ANRV344-PL11-22 ARI 17 April 2008 13:15

Deliberative DemocraticTheory and EmpiricalPolitical ScienceDennis F. ThompsonDepartment of Government, Harvard University, Cambridge, Massachusetts 02138;email: dennis thompson@harvard.edu

Annu. Rev. Polit. Sci. 2008. 11:497–520

First published online as a Review in Advance onMarch 11, 2008

The Annual Review of Political Science is online athttp://polisci.annualreviews.org

This article’s doi:10.1146/annurev.polisci.11.081306.070555

Copyright c© 2008 by Annual Reviews.All rights reserved

1094-2939/08/0615-0497$20.00

Key Words

deliberative democracy, equality, participation, legitimacy,publicity

AbstractAlthough empirical studies of deliberative democracy have prolifer-ated in the past decade, too few have addressed the questions thatare most significant in the normative theories. At the same time,many theorists have tended too easily to dismiss the empirical find-ings. More recently, some theorists and empiricists have been payingmore attention to each other’s work. Nevertheless, neither is likelyto produce the more comprehensive understanding of deliberativedemocracy we need unless both develop a clearer conception of theelements of deliberation, the conflicts among those elements, andthe structural relationships in deliberative systems.

497

Ann

u. R

ev. P

olit.

Sci

. 200

8.11

:497

-520

. Dow

nloa

ded

from

ww

w.a

nnua

lrev

iew

s.or

gby

Roy

al M

elbo

urne

Ins

titut

e of

Tec

hnol

ogy

(RM

IT)

on 0

3/11

/13.

For

per

sona

l use

onl

y.

ANRV344-PL11-22 ARI 17 April 2008 13:15

INTRODUCTION

In a major recent study of deliberative democ-racy, the authors write that “empirical re-search can merely be a helping hand in thebig controversies in democratic theory. But,as a helping hand, empirical research has itsplace” (Steiner et al. 2004, p. 42). The authorsmay be too modest. Some of the best empiricalwork (including theirs) has the potential to of-fer more than a helping hand. But if the handof research is to guide as well as help, it mustbe systematically directed toward the coreproblems in deliberative theory. Empirical in-quiry can more effectively influence—and inturn be influenced by—normative theory ifboth theorists and empiricists proceed with aclearer conception of the elements of deliber-ation. They can then more productively ad-dress two general problems that have not re-ceived the attention they deserve: the conflictsamong those elements, and the relationshipsof the parts of deliberative democracy to itswhole.

At the core of all theories of deliberativedemocracy is what may be called a reason-giving requirement. Citizens and their rep-resentatives are expected to justify the lawsthey would impose on one another by givingreasons for their political claims and respond-ing to others’ reasons in return. (For a surveyof the meaning and variety of theories, seeGutmann & Thompson 2004, pp. 1–39; themost important collections of recent theoret-ical writings include Benhabib 1996, Besson& Marti 2006, Bohman & Rehg 1997, Elster1998, Fishkin & Laslett 2003, Macedo 1999.)Deliberative theorists differ to some extenton what counts as an adequate reason, howextensive the reason-giving forum should be,whether procedural norms are sufficient, andthe desirability of consensus as a goal. But theyagree in rejecting conceptions of democracythat base politics only on power or interest,aggregation of preferences, and competitivetheories in the tradition of writers such asSchumpeter and Downs. These conceptionsdo not give sufficient weight to the process of

justifying to one’s fellow citizens the laws thatwould bind them. (For various statements ofthe contrast with other theories, see Cohen1989; Habermas 1984, 1989, 1996; Gutmann& Thompson 1996.)

Some of the claims of deliberative theoryare not empirical. One of the most importantbenefits that theorists ascribe to deliberativedemocracy is that the decisions it producesare more legitimate because they respect themoral agency of the participants. This benefitis inherent in the process, not a consequenceof it. It is not appropriately subjected to directempirical investigation. But other claims thetheory makes plainly invite empirical inquiry,and theorists themselves were among the firstto undertake empirical studies of deliberation(Chambers 1996, 1998; Dryzek & Braithwaite2000; Fishkin 1999; Fishkin & Luskin 2005;Mansbridge 1980). They treated the empiricalclaims not as assumptions but as hypotheses,many of which in their view required furtherresearch. Then, as deliberative democracy be-came the “most active” area of political the-ory (Dryzek 2007), political scientists joinedthe venture. The result has been a profu-sion of empirical studies, now more numer-ous than the normative works that promptedthem. (For surveys, see Delli Carpini et al.2004; Mendelberg 2002; Ryfe 2005; Steineret al. 2004, pp. 43–52.)

Despite these impressive efforts, much ofthe empirical research by political scientistshas not fully engaged with the normative the-ory. Theorists and researchers often “talk pasteach other” (Neblo 2005). Some researchershave assumed that they can dispose of deliber-ative theory by showing that political discus-sion often does not produce the benefits thattheorists are presumed to claim for it. Theyextract from isolated passages in various the-oretical writings a simplified statement aboutone or more benefits of deliberative democ-racy, compress it into a testable hypothesis,find or (more often) artificially create a sitein which people talk about politics, and con-clude that deliberation does not produce thebenefits the theory promised and may even

498 Thompson

Ann

u. R

ev. P

olit.

Sci

. 200

8.11

:497

-520

. Dow

nloa

ded

from

ww

w.a

nnua

lrev

iew

s.or

gby

Roy

al M

elbo

urne

Ins

titut

e of

Tec

hnol

ogy

(RM

IT)

on 0

3/11

/13.

For

per

sona

l use

onl

y.

ANRV344-PL11-22 ARI 17 April 2008 13:15

be counterproductive. The most insistentlyskeptical work in this mode is Hibbing &Theiss-Morse’s Stealth Democracy (2002). Re-viewing the results of their own focus groupsand other studies of discussion in settings theyconsider deliberative, they argue that “real-life deliberation can fan emotions unproduc-tively, can exacerbate rather than diminishpower differentials among those deliberating,can make people feel frustrated with the sys-tem that made them deliberate, is ill-suited tomany issues and can lead to worse decisionsthan would have occurred if no deliberationhad taken place” (p. 191).

Other studies recognize the limits of theirmethods, and are more qualified in theirconclusions but still present their largelynegative findings as objections to deliberativetheory. In a survey of French citizens aboutgovernment assistance for the unemployed,Jackman & Sniderman (2006) found that de-liberation does not lead to “better groundedjudgments—that is, judgments that reflectone’s considered view of the best course ofaction all in all” (p. 272). Deliberation leads“many people to ideologically inconsistentpositions.” A study of discussions aboutrace in five town meetings in New Jersey(Mendelberg & Oleske 2000) found thatin the integrated meetings (which had thediversity that deliberative democrats seek) thedeliberation failed to lessen conflict, increasemutual understanding and tolerance, or re-duce the use of group-interested arguments.The meetings with all white participantsproduced consensus, but consensus againstschool integration—not the result thatdeliberative democrats presumably favor.Using survey data and focus groups from sixcommunities in the United States and Britain,Conover & Searing (2005) examined the ex-tent to which political discussion satisfies “thestandards set by political theorists: publicity,nontyranny and equality.” They concludethat the discussions “currently fall short of theideals of deliberative democracy,” althoughthey see some potential for improvementin educational reforms. Rosenberg (2007b)

also finds that deliberation fails to providethe benefits that some theorists claim for it,but suggests that a “more collaborative andtransformative” form of deliberation mayhave greater potential.

The objection prompted by thesestudies—that deliberative theory is notrealistic—has never impressed normativetheorists. They believe that it misses thepoint. Theory challenges political reality. Itis not supposed to accept as given the realitythat political science purports to describeand explain. It is intended to be critical, notacquiescent.

Theorists also challenge some of the em-pirical studies on their own terms. Dryzek(2007) sharply criticizes the methods ofHibbing & Theiss-Morse, for example. Heargues that they ignore contrary evidence intheir own survey data, and that they use theconclusions of focus groups, who are in effectdeliberating, to show that citizens do not wantto deliberate.

Yet most deliberative theorists now recog-nize that they cannot ignore empirical stud-ies without retreating into utopianism andrendering the theory irrelevant to ongoingpolitics. Despite his criticism of some of thestudies, Dryzek (2007, p. 250) acknowledgesthat other empirical findings are “quite capa-ble of discomforting theorists.” In the samespirit, even while insisting that “delibera-tive democracy is still in large part a criti-cal and oppositional ideal,” Bohman (1998,p. 422) concludes that the theory has “come ofage” because it has recognized that “the bestand most feasible formulations of deliberativedemocracy require the check of empirical so-cial science.”

The general conclusion of surveys of theempirical research so far is that taken to-gether the findings are mixed or inconclusive(Chambers 1996, p. 318; Delli Carpini et al.2004, pp. 336–37; Janssen & Kies 2005, p. 331;Ryfe 2005; Sulkin & Simon 2001, p. 812). Themain reason for the mixed results is that thesuccess or failure of deliberation depends somuch on its context. The contingent character

www.annualreviews.org • Deliberative Democratic Theory and Empirical Political Science 499

Ann

u. R

ev. P

olit.

Sci

. 200

8.11

:497

-520

. Dow

nloa

ded

from

ww

w.a

nnua

lrev

iew

s.or

gby

Roy

al M

elbo

urne

Ins

titut

e of

Tec

hnol

ogy

(RM

IT)

on 0

3/11

/13.

For

per

sona

l use

onl

y.

ANRV344-PL11-22 ARI 17 April 2008 13:15

of these results may seem to give theoristshope. If only theorists can identify the rightconditions, they can confidently continue toextol the virtues of deliberative democracy(e.g., Gastil & Levine 2005, pp. 273–74; Fung& Wright 2003, pp. 259–60). They can useeven the negative findings to point out de-fects in the system, and support reforms thatwould bring about conditions more favorableto deliberative democracy. When confrontedwith findings that seem to confute his theory,Habermas is unfazed. He reads the “contra-dicting data as indicators of contingent con-straints that deserve serious inquiry and. . .asdetectors for the discovery of specific causesfor existing lacks of legitimacy” (Habermas2006, p. 420). His article is pointedly subtitled“the impact of normative theory on empiri-cal research.” It implicitly relegates empiricalresearch to the job of being merely a helpinghand. In that role, it poses no risk of becominga disruptive voice in the deliberative project.

Theorists should not take too much com-fort from the mixed or contingent characterof the empirical conclusions. The conditionsunder which deliberative democracy thrivesmay be quite rare and difficult to achieve.In a welcome collection that brings togethertheorists and empiricists (Rosenberg 2007a),several of the theorists explicitly take up thischallenge (Warren 2007, Cohen 2007, Dryzek2007).

The most promising approach for empir-ical research would therefore seem to be tocontinue trying to discover the conditionsin which deliberative democracy does anddoes not work well, while paying more at-tention to the question of to what extent theunfavorable conditions could change. Someconditions (such as the absence of genuinelydeliberative forums) might be affected by leg-islative measures or political action at localor national levels. Others (such as inequal-ities of resources) may be products of thesocial and economic structure of particularsystems. Still others (such as the fact of rea-sonable pluralism) may be essential charac-teristics of democratic systems. This poten-

tially fruitful approach would address a centralconcern of deliberative theory—the possibil-ities of its practical realization (for a system-atic statement of a similar approach, see Fung2007a). There would be no guarantee that de-liberative democracy would be vindicated, butwith a more discriminating and wide-ranginganalysis of the conditions that promote or im-pede it, we would have a clearer sense of itsplace in democratic theory and practice.Thisseems a worthy and appropriate project forcollaboration of theorists and empiricists in-terested in deliberative democracy. The aimwould not be reform as such (although theconclusions may be useful to reformers). Itwould be to understand better the extent towhich the values posited by deliberative the-ory can be realized under not only current butalso potential conditions.

