Appendix Data Collection and Analysis

To collect the necessary data for my analysis, I conducted twelve months of fieldwork in Mozambique between July 2010 and July 2012 in five districts and two provincial capitals, and went on a shorter repeat visit in 2016. In Zambézia province, I conducted research in the Nicoadala, Namarrói, and Lugela districts, and in the provincial capital of Quelimane. In Nampula province, I worked in Murrupula and Mecubúri districts and in the provincial capital Nampula (see map in Figure 5.1 for an overview of fieldwork sites). For most of my fieldwork, I was based in Nampula city and traveled to the districts for repeated stays ranging from one to four weeks. In addition, I stayed in the Nicoadala district for one month and in Quelimane for five weeks, both in Zambézia province.

I began my fieldwork in Nicoadala in Zambézia province, as Naparama’s headquarters were in the district town by the end of the war and its main leaders still live in Nicoadala today. Similarly, Murrupula was the center of Naparama activity during the war in Nampula province. In both districts I was introduced to respondents by a trusted person who knew the Naparama combatants living there. The main objective of my work in these districts was to collect background material and data on temporal and spatial variation of militia activity in the two provinces. In addition, evidence from both districts served to explore and test the mechanisms that explain why individuals joined Naparama, allowing me to compare mobilization processes for state- and community-initiated militias.

I selected the remaining field sites depending on the presence or absence of Naparama forces and the severity of the war in that district (see Table 2.2 for an overview of the research design and case selection). I chose Lugela and Namarrói districts in the northwest of Zambézia province because both districts are similar in their geographic, cultural, and wartime characteristics, but the Naparama militia spread to Lugela district in a sustained manner, while it did not to Namarrói. I used evidence from Mecubúri district in the northwest of Nampula province to explore the argument developed in Lugela and Namarrói, since – like Namarrói and Lugela – it was one of the most war-affected districts in the region and one to which Naparama spread.

To reconstruct local histories of war and militia formation, I carried out more than 250 oral histories and semi-structured interviews with community residents, former Naparama combatants, Frelimo soldiers, and Renamo combatants, current and former government representatives, and community leaders, with an average of 46 respondents in each of the 5 districts (see Table A.1). Usually, interviews took place in a semi-private place in the respondents’ compounds, but in cases in which local authorities helped me to organize interviews, we met at the residence of the chief, headman, local administrator, or at the local administrative building. In addition to these interviews in the districts, I spoke to journalists, politicians, and researchers in Maputo, Quelimane, and Nampula about the history of the war and its legacy.

The interviews took between thirty minutes and five hours, sometimes in repeated sessions. If the respondent agreed (which happened in almost all cases), I recorded the interview with a digital audio recorder. I conducted the interviews in Portuguese or with the aid of an interpreter who translated the local language – Chuabo in Nicoadala, Makua in Mecubúri and Murrupula, Lomwe in Namarrói, and Manyawa in Lugela district – into Portuguese. The respondents spoke in the language they were most comfortable with, and sometimes mixed or changed the language during the interview. I did not remunerate respondents for their time, but in cases when I met respondents repeatedly, or when respondents had to wait to be interviewed, I brought small necessities such as food or soap or provided refreshments.

For most of my interviews, I worked with an assistant who translated during the interview and prepared translated transcriptions of the recorded interviews afterward. The assistants I worked with in the field were both Mozambican. My assistant in Nicoadala grew up in a locality in the district, but had lived in the provincial capital Quelimane for several years. He was not known to the respondents but, having grown up in the district, was not a stranger, which proved useful for establishing rapport with the respondents. The assistant I worked with in Mecubúri, Murrupula, Namarrói, and Lugela was originally from Zambézia province but had lived in Nampula province for a long time, thus being able to speak the local languages of both provinces fluently. As someone living in the city, he was perceived more as a stranger than my assistant in Nicoadala (see Chapter 3). In both cases, my assistants did not come from the communities we worked in but possessed the necessary knowledge of cultural codes and social rules that were valuable for setting up interviews and ensuring that respondents felt comfortable. They were also able to explain responses I would not have understood myself due to the social and cultural barriers or lack of context. Both assistants possessed lots of field research experience since they had worked for other researchers before, either translating interviews or conducting surveys. However, the focus on open-ended questions and narratives in interviews was new to both of them, and thus it took several interviews to practice the type of questions to ask and the right moments for translation. All translations from Portuguese to English of quotes used in the book are mine unless noted otherwise.

Before my research assistant and I began each interview, we explained our project and asked for (oral) informed consent to be interviewed. At the beginning and end of each interview, we also encouraged the respondents to let us know if they had any concerns or questions about the objectives of the interview and the project. Asking for questions after the interview was crucial, as some respondents felt more comfortable coming forward with their questions after we built up rapport during the interview. I never asked directly about the experience of or participation in violence and waited for the respondent to divulge such information to protect the respondents from retelling painful experiences. The project was approved by the Human Subjects Committee of Yale University, under the IRB protocol number 110308177.

