Coding decisions on key armed actors, anti-coup demonstrations, violence in Rakhine state, and more
Published on: 2 March 2023 | Last updated: 4 November 2024
Background
Coding political violence in Myanmar poses several methodological challenges given the complexity of the many conflicts in the country. Since the military coup on 1 February 2021, a large number of new actors have emerged, and many new locations have become sites for political violence and demonstrations. This guide aims to explain how ACLED has coded political violence in Myanmar prior to the military coup and amid the ongoing changes to the conflict landscape since the coup.
The military coup in Myanmar was met with nationwide demonstrations against a return to military rule. The military used lethal force to crack down on these demonstrations, killing hundreds of demonstrators. Many anti-coup activists subsequently took up arms in an effort to defend their local communities from the military’s violence, forming resistance groups. Meanwhile, the military has also backed the formation of local militias to fight against anti-coup forces.
ACLED’s coverage of political violence and demonstrations in Myanmar spans from January 2010 to the present. Prior to the coup, the main form of political violence tracked in the ACLED dataset was the ongoing conflicts between the military and various ethnic armed organizations (EAOs), or ethnic resistance organizations (EROs), which have also continued since the coup. In particular, Kachin and northern Shan states were the sites of intense conflict between 2011 and 2018. Kachin state has again been the site of numerous battles between the military and the Kachin Independence Organization/Kachin Independence Army (KIO/KIA) since the coup. The United League of Arakan/Arakan Army (ULA/AA), meanwhile, clashed with the military in northern Rakhine state, and Paletwa township in Chin state, with the most intense period of conflict spanning from 2018 to just after the November 2020 general election. In addition to battles between state forces and EAOs, there are a number of territorial disputes that have led to battles between EAOs themselves, primarily in Shan state.
The complexity of disorder in Myanmar presents unique methodological challenges for recording political violence and demonstrations (for more information about ACLED-wide coding decisions, see ACLED’s Codebook). The key challenges include: identifying and labeling the many actors involved; georeferencing village locations given different transliterations of names; and triangulating data from various sources reporting on events.
How are key armed actors recorded?
State forces include standard military and police, as well as other paramilitaries, which function as part of the state’s security apparatus and are organized under the armed forces. ACLED organizes state actors based on regime years. In Myanmar, the regime years are 1988-2011, 2011-2016, 2016-, and 2021-. After the coup, ACLED has kept the 2016- regime years for state actors who represent the continuation of the National League for Democracy (NLD) government elected by the people in November 2020. This includes members of the Committee Representing Pyidaungsu Hluttaw (CRPH), composed of elected lawmakers ousted in the military coup, and/or the National Unity Government (NUG). The regime years 2021- are used to represent the State Administration Council (SAC) (also the ‘caretaker government’) formed by the leaders of the military coup and the military and police forces controlled by the military junta.
Three notable state forces sub-actors include:
- Military Forces of Myanmar (2021-) Border Guard Force: When the military implemented the Border Guard Force (BGF) scheme in 2009 and 2010, Myanmar military soldiers were integrated into some armed groups that had ceasefires with the military or armed groups that were factions of EAOs which acceded to the scheme. Most BGF groups are located in Kachin, Shan, Kayin, and Kayah states (The Asia Foundation, July 2016). They have fought alongside the military in battles against EAOs.
- Military Forces of Myanmar (2021-) People’s Militia Forces: When the BGF scheme was implemented, some local militias friendly with the military were transformed into People’s Militia Forces (PMFs). Other PMFs existed prior to the BGF scheme. These armed groups do not have regular military soldiers integrated into their forces, but are expected to support military combat operations (The Asia Foundation, July 2016). Many PMFs operate in Shan state and have fought alongside the military in battles against EAOs.
- Police Forces of Myanmar (2021-) Border Guard Police: A prominent state actor in Rakhine state is the Border Guard Police (not to be confused with the Border Guard Force), which focuses on security mainly along the border with Bangladesh. Before being disbanded in 2013, NaSaKa, coded as “Police Forces of Myanmar (2011-2016) Border Area Immigration Scrutinization and Supervision Bureau (NaSaKa),” provided security along the Myanmar-Bangladesh border (International Crisis Group, 16 July 2013).
