Democratically speaking - Can citizen deliberation be considered fair and equal?

Democratically speaking - Can citizen deliberation be considered fair and equal? Staffan Himmelroos Åbo Akademi University Fänriksgatan 3a, 20500 ÅBO...
0 downloads 0 Views 369KB Size
Democratically speaking - Can citizen deliberation be considered fair and equal?

Staffan Himmelroos Åbo Akademi University Fänriksgatan 3a, 20500 ÅBO [email protected]

Paper presented at the ECPR Joint Sessions, Workshop 'Frontiers of Deliberation' April 12 – 17, 2011, University of St. Gallen, Switzerland

Detta är alltså en text som presenterats i samband med ECPR Joint sessions. Jag ville dock även presentera den på detta allmänna seminarium eftersom det är en text som relativt väl representerar mitt avhandlingsarbete överlag. Därmed torde den kunna ge en sådan inblick som gör det möjligt att komma med konstruktiv kritik som kan utnyttjas i själva avhandlingsarbetet. Texten i sig är dock tänkt att kunna utnyttjas som underlag för en artikel. Därmed tas också tips om hur det här bäst låter sig göras emot med tacksamhet.

1

Introduction

The notable deliberative turn democratic theory took in the 1990s has in the last few years become evident in empirical research as well. But, despite this surge of interest in deliberative approaches, the lack of empirical research in certain fundamental areas is surprising, if not to say troubling. According to the proponents of deliberative democracy rationally motivated reasons have the power to shape opinions, i.e. a power to change people’s views and preferences, in order to achieve mutual understanding and unforced consent (Miller 1993). This transformation of opinions is expected to emerge in the light of new information, through the consideration of different perspectives and by reflecting upon one’s own views (Rosenberg 2007). Empirical evidence of how the deliberative process plays out in practice is, however, still relatively scarce. A comprehensive review by David Ryfe (2005), indicates a large gap in the research concerning what actually transpires when people deliberate, while John Dryzek (2007: 237-39) points out that in the case empirical evidence is presented, it is often in the form of general claims, or stylized facts. Ryfe (2005: 54) further observes that to the extent rigorous empirical research has been carried out, the focus has often been on the effects of the deliberative process. Outcomes or effects, like change of opinion, is however not proof of a fair and inclusive deliberative process. The outcome may actually be the result of various social dynamics and underlying power structures, rather than that of an inclusive and enlightened discussion between equals (Ryfe 2005). Since deliberative democracy largely rests on procedural standards, it can hardly be considered democratic if the process is guided by something else than reasoned argumentation, or if some are systematically disadvantaged in their ability to influence the process. We should therefore not only be concerned with whether the process is one of reasoned argumentation, but also look into whether the expectations of a fair and equal process have been met. The aim of this paper is to shed some light on the workings of citizen deliberation through systematic empirical evidence and to be able to do that I am using data that was gathered from a deliberative mini-public experiment. Many scholars with an interest in deliberative democracy, put their faith in citizen deliberation rather than in deliberation among political elites, suggesting that it is among ordinary citizens we are most likely to encounter an unbiased and non-coercive decision-making process based on deliberation (cf. Fishkin 1997, Brown 2006). Mini-publics are expected to provide favorable conditions for citizen deliberation, since they are especially designed with the deliberative environment in 2

mind (Smith 2009). Assuring a constructive environment for deliberation will subsequently allow rigorous testing of the theoretical assumptions. If democratic deliberation is unable flourish within this design, it is unlikely to take root elsewhere. To capture elements of procedural quality I have used a modified version of the content analytical measure Discourse Quality Index, also known as DQI by Steiner et al. (2004).

Democratic deliberation and its critics

According to a generally acknowledged understanding within deliberative theory, ideal deliberation rests on individuals interacting with carefully weighed reasons and morally justified arguments in a context of mutual respect (Habermas 1981, Cohen 1989, Dryzek 1990, Chambers 1996, Bohman 1996, Gutmann & Thompson 1996). What does this imply? First of all, the participants are expected to able to communicate their views in a calm and rational manner, presenting arguments supported by logically sound reasons. However, ideal deliberation requires not only the logical justification of assertions and validity claims; the arguments should also have intrinsic characteristics that make them compelling to others (Cohen 1989). The participants must be able to consider the well-being of others and of the community at large. According to the political philosopher T.M. Scanlon (see Chambers 1996), a reasonable person is not simply the person who has good reasons for her actions; it is also a person who is willing to listen to objections, who is open to suggestions, and who will reevaluate a position in the light of new evidence. Bohman (1996: 63) explains that even if I do not adopt your views, a different perspective may cause me to modify my own reasons, especially if they do not convince those who disagree with me. In order to convince you I may have to take your perspective, and vice versa. Moreover, being open to other opinions and having them influence one’s own views also require a fundamental respect for the other participants and their arguments. Young (2000: 38) maintains that without respectful consideration of other participant’s arguments a deliberative process can never bridge different views or stands that must act alongside each other in every pluralist society. The description of deliberation above seems to require considerable reflective capacities from the ones taking part in the decision-making process. Individuals are assumed to be able to consider and arrange their specific preferences and values relative to their sense of higher order good or at least to be able to incorporate the good of others with their own view. They are also assumed to be able to take the perspective of others and thus fairly 3

