Deliberative critics contend that because societal inequalities cannot be bracketed in deliberative settings, the deliberative process inevitably perpetuates these inequalities. As a result, they argue, deliberation does not serve its theorized purposes, but rather produces distorted dialogue determined by inequalities, not merits. Advocates of deliberation must confront these criticisms: do less-privileged, less-educated, or perhaps illiterate participants stand a chance in discussions with the more privileged, better educated, and well spoken? Could their arguments ever be perceived or weighed equally? This essay presents empirical evidence to demonstrate that, in deliberations that are structured to provide a more level playing field, inequalities in skill and status do not translate into inequalities of influence.
When we think of the greatest orators, we often see men. In many developed democracies, those men are also likely to be white, educated, and privileged-men who had better opportunities from birth. It would be natural to expect these same privileged men to dominate in deliberation; indeed, we have all seen this kind of discursive domination in our own lives. Thus, many critics of deliberation have identified societal inequalities in deliberative settings, from town meetings to the jury room. Compared with the vote, which is explicitly structured to foster equality, deliberation seems destined to perpetuate existing societal inequalities, and perhaps foster greater inequality. These criticisms apply to many contexts; yet in deliberative settings structured to provide a more level playing field, we do not find empirical evidence to support these claims. Since the critics' claims are empirical, it is necessary to examine them empirically. This essay provides empirical evidence to demonstrate that inequality is not a necessary attribute of deliberation.
Deliberative theorists contend that forums for public deliberation provide opportunities for citizens to engage each other in thoughtful discussion; in such settings, they may share competing views and, over time, develop reasoned and considered opinions.1 John Stuart Mill argued that taking part in public functions, such as small town offices or jury duty, serves as a school of public spirit.2 In the case of juries, people, privileged or not, would engage in deliberations together to decide the fate of others.3
More recently, German sociologist and philosopher Jürgen Habermas has envisioned the larger public sphere as a space in which “private people com[e] together as a public.”4 That public sphere can serve as a place for “critical public debate,”5 where public opinion can form and to which all citizens could have access, with freedom to discuss and gather as desired. In his essay “Further Reflections on the Public Sphere,” Habermas discusses how the public debate could be carried through in a variety of today's civil society organizations, including volunteer organizations, churches, and academic institutions.6 Such organizations, he argues, are capable of helping the public engage in the kind of debate and discussion that could produce reflective opinion formation.
Claus Offe has also argued that deliberation could help rehabilitate liberal democracy. He has noted the abundance of literature on the crisis of democracy and even “the end” of democracy.7 Defining the basic elements of liberal democracy as “stateness” (the state's ability to maintain the allegiance of its population and to execute central functions), rule of law, political competition, and accountability, he has illuminated the inherent threats to these key elements. For example, although political competition allows for parties and contenders to have a clear and legitimate winner, the process of competition has created an electoral need to portray candidates through their “personalities.” It has also encouraged politicians to treat their competition in terms of them versus us, in which we are good and honest and they are untrustworthy and evil. The rampant use of such strategies in political campaigns and politics in general has strengthened populist movements and created more divisiveness in society. Offe's remedies include strengthening the people's voice through various forms of participation and improving the public's will formation through deliberation. He argues that deliberation, through reason-giving, listening, and respecting, could bring forth positive effects, such as more informed opinion, internal efficacy, and the ability to widen social inclusion. Offe also suggests using “randomness” in the composition of participants in deliberation to ensure diversity and inclusive representation. Deliberation, he argues, can offer a path to restoring liberal democracy, especially if that deliberation can be institutionalized.
Many deliberative critics find the aims and aspirations of deliberation too lofty. They have argued that, given societal inequalities, deliberation in practice does not come close to reaching those ideals. The critical theorist Nancy Fraser has argued, for instance, that Habermas's public sphere is incapable of “bracketing” societal inequalities: that is, to neutralize inequality in a deliberative setting.8 Even if people voluntarily agree to participate and deliberate “as if” they were equals, it is simply not possible to impose deliberative equality on a social base of inequality.9 The common societal behaviors of men and women play out in deliberative settings. Men tend to interrupt women, speak longer and more often than women, and ignore women in deliberations.10 Men are also more likely to be assertive, while women are more tentative and accommodating.11 Further, Fraser is concerned that group deliberation often transforms what were individual opinions into one single group opinion.12 Societal inequalities then not only unbalance the discussion; they may also, through a dynamic of interrupting and silencing, create inaccurate impressions of group opinion. Finally, certain speech styles, characteristic of the dominant, diminish the value of other's opinions.13 The inherent inequalities in our society cause deliberation to benefit more dominant individuals and groups, while disadvantaging minority individuals and groups.
