Lone organizers: Opposition party-building in hostile places in Tanzania

Abstract

I ask where African opposition parties organize. Party-building is communicative; it involves persuading people to become activists. The literature suggests that opposition parties organize where people are receptive to their messages and build outwards from there. I study Chadema’s opposition party-building through site-intensive fieldwork. Chadema organized primarily in such receptive areas, but also in four unreceptive constituencies. I use these deviant constituencies to refine the literature. Prior theory neglects the heterogeneity of party-building. I decompose party-building into three modes as follows: by touring leaders, branches and concentrating leaders. Concentrating leaders dedicate their organizing to single places. They employ small rallies which afford interactive, individualized and iterative communication. This personalized communication enables them to overcome initial unreceptiveness to their messages. I conclude that opposition parties can organize in unreceptive areas, but only through the personalized methods of these ‘lone organizers’. Altogether, I show how and through whom opposition parties organize in hostile environments.

African opposition party-building follows a common geographic vector. Organizing involves persuading citizens to found branches and run them. The literature suggests that citizens will be receptive to such persuasion insofar as they already support the party, agree with it and trust its messengers (Iyengar and Simon, 2000; Zaller, 1992). However, it also indicates that citizens who are thus receptive to opposition messages are concentrated in areas which are urban, populated by parties’ constructed co-ethnics or home to pre-existing party branches (Horowitz, 2016; Harding and Michelitch, 2019; Letsa, 2019). Therefore, opposition party organizing begins in such receptive areas and grows outwards from them like ink blots spilled on paper. This creates a quandary for opposition parties. Local branches help parties to shape opinion and become popular (Letsa, 2019; Paget, 2019a), but opposition parties can most feasibly establish them where they already encounter most like-minded opinion and popularity. Therefore, in one respect, opposition parties can best organize where they least need to.

Tanzania’s leading opposition party is Chama cha Demokrasia na Maendeleo (Chadema or ‘The Party of Democracy and Development’). Between 2007 and 2015, it organized at an incredible scale (Paget, 2019a). Largely, it followed the spatial organizational trajectory described above. However, it also established local branches in three constituencies in and one near what it calls ‘the Central Zone’, a ruling party stronghold beyond Chadema’s organizational frontier (Collord, 2015). This was among the most hostile environments in which Chadema could have organized by location, demography and electoral history. These are instances of opposition organizing which defied Chadema’s national strategy and the constraints on opposition organizing indicated by the literature. I treat these clusters of Chadema party-building as deviant cases (two cases, two shadow cases).

I argue that Chadema organized in those deviant constituencies by employing a particular mode of organization. I define party-building – or synonymously, party-organizing – as the establishment of party organs such as branches in places in which they were previously absent, and the concomitant recruitment of members as activists for those organs. I critique the literature and suggest that it under-theorizes party-building. It neither specifies who builds the party, nor how they do so. One notable exception is that the literature singles-out MPs as inimical to opposition party-building. It suggests that they develop patron–client relations with activists which undermines party cohesion (Cooper, 2014; Rakner and Van de Walle, 2009).

I distinguish between three modes of party-building by the actors involved and the practices or methods they employ: organizing by (1) touring leaders, (2) branches and (3) concentrating leaders or ‘lone organizers’ who focus on one area or constituency. I argue that each of these modes enables party-building in unreceptive or ‘hostile’ areas to different degrees. The enabling power of each mode turns on two features: the personalization of party-building practices and mobility of the organizers. First, opposition organizing in hostile areas is more effective when it involves personalized methods of communication. These interactive, individualized and iterative methods help opposition parties to recruit activists in hostile environments more than impersonalized methods do. Second, organizers can operate in hostile areas far from party’s branches only insofar as they are mobile.

Each mode of organizing is associated with different combinations of these characteristics. Touring leaders are mobile, but they use impersonalized methods. Branches use personalized methods, but they are immobile. This makes organizing by concentrating leaders is unique. This mode alone combines the mobility and personalized methods which enable opposition organizing in hostile areas. I trace the modes of party-building that Chadema employed in Tanzania. I find that it organized in receptive areas through touring leaders and branches, but that it organized in the Central Zone through the personalized party-building of a small group of lone organizers. Therefore, I conclude that opposition parties can organize in hostile areas, but only through the concentrating leaders. A corollary of this argument is that lone organizers like MPs, far from undermining party-building, can be vehicles for it.

Thereby, I use this deviation of Chadema from the literature – in organizing in the Central Zone – to refine the literature on party-building in sub-Saharan Africa. Asking ‘where did Chadema organize?’ leads me to the further question: ‘how and through whom did Chadema organize here?’ This serves as a lens onto the equivalent wider question: ‘how and through whom do opposition parties organize in sub-Saharan Africa?’ Altogether, I do not make claims about the causal determinants of opposition organizing. Instead, I contribute to the literature by make claims about the practices and actors through which they organize.

