Democracy Transformed : Perceived Legitimacy of the Institutional Shift from Election to Random Selection of Representatives

While democracy remains a firmly-held ideal, the present state of electoral democracy is plagued by growing disaffection. As a result, both scholars and practitioners have shown considerable interest in the potential of random selection as a means of selecting political representatives. Despite its potential, deployment of this alternative is limited by concerns about its perceived legitimacy. Drawing on an inductive analysis of the replacement of elections with random selection in two student governments in Bolivia, we explore stakeholders’ perceptions of the legitimacy of random selection by investigating both their overall support for randomly selecting representatives as well as the views that inform this support. Overall, we find that random selection is indeed accepted as a legitimate means of selecting representatives, with stakeholders broadly preferring random selection and recommending its use in other schools—views which are informed by a critical assessment of random selection’s relative merits. Moreover, we find that perceptions may be affected by contextual factors that extend beyond individuals’ own values. Our findings thus contribute to work on random selection, its contextual embeddedness, and on the values underpinning democratic structures. Author Biography Simon Pek: Simon Pek works as an Assistant Professor of Sustainability and Organization Theory at the Gustavson School of Business at the University of Victoria. Through his research, he explores how organizations and the individuals within them embed social and environmental sustainability into their cultures, strategies, and daily operations. Specifically, he looks at new forms of practicing democracy in organizations, the micro-processes of sustainability-oriented cultural change, and sustainability-related issues in the domain of international business. Jeffrey Kennedy: Jeffrey Kennedy is a doctoral candidate at McGill University’s Faculty of Law, where he researches and writes primarily at the intersections of deliberative democracy and public decisionmaking pertaining to criminal justice. Outside the University, he is actively involved in communitybased projects involving prisoner reintegration and democratic experimentation. Adam Cronkright: Adam Cronkright is one of the co-founders of Democracy In Practice, a non-profit dedicated to democratic experimentation, innovation, and capacity building. His passion for democracy and deliberation has led to a broad base of experience, which includes an independent study of the jury system; dialoguing with members of the 2011 Icelandic Constituent Council; co-facilitating two NYC General Assemblies; co-writing the Spokes-Council Proposal at Occupy Wall Street; and teaching and learning at the democratically-run Brooklyn Free School.


Introduction
While democracy itself continues to be a firmly-held political ideal, the last several decades have seen increasing dissatisfaction with the state of contemporary representative democracy.Citizens' trust in elected representatives and political institutions has diminished and remains alarmingly low (Dalton, 2017).Those in government are regularly perceived as-and indeed often shown to beunrepresentative of the diverse populations they serve as well as unresponsive to their views and interests (Bartels, 2008;Carnes, 2012;Pharr, Putnam, & Dalton, 2000).At the same time, the deliberative turn in democratic theory has articulated new standards and concentrated critiques along those lines.Rather than proceeding from informed public deliberation-a process of exchanging and genuinely reflecting on competing arguments for a common good (Cohen, 1989)-electoral politics has been cited as having a host of shortcomings: rigid partisanship, personal ambition, ignorance and misinformation, and a range of vote-seeking behaviour such as pork-barrelling, pandering to special interests or citizens' baser instincts, and point-scoring at the expense of substance (see e.g.Gutmann & Thompson, 1996;Leib, 2004).
In response to these apparent deficiencies, both academics and practitioners have turned their attention toward novel means of achieving more representative, participatory, and deliberative government.Taking a prominent place within both these shifts has been renewed interest in the use of random selection-referred to otherwise as lot or sortition-as a means of selecting political representatives from among the citizenry.Once an integral component of ancient Athenian democracy where the majority of public offices were filled in such a manner, the practice had largely disappeared from modern political consciousness (Hansen, 1991).However, in light of its potential to combat contemporary ails, the idea has undergone a revival in both democratic theory and practice.
The most visible part of this revival has occurred through the practical development of various 'democratic innovations': novel mechanisms representing "a departure from the traditional institutional architecture [of] advanced industrial democracies" (Smith, 2009, p. 1).Among these, deliberative 'mini-publics'-randomly selected microcosms of a wider population-such as Citizens' Assemblies or Citizens' Juries have become increasingly common features of the democratic landscape and have offered insights for possible reform (Smith, 2009;Smith & Wales, 2000).Less visibly, a growing body of scholarship has explored random selection's theoretical potential to address contemporary politics' shortcomings.Together, this theoretical analysis and real-world experience has led some to advocate for incorporating random selection as a permanent feature of contemporary governance.While some

