The European Parliament Campaign

Transcription

1 FIFTH FRAMEWORK RESEARCH PROGRAMME ( ) Democratic Participation and Political Communication in Systems of Multi-level Governance The European Parliament Campaign Federica Bicchi Robert Schuman Centre for Advanced Studies European University Institute Jean Blondel Robert Schuman Centre for Advanced Studies European University Institute Palle Svensson Department of Political Science University of Aarhus Work in Progress March 2003 Draft text not to be quoted without permission of the author(s).

2 Introductory It is by now widely known that abstention at European Parliament elections is much higher than at national elections, though it is not higher than at local elections in all countries. The only member-states of the European Union in which turnout at European Parliament elections is about the same as at national elections are Belgium and Luxemburg, where voting continues to be compulsory (Table 1). On the other hand, the reasons for the low European Parliament elections turnout are far from clear. Three broad sets of causes have been put forward traditionally in the literature, none of which involves the campaign; yet these classical sets of causes of abstention are far from being empirically satisfactory: other explanatory factors, including the exposure to the campaign, have indeed to be explored. One of the set of classical reasons concentrates on the technical characteristics of the vote, such as whether the vote is compulsory or not, whether it takes place on a weekday or not, whether postal voting is easily granted or not (Franklin et al., 1996, 328-9). These reasons may account to an extent for differences in turnout from one member-state to another, but not for differences in turnout between European and national elections. A second type of argument suggests that European parliament elections are not only less important than national elections but are of a second order (Reif and Schmitt, 1980, 3-44; Reif, 1985). Yet this 'second order' character is far from being proven in so far as it assumes that citizens vote at European elections primarily to influence national politics. The authors of a volume on the 1994 European Parliament election, published in 1998, noted on the

3 2 contrary, on the basis of a systematic empirical analysis of the variables affecting turnout, that... electors do not appear to abstain in European Parliament elections on the basis that they are second-order elections (Blondel, Sinnott and Svensson, 1998, 232). A multivariate analysis further showed that... turnout in European Parliament elections is not simply influenced or not even mainly influenced by national political attitudes... (244-5). The third classical argument has to do with the general feeling that the European Parliament lacks fundamental legitimacy because of a democratic deficit in the European Union (Franklin et al., 1996, 377). This deficit was perhaps manifest in view of the originally very limited powers of the Parliament: with the increase in powers of the Parliament from the mid-1980s onwards, the argument is more difficult to sustain. It appears essentially based on the notion that decisions taken by the Commission and the Council of Ministers suffer from an original sin because these organs are not democratically elected and not democratically accountable. That argument, although widely adopted and often repeated, is difficult to hold. Its basis in theory is not clear, since the governments of the member-states are all democratically elected. The empirical basis is thin, as was pointed out in the volume published in 1998:.. the view which is almost universal among elite circles according to which there is a large democratic deficit in the Union that does not exist in the member states does not appear to be shared by large numbers of European citizens, either because they do not know what the situation really is or because they do not believe that democracy works very well in their own country (ibid., 77).

4 3 Other possible causes for the low turnout had therefore to be discovered: these emerged from the analysis conducted on the 1994 election, which showed that abstention had been largely due to a combination of three sets of factors, lack of interest in politics, lack of knowledge of politics and dislike of European integration (ibid., 240). These three attitudinal variables are interconnected: citizens who lack knowledge and those who dislike the Union tend to be among those who are least interested in politics and in European politics in particular, while, conversely, by and large, a better knowledge is associated with greater support for the Union. These factors pointedly also accounted primarily for voluntary abstention, voluntary abstention being the type which closely measures the difference between European and national elections (ibid., 50-3). Given the part played by these three attitudinal variables, the possible part played by the campaign needs to be closely investigated, since the campaign is the one political activity which is specifically devoted to boosting political knowledge as well as interest. Some evidence in that direction was indeed found in the passive campaign exposure has no direct effect on abstention of any shape or form. What is required is active exposure, that is reading about it or talking about it. Involvement in even one of these activities is associated with a reduction in the odds of voluntary Euro-specific abstention... (ibid., 232). At the 1994 election, however, it was possible to trace only some elements of the influence of campaign on turnout. No analysis of the input of the parties during that campaign could be undertaken nor could the content of messages given through the media be monitored.

