Taking the Goals of Deliberation Seriously : A Differentiated View on Equality and Equity in Deliberative Designs and Processes

Deliberation must be immunized against coercive power by a baseline of equality. But what does the requirement of equality mean, in practice, for organizers designing deliberative events and forums? This question is complicated by the fact that equality is fundamentally about two—at times contradictory—values. On the one hand, the value of universal moral equality, which requires abstracting from social circumstances. On the other hand, the value of equity, which requires attending to social circumstances. Deliberative institutions vary in their capacity to promote one value over the other, or in their capacity to compromise between the two. We argue that negotiating between these twin values should be done with reference to the different goals of the deliberative process (generating legitimate decisions, producing more informed opinions, promoting mutual respect, enabling accommodation, and so on), and with an eye to the trade-offs that achieving particular goals might require. Focusing on civic forums, we review existing research related to three important aspects of design—participant recruitment, the nature of the interaction, and decision-making—and discuss how different designs impact deliberation’s different normative goals. We argue against a totalizing view of deliberation, where unitary institutions try to achieve all of deliberation’s goals at once, and instead discuss how the trade-offs between deliberation’s different functions can be resolved at the system level. We conclude by arguing that practitioners should not try to realize all deliberative goals—including equality and equity—at once, but rather should prioritize the goals they want to achieve, and select institutional rules and practices that optimally achieve these goals. Author Biography Edana Beauvais is a Ph.D. candidate in political science at the University of British Columbia. Her research interests include deliberative democratic theory, political communication, and political psychology. Her dissertation, “Discursive Equality,” considers how cognitive biases (such as stereotypes) can contribute to the internal exclusion of disempowered speakers, and theorizes potential solutions to this problem. Edana is involved in the Participedia project, trying to leverage the potentials of information technology to gain insight into citizen participation in democratic innovations.


Introduction
Deliberative theorists have long stressed that deliberation must be immunized against coercive power by a baseline of equality (Habermas, 1990).But what does the democratic pre-condition of equality mean, in practice, for organizers designing deliberative events and forums?After all, as Bernard Williams (1972) notes, equality is fundamentally about twoat times contradictoryvalues.On the one hand, the value of universal moral equality, which refers to the fundamental sameness of common humanity, requires abstracting from social circumstances.On the other hand, the value of equity, which refers to just distributions of power and resources, requires attending to social circumstances.Deliberative institutions vary in their capacity to promote one value over the other, or in their capacity to compromise between the two.We argue that negotiating between these twin values should be done with reference to the different goals of the deliberative process, with an eye to the trade-offs that achieving particular goals might require, and to the context within which the deliberation takes place.
In the first section of this paper, we discuss some of the central normative goals that discourse achieves in a healthy deliberative system.In the second section, we review existing empirical research on how institutional designs impact deliberation's different goals, including the trade-offs different institutional design choices might require.While there are many examples of deliberative sites in political systems, we restrict our discussion to instances of organized, structured deliberation, or "civic forums" (Karpowitz & Raphael, 2014), because these are instances where practitioners can more easily exert direct influence over design. 1ivic forums include a wide variety of deliberating bodies, such as community policing initiatives (Fung, 2009;Karpowitz & Raphael, 2014), participatory budgeting (Avritzer, 2009), civic intergroup dialogue meetings (Walsh, 2004), and deliberative "mini-publics," such as Deliberative Polls and citizens' assemblies (Fung, 2003;Goodin & Dryzek, 2006;Grönlund, Bächtiger, & Setälä, 2014;Smith, 2009).We consider three important aspects of designparticipant recruitment, the nature of the interaction, and decision-makingand review existing research regarding how different designs impact deliberation's different normative goals.
We conclude by drawing out the implications of our discussion for practitioners and theorists, arguing against a totalizing view of deliberation where unitary deliberative institutions and processes achieve all of deliberation's desired outcomes at once (see Fishkin, 2009).Instead, we show that deliberative theorists and practitioners should ultimately accept that various ideals may sometimes form trade-offs that require thinking about which designs and processes are most appropriate for realizing particular normative outcomes.

