INTRODUCTION
The “rank and file,” or the “armies of volunteers on the ground” (Auerbach et al. Reference Auerbach, Bussell, Chauchard, Jensenius, Nellis, Schneider and Sircar2021, 8) are key to effective ground campaigns in modern elections (Brierley and Nathan Reference Brierley and Nathan2022; Goyal Reference Goyal2023), a pipeline for party leadership (McKenna and Han Reference McKenna and Han2014; Weghorst Reference Weghorst2022), and, therefore, instrumental in shaping party development in the long run (Aldrich Reference Aldrich2011; Thachil Reference Thachil2014).Footnote 1 Despite the crucial role of these personnel in representative democracy, there is a dearth of evidence on how party leaders confront the large human resource management work of recruiting the rank and file before campaigns can actually be run. While political scientists have demonstrated that party leaders exercise a high degree of control in the selection of candidates (Cohen et al. Reference Cohen, Karol, Noel and Zaller2009; Dancygier et al. Reference Dancygier, Lindgren, Oskarsson and Vernby2015; Gulzar Reference Gulzar2021), relatively little is known about whether similar leverage can extend to recruiting the rank and file.
Perhaps one reason for this gap in the literature is the theoretical expectation that without financial incentives, or the prize of a party nomination, leaders have little ability to recruit a high number of quality rank and file. They must, instead, be content with whoever puts themselves forward (Chewning et al. Reference Chewning, Green, Hassell and Miles2024; Enos and Hersh Reference Enos and Hersh2015; Hannah, Reuning, and Whitesell Reference Hannah, Reuning and Whitesell2023; McKenna and Han Reference McKenna and Han2014; Neuenschwander and Foos Reference Neuenschwander and Foos2021). However, this reasoning is in contrast with modern campaigns around the world that run as highly sophisticated organizations targeting individual voters (Cheema et al. Reference Cheema, Khan, Liaqat and Mohmand2023; Enos and Hersh Reference Enos and Hersh2015; Jha Reference Jha2017; Kalla and Broockman Reference Kalla and Broockman2018; Thachil Reference Thachil2014), suggesting that elites are perhaps already finding high-quality people to deliver those campaigns.Footnote 2
This article attempts to fill the gap between practice and research on the recruitment of the rank and file by examining if party leaders indeed possess limited ability in recruiting them in the absence of prized incentives. Doing so is not straightforward. First, party elites in the status quo may already respond to a constrained supply of interested recruits. Examining how their actions shape the rank and file is, therefore, difficult without exogenous variation in effort exerted toward recruitment. Second, the problem of observing these efforts is compounded by a lack of researcher access to internal processes driving the modern party bureaucracy (Cantoni and Pons Reference Cantoni and Pons2021; Gulzar, Hai, and Paudel Reference Gulzar, Hai and Paudel2021; Wantchekon Reference Wantchekon2003).Footnote 3 As a result, scholars note that extant research on ground campaigns examines the effects of direct voter outreach in great detail (Gerber and Green Reference Gerber and Green2000) (e.g., get-out-the-vote campaigns), but the study of strategy behind campaigns is thin (Enos and Hersh Reference Enos and Hersh2015). That is, to the best of our knowledge, there exists no prior research that provides quantitative causal evidence on real-world rank-and-file recruitment efforts by political parties.
We present evidence from a recruitment experiment carried out by a political party in India at scale. We ask if party elites can effectively exercise levers at their disposal to shape the rank and file, or if their hands are tied by whoever puts themselves forward as a volunteer.Footnote 4 The results provide robust evidence that not only can party elites shape the quantity and quality of the rank and file, but they can do so with nonmonetary incentives at hand.
We overcome the empirical and theoretical challenges identified above in several ways. Collaborating with a political party allows us to closely examine how party elites recruit ground activists at scale in the status quo. In anticipation of state elections in Jharkhand, the Aam Aadmi Party (AAP) set out to enlist an army of party workers by distributing pamphlets among nearly 150,000 of the state’s 13-million voters, reaching about 1% of the electorate. We uncover the internal workings of party organizations by using survey data on these recruits and information on Jharkhand demographics to describe status-quo recruitment. Ours is perhaps the largest survey of party rank and file in any developing democracy to date.
We make progress on the problem of causally evaluating the efficacy of party leader efforts in shaping the rank and file by studying a field experiment that the party built into the recruitment exercise. A key problem political parties face in the developing world is the recruitment and retention of women (Goyal Reference Goyal2023; Goyal and Sells Reference Goyal and Sells2023).Footnote 5 In addition to pamphlets recruiting party workers under the status quo where the stories of existing male party workers are highlighted, in select areas party leaders randomly distributed pamphlets which, instead, emphasized the stories of its women workers. This allows us to study how signaling the inclusiveness of the party impacts who decides to join. In addition to this first dimension of messaging, the party also cross-randomized the portrayal of benefits that can accrue to a prospective recruit from joining the party. These treatment pamphlets randomly emphasized four benefits of joining: the party’s ideology around its anti-corruption brand, the chance of becoming a candidate, the policy influence that party workers can have, or gathering political work experience.
We begin the analysis with the key theoretical puzzle of whether efforts by party leaders can shape the rank and file at all, or if their efforts will be moot in the absence of high-powered incentives. During mobilization, prospective recruits were asked to call back later at a number if they were interested in joining the party, our main outcome. On receiving the call, the party administered a brief survey allowing us to characterize the diversity and skills of the recruited pool. Three years later, party leaders carried out a follow-up call enabling the measurement of treatment effects over the long run. We establish a null hypothesis that diverging from the status-quo way of recruiting the rank and file will not yield party members in different quantities or compositions. We test this by comparing the joint significance of all treatments implemented by AAP against status-quo recruitment.
We find that party elite efforts at recruitment yield not only a statistically and substantively different size of the recruit pool but one that differs in its composition as measured through abilities and diversity. We also find that these actions continued to matter 3 years after the recruitment campaign. Taken together, this is compelling evidence that under real-world conditions where the party cannot offer financial incentives, party elites’ efforts can still shape the rank and file in meaningful ways.
While our primary question stressed in general whether party leaders have the ability to move the needle on who is in the rank and file, we also examine secondary questions of whether elites can shape the recruitment—that is, can they pull levers at their disposal to yield individuals who have specific backgrounds or abilities? To answer this, we compare treatment and status-quo pamphlets across two dimensions: signaling features of the party organization such as gender inclusiveness and emphasizing the benefits of joining.
In a context where politics is the domain of men, seeing that a party is inclusive can encourage women and other marginalized communities to join by increasing their perceived sense of political efficacy, likelihood of being appointed to a position of influence, or chances of having a harassment-free work environment. The impacts on the skills of the recruited pool are ambiguous: on the one hand, broadening politics can create space for high-caliber people from marginalized backgrounds, but, on the other hand, the traditional fear is that people from these groups possess few skills.
We first find that pamphlets emphasizing gender inclusiveness of the party were successful at recruiting more rank and file. Second, the treatment was effective at recruiting people from excluded groups, including marginalized castes and tribes and women. The latter is particularly significant because moving the needle on bringing more women into political parties is seen as a especially difficult problem. Third, we find that, contrary to concerns that broadening access will reduce quality, the skills profile of recruits also improves. Fourth, examining long-term impacts, we find that the treatment was remarkably successful at recruiting people who continued to be available for the party up to 3 years later. However, while these long-term effects arise particularly among marginalized castes and tribes, which is significant for broadening politics in the long run, we, nonetheless, observe that the positive effects on the recruitment of women attenuate completely. We conclude that signaling inclusiveness can be effective at making progress on the immediate, seemingly intractable, recruitment of women, but the party must continue to exert effort at maintaining the pipeline.
Besides the impacts of signaling inclusiveness, we also study how varying the salience of potential benefits to prospective recruits affects whether they join the party. The general prediction here is that these efforts will not produce much change because no monetary benefits are being offered and recruits must bear significant costs to work for the party.Footnote 6 Any minimal movement we might observe will be from low-skilled people.
