ACLED’s aim is to capture the forms, agents, dates and locations of political violence and protest as it occurs within developing states.
This page contains information about how the ACLED team collects, cleans, reviews and checks event data, with a focus on what makes ACLED unique and compatible with other data. The process of ACLED coding assures that it is accurate, comprehensive, transparent and regularly updated. ACLED-Africa data is available from 1997 and into real time. ACLED-Asia produces publicly available real-time data and continues to backdate for all states. Data will be posted as it is complete.
ACLED began in 2005 and coded data for six Central African states and three West African states. ACLED-Africa released its fifth version as of January 2015. This version has expanded, altered, updated and corrected information from the four versions that precede this release. Our present coding rules are largely similar to those the beta set established in 2005, but decision rules, geographic reach and content has expanded since. In 2008, ACLED focused on developing a dataset for the fifty least developed countries, with a concentration on African states. These early data are compiled in version 1. Versions 2 for 2011, 3 for 2012, and 4 for 2013 are each revised and extended collections of political violence and protest data occurring in African states. In 2014, ACLED expanded to include ten states in South and Southeast Asia. Covering Pakistan, India, Bangladesh, Myanmar, Nepal, Laos, Thailand, Cambodia, Vietnam, and Sri Lanka, the ACLED-Asia team tracks real time political violence and protest in addition to chronicling political violence between 2010-2014.
ACLED data are coded by a range of experienced researchers who collect information primarily from secondary source information and apply the guidelines outlined in the codebook to extract information from news reports. ACLED codes information by date, location, agent and event type; this means that a battle occurring over the course of a weekend will be recorded as two events, one on each day. Further, if two distinct events (e.g. an act of violence against civilians and a riot) occur in Mogadishu on Jan 2nd, both events are discrete observations in the dataset. As we aim to capture how political violence occurs within states, we ascribe no fatality minimum to be included in our coded events.
ACLED Coding Methodology
Every event is coded using the same rules on who, what, where, and when, to maximize comparability and validity, thorough information. Additional information, such as event ID numbers, precision scores for location and time, notes to give the context of the event, fatality numbers if reported, codes to distinguish between the types of actors, and additional spatial information are also provided in each row of information. ACLED data are released in both excel and csv forms. The most recent version of the dataset is available here.
ACLED data are collected each week after individual coders have scrutinized the information from reports; they are then aggregated and revised by the first coding reviewer, investigated and cross-checked by the second reviewer and then event notes and details are inspected by the third and final reviewer. The process is designed to assure (1) validity through intra- and inter-coder checks; (2) accuracy to correct mistakes in coding; and (3) relevance by determining whether each compiled event constitutes an act of political violence or protest.
Details of the review process can be found here.
Below are a series of definitions and example codes that highlight the detail of each ACLED event. A more detailed description of event types, actors, and other details are available here.
Definition of Political Violence and Protest
Political violence is the use of force by a group with a political purpose or motivation. ACLED defines political violence through its constituent events, the intent of which is to produce a comprehensive overview of all forms of political conflict within and across states. A politically violent event is a single altercation where often force is used by one or more groups to a political end, although some instances – including protests and non-violent activity – are included in the dataset to capture the potential pre-cursors or critical junctures of a conflict.
We code nine event types including three types of battles, violence against civilians, remote violence, rioting (violent demonstrations) and protesting (non-violent demonstrations) and three types of non-violent events.
Battles are violent clashes between at least two armed groups. Battle types are distinguished by whether control of a location is unchanged as a consequence of the event; whether a non-state group has assumed control of a location, or whether a government has resumed control of that location. Battles make up approximately one third of the dataset.
Examples of Battles:
- Libya: Ghat residents and local forces repelled an attempt by Libya Dawn to take over the Ghat airport. Some 25 armoured vehicles carrying Libya Dawn forces were reported to have stormed the Ghat Airport on Friday night.
- Nigeria: Military forces launch an offensive on Boko Haram positions in Damaturu. A total of 14 soldiers were killed in action during the attack.
- Somalia: Two opposite clan militias hailing from Habar-Gedir/Sacad clashed in Galinsor village (30km NE of Cadaado) in the afternoon of 10/01. The two sides exchanged gunfire.
- Sudan: The SAF claims they killed between 41-100 rebel fighters (coded as 41) and took control of Angartu, near Kadugli, during fighting over the weekend, with assistance from the RSF.
- Pakistan: Pakistani paramilitary troops engaged with Balochistan Liberation Army militants in the Pirkoh area of Dera Bugti, Balochistan. 13 militants were killed on a gas field by troops backed by helicopters.
- Myanmar: Fighting flared between the Myanmar military and the SSA-South, with a member of the rebel group accusing government troops of staging an attack on its base in Shan State’s Mauk Mae Township. No casualties were reported.
