Winning, Losing and Satisfaction with Democracy - Wiley Online Library

P O L I T I C A L S T U D I E S : 2 0 0 7 VO L 5 5 , 4 2 5 – 4 4 1 doi: 10.1111/j.1467-9248.2007.00659.x

Winning, Losing and Satisfaction with Democracy André Blais

François Gélineau

Université de Montréal

Université Laval

Previous research has shown that those who won an election are more satisfied with the way democracy works than those who lost. What is not clear, however, is whether it is the fact of winning (losing), per se, that generates (dis)satisfaction with democracy. The current study explores this winner/loser gap with the use of the 1997 Canadian federal election panel study. It makes a theoretical and methodological contribution to our understanding of the factors that foster satisfaction with democracy. At the theoretical level, we argue that voters gain different utility from winning at the constituency and national levels in a parliamentary system, and that their expectations about whether they will win or lose affect their degree of satisfaction with democracy. On the methodological front, our analysis includes a control group (non-voters) and incorporates a control for the level of satisfaction prior to the election. The results indicate that the effect of winning and losing on voters’ satisfaction with democracy is significant even when controlling for ex ante satisfaction before the election takes place, and that the outcome of the election in the local constituency matters as much as the outcome of the national election. They fail to show, however, that expectations about the outcome of the election play a significant role in shaping satisfaction with democracy.

The extant literature has provided ample empirical evidence that support for democratic institutions in liberal democracies varies among citizens. More precisely, the literature has convincingly demonstrated that those who vote for the winning party in electoral contests are generally more satisfied with democracy than those who vote for the losing parties. Yet we know relatively little about how and why this winner/loser gap is produced. We do not know whether winners become more satisfied (and losers more dissatisfied) immediately after the election or whether the difference emerges before the election. We do not know whether losers are more or less dissatisfied than those who do not play the game, non-voters. We do not know whether winning (or losing) in national elections matters more than winning (or losing) in local ones. Finally, we do not know whether expectations about the outcome of the election affect the extent to which winners and losers are satisfied with the way democracy works. This article examines the determinants of the winner/loser gap with the use of the 1997 Canadian Election Study (CES). Building on previous work on the subject, we develop a model that discriminates winners and losers at the local and national levels, includes non-voters and takes into account voters’ expectations about whether they will win or lose. Our results indicate that the effect of © 2007 The Authors. Journal compilation © 2007 Political Studies Association

426

A N D R É B L A I S A N D F R A N Ç O I S G É L I N E AU

winning and losing on voters’ satisfaction with democracy is significant even when controlling for ex ante satisfaction before the election takes place. Exploring the winner/loser gap is crucial for the study of democracy. Democratic contests inevitably produce winners and losers. It is easy for the winners, who support the party or parties that form government, to reach the conclusion that democracy works just fine. Things are not so easy for the losers. They would have preferred some other party to govern, and they may distrust those who were elected. The democratic process is based on the premise that losers consent to their loss and accept that democracy is still working pretty well, despite the unfortunate outcome (Anderson et al., 2005; Nadeau and Blais, 1993). Satisfaction with democracy is therefore closely associated with the legitimacy of the political regime. To use Russell Dalton’s (2004, p. 24) typology of political support, it is an evaluative dimension (not an affective orientation) that deals with regime norms and procedures. In democracies, just as in other types of regimes, legitimacy provides the basis on which the system continues to function. Elections act as one vehicle to foster legitimacy. The current article explores the link between the electoral process and support for existing democratic procedures. It seeks to understand to what extent elections contribute to satisfaction (or dissatisfaction) with the performance of the political regime.

Theoretical Framework Electoral outcomes have been found to have an impact on citizens’ attitudes towards democratic institutions (Anderson and Guillory, 1997; Anderson and Tverdova, 2001; Anderson et al., 2005; Bowler and Donovan, 2002; Clarke and Kornberg, 1992; Fuchs et al., 1995; Ginsberg and Weissberg, 1978; Joslyn, 1999; Listhaug, 1995; Norris, 1999a). The logic is straightforward. Citizens who cast their vote for the winning party are more inclined to display faith in the way democracy works, because they like and/or trust the party/individual who has been elected. Inversely, citizens who cast their vote for any of the losing parties are more prone to display lower levels of support for the political system, because they dislike and/or distrust those who have been chosen to govern. The pattern is well established and makes sense theoretically. But do we know whether it is the fact of winning (losing), per se, that generates (dis)satisfaction with democracy? Almost all prior studies are cross-sectional and compare the views of winners and losers after the election. Two notable exceptions are Christopher Anderson and Andrew LoTempio’s (2002) panel study, which deals with political trust rather than with satisfaction with democracy, and Anderson et al.’s (2005) analysis, which, among other things, examines the aggregate evolution of attitudes among supporters of parties that went from being winners to being losers, or vice versa. Ideally, we would like to know whether it is the election as such that produces the winner/loser gap that we observe after © 2007 The Authors. Journal compilation © 2007 Political Studies Association POLITICAL STUDIES: 2007, 55(2)