In contrast to these predictions, we find that relative to the status-quo pamphlet, party leaders are able to increase the number of new recruits—particularly those who belong to excluded groups and those who possess political skills—if they signal the ideological and potential candidacy returns from joining the party. Signaling policy priorities of the party, however, reduced the recruited pool on all dimensions. A potential interpretation of these effects is that ideological and candidacy benefits accrue regardless of electoral performance, but that policy returns were not credible since the party had a minimal chance of forming government. Finally, unlike the female pamphlet, we observe no evidence that these effects persist in the long run.
While our analysis focuses on recruitment by AAP in India, the results carry lessons for a broad set of cases. First, it is worth noting that the recruitment campaign we study is part of AAP’s party-building process across India. On April 11, 2023, the Election Commission recognized AAP as a national-level party by reserving its election symbol for races across the country.Footnote 7 AAP’s party-building work is typical of India’s dynamic party system where even the nationally ruling party does not have a presence in all constituencies and where parties are constantly battling over new turf and poaching politicians from each other. Second, within India, all parties employ a similar pamphlet distribution method for recruitment (see Appendix A.2 of the Supplementary Material for examples from BJP), and leverage the rank and file in competing door-to-door campaigns before election day (Auerbach et al. Reference Auerbach, Bussell, Chauchard, Jensenius, Nellis, Schneider and Sircar2021; Sahu Reference Sahu2019; Yadav Reference Yadav2019). Third, experts of Indian politics also note that the composition of AAP’s membership is remarkably similar to other, more established Indian parties.Footnote 8 Finally, recruitment is not only an Indian phenomenon, as we describe more in the following section. Just as the Indian Bharatiya Janata Party (BJP), long branded as an upper-caste, elite party, managed to incorporate and promote historically marginalized groups within its ranks (Thachil Reference Thachil2014), the Argentinian Justicialista (Peronist) Party’s leadership was successful in weakening its union links and recruiting workers from low-income communities (Levitsky Reference Levitsky2001). Strategic recruitment of rank-and-file party workers remains an important tool for parties to credibly promote new policies and mobilize new electorates as economies transform, demographics shift, and windows of discourse expand.
Contributions
This article makes theoretical and empirical contributions to the literature on political campaigns (Aldrich Reference Aldrich2011; Auerbach and Thachil Reference Auerbach and Thachil2018; Chewning et al. Reference Chewning, Green, Hassell and Miles2024; Goyal Reference Goyal2023; Hannah, Reuning, and Whitesell Reference Hannah, Reuning and Whitesell2023; Hirschman Reference Hirschman1970; Ichino and Nathan Reference Ichino and Nathan2012; Kalla and Broockman Reference Kalla and Broockman2018; Stokes Reference Stokes1999). First, there is relatively limited work on ground campaigns, particularly from low- and middle-income contexts (Cheema et al. Reference Cheema, Khan, Liaqat and Mohmand2023; Palmer-Rubin, Garay, and Poertner Reference Palmer-Rubin, Garay and Poertner2021; Thachil Reference Thachil2014). We theoretically clarify arguments for why leaders’ ability to affect change in the rank and file can be limited, and how this expectation contrasts with other domains like candidate selection where party leaders are supposed to exert great influence. We test this claim empirically and present at-scale evidence against it.Footnote 9
Second, studies on political campaigns focus on the impacts of ground campaigns on voters rather the strategies behind them as Enos and Hersh (Reference Enos and Hersh2015) note. In at least two ways, our work departs from this by examining strategies of running campaigns. First, we propose and provide evidence that even lacking financial resources, party leader efforts can matter. Second, we speak to the problem that as elite organizations, the personnel of parties are less diverse than the electorate, potentially limiting their ability to reach new groups (Rokkan and Valen Reference Rokkan and Valen1962). Our results indicate that one campaign strategy could be to diversify the party through signaling inclusiveness.Footnote 10 This could have downstream benefits; recent evidence from India shows that the identity of campaigners in parties matters for persuading and mobilizing people (Goyal Reference Goyal2023), though evidence from other contexts suggests that this might not always be the case (Broockman et al. Reference Broockman, Kalla, Ottone, Santoro and Weiss2023; Broockman and Kalla Reference Broockman and Kalla2016).
Third, beyond political campaigns, political scientists have long examined how political parties are organized and the consequences of that for democratic performance (Aldrich Reference Aldrich2011). Recently, scholars have turned their attention toward party personnel, focusing, for instance, on the selection of political candidates (see, e.g., Auerbach and Thachil Reference Auerbach and Thachil2018; Goyal Reference Goyal2023; Gulzar Reference Gulzar2021; Ichino and Nathan Reference Ichino and Nathan2012), and also elite preferences and behavior (Broockman and Skovron Reference Broockman and Skovron2018; Gulzar, Hai, and Paudel Reference Gulzar, Hai and Paudel2021; Pereira Reference Pereira2021) during the candidate nomination and policymaking process. Without a deeper understanding of how parties recruit from the ground up, how they build and motivate these people, and how these people rise through the party’s ranks, it is not possible to fully characterize the core functions and evolution of political parties. Our article presents, to the best of our knowledge, the first quantitative causal study of the initial stage of political selection: the recruitment and motivation of frontline rank and file.
Finally, we also contribute to the study of political parties in India, and South Asia more broadly. Challenging the notion that political parties in the region can be boxed into characterizations of “clientelism” and “weak parties,” scholars have issued a call to unpack the foundations of party-building in South Asia (Auerbach et al. Reference Auerbach, Bussell, Chauchard, Jensenius, Nellis, Schneider and Sircar2021; Chhibber and Verma Reference Chhibber and Verma2018; Mufti, Shafqat, and Siddiqui Reference Mufti, Shafqat and Siddiqui2020). Among other directions, they argue that more work is needed to examine recruitment strategies related to “selective material incentives, ethnicity, ideology, or leadership charisma,” and whether they attract committed and capable party workers (Auerbach et al. Reference Auerbach, Bussell, Chauchard, Jensenius, Nellis, Schneider and Sircar2021, 10).Footnote 11 This article directly answers this call to action.
RECRUITMENT TO PARTY RANK AND FILE
The Party Rank and File and the Aam Aadmi Party
Who are the party rank and file? We differentiate three tiers in political parties: the national leadership, whose responsibilities include candidate nominations, national campaigns, communication, and policy research (Katz and Mair Reference Katz and Mair1993); the heterogeneous mid-level members,Footnote 12 who work at regional levels to organize rallies, mobilize supporters on election day, and make government benefits accessible to voters; and the rank-and-file party activists who campaign door-to-door, attend rallies, work phone banks, and check polling stations on election day (Auerbach et al. Reference Auerbach, Bussell, Chauchard, Jensenius, Nellis, Schneider and Sircar2021; Enos and Hersh Reference Enos and Hersh2015).Footnote 13 The recruitment of thousands of these rank-and-file members is a relatively new phenomenon in political campaigns, gaining traction over the last decade and a half.
The rank and file also play a crucial role in electoral campaigns for AAP, literally the “Common Man’s Party.” AAP formed from the India Against Corruption social movement in 2013 and continues to expand across India. Like most medium-sized parties, AAP has a complex bureaucratic structure to govern its affairs. Figure 1 displays the party’s national- and state-level organizational structure.
Figure 1. Political Units in India and Concurrent AAP Positions
AAP staffed its Jharkhand campaign, the focus of our study, with leaders—called “assembly-in-charges”—that managed overall affairs within the 60 of 81 state constituencies where AAP built its party organization. A total of 523 vice presidents (which we refer to as the “middle-level members”), reported directly to these assembly-in-charges, with multiple vice presidents per assembly. These vice presidents manage workers and party outreach within approximately five village clusters (gram panchayats) each, and are responsible for implementing pamphlet-based recruitment campaigns like the one we study.
