Emerging Trends in The Social and Behavioral Sciences

Deliberative Democracy

Media

Part of Deliberative Democracy

Title

Deliberative Democracy

extracted text

Deliberative Democracy
JAMES FISHKIN

Abstract
At the center of the idea, deliberative democracy is that the “will of the people”
should be based on the consideration of competing arguments about their merits of
each policy choice. Despite this, partisanship and electoral campaigns are designed
to win elections, not promote citizen deliberation. The focus of most modern research
on deliberation by ordinary citizens rather than by representatives or office holders
raises interesting questions about who deliberates, what is the policy context, what
rationale is used, and what criteria should evaluate deliberative processes. Access to
accurate information, relevant arguments, and a representative sample are necessary
for good conditions. Some criteria for evaluation include demographic and attitudinal representativeness, sample size, the opportunity to engage in arguments for and
against proposals, and the elimination of distortion (e.g., misleading information).

The core idea of deliberative democracy is the notion that the “will of the
people” should be based on the consideration of competing arguments considered on their merits about the choice of policy or, as Jurgen Habermas
famously identified the core idea, “the forceless force of the better argument”
(Habermas, 1996). In the past two decades, it has produced a large literature in political theory (for some summary collections, see Bohman & Rehg,
1997; Elster, 1998; Fishkin & Laslett, 2003). This notion of reason-based choice
under good conditions is deceptively simple and will be regarded by many
as little more than common sense. Yet if taken seriously, it challenges the most
widely applied approaches to democracy. Further, it poses problems of institutional design that require further empirical study and experimentation.
This emphasis on deliberation is in tension with the other main approaches
to democracy such as party competition-based models (Schumpeter, 1942) or
mass participatory democracy (bringing decisions about issues or candidate
selection directly to the public in initiatives, referendums, or primaries;
see, e.g., Cronin, 1999.) The tension arises from the fact that neither party
competition nor electoral campaigns for ballot propositions are designed
to promote citizen deliberation. Rather they are designed to win elections
Emerging Trends in the Social and Behavioral Sciences.
Robert Scott and Marlis Buchmann (General Editors) with Stephen Kosslyn (Consulting Editor).
© 2016 John Wiley & Sons, Inc. ISBN 978-1-118-90077-2.

1

2

EMERGING TRENDS IN THE SOCIAL AND BEHAVIORAL SCIENCES

and if elections can be won by misleading or distorting relevant facts or
arguments, that is not a concern. Such strategies are only to be expected
in campaigns. Joseph Schumpeter, the most influential proponent of party
competition-based democracy, was explicit that the same techniques of
advertising and public relations employed in the market should be expected
in the political process. The result, he argued, is a “manufactured” political
will. The concern for some sort of “genuine” political will is a leftover from
“classical” theories of democracy that Schumpeter treats as irrelevant and
hopelessly ill-defined (Schumpeter, 1942). For mass participatory democracy, the tension with deliberation stems partly from the same problem
of competitive political campaigns (whether for ballot propositions or for
candidates in primaries) having similar motivations to those in political
parties. In addition, the citizen in mass society has little incentive to become
informed in large-scale elections, whether ballot propositions or primaries,
or general elections between competing parties. There is plausibility to
the supposition of Anthony Downs that they are “rationally ignorant.” If I
have one vote in millions my individual effect on an election is tiny so my
individual incentive to become informed to cast a more effective vote in that
election is also equally small (Downs, 1957).
The term deliberative democracy may well have been coined in an interpretation of the American Founders’ account of the US congress, especially the
role of the Senate (Bessette, 1994). However, the views of James Madison
and Alexander Hamilton were an idealization of individual representatives deliberating without any concern for political parties. Modern party
competition-based democracies bring party discipline to voting in legislatures, limiting the range of choice for legislators to engage in deliberations on
the merits. Clearly, Madison envisaged deliberation as central to the role of
the Senate. Representatives were to “refine and enlarge the public views” as
Madison said in Federalist 10. However, with modern party discipline and
electoral incentives, the opportunities for deliberation by representatives are
limited by the action of party “whips.”
Most modern research on deliberative democracy has focused on deliberation by ordinary citizens rather than by representatives or office holders
(for some exceptions, see Steiner, Bächtiger, Spörndli, & Steenbergen, 2005;
Uhr, 1998). The ordinary citizens are usually recruited to participate in a
mini-public or microcosm of the population. Such efforts suggest several
questions:
1. Who deliberates?
2. What is the policy context?
3. What is the rationale for others (policy makers or the rest of the public)
to pay attention to their conclusions?
4. By what criteria should the success of such processes be evaluated?

Deliberative Democracy

3

WHO DELIBERATES?
Given low levels of information and public inattention on the part of the
mass public on most issues most of the time (Delli Carpini & Keeter, 1997),
it is hard to attribute much active deliberation to the mass public in its ordinary environment in the normal course of events. There may be dramatic
periods of mass public engagement, what Bruce Ackerman calls “constitutional moments” when there was an enormous widespread and sustained
public debate, but these episodes are rare. He identifies three, perhaps four
such “moments” in American history—the Founding, Reconstruction, the
New Deal (Ackerman, 1991), and perhaps the Civil Rights period (Ackerman,
2014). Even for such cases, we must distinguish mobilization from deliberation. Many people can get engaged to vote or attend demonstrations but are
they considering competing arguments? Are they really thinking about the
issues? Data is sparse, especially for the earlier periods of the constitutional
moments.
One might envision an institutional design that would encourage everyone to deliberate, perhaps before a national election. One example is
the “Deliberation Day” proposal (Ackerman & Fishkin, 2004). On this
approach, all voters are invited to sites around the country for a day of
deliberation—alternating small group discussions and plenary sessions
with competing experts, roughly on the model of a Deliberative Poll, but one
which is scaled up to include the total population. An incentive is paid for
participation for the day’s work of citizenship. Ideally, it would be held on
a national holiday to save expense (perhaps supplanting President’s Day).
If the method of the Poll works with microcosms, random samples of the
population, then the same process could work with the entire population.
A deliberation with the whole population would be far more consequential
because actual public opinion in the society would likely have changed.
In the absence of deliberation by the whole population, there are two main
approaches for a deliberative subgroup. It could be self-selected or it could
be arrived at via some form of scientific sampling. Participatory Budgeting,
begun in Porto Alegre, Brazil, uses self-selection for the local population to
deliberate on the choice of projects for the city. This process has spread to
many countries. It is sometimes lauded as deliberative democracy (Fung); it
is also often viewed as mobilization for specific benefits more than deliberation. In any case, any process based on self-selection is likely to be unrepresentative. Only those who are especially interested volunteer themselves
(World Bank, 2007).
From the standpoint of deliberative democracy, the principal approach over
the past two decades has been to gather a representative microcosm and
then to engage it in good conditions for considering the issue. The basic idea

4

EMERGING TRENDS IN THE SOCIAL AND BEHAVIORAL SCIENCES

goes back to Ancient Athens. Several key institutions used random sampling
implemented with a machine (the Kleroterion) that selected citizens from a
list of those who had previously expressed willingness to engage in public
service. The Council of 500 selected in this way determined the agenda for
what could be voted on in the Assembly, the graphe paranomon was a randomly selected jury of 500 that could prosecute irresponsible participation
in the Assembly and the nomethetai was a final stage of deliberative decision.
After a proposal passed the assembly, it did not immediately become law. It
could only do so if a random sample of 500 or more was convened to hear
the case for and against the proposed law and then after a day of deliberation voted positively on the proposal. Hence, the deliberating microcosms
had a key role in Athenian democracy before, during, and after the direct
democracy of the Assembly.
These ancient institutions exemplify a practical method for implementing
a very different kind of democracy. It is distinct from direct participatory
democracy (as in the Assembly or as in modern referenda), distinct from
party competition democracy implemented through elections and distinct
from the kind of deliberation by representatives envisioned by the American founders (deliberation by representatives on the merits without regard
to party interests or party discipline). A challenge for modern research is
whether some of the positive attributes of this kind of democratic design for
deliberation by a random sample can be adapted to modern conditions.
In modern times, deliberating microcosms chosen by lot have been used
to make recommendations on policy in many countries around the world
(Gronlund, Bachtiger, & Setala, 2014). A particular design, Deliberative
Polling (cdd.stanford.edu), has been applied in 23 countries. In addition, deliberating microcosms such as the Citizens’ Assemblies in British
Columbia and Ontario have directly put propositions on the ballot (Fournier,
van der Kolk, Kenneth Carty, Blais, & Rose, 2011), “citizens’ juries” have been
employed on many issues, especially health care in the United Kingdom.
The list is extensive, but the particular designs vary in important particulars.
For now, it is worth noting the use of random sampling for a deliberating
microcosm as the predominant approach to implementing “deliberative
democracy” by the people themselves, rather than by their representatives.
WHAT IS THE POLICY CONTEXT?
When a microcosm of the public is convened, should its decisions be
binding or advisory to other decisions? While some theorists have argued
that in order to count as deliberation, the conclusions ought to be “binding”
(Gutmann & Thompson, 1996, 1998), there have in fact been few if any such
cases if binding is interpreted strictly as requiring the decision to conform