Achieving Different Normative Outcomes: Understanding the Trade-Offs
Deliberation can achieve a number of distinct goals or functions.One of deliberation's central functions is to produce decisions that are perceived as legitimate by those who are bound by them.Since democratic legitimacy is predicated upon the inclusion of those affected by decisions in processes of decision-making, those affected by decisions must have equal opportunities to participate, and equal (or fair) influence over the outcomes of discourse.Clearly, these values can conflict.In practice, equal opportunity to participate is often interpreted to mean making civic forums open to anyone who wants to join.But inclusion that abstracts from social differences in this way may not produce diverse or even representative deliberating groups.By contrast, inclusion that is attentive to social differencessuch as reserving seats for, or affirmatively recruiting disempowered social group memberscan achieve more diverse or representative deliberating bodies, but at the cost of limiting the openness of recruitment.
In addition to legitimacy, deliberation achieves epistemic and ethical goals.Deliberation's epistemic function refers to discourse's capacity to encourage learning and produce opinions, preferences, and attitudes that are informed by facts, information, and the full range of relevant arguments and concerns.Conventionally, the ethical function refers to whether deliberation generates mutual respect (Mansbridge et al., 2012).We suggest that other ethical functions include promoting mutual recognition, accommodating ethno-cultural or linguistic diversity, and community-building through developing social bonds, feelings of mutual interdependence, and trust both within and across groups.As we have suggested, achieving one function may conflict with other functions in ways that are relevant for deciding how to organize a deliberative event.
Deliberative venues and forums often cannot achieve every deliberative goal simultaneously, since different functions (or different aspects of the same function) can come into conflict.This is not a problem for the overall health of the deliberative system, since different deliberative tasks are "distributed" sequentially across various component parts, which can refer to different moments in a deliberative event, or to different deliberative forums and actors across a deliberative system (Goodin, 2005).The health of the deliberative system is judged according to how well the variegated, interlocking deliberative forums and actorsinformal communication networks, associations, the media, legislatures, and the courts, to name a few -achieve discourse's various normative goals, in the aggregate (Mansbridge, 2015;Mansbridge et al., 2012).Now we will discuss how, in practice, deliberative design can impact which normative ideals are achieved and which are sacrificed in civic forums.

How Institutional Design Impacts Deliberation's Different Aims
In this section, we consider which institutional designs are best suited for achieving the twin values of equality and equity, and discuss the consequences of these designs for other deliberative goals.We consider three important aspects of design: participant recruitment, the nature of the interaction (whether it was facilitated, the communication format, and whether the deliberation is face-to-face or online), and decision-making.For each aspect, we review existing empirical research regarding how different designs impact deliberation's different normative goals.

Participant recruitment: Who gets to deliberate?
With respect to participant recruitment, practitioners should be cognizant of how recruitment impacts opportunities for participation, as well as how recruitment impacts the diversity of the deliberating body.And, of course, organizers should be aware of how opportunities for participation and diversity impact the goals of deliberation, such as the perceived legitimacy of deliberative outcomes, as well as epistemic and ethical outcomes.
Of the different participant-selection methods, self-selection and random selection are the most effective means for achieving equality that abstracts, in different ways, from social circumstance to provide equal opportunities for participation.Selfselected processesprocesses that are radically open to anyonemay be linked to perceptions of legitimacy.This is particularly true in democracies where public conceptions of political legitimacy are heavily influenced by the idea of equal influence defined as "one person, one vote."A study of a Canadian citizens' assembly on neighborhood planning found that local activists perceived selfselection as the most legitimate way to populate a deliberating body, precisely because such processes are open to all interested parties (Beauvais & Warren, 2016).
However, because males, white people, and people of higher socio-economic status tend to participate in talk-centric politics at higher rates (Jacobs, Cook, & Delli Carpini, 2009), self-selection often produces more homogeneous groups that reflect these social inequalities (Urbinati & Warren, 2008;Warren, 2001).The perception that a civic forum is not descriptively representative of the underlying public can also undermine public perceptions of the forum's legitimacy (Beauvais & Warren, 2016).Furthermore, lack of diversity can negatively impact deliberation's epistemic goals, since homogeneity can reduce learning and inhibit opinion change (Barabas, 2004;Bohman, 2006;J. S. Dryzek & Niemeyer, 2008).