At the bottom are the rank and file. These individuals operate in teams, coordinating activities under the supervision of their superiors. Before elections, each rank-and-file member canvasses 20–50 households from electoral rolls, identifying supporters. Citizens have a rare “opportunity to rate the performance of the powerful” (Banerjee Reference Banerjee2017, 42) as millions of party workers in India, and across the world, funnel this feedback upward. With all of this information, the party and rank-and-file members ensure platforms respond to voters’ preferences and supporters turn out on election day.Footnote 14 According to AAP leaders, rank-and-file members’ “main job on [election] day is to help voters find the right polling booth. They have to manage breakfast and lunch for the party. They have to monitor all the booths whether something is going wrong. […] AAP has focused on breaking down [illegal] distribution of liquor and money and party workers [have to] remain observant.”Footnote 15 After elections, some party workers continue their work as liaisons between elites, elected officials, and voters. They provide a crucial democratic “feedback loop,” allowing one elected representative we spoke with to be immediately clued into the issues facing voters in his constituency.Footnote 16
The Calculus of Joining the Rank and File
How do people decide to join the rank and file? One way to conceptualize this decision problem is to assume that individuals want to maximize the utility from how they spend their time. They would, therefore, weigh the benefits and costs of joining the rank and file against other uses of their time and effort.Footnote 17 The first consideration for prospective recruits relates to the benefits they expect to receive from joining the rank and file. Typically, benefits from any job are thought of in terms of financial compensation. As the party rank and file are nonsalaried volunteers, our study focuses on nonpecuniary returns that we detail in a coming section. Individuals also consider the costs of participating. Time costs relate to work done prior to and on election day, and nonelection periods. Recruits for AAP, as other Indian parties, typically bear their own monetary costs associated with travel and food without expectation of compensation. As one veteran worker put it, “The [Biju Janata Dal party] takes care of my needs and has been with me through thick and thin. But I don’t depend on it for a living” (Sahu Reference Sahu2019). Individuals weigh the relative benefits and costs against alternative uses of their time and energy: the opportunity costs. As shown in related literature on candidate selection, individuals with high opportunity costs are considered more “competent,” having higher outside income and more education. Conversely, individuals with limited alternative uses of their time may be less “qualified” and effective at the job of being a party volunteer, yet more likely to sign up.
Why Shaping the Rank and File Is Difficult
At its core, the problem from the party leaders’ perspective is that they need to recruit thousands of competent rank-and-file members before elections. Hannah, Reuning, and Whitesell (Reference Hannah, Reuning and Whitesell2023) write that “all local parties face constraints on limited resources, and the ability to recruit and train new volunteers may be a function of organizational capacity” (931). Weir and Ganz (Reference Weir and Ganz2017) highlight a campaign training document in Ohio which says: “We cannot achieve the sheer volume of what we need in order to win without [volunteers’] help” (8).
On the benefits side, parties cannot afford to pay the large number of people they need to recruit and must rely on volunteerism.Footnote 18 This presents party leaders with a challenge in attracting high-quality workers. Scholars studying the United States noted this challenge even before the mass mobilization of Obama’s 2008 campaign. McKenna and Han (Reference McKenna and Han2014), describing campaigns around 2002, write that “most voter contact operations were contingent on the amount of paid staff a campaign could hire” (33). The authors quote Weir and Ganz (Reference Weir and Ganz2017) to note that this meant campaigns had a hard time absorbing “ordinary people into party-run field efforts” (211) who could mobilize many others in the lead-up to and on election day.
On the costs side, too, party leaders face challenges. People with high opportunity costs have greater skills and education, limiting the pool of high-quality recruits. In addition, the work is challenging—recruits toil day in and day out, knocking on doors, staffing rallies, and fulfilling unplanned tasks that inevitably arise. McKenna and Han (Reference McKenna and Han2014) write that “it’s hard, grueling…work to build a truly grassroots national organization—and that is why so few campaigns choose to do it” (ix). AAP party leaders echoed similar concerns in interviews, saying they delicately balance their demands of volunteers not to overwhelm the rank and file’s finite economic and temporal resources.
Given the limited ability of leaders to pay and these high costs incurred, it is reasonable that few would want this job, and that those who do have limited experience or education. We see this in less institutionalized settings where candidates’ own family and social networks staff elections. But to contest elections institutionally at scale, parties must resolve this recruitment challenge. This motivates our core question: can elites strategically manipulate the size and composition of the rank and file, and if so, how?
How Parties Can Recruit in the Absence of Pecuniary Benefits
In the face of a limited ability to offer high-powered incentives, we ask if signals from party elites of potential nonpecuniary benefits from joining the party can shape who decides to join. As parties institutionalize, leaders can engage in strategic mobilization of the citizenry to develop a robust bureaucratic base for electoral mobilization (Rosenstone and Hansen Reference Rosenstone and Hansen1993). To do so, parties use door-to-door canvassing to easily relay information about how to join (Whiteley and Seyd Reference Whiteley and Seyd1992). Other common modes of outreach include membership forms on websites or at local offices, or mailing lists in high-income contexts (Neuenschwander and Foos Reference Neuenschwander and Foos2021).
Previous work discusses whether the content of mobilization materials can impact how many and which people respond to messaging (Broockman Reference Broockman2014; Broockman et al. Reference Broockman, Kalla, Ottone, Santoro and Weiss2023; Broockman and Kalla Reference Broockman and Kalla2016; Palmer-Rubin, Garay, and Poertner Reference Palmer-Rubin, Garay and Poertner2021; Preece and Stoddard Reference Preece and Stoddard2015). Most, however, focus on the recruitment of political candidates (Broockman Reference Broockman2014) or middle management (Brierley and Nathan Reference Brierley and Nathan2021), instead of the rank and file, perhaps because the large-scale recruitment of the rank and file is a recent phenomenon, as McKenna and Han (Reference McKenna and Han2014), Enos and Hersh (Reference Enos and Hersh2015), and others document regarding the Obama campaign. While the appeals parties make to prospective recruits can take many forms, our focus in this article is on two dimensions. The first is a signal about the features of the party organization—here, a salient dimension is its gender inclusiveness. The second is a signal about the benefits that accrue to the recruit. We also consider the interaction between these two dimensions: they can reinforce each other, or substitute the perceived returns if seen to be in conflict.
Highlighting Inclusiveness of the Party
Political parties are disproportionately staffed by men and a large gender gap in party membership persists in many parts of the world—for instance, Figure 1 in Goyal and Sells (Reference Goyal and Sells2023) shows that parties in Brazil have about a third more men than women as activists, while Table 1 in Goyal (Reference Goyal2023) shows that women report 18% less contact by activists than men in India, and this gap arises with male-only activist teams.Footnote 19 This is perhaps because political party elites have not prioritized the recruitment of women as a key feature of their party organization, a limitation elites are increasingly sensitive to.