Deliberative Democracy

5

to the conclusions of the citizens. Deliberative microcosms do not produce
binding decisions in the way that say a referendum can (although of course
some referenda are also only advisory). The definition of binding, when
clarified comes down to the intent of the deliberators. As Gutmann and
Thompson describe their proposed requirement, participants “intend their
discussion to influence a decision the government will make or a process
that will affect how future decisions are made” (Gutmann & Thompson,
1998). On this interpretation, many deliberating forums are binding.
The deliberating microcosm typically will offer its conclusions as an input
to other decision-making. For example, between 1996 and 1998, there were
eight Deliberative Polls convened in Texas by the electric utilities and the
Public Utility Commission to provide input to the creation of “Integrated
Resource Plans” that would chart how that portion of the state would
satisfy future energy needs. Deliberators increased their level of support
for renewable energy such as wind power, to the point that much higher
percentages of the respondents were willing to pay increased utility bills
to support renewable energy. The Public Utility Commission reacted to
the data and the eight events by supporting increases in wind power and,
in several service territories, slightly increased fees to support the new
energy. Texas dramatically increased its wind power. These consultations
were part of a process by which Texas moved from having the least wind
power among the 50 states to having the most, surpassing California in 2007
(Fishkin, 2009). The context that made this successful was the requirement,
implemented by the Public Utility Commission, that the utilities consult
the public in some way. The Deliberative Poll was adopted for public
consultation in all eight service territories of the regulated utilities because
it offered a method to consult that was representative and evidence based.
Large investments in different forms of energy were at stake and an unrepresentative process or one dominated by bad information could lead to costly
and or bizarre outcomes. Other examples of advisory but near-binding
decisions can be found in the Deliberative Polls in China and Macau. When
the government convenes a Deliberative Poll to consult its citizens about
a policy issue, say which infrastructure projects to build (as in several
projects in China sponsored by local government, see Fishkin, He, Luskin,
& Siu, 2010), or the design of a press council (in Macau see report http: cdd.
stanford.edu/2011/final-report-deliberative-polling-on-the-amendment-ofthe-press-law-and-the-audio-visual-broadcasting-act-in-macao/), there is a
strong impetus to implement the result. After all, there is official sponsorship
and government resources are being spent. It would seem odd to go to all
the trouble of convening a random sample for intensive deliberation and
then to ignore the result.

6

EMERGING TRENDS IN THE SOCIAL AND BEHAVIORAL SCIENCES

Another route to impact is to set an agenda for decisions by the voters. The
two Citizens’ Assemblies in British Columbia and Ontario proposed ballot
measures for electoral reform that went directly on the ballot. While neither
passed, they broke new ground in that a random sample of citizens deliberated for about a year and produced a proposal that the rest of the province
voted on (Fournier et al., 2011). In a less explicit way, the 2011 Deliberative
Poll in California (“What’s Next California”) produced recommendations
both for what could go on the ballot and for decisions by the legislature.
Proposition 31 included six elements from the citizen deliberations, but
other elements as well. The proposition was defeated, but recommendations
for legislation about initiative reform did pass (Fishkin, Kousser, Luskin, &
Siu, 2015).
Setting the agenda for voting by the entire electorate harks back to the role
of the council of 500 in Ancient Athens. It is a very strong form of influence without final decision-making power. A lesser form of influence could
occur through the deliberations producing a voter recommendation. The Citizens’ Initiative Review convenes small random samples to deliberate on the
model of a citizens’ jury (Coote & Lenaghan, 1997) and make a recommendation about ballot propositions. Those recommendations go in the voter
handbook. Deliberative Polls, which employ larger samples, were broadcast
before national referendums in Australia (on the Republic in 1999) and in
Denmark (on the Euro in 2000). Of course, a televised process of voter discussion with results of the citizen deliberations would have to compete with
all the campaign sound bites in a referendum campaign to have influence.
However, the idea is that such deliberating microcosms express the informed
views of ordinary citizens rather than just partisan campaign messages.
WHY SHOULD OTHERS PAY ATTENTION?
The strategy of trying to implement deliberative democracy, at least on
selected issues, through randomly chosen microcosms has the advantage
that it offers a practical method to give voice to the public’s considered
judgments. In that respect, it offers a stark contrast to standard public
opinion polls that, of course, also use random sampling. Because most
voters in their ordinary environments do not have much occasion or
incentive to deliberate, the opinions represented in standard opinion polls
often represent little more than the public’s impressions of sound bites and
headlines. However, when such poll results are reported, policy makers
have, nevertheless, a reason to pay attention. If done well, those polls offer
a snapshot of the actual state of public opinion at the time. This is what
the people are actually thinking, even if they are not thinking a great deal
about the issue in question. There are exceptions of course. The polls might

Deliberative Democracy

7

represent only nonattitudes or phantom opinions as respondents who have
not thought about an issue attempt to offer an answer on the spot to avoid
admitting that they “don’t know” (Bishop, 2005; Converse, 1964). However,
setting aside this complication, standard polls offer policy makers a picture
of the current state of public opinion and if they wish to be responsive to the
public, such information is useful and relevant.
However, the opinions offered by a deliberating microcosm are different
in a key respect. They are not shared by those who have not deliberated.
They offer a counterfactual representation of what the people would think,
presumably under good conditions for thinking about it. Why should policy
makers or other citizens (voters) respond to counterfactual opinion?
The issue is ultimately normative. As we noted earlier, democracy presumes some connection between the public will and what is done. If actual
conditions do not really permit the public to come to an informed judgment,
why not find out what it would think under transparently good conditions,
under conditions where it can really consider competing arguments and get
its questions answered from different points of view. If the reasons for its
considered judgments can then be identified, those reasons should resonate
with other citizens.
On the many issues for which the public is inattentive, why rely on conventional polls rather than a representative and informed judgment from
the deliberations of a good sample? Of course, one might take the position
that public opinion, whether deliberative or not, is irrelevant. Policy making
should be a matter of expert judgment. However, it is the people who have
to live with the consequences of one policy choice or another. Surely, there
is room to take their views into account. Moreover, if one takes that step,
why not take their informed and representative views. Hence the idea of the
deliberating mini-public or microcosm.
The Texas utility cases mentioned earlier provide an illustration. The public
was not very knowledgeable before deliberation about the attributes of various energy choices. In fact, the original discovery of nonattitudes or phantom
opinions had come from a question about electric utilities in a panel study
from the American National Election Study (Converse, 1964). Thus, on some
questions about electric utilities, the public could be expected not to have any
real opinions at all.
In Texas, the low salience and information levels on these issues helped
make the case for Deliberative Polling. The deliberating microcosm offered
an alternative that was evidence based, thoughtful, and representative. To
rely on conventional polling alone would have rendered very big expenditures vulnerable to conclusions from people who had little understanding,
predeliberation, of the trade-offs at stake. Yet to leave the people out of the
decision would, in effect, have substituted values of the experts for the values

8

EMERGING TRENDS IN THE SOCIAL AND BEHAVIORAL SCIENCES

of the public. Hence the case for consulting those affected but in a representative and thoughtful way on the basis of the best information available.
BY WHAT CRITERIA SHOULD SUCH PROCESSES BE EVALUATED?
The claim of the deliberating microcosm for the attention of policy makers
and fellow citizens rests upon a hypothetical inference—these are the conclusions the population would come to if they could somehow consider the
issue in depth under the same good conditions. Hence, the conditions must
be credible as good conditions (e.g., access to good information and relevant
arguments) and the sample must be representative. Consider some criteria
for evaluation building on one or the other of these two basic points:

•
•
•
•

•

Demographic representativeness
Attitudinal representativeness
Sample size
Opportunity to engage policy arguments for and against proposals for
action
Avoiding distortions

DEMOGRAPHIC REPRESENTATIVENESS
To support the hypothetical inference that the population as a whole would
likely come to similar conclusions if it were to deliberate under comparably good conditions, the microcosm needs to be representative at the start. It
needs to be a version of the whole country or the whole city (or whatever the
relevant population may be) in miniature, usually in one place (although for
virtual versions distributed in space but connected electronically, see Fishkin,
2009). But representative in what ways? Demographics in all the standard
categories, class, gender, education, income, and ethnicity are usually among
the most prominent. Some approaches to creating a deliberative microcosm
view such representativeness in demographics as sufficient (Carson, Gastil,
Hartz-Karp, & Lubensky, 2013). Others require attitudinal representativeness as well. An argument for demographic representativeness is that people
from different demographic backgrounds will likely have different views and
interests. Furthermore, as they deliberate they may realize more about their
views and interests. If significant portions of the population were excluded
(say very few women or very few minorities) one could plausibly suspect
that the deliberation might have turned out differently if the microcosm had
been more representative.

Deliberative Democracy

9

ATTITUDINAL REPRESENTATIVENESS
Attitudinal representativeness seems equally important. If the sample is a
representative microcosm of the viewpoints in the population, then it is more
plausible to argue that as participants weigh the arguments from these competing perspectives, their views would represent the hypothetical—the conclusions the population would come to, if they could weigh the arguments
with comparable engagement, seriousness, and high-quality information. If
the microcosm begins with different viewpoints than the population, then the
hypothetical inference loses its plausibility. J.S. Mill described the importance
of attitudinal representativeness in famous comments on the idea of Parliament acting as a “Congress of Opinions:” “where every person in the country
may count upon finding somebody who speaks his mind, as well or better
than he could speak it himself … where those whose opinion is overruled,
feel satisfied that it is heard, and set aside not by a mere act of will, but for
what are thought superior reasons” Mill, 1861). Mill’s account of the deliberative ideal probably applies to Parliaments only rarely, but it does offer a
clear picture of the advantages of ensuring that a deliberative microcosm is
representative in attitudes.
For samples of ordinary citizens, the recruitment of the microcosm should
provide a basis for empirical assessment of how attitudes of participants
compare to the attitudes of the rest of the public. One simple approach is
to ask the questionnaire of potential participants before recruitment allowing comparison of participants and nonparticipants in attitudes as well as
demographics. In addition, having another comparison group who are never
asked to deliberate is useful and can also be used for matching in case there
is any sampling bias. Deliberative Polling typically employs these methods
but most attempts at recruiting deliberative microcosms do not (for a compendium, see Gastil & Levine, 2005).
SAMPLE SIZE
The microcosm needs to be large enough that its representativeness and the
statistical significance of any opinion changes can be meaningfully evaluated. If it is not representative, then why should others pay attention? Its
conclusions might have been determined by strong or idiosyncratic views
at the start. If the changes are not statistically significant, then they might
be mere random noise. We do not know how seriously to take them. Some
microcosms, on the model of a modern jury face both of these difficulties. The
Citizens’ Initiative Review in Oregon has the noble aspiration of engaging a
microcosm of the state’s voters and offering recommendations about how
the electorate as a whole should vote. The recommendations are printed in
the voter handbook. However, the sample size is so small that some of the

10

EMERGING TRENDS IN THE SOCIAL AND BEHAVIORAL SCIENCES

recommendations are simply wiped out by sampling error. For example, the
recommendation for Oregon Proposition 74 on medical marijuana in 2010
was 13 to 11 (Fishkin, 2013).
ARGUMENTS FOR AND AGAINST
The root notion of deliberation is “weighing,” the weighing of competing
arguments or reasons for one course of action or another. Therefore, any
deliberative microcosm needs an effective method of engaging the participants in these competing arguments. One method is to have a stakeholder
advisory group prepare briefing materials that attempt to anticipate the arguments on either side that are current in the policy community of experts.
Ordinary citizens are unlikely to be aware of all of these. However, experts
will usually not be able to anticipate all the concerns of the public. Hence
the public will likely add further arguments, either pro or con. If the public
in the deliberating microcosm has an opportunity to get its factual questions
answered by experts and policy makers, it further facilitates their coming to
an informed judgment. The Deliberative Poll design uses all three of these
methods: the balanced briefing materials, the small group discussions where
people share their own reasons for support or opposition to the various proposals, and access to competing experts who can answer questions from the
small group discussions.
How can we know if participants have actually considered competing arguments? Both quantitative and qualitative data can be gathered, the former
from before and after questionnaires and the latter from transcripts of the
group discussions. These can be coded for balance and argumentation creating new variables that can be incorporated into the quantitative analyses
(Gerber, Bächtiger, Fiket, Steenbergen, & Steiner, 2014; Siu, 2009). This is an
important emerging trend in the study of deliberative democracy.
AVOIDING DISTORTIONS
There are two principal ways in which it has been alleged that deliberations
may go wrong. First there is a long tradition of discussion, mostly in the
jury literature, that deliberations will be dominated by the men, by the more
educated, by those who are more advantaged. They will have preponderant
influence, importing the socioeconomic inequalities in the wider world into
the deliberating microcosm. Part of this is deference to the more advantaged,
part of it is the very skills required for deliberation are thought to be disproportionately distributed to the better advantaged. The more advantaged are
better able to make arguments and because of their social positions, they are
more likely to be listened to (Sanders, 1997; Young, 2000).

Deliberative Democracy

11

Yet the actual empirical study of this problem has not confirmed these speculations. At least with Deliberative Polling, if one looks at the initial positions
of the more advantaged (the more educated, the males, the rich, etc.) on policy indices on the issues deliberated about, there is no pattern of movement
by the small groups in the direction of those positions. The more advantaged
are not dominating the discussions by imposing their views (Luskin, Sood,
Fishkin, & Hahn, 2015; Siu, 2009). This is an area for further study, but it
seems likely that the design of the deliberative process, the role of moderators, balanced materials, and ample opportunities for everyone to discuss,
all play an important role. Further, juries are likely to employ social pressure
to reach a unanimous verdict. In the Deliberative Poll there is by design no
pressure for consensus. Rather the participants’ final considered judgments
are recorded in confidential questionnaires.
A second major distortion is the pattern Cass Sunstein has identified as the
“law of group polarization” (Sunstein, 2002, 2009). If there is an issue for
which there is a midpoint, if the initial position of the group is one side of
the midpoint, the hypothesis is that it will move further away from the midpoint to that side, in effect becoming more extreme. Two basic mechanisms
are offered—imbalance in the argument pool and a social comparison effect.
If most people are on one side of an issue, they are likely to offer more reasons for that side of the argument. As they see others moving in that direction
they will feel pressures to conform. Sunstein and various collaborators have
corroborated this pattern in some jury-like studies (for an overview, see Sunstein, 2009). However, in the Deliberative Polling design, the pattern has not
been found. In an aggregation of the small group patterns in 22 Deliberative Polls, the tendency to move away from the midpoint was <50% (Luskin
et al., 2015; see also Fishkin et al., 2010; Siu, 2009). If the deliberative design
has sufficient balance (briefing materials, moderation, and balanced panels
of experts), it can overcome any imbalance in the argument pool provided
by participants. Moreover, if there are confidential questionnaires, the design
can minimize the social comparison effects.
Why are these two distortions important? Either domination by the
more advantaged or polarization (in Sunstein’s sense) regularly moving
the participants to more extreme positions could plausibly bring into
question whether the conclusions were the result of genuine deliberation
on the merits. It cannot be the Habermasian “unforced force of the better
argument” (Habermas, 1996) if people are trapped in a pattern of group
psychology or implicitly coerced by pressures of group conformity. The
fact that the prevalence of these patterns seems to depend on the precise
design of the deliberative process opens up an area for further empirical
work. The study of multiple projects, aggregating the changes in the small
group discussions to investigate whether or not these distorting patterns