However, any such project is more prob-lematic than this straightforward prospectusmight suggest. No collaboration between the-orists and empiricists is likely to make fur-ther progress until three general problems aremore fully addressed:

� The analytic problem, which re-quires distinguishing the elements ofdeliberation—its concept, standards,and conditions.

� The internal conflicts problem, whichnecessitates recognizing that the con-ditions that promote some values ofdeliberative democracy may undermineother values, including some that delib-erative democrats favor.

� The structural problem, which calls formoving beyond the study of isolated orone-time deliberative experiences andexamining the relationship between de-liberative and non-deliberative prac-tices in the political system as a wholeand over time.

THE ELEMENTS OFDELIBERATION

The empirical studies typically begin with aconcept of deliberation and a list of benefits it

500 Thompson

Ann

u. R

ev. P

olit.

Sci

. 200

8.11

:497

-520

. Dow

nloa

ded

from

ww

w.a

nnua

lrev

iew

s.or

gby

Roy

al M

elbo

urne

Ins

titut

e of

Tec

hnol

ogy

(RM

IT)

on 0

3/11

/13.

For

per

sona

l use

onl

y.

ANRV344-PL11-22 ARI 17 April 2008 13:15

is supposed to produce. These are sometimesdrawn from one or two theories, oftenmodified for the convenience of the research.While claiming (correctly) that deliberativetheories share a common core of values, theempirical studies actually adopt diverse con-cepts of deliberation and examine differentconsequences under a range of conditions.The variations make it difficult to compare thefindings of the studies and relate them to thetheories. That would not be an insuperableobstacle to collaborative work if the variationswere presented within a common framework.We could then say that, given any conceptionof deliberation, the practice is likely toproduce consequences of a certain kind underspecified conditions. But that would not over-come this persistent problem: the elements ofdeliberation are often run together, as in thisdefinition: “. . .we have deliberative democ-racy when, under conditions of equality, inclu-siveness and transparency, a communicativeprocess based on reason. . .is able to transformindividual preferences and reach decisionsoriented to the public good” (della Porta2005, p. 340). Better are those approachesthat distinguish the definition from the eval-uation of deliberation (the “unit of analysis”from the “democratic quality”) (Nanz &Steffek 2005). But even so, the differencesamong the elements remain underanalyzed.

Three elements in the analysis of politicaldeliberation need to be distinguished: concep-tual criteria, evaluative standards, and empir-ical conditions. Each may be understood as adifferent kind of necessary requirement of de-liberation. Conceptual criteria stipulate whatis necessary for a practice to count as delibera-tion. Evaluative standards specify what countsas good (or better) deliberation. Empiricalconditions indicate what is necessary for pro-ducing good deliberation (or less strongly,what may contribute to producing good delib-eration). Each is subject to empirical inquiry,but in different ways.

It is understandable that researchers do notalways distinguish the elements. In deliber-ative theory itself, some of the same values

that specify the concept of deliberation ap-pear as standards that evaluate the practice ofdeliberation, and also as empirical conditionsthat promote it. Take the value of equality: Adiscussion does not count as deliberation atall if one person completely dominates; thediscussion is better deliberation to the extentthat the participation is equally distributed;and the discussion is more likely to be moreegalitarian if the background conditions aremore nearly equal. This interaction is inher-ent in—and a positive feature of—deliberativedemocracy. It reflects its self-correcting char-acter: Equal conditions produce a more equalprocess, which in turn produces more equalconditions, and so on. This dynamic processcan of course work in the opposite direction.It can degenerate as the conditions becomemore and more unequal. Given the poten-tial confusions of dependent and independentvariables that these interactions create, it isall the more important to keep straight thevarious elements of deliberation and to dis-tinguish the different levels of the values thateach expresses.

Conceptual Criteria

Clarifying and limiting the scope of the con-cept is not an effort by theorists to “imposefixed and often narrow definition that effec-tively shuts scholars off from existing politicalrealities,” as some suggest (Button & Mattson1999, p. 612). It is an important step in theanalysis because the choice of the concept de-termines the scope of any inquiry, and thesignificance of any conclusions for normativetheory. Empirical researchers do not have toagree on a single concept of deliberation. Afterall, theorists have not been able to do so. Butif the research is to be relevant to deliberativetheory, researchers must be clear about whatpractice they are investigating. Their charac-terization of that practice must at least partlycoincide with what most theorists regard asthe core of the problem of deliberation. Moregenerally, any inquiry must have the concep-tual resources to be able to specify whether

www.annualreviews.org • Deliberative Democratic Theory and Empirical Political Science 501

Ann

u. R

ev. P

olit.

Sci

. 200

8.11

:497

-520

. Dow

nloa

ded

from

ww

w.a

nnua

lrev

iew

s.or

gby

Roy

al M

elbo

urne

Ins

titut

e of

Tec

hnol

ogy

(RM

IT)

on 0

3/11

/13.

For

per

sona

l use

onl

y.

ANRV344-PL11-22 ARI 17 April 2008 13:15

the particular practice under investigation isdeliberation or only discussion; and, indepen-dently, if it is deliberation, to what extent it isbetter or worse deliberation.

If the concept is too broad—if it includesevery form of political talk (e.g., Cook et al.2007)—the conclusions will have “uncertainbearing” on deliberative theory (Cohen 2007,p. 222). “Everyday talk” and other forms ofpolitical discussion may contribute to devel-oping citizens’ political views and their ca-pacity to make political decisions, and thuscreate conditions that support deliberation, assome researchers on political discussion rec-ognize (Walsh 2003, 2007) and some theoristsemphasize (Mansbridge 2007). But ordinarypolitical discussion should be distinguishedfrom decision-oriented deliberation so thatthe relationships between the practices canbe systematically analyzed. Maintaining thisdistinction should not be taken to imply thatother forms of discussion are somehow lessworthy of a place in deliberative democracy.As the discussion of the problem of struc-ture (below) indicates, it is important to rec-ognize that deliberative democracy includesmany kinds of political interactions other thandeliberation. But we can more clearly retainthe connection to the central aim of deliber-ative theory if we treat these other activitiesas part of a larger democratic process, ratherthan as instances of deliberation per se.

To capture the distinctive character ofthe kind of deliberation that is central tothe theory, researchers would do better toavoid an expansive concept of delibera-tion. They should focus on those featuresof the practice that directly relate to thefundamental problem deliberative theory isintended to address: In a state of disagree-ment, how can citizens reach a collective de-cision that is legitimate? The first two aspectsof the problem, disagreement and decision,characterize the circumstances of deliberativedemocracy. The third, legitimacy, prescribesthe process by which, under these circum-stances, collective decisions can be morallyjustified to those who are bound by them.

It is the key defining element of deliberativedemocracy.

A state of disagreement. Some basic dis-agreement is necessary to create the prob-lem that deliberative democracy is intendedto solve. Several empirical studies recog-nize this criterion, although they use slightlydifferent terms: cross-cutting exposure, orsimply diversity of opinion (Barabas 2004,p. 689; Jackman & Sniderman 2006; Mutz2006, pp. 6, 14, 20, 139). If the participantsare mostly like-minded or hold the same viewsbefore they enter into the discussion, they arenot situated in the circumstances of delibera-tion. They do not confront the problem thatdeliberation is intended to address. That is notto say that discussion among like-minded peo-ple cannot contribute to deliberative democ-racy. Such discussion can help citizens learnmore about the reasons they hold their views,or perhaps even discover that they do notagree as much as they thought they did. Itcan also strengthen the views of group mem-bers and help mobilize the group for moreeffective political action. The effects of dis-cussion among the like-minded can be pos-itive or negative for the democratic process,and its differential consequences merit moreempirical and normative attention. But thiskind of talk should not be confused with dis-cussion among citizens with diverse opinions.Distinguishing the two, as indicated below, isnecessary to recognizing a potential conflictin deliberative democracy.

A collective decision. Deliberative democ-racy is focused on the circumstance in whicha group must make a decision to which allmembers are bound whether they agree withit or not. Although even political deliberationcan have various purposes (see Fung 2007b),its essential aim is to reach a binding de-cision. From the perspective of deliberativedemocracy, other purposes—such as learningabout issues, gaining a sense of efficacy, or de-veloping a better understanding of opposing

502 Thompson

Ann

u. R

ev. P

olit.

Sci

. 200

8.11

:497

-520

. Dow

nloa

ded

from

ww

w.a

nnua

lrev

iew

s.or

gby

Roy

al M

elbo

urne

Ins

titut

e of

Tec

hnol

ogy

(RM

IT)

on 0

3/11

/13.

For

per

sona

l use

onl

y.

ANRV344-PL11-22 ARI 17 April 2008 13:15

views—should be regarded as instrumental tothis aim.

Until recently, nearly all studies—andmuch of the normative theory— investigateddeliberation by ordinary citizens rather thanpoliticians. (Steiner et al. 2004, an impor-tant and welcome exception, is discussed fur-ther below.) Even in most empirical stud-ies of deliberation among ordinary citizens,the participants are not making or influenc-ing actual political decisions. Much of theliterature in the surveys is based on smallgroup discussions and laboratory experiments(Mendelberg 2002). That is a limitation, notonly because what is being studied is severalsteps removed from what deliberative theoryis ultimately concerned about, but also be-cause discussion alone is likely to produce dif-ferent empirical consequences than those ofdecision-oriented discussion. Empirical stud-ies suggest that the differences are significantand their implications not always favorablefor deliberation (Fung 2007b; Janssen & Kies2005, pp. 325; Ryfe 2005, pp. 57, 61). On thefavorable side, if participants believe that theyhave a stake in the outcome and will have tolive with the decision and with their fellowdecision makers, they may take the discussionmore seriously and try harder to reach a de-cision that is mutually acceptable. But know-ing that the discussion ends with a decisionthat counts may have just the opposite effects.Participants may act more strategically, showless tolerance for opponents, and take moreextreme positions. Groups such as juries thatare charged with reaching consequential deci-sions often polarize (Sunstein 2002), whereasFishkin’s “juries” (deliberative polls), in whichthe participants are not asked to reach a col-lective decision at all, are less likely to do so(Fishkin & Luskin 2005, p. 293). Theoristsare not surprised that, when group discussionhas little “possibility of making a real practi-cal difference,” the deliberation is less “criticaland emancipatory” than they might wish (cf.Cohen 2007, p. 234; Rosenberg 2007a).

There is another reason that deliberativetheorists insist on a distinction between dis-

cussion directed toward helping individualsdevelop more informed preferences and dis-cussion directed toward helping groups reacha collective decision (Habermas 1989). Struc-turing a discussion that in effect asks partici-pants, “What do you, as an individual, prefer?”begins to resemble the aggregative democracy(adding up the well-informed preferences ofindividuals) that deliberative democrats crit-icize. Discussions framed by asking partici-pants, “What action should we, as a group,take?” come closer to the deliberative democ-racy (creating a genuinely public opinion) thatthey favor. Some empirical evidence suggeststhat the “frame-shift” toward group ratherthan individual decisions has some of the pos-itive effects that the theorists hope for (Neblo2007b).

