Sie sind auf Seite 1von 17
Europe and the Ballot: Voting Behaviour in Referendums on European Integration SARA BINZER HOBOLT Centre of Intemational Studies, University of Cambridge St, John’s College, Cambridge CB2 1TP, UK sbh28@cam.ac Draft paper. Comments appreciated. Please do not cite without the author's permission Paper prepared for presentation to the panel on “Referendums and Knowledge 4-10) at the 2” Conference of the European Consortium for Political Research (ECPR), jmtegration, Hence, rather than assuming that “one-model-fits-all", this paper suggests that referendum behaviour varies according to the level of EU salience. This paper addresses several interrelated questions on voting behaviour in EU referendums: How salient is the EU in different electorates? How stable are European altitudes and how does attitude stability vary between segments of the population, over time and between countries? And finail ; how do attitudes impact on voting decisions in EU referendums? This paper begins by outlining the existing theories of voting behaviour in referendums on European integration. On the basis of these explanations, a ‘model concerning the impact of attitudes oF political predispositions ~ on voting behaviour is developed. This model suggests that attitudes towards European integration can become stable political predispositions, which act as a filter on elite cues in a referendum campaign. Hence, as EU attitudes become stronger and mote stable, they play a more significant role determining voting behaviour, and ‘second-order election’ factors become less important. In order to test these hypotheses, this paper examines salience levels in Denmark and Ireland, by measuring the media coverage afforded to the EU. Thereafter, panel studies are used to measure the stability of European attitudes by ‘analysing the level response stability and the impact of political knowledge. Finally, this Paper tests the impact of attitudes on referendum behaviour relative to other factors, such as elite cues, in Denmark and freland. Surveys conducted after three Danish and two {ish referendums are used in order to analyse the variations in referendum behaviour, ‘Theories of referendum behaviour The scholarly debate on voting behaviour in referendums on European integration has predominantly been a battle between three competing “schools”: the ‘attitude’ school, the ‘second-order election” school and the ‘utilitarian expectations’ school The first schoo! focuses on individuals’ values and beliefs, and argues that voting behaviour in EU referendums reflects peopte’s underlying broad attitudes towards Europe. This attitude-epproach therefore suggests that it is primarily voters" general views on the EU that drive voting in referendums. (Siune, Svensson and Tonsgaard 1994a, 1994b. Svensson 1994, 2002). Another competing explanation of voting behaviour in EL! referendums is inspired by the ‘second-order’ theory of elections. The important characteristic of ‘second-order’ elections (local and regional elections also fall into this category) is that 1 ist- 'y are regarded of lesser importance than national elections ( order), and consequently voter tumout is lower, protest-voting and voter-switching are | “mechanisms” that determine whether people (voters) choose to fallow the elites. Lupia and MeCubbins argue that the elites are only successful at persuading the public, if People perceive the information provider (the elite) to be trustworthy and Xnowledgeable, Heuristics such as party aligament, reputation and ideology can convey information to the public about the knowledge and credibility of the information Provider. Equally, Zaller refers to political predispositions, that is stable, individual level traits that regulate the acceptance or non-acceptance of the political communications a person receives, allowing this logie, political predispositions are likely to play a decisive role in opinion formation during a referendum campaign, because “predispositions are the critical intervening variable between the communications people encounter in the mass media, on one side, and their statement of Political preferences on the othes” (Zaller 1992:23), Zaller does not go into much detail about where these political predispositions come from or whether they can vary over Hime or between issues. Yet, it seems plausible that the strength of people's Predispositions will vary across issues and over time depending on the level of issue salience. This paper suggests that as people are exposed to an intense flow of information and debate on European integration over a mumber of years, they become more knowledgeable and aware of European issues and form more stable wropiean autitudes, which influence the reception of elite cues. Stronger predispositions will consequently act 2s an important intervening variable in the opinion formation process in referendum campaign, Hence, in