From the party’s perspective, recruiting women is beneficial in the long run because it potentially builds a bench of candidates that better represent the interests of voters. This is especially true in places like India where the gender gap in voting is shrinking. On the more immediate level, parties care about recruiting a more gender-balanced rank and file because of how they might mobilize voters differently. In a context most similar to ours, Goyal (Reference Goyal2023) shows that women staffing ground campaigns can be an important determinant of the campaign’s efficacy at reaching women voters. Qualitative accounts from India also suggest that women workers might be important agents in mobilizing women. In Outlook’s profiles of the rank and file, a leader of West Bengal’s All India Trinamool Congress (TMC) praised a young female worker for playing “a major role in bringing women in villages …into the Trinamool fold.” Her goal was “to ensure every woman in our polling booth casts her vote on [election day]” (Pramanik Reference Pramanik2019). Evidence from other contexts, however, is more mixed and signals the need for more research. For instance, Broockman and Kalla (Reference Broockman and Kalla2016) show that, in the United States, transgender and nontransgender canvassers were both equally effective at reducing transphobia.Footnote 20
From the recruits’ perspective, there are reasons to believe that gender-inclusive recruitment by a party can be successful at attracting women volunteers. First, it could boost the chances that women are recruited at all (Fox and Lawless Reference Fox and Lawless2010). Second, it could increase women’s perceived intrinsic returns from joining, for example, by raising their ambition to participate in politics (Bonneau and Kanthak Reference Bonneau and Kanthak2020) and increasing their sense of political efficacy (Campbell and Wolbrecht Reference Campbell and Wolbrecht2006). Third, women recruits might also anticipate greater instrumental returns from joining—for instance, by increasing the perception that women are likely to be influential in setting the party’s policy (Clayton and Zetterberg Reference Clayton and Zetterberg2018) or being effective if nominated to positions of leadership (Beaman et al. Reference Beaman, Chattopadhyay, Duflo, Pande and Topalova2009). Finally, inclusive recruitment can also lower the costs of joining the party, for example, by improving the perception of on-the-job conditions such as harassment (Håkansson Reference Håkansson2021) and enabling bargaining with gatekeepers, such as parents or husbands (Cheema et al. Reference Cheema, Khan, Liaqat and Mohmand2023).
Beyond women, our conceptualization of “excluded groups” encapsulates gender, caste, and religious groups, all of whom remain underrepresented in the ranks of political parties in India. While the historical reasons for these groups’ exclusion from political and economic spheres differ, the below predictions assume that similar processes occur among diverse individuals when parties signal positive incentives to them.Footnote 21
Highlighting gender inclusivity can, therefore, signal a welcoming party for broader set of marginalized groups. Adida, Davenport, and McClendon (Reference Adida, Davenport and McClendon2016) show that African Americans in the United States respond positively to co-minority (Latino) political cues, while Brulé and Toth (Reference Brulé and Toth2022) show that political quotas along two identity dimensions are more likely to build ethnic and gender solidarity in India. Thus, AAP’s gender-inclusive recruitment material may also attract other marginalized groups through similar channels of change.
That leaves the status-quo, likely male, recruits who would join more if they value inclusiveness of parties. However, it is possible that they become less likely to join if they perceive that the party is no longer speaking in their interests. Taken together, we hypothesize that the overall effect of gender-inclusive recruitment is dependent on how strong this backlash is from non-excluded groups compared to the increase in membership from women, excluded castes, and excluded religions.
What remains unclear is how this recruitment would affect the skill composition of the pool vis-à-vis status-quo recruitment. The logic for this is not straightforward because the predictions depend on the average level of skills in the status quo, as well as the skills of the marginal recruit. There are two possibilities. On the one hand, inducting more marginalized recruits, who are typically also excluded from educational opportunities (Jensenius Reference Jensenius2015), can lower the average skills of the rank and file. On the other hand, if the status quo is biased toward recruiting men, even if their quality is low (or “mediocre” per Besley et al. Reference Besley, Folke, Persson and Rickne2017), it is possible that including excluded groups in the rank and file could actually boost the average quality of the recruits.Footnote 22
Highlighting Benefits from Joining the Party
While the first dimension of recruitment aims to signal the party’s inclusiveness, recruitment materials also allow citizens to directly visualize the personal gains of membership. We now examine how such nonpecuniary returns could tip the threshold for individuals to engage in costly participation. We draw four examples of benefits from a long literature on party organizations: ideology, candidacy, policy, and career. As described above, citizens will weigh these perceived returns against the cost of joining the party to make their decision to join.
To start, parties can signal the broad ideological basis of their brand to recruits. These are typically valence issues that have broad appeal. The literature agrees that some “benefit-seekers” derive value from a party’s ideas (see Stokes Reference Stokes1999, 249 for a review), or join organizations due to psychological attachments or affect (Costa Reference Costa2021). Belonging to what Putnam (Reference Putnam2000) calls “bridging (or inclusive)” networks, like those from a political party offering broad ideas of a community, may appeal in particular to those from excluded groups who lack representation in other political networks. Past work indicates that ideological appeals may be more attractive to people with lower opportunity costs (Shayo Reference Shayo2009), suggesting that ideological signals can result in low-skilled recruits.
Party work also allows individuals to demonstrate their skills and loyalty for the purpose of candidacy. Party leaders exercise considerable power in deciding party nominations (Cohen et al. Reference Cohen, Karol, Noel and Zaller2009; Gulzar, Hai, and Paudel Reference Gulzar, Hai and Paudel2021). Potential future perks of elected office—power, prestige, and personal rents—are attractive to new recruits (Gulzar and Khan Reference Gulzar and Khan2024; Truex Reference Truex2014). Even long-shot candidates can glean intrinsic benefits, such as being known as leaders, just from running (Weghorst Reference Weghorst2022). As parties do not typically advertise candidacy prospects broadly, highlighting this benefit can also induce people from politically nontraditional backgrounds to join the party, both women and excluded ethnic groups who do not typically see themselves reflected in parties’ nominees. But, as was the case above, it is not clear if recruits will possess key skills. On the one hand, high-skilled people from politically excluded groups could join, while, on the other hand, low-skilled people with low opportunity costs could also be induced to show up.
Perhaps the most obvious reason to join a political party is the benefit of actualizing one’s policy preferences (Lupu Reference Lupu2013). This benefit can accrue from joining a party with an aligned platform and working to improve its chances of forming the “winner’s coalition” (Aldrich Reference Aldrich2011). Though the potential to affect policy change may be alluring, highlighting specific priorities can also alienate prospective recruits for two reasons. First, those with divergent preferences on specific policies (versus a more generic “ideology” appeal) might be turned off; and second, highlighting policy preferences without specifying a path to implementation (such as winning sufficient seats to be a part of the governing coalition) can make the appeal seem disingenuous. Nevertheless, as policy agendas are typically framed as pro-poor, it is conceivable they can attract economically marginalized populations. In our case, the policy priorities highlighted by AAP of improved schools, healthcare, and environmental protection were likely to appeal to women, Scheduled Castes, and Scheduled Tribes. The skills of the recruited pool can increase if competent individuals from excluded groups displace the status-quo recruits. It may also decrease if the reverse happens.
Parties may also signal rank-and-file work as a first step toward a political career. As parties become more bureaucratized, professional staff are increasingly critical to their long-term development. But becoming a party worker even before elections allows individuals to immediately gather experience relevant to subsequent jobs, including political ones. We label these types of benefits career incentives. People who take advantage of these opportunities may be referred to as machine-type “professionals” by Wilson (Reference Wilson1962). Career benefits can appeal to a wide range of individuals—as the experience they gain is not necessarily contingent on the party winning—but may appeal most to those with low opportunity costs like excluded groups. This is because those who are already quite skilled may have a high opportunity cost of time, such as by already being fully employed.
How the Two Dimensions Interact
While the focus of this article is on the marginal effects of each dimension, it is possible that the effects of highlighting specific features of the party’s organization (the first dimension) together with the benefits that recruits can gain from joining (the second dimension) could interact in conflicting or reinforcing ways.
First, the two signals could complement each other’s effects. For instance, research on recruiting women for candidacy suggests that a party that signals gender inclusiveness and highlights the potential for candidacy may receive more sign-ups than it would if it only signaled one of those dimensions (Preece, Stoddard, and Fisher Reference Preece, Stoddard and Fisher2016).
In addition, it is also conceivable that signals on one dimension negate the other’s efficacy. For instance, consider the same example as above. First, the salience of gender inclusiveness (especially in contexts like South Asia) could dominate prospective recruits’ attention, rendering additional information a distraction (Taylor and Fiske Reference Taylor and Fiske1978). Second, it could be that the two dimensions are interpreted to be in conflict with one another, resulting in a backlash effect (Rudman and Fairchild Reference Rudman and Fairchild2004). Highlighting women can dissuade status-quo (male) recruits from joining if they do not support broadening the rank and file, as other studies in India have indicated may be the case for women’s increased political rights and leadership (Beaman et al. Reference Beaman, Chattopadhyay, Duflo, Pande and Topalova2009; Brulé Reference Brulé2020).
