Guide to 2023 ACLED Column Changes

On 20 March 2023, ACLED made changes to the structure of the dataset by removing three existing columns and adding three new columns. These changes aimed to eliminate redundancies and reduce potential sources of confusion while also introducing more useful variables for analysis. In addition, some existing columns were repositioned within the dataset to improve the readability of the event data.

Three columns – event_id_no_cnty, data_id, and iso3 – have been removed from the dataset. The decision to remove these columns came after careful consideration and internal reviews of their usefulness. Alongside these column removals, ACLED introduced three new variables to facilitate analysis and provide users with additional information about events. The new columns are disorder_type, civilian_targeting, and tags, all of which are described in detail below.

This guide outlines the aforementioned changes to the ACLED dataset and explains how users can adjust to them. Users should review the guide carefully to ensure the updates do not interrupt their workflow.

Column Removals

event_id_no_cnty

ACLED originally created the event_id_no_cnty column so users could sort event IDs within a single country. However, other columns (e.g. event_date or timestamp) can be used to sort events chronologically and do so more accurately. Because of its redundancy and potential inaccuracy when used for chronological sorting, the event_id_no_cnty column was removed from the ACLED dataset. If a user still requires this column, it can easily be reproduced using the event_id_cnty column (see below for details).

data_id

The data_id column contained auto-generated IDs representing each event’s row within the ACLED dataset. Importantly, the values in the data_id column were not static, but changed each time the data were updated. The dynamic nature of the data_id column created confusion, as some users would mistake the values for static, unique IDs. ACLED removed the column from the dataset to reduce this confusion and prevent the use of an incorrect ID variable in the future. To uniquely identify and track events, users should always use the event_id_cnty column, which contains unique IDs that remain static even as the dataset is updated.

iso3

The ACLED dataset contained several columns indicating the country in which an event occurred. Two columns in particular – iso and iso3 – allowed users to easily join ACLED data with external datasets. The iso and iso3 columns, respectively, provided users with a country’s unique numeric code and three-letter code from the International Organization for Standardization (ISO). These columns provided different, but entirely interchangeable, country identifiers. ACLED therefore removed the iso3 column from the dataset in order to eliminate this redundancy and create space for more useful columns.

Column Additions

disorder_type 

The new disorder_type column provides users with a broader classification of event types. This new classification system will allow users to more easily identify and filter relevant event categories, particularly those that are often used in ACLED methodology documentation and analysis. Each event will be assigned a disorder type based on the event_type and sub_event_type columns:

Note that the disorder_type categories are not mutually exclusive, as the excessive force against protesters sub-event type (a subset of the protests event type) is classified under both political violence and demonstrations.

civilian_targeting

The new civilian_targeting column denotes that violence in an event mainly or solely targeted civilians. Without this column, users can only identify civilian targeting events by applying a combination of filters across event_type, sub_event_type, and various actor columns. The civilian_targeting column eliminates the need for such complex filtering, as it will contain one of two values: “Civilian targeting,” which indicates that civilians were targeted during the event, or blank (null), which indicates that ACLED found no reports that civilians were the main or sole target in the event. The lack of a civilian targeting designation does not rule out the possibility that civilians were affected by violence in the event, however (e.g. as ‘collateral damage’ in the context of a battle or explosions/remote violence event). 

tags

ACLED uses a variety of tags to provide standardized information about events. For example, tags may denote the size of a demonstration, whether women were specifically targeted in a violent incident, or connections to a particular political movement (e.g. “stop the steal” in the United States). Tags were previously included in the notes column within square brackets, which could make it difficult to filter events by tag or to extract tag information (e.g. size of a demonstration). The addition of a standalone tags column facilitates the extraction and analysis of tagged events. All tags that were in the notes column were shifted to the new tags column.

Preparing for Column Updates

Anyone using ACLED data downloaded prior to 20 March 2023 will be using an outdated column structure. Those users will need to take certain steps to ensure that the column changes do not interrupt their workflow, regardless of whether they actively use the specific columns that were removed. Even users who do not currently use the removed columns may be affected by changes in column positions. All users should follow the steps below to review and update any scripts and/or Excel files used to interact with ACLED data. The following examples focus specifically on Excel, R, and Python, but the underlying logic can easily be applied to other platforms and programming languages.