an electorate, where people generally have strong and Stable political predispositions relating to European integration, cite cues and campaign factors will be met with more resistance than in an electorate with very weak political Dredispositions. This may not necessarily mean that in the former case, people are more likely fo vote no (since the elite cues may be consistent with people's political Predispositions), but it will imply that the elites will find it more difficult to sway the voters during a campaign. This model of opinion formation also incorporates elements from the different explanations of referendum behaviour. As with the second-order election school, it assumes that people will follow cues from national elites and that persuasion is more successful when the information provider (govemment) is popular (wustworthy/knowledgeable). However, coatrary to the second-order election school, 1 do not ume that people necessarily have weak (or non-existing) political Predispositions concerning the BU itself Instead, my model suggests thatthe strength of Salience of the EU The issue of salience has figured prominently in many studies of political behaviour of lites and masses. Moreover, this concept has found its way to studies on EU referendums, and it recently played an important role in the debate between advocates of the attitude-school and advocates of the second-order-clection-school. In a recent article, Palle Svensson argues that his stady of Danish referendums undermine the “Franklin- thesis” (second-order election school), because the Danish survey data show that ‘voters can become more knowledgeable on a topic of jereasing salience, and more resistant 10 appeals that contradict their views’ (Svensson 2002:7: ). Mark Franklin (2002) has subsequently acknowledged that, “if any European electorate would have based their referendum votes on well thought-out preferences rather than on positions taken by their Parties, it would certainly have been the Danes’ (Franklin 2002: 752). Hence, both Svensson and Franklin argue that increased salience of the EU has an impact on the opinion formation process, and they imply that political predispositions become more important as salience increases. However, the only problem with the use of the “salience” argument in these studies is that neither of the two authors ~ nor indeed other researchers in the field - have systematically defined or measured the notion of EU salience or tested its actual impact on opinion formation. This paper represents a first attempt to fill this gap in the literature by adopting a new approach to the measurement of BU salience and applying this to the cases of Ireland and Denmark. ‘The concept of salience can be defined as the importance attached to a certain issue or policy area, and EU salience can thus be defined as the importance attached to the EU by the general public, Many studies have shown that both clites and masses may respond differently to issues depending on their level of salience (Epstein and Segal 2000). Yet the problem encountered in all studies of salience relates to the operationalisation and measurement of the concept, The indicator often employed to measure issue salience in the general public is the “most important problem-question” in public opinion surveys in hich respondemts are asked what the most (and second most) important problem facing the nation is, This is indeed a usefull measure when comparing the relative salience of ifferent domestic policy issues, however, itis problematic when it comes to measuring the “absolute” salience of any issue. Applying this measure to EU salience is particularly awkward, because foreign policy issues generally receive a very low ranking in sesponses to these types of qu stions and moreover, the emphasis on the most important one, the likelihood is that we will find considerably lower EU salience in the former case, a8 tabloid newspapers tend to focus on domestic issues. Hence, similar criteria must be employed in choosing national newspapers across countries in order to avoid serious biases. In this analysis, the four criteria of high circulation, quality (broadsheet), national scope, and EU attitude’ were employed in choosing the newspapers (see also Kriesi et al 1995: Appendix). The criteria concerning circulation and national scope were chosen to target newspapers that reach the maximum mumber of readers across the country, The “quality” criterion may be controversial, but it was adopted to target newspapers, which would report more rather than less EU stories. This criterion is likely to “inflate” the measure of EU salience, but this will not affect the cross-temporal and cross-national comparisons of frequencies. Applying these criteria, the Irish Independent was chosen as the newspaper source in Ireland and Berfingske Tidende as the source in Denmark. Both newspapers are popular daily broadsheets and are very comparable in political outlook, format, quality and style of news coverage The Irish Independent is by far the best selling daily newspaper in ireland with an average daily circulation over the past decade of 158,000 copies.’ Berlingske Tidende has been the largest ot second largest broadsheet in Denmark during the last three decades with an average daily circulation of 148,000 copies during the past decade In the actual analysis of these newspapers, the sampling method* was used, Hence, rather than analysing every single copy of the newspapers during the past 30 years, a weekly sample was drawn from each newspaper in the period 1971 to 2001. In this analysis, only the Wednesday editions of the newspapers have been coded.” In the ‘over 1000 issues of newspapers, all articles relating to the EU* were coded in terms of frequency. Figure | illustrates the result of the frequeney coding, that is, how many average daily items relating to the EU appeared in the two newspapers * As the majority of the newspapers in both Ireland and Denmark are fit pro-European, it was decided toanalyse two pro-European newspapers. The second largest daily quality newspaper in Ireland, the /rish Times, has a daily 04,000 copies (1990-2002 average) “The sampling method is used routinely in survey research, and has aiso been applied 10 newspaper analysis (see Kriesi etal. 1995), "The Wednesday issue was chase asa “representative” sample ofthe daily newspapers. This assumption was tested by coding the Monday, Tuesday, Thursday, Friday and Saturday issues of tbe newspaper in (wo years (1985 and 1995) and comparing with the Wednesday data. The results showed thatthe vartance in ihe frequency coding was less than 58%, and hence the Wednesday issues ean be regarded us 2 representative sample "This included every article about EU institutions, EL policies or individual EU personalities ( rational commissioner or MEP), but excluded articles ox! European countries when no reference was the EU (e lation of o wade ‘8 German election) Se Table 1 EU salience in Denmark and Ireland TS EE Ireland Denmark 1970-74 low lowimedium 1975-79 low low 1980-84 low low/medium 1985-89 fow mediuin/high 1990-94 low high 1995-99 tow high 2000-01 lowimedium igh Note: Low « 1.$ articles per day, Medium 1.5.2.5 articles per day, High > 2.5 anticles per day Source: irish Independent and Bestingske Tidende Table | presents a summary of the development in EU salience during the past three decades, and illustrates that whereas salience has increased very little, ifat all, in Ireland since 1970, it has increased significantly in Denmark. Whilst it would be interesting to ‘examine why EU salience varies between countries and over time, that is not the aim of this paper. Rather this paper has set out to examine the impact of EU salience on voting behaviour in referendums on European integration. In the hypotheses presented above, it ‘was argued that an increase in EU salience creates more stable EU attitudes that in tun have a greater impact on voting behaviour. In the following ‘ction, the stability of BU attitudes in Denmark and Ireland are examined. Stability of European attitudes An increase in EU salience should lead to stronger and more stable attitudes towards European integration. One way of testing the strength and stability of EU attitudes is to measure the level of response (in| tabitity in panel studies. Only around half of the respondents in panel studies normally manage to state the same opinion on successive interviews (Zaller 1992; Converse 2000), In the literature on public opinion, the origins and nature of this response instability have been subject to much debate. In Converse’s famous monograph (1964), he documents an extensive instability in mass public opinion, and explains this by using a ‘black-and-white’ model in which the attitudes of a small seginent of the public are modelled as fixed while the attitudes of the vast majority are modelied as purely random, He concludes that the majority of people hold “non- itudes’ towards pol Many scholars have challenged Converse’s non- attitude thesis attributing instability to the measurement error that stems ftom “fuzzy ee Table 2 Consistency in responses on European integration? Denmark 1972 Denmark 1992 iretand 2001.02 Aittedes towards mambershio a5 7 sees Attitudes towards future develonment ofthe EU 56 a Serer N 708 104 188 e-voded into agreeidsagree answers ** Notre-cdk five options (very much are, agree, neither agree nor sare, disspoe, very mck dssgas) Sources: EF -undersagelsen 1972 DDA 194, Folkeafitemningen om Maastichtaftalon 1992 DDA 1743 & Panel Study of Nice referendums 2001 82002 Table 2 shows significant variations in attitude consistency levels in Denmark over time and between Denmark and Ireland. The lowest levels of attitude stability are found in Ireland. In 2002, only just above half of the re-coded responses were stable over