. The most consistent issue to enter in is the environment, although only in Sweden does it jump to almost the most prominent role. Mostly the entry of new issues is signalled by modest increases in attention to a topic which brings it to the bottom of the leading ten. There is a hint from the evidence on party programmes that issue change is a function of economic, social and "external" political changes, particularly in international affairs. This would be a reassuring conclusion as it would demonstrate that parties do function as transmitters and debaters of real problems, which governments may then be stimulated to take action on. It would be nice if it were so, since it would buttress our faith in the ability of democracy to cope with contemporary problems. There is, however, one piece of counter-evidence from the programmes which we ought to consider; and two general points to make on the way in which the "real world" gets reflected in issues. The counter evidence is the general absence of references to the European Community in most of the countries directly affected by it. Only in Italy and Belgium does it appear as a rising issue. Perhaps, however, this is because it is not controversial in many places -though this would not be true for Britain and Denmark in the 1970s. The reflection I want to add on the genesis of issues, is how far the environment is really an "objective" problem creating immediate difficulties for people, and how far it is an issue created by some scientists and the media. Objectively, in the countries considered, most people probably suffer less, directly, from pollution than in the smoke and smog-shrouded cities of 50 years ago. Yet it is clearly a "rising" issue. Declining international tension is also not experienced directly -though it is true that in the earlier postwar period wars were actually being fought. And environmental change also seems to be a real process happening "out there", as well as a subject of concern for scientists and media. The media do presumably, most of the time, reflect the "real world"? Another general aspect of issue-generation at the international level is the effect of imitation and transmission. Although they may be insulated to some extent, no national politics takes place entirely in a vacuum. In particular what is said and debated in the United States has an enormous impact elsewhere -probably it is an asymmetric relationship. Thus it may be that the environment issue was touched off by objective conditions in the United States where the big cities probably are more polluted than 50 years ago; and then spread as an issue to Western Europe even though the cities are cleaner than they were. Whatever the plausibility of this particular speculation, there is no doubt that imitation and transmission effects need to be reckoned with. ii) Changes in Newspaper Accounts of Campaign Issues Further evidence on the processes behind issue-change comes from another study, of campaign issues mentioned by newspapers over 23 countries in the postwar period, up to 1980 (Budge and Farlie, 1983, pp. 35-41). Codings here were made on a different (though broadly compatible) categorization to those of electoral programmes (also based on saliency ideas) into 14 broad categories, including candidate reactions (see Table 3). They are not directly comparable with the programmatic issues discussed earlier but do give an alternative estimate of postwar trends to compare with them. Table 9 first of all gives an overall summary of the issues that emerged in elections over the whole postwar period within each country. From this we can see (always constrained by the type of classification made) that the most widespread type of issue which enters into the great majority of elections in all countries, in nerwpaper terms, is government records. This is hardly surprising, since elections are designed to choose a government and the most immediate ground for choice is how well the contestants have performed in the past or seem likely to perform in the future. The record has always been assessed primarily in economic terms but there is an increasing concentration on this in the later postwar period. In most countries government record is as predominating a concern as it is overall. In Belgium, however, it enters into only six out of ten elections -buried very often by the disruptive linguistic divisions of the 1970s. Switzerland is the other country where government record is not prominent, but this is a reflection of the absence of national issues from many Swiss elections rather than a downgrading of government record in relation to other types of issue. The personal characteristics of candidates (usually for the chief executive office) enter into slightly more than half the elections. Again it is unsurprising that the qualities of potential leaders should form such a widespread basis of assessment. Being less widespread than government record, however, there is more room for variation between