Ethnic Armed Organizations (EAOs) have been coded using the names and abbreviations of both their political wings and armed wings. While presumably only the armed wings are involved in battles, some sources only refer to the political wing in their reporting. Thus, to avoid confusion for both researchers and users, both the political wing and armed wing names are included when coding an EAO actor. For example, the Kachin Independence Organization (KIO) is the political wing of the Kachin Independence Army (KIA). This group has been coded as “KIO/KIA: Kachin Independence Organization/Kachin Independence Army.”
In some cases, where there has been splintering of EAOs, ACLED has clarified the actor by using years. For example, this is the case for the DKBA in Kayin state, where ACLED has distinguished three different DKBA groups (Karen Human Rights Group, September 2016). When the military’s BGF scheme came into effect in 2009 and 2010, the original DKBA founded in 1994 (coded as “DKBA (Buddhist): Democratic Karen Buddhist Army (1994-2010)”) was split, with one faction becoming BGF and another continuing under the name Democratic Karen Benevolent Army (coded as “DKBA (Benevolent): Democratic Karen Benevolent Army (2010-)”). The “DKBA (Benevolent): Democratic Karen Benevolent Army (2010-)” group signed the National Ceasefire Agreement (NCA) in 2015. This group later split in two following several battles against the military on the Asia Highway in 2015. In early 2016, the new splinter faction named itself the Democratic Karen Buddhist Army (coded as “DKBA (Buddhist): Democratic Karen Buddhist Army (2016-)”) after the original group’s name. Media sources do not always clarify which faction of the DKBA they are reporting on, but ACLED researchers use additional information (e.g. names of commanders, locations of battles, etc.) to decide on the appropriate group to code.
ACLED also records alliances among EAOs. There are currently two key military alliances among the EAOs. These include:
- Northern Alliance (NA-B): Formed in late 2016, it is composed of four EAOs: KIO/KIA, ULA/AA, Palaung State Liberation Front/Ta’ang National Liberation Army (PSLF/TNLA), and the Myanmar National Truth and Justice Party/Myanmar National Democratic Alliance Army (MNTJP/MNDAA).
- Brotherhood Alliance: Formed in 2019, it is composed of the PSLF/TNLA, MNTJP/MNDAA, and ULA/AA.
In cases where all four groups making up the Northern Alliance fight together, “NA-B: Northern Alliance” is coded as the main Actor. In cases where only two or three of the groups are identified by the media as fighting together, those specific groups are coded as the main Actor and Associated Actors. The same holds true for the Brotherhood Alliance.
What new actors have emerged after the 2021 military coup?
Anti-coup resistance forces
After the military crackdown on peaceful protests, locals began to take up arms against the military in an effort to defend their communities from further military violence. This led to the emergence of anti-coup resistance groups. Some of these groups have called themselves people’s defense forces, while some have adopted other names (e.g. Chinland Defense Force (CDF)). On 5 May 2021, the NUG, formed by elected lawmakers overthrown in the military coup, formed the People’s Defense Force (PDF). While some anti-coup resistance forces have aligned themselves with the NUG’s PDF, others have not, though they share the same goal of resisting the military coup (Wilson Center, November 2021).
A number of resistance groups have formed alliances with each other as well (coding of these alliances follows the same logic for EAOs described above). Resistance groups have also joined with EAOs in several places in clashes with military forces.
ACLED codes all anti-coup resistance forces (i.e. the NUG’s PDF and those groups named and unnamed as described below) with an Interaction code of ‘Political Militias’ (3) to reflect armed, organized groups created for the furtherance of a political purpose. Such groups operate in conjunction or in alliance with a political elite – in this case, the government formed by those elected in the 2020 elections (for more on ACLED coding methodology, see ACLED’s Codebook).
