Europolis and the European public sphere: Empirical explorations of a counterfactual ideal

Abstract

Is there—or could there be—a Europe-wide public sphere? Some argue that one already exists, others that none is attainable. This debate turns on what it means to have one—on how much (and what kinds of) cross-border ‘discussion’ and public input it must entail. An ambitious European public sphere would involve more truly Europe-wide collective will formation and political accountability. This article attempts to move beyond speculation, with a discussion on an ambitious version of a European public sphere. Participants' opinions and vote intentions in Europolis were gauged before and after deliberating. This enables us to probe a double counterfactual: what if there was a more ambitious European public sphere, and what if European Parliamentary elections were consequently more deliberative.

Keywords

Actors and institutions decision-making mode deliberative democracy deliberative polling European Parliament European public sphere public opinion research and innovation voting

Is there—or could there be—a Europe-wide public sphere? Some argue that one already exists, while others claim that none is attainable. In large part, however, this debate turns on what it means to have one—on how much (and what kinds of) cross-border ‘discussion’ and how much (and what kinds of) cross-border public input it must entail.

Surely a pale version of a European public sphere already exists. Some people travel and talk, in shared languages, face to face. They may also ‘talk’ telephonically, online, or via social media. Print, broadcast, and online media cover sufficiently salient stories that are either Europe-wide or about other European countries. In addition, European Parliamentary elections provide a limited form of Europe-wide democratic accountability. Yet there are limits. Cross-national ‘discussion’ is limited by the nontrivial proportions of each country’s citizenry who know no other language and never talk meaningfully with foreigners at home or abroad; by the print, broadcast, and online media’s tendency to focus preponderantly on domestic concerns; and by the nature of European Parliamentary elections—low-turnout, low-information, and focused chiefly on domestic issues. They are, in a word, ‘second-order.’¹

A more ambitious European public sphere would involve more truly Europe-wide collective will formation and political accountability. There would be much more common discussion of shared policy issues across linguistic and national boundaries and more voting or other forms of political action based on the opinions thus formed about those issues. This is a public sphere that surely does not yet exist but possibly could, at least at some point in the long term.

The unitary public sphere applied Europe wide, sometimes referred to as ‘public sphere heavy’ has been largely a foil for empirical investigations of the actual extent to which there is interpenetration of European topics in national discussions and on whether media in different countries are treating the same topics in the same way or from their own national perspectives (de Vreese and Schmitt, 2007; Koopmans and Erbe, 2004; van de Steeg, 2002). There have also been studies of the salience of Europe, the development of European identification or citizenship as a precondition for European wide communication and on whether new media have allowed some topics to more easily transcend national boundaries (de Vreese and Schmitt, 2007). The ‘public sphere heavy’ ideal has inspired empirical study of the more modest communicative interactions in a European ‘public sphere light.’

But the ‘heavy’ or unitary version should not be simply dismissed or treated as a straw man because it is currently so far from reality. It has continuing appeal because of its simplicity and clarity as a democratic ideal for Europe. Perhaps it is merely a ‘cosmopolitan temptation’ (Schlesinger, 2007). Yet it is also the picture that influential critics of European Union (EU) wide democracy inevitably invoke (see e.g. Grimm, 1995). In this paper, we treat it seriously, try to uncover its underlying logic, and then explore its implications in a microcosmic experiment. In doing so we move from thought experiments to a real experiment in the context of the 2009 EU Parliamentary Elections.

Why would it be useful to do this in microcosm, when the full scale version is obviously so far from realistic realization? First, the microcosmic version, if credible, is itself a useful addition to the tool kit of democratic practices. It is intended to offer a representation of what the public would think under good conditions for thinking about the issue. Of course, in the case of Europe, the very existence of the relevant public is part of the issue that is contested, part of the challenge facing the unitary ideal. Yet such an experiment allows us to bring into focus a picture of what a unitary public sphere for all 27 countries would be like.²

First, consider a modest use case. If the microcosmic version can be made to work, it can add to the EU’s tool kit for democratic consultation because it can be convened to make recommendations for specific policy challenges. Such an action is not unprecedented in other parts of the world. In British Columbia and Ontario, randomly selected ‘citizens assemblies’ were convened to propose ballot propositions for electoral reform (Warren and Pearse, 2008). More recently, the Japanese government convened a national Deliberative Poll (DP) to make recommendations for how to resolve the nearly intractable issue of Japan’s future energy policy after the Fukushima disaster. The preferred option in the DP was accepted by the government at the time. In Texas, a series of DPs—deliberating microcosms chosen by random sampling—were convened by the state Public Utility commission to deliberate about energy choices for the state. These deliberations led to large increases in the investment in wind power, making Texas the leading state by 2007 (Fishkin, 2009). Prominent European scholars have proposed institutionalizing randomly selected ‘citizens assemblies’ in Europe for certain purposes (see Schmitter and Treschel, nd).

Convening a random sample to deliberate on legislative and policy issues has a long history. It goes back to ancient Athens, where the randomly selected Council of 500 set the agenda for the Assembly and where randomly selected legislative commissions, or nomethetai, would make the final decision on legislation after hearing the arguments both for and against a new law (Hansen, 1991). If such randomly selected microcosms could be made to work at the European level, they could be convened to provide recommendations (or even decisions) to respond to difficult choices that might assist with ameliorating the ‘democratic deficit’. In a time of democratic experimentation in which EU institutions have changed profoundly over the last decade, the use of such a tool is not unthinkable, especially for difficult problems.

In all these cases, it should be clear from the outset that the random samples will likely become more informed and engaged than the rest of the public. Yet that is, in a way, the point. After extensive deliberation, their views provide a recommendation for the rest of the public to vote on or for governments to act on. The idea is that the microcosm should represent what the public would think under good conditions for thinking about it and for getting its questions answered. Most members of the public most of the time are not seriously engaged in public issues. They may be, to use the classic phrase ‘rationally ignorant’ (Downs, 1957). European issues may be a context where this is even more often the case than in other policy areas. Yet in the DP, the participants each have a significant role and the hope is that they may be effectively motivated to seriously engage the arguments. The microcosm thus offers a ‘what if’: what if the people were more informed and engaged, what would they recommend under transparently good conditions for thinking about the problem?