countries. The United Kingdom, West Germany, India and Sri Lanka have candidates entering as a major considerational into all elections, followed by Australia (eleven out of thirteen), Canada (ten out of thirteen), Ireland and Israel (nine our of ten) and the United States (six out of nine). On the whole it seems that countries with a broadly Anglo-Saxon parliamentary tradition and tendencies to a two-party system tend to focus attention on candidates (however, West Germany does not follow the Anglo-Saxon tradition and Israel has not until very recently had anything like two-party competition),. Democracies with least stress on candidates are Denmark, Finland, Iceland, Italy, Japan, Norway, Switzerland and the Low Countries. These are on the whole the smaller North European countries, with the significant exceptions of Italy and Japan. Perhaps the factionalism of the dominant party in the last two countries prevents national leaders appearing as more than nominees of special interests. The institutionalization of the parties, and traditions of collective leadership in the smaller North European democracies, may inhibit assessment through individual personalities. Foreign relations appears as an important issue in rather less than half the elections. It is particularly important in the United States -naturally, in a country at the centre of a world system of alliances which has been engaged in two major conflicts within the last thirty years. The United Kingdom (six out of ten) and France (there out of seven) have been in a similar situation. Foreign relations are also important to smaller countries next to a powerful threatening neighbour, such as Finland vis a vis the USSR (eight out of eleven elections), West Germany (five out of seven), Japan (six out of nine), India (three out of seven) and Norway (six out of eight). Iceland (six out of twelve) and Canada (five out of thirteen) have been particularly concerned over relations with the United States. The last leading issue type again cropping up in nearly half the elections is socioeconomic redistribution, at the core of which stand welfare policies. In Australia these were debated in eleven out of thirteen elections, attaining almost similar prominence in the United Kingdom (seven out of ten). They were raised also in more than half the elections held in Italy, India and, rather surprisingly, the United States. Elsewhere they appeared in a significant number of elections, except in Canada (one out of thirteen) France (two out of seven), Sri Lanka (one out of five) and Switzerland (no elections at all). Possession either of a Socialist party or strong two-party competition appears to contribute to the raising of this issue, although the exceptions indicate that the relationship is not strong. All remainings issue types appear in the range between fourteen and forty-five elections (i.e. between about 8 per cent and 20 per cent of the total). In all cases, low general representation stems from their uneven distribution within different countries. All are raised frequently in one or two countries, but do not appear at all elsewhere. Thus civil order has been a recurring concern in France and West Germany (three out of seven elections) and peculiarly -because of fears about communist subversionin almost a third of Australian elections. Constitutional changes have been debated a great deal in Belgium (six out of ten elections), primarily because of the need to conciliate Flemish speakers by bestowing greater autonomy. Defence has been of particular concern in Canada, where it unleashes tensions between francophones and anglophones, and towards the United States; in Denmark and the United Kingdom debate has revolved around the overall level of defence expenditures within a precarious economy. Religion has entered into politics in Italy, Netherlands and Luxembourg through the Christian parties there and in Sri Lanka through the fears of the Buddhist priesthood about its position. Ethnic rivalry has dominated Belgian politics over the last ten years, and has emerged in Britain and Israel through problems of assimilation posted by recent immigration. Regional issues are related to ethnic tension in Belgium, while in Canada the position of French-speakers is symbolised by the power of Quebec. Farming problems are important to Finland, not only because of the relatively greater number of the rural population, but because of the pivotal position of the Agrarian Party. Nationalisation and associated powers of government control have been particularly controversial in the United Kingdom, but also in Sweden, Norway and the Netherlands, while its converse -an emphasis on individual freedom and initiativewas prominent in Denmark even before the dramatic emergence of Glistrop's anti-tax party in 1973. Government regulation has emerged in about a quarter of elections in Finland, the United Kingdom and the Netherlands, mainly in relation to strikes. Since these are an endemic problem in modern