Summary of Empirical Predictions
Our null hypothesis from the discussion in “Why Shaping the Rank and File Is Difficult” is that party leader effort is not likely to change the size and composition of the rank and file. Nevertheless, we discussed that if party elites choose to mobilize nonpecuniary levers at their disposal, the literature can offer us some predictions on the likely direction of these effects. We summarize these in Table 1. Overall, highlighting benefits should not decrease the total number of new recruits, and the key tradeoff party leaders may face is that diversifying the party may decrease the average skill of recruits.
Table 1. Potential Direction of Effects According to the Literature
AAP’S STATUS-QUO RECRUITMENT IN JHARKHAND
We examine AAP’s recruitment campaign run in early 2018 in anticipation of Jharkhand’s state and national elections in 2019 and 2020. As their first foray into Jharkhand politics, their aim was to build a cadre of local-level recruits in the state.
AAP leadership proceeded in their usual manner of recruiting local party workers by distributing pamphlets and discussing the points on the pamphlet when doing so. Their status-quo campaign closely resembled those of other parties across India, where recruitment is characterized by the mass, face-to-face dissemination of printed materials plastered with the faces of prominent party (male) leaders, and personal conversations (see Appendix A.2 of the Supplementary Material).
Pamphlets were double-sided and A4-sized, and appear in Figure 2 and in Appendix A.1 of the Supplementary Material. One side displayed the party president, highlighting the party’s achievements in its national stronghold of Delhi and its brand (a broom icon serves as its official election symbol and represents its anti-corruption stance). On the other side, the pamphlet offered a current party member’s photo and a short biography above a broad description of what becoming a party member could look like. Pamphlets made little attempt to signal the party’s inclusion to women or non-Brahmins, as all photos depicted relatively light-skinned men of ambiguous-to-high Hindu caste. Below was a phone number that a prospective recruit could call and hang up to signal interest (locally this is referred to as giving a “missed call”). AAP, like other parties in India that maintain contact with members via phones, would then use the list of phone numbers to ask interested individuals to complete a survey as their first party task.
Figure 2. Front and Back of Baseline Pamphlet
The photographs in Figure 3 show scenes of face-to-face pamphlet distribution and the accompanying conversations about the points in the pamphlets as they occurred in people’s homes and in public areas from the summer and fall of 2018. In total, the party distributed about 150,000 pamphlets.Footnote 23 AAP Jharkhand mid-level members (vice presidents) who handed out pamphlets in their assigned areas on the ground were mostly men. Given social norms, it is likely that most conversations these mid-level members had were with men, too.
Figure 3. Pamphlet Distribution on the Ground in Jharkhand
Partnering directly with a political party allows us to descriptively characterize this status-quo recruitment drive with multiple waves of surveys. In subsequent sections, we describe whether the party can strategically manipulate this pool of people. Appendix B of the Supplementary Material provides an extended discussion on ethics.
Data
For the main outcomes, we combine data from a Dataset on Pamphlet Distribution and a Phone Number Database compiled during the experiment. The first records the number of pamphlets that each mid-level member (vice president) handed out. AAP confirmed through weekly phone calls the number of pamphlets each mid-level member handed out, and all pamphlets received were distributed in the field. To study the efficacy of each pamphlet handed out, we match pamphlets distributed to phone calls received from new members. This was possible because each treatment arm had a unique phone number associated with it.
To begin onboarding those who expressed interest, AAP set up a call center in Delhi to administer a survey that would be party recruits’ first task. This Onboarding Survey collects information on new members’ demographics, policy preferences, and previous political engagement for usage by party leaders. Senior party leaders told researchers that members recruited in this 2018 drive led door-to-door campaigning and election-day mobilization in the 2019 elections, and remained at the disposal of the party. AAP also conducted a Long-Term Retention Survey at the 3-year mark (2021–22) to those reached in the first wave to assess their utility in serving across electoral cycles. See Appendix C of the Supplementary Material for further details.
Outcomes
We examine three categories of outcomes. First, we analyze whether or not a pamphlet yielded a “new recruit” (measuring the “size” of the pool). The primary measure of size is a binary outcome that indicates whether or not we can assign an individual who completed the onboarding task to a pamphlet. This behavioral outcome reflects multiple stages of costly actions to join a political party, from talking to a party member during pamphlet distribution, providing a missed call after the fact, and completing their first party task (a detailed intake survey that provided AAP leaders with local knowledge).Footnote 24
Second, we analyze the diversity of recruits (their “type”), again using their phone survey responses. The new member “type” variable is a measure of whether the new member is a woman, a member of a politically and socially excluded caste group (a member of Scheduled Castes and Tribes or the Other Backward Classes), or a member of a politically excluded religious group (Muslim, Christian, Sarna, and Sikh). If the pamphlet is assigned to a new member who has any of these characteristics, we code the pamphlet as having recruited a member from an “excluded group.” We also disaggregate by gender, caste, and religious exclusion.
Third, we study the political experience of recruits (their “skills”). The “skills” variable is a summary measure of new members’ prior history of employment, education, and political participation history (voting at elections and volunteering for parties). If we can assign a pamphlet to a new member with any of these characteristics, we code the pamphlet as having recruited a “skilled” member. We also study each outcome independently.
Importantly, all of our outcomes are defined as binary indicators equaling one if a pamphlet can be assigned to an individual with a particular characteristic, and zero otherwise. We avoid the problem of posttreatment bias by not conditioning on having recruited a new member, a standard approach in the literature. For ease of interpretation, we multiply all outcomes by 1,000 so that they can be interpreted as the number of new members recruited per one thousand pamphlets distributed.
Descriptive Patterns in Status-Quo Recruitment
Table 2 presents the results of the recruitment drive with further details in Supplementary Figure A.3. First, we analyze the size of the recruit pool achieved using AAP’s typical strategy. Under status-quo recruitment, AAP received 7.5 new members who completed the onboarding survey task for every one thousand pamphlets distributed, which tracks rates for similar campaigns. Three years later, across two elections, 1.6 members per one thousand pamphlets distributed remain eager AAP workers. This figure represents a high bar on continued interest in the party given changing mobile phone numbers and that these individuals considered themselves active members who were still ready to volunteer.
Table 2. Summary Statistics on the Group Recruited by Baseline Pamphlet
Note: The table shows, for each variable, the mean value, standard deviation, and number of non-missing observations in the status-quo recruitment group along the dimensions of Size, Type, and Skill. Each variable has been multiplied by 1,000 so that each reflects the number of individuals recruited per one thousand pamphlets. Italics denote an index.
Second, looking at the types of recruits the baseline pamphlet attracted, only 3.5% of new recruits were women, while 68% of them were members of groups traditionally excluded from politics in India and 34.6% of them were members of politically marginalized religions. These recruits broadly reflect the demographics of the AAP Jharkhand mid-level members who recruited them in terms of Scheduled Caste, Christian, Hindu, and Sikh identities. However, Scheduled Tribes are underrepresented, while Muslims are overrepresented, relative to both the Jharkhand average and the mid-level members. Finally, more women were recruited than existed at the middle party tier.
Examining skills of recruits, we find that for every one thousand pamphlets distributed, about 7.4 “skilled” members join. In addition, 45.9% of AAP’s status-quo recruits are employed and 43.7% have at least a high school education. While the majority of these status-quo recruits had voted prior to joining AAP (67.1%), only 40.7% had volunteered at another political campaign. Finally, recruits and mid-level members are both more likely to have received education than the average Jharkhandi.
Our overall takeaway is that this status-quo strategy has mixed efficacy vis-à-vis recruiting individuals who display high competence and diversity. The key challenge for party leaders is to expand the pool of recruits joining via status-quo methods while not compromising—and even potentially improving—who ends up in the pool.