A second and separate use, probably more important over the long term, is to clarify the appropriate democratic ideal that might guide democratic reform in the EU. As noted, there is a widely perceived ‘democratic deficit’ in the EU. Democratic reform is facilitated when one has a coherent picture of what the reforms ought to aim at over the long term. Does the unitary version of the European wide public sphere constitute such an ideal, or does it decompose or discredit itself once one tries to realize it? By replacing utopian speculation with empirical evidence, we can contribute to this debate.

There are many critiques of democracy in the EU. Some focus on the low participation and the lack of competitiveness in EU Parliamentary elections (Marsh and Mihkhaylov, 2010). Some focus on the lack of transparency in EU decision-making. Others focus on the low level of public understanding and the low salience for most European issues (Eriksen and Fossum, 2000; Kohler-Koch and Rittberger, 2007). Yet from the perspective of what has come to be called ‘deliberative democracy’, these critiques focus on only part of the problem. They do not, as their authors acknowledge, precisely hit the mark. From the perspective of deliberative democracy (see Bohman and Rehg, 1997; Elster, 1998; Fishkin and Laslett, 2003) the democratic deficit is a deficit in reason-based collective will formation (Habermas, 2009, 2012; Offe, 2011). In this view, democracy requires that the ‘will of the people’ be meaningful and so the people need some sort of effective opportunity to think and consider competing reasons for what should be done. We say ‘effective opportunity’ because the public, having information available in theory, is not usually motivated to exercise such opportunities. Other forms of democratic practice do not have this requirement. Even when election campaigns are highly competitive, they can be largely devoid of any coherent substance or reason giving. Campaigns can degenerate into competing efforts merely to mobilize and mislead the public. One can have high participation, with forms of participatory democracy, such as referenda, where the public has little knowledge of the issues. One can also have transparent decision-making but very little thinking and discussion outside of elite circles. The unitary version of the European wide public sphere contributes something distinctive to the debate over the democratic deficit in the EU. It is a concern for reason-based collective will formation on the part of the people themselves, rather than merely on the part of policy relevant elites.

The unitary version is a further specification of a distinctive, deliberative perspective on democracy in the European context. Perhaps it is inspiring or perhaps it is inappropriate for the EU (Moravscik, 2006). Yet an institutional design for exploring what it would mean can assist both theory and practice. The DP is an institutional design that aspires to convene such a process with a microcosm of the people. Can the microcosmic version work on a pan-European basis? What would it mean for it to work? In this paper, we propose some criteria and where data are available, we apply them to the pilot version offered by Europolis.

Even though the unitary ideal of the public sphere is far removed from current practice, if it is a coherent and defensible ideal, one that can inspire democratic reform, then we can think over the long term how its realization might be approached. There are credible arguments about the incentives for the spread of a unifying language (usually English) within the EU over the long term (see de Swann, 1993; Van Parijs, 2011). In addition, technology is already making cross border communication easier on European issues. Virtual communication and social media are not limited by the compartmentalization of national broadcasters and national newspapers. It is hard to predict the long term limits of such trends. If the unitary ideal offers a compelling picture for the long term democratic future of the European public sphere, at least on EU issues (see de Vreese and Schmitt, 2007 for an effort to distinguish European issues), then it is arguable that it might well be approached or eventually approximated.

A European wide public sphere

What are the elements of an ambitious version of a European wide public sphere that we might want to see successfully piloted in microcosm? Here we can piece together the logic of such an undertaking, attempting to distill a coherent picture from a large literature.

The unified version of the European wide public sphere should be:

Inclusive: It should be representative of citizens from throughout the EU. By modern norms of political equality, if a European wide public sphere were to serve a democratic function, we can assume that it should be open to all members on the same basis. More specifically, if some groups or countries were left out, that would be a distortion of the dialogue. Hence any version or representation of the European wide public sphere should either engage the whole population or it should engage a representative microcosm. Put another way, it should avoid ‘participatory distortion’ (Verba et al., 1995). A good random sample is one way of doing so.

Dialogic: It should allow for an active discussion weighing competing reasons in an atmosphere of mutual respect shared by the participants from throughout the EU. An atmosphere of mutual respect is postulated so that participants can listen to the reasons offered by others as well as express them in some confidence that they will be heard.

Informative: It should allow the participants to become more knowledgeable.

Deliberative: It should allow the participants to arrive at their considered judgments about what should be done on selected policy questions after considering alternative reasons or the pros and cons of competing options.

Undistorted: It should not be distorted by inequalities that would put into question the process by which the participants were coming to their judgments of the issue. For example, if the process was dominated by the more advantaged so that the conclusions that emerge were the ones that are pushed by the most advantaged and then imposed on the most vulnerable, then the credibility of the deliberative process would have been undermined. This criterion has to do with a rough equality among those who participate in the process and not the equality of the selection process for participants (as in criterion 1 above).

Consequential: It should take place in a context of accountability by which the conclusions it reaches can be connected to decisions by political actors or policy makers.

The motivation for these conditions is normative. Lack of any of them would undermine the appeal and the clear vision of the European wide public sphere. First, lack of inclusiveness would provide an obvious objection that relevant groups or perspectives have been excluded. Leaving out some countries or some widely held perspectives within countries raises the question of whether the conclusions might be very different if those views had been included in the dialogue. Second, if the process is not dialogic, then competing reasons will not have been offered or responded to, nor would they have been listened to so that their merits could actually be considered. Without a dialogue considering competing reasons for policies, the goal of a reason based process of collective will formation would be lost. Third, if the process is not informative, then it is vulnerable to being determined by flawed, misleading or inaccurate information. The basic idea is to see what people would think after they considered reasons under good conditions; access to reasonably good information is clearly one of those conditions. If participants have been misled or are just so uninformed that they cannot understand the issues, then the aspirations of the deliberative process have been defeated. Fourth, if the process is not deliberative, then competing reasons for the policy options will not have been considered and weighed and so the goal of piloting a more authentic process of public will formation would have been undermined. The root of deliberation is ‘weighing’ and the participants would, ideally, be weighing the merits of competing considerations or trade-offs in deciding what they think should be done. Fifth, if the process is distorted by inequalities, then it would be reasonable to question whether people are reaching their conclusions on the merits rather than just acquiescing in conclusions that have been imposed on them by those with high status or power. The literature about the public sphere emphasizes the possibility of ‘bracketing’ inequalities, seeing if people could reason so that the inequalities were not distorting the route to their conclusions (Fraser, 1992; Sanders, 1997; Young, 2002). Another way of putting this criterion is therefore that it should more or less ‘bracket’ the inequalities, so that there is a relatively equal discussion regardless of, say, class and education. Sixth, the process should be consequential. It needs to be more than just talk. Ideally it is connected to some mechanism of accountability for decisions (Fraser, 1992; Guttman and Thompson, 1994; Offe, 2011).

