Dynamics of Democratic Satisfaction in Transitional Settings

Abstract

How do campaigns and elections affect evaluations of democracy in transitional settings? Little research has been conducted on this topic in transitional settings, where citizens’ evaluations of regime performance might be especially fluid and where popular dissatisfaction with democratic institutions could inhibit political development. Furthermore, previous studies’ reliance on cross-sectional data has made it difficult to make valid causal inferences in this area. Drawing upon unique data from the 2010–2011 Uganda Campaign Panel Study, the first panel on campaign effects in Africa, we measure changes in individuals’ satisfaction with democracy. We find that individuals who supported a losing presidential candidate are more likely to show decreases in satisfaction over the course of the campaign than those who supported the winner. Importantly, we find that previous political participation does not necessarily result in increases in democratic satisfaction: rather, it is associated with decreased satisfaction for losers. Finally, there is no evidence that declines in satisfaction with democracy are associated with decreased support for democracy as a regime. This study furthers understanding of political development by identifying factors that affect how citizens evaluate their political regime.

Keywords

Losers’ consent satisfaction with democracy participation African politics democratic transitions

Recent years have seen significant growth in the literature on nominally democratic institutions.¹ Even hybrid or authoritarian regimes frequently have parties and legislatures, which ostensibly allow rulers to co-opt adversaries (Blaydes 2011; Gandhi 2008; Gandhi and Przeworski 2007; Magaloni 2006, 2008; Svolik 2012). Such institutions do not mean that a country is on an inexorable march toward full democracy, although they might increase the likelihood of democratic consolidation, were a regime change to occur (Brownlee 2009; Bunce and Wolchik 2011; Howard and Roessler 2006).

Despite increased attention, there is little scholarship on citizens’ evaluations of nominally democratic institutions. Theoretically, citizen satisfaction could impact institutions’ survival, as incumbents’ willingness to maintain them might diminish if popular support dissipates. In short, satisfaction with the performance of such institutions could be critical. However, although studies have examined the determinants of satisfaction in new democracies (Anderson et al. 2005; Anderson and Tverdova 2001; Curini et al. 2012; Dahlberg, Linde, and Holmberg, forthcoming; Doorenspleet 2012), very few have included transitional countries. Such systems, which are defined as having undergone recent political reforms while not achieving democratic status, could be particularly vulnerable to institutional collapse. However, popular satisfaction with nominally democratic institutions could lead citizens to defend their existence and perhaps to demand more meaningful participation and accountability. In short, there is a particular need to study determinants of attitudes about institutions in less-than-fully-democratic settings because of the potential for democratic development or decay there, yet these are the very cases in which our understanding of regime satisfaction is most limited.

Here, we focus on how elections in transitional settings impact satisfaction with democratic institutions. First, we hypothesize that electoral outcomes in transitional settings will function much as others have found they do in more competitive ones: supporters of losing candidates will have lower satisfaction after the election than they did previously (see, inter alia, Anderson et al. 2005). This relationship will hold despite the fact that losing should hardly be surprising to opposition supporters in such environments, where incumbents have often had serious advantages for years; the psychological effect of losses might be cumulative, elections might highlight the only-nominal status of democratic institutions, and opposition supporters’ expectations for change might have been raised by the fervent campaigning that often precedes elections.

Second, unlike others who have studied participation, we expect that its effects on satisfaction will be contingent on political status. Although existing theory posits that participation might have myriad democracy-enhancing benefits, we anticipate that, for losers, higher levels of participation will be associated with decreased, rather than increased, satisfaction. More-participatory losers experience greater psychological costs than their less-involved counterparts.

We test these hypotheses with panel data collected in Uganda surrounding the 2010–2011 campaign. Uganda fits the “transitional” classification well, in that it has recently undergone significant political reforms—it only began allowing parties to compete in elections in 2006, after a two-decade ban—whereas most observers characterize the country as facing serious democratic deficits (Tripp 2010).

Our results indicate that, over the course of the study period, important inter- and intraindividual variation in Ugandans’ democratic satisfaction existed. As expected, voting for a losing presidential candidate is associated with significant declines in satisfaction. The result holds for individuals who switched to an opposition candidate during the campaign, as well as for those who remained opposition supporters throughout. The latter finding is important, in that it suggests that the estimated relationship between loser status and satisfaction cannot be explained by individuals’ voting against the incumbent (i.e., being a loser) because of their declining satisfaction. In addition, we find that the effects of participation on satisfaction are conditional: for losers, increased participation is associated with larger decreases in satisfaction. For winners, participation has no estimated effect.

Our paper makes a number of important contributions to the study of regime evaluations. First, our attention to a transitional setting is unique; while other studies have included transitional settings among their cases, ours is the first of which we are aware to focus specifically on how elections might impact satisfaction in regimes with nominally democratic institutions. Second, unlike extant literature, we examine how the effects of participation on satisfaction differ according to political status. Our findings of conditionality suggest that previous works, which have generally been characterized by inconsistent or null results, warrant re-examination. Finally, we make a methodological contribution in our use of panel data in a transitional setting. Studies of democratic satisfaction generally have relied on data collected at one time point or, in some cases, at various time points across different samples. However, it is impossible with such data to measure how individuals’ assessments of regime performance change over time. In addition, cross-sectional data make valid causal claims difficult (Norris 2014; Singh, Karakoç, and Blais 2012). Some have attempted to address individual-level change and endogeneity using panel data (Banducci and Karp 2003; Blais and Gélineau 2007; Curini et al. 2012; Singh, Karakoç, and Blais 2012), but, with the exception of Esaiasson (2011), who includes Russia among his cases, these studies have been conducted exclusively in democracies. Our panel data allow us to identify causal relationships between political status, participation, and regime evaluations in transitional settings more precisely.

Determinants of Democratic Satisfaction

The late David Easton’s (1965, 1975) contributions on the importance of diffuse support spurred a cottage industry of research on popular attitudes about regimes. One of the most widely studied dimensions has been democratic satisfaction (Norris 1999, 2011), which refers to citizens’ evaluations of the performance of institutions associated with democracy, such as parties, legislatures, and elections. This concept, which is distinct from support for democratic principles, is considered central to institutional stability and development (Almond and Verba 1963; Diamond 1999; Linz and Stepan 1996).