W I N N I N G, L O S I N G A N D S AT I S FAC T I O N

427

the election.We would have more confidence in saying that the gap is caused by the outcome of the election if we find that there was no such gap before the election, that it emerged only after the election had taken place.1 Should we then expect the winner/loser gap to emerge only after the election? Not necessarily. Winners may well feel good about democracy even before the election because they expect to win. The same logic applies to losers. The current study, contrary to previous ones, looks at satisfaction with democracy immediately before and immediately after an election. In addition, we argue that in order to understand how winning or losing affects one’s evaluation of democracy it is important to consider as well those that are neither winners nor losers, that is, those who did not vote. Non-voters do not experience the ‘fact’ of winning or losing, and they constitute an interesting group with which to compare winners and losers. Previous studies have also failed to account for the fact that there is typically more than one contest in a given election. In the United States, the country that has been most systematically studied, there is a contest for who will become president, but also contests to decide who will become members of Congress. In parliamentary regimes, there are local contests for who will become the Member of Parliament in each constituency as well as a national contest to decide which party will form the government.2 It is quite possible for a voter to win in one contest and to lose in another. In their study, Anderson and LoTempio (2002) made a useful distinction between winning (or losing) the presidential vs. the congressional election, and they report that only the presidential election seems to matter. There is no comparable analysis for parliamentary regimes, distinguishing winners and losers at the local and national levels. The current study fills that gap. In a parliamentary regime, the stakes are arguably higher in the national contest. After all, it is the party that elects the most candidates that forms the government. For this reason, we expect that winning/losing in the national contest will have a greater effect on satisfaction than winning/losing in the local constituency. Expectations about the outcome of the election may also play an important role in the way winners and losers evaluate their democratic institutions. Because ‘outcomes do not differ systematically (i.e. regularly or predictably) from what people expect them to be’ (Sargent, 2002), as proposed by the rational expectations theory (Muth, 1961), we suggest that only voters who wrongly assessed the chances of their party to win or lose will change their assessment of democracy after the election. As a result, a gap should only emerge after the election among the winners who expected to lose, and the losers who expected to win. Accordingly, voters who correctly evaluated the chances of their party winning/losing the election should not be more/less satisfied with democracy after the election, since the outcome reflects their expectations. In contrast, winners who expected to lose should be more satisfied with democracy, while losers who expected to win should feel more negative. © 2007 The Authors. Journal compilation © 2007 Political Studies Association POLITICAL STUDIES: 2007, 55(2)

428


Model and Data In order to explore the questions outlined above, we use the 1997 Canadian Election Study (CES). The study includes a rolling campaign cross-section survey of 3,949 respondents and a post-election re-interview of 3,170 of them. The dependent variable consists in the standard question tapping people’s degree of satisfaction with the way democracy works in Canada (see Appendix for a detailed description of the variables used in the analysis). The variable ‘satisfaction with democracy’ is obviously an imperfect measure of support for the political regime. It is a measure of ‘regime performance’ (Dalton, 2004; Norris, 1999a; 1999b). As Jonas Linde and Joakim Ekman cogently point out (2003, p. 405), satisfaction with democracy should not be used as an indicator of support for the principles of democracy: ‘it is an item that taps the level of support for how the democratic regime works in practice’. Expressing further concerns with the variable, other authors have noted that the indicator is correlated with specific measures of support such as executive approval or partisan preference (Canache et al., 2001; Klingemann, 1999; Kornberg and Clarke, 1994). While a strong correlation between several survey questions does suggest some conceptual overlap, by no means does it entail that the different measures capture the exact same theoretical phenomenon (Anderson, 2002).And indeed, with only one notable exception (Canache et al., 2001),3 no one has completely discredited the indicator for the above-mentioned reasons. Hence,‘in the absence of a better item, the satisfaction with democracy measure is a reasonable (albeit imperfect) indicator that we can use to test our theories’ (Anderson, 2002). The question measuring satisfaction with democracy was asked in the campaign survey and repeated in the post-election questionnaire. We can thus compare respondents’ views before and after the election. By including the pre-election measure in our model, we are effectively measuring change in satisfaction with democracy after the election. The Liberal party won 155 seats out of 301 and went on to form the government. Respondents who reported to have voted for the Liberal party are therefore construed to be ‘national winners’; those who voted for another party are identified as ‘national losers’. The 1997 CES contains information about respondents’ constituency, which enables us to establish whether individual voters won or lost in their own constituency.4 We can thus distinguish ‘local’ from ‘national’ winners and losers. The analysis also takes into account those who did not vote. The campaign questionnaire included questions tapping respondents’ perceptions of the various parties’ chances of winning, both nationally and in their local constituency. These questions allow us to distinguish, among both winners and losers, those who expected to win from those who did not. Another way of assessing voters’ expectations with regards to the outcome of the electoral contest is to divide losers along the political parties for which they voted. In doing so, we © 2007 The Authors. Journal compilation © 2007 Political Studies Association POLITICAL STUDIES: 2007, 55(2)