Random Selection of Political Representatives: Prior Work and Present Problems
The Potential of Random Selection: Theoretical Advantages and Disadvantages While in some respects still in its early stages of development (Stone, 2016), scholarly interest in the random selection of representatives as a democratic practice has increased considerably over recent decades in light of its potential for realizing democratic values.Indeed, contemporary attention to random selection is largely a story of its potential toward that end, with most scholarship in this area being either directly concerned with or informed by identified advantages and disadvantages.
In terms of advantages, random selection is first thought to express and give effect to values of fairness and equality in an especially compelling way.Impartial with respect to personal characteristics, random selection gives all citizens an effectively equal chance of being selected (Buchstein, 2010).Accordingly, it clearly expresses the idea that all citizens are politically equal and encourages this view among citizens (Mulgan, 1984).Moreover, it actively fosters equality through a fair and even distribution of the goods (and responsibilities) of public service, including opportunities for development or remuneration (Barber, 1984;Engelstad, 1989).
Second, through the same mechanism, random selection distributes positions proportionally across the demographic profile of a population, producing descriptively representative political bodies (Carson & Martin, 1999).Accordingly, it both "mitigate[s] the oligarchical tendencies of representation" (Barber, 1984, p. 291) and facilitates proportionate representation of traditionally underrepresented demographics, such as women or visible minorities (Mueller, Tollison, & Willett, 1972).Third, with regular use, random selection is thought to create a more direct and participatory form of democracy (Dowlen, 2009).For one, the logic of random selection involves no rationale for successive terms in office, allowing frequent rotation of political decision-makers and inhibiting the formation of an entrenched political class (Carson & Martin, 1999;McCormick, 2006).Moreover, random selection lowers the threshold for participation, removing the personal risks and resource demands of intensive or drawn-out electoral contests (Carson & Lubensky, 2009;Goodall & Osterloh, 2015).
Fourth, random selection may enhance the deliberative, public-oriented quality of decision-making.Its impartiality is thought to inhibit corruption (Buchstein, 2010) while also preserving humility-and therefore responsiveness-among those selected (Guerrero, 2014).Selected representatives are themselves likely to be more impartial absent political debts or pressures to cater to partisan or special interests in seeking re-election, and are therefore free to revise their opinions and take a longterm view of policy decisions (Burnheim, 1985;Vandamme & Verret-Hamelin, 2017).Moreover the tendency toward descriptive representation is likely to reduce the overrepresentation of ambitious, prestige-seeking individuals (Guerrero, 2014).
Similarly, the consequent diversity of perspectives is thought to offer epistemic advantages in decision-making (Landemore, 2013) while the mechanism itself sidesteps the problem of uninformed voting and its consequences (Mueller et al., 1972).
Fifth, random selection is considered to be less socially divisive, avoiding the social conflict, polarization, and partisanship that often accompanies elections and creates lasting issues such as legislative deadlock (Engelstad, 1989).Lastly, random selection is touted as a more efficient selection process than elections, requiring much less time, effort, and resources to conduct than drawn-out campaigns aimed at swaying voters or facilitating public debate (Engelstad, 1989).
Of course, some of these same characteristics-the lack of professionalization, intentional (self-)selection, or political pressures-have been highlighted as potential disadvantages.First, the "inherent amateurism" (McCormick, 2006, p. 156) of shorter-term, non-professional politicians is contrasted with the view that elections theoretically allow for the selection of the more capable or talented individuals to serve and build specialized capacities over time (Burgers, 2015;Carson & Martin, 1999).Secondly, some have speculated that randomly selected representatives may feel less responsibility or moral obligation to their office or public good (Carson & Martin, 1999), either because random selection precludes the intentional (self-)selection of those most committed to the public good (Burgers, 2015;Mansbridge, 1999) or because elections uniquely provide an institutionalized accountability mechanism through the prospect of re-election, incentivizing good performance (Burgers, 2015).Other widely perceived benefits associated with elections might also be seen as diminished or absent in random selection.While random selection provides greater room for meaningful participation as representatives, it may nonetheless diminish the widespread participation and contestation that comes through voting for representatives (Vandamme, Jacquet, Niessen, Pitseys, & Reuchamps, in press).Similarly, absent representatives' need to seek public awareness and support, random selection may forego the benefits associated with campaigns, including the way in which they can engage the citizenry, renew their support for the regime, educate individuals on both politics and political issues, and encourage further conversations among the citizenry (Hart, 2009).
For a number of scholars at least, the perceived balance of these advantages and disadvantages has generated interest in institutionalizing some role for randomly selected bodies within contemporary politics and consequently an increasing number of proposals for implementation.The modern trend in this regard emerged in the 1970s and 1980s (see, e.g.Dahl, 1970Dahl, , 1989)), though proposals have since become increasingly common through a new wave of participatory and deliberative democrats (Snider, 2007).While the more modest proposals envision randomly selected bodies playing an advisory role-to elected representatives (Dahl, 1970), the general public (Gastil, 2000), or both simultaneously (Fishkin & Luskin, 2006)-the more ambitious propose randomly selected bodies with actual decisionmaking powers (Lucardie, 2014).