My study of these four deviant cases is nested within my study of Chadema as an extreme case of opposition organizing. The scale of Chadema’s organizing made its party-building practices particularly pronounced, and this aided my concept-development and data-generation. I used my research to develop theory (George and Bennett, 2005). I did not fix data collection procedures that would standardize the operationalization of variables in advance. Instead, I construct variables qualitatively in the text (George and Bennett, 2005: 28). This was a deliberate choice to preserve the exploratory character of my research. This enabled me to develop a grounded theory (Charmaz, 2006). As I do not make inferences about causal determinants of party organizing, I do not employ a process tracing methodology as defined in the literature. However, I do nonetheless trace processes, in particular organizational practices, and I generate substantive empirical evidence to demonstrate that those practices were at work.

To generate that theory and evidence, I conducted eight months of field research between January and December 2015. In this article, I draw primarily on four forms of field research. First, I interviewed 11 past and present Chadema Central Committee members and six further high-level officials. Second, I interviewed 14 Chadema parliamentary candidates (of which, five Central Committee members), including four who ran in the Central Zone. Third, I visited 35 Chadema organs at the district, ward, branch and foundation levels, primarily in eight constituencies spread evenly across four regions: Mwanza, Singida, Dodoma and Mbeya regions. Fourth, I was a latent participant observer of the 2015 election campaign. There, I witnessed some of the party-building methods which had been employed in the preceding eight years. I attended 42 rallies, including 34 parliamentary candidate rallies; and I accompanied five canvassing trips. This diversity of evidence is crucial for both the qualitative construction of variables and the generation of evidence about organizing practices. First, my multiple sources of evidence enabled me to fill gaps left by any individual sources. Second, this variety of sources enabled me to verify the claims of individuals.

Tanzania is ruled by an electoral-authoritarian regime (Paget, 2017, 2021). I anonymize all interviewees save those whose political views are publicly known to protect them from the state. Quotations are expressed in English; all quotes originally given in Swahili were translated professionally. I was aided by research assistants who wish to remain anonymous.

Where opposition parties organize

Many opposition parties in sub-Saharan Africa are organizationally weak (Rakner and Van de Walle, 2009). In particular, they lack local presence (Krönke et al., 2020; Randall and Svåsand, 2002). This places them in stark contrast to ruling parties which, with the benefit of incumbent largess and state infrastructure, often achieve widespread organizational presence (Randall and Svåsand, 2002). That opposition organizational weakness varies sub-nationally; most opposition parties have local presence in some places, but not in many others (Horowitz, 2016; Letsa, 2019; McLellan, 2020). Nonetheless, most studies analyse opposition organizing at the national-level. ¹ This is a significant limitation of the literature, which consequently only identifies national-level determinants of opposition party-building (LeBas, 2011; Riedl, 2014; Paget, 2019a). Consequently, spatial patterns of opposition organizing have become the subject of recent research (Bob-Milliar, 2019; McLellan, 2020; Paller, 2019). I theorize where parties organize by building logically on the implications of adjacent literatures.

Party-building involves recruiting activists, either party members-as-activists or extra-party activists. Therefore, party-building is a communicative practice; more precisely, it is a persuasive one. I do not use the term persuasion’ and ‘mobilization’ in the sense fixed in the political science literature of ‘changing policy preferences’ and ‘increasing turnout’. Following the political communication literature, I recognize that parties change opinions and behaviour by informing, priming and framing too (Iyengar and Simon, 2000; Scheufele and Tewksbury, 2007). I draw on this broader conception of ‘persuasion’. Below, I use ‘persuasion’ as a shorthand for ‘convincing people to become party activists through communication’.

The most common distinction drawn in African electoral political geography is between party ‘swing’ and ‘core’ areas (Horowitz, 2016). This distinction is pertinent in the study of campaign strategy because it focuses on the current levels of party support relative to rival parties. Therefore, it bears on the probability that a marginal increase in campaign effort will tip party support into a plurality. However, it is not the most analytically incisive distinction if studying party organizing, when the most pertinent question is the effect of marginal organizational effort, irrespective of levels of prior organization relative to rival party organization. Therefore, I develop an alternate distinction between ‘receptive’ and ‘hostile’ areas below.

The comparative literature shows that the effects of party–citizen political communication are mediated by several factors, chiefly the citizen’s prior support for the party, congruence between the party’s message and the citizens’ prior beliefs and the citizen’s trust in the medium (Iyengar and Simon, 2000; Zaller, 1992). In other words, people are more receptive to messages insofar as they already support the party, they agree with its messages and they trust the messenger. Insofar as an area is populated by people that meet any and each of these three criteria, it will be receptive to party messages and be a ‘receptive area’. Insofar as it is not thus populated it will be a ‘hostile area’.

The Africanist literature finds that ruling party support is national (Wahman, 2015); it is widely dispersed and present in all areas. Therefore, even when incumbents lack prior organizational presence, citizens will be receptive to their messages and so they will be able to organize across space. By contrast, the same literature suggests that people who are receptive to opposition messages are geographically concentrated. Specifically, it suggests that citizens will be receptive to opposition messages in three non-exclusive types of location: co-ethnic areas, urban areas and areas where the party is already organizationally present.