consider the claims of another person in that person’s terms. As well as, consider not only the personal value of specific actions or outcomes, but also the common value of general principles of interaction. (Rosenberg 2004) Many are, thus, skeptical of whether real-world deliberation can be compared to the ideal at all. The Habermasian ideal speech situation, where the normative ideals of deliberation are expected to prevail, has been strongly criticized for relying on an assumption that under ideal conditions language would be fully transparent, i.e. its meaning is accessible to all (Kohn 2000: 410). Dryzek & Niemeyer (2008: 2) point out that the problem with discourses is that they constrain as well as enable thoughts, speech, and action. Every discourse embodies some conception of common sense and acceptable knowledge, and thus, it may act as an expression of power by recognizing some interests as valid while repressing others. It has been suggested that the Habermasian take on rational discourse represents a form of communication that is characteristic of some groups, while it tend to exclude others. Sanders (1997: 2) contend that those who are less likely to present their arguments in ways that we recognize as characteristically deliberative, are likely to be those who already are underrepresented and systematically disadvantaged in formal political institutions, namely women, racial minorities, and poorer people. Education is another aspect which is likely to cause inequalities in deliberation, just as it does in other forms of political participation (cf. Verba 2001). The well educated are more likely to have sophisticated reasoning skills, while people with little education lack access to occupations where reasoning and public speaking skills develop (Mendelberg 2002). According to Mendelberg (2002) people also vary in their need for cognition, i.e. their motivation to think in depth about the essential merits of a message. It has been established that people high in need for cognition not only generate more arguments, but that they also generate more well-reasoned arguments to support their claims. Critics have even stated that, “it is hard to avoid the suspicion that deliberative democracy is the ‘democracy’ of elite intellectuals” (Posner 2003: 136). This skepticism shoot from a belief that deliberative democracy is far from as egalitarian as it presents itself. Another common concern related to this challenge is that alternative forms of participation tend to engage the already politically active that possess the necessary skills to participate in public life, while the ones who are not active in first place will be ever more marginalized (Ryfe 2005). Since the dynamics of the deliberative process inevitably must be affected by what one feels is at stake, we cannot overlook what motivates participants in a deliberative process. If you feel strongly about an issue, it is likely to make you participate more actively, while if you are less certain of what to think you may easily take the backseat in the debate. This may 4

be quite problematic for deliberation since, according to Mutz (2006), people who are passionate about an issue seldom have the will to ponder different alternatives or to consider other people’s views. If people with strong views come to dominate the discussion, deliberation might actually result in deeper conflict, rather than a bridging of conflicting views.

Research design

This paper revolves around two research questions. First of all, what does citizen deliberation actually look like? As we have seen there was quite a deep gulf between the normative ideal and the real-world challenges put forward by its critics. To be able to examine the deliberative process in the light of these diverging perspectives we need operationalizations that allow us to empirically test the width of this gap. Secondly, does everyone have an equal standing in the process of deliberation? What is equal deliberation, is it only the equal opportunity to make your voice heard? Since a legitimate decision according to deliberative theory is supposed to be the result of a process involving everyone affected, it probably requires a bit more than that (Manin 1987: 352). Knight and Johnson maintain in their article What sort of equality does deliberative democracy require (1997) that democratic deliberation requires equal opportunity of access to political influence. Everyone should be recognized as having equal standing through the process, where they can propose issues for the agenda, propose solution to the issues on the agenda, and offer reasons in support or criticism of proposed solutions. This suggests that it is something more than mere access to the process, and that the kind of equality required for democratic deliberation will have both procedural and substantive dimensions. I have simply defined a fair and equal process as one where all participants (everyone affected) have the capacity to take part in an exchange and evaluation of reasoned arguments. Furthermore they must not be unfairly disadvantaged in this process by deficiencies that are due to conditions or circumstances beyond their control. That is they should be able to argue and reflect on different perspectives in a context of mutual respect and their contribution must not be undermined due to systematic disadvantages.

Methods for capturing democratic deliberation To able to gather empirical data we need to operationalize our theoretical dimensions. We need to know the something about the both the quantity and the quality of different elements 5