Violence against civilians involves violent attacks on unarmed civilians. These acts comprise a third of the collected data.
Examples of violence against civilians events:
- Nigeria: Two female suicide bombers, suspected as Boko Haram, detonated themselves up at a Yobe market (along Muhammed Idriss Way in Potiskum) after detonating bombs wired to their bodies, killing 39. The girls were said to be between 15 and 17-years-old.
- India: A Communist Party of India (Marxist) activist was hacked to death by Rashtriya Swayamsevak Sangh activists at Chukku Bazaar, near Thrissur district, on Sunday night.
- South Sudan: 8 people were killed in a revenge attack by Pagoor youth on Tieptiep payam in Cueibet county, allegedly due to the release of a Tieptiep clan member who killed one of the Pagoor community. 2 others were injured, including the paramount chief.
- Sudan: 3 villagers were killed in attacks on villages near Al Fasher by RSF militia-men. A number of people were wounded, 10 were abducted, and large numbers of livestock were also stolen.
Non-violent activity includes incidences of looting, peace-talks, high profile arrests, recruitment into non-state groups etc, and accounts for 6% of the total dataset
Examples of non-violent events:
- Rwanda: Rwandan genocide suspect Jean Paul Birindabagabo is arrested and extradited to the ICC.
- Bangladesh: In an overnight raid in Chittagong, police arrested 82 activists of Jamaat-e-Islami and its student wing Islami Chhatra Shibir in addition to acquiring firearms and bullets.
- Nigeria: Police discover and dismantle a bomb targeting the Owerri Capital Development Authority (OCDA) office in Owerri, Nigeria.
- South Sudan: A peace conference has started in Mundri, South Sudan to resolve disputes between pastoralists and local farmers in the area following clashes which began in October 2014. (note: ACLED codes only the first announcement of ongoing peace talks.
- Chad / Nigeria: Chad has deployed a large military force, including at least 400 vehicles and helicopters, to assist Nigeria and Cameroon in their fight against Boko Haram. The vehicles arrived in Kousseri on Jan.17.
Non-violent takeover of territory is also under 1% of the dataset, as is headquarter or base establishment.
Examples of non-violent takeover events:
- Somalia: Al Shabaab forces take control of three Bay villages (Hagarko, Barbaare – near Qansadheer – and Buulo Barako) following the withdrawal of government forces.
- Libya: The Islamic State in Iraq and the Levant [Islamic State, IS, ISIS/ISIL] known by its [Arabic] acronym Da’ish, has officially announced the establishment of its branch in Libya and based in Benghazi.
Riots assume 7.5% of the dataset.
A riot is a violent demonstration, often involving a spontaneous action by unorganized, unaffiliated members of society
Examples of riot events:
- Burkina Faso: Rioters set fire to Truegold mining materials in Noogo. Police forces used tear gas to take control of the area while firefighters attended to the scene.
- Mauritania: After three anti-slavery activists received prison sentences, dozens of their supporters stormed the courthouse and smashed police van windows. Police used tear gas to disperse them, leaving four injured. Clashes between police and anti-slavery activists expanded from the courthouse to the civil prison area, causing closure of the city market. Rioters surrounded police vehicles and police used tear gas to disperse them.
- Algeria: Thousands of Algiers’ inhabitants went out after the prayer on Friday to denounce the repeated attacks on Islam. Police intervened to disperse the demonstrators, who reacted by throwing stones and bottles. A few police officers were wounded.
- Nepal: Two student unions affiliated to ruling CPN-UML and main opposition UCPN (Maoist) clashed over the appointment of Campus Chief at Lainchour-based Amrit Science Campus.The students also vandalised the office of the Campus Chief.
- Bangladesh: Hartal supporters vandalised at least six vehicles and exploded several crude bombs in Paltan on Sunday morning. No injuries were reported. According to the police, the rioters were activists of Jamaat-e-Islami and its student wing Islami Chhatra Shibir.
Protests are non-violent demonstrations, involving typically unorganized action by members of society. They comprise 14% of the total dataset.
Examples of protest events:
- Sierra Leone: Health workers at the PTS I Hastings Ebola Treatment Center in Freetown protested against non-payment of their December hazard incentives.
- Zimbabwe: More than 200 villagers stage a demonstration at Masvingo Central Police Station against failure by Chiefs Musara and Chikwanda to resolve a boundary dispute. 8 of the demonstrators are arrested.
- Cambodia: More than 50 CNRP officials seeking the removal of lawmaker Ke Sovannaroth as the party’s head in Siem Reap province, demonstrated for two hours outside the opposition’s headquarters in Meanchey district. The protesters accused Ms. Sovannaroth of nepotism and spending irregularities.