Europolis

Here we attempt to move beyond speculation, using data from Europolis, a Europe-wide DP, conducted just before the European Parliamentary elections of 2009, to shed empirical light on an ambitious version of a European public sphere (for more on Deliberative Polling see Fishkin, 2009; see also Luskin et al., 2002 for the first empirical application). The participants—a random sample from all 27 EU member-states—spent time learning, thinking, and talking about issues affecting the whole continent. We gauged their opinions and vote intentions before and after deliberation.

Europolis gathered a microcosm of the European public together for one weekend, from Friday to Sunday, to discuss two issues: climate change and immigration. Participants discussed these issues in randomly assigned small groups, led by trained moderators, and questioned balanced panels of experts in plenary sessions with questions that had been agreed to in small group discussions. The random assignment was stratified to ensure a manageable amount of language diversity in each group. All the groups employed simultaneous interpretation.

At the end, after three days of alternating small groups and plenary sessions, the participants completed an expanded version of the same questionnaire they completed previously (both on first contact and on arrival.) In all, they were interviewed four times: at home (T₁), on arrival (T₂), on departure (T₃), and several weeks after the election (T₄). Some of the policy attitude questions were asked in all four waves, but some were only asked at T₁ and T₄, while a more extensive battery was asked via self-completion questionnaires at T₂ and T₃. Subjects who were randomly assigned to the control group were interviewed at T₁ and T₄. TNS, which administers the Eurobarometer, conducted the sampling and interviewing. A detailed discussion of the research design can be found in the Introduction to the symposium.

There were detailed briefing materials offering arguments for and against various policy choices on the two issues. These briefing materials were supervised by an elaborate advisory group that vetted the materials for both balance and accuracy. More on the process can be found in ‘Introduction: The EuroPolis Deliberative Poll’ in this special issue.

How did Europolis do on our six criteria? The evidence is incomplete. On some criteria, we have data that speak directly to the issue. On others, we have only partial and suggestive evidence. Other papers in this symposium focus on certain criteria in more detail because it would take a considerable effort to examine the necessary qualitative data from the small group discussions. Our aspirations in this paper are thus limited by the available data and by space. Nevertheless, we think a suggestive and coherent picture does emerge.

Does Europolis offer a reasonable approximation of the unitary model of the European wide public sphere? Depending on our answer, what do we learn about deliberative democracy and about EU politics?

Inclusion: How representative?

The representativeness of the 348 participants can be checked by comparing them to the 4036 ‘nonparticipants’ (T₁ interviewees who did not attend the event). In age, class and other demographics, the participants and nonparticipants were very similar, although men were slightly over-represented among participants (54%). The participants’ and nonparticipants’ left-right self-locations were virtually identical. The participants did start with somewhat less restrictive attitudes toward immigration, but had roughly the same attitudes as the nonparticipants about climate change. The participants were also more interested in politics, had a stronger sense of civic duty, included somewhat more people intending (at T₁) to vote for the center right group of parties, the European People’s Party (EPP), and somewhat fewer intending to vote for the Party of European Socialists (PES). These modest differences are unlikely, in our view, to affect the results.

On the two specific policy issues, immigration and climate change, we constructed indices that offer a basis for comparison at T₁, both between participants and nonparticipants and between participants and the separate random sample of the control group.

The immigration index consisted of nine questions scaled so that support for policies that were more welcoming toward immigrants were at the top of the scale and more restrictive policies were at the bottom. The questions included whether illegal immigrants should be eligible for national health care, whether the children of illegal immigrants should be eligible to attend public school, whether border controls should be enforced, whether decisions about what immigrants to admit should take account of what countries they are from, whether penalties should be imposed on employers who hire illegal immigrants, whether immigrants from non-EU countries needed to be Christian, white, committed to the national way of life or from a similar culture (as the country admitting them). The Chronbach’s alpha for the index is .736.

The Climate Change index consists of two questions: whether climate change is a serious problem and whether ‘we should do everything possible to combat climate change even if that hurts the economy or whether we should do everything possible to maximize economic growth, even if that hurts efforts to combat climate change.’ The former option, in support of climate change efforts, was 10 on a 0 to 10 scale and the latter option in favor of the economy was 0 on a 0 to 10 scale. These are the only two questions available at both T₁ and T₄. There are a number of related questions at T₂ and T₃ but since we are using the treatment control comparisons at T₁ and T₄, we have limited choice. This index has a Chronbach’s alpha of .5.

Both indices are listed in online appendix A along with the changes in the individual items.

Table 1 shows the difference between participants and non-participants on the two indices at time 1. There is no significant difference on the Climate Change index. On the immigration index, there is a significant difference. The participants are somewhat more welcoming to immigrants than the nonparticipants, but still not far from the midpoint of the scale.

Table 1.

Attitudinal representativeness for indices.

	Participants	Nonparticipants	Sig.
Immigration index	.583	.541	.000
Climate change index	.668	.657	.201
N	348	4036

Note: An independent test was used to compare the indices between participants and nonparticipants at time 1.

In order to assess whether any selection bias in the recruitment affected the results, we conducted a propensity matching analysis parallel to that employed by all the other contributions to this symposium. This analysis is detailed in the online appendix. It shows that the differences if matching were employed would be modest.

As a separate check on representativeness, the participant sample was checked against Census and other data at the European level for the general population. This analysis is discussed in the Introduction, but it shows a representative microcosm on basic demographics and no significant differences on a left-right spectrum with Eurobarometer data. While the microcosm was not perfect, it was generally a good representation of the aspiration to put Europe in one room.

Was it dialogic?

We are limited about what we can say about this criterion, apart from self-reports. The best way to investigate the extent to which people were really communicating with each other and sharing reasons would be to look at the transcripts and code them with a method such as the Discourse Quality Index (DQI). That is what has been done in one of the other papers in this issue (Gerber et al., 2014). Within this paper, it is worth briefly summarizing the self-reports, especially since they are consistent with the picture from the DQI. One can argue about whether the effects are large or small but the DQI analyses clearly show movement based on arguments offering reasons—what the authors call ‘deliberative persuasion’ on the difficult substantive issue of legalizing illegal immigrants (see Gerber et al., 2014).