12

EMERGING TRENDS IN THE SOCIAL AND BEHAVIORAL SCIENCES

hold with various designs is an important emerging trend in the study of
deliberative democracy. The whole empirical research program connected
with deliberative democracy is designed to make it practical. The empirical
study allows for an assessment of the degree to which a theoretical concept
can actually be implemented in real democratic practices.
REFERENCES
Ackerman, B. (1991). We the people: Foundations: We the people (Vol 1). Cambridge, MA:
Harvard University Press.
Ackerman, B. (2014). The civil rights revolution: We the people (Vol 3). Cambridge, MA:
Harvard University Press.
Ackerman, B., & Fishkin, J. (2004). Deliberation day. New Haven, CT: Yale University
Press.
Bessette, J. (1994). The mild voice of reason. Chicago, IL: University of Chicago Press.
Bishop, G. (2005). The illusion of public opinion. Lanham, MD: Rowman and Littlefield.
Bohman, J., & Rehg, W. (Eds.) (1997). Deliberative democracy: Essays on reason and politics. Cambridge, MA: MIT Press.
Carson, L., Gastil, J., Hartz-Karp, J., & Lubensky, R. (Eds.) (2013). The Australian citizens’ parliament and the future of deliberative democracy. State College, PA: Penn State
University Press.
Converse, P. E. (1964). The nature of belief systems in mass publics. In D. Apter (Ed.),
Ideology and its discontents. New York, NY: The Free Press of Glencoe.
Coote, A., & Lenaghan, J. (1997). Citizens’ juries: Theory into practice. London, England:
Institute for Public Policy Research.
Cronin, T. (1999). Direct democracy: The politics of initiative, referendum and recall. Cambridge, MA: Harvard University Press.
Delli Carpini, M., & Keeter, S. (1997). What Americans know about politics and why it
matters. New Haven, CT: Yale University Press.
Downs, A. (1957). An Economic Theory of Democracy. New York: Harper and Row.
Elster, J. (Ed.) (1998). Deliberative democracy. Cambridge, MA: Cambridge University
Press.
Fishkin, J. S. (2009). When the people speak: Deliberative democracy and public consultation.
Oxford, England: Oxford University Press.
Fishkin, J. S. (2013). Deliberation by the people themselves: Entry points for the public voice. Election Law Journal, 12(4), 490–507.
Fishkin, J. S., He, B., Luskin, R. C., & Siu, A. (2010). Deliberative democracy in an
unlikely place: Deliberative polling in China. British Journal of Political Science,
40(2), 435–448.
Fishkin, J. S., Kousser, T., Luskin, R. C., & Siu, A. (2015). Deliberative agenda setting:
piloting reform of direct democracy in California. Perspectives on Politics, 13(4),
1030–1042.
Fishkin, J., & Laslett, P. (Eds.) (2003). Debating deliberative democracy. Oxford, England:
Blackwell.

Deliberative Democracy

13

Fournier, P., van der Kolk, H., Kenneth Carty, R., Blais, A., & Rose, J. (2011). When
citizens decide: Lessons from citizen assemblies on electoral reform. Oxford, England:
Oxford University Press.
Gastil, J., & Levine, P. (Eds.) (2005). The deliberative democracy handbook. San Francisco,
CA: Jossey-Bass.
Gerber, M., Bächtiger, A., Fiket, I., Steenbergen, M. R., & Steiner, J. (2014). Deliberative and non-deliberative persuasion: Opinion change in a pan-European deliberative poll (Europolis). European Union Politics, 15, 410–429.
Gronlund, K., Bachtiger, A., & Setala, M. (Eds.) (2014). Deliberative mini-publics:
involving citizens in the democratic process. Colchester, England: ECPR Press.
Gutmann, A., & Thompson, D. (1996). Democracy and disagreement. Cambridge, MA:
Harvard University Press.
Gutmann, A., & Thompson, D. (1998). Why deliberative democracy? Princeton, NJ:
Princeton University Press.
Habermas, J. (1996). Between facts and norms: Contributions to a discourse theory of law
and democracy. Cambridge, MA: MIT Press.
Luskin, R. C., Sood, G., Fishkin, J., & Hahn, K. (2015). Deliberative distortions? Homogenization, polarization, and domination. Working paper, Center for Deliberative
Democracy, Stanford University
Mill, J. S. (1861). Considerations on representative government. London, England: Parker,
Son and Bourn.
Sanders, L. M. (1997). Against deliberation. Political Theory, 25, 347–376.
Schumpeter, J. A. (1942). Capitalism, socialism and democracy. New York, NY: Harper
and Row.
Siu, A. 2009. Look who’s talking: Examining social influence, opinion change and argument
quality in deliberation (PhD dissertation), Stanford University.
Steiner, J., Bächtiger, A., Spörndli, M., & Steenbergen, M. R. (2005). Deliberative politics
in action: Analyzing parliamentary discourse. Cambridge, MA: Cambridge University Press.
Sunstein, C. R. (2002). The law of group polarization. The Journal of Political Philosophy,
10(2), 175–195.
Sunstein, C. R. (2009). Going to extremes: How like minds unite and divide. Oxford,
England: Oxford University Press.
Uhr, J. (1998). Deliberative democracy in Australia. Cambridge, MA: Cambridge University Press.
World Bank. 2007. Enhancing the social and economic sustainability of participatory budgeting in Porto Alegre, Brazil. Final report.
Young, I. M. (2000). Inclusion and democracy. Oxford, England: Oxford University
Press.

JAMES S. FISHKIN SHORT BIOGRAPHY
James S. Fishkin holds the Janet M. Peck Chair in International Communication at Stanford University where he is a professor of communication,

14

EMERGING TRENDS IN THE SOCIAL AND BEHAVIORAL SCIENCES

professor of political science (by courtesy), and the director of the Center for
Deliberative Democracy. His work focuses on deliberative democracy and
democratic theory in books such as When the People Speak (2009), Deliberation Day (2004 with Bruce Ackerman), and Democracy and Deliberation
(1991). He originated Deliberative Polling as a method of public consultation
in 1988. He began to apply it in collaboration with Robert C. Luskin in 1994
and has since spread it, various collaborators, to projects in 23 countries. For
more on Deliberative Polling see cdd.stanford.edu. He has been a Guggenheim Fellow, a Fellow of the Center for Advanced Study in the Behavioral
Sciences at Stanford, and a Fellow of the American Academy of Arts and
Sciences.
RELATED ESSAYS
Understanding American Political Conservatism (Political Science), Joel D.
Aberbach
Party Organizations’ Electioneering Arms Race (Political Science), John H.
Aldrich and Jeffrey D. Grynaviski
Economic Models of Voting (Political Science), Ian G. Anson and Timothy
Hellwig
The Underrepresentation of Women in Elective Office (Political Science),
Sarah F. Anzia
Public Opinion and International Conflict (Political Science), Adam J.
Berinsky
Neoliberalism (Sociology), Miguel Angel Centeno and Joseph N. Cohen
Political Advertising (Political Science), Erika Franklin Fowler
Government Formation and Cabinets (Political Science), Sona N. Golder
Presidential Power (Political Science), William G. Howell
Group Identity and Political Cohesion (Political Science), Leonie Huddy
Racial Disenfranchisement (Political Science), Vincent L. Hutchings and
Davin L. Phoenix
Genetic Foundations of Attitude Formation (Political Science), Christian
Kandler et al.
Women Running for Office (Political Science), Jennifer L. Lawless
Civic Engagement (Sociology), Peter Levine
From Individual Rationality to Socially Embedded Self-Regulation (Sociology), Siegwart Lindenberg
Political Inequality (Sociology), Jeff Manza
Participatory Governance (Political Science), Stephanie L. McNulty and Brian
Wampler
Money in Politics (Political Science), Jeffrey Milyo
Implicit Attitude Measures (Psychology), Gregory Mitchell and Philip E.
Tetlock

Deliberative Democracy

15

Feminists in Power (Sociology), Ann Orloff and Talia Schiff
Evolutionary Theory and Political Behavior (Political Science), Michael Bang
Petersen and Lene Aarøe
Electoral Authoritarianism (Political Science), Andreas Schedler
Does the 1 Person 1 Vote Principle Apply? (Political Science), Ian R. Turner
et al.
Information Politics in Dictatorships (Political Science), Jeremy L. Wallace
Constitutionalism (Political Science), Keith E. Whittington
Political Psychology and International Conflict (Sociology), Rose McDermott



Deliberative Democracy
JAMES FISHKIN

Abstract
At the center of the idea, deliberative democracy is that the “will of the people”
should be based on the consideration of competing arguments about their merits of
each policy choice. Despite this, partisanship and electoral campaigns are designed
to win elections, not promote citizen deliberation. The focus of most modern research
on deliberation by ordinary citizens rather than by representatives or office holders
raises interesting questions about who deliberates, what is the policy context, what
rationale is used, and what criteria should evaluate deliberative processes. Access to
accurate information, relevant arguments, and a representative sample are necessary
for good conditions. Some criteria for evaluation include demographic and attitudinal representativeness, sample size, the opportunity to engage in arguments for and
against proposals, and the elimination of distortion (e.g., misleading information).

