Data collection took place between March and May 2019. In each country, a team consisting of local researchers either from, or familiar with, these regions was trained during a two-day workshop to conduct structured key informant interviews (KIIs) and focus group discussions (FGDs) in the selected municipalities. In principle, all researchers followed the same interview and FGD guides, although the workshops were used to tailor the questionnaires to local circumstances and sensitivities. In addition, the training workshops served to map the local authority structures and community dynamics to identify key respondents for the structured interviews and FGDs.

In every municipality, the local research teams conducted between 20-25 structured KIIs with traditional authorities, religious figures, state authorities (to the extent that these were present, such as mayors, security forces and the judiciary), non-state authorities such as leaders of armed groups and self-defence groups (again, to the extent they were present), and civil society organisations. The structured nature of the interviews allowed for the comparison of findings across municipalities and countries. In addition, the local research teams conducted between 1-4 FGDs in each municipality with 5-15 respondents per session. Focus group participants explicitly included women, young people and underrepresented groups. In total, the study uses data obtained through 323 KIIs and 34 FGDs. The interviews and FGDs were conducted in French, Arabic and other local languages.

The local researchers provided French and Arabic transcripts of the KIIs and FGDs to the Clingendael project team. Where possible, we aimed to receive verbatim transcripts but recording the interviews was often not possible due to the sensitive nature of the questions and the fact that data collection took place in local areas experiencing ongoing violence and insecurity. Receiving the interview notes also proved challenging due to the limited availability of communication means in remote regions, to the local researchers having limited to no access to electricity, and to the outbreak of violent conflict in Tripoli in early April 2019, as well as massive flooding in the Ghat region (the latter two issues preventing the local researchers from travelling to other locations to type out their interview notes). The local researchers had to resort to creative solutions like sending photographs of their handwritten notes via WhatsApp (Al-Barkat) and sending their handwritten notes and audio recordings via UN Humanitarian Air Services flights (Mali) and buses (Niger) to Bamako and Niamey respectively so that they could be typed up there by other members of the local research teams. To the extent possible, the local researchers checked the transcripts before sending them to the Clingendael project team.

The abovementioned solutions all posed different risks to the data. To ensure the quality and consistency of the data, the Clingendael project team sought to engage with the local researchers on a daily to weekly basis (depending on the available lines of communication – either through direct WhatsApp contact or through the local research coordinators) to follow up on data collection issues and answer questions from the field. The project team asked the local research teams to send transcripts of their interviews on a daily to weekly basis, again to monitor the quality and consistency of the data. This allowed the team to identify data collection issues early on and to address them as soon as possible.

To ensure the validity of the data, the Clingendael project team, VNG International, and Bureau Ensemble organised 1-2 rounds of follow-up workshops and meetings in the three countries to discuss the initial findings and recommendations with the local research teams, as well as with a wide range of experts, (traditional) authorities, NGOs and members of the international community (the latter of which are all potential end users of the project’s findings).[401] As well as data validation, many of these conversations took the form of key informant interviews that allowed the Clingendael project team to fill gaps in the data and collect additional information. In Libya, for example, the Libyan researchers engaged in a participative scenario exercise mapping local authority structures under different scenarios that proved insightful into local dynamics but perhaps even more so into local conflict lines that ran across the local research team itself.

So than an analysis could be undertaken, the project team coded all transcripts using NVivo software. A codebook was developed to enable the coding of the transcripts – followed by an intercoder reliability exercise to ensure consistency across the different coders. We coded 6,426 text segments. The coding focused on identifying the main governance actors identified by the respondents (cases), the relationships between the different actors (relationships), and the main dimensions of legitimacy (rules, values, functions – see conceptualisation in Chapter 1.1) and challenges/changes to this legitimacy in the municipalities under study (nodes). The coding allowed for the creation of network sociograms that we have published online using the kumu.io software.[402] The online sociograms provide access to some of the key data points underlying our analysis of the relationships between different governance actors in the municipalities under study here. They can be found here: link.