There is a large literature on the determinants of democratic satisfaction in established (Anderson and Guillory 1997; Banducci and Karp 2003; Bernauer and Vatter 2012; Blais and Gélineau 2007; Craig et al. 2006; Curini et al. 2012; Singh, Karakoç, and Blais 2012; Wagner, Schneider, and Halla 2009) and new democracies (Anderson et al. 2005; Anderson and Tverdova 2001; Chang, Chu, and Wu 2014; Curini et al. 2012; Dahlberg, Linde, and Holmberg, forthcoming; Doorenspleet 2012). However, the literature identifies other cases—transitional regimes—that are like new democracies, in that they have recently undergone major political reforms, but which are distinct in that they are not considered to meet fundamental democratic criteria (Anderson et al. 2005).² Such settings are likely to have democratic institutions, but their status as hybrid or authoritarian regimes demonstrates that these institutions do not provide for level political competition or respect for individual rights.

Despite the prevalence of such transitional cases—they are particularly common in Africa, Latin America, and the former Soviet Union—few scholars of democratic satisfaction have focused on them, and those who have include them with democracies in their studies (Anderson et al. 2005; Bratton and Chang 2006; Bratton et al. 2005; Esaiasson 2011; Greenberg and Mattes 2013; Mattes 2014; Mattes and Bratton 2007; Norris 2011, 2014; Singh 2014). We argue that there is a need for greater study of these cases specifically, given their prevalence and the often-fragile nature of nominally democratic institutions there. If popular satisfaction with these institutions declines, elites may be tempted to abolish them.

Elections and the campaigns preceding them should be obvious focal points in attempts to discern the determinants of satisfaction with democracy. First, they act as separating mechanisms, dividing the population into winners and losers.³ This separation may, in turn, affect satisfaction via psychological (i.e., winners receive an emotional high from the results, whereas losers have the opposite response) or pragmatic (i.e., winners expect favorable policies or distributions, whereas losers anticipate deleterious consequences) pathways. And second, campaigns mark periods in which citizens’ interactions with political actors are most frequent and sustained. They also often mobilize participation, by providing opportunities for citizens to influence politics. Consequently, when examining effects of elections on satisfaction, there are advantages to a multifaceted approach, studying (1) how citizens respond to outcomes, as well as (2) their attempts to shape politics. Attention to the latter is missing from most studies of satisfaction. In the following sections, we hypothesize how these two aspects of campaigns and elections—their creation of winners and losers and citizens’ increased opportunities for political involvement—affect democratic satisfaction.

Creating Winners and Losers

There is an extensive literature on how support of a winning or losing candidate is correlated with institutional satisfaction in democracies (Anderson and Guillory 1997; Anderson and Tverdova 2001; Banducci and Karp 2003; Beaudonnet et al. 2014; Blais and Gélineau 2007; Craig et al. 2006; Enyedi 2014; Singh, Karakoç, and Blais 2012), as well as some studies that include both democratic and transitional cases (Anderson et al. 2005; Curini et al. 2012; Dahlberg, Linde, and Holmberg, forthcoming; Esaiasson 2011; Howell and Justwan 2013; Mattes and Bratton 2007; Norris 2014; Singh 2014). In most instances, winners demonstrate significantly higher satisfaction than losers.

Scholars have posited two main explanations for such gaps. First, elections are emotionally charged, and their outcomes might generate psychological effects (Anderson et al. 2005). The positive feelings winners experience might promote satisfaction, whereas losers’ disappointment might generate cynicism. Second, losers might express greater dissatisfaction with democracy (relative to winners) because they believe that the government is less likely to pursue their preferred policies (Curini et al. 2012).

However, the extant literature faces two significant limitations, in terms of its application to transitional settings. First, there is a paucity of theoretical attention to why election outcomes might generate shifts in satisfaction in situations with nominally democratic institutions. Given often long-standing incumbent stability in such environments, we might not expect a particular election to have much influence on satisfaction. And second, data used for such studies are cross-sectional, which makes it impossible to determine whether losses from a particular election generate larger gaps in satisfaction between groups than had existed previously. This is especially important in environments with nominally democratic institutions, where turnover is rare. It is possible that winner–loser gaps in such situations occur not primarily because of the outcome of a studied election, but because disenchantment among a particular group existed prior to the election, for reasons such as ideological or programmatic differences with the incumbent, or perceived discrimination at the hands of the government. In other words, significant winner–loser gaps in such environments are not necessarily evidence that elections decrease satisfaction among those who voted for the losing side.

A priori, we might expect that election outcomes actually do not cause significant changes in satisfaction in settings with nominally democratic institutions. Given the structural biases against their side, and a likely multielection record of losses, losing yet again should not surprise opposition supporters,⁴ nor should it cause them to re-evaluate their political standing significantly. According to this perspective, opposition adherents’ satisfaction is likely to be low before the election and remain unchanged after.

However, election losses might cause decreased satisfaction in such settings, for at least three reasons. First, the effect of repeated losses might be cumulative (Curini et al. 2012); even if the nth defeat is largely expected, it still might generate higher levels of dissatisfaction than n − 1 defeats had. Second, elections and the campaigns preceding them might constitute focal points, reminding opposition supporters of the only-nominal nature of democratic institutions or highlighting previously unrecognized flaws. Finally, the campaign period, with the concomitant crowds at rallies and ubiquitous posters, might raise opposition supporters’ expectations that public support is in their favor, and that victory is now actually possible. A loss will thereby result in a significant drop in satisfaction. From this perspective, we can hypothesize as follows:

Hypothesis 1 (H1): Individuals who voted for a losing candidate (“losers”) will report decreased satisfaction with democracy, relative to those who voted for the winner (“winners”).

Political Participation

In studying the determinants of satisfaction, it is important not just to consider how individuals respond to political events but also how they try to shape them. In other words, participation might also affect satisfaction. In line with the existing literature, we define participation as the extent to which citizens are engaged in activities for political expression (Norris 2002). The theoretical literature often extols the democracy-enhancing effects of participation (Almond and Verba 1963; Brehm and Rahn 1997; Ikeda, Kobayashi, and Hoshimoto 2008; Lindberg 2006; Norris 2002; Putnam 1993), with Bratton et al. (2005) contending that participation shapes the formation of regime evaluations. Participants might gain greater cognitive awareness—through increased knowledge about public affairs and democratic principles—and a heightened sense of efficacy, which may in turn boost regime satisfaction (Bratton et al. 2005; Finkel 1987).