First, ethnicity is a significant determinant of party support in sub-Saharan Africa (Bratton et al., 2012; Cheeseman and Ford, 2007), albeit contested (Harding and Michelitch, 2019). Opposition party support is especially mono-ethnic (Cheeseman and Ford, 2007) and concentrated geographically (Wahman, 2015). Therefore, people who share an ethnic identity with which an opposition party is associated (co-ethnics) will be more receptive to its messages, other things being equal.

Second, research shows that opposition parties and urban residents share policy positions. Opposition parties choose policies close to those of urban residents (Resnick, 2013). Equally, urban residents hold opinions about liberal democracy (Letsa, 2019) which incline them to opposition parties (Bleck and Van de Walle, 2013), especially the young and the male (Macdonald, 2018). Research, which is contested (Nathan, 2019), suggests that urban residents’ support for political parties is guided more by these programmatic considerations. Urban residents are, variously, more critical, more exposed to media, better educated, less guided by ethnicity, less partisan and free to choose between competing clientelist patrons (Harding and Michelitch, 2019; Macdonald, 2017; Resnick, 2013). Finally, urban residents are more likely to support opposition parties (Bratton et al., 2012; Resnick, 2013). Altogether, urban residents will be more receptive to its messages, other things being equal.

Third, people located near opposition party organs will be exposed to opposition messages and by degrees adopt partisan opinions (Letsa, 2019). They will also become familiar with opposition activists. Therefore, they too will be more receptive to opposition messages. A corollary of these claims is that opposition parties will concentrate their party-building in these receptive areas, where they will be most effective. Therefore, they will organize outwards from these areas, expanding their party territory by advancing an organizational frontier. While this theory is untested, this spatial pattern of opposition organizational presence is documented or otherwise treated as an empirical fact in several studies (Horowitz, 2016; Macdonald, 2017; McLellan, 2020; Paget, 2019a; Wahman, 2015).

How and through whom opposition parties organize

This forms a starting-point for my theory-development. However, this starting point does not differentiate between who organizes or how they do so. Neither does the extant literature. It theorizes organizing practices through metaphors of ‘investment’ and ‘construction’ enacted simply by ‘parties’. ² This is a shortcoming in the theorization of party-building which oversimplifies causal claims. A partial exception to this pattern is the study of MPs. The literature claims that MPs often frustrate party-building. They foster relationships with citizens based on clientelism or charisma which are independent of their membership of parties and erode opposition strength (Cooper, 2014; Rakner and Van de Walle, 2009; Randall and Svåsand, 2002). This suggests that MPs undermine opposition organizing.

I refine the literature by disaggregating party-building into three modes. Each of these modes corresponds to a set of actor-roles and practices. Each of these modes is ‘grounded’ concepts developed from the insights of my field research (Charmaz, 2006).

Organization by touring leaders involves a schedule of rallies. Speakers at these rallies draw people to attend and recruit them as members. A follow-up team contacts the new members, selects a provisional committee from among them and renders it a party organ. This involves producing rallies without the help of local branches who normally complete the accompanying labor-intensive tasks (Paget, 2020). Therefore, this mode of party-building is typically conducted by senior politicians with the fame, persuasive ability or resources to make rallies work in the absence of branches. It is relatively labor-light to execute in one place, and so can be conducted by a small group of such politicians across localities through a national tour.

Organization by branches involves street-level action. Campaigners convene small rallies and indoor meetings, canvass and mount posters. Thereby, they recruit members. They assign these members to an existing local organ, or they select a provisional committee from among them and render that committee a party organ. This mode of party-building is labor-intensive. That heavy labor demand is met by networks or party-aligned activists who are normally locally-based. They often take the form of party members in pre-existing nearby local organs such as branches. Equally, they can take the form of extra-party associations and assemblages (Nielsen, 2012).

Organization by concentrating leaders involves street-level action too. Lone organizers convene small rallies and indoor meetings. They recruit members through this action, form a provisional committee and render it a party organ. This mode of party-building is labor-intensive. Lone organizers meet that heavy labor demand through the commitment of large amounts of personal time, making organizing a full-time endeavour over months or years. Equally, they lighten this load through their superior persuasive ability or the expenditure of resources. Typically, this mode is carried-out by senior politicians who concentrate on building the party in one place. Often, this politician is an aspirant for elected office, such as a prospective parliamentary or councillor candidate.

Therefore, party-building is heterogeneous; it takes many forms. This variety in party-building has ramifications for where parties can build, which turn on two features of each mode of organization: the mobility of the organizers and the personalization of communication methods. Mobile actors, thus defined, are able to organize far from the party’s pre-existing structures. Immobile party actors, by definition, are not thus able.