A lack of diversity can also undermine deliberation's ethical functions, such as the accommodative function.Because cross-cutting associations promote intergroup interaction and trust (Warren, 2001), homogeneous forums can potentially inhibit deliberation's ability to promote intergroup trust and accommodation.With respect to deliberation's accommodative function, Lindell et al. (2015) find that the total absence of any participants with an immigrant background in groups deliberating about immigration policy was associated with polarization in an anti-immigrant direction.Mendelberg and Oleske's (2000) research suggests that the potential for group polarization is exacerbated when opposite sides of a debate are represented by different groups.However, Grönlund, Herne, and Setälä (2015) find that the threat of polarization can be mitigated by the presence of facilitators who stress deliberative norms.
Like self-selection, drawing a random sample of participants abstracts from social circumstances, and also preserves the value of equal opportunity to participate, in the sense that all members of the underlying population have an equal probability of being selected.Furthermore, allotment should produce a cross-section of the public not otherwise seen in small-group deliberations (Fishkin, 2003;Smith, 2009), and so potentially addresses some of the epistemic and ethical concerns surrounding the lack of diversity that can accompany self-selection.Caluwaerts and Kavadias's (2014) experimental research suggests that diversity correlates with higher-quality deliberation, as measured by the Deliberative Quality Index (see Steiner, Bächtiger, Spörndli, & Steenbergen, 2004).Hélène Landemore (2013Landemore ( , p. 1211) ) suggests that random selection is the best way to promote the "cognitive diversity" of deliberating groups.Cognitive diversity refers to a diversity of ways of seeing the world, and a diversity of ways of interpreting and solving problems.Cognitive diversity helps ensure participants hear diverse opinions, learn about opposing views, and hear counterarguments to their own opinions.
But random selection may still exclude members of minority groups.This is particularly true when the deliberating body is small, since small random draws are unlikely to be representative (Karpowitz & Raphael, 2014).Random-stratified sampling (to over-represent particular group members) or purposive sampling and recruitment (to include only specific group members, such as for enclave deliberation) are better tools for achieving equity-based inclusion, if one's goal is to achieve deliberation that is attentive to social differences.Random stratified sampling is often used to ensure the deliberating body proportionally represents an underlying public on certain socio-demographic criteria.Practitioners should think in advance about who is affected by the outcomes of deliberation, and consider the kinds of interests that need to be represented in the deliberating body.And finally, since proportional representation does not guarantee that members of disempowered groups will have equal (or even proportionate) influence over the outcomes of deliberation (Gerber, 2013;Karpowitz & Mendelberg, 2014), purposive sampling and recruitment may be used to ensure certain group members (such as those particularly affected, or the disempowered) are over-represented, or even singularly represented in enclave deliberation.
And although homogeneous groups may not be ideal for promoting the ethical function of inter-group trust and accommodation, they are excellent venues for promoting ethical aims of intra-group trust and interdependence among the disempowered.For instance, consider the sum of discursive forums and talk-based interactions in disproportionately African-American neighborhoods that constitute what has been described as the "black public sphere" (Collective, 1995;Squires, 2002).Discourse in the black public sphere encourage feelings of mutual respect, interdependence, and trust among African-Americans, which is essential for increasing efficacy and enabling collective action.And other research shows that enclave deliberation among the disempowered can produce the same epistemic benefits as more heterogeneous groups (Karpowitz & Raphael, 2014;Karpowitz, Raphael, & Hammond, 2009).

The Nature of the Deliberative Interaction
The nature of deliberative interactionsthe nature of facilitation, communication format, and whether deliberation takes place face-to-face or onlinecan directly impact deliberation's democratic, epistemic, and ethical aims.With respect to facilitation, unfacilitated deliberation can achieve a more abstract, laissez-faire ideal of equality of opportunity for discursive participation that ignores social circumstances.However, facilitation not only ensures that all participants have an equal chance to speak (achieving a more concrete equality of opportunity for participation), it also ensures that different sides of the debate are heard (achieving equitable participation) (Landwehr, 2014).