However, participation has only factored into a handful of studies on satisfaction (Blais and Gélineau 2007; Bratton and Chang 2006; Bratton et al. 2005; Curini et al. 2012; Dahlberg, Linde, and Holmberg, forthcoming; Doorenspleet 2012; Mattes 2014; Singh 2014; Singh, Lago, and Blais 2011). We expand on this literature in four important ways. First, most of these studies examine the effects of participation in democracies, with only a small number including transitional settings among their cases (Bratton et al. 2005; Mattes 2014; Singh 2014). Next, all use cross-sectional rather than panel data, which makes it difficult to establish causality; a correlation between participation and satisfaction does not necessarily indicate that participation causes higher or lower satisfaction, as individuals might adjust their participation rates because of their pre-existing level of satisfaction. Third, many of these studies focus only on one type of participation: voting.⁵ Given that that particular activity is often nonautonomous in transitional settings (Bratton et al. 2005), it is particularly important to study myriad possible actions. Our study includes a wider range of activities than any other.

Finally, existing studies do not consider that the effects of participation on satisfaction might be conditional on political status. One view would hold that participation increases satisfaction, for winners and losers alike, by providing individuals with a sense of ownership of institutions and inculcating greater trust. We theorize, however, that the effects of participation on satisfaction will vary depending on winner/loser status. We expect that greater participation will be associated with decreases in satisfaction, rather than increases, among losers. Two arguments drive our expectation. First, active losers might be more psychologically affected than nonactive ones by the election results, given the commitments they have made to politics in time, energy, and resources. Second, active losers might be more aware of antiopposition abuses that occurred during the campaign and thus express greater dissatisfaction with the regime. Therefore, we hypothesize as follows:

Hypothesis 2 (H2): As political participation increases, losers’ democratic satisfaction will decrease.

Case

Given our interest in how elections affect satisfaction in transitional environments, we test our hypotheses in a setting in which significant political reform has recently occurred, yet in which institutions are only nominally democratic.⁶ The 2011 election in Uganda fits these criteria well. First, that election marked only the second multiparty presidential balloting in Ugandan history. A 2005 referendum reintroduced multiparty elections after a nearly twenty-year experiment with “no-party democracy” (Carbone 2008).

Next, recent years have been marked by crackdowns on journalists and media houses, significant resource advantages for the ruling National Resistance Movement (NRM), and politicized control of electoral management bodies (Conroy-Krutz and Logan 2012; Mwenda and Tangri 2005; Tripp 2010). President Yoweri Museveni had won the previous three elections by margins of between twenty-two and fifty-one points, and the opposition’s inability to field a consensus candidate in 2011 made turnover seem further out of reach. Freedom House has never rated Uganda lower than a 4.0; it was considered “partly free” at the time of the 2011 election.

There are reasons to expect a priori that elections will not result in significant changes in satisfaction in Uganda. Long-time opposition supporters were unlikely, by 2011, to be surprised by yet another loss. Museveni has been in power since 1986 and has never been seriously threatened electorally. Heading into 2011, he faced a divided opposition—an attempt at coordination had collapsed—while the economy was enjoying robust growth and the brutal Lord’s Resistance Army had recently been exiled. In short, we might expect that opposition supporters’ satisfaction would be relatively low, but be unaffected by another defeat.

However, long-time losers might still have seen satisfaction decline further. Although they were unlikely to believe that their policy or distributional interests would be further weakened by another loss—they likely already believed that they were not being represented—the cumulative psychological effects of losing a fourth straight time to Museveni might have been significant enough to catalyze greater dissatisfaction. In short, a significant decline even in a case like Uganda would offer support—albeit indirect—for psychological explanations for winner–loser gaps in satisfaction.

Data

As previously discussed, panel data are important in measuring more precisely how elections affect satisfaction, as well as in alleviating identification problems. We use original data from the Uganda Campaign Panel Study (UCPS), which included three waves: one before campaigning had begun in earnest (November 2010), one late in the campaign (January–February 2011), and one after voting was completed (March–April 2011). From nine sampled parliamentary constituencies, 660 individuals completed all three waves. Details, including sampling procedures and attrition rates, are available in Online Appendix A (http://prq.sagepub.com/supplemental/).

Dependent Variable

We measure satisfaction with the following: “Overall, how satisfied are you with the way democracy works in Uganda? Are you very satisfied, fairly satisfied, not very satisfied, or not at all satisfied?”⁷ This question was asked in Waves 1 and 3. The variable is coded 0 to 3, with higher values indicating greater satisfaction. The mean value declined significantly between Waves 1 (2.03) and 3 (1.91) (p = .00).

We create a variable measuring change in satisfaction by subtracting the wave 1 from the wave 3 measure. Δ Dem Satisfaction ranges from −3 to 3, with positive values representing increased satisfaction and negative values representing decreased satisfaction. A plurality (44%) registered no change, while 34 percent and 23 percent reported decreases and increases, respectively.

Winner/Loser Status

To test H1, we generate a dummy—Loser—in which respondents who reported voting for any of the seven unsuccessful presidential candidates are coded as 1,⁸ with all others coded as 0. We also generate a dummy—Abstain—to indicate nonvoters.⁹ Sixteen percent of respondents are coded as losers, whereas the same proportion is coded as abstainers.¹⁰ The reference category is winners.¹¹

Participation

We measure participation through questions, asked in wave 2,¹² in which respondents indicated whether they had recently attended (1) a local government meeting, (2) protest, or (3) campaign rally; voiced their attitudes to a (4) newspaper, (5) Internet site, or (6) radio station; talked about politics with a (7) local government official, (8) Member of Parliament, or (9) traditional leader; and advocated for a favorite candidate (10) to family and friends, or (11) to strangers, and their frequency of participating in these activities. We create a scale from the standardized variables (Participation),¹³ indicating the extent of each respondent’s participation.¹⁴

An initial look at the data lends support to our hypotheses (Table 1). Panel 1 reports changes in satisfaction by political status. As expected, winners had the highest satisfaction in the postelection survey (2.11, out of 3), whereas losers had the lowest (1.43).¹⁵ Losers experienced significant declines (p = .01), and their mean in wave 3 was 14.4 percent lower than it had been in wave 1.¹⁶ Forty-six percent of losers saw their satisfaction decline, and the mean change for this group was −.25 points. These results lend initial support to H1.