I defined party organizing above as the establishment of branches and the recruitment of activists. I asserted that it is achieved through communication, whereby people are persuaded to become party activists. To define communication as personalized and impersonalized, I draw on the three features of communication enumerated in the seminal work of Lazarsfeld et al. (1948): (1) interactivity, (2) individualization and (3) iteration. Here defined, communication is interactive when it is conversational, when participants take turns, change roles and respond to each others’ messages (Isotalus, 1998). It is individualized when the content of the message is tailored by the speaker to individual listeners rather than a large audience (Lazarsfeld et al., 1948). It is iterative when there are successive episodes of communication between the same persons.

Lazarfeld, Berelson and Gaudet generate these features to distinguish between face-to-face and mediated communication, which they argue do and do not bear these characteristics, respectively. Recent work distinguishes between modes of face-to-face communication and argues that canvassing bears these characteristics while large rallies do not (Paget, 2019b). Following this work, I define organizing as personalized insofar as it involves (1) interactive, (2) individually-tailored and (3) iterative communication (Nielsen, 2012). Canvassing and rallies afford such personalized organizing insofar as they are small. I define organizing as impersonalized insofar as it involves (1) one-way, (2) uniform and (3) single-shot communication. Rallies afford only impersonalized organizing insofar as they are large. Following the literature, I stipulate that the effects of impersonalized organizing are mediated by peoples’ receptiveness, as defined above, more than the effects of than personalized organized (Lazarsfeld et al., 1948; Nielsen, 2012). Therefore, personalized organizing better enables opposition parties to organize beyond receptive areas.

In political science, especially Africanist political science, ‘personalized’ has been used as a synonym for ‘patrimonial’ or ‘neo-patrimonial’. This refers to a relationship between a politician and a citizen as the ties between a patron and a client based on either charisma or material exchange. Personalized communication as I define it is distinct from these meanings of ‘personalized’ because it necessarily involves persuasion rather than material exchange, and because it involves mobilization of activists for a party, rather than an individual.

These organizational modes are not operationally incompatible; each can be employed alongside the others as well as independent of the others. Nonetheless, of the three modes of party-building described above, only one enables opposition organizing in hostile areas. Touring leaders are mobile. However, they use impersonalized methods, and this blunts their ability to organize in hostile areas, especially because people who are unreceptive to a party’s message are less likely to attend their rallies (Paget, 2019b). Party branches use personalized methods but lack the mobility to operate far from their locations in receptive areas. By contrast, organization by concentrating leaders combines the mobility of leaders with the personalized methods of branches. Therefore, they enable party-building in hostile areas which are remote from the party’s organizational territory. Far from frustrating party-building, actors like MPs can be crucial to it. Altogether, I contend that opposition parties can organize in hostile areas, if lone organizers dedicate extensive time to party-building using personalized methods.

In the following sections of this paper, I substantiate these categories and this argument through a study of Chadema’s organizing. In next section, I enrich these concepts. I present Chadema’s party-building in urban and co-ethnic areas receptive to its message. I show that it organized in these areas primarily through touring leaders and branches. In the subsequent section, I trace how concentrating leaders built the party. From this evidence, I conclude that Chadema organized in four isolated constituencies in or near the Central Zone primarily through lone organizers using personalized methods of communication.

Operation Sangara

In 2006, Chama cha Mapinduzi’s (CCM, ‘The Party of the Revolution’) organization was unparalleled. It had, and has, a nation-spanning hierarchy of branches (Morse, 2014) and remained an exemplar dominant party (Makulilo, 2012; Randall and Svåsand, 2002). By contrast, Chadema was poorly organized (Mmuyu and Chaligha, 1994). By 2015, the organizational landscape was transformed. Chadema claimed that it had acquired a structure of 51,947 foundations and 16,359 branches (Slaa, 2014); these figures were indicatively supported by post-election survey data, which showed that Chadema canvassed as many or more people than CCM (Paget, 2019a). They have continued this party-building since then (Kwayu, 2019). In Chadema’s organizational expansion from 2007 to 2015, three modes of party-building can be traced. In this section, I introduce two of them.

Chadema’s leadership raised money to fund a long period of party-building (Paget, 2019a). In 2007, Chadema initiated a party-building program called Operation Sangara (Babeiya, 2011). It dispatched a motorcade upon a regional tour, containing the party’s best-known leaders. This convoy convened successive rallies intended ‘to introduce the party to the public…’ (Kigaila, 2015b). However, while imparting their messages to citizens, Chadema also recruited members. Then-Youth Wing Secretary-General Deogratias Munishi described that at rallies ‘You recruit members. You organize them to elect their leaders’ (Munishi, 2015). Benson Kigaila, Head of Chadema’s Directorate for Organization, Training and Zonal Administration, recalled that where organizational initiative was lacking, the convoy would provide further direction ad hoc (Kigaila, 2015b). While these practices were organizational, they were nonetheless impersonalized. They involved large rallies which were often stand-alone events with little follow-up.