Perceived Legitimacy: A Barrier to Reform
Whatever their particular design or context, however, a central issue has been the question of whether or not the governed would in fact perceive such reforms as legitimate-a distinct question from the philosophical one of objective or normative legitimacy (Parkinson, 2003).Independent of whether or not there are compelling democratic advantages to the random selection of representatives, scholars have nonetheless recognized that its perceived legitimacy is crucial to its feasibility (Font & Blanco, 2007;Lucardie, 2014).Accordingly, they suggest that without broad public support such bodies would command limited respect and in turn exert little influence (Buchstein & Hein, 2009).
Given a lack of public familiarity, it is unsurprising that both scholars and practitioners have at times expressed intuitive doubts that random selection would be perceived as a legitimate or rational mechanism for selecting representatives (Buchstein & Hein, 2009;Carson & Martin, 1999).Some have thus suggested that perceptions remain an obstacle to institutionalizing random selection, though they also recognize the contingent nature of such perceptions.Accordingly, Buchstein and Hein (2009) conclude that their own proposal "would gain in democratic legitimacy only if the political culture changed to the point where the rational potential of the lot was recognized" (p.49).Toward that end, Carson and Martin (1999) suggested that the perceived legitimacy of random selection may be culturally specific and positive perceptions may depend on "familiarity, discussions of random selection versus alternatives, and overcoming vested interests opposed to [it]" (p.38).
As it stands, however, little is actually known about public perceptions of the random selection of representatives nor what considerations inform them.Indirect insight might be gleaned through varying degrees of inference.For instance, research shows that participants regularly point to the more inclusive diversity of mini-publics-a fact tied closely to random selection-as a positive, if not the most important, feature of these initiatives (Carson, 2010;Curato & Niemeyer, 2013;Font & Blanco, 2007).Research also demonstrates that participants view minipublics positively overall and support their wider deployment (Font & Blanco, 2007;Kuper, 1997).Citizens' Assemblies overall have been shown to be perceived positively by the general public as well, being seen as legitimate, influential, and trustworthy (Cutler, Johnston, Carty, Blais, & Fournier, 2008;Levy, 2010).Cutler and colleagues (2008)  explainable by other features or characteristics of these initiatives-such as effective facilitation or the provision of high quality information by independent experts.Indeed, in their study, Cutler and colleagues (2008) demonstrated that overall perceived legitimacy was arrived at for different reasons by different people: while for some the 'ordinariness' of representatives was determinative-a fact highly attributable to random selection-others were persuaded by the gained expertise of citizen representatives-a fact which has less of an inherent connection.In all, and while perceptions of modes of selection are inevitably contextual, conclusions drawn from overall perceptions require some qualification in that these perceptions are not necessarily linked to random selection specifically.
More direct research comes from the 2009 Australian Citizens' Parliament-a fourday initiative involving 150 randomly selected Australian citizens deliberating and making recommendations for an improved political system (Dryzek, 2009).Here, Lubensky and Carson (2013) found that many participants believed that random selection "made the process fairer," (p.42), while Hartz-Karp and colleagues (2010) relayed that participants pointed out that the fact that "each Australian citizen had an equal chance to participate… gave the whole process a kind of legitimacy unusual to most public forums" (internal quotations omitted) (p.363).Even here, however, possible differences between the studied initiative and those proposed for institutionalization potentially limit the former's generalizability.
Both the former and those previously surveyed were temporary, discrete initiatives, having only recommendatory power, while the frontier of proposed reforms involves standing bodies with dynamic functional responsibilities and decisionmaking authority.Accordingly, citizens may support random selection in one context but not the other, suggesting the need for research in contexts which share the latter features.
Recent work by Vandamme and colleagues (in press) has contributed to bridging this gap, using surveys to assess support among both elected representatives and the general public for a legislative chamber wholly or partially constituted through random selection.While elected officials were highly critical of the idea-opposing both propositions with a significant majority-the general public was more ambivalent.Of those adopting a position, support was split: while 40% opposed a fully randomly selected chamber in comparison with 29% in support, 47% supported a mixed chamber, with only 25% opposed.With a notable number remaining neutral on the propositions, the authors note that there exists significant space for public opinion to shift if the debate intensifies, and echo Carson and Martin in suggesting that, considering a lack of public familiarity, the popularity of the idea could increase if it were to become a more prevalent practice.Research by Dimitri Courant moves closer to this reality by examining the actual use of random selection since 1969 to constitute a standing, albeit advisory, body within the French military (Courant, in press).While proposals by the body have been particularly influential, Courant echoes Vandamme and colleagues in noting that elected officials have come to assert illegitimacy because members are not elected, although perhaps for political reasons.However, Courant's study does not include an investigation of the perceptions of those being represented by the body.
Despite this important work, there remains an overall dearth of information on public perceptions of random selection, necessitating further investigation.In light of the above review, research is needed which not only explores perceptions of randomly selecting representatives specifically, but does so in regard to standing bodies with some decision-making authority.Moreover, in light of scholars' speculation about the connection between familiarity and perceptions, research which examines perceptions in the context of actual practice, and especially that which uses a more in-depth methodology to explore underlying considerations, would provide an especially important contribution.While based in a unique educational context subject to the limitations we elaborate on in our discussion section, the present study takes important steps toward addressing each of these needs.In doing so, it not only provides evidence that the random selection of representatives can, under certain conditions, be perceived as legitimate, but also provides novel insights into how context may shape these perceptions, all of which should inform future research and real-world initiatives involving the random selection of representatives.