Political theorist Iris Marion Young has similarly argued that “speech that is assertive and confrontational is here more valued than speech that is tentative, exploratory, or conciliatory. In most actual situations of discussion, this privileges male speaking styles over female.”14 Moreover, if Habermas were right and the essence of good deliberation were “the force of the better argument” restricted to reason alone, then deliberation would disadvantage people who typically use other forms of argumentation, such as telling stories and sharing experiences.15 Participants who are less privileged may be particularly likely to share their arguments in the form of stories. Thus, even if participants from varying socioeconomic classes did deliberate together, there would be no guarantee that they would actually listen to and understand one another.16
A further argument against deliberation stresses the undue influence of those who dominate deliberations verbally.17 Because there is limited time in any given deliberation, the participants who use the most time and are most articulate in sharing their opinions are likely the most successful in persuading their group members. The evidence from jury studies shows both how the more socially advantaged members speak more and how the sheer quantity of remarks from participants is correlated with influence on fellow jurors.18 Given these findings, the results of deliberations should favor the more privileged, contrary to the aims and aspirations of deliberative theorists.
The most common point of reference for deliberation is the jury. Accordingly, many scholars have used juries (usually mock juries) to examine group dynamics, behaviors, and decision-making mechanisms.19 Correlational evidence from jury studies has supported deliberative critics. Most of the jury research on gender has found that men participate significantly more than women.20 Men are also more likely to reference facts, dispute facts, and discuss organizational matters, whereas women are more likely to discuss the consequences of verdicts. Significant differences also appear between the more- and less-educated jurors. More-educated jurors not only participate more than less-educated jurors, but also discuss legal and factual issues more often. But discussing more factual issues does not necessarily mean that the more- educated jurors bring forth more facts. Researchers have found that the number of facts shared by the more- and less-educated jurors differed only marginally in favor of the more educated.21
Research on juries has also shown a relationship between juror occupation and income and the likelihood of being selected as foreperson.22 Across the jury research, the foreperson was almost always male and usually was more educated, had a higher income, and had a higher-status occupation.23 Typically, income and occupation are highly correlated with education, and the relationship is no different in juries. Jurors with higher-status occupations and higher income have been shown to participate in deliberations more than their counterparts.24 Further, men with higher-status occupations and higher income participated more than women who possessed the same criteria.25 The evidence from jury studies suggests why deliberative critics fear the negative consequences of introducing more citizen deliberation into democracy.
Until recently, there has been little empirical evidence on forms of citizen deliberation other than juries and mock juries. This essay presents one of the most comprehensive analyses of deliberation to date, using both quantitative and qualitative correlational evidence drawn from five nationally representative Deliberative Polling projects in the United States, consisting of four online Deliberative Polls and one face-to-face Deliberative Poll. Across these five projects, the study included 1,474 participants and ninety-nine small groups. Typically, empirical results, if available, derive from much smaller samples with fewer participants. This study made possible a more systematic and thorough analysis.