The criterion specifying that deliberationshould be decision-oriented does not im-ply that studies of groups that only discusspolitics, such as Fishkin’s deliberative polls,are not relevant to the study of deliberativedemocracy., The experiments conducted byFishkin and colleagues have been among themost cited in the literature of deliberative the-ory and practice. (Also, some of his more re-cent projects have involved groups that makedecisions or advise decision makers. See thereports and papers of his Center for Delibera-tive Democracy at http://cdd.stanford.edu.)Although participants in discussions of thiskind may not make collective decisions, theymay be seen as taking part in an early phaseof a process that leads to a deliberative deci-sion. Like subjects in some other studies ofpure discussion, Fishkin’s subjects are prepar-ing for (or can be seen as modeling citizenswho are preparing for) the making of po-litical decisions for the collectivity. Fishkindoes not ask his subjects to make a col-lective decision, evidently because he wantsto mitigate the pressures toward conformityand encourage a greater capacity for inde-pendent judgment. But this raises a questionthat reveals a potential conflict in deliberativedemocracy. To what extent is independentjudgment compatible with making collective

www.annualreviews.org • Deliberative Democratic Theory and Empirical Political Science 503

Ann

u. R

ev. P

olit.

Sci

. 200

8.11

:497

-520

. Dow

nloa

ded

from

ww

w.a

nnua

lrev

iew

s.or

gby

Roy

al M

elbo

urne

Ins

titut

e of

Tec

hnol

ogy

(RM

IT)

on 0

3/11

/13.

For

per

sona

l use

onl

y.

ANRV344-PL11-22 ARI 17 April 2008 13:15

decisions? To answer that question, we need tostudy deliberation that leads directly to bind-ing decisions.

Studies that examine opinion formationmore generally can also be relevant to thestudy of deliberative democracy. For ex-ample, some experiments suggest that thecross-cutting discussion favored by deliber-ative democrats may protect ordinary citi-zens against manipulation by elites. Individ-uals who discuss a political issue in “mixed”groups (in which the members have been ex-posed to conflicting perspectives on the issue)are less vulnerable to elite framing effects (thetendency to focus only on the subset of con-siderations that politicians and other leadersprefer to emphasize) (Druckman & Nelson2003). Thus, the larger democratic processthat ends in a collective decision includes mul-tiple stages and various sites. All may be rele-vant to deliberative theory and are worthy ofstudy, but they should be kept distinct so thattheir interrelationships and their role in theprocess as a whole can be more systematicallyinvestigated.

The legitimacy of the decision. Giventhese circumstances (the need for a collectivedecision in a state of disagreement), deliber-ative democracy seeks a conclusion that is le-gitimate. The criterion of legitimacy is notonly or mainly an empirical one. For a law tobe legitimate, it is neither necessary nor suf-ficient that most citizens feel that it is. But itis necessary that citizens take part in a processaimed at producing laws that are mutually jus-tifiable to all citizens. Hypothetical legitimacyis not sufficient. Thus the primary conceptualcriterion for legitimacy, and the most impor-tant distinguishing characteristic of deliber-ation, is mutual justification—presenting andresponding to reasons intended to justify a po-litical decision (Cohen 1989, 2007; Gutmann& Thompson 2004; Mendelberg 2002).

Theorists hold more or less expansive no-tions of this reasoning process (cf. Cohen2007 and Mansbridge 2007), but mostagree on its essential characteristics: public-

spiritedness, equal respect, accommodation,and equal participation.

Public-spirited reasoning is directed to-ward the collective good of the group that willbe bound by the decision, even if the reasonsalso refer to other goods. Assertions of powerare not justifications at all, and claims of selfinterest alone, though admissable, are not suf-ficient. Studies that distinguish arguing frombargaining, and identify deliberation with theformer, capture many of the relevant featuresof mutual justification (Holzinger 2005; Risse2000; Ulbert & Risse 2005), although mostrecognize that arguing and bargaining oftengo together in actual political discussion. Mosttheorists would include almost any kind of ap-peal, provided that it is not merely or finallybased on self or group interest. After all, eventhe philosophical versions of the theories areabout politics, not philosophy. Furthermore,the appeal beyond self interest does not haveto be sincere if it is plausible on the merits; ac-tual arguments are what matter, not motives(except insofar as the motives are predictorsof future arguments). More broadly, researchshould focus not on deliberative intentionsbut on institutional functions (Warren 2007,pp. 275–77). Empirical researchers thereforeshould not worry, as some evidently do, aboutformulating an independent test for sincer-ity or truthfulness (Bachtiger & Steiner 2005,pp. 162–64; Steiner et al. 2004, pp. 19–20, 56,166).

In addition to public-spiritedness, the rea-soning must show respect to the participantsand their arguments, even if it challenges thevalidity of the claims. In mutual justification,deliberators present their arguments in termsthat are accessible to the relevant audience,and respond to reasonable arguments pre-sented by opponents.

The requirement of accommodationmeans that the reasoning must keep openthe possibility of cooperation on other issues,even if the deliberators do not specificallypropose alternatives or initiate collaboration.

Equal participation requires that no oneperson or advantaged group completely

504 Thompson

Ann

u. R

ev. P

olit.

Sci

. 200

8.11

:497

-520

. Dow

nloa

ded

from

ww

w.a

nnua

lrev

iew

s.or

gby

Roy

al M

elbo

urne

Ins

titut

e of

Tec

hnol

ogy

(RM

IT)

on 0

3/11

/13.

For

per

sona

l use

onl

y.

ANRV344-PL11-22 ARI 17 April 2008 13:15

dominate the reason-giving process, even ifthe deliberators are not strictly equal in powerand prestige.

Notice that none of these requirements de-mands that deliberators use only pure reasonin their discussions. Most theorists regard af-fective appeals, informal arguments, rhetori-cal speeches, personal testimony and the likeas important ingredients in the deliberativeprocess. They do not assume that only argu-ments that would satisfy philosophers will orshould carry the day. One of the most citedexamples of successful deliberation involvesSenator Carol Moseley Braun’s highly emo-tional appeal that brought about the defeatof an amendment to renew the Daughtersof the Confederacy’s patent on their insignia(Gutmann & Thompson 1996, p. 135). Exag-gerating the kind of rationality that deliber-ative theory requires is a common mistake.Much as Schumpeter criticized a “classicaltheory of democracy” that no theorist actu-ally held (Pateman 1970), some critics nowattribute to deliberative theory assumptionsabout rational citizens that no theorist accepts(even as an ideal). Without attention to whattheorists actually write, empirical researchersrisk creating a caricature of the theory. Evena researcher as careful as Mutz (2008) doesnot always avoid this pitfall. In one sectionof her contribution to this volume, she criti-cizes deliberative theory—without citing anytheorists—for assuming that only the messageshould matter in deliberation, not the char-acteristics of the speaker and the listener orthe context in which the communication oc-curs. No major theorist makes such an as-sumption. Many explicitly address how factorsother than argument quality can (and should)bring about opinion change.

The normative requirements that charac-terize deliberation are intended to be rela-tively minimal criteria. By differentiating de-liberation from other kinds of discourse, theyare meant to isolate the practice that is thesubject of the inquiry. They do not includeall forms of political discussion, and they donot assume that the empirical consequences

of deliberation are necessarily beneficial. Thecriteria could be formulated somewhat differ-ently depending on the purposes of the in-quiry, but however formulated they shouldbe weaker than their corresponding evalua-tive standards, which impose more stringentdemands on deliberation.

Evaluative Standards

The evaluative standards provide a basis forjudging the quality of the deliberation identi-fied by the conceputal criteria. The closer theactual deliberation comes to meeting the stan-dards, the better it is in terms of deliberativetheory. The standards are sometimes calledideals because theorists assume that althoughthey guide actual discussion they can never befully realized (see Thompson 1970, pp. 45–51,86–119). But they should not be understood asvalues derived from ideal theory or from anyexternal theoretical source (Habermas 2005,p. 385). They are implicit in political practice,presupposed by the political communicationthat takes place in actual democracies. Evenwhen participants fail to meet the standards,their attempts to communicate acknowledgethe significance of the standards. The failures(or the partial successes) of the participantscan be adequately understood only in termsof the standards.

Although this claim about the implicationsof a practice may be largely theoretical (asit involves interpreting what a given prac-tice presupposes), empirical examination of aclosely related question could be illuminating.To what extent do participants themselves ex-plicitly endorse the standards? Speakers mayneed to presuppose some basic standards ofcommunication in order to have any discus-sion at all, but they do not have to presup-pose standards that are specifically delibera-tive. Some may come to a forum merely toadvocate their positions vigorously and tobring pressure on officials. When citizenswith that kind of aim come together inthe same forum with citizens who wish todeliberate, the result may frustrate the aim

www.annualreviews.org • Deliberative Democratic Theory and Empirical Political Science 505

Ann

u. R

ev. P

olit.

Sci

. 200

8.11

:497

-520

. Dow

nloa

ded

from

ww

w.a

nnua

lrev

iew

s.or

gby

Roy

al M

elbo

urne

Ins

titut

e of

Tec

hnol

ogy

(RM

IT)

on 0

3/11

/13.

For

per

sona

l use

onl

y.

ANRV344-PL11-22 ARI 17 April 2008 13:15

of deliberative democracy. This conflict be-tween different expectations (different viewsabout how the political discussion should beevaluated) is a fruitful area for further research(see Karpowitz 2003).

As more stringent versions of the concep-tual criteria, the evaluative standards may de-mand more of what the criteria require (e.g.,more frequent or sustained appeals to thecommon good), or a more robust form of whatthey require (e.g., a more active form of ac-commodation). The standards that apply tothe discourse itself are most plainly amplifica-tions of the conceptual criteria. The standardof public-spiritedness simply demands moreof the same kind of other-regarding reasoningthat characterizes the minimal form of delib-eration. (It does not rule out affective, rhetori-cal, or informal appeals.) To assess more of thecognitive content in the deliberation, someresearchers advocate a standard that focuseson the quality of the information the delib-erators use—for example, the extent to which“members of Congress rely on informed rea-sonable beliefs about. . .the impact of pro-posed policies” (Mucciaroni & Quirk 2006,p. 5).

As for equal respect, the evaluative stan-dard also simply requires more than itsconceptual counterpart. Deliberation rankshigher if more participants more often usearguments that the criterion emphasizes. Ac-commodation, as an evaluative standard, pre-scribes a positive effort toward cooperation,not merely avoiding actions that obstruct it,as required by the conceptual criterion. Itcorresponds to what some deliberative the-orists have called the “economy of moral dis-agreement,” in which citizens seek the ratio-nale that minimizes rejection of the positionthey oppose and try to find common groundon related policies (Gutmann & Thompson1996, pp. 84–94). This standard may be dif-ficult to operationalize, but some researchershave begun to develop empirical methods foridentifying and measuring what they deem“constructive politics” or “deliberative reci-procity,” which are intended to capture much

of the content of the idea of accommodation(Steiner et al. 2004, pp. 59–60, 107–9, 178–79;Weale et al. 2007).

The evaluative standard of equal partici-pation goes beyond its conceptual cousin. Itrefers to a stronger and somewhat differentphenomenon. The standard applies not onlyto the discourse itself but also to its distribu-tion. Equality is obviously a complex idea, andtheorists disagree about what kind of equalitythey think most important or relevant to de-liberation (Cohen 2007, Knight & Johnson1997, Thompson 2008). But most agree thatthe more the deliberation is influenced by un-equal economic resources and social status,the more deficient it is. That is because delib-erative democracy is based on a moral prin-ciple of reciprocity, a form of mutual respectthat requires treating citizens as equals (evenif, or especially if, they are not equal in power).Its justification is not primarily empirical, al-though the extent to which it is satisfied canbe investigated empirically. The general stan-dard of equality is applied both to the distribu-tion of membership in the deliberative bodyand to the patterns of participation in the de-liberation itself. Equal opportunity, randomselection, proportional representation, repre-sentative sampling, and equal time are amongthe versions of the standard that may be ap-plied to assess equality.