429

can distinguish individuals who voted for parties that previously controlled the government (the Progressive Conservative party) or had some chance of forming the government (the Reform party) in the near future from those who voted for ‘minor’ parties.5 Supporters of minor parties may accept their defeat more easily because they were resigned from the beginning to the fact that they would not win. Our model also incorporates other factors that are believed to affect citizens’ assessments of democratic performance. Numerous accounts exist to support the claim that those who feel that their own personal financial situation and/or the overall economy are in good shape are more prone to evaluate positively their democratic regime (Clarke et al., 1993; 2004; Finkel et al., 1989; Franz, 1986; Kornberg and Clarke, 1994; Listhaug and Wiberg, 1995; Weatherford, 1984; 1987). To assess this possibility, we constructed a scale combining egocentric, sociotropic, retrospective and prospective evaluations of the economy. One’s evaluation of how well or badly democracy works can also reflect more general orientations towards politics and politicians. More specifically, those who distrust politicians in general are likely to display lower levels of satisfaction with the democratic process.We constructed to that effect a cynicism variable that taps citizens’ overall feelings about politicians. The model finally includes dummy variables to control for partisanship, region, age, gender, education and income.6 As already indicated, we examine the impact of winning and losing on one’s evaluation of how well or badly democracy works in the context of the 1997 Canadian election.We have chosen that specific election because the 1997 CES is the first study to have measured satisfaction with democracy before and after the election among the same respondents, and to include all the information that is required to test our hypotheses. While further research will be needed to determine whether the patterns uncovered here hold across space and over time, we should note that the overall level of satisfaction observed is not atypical of what is usually found in contemporary democracies (Fuchs et al., 1995, table 11-4, p. 341; Klingemann, 1999, table 2.10, p. 50).

Findings The analysis starts with an examination of the evolution of satisfaction with democracy over the course of the campaign and after the election. Figure 1 reports seven-day moving averages of the daily percentage of respondents declaring being fairly satisfied or very satisfied with the way democracy works in Canada. Overall level of satisfaction remains remarkably stable throughout the campaign, it jumps substantially immediately after the election and it remains at higher levels thereafter. The fact that we observe a clear increase in satisfaction after the election strongly suggests that the election itself made people more satisfied with the way democracy works. © 2007 The Authors. Journal compilation © 2007 Political Studies Association POLITICAL STUDIES: 2007, 55(2)

430


Figure 1: Satisfaction with Democracy Before and After the Election

Note: Entries are moving averages (over a period of seven days; three days before and three days after) of the aggregated percentage of respondents who are somewhat or very satisfied with the way democracy works in Canada.

The real question, however, is whether winning as such made people more satisfied. Let us look, then, at how people felt before and after the election. We compare the level of satisfaction with democracy among five groups, the first four groups corresponding to those who won or lost in both electoral contests, or won only nationally or only locally. The fifth group corresponds to non-voters. Table 1 (second row) confirms that, after the election, winners are more satisfied than losers. It also shows that those who win at both levels are more satisfied than those who win only at one level. The same logic applies to losers. This being said, winning or losing at the national level seems to matter more than winning or losing at the local level. Finally, losers’ assessments of democracy are as positive as those of non-voters. Was that pattern ‘produced’ by the election or did it exist before the election? Were the eventual winners ‘happier’ even before they won? Table 1 (first row) suggests that there was indeed such a pattern: those who were about to win were already more satisfied than the eventual losers. Also noteworthy, satisfaction with democracy increased after the election among losers and non-voters, as well as among winners (Table 1, third row). Should we then conclude that winning as such does not make a difference? Not necessarily.We need to take into account other factors that may affect satisfaction with democracy. The implication, however, is that in order to ascertain the specific impact of winning, it is imperative to control for the level of satisfaction before the election. The multivariate regression models presented in Table 2 do precisely that.7 The first model seeks to explain satisfaction with democracy with prior levels of © 2007 The Authors. Journal compilation © 2007 Political Studies Association POLITICAL STUDIES: 2007, 55(2)

431


Table 1: Satisfaction with Democracy among Winners, Losers and Non-Voters

Before After Difference N

Win federal/win constituency

Win federal/lose constituency

Lose federal/win constituency

Lose federal/lose constituency

Non-voters

0.645 0.738 +0.09

0.634 0.688 +0.05

0.455 0.587 +0.13

0.526 0.579 +0.05

0.493 0.577 +0.08

373

206

428

624

503

Note: Entries are mean responses. Satisfaction with democracy = 0; 0.33; 0.66; 1.

satisfaction, cynicism, evaluation of the incumbent’s performance and a host of socio-demographic variables. As per our expectations, the results suggest that the likelihood of being satisfied with the way democracy works is enhanced by prior levels of satisfaction with democracy, positive evaluations of the economy, liberal partisanship and by being born in a foreign country. Also in line with our expectations, cynicism is found to reduce the likelihood of being satisfied with democracy. Model 2 of Table 2 provides a more rigorous test of the effect of winning and losing, while controlling for prior satisfaction and the other determinants of satisfaction. Here, the effect of winning and losing is added to the model. We created four groups (win federal/win constituency, win federal/lose constituency, lose federal/win constituency, lose federal/lose constituency) depending on whether voters won or lost in their constituency and at the national level. Those who won at both levels (win federal/win constituency) are the reference group. We find that, net of all other factors, full losers (lose federal/lose constituency) and non-voters are more likely to become dissatisfied than full winners (at both levels). Interestingly, those who win at only one level and lose at the other emerge as similar to full winners; the coefficients for those voters are not statistically different from zero, therefore substantively not different from those who won at both levels, the reference group. The implication is that Canadian voters react to the local outcome as well (and as much) as to the national one. This is a surprising result. Our expectation had been that Canadians would focus on the national election, just like Americans who appear to be affected solely by the result of the presidential contest (Anderson and LoTempio, 2002). Another interesting observation from column 2 of Table 2 is the fact that the significance of the liberal partisanship variable vanishes when we control for the fact of winning and losing. This suggests that liberal supporters are more likely to be satisfied with democracy because they win their electoral contest, and not simply because they are biased in favor of the incumbent party. © 2007 The Authors. Journal compilation © 2007 Political Studies Association POLITICAL STUDIES: 2007, 55(2)