in the debate. Since we are interested in their distribution among different individuals, in order to establish whether the deliberative process can be considered fair and inclusive, some type of content analysis would seem the most appropriate. The fundamental idea of content analysis is to deduce speech or text to a number of categories and it offers an opportunity to look at both frequencies of different categorizations, as well as, how they are connected to particular actors (Krippendorf 2003, Brydér 1985). While there have been other attempts at developing content analytical measures for the analysis of the deliberative process (see Dutwin 2002, Holzinger 2004, Stromer-Galley 2007 etc.1), the most interesting and perhaps most advanced instrument developed for this purpose is the Discourse Quality Index (DQI for short) by Steiner et al. (2004). The DQI was originally developed as a discourse measure for deliberative quality in parliamentary debates, and while it mainly draws on Habermasian discourse ethics, other views of deliberative democracy have also been incorporated in the measure as well. The strong and apparent connections to the core of deliberative theory combined with its functional outlook sets it apart from most, if not all, other attempts at analyzing the deliberative process. It has also been met with considerable support from deliberative theorists (Habermas 2005, Thompson 2008), which can be considered a strong indication of its validity. But no matter how valid the DQI is as a measure of deliberative quality, it has certain weaknesses when it comes to examining citizen deliberation. The DQI is quite distinctly shaped by what it was set out to analyze, namely parliamentary debates. Deliberation involving citizens can, however, be quite different from parliamentary debates (see Wales et al. 2009, Himmelroos 2010). The communication between citizens tends for instance to be more dialogical than parliamentary debates, where the interaction easily becomes quite structured and monological. Whereas the argumentation certainly can be quite erratic in citizen deliberation, it is usually also more dynamic and organic in its development. Citizen deliberation will very likely include a greater degree of both misunderstandings and cooperation before any kind of common definitions or goals can be set up. The original DQI was composed of seven indicators: Participation, Level of Justification, Content of Justification, Respect toward groups to be helped (empathy), Respect toward the demands of others, Respect toward counterarguments, and Constructive politics. Based on a pilot study using DQI in the context of citizen deliberation (Himmelroos 2010), some indicators have been modified, while others that can be measured more accurately in

1

Not including all the measures developed for text-based internet communication (cf. Jansen & Kies 2004)

6

other ways have been left out altogether. These modifications mainly concern the respect indicators, since they, are predisposed by the purpose of the original measure, i.e. capturing discourse quality in a parliamentary setting. In most cases, lay citizens are not as easily divided into factions or group interests as are elected representatives, at least if the issue at hand doesn’t concern a situation where there is an open conflict. This lack of distinct divisions makes it difficult to identify whether the participants are truly considerate of views different to their own. By this I mean that in cases where distinct groups or factions can’t visibly be discerned, it will be extremely hard to tell whether participants are respectful toward a view different from their own, or if they just concur because they share the same view. When examining discourse quality, a distinction is made between relevant and irrelevant parts in each speech, and only speeches with relevant parts are coded. A relevant part is one that contains a demand, i.e. a proposal on what should be done (or not). The emphasis on demands by Steiner et al. (2004) stems from the idea that demands stipulate what ought to be done, and this normative character puts them at the center of discourse ethics (Steiner et al. 2004: 55). When it has been established that a speech includes a demand, it is evaluated according to the now four categories of a discourse measure where the characteristics of citizen deliberation have been taken into account: (I) Level of Justification, (II) Content of Justification, (III) Respect and (IV) Reciprocity. Whatever the form of the (I) justification, the crucial point is that the participants present their arguments in a logically coherent way, so that other participants can understand the argument. The tighter the connection between premises and conclusions, the more coherent the justification is and the more useful it will be for deliberation (Steiner et al. 2004: 21). According to the deliberative ideal, arguments should, however, not only be logical and coherent, but also expressed in the terms of the common good. (II) Content of justification aims to capture the reason behind one’s argumentation. Are you only looking out for your own interests or are you considering others’ as well. Appeals to the common good can take different forms. On the one hand, the common good may be stated in utilitarian terms, i.e. as the best solution for the greatest number of people (Mill 1859). The common good may also be expressed through the difference principle in the sense that the common good is best served if the least advantaged are helped (Rawls 1971). In order to generate an environment where arguments can be appreciated at least a modicum of respect is needed. (III) Respect is a measure of how considerate participants are of each other. But, it may also act as a measure of empathy, as it reveals attitudes towards 7

different persons or groups under discussion. As noted before, citizen deliberation may not be as factional as parliamentary debates, and in this case respect is perhaps best determined by the lack thereof, in cases of explicit disrespect. The use of say, inflammatory rhetoric is an indicator that interlocutors are not very interested in finding agreement, persuading their opponents or easing conflict (Chambers 1996: 209). The level of respect is therefore captured by looking for instances of disrespect. Two types of respect are included in this measure, internal and external; internal referring to respect between participants and external referring to comments of persons or groups outside the small group. For all empirical purposes the two measures of respect have however been collapsed into one general respect indicator, since the cases of internal disrespect are too infrequent to be evaluated on their own. Deliberation also relies on the fact that everyone reflects on what is being said and evaluates each argument and the way it relates to their own opinions (Bohman 1996) The last indicator (IV) reciprocity, can therefore perhaps best be understood by looking at how participants treat arguments that contradict their own, or if they touch upon them at all (Steiner et al. 2004). Another way to identify whether the participants act in a reciprocal manner would be to see if different, or conflicting, demands are weighed or compared by a participant. Table 1. Indicators of Discourse Quality

I. Level of Justification 0: no justification, presents only his/her point of view 1: inferior justification; conclusion(s) embedded in (an) incomplete inference(s), no linkage is made as to why X will contribute to Y 2: qualified justification; one conclusion embedded in a complete inference, other incomplete inferences may be present, a linkage is made as to why X will contribute to Y 3: sophisticated justification; more than one conclusion, each embedded in a complete inference

II. Content of Justification 0: explicit statement concerning group/ self- interest 1: neutral statement; no reference to group or self- interest, but no reference to common good either 2a: explicit reference to common good in utilitarian or collective terms 2b: explicit statement in terms of the common good with reference to the difference principle