Whatever the standard, one of the mostconsistent empirical findings is that unlessspecial measures are taken, membership andparticipation are likely to be significantly un-equal (Delli Carpini et al. 2004, Mansbridge1980, Mendelberg 2002). Although critics re-peatedly brandish the findings of inequality todeclare deliberative democracy fatally flawed,most deliberative theorists are neither sur-prised nor discouraged. They believe that de-liberation itself can help expose unjust in-equalities in politics and that the findings canserve as a justification for leaders who wouldundertake special measures to counteract theinequalities—such as requiring proportionalrepresentation of disadvantaged groups in de-liberative bodies.

506 Thompson

Ann

u. R

ev. P

olit.

Sci

. 200

8.11

:497

-520

. Dow

nloa

ded

from

ww

w.a

nnua

lrev

iew

s.or

gby

Roy

al M

elbo

urne

Ins

titut

e of

Tec

hnol

ogy

(RM

IT)

on 0

3/11

/13.

For

per

sona

l use

onl

y.

ANRV344-PL11-22 ARI 17 April 2008 13:15

Empirical research that simply reinforcesthe general conclusion that deliberation fallsshort of the standards of equality is thereforenot very illuminating. Research that showsspecifically what conditions and changesmight mitigate inequality can be useful. Evenmore valuable, and less common, is researchcomparing the inequalities in deliberative fo-rums with the inequalities in other politicalsettings. Because so much of democratic pol-itics is pervaded by inequality, the more fun-damental question is comparative: To whatextent do deliberative forums satisfy variousstandards of equality more or less effectivelythan other political processes?

It might seem that these standards (andtheir corresponding criteria) are too compli-cated for fruitful empirical analysis, but sim-ilar and no less complex measures have beenadopted by some political scientists with il-luminating results. The most systematic at-tempt to operationalize principles for iden-tifying and evaluating deliberation is at thecenter of the study of parliamentary dis-course by Steiner and his associates. Morethan most empiricists, Steiner et al. (2004,pp. 52–61, 170–79) have made a serious andwell-informed effort to capture what theoristsregard as core elements of deliberation. Their“discourse quality index” comprises codingcategories intended to track principles theyfind in deliberative theory: Level of justi-fication (a reason, conclusion, and link be-tween them), content of justification (appealto common versus group interest), respecttoward groups to be helped (empathy), re-spect toward the demands of others (articu-lated regard for an opponent’s proposal or ar-gument), respect toward counterarguments (apositive statement about an opponent’s argu-ment against one’s conclusions), constructivepolitics (presentation of an alternative or me-diating proposal), and participation (absenceof interruptions).

Their categories measuring the level andthe content of discourse track well both thecriterion and standard of public-spirited rea-soning. If the categories are not found to some

degree (if they are coded 0 or 1), the discourseshould not count as deliberative at all, orshould not be regarded as sufficiently deliber-ative to be worthy of evaluation. Higher scoresqualify as better deliberation. The three “re-spect” categories are useful interpretations ofequal respect. The “constructive politics” cat-egory demands more than the minimal ac-commodation criterion and is better treatedas an evaluative standard. The “participation”category is a curious measure: It seems largelysubjective and culturally variable. It dependson whether the speaker thinks the interrup-tion is significant. A measure of speaking time(by gender, race, education) might be a moreuseful test of equal participation.

We need a second set of evaluative stan-dards in order to assess the effects on par-ticipants. Empirical studies have more to sayabout these effects than about most other as-pects of deliberative democracy because theeffects track familiar categories of social sci-ence research. It is relatively straightforwardto test whether deliberation (under certainconditions) increases political knowledge, asense of efficacy, and other standard sur-vey items. The very familiarity can lead re-searchers astray, however. From the perspec-tive of deliberative theory, knowledge of thepolitical views of other participants, for exam-ple, is as important as knowledge of issues. Ifyou are to respect your fellow participants—and even more if you are to be open to theirpersuasion—you have to understand theirviews and their reasons for holding them. Yetmost studies tend to concentrate on whetherparticipants are better informed about pub-lic policies, candidates’ positions, and govern-ment actions.

Another potential danger of familiarityis that the empirical tests may be mistak-enly identified with the similarly named butsubstantially different normative standards.Normative concepts of legitimacy (for exam-ple, a decision that cannot be reasonably re-jected by individuals seeking fair terms ofcooperation) are not the same as empirical le-gitimacy, which is sometimes called a sense of

www.annualreviews.org • Deliberative Democratic Theory and Empirical Political Science 507

Ann

u. R

ev. P

olit.

Sci

. 200

8.11

:497

-520

. Dow

nloa

ded

from

ww

w.a

nnua

lrev

iew

s.or

gby

Roy

al M

elbo

urne

Ins

titut

e of

Tec

hnol

ogy

(RM

IT)

on 0

3/11

/13.

For

per

sona

l use

onl

y.

ANRV344-PL11-22 ARI 17 April 2008 13:15

legitimacy (for example, a decision that is ac-cepted because of a favorable attitude towardor trust in government). A decision may be le-gitimate in the normative sense but lack em-pirical legitimacy. Conversely, a decision maybe normatively illegitimate but still perceivedas legitimate. These possibilities are worth ex-ploring as examples of potential conflicts ofthe kind discussed below. But the explorationcannot get started if the normative and empir-ical concepts are not carefully distinguishedwhen the evaluative standards are applied.

The third set of standards concerns theoutcome of the deliberation. The least prob-lematic of these standards—and those forwhich empirical research is likely to be themost useful in assessing outcomes—are sim-ply the composites of the standards that mea-sure the effects on individuals. For example,the extent and distribution of the change inpolitical knowledge in the group as a wholecan be presented and analyzed in any assess-ment of the effects on individuals. For cer-tain issues (those that turn mainly on mattersof fact and similarly objective consequences),empirical studies may help assess the qualityof the outcome. This approach would be mostrelevant to those theorists who believe that de-liberative democracy has substantial epistemicvalue (Estlund 1997, Marti 2006).

However, some other standards sometimesdeployed for assessing outcomes should beused for empirical research only with greatcaution, if they are used at all. The stan-dard most commonly invoked is consen-sus (Karpowitz 2003, Mendelberg & Oleske2000). Its problems are normative and empir-ical. First, there is no consensus among de-liberative theorists themselves that consensusshould be the goal of deliberation (Gutmann& Thompson 2004). Exposing and even in-tensifying disagreements may be desirable inmany circumstances. Second, it is difficult em-pirically to distinguish consensus from com-promise (Steiner et al. 2004, pp. 91–92). Someindirect evidence suggests that discussion maychange people’s minds and move a grouptoward greater agreement, but the effect is

not observed because it is latent and delayed(Mackie 2006). People are more likely to saythat they accept a decision as a compromisethan that they have changed their mind. Em-pirical evidence may help evaluate the extentto which deliberators change their minds, butthe goal should not be to determine whetherthe deliberation achieves consensus. (For amore favorable view of consensus as part ofa research strategy, see Neblo 2007a.)

Another outcome standard for which thehand of empirical research has not been veryhelping is the justice of the decision. Themost systematic study of the capacity of de-liberation to produce just outcomes in actualpolitical settings finds no significant relation-ship between the quality of the discourse (asmeasured by the index cited above) and weakegalitarian decisions (as indicated by the ex-tent to which they help the least well off)(Sporndli 2004). The outcomes seem to bebest explained by the pre-existing preferencesof the majority, which may suggest that thedistribution of power has a greater effect thanthe quality of the reasoning. But this studyis quite limited in scope. It is based on cod-ing speeches in 20 debates that took placefrom 1971 through 1982 in the 22-memberGerman Mediation Committee, an unusualconstitutional body that meets in private andmakes recommendations to resolve disputesbetween the two federal legislative chambers.Moreover, the criteria for weak egalitarianismare questionable. They are an exiguous inter-pretation of only one specific type of justiceand do not take into account whether a pro-posed policy is more or less egalitarian thanthe existing policy it would replace. More gen-erally, the empirical challenges of isolating theeffects of the deliberation on the justice of theoutcome are formidable because the interven-ing causes are even more numerous and com-plex than in the causal paths of the other ef-fects of deliberation.

A further normative problem complicatesempirical inquiry into the justice of outcomes.Deliberative theorists disagree not only onstandards of justice but also on whether

508 Thompson

Ann

u. R

ev. P

olit.

Sci

. 200

8.11

:497

-520

. Dow

nloa

ded

from

ww

w.a

nnua

lrev

iew

s.or

gby

Roy

al M

elbo

urne

Ins

titut

e of

Tec

hnol

ogy

(RM

IT)

on 0

3/11

/13.

For

per

sona

l use

onl

y.

ANRV344-PL11-22 ARI 17 April 2008 13:15

substantive standards should be part of delib-erative democracy (Gutmann & Thompson2004). Yet all agree that to some extent thenature of justice should itself be the subjectof deliberation. This creates what might becalled a problem of normative endogeneity.The standard used for evaluating the deliber-ative process is influenced by the process itself.What principle of justice should be accepted,and how it should be interpreted in particularcases, is supposed to be partly decided in thatprocess.

Empirical Conditions

The aspect of deliberation about which em-pirical inquiry has potentially the most to sayis the set of conditions that are necessary for,or at least contributory toward, good deliber-ation. As we have seen, good deliberation ismultifaceted, and any empirical inquiry intothe conditions that support it needs to be clearabout which evaluative standards are at issue.Composite standards (which combine severaldifferent measures of the quality of delibera-tion) may be appropriate for some purposes,but using separate standards is more useful foridentifying conflicts and trade-offs. This dis-aggregated approach is consistent with whatMutz (2008) in this volume calls middle-rangetheory—and has more in common with it thanshe acknowledges. But as indicated below, theapproach supports (what she and some othersneglect) the study of the structure of deliber-ative democracy beyond individual instancesand particular conditions of deliberation.

With respect to empirical conditions, a dis-aggregated approach can help determine theextent to which satisfying one standard entailsfalling short on another because each requiresincompatible empirical conditions. For exam-ple, the more that trained facilitators lead adiscussion, the better the quality of the dis-course and the more participants learn, butthe less equal the participation is likely to be.Bringing the deliberation closer to officialswho are actual decision makers (enhancingone of the circumstances of deliberation) can

generate another kind of inequality. Withoutcareful planning and strong independent con-trol, the officials tend to dominate; and whenthey do not, the citizens often simply use theoccasion to criticize the officials or to advo-cate their own cause (Button & Mattson 1999,Ryfe 2002, Weatherford & McDonnell 2007).More encouraging is an important ongoinginvestigation of the interaction of membersof Congress with their constituents (Esterlinget al. 2007). In discussions with their congres-sional representative about immigration pol-icy, constituents “gain knowledge that is usefulto make accountability effective.”

Clearly the range of possible empiricalconditions is large, and part of the challengeof empirical inquiry is to decide which areworth investigating. Examples of two types—institutional and cultural—can illustrate thechallenges that confront such research.

