creating better citizens? effects of a model citizens ... · ested in this type or level of...

23
Creating Better Citizens? Effects of a Model Citizens’ Assembly on Student Political Attitudes and Behavior JOSEPH GERSHTENSON GLENN W. RAINEY, JR. JANE G. RAINEY Eastern Kentucky University Perceiving political engagement to be dangerously low among American citizens, many political science professors in recent years have attempted to promote engage- ment and ‘‘healthier’’ political attitudes. The effectiveness of these efforts appears variable and generally quite modest. Following the model of Canadian citizens’ assemblies, we taught a course called Citizens’ Assembly on Critical Thinking about the United States (CACTUS) in spring 2008 in which students considered the ques- tion: ‘‘Is it time to change the way we elect the President of the United States?’’ Because the course employs a form of deliberative democracy CACTUS might be anticipated to encourage engagement. We use a pre-post survey design to measure attitudes of both CACTUS (treatment group) and other (comparison groups) stu- dents to examine this. We find that both CACTUS and students enrolled in other political science courses experienced modest growth in their political engagement. More notably, CACTUS students became more extreme in their party identifi- cation, ideology, and issue positions and became more supportive of the existing electoral system. We suspect these findings are attributable to the nature and content of CACTUS. Our findings have important implications for future efforts to promote political engagement and for measuring the effects of those efforts. Keywords citizens’ assemblies, civic engagement, deliberative democracy, political engagement Masses make noise, citizens deliberate; masses behave, citizens act; masses collide and intersect, citizens engage, share and contribute. At the moment when masses start deliberating, acting, sharing, and contributing, they cease to be masses and become citizens. Only then do they ‘‘participate.’’ (Barber 1984, 155) How to transform masses into citizens in an age of declining political engage- ment is widely discussed by both politicians and political scientists. Among the pro- posed solutions is deliberative democracy both as means and end: Masses become citizens by learning how to deliberate and practice deliberation as a part of political life. Advocates of deliberative democracy assume that by participating in such Address correspondence to Joseph Gershtenson, Eastern Kentucky University, 113 McCreary Hall, 521 Lancaster Avenue, Richmond, KY 40475-3102. E-mail: joe.gershtenson@ eku.edu Journal of Political Science Education, 6:95–116, 2010 Copyright # 2010 Taylor & Francis Group, LLC ISSN: 1551-2169 print=1551-2177 online DOI: 10.1080/15512161003708129 95

Upload: others

Post on 09-Sep-2020

3 views

Category:

Documents


0 download

TRANSCRIPT

Page 1: Creating Better Citizens? Effects of a Model Citizens ... · ested in this type or level of political participation (e.g., Hibbing and Theiss-Morse 2002), and more empirical research

Creating Better Citizens? Effects of a ModelCitizens’ Assembly on Student Political

Attitudes and Behavior

JOSEPH GERSHTENSONGLENN W. RAINEY, JR.JANE G. RAINEY

Eastern Kentucky University

Perceiving political engagement to be dangerously low among American citizens,many political science professors in recent years have attempted to promote engage-ment and ‘‘healthier’’ political attitudes. The effectiveness of these efforts appearsvariable and generally quite modest. Following the model of Canadian citizens’assemblies, we taught a course called Citizens’ Assembly on Critical Thinking aboutthe United States (CACTUS) in spring 2008 in which students considered the ques-tion: ‘‘Is it time to change the way we elect the President of the United States?’’Because the course employs a form of deliberative democracy CACTUS might beanticipated to encourage engagement. We use a pre-post survey design to measureattitudes of both CACTUS (treatment group) and other (comparison groups) stu-dents to examine this. We find that both CACTUS and students enrolled in otherpolitical science courses experienced modest growth in their political engagement.More notably, CACTUS students became more extreme in their party identifi-cation, ideology, and issue positions and became more supportive of the existingelectoral system. We suspect these findings are attributable to the nature andcontent of CACTUS. Our findings have important implications for future effortsto promote political engagement and for measuring the effects of those efforts.

Keywords citizens’ assemblies, civic engagement, deliberative democracy,political engagement

Masses make noise, citizens deliberate; masses behave, citizens act; massescollide and intersect, citizens engage, share and contribute. At the momentwhen masses start deliberating, acting, sharing, and contributing, theycease to be masses and become citizens. Only then do they ‘‘participate.’’(Barber 1984, 155)

How to transform masses into citizens in an age of declining political engage-ment is widely discussed by both politicians and political scientists. Among the pro-posed solutions is deliberative democracy both as means and end: Masses becomecitizens by learning how to deliberate and practice deliberation as a part of politicallife. Advocates of deliberative democracy assume that by participating in such

Address correspondence to Joseph Gershtenson, Eastern Kentucky University, 113McCreary Hall, 521 Lancaster Avenue, Richmond, KY 40475-3102. E-mail: [email protected]

Journal of Political Science Education, 6:95–116, 2010Copyright # 2010 Taylor & Francis Group, LLCISSN: 1551-2169 print=1551-2177 onlineDOI: 10.1080/15512161003708129

95

Page 2: Creating Better Citizens? Effects of a Model Citizens ... · ested in this type or level of political participation (e.g., Hibbing and Theiss-Morse 2002), and more empirical research

deliberative exercises as ‘‘citizen juries,’’ citizens become more politically aware andknowledgeable, developing ‘‘healthier’’ political attitudes such as increased faith intheir ability to understand politics and to participate meaningfully, as well asincreased faith in the responsiveness of governmental officials to public concernsand increased trust in government.

In turn, the system is strengthened by a knowledgeable citizenry, motivatedmore by communal thinking and less by pure self-interest (Harriger and McMillan2007). Colby et al. (2007, 27) find that ‘‘participating in deliberations about politicaloutcomes leads people to discard inaccurate perceptions of the facts and rigidly heldpolitical views, and may yield a greater sense of the overall legitimacy oracceptability of the ultimate outcomes, even if one personally disagrees with them.’’Still, the deliberative vision is not immune from criticism ranging from concernsabout ‘‘groupthink’’ and impracticality to indicators that the public are not inter-ested in this type or level of political participation (e.g., Hibbing and Theiss-Morse2002), and more empirical research is needed to determine the credence of the bodyof normative theory about and exhortatory praise of deliberative democracy.

There is an accumulating body of knowledge about what can happen to atti-tudes about participation, efficacy, trust, and other indicators of political engage-ment as a result of deliberative practice in an academic setting. For example, theAPSA Standing Committee on Civic Education and Engagement found deliberativeactivities to be a potentially promising way to contribute to the creation of a morepolitically engaged citizenry. Indeed they defined the most ‘‘mature’’ engagementas that which is deliberative (Macedo 2005, 61), and they called on Americans tolearn from other countries. Taking up this challenge, we report in this article on adeliberative exercise at Eastern Kentucky University modeled on actual citizen-empowering experiments in Canada.

The Canadian Models

A particularly innovative effort to implement deliberative democracy occurred inCanada in 2004 when the government of British Columbia convened, mandatedand funded a Citizens’ Assembly on Electoral Reform. The process consisted oflearning, public hearings, and deliberation phases extending over 10 months. The161 members, chosen through a process of random and self-selection, met biweeklyon spring and fall weekends with public hearings in the summer. The process wasfollowed by a binding public referendum on the Assembly’s recommendations forchanging the electoral system. The process was later replicated in Ontario (see:‘‘British Columbia Citizens’ Assembly on Electoral Reform’’ n.d.; ‘‘Ontario Citizens’Assembly on Electoral Reform’’ n.d.). These citizens’ assemblies, created to addressboth electoral reform and declining civic engagement, were grounded in an aware-ness of contemporary theory and practice involving deliberative democracy, andtheir formats reflected the deliberative style praised by advocates of deliberativedemocracy as a ‘‘higher order’’ of citizen involvement (Rainey and Rainey 2006,2008; Warren and Pearse 2008).

Citizens’ Assembly for Critical Thinking about the United States

At Eastern Kentucky University the Canadian model was adapted for classroom usein the form of the Citizens’ Assembly for Critical Thinking about the United States

96 J. Gershtenson et al.

Page 3: Creating Better Citizens? Effects of a Model Citizens ... · ested in this type or level of political participation (e.g., Hibbing and Theiss-Morse 2002), and more empirical research

(CACTUS). Taught first in spring, 2008, CACTUS can count toward generaleducation, the political science major, or a university elective. Any students in theuniversity are permitted to enroll as the course has no prerequisites. In contrast withother student deliberative exercises that have been staged in as little as one to a fewdays with few informational components (e.g., Morrell 2005), it is a full semestercourse, meeting weekly for 16 weeks.