5 4 The aim of this Chapter and of the following is partly to fill this gap. The first section of the present Chapter examines the general conditions under which campaigning can be expected to affect turnout. In the second section, the extent to which the citizenry at large is affected is analysed on the basis of survey data, both in the Union as a whole and in the different member-states. The input of the parties in the campaign is then described together with the extent to which these variations were reflected in the press. The following Chapter is devoted specifically to the campaign on television. I How European election campaigning may affect voting Why campaigning is becoming more important in order to understand electoral behaviour, in particular at European Parliament elections. For a very long time, electoral behaviour analysis concentrated its attention on socio-economic characteristics, although the Michigan studies, from the 1960s, emphasised psychological aspects of voting, especially by means of party identification. Such cleavages as class, religious practice, age, gender and education, have thus been regarded as key explanatory factors of voting behaviour (Miller and Shanks, 1996, 212-4). In Europe at least, turnout had rarely drawn the attention of political scientists, but to the limited extent that it did, at least some of these social variables were also regarded as significant (Blondel, Sinnott and Svensson, 1998, 20-22). Towards the end of the twentieth century, however, it became increasingly clear that these socio-economic factors were playing a decreasing part in accounting for voting patterns. Among the reasons

6 5 for such a change is for instance the emergence of a large middle class of citizens who are neither rich nor poor, are less closely connected to one or another of the political parties and are probably also relatively unconcerned with politics in general. The traditional links between citizens and parties have not ceased to exist altogether, admittedly, as only a segment of the voters, even if it is a large one, is genuinely independent : but, alongside traditional (long-term) expressiveness in electoral behaviour, an instrumental (and short-term) approach to voting is becoming widespread. In such a context, election campaigns are likely to play a significant part. This may have been the case in the past, if one judges from historical impressions about the role campaigns had in some countries, Britain for instance, in the middle part of the nineteenth century. However, as the vote for parties became anchored in socio-economic cleavages such as class or religion in many European countries, campaigning seemed to lose some of its relevance even in the eyes of politicians, let alone of observers. The trend away from campaigning came to be reversed somewhat in the last decades of the twentieth century. The leaders of the established parties seemed able to modify, on the basis of a strong popular appeal, the allegiance patterns of many. Mainly in the United States, the part played by individual candidates came to be fully recognised by politicians, observers and even academics (Sorauf, 1988, 504-5): large proportions of voters were seen to be casting their ballots independently from the party with which they claimed to be identified. While individual candidates did not play or have not as yet played - a similar part in Europe, new parties based on hitherto unknown personalities have emerged and made, sometimes very

7 6 rapidly, substantial inroads in the traditional electorate of their more established rivals. In so doing they have contributed to a reduction of the part played by long-term party allegiance based on socio-economic cleavages. In this process, election campaigns have been a key instrument used by these new parties to appeal to electors. In a general context in which expressive voting is in decline, electors seem more likely to have more doubts at the European level than at the national level about casting their ballot. As, in this case, less interest and less knowledge about the contest will tend to tilt the balance towards abstention, the campaign would appear likely to be one of the main ways to counter such a tendency. The three sets of factors which, as we noted, appear to account more specifically for non-voting at European parliament elections, lack of interest, lack of knowledge and distrust of the European Union, are clear indications of the fact that citizens are very unclear about the significance of European Parliament elections. It is not surprising that this should be the case. The amount of information which percolates down to the average elector about what goes on in Brussels is so minimal that it borders on being almost non-existent. Electors have to make a major effort to find out what are the activities of the European Union, except in a few welladvertised cases. Such an effort is well beyond what the average citizen, typically already not very interested in politics, is likely to be willing to make. Indeed, it is surprising so surprising that the veracity of the answers seems questionable that a substantial proportion of respondents should state, as they regularly do in reply to Eurobarometer questions, that they rely on the Commission, the Parliament, the Council of Ministers or the Court of Justice. It is hard to believe that many respondents are in a position to assess what these institutions do: the judgements which are passed are

8 7 therefore likely to be given on spec and not on the basis of concrete examples which respondents have in mind. So few such examples are indeed at their disposal. This is why one might expect the election campaign to be potentially significant: this seems indeed to follow from the fact that interest in and knowledge of European politics are statistically associated (Blondel, Sinnott and Svensson, 1998, ). The campaign can instil some knowledge; that knowledge is likely to stimulate interest. Both the knowledge acquired and the interest generated are in turn likely to alter views about the Union: what are literally prejudices, given the limited knowledge of the electors concerned, may therefore be reduced. The key occasion when these effects seem likely to occur is during and through the election campaign. One must therefore examine whether and, if so, to what extent the election campaign does affect the propensity of electors to vote at European Parliament elections. Why the study of the impact of campaigning at the European Parliament elections is likely to be revealing. Campaigning can thus be expected to play a part in increasing turnout in European Parliament elections (and possibly, more and more, in national elections as well for the reasons which were outlined earlier). Moreover, the study of the effect of campaigning at European election level is also likely to be revealing because of the intrinsically comparative nature of any study of the European electorate at large. At the national level, campaigning, with respect to the main European political parties at least, has become highly centralised and consequently rather uniform. Admittedly, as was noted earlier, new parties have tended to emerge and these have somewhat modified