The most difficult empirical condition tospecify is equality. It refers to the resources,including talents, status, and power, that par-ticipants bring to the deliberation (as distinctfrom the membership and participation thatare the subject of the evaluative standards). Ifequality of resources were a requisite for de-liberation, then deliberative democracy wouldfail from the start. For many deliberative the-orists, one of the main points of deliberativedemocracy is to expose inequalities to pub-lic criticism and create less unjust conditionsin the future. They believe that deliberativepolitics can provide a more level playing fieldfor the disadvantaged because, compared tocompetitive or other forms of interest grouppolitics, it does not track so closely the exist-ing distributions of power in society. Still, un-equal resources are likely to produce unequalparticipation in the deliberative forum. Apartfrom the studies that use ascriptive character-istics as indicators, almost no empirical workinvestigates how great this effect is likely tobe. We do not even know whether it is true,as some theorists plausibly argue, that undermany conditions deliberation is less affectedby prevailing inequalities than power-basedmodes of decision making.

www.annualreviews.org • Deliberative Democratic Theory and Empirical Political Science 509

Ann

u. R

ev. P

olit.

Sci

. 200

8.11

:497

-520

. Dow

nloa

ded

from

ww

w.a

nnua

lrev

iew

s.or

gby

Roy

al M

elbo

urne

Ins

titut

e of

Tec

hnol

ogy

(RM

IT)

on 0

3/11

/13.

For

per

sona

l use

onl

y.

ANRV344-PL11-22 ARI 17 April 2008 13:15

The most discussed empirical condition ispublicity—the requirement that the delibera-tive forum be open to scrutiny by citizens ei-ther directly or through the media (Chambers2004, 2005). Deliberative theorists do not in-sist that all deliberation take place in public,only that the second-order decision to delib-erate in private be subject to public deliber-ation at some stage (Gutmann & Thompson1996, pp. 96, 104, 117). But most theoristsemphasize the salutary effects of making first-order arguments in public. They assume thatsuch arguments will be more public-spiritedand mutually respectful if made in the open.When speakers have to defend their propos-als and preferences before a large and diverseaudience, they are more likely to appeal tomore general principles and to take seriouslytheir opponents’ views (Benhabib 1996, p. 72;Cohen 1997 pp. 76–77; Elster 1998, p. 12;Goodin 1992, pp. 124–46). Even in interna-tional negotiations, the appeal to third par-ties that public deliberation makes possiblecan bring out “universal principles” (Ulbert& Risse 2005, pp. 358–59).

Yet empirical research has also confirmedwhat common sense suggests: In many cases,politicians who deliberate in private are moreinclined to make candid arguments, recog-nize complexities, and offer concessions (seeChambers 2004, 2005). Moreover, even if pri-vate discussions present more opportunitiesfor capture by special interests and for collu-sion among parties against the public interest,greater transparency often does not help, sim-ply because most citizens do not pay attention(Curtin 2006).

These mixed results (and assumptions)suggest that further research could help de-termine in which settings and for which is-sues publicity would promote or underminedeliberation. In conducting this research, theempirical condition of publicity must be dis-tinguished from the normative requirementthat deliberation be conducted in terms ofpublic reason. This requirement, the scope ofwhich is controversial among theorists, is aconceptual criterion of deliberation, as noted

above in the discussion of public-spiritedness.The two should be kept distinct because theempirical condition of publicity may affectthe extent to which the conceptual require-ment can be satisfied. It is important to knowwhether this hypothesis or its opposite is valid:The more public the discussion, the morelikely the participants are to use public rea-son, and the more likely the discussion is to bedeliberative.

In a thoughtful review of the theory and ev-idence on publicity, Chambers (2005, p. 256)suggests that in addressing such hypotheses,future research should adopt “a more nuancedidea of publicity and its effect on speakers.”She credits empirical approaches with show-ing the need to distinguish three kinds ofeffects that publicity may have on public rea-son. Publicity can promote (a) rationality—justifying one’s beliefs, articulating premisesand conclusions, taking account of oppos-ing points of view; (b) generality—appealingto the common good or the general inter-est; and (c) plebiscitary reason—appealing towhat seems to be the common good, but with“shallow, poorly reasoned pandering to theworst we have in common” (p. 260). Publicforums, she suspects, are more prone to irra-tionality and plebiscitary reason, whereas pri-vate discussions are more vulnerable to cap-ture by special interests and may not evenavoid plebiscitary reason completely (but seeSteiner et al. 2004, pp. 128–31, 165; and thecomment by Chambers 2005, p. 263).

The most significant analysis of institu-tional conditions at the level of the po-litical system is found in an ambitiousstudy of parliamentary discourse in Germany,Switzerland, the United Kingdom, and theUnited States (Steiner et al. 2004). Using thediscourse quality index cited above, Steinerand his colleagues found that “institutionaldesign matters for the quality of political dis-course” (p. 135). Issues make a difference:Deliberation is less successful when opinionis extremely polarized, as on the questionof abortion. But for many other importantissues, institutional conditions are significant.

510 Thompson

Ann

u. R

ev. P

olit.

Sci

. 200

8.11

:497

-520

. Dow

nloa

ded

from

ww

w.a

nnua

lrev

iew

s.or

gby

Roy

al M

elbo

urne

Ins

titut

e of

Tec

hnol

ogy

(RM

IT)

on 0

3/11

/13.

For

per

sona

l use

onl

y.

ANRV344-PL11-22 ARI 17 April 2008 13:15

Among the conditions favorable to deliber-ation are coalition cabinets, multiparty sys-tems, proportional representation, veto pro-visions, and second-chamber debates.

The most interesting conclusion for delib-erative theory is that the variation in the insti-tutional conditions has different effects on thedifferent aspects of deliberation (Steiner et al.2004). “Respect” is most affected by the con-ditions, “level of justification” draws mixedresults, and “constructive politics” shows lit-tle variation across institutional factors (ex-cept in the difference between first and sec-ond chambers). The authors speculate thatthe persistence of “positional politics” (ratherthan “constructive politics”) is due to generalfeatures of the legislative process. In other set-tings, such as civic forums where partisanshipis less prominent and initial positions have notcrystallized, the discourse may display moreconstructive attitudes. Even if this turns out tobe true, deliberative theorists may still be con-cerned about the lack of constructive politicsin legislatures, which are after all importantdecision-making bodies.

Cultural conditions are no doubt impor-tant but have only recently received seriousattention from researchers specifically inter-ested in deliberative democracy. Although de-liberation is less necessary to the extent thatthe participants agree on political issues to bedecided, deliberation may not be possible at allif the participants do not agree on the frame-work for discussion (if some believe for ex-ample that only violence can resolve the dis-putes). Some cultural consensus on the valueof settling disputes by mutual accommoda-tion is probably necessary. That would sug-gest deliberation is not possible in segmentedsocieties and in many international disputes,where the parties are divided by deep culturaldifferences about how to deal with fundamen-tal disagreements. Surprisingly, several recentstudies have shown that deliberation does takeplace in divided societies, and potentially ata higher level than many expect (O’Flynn2006). Similarly, several scholars contend thatthe favorable outcomes of some important in-

ternational negotiations cannot be explainedwithout reference to the efficacy of delibera-tion (“arguing” as distinct from “bargaining”)(Risse 2000, Schimmelfennig 2001, Ulbert &Risse 2005). Dryzek (2006) sees significantpossibilities for deliberation in large-scale in-ternational forums, although only to the ex-tent that the discussion takes place in interna-tional civil society at some distance from theexercise of sovereign power.

THE CONFLICTS INDELIBERATIVE DEMOCRACY

One of the most important reasons for disag-gregating the elements of deliberative democ-racy in the way suggested here is to exposepotential conflicts in its theory and practice.These are neglected by much of the theoryand research. There is a tendency, evident inmuch of the work cited above, to treat de-liberative democracy as a cohesive set of val-ues that are jointly realized or jointly fail tobe realized. The benefits of deliberation arepresumed to go together: As citizens engagein deliberation, they learn more about the is-sues, gain respect for opposing views, employmore public-spirited arguments, and so on.Or if citizens fail to deliberate, they learn less,disrespect more, pursue self-interested goals,and so on. We miss the complexity and powerof deliberative democracy if we do not recog-nize the possibility that its elements may con-flict with one another, that not all the goodsit promises can be secured at the same time,and that we have to make hard choices amongthem. We miss the opportunity to pursue em-pirical work more relevant to normative the-ory because one of the most fruitful contri-butions of empirical research is to expose thenature and extent of these conflicts.

The potential for this kind of contribu-tion is illustrated by the conflict betweentwo major values in deliberative theory—participation and deliberation itself (seeAckerman & Fishkin 2004, pp. 289–301;Cohen & Fung 2004). Among democratictheorists, the turn toward deliberative theory

www.annualreviews.org • Deliberative Democratic Theory and Empirical Political Science 511

Ann

u. R

ev. P

olit.

Sci

. 200

8.11

:497

-520

. Dow

nloa

ded

from

ww

w.a

nnua

lrev

iew

s.or

gby

Roy

al M

elbo

urne

Ins

titut

e of

Tec

hnol

ogy

(RM

IT)

on 0

3/11

/13.

For

per

sona

l use

onl

y.

ANRV344-PL11-22 ARI 17 April 2008 13:15

has not displaced participatory theory. Al-though elitist versions of deliberative the-ory look with suspicion on citizen involve-ment in decision making, most deliberativedemocrats favor greater participation by cit-izens, if not in the deliberation itself then atleast in judging the deliberation of their rep-resentatives. Rather than transcending partic-ipatory theory, many deliberative democratssee themselves as extending it. To the stan-dard list of political activities in which citizensparticipate—voting, organizing, protesting—they add deliberating.

The most common empirical challengeto participatory theory has taken the sameform as the challenge to deliberative theorymentioned above. The theory is unrealistic,critics say, because most citizens are not po-litical animals. They do not want to partic-ipate in politics, just as they do not want todeliberate about politics (Hibbing & Theiss-Morse 2002).This objection against partici-patory theory misses the point, just as it doesagainst deliberative theory. Participatory the-ory deplores the lack of participation in anycurrent political system, just as deliberativetheory condemns the lack of deliberation. Themore penetrating (and ultimately more con-structive) empirical challenge to any norma-tive theory seeks to show that the values thatit prescribes conflict in practice. The theoryfalters not because current democracies failto realize its values but because one of its val-ues cannot be fully realized without sacrificingone of its other values. Such a conflict is espe-cially disturbing if the principles are equallyindispensable to the theory. Standard trade-off techniques and pluralist approaches thenoffer no ready solution.

That conflict is the challenge that Mutz(2006) poses to deliberative theory in herstudy of political networks in the UnitedStates. She supplements her database of threenational surveys with comparative studies inother countries. An admirable feature of herbook is its selective engagement with delib-erative theory. She does not try to test thewhole theory—a “large package of variables

all rolled into one concept” (p. 6). Her ap-proach is disaggregated in the sense describedabove. She focuses instead on a specific nec-essary criterion of deliberation: the exposureto oppositional political perspectives throughpolitical talk. This cross-cutting exposure cre-ates what she calls diverse political networks,which satisfy one of the key conceptual crite-ria set out above for deliberation (a state ofdisagreement). Mutz finds that deliberationunder these circumstances provides some ofthe benefits that theorists hoped for: recogni-tion of the legitimacy of opposing viewpoints,greater tolerance, and greater empathy for po-litical opponents (pp. 84–86). Yet these delib-erative benefits come at a high participatoryprice. Her research suggests that the more cit-izens discuss politics with people whose viewsdiffer from theirs, the less likely they are toengage in political activity (pp.˜89–124). Themore they deliberate, the less they partici-pate. The moderate attitudes encouraged bydeliberation weaken some of the most power-ful incentives to participate. Opponents seemless like enemies; mobilizing to bring abouttheir defeat seems less urgent. Unlike citi-zens who talk mostly with like-minded com-patriots, deliberating citizens find themselvescross-pressured, and their views challengedrather than reinforced.