CACTUS was developed as part of the university’s Quality Enhancement Planto develop ‘‘informed, critical, and creative thinking’’ and the ability to ‘‘communi-cate effectively’’ but also as a deliberative arena for students to develop politicalknowledge, participatory skills, and commitment. Disagreement and deliberationoccurred in a structured, moderated format intended to keep the group on task.The 28 students addressed the question, ‘‘Is it time to change the way we elect thePresident of the United States?’’ They studied models and proposals for electingthe President of the United States and recommended whether the current processshould be retained or another method should be adopted CACTUS Web site, n.d.).

The first third of the semester consisted of a learning phase during whichstudents studied the presidential electoral process via lectures and readings on theElectoral College and on deliberative democracy and were examined in a regularclass. Small group sessions, facilitated by political science students, allowed fordiscussion, clarification, and weighing of trade-offs. The learning phase (and sub-sequent phases) also included a discussion board on which students were requiredto post comments. Instructors and facilitators assumed a neutral stance and guestspeakers presented a range of perspectives.

During a public hearings phase in mid-semester other students and faculty statedtheir opinions before panels of CACTUS members. Each CACTUS member sat on apanel of four or five students and listened to several speakers, with the remainder ofthe class as the audience. A public discussion board was also launched on theCACTUS Web page.

The last third of the semester was the deliberation phase during which theAssembly worked together to build two models—a reformed Electoral Collegemodel and a popular vote alternative. The purpose of this format was to keep every-one working together throughout, rather than letting the exercise become an adver-sarial debate. After deliberating, the Assembly voted 18-8 in favor of their reformmodel, and after further deliberation, voted 18-7 to recommend replacing the currentElectoral College with their reform model. Lastly, they wrote a Final Report and areferendum question. Via campus-wide e-mail balloting, the CACTUS reform pro-posal won approval 266-114.

No such experiment takes place in a vacuum, and other factors may have con-tributed to CACTUS (and other) student outcomes. Four deserve mention. First wasthe 2008 Presidential primary campaign (which led to a discussion of the nominatingprocess in CACTUS) with growing ‘‘Obamamania’’ among young people. Second, apolitical science professor took students from her presidency class, including severalCACTUS students, to South Carolina for several days to work in South Carolina’sDemocratic primary. They attended parties, shook candidates’ hands and engagedin other emotionally compelling activities. Third, for the first time in years theKentucky Democratic presidential primary, coming in late May, promised to makean impact on the nomination. Finally, several CACTUS students joined in localefforts to save a city recreation area from encroachment by an RV camp. Theyreported on this at CACTUS meetings, urging others to get involved.

Effects of a Model Citizens’ Assembly 97

Page 4: Creating Better Citizens? Effects of a Model Citizens ... · ested in this type or level of political participation (e.g., Hibbing and Theiss-Morse 2002), and more empirical research

Though much smaller and less diverse than its real-world Canadian counter-parts, CACTUS drew from a variety of majors and class levels. Only eight memberswere political science majors. The time committed to CACTUS sessions was muchless than the Canadian models but spread over 16 weeks it did allow for intensiveperiods of learning and of deliberation. The challenge of building CACTUS mem-bers into a deliberative ‘‘community’’ was greater because their meeting space wasnot physically set apart from their daily campus environment, but conscious effortswere made to compensate for this. The ‘‘real’’ Citizens’ Assemblies had the empow-ering component of the promise of binding public referenda on their recommenda-tions. CACTUS students had the motivation of a grade plus a desire to prevail inthe university referendum.1

Civic and Political Engagement: Conceptual and Measurement Issues

Political engagement is routinely used as an umbrella term for a variety of behaviorsand attitudes; an umbrella so broad that it sometimes overlaps with just about every-thing people do—‘‘skills, behaviors, knowledge, values, attitudes, motivations,dispositions and self-perceptions (p. 57)’’ in one application (Ehrlich 2000a). Rel-evant indicators include political knowledge and interest, attitudes toward politicsand government (e.g., trust=cynicism, support, involvement in public affairs, as wellas interest in news, political knowledge, voting, anticipated interest in running forpolitical office or working in political campaigns, and both internal and externalpolitical efficacy), as well as subjective sense of the importance of communityinvolvement, and actual involvement in volunteer or community service work,among others (Ehrlich 2000b; Elder, Seligson, and Hofrenning 2007; Furlong andScheberle 2006; Huerta and Jozwiak 2008).2

Attempts to explain fundamental attitudes such as political trust and efficacyand citizen interest and participation in government affairs typically produce modesteffects and weak explanatory models with coefficients of determination often in arange from .05 to .30 (e.g., Hetherington 1998; Mishler and Rose 2001; Hibbingand Theiss-Morse 2002). Associations suggest complex interrelationships that maybe superficially paradoxical. Increased knowledge, for example, has sometimes beenassociated with decreased trust (e.g., Delli Carpini and Keeter 1996; Harriger andMcMillan 2007). The impact of policy and institutional factors may be greater forparticipation than for trust (Kenski and Stroud 2006). And while overall economicconditions can significantly contribute to explanatory models of trust (Hetherington1998) the relationships are not consistent. While substantial data support the argu-ment of Hibbing and Theiss Morse that a large, uninterested majority of citizens are‘‘fixated on themselves’’ (Hibbing and Theiss-Morse 2002, 80), other results fromcontrolled analysis show they are often more sociotropic than egocentric in theirreactions to economic issues (e.g., Mishler and Rose 2001), and the significantimpacts of economic conditions must be juxtaposed against overall negative reac-tions that occur in positive economic climates (e.g., Hibbing and Theiss-Morse2002, 63–64). Clearly, there are powerful forces at work that we do not fullyunderstand.

It is therefore not surprising that attempts to promote civic engagement inhigher education have also tended to produce modest and=or narrowly focusedeffects, and those approaches that do appear to produce some significant impactsare typically intensive in labor, resources, and=or content. In addition to the

98 J. Gershtenson et al.

Page 5: Creating Better Citizens? Effects of a Model Citizens ... · ested in this type or level of political participation (e.g., Hibbing and Theiss-Morse 2002), and more empirical research

extended engagement projects already discussed, examples include extended fieldexperiences such as participation in primaries and elections (e.g., Elder, Seligson,and Hofrenning 2007), school-wide or nation-wide exercises in dialogue or civicaction (Longo, Drury, and Battistoni 2006), or enhanced class reading and analysisapproaches, such as the New York Times project at Texas A&M (Huerta andJozwiak 2008). While education in general has been associated with increasedpolitical interest and participation, there is some evidence that political scienceand policy-related courses are associated with larger changes; although difficultyestablishing experimental controls has made it hard to measure the effects ofself-selection among the students (e.g., Beaumont et al. 2006; Furlong and Scheberle2006).

Even the effects produced by relatively intensive approaches may producechanges only in particular attitudinal or behavioral niches. Involving students inprimaries, for example, appeared to engender a substantial increase in students’interest in news and events and more modest increases in expressed interest inrunning for office but produced only mixed changes in political efficacy and no‘‘gains’’ at all in political trust or attitudes toward political officials, or in belief thatpolitical involvement can solve major issues and produce tangible results (Elder,Seligson, and Hofrenning 2007). Using intensive reading and discussion of newspa-pers in classes appeared to produce increased interest in news and more positivesubject attitudes about the importance of ‘‘community involvement’’ but did notproduce more ‘‘positive attitudes toward politics’’ (Huerta and Jozwiak 2008). Onthe other hand, reviewing the outcomes of the 21 courses and programs under theaegis of the Political Engagement Project, Beaumont et al. (2006) found significantincreases among students across four different dimensions of civic engagement,moderate for those with low prior political interest and weak for those that enteredwith political interest already high. At the same time, the authors were relieved thatthe experiences of the students produced no significant changes in ideological orpartisan identity, thereby sparing the academic community from accusations ofindoctrinating the nation’s youth. But this also raises the question of what exactlythe students are getting when they become more politically engaged without everquestioning or changing their own ideological or partisan identities—perhaps littlemore than engagement for its own sake.