economies, we may expect this type of issue to become increasingly prominent in future. Even generally under-represented issue types thus emerge as important for the analysis of elections in at least one country. Issue types vary not only across countries but over time -some recurring more frequently in later elections, some declining in saliency. Table 10 traces the incidence of issues in each election "period" of the postwar years. Such periods have been selected simply to group contiguous elections, without any deeper theoretical justification. They seem short enough to capture significant changes. Government record is such a ubiquitous type of issue that no particular trend can be traced; if anything it appears to have become an even more widely used basis of assessment in recent years than formerly. The same can be said the more strongly of candidates -from a third of elections in the late 1940s to three-quarters since the mid-1970s. Foreign relations appear more sensitive, as one might expect, to the state of international affairs -relatively more prominent in the late 1940s with the Cold war, then in the late 1950s and early 1960s with detente, then from 1565 to 1974 with Vietnam and its aftermath. Socioeconomic redistribution appears in a quarter to a half of the elections in each time-period, attaining most prominence in the early 1950s during the first wave of postwar prosperity; and least prominence more recently. Of the less generally recurring issues, civil order comes to much greater prominence in the politically disturbed periods after 1965, with student demonstrations and terrorism. Constitutional issues come to prominence earlier, from 1955. Defence shows a rising trend to the early 1960s, and a slow decline thereafter. Religious-moral questions remain at a low level throughout, with a slight peak in the late 1950s. Reflecting the growth of separatism and autonomist movements in Europe, and the increasing numbers of foreign workers in all countries, ethnic issues show a distinct rise over the postwar period. Regionalism reached a sudden peak in 1965-74 but an indication of decline is seen in the thirty-three elections out of thirty- six in which it failed to appear in 1975-79. Urban-rural questions, on the other land, show no distinct tendencies, though they reached their highest in 1970-74. The three non-redistributive class issues all show a tendency to grow in prominence over the postwar years. The extent of government control (involving such questions as nationalisation) as much debated during the postwar reconstruction, slipped almost out of sight in 1960-64, then returned to prominence in the 1970s with the breakdown of Keynesian approaches to economic management and the advent of neo-capitalist modes of thought. A broadly similar pattern, though different in detail, ca, be seen for the antonym of government control -individual initiative and freedom. Regulation, increasingly discussed in relation to industrial unrest and strikes, again rises sharply to prominence in the 1970s. Most of these tendencies are understandable in terms of a postwar history which in most countries involved initial reconstruction, followed by a relaxed enjoyment of restored prosperity from about 1955 to 1965, succeeded in turn by the increasing problems of a complex economy and intergroup tensions connected with this (foreign affairs proceeding, meanwhile, from a consolidation of Cold War alliances, through detente, to a concern with American involvement in Vietnam, and further dissolution of previous alliances). That the appearance and disappearance of election issues can be related to general tendencies, outside the control of politicians in any one country, carries implications for analyses of campaign strategies. For if politicians have only a limited ability to emphasise or deemphasise issues, which become independently salient in elections, one can attribute only a partial influence over victory or defeat to their actions. Rather than seeing their strategy as determining the outcome, we must ask instead whether they made the most of their appeal under given circumstances which they could not wholly (or perhaps even substantially) affect. The inference that salient election issues relate broadly to significant events in the outside world and are not sham debating topics designed to divert attention form real problems, is consistent with the evidence and is a cheering conclusion for the democrat, if somewhat daunting for ambitious politicians. iii) Future Research It must be said that all the evidence reviewed above is inferential and highly circumstantial. We look at changing patterns of issues (or more exactly, of various categorizations of issues) and interpret them in terms of changing conditions in the world around them. That is quite suggestive but can only be a first cut at the problem of issue generation. Further efforts need to be made in regard to the systematic collection of indicators of political and