9 8 the panorama of political contests: but even these parties tend to be centralised. In contrast to the United States, personalised campaigning at the level of candidates is rare, indeed almost nonexistent in Europe at the national level. Methods of campaigning may have changed somewhat, but these changes have tended to affect almost to the same extent at any rate all the main competing organisations. It is therefore difficult to measure precisely the effect of campaigning either on turnout or on the voting decision in the context of national elections. Admittedly, at national elections, while all or nearly all look at television, not all look at television to the same extent; nor do they all look at the same television channels. Moreover, not all citizens read the same newspaper; nor do they all read political news in the newspaper which they read. Perhaps above all, there are major variations in the extent to which citizens discuss politics with friends during election campaigns. Yet a common set of rules and of practices characterises national elections: the only significant exceptions occur in countries in which regional differences are most marked, as above all in Belgium, and also in parts of the United Kingdom, of Spain and, to an extent, of Italy. At the level of European Parliament elections, on the contrary, there is no uniformity in the character of the campaign. National practices obtain: these automatically result in major differences between countries. Forms of campaigning vary: canvassing, for instance, does not take place everywhere across the Union. Rules about party finance differ and affect campaigning. At the level of the media, too, rules and practices vary: each country treats the European election in the same way as it treats national elections. Thus the manner in which the campaign is reported in news bulletins

10 9 or in other types of programmes (such as debates among political representatives) and in party political broadcasts tends to vary from country to country. The basis for the analysis of the effect of campaigning at European elections is thus much richer than it is at the national election level, as if experiments were made in the way they are occasionally made in the United States, typically at the local level - to discover the impact of candidates when different techniques are used in different parts of the districts which they contest. Admittedly, the study of campaigning reported here has had to be, mainly for financial reasons, rather limited in scope. It was not possible to monitor the activities of the parties during the election itself: that inquiry had to be undertaken later. Party officials were asked to recall what had been done and what had been the impact; not surprisingly, there may be variations in the degree of accuracy of such a recall. Nor was it possible to monitor the content of all the campaign reporting made by all television stations across Europe: only the news broadcasts, and indeed for a limited period, were covered and only some channels were studied. Yet these analyses, in conjunction with the results of the surveys of the electorate undertaken a few months after the election, give a picture of the relationship between the three key types of actors which need to be examined if the impact of the campaign is to be assessed, the providers, which are the parties, the intermediaries, which are the media, and the customers, who are the electors. The characteristics of the campaign: facilitation and mobilisation on both a personal and a contextual basis. Emphasis has been placed throughout this volume on the major distinctions to be made, first, between mechanisms which facilitate the vote and mechanisms which

11 10 mobilise the electors and, second, between the personal and the contextual bases of turnout and abstention. These distinctions obtain in the context of the campaign as well: indeed, as interest in and knowledge of European politics are low, facilitation and mobilisation during the campaign can be expected to play a significant part in European Parliament elections. In the context of the act of voting itself, facilitation refers to the sets of mechanisms which may be introduced to render this act easier to accomplish: rules about postal voting are an obvious example, as is the extension of the hours of polling. On the other hand, also with respect to voting, mobilisation refers to the ways in which, by means of various forms of party propaganda, electors are induced to register their vote for one of these parties. In the context of the election campaign, what facilitation and mobilisation consist of is rather different. Facilitation refers to the processes by which electors come to be in a position to be better informed on what the election is about: it lifts part of the veil of ignorance within which most electors find themselves with respect to European Union matters. Meanwhile, mobilisation, in the context of the election campaign, does not aim at inducing electors to make a physical move or to fill in and post a postal ballot: it consists in inducing electors to become motivated to understand better what the election is about, to be more prepared to listen to or read what the media say about it and to be more inclined to talk to family or friends about the election. These acts are as positive as the act of voting itself; but they are more intellectual and their action covers a longer period.

12 11 We noted earlier that during the election campaign three sets of actors are involved, the parties, the media and the electors. In principle, facilitation seems primarily achieved by the media and mobilisation seems the preserve of the parties, the media being regarded as intermediaries, perhaps even, somewhat oversimplistically, as neutral intermediaries, while the parties are the providers : they are the bodies from which the messages are expected to come. This division of labour is not as clear-cut in practice. Some media at least are closely tied to the parties. This was and still is to an extent the case with the press; this is even to a limited extent the case with radio, though not, at any rate in Western Europe, ostensibly at least, with television. Mobilisation is thus achieved, in some cases, by the media, and especially by part of the press, acting under the supervision of the parties, admittedly. Moreover, those who work for the media are not all wholly satisfied to be mere facilitators: they want to present points. This means, to a lesser or a greater extent, admittedly, but to some extent inevitably, a degree of mobilisation. While it may be the case, by and large, that the media (except part of the press) facilitate while the parties mobilise, there is some overlap and perhaps a combination of both aspects even where there is not deliberate collusion. These matters are somewhat complicated by the other distinction in need of examination, that between the personal and contextual aspects of the voting process. The contextual aspects are manifest: they are constituted by the institutional structures of both parties and media. Personal elements are also present, however: in the case of the campaign, these personal characteristics do not apply to voting itself, but, as with facilitation and mobilisation, they