The long-term value of political engagement for its own sake, divorced from theresolution of specific issues of governance, may be quite dubious. If university stu-dents tend to respond to civic engagement exercises with increased interest in obser-vation (news) and process (elections) but see no hope of resolving substantive issuesor of producing trustworthy government, it is behavior consistent with more generalfindings. A substantial majority of all adults appear to be characterized by substan-tial rigidity and self-fixation as a result of the enduring effects of socialization (DelliCarpini and Keeter 1996; Kam and Palmer 2008; Kerlin 2005) and by a natural tend-ency to avoid confrontation over substantive matters and to delegate that chore to‘‘empathetic, non-self-interested decision makers’’ (p. 161). (Hibbing and Theiss-Morse 2002), and even their genetic attributes (Fowler and Dawes 2008). The con-ditions under which generalized as opposed to niche impacts on political attitudesand behaviors may be fostered is therefore a crucial issue for those who wouldattempt to engineer civic engagement. It speaks directly to the question of whetherlogical and analytical processes can be used to overcome political rigidity andself-fixation. Nonetheless, students of such behavior do not entirely despair. Hibbing

Effects of a Model Citizens’ Assembly 99

Page 6: Creating Better Citizens? Effects of a Model Citizens ... · ested in this type or level of political participation (e.g., Hibbing and Theiss-Morse 2002), and more empirical research

and Theiss-Morse, for example, suggest (2002, 225) that ‘‘exposing students to therange of issue interests of people across the United States and simulations illustratingthe challenges of coming to agreement in the face of divided opinion is whatwe need.’’

CACTUS is designed to pursue such exposure through its combination ofacademic study and group-based resolution of a prominent process or policy issue.In accordance with the concerns expressed by Beaumont et al. (2006, 266), inclusionof the assembly as a course in the general education curriculum, with no prerequi-sites, is intended to encourage broad representation from the student body, withthe hope that significant numbers of participants will not already be ‘‘primed’’ forparticipation. In assessing the outcomes, our attention is focused on both changesin specific niche attitudes—e.g., attitudes toward electoral systems—and more gener-ally enabling changes in such dimensions as political efficacy, knowledge, trust andcynicism, and intended or actual participation, and their interaction with basic socialidentity as reflected in demographic backgrounds. These latter attitudes also seem tobe dimensions whose importance is emphasized by Beaumont et al. (2006). Finally,comparison of the outcomes between CACTUS and other classes will allow at leastsome degree of ‘‘before and after’’ comparison across different treatments.

Research Questions

The issue chosen for the pilot administration, whether to change the method ofelecting the President of the United States (implicitly meaning whether to keep, toamend, or to abolish the Electoral College), is a process issue, nominally familiarand important to a wide segment of the public, especially in the light of the contro-versy over the 2000 election and subsequent discontent with the Bush administration.We would expect it to have appealed and offered a certain psychological comfort tothose participants whom Hibbing and Theiss-Morse might characterize as processfocused. However, we also had to wonder whether it would encourage narrowlyfocused interest and situation-specific efficacy (e.g., Morrell 2005), despite theextended learning period, and whether increased knowledge will contribute to inter-nal but not external efficacy, as Delli Carpini and Keeter found (1996, 183). Or,would there be spillover effects – i.e., will ‘‘understanding the process’’ lead peopleto ‘‘be more sympathetic’’ (Hibbing and Theiss-Morse 2002, 68) in a broader rangeof political interests and behavior?

For this initial overview of the effects of CACTUS and other courses ourresearch questions are focused on determining whether significant ‘‘before and after’’changes in key variables of interest did in fact occur, whether they support inferencesregarding the differential impact of CACTUS and the comparison courses, andwhether there are reasons to suspect that these impacts are general as opposed tosituation specific. More specifically we ask:

. Are there any significant changes in political interest, political knowledge, internalefficacy, external efficacy, trust=cynicism, and intent to participate, and are thoseeffects mediated by variables such as ideology and partisanship?

. Are there any significant changes in attitudes toward the specific object(s) of theengagement exercise—elections and the Electoral College?

. Is there significant variation in the direction and magnitude of changes experi-enced by CACTUS students compared to other students?

100 J. Gershtenson et al.

Page 7: Creating Better Citizens? Effects of a Model Citizens ... · ested in this type or level of political participation (e.g., Hibbing and Theiss-Morse 2002), and more empirical research

. Did the CACTUS experience affect, either positively or negatively, the parti-cipants’ tolerance for confrontation and negotiation over issues (albeit, in thisinitial iteration, a process issue)?

Data and Methods

To examine the extent to which participation in CACTUS led students to become‘‘better’’ citizens in terms of their political attitudes and behavior, we employ aquasi-experimental (field experiment) design. Specifically, we use a pre-post designin which attitudes are measured through survey responses by both CACTUS (treat-ment group) and other (comparison) groups. Our initial research design envisionedcomparing CACTUS students to a sample of approximately 500 students drawn ran-domly from the general student population. Despite the offer of a cash incentive, wewere unable to obtain this control group; students simply did not show up to com-plete the survey. (Anecdotal evidence suggests that this is a common occurrence indoing research on a college population.) Thus, we turned to our colleagues to gen-erate other groups of students for comparisons. Ultimately, we were successful indeveloping two comparison groups. The first consists of students enrolled in bothintroductory and upper-division political science courses and therefore includes stu-dents ranging from our finest majors to those taking the course simply to fulfill arequirement. The second comparison group is students from an upper-division for-ensics course (comprised exclusively of forensic science majors) that was team-taughtby a faculty member from the Department of Government and a colleague fromanother department.

The nature of our design permits us to evaluate changes in the attitudes and beha-viors of CACTUS students and to compare changes among these students to changesexperienced by students in other political science courses and to students in the for-ensics course (most of whom have never had a political science course in college).We offer, however, two notes of caution with regard to this design. First, as notedabove, we do not have a classic experimental design with random assignment of sub-jects to treatment and control groups. Rather, we have students self-selecting intogroups (by virtue of enrolling in the relevant courses), and, especially since one ofthe groups is comprised solely of individuals in an upper-division class of a specializedand technically oriented major, we might observe systematic differences between ourgroups. Table 1 describes the demographic characteristics of the three groups. Asshown, there are no statistically significant differences between CACTUS studentsand the other political science students. On the other hand, the forensics studentsdo exhibit some meaningful differences: They lack racial diversity, are predominantlyfemale, have higher grade point averages and tend to be older and farther along intheir academic careers than other students. Given these results, we feel more confidentin regarding the students enrolled in other political science courses as a ‘‘control’’group than we do with the forensics students. Yet, including forensics students inthe design offers some potentially interesting comparisons. For example, given theirmore advanced standing and the fact that their course has no explicit political contentor focus, we might anticipate that the forensics students would experience less changein their political attitudes than would students in the other groups.

Our design also requires some caution with regard to conclusions about caus-ality. As opposed to a pure experiment conducted in an isolated lab setting, thisstudy took place in the context of a broader environment with students taking other

Effects of a Model Citizens’ Assembly 101

Page 8: Creating Better Citizens? Effects of a Model Citizens ... · ested in this type or level of political participation (e.g., Hibbing and Theiss-Morse 2002), and more empirical research

courses, interacting with other individuals, working, etc. Perhaps most notably, thestudy took place during a period in which there was a relatively hotly contestedbattle for the Democratic presidential nomination.

Surveys were administered the first week of classes in the spring 2008 semesterand again during the last two weeks of the semester. The two surveys included anumber of identical items because we are interested in a change in attitudes. Weare particularly interested in ‘‘democratic’’ attitudes, expecting that CACTUS stu-dents may experience greater ‘‘growth’’ in such attitudes relative to their counter-parts in other political science courses and the forensics course. While we wroteseveral of our own survey items, many were drawn directly from otherwell-established political surveys, most notably the American National ElectionStudies (NES). For the sake of clarity and simplification, we group the items intothree categories.

Table 1. Characteristics of different student groups

Variable (range)CACTUSstudents

POLstudents

FORstudents

Year in School(1–4)

Mean 2.360 2.205 3.6921

Std Deviation 1.287 1.106 0.471n 25 190 26

Age (17–50) Mean 20.44 21.46 22.582

Std Deviation 2.47 4.95 3.95n 25 189 26

GPA (1–8) Mean 5.960 6.419 7.2311

Std Deviation 1.457 1.263 0.710n 25 186 26

Gender (0–1) Mean 0.320 0.437 0.7311

Std Deviation 0.476 0.497 0.452n 25 190 26

Race (Black)(0–1)

Mean 0.040 0.047 0.03

Std Deviation 0.200 0.213 0.0n 25 190 26

Family Income(1–7)

Mean 4.895 5.350 4.813Std Deviation 1.629 1.553 1.328n 19 143 16

Live on Campus(0–1)

Mean 0.560 0.505 0.423Std Deviation 0.507 0.501 0.504n 25 190 26

Childhood Politics(1–5)

Mean 3.440 3.137 2.808Std Deviation 1.044 1.187 1.357n 25 190 26

Notes: Cell entries are for all students in the specified type who completed both pre- andpostsurveys. Significance tests are two-tailed independent sample t tests. Text of the surveyitems is available on request.