Each of these five projects covered a different discussion topic, including U.S. foreign policy (online in 2002 and face-to-face in 2003), primary elections (online in 2004), general elections (online in 2004), and health care and education (online in 2005).26 The four online projects were conducted through voice-only, like a typical phone conference but with the addition of web-based software that enabled participants to see their own and other group members' icons on screen and queue up to speak. This software also allowed moderators to mimic in-person moderation by interrupting speakers with pause/mute buttons and moving people around in the queue. Instead of a full-day or weekend deliberation, the discussions in the online Deliberative Polls were spread out over a few weeks, on a weekly or biweekly basis, with each session taking approximately 1.25 to 1.5 hours. Like the in-person deliberations, the online sessions included plenary sessions in which experts and/or policy-makers would answer questions developed by participants in the small group discussions. Instead of a live plenary session, the plenary sessions were prerecorded and then listened to together as a group. In total, the small group and plenary session discussions took between five to six hours, similar to a one-day face-to-face event. All small groups were recorded and transcribed for qualitative and quantitative analyses. All projects also had control groups that were surveyed at the same time as the deliberative groups but did not receive the deliberative “treatment.”27
To directly address concerns regarding participation inequality in deliberation, the analyses examined the number of words spoken, minutes used, and statements made by each person in each small group discussion. The expectation from jury studies was that economically and educationally privileged white males would dominate the time in the deliberations, leaving less time for others in the group to participate. In these Deliberative Polls, however, there were no statistically significant differences between the participation levels of men and women: in total minutes used, total statements made, or total words spoken. Income, age, and race produced some statistically significant differences, but not in a consistent pattern of dominance. Those with higher incomes took more time in the discussion and used more words, but spoke less frequently.28 Comparing participants over and under fifty years old yielded statistically significant differences only on the topics of health care and education. Even here, although those over fifty years old contributed significantly more statements and used more time, they did not use more words in the discussions. Like age, the results for race, coded as white and nonwhite, were mixed. In the discussion of the candidates in the presidential primaries, white participants contributed statistically significantly more statements than nonwhite participants, but did not use more words or time. In the discussion of issues relating to health care and education, however, nonwhite participants used more time, words, and statements than the white participants.
In short, in these five Deliberative Polls, the more- privileged participants did not consistently dominate the deliberations. Only one Deliberative Poll produced an instance in which one group used more time, words, and statements. In this instance, the demographic variable was race. But in this case, nonwhite participants contributed more than white participants in the discussions of both health care, with a specific focus on the rising costs of health care and universal health care, and education, with a specific focus on educational standards, testing, No Child Left Behind, vouchers, and charter schools. In this deliberation, on average across the small groups, nonwhite participants used twenty minutes each of the allotted deliberation time, while white participants used thirteen minutes each. (This difference is statistically significant at the highest level: p=0.000. Each small group session allowed for between fifty-five to seventy minutes of discussion time.) Nonwhite participants collectively used an average of 2,587 words, whereas white participants used 1,742 words (p=0.002). Nonwhite participants also contributed more statements than white participants: an average of sixteen versus twelve statements, respectively. In short, the nonwhite participants made more individual statements, using more words and time per statement.
Although the participation measures show no consistent domination by the more-privileged participants, that may be because more-privileged participants have more experience in deliberative settings, and thus their contributions influenced their group members' opinions more effectively, bringing opinions closer to their own. The analysis used to test for this possibility did so by quantifying the pre- and post-deliberation opinions of participants to see whether their opinions moved after deliberation and, if so, in what direction. The analyses examined the proportion of participants in each small group that moved either closer or further away (in a binary indicator) from the opinions of more-privileged participants. If, after deliberation, participants in a small group moved closer to the opinions of the more privileged, the movement was coded as a 1; if further away, a 0. Across all five Deliberative Polls, on average, the proportion of small groups that moved closer to the opinions of the more privileged was about 0.500–a coin flip.
For this analysis, “more privileged” means being either male, more highly educated (having an undergraduate four-year college degree or beyond), with a higher income ($60,000 or higher annual income), or white. On average, across the five Deliberative Polls, the proportion moving toward the average male opinion was 0.515; toward the average higher-educated opinion 0.542, toward the average higher-income opinion 0.526, and toward the average white opinion 0.484. The range of opinion change toward the more privileged in these five Deliberative Polls was fairly narrow, with the exception of the movement toward the higher educated. The range for the proportion of small groups moving toward the opinions of the more highly educated was 0.448 to 0.714, toward the opinions of males 0.438 to 0.558, toward the opinions of those with more income 0.479 to 0.617, and toward the opinions of whites 0.438 to 0.563. In short, we see no consistent movement toward the more privileged in these Deliberative Polls.