13 12 relate to the ability or the desire to display interest in the events which are taking place and to obtain information about these events. Such personal characteristics as education, age, gender, occupation, and religious appartenance may be expected to affect at least to an extent the propensity of citizens to be better informed and to seek the relevant information. The analysis of the activities of the political parties and of the media in the context of election campaigns thus gives an opportunity to determine in greater detail how far these providers and these intermediaries facilitate and/or mobilise electors. Prima facie, providers and intermediaries seem to do so but to achieve these objectives to a limited extent only in the context of European Parliament elections, since abstention is high and interest in and knowledge of European affairs are low. What has to be determined is whether and, if so, to what extent this is because electors are very little affected by or even been made aware of the election campaign. Such an exploration constitutes a further area in which comparisons across countries are likely to be revealing, since there will inevitably be cross-national variations in the character of the campaign. As European elections take place simultaneously in a number of countries, differences in interest and knowledge displayed during the campaign can be monitored and accounted for, thus making it possible to assess how deficiencies in this respect could at least partly be overcome and improvements could take place. Limitations in the analysis: data and methodology. Yet the effect which the campaign, in its various forms, may have on turnout can be determined with precision only if, first, the level of interest and

14 13 the level of knowledge of electors is measured both before and at the end of the European election campaign: this means interviewing a panel of electors at least at two points in time or having a two-wave survey. Such an aim not be achieved with respect to the 1999 European parliament election. Moreover, as noted earlier, the survey which took place in 1999 was administered only several months after the election itself: errors in recall could be expected to be substantial while no information was obtained as to whether the respondents interest and knowledge changed during the period of the campaign. Three main conclusions can be drawn from the information obtained, however. First, as turnout at the 1994 and 1999 election can be compared, one can see how far there were variations in the effect of the campaign as recorded by the respondents in these two occasions. If the trend is broadly similar, it seems permissible to conclude that the campaign did indeed have in 1999 the effect which respondents recalled that it had, while, if differences in the reactions of respondents in 1994 and in 1999 emerge, conclusions about the effect of the campaign have to be more tentative. Second, one can compare interest and knowledge among those who stated that they had followed the campaign and among those who did not and indeed among respondents who were exposed to the campaign in a number of ways. Finally, cross-national variations provide a major basis for comparisons. The coming section of the Chapter is thus devoted to the analysis of the role of the campaign as recorded by the respondents. The third section then examines the activities of parties as providers during the campaign. Admittedly, as stated earlier, party activity was monitored only on the basis of a recall and of a recall

15 14 which occurred over a year after the 1999 European election. Officials in all the main parties across the Union were interviewed about the nature of the campaign conducted by their party; they were also asked to reflect on what seemed to them to have been the impact of the campaign. The survey of party officials thus provides a means of obtaining a broad picture of the activities of the parties and therefore of the extent of seriousness in which the European Parliament election was held in the member-states of the Union. The analysis of the impact of television takes place for the most part in the following Chapter: it is based on the content analysis of a large sample of the news programmes which went on the air in the last two weeks of the campaign. Since the respondents exposure to the campaign is examined generally in this Chapter, findings on television news reporting described in the next will be related to the extent to which they contributed to modifying and almost certainly increasing - the level of interest and of knowledge of electors. II

16 15 Citizens and the European Parliament election campaign Forms, state and development of campaign exposure. People become aware of an election campaign in various ways; they learn about candidates and parties running for office through various channels. Some may be very interested and seek information by reading about the election in the press; others may become aware of the election by seeing posters on walls. Some may actively take up issues and discuss the election with family and friends; others may come across the campaign accidentally when they open their television set to see the news generally or merely to be entertained. In the 1994 and 1999 Eurobarometer surveys campaign exposure was measured by means of the same six campaign modes; three additional modes were added in the 1999 survey, but these cannot be taken into account to compare the two election. Moreover, as the entry of three new member states occurred in 1995 only, over time comparisons can be made among the 12 old member-states only. Table 1 gives an overview of the channels of campaign exposure in 1994 and 1999 (1). TABLE 1 ABOUT HERE Remarkably, there was an increase in campaign exposure from 1994 to 1999 in the old member-states: this contrasts with the decline in turnout which occurred between 1994 and All forms of communication did show increases, but these were largest in connection with the extent to which party workers call on private homes (83 per cent) and smallest with respect to radio and television (15 per cent). It does not follow that the campaign improved in 1999;