III. Respect 0: disrespect; explicitly negative statement concerning either the group/person under discussion or the other participants and their views 1: implicit respect; no explicitly negative statement concerning the group/person under discussion nor the other participants and their views

IV. Reciprocity 0: No reference to demand of others or self-linked 1: Reference to demand of others 2: Considers counter-argument in own argumentation or compares/weighs different arguments

8

Where to look for deliberation and what to look out for It is important to recognize that deliberative virtues may not emerge naturally. It has been argued that the revising character of deliberation, i.e. the power to transform views, would be undermined if the reasons were not publicly discussed (Gutmann & Thompson 1996, Elster 1997). The light of publicity is supposed to neutralize power and to protect the discursive process from collapsing into subtler forms of power or illicit consensus not grounded in a fair and equal process (Pyrcz 2004: 183, Parkinson 2006: 100). But, as e.g. Naurin (2005) has shown publicity may affect deliberation in different ways. The arena where deliberation takes place makes a difference, some arenas may constrain while others promote deliberative behavior. According to James Fishkin (2009) this would speak in favor of citizen deliberation, since ordinary citizens have less opportunity to bargain and act strategically than do political elites. In a citizen forum participants do not hold an office for which they may run for in an election. While the public forum may tempt the political elites to make grand gestures and resort to empty promises, ordinary citizens have neither reason nor opportunity to spin doctor their positions in order to gain advantage. We can usually assume that their responses to welldesigned deliberative consultations are genuine. (Fishkin 2009: 40) We should, furthermore, hold in mind that deliberative interaction generally takes place in groups and the success of deliberation is, thus, not only affected by the individual abilities and capacities of its participants, there might also be powerful social dynamics and group processes influencing the democratic procedure in such small groups (Gastil 1993). Even when deliberation is convened under the guise of promoting shared understanding or common ground, it is hard to imagine that members of dominant groups and members of subordinate groups strive for unity in an identical fashion, or that the airing of differences has identical implications from both majority and minority points of view (Walsh 2008: 58). Sunstein (2003) has for instance argued that opinions in group processes tend to shift toward the group norm. Karpowitz and Mendelberg (2007) underline the importance of understanding the effects of groups and group-level processes for deliberative dynamics, and point out the risk of deliberative outcomes being more a result of group composition than of reasonable argumentation. Sanders (1997:4) argue along similar lines saying that some people might be ignored no matter how good their reasons are, no matter how skillfully they articulate them,

9

because one cannot counter a pernicious group dynamic with good reason. Inequalities of various kinds will reduce the need for powerful participants to shift their perspectives. While what is different, distinctive or uncommon may be articulated, it is not, attended or acknowledged (Sanders 1997: 9). Since unequal deliberators will not be able to participate effectively, their dissenting points of view will tend to be assimilated to the contributions of the more effective participants (Bohman 1996: 65).

A citizen deliberation experiment in Turku 2006 The empirical analysis in this paper is based on an experiment held in November 2006. The design of citizen deliberation experiment was that of a deliberative mini-public (cf. Fung 2003 or Goodin & Dryzek 2006). The topic was nuclear power, or more specifically, the participants were asked to make a decision on the question “Should a sixth nuclear power plant be built in Finland?”2 The citizen deliberation experiment began by forming a random sample of 2500 adults from the Turku region, in south-western Finland. The final target sample for the experiment was 144 people, that is, 12 small groups consisting of 12 participants

each.

Of

the

invited,

135

participants

finally

showed

up

(10-12

participants/group). To counter the treacherous effects of social and group dynamics, as well as, other inequalities that may distort the deliberative process, different counter-measures were applied (cf. Smith 2009). Random sampling was used to bring about an inclusive process where all relevant views are represented. Besides random sampling and group allocation, participants received a neutral and balanced information material that contributed to their deliberative capacity by equipping them with basic command of the issue at hand. They also met with experts representing different interests to help them in their search for additional information and arguments, before they took part in facilitated group discussions. In the small group discussions they were encouraged to be respectful of and attentive toward other participants. Each participant was asked to come up with an issue, related to the main topic, for a common agenda that the discussion was to be based on. The trained facilitators overlooking the process were instructed to intervene only if the discussion halted or to encourage participants to speak up. The intention was to generate a free-flowing but, constructive deliberative environment, with a fair and balanced discussion. These measures leave us with something of a most likely

2

In 2006, the construction of the fifth nuclear power plant was under way in Finland, but there was already some discussion on the construction of a sixth nuclear power plant. In 2010 the Finnish parliament approved the construction of two new nuclear power plants.

10

case design, an environment where the participants have the greatest probability of attaining democratic deliberation. Democratic innovations such as mini-publics also provide a good opportunity to capture the process of citizen deliberation. All 12 small group discussions were recorded. But, unfortunately not all the group discussions were of a standard that would allow for proper comparisons. Recording 12 simultaneous discussions is not without its hazards and due to technical challenges (varying audio quality, two recordings failing at different points etc.) only eight of the small groups could produce transcriptions at the required level of detail, i.e. all speeches were captured and could by voice recognition be tied to the participant in question. The data I have analyzed is, thus, comprised of 8 small groups with 90 participants and total of 4882 individual speeches. The citizen deliberation experiment was designed around two decision-making methods (vote vs. consensus). The data analyzed here represents the part of the discussion before the decision-making commenced, around 3 hours where all groups had a matching treatment (3351 speeches).