432


Table 2: Satisfaction with Democracy After the Election (1) Prior level of satisfaction

1.908*** (0.164)

Win federal/lose constituency Lose federal/win constituency Lose federal/lose constituency Non-voter Liberal ID PC ID NDP ID Reform ID Bloc québécois ID Economy Cynicism Atlantic Quebec West Age Male School Foreign-born Cut 1 Cut 2 Cut 3 Number of obs Log likelihood LR chi2 Pseudo R-square

0.275*** (0.085) 0.001 (0.101) 0.019 (0.126) -0.122 (0.126) 0.117 (0.142) 0.666*** (0.173) -0.569*** (0.165) 0.007 (0.109) -0.003 (0.100) 0.017 (0.078) -0.083 (0.230) 0.003 (0.066) 0.136 (0.166) 0.329*** (0.096) -0.741 (0.260) 0.340 (0.253) 2.311 (0.266) 1,922 -1,801.92 322.26*** 0.15

(2) 1.900*** (0.162) -0.199 (0.129) -0.193 (0.135) -0.447*** (0.120) -0.352** (0.125) 0.162 (0.094) 0.060 (0.104) 0.089 (0.128) -0.142 (0.132) 0.099 (0.148) 0.625*** (0.173) -0.503** (0.167) 0.042 (0.115) 0.008 (0.105) 0.024 (0.089) -0.125 (0.245) 0.013 (0.066) 0.149 (0.170) 0.307** (0.097) -1.023 (0.281) 0.068 (0.275) 2.062 (0.288) 1,922 -1,785.42 383.1*** 0.16

Note: Entries represent unstandardized coefficients with robust standard errors. *p ⱕ 0.05; **p ⱕ 0.01; ***p ⱕ 0.001. PC = Progressive Conservative Party NDP = New Democratic Party

© 2007 The Authors. Journal compilation © 2007 Political Studies Association POLITICAL STUDIES: 2007, 55(2)


433

In order to illustrate the results presented in Table 2, we computed the predicted probability of being satisfied (fairly and very satisfied) with the way democracy works after the election.8 This exercise indicates that voters who won in either electoral contest (or both) (0.79) are significantly more likely to be satisfied with democracy than individuals who lost at both levels (0.71) or abstained (0.74). While these results are highly interesting, they do not say much about the underlying mechanism that could explain why winners feel more positive about democracy, and losers feel more negative, after the election.9 A plausible interpretation is that among winners, those who did not expect to win should become more satisfied with democracy after the election, while those who ‘knew’ they were going to win should not change their assessment of democracy. Among losers, those who did not anticipate the loss should become more cynical about democracy, while those who ‘knew’ they would lose should exhibit little change. On the whole, we should observe little change in satisfaction after the election among those who correctly expected to win or lose. Conversely, we should observe greater shifts in satisfaction among voters whose expectations proved to be wrong. Table 3 tests these hypotheses using expectations about the outcome of the election. We distinguish four groups of voters: optimistic winners, pessimistic winners, optimistic losers and pessimistic losers. Voters were construed to be ‘optimistic’ when the chances of winning they gave their party in either the national or constituency contests were at least 50 per cent, and to be pessimistic otherwise.10 Winners were defined as those who won either at the national or local level (or both) and losers as those who lost in both contests. Optimistic winners are the base category. Here, non-voters are excluded from the analysis as it is not possible to label them as optimists or pessimists. We expect ‘pessimistic’ winners to be more satisfied with democracy than their optimistic counterparts after the election as they are favorably surprised by the outcome.We expect ‘optimistic’ losers, who anticipated to win, to become more antagonistic about democracy than their pessimistic counterparts as a display of their disappointment with the outcome. The results presented in Table 3 do not confirm our predictions. Consistent with the findings reported in Table 2, losers became more negative about democracy after the election. However, the difference between optimistic and pessimistic voters, among either winners or losers, does not appear to be statistically significant.11 As a result, our analysis does not offer support for the hypothesis that winners who were not very hopeful to win become more supportive of democracy after the election, or that losers who were hopeful to win become less supportive of democracy. These results rather suggest that expectations about the outcome do not substantially affect voters’ assessment of the functioning of democracy. Another way of measuring the impact of expectations about the outcome is to divide losers along the parties they voted for. Accordingly, losers who voted for © 2007 The Authors. Journal compilation © 2007 Political Studies Association POLITICAL STUDIES: 2007, 55(2)

434


Table 3: Expectations and Satisfaction with Democracy after the Election (3) Prior level of satisfaction Pessimistic winner Optimistic loser Pessimistic loser Liberal PC NDP Reform Bloc québécois Economy Cynicism Atlantic Quebec West Age Male School Foreign-born Cut 1 Cut 2 Cut 3 Number of obs Log likelihood LR chi2 Pseudo R-square

1.959*** (0.190) -0.103 (0.150) -0.413*** (0.106) -0.290** (0.104) 0.151 (0.101) 0.096 (0.117) 0.109 (0.154) -0.215 (0.137) -0.057 (0.156) 0.593** (0.201) -0.487** (0.185) -0.006 (0.121) 0.054 (0.112) -0.043 (0.096) -0.250 (0.263) 0.030 (0.076) 0.036 (0.191) 0.329*** (0.103) -1.082 (0.308) 0.074 (0.298) 2.056 (0.312) 1,514 -1,392.16 297*** 0.16