We may now ask whether increased levels of participation themselves might influence participation. That is, regardless of socioeconomic status, does simply participating more in the deliberation cause group members to move toward your views? To answer this question, the analyses applied “participation weights” to each participant, based on the amount of time, number of words, and number of statements used. Using these participant weights, the regression model used three explanatory variables: the weighted opinion of each small group (without the individual's own opinion), each individual's opinion, and each individual's knowledge. The dependent variable was individual level change in opinion. (The explanatory variables of the weighted opinion of the group and each individual's opinion were measured before deliberation; the variable of each individual's knowledge was measured after deliberation.)
If the standard social patterns revealed in the jury studies held up, the regression analyses should have yielded a statistically significant and positive association between the weighted group mean opinion (group mean with participation weights applied) and individual change in opinion. In these five Deliberative Polls, however, none of the coefficients were statistically significant. Only two of the five coefficients were in the hypothesized positive direction.29 In short, participants with higher participation scores did not influence outcomes any more than participants with lower participation scores.30
The final piece of empirical evidence from these data speaks to the possibility, suggested by some critics of deliberation, that those who are less privileged will be less likely to deliberate by giving explicit justifications and reasons (“argument quality”). Some earlier experiments interviewed experts and nonexperts to determine the ability of persons in these two groups to offer and counter arguments. These experiments found that most interviewees, both expert and nonexpert, were capable of defending their arguments.31 Other experiments, however, have found that citizens who had previously engaged in political conversation were more likely to offer reasoned arguments in deliberation.32
Using the transcripts of these five Deliberative Polls, the analysis examined how well participants defended their views according to whether or not they offered reasons or justifications for their statements. The amount of justification was categorized as 1) statements without reasons; 2) statements with one reason; and 3) statements with two or more reasons. For example, the simple statement “I support free trade” would be coded as a statement without reasons. The statement “free trade is harmful because it takes away jobs from the United States” would be coded as a statement with one reason. The statement “the government should consider universal health care because millions of Americans are uninsured and governments in other countries provide universal health care for their citizens” would be coded as giving two or more reasons.
It turns out that gender, income, and political affiliation did not predict reasoned arguments in these five small group deliberations. The significant explanatory variables were predeliberation knowledge, education, race, and age. But in the case of the social variables–education, race, and age–the association did not fall in the predicted direction. Predeliberation knowledge did perform as predicted. As one might expect, the small groups with more participants with higher predeliberation knowledge scores produced the greatest number of reasoned arguments. These groups contributed more statements with two or more reasons, more statements with one reason, and fewer statements with no reasons. Education, however, revealed a pattern contrary to prediction. The more-educated small groups offered fewer reasons in their arguments. That is, they contributed fewer statements with two or more reasons, fewer statements with one reason, and more statements with no reasons. Thus, although the more highly educated groups were more likely to contribute opinions to the deliberations, they were not as likely to give the reasons for these opinions. Race, too, showed a pattern contrary to expectations: the less privileged offered more reasons in the deliberations. The small groups with more white participants contributed significantly fewer statements with one reason and more statements without any reasons. Groups with more whites also contributed fewer statements with two or more reasons, but this difference was not statistically significant. Finally, age also performed against traditional privilege-based predictions, although only one of these associations was statistically significant. The younger small groups (that is, having the greatest number of participants under fifty years old) were more likely to contribute statements with two or more reasons. They were not significantly more likely to offer one reason or no reasons. Summing up, although predeliberation knowledge did perform as expected, in these five Deliberative Polls, the groups with more white, highly educated, and older participants did not provide as many reasons for their arguments as the groups with more nonwhite, less-educated, and younger participants. If offering reasons is the quintessential Habermasian characteristic of good deliberation, it seems that the less traditionally privileged groups in these Deliberative Polls acted in the most Habermasian manner.
It is possible that those who believe their statements carry more weight because of their social position feel less need to offer reasons. Those with less power, conversely, might believe that they are required to furnish more justifications for their views. For the same reason, the number of words uttered might not necessarily be a sign of power. Often the most powerful in a meeting speaks least. These possibilities are open to future investigation. At the moment, we can say only that on the obvious and surface dimensions–those that earlier studies of juries have measured and that have informed the conclusions of earlier critics of deliberation–the predicted patterns do not emerge.