Table 1.

Change in Democratic Satisfaction (Wave 1 to Wave 3), by Group.

	n	Wave 1	Wave 3	Δ%	p
Panel 1: By presidential vote
Nonvoters	88	1.80	1.64	−8.9	.23
Winners	424	2.20	2.11	−4.1	.04
Losers	101	1.67	1.43	−14.4	.01
Panel 2: By activity level
Lower activity group	319	1.97	1.81	−8.1	.01
Higher activity group	313	2.13	2.01	−5.6	.02
Panel 3: By vote and activity level
Nonvoters
Lower activity	60	1.72	1.48	−14.0	.19
Higher activity	28	1.96	1.96	0	NA
Winners
Lower activity	206	2.11	2.00	−5.2	.09
Higher activity	218	2.28	2.20	−3.5	.23
Losers
Lower activity	42	1.62	1.45	−10.5	.23
Higher activity	59	1.71	1.41	−17.5	.01

Results presented only for respondents who answered questions regarding democratic satisfaction in waves 1 and 3. Last column represents results from Wilcoxon signed ranks tests of interwave differences by group.

Next, we examine changes in satisfaction by activity level (panel 2). To create two roughly equal groups, we divide our sample between those with Participation scores below (“lower activity”) and above (“higher activity”) the median. Both groups saw their mean satisfaction decline significantly. At the individual level, nearly equal numbers of lower (34%) and higher activity (34%) respondents saw their satisfaction decline, and their mean declines were statistically indistinguishable: −.16 and −.12 points for lower and higher activity respondents, respectively (p = .69). On its face, this suggests that participation does not affect changes in satisfaction.

Finally, panel 3 reports changes in satisfaction by activity level and political status. We see evidence of a meaningful interaction: higher activity winners saw no changes in satisfaction, whereas lower activity winners saw a slight decline (p = .09). For losers, however, participation seems to have made satisfaction sink even further. As predicted, higher activity losers saw a large decrease in their mean satisfaction (17.5%), an interwave difference that is statistically significant at p = .01. Less-active losers, however, saw no significant decrease in satisfaction. In short, we find evidence supporting H2.

Analysis

To conduct more systematic tests, we estimate a series of ordinary least squares (OLS) models in which Δ Dem Satisfaction is regressed on measures of political status and participation. We also interact Participation with Loser and Abstain separately to test H2. We expect the coefficient on Loser will be significant and negative, indicating that individuals who voted for the opposition saw larger decreases in satisfaction than those who voted for the victorious incumbent (the reference group). In addition, we expect the coefficient on the interaction between Participation and Loser will be negative, indicating that increased participation among losers further depresses satisfaction.

We also control for three sets of factors that might affect satisfaction.¹⁷ First, given that others have argued electoral integrity affects regime evaluations (Birch 2011; Bratton and Houessou 2014; Craig et al. 2006; Greenberg and Mattes 2013; Mattes 2014; Norris 2014; Rose and Mishler 2009), we control for reported experience with two kinds of malpractice: resource distributions and violence. Respondents in wave 3 were asked whether they had witnessed candidates making distributions during the campaign and, if so, to assess the number of people in the village or neighborhood receiving them. We create Distribution, in which 0 represents no reported distributions, 1 “only one or two people” being recipients, 2 “a few people,” and 3 “a lot of people.”¹⁸ A dummy (Violence) indicates whether the respondent reported local election-related violence.¹⁹

Second, likelihood of attitudinal change might be affected by political sophistication—namely, we would expect that more-sophisticated individuals would be less likely to change their assessment because of their stronger precampaign dispositions. We include a variable controlling for Political Knowledge,²⁰ as well as measures of Political Interest,²¹ and frequency of accessing news from Radio, Television, and Newspaper.²²

Finally, we test an alternate set of explanations for changes in satisfaction: changes in economic outlook or access to services. First, improvements or declines in economic outlook could result in increases or decreases, respectively, in satisfaction. We thus include Δ Economic Outlook, which is calculated by subtracting a respondent’s economic outlook at wave 1 from outlook at wave 2. Assessments range from 0 (“very bad”) to 4 (“very good”). Mean assessments did not change significantly between waves 1 (2.66) and 2 (2.68) (p = .75). Forty-three percent did not change their assessment, whereas an additional 30 percent only increased or decreased their assessment by one point. Second, improved access to services, which was reportedly a focus of the government in the election’s run up (Conroy-Krutz and Logan 2012), could increase satisfaction. We, therefore, include an index of the number of recent improvements in services that the respondent reported in his or her area.²³

The models also include a measure of the dependent variable at wave 1 (Dem Satisfaction W1) to deal with concerns over regression to the mean (Finkel 1995, 8; Markus 1979), as well as controls for respondent sex (Female), age (Age), urban/rural residency (Urban), household wealth (Wealth),²⁴ and completion of primary school (Completed Primary).

First, we run a baseline model without interactions, to get a sense of the unique relationships between our variables of interest and satisfaction. Results are reported in Table 2 (model 1). First, Loser is associated with significant declines in satisfaction relative to changes experienced by those who supported the winner (the excluded category).

Table 2.

Determinants of Changes in Regime Satisfaction.

	Model 1	Model 2	Model 3
Loser	−.54 (.11)***		−.27 (.16)*
Opp constant		−.59 (.12)***
Opp new		−.44 (.13)***
Abstain	−.31 (.12)**		−.39 (.20)**
Abstain constant		.02 (.15)
Abstain new		−.56 (.14)***
Participation	.00 (.01)	.00 (.01)	.01 (.01)
Participation × Loser			−.03 (.01)**
Participation × Abstain			.02 (.04)
Violence	−.09 (.10)	−.10 (.10)	−.08 (.10)
Distribution	−.04 (.03)	−.04 (.03)	−.04 (.03)
Political knowledge	−.02 (.02)	−.01 (.02)	−.02 (.03)
Political interest	.06 (.04)	.06 (.04)	.06 (.04)
Radio exposure	−.07 (.03) *	−.06 (.03)*	−.07 (.03)*
Television exposure	−.02 (.05)	−.03 (.05)	−.02 (.05)
Newspaper exposure	−.01 (.07)	−.01 (.07)	−.01 (.07)
Δ Economic outlook	−.02 (.03)	−.02 (.03)	−.02 (.03)
Service improvements index	.09 (.04)**	.09 (.03)**	.09 (.04)**
Age	.00 (.00)	.00 (.00)	.00 (.00)
Female	.09 (.07)	.09 (.07)	.07 (.07)
Urban	−.37 (.08)***	−.37 (.08)***	−.38 (.08)***
Wealth	−.01 (.03)	−.01 (.03)	−.01 (.03)
Completed primary	.22 (.05)**	.22 (.08)**	.20 (.08)*
Democratic satisfaction W1	−.94 (.05)***	−.93 (.05)***	−.95 (.05)***
Constant	1.95 (.23)***	0.90 (.23)***	1.98 (.23)***
n	518	518	518
R ²	.50	.51	.51

Clustered robust standard errors by enumeration area (the PSU) (N = 90). Opp = opposition.

p < .10. **p < .05. ***p < .01, for two-tailed tests.