Turning to the self-reports, the event evaluations offer indications of successful communication and mutual respect. Participants were asked to evaluate the event as a whole on a 0 to 10 scale, where 10 indicated the event was extremely valuable. On this evaluation question, 86% rated it at 8 or above, and 59% gave the event a perfect 10. The ratings of the plenary sessions both with politicians and with experts were also high, with each being found useful by 74%. The ratings of the small group discussions were still higher, with 92% of participants finding them useful.

These evaluations also offer hints of deliberative quality. On average, the participants considered the event extremely balanced. Of those who said that they had read more than half of the briefing materials (a large majority of the participants), roughly two-thirds saw them as balanced, and only 11% saw them as clearly favoring some positions over others. Similarly, 69% agreed that their small group moderator tried to make sure that opposing arguments were considered, while 86% disagreed that the moderator sometimes tried to influence the group with her/his ideas.

The participants also considered the quality of the discussion to be high. More than 60% saw their fellow group members as participating equally in the discussion. Almost 90% thought that they had ample opportunity to express their own views. Furthermore, these expressions seemed to take place in an atmosphere of mutual respect, as 84% felt that their fellow participants ‘respected what I had to say, even when they didn’t agree.’ They also saw their fellow participants as taking their roles seriously: 87% saw them as ‘express[ing] what was truly on their mind.’ Only 18% considered that they ‘expressed strong views without offering justifications’, while only 23% considered that they ‘had made up their minds [so that] the discussion had little effect on them.’

The experience of meeting and talking with other people from all across the continent and from all walks of life also had an impact on participants, as 81% thought that they learnt ‘a lot about people different from me—about who they are and how they live.’

Was it informative?

The participants clearly learned a great deal about both immigration and climate change—and also about the EU. They were asked nine knowledge questions, three each about each of those three topics. For each topic, two of those three questions were first asked in the initial interview, while the remaining one was first asked only at the beginning of the event some weeks later. Since the participants begin learning from the moment they are initially interviewed and invited to the event (and are sent the briefing materials well in advance), the six items first asked in the initial interview show a distinctly greater knowledge gain than the three asked only on arrival (16.5% versus 7.5%). The participants presumably learned about as much on the latter as on the former; it is just that on the latter our earliest measurement (on arrival) occurs too late to capture all the learning. The knowledge gains for all the items are displayed in Table 2.

Table 2.

Knowledge gain (based on all available items).

Knowledge items	T₁	T₂	T₃	T₃–T₁	p
Q43. Is the main decision-making body of the European Union the … ? (Council of Ministers)	10.1	15.8	23.6	13.5	.000
Q44. Only one of the following statements about the European Parliament is false. Which one is it? (It passes all EU laws)	11.8	22.4	23.6	11.8	.000
Q45. Is the European Union represented on the international stage by the … ? (European Commission)	–	39.9	46.0	6.1	.028
Q46. Which of the following is true of Blue card workers? (They must have university education)	6.6	23.9	30.5	23.9	.000
Q47. Which of the following is true about the ways in which immigration policy is currently made? (The EU sets the basic rules about entry and residency requirements)	22.1	36.5	46.8	24.7	.000
Q48. Which of the following is true of the EU's immigrants? (Most illegal immigrants enter the EU legally but outstay their visas)	–	44.5	56.7	12.1	.000
Q49. The percentage of the EU's total energy consumption that comes from fossil fuels (coal, gas or oil) is about … ? (80%)	22.4	18.4	30.2	7.8	.017
Q50. Which of the following produces the most greenhouse gases? (China)	45.7	49.7	63.2	17.5	.000
Q51. Which of the following is true about wind power in the European Union? (Wind power's share of EU energy consumption is increasing by about roughly 30% a year)	–	15.5	19.8	4.3	.032
Knowledge Index 1 (9-item)		29.6	37.8	18.0	.000
Knowledge Index 2 (6-item)	19.8	27.8	36.3	16.5	.000

Note: The entries are mean percentages answered correctly. The nine-item index includes all the knowledge questions both before and after deliberation. The six-item index includes only the six asked at T₁ and not the three asked only at T₂.

There was also a noticeable difference in how much the participants learned by topic. They learned most about immigration (a 20.2% before–after knowledge gain), then most about the EU (10.5%), and least—though still very significantly—about climate change (9.9%). Among the nine items that were asked at T₂ and T₃, there was an average gain of 18%. The question with the highest increase was about immigration regarding who set the basic rules about entry and residency requirements, 24.7% more participants answered this question correctly after deliberation. All these numbers are probably underestimates, because the indices include items measured only from arrival and there was likely learning in preparation for the event not captured at T₂.

Was it deliberative?

The participants discussed policy options applying to two issues, climate change and immigration. In addition, they discussed their views of political parties and registered voting intentions in the final questionnaire.

The briefing materials that formed the basis for the small group discussions had arguments for and against each policy option. These arguments were vetted by a distinguished advisory group for balance and accuracy. The participants also had the opportunity to compose agreed questions, for each small group, that were directed to balanced panels of competing experts in the plenary sessions. This process is described in more detail in the Introduction. With all this preparation, did anything happen?

On the two issues, there were significant changes on both. The participants became significantly more open or welcoming to immigrants and they became significantly more determined to engage in action on climate change.

Table 3 shows the changes from T₁ (first contact) to T₃ (at the end of the weekend). The participants were also re-interviewed some weeks after the election at T₄, at the same time as the control group. One way of thinking about whether the broad deliberative treatment of participating in the discussions had an effect would be to compare the change in the treatment group with the change in the control group. We can see that in Table 4. On both immigration and climate change there were significant changes from T1 to T4. In both cases those changes hold up in comparison to the control group. On immigration, participants became significantly more accepting of immigration while the control group moved somewhat in the other direction. On climate change, the participants became significantly more supportive of serious action, while the control moved in the same direction, but less so. The difference of differences for both indices is highly significant (in both cases by a two-tailed test).

Table 3.

Indices before and after deliberation.

Index	T₁	T₃	T₃–T₁	p
Immigration	.583	.618	.035	.000
Climate change	.668	.747	.079	.000

Table 4.

Treatment versus control group by indices at T1 and T4.

	T1T	T4T	T4T–T1T	p	T1C	T4C	T4C–T1C	p	(T4T–T1T) – (T4C–T1C)	p
Immigration index	.583	.602	.019	.008	.545	.539	−.006	.314	.025	.002
Climate Change index	.668	.719	.051	.000	.650	.667	.016	.028	.034	.012
N	348	348			722	722

Was it distorted by inequalities?