EXPERIMENTALLY SHAPING THE RANK AND FILE
AAP leaders embedded an experiment and multiple surveys into status-quo recruitment procedures. These allow us to make progress on the theoretical question of whether specific actions by party leaders affect recruitment.
Description of Treatment Pamphlet Design
AAP created 10 pamphlets cross-randomizing two sets of recruitment treatments. The first dimension of the treatment varies inclusiveness along gender lines within AAP’s ranks by randomizing whether a female or male party member’s face and story is featured on the pamphlet. We refer to this in shorthand as the “female” treatment in the analysis below. The second dimension varies perceptions of four benefits of joining the party. The “candidacy” message primes on the possibility of becoming a political candidate, the “career” message on advancing one’s career and skills through party work, the “ideology” on broad appeals to AAP’s “new kind of politics” that relates to anti-corruption, and the “policy” on Jharkhand-specific development priorities.Footnote 25 Combined with the status-quo pamphlet discussed in the previous section, there were five types of pamphlets under this second dimension of treatments. Cross-randomizing the two dimensions yields 10 treatment conditions that we show in English in Figure 4, and in the original text in Appendix A.1 of the Supplementary Material.
Figure 4. Translations of Treatment Pamphlets
Randomization
Table 3 presents the randomization schedule. Out of 60 assembly constituencies that AAP decided to mobilize in, 29 were assigned to the female treatment. The randomization of the second dimension of treatment, the benefits messaging treatment, occurred at the lower level of mid-level members (vice presidents). This procedure is described in Appendix A.5 of the Supplementary Material. Appendix B of the Supplementary Material discusses that some of the authors worked as unpaid consultants during the campaign, ensuring that randomization was consistent with academic best practices. Appendix E.1 of the Supplementary Material presents evidence of balance in pamphlet distribution across the treatment conditions.
Table 3. Number of Units in Each Treatment Condition
Note: We show in the 60 assembly constituencies with AAP operations the number of mid-level members (vice presidents) assigned to each treatment condition.
Estimation for Recruitment Experiment
We create a dataset where each distributed pamphlet is an observation. This allows us to study, for instance, the probability of recruiting new members by creating an outcome variable that equals one if a pamphlet yielded a member, and zero otherwise. We can also study characteristics of the new members, say gender, by defining a variable that equals one if the member who completes the onboarding survey task is a woman, and zero otherwise (including if the individual who received the pamphlet did not ever give a missed call or did not participate in the survey).
We estimate two types of regressions:
$$ \begin{array}{r}{Y}_{i,a}={\hat{\gamma}}_1{Female}_{i,a}+{\epsilon}_{i,a},\end{array} $$
$$ \begin{array}{l}{Y}_{i,v}={\alpha}_a+{\hat{\delta}}_1Ideolog{y}_{i,v}+{\hat{\delta}}_2Candidac{y}_{i,v}\\ {}\hskip2.80em + {\hat{\delta}}_3Polic{y}_{i,v}+{\hat{\delta}}_4Caree{r}_{i,v}+{\epsilon}_{i,v},\end{array} $$
where each observation is a pamphlet i belonging to vice president area v or constituency assembly a.
$ {Y}_i $
is a binary outcome for whether pamphlet i yielded a new member who completed the onboarding survey task.
$ {\gamma}_1 $
measures the aggregate impact of the female pamphlet treatment
$ Femal{e}_{i,a} $
randomized at the assembly constituency level.
$ {\delta}_j $
identifies the treatment effect of a
$ Pamphlet\hskip0.3em Typ{e}_{i,v} $
randomized at the vice president area level. The hats on treatment coefficients signify that we focus on local average treatment effects throughout the article where we instrument the actual treatment received by the treatment assigned, though very high compliance ensures that the two effects are similar.Footnote
26
$ {\alpha}_a $
are assembly constituency fixed effects that serve as blocks for the randomization of the four benefits.
We cluster standard errors according to treatment assignment: assembly constituency in specification (1) and mid-level member (vice president) for specification (2). The clustering strategy, however, is slightly more involved. While AAP put different phone numbers for each pamphlet type (i.e., candidacy-female, candidacy-male, career-female, and so on), it did not do so for each specific vice president (assembly) area within each treatment arm. This means that while we are able to match new members to the correct treatment arm (by linking them to the number they called), we are not able to place all new members in a specific vice-president (assembly) area v. Footnote 27 To resolve this, we assume that the probability that a particular call from a new member who was recruited by a particular vice president varies in proportion to the number of pamphlets distributed by vice president v (or in assembly constituency a), within treatment arms. We then bootstrap this assignment process to calculate the standard errors as explained in Appendix D of the Supplementary Material.Footnote 28
Pre-Analysis Plan
Prior to receiving the data from this experiment, we specified our plan for the analysis. We follow our pre-analysis plan in the construction of outcomes and indices and in the estimation of standard errors. Only at a few points did we depart from this analysis plan, which we summarize in Supplementary Tables F.5 and F.6. Our most important detour is to include constituency fixed effects in our analysis of campaign messages’ impact. We do so because the recruitment message treatment was block-randomized at this level. Given that our estimates are already quite precise, we do not need to include further controls.
EXPERIMENTAL RESULTS
Can Party Leaders Shape the Rank and File?
We first consider the primary question for this article: can party elites shape the rank and file? We set up a null hypothesis that any effort exerted by party elites toward shaping the party member pool will not yield a quantity or composition that differs from status-quo recruitment. We test this null hypothesis by comparing the joint significance of all treatments implemented by AAP against status-quo recruitment, such that,
$ {H}_0\hskip-2.77695pt :{\delta}_i={\delta}_j=0\hskip2.77695pt \forall i,j. $
The alternative hypothesis is that at least one of the treatments has an effect on the number and types of volunteers recruited.
Table 4 presents the results on the number of new members recruited, the number of new members who belong to excluded groups that were recruited, and the number of skilled new members recruited. We present results from the onboarding survey in columns 1–3, and the long-term retention survey in columns 4–6. In all cases, with large F-statistics (and
$ p<0.01 $
), we reject the null hypotheses that none of the treatments recruited a different number of rank and file than the status-quo strategy.Footnote
29
Table 4. Do Party Leader Efforts Affect the Rank and File?
Note: This table shows the mean value of the outcome (multiplied by 1,000) for each treatment arm. The bottom panel reports F-statistics and p-values for a test of the joint orthogonality of the treatments against Baseline (Male) condition. Standard errors are clustered at the vice president (mid-level member) level. Supplementary Tables G.7 – G.8 show joint orthogonality tests for the components of Excluded Group and Skilled Members.
Taken together, against concerns that political elites have little power in influencing who takes on the laborious task of party work, these results provide compelling evidence that elites’ interventions can directly influence the number and types of people who join parties and that their efforts can have impacts in the long term.
Does Varying the Portrayal of the Rank and File Affect Recruits?
Next, we move to the secondary research question of how the specific levers at the hands of party elites shape the composition of the rank and file. We compare each type of pamphlet against the baseline recruitment strategy.
Impacts on Recruitment at Onboarding
Size of the recruit pool
Beginning with the onboarding survey, we first present results for the size of the recruit pool, meaning the number of individuals who expressed initial interest by giving a missed call and who completed the first party member task of an intake survey used to aid AAP’s understanding of local electoral dynamics. We show in the top of the left panel of Figure 5 that highlighting AAP’s female party workers on recruitment pamphlets, instead of the status-quo men, increases the number of new members who completed the onboarding survey task before the state elections by 19% (
$ p<0.01 $
).
Figure 5. Impact of Different Treatment Conditions on Main Outcomes
Note: The figure represents coefficients with 95% confidence intervals for each treatment condition in the onboarding and long-term retention surveys. The coefficients can be interpreted as the change in the number of new recruits by type (e.g., all new rank-and-file recruits; a member of an excluded group; or a skilled recruit) generated for every 1,000 pamphlets distributed. Results for this figure at onboarding are reported in Supplementary Table G.9, Table 5, and Supplementary Table G.12, respectively, as well as Supplementary Table G.14 for the long term.