These results do not demonstrate that winner/loser status is causing changes in satisfaction, however. It is also plausible that individuals’ satisfaction declined for reasons unrelated to political status, and that this caused them to vote against the incumbent (i.e., the winner). To check this, we split the variable Loser into opposition voters who had indicated intention to vote for the opposition in wave 1 (Opp Constant) and those who had earlier indicated that they either would not vote, did not know how they would vote, or would vote for the incumbent (Opp New). If the relationship between Opp Constant and the dependent variable is negative and significant, we can be more confident that political status is driving changes in satisfaction, rather than vice versa.

When we include these two dummies in place of Loser (model 2), the coefficient on Opp Constant is negative and significant. In other words, even opposition voters who had supported an opposition candidate in the first wave experienced declines in satisfaction. Importantly, declines in satisfaction are not likely to explain why these respondents ended up voting for the opposition (i.e., as losers).²⁵

Returning to model 1, the coefficient on Abstain is also negative and significant, indicating that individuals who did not vote saw larger satisfaction declines than winners. As with Loser, however, this result does not demonstrate causality, as individuals whose satisfaction declined might have decided not to vote as a result. Therefore, we again create two dummies, one indicating the respondent did not plan to vote in wave 1 and actually did not vote (Abstain Constant), and another indicating the respondent planned to vote in wave 1 but did not actually vote (Abstain New). Again, if the coefficient on Abstain Constant is negative and significant, these endogeneity concerns should be mitigated. Here (model 2), the coefficient for Abstain New is negative and significant, whereas that for Abstain Constant is not significant. Thus, we are not able to alleviate endogeneity concerns for any subset of the abstainers, and it remains possible that many of those who did not vote chose to do so because their satisfaction declined.

Next, the coefficient on Participation in model 1 is not significant. However, in model 3 we add terms in which we interact Participation with our status dummies. Participation remains insignificant, whereas Loser remains negative and significant. Importantly, the result for the interaction between Participation and Loser is negative and significant. This indicates that, as participation increases, the negative effect of having voted for a loser on changes in satisfaction increases, as well. To put it differently, losers who were more involved expressed greater decreases in satisfaction than losers who were less engaged.

Figure 1 displays this relationship. At the lowest levels of participation, changes in satisfaction between losers and winners are indistinguishable. As participation increases, losers experience significant declines in satisfaction. However, greater participation does not result in significant changes in satisfaction among winners.

Figure 1.

Effects of Political Status, Conditional on Participation.

Finally, we consider the controls. First, the consistently negative and significant coefficient on Dem Satisfaction W1 suggests a potential ceiling effect, in that higher early-campaign levels of satisfaction are associated with decreases in satisfaction between waves 1 and 3 (i.e., a regression to the mean). We find no evidence that electoral malpractice, as measured here, affected satisfaction: neither Violence nor Distribution is significant in any model. Increases in Service Improvements are associated with significant increases in satisfaction across all three models, suggesting that investments in respondents’ local areas—usually undertaken by the government—improved regime evaluations. Among the media access measures, only Radio is statistically significant: across all three models, individuals who reported greater frequency of accessing the news through the radio experienced larger decreases in satisfaction. Completed Primary is positive and significant across all models; better-educated respondents saw larger increases in their satisfaction. And urbanites saw larger decreases in satisfaction than their rural counterparts across models. No other controls are significant in any of the models.

Our main results are robust when controlling for partisanship and voting as an activity and after checking for attrition bias. First, scholars have found that the effects of winning and losing may be mediated by partisanship, with individuals who feel close to a party having stronger responses to election outcomes than those who do not (Anderson et al. 2005, 78–80). To test this, we generate indicators of partisan attachment, based on respondents’ favored party and their level of trust of that group.²⁶ We then rerun model 1, including two dummies—strong partisans and weak partisans, with nonpartisans as the reference—which are interacted with our indicators of political status. If strong partisanship has an “amplifying” effect, to use the terminology of Anderson et al. (2005), we should expect a negative coefficient on the interaction between the strong-partisan and loser dummies (i.e., strongly partisan losers’ satisfaction decreases more than nonpartisan losers’).²⁷ The coefficient on Loser remains negative and significant, even when controlling for partisanship, and the coefficient on the indicator for strong partisanship is positive and significant, which suggests that individuals who were more attached to parties had generally larger increases in satisfaction than those who were not (Online Appendix G). However, none of the interactions are significant, meaning that we cannot say that partisan losers saw changes in satisfaction different from those experienced by nonpartisan losers.²⁸

Next, some have found that voting can affect regime evaluations (Ginsberg and Weissberg 1978). Examining this requires we use alternate measures of political status, as our categories of winners and losers only include voters. In the absence of a measure of preferred victor, we use partisanship as a rough proxy. NRM partisans are treated as winners, and those who supported an opposition party as losers. We rerun models 1 to 3 using an indicator for opposition support (Opposition) in place of Loser and one for nonpartisanship (Nonpartisan) in place of Abstain (Online Appendix H). In addition, we measure whether or not the respondent voted in the presidential election (Voted). The results are similar to those in the original analyses, with Opposition associated with satisfaction decreases, and a negative and significant interaction between Opposition and Participation.²⁹ Voting, however, has no significant effect.

Finally, our results are also robust to checks for the possibility that individuals did not drop out of the panel randomly (Online Appendix I).