This party-building focused on specific areas. Chadema takes ten self-defined zones, ³ rather than Tanzania’s 31 administrative regions ⁴ as its primary geographic referents (Chadema, 2020; Kigaila, 2015; Lissu, 2015; Munishi, 2015). Figure 1 displays a labelled map of Tanzania’s regions, colour-coded by the zones into which Chadema sorts them. I omit labels and coding for the semi-autonomous archipelago of Zanzibar for simplicity of presentation. In 2007, Chadema leaders said that they were best organized in two such zones. The first was the Coastal Zone, which contains Dar es Salaam where its national headquarters were based (Baregu, 2015; Kimesera, 2015; Munishi, 2015). The second was the Northern Zone, especially Kilimanjaro Region and parts of neighboring Arusha Region (Kimesera, 2015a; Mrema, 2016). Chadema disavows ethnic politics, but many of its leaders in this period hailed from Kilimanjaro and Arusha Regions and nearby; both Chagga ethnicity and a partially-constructed regional ‘Northerner’ identity have been sometimes associated with Chadema.¹² Therefore, Chadema leaders said that they were best organized and most popular in areas which might be analysed as the primary urban centre and a rural co-ethnic area nested within a wider co-regional area.

Figure 1.

Map of mainland Tanzania ⁵ by region, ⁶ colour-coded by Chadema’s zones (Chadema, 2020) ⁷ .

In Operation Sangara, Chadema focused initially on Mwanza Region in Victoria Zone and the surrounding Serengeti Zone (Babeiya, 2011). Elite interviewees attest to this (Kabwe, 2018; Kigaila, 2015) and I verified their claims in my ground research. Chadema members and low-level officeholders in Mwanza reported joining Chadema in 2010 (Anonymous, 2015o; Fieldnotes, 2015d), some specifically after attending Operation Sangara rallies (Anonymous, 2015v). Subsequently, Chadema turned principally to what it calls the Nyasa Zone, especially Mbeya Region, and the Northern Zone (Anonymous, 2015; Kigaila, 2015; Kimesera, 2015; Munishi, 2015). Chadema members in Mbeya Region reported joining or first running for party positions during Operation Sangara (Anonymous, 2015; Anonymous, 2015h; Anonymous, 2015u). After Dar es Salaam, the largest cities in Tanzania are Mwanza, Arusha and Mbeya, located in the Victoria Zone, Northern Zone and Nyasa Zone, respectively (National Bureau of Statistics, 2012). ⁸ Therefore, in accordance with the literature, Chadema organized outwards from urban centres (Babeiya, 2011), in a co-regional area, and outwards from a co-ethnic area. Robert Macdonald’s ethnographic study describes Chadema reach emanating from urban areas outwards (Macdonald, 2017).

Subsequently, Chadema refined their party-building practices. This involved operations closer to the ground, in which they concentrated further on instituting party organs. Kigaila described the revised method:

We had a team doing the preparation… They [politicians] do a rally… you have the training session for members. After the training, you have the election for the leaders… [then] the leaders will be trained (Kigaila, 2015b).

These sequential preparations, rallies, trainings and elections made the organizing more personalized, as defined above; it was iterative and involved increasingly smaller groups where interaction and individualization became possible. Chadema called this program Movement for Change (M4C) or Chadema ni Msingi (‘Chadema is the Foundation’) interchangeably.

In M4C, Chadema returned to and built on its organizational gains in Victoria, Serengeti, Northern and Nyasa Zones (Anonymous, 2015a, Anonymous, 2015b, Fieldnotes, 2015; Kigaila, 2015; Mbilinyi, 2015; Munishi, 2015; Shaba, 2015; Silinde, 2015). Consistent with the theory described above, it also employed these more personalized party-building methods to organize in less receptive areas. It concentrated on four new regions (Munishi, 2015): Mtwara, Lindi, Morogoro and Iringa. Chadema had little prior organizational presence in these regions beyond Iringa Town (Kigaila, 2015b; Kilimwiko, 2015). Moreover, they lacked the urbanity and proximity to ethnic strongholds that previous target areas had enjoyed. Chadema hoped to capture discontent related to perceived exploitation of natural gas in Mtwara and cashew-nut farmers in Lindi (Kilimwiko, 2015). Therefore, these areas may have been somewhat receptive to Chadema’s organizational message. Nonetheless, Chadema’s leaders organized there partly because the refinement of their touring leader party-building described above made it more personalized.

Alongside this organizing by party leaders, Chadema branches organized. Some did this of their own initiative between 2007 and 2010. However, branch party-building became more frequent and routine between 2011 and 2015 as a second prong of M4C. Chadema dispensed instructions to every branch through a train-the-trainers program which is described further elsewhere (Paget, 2019a). The trainers gave branches the following instructions: ‘get us new membership’ (Slaa, 2015). Then-parliamentary candidate Susan Kiwanga described that how branch organizing worked. Ward-level staff members were trained to visit ‘villages or streets and gather people, gather their neighbors. Register them…People buy cards’. (Kiwanga, 2015).