Analyzing democratic deliberation To assess democratic deliberation I am looking at deliberative activity and quality. Activity is simply the amount of speeches each participant produced. The discourse quality measure, on the other hand, relies on an idea that deliberative actions can be placed on continuum from weak deliberation, with insufficient justification and disrespect, to ideal deliberation, with sophisticated justifications and respectful reciprocal communication. Each speech can be placed anywhere on this scale and thus give us an understanding of how close the discussion is to ideal deliberation. The higher a score a speech act receives on the four discourse quality indicators the closer they are to the ideal of communicative rationality. What are the determinants of activity and discourse quality within a deliberative process? Democratic deliberation is as we have seen associated with different characteristics, which individuals may possess to a varying extent. Besides some basic socio-demographics (age, gender and education) I have included five other independent variables. Within this context it is important to examine the participants understanding of issue at hand and their understanding of how the political system works. We should also look into whether earlier experience from small group decision-making affect participation. The former can be considered a more refined instrument for capturing cognitive capacity, than merely looking at education. The knowledge variable is based on a quiz where the participants’ familiarity with the nuclear energy issue as well as a few items on general political knowledge was tested. The 11

latter is based on whether the participant is an active member in an interest association or political party. Being an active member will probably have given you some experience of how small group deliberation (see Gastil 1993). It also acts as a test of whether the deliberative process favors the ones that already are politically active. What motivates people to be debate actively? Is it because you believe that you are able to influence the political process or perhaps because you have strong feelings concerning the issue at hand. Is there a real risk that people high on motivation, are low on deliberative intent? To examine this I will look into whether internal or external efficacy had any effect on deliberative activity and quality. Together with a measure that describes the intensity of each participant’s opinions on different energy related issues (see appendix), they can help us understand how different impetuses affects the deliberative process. Apart from their individual characteristics, the participants’ capacity to take part in the deliberative process may also be affected by different contextual factors that potentially could give the participants very different starting points. This issue can easily include a myriad of variables, but I will focus on two that are especially relevant to this study. The first relates to the design of the experiment where two different decision-making mechanisms, vote and common statement, where used. Even if the part of deliberative process analyzed here only concerns the part where small groups from both treatments had a parallel design (general discussion based on their common agenda), the participants in the common statement groups knew that they would eventually have to come to some form of agreement. Something the ones with a vote treatment did not have to worry about. The other variable that concerns the context of the deliberative process relates to position from which each individual where able to argue their case. Did they represent the majority or the minority view on nuclear power within their own small-group?

Results from the citizen deliberation experiment

The statistical analyses of the data collected in the citizen deliberation experiment are 12

conducted at two levels: I analyze developments at the level of the speeches and at the aggregated level of the individual participant. I will begin by presenting some descriptive findings that are easily retrieved from the data at the speech level. The purpose is simply to give an overall picture of what the deliberative process entails and what extent the process as a whole have managed to fulfill the assumptions presented in the theory. Besides looking at the data as a whole, I intend to examine some basic characteristics. Thereafter I examine the results at the participant level, first by looking at how they fared in general, as well as, possible differences between them. Finally I will conduct a more systematic analysis, where the all above-mentioned variables for individual deliberative capacity are controlled for.

The general level of deliberation As we can see from table 2 roughly one third (35.5%) of the speeches included a demand. It is quite hard to tell whether this number should be considered good or bad. What can be said is that it leaves a lot of room for other forms of communication than deliberative argumentation to prosper in the deliberative process3. And if you think about it, it seems quite logical that argumentative modes come and go in a discussion. This being a forum for ordinary citizens it is also quite natural that they from time to time withdraw from a more argumentative mode to a more explorative one where facts are shared and common knowledge is pooled. Most proponents of deliberative democracy readily admit that there will always be some constraints limiting the ideal of a communicative rationality. James Fishkin (1995: 41) thinks that in order to embrace a more realistic understanding of deliberation, “a great deal of incompleteness must be tolerated”, and Habermas (1996: 323) himself concedes that “rational discourses have an improbable character and are like islands in an ocean of everyday praxis”. What was the discourse quality like in the cases where there was a demand? In most of the cases there was at least some attempt at justifying ones view to the other participants. Only in 12 percent of the cases was there no form of justification to a demand. 46.4 percent of the demands were supported by inferior justifications and in 41.6 percent of the cases there was at least one qualified justification.

3

This can be examined with the help of other measures like IR, a more general content analytical measure designed to capture all kinds of interaction, deliberative as well as dysfunctional.