Note: Entries represent unstandardized coefficients with robust standard errors. *p ⱕ 0.05; **p ⱕ 0.01; ***p ⱕ 0.001. PC = Progressive Conservative Party NDP = New Democratic Party © 2007 The Authors. Journal compilation © 2007 Political Studies Association POLITICAL STUDIES: 2007, 55(2)


435

parties that had no aspiration or realistic chances of forming the government, that is, the NDP (New Democratic Party) and the Bloc québécois, should not be as disappointed as those who voted for losing parties that had been in government in the past (the Conservatives) or had hopes of replacing the Liberals in the near future (the Reform party). The reason is simple: the former most probably did not expect their party to win both the federal and constituency contests. In order to test that hypothesis, we created three groups: winners (the reference category), losers who voted for parties with no aspiration to govern (NDP and Bloc québécois) and losers who voted for parties with some aspiration to govern (PC and Reform). The results presented in Table 4 are ambiguous.12 While losers who voted for the PC or the Reform are found to exhibit less satisfaction than winners (i.e. Liberal voters, the reference group), their level of satisfaction with democracy is not statistically different from those who voted for the NDP or the Bloc québécois.13 From this, we conclude that there is only weak support for the hypothesis that losers who support parties that have no aspiration to form a government are less dissatisfied than losers who support parties with such ambition. All in all, our findings concerning the role of expectations are somewhat disappointing.We performed two different tests, reported in Tables 3 and 4. In Table 3, we distinguished optimistic and pessimistic voters (among both winners and losers) and did not observe any significant difference between the two groups. In Table 4, we divided losers into two groups: those supporting parties with no aspiration to form a government and those supporting parties that have formed a government in the past or could be in government in the not-too-distant future. While the former group emerges as somewhat less dissatisfied than the latter, the difference is not strong enough to reach conventional levels of statistical significance. Clearly more work is needed to explore this question. But for the time being the plausible intuition that the impact of winning or losing is conditional on whether one correctly anticipated the outcome of the election is not borne out by the data.

Conclusion Our analysis started by asking whether voters’ satisfaction with democracy is contingent upon the actual outcome of the electoral contest. In doing so, we proposed theoretical and methodological improvements to past attempts at explaining individual satisfaction with democracy. The results of our analysis confirm that winners are more satisfied than losers with the way democracy works. This is consistent with prior findings (Anderson and Guillory, 1997; Anderson and Tverdova, 2001; Anderson et al., 2005; Bowler and Donovan, 2002; Clarke and Kornberg, 1992; Fuchs et al., 1995; Ginsberg and Weissberg, 1978; Joslyn, 1999; Listhaug, 1995; Norris, 1999b). Contrary to previous research, however, we show that the eventual winners were already more satisfied than the © 2007 The Authors. Journal compilation © 2007 Political Studies Association POLITICAL STUDIES: 2007, 55(2)

436


Table 4: Parties and Satisfaction with Democracy After the Election (4) Prior level of satisfaction Losers – PC and Reform voters Losers – NDP and Bloc voters Liberal ID PC ID NDP ID Reform ID Bloc québécois ID Economy Cynicism Atlantic Quebec West Age Male School Foreign-born Cut 1 Cut 2 Cut 3 Number of obs Log likelihood LR chi2 Pseudo R-square

1.964*** (0.190) -0.443*** (0.098) -0.161 (0.132) 0.158 (0.100) 0.138 (0.120) 0.007 (0.165) -0.184 (0.134) -0.101 (0.161) 0.600** (0.201) -0.467* (0.184) -0.020 (0.121) 0.038 (0.112) -0.070 (0.097) -0.193 (0.263) 0.045 (0.075) 0.067 (0.190) 0.318** (0.103) -1.018 (0.307) 0.138 (0.295) 2.125 (0.309) 1,514 -1,390.09 291.76*** 0.16

Note: Entries represent unstandardized coefficients with robust standard errors. *p ⱕ 0.05; **p ⱕ 0.01; ***p ⱕ 0.001. PC = Progressive Conservative Party NDP = New Democratic Party © 2007 The Authors. Journal compilation © 2007 Political Studies Association POLITICAL STUDIES: 2007, 55(2)


437

eventual losers during the campaign, before the election took place. Notwithstanding this, we find that, everything else being equal, after the election satisfaction with democracy increases more among winners and less among losers. We also show that the outcome of the election in the local constituency matters as much as the national outcome. Our findings are slightly different from those of Anderson and LoTempio (2002). In their study of political trust in the United States, they indicate that losing the presidential election does make one less trusting, but that losing the congressional election does not make a difference. In our study of the Canadian parliamentary system, we find that electoral outcomes only have a significant effect when voters win or lose in both the constituency and national contests. It remains to be seen whether this unforeseen pattern is peculiar to the Canadian case, with its strong regional cleavages (Gidengil et al., 1999), or is common in parliamentary regimes with single-member districts, like Britain and Australia. Previous work in the area (with the important exception of Anderson and LoTempio, 2002) has tended to assume that there is only a single contest in any given election. The typical situation is quite the opposite. Most of the time there are many concurrent elections (and sometimes referenda as well). Our results suggest that we need to decipher and factor in the relative impact of these various contests. Finally, our analysis has explored how expectations about the outcome of the election affect the way individuals evaluate the democratic process. Drawing upon rational expectations theory, we suggested that winners who did not expect to win would become more supportive of democracy after the election and losers who thought they would win would become more antagonistic about democracy. Our analysis found little support for this claim. This is a surprising finding since there are plausible theoretical reasons to suppose that one’s reaction to the outcome of an election hinges on one’s prior expectations. Further research is required to determine whether this is a general pattern or not. All things considered, however, it seems that satisfaction with democracy increases after the election among losers and non-voters as well as among winners, thus suggesting that the end of the electoral cycle has a positive effect on everyone’s evaluation of the democratic process, whether they won or lost, or even abstained. This would suggest that there is a honeymoon effect not only for the elected government, but also for democracy writ large.