Even within facilitated civic forums, facilitator style is important for achieving epistemic or ethical goals.Kara Dillard (2013, p. 220) uses the three-category distinction of passive facilitators, who play a "turn-taking enforcer" role, moderate facilitators, who play a "designated driver" role (moving the conversation along, without adding new interpretations), and involved facilitators, who play a "quasiparticipant" role (editorializing or interpreting the conversation).Dillard (2013, p. 232) suggests that passive facilitators may more effectively achieve ethical aims such as reinforcing "democratic values," whereas moderate and involved facilitators may more effectively achieve epistemic aims of learning about opposing views.Moderate or active facilitators may be especially useful for achieving epistemic and ethical aims in homogeneous groups.By offering hypotheticals ("What would you reply if someone argued that because of p, q?") facilitators can create awareness of opposing views or empathetic responses even in homogeneous groups (Landwehr, 2014), through a kind of "discursive representation" (Dryzek & Niemeyer 2008).
Another related aspect of the deliberative interaction is the communication format.Bernard Manin (2005) forcefully argues that proper deliberative interaction should be about debate rather than discussion (the standard format of organizing citizen deliberation).Debate, says Manin, advances the quality of argumentation, and thus deliberation's epistemic aims.By presenting and defending one's case maximally, a debate format discloses the merits of specific positions.And by challenging other arguments and positions maximally, it unravels inconsistencies, unearths new and unconsidered facts, and forces participants to provide more, and more robust, reasons for their positions.In a laboratory experiment, David M. Schweiger et al. (1986) compared the effectiveness of contestatory discussion formatsdialectical inquiry and devil's advocacyversus consensus approaches in group decisionmaking.Results showed that both dialectical inquiry and devil's advocacy led to a higher level of critical evaluation of assumptions and better-quality recommendations than the consensus treatment.
On the other hand, feminists and difference democrats have claimed that contestatory forms of deliberation may suppress marginalized voices.Rational debate is a highly demanding communication format requiring sophisticated cognitive as well as rhetorical abilities on the part of the contestants, thus being potentially exclusionary (Young, 2000).Psychologists have shown, for instance, that there is a gender bias when roles such as the devil's advocate are assigned.Women's reputations can be harmed when they verbally challenge men (Sinclair & Kunda, 2000).In other words, while contestatory deliberation may help to reduce epistemic bias, it may do so at the expense of equity of specific social and cultural groups.Summarizing their findings on gender inequality in deliberation, Mendelberg et al. (2014, p. 18) compellingly argue "that egalitarian discussion rests not on adversarial but on supportive communication, which lifts women's authority." Again, it may be difficult to maximize all the aims of deliberation simultaneously in a single forum.Focusing on the Australian Citizens' Parliament, Curato et al. (2013) show that inclusiveness was "produced" by a specific mediation technique deployed by facilitators, namely "appreciative inquiry," which aims to emphasize strengths that people can build on.However, the authors find that "the prioritization of the appreciative and constructive aspects of deliberation restrained the CP's [Citizens' Parliament's] potential for generating epistemically superior proposals" (Curato et al., 2013, p. 11).Of course, as we will discuss in greater detail in our conclusion, not every civic forum needs to achieve every deliberative goal simultaneously.A healthy "division of labor among different elements" (Karpowitz & Raphael, 2014, p. 21) within a deliberative system should ensure that while some civic forumssuch as those that promote supportive communicationencourage internal inclusion and produce stronger ethical outcomes, other civic forumssuch as those that promote contestatory communicationproduce better epistemic outcomes.
A final determinant of the nature of interaction is whether it is face-to-face or online.Grönlund et al. (2009) found that participants who were assigned to either face-to-face deliberations or online deliberations both learned.Furthermore, with respect to opinion change, both groups' opinions moved on the same issues and in the same direction when change occurred.Other studies suggest that while online deliberation is useful, face-to-face deliberation may achieve more deliberative aims.Knobloch & Gastil (2015) find that after deliberation, face-to-face deliberators reported higher levels of political engagement along a number of measures, including paying more attention to public affairs, and talking to others about politics and who to vote for; they also reported higher levels of volunteerism and community engagement.By contrast, online deliberators only reported higher levels of attention to public affairs and talking about issues (but not voting decisions), with no increase in volunteerism or community engagement.Online deliberations have other potential pitfalls, particularly with respect to the democratic aims of equality and equity, since Internet access and technology skills are unequally distributed throughout the population (Grönlund et al., 2009).Existing research suggests that online political participation may replicate many longstanding participatory inequalities (Schlozman, Verba, & Brady, 2010).