One of the common criticisms of deliberation is that the more advantaged will dominate the discussions, imposing their views on the rest of the participants (Sanders, 1997; Young, 2002). Some have even argued that the problem is so intractable that inequalities cannot be ‘bracketed’ in the public sphere; rather they must be eliminated (Fraser, 1992). EuroPolis allows a look at this problem empirically in the context of a European mini-public. If the more privileged were to dominate, we could expect a predictable pattern in which the small groups overwhelmingly moved in the direction of the initial positions of the more advantaged. Hence it is worth exploring whether the deliberations in this European wide microcosm of the public sphere suffer from such a distortion. One might imagine that the challenge of engaging across the barriers of language and nationality on difficult policy issues would provide great advantages to the more privileged. They would be deferred to by others and they might be more comfortable with the cross national context of policy discussion. If there were an overwhelming movement in the direction of their initial positions, then that would raise questions about the extent to which the substance of the deliberations was determining the final opinions as opposed to distortions from these social inequalities.

While there are many detailed group level analyses that could be conducted, we believe we can assess the issue of distortion from inequalities by looking at the group issue combinations before and after deliberation and seeing whether the mean levels in each group and on each issue moves toward or away from the initial position of the more advantaged in that small group. For each of the two indices, immigration and climate change, Table 5 shows the proportion of the 25 small groups that move toward or away from the initial position of the males, the females, the upper (and upper middle) class positions and the lower (and lower middle) class positions before deliberation.³ Since this analysis concerns the small group discussions we use the measures at T₂ (on arrival just before deliberation) and T₃ (at the end of the weekend).

Table 5.

Inequality and the small groups.

	Toward T2 males’ position	Toward T2 females’ position	Toward T2 upper- & upper-middle class position	Toward T2 lower-middle/ working class position	Group’s T3 – T2 mean attitude change
Immigration index	.32	.64	.52	.44	.026
Climate change index	.56	.48	.48	.56	.038
Average	.44	.56	.50	.50

Note: Entries, excepting the last column, are proportions of the small groups moving in the indicated direction.

If there were a pattern of domination in the group discussions by which the advantaged imposed their views, then the movements would be mostly in the direction of the initial positions of the more advantaged. However, precisely 50% of the group issue combinations, averaged over the two indices, moved toward the initial positions of the upper and upper middle class respondents. Only 44% of the group issue combinations moved in the direction of the initial male positions. Clearly there was no pattern of domination by the more advantaged.

Was it consequential—For voting?

A key question for piloting a European wide public sphere in the context of electoral choice is whether the post-deliberation policy preferences and political viewpoints are connected to the final voting intentions. We are especially interested in any effects on voting from the post-deliberation policy attitudes on climate change and immigration. Yet the basic political orientations captured by the left-right self-placement are of interest as well. Ideally a citizen who considered the arguments and became informed about the issues would connect her policy attitudes and basic political viewpoints to her voting decisions. Given the spectacular change in voting intentions of the participants for the Greens we are especially interested in whether their post-deliberation voting support has any connection to post-deliberation substantive policy positions or political views.

In the case of the Greens, there does seem to be support for such a connection. Table 6 shows the estimates of multinomial logit models with the immigration and Climate Change indices and left-right self-placement as explanatory variables, and voting for the Greens as the dependent variable. Those who chose other parties or no parties at all constitute the omitted category. The tests presented for Table 6 are one-tailed as we expect that climate change, immigration and left right support will have effects on support for the Greens.

Table 6.

Explaining the Green Vote T₁ and T₃.

Parties		T₁			T₃
Parties		b	S. E.	p	b	S. E.	p
GREENS	Immigration	4.009	1.863	.031	3.758	1.333	.005
	Climate change	3.942	1.542	.011	3.932	1.025	.000
	Left-right	−0.888	1.119	.426	−1.689	0.678	.013
	Intercept	−8.041	1.987	.000	−6.451	1.304	.000
	N	330			330
	χ2	21.98			51.23
	Pseudo R²	.157			.170

Note: Reference categories are participants who selected other parties or did not offer a selection.

Separate analyses conducted with the inclusion of demographic variables did not alter the picture, so we have limited our presentation to this version, confining the explanatory variables to these more proximate ones.

Note that the immigration index is defined so that higher scores indicate a more welcoming position toward immigrants and lower scores indicate a more restrictive position. Hence the significant results for immigration, both before and after deliberation, show a connection in the correct direction for the Greens, who advocated a welcoming or more tolerant position on immigration. Similarly, the Climate Change index is defined so that higher scores indicate more support for action to resolve the problem despite the costs. The significant results for climate change, both before and after deliberation are also in the right direction. Lastly, the Left-Right index is defined so that higher scores indicate a more conservative position. The significant result after deliberation is also in the correct direction for the Greens. These post-deliberation results indicate the levers of opinion change supporting the final voting preferences. As noted earlier, the percentage supporting the Greens rose more than three-fold after deliberation (T₃).

Of course, the probability of a coefficient estimate being likely or unlikely to have occurred under a null hypothesis of no effect (i.e. being statistically insignificant or significant) says nothing about the effect’s magnitude. The effects, moreover, are not the coefficients (or any other constants)—not in a nonlinear (and nonadditive) model like the logit model here. Thus to give a sense of the magnitudes of the effects on the probability of voting Green, Table 7 reports the estimated first-differences between participants at the most propitious (pro-immigration, favoring action to combat climate change, and identifying with the left) and most unpropitious extreme (anti-immigration, opposing action to combat climate change, and identifying with the right on each of the regressors. The first row gives the first-differences for attitudes toward immigration, the second row the first-differences for attitudes about acting to combat climate change, and the third row first-differences for left-right self-identification. Let Imm and CC denote immigration and climate change attitudes, and LR denote left-right self-location. Each column supplies a numerical scenario for the other two variables: that they both extremely unpropitious (CC = 0, LR = 1), both somewhat unpropitious (CC = .25, LR = .7) , both neutral (CC = LR = .5), both somewhat propitious = (CC = .75, LR = .25), or both extremely propitious (CC = 1, LR = 0).

Table 7.

First-differences.