13

Table 2. Summary of speech activity and discourse quality at the speech level, with some basic socio-demographic components used to display the overall level of deliberation

46.4

35.2

6.4

2.5

89.9

3.9

3.7

93.0

7.0

32.5

57.6

9.9

(143)

(552)

(418)

(76)

(30)

(1069)

(46)

(44)

(1102)

(87)

(386)

(685)

(118)

Age 18-29

14.3

41.8

12.4

37.3

43.8

6.5

1.5

89.1

4.5

5.0

5.0

95.0

17.4

66.7

15.9

(15)

(480)

(201)

(25)

(75)

(88)

(13)

(3)

(179)

(9)

(10)

(10)

(191)

(35)

(134

(32)

30-44

28.1

33.4

9.6

46.2

38.2

6.1

2.6

91.4

2.9

3.2

5.7

94.3

35.4

54.1

10.5

No reference/ Self-linked

Difference principle

12.0

(N=1189)

Implicit respect

Utilitarian

35.5

(N=3351)

Disrespect

Neutral

100.0

(N=90)

All participants

No Justification

Group/Selfinterest

Reference to counterarg./weighing

Reciprocity

Sophisticat.

Respect

Qualified

Content of justification

Inferior

Relevant speeches

Speeches

Demographics

Level of justification

Reference to others dem.

Indicator

(23)

(941)

(314)

(30)

(145)

(120)

(19)

(8)

(287)

(99)

(10)

(18)

(296)

(111)

(170)

(33)

45-64

31.0

43.1

13.0

49.2

29.7

8.1

3.4

88.0

4.7

3.9

7.0

93.0

32.0

59.9

8.1

(34)

(1040)

(384)

(50)

(189)

(114)

(31)

(13)

(338)

(18)

(15)

(27)

(357)

(123)

(230)

(31)

65+

26.6

32.6

13.1

49.3

33.1

4.5

2.1

91.4

3.5

3.1

8.3

91.7

40.3

52.1

7.6

(18)

(890)

(290)

(38)

(143)

(96)

(13)

(6)

(265)

(10)

(9)

(24)

(266)

(117)

(151)

(22)

Gender Male

67.3

34.8

13.1

43.4

36.6

6.9

2.3

90.8

3.8

3.1

7.4

92.6

33.0

56.7

10.3

(48)

(2255)

(785)

(103)

(341)

(287)

(54)

(18)

(713)

(30)

(24)

(58)

(727)

(259)

(445)

(81)

Female

32.7

36.8

9.9

52.2

32.4

5.5

3.0

88.1

4.0

5.0

5.2

94.8

31.4

59.4

9.2

(42)

(1096)

(404)

(40)

(211)

(131)

(22)

(12)

(356)

(16)

(20)

(21)

(383)

(127)

(240)

(37)

Education -9yrs

9.4

37.7

10.9

55.5

31.1

2.5

1.7

90.8

1.7

5.9

10.1

90.0

35.3

61.3

3.4

(14)

(316)

(119)

(13)

(66)

(37)

(3)

(2)

(108)

(2)

(7)

(12)

(107)

(42)

(73)

(4)

12yrs

43.4

34.3

12.2

44.3

36.9

6.6

3.0

90.0

3.8

3.2

7.6

92.4

32.1

56.5

11.4

(37)

(1456)

(499)

(61)

(221)

(184)

(33)

(15)

(449)

(19)

(16)

(10)

(215)

(160)

(282)

(57)

15yrs

18.0

37.3

10.7

48.0

34.2

7.1

3.6

88.4

3.6

4.4

4.4

95.6

27.6

61.3

11.1

(19)

(604)

(225)

(24)

(108)

(77)

(16)

(8)

(199)

(8)

(10)

(10)

(215)

(62)

(138)

(25)

17yrs-

21.8

37.8

10.5

45.1

36.1

8.3

1.4

91.0

5.1

2.5

2.9

97.1

34.3

56.3

9.4

(15)

(732)

(277)

(29)

(125)

(100)

(23)

(4)

(252)

(14)

(7)

(8)

(269)

(95)

(156)

(26)

Note: Years of education is an estimation based on their educational degree.

14

As mentioned before, according to deliberative theory arguments should not only be logical and coherent, but also expressed in terms of the common good. When it comes to citizen deliberation it seems like pleads to the common good are quite infrequent, but on the other hand neither were their many demands expressed in terms of the participants own selfinterest. Actually close to 90 percent of the demands were not expressed with anyone’s interest in mind. But, what does these numbers actually tell us? If we compare them to what Steiner et al. (2004) found in their study using DQI in a parliamentary setting, there are certainly some differences, but also a few similarities. The most distinct differences appear when you compare citizen deliberation to plenary debates. In the plenary debates 88 percent of the demands were supported by at least a qualitative demand, twice as much as for the smallgroups. What perhaps should be pointed out is that on an average there were more than 60 demands with qualified or sophisticated justifications in each small-group. In other words, even if the share of qualified justifications might not seem particularly large, quite a few reasoned arguments were put forth. The citizens in the small groups only made an appeal to the common good in 7.6 percent of the cases, while the number for plenary debates according to Steiner et als. (2004) study is over 30 percent. If we, however, choose to compare citizen deliberation to the nonpublic debate in committees, a perhaps better comparison to small group discussions, the differences are noticeable smaller. The members of parliament still make more qualified justifications in support of their arguments than ordinary citizens do, but the difference is less striking (committees 60% vs. 42% for citizens). And if we look at the number of demands that include neither an appeal to common good nor self-interest, it is actually very similar, with the small groups at 89.9 percent and the committees at 86 percent. There are a number of occasions in each group where participants are clearly disrespectful, but in general the respect seems to be quite high. Signs of reciprocal deliberation among the participants appear, on the other hand, to be somewhat lacking. Even though the participants in general seem attentive to each other’s demands with two thirds of the demands being linked to a previous demand one way or the other. But, only 10 percent of the demands include a response to a counter-argument or a comparison of different demands, which is significantly less than the amount of demands that include no connection at all (32.5%). Since the indicators for respect and reciprocity are different from the ones used in the original DQI no direct comparisons can be made here, but the evidence of the participants actually weighing arguments and sorting them out like it is portrayed in the theory, seems to 15

