Can Deliberative Democracy Work in Hierarchical Organizations ?

Some measure of equality is necessary for deliberative democracy to work well, yet empirical scholarship consistently points to the deleterious effect that hierarchy and inequalities of epistemological authority have on deliberation. This article tests whether real-world deliberative forums can overcome these challenges. Contrary to skeptics, it concludes that the act of deliberation itself and the presence of trained moderators ameliorate inequalities of epistemological authority, thus rendering deliberative democracy possible, even within hierarchical organizations.


Introduction
Democratic theory took a deliberative turn in the 1990s, such that aggregative conceptions of democratic legitimacy were abandoned for those emphasizing the importance of individuals participating substantively in collective decisions.The literature produced from this turn has tackled many thorny theoretical issues inhering in deliberative democracy.Empirical questions-over just how, when, and where deliberative democracy can and cannot operate-have received less attention. 1Much of the scholarship still orients itself, rightly or wrongly, toward theoretical questions rather than empirical.Absent the feedback loops from the empirical to the theoretical scholarship, many assumptions of deliberative democratic theory remain untested.Proponents assume, for example, that a certain level of equality exists among participants, while hierarchy and status differences can be harbored from deliberation.Scholarship from communication studies and psychology consistently point, however, to the deleterious effects that hierarchy and inequalities in status have on deliberation.Can real-world deliberative practices overcome the hierarchy hurdle?This article presents findings from a deliberative experiment that tested just this question. 2While limited in scope, the experiment revealed that deliberation itself and the presence of trained moderators ameliorated the effects of status differences.This article also points to fertile ground for further empirically oriented deliberative democracy research.