Research Strategy and Context
In this paper, we adopted an emergent and inductive approach to study students' and staff members' perceptions of the shift from election-based to random selection-based student government structures at two schools in Bolivia: termed here School A and School B to protect the confidentiality of participants.School A is an elementary school located in a rural region of Bolivia, while School B is an evening high school located in an urban area.Beginning in 2014, Democracy In Practice, an organization dedicated to real-world democratic experimentation, began working with both schools to redesign their student governments around practices of random selection, rotation, and horizontality.The principals of both schools approved the initial student government projects as well as the research study on which this paper is based.
Prior to the above projects being implemented, both schools had similar student government structures that were centered on elections, full-year terms, and hierarchical roles.The student government structure at School A was based on a system in which students annually voted for parties who campaigned to form the student government for a one-year term.Each party had to have an equal number of males and females, and only students in upper grades were eligible.The student government structure at School B was similar, though involved no gender quotas and, according to one interviewee, operated only twice in the past 28 years.The new student government structures at both schools were similar in that they were centered on the random selection and rotation of representatives, and horizontal decision-making within the governments rather than traditional hierarchical roles.Their mandates included advocating for student interests, making and implementing decisions on various issues affecting those interests, planning and organizing events, and rolling out programs such as establishing a new library and creating student identification cards.
At the time of our study, the student government at School A consisted of eight randomly selected members.Stratified sampling was used to ensure that the eight members were spread across grades and that genders were equally represented over time.Students could only be randomly selected for one term per school year.Student governments were rotated every three months, creating three terms per year.Finally, in addition to the eight representatives selected every term, replacements were randomly selected in case representatives had to leave their positions during their three months in office.The new student government at School B consisted of twelve members.Stratified sampling was used to ensure gender balance; however, it was not used with regard to grade levels.Student governments here were rotated approximately every four months, resulting in two terms per school year.School administrators did not have formal veto powers over the student governments' decisions, but could withhold support for decisions and initiatives they did not agree with.Formal teacher engagement was generally limited, and when present, it was focused on supporting the student governments' work and ideas as opposed to influencing the direction of their work.
Both the implementation of the new government structures and the work of the student governments were regularly supported by two members of the organization working on the ground in the schools.Regarding the former, the organization proposed the new democratic practices and, upon their approval, launched several workshops which both introduced the new structure and practices to the entire student population and facilitated dialogue about what areas of concern the student governments ought to address.Following these workshops, the organization conducted the first student lottery and trained students to run subsequent ones on their own with limited guidance and support.Once the governments were selected, members of the organization helped orient and build the capacity of new members of the government and were present for most meetings to serve as a general resource, helping with limited tasks such as proofreading documents, making suggestions, and congratulating and recognizing students at the end of their terms.Members of the organization had no vote, and had to be called on by student facilitators to speak at meetings.They also did not have any authority over the student governments or any other students at the schools: they had no role in influencing students' grades, could not discipline any students, could not add or remove any members of the student government, and did not provide teachers or administrators with any formal assessments of the student government members' performance.

Data Sources
In this paper, we drew from two sources of data that were gathered as part of a larger study on the implementation of these novel democratic practices at the schools: in-depth semi-structured interviews and observations.Between 2015 and 2016, the third author conducted a total of 67 interviews with 66 students and staff at the schools, including interviews with students that had and had not participated in the student government.All interviews were conducted in Spanish. 1 Students and staff at both schools were informed of the study, and all those who were interested and provided consent and/or assent were interviewed.At School A, this involved 21 interviews with students, and 7 with staff, with one of the 7 staff members being interviewed twice.At School B, this involved 30 interviews with students, and 8 with staff.The scope of the interviews included participants' interpretations and perceptions of various aspects of the new student government system, their experiences engaging with or participating in the student government, and perceived facilitators and barriers of the changes in government structures.In addition to these interviews, the third author observed many of the meetings and operations of the student government and documented his observations in typed field notes.

Data Analysis
We analyzed our corpus of data using an inductive approach, identifying emergent themes and patterns through comparing and contrasting fragments of the data (Glaser, 1978;Glaser & Strauss, 1967).With a particular focus on participants' perceptions of the new practices, the first author engaged in a process of open coding and constant comparison to identify broader categories.After identifying a complete set of categories of advantages and disadvantages of random selection (which included 11 categories of perceived advantages and 5 categories of perceived disadvantages), the first and second authors compared these emergent categories with existing theoretical work on this topic, grouping similar categories of perceived advantages and disadvantages where appropriate under broader headings and ultimately identifying areas of both coincidence and novelty that we report in discussing our findings.Expressing our findings this way was intended to not only enhance analytical clarity and flow, but also improve the dialogic potential between our findings and theoretical research which frequently operates in the currency of random selection's advantages and disadvantages.Finally, we coded all participants' responses to assess whether they preferred random selection or elections and whether they recommended that these democratic practices be used at other schools, which we used to investigate their overall preferences for random selection versus elections.

Strategies for Ensuring Rigor
We adopted several strategies to ensure rigor.First, we conducted several member checks with six diverse participants from each school after completing our data analysis to confirm that our findings resonated with the experiences of participants at the schools.
Second, to increase the reliability of our analysis, we conducted an assessment of inter-rater reliability-a process aimed at evaluating the level of similarity when different coders code the same data (Campbell, Quincy, Osserman, & Pedersen, 2013).Based on the practices suggested by Campbell and colleagues (2013), we selected a sample of nine interview transcripts, with the first author highlighting all data fragments (27 in total) which had a code relevant to our research question.The third author then coded these 27 data fragments using a codebook containing the category names and definitions prepared by the first author.Both authors matched on 20 of the 27 data fragments-a score of 74%.The first author also created a table of each participants' overall recommendations about random selection versus elections, rotation versus year-long terms, and the appropriateness of other schools adopting the same program of new democratic practices.He blanked out his codes for ten of the participants, which the third author completed using the set of codes used throughout the document.The two authors matched on 29 out of the 30 responses, resulting in a score of 97%.In both activities, the authors discussed any discrepancies and agreed on a shared interpretation.
Third, we undertook a reflexivity exercise in which the third author engaged in intersubjective reflection (Finlay, 2002), where he analyzed his work and interactions to explore how these could have influenced participants' perceptions and interpretations.We identified two overarching means of potential influence.
The first was through workshops and meetings in which students discussed the nature and merits of the new democratic practices with project organizers, including the third author, as part of the roll-out of the new student government structures.
Given their educational nature, these likely had an influence on participants' perceptions.However, these events generally took place at the outset of the project, approximately a year and a half prior to the interviews, and included critical views as well.Accordingly, it seems unlikely that the content of those meetings was simply reproduced by students at the point of interview.The second means was the organizational members' working relationship with the students and staff which created a positive bond and could have led to participants answering in socially desirable ways instead of expressing their true opinions-a phenomenon commonly referred to as social desirability bias (Grimm, 2010).Ultimately, we do not feel this had an adverse impact on our results as the third author maintained an open demeanor and regularly insisted that participants be as open and honest as they like, particularly in terms of expressing dissent.Moreover, many participants did speak for one of the practices while being against others, suggesting that they were not tailoring their responses to be socially desirable.