socioeconomic change (both national and international) in order to relate them through regressions to changes in programmatic content. This should answer more conclusively some of the questions raised by the analysis reported here. 9. THE CONSEQUENCES OF POLITICAL ISSUES Strictly speaking, the question of what effects issues produce is beyond the remit of this discussion, whose major focus is on the way in which one type of issue succeeds another. To leave it at that however is to leave many questions hanging in the air. We should not after all consider issue change important did we not think it carried further consequences for election outcomes, party strategies and government activity. Each of these is considered briefly, in turn: a) Election outcomes There was a great debate in the field of American voting studies about the extent of "issue-voting" compared to voting on the basis of "Party Identification", in the early 1970s (reviewed by Margolis, 1977). This partly derived from the fact that explanations of voting have to cope with two phenomena: i) the large number of voters who vote t he same way from election to election ii) the smaller, but possibly increasing number who vote differently. In many studies i) is associated with voting on the basis of a basic identification with the party uncontaminated with issues and ii) with voting under the influence of issues (cf, among many others, Pomper, Brody, Page and Boyd, 1972). Obviously, however, one could associate both effects with issue voting. From both spatial and non-spatial perspectives, if a party makes only marginal adjustments to policy and voters do not substantially change their opinions, most of its previous supporters will end up voting for it in the next election. Explaining all voting in terms of reactions to issues is neater and more internally consistent than using one type of factor to account for stability and quite another to account for change. All rational choice accounts make issue based assumptions. Budge and Farlie, 1983, made an attempt to develop a thorough-going issue-based account of voting which (unusually) generated, advance predictions of 10 national elections which performed reasonably well. Current predictive "economic" models of elections (Tufte, 1975, 1978; Sanders, Ward and Marsh, 1987, 1991) are also, of course, issue-based though emphasising effects of only one type of issue (economic). Thus, there seem strong grounds for assuming that issues have a determining, if not exclusive, influence over election outcomes: and that election change is the effect of issue-change. b) Party Strategies If this is the case then what parties do about issues strongly affects their electoral fortunes and participation in government. The whole tradition of modelling and spatial analysis which relies exclusively on issue spaces and issue effects seems to be on the right lines. Above I have suggested one direction that such analyses could take, in terms of saliency theory. As government record is itself a major issue, what governments do in office will affect their electoral fortunes, and vice-versa. c) Governement Formation Most models of coalition-bargaining have moved ever closer to using pure policy-factors rather than office-seeking ones to explain government formation. Recent research using election programmes to measure party policy positions and government programmes to represent government policy, have given only limited support to the idea that policy-proximity or strategic positioning in issue-space provide a basis for coalitions. Of 10 different policy-based models tested (two indirectly median-legislator ones) the most successful attained only a 50 per cent success rate and 50 per cent efficiency rate over postwar governments in 12 countries. Moreover, party programmes related only inconsistently to government programmes -only in 4 out of 8 countries were there correlations of any magnitude between the two (Laver and Budge 1922, Chapter 14). d) Government Outputs These negative results could be explained, however, by government programmes not being good indicators of policies actually pursued by government. Much stronger relations are discovered between parties' programmatic stands and actual government outputs, as measured by percentage expenditures (Budge and Hofferbert, 1990: Hofferbert and Klingemann, 1990). Parties in coalitions may well affect outputs through their tenure of specific ministries for which they have a strong preference linked to their particular policy concerns (Budge and Keman, 1990: Chapter 4). Specific ministries may thus be an intermediate goal for parties as a means to affecting policy outputs (Budge and Keman, 1990, Chapter 5). This undermines established perceptions of ministries as interchangeable coinage (Browne and Franklin, 1972) in favour of certain ministries having a particular value for a party (Budge and Herman, 1978). Besides saving the idea of a reasonably pure policy space as the venue within which coalition parties operate, the more general findings on the linkage between emphases in election programmes and the expenditures made by Governments, support mandate theories of democracy (cf Downs, 1957) in which electors choose between parties on the basis of their alternative programmes for government; and parties carry through the programmes in office, thus effecting majority or at least plurality preferences. 