ACLED codes such groups based on the names they give themselves and what is reported in the sources used by ACLED. When a name is only reported in Burmese, ACLED transliterates the Burmese name and creates an actor with the transliterated name. ACLED’s English translation of that name is also included. Where the group has translated their name into English themselves, that English name is coded alone without the transliteration.
When a more specific name of an actor is not given in a source, the actors are coded accordingly:
-
- PDF: People’s Defense Force: This actor represents the PDF formed by the NUG on 5 May 2021. The PDF operates under the NUG’s Ministry of Defense. It is coded when a PDF group, otherwise unnamed, is reported to have been formed by or is reported to be acting under direct orders of the NUG. When a group is reported to have acted as part of announced NUG operations, it is also coded. To date, two main operations have been announced: Operation Pyan Hlwar Aung and Operation Nan Htike Aung.
- Unidentified Anti-Coup Armed Group: This is a generic actor created by ACLED for use in the dataset that is coded for unidentified anti-coup resistance forces formed after 1 February 2021, referred to in the source by common terms like ‘people’s defense forces,’ ‘local defense forces,’ ‘local resistance forces,’ or ‘local guerilla group,’ but not explicitly reported as operating under the NUG and not referred to in the reporting by a specific name. The terms used in the reporting on such groups are captured in the Notes column.
- People’s Defense Force – [Location]: These actors are coded for the local autonomous PDF groups with formal names that add the location to the PDF moniker and are explicitly mentioned in the source. They may or may not be affiliated with the NUG. Thus, locations are only coded when they are part of the group’s name, not based on PDF activity in a specific location. This is so that the emergence of multiple PDF groups in one location can be coded based on the names that the groups give themselves.
- Example: If the source says “the Myingyan People’s Defense Force fought with the military,” the actor is coded as “People’s Defense Force – Myingyan.”
A single event may be reported in different ways by different sources. In cases where at least one source identifies the formal name of the group involved in an event, that name is coded as an actor for that event. ACLED triangulates reporting on events in an effort to include the name of the groups involved whenever possible.
The number of resistance groups that have emerged after the military coup is significant. ACLED records announcements of armed group formations as ‘Strategic developments’ where possible when reported on in the sources being covered. However, as ‘Strategic developments’ are coded using a methodology distinct from other events types, the list of resistance groups in the ‘Strategic developments’ category is not exhaustive (for more, see this methodology primer). As ACLED has noted in an earlier report, the announcement of a group’s formation is largely indicative of the adoption of armed struggle as a tactic of resistance and not indicative of group size or a large supply of weapons. Further, not all armed groups that announce their formation are necessarily involved in political violence. Also, not all groups who are involved in political violence make their formation known.
ACLED primarily tracks the number of armed resistance groups that are considered active, meaning that they have engaged in organized political violence, including ‘Battles,’ ‘Explosions/Remote violence,’ and/or ‘Violence against civilians’ events. A year after the 2021 coup, ACLED has recorded hundreds of distinct resistance groups engaged in such activity.
Unidentified armed groups
After the 1 February 2021 coup, the activity of unidentified armed groups has risen. The military and anti-coup resistance groups have traded blame in many cases where actors are unidentified. In many cases, groups may intentionally act anonymously. In cases where the source makes no reference to local defense forces, people’s defense forces, nor gives any indication of the involvement of such groups or any other identifiable actor, the actor is coded using “Unidentified Armed Group (Myanmar).” This is used in all cases in the data (pre- and post-coup) where actors are anonymous or unknown, in line with general ACLED methodology. “Unidentified Armed Group (Myanmar)” is coded with an Interaction code of ‘Political Militias’ (3), per ACLED methodology.
Thus, “Unidentified Armed Group (Myanmar)” does not represent the activity of groups that are described as unidentified people’s defense forces or other local defense forces (see above for more on how unidentified people’s defense forces and local defense forces are coded). If any group later claims to have perpetrated a certain event, or their involvement is subsequently reported, that event is updated, and the name of the group is included. If the event is later reported to be attributable to unidentified resistance groups, then the actor is changed accordingly, as described above. (See this methodology primer on how to keep your ACLED dataset updated to reflect such updates to the data.)