Decision-making
It should be noted that many civic forums do not ask participants to come to a decision at all.Whether or not participants need to reach a decision depends, of course, on the purpose and goals of deliberation.If the purpose is to help overcome conflicts related to identity and values through dialogue, and to achieve ethical aims of increased intergroup trust and mutual respect, then it may be best if participants do not take a position or reach a decision.Many civic dialogue groups, such as the Public Conversations Project, do not ask participants to come to a decision.
But if the goal of deliberation is to come up with policy proposals, or to achieve epistemic aims, then it may be best if participants take a stand and reach a decision.Simply asking participants to take a position at the start of deliberation produces higher deliberative quality and more learning, but there is less opinion change as compared to when participants do not take a position in advance (Baccaro, Bächtiger, & Deville, 2014).In terms of decision-making, participants may reach decisions individually, as is the case with Deliberative Polling (Fishkin, 2003).Alternatively, participants may reach collective decisions, as is the case with participatory budgeting in Porto Allegre (Avritzer, 2009) and the Canadian citizens' assemblies on electoral reform (Warren & Pearse, 2008).Whether or not decisions are made individually or collectively, it seems that deliberative decisionmaking works.As long as subjects have enough information, deliberative decisionmaking is better than what happens when citizens merely vote without first talking (Neblo, 2007).For instance, deliberation reduces the explanatory effect of demographic characteristics and negative affect on opinions.
With respect to reaching collective decisions, does the decision-making method matter?It seems likely.After all, Existing empirical evidence suggests the decisionmaking method (for instance, consensus or majority vote) plays an important role in influencing which of deliberation's normative aims are achieved, both directly and indirectly by interacting with the composition of the deliberating body.There is a longstanding and popular argument in analytical politics that a consensus constraint may have perverse effects on the deliberative process and its outcomes.Douglas Rae (1975Rae ( , p. 1273)), for instance, mentions that when unanimous agreement is to be achieved, there are "incentives for persons indifferent or even modestly favorable to change to falsify their preferences as a means of extortion."This argument has been expanded and formalized by Austen- Smith and Feddersen (2006), who claim that veto power and unanimous voting rules create incentives for some actors to conceal information, making information from all discourse participants suspect.Consequently, the deliberative process can break down under unanimity rule (see also McGann, 2006).
But this account has been challenged in recent years, both theoretically and empirically.Austen- Smith and Feddersen (2006) and Mathis (2011) formally demonstrate that the unanimity rule is superior in achieving full opinion-sharing and information revelation, under the (plausible) condition that individuals may provide verifiable evidence for their private information.There is also initial evidence that a consensus constraint produces both higher-quality argumentation (Caluwaerts, 2012) and induces more learning (Grönlund, Setälä, & Herne, 2010).By the same token, Grönlund et al. (2010) also found evidence for an ethical role of the consensus rule, as they found political trust and readiness for collective action increased more in a consensus group.Morrell's (2005) experiment compares groups that did not deliberate, reached a decision via voting, or reached a decision via agreement.He found that regardless of the decision-making method, deliberation fails to increase internal efficacy, but he also found higher levels of "situationally specific" internal efficacy (efficacy specific to the deliberative process) among deliberators in the consensus condition.