Theoretical Expectations for Equality
Some measure of equality is necessary for deliberative democracy to work well, according to most of its proponents, but one finds various shades of equality discussed in the literature.Jim Fishkin's Democracy and Deliberation posits a broadly conceived notion of equality that "grants equal consideration to everyone's preferences and which grants everyone appropriately equal opportunities to formulate preferences on the issues under consideration" (1991: 30-1).Advancing his case for the liberal state, Bruce Ackerman calls for "neutral dialogue" between citizens where, "No reason is a good reason if it requires the power holder to assert: (a) that his conception of the good is better than that asserted by any of his fellow citizens, or (b) that, regardless of his conception of the good, he is intrinsically superior to one or more of his fellow citizens" (1980: 11; see also Cohen 1989;Elster 1998;Gutmann & Thompson 1996;Saward 2001;Chambers 2005).Jurgen Habermas, whose scholarship proved central to the deliberative turn, envisions discourse free of constraints and distortions-an ideal speech situation-that guarantees an individual's "participation in all deliberative and decisional processes" and "provides each person with equal chances to exercise communicative freedom to take a position on criticizable validity claims" (1996: 127).
These permutations on equality as a precondition for deliberation exist, in part, because theorists differ on the ultimate point of deliberation.Some think it produces better decisions or greater adherence to decisions.Others claim that it uncovers preferences or opinions that might otherwise remain veiled through aggregative politics.Still others see deliberative exercises promoting some combination of these.The theoretical ends of deliberative democracy are indeed varied, but the ends of real-world deliberative forums are particularized to the people who assemble, the topics discussed, and the contexts in which the forums occur.The forum discussed in this article consisted of a university community deliberating how the institution could promote greater academic excellence and engagement on campus.It raised complex, systemic questions about institutional mission, campus culture, resource allocation, and curricular and co-curricular matters.Any changes prompted by the forum would require the coordinated efforts of many individuals and constituencies across the university.Thus, the kind of equality needed in this forum was one where 1) every member of the community had the opportunity to participate, 2) once deliberating, every individual had an equal opportunity to share opinions and listen to others, and 3) individuals had equal opportunity to have their arguments acknowledged by the group.
Deliberative democratic theorists recognize the importance of equality as a precondition for effective deliberation, yet readily admit their calls are often aspirational.Fishkin, for instance, describes Habermas' ideal speech situation as "hypothetical" and "utterly utopian" and concludes that the real world presents institutions "closer to the nondeliberative end" (1991: 36-7).Gutmann and Thompson concede to the elitist critique of deliberation-that deliberative ability corresponds to social, education, and economic status-but conclude that deliberation still offers the best way for diminishing the effects of status and inequality.In the end, the gap they see between deliberative theory and practice is surmountable and "narrower than in most other conceptions of democracy" (1996: 357).They may be correct.On the other hand, the gap between the equality presumed to be theoretically necessary for effective deliberation and the equality available in practice may be wider than some think.
Consider, for example, the challenge of implementing effective deliberative practices within hierarchically ordered organizations, such as universities, corporations, or the military (see Gerencser 2005).Differences in status and the inequalities of power that flow from those status differences are necessary and inevitable for the institution to function, yet they pose real challenges when its members deliberate.The challenges are all the more acute when the deliberation concerns the institution's purpose and future.Consider an assembly-line worker discussing a corporation's business plan with the chief executive officer, or an entry-level soldier discussing war strategy with the commanding general, or the first year college student discussing post-tenure review policies with a university provost.In each situation, the inequalities of power stemming from status differences may discourage the sort of equality envisioned by deliberative democrats.
Organizational theorists have noted that such hierarchy naturally produces inequalities of power within small groups (Lee & Tiedens 2002).These inequalities of power clearly affect group dynamics, including the ways individuals interact and communicate.High-status individuals participate more frequently than low-status individuals because contributions from the former are perceived as more valuable and more likely to produce optimal group decisions.By way of example, Skvoretz (1981) noted differences in participation rates among doctors and nurses during hospital administrative meetings.Pauchet (1982) described how senior faculty dominated discussion at university faculty meetings as their junior counterparts remained relatively silent.Documented in these and other contexts, individuals tend to make social comparisons within small groups that orient participation rates (see Berger & Zelditch 1977;Balkwell 1994;Knottnerus 1997).
Hierarchy wrecks havoc on more than participation rates.High-status persons tend not to individuate low-status persons in group conversations, in part to maintain their position and authority (Goodwin, Gubin, et al. 2000).In turn, low-status individuals tend to apply stricter standards to themselves than to highstatus individuals when evaluating ability and judgment (Foschi 1996).Finally, scholars have found that power advantages may also lead to stereotyping and discrimination (Sachdev & Bourhis 1991).
Thus, deliberative processes are potentially hampered when they occur within hierarchically ordered organizations.Absent rules or structures that put participants on a more level playing field, deliberations could be plagued by what Lynn Sanders calls inequalities of epistemological authority, which she defines as the capacity to "evoke acknowledgement of one 's arguments" (1997: 349).This concept gets at the fact that some people have an easier time than others having the group acknowledge their arguments and treat those arguments as authoritative, not because they are necessarily correct, but because they come authentically from a member of that organization.True deliberation requires that all participants, regardless of position within the organization's hierarchy, have equal opportunity to have their arguments acknowledged by the group.Each must feel equal authority going into a deliberation, but these conditions may not be present.Sanders points to a number of sources for these inequalities, including gender, race, and status.
On the one hand, it is not surprising to find inequalities of epistemological authority within hierarchically ordered organizations.Differences in status and power often reflect real differences in participants' abilities to articulate and act upon their thoughts about the institution.CEOs, military generals, and university provosts possess more epistemological authority within their institutions at least in part because they have a view of the whole to an extent that the assembly-line worker, private, or undergraduate student does not.
That does not mean that these high status members would be uninterested in deliberating with lower-status members in a setting where hierarchy was less pronounced and consequential.Doing so, they might expect to make better decisions, they might secure greater adherence to the decisions they made, and they may capitalize on discrete knowledge or competencies that otherwise remain untapped within an organizational hierarchy.
The presence of inequalities of epistemological authority among participants in deliberative democratic forums runs counter to the theoretical presumption that many of its proponents make-that deliberation is a collective conversation among co-equals about issues of common concern.When the forums occur within hierarchically ordered organizations, such inequalities might be expected, but their effects are corrosive all the same.Thus, a major empirical question facing deliberative democracy is whether or not it can overcome inequalities of epistemological authority stemming from organizational hierarchy and status differences?