Pro-military militias
When demonstrations against the coup began in February 2021, military supporters were organized and encouraged to attack peaceful protesters (Guardian, 25 February 2021). Later, as anti-coup resistance groups emerged, the military stepped up its support of pro-military militias; namely, groups called Pyu Saw Htee after a Bagan-era king (Irrawaddy, 11 June 2021; Frontier Myanmar, 14 July 2021; International Crisis Group, 6 April 2022).
Pyu Saw Htee groups can act alongside state forces or alone. Pyu Saw Htee groups are coded as Associated Actors where they are reported to be acting alongside state forces. They are coded as the main Actor when they are reported to act alone. As Pyu Saw Htee groups are considered to be pro-government militias, they are coded with an Interaction code of ‘Political Militias’ (3). As additional military-backed militias emerge (Radio Free Asia, 10 May 2022), they are coded similarly.
Alleged and suspected members or informants of armed groups
After the military coup, alleged or suspected members of both military-backed militias and resistance groups have been targeted. There are concerns, however, over the veracity of allegations that people are members of certain armed groups (The Straits Times, 31 August 2021). As such, ACLED codes such events – where the source notes that the target of violence is an alleged or suspected member of a group – as ‘Violence against civilians.’ If reporting indicates they are a known member of an armed group, they are assumed to be armed, and the event is coded as an ‘Armed clash’ with the actor coded as described above.
This means that when a person is only reported by sources as a suspected member of a pro-junta militia, they are not definitively assumed to be members of the group, given concerns around the veracity of such claims; as a result, they are coded as civilians. In such cases, the group is not coded in the Associated Actor column. The Notes column can be reviewed to determine which events fall into this category.
A similar logic is used with regard to resistance groups where sources report that the military has falsely accused civilians of being involved with resistance groups (Myanmar Now, 3 December 2021). Again, such cases are coded as ‘Violence against civilians’ with the relevant information included in the Notes column, without the resistance group coded in the Associated Actor column.
Likewise, when alleged or suspected informants (ဒလန်) for the military or police are targeted, ACLED does not code the military or police as an Associated Actor, though the allegation is included in the Notes column. In situations where the source indicates that known informants for the military or police are targeted, “Civilians (Myanmar)” is coded as Actor 2, and the military or police is coded as an Associated Actor to track the person’s affiliation with the military or police.
In all cases, events are regularly reviewed and updated as new information becomes available in order to capture the most correct event type and/or actor affiliations.
How are anti-coup demonstrations recorded?
One of the first forms of resistance used by the public against the military coup was banging pots and pans at set times each evening. These coordinated pot banging protests are coded as ‘Protests’ events in the data. Similar pot banging ‘Protests’ events have also been coded in other regions of ACLED coverage, such as Brazil and Spain in the early days of the COVID-19 pandemic.
Demonstrations are recorded regardless of their duration. The flash-mob-type guerilla protests that have been occurring since the military crackdown on protesters are recorded as ‘Protests’ in the data.
As the military began killing peaceful protesters, public funerals and vigils took on a political tone. At some of these, mourners made the three-finger Hunger Games-inspired salute, held signs, and shouted protest slogans. These types of events are coded as ‘Protests.’ A similar strategy is used in coding such politicized funerals in other regions of ACLED coverage, such as Ireland.
Peopleless protests (လူမဲ့သပိတ်), in which material objects are set in protest formation in public spaces, are not coded as ‘Protests’ events as ACLED only codes protests where at least three people are physically present. Instead, these events are captured in the data as ‘Strategic developments’ (for more on ‘Strategic developments,’ see this methodology primer). The use of peopleless protest emerged in response to the violent military response to demonstrations to prevent demonstrators from being targeted.
How have fatalities been coded post-coup?