The empirical evidence presented in this essay has demonstrated that, when citizen deliberations are well structured, the many social patterns that we might expect from inequalities in the world around us are, to some degree, negated. The more privileged do not consistently dominate deliberations, nor are their opinions more influential than their fellow group members. The participants who speak most frequently and for the longest duration in the conversations also have no greater influence over the rest of the group. The idea that the more privileged will be more capable of engaging in the reasoned exchange of justification is also, in this setting, not true.
What characteristics of these deliberations made them differ from the predictions of deliberation's critics? The relevant characteristics are clear and replicable. These deliberations began with a selection from the citizenry that was closer to random than most such endeavors, such as jury studies, and far more reflective of the citizenry than the typical citizen forum, which attracts self-selected persons, often activists or the extremely interested. This deliberative design included sending the participants, before they arrived, informational materials that were balanced and agreed upon by both sides of the issue at stake. Because the deliberations were special occasions and the citizens involved were aware of being personally selected for participation, they typically read some of the materials they were sent, thus becoming aware, more or less, of information on the side of the issue that they might usually oppose. During the deliberative process, participants also engaged with experts on both sides, an experience that again exposed them not only to much more information than they usually would have access to, but particularly to information on the side opposing their predisposition. The small group deliberation took place in two stages, one of which had time dedicated to preparing questions for the experts in the plenary sessions. Since not all questions could be asked, small groups had to carefully consider which questions would be most useful in helping them form their own opinions. They indirectly learned about one another's views through this common task, but the process was designed so that they were focused on developing their group's questions and were not likely to confront one another on opposing views at this stage. Finally, and perhaps most important, small group discussion moderators were trained explicitly to solicit opinions from everyone in the group, encourage all participants to speak, and maintain a tone of courtesy and nonconfrontation in pursuit of mutual understanding. With these conditions in place, the predicted dynamics of domination by the more privileged did not appear.
This examination of the empirical evidence from five projects–with 1,474 participants and ninety-nine small groups–is, to my knowledge, the largest study of the internal dynamics of deliberation ever undertaken. It measured class, race, gender, and previous knowledge, and traced the effects of these variables on several forms of participation and influence. The conclusion–that in fairly easily replicable circumstances, the expected forms of privilege have little effect on participation and no effect on influence–ought to undermine the automatic association of deliberation and inequality.
Amy Gutmann and Dennis Thompson, Why Deliberative Democracy? (Princeton, N.J.: Princeton University Press, 1996); James Fishkin, Democracy and Deliberation: New Directions for Democratic Reform (New Haven, Conn., and London: Yale University Press, 1991); James Fishkin, The Voice of the People: Public Opinion and Democracy (New Haven, Conn., and London: Yale University Press, 1995); and James Fishkin, When the People Speak: Deliberative Democracy and Public Consultation (Oxford: Oxford University Press, 2009).
John Stuart Mill, Considerations on Representative Government (New York: Prometheus Books, 1991 ), 79.
Jürgen Habermas, The Structural Transformation of the Public Sphere: An Inquiry into a Category of Bourgeois Society, trans. Thomas Burger (Cambridge, Mass.: The mit Press, 1989), 27.
Jürgen Habermas, “Further Reflections on the Public Sphere,” in Habermas and the Public Sphere, ed. Craig Calhoun (Cambridge, Mass.: The MIT Press, 1996), 421–461.
Claus Offe, “Crisis and Innovation of Liberal Democracy: Can Deliberation Be Institutionalised?” Czech Sociological Review 47 (3) (2011): 447–473.
Nancy Fraser, “Rethinking the Public Sphere: A Contribution to the Critique of Actually Existing Democracy,” Social Text 25/26 (1990): 55–80; and Nancy Fraser, “Transnationalizing the Public Sphere: On the Legitimacy and Efficacy of Public Opinion in a Postwestphalian World,” in Identities, Affiliations, and Allegiances, ed. Seyla Benhabib, Ian Shapiro, and Danilo Petranovic (Cambridge: Cambridge University Press, 2007), 45–66.
Fraser, “Rethinking the Public Sphere.”
Ibid. For more recent work on these behaviors, see Christopher Karpowitz and Tali Mendleberg, The Silent Sex: Gender, Deliberation, and Institutions (Princeton, N.J.: Princeton University Press, 2014).