Research Design, Variables, and Hypotheses
Our experiment took place during a half-day deliberative forum.Over 500 university students, faculty, and administrators participated in the deliberation, modeled on the popular and well-established National Issues Forum (NIF) format.NIF events invite participants to discuss complex social, political, and economic issues in small groups.Participants typically consider three different (competing) ways of approaching the issue and identify the merits, demerits, and tradeoffs associated with each approach.The NIF model asks all participants to read in advance an informational booklet about the issue and possible approaches, giving everyone a common knowledge base.No one is identified before or during the small group discussions as an expert or having specialized knowledge.All are presumed equal in their capacity and authority to attend and contribute to the discussions.Participants were randomly assigned to the groups, which averaged 10 to 12 people.
National Issues Forums use trained moderators to encourage broad participation in the small groups and to solicit perspectives that have not been voiced.In this way, NIF moderators aid the group in vetting fully the topic.Achieving these goals requires that they not participate in the deliberation and remain neutral toward the issue, the competing approaches, and participants' comments.Roughly half of the groups in our forum had NIF moderators.The other half had moderators who actively participated in the deliberations, were not neutral toward the issues, approaches, or other participants' comments, and did nothing to foster broad participation.The moderators were randomly assigned to the small groups as well.
The composition of the small groups reflected various hierarchical relationships at the university: student-faculty, student-administrator, and student-faculty-administrator.A number of groups had only students.Given the small group size and based upon pre-forum testing with students, we decided that the presence of a single faculty member or administrator in a group was sufficient to introduce a hierarchical element into the deliberations, although many groups had several high-status members.3Although our research design was constrained by the unavoidable reality that the pool of higher status participants (faculty and administrators) was much smaller than the pool of lower status participants (students), we expected many students would react to the small group deliberations not unlike a university classroom, such that they were cognizant of status differences given the presence of a single higher status participant.
Forum participants completed a brief survey prior to their small group deliberations that measured their own epistemological authority.A post-forum survey also was administered that matched the pre-forum questionnaire.Appendix B includes both surveys.The matched response sets were coded for various participant and small group characteristics.Using a five-point Likert scale, participants indicated how strongly they agreed or disagreed with statements that captured different dimensions to epistemological authority, with higher scores indicating greater agreement.We were unaware of extant survey questions that measure epistemological authority in deliberative forums, so we constructed our own.These questions were not exhaustive by any means and future research would do well to evaluate them in other settings and against alternative questions.
We thought it important that the survey questions explore facets of epistemological authority germane to our forum's participants.This required sensitivity to how status might affect participants before and during the forum.
Asking students, faculty, and administrators to deliberate on how their university should promote greater academic excellence and engagement would raise delicate subjects: Are faculty demanding enough from students in the classroom?Is the students' social scene having a deleterious effect on academics?What steps can administrators take to improve the institution's reputation and how should they allocate resources to promote greater academic excellence?These questions asked participants to turn critical eyes on others and themselves, as they considered what role they could play in a larger effort at institutional reform.Our background research revealed two important considerations.First, students, faculty, and administrators would exhibit varying levels of comfort talking about these issues.Second, people disagreed over how important student opinion and involvement were to administrators and faculty.
One can see how status differences and the inequalities of epistemological authority that stem from them could hamper deliberation.Status differences could feed discomfort, a feeling that some care more about the institution than others, or a sense that some are more responsible for implementing reforms than others.Given these contextual variables, epistemological authority would come if, at the least, participants felt comfortable talking about the topic with others, and perceived that others cared about their opinions.
Regarding the first criterion, we anticipated that participants may be more or less comfortable talking about certain topics.Students may have greater confidence, for instance, conveying the state of campus culture, while faculty may have a better grasp of classroom expectations, and administrators may feel more comfortable discussing curricular reform.Thus, the survey posed three comfort questions: how comfortable they were talking about the forum's theme of academic excellence (henceforth in the tables and charts, Comfort-Excellence), how comfortable they were talking about the university's identity and reputation (Comfort-Identity), and how comfortable they were talking about campus culture (Comfort-Culture).
The second dimension, less introspective and more projective, explored whether a participant thought other status groups cared about their views.Two care questions were posed: whether professors care about student views (Care-Professor) and whether student views matter to the university's administrators (Care-Administrator).
The comfort and care questions are sensitive to different facets of a person's epistemological authority.The former offers an internal measure; that is, how easy or difficult a person perceives contributing to the deliberation.The latter offers an external measure, asking how they perceive high status participants' attitudes toward lower status participants.The comfort and care questions inevitably covary and perhaps covary by status.One could imagine a lower status participant perceiving that a higher status participant cared about his or her opinion, and thus, felt more comfortable participating; alternatively, a lower status participant's comfort could decrease after perceiving indifference from a higher status group member.On the other hand, a higher status person's comfort does not likely correlate positively with that person's perception that lower status participants are showing regard for their opinion.Status alone may account for higher status participants feeling more comfortable.The pre-and post-forum responses to the foregoing questions were analyzed for any statistically significant shifts, which could indicate that the deliberative experience itself shaped these facets of epistemological authority.
Two additional post-forum questions evaluated the quality of the deliberation.The first asked about the participation rate within the small group, specifically whether discussions were dominated by a few individuals.The second asked whether the respondent thought everyone's opinions were equally valued.
Both measure equality of epistemological authority within the respondent's small group.These last two questions move respondents from thinking about how they felt before and during the deliberation to evaluating observable group dynamics: who offered comments and how did the group treat those comments?These questions enabled us to measure the moderators' effect on deliberation.Thus, we posed the following hypotheses.

H1: Deliberation Hypothesis
We first examine the effect that the act of deliberation itself has on epistemological authority.Does simply bringing together members of a hierarchically ordered organization for a deliberation engender greater epistemological authority?Status differences present prior to the forum should decrease as a result of the deliberation.We test such an effect by comparing individual participants' pre-and post-forum responses.If the deliberation had the anticipated effect, post-forum responses at the individual level would show increases in epistemological authority evidenced by greater comfort talking about the topic and a heightened sense of regard for others' opinions.A participant's comfort should increase because the venue invites opinions from all stakeholders and offers a salubrious setting for sharing those opinions.In our forum, participants were given permission to talk about a thorny topic that may be plumbed at shallower depths absent the deliberative space.We anticipate lower status participants will observe higher status participants listening and sharing their own opinions, and thus should perceive higher status participants caring more about lower status participants' opinions.We expect this will be present only for students in the hierarchical groups.We also expect that higher status participants will report themselves caring more about lower status participants because of their presence and involvement in the small group.We expect this effect to hold for all status categories, regardless of moderator type.