I documented this organizing by branches. Consistent with Kiwanga’s description, they recruited members by canvassing (Anonymous, 2015b, 2015i), convening public meetings, convening private indoor-meetings (Anonymous, 2015b, 2015d), mounting posters (Fieldnotes, 2015b) and addressing street-corner conversation groups called vijiwe (Anonymous, 2015r, 2015v). I witnessed that these practices involved personalized communication: namely, organizers visited the same groups repeatedly and engaged in small-group conversations which were individualized and interactive. For example, as I accompanied Chadema canvassers, I learnt how they spoke to people. I witnessed them engage people in steady-paced conversations (Fieldnotes, 2015b, 2015c, 2015d). I observed them altering (or tailoring) the content of their messages (Anonymous, 2015, Fieldnotes, 2015a, Fieldnotes, 2015b, Fieldnotes, 2015c). Lastly, I noticed that often, canvassers knew the people they visited and had visited them before, iteratively (Fieldnotes, 2015c, 2015d).

To summarize, my field research indicates that, consistent with the literature, Chadema focused its organizing in receptive areas. The scale of their organization achievement is illustrated by data from Mwanza Region. By 2015, Chadema had 171 foundations in Ilemela District alone (Anonymous, 2015m) which I saw first-hand (Anonymous, 2015b, 2015d, 2015i, 2015v). It had a further 252 foundations spread across the adjacent Magu and Sengerema Districts (Anonymous, 2015t). Macdonald’s study corroborates this strength in Mwanza Region.

However, elsewhere, Chadema had acquired less organizational reach. Kigaila’s assessment was that in 2010 ‘The Central [Zone] was weak; Singida, Dodoma, Morogoro’ (Kigaila, 2015b). By 2015, Chadema leaders claimed that they were organized nationwide (Kigaila, 2015a; Slaa, 2015). Nonetheless, my first-hand observation suggested that Chadema was often organizationally absent in the Central Zone, especially in Singida and Dodoma. For example, Kigaila said that Dodoma Town was Chadema’s organizational strong-point in the region (Kigaila, 2015a). However, in Dodoma City, Chadema had few branches; by attending opening ceremonies, I witnessed that it was literally opening branches a month before the 2015 election (Anonymous, 2015k; Fieldnotes, 2015g). Likewise, in Iramba District (Singida Region), Chadema reported 30 functioning branches out of a possible 50 (Anonymous, 2015s). However, at Chadema rallies in Iramba, I observed on three occasions that no local activists aided rally-production (Paget, 2020). Their absence suggests that there may have been fewer branches in practice (Fieldnotes, 2015i; Fieldnotes, 2015j). Macdonald’s study corroborates Chadema’s absence elsewhere in Dodoma Region (Macdonald, 2017).

In this section, I have empirically demonstrated that Chadema organized primarily through two modes of party-building between 2007 and 2015: organizing by touring leaders and organizing by branches. I developed these two concepts by distilling the processes which I have described above. By placing them in context, I have enriched them empirically and drawn out some idiosyncrasies, which in the name of abstraction, I omitted in my prior conceptual rendering. These modes of organizing are conceptually distinct. However, they are also interrelated. Organizing by party leaders engendered the establishment of branches, which subsequently went about establishing further branches. In effect, organizing by touring leaders multiplied the organizing by branches. Lastly, consistent with the literature, I have shown that Chadema focused organizing by touring leaders and branches in receptive areas. This serves as a foil against which I contrast my study of deviant party organizing in the Central Zone, which I turn to in the next section.

Beyond the frontier

In Singida East, ⁹ Chadema used to have almost no organizational presence at all. Chadema Central Committee member Tundu Lissu, who won Singida East in 2010 and 2015, recalled ‘There was no Chadema in Singida then [2008]. There were no branches, there was no leadership, there was nothing…’ (Lissu, 2015). In seven years, this changed. Lissu said in 2015 that ‘We are everywhere…You wouldn’t go to a village in which there is no Chadema branch chairman and secretary’ (Lissu, 2015). Other Chadema activists in Singida East affirmed this characterization. One recalled that in 2010:

Previously, there were some wards you would have found that there is no branch leadership but now in most of the wards you will find that branch and foundation leadership exists (Anonymous, 2015o).

They went on to report that Chadema had 50 foundations across the district, complemented by an advanced district-level campaign infrastructure (Anonymous, 2015o). These claims were corroborated by my site-visits, where I witnessed activists producing rallies (Fieldnotes, 2015h). Collord (2015) provides corroboratory evidence that Chadema organized widely in Singida East.

Singida East was a particularly hostile environment in which for Chadema to organize. The demographics were against Chadema. Compared to the Tanzanian mean, radio penetration was low, educational attainment was low and population-density was low (National Bureau of Statistics, 2012). Meanwhile, CCM was well-organized (Anonymous, 2015f). Chadema organized in Singida East through the personalized methods that Lissu employed. Without his efforts, Chadema would have not achieved this organizational presence. Lissu’s organizing in Singida East is an exemplary case of a mode of organizing by concentrating leaders found in several constituencies in the Central Zone and beyond. I interviewed Lissu formally in 2015, spoke with him informally on three further occasions, made two extended visits to his constituency during the 2015 campaign and saw him address three separate rallies, two of which were in Singida (Fieldnotes, 2015a, Fieldnotes, 2015b). I draw on this evidence below.