Appendix: Description of Variables Satisfaction with democracy (pre- and post-election): a 4-point scale running from zero (not satisfied at all) to 1 (very satisfied) built with the answer to the following question:‘On the whole are you very satisfied, fairly satisfied, not very satisfied, or not satisfied at all with the way democracy works in Canada?’ Win federal: a dummy variable identifying voters who reported voting for the winning party in the federal contest, i.e. the Liberal party. Non-voters are given the value 0. © 2007 The Authors. Journal compilation © 2007 Political Studies Association POLITICAL STUDIES: 2007, 55(2)

438


Lose federal: a dummy variable identifying voters who reported voting for any of the losing parties in the federal contest. Non-voters are given the value 0. Win constituency: a dummy variable identifying voters who reported voting for the winning party in their constituency contest. The winning party of each constituency was associated with individual voters on the basis of their reported postal codes (see Note 2). Non-voters are given the value 0. Lose constituency: a dummy variable identifying voters who reported voting for any of the losing parties in the constituency contest. The winning party of each constituency was associated with individual voters on the basis of their reported postal codes (see Note 2). Non-voters are given the value 0. Non-voter: a dummy variable identifying individuals who reported not voting. Optimism: a dummy variable identifying voters who expected the party they voted for to win in at least one electoral contest. Voters were construed to be optimistic if the chances of winning they gave the party they voted for were at least 50 per cent in either the federal or the constituency contests. Non-voters are excluded from this portion of the analysis and are thus coded as missing. Winners: a dummy variable identifying voters who won in at least one contest (i.e. win federal/win constituency, win federal/lose constituency or lose federal/ win constituency). Party identification: a series of dummy variables identifying individuals who feel fairly/very strongly about a political party (Liberal, PC, NDP, Reform and Bloc québécois). Non-voters are given the value 0. In the analysis, the reference category is individuals who do not feel fairly/very strongly about any party. Economy: a 9-point scale running from zero (negative) to 1 (positive) constructed from an additive index made up of the following questions:‘Financially, are you better off, worse off, or about the same as a year ago?’,‘Do you think that a year from now you will be better off financially, worse off, or about the same as now?’,‘Over the past year, has Canada’s economy gotten better, gotten worse, or stayed about the same?’, ‘What about the next 12 months? Will Canada’s economy get better, get worse or stay about the same?’ Cynicism: a 17-point scale running from zero (not at all cynical) to 1 (very cynical) constructed from an additive index made up of the following questions: ‘For each of the following statements, please tell me if you strongly agree, somewhat agree, somewhat disagree, or strongly disagree: a) Those elected in Parliament soon lose touch with people, b) I don’t think the government cares much what people like me think, c) Politicians are ready to lie to get elected’;‘Do political parties keep their election promises most of the time, some of the time, or hardly ever?’ © 2007 The Authors. Journal compilation © 2007 Political Studies Association POLITICAL STUDIES: 2007, 55(2)


439

Regional dummy variables: a series of dummy variables identifying the region of residence of the respondents, i.e.Atlantic (comprised of Newfoundland, Prince Edward Island, Nova Scotia and New Brunswick), Quebec, Ontario and West (comprised of Manitoba, Saskatchewan, Alberta and British Columbia). Ontario is not included in the models as it is used as the base category. Age: actual age of respondent (continuous variable divided by 100). Male: a dichotomous gender variable coded 1 for male and zero for female. School: an 11-point scale running from zero (no schooling) to 1 (master degree or doctorate). Foreign-born: coded zero if respondent was born in Canada, coded 1 otherwise. (Accepted: 13 June 2006) About the Authors André Blais, Department of Political Science, Université de Montréal, C.P. 6128, Succ. Centre-ville, Montréal, Québec H3C 3J7, Canada; email: [email protected] François Gélineau, Department of Political Science, Université Laval, Québec, Québec G1K 7P4, Canada; email: [email protected]