Findings
In this section, we report on the perceived legitimacy of random selection as determined by both our analysis of participants' overall support for random selection relative to elections, and their perceptions of the advantages and disadvantages of random selection relative to elections, taken as a value-based assessment which informed their overall support.

Participants' Overall Support for Random Selection
We investigated participants' overall support for random selection relative to elections primarily through two questions.First, we coded participants' responses to the question about their preferences regarding selection methods to assess whether they preferred random selection or elections.Of the 65 participants who were asked this question, 2 were unsure (3%), 14 preferred elections (22%), and 49 preferred random selection (75%).While a relatively small sample size, this does suggest that students preferred randomly selecting student governments over electing them.Second, we asked students whether they would recommend the new government structure (which includes random selection, rotation, and horizontal decision-making).Of the 59 participants asked, 57 (97%) broadly recommended that other schools adopt the government structure, and 2 were unsure (3%).Of the 57 who broadly recommended the new government structure, 24 made explicit comments related to whether they would recommend elections or random selection (the remainder only made a broad recommendation for the overall structure).Of these, 21 recommended random selection (88%), 2 recommended elections (8%), and 1 was unsure (4%).Some participants perceived random selection as a means of "break[ing]…the tradition of selection based on votes" more generally and suggested using it in other contexts as well, such as for selecting members of the Parent-Teachers Association, or even for distributing homework.Indeed, for choosing the designated students that help with in-class tasks and activities, one staff member shared: "We always selected them using elections, because we had never used random selection.But this year we began selecting them through random selection." Notably, of the 14 participants who preferred elections over random selection, 13 were at School B (93%).While not a direct focus of our research, we identified three possible explanations for this discrepancy.First, School A is in a rural area, which, according to a staff member, tend to be more collective-oriented.Youth there thus appeared to be more inclined toward democratic practices like random selection that rely less on notions of social distinctiveness: "The kids from the countryside have a better sense of community and of community leaders, that the leader isn't the highest authority, like here in the city." Second, School A is smaller than School B, and almost all students were able to enter student government at some point over the previous two years.Greater proximity to the student government-either through direct participation or relationships with participating peers-may thereby contribute to one's familiarity with or attention to student government structures.Third and relatedly, while School A serves children and youth, School B serves an older population, many of whom have competing obligations and attend school less regularly, limiting communication about, familiarity with, or attention to the student government.
Inconsistent attendance was also cited by participants as decreasing the overall functionality and effectiveness of the student government at School B, which might also have led to conclusions that representatives elected from the most committed would be more desirable in that setting.
With respect to the above, participants frequently justified their overall support by referring to their perceptions of the relative advantages and disadvantages of random selection.We discuss these in the following sections.