be quite thin. According to Shawn Rosenberg (2004) the systematic thinking that such an act requires is quite uncommon and this outcome would seem to support that notion. These numbers will, however, only give us an overall understanding of how the deliberative process went. To get a better understanding of the workings of citizen deliberation we need to know is how these numbers are distributed among different characteristics. We begin by looking at some basic socio-demographic characteristics at the speech level. Basic as they may be, gender and education have generated a vibrant debate at the theoretical level (Sanders 1997, Young 2000 etc.). Looking at the total number of speeches made, it is quite evident that the men were more active than women during the process, even when we take into account that they were in a small majority. They made almost two thirds of all speeches and they were almost twice as likely to speak up. Similar differences are not detectable from the other characteristics even though the oldest age group and the most educated are more active than others, while the youngest and the least educated are the least active. The relationships for these variables appears however not to be linear. A better for sign for deliberation is the fact that there appears to be no clear patterns when it comes to how prone the participants are to argumentation (producing relevant speeches). According to assumptions in deliberative theory men would have a greater aptitude for the rational ideals of deliberation, but at first look there seems to be little evidence supporting that notion. Yes, men seem to have produced slightly better justifications for their arguments, but this is the only discernable difference, and it is not even significant one (independent samples t-test of average level of justification). Moving over to the other characteristics we see that especially in the youngest group there seems to be a real aptitude for reasoned argumentation. More than half of the demands are supported by at least a qualified justification, compared to only 37.6 percent in the eldest. No clear patterns can be discerned for education other than that the least educated are less likely to make qualified or sophisticated justifications to support their claims than others. However, once the level of education rises above the lowest level the differences seem to vanish. The indicator for content of justification is hard to examine because of the heavy inclination towards neutral justification preventing any clear differences from emerging. The respect indicator suffers from a similar fate, but a few minor trends can be discerned non-theless. More education and younger age appears to amount in more respectful interaction. Among the least educated every tenth claim includes some kind of disrespectful remark, while the equivalent number for the ones with the highest education is below three percent. 16

When looking at the overall level of reciprocity some reservations were made concerning the quality. Is it a universal problem or can it be explained by certain characteristics? A few diverging patterns emerge quite readily. Youth seems again to be a predictor for deliberative behavior. In the youngest age group more than 80 percent of the claims have a connection to a previous one and the share of claims include considerations of counterarguments or comparisons of different viewpoints is clearly above average. Compared to oldest age group where more than 40 percent of the demands have no connection to a previous argument and only 7.6 percent of the claims are of the more sophisticated kind, the difference is if not staggering, at least notable. Before we move over to examine participants’ individual capacity to make their voice heard in the deliberative process, and their individual capacity for rational discourse, we need to give some thought to the fact every deliberative process takes place within a certain context. As I said before I will focus on two contextual variables, the two treatments of the experiment and each participants stand on the topic of the debate (sixth nuclear power plant) compared to the small group as a whole. Table 3. Difference between treatments and effects of position relative to group norm tested by means of t-test All participants Treatment Vote treatment Common statement Position (group norm) In minority In majority

N 45 45 31 55

Speech activity (N=90) Mean stdev Mean dif. 32,8 26,7 -8,87 41,7 32,8 (n.s.) 37,8 34,8

27,9 28,6

3,0 (n.s.)

N 44 41 29 52

Discourse quality(N=85) Mean stdev Mean dif. 2,52 0,25 0,02 2,50 0,28 (n.s.) 2,55 2,50

0,31 0,24

0,05 (n.s.)

The difference between the two treatments did not seem to have any significant effects for this part of the deliberative process, neither on the activity nor on the discourse quality. This finding corresponds with Setälä et als. (2010) study based on survey data from the same citizen deliberation experiment. They found that the two treatments seemed to have little effect on how opinions developed. Neither is there any evidence of the majority strongly dominating the minority view. The ones representing the minority view in each small group does not seem to have been sidelined, or they have at least been able voice their opinions, just as well as the majority. While this confirms that minority views have been articulated, it does, however, not tell us whether they have enjoyed any real influence in the debate.