The conflict between participation and de-liberation does not of course express a univer-sal law. We do not yet know enough about howgeneral the conflict is—under what specificconditions it is more or less likely to appear.Is it more likely in discussions about certainkinds of issues? Is it more common in dis-cussions among ordinary citizens than amongpolitical leaders? Is it more frequent in infor-mal interactions than in formal institutions?Can forums for deliberation be structured inways to avoid or mitigate the conflict? Someavailable research bears on these questions,but political scientists have much more workto do before democratic theorists would giveup affirming both of these values. The em-piricist’s answer to the theorist’s general ques-tion should prompt the theorist to ask more

512 Thompson

Ann

u. R

ev. P

olit.

Sci

. 200

8.11

:497

-520

. Dow

nloa

ded

from

ww

w.a

nnua

lrev

iew

s.or

gby

Roy

al M

elbo

urne

Ins

titut

e of

Tec

hnol

ogy

(RM

IT)

on 0

3/11

/13.

For

per

sona

l use

onl

y.

ANRV344-PL11-22 ARI 17 April 2008 13:15

specific questions. The theorist needs the an-swers in order to evaluate how serious theconflict of values is, and what steps are worthtaking to overcome it. The helping hand cangesture in new directions.

This conflict is only one among manypossible tensions in deliberative theory thatwould repay further empirical and norma-tive analysis. Several are implicit in the earlierdiscussion of the elements of deliberation.Equal participation may lower the quality ofthe deliberative reasoning. Publicity may dothe same. Public deliberation may also beless conducive to mutual respect than pri-vate discussion. Decision-making authoritymay encourage polarization and positionalrather than constructive politics. Some fur-ther conflicts are suggested by recent stud-ies of institutional conditions. Consensus sys-tems (grand coalitions, multi-party structures,veto powers) tend to produce better delib-eration than competitive systems, but at thecost of less transparency in policy makingand less accountability of officials (Steineret al. 2004). More generally, “if the goal isrespectful deliberation that also entails argu-mentative change, then the actor relationshipsshould not be too competitive, actors shouldnot have bound mandates, bodies should besmall and explicitly geared toward rationaldiscussion and reflections. . .” (Bachtiger et al.2007, p. 98). This may not mean that deliber-ative democrats are forced to endorse a “pre-modern and gentlemen’s club model of poli-tics,” but it does suggest that they need to faceup to the tensions that empirical research ex-poses among their key values, and refine theirtheories to help decide the extent to whichone value should be sacrificed for another.

Empirical research thus may pose somechallenging questions and even offer someprovocative answers, but it does not have thelast word. Exposing conflicts among values isan important step, but it is closer to the begin-ning than the end of fruitful research into thenature of deliberative democracy. If participa-tion and deliberation (or any other key valuesin deliberative democracy) stand in conflict,

we still have to decide under what conditionswhich value should have priority, and whichcombination of the values is optimal. That de-cision depends partly on considerations thatare not primarily empirical—such as concep-tions of human dignity and understandings ofthe fair terms of social cooperation. The de-cision is usually not a matter of simple trade-offs but also entails a further normative choiceof the best method for dealing with the con-flicts (Goodin 1995). It is normative theorythat ultimately determines the significance ofany conflict and the appropriate mode of deal-ing with it. In that respect, it has the upperhand.

THE STRUCTURE OFDELIBERATIVE DEMOCRACY

Most of the empirical research on delibera-tive democracy has focused either on a singleepisode of deliberation, as in one-time groupdiscussions, or on a continuing series with thesame group or in the same type of institu-tion. This limitation is understandable. Thechallenges of conducting research on discretecases is formidable enough without attempt-ing to relate the findings to deliberation inother parts of the political system, let alone tonondeliberative practices in the system. Yetdeliberative theory is ultimately concernedwith the democratic process as a whole, andtherefore with the relationships of its parts tothe whole. The theory’s approach is decidedlynot “one size fits all,” nor even “deliberationall the way down.” Most deliberative theoristsrecognize not only that the practice of delib-eration may take different forms in differentparts of the process, but also that it is only oneof many desirable modes of decision making.Deliberative democracy is more than a sum ofdeliberative moments. Deliberative theoristsmake room for such activities as interest groupbargaining and political protests (Mansbridge2007), but most insist that their role—andthe form they take—be justified at some stagefrom a deliberative perspective (Gutmann &Thompson 2004, Habermas 2005).

www.annualreviews.org • Deliberative Democratic Theory and Empirical Political Science 513

Ann

u. R

ev. P

olit.

Sci

. 200

8.11

:497

-520

. Dow

nloa

ded

from

ww

w.a

nnua

lrev

iew

s.or

gby

Roy

al M

elbo

urne

Ins

titut

e of

Tec

hnol

ogy

(RM

IT)

on 0

3/11

/13.

For

per

sona

l use

onl

y.

ANRV344-PL11-22 ARI 17 April 2008 13:15

How deliberation should be allocatedwithin the democratic process thus becomesa key question. It is a question that has notreceived sufficient empirical or normative at-tention. Deliberative democrats should moreseriously “think about legitimacy across mul-tiple deliberative moments and the wider de-liberative system” (Parkinson 2006, p. 174).They need to confront more systematicallythe structural problem of the division of la-bor in deliberative democracy. It is possibleto study the structure with the middle-rangetheories that Mutz (2008) recommends, butonly if the theories are ultimately related tothe democratic process as a whole. This ef-fort is not to be confused with the testing ofa “grand theory.” It is simply to try to ensurethat research is relevant to a central concern ofpolitical science and political theory—the per-formance of the political system. That meansthat we must go beyond the study of the inter-actions of small groups considered separately,and examine how those interactions relate tothe larger political institutions. It also meansthat we need to know more about how delib-eration relates to (not only compares to) othermodes of decision making.

Three approaches to the structural prob-lem merit examination. The first is what hasbeen called distributed deliberation (Goodin2005). Different aspects of the “deliberativetask” are assigned to different institutions,which can then be held to different delibera-tive standards. Party caucuses are expected tosatisfy standards of candor; parliamentary de-bates are better at reasonable arguments; thepublic at large in elections supposedly comecloser to achieving the common good; andpostelection bargaining can reach mutuallyacceptable compromises. A similar division oflabor might be derived from the conclusionthat parliamentary committees are better atfostering mutual respect whereas plenary ses-sions are better at articulating the public in-terest (Steiner et al. 2004).

The advantage of the idea of distributeddeliberation is that, by emphasizing that notall institutions are equally effective at pro-

moting all aspects of deliberation, it invitesfurther research into the comparative advan-tage of each institution. But it also faces dif-ficulties. The various functions of delibera-tion cannot be kept as distinct in practice asthe model implies. Arguing and bargaining,for example, are difficult to distinguish em-pirically, and their segregation in separate in-stitutions is normatively questionable. Evenif a division of labor can be sustained, thereis the further difficulty of how the dispersedfunctions are to be coordinated to create a rec-ognizably deliberative system. Integrating thefunctions into a coherent structure remains achallenge for both normative theory and em-pirical research.

A second approach—decentralized delib-eration—avoids this difficulty by keeping thepractice of deliberation itself in one piece.Rather than distributing different deliberativefunctions to different institutions, it createsunified deliberative processes in many differ-ent bodies. Instead of dividing deliberation, itdivides the deliberators. The most prominentexample of this approach is the Porto AlegreParticipatory Budget (Baiocchi 2005, Fung2007b, Gastil & Levine 2005). Citizens ineach of the 16 districts of this Brazilian city as-semble annually to consider the budget priori-ties on policies ranging from sewage and hous-ing to health care and transportation. The re-sults of the “reflective preferences” generatedin these and other sessions are then aggre-gated into a single city budget.

The process certainly increases participa-tion and evidently produces genuine delib-eration about the issues in each district andneighborhood. The difficulty is that the de-liberation focuses on “very local goods andneeds” and does not dispose “citizens to thinkabout the greater good of the city, the justtrade-offs between jurisdictions or the goodof the city through the long arc of time” (Fung2007b, p. 179). The final budget is producedby an aggregative process, relegating delibera-tive democracy to a supporting role. If this ap-proach is to be viable, its proponents must findmore effective ways to encourage a broader

514 Thompson

Ann

u. R

ev. P

olit.

Sci

. 200

8.11

:497

-520

. Dow

nloa

ded

from

ww

w.a

nnua

lrev

iew

s.or

gby

Roy

al M

elbo

urne

Ins

titut

e of

Tec

hnol

ogy

(RM

IT)

on 0

3/11

/13.

For

per

sona

l use

onl

y.

ANRV344-PL11-22 ARI 17 April 2008 13:15

perspective in the local deliberations and tointegrate the decentralized bodies into a delib-erative process at central levels of the politicalsystem. (For a discussion of how mini-publicscan influence decisions in the larger politicalsystem, see Goodin & Dryzek 2006.)

Iterated deliberation, the third approach,addresses this problem of integration. Italso exemplifies one of the most importantfeatures of deliberative democracy—its dy-namic capacity for self-correction (Gutmann& Thompson 2004). The process of iterateddeliberation typically takes this form: A po-litical body (which may or may not be de-liberative) proposes a policy to a deliberativebody, which returns a revised version of thepolicy to the original body. That body re-vises the policy again and submits it for fur-ther consideration to the deliberative bodybefore it is enacted. This loop may continuethrough multiple phases and may be expandedto include other institutions. If the institutionshave different capacities, iterated deliberationcan reap the benefits of the division of laborin distributed deliberation without the costs.

The process for setting health care priori-ties in Oregon in the early 1990s is sometimescited as an example of iterated deliberation.(For various interpretations of that process,see Fung 2007b, Gutmann & Thompson1996.) The state’s Health Services Commis-sion created a priority list of conditions andtreatments. After the list was widely criticized,the commission consulted with a number ofcitizen bodies and other institutions, some ofwhich were deliberative. A substantially re-vised list was considered by the commissionand then by the legislature. The final resultfell short of what many would have liked, butit was better than the earlier proposals. Theprocess itself was certainly flawed in many re-spects, but it too was by deliberative standardsan improvement over what had gone before.

The general question remains, however:how to incorporate the need for expertiseand technical administration in a delibera-tive democracy (see Richardson 2003). Giventhe potential of iterated deliberation (and the

fact that it mirrors one of the key character-istics of deliberative democracy), we need tolearn more about the conditions under whichit works well. For which issues is it more or lesssuitable, which institutions should play whatroles in promoting it, and what is the optimumpoint for bringing it to closure?

All three of these approaches to dividingdeliberative labor (distributive, decentralized,and iterated deliberation) imply ambitiousprojects of institutional design. How shouldthe designers choose their approach? The nat-ural answer is: deliberatively, through a pro-cess that might be called meta-deliberation.Some deliberative theorists distinguish thepractice of deliberation from the conceptionof deliberation. They do not insist that everypractice in deliberative democracy be delib-erative but rather that every practice shouldat some point in time be deliberatively justi-fied (Gutmann & Thompson 2004, Macedo1999). The question of the place of delibera-tion in the larger process should be open todeliberative challenge itself.