10. CONCLUSIONS We have, therefore, the promise from this line of research of a totally issue-based empirically-grounded theory of the functioning of party democracy, from party strategies through election outcomes to government functioning. Not only does this provide us with the prospect of a more exact and testable account of the processes involved, it is also in line with philosophical and popular justifications of the greater choice offered by democracy as a political system. It helps to emphasise the importance of studying the emergence of issues along the lines undertaken here, as they constitute the agenda for government action on societal problems at a particular conjuncture. All this underlines the basic importance of better definitions and measures of political issues themselves, of their effects, and of the influences which shape them and change them. It is this last aspect of issues which is currently the least studied and yet potentially the key to the ability of democratic political processes to respond to their environment and keep themselves and their populations going. It is this which makes studies of issues and agenda-setting particularly valuable at the present time, when vast new populations are turning to democratic methods of tackling and resolving the major problems which threaten to overwhelm them. Table1 Table8 Table9 Table10 Figure1 BIBLIOGRAPHY BELL, Daniel: 1962, The End of Ideology. New York, Random. BROWNE, E.; FRANKLIN,M.: 1972, "Coalition Payoffs in European Parliamentary Democracies". American Political Science Review 67, p. 453-64. American Enterprise Institute. Washington, Election Studies, various dates. BUDGE, Ian; FARLIE, D.J.: 1977, A Voting and Party Competition. London, Wiley. 1978, "The Potentiality of Dimensional Analyses". European Journal of Political Research 6, p. 203-31. 1983, Explaining and Predicting Elections: Issue-Effects and Party Strategies in 23 Democracies. London, AIlen and Unwin. BUDGE, Ian; ROBERTSON, D.; HEARL, D.J.: 1987, Ideology, Strategy and Party Change: Spatial. Analyses of Post-War Election Programmes in 19 Democracies. London and Cambridge, Cambridge University Press. BUDGE, Ian; HERMAN, V.: 1978, "Coalitions and Government formation". British Journal of Political Science 8, p. 459-77. BUDGE, Ian; KEMAN, Hans: 1990, Parties and Democracy: Coalition Formation and Government Functioning in 20 Countries, 1950-1988. Oxford, O.U.P. DOWNS, Anthony: 1954, An Economic Theory of Democracy. New York, Harper. HOFFERBERT, R.I.; INGLEHART, R.: 1990, "Post-Materialist Values and Party Programs in Democratic Europe". Paper for Midwest Political Science Convention. Chicago 1990. KIRCHEIMER, Otto: 1966, "The Transformation of the Western European Party System", pp. 177-200 in J. LAPALOMBARA and M. WEINER (eds.): Political Parties and Political Development. Princeton, Princeton University Press. Keesing's Contemporary Archives. London 1932 - Present. KELLEY, S.; MIRER, T.: 1974, "The Simple Act of Voting". American Political Science Review v. 68, p. 572-91. LIPSET, S.M.; ROKKAN, Stein: 1967, Party Systems and Voter Alignments. New York, The Free Press. LAVER, M.J.; BUDGE, Ian (eds.): 1992, Party Policy and Government Coalitions in .10 Democracies. London, Macmillan. Nuffield General Election Studies. Ed. De Butler with various collaborators. London, Macmillan. MARGOLIS, M.: 1977, "From Confusion to Confusion: Issues and the American Voter 1956-72". APSR 67, p. 342-56. POMPER, G.M.; BRODY, R.A.; PAGE, B.I.; BOYD, R.W.: 1972, "Issues and American Voters 1956-1968". APSR 66, p. 450-69. RALLINGS, Colin: 1987, "The Influence of Election Programmes: Britain and Canada 1945-1979" in BUDGE, ROBERTSON, HEARL (eds.) Chapter 1. RIKER, W.H. (ed.): 1993, Agenda Formation. Ann Arbor, Michigan, University of Michigan Press. ROBERTSON, David: 1976, A Theory of Party Competition. London, Wiley. SANDERS, D.; WARD, H.; MARSH, D.: 1987, "Government Popularity and the Falklands Effect". British Journal of Political Science 17, p. 135-161. 1991, "Government Popularity and the Next General Election". Political Quarterly 62, p. 235-260. SARTORI, G.: 1972, "European Political parties: The Case of Polarized Pluralism" in J. LAPALOMBARA, M. WEINER (eds.): Political Parties and Political Development. Princeton, P.U.P., p. 137-76. TUFTE, E.R.: 1975, "Determinants of the Outcomes of Mid Term Congressional Elections". American Political Science Review 69, p. 812-29. 1978, Political Control of the Economy.Princeton, N.J.P.U.P. WHITE, T.H.: (Various Dates), The Making of the President. New York, Atheneum.