H2: Status Hypothesis
We anticipate that epistemological authority varies among participants, in part, by status.Higher status individuals are thought to enter a deliberation ascribing more epistemological authority to themselves than lower status individuals give themselves.This should be evident in this project with higher status individuals (faculty and administrators) feeling more comfortable talking to others about the issues and reporting a greater willingness to listen than lower status participants (students) report.We further anticipate that the deliberation's effect will vary by status, with lower status participants exhibiting a greater shift in epistemological authority than higher status participants.

H3: Moderator Effects Hypothesis
Scholars have long recognized the central role that moderators play in deliberative forums (see Karpowitz & Mansbridge 2005;Levine, Fong, & Gastil 2005).We examine the influence of NIF moderators on epistemological authority within small group deliberations.We hypothesize participants deliberating with NIF moderators will report greater post hoc epistemological authority than those deliberating with non-neutral, participatory moderators.Given the presence of status differences within the groups, we also expect the moderator effect will vary by status.NIF moderators are expected to foster more epistemological authority in lower status participants than higher status participants because the former are thought to benefit more from such a deliberative environment.
The next section explores answers to these hypotheses.Our analysis operates at the individual level rather than by small group.There are several reasons for this.To begin, the first two hypotheses really concern individual participants-how individuals assess their epistemological authority prior to the event (H1), how the deliberation influenced those assessments (H1), and whether its effect varied by status (H2).These hypotheses are best tested using individual level data because they concern individuals' preconditions and reactions to the forum.Second, individual level data produce more reliable results because of the larger sample size.If we analyzed the data at a group level and disaggregated them according to group composition and moderator type, we would be left with small sample sizes that could produce spurious results.Third, individual level analysis still enabled us to assess moderator effects and whether they varied by status (H3).While our unit of analysis is the individual participant, we test for group effects at several points.

FINDINGS
Were inequalities of epistemological authority brought to the deliberation?
Before analyzing the impact that the deliberative process and NIF moderators had on epistemological authority among participants, we first examine if participants came to the event with a priori conceptions of status and if these conceptions shaped the epistemological authority they ascribed to themselves and others.Second, we determine if the epistemological authority that each participant came with correlated to his or her status.
The pre-forum data, presented in Table 1, provide strong, statistically significant evidence that participants were conscious of status and ascribed to themselves levels of epistemological authority commensurate with their status.Reflected in the mean response scores, students showed the least comfort talking about academic excellence, the university's identity and reputation, and the campus culture, posting statistically significant differences with faculty and administrators.Mean scores for administrators were lower than faculty scores on the comfort questions, although none statistically significant. 4s to the care questions, if respondents brought to the deliberation anticipated notions of status, lower status participants (students) should report that higher status participants (faculty and administrators) cared less about student views than higher status individuals self-report.As hypothesized, students scored lower than faculty and administrators on both questions, meaning that they thought faculty and administrators cared less about student views than faculty and administrators reported themselves; moreover, the differences of means between students and faculty and students and administrators were statistically significant.These results offer further evidence that status differences contributed to variations in a priori epistemological authority.
Students reported consistently different scores on the epistemological authority questions from faculty and administrators, but differences of means between faculty and administrators were not statistically significant on any measure.This suggests that the hierarchical structure shaping this particular deliberation was two-tiered (students and faculty/administrators), rather than three tiered (student and faculty and administrator).
These pre-forum results show that an individual's comfort level going into the deliberation varied by status, with higher status participants feeling more comfortable than lower status participants.Participants ascribed to themselves different levels of epistemological authority based, in part, on status.Linking status with epistemological authority in this way confirms that participants were cognizant of and brought to the deliberation the university's hierarchical power structures.Participants may not have thought intentionally about status differences before the deliberation, but the pre-forum results bear out the connection to epistemological authority.