Perceived Advantages of Random Selection
In our analysis, we identified a range of perceived advantages of random selection that illustrate the considerations that informed participants' support.First, participants' perceptions echoed the notion that random selection expresses and gives effect to core values of equality and fairness.Participants reported that a certain kind of person typically won elections in the past to the exclusion of others: those who were popular, had strong communication skills, and were perceived to be stronger leaders.A staff member reflected on how students with strong communication skills were typically elected: We select randomly so that people are chosen based on luck, and if we elected, we always elect those that know how to speak well.Even among ourselves [the teachers] this occurs."They know how to speak," or "You, you should take charge", we say, and those [teachers] that speak the least never participate.
Not only were specific types of students often selected in elections, but they could also feel more entitled to a role in student government.A student reflecting on those who ran for office at her previous school recounted: "It seems that they feel more capable and more powerful and that they are the only ones that are able to enter, and that this is the way it should be." Random selection helped overcome this inequality of opportunity through its impartiality, with no preference being given to any individual or group of individuals with certain characteristics or relationships.It opened the door for "any student" to participate, including those that "are very shy," "quiet," or "don't talk as much."Accordingly, a student who had recently served his first ever term as a member of the student government reflected that the use of random selection was the only reason he was able to participate: "The truth is that I wouldn't have been able to be here.If we hadn't used random selection, I wouldn't be here… [with elections], people would choose between friends." Increasing the equality of opportunity extended access to the valuable learning opportunities found in student government to different students than traditionally had those opportunities.In our analysis, we found that student government members reported developing a wide range of skills including teamwork, leadership, public speaking, and critical thinking.In this respect, the use of random selection was perceived as a way for all types of students to gain access to these learning opportunities: "One can be selected without any experience and can gain experience and learn because of random selection."This was seen as a reason for recommending that other schools use random selection, as noted by another student: "It should be done using random selection there too, because other students can improve and learn like we did, to lose their fear of public speaking."Relatedly, random selection exposed more students to the reality of governing and helped students empathize with those representing them: I think it's good in the sense that one doesn't only look on from their armchair, as the governed, because there are many who only criticize and make critiques and nothing else, because they never put themselves in the place of others.So, in this situation, using random selection and rotation helps with this, so that students put themselves in the place of others.
Second, participants' perceptions reflected the notion that random selection gives rise to political bodies that are more representative of the populations from which they are drawn.As alluded to above, the previous student government structure was not descriptively representative with respect to students who were less popular, had weaker communication skills, and were not perceived to be stronger leaders.
Random selection increased the representation of these students, which was observed by the research team and noted by participants.A staff member reflected on how things evolved over time: "There is [now] a big difference with everything.
In previous years, we elected those who always participated, those who spoke, those who were the most active Fourth, participants' perceptions echoed the notion that random selection can enhance the deliberative nature of political decision-making, both by avoiding problems associated with voting and by fostering more public-spirited student governments.Participants perceived that random selection reduced faulty decisionmaking by avoiding voting based on ignorance, misinformation or even apathy.
Elections were seen to incentivize parties to make campaign promises that they could not live up to, which led to students voting based on false promises: For me, I don't like elections.It's like we would be giving many things so that people would vote for a particular person.As if they would be giving something they can't accomplish.Saying things that you maybe can't accomplish.Candidates will say many things and in the end, they will follow through with maybe just one of those things.
Elections were also not always taken seriously.One participant from School B relayed that some students purposefully elected candidates who they thought would be unsuccessful in the role so as to tease them later: Student: I would prefer random selection because what we do at times with votes is we elect somebody to make fun of them because they won't be able to do the job.Interviewer: So, at times you elect somebody who you think won't be able to do a good job?Student: Exactly.They elect a person and think [that person] will fail and then they will laugh about everything.
Instead, random selection sidestepped such problems, leaving decisions about plans and priorities to be made by student government members through more deliberative processes.
Moreover, randomly selected governments were perceived as more likely to act in the interest of those they represent.Reflecting on the differences between the two government structures, a student stated that "there is a big difference.Parties don't do hardly anything to benefit the school and students, they just talk and have meetings among presidents and decide things among themselves and don't ask the opinions of the students-it's only among themselves." Fifth, random selection was seen by participants as less prone to conflict and social divisiveness among students, avoiding the tensions and hurt feelings brought on by campaigns.They noted that elections are based on arguments and campaigns in which candidates pit themselves against each other, with one student saying that "if students would vote, they would be arguing for this or that."These disagreements could cause resentment within the student body; however, "with random selection, everyone would go forward without resenting each other." Finally, participants viewed random selection as more efficient, particularly in that it requires much less time.One student reflected that "we delay sometimes when we do it with votes," and another noted that "using random selection is good, because with elections we delay more in selecting [student government members]." While the above perceptions echo previously theorized advantages of random selection, participants perceived two additional advantages of random selection that, to our knowledge, are absent or are undertheorized within published literature.First, randomly selected leaders were seen as having greater motivation to make an impact.Many students held that their previous elected governments achieved little and regularly fell below the expectations they set in their campaigns.A staff member reflected on how random selection motivated student government members to be more active in their roles: For me, selections that are rotative and random are much better because the students see the necessity and obligation to accomplish things and not let their classmates down, to be seen as more responsible, and they are more concerned about achieving activities.
Second, random selection was viewed as more enjoyable than elections.In this respect, students viewed the process of random selection-a public drawing of colored beans from an urn-as an entertaining group event.One student, for instance, remarked simply: "I like drawing beans-it's fun."In contrast, no participants described the electoral process as fun.

Perceived Disadvantages of Random Selection
Alongside advantages, students also perceived some disadvantages.First, participants noted that random selection inhibits the selection of those considered most capable-a fact relevant to them given perceived differences of competence among students.Accordingly, in selecting representatives, a student remarked that they "prefer voting, because we vote knowing who speaks well and is sociable."Another student echoed this sentiment, saying "I'm not in favor.I prefer a thousand times [that we select] by vote, because there are people who are capable." Second, participants expressed the view that randomly selected representatives may feel less responsibility or moral obligation to their office.As noted above, while some participants believed that those selected randomly are accountable, others suggested that students might enter the lottery without being sure they can complete their term, resulting in them being less committed to their role.One student reflected: "It's good, but at times it isn't.There are people who enter inadvertently and do not have time and make a joke of it all.While if you use elections, [the candidates] want to participate and do something productive in the school, and they participate wholeheartedly."Furthermore, in contrast with random selection, elections require candidates to have a clear, previously established platform and direction.One staff member suggested that "a positive part of the previous system could be that everyone presented a work proposal, elaborated a project, and had to defend their work plan for their whole term of office."Given that peers are aware of this platform, elected representatives may sense greater pressure to act responsibly.
Again, while these perceived disadvantages largely echo prior theorizing, stakeholders also perceived two disadvantages absent or undertheorized in prior theory.First, some participants perceived that randomly selected individuals have less popular support than those selected through election, where representatives are intentionally chosen.One student argued that it is important for student representatives to have popular support, and that elections help ensure that the student body supports their representatives: "To me, voting seems better.With voting, everyone agrees, the whole school agrees that it will be this group [of representatives]."A staff member echoed this sentiment: "I think it would be better if the election is done [by voting] because the student government is something that will represent all students and everyone should agree with the work of these students." Second, random selection was seen to inhibit personal control over one's chances at serving on the student government.One student reflected that he supported elections for this reason: "Some people want to enter the student government and have bad luck when trying to enter."In this sense, random selection reduces people's own agency and opportunities to increase their likelihood of being able to attain a role in government.