1Significantly different from both CACTUS and POL students at p< .05.2Significantly different from CACTUS students at p< .05.3Significantly different from POL students at p< .05.

102 J. Gershtenson et al.

Page 9: Creating Better Citizens? Effects of a Model Citizens ... · ested in this type or level of political participation (e.g., Hibbing and Theiss-Morse 2002), and more empirical research

The first category most directly taps into ‘‘democratic’’ attitudes and includes apolitical interest index, a political knowledge index, internal and external efficacyindexes, a question asking what percent of the time the student trusts the nationalgovernment, and items about voter registration and the likelihood of voting in theNovember 2008 election.3 The second category of items relates to the extent to whichrespondents exhibit extreme versus moderate attitudes. Specifically, we examinecommitment to party identification by seeing how strongly Democratic or Republi-can students are, and how committed they are to self-identified ideology (howstrongly liberal or conservative) and to ideological extremity as gauged by positionson four specific policy issues: reducing government spending by providing fewer ser-vices (including health and education), support for defense spending, support for agovernment-sponsored insurance plan, and the government’s perceived responsi-bility in ensuring that all individuals have jobs and good standards of living.4

The final category of items taps into students’ attitudes toward the electoral sys-tem. It includes two items related to assessments of the electoral system as a whole:whether elections are basically fair even if the outcomes are not always those desiredby the respondent and whether the system of elections needs a lot of improvement. Italso has three items specifically regarding presidential elections: whether the Presi-dent should be chosen by popular vote, whether the Electoral College is a fair way toelect the President, and whether the Electoral College is better than alternatives evenif it is imperfect. Finally, we include in this category an item asking students theimportance of seeking consensus in a democracy. Given the nature of CACTUS(procedurally there is some emphasis on achieving consensus) and the depth of focuson elections, particularly Presidential elections, we anticipate that there may begreater differences between CACTUS participants and other students on these items.

As outlined, our research questions ask if students become ‘‘better’’ citizens bytaking political science courses and if students enrolled in CACTUS experience moregrowth in this regard than do other students. To answer these questions, we examinechange in average student responses from the pre- to the postsurvey and differencesbetween CACTUS and other students. In addition, because CACTUS students maysystematically differ from other students in ways that affect political attitudes, we useregression analysis to examine potential variation in growth of ‘‘democratic’’ valueswhen controlling for other possible determinants.

Results

Several broad themes emerge from the results of our empirical analyses. First, wewitness general normatively desired changes in students’ political attitudes fromthe beginning to the end of the semester: Students do appear to become ‘‘better’’ citi-zens by virtue of taking college courses. On the other hand, a second theme is thatthis growth by students is not uniform. Students experience greater changes on someattitudinal dimensions than they do on others, and there is variation between stu-dents enrolled in political science courses and those in the forensics course. In gen-eral, students taking political science courses experienced greater growth in their‘‘democratic’’ attitudes than did forensics students. Another theme is that attitudinalchanges and differences across students may be depressed by both ‘‘ceiling’’ effectsand by sample sizes. CACTUS students generally entered the semester with higherlevels of political engagement and more positive political attitudes and thereforehad less room for growth. In addition, the small number of students in CACTUS

Effects of a Model Citizens’ Assembly 103

Page 10: Creating Better Citizens? Effects of a Model Citizens ... · ested in this type or level of political participation (e.g., Hibbing and Theiss-Morse 2002), and more empirical research

(and in the forensics course) means that it is more difficult to witness change that isstatistically significant even if the change seems to be substantively meaningful. Afinal theme is that our results point to the existence of some targeted or specificeffects. Thus, while all political science courses may produce ‘‘better’’ citizens inmany regards, CACTUS, presumably by virtue of its focus, had a greater impacton attitudes toward the electoral system.

Table 2 illustrates several of these themes. It shows, first, general changes in nor-matively desired directions for the political interest, political knowledge, and internaland external efficacy indexes. Students became more interested, knowledgeable, andefficacious over the course of the semester. However, we also witness substantialvariation in these changes. Forensics (FOR) students actually experienced (statisti-cally significant) declines in both their political interest and external efficacy. Thetable also highlights variation in changes for both CACTUS and other politicalscience (POL) students. Both groups experienced far greater growth in politicalknowledge than in the other indexes and both experienced the least growth withregard to external efficacy. While CACTUS and other political science studentsmoved in the same directions on all four of these indexes, the magnitudes of thesechanges were not identical. Perhaps most notably, despite entering the semester withhigher levels of political interest and internal efficacy (and thus being more suscep-tible to ceiling effects), CACTUS students experienced more growth in these dimen-sions (5.38% and 4.81%, respectively) than did their counterparts (3.12% and 3.76%,respectively).5 Ceiling effects are even more evident with regard to the political par-ticipation variables. Quite simply, both CACTUS and the other political sciencecourses began the semester with such large numbers of students indicating that theywere registered to vote and were definitely going to vote that substantive and=orstatistically significant increases on these variables were virtually impossible.

As mentioned above, increases in external efficacy were smaller for bothCACTUS and other political science students than were increases in interest, knowl-edge, and internal efficacy. Thus, as students learn more about the political systemthey are not necessarily inclined to view public officials and the government as sig-nificantly more responsive to citizens. These findings on external efficacy suggestthat political science courses do not have the effect of encouraging ‘‘blind’’ faithin the government. This assessment is supported by the findings regarding politicaltrust. All students experienced some decline in their trust of both the federal andstate governments, and among CACTUS students (who began the semester withlower levels of external efficacy and trust than did other students) these declines wereboth statistically and substantively significant.

Unlike political interest, political knowledge, political efficacy (particularlyinternal), and political participation, normative assessments regarding partisanshipand ideology are much less straightforward. What does it mean to be a ‘‘better’’ citi-zen? Should one be more partisan, more ideological, and more extreme on issues orshould one be more moderate? Regardless of one’s answer to these queries, the effectof college courses on students’ partisanship and ideology is an interesting and openquestion. As shown in Table 3, both CACTUS students and others became morecommitted in terms of their party identification. However, the absolute and percent-age magnitude of this extremitization for CACTUS students was far greater thanit was for others (and the change is statistically significant only for CACTUSstudents). Similarly, CACTUS students became significantly more extreme, by a full15.76%, in their self-professed ideologies and in their positions on specific policy

104 J. Gershtenson et al.

Page 11: Creating Better Citizens? Effects of a Model Citizens ... · ested in this type or level of political participation (e.g., Hibbing and Theiss-Morse 2002), and more empirical research

issues. These changes stand in contrast to other students. Students in non-CACTUSpolitical science and forensics courses became less extreme in their ideologies. And,while other political science students experienced a statistically significant increase in

Table 2. Political interest, knowledge, efficacy, trust, and participation

Variable (range)CACTUSstudents

POLstudents

FORstudents

Political Interest(3–17)

Pre 12.640 11.016 9.923Post 13.320! 11.360!! 9.308!

Percent Change 5.38 3.12 "6.20n 25 188 26

Political Knowledge(0–10)

Pre 6.227 5.784 5.889Post 7.027!!! 6.603!!! 6.551!!

Percent Change 12.85 14.15 11.25n 25 188 24

Internal Efficacy(3–21)

Pre 15.800 14.610 14.308Post 16.560 15.159!!! 14.500Percent Change 4.81 3.76 1.34n 25 186 26

External Efficacy(5–35)

Pre 19.160 20.027 21.846Post 19.560 20.444 20.077!!!

Percent Change 2.09 2.09 "8.10n 25 185 26

Registered to Vote(0–1)

Pre 0.960 0.863 0.923Post 0.960 0.899! 0.962Percent Change 0.00 4.15 4.17n 25 188 26

Likelihood of Voting(1–5)

Pre 4.800 4.674 4.615Post 4.880 4.686 4.436Percent Change 1.67 0.27 "5.83n 25 188 26

Trust Federal Government(0–100)

Pre 53.440 57.759 59.000Post 44.000! 56.214 58.077Percent Change "17.66 "2.68 "1.56n 25 187 25

Trust State Government(0–100)

Pre 56.600 58.653 59.091Post 49.354! 58.267 57.391Percent Change "12.80 "0.66 "2.88n 22 176 25

Notes: Pre- and Post-cell entries are mean values for all students in the specified type whocompleted both pre- and postsurveys. Significance tests are two-tailed dependent (matched)sample t tests and the n indicates the number of students used in the t test. Variables codedso higher values indicate greater interest in politics, greater knowledge, higher efficacy, regis-tered to vote, greater likelihood of voting, and greater trust. Interest, knowledge, and internaland external efficacy are additive indexes of 3, 10, 3, and 5 items, respectively. Text and codingof the individual items is available on request.