Iris M. Young, “Communication and the Other: Beyond Deliberative Democracy,” in Democracy and Difference, ed. Seyla Benhabib (Princeton, N.J.: Princeton University Press, 1996), 60–74.
Ibid; and Iris M. Young, Inclusion and Democracy (New York: Oxford University Press, 2000).
Lynn M. Sanders, “Against Deliberation,” Political Theory 25 (3) (1997): 347–373.
Phoebe C. Ellsworth, “Are Twelve Heads Better Than One?” Law and Contemporary Problems 52 (1989): 205–224; Reid Hastie, Steven Penrod, and Nancy Pennington, Inside the Jury (Cambridge, Mass.: Harvard University Press, 1983); Valerie P. Hans and Neil Vidmar, Judging the Jury (New York: Plenum Publishing Corporation, 1986); Rita M. James, “Status and Competence of Jurors,” The American Journal of Sociology 64 (6) (1959): 563–570; Charlan Nemeth, Jeffery Endicott, and Joel Wachtler, “From the ‘50s to the ‘70s: Women in Jury Deliberations,” Sociometry 39 (4) (1976): 293–304; and Fred L. Strodtbeck and Richard D. Mann, “Sex Role Differentiation in Jury Deliberations,” Sociometry 19 (1956): 3–11.
Hans and Vidmar, Judging the Jury.
Hastie, Penrod, and Pennington, Inside the Jury.
Fred L. Strodtbeck, Rita M. James, and Charles Hawkins, “Social Status in Jury Deliberations,” American Sociological Review 22 (6) (1957): 713–719.
Hastie, Penrod, and Pennington, Inside the Jury; and Hans and Vidmar, Judging the Jury.
In respective order: James Fishkin, “Realizing Deliberative Democracy: Virtual and Face to Face Possibilities,” Electronic Working Papers Series, Paper No. DEG 007 (Vancouver: W. Maurice Young Centre for Applied Ethics at the University of British Columbia, 2005); ibid; Center for Deliberative Democracy, “PBS Deliberation Day Shows Americans' Views on Iraq, Trade,” January 24, 2004, http://cdd.stanford.edu/2004/pbs-deliberation-day-shows-americans-views-on-iraq-trade/; Center for Deliberative Democracy, “PBS Deliberation Day on National Security and the Global Economy,” October 16, 2004, http://cdd.stanford.edu/2004/pbs-deliberation-day-on-national-security-and-the-global-economy/; and Center for Deliberative Democracy, “By the People: A National Deliberation on Healthcare and Education,” September 6, 2005, http://cdd.stanford.edu/2005/by-the-people-a-national-deliberation-on-healthcare-and-education/.
The data sets for these analyses are available to researchers at the Center for Deliberative Democracy, https://cdd.stanford.edu/. This research is part of a larger project that will compare online and face-to-face deliberations. In the analyses that follow, the groups reported on averaged a composition of 52 percent men, 37 percent high income, and 18 percent nonwhite, with an average age of forty-eight years old. As a point of comparison, the 2004 National Election Study's group was 47 percent male, 40 percent high income, and 28 percent nonwhite, with an average age of forty-seven years old.
Of the two Deliberative Polls that were analyzed, income was unavailable for one. Therefore, this result is based on only one Deliberative Poll.
A further note on the regression analyses: The analyses were conducted on two of the five Deliberative Polls, because the transcripts for three of the Deliberative Polls did not have individual identifiers, which made it too difficult to identify participants. Without identifiers, the regression analyses were not possible. And as mentioned, of the two Deliberative Polls analyzed, income was unavailable for one. Therefore, one Deliberative Poll had three explanatory variables for the regression analyses and another had two explanatory variables for the analyses, making a total of five explanatory variables or five coefficients from the regression analyses.
The analyses were also conducted using unweighted variables. The results were similar, indicating that participation levels are not a predictor of opinion changes.
Deanna Kuhn, The Skills of Argument (New York: Cambridge University Press, 1991).
Joseph N. Cappella, Vincent Price, and Lilach Nir, “Argument Repertoire as a Reliable and Valid Measure of Opinion Quality: Electronic Dialogue during Campaign 2000,” Political Communication 19 (1) (2002): 73–93.