Contributions to Research
While we study a somewhat unique context, we argue that our findings offer a number of contributions to the growing body of research on random selection.First and foremost, the combination of the above findings-overall support for random selection informed by a critical assessment of its merits-suggests that legitimacy was in fact attained in the eyes of a majority of stakeholders.Accordingly, our findings suggest that, under certain conditions and in certain contemporary contexts, random selection can indeed be perceived as a legitimate means of selecting representatives for standing bodies with direct decision-making power.
While previous studies offer some indirect indication that randomly selected bodies might receive public acceptance, our results offer more direct evidence and are derived from actual conditions involving standing bodies with decision making authority.Our findings thus offer a clear reply to doubts within scholarship regarding perceived legitimacy and bolster the prospects of random selection as a viable democratic practice.The fact that random selection can be perceived as a legitimate, even preferable, democratic practice in the studied context suggests that it should perhaps feature more prominently among scholarly efforts to respond to popular disillusionment with contemporary representative democracy.
Findings regarding stakeholders' perceived advantages and disadvantages make further contributions to existing literature on the perceived legitimacy of random selection by revealing considerations that inform such perceptions.In this way, these results connect to and build on previous studies regarding the values that are thought to inform public perceptions of innovative democratic institutions.While other studies have pointed to the way in which support for mini-publics is tied to whether they are perceived to realize values of representativeness, expertise, fairness, and impartiality (Cutler et al., 2008;Levy, 2010), our findings suggest that other democratic values are also at play.Although the perceived advantages and disadvantages highlighted in this paper reaffirm the importance of the above, they also point to the significance of other democratic values such as participation, equal opportunity, and social harmony; more practical considerations like efficiency; and even the degree to which democratic processes or rituals capture the interest of those they implicate.In this way, these findings contribute to a more holistic understanding of citizens' democratic values and can inform the way in which future institutional design is approached and evaluated.Certainly, these values and considerations may carry different weight, and their perceived importance may indeed vary from one demographic to the next (Cutler et al., 2008); additional research is thus needed to add nuance in this respect.
At the same time, the advantages and disadvantages not perceived by stakeholders, as seen in contrast with those previously theorized, offers insight into the way in which context might shape perceptions.While in our paper stakeholders' assessments of random selection's advantages and disadvantages largely echoed prior theorizing, their perceptions did not exhaust the list pointed to by scholars, omitting, for example, advantages in relation to ideological partisanship or the influences of special interests.Given the particular setting, however, this is unsurprising.In this regard, our findings illustrate the fact that the particular advantages and disadvantages perceived-or not perceived-by stakeholders is likely influenced by the social context in which random selection occurs.Given the school context explored, it is unsurprising that stakeholders identified advantages that centred on the issues or values more pertinent to their context, such as participation, social cohesion, and distribution of learning opportunities.
Lastly, while our findings suggest that random selection can be seen as legitimate, they also suggest that there are important contextual factors that might influence perceptions.Previous writing from Carson and Martin (1999), Buchstein and Hein (2009) and Vandamme and colleagues (in press) suggested that perceptions of random selection are dynamic and may be influenced by factors such as familiarity, critical discussion, and culture more generally.While these findings are limited, they do lend some support to the idea that both familiarity and group culture may be influential.Despite maintaining overall support at both sites, random selection of representatives enjoyed greater support in the school community where stakeholders had greater proximity to the practice and its outcomes, where attendance was more consistent, and where local norms were more collectiveoriented.This was the case despite the fact that both schools engaged in critical discussion about random selection at the outset of their respective projects.Accordingly, it might be suggested that while theoretical familiarity with random selection may contribute to stakeholders' perceptions, both the ongoing experience with the practice as well as the group political culture in which it operates remain important factors.
Relatedly, the fact that random selection garnered greater support in the school community with more reliable attendance suggests that the social context and the impact that it has on random selection's practical merits is also influential.Given that random selection does not allow for the intentional (self-)selection of the most interested or committed, some participants may have viewed random selection less favourably in a context where commitment and reliability are at a premium, and thus perceived the practice as less appropriate overall.In this way, perceptions may be tied to circumstantial value, and efforts to determine when random selection may be perceived as legitimate might be effectively connected to emerging efforts to delineate when exactly random selection is an appropriate democratic practice (see e.g.Stone, 2016).While these findings offer some insight in this regard, further research ought to explore more closely the conditions necessary for random selection to be perceived as legitimate.
In addition to the above, our findings may offer some contributions to work on random selection not concerned directly with perceptions of legitimacy: in particular, scholarship regarding advantages and disadvantages of random selection more generally.First, our findings indicate that stakeholders perceived many previously theorized advantages and disadvantages of random selection-a fact which demonstrates that such advantages and disadvantages are of interest to more than simply academics, being grasped and understood by stakeholders on the ground as well.To the extent that perceptions reflect reality, this might also be taken to lend some empirical weight to theory.Second, our findings also articulate perceived advantages and disadvantages that, to our knowledge, do not appear in the literature or are at least undertheorized.Here, our findings might serve to instigate further theoretical attention to these points.