Decision rules are also relevant in the context of the equity-equality debate, albeit in more complicated ways.This is because the decision-making method seems to interact with the composition of the deliberating group.For instance, when disempowered group members' descriptive representation is low, consensus decision-making can increase disempowered group members' deliberative authority (Karpowitz & Mendelberg, 2014;Karpowitz, Mendelberg, & Shaker, 2012;Mendelberg et al., 2014).Caluwearts and Kavadias (2014) also found that consensus decision-making increases deliberative quality among homogenous groups.2

Conclusion
As our discussion makes clear, no single deliberative institution or process can achieve all of deliberation's desired outcomes at once.Different institutional designs achieve some outcomes better than others, and different normative outcomes may require trade-offsincluding the trade-off between the democratic prerequisites of universal equality, which requires abstracting from social circumstances, and equity, which requires attending to social differences.But as Karpowitz and Raphael (2014, p. 338) note, "in a democracy, no single institution should be tasked with representing public opinion or the polity as a whole, including civic forums."Different deliberative tasks are distributed across various agents and forums within a deliberative system.And different tasks can even be distributed across different moments within a single deliberative process.
When deciding how to structure deliberation, practitioners must clearly think through the different goals they hope the deliberations will achieve, and be cognizant of the trade-offs that achieving certain goals might require.Karpowitz and Raphael's contribution in this special edition offers suggestions for how to achieve different ideals of inclusion.The authors offer practical advice on how to achieve compromises between equal opportunities for participation and equitable inclusion, depending on whether a forum is oriented toward holistic goals (to the public as a whole) or relational goals (such as relations between groups or communities), and depending on the openness of the forum.
In regard to the nature of deliberation, there is no contradiction when it comes to facilitation; facilitated deliberation is better than unfacilitated deliberation for ensuring both that everyone can speak, and that all sides of the debate are heard.However, the type of facilitationpassive, moderate, or activemight vary depending on the phase of deliberative process.At the start of deliberative processes, passive facilitation may be preferable, because this can free participants to use narrative and storytelling to build rapport through self-disclosure (Ryfe, 2006).In latter stages of deliberation, moderate and active facilitation may be preferable, since this can "push participants to justify ideas and critically evaluate the ideas of others" (Dillard, 2013, p. 231).
With respect to the communication format, supportive, rather than contestatory modes of speech may be best for achieving equality's twin values, but evidence groups).The authors only find a positive relationship between consensus decision-making and deliberative quality in homogeneous groups.
suggests participants learn more through contestatory argumentation.Of course, the communication format can vary across civic forums within a deliberative system.Civic forums designed to promote ethical goals, such as fostering mutual respect, tolerance, and trust, should probably adopt more supportive modes of speech.By contrast, civic forums designed to promote epistemic outcomes, such as learning or collective decision-making, should probably adopt more contestatory modes of speech.Furthermore, it is possible to achieve both communication formats by sequencing deliberative moments within a single civic forum.For instance, within cross-cutting civic forums, small breakout groups can be used to promote enclave deliberation among the disempowered (Karpowitz & Raphael, 2014).Practitioners might vary the communication format depending on whether group composition is cross-cutting or homogeneous.
With respect to whether deliberation should be online of face-to-face, given existing inequalities in technological skills and access to the Internet, to date purely face-to-face deliberation is more effective than purely online deliberation for achieving the goals of equality and equity.Of course, civic forums do not have to be purely face-to-face or online.Increasingly, forumssuch as the Australian Citizens' Parliamentare effectively integrating the two formats (Dryzek, 2009).
Whether or not practitioners require that participants reach a decision, and, if so, how they reach a decision, depends upon the purpose of the civic forum.For civic forums such as the Public Conversations Project, which are designed to achieve epistemic aims like reconciliation, or promoting mutual respect across intergroup divides, it may not be necessary or desirable to ask participants to take positions or reach decisions.By contrast, for civic forums such as the Canadian citizens' assemblies on electoral reform, or Porto Alegre's Participatory Budgeting, where participants are expected to learn about complex issues and come up with implementable policy proposals, decision-making is necessary.The nature of decision-making should be attentive to how decision-making rules interact with the composition of the deliberating body to influence participants' discursive authority.For instance, when disempowered group members' descriptive representation is low, consensus decision-making is likely the best tactic for increasing disempowered group members' deliberative authority (Karpowitz & Mendelberg, 2014;Karpowitz et al., 2012;Mendelberg et al., 2014).In sum, future deliberative designers should not try to realize all deliberative goalsincluding equality and equityat once, but rather should prioritize the goals they want to achieve, and select institutional rules and practices that optimally achieve these goals.