Does deliberation increase epistemological authority among participants? Do its effects vary by status?
We hypothesized that deliberation itself diminishes inequalities of epistemological authority.The simple process of sitting in a small group and discussing a common issue and possible solutions was expected to engender greater epistemological authority.Higher status participants who may consider others' views less informed, less thoughtful, and less authoritative may hold to these assessments with less vigor after the deliberation.Similarly, lower status participants who may have entered the deliberation thinking that their views were less informed, less thoughtful, and less authoritative would come away with greater epistemological authority.
To assess this hypothesis, paired T-tests were conducted on individuals' pre-and post-forum responses to the epistemological authority questions.Chart 2 presents the mean pre-and post-forum results for all respondents, with the difference of means and significance level recorded above each pairing.The hypothesized result is evident in each question: the deliberation made participants feel more comfortable talking about the issues and more confident that higher status participants were concerned about lower status participants' views.The largest shift occurred on respondents' comfort level talking about academic excellence (nearly half a scale category) and secondly about campus culture (over a third of a scale category).Smaller shifts occurred in the remaining pairs.The difference of means tests were significant on all of the comfort questions and one of the care questions.
The deliberation's effect at fostering epistemological equality was also anticipated to vary by status.Lower status participants were expected to shift more than higher status participants because the organization's hierarchy conditions the former to feel less comfortable talking about these issues than the latter.Chart 3 reports the difference of means scores by status.The data lend strong support to the status hypothesis.Students lodged larger pre-to post-forum shifts on four of the five questions than faculty and administrators.The deliberative process had significant influence on student comfortableness, where statistically significant increases were reported for all comfort questions.More modest change occurred in student opinions toward whether professors and administrators were interested in student views.The deliberation's effect on students varied little based on group type; that is, whether students were assigned to a student-only group or a mixed (hierarchical) group.Students in hierarchical groups recorded statistically significant, positive difference of means scores on all three comfort questions and the care-administrator question.Those in studentonly groups also recorded statistically significant, positive difference of means scores on the three comfort questions, but neither of the care questions. 5 The magnitude of the shifts was slightly larger for students in the hierarchical groups, but not enough to suggest a group effect.Thus, deliberation fostered greater epistemological authority for students across group type.We hypothesized that students in hierarchical groups would record larger shifts on the professors and administrators care questions than students in student-only groups because they would witness firsthand higher status participants listening to and participating in the discussion.Some support emerged for this as well.While neither of the care questions was significant for the student-only groups, students in the hierarchical groups lodged a significant and positive increase on the administrators care question.
As to the faculty and administrators, shifts in the comfort questions were not significant, whereas they were on the two care questions.Thus, higher status participants went away from the deliberation feeling no more comfortable talking about the issues, but they had stronger convictions that they were interested in what lower status participants thought.
The results in Chart 3 support the status hypothesis.It makes sense that higher status individuals entered this deliberation feeling comfortable talking about the issues and left feeling much the same.Indeed, it would be quite surprising if they reported much change at all on the comfort questions.The process should have its greatest effect on those who come into it feeling less comfortable and who think that higher status individuals are less concerned with their views.
To generate a meta-level measure for status effect, differences of means for the paired questions were summed for each status group.Acknowledging that not all of the differences of means are statistically significant, scaling these summed figures enables a rough comparison between status groups.Calculating this "sum shift" figure proceeds as follows: Students registered a larger sum shift (Σ Student =1.19) than faculty and administrators (Σ Fac./Admin.=.93).This measure supports the hypothesis that the deliberation's effect on participants' epistemological authority varied inversely with status.We are confident these results reflect status differences stemming from participants' relative positions within the university, but we recognize that age may have played a role as well.Teasing out whether age, institutional status, or some combination explains these variations is difficult because of the colinearity across the participant pool between institutional status and age.An overwhelming majority of the university's undergraduate student population is traditional age.Student participants were uniformly younger and lower status than faculty and administrators.Thus, we did not collect age data, figuring that it would shed little additional like on epistemological authority.This approach would be ill-advised, of course, if measuring epistemological authority within institutions where the colinearity between status and age is not as strong.One can imagine a very different dynamic in a hierarchical organization, such as a corporation, where an employee's age does not necessarily track status.
There is, however, additional evidence that status, not necessarily age, was at play in this forum.First, we found statistically significant differences between faculty and administrators on two of the epistemological authority questions.These two status groups are certainly closer if not coterminous in age, but these findings indicate that age alone does not account for epistemological authority.Second, although not discussed in this article, we found statistically significant differences on epistemological authority among students based on classificationfirst-year, sophomore, junior, and senior.The students' age differences were modest-generally no more than four years-but the status differences that come with classification were rather pronounced.Thus, while recognizing the challenges distinguishing between age and institutional status in this experiment, we think the latter proved crucial.Future research would do well exploring the relationship between age and status in different institutional settings.
Regardless of the ultimate source of the inequalities of epistemological authority, these collective findings offer strong support for the deliberation and status hypotheses.The act of deliberating engendered greater epistemological authority among all participants, while the deliberation's effects varied by status.This challenges critics of deliberative processes who argue that inequalities of epistemological authority within small groups thwart effective deliberation.Our experiment found just the opposite.
Differences in status were not insurmountable.Just as James Madison's solution to the evil effects of factions was the creation of more factions, deliberative democracy may offer a built-in solution to its equality challenge: more deliberation.