Lissu said he set out with the following objective: to literally build the party from the ground’ (Lissu, 2015). He recalled that ‘From January 2008 I was on the campaign trail, and I did not stop until 31st October 2010’ (Lissu, 2015). His method was to arrive, often unannounced; improvise a meeting; address the public; and recruit party members. Lissu described that ‘The first few months [we] were doing everything. You broadcast the meetings. You get the chairs’ (Lissu, 2015). Opposition leaders struggle to draw audiences. I witnessed another charismatic Chadema leader repeatedly fail to gather crowds (Fieldnotes, 2015j). Lissu said he overcame these challenges with persistence.

I gave myself a timetable that allowed me to be in Singida for two weeks every month. So it is meeting from village after village after village after village. Finish all the villages. Start all over again (Lissu, 2015).

Therefore, his methods were iterative. He clarified that by 2010, ‘I had been in every village at least three times’. (Lissu, 2015) Lissu believed that this repetition was crucial. He described that ‘perceptibly, things started to change, gradually. The message started sinking. People started joining [the party]’ (Lissu, 2015). Lissu also mitigated this problem by going ‘to places where there were people. I went to drink places. I went to market places’ (Lissu, 2015). I saw other Chadema parliamentary candidates using this method (Fieldnotes, 2015i, 2015e). They explained that they went to markets and bars ‘to find a place where people will gather easily’ (Mkisi, 2015).

His descriptions make clear that his meeting involved personalized communication. He said ‘Our meetings have always been massive learning classes, they are political literacy classes. We argue these issues. They ask questions’ (Lissu, 2015). These didactic and dialogic elements make his communication tailored and interactive, respectively, as defined above. Indeed, Lissu clarified that his method was persuasive; he said that ‘[I] argued my case in front of the people’ and elaborated that ‘The political argument you win by articulating issues’ (Lissu, 2015). Altogether, Lissu testifies that he organized in Singida East by using personalized methods of communication to convince people to become party activists.

The case of Singida East might have been idiosyncratic. Lissu is a leading politician in Chadema. In fact, in 2020, he became its presidential candidate (Paget, 2021). Therefore, his mode of organizing might have little relation to Chadema organizing elsewhere. However, I followed Chadema’s parliamentary candidate for Kwimba constituency in 2015, Babila Shilogela, who had organized in a similar way. Geographically, Kwimba District falls beyond the Central Zone. It lies on the South-Eastern tip of Mwanza Region. However, while Kwimba was nominally in the Victoria Zone, in 2010 it displayed many typical features of the Central Zone. It has low education, radio penetration and population density (National Bureau of Statistics, 2012). It is relatively remote from urban centres. A Chadema official admitted that Chadema had struggled to organize there (Anonymous, 2015p). Another attributed this to its remoteness (Anonymous, 2015e).

Much of the organizational progress that Chadema made between 2010 and 2015 was driven by Shilogela. He said that ‘I did M4C. I was in charge of M4C for the district’ (Shilogela, 2015a). Like Lissu, Shilogela started far in advance of the election. A Chadema ward official recalled that Shilogela began to organize the party from 2010–2012. Shilogela told me that ‘I was moving around the district’ (Shilogela, 2015a). A ward official corroborated that Shilogela ‘visited the whole of Kwimba’ (Anonymous, 2015r). He returned in 2013 and continued until 2015. His method closely approximated Lissu’s, and, he told me, primarily involved a succession of small rallies (Shilogela, 2015a). A Chadema ward chairperson described that ‘This young man…travels by vehicle and he is in charge of the Public Address System and calls the people, he also talks to them’ (Anonymous, 2015j). In other words, he held successive small rallies, which by virtue of their smalless, were interactive and individualized as defined above.

Like Lissu, Shilogela also placed iterative, incremental persuasion at the heart of this strategy. A Chadema ward official explained that he joined Chadema ‘Because I [was] impressed by the good policies of the party, the performance, and its goals…Chadema has enlightened us’ (Anonymous, 2015r). He said that Shilogela ‘has been telling us that we need to change the mind-sets of people first and then slowly we shall be instilling the spirit of change as we go along’ (Anonymous, 2015r). Shilogela’s organizing was intended to recruit members. The official said ‘Through volunteering and sacrificing our time we have managed to raise the number of members at ward level’ (Anonymous, 2015r).