17 16 indeed, overreporting of campaign exposure may have taken place in 1999, since the interviews occurred several months after the event. Over time patterns are fairly consistent in the old memberstates. Television and radio were by far the most frequent agents of campaign exposure at both elections, followed by three other channels, newspaper coverage, election leaflets, and advertising. Discussions with family, friends and colleagues affected fewer citizens, while canvassing was the channel which affected the smallest number. Campaign exposure was in most cases higher in the new memberstates than in the old member-states, except for canvassing, which is mainly an Anglo-Saxon phenomenon, and contact by phone (one of the campaign modes included in the 1999 survey only), which was most widespread in Greece and Ireland. Yet higher campaign exposure in the new member-states was not associated with a higher turnout than in the old member states: there may have been more overreporting of campaign exposure in the new member-states, perhaps in order not to display a lack of interest in what was in these countries a novel form of activity. There is apparently no direct or simple relationship between exposure to the campaign and the turnout at the election. People may become aware of the campaign through two or more modes of communication, but the number of these modes may be less important than the kind of campaign exposure: modes which require a more active involvement may affect turnout more than those which require merely passive following.

18 17 A combined measure was therefore constructed to determine what can be referred to as the quality of campaign exposure. While reading about the election in the newspapers and discussing the election with family, friends and colleagues entail a more active involvement, viewing television and listening to radio, receiving leaflets, reading advertisements and being contacted by party workers are more passive instruments (Blondel, Sinnott and Svensson, 1998, 145f). Respondents were divided into four groups, those who did not experience the campaign through any of the six modes ( none ), those whose exposure to the campaign was in passive modes only ( passive ), those who experienced the campaign through involvement in one of the two active modes ( partly active ) and those who experienced the campaign through both active modes ( fully active ). TABLE 2 ABOUT HERE As Table 2 shows, campaign exposure was higher in 1999 than in 1994, both in all member-states (the fully active rose from 16 to 33 percent and the partly active from 37 to 40 percent) and in the old member-states (the fully active rose from 16 to 31 percent and the partly active from 37 to 39 percent); campaign exposure was also higher in the new member-states than in the old memberstates. This suggests that even if the intensity of campaign exposure is taken into account, no direct link exists between campaign and turnout at the aggregate level. To obtain a better grasp of the relationship between campaign and turnout we need therefore to turn to an individual level analysis. Personal characteristics and campaign exposure. The ability or desire to display interest in elections and to obtain information about the parties and candidates in the campaign may be affected by personal

19 18 characteristics of the citizens. Factors such as gender, age, education or occupation may play a part, as we noted (Table 3). TABLE 3 ABOUT HERE Men are indeed slightly more exposed to the campaign than women, a characteristic which, perhaps somewhat surprisingly, became larger through the 1990s (gamma.07 in 1994 and.11 in 1999). Age affects little, on the other hand: citizens scarcely became increasingly aware of European Parliament elections as they experienced previous campaigns (gamma.03 in 1994 and.02 in 1999). The tendency for the middle aged (40-49 years old) to be more involved is not strong enough to refer to a curvilinear relationship, as occurs with turnout and age groups. On the other hand, a strong relationship exists between levels of education and campaign exposure: the better educated citizens are, the more they are actively exposed to the campaign (gamma.14 in 1994 and.13 in 1999). Finally, a relationship also exists between occupation and campaign exposure, but it is not across the board. There was no significant relationship (at the.01 level) in 1994 (gamma.00), but a significant relationship was found in 1999 (gamma.10); farmers and the urban self-employed tend to be more involved, as well the upper white collar citizens and students. Whereas, in the case of this last group, the relationship may be a by-product of education, in the case of the other groups, the relationship may be more instrumental and be due to the Common Agricultural Policy and to the Internal Market. These relatively weak relationships with socio-demographic characteristics and the possible links with knowledge and interest suggest that individual mobilisation may be more important than individual facilitation with respect to campaign exposure: the data