Do NIF moderators foster greater epistemological authority?
Having affirmed that the forum itself generated greater epistemological authority among participants, this section explores whether the rules for deliberation had a similar effect.We put the NIF model-small group discussions facilitated by neutral, non-participating, trained moderators-to the test, determining whether NIF moderators fostered greater epistemological authority among participants, and therefore, better deliberation, than non-neutral, participating moderators (non-NIF moderators).It was hypothesized that NIF moderators would create higher levels of epistemological authority in deliberative settings where status differences exist.
The moderator effect was explored using the comfort and care questions, as well as two additional post-forum questions.The first asked respondents if their small group conversations were dominated by a few individuals.The second asked if everyone's opinions were equally valued.Both address the distribution of epistemological authority in the group.A conversation dominated by a few individuals or one where opinions were not equally valued indicates an inequality of epistemological authority within that group.Conversely, a deliberation involving more of the small group and one where opinions were equally valued suggests greater parity in epistemological authority.
Table 4 records the mean responses and differences of means for these two questions by status, moderator type, and group type.Recall, higher mean scores on the five-point Likert scale indicate stronger agreement with the statement, such that higher scores on PostQ4 suggest less equality of epistemological authority and lower scores on PostQ5 suggest more equality.
The results support the hypothesis that deliberations facilitated by NIF moderators were characterized by greater parity in epistemological authority than non-NIF moderators.
When looking at responses from all participants, statistically significant differences in the hypothesized direction emerged on both questions based on moderator type.Participants with NIF moderators reported broader participation and more respect for others' opinions than non-NIF moderated participants.Moreover, the moderator effect held true for all students, regardless of whether they were in student-only or hierarchical groups.Faculty and administrators also shifted in the hypothesized direction based on moderator type.In sum, NIF moderators appeared to foster deliberations characterized by broader participation and more equal treatment of opinions than their non-NIF counterparts, regardless of group composition.
Comparing respondents' answers on the comfort and care questions offers another cut at the moderator effect.Given earlier results showing that  deliberation itself fostered epistemological authority, we expected respondents in both NIF-moderated and non-NIF moderated groups to reflect this.However, NIF moderators were hypothesized to foster more epistemological authority among participants than non-NIF moderators.Evidence supporting this claim would include larger shifts in the comfort and care questions and a larger sum difference of means among individuals in NIF groups than non-NIF groups.
Chart 5 presents the relevant differences of means for the paired comfort and care questions.
Respondents in both NIF and non-NIF groups evidenced positive difference of means results, with one modest exception (NIF-moderated responses to the "administrators care" question).The shifts that occurred with NIF group respondents were larger than non-NIF group respondents for three of the five paired questions.Those deliberating with NIF moderators felt more comfortable deliberating than respondents in non-NIF groups.Thus, NIF moderators generated more epistemological authority among their participants than did non-NIF moderators.
The moderator effect was less clear on the two care questions.When it came to the "professors care" question, NIF group respondents barely shifted in pre-and post-forum responses (pre-x =2.69, post-x =2.72).Non-NIF group respondents shifted more and in a manner that was statistically significant (prex =2.59, post-x =2.77).This suggests that the presence of NIF moderators did not underscore the idea that professors cared about student views.On the other hand, non-NIF moderators seemed to have that effect.This pattern repeated itself with the "administrators care" question.The shifts for NIF group respondents were marginal, statistically insignificant, and not in the hypothesized direction (pre-x =2.29, post-x =2.25), whereas non-NIF group respondents had a significant shift in the anticipated direction (pre-x =2.17, post-x =2.33).These results are admittedly counterintuitive and appear to contravene the hypothesized moderator effect.That non-NIF group respondents' results for the two care questions mirror each other suggests that they may not be aberrations.
A further counterintuitive result comes from calculating the "sum shift" on the comfort and care questions for respondents by moderator type: NIF moderators generated larger shifts on the three comfort questions (the first three figures in the equation), as hypothesized, but lost significant ground to non-NIF moderators on the two care questions (the last two figures).Non-NIF moderators produced larger sum shifts on the epistemological equality questions than NIF moderators, at first blush.

NIF Groups Non-NIF Groups
To make sense of these unexpected results a more nuanced understanding of the effect that moderators have on epistemological authority in this small group deliberation is necessary.
NIF moderators made participants feel more comfortable talking about the relevant issues, but they did not change participant views about whether and how much professors and administrators cared about student views.Given the tasks that NIF moderators accomplished-encouraging broad participation, enforcing the rules of deliberation, preventing a few voices from dominating the conversation, and drawing out arguments otherwise left unarticulated-it makes sense that they would alter the deliberative environment by making it more hospitable for participants, particularly lower status participants.It may be more difficult, if not beyond the role's intent, however, for the NIF moderators to move beyond the deliberative environment and actually shape perceptions of how higher and lower status participants think about each other.This conclusion is consistent with results from Chart 3 on the care questions.Recall we hypothesized that the magnitude of the deliberation's effect would vary indirectly with participant status.Results for the care questions did not support this hypothesis.Students did not shift, in statistically significant ways, more than faculty and administrators.Coupling this with the moderator effect results points to a more complicated epistemological authority dynamic.