Since data sources can have distinct biases (Davenport and Ball Reference Davenport and Ball2002), I aimed to collect data from a variety of sources. To supplement and cross-check the information from the interviews, I collected documents from the provincial government archives in Quelimane (Zambézia province) and Nampula city (Nampula province). I collected, digitally preserved, and organized more than 10,000 of government documents produced during the period 1975–94 by state and party agencies on the provincial and district level. The collection of documents includes monthly reports about political, economic, and social developments in the districts; reports about special visits by administrative personnel to the districts; and radio messages from the districts to the provincial government about special occurrences. Most of these documents have never been used by other researchers.Footnote ¹

I selected documents for digital preservation depending on their relevance to my research; that is, I made sure to collect all the reports I found that provided information about violent events, security, military affairs, and relevant social affairs such as communal villages and the effects of war on daily life. To illustrate the coverage of the collected documents, Figures A.1 and A.2 provide an overview of the number of documents by location and by year, collected in the archive in Quelimane, Zambézia province. The figure for “Zambézia” in Figure A.1 is high as the provincial office often summarized the monthly reports from the district. The figure for 1992 and 1993 reflects reports collected on the violation of the ceasefire and peace accord that was signed on October 4, 1992.

Figure A.1. Number of documents collected from the Zambézia archive by district

Figure A.2. Number of documents collected from the Zambézia archive by year

A research assistant and I created a dataset of violent events using the archival reports from Zambézia province. The codebook builds on other efforts to collect violent event data, such as the Uppsala Conflict Data Program’s (UCDP) Georeferenced Event Dataset (Sundberg and Melander Reference Sundberg and Melander2013), the Armed Conflict Location and Event Dataset (ACLED) (Raleigh et al. Reference Raleigh2010), and the Sierra Leone Local-Location Event Dataset (Bruijne Reference Bruijne2014). This ensures that the dataset is compatible with other quantitative data collection efforts and can be combined with existing datasets to extend the analysis. The dataset for Zambézia province includes over 1,300 events from the time period between 1974 and 1994. Each event is categorized into different types of violence so that the analysis can differentiate between violence against civilian and military targets and demonstrate patterns of violence across space and time. The dataset focuses on the following actors: government armed forces, rebel forces, state-initiated militias, and community-initiated militias. The typology of events the dataset includes builds on ACLED, which focuses on changes in territorial control (Raleigh et al. Reference Raleigh2010). Battles can result in changes in territorial control or allow a conflict party to remain in control. The dataset also codes violence against civilians as a separate category.

Gaining access to the archives to collect government documents was difficult. Government documents in Nampula and Zambézia provinces from the time of the war are not stored in official archives, and the exact location of the archives and the rules for access were unknown to many government employees. The archives are not organized, so I had to sift through many documents that were not of relevance to my project. In Nampula, documents were stored in boxes in a garage that was part of the provincial finance division guesthouse where they suffered damage from heat, water, dust, and rats. I found similar conditions in Quelimane. Although the building in Quelimane in which the documents were stored also housed documents from other provincial departments, the air-conditioning unit that should have preserved the documents was broken.Footnote ²

Although my collection of documents spans many districts and time periods, it may contain some bias due to the difficult conditions in which the documents are stored and also due to time constraints, but it is difficult to ascertain how systematic the bias is. The office of the district administrator, the police, the security agency, and the military stationed in the district produced reports, but the level of detail in the reports varied. Some reports covered all security-related events, while others left out violent events I knew about from other sources. In general, though, they all had the same structure, as the provincial government provided the districts with guidelines to prepare such reports. Another source of bias may include deliberate selective reporting by local administrative staff. I attempted to deal with these potential sources of bias by cross-checking information from different reports and triangulating that information with evidence from interviews and newspapers. All documents were written in Portuguese; the translations of quotes from the documents are all mine.

In addition to the documents from the archives in Nampula and Quelimane, I collected newspaper and magazine articles; development plans of the districts I worked in that provide basic geographic, administrative, social, cultural, and economic information; and presidential speeches and TV reports. I obtained this material in the archive of the Mozambican News Agency (Agência de Informação de Moçambique), in the National Historical Archives (Arquívo Nacional Histórico), and from the national TV channel Televisão Moçambique (TVM), all in Maputo, and from Yale University Library.

From the interviews and documents, I obtained detailed information on community histories; levels of violence and territorial control by insurgents and the government; motivations for joining the militia; the relationship between the militia, government, and the population; and current activities of former militia members who demand pensions from the government for their wartime contribution.