Taking political decisions necessarily generates winners and losers. The challenge in a democracy is to obtain losers’ consent. In the context of elections, losers need to “somehow, overcome any bitterness and resentment, and be willing (…) to accept the outcome of the election” (Anderson et al., Reference Anderson, Blais, Bowler, Donovan and Listhaug2005: 4). Obtaining losers’ consent after elections is not easy, however, as evidenced from a large body of literature that demonstrates important differences in satisfaction with democracy between winners and losers of elections (Dahlberg and Linde, Reference Dahlberg and Linde2017; Daoust and Nadeau, Reference Daoust and Nadeau2023). Winner-loser effects in a democracy are not limited to elections. When using other procedures to take decisions, including letting experts decide and organizing referenda, citizens’ acceptance of the decision also appears to vary as a function of whether the outcome matches individuals’ preferences (Esaiasson et al., Reference Esaiasson, Persson, Gilljam and Lindholm2019; Marien and Kern, Reference Marien and Kern2018).
In this research note, we examine losers’ reactions to the votes taken in a citizen deliberative assembly. In such an assembly, through the process of deliberation and dialogue, citizens are invited to “take into account the views of others” (Dowding, Reference Dowding2018: 242). This feature of citizen deliberative assemblies implies that, in contrast to elections or referenda, they are “less polarizing in nature and (…) enhance consensus seeking” (Marien and Kern, Reference Marien and Kern2018: 875). At the same time, however, participants in citizen assemblies are typically encouraged to duly consider the consequences of all the options that are presented to them and to form personal opinions based on their own value judgments about the relative importance of these consequences. In this context, people are bound to reach different conclusions about the decision that should be taken. Any vote on a specific option that is presented to a citizen deliberative assembly will therefore be nonconsensual, producing a majority and a minority, and in this way generating winners and losers. The question that motivates our study is whether, during citizen deliberative assemblies that include a vote, the minority (the losers) accepts the decision reached by the majority.
We know little about how participants in a deliberative assembly react when they realize that they lost, and that a majority voted against their position. Summary reports of deliberative assemblies and similar participatory processes often report on participants’ satisfaction with the process. However, we lack systematic analyses of the presence of winner/loser gaps in satisfaction assessing this pattern and examining its robustness.
There is some work that suggests that when a vote is introduced in a deliberative process it creates opposition between participants (Felicetti et al., Reference Felicetti, Niemeyer and Curato2016). A conjoint experimental study furthermore indicates that citizens are more supportive of assemblies that take a decision which is in line with individuals’ personal preferences on an issue (Goldberg and Bächtiger, Reference Goldberg and Bächtiger2023). This work suggests that the public's support for relying on deliberation to reach decisions depends in part on the outcome, but we do not know how much agreeing with the outcome of an assembly matters among individuals who participate in the assembly and actively deliberate about the merits and limitations of various options.
The only research that has examined this process among participants of a citizen deliberative assembly is, to the best of our knowledge, Fournier et al.'s (Reference Fournier, Van Der Kolk, Cart, Blais and Rose2011) book about citizen assemblies on electoral reform in British Columbia (Canada), the Netherlands and Ontario (Canada). It devotes one short paragraph to this question. The authors find that “satisfaction with the assembly's decision topped 80% among those who were on the losing sides of the debates (…). The fact that those whose preference had not been endorsed by the assembly rallied to the wider group's decision suggests members considered the deliberative and decision-making process had been conducted in a just and equitable way” (Fournier et al., Reference Fournier, Van Der Kolk, Cart, Blais and Rose2011: 46). These assemblies were unique, however. They lasted for months, the members had time to meet and get to know each other, and they developed a strong attachment to the institution (Fournier et al., Reference Fournier, Van Der Kolk, Cart, Blais and Rose2011: 47).
This leads to the question whether small-scale deliberative initiatives that typically last one weekend can similarly result in decisions that are widely accepted by both those on the winning and the losing ends of a decision. How easy or difficult is it for losers to accept that the informed opinion they developed after listening, thinking, and discussing, is not shared by the majority in the group? We tackle this question in this research note.