Does the moderator effect vary by status?
Having established the capacity of NIF moderators to foster greater epistemological authority, just who within the small groups shifted opinion?Did the NIF moderator effect vary among participants according to status?We hypothesized that shifts in epistemological authority would be present among both lower and higher status participants, but that lower status individuals would record larger shifts.Lower status individuals were thought to benefit more from the deliberative environment that NIF moderators create because they feel less authoritative going into the deliberation than higher status participants.This would be evidenced in lower status participants recording larger difference of means scores on the pre-and post-forum epistemological authority questions than higher status participants.
Chart 6 reports the difference of means scores on the paired comfort and care questions by status.Remember, the larger the difference of means, the larger the NIF-moderator effect on that status group.The data provide solid support for this hypothesis.
NIF moderators had a larger effect on student scores vis-à-vis faculty/administrators for all three comfort questions, but little effect on the care questions.Neither status group moved much on the care questions, further underscoring a pattern in the survey data that the comfort and care questions tap different components to epistemological authority.NIF moderators clearly made participants more comfortable about the deliberative process and the extent to which this occurred varied, as hypothesized, by status.However, NIF moderators were generally less efficacious changing perceptions of how participants viewed others, particularly how higher status views were perceived by lower status participants.
The one exception from Chart 6, of course, is the faculty and administrator shift on the "professors care" question.This result is best understood in tandem with faculty and administrators in non-NIF groups.Table 7 shows that faculty and administrators as a status group, regardless of moderator type, shifted significantly on the "professors care" question: mean shift for those in NIF groups (.350) and non-NIF groups (.400).The most likely explanation, therefore, is that faculty and administrators had jaundiced pre-forum expectations about professors' attitudes toward students that did not match what they heard at the forum.The deliberation generated the shift and washed out any moderator effect, at least on this question.Such an explanation does not negate earlier support for the moderator effect.The case is still strong that NIF moderators have less effect on higher status participants.
The hypothesized moderator effect receives further support when NIFmoderated respondents are compared, by status and group type, to non-NIF moderated respondents.Table 7 reports the relevant differences of means.Turning first to the student-only groups, NIF moderated students reported large and statistically significant shifts on the three comfort questions, while non-NIF participants did not.The only significant shift occurring with respondents in student-only non-NIF moderated groups was on the administrators care question, where the mean score actually dropped.In other words, the absence of higher status participants in student-only groups with non-NIF moderators exacerbated the belief that administrators did not care about student views on academic excellence.Contrast this with results for students in hierarchical groups, where both NIF and non-NIF moderated sessions fostered greater epistemological authority.This is evidenced by statistically significant, positive results on the three comfort questions.Unlike the student-only groups, however, students in the hierarchical groups agreed more that administrators cared about student views.This shift likely reflects the effect of deliberation per se rather than any moderator effect.It also should be noted that students in hierarchical groups shifted more in their comfort level in NIF groups (two of the three comfort questions) when a NIF moderator was present than a non-NIF moderator.This further underscores the influence of NIF moderators on epistemological authority.
Results for the moderator effect on faculty and administrators offer a mixed bag.Higher status participants in NIF groups lodged statistically significant shifts on two of the comfort questions (comfort-excellence and comfort-culture), while only comfort-culture was significant and positive in non-NIF moderated groups.Both NIF and non-NIF groups also showed significant and positive results on the professors-care question, although the shift was larger for non-NIF groups, as it was on the comfort-culture question.These could be byproducts of NIF moderators doing their jobs well.By minimizing the effects of hierarchy-including as many from the group and ensuring higher status participants did not dominate the conversation-faculty and administrators in NIF-moderated groups may have felt less comfortable than their unchecked colleagues in non-NIF-moderated groups.