Shilogela also contributed his personal resources to facilitate the recruitment of members and the organization of party organs. He bought party cards and core party equipment and distributed them across the constituency. He provided the vehicle and paid for fuel (Shilogela, 2015b). Another Chadema ward official remarked that ‘really Shilogela has done a lot’ (Anonymous, 2015a). I encountered several party activists who had been drawn to Chadema by Shilogela’s persuasion specifically (Anonymous, 2015r). The same official told me that he joined Chadema because ‘he [Shilogela] once addressed the rally during the M4C campaign and I was attracted by his preaching’ (Anonymous, 2015a). I witnessed a Shilogela campaign rally and witnessed the effect of his oratory first-hand (Fieldnotes, 2015l).

By the election in 2015, a CCM official judged that Chadema’s party-building remained incomplete in three wards (Anonymous, 2015w). However, a ward chairperson reported that Chadema had branch chairpersons in 60 villages (Anonymous, 2015j). Shilogela was commonly attributed the credit for this organizational expansion. The same ward chairperson said ‘I thank this young man, he has helped us with our visits across the district. He has really been helpful’ (Anonymous, 2015j).

Overall, Chadema organized less in Kwimba than Singida East, and the organizational improvement is less surprising, given that Kwimba is within Mwanza’s regional orbit. Nonetheless, the evidence of Shilogela’s party-building practices suggests that Chadema organized in Kwimba through Shilogela’s personalized organizing. This similarity between Lissu and Shilogela’s party-building practices is important. It suggests that the case of Singida East is not peculiar, but emblematic of a mode of party-building.

Indeed, the external validity of the practices drawn from these cases to the rest of Tanzania, and the concepts developed from them are further strengthened by Central Zone shadow cases. In Mlimba Constituency (Morogoro Region), Susan Kiwanga organized Chadema. She recalled that in 2010, ‘There was a structure but not a big structure’ (Kiwanga, 2015), but that ‘today [2015] there is a village chairperson in every village’ (Kiwanga, 2015). She won the seat that year. She told me that between 2011 and 2015, she endeavored to ‘form leaders at the grass roots’ (Kiwanga, 2015). She explained that ‘In each village, I do meetings, rallies - and underground meetings’ (Kiwanga, 2015). She explained that through these meetings, she reached out to different sets of citizens. Underground meetings were focused on party activists in particular (Kiwanga, 2015). ‘Meetings’ were small group conversations. ‘Some people come to our rallies, and some people they stay at home, so then you hold meetings to speak to those other people’ (Kiwanga, 2015). These meetings and underground meetings in particular feature the smallness which is typical of personalized communication as defined above.

Similarly, Kigaila, described above, adopted similar organizational methods in Dodoma Town (Kigaila, 2015b). He convened closed-door meetings where he met with small groups and persuaded them to become party activists (Kigaila, 2015b). I accompanied him to one such meeting (Fieldnotes 2015k). They had been invited by Kigaila and one of his party activists. The meeting was long, and like Lissu’s meeting, it was dialogic. Kigaila spoke for 30 minutes, and then he engaged in an unscripted conversation with the attendees for over 90 further minutes, which was interactive and individualized (Fieldnotes, 2015k). Kigaila told me that this was the third meeting with this group (Fieldnotes, 2015k). He and others explained that they had made only partial progress by 2015 because they had only recently started (Anonymous, 2015a, Anonymous, 2015b, Anonymous, 2015c; Kigaila, 2015).

Conclusion

These four cases were all instances of Chadema organizing against the odds. The implication of the literature is that opposition parties do not organize in unreceptive areas, but in those constituencies, Chadema did. They illustrate what is possible; they show that opposition parties can organize in hostile terrain. I use these deviant cases to refine the literature.

My starting-point is that party-building is under-theorized. The literature does not specify who organizes or how. Therefore, it insufficiently expresses the variety in party-building. I decompose opposition party-building into three modes. I generated these grounded categories through exploratory research in which I used site-intensive research to trace and construct practices. Thereby, I assure as much as possible the internal validity of my research. However, this case-study research cannot determine whether these categories capture the forms and variety of party-building practices elsewhere. Therefore, the external validity of these categories is for future research to explore.

Nonetheless, there are tentative signs that opposition parties organize through modes of party-building which approximate those developed here. For example, the Zambian opposition leader initiated an out-of-campaign season rally tour in 2008 (Cheeseman and Hinfelaar, 2009: 22). Eight years later, another Zambian opposition leader followed suit (Beardsworth, 2020). In Burkina Faso, an opposition party, organized rapidly by delegating party-building to aspiring parliamentary candidates in a manner which closely resembles lone-organizing (Bertrand, 2019).

By distinguishing between modes of party-building, I explain how Chadema organized in the Central Zone. Opposition parties organize in hostile areas through personalized communication by lone organizers. These leaders concentrate their efforts on building the party in one place, meeting the labor demands of personalized communication through the commitment of extensive personal time. They lighten those demands through the application of their personal resources and persuasive abilities. This explains how and through whom opposition parties organize in hostile environments. However, it does not explain what causes opposition parties to do so. Nor does it verify the effectiveness of these methods. These questions fall beyond the scope of the paper. They are gaps which future research might seek to fill.