A Citizen Assembly on Local Electoral Democracy
A citizen assembly on issues related to local electoral democracy was organized in June 2023 in the city of Longueuil, a relatively large suburb (population of about 250,000) of Montreal, Canada. There were about one hundred participants, who agreed to devote a full weekend to discussing four potential reforms related to the way local elections are conducted in that city.Footnote 1 The participants were recruited through an invitation that was initially sent by postal letter to a random selection of residential addresses in the city of Longueuil and, in a second step, through social media. To be eligible to participate in the assembly, interested participants had to be eighteen years or older, residents of Longueuil and be comfortable discussing in French. The group was representative of the population of Longueuil in terms of age and gender, but the lower educated were underrepresented (see Appendix B).Footnote 2
The participants were invited to a welcoming reception on Friday evening, where they received more information about the objectives and goals of the assembly. The participants were also asked to fill out a paper survey on Friday, which provided baseline measures of participants’ characteristics, political attitudes and opinions on the issues that would be discussed during the assembly. Saturday and Sunday were divided into four three-hour segments, each dealing with a specific reform on which the assembly had to vote. Participants completed another survey at the end of the assembly.
The four reforms that were debated were: (1) introducing internet voting, (2) making voting compulsory, (3) giving permanent residents the right to vote and (4) holding local elections at the same time as provincial (that is, regional) elections. In selecting the topics to be discussed in the deliberative assembly, we opted for institutional issues related to the conduct of local elections. This focus was partly driven by our own interest in electoral rules and reform, and partly driven by the political and societal relevance of these topics. Local and provincial authorities are particularly interested in institutional reform given the low and declining turnout in municipal elections, and Elections Quebec had already initiated a pilot project about internet voting. The final selection of topics was informed by discussions with the institutional partners of the assembly, to ensure the reforms would be of interest to them. While we do not have measures of the salience of the topics that were discussed at the assembly, it is fair to assume that these are low salience issues that are not discussed extensively in the media. Therefore, our findings may not apply to high salience or polarizing issues like other citizen assemblies (for example, the UK Climate Assembly, the Irish Citizens’ Assembly and so on).
The participants knew that each segment would end with a vote on the issue, they knew the exact wording of the question (see Appendix A) and they knew that they would deliberate on the issue. The fact that the assembly discussed and voted on four different reforms provides a unique setting for studying the connection between winning/losing and support for citizen deliberative assemblies, as it results in different dosages of being on the winning or losing side of the decisions taken by the assembly.
The project was the initiative of our research group. Though it was partly funded and supported by the city of Longueuil, Elections Quebec and the Quebec department of municipal affairs, it was made clear to the participants that the assembly possessed no mandate, and that there was no commitment to follow up on the decisions they would reach, especially as the rules regarding the conduct of local elections are the responsibility of the Quebec government.
Deliberation took place in three steps. First, an expert political scientist made a short presentation about the reform, explained where it existed, how it worked and presented the main arguments for and against in an easy to understand and balanced fashion. The presentation was followed by a Q&A session. In the second stage, members met in small groups of ten to twelve participants, where they continued to discuss the issue and exchanged viewpoints in the presence of a discussion facilitator. The final step before the vote was a plenary session in which the participants could ask further questions to the expert and express their opinions. The vote was secret and took place using paper ballots.
Even though there is concern that having participants vote in a deliberative assembly fuels disagreement and reduces participants’ deliberative capacity (Felicetti et al., Reference Felicetti, Niemeyer and Curato2016), in line with other deliberative assemblies (for example, Fournier et al., Reference Fournier, Van Der Kolk, Cart, Blais and Rose2011; Kamenova and Goodman, Reference Kamenova and Goodman2013; Muradova et al., Reference Muradova, Walker and Colli2020), we asked participants to vote on concrete proposals for democratic reform. We did so for two main reasons. First, by introducing a proposal on which participants would have to vote, we directed the participants and the debate towards a well-articulated idea. Second, the vote provided us with an outcome of the deliberative assembly that would be easy to communicate to interested authorities, such as the municipal and provincial governments.
Throughout the deliberation, we insisted that there were no good or bad answers, that there were sound reasons for and against each of the four proposed reforms and that we wanted the participants to form their own personal opinions on the different issues. There was no pressure or incentive to reach a group consensus.
The set-up and conduct of the assembly were inspired by the Deliberative Polls of Fishkin and Luskin (for example, Fishkin et al., Reference Fishkin, Luskin and Jowell2000; Luskin et al., Reference Luskin, Fishkin and Jowell2002), and the amount of time that participants spent on each of the topics is comparable to what is common in such Deliberative Polls. Devoting three hours to listen to experts, discuss with other people, and reflect about all the arguments is substantially more than what people commonly do about a political topic. For those reasons, and even though the time for discussion was shorter than in other types of deliberative exercises (for example, Farrell et al., Reference Farrell, O'Malley and Suiter2013), we describe the gathering as a deliberative assembly.
The recruitment process involved participants self-selecting into participating in the assembly, implying that participants are not representative for the overall population. As an indication that there was indeed self-selection, we note that the average political interest of participants in the assembly is 7.8 (measured on a scale from 0 to 10), which is higher than the average for citizens in Quebec (for example, it is 6.5 in the 2022 Quebec Election Study). The assembly thus was conducted among citizens who are, on average, more interested in politics. Later on, we come back to this point and discuss the implications of self-selection for our findings.