17

Deliberation at the participant level Much of the data presented above will simply not tell us that much until we look at how it is distributed among the participants. Table 4 below includes a number of measures of centrality and variability displaying the deliberative activity of the individual participants. Although these measures will not tell us what makes people act in a certain way, it can at least give us a better understanding of what the level of activity and deliberative quality can look like. Table 4. Overview of participation at the level of the individual, using measures of centrality and deviation. Indicator ‾ x

All participants s ‫ݔ‬෤

Max.

Min.

All Speeches

37.2

31.5

30.1

142

0

Relevant speeches

13.2

11.0

10.2

51

0

Discourse quality (0-4)*

2.51

2.52

0.27

3.11

1.70

*The discourse quality is combination of the four indicators. All variables coded to vary between 0-1.Qualified and sophisticated arguments, as well as references to common good (utilitarian + difference principle) have been collapsed to a single value (1).

When we look at the how the participants in the citizen deliberation experiment on nuclear power took part in the debate, we can see that the activity measured in number of speeches varies quite a bit. The most active member in any of the small groups has made 142 speeches within the time frame for this analysis, while two participants did not produce a single speech. Apart from the participant with 142 speeches we have two other extreme outliers (see box plot in appendix). Both with a fair amount over 100 speeches. In all groups the difference in activity between the most and the least active was at least 10 times. It is quite easy to see that some participants can be quite dominating, but how active should you have been to have had any influence at all on the debate? One, five, twenty speeches? This is where things become difficult to assess, but to give you some idea of activity is distributed at lower end of the scale, it can be revealed that 14 participants produced seven or less speeches each during the discussion. Belonging to this group basically means that you participated when you were encouraged to do so by the facilitator or a fellow participant. A look at the more argumentative side of the debate shows that five participants produced no relevant speeches at all and a total of eleven participants made less than 3 speeches including a demand during the debate. The considerable variation between participants must be considered a cause for concern, since studies of juries and other small groups indicate that individuals who speak the most are likely to be viewed as the most persuasive by other group members and that is the quantity of remarks, not their quality that seems to drive these perceptions (Sanders 1997: 11).

18

If the variation in activity is this considerable it might inevitably affect the prospect for fair and equal deliberation. Turning our attention to discourse quality at the individual level we can see from table 4 that the level of discourse quality ranges from 1.70 to 3.11 (box plot in appendix). With a variable that goes from zero to four that may not seem as such a great variance at first glance. This has partly to do with the fact that some of the discourse quality indicators tended to lean heavily towards one value. What does it mean to have scored 1.70, 3.11 or the average of 2.51? To obtain the average score of 2.51 what you need to do is basically not be disrespectful (1.00), connect to prior arguments (.50), have a neutral reason for your justification (.50) and give mostly inferior justifications (0.50) in support of your arguments. The average participant is as we can see, although perfectly capable of civilized debate, not particularly deliberative, at least from the perspective of rational discourse. The one that scored 3.11, on the other hand, should be considered quite, if not very deliberative. Arguments would be mostly qualified or sophisticated, quite a few of them where different perspectives are weighed or counter-arguments taken into account, as well as, some where there are references to the common good. On the other hand if you score as low as 1.70 your justifications are likely to have been inferior or non-existent, there are likely to have been a few cases of disrespect or an unwillingness to take others arguments into account.

A systematic analysis deliberative capacity at the participant level The measures above are, however, insufficient if one aims to more systematically examine the characteristics that influence discursive activity and quality, i.e. if some people based on certain set of characteristics are more or less likely to fulfill the requirements of democratic deliberation. The more systematic analysis is based on four different models. The first includes only the socio-demographic variables, the second includes besides the sociodemographic variables, two additional variables that can tell us something of the individuals’ capacity to take part in discussion. The third model includes besides the basic sociodemographic variables a set of variables that are expected to explain the motives behind participation, while the fourth and final model includes all variables from the prior models.

19

Table 5. Activity in small-group discussion (linear regression) Model 2. Capacity 1. Sociodemo. B B (standard error) (standard error) 45.46** 38.51** Age/100 (18.66) (19.31) 18.87*** 14.31** Gender (male) (5.73) (6.41) Educationa 35.19*** 26.58** (12.03) (12.80) Knowledgeb 30.44 (20.12) Active memberc 7.60 (6.59) Internal efficacyd External efficacyd Opinion Intensitye R2 (adj.) N

0.22 (0.19) 84

0.26 (0.21) 84

3. Motive B (standard error) 38.19** (19.14) 15.47** (6.01) 21.25* (12.74)

33.76*** (12.60) 2.54 (11.51) 13.94 (13.88) 0.29 (0.24) 82

4. Final B (standard error) 31.27 (20.24) 12.61* (6.67) 17.32 (13.25) 19.34 (20.62) 6.89 (6.88) 29.73** (13.00) -1.83 (11.99) 10.24 (14.18) 0.31 (0.24) 82

Note: All independent variables coded to vary between 0-1 a Education: -9yrs (0), 12yrs (0.33), 15yrs (0.67), 17yrs- (1) b Knowledge (15 items in additive index, see www.dce.abo.fi for knowledge items) c Active member of interest association (dummy variable) d Internal & externa efficacy (additive indexes based on two questions from the survey, see appendix) e Opinion intensity (additive index based on survey items, see appendix) Levels of significance: *

Suggest Documents