Ramifications for Support for Democracy

Last, we consider the broader ramifications of these changes in democratic satisfaction on support for democracy in Uganda. The literature is divided on the relationship between satisfaction and support, with some theoretical (Linz and Stepan 1996) and empirical (Stoker 2006) literature taking a pessimistic approach, contending that, when satisfaction with the performance of democratic institutions is low, citizens’ diffuse support for them will be low, as well. In Uganda, the fact that the largest decreases in satisfaction occur among individuals with previous capacity and inclination for participation might be particularly problematic for future democratic prospects, as politically active losers may lose faith in the political system and resort to nondemocratic methods, such as violence or support for military intervention, to alter the rules of the game.

Others suggest, perhaps more optimistically, that individuals might be dissatisfied with democratic institutions but still support their maintenance (Doorenspleet 2012; Gunther et al. 2007; Norris 2011). In fact, some have found higher rates of dissatisfaction might spur individuals to be even more supportive of democracy (Bratton et al. 2005; Moehler 2009; Qi and Shin 2011). If this relationship were to hold in Uganda, the fact that the largest decreases in satisfaction occur among the most politically active could augur well.

To measure how the election affected support for democracy, we draw upon questions in which respondents were asked how acceptable they would find (1) bans on organizations undermining government policies, (2) limits on speech, (3) military rule, and (4) one-party rule. Each is measured on a 5-point scale, with respondents indicating that they would reject the practice or regime “strongly” (4) or “somewhat” (3), “neither agree nor disagree” (2), or embrace the practice “somewhat” (1) or “strongly” (0). Higher values, therefore, indicate greater democratic support. The questions were asked in the first and third waves. We report changes in each variable, by political status, in Online Appendix J.

As in previous analyses, we run three models for each dependent variable: a baseline, one checking for endogeneity related to winner/loser status, and one including interactions between political status and participation. Results are presented in Table 3. We find no evidence that losing is associated with shifts against democratic principles. Rather, the coefficient on Loser is insignificant in three of the four baseline models (models 4, 7, & 10),³⁰ and losers experienced significantly larger increases in their rejection of single-party systems than winners did (model 13). This result holds when checking for endogeneity (model 14), although it drops below 90 percent significance (p = .138) when interacting political status with participation (model 15).³¹ Again, given the NRM’s status as a de facto single-party ruler during the “no-party democracy” period (1986–2006) and its subsequent status as a clearly dominant party, opposition voters’ increased rejection of rule by a single party makes sense.

Table 3.

Determinants of Changes in Support for Democratic Regimes.

	Antiorganization bans			Antispeech limits			Antimilitary rule			Anti single-party system
	4	5	6	7	8	9	10	11	12	13	14	15
Loser	.30 (.21)		.06 (.31)	.16 (.13)		.28 (.21)	−.05 (.20)		−.13 (.32)	.65*** (.19)		.47 (.31)
Opp constant		.32 (.26)			.06 (.21)			.26 (.27)			.92*** (.20)
Opp new		.28 (.25)			.12 (.18)			−.24 (.26)			.41* (.24)
Abstain	−.01 (.22)		.06 (.36)	.17 (.11)		.21 (.17)	.17 (.24)		.15 (.41)	.38 (.27)		.73 (.45)
Abstain constant		−.01 (.33)			.01 (.18)			.03 (.30)			.10 (.32)
Abstain new		−.11 (.19)			.27** (.12)			.17 (.30)			.58* (.29)
Participation	−.02 (.01)	−.02 (.01)	−.02 (.01)	.01 (.01)	.01 (.01)	.01 (.01)	−.02 (.01)	−.02 (.01)	−.02 (.02)	−.03* (.01)	−.03* (.01)	−.03 (.02)
Participation × Loser			.03 (.03)			−.02 (.02)			.01 (.04)			.02 (.03)
Participation × Abstain			−.02 (.05)			−.01 (.02)			.00 (.06)			−.08 (.06)
Constant	1.84*** (.40)	1.84*** (.40)	1.84*** (.41)	2.83*** (.25)	2.84*** (.25)	2.82*** (.26)	2.43*** (.43)	2.47*** (.43)	2.44*** (.44)	2.63*** (.41)	2.68*** (.42)	2.56*** (.41)
n	478	478	478	511	511	511	474	474	474	501	501	501
R ²	.50	.50	.50	.54	.54	.54	.47	.47	.47	.50	.51	.51

Clustered robust standard errors by enumeration area (the PSU) (N = 90). All models include controls for violence, distribution, age, sex, urban/rural setting, wealth, education, knowledge, political interest, media access, change in economic outlook, local service improvements, and wave 1 measure of the dependent variable. Opp = opposition.

p < .10. **p < .05. ***p < .01, for two-tailed tests.

In sum, even though we find that voting for a loser is associated with decreases in satisfaction—and that these declines are largest among the most politically active—these analyses suggest that losers are no more likely to also turn against democracy as a regime. These findings also highlight the possibility that elections and campaigns may have contrasting effects on different dimensions of political support, thereby supporting the more-optimistic literature on the topic: while losers may experience declines in some regime orientations, such as evaluations of democratic performance, other outcomes, including support for democratic principles, may be less affected. Although our data do not allow us to predict the effects of diminished satisfaction on future behavior—a potentially worthy topic for future research—these results provide some hope that such behavior would not be targeted toward dismantling democratic institutions.

Conclusion

Satisfaction is one of the most important components of regime support, but our understanding of the factors underpinning it is incomplete, especially in regard to contexts outside of democracies. We use panel data from Uganda, a country that recently transitioned to multiparty politics yet is still marked by institutions that are only nominally democratic, to measure how individuals’ satisfaction with democracy changed over the course of a campaign and election, and to identify the factors influencing changes that did occur. Our study offers an innovation in its use of panel data to study changes in satisfaction in such a setting, and doing so allows us to better identify causal relationships.

We make important methodological, theoretical, and empirical contributions in this paper. First, existing studies of satisfaction outside of established democracies have been primarily correlational, as they have used cross-sectional survey data to identify how various factors—primarily among them, winner/loser status—are related to satisfaction. Although these studies tend to find that winners have higher satisfaction than losers, they are generally unable to identify a causal effect of losing on lower satisfaction. Panel data allow us to identify a relationship between voting for a loser and experiencing declines in satisfaction. Furthermore, our finding that these declines occur even among individuals who were supporters of an eventual loser for the duration of the campaign suggests that the experience of losing induces declines in satisfaction.