Limitations and Directions for Future Research
Our paper has noteworthy limitations that set the stage for further research.The first of these derives from our somewhat unique research context-student governments-which may limit the directness with which our findings apply to governance at the broader societal level.Certainly, the issues that student governments deal with are less complex and have lower stakes than the issues commonly faced by governments at the broader societal level.Moreover, the actions of student governments can be subject to formal and informal controls from teachers and administrators (McFarland & Starmanns, 2009).Accordingly, considerations around competence, accountability, or expertise may be more salient when related to other governing bodies than they were here.Schools also have an ingrained educational focus, potentially increasing the salience of considerations related to opportunities for learning and development relative to other contexts.Furthermore, schools are also smaller than other jurisdictions at a broader societal level, resulting in greater possibilities for students to be randomly selected and for the process of random selection to be more familiar, engaging, and fun than it might be at a municipal, state, or national level.
Nonetheless, based on the analysis of McFarland and Thomas (2006), student governments can be considered a "politically salient youth organization" (p.418) that "closely correspond[s] with adult political activities" (p.403).As with broader societal governments, student governments are standing bodies that exert some control and influence over their constituents through their decisions.Indeed, the functions of student governments, which tend to include organizing activities and events, advocating for students and their interests, making recommendations on school policies and practices, and mediating concerns between students and staff (McFarland & Starmanns, 2009), are analogous to the core tasks undertaken by governments at the broader societal level.The very fact that the norms for selecting governments typically mirror those of government more generally is also telling and, in combination with the above, suggests that a valid, albeit qualified, comparison can be made.Consequently, while we argue that our findings offer evidence of the potential for the random selection of representatives to be perceived as legitimate in the context of standing, decision-making bodies, we also argue that the above similarities and differences must be taken into account when assessing their generalizability.
Second, there remains the possibility that the perceptions we documented could in part be influenced by other variables.As discussed in our section on reflexivity, students' perceptions were potentially influenced through workshops and meetings with members of the organization.Additionally, participants' perceptions of random selection may have been influenced by the performance of the student government overall-a consideration which may indirectly incorporate the influence of parallel changes in the student government program, such as the capacity-building and support provided by the organization and the additional democratic practices of rotation and horizontal teams.This may be unavoidable in any case given the fact that institutional change never occurs in a vacuum, but it is nevertheless worth taking into account.
Third, as discussed in our section on reflexivity, there is the possibility that participants were influenced by a desire to provide socially desirable answers to the third author, given the pre-existing working relationship between organizational members and participants.As we argued earlier, we do not feel this had an adverse effect on our results for two main reasons: the third author maintained an open demeanor and regularly insisted that participants be open and honest, particularly in terms of expressing dissent, and many participants spoke for one of the practices adopted as part of the new governance structure and against another.
We thus recommend that future work be undertaken in different contexts, adopting a more controlled research design.In addition to overcoming the limitations of our design, future research could build off our contributions by investigating topics including additional contextual factors that may influence perceptions of the legitimacy of random selection, the relative salience of various considerations and their impact on overall perceptions of legitimacy, and the more novel advantages and disadvantages identified in this paper.

Conclusion
Scholars, practitioners, and citizens are increasingly seeking new ways to address the problems facing contemporary representative democracy.Based on its potential advantages, random selection has seen a significant growth in interest over the past few decades.Yet, despite this interest, concerns about the perceived legitimacy of randomly selecting representatives coupled with a lack of knowledge in this area hinder its further development and deployment.In this study, we helped address this by investigating stakeholders' perceptions of the shift from elections to random selection in selecting student government members in two Bolivian schools, finding that random selection was indeed perceived as a legitimate means of selecting representatives by a variety of stakeholders.Such perceptions were informed by a critical evaluation of random selection's merits on the ground and seemingly influenced by contextual factors.As a novel study in an emerging area of democratic thought and practice, we believe that these findings make important contributions to present and future work related to random selection.
[…]timid boys and girls were never chosen."In addition, random selection was seen to foster greater representativeness of women in government in School B, which did not have a gender-based quota in the past.In contrast, elections tended to prioritize males: "I'm not in favor of what the majority of schools are doing with parties, it's not good.There is no gender equality with respect to those who can enter [the student government]."Third, participants' perceptions reflected the view that random selection creates a more participatory form of democracy by lowering the threshold to stand for selection in two ways.First, random selection lowered the personal risks people face through elections.It avoids those who lose elections feeling a sense of rejection: "If somebody receives fewer votes they feel disappointed […] while with random selection, anybody can enter."Second, it also avoids interested students feeling discomfort by being under the spotlight and being judged by their peers.
One student, for instance, reported that this type of situation would "make [him] feel uncomfortable because [he] wouldn't be able to enter without having the admiration of [other students]."Accordingly, participation in student government increased substantially.As noted earlier, School B had only elected student governments twice during the past 28 years.One student from this school reflected: "I think [random selection] is good, because from what I saw there weren't many candidates that wanted to enter the student government."During the study period, there was a high level of interest in participating in the student governments at both schools.There were more positions open per year than in the past due to the use of rotation, and all positions always had more than enough interested volunteers.