		T₁					T₃
		Scenarios for the other two regressors					Scenarios for the other two regressors
		Extremely unpropitious	Somewhat unpropitious	Neutral	Somewhat propitious	Extremely propitious	Extremely unpropitious	Somewhat unpropitious	Neutral	Somewhat propitious	Extremely propitious
Greatest possible difference to the probability of voting Green	Immigration	.007	.023	.074	.210	.461	.012	.047	.167	.439	.701
	Climate change	.007	.022	.072	.207	.460	.014	.054	.180	.452	.712
	Left/Right	.000	.001	.010	.062	.203	.001	.009	.055	.250	.386

Note: Entries are differences in the predicted probabilities of voting for the Greens. The difference is between the probability when the row variable is at its theoretical maximum (1) and the probability when it is at theoretical minimum (0), in each column under the indicated numerical scenario for the other two regressors. The extremely propitious values are 1 for CC and Imm and 0 for LR, the extremely unpropitious ones the reverse.

For instance, the upper left-hand corner cell contains

\overset{\land}{p} (y_{i} = 1 | Imm = 1, CC = 0, LR = 1) - \overset{\land}{p} (y_{i} = 1 | Imm = 0, CC = 0, LR = 1)

where

\overset{\land}{p} (y_{i} = 1 | Imm = 1, CC = 0, LR = 1)

is the estimated probability (based on the estimated coefficients in Table 6) of voting Green (

y_{i} = 1

) for people who are completely opposed to combating climate change (CC = 0), at the extreme right (LR = 1), and completely pro-immigration (Imm = 1) minus the estimated probability for people who are likewise completely opposed to combating climate change (CC = 0) and at the extreme right (LR = 1) but completely anti-immigration (Imm = 1). The third entry in the same row contains the parallel difference with CC and LR held constant at .5 instead of 0, and so on. Each entry indicates the greatest possible difference that any change in given regressor (from its minimum to its maximum) can make to the probability of voting Green under each of the five column scenarios for the other two regressors.

The results suggest that attitudes toward climate change and immigration can make a very large difference to the probability of voting Green when the other two variables also favor doing so; that left-right self-location can make a somewhat smaller difference, again when the other two variables favor voting Green; and that all three variables can make a substantially bigger difference after deliberation (at T₃) than before (at T₁).⁴

Table 8 shows the same analysis as Table 6 at T₁ and T₄ both for the treatment group of deliberators and a separate control group of 682 that was interviewed only at T₁ and then again just after the elections, at T₄. The control group did not deliberate in the Europolis event but like all citizens would have possibly had exposure to the campaign. The tests presented are one-tailed. For the deliberators, one can see connections post-deliberation between immigration, climate change and the left-right dimension for the Greens. For the control group only the left-right dimension is significant. In general, we think the spectacular rise in support for the Greens after deliberation, combined with this evidence of a coherent connection to substantive views, shows a deliberative effect on voting in the case of the Green party.

Table 8.

Explaining the Green Vote, T1 and T4.

Participants
		T1			T4
Parties		b	S. E.	p	b	S. E.	p
GREENS	Immigration	3.878	1.864	.037	4.184	1.524	.006
	Climate change	3.965	1.539	.010	4.335	1.254	.001
	Left-right	−0.833	1.112	.454	−1.923	0.817	.019
	Intercept	−7.943	1.984	.000	−7.360	1.537	.000
	N	315			315
	χ²	21.69			50.44
	Pseudo R²	.157			.214
Control group
		T1			T4
Parties		b	S. E.	p	b	S. E.	p
GREENS	Immigration	0.323	1.456	.829	1.927	1.179	.102
	Climate change	1.717	1.178	.145	0.902	1.004	.369
	Left-right	−2.280	0.943	.016	−1.900	0.776	.014
	Intercept	−3.891	1.314	.003	−3.956	1.090	.000
	N	675			675
	χ²	9.72			12.43
	Pseudo R²	.056			.052

Note: The reference category is other parties or no party selected.

Voting and the major parties

The situation is murkier with the other, mostly larger parties. Table 9 shows mulitinomial logit estimates for models explaining support for each of the four main party groupings at T₁ and T₃. The tests presented are one-tailed. At both T₁ and T₃ the coefficient estimates are generally biggest (in absolute value) for the Greens, the only exception being the coefficient estimate for left-self placement at T₁, which is smallest for the Greens. After deliberation (at T₃), that coefficient estimate, too, becomes biggest for the Greens. For the Greens, both climate change and immigration matter at T₁ and continue to matter at T₃. All the significant coefficient estimates have the right sign. For the other parties we see that the participants connected their left-right self-placement correctly with support for the PES (Socialists) and EPP (center right) but there was no connection for the small Alliance of Liberals and Democrats for Europe (ALDE) contingent. Furthermore, climate change and immigration did not have the clear effect on support of these parties that it did for the Greens. In many ways this is not surprising. The Greens had well-defined and strong positions on both climate change and immigration. The other parties were more ambiguous.

Table 9.

Voting and the major parties.

Parties		T₁			T₃
Parties		b	S. E.	p	B	S. E.	p
PES	Immigration	1.454	1.299	.263	0.647	1.397	.643
	Climate change	−1.088	0.825	.187	−1.157	1.017	.255
	Left-right	−3.152	0.865	.000	−2.078	0.850	.014
	Intercept	−0.449	1.107	.685	1.501	1.269	.237
EPP	Immigration	0.430	0.949	.650	−1.679	1.243	.177
	Climate change	−0.905	0.702	.197	−0.698	0.945	.460
	Left-right	2.843	0.683	.000	3.338	0.823	.000
	Intercept	−2.220	0.927	.017	0.953	1.170	.935
GREENS	Immigration	4.478	1.909	.019	3.414	1.580	.031
	Climate change	3.499	1.576	.026	2.878	1.215	.018
	Left-right	−0.845	1.203	.482	−1.463	0.895	.102
	Intercept	−7.463	2.034	.000	−4.021	1.536	.009
ALDE	Immigration	0.670	1.605	.677	0.014	1.810	.994
	Climate change	−0.470	1.177	.690	−2.107	1.313	.109
	Left-right	1.540	1.116	.167	0.506	1.126	.653
	Intercept	−3.158	1.549	.041	0.381	1.634	.816
Other Party	Immigration	2.447	1.728	.157	−1.042	1.552	.502
	Climate change	−1.717	1.100	.119	−2.865	1.151	.013
	Left-right	0.311	1.125	.782	0.767	0.998	.442
	Intercept	−2.824	1.547	.068	1.795	1.409	.203
	N	330			330
	χ2	81.47			127.91
	Pseudo R²	.090			.114