!Significantly different from presurvey at p< .10.!!Significantly different from presurvey at p< .05.!!!Significantly different from presurvey at p< .01.

Effects of a Model Citizens’ Assembly 105

Page 12: Creating Better Citizens? Effects of a Model Citizens ... · ested in this type or level of political participation (e.g., Hibbing and Theiss-Morse 2002), and more empirical research

their issues extremity, the magnitude of their increase (both in absolute and percent-age terms) was less than half that of CACTUS students.

We speculate that increased extremity among CACTUS students may be afunction of two factors. First, extant literature demonstrates that individuals withhigher levels of political interest and knowledge are often more extreme. Thus,gains in interest and knowledge for CACTUS students might also have led togreater extremity. However, this explanation would have a difficult time account-ing for why other political science students, who also experienced increases in inter-est and knowledge, became more moderate in their self-identified ideologies. Wetherefore suspect that greater extremity among CACTUS students may be afunction of the nature of the course. Specifically, CACTUS involved high levelsof student interaction with one another in small groups, and discussion of politicalissues in this setting may have encouraged students to become more committedto their preexisting dispositions even though the discussions were designed to benonadversarial.

The nature of CACTUS might be expected to influence other political attitudesas well. The citizens’ assembly model emphasizes deliberation with the aim of reach-ing a collective decision. The vision is that this collective decision will not be achievedthrough adversarial debate and a majority vote but through a greater degree ofcooperation and persuasion. Ultimately, there is an implicit if not explicit promotion

Table 3. Partisan, ideological, and issue extremity

Variable (range)CACTUSstudents

POLstudents

FORstudents

Extremity of PartyIdentification(0–3)

Pre 1.792 1.647 1.462Post 1.960!! 1.692 1.500Percent Change 9.40 2.70 2.63n 24 185 26

Extremity of Self-Identified Ideology(0–3)

Pre 1.417 1.449 1.417Post 1.640!! 1.389 1.304Percent Change 15.76 "4.18 "7.93n 24 172 23

Extremity of IssuePositions (0–12)

Pre 4.583 3.851 3.680Post 5.304 4.130!! 3.522Percent Change 15.73 7.24 "4.30n 22 152 23

Notes: Pre- and Post-cell entries are mean values for all students in the specified type whocompleted both pre- and postsurveys. Significance tests are two-tailed dependent (matched)sample t tests and the n indicates the number of students used in the t test. Variables codedso higher values indicate more extreme party identification (i.e., strong Democrat=strongRepublican), more extreme ideology (i.e., very liberal=very conservative), and more extremeissue positions (in a consistent direction). The extremity of issue positions measure is createdfrom a battery of four issue items. Text and coding of the individual items is available onrequest.

!Significantly different from presurvey at p< .10.!!Significantly different from presurvey at p< .05.!!!Significantly different from presurvey at p< .01.

106 J. Gershtenson et al.

Page 13: Creating Better Citizens? Effects of a Model Citizens ... · ested in this type or level of political participation (e.g., Hibbing and Theiss-Morse 2002), and more empirical research

of decision making through consensus. It is somewhat surprising, therefore, thatupon examining attitudes toward the electoral system and the role of consensus,we found, as shown in Table 4, that CACTUS students actually experienced adecline (of 7.5%) in the importance they placed on seeking consensus in a democracywhile other political science students and the forensics students’ attitudes remainedessentially unchanged.

Table 4. Attitudes toward electoral system and consensus

Variable (range)CACTUSstudents

POLstudents

FORstudents

Important to SeekConsensus (1–7)

Pre 5.320 5.048 5.115Post 4.920 5.079 5.077Percent Change "7.52 0.63 "0.75n 25 188 26

Elections BasicallyFair (1–7)

Pre 4.480 4.771 5.115Post 5.120! 4.984!! 4.538!!

Percent Change 14.29 4.46 "11.28n 25 187 26

System of ElectionsNeeds Improvement(1–7)

Pre 3.080 3.226 3.692Post 3.160 3.360 3.154!!

Percent Change 2.60 4.14 "14.58n 25 189 26

President Should beChosen by PopularVote (1–7)

Pre 2.720 2.831 2.885Post 4.240!!! 2.819 2.654Percent Change 55.88 "0.14 "8.00n 25 187 26

Electoral College FairWay to Elect President(1–7)

Pre 3.760 3.746 4.000Percent Change 4.640!! 3.905 3.192!!!

n 25 189 26Electoral College Betterthan Alternatives(1–7)

Pre 3.920 4.153 4.192Post 4.640! 4.289 3.885Percent Change 18.37 3.28 "7.34n 25 189 26

Electoral Fairness=Support for ChangeIndex (5–35)

Pre 19.800 20.274 20.500Post 23.480!!! 20.626 19.115!

Percent Change 18.59 1.73 "6.75n 25 183 26

Notes: Pre- and Post-cell entries are mean values for all students in the specified type whocompleted both pre- and postsurveys. Significance tests are two-tailed dependent (matched)sample t tests and the n indicates the number of students used in the t test. Variables codedso higher values indicate greater sense of electoral fairness, less support for electoral reform,and more support for the importance of consensus. The electoral fairness=support for changeindex is an additive index of the items preceding it in the table. Text of the items is available onrequest.

!Significantly different from presurvey at p< .10.!!Significantly different from presurvey at p< .05.!!!Significantly different from presurvey at p< .01.

Effects of a Model Citizens’ Assembly 107

Page 14: Creating Better Citizens? Effects of a Model Citizens ... · ested in this type or level of political participation (e.g., Hibbing and Theiss-Morse 2002), and more empirical research

Like partisanship and ideology (and beliefs about the importance of consensus),normative expectations regarding attitudes toward the electoral system are notstraightforward. As one example, consider the item asking respondents if the presi-dent should be chosen by popular vote. It is not evident whether ‘‘better’’ citizensshould be supportive of this or not. Nevertheless, political science courses in generaldevote some attention to electoral systems and CACTUS in particular had presiden-tial elections as its primary focus. Thus, we would anticipate possible changes in stu-dent attitudes, particularly when compared to students not taking a political sciencecourse, on these items. Table 4 reveals the existence of some attitudinal changes aswell as variation in both the magnitude and direction of those changes. The groupexperiencing the least change is the non-CACTUS political science students. Whilethose students generally became more supportive of existing electoral mechanisms(note that the variables in Table 4 are coded such that higher values indicate greatersense of fairness and less support for electoral reform), the magnitude of the shift intheir attitudes is rather muted and the change is only statistically significant withregard to assessments of the fairness of elections.

As anticipated, CACTUS students experienced greater changes in their attitudesabout the electoral system. While these students experienced moderate increases intheir beliefs that elections are basically fair even if they don’t always agree with the out-comes and that the election system does not need substantial improvement, the mag-nitude of the change in their attitudes about Presidential elections specifically are trulynotable. There was an 18.4% increase in the belief that the Electoral College is betterthan alternatives, a 23.4% increase in the belief that the Electoral College is a fair wayto elect the President, and a staggering 55.9% decline in support for choosing the Presi-dent by popular vote—this despite the neutral stance assumed by the instructors andfacilitators both during the learning phase and while chairing the deliberations.

What is perhaps most surprising about the results in Table 4 is the attitudinalshift by the forensics students. Given the fact that the course is not designed to coverpolitical topics (including electoral politics), we anticipated little change among thosestudents. Yet, these students moved in the opposite direction of CACTUS studentsduring the semester, becoming less supportive of existing electoral arrangements.And the magnitude of their changes rivaled those of the CACTUS students in someinstances. One possible explanation for these changes is the political context duringthis time, namely the ongoing presidential nomination battle in the DemocraticParty. This explanation, however, offers little help in accounting for change in beliefsabout the Electoral College being a fair way to elect the President since mediadiscussion focused on nomination processes. We believe the more likely force atwork here is the fact that a number of the forensics students appeared at the publichearings of CACTUS to speak about the method of Presidential selection. Studentsin the course were offered extra credit for participating in a ‘‘public-speaking’’ eventand nine took the opportunity to do so in CACTUS. The students’ presentationswere no more than five minutes in length and they were told explicitly that theydid not need to do research on the topic. It was evident, however, that most hadgathered some minimal amount of information about the Electoral College; andwe suspect that this cursory introduction to electoral systems produced attitudinalchanges that would not have otherwise occurred.