ACLED emphasizes that fatality estimates are often a poor approximation of a conflict’s form and impact. They are often debated and can vary widely. ACLED has maintained the same methodology for coding fatalities pre- and post-coup. Where there are diverging accounts of fatalities, the Notes column includes clarifications on various reported fatality numbers, as needed. As the ability of media outlets to independently verify claims has been severely constrained, the group providing the fatality information within a source is stated within the Notes when there are conflicting claims. The lowest number of fatalities reported is coded while taking into account the reliability of sources. When more information about a particular event becomes available, including updated information on fatalities, the relevant columns in the data are updated accordingly (for more, see ACLED’s Coding and Review Process).
Fighting between the military and resistance groups increased significantly beginning in early September 2021. While Myanmar has traditionally seen an increase in battles after the rainy season ends around October, the increased fighting also coincided with the NUG’s declaration of a “people’s defensive war” as well as increased military operations (Guardian, 7 September 2021). Along with the increase in events, there was a marked increase in the number of reported fatalities.
While some ‘Battles’ and ‘Explosions/Remote violence’ events have been reported to result in a higher number of fatalities than has traditionally been reported in Myanmar, engagement by the ACLED team with external experts has concluded that these fatality numbers need not be adjusted (i.e. ACLED does not assume that they are regularly purposefully and/or systematically over-inflated by media sources that ACLED draws from). While some events involving clashes between the Myanmar military and armed resistance groups have resulted in high fatalities, with the Myanmar military often incurring a higher proportion of the total fatalities, this is possibly due to the resistance groups’ growing proficiency with mine and IED ambushes, and the familiarity of these groups with the geography of their hometowns.
Nevertheless, as emphasized above, fatality numbers are frequently the most biased and poorly reported component of conflict data, so these numbers too should be considered with such caveats. For example, self-reporting by armed groups – including state forces and others – on events may lead to the inflation of enemy fatalities or deflation of their own fatalities. Given the current conditions for reporters in Myanmar, estimates given by such groups and reported by media sources often cannot be independently verified. The number of armed resistance groups that have emerged and the geographical spread of the fighting also contribute to the difficulties in verifying fatalities reported by armed groups or locals. The Myanmar military often denies reports of battlefield fatalities, but it is their own crackdown on the media which has prevented further investigation into the fatalities that are reported.
In short, ACLED does not have a separate mechanism to verify reported fatalities on the ground. The fatalities coded in the data are those reported by the trusted sources ACLED draws from each week. The source(s) for any event is cited in the Source column. ACLED fatality estimates are thus based on the most conservative estimate of reported fatalities.
When events are noted to have resulted in fatalities, but the number of fatalities is unknown, ACLED codes unknown fatalities as 3. ACLED has taken a conservative approach to coding unknown fatalities in Myanmar. (For more on ACLED’s fatality methodology, and how unknown fatalities are coded, see this guide.)
Forced portering and human shields
The Myanmar military has a history of abducting civilians and forcing them to porter and/or act as human shields when engaging in battle with EAOs and resistance groups (Reuters, 11 June 2021; DVB, 21 September 2011). This type of violence is captured in the dataset under the ‘Abduction/forced disappearance’ sub-event type. If the civilians involved are later killed during a battle or other event, the fatality is included in that event. If other armed groups engage in similar violence, it is coded under the ‘Abduction/forced disappearance’ sub-event type as well.
How has violence against the Rohingya in Rakhine state been coded?
The rise of Buddhist nationalist sentiments, often encouraged by the military, led to significant violence targeting Muslim communities in recent years — particularly the Rohingya Muslim community in Rakhine state. Coding the violence against the Rohingya population has posed additional challenges. The ACLED dataset covers both the communal clashes that occurred between Rakhine and Rohingya communities in Rakhine state in 2012, as well as the violence against the Rohingya population perpetrated by the military in October 2016 and August 2017.
The mass violence against Rohingya civilians in 2017 — which involved mass killings, mass rape, and the razing of villages — has been especially difficult to capture. This is due to the military restricting access to northern Rakhine state for journalists and aid workers. Attempts to report on the violence led to imprisonment, most notably for two now-released Reuters journalists who reported on the mass killing in Inn Din village (Myanmar Times, 7 May 2019).