CONCLUSIONS
In Why Deliberative Democracy?, Amy Gutmann and Dennis Thompson assert, "The future of deliberative democracy … depends on whether its proponents can create and maintain practices and institutions that enable deliberation to work well " (2004: 59).This research recognizes a potential hurdle for deliberative processes working well in hierarchically ordered organizations: inequalities of epistemological authority based on status differences.These inequalities may create an unlevel playing field, a condition deliberative democratic theorists find detrimental to effective deliberation.By identifying this potential hurdle and then studying just how high of a barrier it is, in fact, this research tells something about the extent to which deliberative processes can or cannot operate within hierarchically ordered organizations.This is a significant question because many organizations, groups, and communities are not defined by egalitarianism and their members would not necessarily enter a deliberation thinking of others as coequals.It is important, therefore, to understand better the dynamic between deliberation and hierarchy.This is by no means a definitive study.As a single case study its conclusions are heavily contextualized, but it offers some intriguing findings that prompt, we hope, further inquiries using other hierarchically ordered organizations and testing the efficacy of other deliberative rules against the NIF model.To the extent that epistemological equality is a condition for effective deliberation, further empirical research is warranted.
The project first asked whether individuals within a hierarchically ordered organization ascribe to themselves higher or lower levels of a priori epistemological authority based on their status within the organization.Strong support was found for this claim when comparing students, faculty, and administrators.
The project next explored whether the deliberative process ameliorated inequalities of epistemological authority caused from status difference.Exploring this issue tells something about limits-where deliberative democracy may or may not work.Knowing an empirical boundary of this sort is an important finding for those promoting the practice and for theorists writing about it.It contributes to our understanding of why deliberative forums work well in certain settings and less so in others, and it may also prompt further investigations into mitigating status differences or forging new deliberative formats apropos hierarchical organizations.
The data indicate that the deliberative forum employed in this project handled the status differences present in the university.It demonstrated that merely bringing individuals together in a deliberative setting fostered greater epistemological authority.
This proved true for students, faculty, and administrators alike, although students tended to shift more than faculty and administrators.That is, the deliberation had its greatest effect on the students, empowering them to feel more comfortable talking with higher status participants.These findings suggest that proponents of deliberative democratic forums may be bold in their claims, particularly as to the contexts in which such forums may be efficacious.Having demonstrated that it handled status differences present in a university, one might expect it could be employed in other hierarchically ordered organizations.Such an expectation demands further research.
This study also considered whether the rules for deliberation shaped its quality and content.In particular, it put the National Issue Forum's moderating style to the test.Before embarking on research that tests more exotic rule sets, the efficacy of the NIF format-now twenty-five years old-is a logical and necessary place to start.Do NIF moderators encourage a more level playing field, a condition the deliberative democratic theorists find so critical?In this experiment, yes.NIF moderators encouraged broader participation in small group discussions and were more successful in having participants view their deliberations as ones where everyone's views were not just valued, but equally valued.Participants who deliberated with a NIF moderator also recorded being more comfortable in their deliberations.This project affirms the utility of NIF moderators in hierarchical deliberations.Having said that, the specter of using more intrusive rules remains.It could be the case that a different set of moderator rules-one that more explicitly addresses hierarchy and status difference-such as dictating speaking order or rationing the frequency of participation based on status, may be more efficacious.That question remains unanswered.At the very least, this study provides an important baseline for evaluating other rule sets.Finally, this study demonstrated that the NIF moderator effect varied in some important ways by status.Students responded the most to having a NIF moderator present, particularly in terms of their comfort level.Although variation in status did not account for results on some specific questions, the aggregate picture supported the hypothesized relationship.
If Gutmann and Thompson's admonition is correct-that effective processes are necessary for deliberative democracy to thrive-this research affirms the utility of small group deliberative settings and NIF moderators when the deliberation takes place within hierarchically ordered organizations.Given the prevalence of hierarchy and status differences in contemporary society and in the many settings individuals organize themselves, affirming the efficacy of deliberation and NIF moderating is an important step.
Appendix A: Further Details on Research Design This project tested the effect of two variables on epistemological equality in hierarchically ordered organizations: the act of deliberating itself and moderator type.Small groups reflected one of five different hierarchical relationships and one of two moderator types.Organizing the groups in this manner captured the various hierarchical relationships within the university, and with the pre-and post-forum surveys, it also generated pooled, cross-sectional survey data that address the two independent variables.Participants and moderators were assigned to groups randomly.
Several efforts were taken in structuring the deliberation to ensure that participants recognized any hierarchy in the group, should it be present and should it matter to them.First, participants wore differently colored name badges that reflected their categorical membership.Students wore red badges.Faculty wore blue badges, while administrators and staff had gold name tags.Second, and most important, all moderators started their discussions by having group members identify themselves and their positions at the university.Participants were not informed of their moderator type.Every effort was made to have four groups for each composition-moderator pairing, but changes in actual attendance on the day of the forum produced some variation.For this reason, we combined survey data from the student-faculty, student-administrator, and student-facultyadministrator groups when conducting the analyses.Over 500 people attended the forum.Not everyone stayed for the entire forum or completed the pre-and post-forum surveys as instructed, resulting in some slippage between the number of attendees and survey participants.The table below reports the number of attendees who completed surveys.These numbers illustrate the research design's emphasis on constructing small groups composed primarily of students and then including a small number of faculty, administrators, or staff to create hierarchically ordered groups.

Table 1 :
Pre-Forum Indicators of Epistemological Authority By StatusPre-Q6 Comfort-Excellence: I am comfortable talking with other members of the university community about academic excellence.

Survey Please tell us how much you agree or disagree with the following statements…
http://www.publicdeliberation.net/jpd/vol4/iss1/art14

tell us how much you agree or disagree with the following statements …
Rank order (1, 2, 3) the approaches, in terms of which would best promote academic excellence.