20 19 are not such that these points can be fully established, however. Questions about political interest in general and in particular interest in European matters as well about party attachment were asked in 1994 but not in Yet, as involvement in political discussions (for 1999) may be taken as an indicator of political interest and knowledge, attitudes to European unification and satisfaction with democracy in the EU can be explored in a comparable manner (Table 4). TABLE 4 ABOUT HERE Campaign exposure is closely related to political interest, both in political matters in general and in European matters in particular (gamma.31 and.28 respectively). Similarly, those who are frequently engaged in political discussions are more actively involved in the campaign than those who never discuss politics or are unsure as to whether they do (gamma.36). The relationship is equally strong between campaign exposure and political knowledge about European affairs (2): the better informed people are, the more they are actively involved in the campaign (gamma.28 in 1994 and.31 in 1999). Whereas it seems reasonable to treat political interest and discussion, which are general political orientations, as the cause and campaign exposure as the effect, what is cause and effect with respect to European knowledge is less clear. The level of information is measured specifically in relation to European matters and it seems more sensible to view the relationship between knowledge and campaign exposure as being of a reciprocal character. Attitudes to European integration are also related to campaign exposure, but less strongly, in particular in 1994 (gamma.18 in 1994 and.25 in 1999). The more positive people are towards European

21 20 integration and the more they find membership of the European Community a good thing, the more they are actively exposed to the campaign before the election of the European parliament. The strength of this relationship is at about the same level as the strength of the relationship between party attachment and campaign exposure (gamma.17) and it is stronger than the relationship with satisfaction with the working of democracy in the European Union (gamma.07 in 1994 and.12 in 1999). Education is related to the extent to which people are exposed to the campaign, as we saw; since we just noted that exposure to the campaign is also related to political interest, to knowledge and to some extent to attitudes towards European integration, these relationships might be merely or mainly an effect of the education of the respondents. A control for education, however, shows that those who are more interested in politics, who discuss politics frequently, who are well informed about European matters and who are positively inclined towards the EU are also more actively involved in the European Parliament campaign at all levels of education: this means that individual political mobilisation is important for the involvement in the campaign. In sum, citizens were found to be more actively involved in the campaign in 1999 than in 1994, both in the old and in the new member-states. This involvement was also, in both 1994 and 1999, more widespread among the well educated, as well as among those who are politically interested and knowledgeable and but less strongly among those who are positive towards European unification. We need now to consider turnout to see whether and to what extent it is associated with campaign exposure.

22 21 Campaign exposure and turnout. The 1994 study showed that campaign exposure and turnout were positively related: the more actively citizens were involved in the campaign, the more they voted (Blondel, Sinnott and Svensson, 1998, 160). A similar relationship was also found in 1999, indeed more strongly (Table 5). TABLE 5 ABOUT HERE Whereas turnout increased 21 points from the lowest to the highest level of exposure in 1994 (3) it rose 43 points in 1999, with the strongest relationship at the lowest level of turnout. In 1994, 83 percent of the citizens who were fully active in the campaign did vote: in 1999, only 76 percent did so, but this relatively small decline is in part due to the new member-states, since, in the old member-states, 79 percent of the fully active did vote in 1999 (as against 68 percent only in the new member-states). The campaign was, however, as successful in the new member-states as in the old member-states in increasing turnout. The increase in turnout between those who were least involved in the campaign and those who were most involved was nearly 50 points (from 19 to 68 percent) in the new member-states and over 40 points (from 35 to 77 percent) in the old member-states. As far as the citizens in the new memberstates are concerned, one can only surmise that more of them abstained because they were less convinced than citizens in the old member-states at the same level of exposure of the importance of European Parliament elections. Admittedly, it was noted in the 1994 study that the relationship between campaign exposure and turnout could be spurious as both turnout and campaign exposure could be a function of interest in politics; it was shown that the campaign was primarily effective

23 22 in inducing to vote those who were least interested in politics but had relatively little effect among those who had some interest in politics (Blondel, Sinnott, Svensson, 1998, 160). This point cannot be ascertained in the context of the 1999 election: although political discussion is undoubtedly an indication of political interest, it is also a means of political communication. Since discussing the campaign is one of the indicators of active involvement in the campaign, that indicator cannot be used as evidence of political interest in this context. On the other hand, one can explore the extent to which politically relevant knowledge has an effect on turnout and the extent to which the campaign has an independent effect on political knowledge (Table 6). TABLE 6 ABOUT HERE Both in 1994 and 1999 the campaign had an impact on turnout at all levels of political knowledge. That impact was strongest among those with little or no knowledge; it was also stronger in 1999 than in Remarkably, the turnout is as high among those who are fully active without political knowledge as among those who are fully active and well informed: turnout is therefore likely to increase when campaign activities increase. The campaign also has an impact also among the well-informed, thus showing that the relationship between knowledge and campaign is reciprocal. Knowledge is acquired as one becomes involved in the campaign, while some knowledge of European matters facilitates taking part in the campaign. This interaction stimulates turnout, the highest level of turnout being found among those who are both very active and well-informed, since among them between eight and nine out of ten do vote. That relationship weakened between 1994 and 1999,

