New, extended issues of ACCORD’s overviews on conflict events based on ACLED data

Issues covering the year 2017 are now available. The new overviews are extended to show more aspects of the data. The overviews are published by ACCORD and based on data collected by the Armed Conflict & Location Event Data Project (ACLED).

In January 2018, ACLED released data collections for a range of additional countries, with Afghanistan, Iraq and Syria being among them.

The new format of the overviews produced by ACCORD

  1. displays the data on the geographical detail level of the province
  2. complements the sum of fatalities with the sum of events with at least one fatality

Why we chose the province level

ACLED geographically codes the event data with three degrees of precision: If the precise town or nearer area of an event is reported, the highest level of precision is coded. If only the wider region is known, a medium level us coded (using, for instance, the district capital's coordinates for the event when only the district is known). If only the province in which an event took place is known, the lowest precision is known and the province capital's coordinates are used for the event.

How precisely events can be located depends on the country and on reporting by sources. For instance, ACLED's precision levels show that for Afghanistan, only 22% of the collected events are coded with the highest precision level. For about 19% of events only the event's province was reported (the events are therefore coded with the province capital's coordinates).

The distribution of precision levels varies from country to country and within a countries. To avoid the danger of presenting a false semblance of accuracy, ACCORD now only uses the province level for the overviews.

Please note that this method of geo-coding also affects the list of conflict events in the overviews: district or province capitals can be used to represent the actual (but unknown) location of an event.

Why we show two sums related to fatalities

ACLED advises extreme caution when using fatality numbers. "Fatality information is the most biased, and least accurate, part of any conflict report and extreme caution should be employed when using any fatality number to show patterns." (ACLED: Fatalities, 28 December 2017)

ACLED states to use conservative estimates. ACLED uses the context available in the reporting sources to estimate the number of fatalities for those events for which no exact number is reported ("10" for "several" fatalities, "100" if "hundreds" are mentioned, etc.).

This is why ACCORD now complements the sum of fatalities with the sum of events with at least one reported fatality.

You can find ACCORD's overviews on ACLED data with the following link:

https://www.ecoi.net/en/document-search/?srcId[]=10979&documentTypeId[]=11595

Issues covering the first quarter 2018 will be published soon. Update 27 June 2018: The overviews for the first quarter have now been published.

For additional details on these overviews, see our initial blog post.

We thank ACLED for the kind permission to use their datasets.

About this blog