While substantive issues about the arguments for and against general policy orientations for climate change and immigration were the explicit subjects of deliberation, the party positions on these issues were not an explicit topic. A section at the back of the briefing document included excerpts from party manifestos on these issues but these were ambiguous in most cases. In contrast, the Greens proposed the EU ‘to commit to emissions reductions of 40% at the very least by 2015 and of 90% by 2050.’ They also proposed ‘a combination of ambitious and binding targets, of incentives and of public investments into green technologies and services [which] can help create millions of green jobs in Europe … at a time of economic slowdown.’ The Greens made a strong commitment to renewables and rejected nuclear energy. On immigration, the Greens said that they ‘oppose the siege mentality of Fortress Europe’. They see ‘immigration as an opportunity’ and recommend ‘positive-minded policies that will allow people … to come here legally and efficiently … with equal rights and equal pay, as well as the opportunity of European citizenship.’ In addition: the Greens ‘oppose repressive laws on returning unauthorized migrants’ and ‘inhumane or xenophobic legislation’. The Greens also proposed ‘more efficient cooperation among the EU countries’ to ‘fight the despicable trafficking of men, women and children.’

While the other parties offered some hints, none had unambiguous and strong positions that were comparable. The closest was the EPP on climate change, which proposed ‘international dialogue in order to reach a decision on a post-2012 agreement by the end of 2009 in Copenhagen.’ The EPP supported a 20% target for renewables by 2020. On immigration, the EPP ‘proposes to fight illegal migration at the EU level, starting from the needs, the capacity and the priority set by each member state.’ On the one hand the EPP supported ‘a fair but firm illegal migrants return policy’, but it also ‘proposes to implement the European preference to tackle skill shortages in the Member States and encourage intra EU migration.’ Thus the EPP had elements on both sides of the immigration debate while offering modest support for action on climate change.

The PES manifesto quoted in the briefing materials offered ‘to establish common standards for legal migration into the European Union’ and to coordinate ‘European efforts to combat illegal migration’ by ‘strengthen [ing] cooperation with third countries.’ The manifesto quotations offered something for both sides. On climate change, the socialists supported ‘the achievement of a new global climate agreement for the post-2012 period’ aimed at granting ‘a 30% global target for emissions reductions by 2020.’ It also proposed to establish ‘a global energy and development forum’ and ‘to increase EU support for developing countries to fight … climate change.’ Note that the support for global targets would depend on ‘a new global climate agreement’ and lacking such an agreement would not have any direct policy implications for the EU. A new ‘forum’ and some assistance for developing countries both offer some symbolism but neither directly deals with the kinds of policy tradeoffs for EU level decisions discussed in the deliberations for climate change.

The only statement on the immigration issue for the ALDE (identified as the ELDR in the manifestos) was ‘the introduction of an EU “blue card” system administered by each member state to ensure measured economic migration for the benefit of EU citizens.’ No statements for this party grouping were included for climate change. Hence even if a participant had been motivated to try and place the ALDE on these issues, there was really no clear basis provided for doing so (see Europolis, 2009).

In sum, it seems that the one clear opportunity given to the deliberators to connect policy positions with their voting intention was provided by the Greens. The participants clearly exercised this opportunity. The murkiness of the rest of the voting intention story, apart from the effect of general left-right orientations, is really testimony to the limitations of a field experiment conducted during an actual election. The project was limited by the ambiguities of position on the part of the actual party groupings in an actual (second-order) election.

Conclusion

Europolis was the first test of the unitary public sphere in the context of an actual election. It was an empirical exploration, in a quasi-experiment, of a counterfactual ideal. It was certainly not intended to be predictive of actual voting behavior on the part of those millions of voters who were not subject to the treatment. Therefore what purpose did it serve?

The process was reasonably successful at fulfilling our six criteria. The evidence has gaps but the picture that emerges, despite all the differences of language and nationality, is that it is indeed possible for the people of Europe to deliberate together across the many barriers of language and nationality, to weigh competing arguments, become more informed and arrive at considered judgments both about policy issues and voting intention. The one clear question mark on our list was the ability of the participants to coherently connect their policy views with voting intentions for any parties other than the Greens. We think a major factor is the ambiguity and elusiveness of the substantive positions of the European party groupings. Perhaps as the elections to the Parliament come to be seen as more consequential, and as the elections become more competitive, parties will eventually be motivated to further specify their positions. Then again, perhaps not. There are well known electoral benefits to ambiguity (Page, 1976). But for the European wide public sphere to take the final step of connecting considered judgments with electoral choice, there have to be clear policy differences among the competing parties. Here again, the ‘what if’ of this quasi-experiment faced the limitations that it took place in a real-life context of an actual election.

Our discussion of the six criteria and their application to Europolis suggest three conclusions. First, the mini-public or micro-cosmic version of the EU wide public sphere is a viable democratic tool, something that even critics of deliberative democracy grant in principle (see Moravscik, 2006 admitting this possibility). If there is at some point enough interest in bringing the people of Europe into decisions in a thoughtful and representative way, then the mini-public version can be convened, just as the nomethetai in ancient Athens or the Citizens' Assemblies in Canada.

Second, the actual convening of a European wide public sphere in microcosm among the citizens of Europe does a great deal to dramatize the gap between the mini-public and the actual conditions of the European publics in mass society. The dialogue was facilitated by simultaneous interpretation in 22 languages. Obviously this kind of shared public dialogue across the boundaries of language and nationality will not take place on a mass scale in the foreseeable future.

Yet does that undermine the European wide public sphere in its unified version, as an ideal? This is a contested issue but it suggests a third conclusion. If an ideal is not coherent in theory, then one may be left with ‘ideals without an ideal’ with a picture of where policies and reforms should go that does not hold together and, in its incoherence points to different directions for incremental change (Fishkin, 1992). That is not the case with the European wide public sphere. It does not self-destruct or undermine itself. It is just very distant from current practice. We can thus think about ways of approaching it, improving the European wide dialogue, improving the information given to voters, improving trans-cultural communication and citizen engagement across national boundaries so that this ideal, coherent but very distant, is incrementally approached.

Lastly, it is worth distinguishing the counterfactual elements of Europolis that apply especially to Europe from those that apply to any large scale DP or microcosmic deliberation. The European citizenry comprises the electorates of 27 countries (now 28), with 22 languages. In addition, the public spheres of each of these countries are naturally focused on the issues that are germane to national decision. The public would therefore be ill prepared for European issues even if there were no differences in language. Yet the fact that the public sphere is in silos by nation state and language exacerbates the problem. When a national DP takes place in a given country, in the US or Japan for example, it is building on a national discussion on the issue in question. In the case of the EU, the lack of such a shared discussion across all the states is one of the motivations for the experiment.