time, and the “real” response stability was even lower (35-38 per cent), In a panel study carried out in Denmark in 1972, attitude stability was slightly higher. As we would expect with the increase in salience of the EU, attitude stability increased over the next two decades and in 1992 more than two-thirds of the respondents gave consistent answers to European integration questions over time, Hence, these limited data broadly Support the proposition that the higher the level of EU salience, the stronger and more stable the attitudes towards European integration become. A comparison with response instability on other attitude questions illustrates that the high degree of response 'nstability in Ireland in 2001 and in Denmark in 1972 is not just a result of measurement error, but aiso a reflection of weak EU attitudes. Response stability on an attitude question on abortion'® (a very salient issue in Ireland) was for example 63 per cent, which is considerably above the 53/56 per cent consistency in the re-coded EU Guestions. Similarly, the response stability on questions concerning the degree of party identification reached 68 per cent in the Danish penel study in 1972, and concrete party identification (which party) was as high as 82 per cent. This suggests that high respor stability also reflects stronger attitudes/more salient iss Question wordings are as follows: Denmark 1972: Denmark should leave the EEC: Should the EEC become a United States of Europe? Denmark 1992: For or against membership of the Common Market: Should the EU become a United States of Europe”; Irland 2002; Irland should protects independence {fom the BU; Ireland should not give any more powers to the EU ory for the abortion question consisted of 4 seale from The answer es 1 to 10, which was recoded into 8 three-caiezory option (1-3, 4-6 ancl 7-10) before the response stability was measured 2 individuals have more consistent attitudes towards the EU, and this may also make them ‘more resistant to elite cues that ate inconsistent with these attitudes. Unfortunately, the Irish panel study did not include enough variables on demographics and knowledge to produce a similar stratification according to levels of political awareness. Table 4 Political awareness and response stability ~ Denmark 1992 (%) Political awareness Response stability Low Medium Hi very inconsistent 7 3 ° somewinat Inconsistent 35 28 mostly consistant “8 56 very consistont 10 15 Source: Folkeafstemningen om Maastrichtaaien 1992 DDA 1743, in sum, the results presented in this section are consistent with our expectations based on the salience data: the Danes have more stable attitudes towards the BU, and the stability and strength of EU attitudes have increased during the last three decades. Curreat Irish EU attitudes, however, are relatively unstable. This would lead us to expect that attitudes have the most prominent impact on voting behaviour in recent Danish referendums, whereas elite cues and government performance are more significant in earlier Danish and all Irish referendums. This proposition will be tested in the following section, Referendum behaviour in Denmark and fretand This section evaluates the model of voting behaviour in EU referendums, In particular, it ‘0 examine the strength of the explanations put forward in the literature concerning, attitude factors, second-order election factors and utilitarian expectation factors. Moreover, it will indirectly test the proposition that the relative explanatory power of these factors depends on an intervening factor, namely the salience level of the EU. model Several post-referendum surveys in Denmark and ireland are used to test on Bi Although both countries have held six referendums uropean integration, it has unfortunately only been possible to examine five of these referendums in this paper, due to the limited availability of adequate survey data. This model of referendum behaviour has a dichotomous dependent variable {yevno vote). Only respondents who actually voted are included in the analysis, since the aim is to explain and predict actual voting behaviour, The key explanatory variables expected that European attitudes become a more significant explanatory variable as the salience of the EU increases (hypothesis 3). Hence, on the basis of the salience and attitude stability data, we would predict that attitudes are more significant in more recent Danish referendums (post-SEA) than in earlier ones, and that attitudes are more significant in Denmark than in Ireland, In tab after the Danish EEC ¢ mode! is tested using public opinion data ftom a survey conducted accession referendum in 1972. The Danish EEC accession was ‘accepted by a large majority (by Danish standards) of 63.3 per cent of the voters, and had a high tumout of 90,3 per cent. The vast majority of political parties were in favour of accession, although the Social Democratic Party was divided on