Notes 1 Another possibility is that the outcome of the preceding election has an impact on satisfaction with democracy. For example, voters who vote for the losing party in consecutive elections may be more critical of the way democracy works than voters who consistently vote for winning parties. Our model, however, already controls for the level of satisfaction with democracy prior to the election, which captures the influence of previous elections. Furthermore, the only way to establish whether voters won or lost in the previous election is to rely on vote recall, the reliability of which is dubious (Himmelweit et al., 1978). 2 In a parliamentary regime with single-member district plurality such as Canada, each voter participates in the two contests by casting a single vote for one candidate/party. Voters can win at the district level by voting for the candidate who receives the plurality of the vote. Voters can win at the national level by voting for a candidate whose party elects a plurality of candidates in all the districts. 3 Note that Canache et al.’s conclusions are based on data collected in newer democracies, where the link between specific support and satisfaction with democracy should be much stronger because of the limited basis of diffuse support (Anderson, 2002). 4 Respondents’ constituency could be identified with their postal code. The mailback survey respondents indicated their complete postal code. For those who did not return the mailback survey, the constituency was identified on the basis of the FSA (forward sortation area) of the postal code. Some respondents refused to give their FSA, some only gave a partial FSA code and others gave codes which did not make any sense. As a result, the 1997 CES identifies the respondent’s constituency for about 61 per cent of the sample (2,400 respondents). In order to establish whether the individual voter won or lost the election in their constituency, we used the official electoral results. 5 The Liberals and the Progressive Conservatives are the only parties to have formed the government in Canadian history. From the 1960s to the 1990s, the NDP obtained between 15 and 20 per cent of the vote ( Johnston et al., 1992). The party system was shattered in 1993 when the Progressive Conservatives, who had formed the government since 1984, won only 16 per cent of the vote and two seats and the NDP got only 7 per cent of the vote and 9 seats. Two new parties emerged: the Reform party, which replaced the Progressive Conservatives as the main challenger on the right with 19 per cent of the vote and 18 per cent of the seats; and the separatist Bloc québécois, which ran candidates only in Quebec and obtained 49 per cent of the vote in that province, enough to win 57 seats and to form the official opposition ( Johnston et al., 1996). The Bloc could not win more than 75 seats (the total number of seats in the province of Quebec), and so the most serious challenge to the Liberals in 1997 was the Reform party, which went on to form the official opposition, but the fight on the right was not over since the Conservatives had a longer tradition and more party identifiers than the Reform party and the latter was plagued © 2007 The Authors. Journal compilation © 2007 Political Studies Association POLITICAL STUDIES: 2007, 55(2)

440


by an image of ‘extremism’ (Nevitte et al., 2000). The Progressive Conservatives and the Reform party (which had become the Alliance in 2000) merged to become the Conservative party in 2004. 6 See Blais et al. (2002) for a discussion of socio-demographic cleavages in Canadian politics. 7 Given the limited range of our dependent variable, we rely upon ordered probit regression analysis to estimate our models of individual satisfaction with democracy.All our tables report robust standard errors, obtained by using the Huber-White sandwich estimator of variance to address potential problems of heteroskedasticity. 8 In order to isolate the conditional effect of a single dummy variable, we first set the variable of interest at 0, while letting the values of all the other variables fluctuate. Second, we computed the probability of being satisfied (somewhat or strongly) with democracy for each individual with the help of the ‘predict’ command in Stata. Third, we computed the average probability for the entire sample. We then repeated these three steps while setting the variable of interest at 1. 9 Because we are controlling for satisfaction just before the election, we are tapping the change in satisfaction that is produced by the election itself. 10 Other operationalisations of this concept were also explored, but generated similarly inconclusive results. For example, following one anonymous reviewer’s suggestion, we tried dividing the losers into three groups rather than two. In doing so, we compared losers with low (0 to 33 per cent), medium (34 to 66 per cent) and high (67 to 100 per cent) expectations about the victory of their political party. Similarly to the results reported in Table 3, there seemed to be a negative correlation between satisfaction and expectations among losers, as the magnitude of the coefficients increased with the level of expectations. Unfortunately, however, none of the coefficients were statistically different from one another. 11 This was established by performing a Wald test on each pair of coefficients to assess whether their difference was statistically significant. 12 As in the analysis reported in Table 3, and for the same reason, non-voters are not included in the analysis. 13 The Wald test indicates that the difference between the two loser groups is significant only at the 0.07 level.

References Anderson, C. J. (2002) Good Questions, Dubious Inferences, and Bad Solutions: Some Further Thoughts on Satisfaction with Democracy. Unpublished draft, Maxwell School of Citizenship and Public Affairs, Syracuse University. Anderson, C. J. and Guillory, C. A. (1997) ‘Political Institutions and Satisfaction with Democracy: A Crossnational Analysis of Consensus and Majoritarian Systems’, American Political Science Review, 91 (1), 66–81. Anderson, C. J. and LoTempio, A. J. (2002) ‘Winning, Losing and Political Trust in America’, British Journal of Political Science, 32 (2), 335–51. Anderson, C. J. and Tverdova, Y.V. (2001) ‘Winners, Losers, and Attitudes about Government in Contemporary Democracies’, International Political Science Review, 22 (4), 321–38. Anderson, C. J., Blais, A., Bowler, S., Donovan, T. and Listhaug, O. (2005) Losers’s Consent: Elections and Democratic Legitimacy. Oxford: Oxford University Press. Blais, A., Gidengil, E., Nadeau, R. and Nevitte, N. (2002) Anatomy of a Liberal Victory: Making Sense of the Vote in the 2000 Canadian Election. Peterborough: Broadview Press. Bowler, S. and Donovan, T. (2002) ‘Democracy, Institutions and Attitudes about Citizen Influence on Government’, British Journal of Political Science, 32 (2), 371–90. Canache, D., Mondak, J. J. and Seligson, M. A. (2001) ‘Meaning and Measurement in Cross-national Research on Satisfaction with Democracy’, Public Opinion Quarterly, 65 (4), 506–28. Clarke, H. D. and Kornberg, A. (1992) ‘Do National Elections Affect Perceptions of MP Responsiveness? A Note on the Canadian Case’, Legislative Studies Quarterly, 17 (2), 183–204. Clarke, H. D., Dutt, N. and Kornberg, A. (1993) ‘The Political Economy of Attitudes Toward Polity and Society in Western European Democracies’, The Journal of Politics, 55 (November), 998–1021. Clarke, H. D., Sanders, D., Stewart, M. C. and Whiteley, P. (2004) Political Choice in Britain. Oxford: Oxford University Press. Dalton, R. J. (2004) Democratic Challenges, Democratic Choices: The Erosion of Political Support in Advanced Industrial Democracies. Oxford: Oxford University Press. Finkel, S. E., Muller, E. N. and Seligson, M. A. (1989) ‘Economic Crisis, Incumbent Performance and Regime Support:A Comparison of Longitudinal Data from West Germany and Costa Rica’, British Journal of Political Science, 19 (3), 329–51. © 2007 The Authors. Journal compilation © 2007 Political Studies Association POLITICAL STUDIES: 2007, 55(2)