24 23 however, a point which provides a partial explanation for the paradox of a combination of higher campaign exposure and lower turnout at that election: if citizens were not really as actively involved in the campaign in 1999 as they claim, high knowledge and high involvement, even combined, could not lift the turnout as much in 1999 as in Individual characteristics such as political knowledge and exposure to the campaign thus have an impact on turnout at European Parliament elections. The campaign s strongest impact is among those who are least politically engaged, but in a political context in which there is no prior experience of the European Parliament, as in the new member-states, the starting point at which the campaign has an impact on turnout is lower. In general, the political context may be crucial for the impact which the campaign may have on turnout: it seems therefore appropriate at this point to take a closer look at the impact on turnout of the campaign among the various countries of the European Union. Cross-national variations in citizens reaction to the campaign. Very large differences were found to exist among member-states in terms of campaign exposure in 1994: 80 percent of the respondents stated that they were actively involved in Denmark, but only 24 percent in Spain. In 1999, Swedish respondents were the most actively involved and the respondents of the other two new member-states followed, together with Ireland. Overall, the ranking of the countries was rather similar at the two elections, although Britain moved from occupying a middle position to one near the bottom and respondents in Portugal, Belgium, France and Italy were somewhat more involved in the campaign in 1999 than in Overall, citizens tend therefore to be more involved in European Parliament campaigns in Ireland, Greece or

25 24 Luxembourg than in Spain or Britain. Whether this has to do with the extent to which political parties conducted an active campaign or not needs still to be documented (Table 7). TABLE 7 ABOUT HERE Overall, throughout the European Union, as we saw, the more actively potential voters are exposed to the campaign, the more they tend to turn out to vote and this higher turnout is not merely the result of a greater knowledge about European affairs: it is linked to an independent mobilisation effect of the campaign. One may therefore hypothesise that differences in turnout among EU membership countries may at least partially be explained by variations in campaign exposure: the higher the exposure to the campaign, the higher the expected turnout. Yet, although a strongly positive relationship between campaign exposure and turnout is thus to be predicted in all EU membership countries, the impact is not the same everywhere. It seems prima facie that the relationship between campaign and turnout is, for instance, weaker in Spain and Britain where citizens are less involved in campaign activities than member-states - than in Portugal, Belgium, France and Italy where the effect of the campaign increased more than in other member-states between 1994 and 1999 (Table 8). TABLE 8 ABOUT HERE There are indeed substantial variations from country to country in the extent to which campaign exposure is related to turnout: it is more than double in Ireland, Sweden, Britain, Germany, Finland, Austria and Denmark than what in Belgium and Luxembourg; generally, the relationship is strongest in northern member-states (except

26 25 Germany) and in new member-states. If figures in Tables 9 and 10 are compared, as is shown in Figure 1, a continuous relationship between campaign exposure and the strength of campaign impact on voting is found to exist for most EU membership countries. The exceptions are Spain and Britain, where campaign exposure is lower than what should be expected from the strength of campaign impact on voting, and Greece, Belgium and Luxembourg, where the apparent effect of the campaign on voting is lower than would be expected on the basis of campaign exposure. These exceptions can be accounted for to an extent, however. The lower impact of the campaign in Greece, Belgium and Luxembourg relates to the fact that these countries have or recently had compulsory voting: these are therefore countries in which the campaign is bound to have less impact on turnout, as turnout is high for other reasons. Spain and Britain are opposite cases: a more intensive campaign would seem likely to raise the level of turnout to a similar level of that of other member-states. FIGURE 1 ABOUT HERE Thus the campaign had an important effect on turnout in both 1994 and While campaign involvement was larger among the well educated as well as among the politically interested and the knowledgeable, the strongest impact on turnout was among those otherwise least mobilised. Increased turnout resulting from campaign exposure is not merely the result of higher knowledge about European affairs: the effect is sui generis. A more intensive campaign can thus be expected to raise turnout in most EU member-states, but to do so particularly in countries such as Spain and Britain, where the level of campaign exposure has been stable but low, and not where, as in Greece, Belgium or Luxembourg, high turnout results from particular procedural arrangements.