the issue. Only the small patties on the left actually recommended a no-vote, The majority of civil society organisations and the media also advocated a yes-vote. Although the EEC received considerable coverage in the weeks leading up to the referendum, the issue was still relatively new and unfamiliar to the voters, and it is unlikely that voters had established Strong underlying predispositions relating to the BEC, The data on stability of EU attitudes also reveal a relatively high level of response instability in 1972 (table 2) ‘Table $ Predicting the yes-vote in the Danish Accession referendum Bull model implied mo. Aide) OO Sinplfied model ‘Manual workers O21 014 | Manat workers 047 018 0.011 y 125 fw ist Source: Observe pravevalg 1972 DDA 909 Table 6 Predicting the yes-vote in the Danish Maastricht referendum si RSE Si Pro-Eaopean tides (000 | Pro-European anaes is 2.000 Government sisatsfcion 038 007 600 | Government dssasition 0.38 000 ‘Yor for yes-paty 2 000 | Vote ores party 221030900 Sex ae) 036 on | Fucation ara | i 026 038 488 Mediu 02103808 occupation 098 Executive 0010480981 Professions 02 0290335 Manual workers 013 033, 0689 see 007004 0093 | Age 007 008 904% Conse 159 ast 0010 | Constant 138034 0.000 y oo or Mupetterte R Square 086 ass Moret prodision 81.50% si. Source: Folkeafstemningen om Maastrichtafialen 1992 BDA 1743, ‘The Danish Maastricht referendum is widely seen as marking the end of the “permissive consensus’ at least in Denmark, The Danish and European political establishments were taken by surprise when @ small majority of the Danish voters (50.7 per cent) rejected the Treaty on 2 June 1992, The Treaty ratification was supported by the vast majority of the Danish Parliament, and only the parties on the fringes of the political spectrum, th Socialist Peoples’ Party and the Progress Party, together with the very small Christian Democratic party recommended a no-vote. Most civil society organisations were also in favour of a yes-vote. Yet, whilst public support for E membership had in reased during the past two decades (although it was stil nong the lowest in the EU), support for further integration had declined (Mannheim urobarometer Trend File), Table 6 shows that party cues are still an important explanatory variabl the data show that voters who are dissatisfied with Equally government (a centre- right coalition government at the time) are more fikely to vote no, As predicted, pro- European attitudes have also become a significant explanatory vaciable in. this ferendum, People who are generally in favour of further integration are more likely to vote yes. Parti dissatisfaction and attitudes are the only si anship, goverame! ‘Table 7 Predicting the yes-vote in the EURO-referendum Bose Sg a Sig, Pro-Europeam attitudes 2.97 020.000 | Pro-Europsan anaes 287 025 amo pany 059052 0.283. | Vote foe yes party 995 046 038 Secale) 08038 064 Education 0428 High 094 04s 28 Meum oss os) 028s ce 0.380 Executive 088 057 a2 Professions O78 036 a6! sal workers 4430620491 om 000 as Constant oat 060 0.495 | Constant 0x7 N a2 902 Nagettrte R Spare ors ons 2ecorrectprodition $9206 5.90% Source: EURO-afsternningen 2000 DDA 4013, AS in previous referendums, the Euro-referendum was widely supported by the Danish establishment, including tade unions, business organisations and political parties representing more than three-quarters of the seats in parliament. The parties on the left of the political spectrum, the Unity List and the Socialist People’s Party, as well as the far right party, the Danish People's party, recommended a no, and so did the smali centre party, the Christian People's Party. Yet, in spite of this, only 47 per cent of the voters said yes to the new currency ‘Table 7 shows that in the full model, party cues are no longer important,'* whilst European attitudes are the only statistically significant variable. At the time of the referendum, only around 45 per cent of the voters were in favour of @ common currency. and this had fluctuated between 34 and 53 per cent during the previous ten years Public support for a European Political Union or a United States of Europe remained low. Hence, it seems likely that the strong political predispositions against further Political integration in Europe among a majority of the electorate meant the elite cues were not readily accepted by the voters. Consequently, the result was a rejection of the * In spite ofthis, have chosen to inchude them inthe S tevel. The substantive effets of pany cue Mannheim Burabsro fied model where they ave significant atthe government satisfaction, These changes in the explanatory variables ean all be regarded as plausible and table 8 illustrates their predicted impact on the yes-vote.? when all other variables ace set to their mean value, As expected, the changes in partisanship and the govemment satisfaction have the most pronounced impact on the vote in 1972, whereas