Second, we examine how portraying benefits from joining the party impacts the total number of new members. We find that the rate of new members at onboarding increases by about 18% (
$ p<0.10 $
) and 33% (
$ p<0.01 $
) when the party highlights ideology and candidacy, respectively, but that it decreases by about 26% when the pamphlet displays policy plans (
$ p<0.05 $
). Offering career benefits does not seem to move the needle on recruitment.
Diversity of the recruit pool
Next, we focus on whether different messages can shape the types of new recruits. The middle panel in Figure 5 shows effects on an overall index of recruiting politically excluded groups—women, excluded castes and tribes, and excluded religions. We find that when women’s stories are highlighted on the pamphlet, recruits from politically excluded groups increase by 22%. In Table 5, we decompose this index into its component parts. We observe that the effects are first driven by a 91% increase in women joining, though we note that the number of women joining with the status-quo male pamphlet was very low. In addition, we also find a 30% increase in sign-ups by people belonging to excluded castes and tribes.
Table 5. Unpacking Impacts on Excluded Groups
Note: The unit of observation is at the pamphlet level. In column 1, the dependent variable is a binary indicator of whether or not a missed call from a rank-and-file recruit who belongs to a politically excluded group has been matched to a pamphlet. Excluded group is coded 1 if the respondent is either a woman, belongs to an excluded caste or tribe (ST, SC, OBC), or to an excluded religion (non-Hindu). The other columns show results for each component. Robust standard errors reported in parentheses are clustered at the assembly level in Panel A and at the mid-level-member level in Panel B. Standard errors are obtained by bootstrapping potential assignment of a pamphlet to different mid-level members five thousand times. ***
$ p<0.01 $
; **
$ p<0.05 $
; *
$ p<0.1 $
.
Regarding the impact of messages highlighting different returns from joining, we find, as with the total number of new recruits who completed the onboarding survey task, the ideology and candidacy messages increase the rate of politically excluded new recruits by about 27% and 31%, respectively, and the policy treatment reduces enrollment from this group by 23% at onboarding. Interestingly, most of these effects emerge among excluded castes and religions, but we do not detect a statistically significant change in women recruits (see Table 5).
Skills of the recruit pool
Finally, we test the extent to which different campaign messages attract individuals with skills that could be important for organizing political campaigns. While status-quo recruits on average tend to be employed, most of them have less than a high school education and have no experience with political volunteering. Figure 5 shows results for the skilled recruitment index in the right panel. We see that signaling gender inclusiveness increased the number of onboarded skilled individuals by 20%. We present tabular results in Supplementary Table G.12. These politically experienced volunteers can be crucial for a party when it is building a new, nimble political organization.
Now, examining the impact of different benefits messaging, we find that skilled recruits at onboarding are persuaded when candidacy is salient. The salience of policy benefits proved to dissuade skilled recruits initially. We show positive effects of ideology and candidacy for the component parts of this skill index in Section G.4 of the Supplementary Material at onboarding.
Impacts on Long-Term Retention
In addition to recruitment in the current electoral cycle, our data also allow us to see how treatment effects persist. Figure 5 also presents effects from the long-term retention survey on these individuals who remain reachable by and willing to dedicate their time to the party three years later.
In general, we find that the effects are attenuated at the 3-year mark which is expected in these types of persuasion treatments. However, examining the two groups of treatments, we find striking evidence that the gender inclusiveness treatment continues to exhibit effects even at the 3-year mark—we observe a higher number of recruits in this condition relative to the male pamphlet. There are also more excluded and skilled recruits. When we decompose the excluded groups results, we note two further patterns. First, the long-term effects primarily arise among the excluded castes and tribes category, where the latter is the primary marginalized group in our setting, suggesting that signaling inclusiveness can be particularly effective in recruiting individuals from marginalized backgrounds. However, second, we note that the initial positive results on recruiting women are completely attenuated (see Supplementary Table G.15). This second result is particularly important. On the one hand, our results demonstrate that parties are indeed able to recruit more women, but on the other hand, they suggest that party leadership might need to continuously put effort into retaining the women they have recruited to prevent a leaky pipeline (Thomsen and King Reference Thomsen and King2020).
In addition to diversifying the pool of volunteers, the gender-inclusive pamphlet also recruited individuals who were 50% more likely to stay involved in the party 3 years later. These politically experienced volunteers are important for a party seeking to both recruit and retain skilled workers across multiple electoral cycles.
Joint Impacts of Signaling Inclusiveness and Highlighting Benefits
Next, we examine if recruitment messages across the two dimensions (inclusiveness and benefits) interact to produce effects. Figure 6 shows the average number of new recruits per one thousand pamphlets distributed for each treatment condition.Footnote 30 We find two key results.
Figure 6. The Impact of Decomposed Treatments on Number of Recruits (Per 1,000 Pamphlets)
Note: The figure shows the number of new recruits per one thousand pamphlets distributed for each treatment condition. “M” and “F” on the labels indicate whether the pamphlet shows male or female photos, the gender-inclusive treatment dimension. All coefficients report local average treatment effects with 95% confidence intervals. Coefficients for testing the effectiveness of all female treatments against all baseline male treatments are reported in Supplementary Table G.9. All other results are reported in Supplementary Table G.10.
First, with a baseline message, emphasizing women’s stories on the pamphlets yields a higher number of recruits than the status-quo male pamphlets (
$ p=0.041) $
. This effect also carries over to all male versus female pamphlets (
$ p=0.002 $
). Averaging across all messages emphasizing benefits, female pamphlets recruit more rank and file than male pamphlets, suggesting that a backlash channel does not dominate.
Second, examining the interaction of the two dimensions specifically, we find that, within male pamphlets, emphasizing benefits versus the status-quo baseline message changes the rate of callbacks, but there is little evidence of a similar effect within female pamphlets—the joint orthogonality F-statistic is large and significant at the 5% level for the first but not for the second test.Footnote 31 This suggests that once a woman’s story is highlighted, the marginal returns to emphasizing further benefits are limited, which is consistent with the salience story we detailed in the section “How the Two Dimensions Interact.”
Discussion
Overall, three high-level takeaways from AAP’s recruitment experiment challenge conventional beliefs. First, against concerns that the high opportunity costs will completely inhibit any participation, nonmonetary incentives convinced more and higher-quality recruits to join the party. Second, against concerns that these forms of persuasion have limited effects on behavior, particularly in the long term, these effects persisted into a subsequent election cycle. This indicates the party leaders’ ability both to assemble an immediate workforce and also guide the long-term development of the party. Third, our results demonstrate that messages by themselves, rather than the identity of the worker who handed them out, can be powerful in increasing the diversity of the rank and file, suggesting a scalable strategy for parties wishing to broaden their existing appeal.Footnote 32
A key concern from the literature, as reflected in Table 1, is that broadening the political pipeline will diminish the competence of the rank and file, a concern shared by leaders of AAP. Yet, as we discover via the experiment, the messages that are most effective at reaching politically marginalized groups also mobilize high-skilled workers, suggesting that the perceived conflict between skill and diversity may be overinflated. In fact, we further decompose our results on included groups and low-skilled workers and do not find any consistent effects on those populations, further suggesting that treatments are most effective on the specific populations we have identified (See Supplementary Tables G.11 and G.13).