The project also explored whether there could be a coherent connection between post-deliberation political views and voting intention in a European context. It was a test of European wide public will formation, or deliberative democracy. It is arguable that even national elections, in modern polities, rarely exhibit anything approaching deliberative will formation. Campaigns rely on the persuasion industry and mobilization in order to engage in a Schumpeterian ‘competitive struggle for the people’s vote.’ Parties attempt to win campaigns whether or not they can offer good reasons for their campaign positions or policy choices. Modern democracies depend primarily on party competition, not deliberation. Thus the idea of Europolis is idealistic, or counterfactual, both in the sense that it applied to Europe and in the sense that it aspired to reason-based collective will formation, albeit at the European level. The fact that such an aspiration could be realized to even a partial degree is striking and may well be useful in the continuing normative and empirical debate embodied by the EU and its future democratic aspirations.

Footnotes

Authors’ note

Deliberative Polling is a registered trade mark of James S Fishkin. It is trade marked for quality control and supports the research at the Center for Deliberative Democracy at Stanford.

Acknowledgments

The authors would like to thank the editors and reviewers for their comments. We would also like to thank Gaurav Sood for his insights and help with some of the analyses.

Notes

References

Bohman

Rehg

(1997) Deliberative Democracy: Essays on Reason and Politics, Cambridge, MA: MIT Press.

de Swann

(1993) The evolving European language system: A theory of communication potential and language competition. International Political Science Review 14(3): 241–255.

de Vreese C and Schmitt H (eds) (2007) A European Public Sphere: How Much of It Do We Have and How Much Do We Need? Connex Report Series No. 2. Available at: http://www.mzes.uni-mannheim.de/projekte/typo3/site/fileadmin/BookSeries/Volume_Two/Volume_2_Public_sphere_online.pdf (accessed 21 March 2014).

Downs

(1957) An Economic Theory of Democracy, New York, NY: Harper and Row.

Elster

(1998) Deliberative Democracy, Cambridge: Cambridge University Press.

Eriksen

Fossum

(2000) Democracy in the European Union: Integration Through Deliberation, London: Routledge.

Europolis (2009) Briefing Materials. Available at: http://cdd.stanford.edu/docs/2009/2009-europolis-EN.pdf (accessed 21 March 2014).

Fishkin

(1992) The Dialogue of Justice: Toward a Self-Reflective Society, New Haven, CT: Yale University Press.

Fishkin

(2009) When the People Speak: Deliberative Democracy and Public Consultation, Oxford: Oxford University Press.

10.

Fishkin J and Laslett P (eds) (2003) Debating Deliberative Democracy. Oxford: Blackwell.

11.

Fraser

(1992) Rethinking the public sphere: A contribution to the critique of actually existing democracy. In: Calhoun

(ed) Habermas and the Public Sphere, Cambridge, MA: MIT Press, pp. 109–142.

12.

Gerber M, Bächtiger A, Fiket I, et al. (2014) Deliberative and non-deliberative persuasion: Mechanisms of opinion formation in Europolis. European Union Politics 15(3): 410–429.

13.

Grimm

(1995) Does Europe need a constitution? European Law Journal 1(3): 282–302.

14.

Guttmann

Thompson

(1994) Why Deliberative Democracy?, Princeton: Princeton University Press.

15.

Habermas

(2009) Europe: The Faltering Project, Cambridge: Polity.

16.

Habermas

(2012) The Crisis of the European Union: A Response, Cambridge: Polity.

17.

Hansen

(1991) The Athenian Democracy in the Age of Demosthenes, Oxford: Blackwell.

18.

Kohler-Koch

Rittberger

(2007) Debating the Legitimacy of the European Union, Lanham, MD: Rowman and Littlefield.

19.

Koopmans

Erbe

(2004) Towards a European public sphere? Vertical and horizontal dimensions of Europeanized political communication. Innovation 17(2): 97–118.

20.

Luskin RC, Fishkin JS and Jowell R (2002) Considered opinions: Deliberative polling in Britain. British Journal of Political Science 32(3): 455–487.

21.

Marsh

(1998) Testing the second-order election model after four European elections. British Journal of Political Science 28: 591–607.

22.

Marsh

Mikhaylov

(2010) European parliament elections and EU governance. Living Reviews in European Governance 5(4): 5–30.

23.

Moravscik A (2006) What can we learn from the collapse of the European constitutional project: A response to eight critics. Notre Europe. Available at: http://www.notre-europe.eu/media/moravcsik-reponse-en_01.pdf?pdf=ok (accessed 21 March 2014).

24.

Offe

(2011) Crisis and innovation of liberal democracy: Can deliberation be institutionalized? Czech Sociological Review 47(3): 447–472.

25.

Page

(1976) The theory of political ambiguity. American Political Science Review 70(3): 742–752.

26.

Reif

Schmitt

(1980) Nine second order national elections – a conceptual framework for the analysis of European election results. European Journal of Political Research 8(1): 3–44.

27.

Sanders

(1997) Against deliberation. Political Theory 25(3): 347–377.

28.

Schlesinger

(2007) A cosmopolitan temptation. European Journal of Communication 24(4): 413–426.

29.

Schmitter P and Treschel A (nd) Green Paper on the Future of Democracy in the European Union for the Council of Europe. Available at: http://www.eui.eu/Documents/DepartmentsCentres/SPS/Profiles/Schmitter/GreenPaper.pdf (accessed 21 March 2014).

30.

Van de Steeg

(2002) Rethinking the conditions for a public sphere in the European Union. European Journal of Social Theory 5(4): 499–519.

31.

Van Parijs

(2011) Linguistic Justice for Europe and for the World, Oxford: Oxford University Press.

32.

Verba

Schlozman

Brady

(1995) Voice and Equality: Civic Voluntarism in American Politics, Cambridge, MA: Harvard University Press.

33.

Warren M and Pearse H (eds) (2008) Designing Deliberative Democracy: The British Columbia Citizens Assembly. Cambridge: Cambridge University Press.

34.

Young

(2002) Inclusion and Democracy, Oxford: Oxford University Press.

Supplementary Material

Please find the following supplemental material available below.

For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.

For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.

0.62 MB

0.00 MB

0.09 MB