a shift in attitudes has only a negligible effect. This has changed by 1992, where the EU attitude variable has the most significant impact ~ a % point shift in attitudes in both increases the yes-vote from SO to 67 per cent. Yet as we can see, chang partisanship and government satisfaction could also have secured a comfortable yes In the 2000 Euro-referendum, the impact of changes in partisanship has disappeared, yet, a shift in attitudes is predicted to majority in 1992 according to these simulations. have a quite remarkable impact on the referendum outcome, Hence, these simulations confirm the proposition regarding the increasing impaet of European attitudes and the diminishing effect of elite cues, as salience increases. How do the results from the Danish referendum surveys compare with Irish data? The Irish salience data revealed that the salience of the EU has remained low until very recently (table 1). This is also reflected in the relatively high level of response instability fon EU attitude questions (table 2). Hence, we would expect that in the recent referendums on the Nice Treaty, underlying attitudes would play a limited rote, whereas elite cues, government satisfaction and campaign events would be of higher importance. ‘The Irish voters” rejection of the Nice Treaty in the 2001 referendum came as a shock to the Irish and European political establishments. The Irish public has traditionally been one of the most pro-European people in Europe. Moreover, almost the ‘entire establishment was in favour of the Nice Treaty. In Parliament, only two small parties, Sinn Féin and the Green Party, openly recommended a no-vote, and trade unions, employer organisations and the media were in favour of the Treaty. Yet, the campaign effort by the yes-side was very limited and characterised by a high level of complacency. The Minister of European Affairs, Dick Roche, has later conceded this. ‘There was very jittle political campaigning in the first referendum, J think when you have situation where the government and the political parties representing 96 per cent of the elected representatives say yes, where the main pillars of public opinion ~ the newspapers - say yes, where the trade union movement says yes-ish, where the business community and farming 1 Danish sur les are almost accurate. The actualy 49.3 por cent in 1992 and 45.9 The yes-votes in roferendaumns were 63.3 per cent in 19 in Ireland, which has been well above 60 per cent for the past decade, it seems unlikely that attitudes alone would have led to a rejection of the Treaty. However, it is plausible that a combination of weak elite cues and weak political predispositions would result in a highly unpredictable outcome, as was the case Following this line of argument, it is hardly surprising thet an intensification of elite cues in favour of the Treaty in the second Nice referendum campaign could potentially lead to a significant change in voting behaviour, After the defeat in the first Nice referendum, the Irish government decided to hold a second referendum on the same issue.* But this time the political parties and civil society organisations launched an intensive and well-resourced campaign in favour of the Nice Treaty. The result was an increase in turnout to 49 per cent and a swing of 17 percentage points in favour, which secured a 63 per cent yes-vote. Such a sizeable swing implies that voters had relatively ‘weak political predispositions. It has been argued that the increase in the yes-vote is merely due to the increase in the tumout, Yet, an analysis of the data shows that the increase in the yes-vote was not merely a result of “new” voters, but equally of a swing among voters who tumed out in the first referendum, The Irish panel study of the Dublin region, for example, shows that almost 20 per cent of those respondents who voted in both referendums changed their votes. 15 per cent of the voters changed their voting behaviour from no in Nice 1 to yes in Nice 2 according to this survey, and this supports the suggestion that a substantial amount of people were persuaded to change their mind during the second campaign. If the Irish bad strong underlying attitudes towards Europe, it seems unlikely that the elites would be able to sway so many voters to vote differently on the same question. Comparing with a similar scenario in Denmark, the rejection of the first Danish Maastricht referendum can be attributed to an underlying resistance towards more political union (illustrated in the survey data) that made it hard for the elites to sway the voters during the campaign. In the first Irish Nice referendum, however - conducted in a low-salience and low-knowledge context ~ an effective no- have caused the pai rather than attitudes, were the most important determinants of the Irish no, whereas th campaign coupled with complacency from the yes-side are likely « final rejection of the ballot