27 26 III The European Parliament election campaign of the political parties If campaign exposure affects the propensity to vote, what determines campaign exposure becomes in turn important to assess. Two factors seem to play a critical part in this respect, the content of the campaign material and the commitment of the parties to spread that material. This meant looking at the party manifestoes, on the one hand, and, on the other, at the way the parties conducted their campaign. The analysis of manifestoes did not raise any particular problems; not so the assessment of the role of the parties, as a retrospective analyses entailed asking officials of each party to describe the involvement of their organisation. Given the danger of exaggeration or of fading memories, a control mechanism was required: this could be provided by the way the press, in each country, judged the quality of the campaign. This section of the Chapter thus examines successively the content of the manifestoes, the assessment by party officials on the way their party conducted the campaign and the report of the press as to what the campaign had been. Party manifestoes at the 1999 European Parliament election. Transnational parties and the manifestoes. The transnational European parties endeavoured to draft manifestoes for the 1999 election. This was the case in particular of, the Party of European Socialists (PES), the European People s Party (EPP), the European Liberal, Democratic and Reformist party (ELDR), the Greens and the European Free Alliance, all of which published European programmes spelling out common principles and policies for the EU. The programme of the Socialists was drafted with an eye to a straightforward

28 27 application to the national campaigns, in order to avoid the contradictions which had marred a similar exercise in 1994: at the time, French and British socialists had signed declarations which could not be defended at the national level. For the 1999 election, the European Socialist party produced the most carefully prepared document of their history and declared that their campaign would be entirely based on it. In spite of these efforts, large differences, often on a national basis, remained, both among the Socialists and even more among the other parties. Several of these stated only that they did 'associate' themselves to European programmes instead of directly adopting them. Very few parties campaigned on the manifesto prepared by the transnational parties, although several referred to it. Exceptions in this respect were the Greek PASOK and the Dutch Labour party, which used the European Socialist manifesto while nuancing their position in the campaign itself. Party manifestoes for the 1999 European election tended therefore to hint at transnational programmes and at times to vary in length accordingly. Short national manifestoes were meant to be summaries and the national translation of European programmes (4). This was the case mainly for Socialist parties, but also for instance for the Dutch Liberal Party or the Flemish Christian Democrats. Thus transnational manifestoes neither replaced nor profoundly modified national party manifestoes. The varied size of manifestoes. The length of these manifestoes varied markedly (Table 5b). Those produced by the Finnish Christian League (300 words) or the Italian Popular Party (575 words) were short leaflets; others had the size of books, and not always small ones, as in Spain those of the Popular Party ( words) or of the United

29 28 Left party ( words) (5). Yet, in general, manifestoes tended to be short. There were some country differences in this respect: the documents tended to be short in Finland and long in Belgium, Italy and Spain. To be for or against Europe does not seem to have a consistent effect on length: in Austria, manifestoes were longer in the case of the more European-oriented parties than in the case of the 'Eurosceptic' Austrian Freedom Party which issued a 1600 word manifesto; but the opposite wss true in the Netherlands, where the more critical parties, such as the Socialist Party, the Calvinist Coalition or the Green Left, had the longest party programmes. The seven broad fields of manifestoes. Content was analysed on the basis of the grid elaborated by the Party Manifesto Research Project (Budge et al., 1987), with some items being added to cover matters which were specific to the European Union and/or were of significance at the time of the election, such as the Euro or Kosovo. Seven broad issue-areas were distinguished (see Appendix). These issue-areas were European questions, such as the structure of the European Union, its philosophy, its membership and the characteristics of European elections, foreign affairs and defence, including relations with NATO and with the Balkans, the government of the European Union, covering cases of corruption, the reform of the Commission and subsidiarity, economy and finance, the environment, also being concerned with food protection, social policy and education and culture. There were also substantial variations on a national basis in the space devoted to these issue areas. In Austria, the topic of Europe was central, the manifestoes having devoted 16 percent more space to that topic than the European average. Enlargement, structure and philosophy of the EU were the key issues developed in these

30 29 manifestoes. Other countries in which the parties also gave a high attention to Europe as such were Denmark (which also scored 16 percent above EU average) and the Netherlands (which scored 8 percent above that average). On the other hand, in Belgium, Greece, Ireland, Italy and Sweden, the space devoted to Europe was relatively small and between 9 and 5 percent under the European average. If space devoted to Europe is regarded as an indicator of the presence of a sharp pro- and anti-europe cleavage, it would follow that such a cleavage is absent in Belgium, Greece, Ireland, Italy and Sweden, while in Austria, Denmark and the Netherlands that cleavage apparently structured part of the debate, at least in the manifestoes (6). There was less variation with respect to the government of the European Union ; these related mainly to rather limited questions about the future of European institutions. Economy and finance was the issue to which most space was devoted on average in the party manifestoes. There were country differences, however. Britain and Greece, both in a difficult position with respect to the Euro, were the two countries where these questions were given the most space. Meanwhile, in Germany, Belgium and Sweden, the central issues revolved around social policy, especially with respect to employment and the control of population flows. In Sweden, immigration, asylum seekers and drugs trafficking received an extensive coverage, because of Schengen and Europol cooperation matters. In France, the various issue-areas were treated rather evenly, except for the Communist party which devoted an above average space to the environment. These differences in the coverage of manifestoes demonstrate that debates continue to differ markedly from one memberstate to another. European politics tends therefore naturally to be