If this approach seems more an abstractidea than an institutional proposal, considerthe citizens’ assemblies that are being usedto reform the electoral system in severalprovinces in Canada, and in the Netherlands.Some 160 citizens, chosen more or lessrandomly, met weekends for nearly ninemonths in 2004–2005 in British Columbiato decide whether to recommend replacingthe current majoritarian system with someversion of a proportional system. Eventhough the question was technically complexand potentially divisive, the members ofthis Citizens’ Assembly, by all accounts,managed to deliberate effectively (Blais et al.2008, Thompson 2008, Warren & Pearse2008). The members of the Assembly werenot deliberating directly about the role ofdeliberation in their political system, butthey were deliberating about institutionalchanges that could substantially affect thatrole. If such assemblies can successfullydiscuss and decide about electoral systems,they may also be capable of considering other

www.annualreviews.org • Deliberative Democratic Theory and Empirical Political Science 515

Ann

u. R

ev. P

olit.

Sci

. 200

8.11

:497

-520

. Dow

nloa

ded

from

ww

w.a

nnua

lrev

iew

s.or

gby

Roy

al M

elbo

urne

Ins

titut

e of

Tec

hnol

ogy

(RM

IT)

on 0

3/11

/13.

For

per

sona

l use

onl

y.

ANRV344-PL11-22 ARI 17 April 2008 13:15

issues of institutional design. They couldprovide a partial answer to the question of howto decide deliberatively what place delibera-tion should have in deliberative democracy.

A citizens’ assembly does not entirely es-cape the general problems inherent in thedivision of deliberative labor. The delibera-tion in the assembly is quite different fromthe deliberation in the public, who in theBritish Columbia case ultimately vote in areferendum on the assembly’s recommenda-tions. (The assembly’s deliberation is also dif-ferent from the deliberation in the legislature,which in other circumstances may be the fi-nal decision-making body.) Members of an as-sembly engage, on relatively equal terms, in aprocess that the electorate can never hope tomatch. Members reach conclusions for rea-sons that most ordinary voters are not likely tofully appreciate. Designed to reduce the gapbetween citizens and experts, the process it-self can reproduce the problem that it was in-tended to overcome. But the deliberative gapmay be partly bridged if voters are prepared totrust the judgment of members. If the work ofthe members is made accessible, voters candeliberate about the fairness of the processrather than its results. The normative ques-tion is whether this is an adequate form of de-liberation, and the empirical question is underwhat conditions citizens are likely to engageeffectively in this form of deliberation.

Deliberative theorists have only begun toanalyze the possibilities of meta-deliberationin such institutions as the citizens’ assembly.Any normative assessment will need to beinformed by the full repertoire of empiricalinquiry, including case studies, interviewswith participants, opinion surveys, andcontent analysis. The key question here, aswith all the approaches to dealing with theproblem of the division of labor in delibera-tive democracy, is what are the most effectiveand desirable relationships among the variousbodies that operate within the structure ofdeliberative democracy—those designed todeliberate, as well as those constituted todecide in other ways.

CONCLUSION

The study of deliberative democracy, like itspractice, has tended to observe a division oflabor. The division of deliberative labor mayor may not serve the practice well—that isone of the questions flagged above for fur-ther inquiry. But it is clear that a rigid divi-sion between normative and empirical inquirydoes not serve the study of the practice well atall—that is one of the conclusions this reviewsuggests. Normative and empirical inquiry aredistinct, and justifiably so. Their methods andagendas diverge, appropriately so. But ourunderstanding of deliberative democracy willfall short until theorists and empiricists takegreater steps to bridge this division. Somemight say that unless philosophers becomepolitical scientists, or unless those who noware political scientists become philosophers,there can be no end to troubles in our disci-pline; only then will the theory of deliberativedemocracy see the light of day. Indeed, someof the most valuable recent work on delibera-tive democracy is being produced by youngerscholars (such as Bachtiger, Fung, and Neblo)who are as proficient in normative theory asin empirical social science.

But even with the division of labor that islikely to persist, collaboration can still go for-ward constructively if theorists and empiri-cists systematically engage with each other’swork. The questions each side poses can stim-ulate answers that raise further questions,which in turn identify new problems—an iter-ative program of research that would mirrorthe deliberative process itself. Such a programis more likely to be enlightening if theoristsand empiricists proceed with a common ana-lytic understanding of the elements of delib-eration, a deeper appreciation of the conflictswithin deliberative democracy, and greater at-tention to its structure. They need to study thedeliberation that takes place not only amongcitizens but also between citizens and theirrepresentatives, and among representativesthemselves. Theory and empirical researchmight then more often progress hand in hand.

516 Thompson

Ann

u. R

ev. P

olit.

Sci

. 200

8.11

:497

-520

. Dow

nloa

ded

from

ww

w.a

nnua

lrev

iew

s.or

gby

Roy

al M

elbo

urne

Ins

titut

e of

Tec

hnol

ogy

(RM

IT)

on 0

3/11

/13.

For

per

sona

l use

onl

y.

ANRV344-PL11-22 ARI 17 April 2008 13:15

DISCLOSURE STATEMENT

The author is not aware of any biases that might be perceived as affecting this review.

LITERATURE CITED

Ackerman B, Fishkin J. 2004. Deliberation Day. New Haven, CT: Yale Univ. PressBachtiger A, Sporndli M, Steenbergen M, Steiner J. 2007. Deliberation in legislatures: an-

tecedents and outcomes. See Rosenberg 2007a, pp. 82–100Bachtiger A, Steiner J. 2005. Introduction. Acta Polit. 40:153–68Baiocchi G. 2003. Participation, activism, and politics: the Porto Alegre experiment. See Fung

& Wright 2003, pp. 45–76Barabas J. 2004. How deliberation affects policy opinions. Am. Polit. Sci. Rev. 98:687–701Benhabib S, ed. 1996. Democracy and Difference: Contesting the Boundaries of the Political. Prince-

ton, NJ: Princeton Univ. PressBesson S, Marti JL, eds. 2006. Deliberative Democracy and its Discontents. Hampshire, UK:

AshgateBlais A, Carty RK, Fournier P. 2008. Do citizens’ assemblies make reasoned choices? See

Warren & Pearse 2008, pp. 127–44Bohman J. 1998. Survey article: the coming of age of deliberative democracy. J. Polit. Philos.

6:400–25Bohman J, Rehg W, eds. 1997. Deliberative Democracy: Essays on Reason and Politics. Cambridge,

MA: MIT PressButton M, Mattson K. 1999. Deliberative democracy in practice: challenges and prospects for

civic deliberation. Polity 31:609–37Chambers S. 1996. Reasonable Democracy: Jurgen Habermas and the Politics of Discourse. Ithaca,

NY: Cornell Univ. PressChambers S. 1998. Contract or conversation: theoretical lessons from the Canadian constitu-

tional crisis. Polit. Soc. 26:143–72Chambers S. 2004. Behind closed doors: publicity, secrecy and the quality of deliberation.

J. Polit. Philos. 12:389–410Chambers S. 2005. Measuring publicity’s effect. Acta Polit. 40:255–66Cohen J. 1989. Deliberation and democratic legitimacy. In The Good Polity: Normative Analysis

of the State, ed. A Hamlin, P Pettit, pp. 17–34. Cambridge, UK: Basil BlackwellCohen J. 1997. Deliberation and democratic legitimacy. See Bohman & Rehg 1997, pp. 67–91Cohen J. 2007. Deliberative democracy. See Rosenberg 2007a, pp. 219–36Cohen J, Fung A. 2004. Radical democracy. Swiss J. Polit. Sci. 10:23–43Conover PJ, Searing DD. 2005. Studying ‘everyday political talk’ in the deliberative system.

Acta Polit. 40:269–83Cook FL, Delli Carpini MX, Jacobs LR. 2007. Who deliberates? Discursive participation in

America. See Rosenberg 2007a, pp. 25–44Curtin D. 2006. Framing public deliberation and democratic legitimacy in the European Union.

See Besson & Marti 2006, pp. 133–56della Porta D. 2005. Deliberation in movement: why and how to study deliberative democracy

and social movements. Acta Polit. 40:336–50Delli Carpini MX, Cook FL, Jacobs LR. 2004. Public deliberation, discursive participation,

and citizen engagement: a review of the empirical literature. Annu. Rev. Polit. Sci. 7:315–44Druckman JN, Nelson KR. 2003. Framing and deliberation: how citizens’ conversations limit

elite influence. Am. J. Polit. Sci. 47:729–45

www.annualreviews.org • Deliberative Democratic Theory and Empirical Political Science 517

Ann

u. R

ev. P

olit.

Sci

. 200

8.11

:497

-520

. Dow

nloa

ded

from

ww

w.a

nnua

lrev

iew

s.or

gby

Roy

al M

elbo

urne

Ins

titut

e of

Tec

hnol

ogy

(RM

IT)

on 0

3/11

/13.

For

per

sona

l use

onl

y.

ANRV344-PL11-22 ARI 17 April 2008 13:15

Dryzek JS. 2007. Theory, evidence, and the tasks of deliberation. See Rosenberg 2007a,pp. 237–50

Dryzek JS. 2006. Deliberative Global Politics. Cambridge, UK: PolityDryzek JS, Braithwaite V. 2000. On the prospects for democratic deliberation: values analysis

applied to Australian politics. Polit. Psychol. 21:241–66Elster J, ed. 1998. Deliberative Democracy. Cambridge, UK: Cambridge Univ. PressEsterling KM, Neblo MA, Lazer DMJ. 2007. Means, motive & opportunity in becoming informed

about politics: a deliberative field experiment. Presented at Annu. Meet. Am. Polit. Sci. Assoc.,Chicago, Aug. 30–Sep. 2. Available online as a Program on Networked Governance work.pap., http://wwwksg.harvard.edu/netgov/files/png workingpaper series/PNG07-006.pdf

Estlund D. 1997. Beyond fairness and deliberation. See Bohman & Rehg 1997, pp. 173–204Fishkin J. 1999. Toward a deliberative democracy: experimenting with an ideal. In Citizen

Competence and Democratic Institutions, ed. S Elkin, KE Soltan, pp. 279–90. State College:Penn. State Univ. Press

Fishkin JS, Laslett P, eds. 2003. Debating Deliberative Democracy. Oxford, UK: BlackwellFishkin JS, Luskin RC. 2005. Experimenting with a democratic ideal: deliberative polling and

public opinion. Acta Polit. 40:284–98Fung A. 2007a. Democratic theory and political science: a pragmatic method of constructive

engagement. Am. Polit. Sci. Rev. 101:443–58Fung A. 2007b. MiniPublics. See Rosenberg 2007a, pp. 159–83Fung A, Wright EO. 2003. Deepening Democracy: Institutional Innovations in Empowered Partici-

patory Governance. London and New York: VersoGastil J, Levine P. 2005. The Deliberative Democracy Handbook. San Francisco: Jossey-BassGoodin RE. 1992. Motivating Political Morality. Oxford, UK: BlackwellGoodin RE. 1995. Political ideals and political practice. Br. J. Polit. Sci. 25:37–56Goodin RE. 2005. Sequencing deliberative moments. Acta Polit. 40:182–96Goodin RE, Dryzek JS. 2006. Deliberative impacts: the macro-political uptake of mini-publics.

Polit. Soc. 34:219–44Gutmann A, Thompson D. 1996. Democracy and Disagreement. Cambridge, MA: Harvard Univ.

PressGutmann A, Thompson D. 2004. Why Deliberative Democracy? Princeton, NJ: Princeton Univ.