question. It could therefore be argued that factors, " Although the ballot question was extended fo include two further subsections in addition to the ticle 29.4.7 fy the Nice Treaty. The additional proposals related to enfumncad ¢ ‘(Aticle 29.4.8) and a constitutional ion on Ireland jinn the s-cal clause! acceptance of amend arate vote on cach simulations are presented. As in the analysis of Danish referendums, the predicted impact of changes in partisanship, government satisfaction and attitudes are cafculated Unfortunately, the distribution of yes and no-voters in the Irish survey sample is not as accurate as in the Danish, and this may affeet the validity of the a alyses presented below. In the Irish survey sampies the yes-vote is 41 per cent in 2001 and 74 per cent in 2002 ‘Table 11 Simulated impact of selected variables on the vote in Irish referendums Yeor Variable Chango in variable Resulting change in Confidence yesvote Interval at 95% oot Pantisansrip Now foes t0 I yezon pares att0.42% sw EVatitudoe 120 stated cvetan me OES sto 2002 Partisanship Nove fer 0 1% yea ep 74078% ree EV atitudos pct sveton) ne po Tato 80% ree Goverment satsfion __1 pot ean don reeset 710 81% 7885 Source: Treaty of Nice Referendurn 2001 and 2002 surveys ible 11 predicts that both (positive) changes in partisanship and in FU attitudes would have a positive impact on the yes-vote in the first Nice referendum, yet not sufficient to change the outeome of the final referendum (again, this is also due to the deflated yes- vote in this sample). Equally, partisanship and attitudes have an impact in the second Nice referendum, however, interestingly, the most significant impact on the vote is caused by a shift in government satisfaction, This could indicate that, as suggested, “second order” factors were more significant in the Irish referendums than in the more recent Danish referendums. The results in tables 9 to 11 thus seem to confirm the prediction in the hypotheses 1 and 2 relating to the importance of partisan cues and government satisfaction on an individual level. On an aggregate level, however, the rejection of the Nice Treaty in the first referendum give the impression of exeerly the opposite, since the majority of parties recommended a yes and the government was very ide may need to be popular at the time, Hence, the level campaign intensity on the taken inte account as an additional explanatory variable in order to interpret the voting behaviour in these referendums, since this most probably had s major impact on both the ent and 63 per cent in 2002. The higher level of inaccura The actual yes-vere in 2 may be related 10 the i rah elections yet, these heuristics are often related to issues of national politics, such as partisanship and government popularity, rather than Furopean issues. Hence, the model proposed in this paper borrows elements from the ‘attitude’ as well as the ‘second-order-election’ explanations of referendum behaviour, It suggests that the relative impact of European attitudes and national political factors depends on whether BU attitudes have become stable political predispositions as a resuit of fong-term intensive coverage and debate on the European affairs. To test this model, this paper has examined salience levels in Denmark and Ireland, by measuring the media coverage afforded to the EU, and analysed the strength of European attitudes by determining response stability in pane! studies, The data clearly show that whilst EU salience has increased dramatically in Denmark over the past three decades, it has remained relatively tow in Ireland, This is aiso reflected in the analysis of the stability of European attitudes in Denmark and Ireland, which illustrates stronger and more stable EU attitudes in the more recent Danish panel study. In the actual analyses of voting behaviour in Danish and trish referendums, the propositions of the model are also supported. Partisan cues and government satisfaction were important variables in all referendums (except the Euro- referendum), whilst EU attitudes have become an increasingly important factor in Danish referendums over the years. Hence, as expected, attitudes are a stronger predictor of voting behaviour in electorates with a higher level of EU salience and more stable attitudes towards European integration than in electorates with low EU salience and highly unstable EU attitudes, The results based on the irish survey data from the two Nice referendums were more ambiguous. Both partisanship, government satisfaction and attitudes were important in both referendums. Yet, this does not explain the large swing between the two referendums. An important additional explanatory factor may be the intensity of (pro-EU) elite cues in the two referendums. The level of pro-EU campaign tensity changed dramatically between the two campaigns, and this is likely to have affected the reception of cues and the willingness