In terms of the outcomes of the assembly, most participants voted in favor of internet voting, the right to vote for permanent residents and simultaneous (provincial and local) elections, and voted against compulsory voting. Table 1 shows the official results of the votes. We are interested in the reactions of those who were on the losing side on the various votes, and we compare them to the reactions of those who were on the winning side.
Table 1. Outcomes of the four proposals
Note: Row percentages are shown in parentheses. Throughout the assembly, there was some variation in the total number of participants present during the vote. See Appendix A for the question wording.
The Participants’ Assessments of the Assembly
To identify which participants are “winners” and “losers” for each of the decisions taken by the assembly, we make use of the data from the surveys administered just before and at the very end of the assembly. Specifically, the final survey asked respondents to report how they voted on each of the four proposals. We follow a common approach in the literature on winner-loser gaps and focus on whether the participant sided with the majority (winners) or the minority (losers) (Anderson et al., Reference Anderson, Blais, Bowler, Donovan and Listhaug2005; Stiers et al., Reference Stiers, Daoust and Blais2018). A total of 101 citizens took part in the final survey. Table 2 shows the number of reported losers for each of the votes, which ranges between 14 per cent and 36 per cent. In terms of respondents’ status across the four proposals, 42 (42%) participants never lost, 38 (38%) lost once, 14 (14%) lost twice and 7 (7%) lost three times. No participant was on the losing side on each of the four votes. (Appendix A shows the exact wording of the survey questions used for this analysis.) A comparison of the distributions in this table with those of the official outcomes of the votes on the four proposals (see Table 1) indicates that the reported votes match the actual votes cast during the assembly closely, suggesting that respondents reported their voting decisions accurately.
Table 2. Winners and losers, by outcomes
As a measure of consent, we asked the participants how satisfied they were with the overall conduct of the assembly.Footnote 3 The item captures a general level of satisfaction with the deliberative process, very much like the indicators of satisfaction with the functioning of democracy that are used in the literature that studies electoral winner/loser effects (Anderson et al., Reference Anderson, Blais, Bowler, Donovan and Listhaug2005). Our goal with this indicator was to tap the participants’ general evaluation of the deliberative process, in the same way that the standard satisfaction with democracy item measures an overall assessment of the functioning of democracy using a single-item indicator (Daoust and Nadeau, Reference Daoust and Nadeau2023). It is possible, however, that it might also capture citizens’ appraisals of the material organization of the citizen assembly (for example, lunches were provided, which was appreciated by the participants). To partially address concerns about the single item outcome variable, the questionnaire also included three additional questions about evaluations of more specific elements: expert presentations, small group discussions and plenary discussions. While we focus on the overall assessment, we also replicated the results for the more specific evaluations.
Figure 1 shows the relationship between the number of defeats a person experienced and the degree of satisfaction with the process. The grey circles in the graph represent individual participants, while the black coefficients and 95 per cent confidence intervals indicate the mean level of satisfaction in each category of respondents—grouped by how many losses they experienced. We observe a remarkable degree of satisfaction. Only two individuals expressed dissatisfaction (indicating they were “not very satisfied”) and these persons had lost only one vote. The percentage of respondents who report being very satisfied with the assembly is lower among those who lost three times, but even in this group a majority (4 out of 7) were very satisfied with the conduct of assembly. In other words, we find no evidence of an association between how often a person was on the losing side and their evaluations of the conduct of the assembly when focusing on the bivariate relationship (Kendall's tau is .00).
Figure 1. Overall satisfaction with the conduct of the assembly by the number of losses.
The same pattern holds when we look at each of the votes separately. More precisely, at least two-thirds of losers on each vote are very satisfied with the overall conduct of the assembly.Footnote 4 All in all, the losers expressed strong support for the deliberative process.
We can test more formally the existence of a winner-loser gap through OLS regression analyses. The dependent variable is satisfaction with the conduct of the assembly. For the regressions, we treat satisfaction as a continuous variable, ranging from 0 for those not at all satisfied to 1 for those very satisfied. We consider two independent variables to examine the presence of winner-loser gaps. As a first indicator, we focus on the proportion of votes on which the individual was on the losing side, which runs from 0 to 1.Footnote 5 Additionally, we assess the impact of a series of dummy variables capturing whether the person lost on a specific vote.