The core idea of deliberative democracy is the notion that the “will of the
people” should be based on the consideration of competing arguments considered on their merits about the choice of policy or, as Jurgen Habermas
famously identified the core idea, “the forceless force of the better argument”
(Habermas, 1996). In the past two decades, it has produced a large literature in political theory (for some summary collections, see Bohman & Rehg,
1997; Elster, 1998; Fishkin & Laslett, 2003). This notion of reason-based choice
under good conditions is deceptively simple and will be regarded by many
as little more than common sense. Yet if taken seriously, it challenges the most
widely applied approaches to democracy. Further, it poses problems of institutional design that require further empirical study and experimentation.
This emphasis on deliberation is in tension with the other main approaches
to democracy such as party competition-based models (Schumpeter, 1942) or
mass participatory democracy (bringing decisions about issues or candidate
selection directly to the public in initiatives, referendums, or primaries;
see, e.g., Cronin, 1999.) The tension arises from the fact that neither party
competition nor electoral campaigns for ballot propositions are designed
to promote citizen deliberation. Rather they are designed to win elections
Emerging Trends in the Social and Behavioral Sciences.
Robert Scott and Marlis Buchmann (General Editors) with Stephen Kosslyn (Consulting Editor).
© 2016 John Wiley & Sons, Inc. ISBN 978-1-118-90077-2.

1

2

EMERGING TRENDS IN THE SOCIAL AND BEHAVIORAL SCIENCES

and if elections can be won by misleading or distorting relevant facts or
arguments, that is not a concern. Such strategies are only to be expected
in campaigns. Joseph Schumpeter, the most influential proponent of party
competition-based democracy, was explicit that the same techniques of
advertising and public relations employed in the market should be expected
in the political process. The result, he argued, is a “manufactured” political
will. The concern for some sort of “genuine” political will is a leftover from
“classical” theories of democracy that Schumpeter treats as irrelevant and
hopelessly ill-defined (Schumpeter, 1942). For mass participatory democracy, the tension with deliberation stems partly from the same problem
of competitive political campaigns (whether for ballot propositions or for
candidates in primaries) having similar motivations to those in political
parties. In addition, the citizen in mass society has little incentive to become
informed in large-scale elections, whether ballot propositions or primaries,
or general elections between competing parties. There is plausibility to
the supposition of Anthony Downs that they are “rationally ignorant.” If I
have one vote in millions my individual effect on an election is tiny so my
individual incentive to become informed to cast a more effective vote in that
election is also equally small (Downs, 1957).
The term deliberative democracy may well have been coined in an interpretation of the American Founders’ account of the US congress, especially the
role of the Senate (Bessette, 1994). However, the views of James Madison
and Alexander Hamilton were an idealization of individual representatives deliberating without any concern for political parties. Modern party
competition-based democracies bring party discipline to voting in legislatures, limiting the range of choice for legislators to engage in deliberations on
the merits. Clearly, Madison envisaged deliberation as central to the role of
the Senate. Representatives were to “refine and enlarge the public views” as
Madison said in Federalist 10. However, with modern party discipline and
electoral incentives, the opportunities for deliberation by representatives are
limited by the action of party “whips.”
Most modern research on deliberative democracy has focused on deliberation by ordinary citizens rather than by representatives or office holders
(for some exceptions, see Steiner, Bächtiger, Spörndli, & Steenbergen, 2005;
Uhr, 1998). The ordinary citizens are usually recruited to participate in a
mini-public or microcosm of the population. Such efforts suggest several
questions:
1. Who deliberates?
2. What is the policy context?
3. What is the rationale for others (policy makers or the rest of the public)
to pay attention to their conclusions?
4. By what criteria should the success of such processes be evaluated?

Deliberative Democracy

3

WHO DELIBERATES?
Given low levels of information and public inattention on the part of the
mass public on most issues most of the time (Delli Carpini & Keeter, 1997),
it is hard to attribute much active deliberation to the mass public in its ordinary environment in the normal course of events. There may be dramatic
periods of mass public engagement, what Bruce Ackerman calls “constitutional moments” when there was an enormous widespread and sustained
public debate, but these episodes are rare. He identifies three, perhaps four
such “moments” in American history—the Founding, Reconstruction, the
New Deal (Ackerman, 1991), and perhaps the Civil Rights period (Ackerman,
2014). Even for such cases, we must distinguish mobilization from deliberation. Many people can get engaged to vote or attend demonstrations but are
they considering competing arguments? Are they really thinking about the
issues? Data is sparse, especially for the earlier periods of the constitutional
moments.
One might envision an institutional design that would encourage everyone to deliberate, perhaps before a national election. One example is
the “Deliberation Day” proposal (Ackerman & Fishkin, 2004). On this
approach, all voters are invited to sites around the country for a day of
deliberation—alternating small group discussions and plenary sessions
with competing experts, roughly on the model of a Deliberative Poll, but one
which is scaled up to include the total population. An incentive is paid for
participation for the day’s work of citizenship. Ideally, it would be held on
a national holiday to save expense (perhaps supplanting President’s Day).
If the method of the Poll works with microcosms, random samples of the
population, then the same process could work with the entire population.
A deliberation with the whole population would be far more consequential
because actual public opinion in the society would likely have changed.
In the absence of deliberation by the whole population, there are two main
approaches for a deliberative subgroup. It could be self-selected or it could
be arrived at via some form of scientific sampling. Participatory Budgeting,
begun in Porto Alegre, Brazil, uses self-selection for the local population to
deliberate on the choice of projects for the city. This process has spread to
many countries. It is sometimes lauded as deliberative democracy (Fung); it
is also often viewed as mobilization for specific benefits more than deliberation. In any case, any process based on self-selection is likely to be unrepresentative. Only those who are especially interested volunteer themselves
(World Bank, 2007).
From the standpoint of deliberative democracy, the principal approach over
the past two decades has been to gather a representative microcosm and
then to engage it in good conditions for considering the issue. The basic idea

4

EMERGING TRENDS IN THE SOCIAL AND BEHAVIORAL SCIENCES

goes back to Ancient Athens. Several key institutions used random sampling
implemented with a machine (the Kleroterion) that selected citizens from a
list of those who had previously expressed willingness to engage in public
service. The Council of 500 selected in this way determined the agenda for
what could be voted on in the Assembly, the graphe paranomon was a randomly selected jury of 500 that could prosecute irresponsible participation
in the Assembly and the nomethetai was a final stage of deliberative decision.
After a proposal passed the assembly, it did not immediately become law. It
could only do so if a random sample of 500 or more was convened to hear
the case for and against the proposed law and then after a day of deliberation voted positively on the proposal. Hence, the deliberating microcosms
had a key role in Athenian democracy before, during, and after the direct
democracy of the Assembly.
These ancient institutions exemplify a practical method for implementing
a very different kind of democracy. It is distinct from direct participatory
democracy (as in the Assembly or as in modern referenda), distinct from
party competition democracy implemented through elections and distinct
from the kind of deliberation by representatives envisioned by the American founders (deliberation by representatives on the merits without regard
to party interests or party discipline). A challenge for modern research is
whether some of the positive attributes of this kind of democratic design for
deliberation by a random sample can be adapted to modern conditions.
In modern times, deliberating microcosms chosen by lot have been used
to make recommendations on policy in many countries around the world
(Gronlund, Bachtiger, & Setala, 2014). A particular design, Deliberative
Polling (cdd.stanford.edu), has been applied in 23 countries. In addition, deliberating microcosms such as the Citizens’ Assemblies in British
Columbia and Ontario have directly put propositions on the ballot (Fournier,
van der Kolk, Kenneth Carty, Blais, & Rose, 2011), “citizens’ juries” have been
employed on many issues, especially health care in the United Kingdom.
The list is extensive, but the particular designs vary in important particulars.
For now, it is worth noting the use of random sampling for a deliberating
microcosm as the predominant approach to implementing “deliberative
democracy” by the people themselves, rather than by their representatives.
WHAT IS THE POLICY CONTEXT?
When a microcosm of the public is convened, should its decisions be
binding or advisory to other decisions? While some theorists have argued
that in order to count as deliberation, the conclusions ought to be “binding”
(Gutmann & Thompson, 1996, 1998), there have in fact been few if any such
cases if binding is interpreted strictly as requiring the decision to conform