Second, we make a theoretical contribution with our argument that predictions of participation leading to improved regime orientations need to be amended. Participation does not necessarily boost regime support. Rather, winners see no significant increases in satisfaction as their participation increases, whereas losers’ satisfaction declines as they participate more. In short, participation does not necessarily make citizens into contented democrats.

Finally, we also examined the effect of elections on support for democracy. Our findings suggest that although active losers experienced the greatest declines in satisfaction, their support for democracy was largely unaffected, and, in some ways, it increased. Losing did not lead opposition supporters to embrace illiberal alternatives.

These findings should provide some comfort for those concerned about democratic development in Uganda and similar transitional settings. For many theorists, transitional polities are more likely to establish robust democratic institutions when they develop legitimacy (Easton 1975; Linz and Stepan 1996). Our study suggests that although elections may depress satisfaction among political minorities, there are certain fundamental democratic values that members may consider essential and worthy of being defended.

Footnotes

Acknowledgements

Many thanks to all interviewers for their dedicated work on the panel survey, and especially to project manager John Kavuma, and Susan Mukasa, Virginia Nkwanzi-Isingiro, James Kakande, and Patrick Uma at Synovate Ipsos in Kampala. We thank Kate Baldwin, Allyson Benton, Emily Beaulieu, Michael Bratton, Jeremy Horowitz, Pippa Norris, Shane Singh, Jeffrey Wooldridge, and attendees at the 2014 Midwest Political Science Association Annual Meeting and July 2014 Electoral Integrity Project Workshop for helpful advice on the project.

Authors’ Note

The research was approved by Michigan State University’s Institutional Review Board (protocol 10-531) and by the Uganda National Council for Science and Technology.

Declaration of Conflicting Interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: Dr. Conroy-Krutz acknowledges support for the panel study from the National Science Foundation (SES-1024031) and the Center for Advanced Study of International Development at Michigan State University.

Notes

References

Almond

Gabriel

Verba

Sidney

. 1963. The Civic Culture. Princeton: Princeton University Press.

Anderson

Christopher

Blais

André

Bowler

Shaun

Donovan

Todd

Listhaug

Ola

. 2005. Losers’ Consent. New York: Oxford University Press.

Anderson

Christopher

Guillory

Christine

. 1997. “Political Institutions and Satisfaction with Democracy.” American Political Science Review 91 (1): 66–81.

Anderson

Christopher

Tverdova

Yuliya

. 2001. “Winners, Losers, and Attitudes about Government in Contemporary Democracies.” International Political Science Review 22 (4): 321–38.

Banducci

Susan

Karp

Jeffrey

. 2003. “How Elections Change the Way Citizens View the Political System.” British Journal of Political Science 33 (3): 443–67.

Beaudonnet

Laurie

Blais

André

Bol

Damien

Foucault

Martial

. 2014. “The Impact of Election Outcomes on Satisfaction with Democracy under a Two-Round System.” French Politics 12:22–35.

Bernauer

Julian

Vatter

Adrian

. 2012. “Can’t Get No Satisfaction with the Westminster Model?” European Journal of Political Research 51 (4): 435–68.

Birch

Sarah

. 2011. Electoral Malpractice. Oxford: Oxford University Press.

Blais

André

Gélineau

François

. 2007. “Winning, Losing and Satisfaction with Democracy.” Political Studies 55 (2): 425–41.

10.

Blaydes

Lisa

. 2011. Elections and Distributive Politics in Mubarak’s Egypt. New York: Cambridge University Press.

11.

Bratton

Michael

Chang

Eric

. 2006. “State Building and Democratization in Sub-Saharan Africa” Comparative Political Studies 39 (9): 1059–83.

12.

Bratton

Michael

Houessou

Richard

. 2014. “Demand for Democracy Is Rising in Africa, but Most Political Leaders Fail to Deliver.” Afrobarometer policy paper (http://afrobarometer.org/publications/pp11-demand-democracy-rising-africa-most-political-leaders-fail-deliver).

13.

Bratton

Michael

Mattes

Robert

Gyimah-Boadi

Emmanuel

. 2005. Public Opinion, Democracy, and Market Reform in Africa. Cambridge: Cambridge University Press.

14.

Brehm

John

Rahn

Wendy

. 1997. “Individual-Level Evidence for the Causes and Consequences of Social Capital.” American Journal of Political Science 41 (3): 999–1023.

15.

Brownlee

Jason

. 2009. “Portents of Pluralism.” American Journal of Political Science 53 (3): 515–32.

16.

Bunce

Valerie

Wolchik

Sharon

. 2011. Defeating Authoritarian Leaders in Postcommunist Countries. New York: Cambridge University Press.

17.

Carbone

Giovanni

. 2008. No-Party Democracy? Boulder: Lynne Rienner.

18.

Carothers

Thomas

. 2002. “The End of the Transition Paradigm.” Journal of Democracy 13 (1): 5–21.

19.

Chang

Eric

Chu

Yun-han

Wenchin

. 2014. “Consenting to Lose or Expecting to Win? Intertemporal Changes in Voters’ Winner-Loser Status and Satisfaction with Democracy.” In Elections and Democracy, edited by Tomassen

Jacques

, 232–53. New York: Oxford University Press.

20.

Conroy-Krutz

Jeffrey

Logan

Carolyn

. 2012. “Museveni and the 2011 Election.” Journal of Modern African Studies 50 (4): 625–55.

21.

Craig

Stephen

Martinez

Michael

Gainous

Jason

Kane

James

. 2006. “Winners, Losers, and Election Context.” Political Research Quarterly 59 (4): 579–92.

22.

Curini

Luigi

Jou

Willy

Memoli

Vincenzo

. 2012. “Satisfaction with Democracy and the Winner/Loser Debate.” British Journal of Political Science 42 (2): 241–61.

23.

Dahlberg

Stefan

Linde

Jonas

Holmberg

Sören

.2015. “Democratic Discontent in Old and New Democracies.” Political Studies 63(special issue1): 18-37.

24.

Diamond

Larry

. 1999. Developing Democracy. Baltimore: Johns Hopkins University Press.

25.

Doorenspleet

Renske

. 2012. “Critical Citizens, Democratic Support and Satisfaction in African Democracies.” International Political Science Review 33 (3): 279–300.

26.

Easton

David

. 1965. A Framework for Political Analysis. Englewood Cliffs: Prentice-Hall.