Beyond these overall patterns, one of our most important and robust findings is on signaling gender inclusiveness. This treatment first increased total recruitment over the long run, which as we noted, in the section “Highlighting Benefits from Joining the Party,” was not ex ante obvious, because status-quo male recruits could have chosen to exit. In fact, if we look at general caste men, we do not find evidence that this treatment impacted their behavior (see Supplementary Table G.11). Second, while the positive effects on the total number of recruits and those from excluded castes and tribes are apparent 3 years later, AAP struggled to retain female recruits, a problem faced by many parties (Gulzar Reference Gulzar2021; Thomsen and King Reference Thomsen and King2020). This is a bittersweet result. It suggests that signaling inclusiveness has real potential at resolving initial barriers to entry, yet may not solve pipeline issues completely. Finally, the positive impacts on skills also allay concerns related to the recruitment of poor-quality workers.
The findings on highlighting the benefits of joining the party were more short-term in nature, but nevertheless carry important lessons. First, highlighting that future recruits could win the party’s nomination for candidacy proved to be highly attractive. This result is perhaps the least surprising given the literature discussed in our theory section—receiving the party’s nomination is a big prize. Because AAP did not have a realistic chance of winning many seats, an interpretation of this result is that the prospect of a party nomination carries status rewards beyond the specific chances of winning elected office, as we discussed.
Highlighting the policy agenda of the party tells the opposite story of the candidacy benefit. Emphasizing policies reduced the number of recruits. How might we interpret this negative finding? Beyond the simplest explanation that the policies chosen were not the preferred policies of the prospective recruits, another potential interpretation is that recruits saw AAP’s policy promises as lacking credibility, given that the party would at best only be part of the winning coalition with minimal policy influence. The results, therefore, suggest that parties aiming to recruit on programmatic grounds could face a bind if the ambition of their policy goals is out of sync with their current capabilities. This is in line with the idea that politicians prefer to make vague instead of specific promises (Shepsle Reference Shepsle1972).
Alternative Explanation: Party Effort
Besides appealing differently to potential recruits, the treatment effects we have identified could additionally emerge through differential efforts by party personnel in charge of running the campaign, that is, state/assembly-level and mid-level party workers. We test for this through several pieces of evidence, which appear in Appendix E of the Supplementary Material.
We begin by examining the efforts that senior party leaders put into the recruitment campaign and report two results. First, we observe that all pamphlets that were assigned to be distributed were in fact distributed and that the number of pamphlets distributed is balanced across treatment arms (see Supplementary Table E.2). Second, we examine how senior party leaders assigned work to subordinate mid-level members. We find that the number of locations assigned to each mid-level member to canvass for recruitment does not differ by treatment arm (see Supplementary Figure E.5). These results suggest that senior party personnel did not direct the campaign in a way that put more attention toward specific treatment areas, thereby producing the treatment effects we report.
Next, we examine the efforts of mid-level party members (vice presidents) and report three results. First, we already discussed above that compliance—that the pamphlet given out to vice presidents is the one assigned via randomization—is very high (see Supplementary Table E.3). Second, we observe that these compliance rates do not differ by the treatment assigned (see also Supplementary Table E.3).Footnote 33 Third, we find that noncompliance among local party leaders does not seem to follow a pattern that indicates that they prefer specific pamphlet types (see Supplementary Table E.4). Taken together, these pieces of evidence show that treatment effects were unlikely to emerge because of differential efforts by those who were directly recruiting.
Another remaining explanation could pertain to the actual in-person interaction between mid-level members and prospective recruits. It could be the case that, when the primarily male local cadre is assigned the female pamphlet to distribute, they distribute fewer pamphlets. With some data, we make a case that there are minimal chances of this happening. First, we previously showed that all pamphlets were distributed and that this number is balanced across treatments. Second, the party conducted periodic check-ins with canvassers to assess their progress. Information on the dates of these check-ins and how many pamphlets were distributed at those points allows us to trace the amount of work done by mid-level members across the campaign period. We show in Supplementary Figures E.6 and E.7 that the distribution effort is balanced across treatment arms.
Finally, what remains is an explanation where personnel distribute the same number of pamphlets but exert differential effort in discussing the gender-inclusive pamphlet with prospective recruits or target different groups with different types of pamphlets. Given the scale of the experiment, this is an element that is not directly quantitatively observed by us, so we cannot rule out this completely. In Appendix E of the Supplementary Material, we take a simulation approach to discuss the conditions under the heterogeneous treatment effects would be completely explained by such a story.
CONCLUSION
The first step in many ordinary citizens’ engagement with political organizations is as campaign staff. If elites can shape these rank-and-file members, then parties can be more agile and responsive to changing electoral preferences both through the selection channel—where the identity of party members themselves changes over time—and the preferences channel—where party members remain the same, but their preferences change over time. Demonstrating the possibility of elite intervention is, therefore, an important complement to research stressing historical processes as the key source of variation in who comprises party members.
We discuss how it is expected that, lacking financial incentives, party leaders must throw in the towel and recruit few and perhaps extreme volunteers. Consequently, party elites should have limited control over recruiting the rank and file for ground campaigns, potentially carrying important consequences for electoral outcomes and party development. But, as parties increasingly need to employ sophisticated ground campaigns in modern democracies, what tools can elites deploy to move this equilibrium? Will their efforts bear fruit? To date, the evidence on these questions has remained scant perhaps because of the difficulty of collaborating with parties to study these questions.
In this article, we have assembled evidence from a real political recruitment drive implemented at a large scale to study these questions. We collaborated with AAP in India at a key moment when it sought to build up its party infrastructure, in competition with other parties doing the same. An at-scale recruitment drive, with an embedded experiment, distributed promotional pamphlets to 1% of Jharkhand’s 13-million registered voters. We analyze this unique data source and find that party elites are not necessarily constrained by whoever puts themselves forward for the party; pulling different messaging levers on these recruitment pamphlets allows them to change both the size and the composition of the sign-up pool in statistically and substantively meaningful ways. The long-term evidence suggests that these efforts can matter for party development.
We also find that certain messages work better than others for yielding more recruits, appealing to diverse individuals, and overcoming the high opportunity costs of skilled volunteers. Specifically, our results suggest that, at least in the case of South Asia and AAP, signaling a willingness to include women is particularly powerful in recruiting diverse and capable individuals. This is the only treatment that also has sustained long-term effects. In the shorter run, promising candidacy or appealing to a broad party ideology was more effective than mentioning specific policies or highlighting a career with the party. Given the rise of large on-the-ground campaigns around the world and an increasing push to recruit previously excluded groups into politics, we present fresh evidence on how party elites can use existing levers to incorporate these new voices and helping hands.
SUPPLEMENTARY MATERIAL
To view supplementary material for this article, please visit https://doi.org/10.1017/S0003055424000649.
DATA AVAILABILITY STATEMENT
Research documentation and data that support the findings of this study are openly available at the American Political Science Review Dataverse: https://doi.org/10.7910/DVN/UW9NBY.
Acknowledgements
We thank Adam Auerbach, Jennifer Bussell, Gary Cox, Rafaela Dancygier, Gemma Dipoppa, Tanushree Goyal, Anna Grzymala-Busse, and Beatriz Magaloni for their detailed comments on prior versions of this article. This research would not have been possible without the support of AAP staff in Jharkhand and Delhi, including Rajan Kumar Singh and Ankush Narang. Surbhi Ghai, Emily Katz, Jasneet Kaur, Anzony Quispe, and Ravichandra Tadigadapa provided excellent research assistance. Finally, we thank participants at APSA, NYU-CESS, PacDev, the Clayman Institute, the Frontiers of Clientelism Conference at Harvard, Indian Democracy at Work, the Harvard Experimental Political Science Conference, the Gender and Politics in South Asia workshop at Berkeley, and the Institutional and Organizational Economics Academy for their suggestions.
CONFLICT OF INTEREST
The authors declare no ethical issues or conflicts of interest in this research.
ETHICAL STANDARDS
The authors declare the human subjects research in this article was reviewed and approved by Stanford University under Protocol No. 51207. The authors affirm that this article adheres to the principles concerning research with human participants laid out in APSA’s Principles and Guidance on Human Subject Research (2020) (see Appendix B of the Supplementary Material for further discussion).
Open access
Comments
No Comments have been published for this article.