Deliberative Democracy

5

to the conclusions of the citizens. Deliberative microcosms do not produce
binding decisions in the way that say a referendum can (although of course
some referenda are also only advisory). The definition of binding, when
clarified comes down to the intent of the deliberators. As Gutmann and
Thompson describe their proposed requirement, participants “intend their
discussion to influence a decision the government will make or a process
that will affect how future decisions are made” (Gutmann & Thompson,
1998). On this interpretation, many deliberating forums are binding.
The deliberating microcosm typically will offer its conclusions as an input
to other decision-making. For example, between 1996 and 1998, there were
eight Deliberative Polls convened in Texas by the electric utilities and the
Public Utility Commission to provide input to the creation of “Integrated
Resource Plans” that would chart how that portion of the state would
satisfy future energy needs. Deliberators increased their level of support
for renewable energy such as wind power, to the point that much higher
percentages of the respondents were willing to pay increased utility bills
to support renewable energy. The Public Utility Commission reacted to
the data and the eight events by supporting increases in wind power and,
in several service territories, slightly increased fees to support the new
energy. Texas dramatically increased its wind power. These consultations
were part of a process by which Texas moved from having the least wind
power among the 50 states to having the most, surpassing California in 2007
(Fishkin, 2009). The context that made this successful was the requirement,
implemented by the Public Utility Commission, that the utilities consult
the public in some way. The Deliberative Poll was adopted for public
consultation in all eight service territories of the regulated utilities because
it offered a method to consult that was representative and evidence based.
Large investments in different forms of energy were at stake and an unrepresentative process or one dominated by bad information could lead to costly
and or bizarre outcomes. Other examples of advisory but near-binding
decisions can be found in the Deliberative Polls in China and Macau. When
the government convenes a Deliberative Poll to consult its citizens about
a policy issue, say which infrastructure projects to build (as in several
projects in China sponsored by local government, see Fishkin, He, Luskin,
& Siu, 2010), or the design of a press council (in Macau see report http: cdd.
stanford.edu/2011/final-report-deliberative-polling-on-the-amendment-ofthe-press-law-and-the-audio-visual-broadcasting-act-in-macao/), there is a
strong impetus to implement the result. After all, there is official sponsorship
and government resources are being spent. It would seem odd to go to all
the trouble of convening a random sample for intensive deliberation and
then to ignore the result.

6

EMERGING TRENDS IN THE SOCIAL AND BEHAVIORAL SCIENCES

Another route to impact is to set an agenda for decisions by the voters. The
two Citizens’ Assemblies in British Columbia and Ontario proposed ballot
measures for electoral reform that went directly on the ballot. While neither
passed, they broke new ground in that a random sample of citizens deliberated for about a year and produced a proposal that the rest of the province
voted on (Fournier et al., 2011). In a less explicit way, the 2011 Deliberative
Poll in California (“What’s Next California”) produced recommendations
both for what could go on the ballot and for decisions by the legislature.
Proposition 31 included six elements from the citizen deliberations, but
other elements as well. The proposition was defeated, but recommendations
for legislation about initiative reform did pass (Fishkin, Kousser, Luskin, &
Siu, 2015).
Setting the agenda for voting by the entire electorate harks back to the role
of the council of 500 in Ancient Athens. It is a very strong form of influence without final decision-making power. A lesser form of influence could
occur through the deliberations producing a voter recommendation. The Citizens’ Initiative Review convenes small random samples to deliberate on the
model of a citizens’ jury (Coote & Lenaghan, 1997) and make a recommendation about ballot propositions. Those recommendations go in the voter
handbook. Deliberative Polls, which employ larger samples, were broadcast
before national referendums in Australia (on the Republic in 1999) and in
Denmark (on the Euro in 2000). Of course, a televised process of voter discussion with results of the citizen deliberations would have to compete with
all the campaign sound bites in a referendum campaign to have influence.
However, the idea is that such deliberating microcosms express the informed
views of ordinary citizens rather than just partisan campaign messages.
WHY SHOULD OTHERS PAY ATTENTION?
The strategy of trying to implement deliberative democracy, at least on
selected issues, through randomly chosen microcosms has the advantage
that it offers a practical method to give voice to the public’s considered
judgments. In that respect, it offers a stark contrast to standard public
opinion polls that, of course, also use random sampling. Because most
voters in their ordinary environments do not have much occasion or
incentive to deliberate, the opinions represented in standard opinion polls
often represent little more than the public’s impressions of sound bites and
headlines. However, when such poll results are reported, policy makers
have, nevertheless, a reason to pay attention. If done well, those polls offer
a snapshot of the actual state of public opinion at the time. This is what
the people are actually thinking, even if they are not thinking a great deal
about the issue in question. There are exceptions of course. The polls might

Deliberative Democracy

7

represent only nonattitudes or phantom opinions as respondents who have
not thought about an issue attempt to offer an answer on the spot to avoid
admitting that they “don’t know” (Bishop, 2005; Converse, 1964). However,
setting aside this complication, standard polls offer policy makers a picture
of the current state of public opinion and if they wish to be responsive to the
public, such information is useful and relevant.
However, the opinions offered by a deliberating microcosm are different
in a key respect. They are not shared by those who have not deliberated.
They offer a counterfactual representation of what the people would think,
presumably under good conditions for thinking about it. Why should policy
makers or other citizens (voters) respond to counterfactual opinion?
The issue is ultimately normative. As we noted earlier, democracy presumes some connection between the public will and what is done. If actual
conditions do not really permit the public to come to an informed judgment,
why not find out what it would think under transparently good conditions,
under conditions where it can really consider competing arguments and get
its questions answered from different points of view. If the reasons for its
considered judgments can then be identified, those reasons should resonate
with other citizens.
On the many issues for which the public is inattentive, why rely on conventional polls rather than a representative and informed judgment from
the deliberations of a good sample? Of course, one might take the position
that public opinion, whether deliberative or not, is irrelevant. Policy making
should be a matter of expert judgment. However, it is the people who have
to live with the consequences of one policy choice or another. Surely, there
is room to take their views into account. Moreover, if one takes that step,
why not take their informed and representative views. Hence the idea of the
deliberating mini-public or microcosm.
The Texas utility cases mentioned earlier provide an illustration. The public
was not very knowledgeable before deliberation about the attributes of various energy choices. In fact, the original discovery of nonattitudes or phantom
opinions had come from a question about electric utilities in a panel study
from the American National Election Study (Converse, 1964). Thus, on some
questions about electric utilities, the public could be expected not to have any
real opinions at all.
In Texas, the low salience and information levels on these issues helped
make the case for Deliberative Polling. The deliberating microcosm offered
an alternative that was evidence based, thoughtful, and representative. To
rely on conventional polling alone would have rendered very big expenditures vulnerable to conclusions from people who had little understanding,
predeliberation, of the trade-offs at stake. Yet to leave the people out of the
decision would, in effect, have substituted values of the experts for the values

8

EMERGING TRENDS IN THE SOCIAL AND BEHAVIORAL SCIENCES

of the public. Hence the case for consulting those affected but in a representative and thoughtful way on the basis of the best information available.
BY WHAT CRITERIA SHOULD SUCH PROCESSES BE EVALUATED?
The claim of the deliberating microcosm for the attention of policy makers
and fellow citizens rests upon a hypothetical inference—these are the conclusions the population would come to if they could somehow consider the
issue in depth under the same good conditions. Hence, the conditions must
be credible as good conditions (e.g., access to good information and relevant
arguments) and the sample must be representative. Consider some criteria
for evaluation building on one or the other of these two basic points:

•
•
•
•

•

Demographic representativeness
Attitudinal representativeness
Sample size
Opportunity to engage policy arguments for and against proposals for
action
Avoiding distortions

DEMOGRAPHIC REPRESENTATIVENESS
To support the hypothetical inference that the population as a whole would
likely come to similar conclusions if it were to deliberate under comparably good conditions, the microcosm needs to be representative at the start. It
needs to be a version of the whole country or the whole city (or whatever the
relevant population may be) in miniature, usually in one place (although for
virtual versions distributed in space but connected electronically, see Fishkin,
2009). But representative in what ways? Demographics in all the standard
categories, class, gender, education, income, and ethnicity are usually among
the most prominent. Some approaches to creating a deliberative microcosm
view such representativeness in demographics as sufficient (Carson, Gastil,
Hartz-Karp, & Lubensky, 2013). Others require attitudinal representativeness as well. An argument for demographic representativeness is that people
from different demographic backgrounds will likely have different views and
interests. Furthermore, as they deliberate they may realize more about their
views and interests. If significant portions of the population were excluded
(say very few women or very few minorities) one could plausibly suspect
that the deliberation might have turned out differently if the microcosm had
been more representative.

Deliberative Democracy

9

ATTITUDINAL REPRESENTATIVENESS
Attitudinal representativeness seems equally important. If the sample is a
representative microcosm of the viewpoints in the population, then it is more
plausible to argue that as participants weigh the arguments from these competing perspectives, their views would represent the hypothetical—the conclusions the population would come to, if they could weigh the arguments
with comparable engagement, seriousness, and high-quality information. If
the microcosm begins with different viewpoints than the population, then the
hypothetical inference loses its plausibility. J.S. Mill described the importance
of attitudinal representativeness in famous comments on the idea of Parliament acting as a “Congress of Opinions:” “where every person in the country
may count upon finding somebody who speaks his mind, as well or better
than he could speak it himself … where those whose opinion is overruled,
feel satisfied that it is heard, and set aside not by a mere act of will, but for
what are thought superior reasons” Mill, 1861). Mill’s account of the deliberative ideal probably applies to Parliaments only rarely, but it does offer a
clear picture of the advantages of ensuring that a deliberative microcosm is
representative in attitudes.
For samples of ordinary citizens, the recruitment of the microcosm should
provide a basis for empirical assessment of how attitudes of participants
compare to the attitudes of the rest of the public. One simple approach is
to ask the questionnaire of potential participants before recruitment allowing comparison of participants and nonparticipants in attitudes as well as
demographics. In addition, having another comparison group who are never
asked to deliberate is useful and can also be used for matching in case there
is any sampling bias. Deliberative Polling typically employs these methods
but most attempts at recruiting deliberative microcosms do not (for a compendium, see Gastil & Levine, 2005).
SAMPLE SIZE
The microcosm needs to be large enough that its representativeness and the
statistical significance of any opinion changes can be meaningfully evaluated. If it is not representative, then why should others pay attention? Its
conclusions might have been determined by strong or idiosyncratic views
at the start. If the changes are not statistically significant, then they might
be mere random noise. We do not know how seriously to take them. Some
microcosms, on the model of a modern jury face both of these difficulties. The
Citizens’ Initiative Review in Oregon has the noble aspiration of engaging a
microcosm of the state’s voters and offering recommendations about how
the electorate as a whole should vote. The recommendations are printed in
the voter handbook. However, the sample size is so small that some of the

10

EMERGING TRENDS IN THE SOCIAL AND BEHAVIORAL SCIENCES

recommendations are simply wiped out by sampling error. For example, the
recommendation for Oregon Proposition 74 on medical marijuana in 2010
was 13 to 11 (Fishkin, 2013).
ARGUMENTS FOR AND AGAINST
The root notion of deliberation is “weighing,” the weighing of competing
arguments or reasons for one course of action or another. Therefore, any
deliberative microcosm needs an effective method of engaging the participants in these competing arguments. One method is to have a stakeholder
advisory group prepare briefing materials that attempt to anticipate the arguments on either side that are current in the policy community of experts.
Ordinary citizens are unlikely to be aware of all of these. However, experts
will usually not be able to anticipate all the concerns of the public. Hence
the public will likely add further arguments, either pro or con. If the public
in the deliberating microcosm has an opportunity to get its factual questions
answered by experts and policy makers, it further facilitates their coming to
an informed judgment. The Deliberative Poll design uses all three of these
methods: the balanced briefing materials, the small group discussions where
people share their own reasons for support or opposition to the various proposals, and access to competing experts who can answer questions from the
small group discussions.
How can we know if participants have actually considered competing arguments? Both quantitative and qualitative data can be gathered, the former
from before and after questionnaires and the latter from transcripts of the
group discussions. These can be coded for balance and argumentation creating new variables that can be incorporated into the quantitative analyses
(Gerber, Bächtiger, Fiket, Steenbergen, & Steiner, 2014; Siu, 2009). This is an
important emerging trend in the study of deliberative democracy.
AVOIDING DISTORTIONS
There are two principal ways in which it has been alleged that deliberations
may go wrong. First there is a long tradition of discussion, mostly in the
jury literature, that deliberations will be dominated by the men, by the more
educated, by those who are more advantaged. They will have preponderant
influence, importing the socioeconomic inequalities in the wider world into
the deliberating microcosm. Part of this is deference to the more advantaged,
part of it is the very skills required for deliberation are thought to be disproportionately distributed to the better advantaged. The more advantaged are
better able to make arguments and because of their social positions, they are
more likely to be listened to (Sanders, 1997; Young, 2000).

Deliberative Democracy

11

Yet the actual empirical study of this problem has not confirmed these speculations. At least with Deliberative Polling, if one looks at the initial positions
of the more advantaged (the more educated, the males, the rich, etc.) on policy indices on the issues deliberated about, there is no pattern of movement
by the small groups in the direction of those positions. The more advantaged
are not dominating the discussions by imposing their views (Luskin, Sood,
Fishkin, & Hahn, 2015; Siu, 2009). This is an area for further study, but it
seems likely that the design of the deliberative process, the role of moderators, balanced materials, and ample opportunities for everyone to discuss,
all play an important role. Further, juries are likely to employ social pressure
to reach a unanimous verdict. In the Deliberative Poll there is by design no
pressure for consensus. Rather the participants’ final considered judgments
are recorded in confidential questionnaires.
A second major distortion is the pattern Cass Sunstein has identified as the
“law of group polarization” (Sunstein, 2002, 2009). If there is an issue for
which there is a midpoint, if the initial position of the group is one side of
the midpoint, the hypothesis is that it will move further away from the midpoint to that side, in effect becoming more extreme. Two basic mechanisms
are offered—imbalance in the argument pool and a social comparison effect.
If most people are on one side of an issue, they are likely to offer more reasons for that side of the argument. As they see others moving in that direction
they will feel pressures to conform. Sunstein and various collaborators have
corroborated this pattern in some jury-like studies (for an overview, see Sunstein, 2009). However, in the Deliberative Polling design, the pattern has not
been found. In an aggregation of the small group patterns in 22 Deliberative Polls, the tendency to move away from the midpoint was <50% (Luskin
et al., 2015; see also Fishkin et al., 2010; Siu, 2009). If the deliberative design
has sufficient balance (briefing materials, moderation, and balanced panels
of experts), it can overcome any imbalance in the argument pool provided
by participants. Moreover, if there are confidential questionnaires, the design
can minimize the social comparison effects.
Why are these two distortions important? Either domination by the
more advantaged or polarization (in Sunstein’s sense) regularly moving
the participants to more extreme positions could plausibly bring into
question whether the conclusions were the result of genuine deliberation
on the merits. It cannot be the Habermasian “unforced force of the better
argument” (Habermas, 1996) if people are trapped in a pattern of group
psychology or implicitly coerced by pressures of group conformity. The
fact that the prevalence of these patterns seems to depend on the precise
design of the deliberative process opens up an area for further empirical
work. The study of multiple projects, aggregating the changes in the small
group discussions to investigate whether or not these distorting patterns

