Published on: 2 March 2023 | Last updated: 29 November 2023
Each week, researchers at the Armed Conflict Location & Event Data Project (ACLED) review dozens of English, French, and Kirundi language sources to code political violence and demonstration events across Burundi. To address challenges posed by the country’s media landscape, ACLED uses a wide range of source types from multiple regions with different methods and biases.
Ongoing and deepening restrictions against local and foreign journalism have created obstacles to sourcing events in Burundi (RFI, 21 October 2019). Many international journalists are barred from working in the country, while local journalists experience limited access by the state, media shutdowns, imprisonment, or even attacks. 2019 protests against international journalism by supporters of the ruling party, the National Council for the Defense of Democracy–Forces for the Defense of Democracy (CNDD-FDD), and political decisions to restrict both international and domestic media have revealed the ruling party’s intolerance of media criticism (IWACU, 8 June 2019).
Traditional media outlets (represented by the International, Regional, National, and Subnational Source Scales in the data) have been the targets of increased government censorship since 2015 (VOA, 29 March 2019), when the Burundi Crisis first began.1The Burundi Crisis first began in 2015 following President Nkurunziza announcing a third mandate, despite this being against the constitution. The announcement was met with wide demonstrations and a coup attempt, resulting in shutdown of the internet, telephone networks, and media coverage (for more on the Burundi Crisis, see Raleigh et al. 2016). International media outlets — such as Agence France Presse (AFP), the British Broadcasting Corporation (BBC), Radio France Internationale (RFI), Voice of America (VOA), and Xinhua — formed a much larger portion of ACLED’s coverage until 2015, with declining coverage since then. Regional news sources — such as All Africa, the East African (Kenya) and the Rwandan News Agency — have also been affected in light of increasing tensions and distrust between countries in the East African community (ISS Africa, 24 May 2019). Since 2015, regional and international sources contribute only one-tenth of ACLED’s Burundi coverage. All media outlets operating openly in Burundi — such as Radio Television Nationale du Burundi (RTNB), and Le Renouveau — must operate under strict state guidelines outlined by the National Council of Communication (Freedom House, 2016).2 This is as of December 2019. Although, some national media sources — such as Iwacu and Radio Publique Africaine (RPA) — have minimized this loss by continuing to publish independently and free of state media regulations through underground journalism and without official permission to operate in the country. Regardless, ACLED’s use of traditional media outlets has continually declined since 2015 due to media censorship, and accounts for approximately a quarter of events since 2015.
Besides state censorship, media bias by sources both within Burundi and from those outside the country provides an additional sourcing challenge. Sources like RTNB contain a strong pro-government bias and report very conservative estimates for casualties incurred by state forces, while also downplaying the role of Imbonerakure in violence (RTNB, 26 October 2017), and limiting reports of casualties and clashes in general in order to project a more peaceful image. External media sources, such as the BBC or RPA, often have a more critical posture towards the current government and carry this into their reporting. These biases must be taken into consideration when covering events where biases provide incentives to include higher or lower casualty counts or exclude events altogether (for more information on how ACLED deals with bias in general, as well as other sourcing issues, see our Sourcing Methodology FAQ).
As a result of these restrictions imposed on traditional media, ACLED prioritizes the inclusion of alternative source types, including information from local partners, ‘Other’ sources (i.e. reports by reputable organizations like monitoring groups and research projects), and new media.
In recent years, over half of ACLED’s Burundi coverage has come from ACLED’s local partner. These data are attributed to an ‘undisclosed source’ in the data. This anonymity has been requested by the partner whose data collection comes from anonymous citizen reports which are counter-verified by other citizen reporters before being audited by provincial coordinators. These reports capture events in areas where journalists have been barred from operating. They are effective in capturing political violence carried out by state forces and the Imbonerakure (the armed wing of the current ruling party, the CNDD-FDD) against perceived opponents — which local media would be barred from reporting — as well as communal violence in rural areas of the country — where journalists are not always present.3Unfortunately, ACLED’s anonymous local partner ceased operations in August 2019. This stems from the heightened risk to those working with the organization becoming too grave for most citizen journalists and provincial coordinators to be able to continue, meaning data sharing with ACLED has come to a halt, as of this writing. ACLED’s ‘do no harm’ policy prioritizes the well-being of partner organizations first and foremost. Many of ACLED’s local partner organizations based around the world face similar risks and must determine their ability to pursue data collection in the face of such threats regularly.
ACLED regularly reviews reports from national and international monitoring groups to supplement information or to identify events unreported elsewhere. These reports (categorized under ‘Other’ in the Source Scale in the data) may include reports from organizations such as Human Rights Watch, Amnesty International, or Africa Research Bulletin, for example. Such reports often include in-depth coverage of particular issues surrounding human rights abuses, appearing most often as ‘Violence against civilians’ events. Nearly one-tenth of events come from such sources.4ACLED is currently in the process of trialing a pilot: fostering a network of engagement with individual journalists across the African continent, including in Burundi.
Lastly, new media may provide valuable insights and can help to supplement traditional media in some cases, yet it varies widely in terms of quality, reliability, and bias. ACLED relies upon information from select, trusted new media sources in the Burundian context, which are added to source lists after they have been assessed for reliability. SOS Medias, for example, is used, as are Humura and Inzamba, both in Kirundi — all of which are partners of Radio Publique Africaine (RPA). These sources are especially effective in covering political violence carried out by state forces and the Imbonerakure against political opponents.5These sources have proven especially helpful in light of ACLED’s local partner ceasing operations given that these sources report on similar types of violence. Another example is data from the 2015 Burundi Crowdmap; this initiative, with oversight from Peace Geeks (a Canadian NGO), mapped incidents of violence and protest from secondary local sources including local newspapers and journalist networks, as well as from established journalists’ social media accounts, during the height of the Burundi Crisis in 2015 and into early 2016.