Comparing the forensics students who participated in the public hearings tothose who did not supports this interpretation. For each of the electoral items inTable 4, changes in attitudes of the former group of students exceeds those of the

108 J. Gershtenson et al.

Page 15: Creating Better Citizens? Effects of a Model Citizens ... · ested in this type or level of political participation (e.g., Hibbing and Theiss-Morse 2002), and more empirical research

latter group, and in all but the case of the Electoral College being better than alter-natives, the magnitude of the difference in changes across the groups is substantial.A particularly striking result was that forensics students who did not appear at thepublic hearings experienced only a 1.19% decline in their assessments of the fairnessof elections while students who appeared experienced a 28.57% drop.6

The results in Table 4 suggest that students’ attitudes toward various elements ofthe political system may be differentially impacted by the nature of courses in whichthey are enrolled and the nature of specific activities in which they engage. Namely, tothe extent that a course focuses more heavily on a particular element like elections,attitudes toward that element may undergo greater change than they do in courseswithout as much emphasis. Similarly, among students in a given class, those who par-ticipate in an activity related to some dimension of the political system are more likelyto change their opinions than are their counterparts who do not participate. While wesuspect that this is the mechanism at work in the results of Table 4, we hope to gaingreater confidence by examining change in attitudes toward the electoral system in amultivariate model. Thus, Table 5 presents the results of Ordinary Least Squares(OLS) regression estimations of models in which the dependent variables include eachof the five items measuring attitudes about specific aspects of the U.S. election systemas well as a composite electoral fairness=support-for-change index. Each of the depen-dent variables is coded so that higher values indicate greater sense of fairness in thecurrent electoral system and less support for change. Our primary independent vari-able of interest is a dummy variable that takes on a value of one for CACTUS stu-dents and zero for all other students. Thus, we anticipate the coefficient estimate tobe positive, indicating greater change toward support of the current electoral systemfor CACTUS students relative to other students. We also include two dummy vari-ables for forensics students, one each for students who did and did not participatein the CACTUS public hearings. Like the CACTUS dummy, each of these takeson a value of one for the relevant students and zero for all other students. As controlswe include a host of sociodemographic variables (year in school, age, grade pointaverage, grade in the course in which the surveys were completed, gender, race, familyincome), a social setting variable (living on campus), and a childhood socializationvariable (frequency of talking about politics at home when growing up).

While the regression models fail to explain much of the variation in changes in atti-tudes toward elections (and, in the case of the Electoral College being betterthan alternatives, the model F-value does not reach standard levels of statisticalsignificance), the results in Table 5 at least partially confirm the importance ofthe course in which a student is enrolled. The coefficient estimates for the CACTUS-student variable are significant (and positive) for three of the dependent variables,including the fairness of the Electoral College and the desirability of popular vote asthe means to choose the President. In other words, even after controlling for a host ofother variables,CACTUSstudents becamemore supportive of the status quo in the elec-toral system, particularly as it relates to the selection mechanism for the President.

The results also support our conjecture about the distinction between thoseforensics students who participated in the CACTUS public hearings and thosewho did not. The coefficient estimate for the nonparticipants is insignificant in eachof the models except for the model for that with the index as the dependent variable.On the other hand, those forensics students who did appear before CACTUS becamefar less sanguine about the electoral system with their coefficient estimate beingsignificant in all cases.

Effects of a Model Citizens’ Assembly 109

Page 16: Creating Better Citizens? Effects of a Model Citizens ... · ested in this type or level of political participation (e.g., Hibbing and Theiss-Morse 2002), and more empirical research

Table

5.Exp

lainingchan

gein

attitudes

towardelectoralsystem

DependentVariable

Independent

variab

les

Elections

basically

fair

System

of

electionsneeds

improvement

Presidentshould

bechosenby

popularvo

te

Electoral

college

fairway

toelectpresident

Electoral

college

betterthan

alternatives

Electoral

fairness=support

forchan

geindex

CACTUSStudent

0.414(0.366)

"0.017(0.347)

1.587!

!!(0.388)

0.780!

!(0.376)

3.291!

!!(1.183)

FOR

participan

t"1.558!

!(0.693)

"1.334!

!(0.658)

"1.266!

(0.735)

"1.886!

!!(0.712)

"6.592!

!!(2.238)

FOR

nonparticipan

t"0.505(0.507)

"0.786(0.482)

"0.455(0.538)

"0.578(0.521)

"2.901!

(1.639)

Yearin

School

0.117(0.133)

0.054(0.126)

0.105(0.142)

0.318!

!(0.136)

0.720!

(0.432)

Age

"0.002(0.032)

"0.033(0.031)

"0.079!

!(0.034)

"0.066!

!(0.033)

"0.190!

(0.104)

GPA

"0.198!

(0.105)

0.006(0.099)

"0.019(0.112)

"0.268!

!(0.108)

"0.485(0.343)

Gradein

Course

0.191!

(0.114)

0.079(0.108)

0.145(0.121)

0.215!

(0.117)

0.773!

!(0.370)

Gender

"0.270(0.231)

"0.547!

!(0.219)

"0.272(0.246)

"0.784!

!!(0.238)

"2.007!

!!(0.755)

Race

0.706(0.711)

"0.585(0.676)

"0.595(0.755)

"0.882(0.733)

"1.772(2.302)

Fam

ilyIncome

0.069(0.076)

"0.070(0.072)

"0.053(0.080)

"0.067(0.078)

"0.230(0.246)

LiveonCam

pus

0.058(0.289)

0.122(0.273)

"0.140(0.307)

"0.057(0.295)

0.057(0.940)

ChildhoodPolitics

0.047(0.101)

"0.192!

!(0.095)

"0.268!

!(0.107)

"0.158(0.103)

"0.757!

!(0.330)

Constan

t0.414(1.189)

2.191!

(1.130)

2.772!

!(1.266)

4.023!

!!(1.236)

10.395

!!!(3.897)

F-V

alue

1.70

!1.76

!3.07

!!!

3.61

!!!

0.85

3.83

!!!

Adjusted

R2

0.047

0.051

0.128

0.155

0.170

n170

171

169

171

171

166

Notes:Cellentriesareregressioncoefficientsan

dstan

darderrors

inparentheses.Modelwithoutcoefficientestimates

was

insign

ifican

tas

model.FOR

participan

tan

dFOR

nonparticipan

treferrespectively

toforensics

students

whodid

anddid

notparticipatein

CACTUSpublichearings.

! Significan

tat

p<.10(two-tailtest).

!!Significan

tat

p<.05(two-tailtest).

!!! Significan

tat

p<.01(two-tailtest).

110

Page 17: Creating Better Citizens? Effects of a Model Citizens ... · ested in this type or level of political participation (e.g., Hibbing and Theiss-Morse 2002), and more empirical research

It is also worth noting that a few of the control variables emerge as importantdeterminants of change in attitudes toward elections. Older students did not gener-ally experience the same increases in support for the status quo as did their youngercounterparts. Similarly, individuals who grew up in households where politics wasmore frequently discussed were less likely than those with less exposure to politicsin their childhood to experience increases in their belief that the existing electoral sys-tem exhibits deficiencies. Finally, gender plays at least a modest role. Specifically,females were more likely than males to question the current electoral system at theend of the semester than at the beginning.

Discussion

Normative beliefs about the desirability of politically engaged citizens lead naturallyto discussions of possible remedies to the so-called ‘‘democratic deficit.’’ Among thepossible remedies is deliberative democracy, and it is precisely the effects of studentparticipation in a deliberative democracy exercise that we have examined here. Whilestudents in both CACTUS and other political science courses did become ‘‘better’’citizens during the test semester, the most interesting avenues for speculation regard-ing the implications of this research and for future research are the variations in‘‘growth’’ observed across categories.

As outlined above, both CACTUS and other students showed gains in theirinternal efficacy while exhibiting little or no gains in external efficacy. To someextent, this is consistent with findings in the literature regarding the relative stabilityof attitudes toward government responsiveness. Yet we suspect that the story is morenuanced than this. In particular, we conjecture that the lack of increases in externalefficacy is a function, at least in part, of the nature of political science courses,including CACTUS. Specifically, most courses devote a majority of their attentionto national politics. In CACTUS the focus was on the method for electing the Presi-dent and the ultimate recommendation forwarded by the members of the assemblywould have required a constitutional amendment to implement. Both the textbookand speakers in the course pointed out the virtual impossibility of securing suchan amendment. Hence it is perhaps not shocking that students did not experiencesignificant increases in their assessments of the responsiveness of public officialsand the political system. Had the students instead tackled a local government issueand had contact with local officials during the course, we posit that the likelihood ofsignificant changes in beliefs about the extent to which public officials care aboutwhat the people think and the extent to which those officials quickly lose touch withthe people might have been notably greater.