12

EMERGING TRENDS IN THE SOCIAL AND BEHAVIORAL SCIENCES

hold with various designs is an important emerging trend in the study of
deliberative democracy. The whole empirical research program connected
with deliberative democracy is designed to make it practical. The empirical
study allows for an assessment of the degree to which a theoretical concept
can actually be implemented in real democratic practices.
REFERENCES
Ackerman, B. (1991). We the people: Foundations: We the people (Vol 1). Cambridge, MA:
Harvard University Press.
Ackerman, B. (2014). The civil rights revolution: We the people (Vol 3). Cambridge, MA:
Harvard University Press.
Ackerman, B., & Fishkin, J. (2004). Deliberation day. New Haven, CT: Yale University
Press.
Bessette, J. (1994). The mild voice of reason. Chicago, IL: University of Chicago Press.
Bishop, G. (2005). The illusion of public opinion. Lanham, MD: Rowman and Littlefield.
Bohman, J., & Rehg, W. (Eds.) (1997). Deliberative democracy: Essays on reason and politics. Cambridge, MA: MIT Press.
Carson, L., Gastil, J., Hartz-Karp, J., & Lubensky, R. (Eds.) (2013). The Australian citizens’ parliament and the future of deliberative democracy. State College, PA: Penn State
University Press.
Converse, P. E. (1964). The nature of belief systems in mass publics. In D. Apter (Ed.),
Ideology and its discontents. New York, NY: The Free Press of Glencoe.
Coote, A., & Lenaghan, J. (1997). Citizens’ juries: Theory into practice. London, England:
Institute for Public Policy Research.
Cronin, T. (1999). Direct democracy: The politics of initiative, referendum and recall. Cambridge, MA: Harvard University Press.
Delli Carpini, M., & Keeter, S. (1997). What Americans know about politics and why it
matters. New Haven, CT: Yale University Press.
Downs, A. (1957). An Economic Theory of Democracy. New York: Harper and Row.
Elster, J. (Ed.) (1998). Deliberative democracy. Cambridge, MA: Cambridge University
Press.
Fishkin, J. S. (2009). When the people speak: Deliberative democracy and public consultation.
Oxford, England: Oxford University Press.
Fishkin, J. S. (2013). Deliberation by the people themselves: Entry points for the public voice. Election Law Journal, 12(4), 490–507.
Fishkin, J. S., He, B., Luskin, R. C., & Siu, A. (2010). Deliberative democracy in an
unlikely place: Deliberative polling in China. British Journal of Political Science,
40(2), 435–448.
Fishkin, J. S., Kousser, T., Luskin, R. C., & Siu, A. (2015). Deliberative agenda setting:
piloting reform of direct democracy in California. Perspectives on Politics, 13(4),
1030–1042.
Fishkin, J., & Laslett, P. (Eds.) (2003). Debating deliberative democracy. Oxford, England:
Blackwell.

Deliberative Democracy

13

Fournier, P., van der Kolk, H., Kenneth Carty, R., Blais, A., & Rose, J. (2011). When
citizens decide: Lessons from citizen assemblies on electoral reform. Oxford, England:
Oxford University Press.
Gastil, J., & Levine, P. (Eds.) (2005). The deliberative democracy handbook. San Francisco,
CA: Jossey-Bass.
Gerber, M., Bächtiger, A., Fiket, I., Steenbergen, M. R., & Steiner, J. (2014). Deliberative and non-deliberative persuasion: Opinion change in a pan-European deliberative poll (Europolis). European Union Politics, 15, 410–429.
Gronlund, K., Bachtiger, A., & Setala, M. (Eds.) (2014). Deliberative mini-publics:
involving citizens in the democratic process. Colchester, England: ECPR Press.
Gutmann, A., & Thompson, D. (1996). Democracy and disagreement. Cambridge, MA:
Harvard University Press.
Gutmann, A., & Thompson, D. (1998). Why deliberative democracy? Princeton, NJ:
Princeton University Press.
Habermas, J. (1996). Between facts and norms: Contributions to a discourse theory of law
and democracy. Cambridge, MA: MIT Press.
Luskin, R. C., Sood, G., Fishkin, J., & Hahn, K. (2015). Deliberative distortions? Homogenization, polarization, and domination. Working paper, Center for Deliberative
Democracy, Stanford University
Mill, J. S. (1861). Considerations on representative government. London, England: Parker,
Son and Bourn.
Sanders, L. M. (1997). Against deliberation. Political Theory, 25, 347–376.
Schumpeter, J. A. (1942). Capitalism, socialism and democracy. New York, NY: Harper
and Row.
Siu, A. 2009. Look who’s talking: Examining social influence, opinion change and argument
quality in deliberation (PhD dissertation), Stanford University.
Steiner, J., Bächtiger, A., Spörndli, M., & Steenbergen, M. R. (2005). Deliberative politics
in action: Analyzing parliamentary discourse. Cambridge, MA: Cambridge University Press.
Sunstein, C. R. (2002). The law of group polarization. The Journal of Political Philosophy,
10(2), 175–195.
Sunstein, C. R. (2009). Going to extremes: How like minds unite and divide. Oxford,
England: Oxford University Press.
Uhr, J. (1998). Deliberative democracy in Australia. Cambridge, MA: Cambridge University Press.
World Bank. 2007. Enhancing the social and economic sustainability of participatory budgeting in Porto Alegre, Brazil. Final report.
Young, I. M. (2000). Inclusion and democracy. Oxford, England: Oxford University
Press.

JAMES S. FISHKIN SHORT BIOGRAPHY
James S. Fishkin holds the Janet M. Peck Chair in International Communication at Stanford University where he is a professor of communication,

14

EMERGING TRENDS IN THE SOCIAL AND BEHAVIORAL SCIENCES

professor of political science (by courtesy), and the director of the Center for
Deliberative Democracy. His work focuses on deliberative democracy and
democratic theory in books such as When the People Speak (2009), Deliberation Day (2004 with Bruce Ackerman), and Democracy and Deliberation
(1991). He originated Deliberative Polling as a method of public consultation
in 1988. He began to apply it in collaboration with Robert C. Luskin in 1994
and has since spread it, various collaborators, to projects in 23 countries. For
more on Deliberative Polling see cdd.stanford.edu. He has been a Guggenheim Fellow, a Fellow of the Center for Advanced Study in the Behavioral
Sciences at Stanford, and a Fellow of the American Academy of Arts and
Sciences.
RELATED ESSAYS
Understanding American Political Conservatism (Political Science), Joel D.
Aberbach
Party Organizations’ Electioneering Arms Race (Political Science), John H.
Aldrich and Jeffrey D. Grynaviski
Economic Models of Voting (Political Science), Ian G. Anson and Timothy
Hellwig
The Underrepresentation of Women in Elective Office (Political Science),
Sarah F. Anzia
Public Opinion and International Conflict (Political Science), Adam J.
Berinsky
Neoliberalism (Sociology), Miguel Angel Centeno and Joseph N. Cohen
Political Advertising (Political Science), Erika Franklin Fowler
Government Formation and Cabinets (Political Science), Sona N. Golder
Presidential Power (Political Science), William G. Howell
Group Identity and Political Cohesion (Political Science), Leonie Huddy
Racial Disenfranchisement (Political Science), Vincent L. Hutchings and
Davin L. Phoenix
Genetic Foundations of Attitude Formation (Political Science), Christian
Kandler et al.
Women Running for Office (Political Science), Jennifer L. Lawless
Civic Engagement (Sociology), Peter Levine
From Individual Rationality to Socially Embedded Self-Regulation (Sociology), Siegwart Lindenberg
Political Inequality (Sociology), Jeff Manza
Participatory Governance (Political Science), Stephanie L. McNulty and Brian
Wampler
Money in Politics (Political Science), Jeffrey Milyo
Implicit Attitude Measures (Psychology), Gregory Mitchell and Philip E.
Tetlock

Deliberative Democracy

15

Feminists in Power (Sociology), Ann Orloff and Talia Schiff
Evolutionary Theory and Political Behavior (Political Science), Michael Bang
Petersen and Lene Aarøe
Electoral Authoritarianism (Political Science), Andreas Schedler
Does the 1 Person 1 Vote Principle Apply? (Political Science), Ian R. Turner
et al.
Information Politics in Dictatorships (Political Science), Jeremy L. Wallace
Constitutionalism (Political Science), Keith E. Whittington
Political Psychology and International Conflict (Sociology), Rose McDermott