- Thailand: Around 40 students rallied outside a military court in Bangkok, chanting slogans in support of four Thai men charged over an anti-coup protest and against the use of military courts to try civilians.
Remote violence refers to events in which the tool for engaging in conflict did not require the physical presence of the perpetrator.
Examples of remote violence events:
- Somalia: An RCIED detonated in the outskirts of Khadija Haji village (40km SW of Beled Xaawo) in the afternoon of January 8th. Reports indicate that the target was reportedly for SNG commander. The device seriously injured five civilians.
- Sudan: The SPLM-N has claimed the SAF bombed the area southwest of Kurmuk, killing 2 civilians and wounding 4 others, two seriously, between Jan.12-13. On Jan.13, the Shali area of Kurmuk was bombed specifically.
- Pakistan: Twenty-four militants were killed and five others sustained injuries when Pakistan Air Force fighter aircraft bombed their hideouts in Tirah Valley, Khyber Agency. Security forces targeted hideouts of the Tehreek-e-Taliban Pakistan and Lashkar-e-Islam in Dars Jumaat, Sandana, Takhtakai and Khyber Sangar in Tirah Valley.
- Bangladesh: A bomb was exploded at the Dhaka Metropolitan Magistrate (CMM) Court premises in the city on Thursday afternoon. The bomb was set off at the balcony in front of the courtroom. However, no one was injured in the incident. It is assumed that the BNP militia was behind the attack.
ACLED does not categorize clusters or campaigns of events, largely because political violence, protest and conflict are aggregations of multiple actors, attacks, goals, etc. To classify events as discrete acts in an ‘insurgency’ or ‘terrorism’, ‘civil war’ or ‘livelihood conflict’ is often a reductive choice, leaving out the complex ways in which conflict creates multiple scales of violence and intended outcomes within a territory. ACLED allows users to select on the type of event, type of actor, type of interaction, named actors, location or time period. We do not dictate nor frame a series of conflict events for researchers, but allow users to determine how to aggregate as they see fit.
ACLED recognizes a range of actors including governments, rebels, militias, ethnic groups, active political organizations, and civilians. In ACLED, politically violent actors include rebels, militias, and organized political groups who interact over issues of political authority (i.e. territorial control, government control, access to resources, etc.). A more detailed description of event types, actors, and other details are available here.
A full list of ACLED actor name and type list will be made available soon.
ACLED does not have a fatality threshold for event inclusion; further, it reports fatalities only when a reputable source has relayed that information. It uses the most conservative estimate available, and will revise and correct the totals- upward or downward- when better information comes available.
ACLED collects information from a variety of primary and secondary sources. By aggregating local and international news sources using various databases, ACLED coders track reporting on events concerning political violence and protest in Africa and Asia. Coders use local, state, and international media sources to capture events ranging from protests to battles.
A working paper on sourcing is available here
When confronted with difficult cases where the actors or types of violence are unclear, ACLED coders follow standard rules including 1) finding additional information; 2) how to code an event with multiple dimensions (e.g. a riot that turns into violence against civilians is not coded as two events, but as the ultimate event against civilians); and 3) discussing specifics with researchers to establish if precedents exist. If the event is closer to criminal violence, over political violence where political objectives are the goal, then it is omitted from the observations.
Below are some examples of common difficult cases:
- South Africa: Vigilante violence where two suspected witches are killed in South Africa by a vigilante group.
These events are often predicated on the internal security arrangements within a community. For that reason, vigilante groups function in similar ways to communal groups, as they believe their role to be the delivery of justice, and their group to be the reliable security arrangements within a community.
- Kenya: Pokot cattle rustlers struck Namawanga village inhabited by members of the Luhya community and shot dead a local farmer before stealing three bulls from his home. In revenge, Luhya tribesmen hacked to death a Pokot man after they found him ferrying two passengers to Karnyarkwat market. The killing of the man sparked off protest from members of the Pokot community who thronged to Kapkoi market vowing to take revenge over the killing of their tribesmen.
Multiple events may also be reported. In the case above, there are discrete events in Namawanga, Karyarkwat and Kapkoi. In the first two cases, different communal militias attacked civilians; in the final case, a protest was held in a market to mark the instability brought about by the first events.
ACLED is designed as a comprehensive dataset on political violence and protest. It aims to capture political violence that remains absent from event datasets which define conflict as civil war, or other limited categories. To demonstrate how this differs in terms of usable data, please see a previous comparison of ACLED and other sets here and here. See below for a mapped comparison between UCDP-GED data and ACLED (restricted sample from 2008-2010). We use the full range of UCDP event types; further information can be accessed on that project’s website.