While in the aftermath of the violence, many organizations were able to document the abuses that took place in the region by interviewing those who fled to Bangladesh, there are challenges to coding the reports based on these interviews. Some organizations have skillfully used survey data to estimate the aggregated number of fatalities and rapes during the violence (Medecins Sans Frontieres, 12 December 2017; Physicians for Human Rights, 30 August 2018). The lack of precise time and location information in some cases, however, does not allow for the disaggregation of all reported incidents into individual, discrete events per ACLED methodology. ACLED only codes those events for which there is sufficient detail within published reports with regard to the event date (when it happened), actors (who it involved), and location (where it happened). ACLED continues to research and revise events as additional information is made available.
How are locations recorded in Myanmar?
There are five administrative (Admin) levels in Myanmar:
- State/Region
- District
- Township
- Village Tract/Town
- Village/Ward
Villages are coded with Geoprecision 1 when the source reports that an event occurred in the village. Wards within towns are not coded (the town is instead coded at Geoprecision 1). When a source notes the event occurred near a village, the village is coded with Geoprecision 2. If an event is noted as occurring between two villages, one of the villages is coded and Geoprecision 2 is recorded.
When information about which village or town an event took place in is not included in the source, village tracts, townships, and districts are all recorded with Geoprecision 2. When an event location cannot be coded below Admin1, the location is coded at a natural location where possible, and, if not, at the capital of the state/region with Geoprecision 3. No events are coded at fuzziness beyond the Admin1 level (e.g. at the country level).
Coding villages in Kachin and Shan states can be especially challenging given the lack of standard English transliterations used in sources. The English transliteration for the Shan language name of a village can differ from the English transliteration for the Burmese language name of the village. Often, the Burmese language name for the village is a transliteration of the Shan name. For example, Mong is commonly part of village and township names in Shan state, but the Burmese transliteration is written မိုင်း, which produces a sound more akin to Mine in English. ACLED has identified common transliterations across the different languages; when it is still unclear which village is being referenced, the township is coded.
A similar challenge is seen when coding for locations in Kayin (Karen) state, where the location name in the reporting is often the Kayin language name. Reporting by Kayin media on events in Kayin state also often reflects the administrative boundaries of the Kayin nation as envisioned by many Kayin people, and thus does not always align with the central government’s administrative boundaries. ACLED compares maps reflecting both boundaries in order to code the event location. The location is coded using the English transliteration of the Burmese language name to standardize the language used to code locations across the dataset. The administrative boundaries recorded in the Admin columns are determined by the geolocation and the corresponding administrative division according to the Myanmar Information Management Unit boundaries data.
ACLED primarily uses location data developed by the Myanmar Information Management Unit (MIMU) to code village and township locations in Myanmar. Where different spellings for a location are possible between what is reported in a source and the MIMU locations data, the data are reviewed for both English and Burmese spelling variations.
There are cases where a village name is given, but the township information is not provided. This poses a challenge as there are villages with the same name in different townships. In such cases, the village that corresponds to areas of known conflict is recorded. In some sources, places that are not townships are reported as being townships; most often these locations are village tracts. If the geo-coordinates for the village cannot be found, but the township is known, the township is recorded as the location of the event. In cases where the township location is not given and the village location cannot be found due to transliteration issues, the location is set at a township where fighting is known to occur between the two armed groups. This is determined based on previously coded events and, in some cases, the battalion locations of the armed groups involved. Such events are set at Geoprecision 3.
How are events sourced?
Each week, ACLED researchers review dozens of English and Burmese language sources to code political violence and demonstration events. One challenge for sourcing events in Myanmar is that journalists have been restricted from accessing and reporting from many conflict locations in the country. Journalists who have reported on such conflicts have been subject to judicial harassment and imprisonment for their reporting (Committee to Protect Journalists, 31 July 2019).