In addition to variations depending on the level of government (national v. statev. local) discussed, we believe that the topic of CACTUS can affect results in otherways. In examining Presidential elections in 2008, CACTUS students were tacklingwhat we regard as a process issue. While some course materials addressed potentialpolicy implications of various election methods, the focus was certainly on theprocess itself and the fact that CACTUS students experienced significant changesin their attitudes toward Presidential elections, becoming more supportive of theElectoral College as opposed to direct popular vote, appears to reflect this. Thischange seems important and suggests some modicum of hope. As Hibbing andTheiss-Morse (2002) convincingly argue, average citizens do not want increased citi-zen involvement in policymaking. Rather, they would like to place policymaking in

Effects of a Model Citizens’ Assembly 111

Page 18: Creating Better Citizens? Effects of a Model Citizens ... · ested in this type or level of political participation (e.g., Hibbing and Theiss-Morse 2002), and more empirical research

the hands of empathetic, non-self-interested decision makers (ENSIDs). Yet, mostAmericans question the possibility for the emergence of ENSIDs given current polit-ical processes. Essentially, it is the process by which public officials are chosen andmake decisions that is the primary focus and source of discontent for citizens. Ourfirst CACTUS iteration suggests that improving public knowledge about politicalprocesses may be one way to bolster faith in those processes.

Yet, our findings also suggest that the necessary gains in knowledge might berelatively substantial. In particular, the significant decrease in support for the existingelectoral system among those forensics students who appeared at the CACTUS publichearings suggests an important distinction between very shallow awareness and amore sophisticated understanding of issues under consideration. The forensics stu-dents who spoke before CACTUS seem to have obtained some minimal informationabout the Electoral College through ‘‘Googling’’ the term or some similar ‘‘research’’method. While we cannot be certain about their understanding of the Presidentialelection system at the outset, presumably these students experienced a visceral reac-tion upon discovering that citizens do not directly choose the President. It is fair toassume that the CACTUS students made much greater gains in their understandingof elections since they spent the entire semester immersed in exploring multiple dimen-sions of the process. And, the reaction of these students to the forensics students whoappeared before them illustrated this. The CACTUS students regarded their visitorsas making uninformed arguments that revealed a lack of real understanding. Thus, toreiterate, we believe that the depth of knowledge about, and understanding of, anissue has important implications for attitudes and this should be kept in mind asprofessors consider trade-offs between breadth and depth in their courses.

Significant gains in knowledge about political processes may boost faith in thoseprocesses. This, in turn, should produce increased confidence in government as awhole. We note, however, that such spillover effects are not immediately apparentin our results. As discussed above, external efficacy did not increase and CACTUSstudents actually experienced declines in their trust in government. If citizens are moreconcerned with political process than with policy, it is also possible that focusing on apolicy debate such as abortion, gun control, or health care in CACTUS mightaccomplish nothing in terms of improving citizen assessments of the political system.

Our findings with regard to partisanship and ideology raise additional questionsabout both the effects of deliberative exercises and about the implications ofdeliberation more generally. Despite the fact that the method for electing thePresident does not fall neatly along partisan and ideological divides, we witnessedincreased extremitization of CACTUS students in both their partisan and ideologicalcommitments. We speculated above that this may have been a function of the natureof the course. In other words, deliberation may tend to produce polarization andthe hardening of preexisting dispositions. If this is the case with a topic such asPresidential elections, we suspect that the effects would be magnified if students wereto deliberate about policy issues. On its face, polarization would appear to be anegative outcome. After all, many citizens abhor polarization in government andpolarization may have adverse consequences for the functioning and output of polit-ical institutions. In this vein, it is worth noting that CACTUS students experienced adecline (of 7.4%) in the importance they placed on seeking consensus as an elementof democracy.

Although increased polarization and decreased emphasis on the importance ofconsensus may seem troubling at first, we note that they need not be normatively

112 J. Gershtenson et al.

Page 19: Creating Better Citizens? Effects of a Model Citizens ... · ested in this type or level of political participation (e.g., Hibbing and Theiss-Morse 2002), and more empirical research

problematic. It is possible that individuals become more extreme in their partyattachments and ideological positions while simultaneously gaining an awarenessof, and perhaps appreciation of, the existence of differences in opinions. Americanstypically have unrealistic beliefs about the extent to which their own views are heldby others (see, for example, Hibbing and Theiss-Morse 2002). Deliberative exercisesby nature expose people to individuals with differing opinions and force individualswho do not agree with one another to attempt to explain their positions and arrive atsome decision. In this process, it is reasonable to suspect that individuals may gainpolitical tolerance and respect, and a willingness to accept that opinions other thantheir own are not necessarily evil, only different. If this is the case, the decline in theimportance of consensus experienced by CACTUS students may reflect an aware-ness of the existence of true differences in opinion and the necessity for negotiationand compromise in reaching collective decisions. In other words, it could simplyillustrate a belief that consensus is often not possible and it is unproductive to seekto force consensus. This interpretation of our results suggests that deliberative exer-cises may fit into prescriptions forwarded by Hibbing and Theiss-Morse (2002).Specifically, our experience with CACTUS suggests that such an extended deliberat-ive exercise is particularly well suited for making students aware of diverse opinions(although not necessarily across a broad range of issues) and for demonstratingdifficulties in reaching collective decisions.

Unfortunately, we do not have data allowing us to directly examine whetherCACTUS participants did in fact gain awareness and appreciation of differencesin opinions and the challenges of decision making. In future iterations of CACTUSwe will seek to obtain such data. In addition, we have many other unansweredquestions. For example, we have little information about variation in the effectsof CACTUS (and other political science courses) on political attitudes and behavioras a function of preexisting levels of political knowledge, personality traits, etc. Quitesimply, there may well be intervening variables that make deliberative exercises moreefficient in producing ‘‘better’’ citizens among some groups of individuals thanamong others. Finally, we have no way yet of knowing the extent, if any, oflong-term implications of participation in CACTUS. Will the gains in politicalknowledge and internal efficacy persist? Will the students be more likely to bepolitically active in the future (i.e., will they experience ‘‘citizenship deferred,’’identified as a recurring theme among Harrigan and McMillan’s (2007) students)?Answers to all of these (and many more) questions have important implicationsfor the utility of CACTUS and other variants of deliberative democracy.

Notes

1. While we have not found other university-level programs that employ a format quitelike that of CACTUS, several other experiments in political engagement offer valuablecomparisons. One is theWake Forest ‘‘Democracy Fellows’’ project, a four-year program usingan intensive Freshman seminar on deliberative democracy for 30 students, followed by othervaried projects of increasing scope, and a campus-wide exercise for students who had short-term exposure to deliberation (Harriger andMcMillan 2007). Another is the 21 courses of vary-ing designs, all encompassing ‘‘pedagogies of engagement,’’ in the Political Engagement Project(Colby et al. 2007). Both were reported to produce in some degree the desired changes—e.g., inpolitical involvement, knowledge, analysis, skills, and efficacy—but with important variationsthat the broader sweep of research on political attitudes might lead us to expect: For example,the participants in the Democracy Fellows program emerged ‘‘less positive about the political

Effects of a Model Citizens’ Assembly 113

Page 20: Creating Better Citizens? Effects of a Model Citizens ... · ested in this type or level of political participation (e.g., Hibbing and Theiss-Morse 2002), and more empirical research

system’’ after (because of?) gains in political knowledge, and the Political Engagement Projectproduced the greatest gains in the students starting with the least interest.

2. Attempts to assess the effects of such programs must cope with endemic issues of con-ceptualization, measurement, and sampling and participation. Declining civic participationand interest might reflect cynicism, or simply indifference—not at all the same things. Politicalengagement is reasonably understood to be multidimensional, but the distinction betweenpolitical skills and engagement on one hand and general social skills and citizen activism onthe other is frequently blurred. For example, the abilities of students to engage in a local com-munity civic project, to reach a group decision about a class grading procedure, or to organizegroup meetings and to speak in public are casually construed as the development of important‘‘political’’ skills (Beaumont et al. 2006; Longo, Drury, and Battistoni 2006). Some of the‘‘positive’’ findings obtained may result from stretching the meaning of political engagementas much or more than from actual changes in behavior. Longo et al. are so enthusiastic aboutthe results of 30 in-depth interviews about such activities that they conclude ‘‘co-curricular’’approaches are inherently superior to traditional classroom settings and proudly cite onestudent’s contempt for classroom knowledge (Longo, Drury, and Battistoni 2006, 325). Notethat this finding is not very consistent with the across-the-board benefits claimed by Beaumontet al. Finally, as we discovered ourselves, self-selection among participants can be a majorconsideration in civic education and deliberation programs.