Table 3 shows the results of OLS regression analyses.Footnote 6 The estimates in Table 3 indicate that neither the proportion of losses (column 1) nor the specific loss dummy variables (column 2) correlate significantly with the participants’ overall satisfaction with the citizen assembly.Footnote 7 The regression approach also allows us to account for several sociodemographic characteristics and political attitudes that can correlate with opinions about a deliberative process, as well as opinions about the proposed reforms. To check the robustness of our findings, we add the following control variables: age, gender, education, political interest and satisfaction with democracy (columns 3 and 4).Footnote 8 Doing so does not alter the main conclusion, that being on the losing end of the votes cast in the citizen deliberative assembly does not shape participants’ satisfaction with the assembly. We thus find that winners express very positive judgments about the conduct of the assembly and that losers fully agree with them.
Table 3. OLS regressions predicting satisfaction with the conduct of the assembly
Note. Standard errors in parentheses
*p < 0.05, ** p < 0.01, *** p < 0.001
As can be seen from the results shown in Appendix D, the patterns are largely the same when focusing on evaluations of different aspects of the citizen assembly. This is reassuring and signifies that our main results are not driven by participants expressing satisfaction with the way the assembly was practically organized but rather reflect an assessment of the decision-making process and procedures.
Given the small sample size, there might be a concern that our design lacks the statistical power to detect meaningful differences in participants’ satisfaction. The estimates in Table 3, however, are close to zero—in particular when we consider the effect of the proportion of losses that participants experienced. In other words, leaving aside statistical significance, our estimates seem to be “precisely null.” That said, as an additional test, we stacked participants’ responses on the four different measures of satisfaction with the conduct of the citizen assembly (the general measure and three specific measures), hence increasing the number of observations and variance in the outcome measure. As can be seen from the estimates that are reported in Appendix E, however, this test still shows an insignificant and substantively very small association between losing and satisfaction.
As indicated above, the participants of the citizen assembly—because they opted into participating in a discussion on electoral reforms—are highly politically interested. For this self-selection effect to bias our findings, however, political interest would need to correlate strongly with democratic satisfaction. Reassuringly, political interest and satisfaction are only weakly correlated. This holds for Quebec citizens overall, with political interest and satisfaction with democracy in Quebec only correlating at .05 in the 2022 Quebec Election Study. This also holds for our sample, as the correlation between political interest and satisfaction with the conduct of the assembly is also low—at .14.
Our findings might not generalize to deliberation around highly contested or polarizing issues either. On this point, it should be noted that even though citizens’ views about electoral reforms are not strongly polarized, they are not consensual either, as participants had meaningfully different views about these issues before the start of the assembly. During the debate about whether permanent residents should have the right to vote in local elections, we also observed some degree of emotion among the participants and strong criticism of the “other” side in the debate. This was likely triggered by the link between this issue and the issue of immigration, which is a polarized issue in Quebec and elsewhere (Bélanger et al., Reference Bélanger, Daoust, Mahéo and Nadeau2022; Gagnon and Larios, Reference Gagnon and Larios2021).
Conclusion
Losers’ consent is fundamental in a democracy. There is a vast literature about what kinds of individuals, in what kinds of context, accept electoral defeat. But the issue of losers’ consent goes beyond the electoral realm. In all kinds of decision-making, people are bound to disagree about what should be done, and the hope is that if there is a vote those who are in the minority (the losers) will agree that the process was fair even though they do not like the outcome of the vote.
In this research note, we examined losers’ reactions in a citizen assembly held in Canada. Our study shows that it is possible and fruitful to ascertain whether a winner/loser gap does or does not emerge in more direct forms of democratic engagement, beyond electoral settings. We have provided evidence that at least in some contexts deliberative democracy can succeed in convincing the losers that the decision-making process is satisfactory.
Our findings are in line with those of Fournier et al. (Reference Fournier, Van Der Kolk, Cart, Blais and Rose2011) who report that those who were on the losing side in the decisions made by the citizen assemblies on electoral reform in British Columbia, Ontario and the Netherlands still expressed strong satisfaction with the deliberative process. Deliberative assemblies appear to enjoy a good deal of support, which is recognized even by those who disagree with their decisions.
This suggests that citizen assemblies enjoy a high degree of support among those who participate in them. How broad consent is remains an open question, however. We cannot dismiss the possibility of a selection bias; that is, that those who are positively disposed towards deliberation are much more prone to participate in such exercises. Furthermore, the electoral reform issues that were debated in the assembly were not “hot” issues about which people have strong emotional feelings, which may have facilitated losers’ consent. More research is clearly needed to better understand the contours of losers’ consent in different forms of deliberation. Our findings, however, suggest that there is a reservoir of support for deliberative democracy.
Supplementary material
The supplementary material for this article can be found at https://doi.org/10.1017/S0008423924000775
Open access