Despite these challenges, several sources still manage to report on the daily battles between the military and various ethnic armed groups, as well as between ethnic armed groups themselves. With the growth of ‘ethnic media’ outlets, subnational sources have been able to produce more in-depth reporting from conflict areas (Burma News International, 23 September 2019). For example, sources such as Kachinland News or the Shan Herald Agency for News have detailed reporting on clashes in their respective states. As a result, subnational sources have been favored to code events over national and international sources where available, as they tend to have better access to conflict areas and stronger connections with the people and groups in those areas. This results in more detailed and timely reporting.
On the other hand, some national sources have also been able to access these conflict zones with some regularity, despite the risks. This, in addition to their coverage of demonstrations, means that national sources are also widely sourced. National sources such as the Democratic Voice of Burma have been primarily used to code the demonstrations across the country after the military coup.
A number of other source types are reviewed as a means of supplementing data gaps and triangulating details (for details on ACLED’s overall sourcing methodology, see FAQs: ACLED Sourcing Methodology). As one example, ACLED partnered with a local organization, Myanmar Peace Monitor (Burma News International), to improve coverage of ‘Battles’ events in several states. Data from the Myanmar Peace Monitor have been particularly useful in supplementing coverage of conflicts involving the PSLF/TNLA in Shan state. Local partner data are coded as ‘Local partner’ in the Source Scale column of the data.
Aside from traditional media reports and data from local partners, ACLED regularly sources reports by the UN, international monitoring groups like Human Rights Watch and Amnesty International, and local human rights organizations. These reports (assigned source scale ‘Other’ in the data) contribute to ACLED’s coverage of violence against civilians (VAC). For example, reports from the Burmese Women’s Union and the Kachin Women’s Association Thailand have contributed to the coding of events concerning violence targeting women in conflict zones. Such reports likewise allow for events to be revised where more information has become available since the initial traditional media report was published. In this way, ACLED’s data are continually improved upon as new information comes to light (for more, see ACLED’s Coding and Review Process).
Lastly, while ‘new media’ can be a significant supplemental source in some contexts, it varies widely in terms of quality. For Myanmar, ACLED’s use of new media sources has been limited, especially in light of concerns around the prevalence of ‘fake news’ (Wired, 6 July 2018) and fake accounts (CNET, 22 August 2019) on social media platforms like Facebook, coupled with the difficulty in verifying information reported on such platforms. As such, ACLED has deferred to traditional media and reports from reputable organizations to code events and kept the use of new media sources minimal.
Sourcing after the military coup
The media environment in Myanmar has deteriorated significantly since the military coup. The military has targeted journalists reporting on anti-coup activity (Reporters Without Borders, 1 October 2021), resulting in many media outlets being forced to shut down or operate covertly or in exile. Journalists have been arrested and tortured (Radio Free Asia, 29 June 2021). As high-profile media groups have been targeted, smaller, local groups have emerged to report on events, at great risk to their personal safety (Frontier Myanmar, 28 May 2021).
ACLED continues to update its sourcing list as the conflict and reporting landscape in Myanmar changes. Since the coup, ACLED has added new sources (such as Myanmar Pressphoto Agency and Myanmar Labour News) to its weekly sourcing. In order not to introduce an artificial increase in events by adding a source immediately, any potential new source is carefully reviewed and back-coded before it is added to the real-time source list (for more on ACLED’s sourcing methodology, see FAQs: ACLED Sourcing Methodology; for more on how new sources are added, see this methodology primer).
The total number of events coded by ACLED in Myanmar in 2021 surpasses the total number of events coded over the previous decade (2010 to 2020). Given the fast pace of events after the coup, and ACLED’s weekly real-time data release schedule, additional information on coded events may become available only after an event has been coded. As such, events in the dataset are updated as additional information becomes available. Events are also routinely checked against reports published by local groups tracking similar forms of violence (such as the Assistance Association for Political Prisoners (AAPP), which tracks arrests and state violence against civilians since the coup) in order to ensure the most accurate information available is coded.