to vote in favour of the Treaty (or vote at all), For this reason, an improved model could include an additional explanatory variable of campaign intensity to improve the substance of the findings and the level of variance explained. Yet, even the limited model presented in this paper reveals that whilst the voters’ pay attention to national politicians and other elites, European attitudes have the potential to play an increasingly important role in EU referendums. » # References Achen, Christopher H. (1975) ‘Mass Political Attitudes and the Survey Response’ American Political Science Review, 69:1218-1231 Coverse, Philip E. (2000) ‘Assessing the Capacity of Mass Blectorates’. Annual Review of Political Science, 3:331-353 Converse, Philip E. (1964) “The nature of belief systems in mass publics’ in D, Apter (ed.} ideology and discontent, New York: Free Press. Craig, Stephen C., James G. Kane and Michael D. Martinez (2000) ‘Ambivalence, Attitude Strength and Response Instability: A Two-Wave Study of Abortion Attitudes in Florida’. Paper presented at the Annual Meeting of the American Political Science Association (APSA), Washington, 31 August-3 September 2000, Craig, Stephen C., James G, Kane and Michael D. Martinez (2001) ‘Information and Voting: A Panel Study’. Paper presented at the Annual Meeting of the American Political Science Association, San Francisco, 30 August-2 September 2001 Dearing, James W. and Everett Rogers (1996) Agenda Setting, Thousands Oaks, CA: Sage Epstein, Lec and Jeffrey A. Segal (2000) ‘Measuring Issue Salience’ in american Journal of Political Science, 44:66-83 Franklin, Mark (2002) ‘Learning from the Danish case: A comment on Palle Syensson’s critique of the Franklin thesis, in European Journal of Political Research 41, 751 Franklin, Mark and Christopher Wlezien (2002) ‘Reinventing election studies” in Electoral Studies 21, 331-338. Franklin, M,, C. van der Bijk and M, Marsh (1995) ‘Referendum outcomes and trust in government: Public support for Europe in the wake of Maastricht’ in West European Politics, 18:3 Franklin, M., M. Marsh, and C.Wlezien (1994) “Attitudes toward Europe and Referendum Votes: A Response to Siune and Svensson’ in Electoral Studies, 13:2, 117- 1 Franklin, Mark, Michael Marsh and Lauren McLaren (1994) *Uncorking the Bottle: Popular Opposition to European Unification in the Wake of Maastricht’, Journal of Common Market Studies, 32:4 fion, Public Gabel, Matthew J, (1998a) Interest and Integration. Market Liberal Opinion and European Union, Ann Abor: The University of Michigan Pri Gabel, Matthew J. (19985) ‘Public Support for European Integration: An Empirical Test of Five Theories’ in The Journal of Politics, 60:2, 333-54 Gabel, M. and H.D. Pelmer (1995) “Understanding variation in public support for European integration’ in European Journal of Political Research, 27, 3-19 Gary, John, Michaei Marsh, Fiachra Kennedy and Richard Sinnott (2002) “Voting Behaviour in EU Referendums: A Dublin Case Study’, Not published. Hill, Jennifer 1. and Hanspeter Kriesi (2001) ‘Classification by Opinion-Changing Behaviour: A Mixture Model Approach’. Political Analysis, 9:4, 301-324 King, Gary, Michael Tomz and Jason Wittenberg (2000) ‘Making the Most of Statistical Analyses: Improving Interpretation and Presentation’. American Journal of Political Seience, 44:2, 347-61 Kriesi, Hans Peter, Ruud Koopsmans, Jan Willem Dyvendak and Marco G. Giugni (1995) New Social Movements in Western Europe: A Comparative Analysis, London: UCL Press Krosnick, Jon A. (1988) “Attitude Importance and Attitude Change’, Journal of Experimental Social Psychology, 24:240-255. 5, Krosnick, Jon A. and Duane F. Alwin (1989) ‘Aging and Susceptibility to Attitude Change’ in Journal of Personality and Social Psychology, 57:3, 416-4 Lippmann, Waiter (1922) Public Opinion: New York: Macmillan Lupia, Arthur and Mathew D. MeCubbins (1998) The Democratic Dilemma. Can Citizens learn what they need to know? Cambridge: Cambridge University Press. Pierce, John C. and Douglas D. Rose (1974) ‘Nonattitudes and American Public Opinion: The Examination ofa Thesis’. American Political Science Review, 68:626-659 Shaw, Donald and Shannon Martin (1992) ‘The fumction of mass media agenda setting’ in Journalism Quarterly 69, 902-920. Sides, Jobn (1999) “Built on Rock or Sand? The Stability of Religiosity and Attitudes Towards Abortion’, Institute of Governmental Studies Working Paper 99-13, Berkeley. Siune, Karen, Palle Svensson, and Ole Tonsgeard (1994a) Fra et nej til et ja. Aartus: Politics Siune, Karen, Palle Svensson, and Ole Tonsgaard (1994b) ‘The EU: The Danes said ‘No” in 1992, but ‘Yes’ in 1993: How and Why?” in Electoral Studies, 13:2. 107-116 Svensson, Paite (2002) ‘Five Danish referendums on the European Community and European Union: A ctitical assessment of the Franklin thesis’ in European Journal of Political Research 41, 733-750

Das könnte Ihnen auch gefallen