PressHabermas J. 1984. The Theory of Communicative Action. 2 vols. Transl. T McCarthy (from

German). Boston: BeaconHabermas J. 1989. The Structural Transformation of the Public Sphere. Cambridge, MA: MIT

PressHabermas J. 1996. Between Facts and Norms: Contributions to a Discourse Theory of Law and

Democracy. Transl. W Rehg (from German). Cambridge, MA: MIT PressHabermas J. 2005. Concluding comments on empirical approaches to deliberative politics. Acta

Polit. 40:384–92Habermas J. 2006. Does democracy still enjoy an epistemic dimension? Commun. Theory

16:411–26Hibbing JR, Theiss-Morse E. 2002. Stealth Democracy: Americans’ Beliefs about How Government

Should Work. Cambridge, UK: Cambridge Univ. PressHolzinger K. 2005. Context or conflict types: which determines the selection of communication

mode. Acta Polit. 40:239–54Jackman S, Sniderman PM. 2006. The limits of deliberative discussion: a model of everyday

political arguments. J. Polit. 68:272–83

518 Thompson

Ann

u. R

ev. P

olit.

Sci

. 200

8.11

:497

-520

. Dow

nloa

ded

from

ww

w.a

nnua

lrev

iew

s.or

gby

Roy

al M

elbo

urne

Ins

titut

e of

Tec

hnol

ogy

(RM

IT)

on 0

3/11

/13.

For

per

sona

l use

onl

y.

ANRV344-PL11-22 ARI 17 April 2008 13:15

Janssen D, Kies R. 2005. Online forums and deliberative democracy. Acta Polit. 40:317–35Karpowitz CF. 2003. Public hearings and the dynamics of deliberative democracy: a case study. Pre-

sented at Annu. Meet. Midwest Polit. Sci. Assoc., ChicagoKnight J, Johnson J. 1997. What sort of equality does deliberative democracy require? See

Bohman & Rehg 1997, pp. 279–320Macedo S, ed. 1999. Deliberative Politics: Essays on Democracy and Disagreement. Princeton, NJ:

Princeton Univ. PressMackie G. 2006. Does democratic deliberation change minds? Polit. Philos. Econ. 5:279–303Mansbridge J. 1980. Beyond Adversary Democracy. New York: Basic BooksMansbridge J. 2007. “Deliberative democracy” or “democratic deliberation.” See Rosenberg

2007a, pp. 251–71Marti JL. 2006. The epistemic conception of deliberative democracy defended. See Besson &

Marti 2006, pp. 27–56Mendelberg T. 2002. The deliberative citizen: theory and evidence. In Research in Micropolitics:

Political Decisionmaking, Deliberation and Participation, ed. MX Delli Carpini, L Huddy, RShapiro, 6:151–93. Greenwich, CT: JAI Press

Mendelberg T, Oleske J. 2000. Race and public deliberation. Polit. Commun. 17:169–91Mucciaroni G, Quirk PJ. 2006. Deliberative Choices: Debating Public Policy in Congress. Chicago:

Univ. Chicago PressMutz DC. 2006. Hearing the Other Side: Deliberative vs Participatory Democracy. Cambridge and

New York: Cambridge Univ. PressMutz DC. 2008. Is deliberative democracy a falsifiable theory? Annu. Rev. Polit. Sci. 11:521–38Nanz P, Steffek J. 2005. Assessing the democratic quality of deliberation in international

governance: criteria and research strategies. Acta Polit. 40:368–83Neblo MA. 2005. Thinking through democracy: between the theory and practice of deliberative

politics. Acta Polit. 40:169–81Neblo MA. 2007a. Family disputes: diversity in defining and measuring deliberation. Swiss

Polit. Sci. Rev. 13:527–57Neblo MA. 2007b. Change for the better? Linking the mechanisms of deliberative opinion

change to normative theory. Presented at Annu. Meet. Int. Soc. Polit. Psychol., Toronto,Jul. 2005. Work. pap. available at http://polisci.osu.edu/faculty/mneblo/papers.htm

O’Flynn I. 2006. Deliberative Democracy and Divided Societies. New York: Palgrave MacmillanParkinson J. 2006. Deliberating in the Real World: Problems of Legitimacy in Deliberative Democracy.

Oxford, UK: Oxford Univ. PressPateman C. 1970. Participation and Democratic Theory. Cambridge, UK: Cambridge Univ. PressRichardson H. 2003. Democratic Autonomy: Public Reasoning about the Ends of Policy. New York:

Oxford Univ. PressRisse T. 2000. Let’s argue! Communicative action in world politics. Int. Organ. 54:1–39Rosenberg S, ed. 2007a. Deliberation, Participation and Democracy: Can the People Govern? Bas-

ingstoke, UK: Palgrave MacmillanRosenberg S. 2007b. Types of discourse and the democracy of deliberation. See Rosenberg

2007a, pp. 130–58Ryfe DM. 2002. The practice of deliberative democracy: a study of sixteen organizations. Polit.

Commun. 16:359–78Ryfe DM. 2005. Does deliberative democracy work? Annu. Rev. Polit. Sci. 8:49–71Schimmelfennig F. 2001. The community trap: liberal norms, rhetorical action, and the Eastern

enlargement of the European Union. Int. Organ. 55:47–80

www.annualreviews.org • Deliberative Democratic Theory and Empirical Political Science 519

Ann

u. R

ev. P

olit.

Sci

. 200

8.11

:497

-520

. Dow

nloa

ded

from

ww

w.a

nnua

lrev

iew

s.or

gby

Roy

al M

elbo

urne

Ins

titut

e of

Tec

hnol

ogy

(RM

IT)

on 0

3/11

/13.

For

per

sona

l use

onl

y.

ANRV344-PL11-22 ARI 17 April 2008 13:15

Sporndli M. 2004. Diskurs und Entscheidung. Eine empirische Analyse kommunikativen Handelns imdeuatschen Vermittlungsausschuss. PhD diss., Univ. Bern. Weisbaden: VS Verlag fur Soz.wiss.Summarized in Steiner et al. 2004, pp. 157–62

Steiner J, Bachtiger A, Sporndli M, Steenbergen M. 2004. Deliberative Politics in Action: Cross-national Study of Parliamentary Debates. Cambridge, UK: Cambridge Univ. Press

Sulkin T, Simon A. 2001. Habermas in the lab: a study of deliberation in an experimentalsetting. Polit. Psychol. 22:809–26

Sunstein C. 2002. The law of group polarization. J. Polit. Philos. 10:175–95Thompson DF. 1970. The Democratic Citizen: Social Science and Democratic Theory in the 20th

Century. Cambridge, UK: Cambridge Univ. PressThompson DF. 2008. Who should govern who governs? The role of citizens in reforming the

electoral system. See Warren & Pearse 2008, pp. 20–49Ulbert U, Risse T. 2005. Deliberately changing the discourse: What does make arguing effec-

tive? Acta Polit. 40:351–67Walsh KC. 2003. Talking about Politics: Informal Groups and Social Identity in American Life.

Chicago: Univ. Chicago PressWalsh KC. 2007. The democratic potential of civic dialogue. See Rosenberg 2007a, pp. 45–63Warren ME. 2007. Institutionalizing deliberative democracy. See Rosenberg 2007a, pp. 272–88Warren M, Pearse H. 2008. Designing Deliberative Democracy: The British Columbia Citizens’

Assembly. Cambridge, UK: Cambridge Univ. Press. In pressWeale A, Bicquelet A, Bara J. 2007. Debating abortion, deliberative reciprocity and parliamentary

advocacy. Work. pap., Dep. Gov., Univ. EssexWeatherford MS, McDonnell LM. 2007. Deliberation with a purpose: reconnecting commu-

nities and schools. See Rosenberg 2007a, pp. 184–215

520 Thompson

Ann

u. R

ev. P

olit.

Sci

. 200

8.11

:497

-520

. Dow

nloa

ded

from

ww

w.a

nnua

lrev

iew

s.or

gby

Roy

al M

elbo

urne

Ins

titut

e of

Tec

hnol

ogy

(RM

IT)

on 0

3/11

/13.

For

per

sona

l use

onl

y.

AR344-FM ARI 8 April 2008 18:44

Annual Review ofPolitical Science

Volume 11, 2008Contents

State FailureRobert H. Bates � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � �1

The Ups and Downs of Bureaucratic OrganizationJohan P. Olsen � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � 13

The Relationships Between Mass Media, Public Opinion, and ForeignPolicy: Toward a Theoretical SynthesisMatthew A. Baum and Philip B.K. Potter � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � 39

What the Ancient Greeks Can Tell Us About DemocracyJosiah Ober � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � 67

The Judicialization of Mega-Politics and the Rise of Political CourtsRan Hirschl � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � 93

Debating the Role of Institutions in Political and EconomicDevelopment: Theory, History, and FindingsStanley L. Engerman and Kenneth L. Sokoloff � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � �119

The Role of Politics in Economic DevelopmentPeter Gourevitch � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � �137

Does Electoral System Reform Work? Electoral System Lessons fromReforms of the 1990sEthan Scheiner � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � �161

The New Empirical BiopoliticsJohn R. Alford and John R. Hibbing � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � �183

The Rule of Law and Economic DevelopmentStephan Haggard, Andrew MacIntyre, and Lydia Tiede � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � �205

Hiding in Plain Sight: American Politics and the Carceral StateMarie Gottschalk � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � �235

Private Global Business RegulationDavid Vogel � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � �261

Pitfalls and Prospects in the Peacekeeping LiteratureVirginia Page Fortna and Lise Morje Howard � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � �283

v

Ann

u. R

ev. P

olit.

Sci

. 200

8.11

:497

-520

. Dow

nloa

ded

from

ww

w.a

nnua

lrev

iew

s.or

gby

Roy

al M

elbo

urne

Ins

titut

e of

Tec

hnol

ogy

(RM

IT)

on 0

3/11

/13.

For

per

sona

l use

onl

y.

AR344-FM ARI 8 April 2008 18:44

Discursive Institutionalism: The Explanatory Power of Ideasand DiscourseVivien A. Schmidt � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � �303

The Mobilization of Opposition to Economic LiberalizationKenneth M. Roberts � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � �327

CoalitionsMacartan Humphreys � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � �351

The Concept of Representation in Contemporary Democratic TheoryNadia Urbinati and Mark E. Warren � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � �387

What Have We Learned About Generalized Trust, If Anything?Peter Nannestad � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � �413

Convenience VotingPaul Gronke, Eva Galanes-Rosenbaum, Peter A. Miller, and Daniel Toffey � � � � � � � � �437

Race, Immigration, and the Identity-to-Politics LinkTaeku Lee � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � �457

Work and Power: The Connection Between Female Labor ForceParticipation and Female Political RepresentationTorben Iversen and Frances Rosenbluth � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � �479

Deliberative Democratic Theory and Empirical Political ScienceDennis F. Thompson � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � �497

Is Deliberative Democracy a Falsifiable Theory?Diana C. Mutz � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � �521

The Social Processes of Civil War: The Wartime Transformation ofSocial NetworksElisabeth Jean Wood � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � �539

Political Polarization in the American PublicMorris P. Fiorina and Samuel J. Abrams � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � �563

Indexes

Cumulative Index of Contributing Authors, Volumes 7–11 � � � � � � � � � � � � � � � � � � � � � � � � � �589

Cumulative Index of Chapter Titles, Volumes 7–11 � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � �591

Errata

An online log of corrections to Annual Review of Political Science articles may be foundat http://polisci.annualreviews.org/

vi Contents

Ann

u. R

ev. P

olit.

Sci

. 200

8.11

:497

-520

. Dow

nloa

ded

from

ww

w.a

nnua

lrev

iew

s.or

gby

Roy

al M

elbo

urne

Ins

titut

e of

Tec

hnol

ogy

(RM

IT)

on 0

3/11

/13.

For

per

sona

l use

onl

y.