27.

Easton

David

. 1975. “A Re-assessment of the Concept of Political Support.” British Journal of Political Science 5 (4): 435–57.

28.

Enyedi

Zsolt

. 2014. “The Discreet Charm of Political Parties.” Party Politics 20 (2): 194–204.

29.

Esaiasson

Peter

. 2011. “Electoral Losers Revisited—How Citizens React to Defeat at the Ballot Box.” Electoral Studies 30 (1): 102–13.

30.

Finkel

Steven

. 1987. “The Effects of Participation on Political Efficacy and Political Support.” Journal of Politics 49 (2): 441–64.

31.

Finkel

Steven

. 1995. Causal Analysis with Panel Data. Beverly Hills: SAGE.

32.

Gandhi

Jennifer

. 2008. Political Institutions under Dictatorship. New York: Cambridge University Press.

33.

Gandhi

Jennifer

Przeworski

Adam

. 2007. “Authoritarian Institutions and the Survival of Autocrats.” Comparative Political Studies 40 (11): 1279–1301.

34.

Ginsberg

Benjamin

Weissberg

Robert

. 1978. “Elections and the Mobilization of Popular Support.” American Journal of Political Science 22 (1): 31–55.

35.

Greenberg

Ari

Mattes

Robert

. 2013. “Does the Quality of Elections Affect the Consolidation of Democracy?” In Voting and Democratic Citizenship in Africa, edited by Bratton

Michael

, 239–52. Boulder: Lynne Rienner.

36.

Gunther

Richard

Montero

José Ramón

Puhle

Hans-Jürgen

, eds. 2007. Democracy, Intermediation, and Voting on Four Continents. Oxford: Oxford University Press.

37.

Hardy

Melissa A.

1993. Regression with Dummy Variables. Thousand Oaks: SAGE.

38.

Howard

Marc Morjé

Roessler

Philip

. 2006. “Liberalizing Electoral Outcomes in Competitive Authoritarian Regimes.” American Journal of Political Science 50 (2): 334–43.

39.

Howell

Patrick

Justwan

Florian

. 2013. “Nail-Biters and No-Contests.” Electoral Studies 32 (2): 334–43.

40.

Ikeda

Ken’ichi

Kobayashi

Tetsuro

Hoshimoto

Maasa

. 2008. “Does Political Participation Make a Difference?” Electoral Studies 27 (1): 77–88.

41.

Izama

Angelo

Wilkerson

Michael

. 2011. “Museveni’s Triumph and Weakness.” Journal of Democracy 22 (3): 64–78.

42.

Klingemann

Hans-Dieter

. 1999. “Mapping Political Support in the 1990s: A Global Analysis.(pp.31-56)” In Critical Citizens, edited by Norris

Pippa

. New York: Oxford University Press.

43.

Lindberg

Staffan

. 2006. Democracy and Elections in Africa. Baltimore: Johns Hopkins University Press.

44.

Linz

Juan

Stepan

Alfred

. 1996. Problems of Democratic Transition and Consolidation. Baltimore: Johns Hopkins University Press.

45.

Magaloni

Beatriz

. 2006. Voting for Autocracy. New York: Cambridge University Press.

46.

Magaloni

Beatriz

. 2008. “Credible Power-Sharing and the Longevity of Authoritarian Rule.” Comparative Political Studies 41 (4/5): 715–41.

47.

Markus

Gregory

. 1979. Analyzing Panel Data. Beverly Hills: SAGE.

48.

Mattes

Robert

. 2014. “Electoral Integrity and Democratic Legitimacy in Africa.” In Advancing Electoral Integrity, edited by Norris

Pippa

Frank

Richard

Coma

Ferran Martínez i

, 211–28. Oxford: Oxford University Press.

49.

Mattes

Robert

Bratton

Michael

. 2007. “Learning about Democracy in Africa.” American Journal of Political Science 51 (1): 192–217.

50.

Mishler

William

Rose

Richard

. 2001. “What Are the Origins of Political Trust?” Comparative Political Studies 34 (1): 30–62.

51.

Moehler

Devra

. 2009. “Critical Citizens and Submissive Subjects.” British Journal of Political Science 39 (2): 345–66.

52.

Mwenda

Andrew

Tangri

Roger

. 2005. “Patronage Politics, Donor Reforms, and Regime Consolidation in Uganda.” African Affairs 104 (416): 449–67.

53.

Norris

Pippa

. 1999. Critical Citizens. New York: Oxford University Press.

54.

Norris

Pippa

. 2002. Democratic Phoenix. New York: Cambridge University Press.

55.

Norris

Pippa

. 2011. Democratic Deficit: Critical Citizens Revisited. New York: Cambridge University Press.

56.

Norris

Pippa

. 2014. Why Electoral Integrity Matters. New York: Cambridge University Press.

57.

Putnam

Robert

. 1993. Making Democracy Work. Princeton: Princeton University Press.

58.

Lingling

Shin

Doh Chull

. 2011. “How Mass Political Attitudes Affect Democratization.” International Political Science Review 32 (3): 245–62.

59.

Rose

Richard

Mishler

William

. 2009. “How Do Electors Respond to an ‘Unfair’ Election?” Post-Soviet Affairs 25(2): 118-36.

60.

Singh

Shane

. 2014. “Not All Election Winners Are Equal.” European Journal of Political Research 53 (2): 308–27.

61.

Singh

Shane

Karakoç

Ekrem

Blais

André

. 2012. “Differentiating Winners.” Electoral Studies 31 (1): 201–11.

62.

Singh

Shane

Lago

Ignacio

Blais

André

. 2011. “Winning and Competitiveness as Determinants of Political Support.” Social Science Quarterly 92 (3): 695–709.

63.

Stoker

Gerry

. 2006. Why Politics Matters. Basingstoke: Palgrave Macmillan.

64.

Svolik

Milan

. 2012. The Politics of Authoritarian Rule. New York: Cambridge University Press.

65.

Tripp

Aili Mari

. 2010. Museveni’s Uganda: Paradoxes of Power in a Hybrid Regime. Boulder: Lynne Rienner.

66.

Wagner

Alexander

Schneider

Friedrich

Halla

Martin

. 2009. “The Quality of Institutions and Satisfaction with Democracy in Western Europe—A Panel Analysis.” European Journal of Political Economy 25 (1): 30–41.