441

Franz, G. (1986) ‘Economic Aspirations,Well-Being and Political Support in Recession and Boom Periods: The Case of West Germany’, European Journal of Political Research, 14 ( January), 97–112. Fuchs, D., Guidorossi, G. and Svensson, P. (1995) ‘Support for the Democratic System’, in H.-D. Klingemann and D. Fuchs (eds), Citizens and the State. New York: Oxford University Press, pp. 323–53. Gidengil, E., Blais, A., Nadeau, R. and Nevitte, N. (1999) ‘Making Sense of RegionalVoting in the 1997 Federal Election: Liberal and Reform Support Outside Quebec’, Canadian Journal of Political Science, 32 (2), 247–72. Ginsberg, B. and Weissberg, R. (1978) ‘Elections and the Mobilization of Popular Support’, American Journal of Political Science, 22 (1), 31–55. Himmelweit, H., Biberian, M. and Stockdale, J. (1978) ‘Memory for Past Vote: Implications of a Study of Bias in Recall’, British Journal of Political Science, 8 (3), 365–75. Johnston, R., Blais, A., Brady, H. E. and Crete, J. (1992) Letting the People Decide: Dynamics of a Canadian Election. Stanford CA: Stanford University Press. Johnston, R., Blais, A., Brady, H. E., Gidengil, E. and Nevitte, N. (1996) ‘The 1993 Canadian Election: Realignment, Desalignment or Something Else?’. Paper presented at the annual meeting of the American Political Science Association, San Francisco. Joslyn, M. R. (1999) ‘Opinion Change after the Election’. Paper presented at the annual meeting of the American Political Science Association, Boston MA, September. Klingemann, H.-D. (1999) ‘Mapping Political Support in the 1990s:A Global Analysis’, in P. Norris (ed.), Critical Citizens: Global Support for Democratic Governance. New York: Oxford University Press, pp. 31–56. Kornberg, A. and Clarke, H. (1994) ‘Beliefs about Democracy and Satisfaction with Democratic Government: The Canadian Case’, Political Research Quarterly, 47 (September), 537–63. Linde, J. and Ekman, J. (2003) ‘Satisfaction with Democracy: A Note on a Frequently Used Indicator in Comparative Politics’, European Journal of Political Research, 42 (3), 391–408. Listhaug, O. (1995) ‘The Dynamics of Trust in Politicians’, in H.-D. Klingemann and D. Fuchs (eds), Citizens and the State. New York: Oxford University Press, pp. 261–98. Listhaug, O. and Wiberg, M. (1995) ‘Confidence in Political and Private Institutions’, in H.-D. Klingemann and D. Fuchs (ed.), Citizens and the State. New York: Oxford University Press, pp. 298–323. Muth, J. A. (1961) ‘Rational Expectations and the Theory of Price Movements’, Econometrica, 29 (6), 315–35. Nadeau, R. and Blais, A. (1993) ‘Accepting the Election Outcome: The Effect of Participation on Losers’ Consent’, British Journal of Political Science, 23 (October), 553–63. Nevitte, N., Blais, A., Gidengil, E. and Nadeau, R. (2000) Unsteady State: The 1997 Canadian Federal Election. Oxford: Oxford University Press. Norris, P. (1999a) ‘Introduction: The Growth of Critical Citizens?’, in P. Norris (ed.), Critical Citizens. New York: Oxford University Press, pp. 1–30. Norris, P. (1999b) ‘The Political Regime’, in H. Schmitt and J. Thomassen (eds), Political Representation and Legitimacy in the European Union. Oxford: Oxford University Press, pp. 257–72. Sargent, T. J. (2002) ‘Rational Expectations’, in D. R. Henderson (ed.), The Concise Encyclopedia of Economics. Available from: http://www.econlib.org/library/Enc/RationalExpectations.html [Accessed 23 February 2005]. Weatherford, S. (1984) ‘Economic “Stagflation” and Public Support for the Political System’, British Journal of Political Science, 14 (April), 187–205. Weatherford, S. (1987) ‘How Does Government Performance Influence Political Support?’, Political Behavior, 9 (1), 5–28.

© 2007 The Authors. Journal compilation © 2007 Political Studies Association POLITICAL STUDIES: 2007, 55(2)