3. The political interest index is created from three items: (a) general interest in infor-mation about government and politics, (b) how closely one pays attention to informationabout what’s going on in government and politics, and (c) interest in politics relative to mostAmericans. The political knowledge index is based on the number of correct responses to itemsasking students to identify the job or political office held by Nancy Pelosi, Dick Cheney, Gor-don Brown, and John Roberts, Jr., and items asking about party control of the U.S. Houseand Senate, about the number of full terms a President may serve, the voting age, the numberof justices on the U.S. Supreme Court, and the names of the three branches of the Americangovernment. The internal efficacy index adds responses on three items: (a) the extent to whichrespondents feel they have a good understanding of important issues facing the country, (b)the extent to which respondents feel that politics is too complicated for them to understandwhat is going on, and (c) the extent to which respondents feel they have opinions worth listen-ing to. The external efficacy index is created from the following five items: (a) whether peoplelike them have any say about what the government does, (b) if the government pays attentionto what the people think when it decides what to do, (c) if elections make the governmentpay attention to what the people think, (d) if public officials care what the people think,and (e) if public officials lose touch with the people quickly. All of the individual variablesand the indexes are coded so that higher values indicate greater interest, knowledge, efficacy,trust, and participation. The full text of these and other items on the surveys is available onrequest.

4. These measures of extremity are all created by ‘‘folding’’ responses to items aroundtheir midpoints. For example, for ideological self-identification, the midpoint of the scale isfour (moderate, middle of the road). A respondent identifying oneself as very liberal (1) orvery conservative (7) would have a value of three for ideological extremity (the absolute valueof one’s self-placement, one or seven, minus the midpoint, four). For extremity of party identi-fication and of issue positions, higher values similarly indicate greater extremity.

5. In addition to the ceiling effects theme, changes in political interest and internal effi-cacy illustrate the statistical significance theme. Namely, while the absolute and percentagegrowth for CACTUS students outstripped those of other political science students in thesedimensions, smaller sample sizes for CACTUS students produces lower, or insignificant,results on matched-pairs difference of means tests for those students.

6. The percentage decreases for the forensics students who did and did not appear at theCACTUS public hearings for each of the electoral items in Table 4 are, respectively, as fol-lows: fairness of elections, 28.57%, 1.19%; election system needs improvement, 20.59%,11.29%; President should be elected by popular vote, 13.64%, 5.66%; the Electoral Collegeis a fair way to elect the President, 36.59%, 9.52%; the Electoral College is better than alter-natives, 7.89%, 7.04%; electoral fairness=support for change index, 22.83%, 6.61%. None ofthe changes for the students not participating in the public hearings are statistically significantwhile changes for the students who did participate are significant at the .05 level or better for

114 J. Gershtenson et al.

Page 21: Creating Better Citizens? Effects of a Model Citizens ... · ested in this type or level of political participation (e.g., Hibbing and Theiss-Morse 2002), and more empirical research

the fairness of elections, Electoral College is a fair way to elect the President, and the electoralfairness=support for change index.

References

Barber, Benjamin R. 1984. Strong Democracy: Participatory Politics for a New Age. Berkeley:University of California Press.

Beaumont, Elizabeth, Anne Colby, Thomas Ehrlich, and Judith Torney-Purta. 2006. ‘‘Pro-moting Political Competence and Engagement in College Students: An Empirical Study.’’Journal of Political Science Education 2: 249–270.

‘‘British Columbia Citizens’ Assembly on Electoral Reform.’’ n.d. (Accessed on March 1,2009) http://www.citizensassembly.bc.ca/public

Citizens’ Assembly for Critical Thinking about the United States (CACTUS). n.d.‘‘Mandate.’’ www.cactus.eku.edu

Colby, Anne, Elizabeth Beaumont, Thomas Ehrlich, and Josh Corngold. 2007. Educatingfor Democracy: Preparing Undergraduates for Responsible Political Engagement. SanFrancisco: Jossey-Bass.

Delli Carpini, Michael X. and Scott Keeter. 1996. What Americans Know About Politics andWhy It Matters. New Haven: Yale University Press.

Ehrlich, Thomas. 2000a. ‘‘Learning About Learning from Alverno.’’ Change 33: 55–58.Ehrlich, Thomas. 2000b. Civic Responsibility and Higher Education. Westport, CT: American

Council on Higher Education and the Oryx Press.Elder, Laurel, Andrew Seligson, and Daniel Hofrenning. 2007. ‘‘Experiencing New Hamp-

shire: The Effects of an Experiential Learning Course on Civic Engagement.’’ Journalof Political Science Education 3: 191–216.

Fowler, James H. and Christopher T. Dawes. 2008. ‘‘Two Genes Predict Voter Turnout.’’Journal of Politics 70: 579–594.

Furlong, Scott and Denise Scheberle. 2006. ‘‘The Role of Political Science Courses in CivicEngagement.’’ Paper presented at the annual meeting of The Midwest Political ScienceAssociation, Palmer House Hilton, Chicago, Illinois, April 20.

Harriger, Katy J. and Jill J. McMillan. 2007. Speaking of Politics: Preparing College Studentsfor Democratic Citizenship through Deliberative Dialogue. Dayton, OH: KetteringFoundation Press.

Hetherington, Marc J. 1998. ‘‘The Political Relevance of Political Trust.’’ American PoliticalScience Review 92: 791–808.

Hibbing, John R. and Elizabeth Theiss-Morse. 2002. Stealth Democracy: Americans’ Beliefsabout How Government Should Work. Cambridge: Cambridge University Press.

Huerta, Juan Carlos and Joseph Jozwiak. 2008. ‘‘Developing Civic Engagement in GeneralEducation Political Science.’’ Journal of Political Science Education 4: 1–24.

Kam, Cindy D. and Carl L. Palmer. 2008. ‘‘Reconsidering the Effects of Education onPolitical Participation.’’ Journal of Politics 70: 612–631.

Kenski, Kate and Natalie Jomini Stroud. 2006. ‘‘Connection between Internet Use andPolitical Efficacy, Knowledge, and Participation.’’ Journal of Broadcasting and ElectronicMedia 50: 173–192.

Kerlin, Mary. 2005. ‘‘Promising Approaches for Strengthening Civic Education.’’Monograph: Educating for Democracy; California Campaign for the Civic Mission ofSchools; Center for Civic Education.

Longo, Nicholas V., Christopher Drury, and Richard M. Battistoni. 2006. ‘‘CatalyzingPolitical Engagement: Lessons for Civic Educators from the Voices of Students.’’ Journalof Political Science Education 2: 313–329.

Macedo, Stephen. 2005. Democracy at Risk: How Political Choices Undermine Citizen Partici-pation. Washington, DC: Brookings Institution Press.

Effects of a Model Citizens’ Assembly 115

Page 22: Creating Better Citizens? Effects of a Model Citizens ... · ested in this type or level of political participation (e.g., Hibbing and Theiss-Morse 2002), and more empirical research

Mishler, William and Richard Rose. 2001. ‘‘What are the Origins of Political Trust: TestingInstitutional and Cultural Theories in Post-Communist Societies.’’ Comparative PoliticalStudies 34: 30–62.

Morrell, Michael E. 2005. ‘‘Deliberation, Democratic Decision-Making, and Internal PoliticalEfficacy.’’ Political Behavior 27: 9–69.

‘‘Ontario Citizens’ Assembly on Electoral Reform.’’ n.d. http://www.citizensassembly.gov.on.ca/en/default.asp

Rainey, Jane G. and Glenn W. Rainey, Jr. 2006. ‘‘Innovation in Civic Engagement: The Brit-ish Columbia Citizens’ Assembly on Electoral Reform.’’ Southern Political Science Associ-ation annual meeting, Atlanta, GA, January 5–7.

Rainey, Jane G. and Glenn W. Rainey, Jr. 2008. ‘‘Civic Education, Civic Engagement, andCitizens’ Assemblies: A Sober Second Look.’’ Southern Political Science Association annualmeeting, New Orleans, LA, January 10–12.

Warren, Mark E. and Hilary Pearse, eds. 2008. Designing Deliberative Democracy: The BritishColumbia Citizens’ Assembly. New York: Cambridge University Press.

116 J. Gershtenson et al.

Page 23: Creating Better Citizens? Effects of a Model Citizens ... · ested in this type or level of political participation (e.g., Hibbing and Theiss-Morse 2002), and more empirical research

Copyright of Journal of Political Science Education is the property of Routledge and its content may not becopied or emailed to multiple sites or posted to a listserv without the copyright holder's express writtenpermission. However, users may print, download, or email articles for individual use.