OSCE monitors and publishes nearly on a daily bases reports on the implementation of the Minsk 2 agreements. Or to what degree the ceasefire is violated. Graphics that are based on those data and about how they were made.
OSCE’s coverage/reporting of all ceasefire is limited. Due to lack of personal or restricted access. Therefore data must be considered as sample. – It’s not a accurate protocoll of all caesefire violations. It might be representative in respect of relative changes over time (weeks), not that much in respect of absolute numbers.
– Event (ceasefire violation) is coded as ‘1’ if its location is known (‘Location of event’).
– Date and geographic ‘SMM position’ of reported incident (who controls the territory).
– Some observation posts report several incidents on different locations per day (high military activity around a observation post). Each of those incidents, reported by OSCE are coded as ‘1’. If there are several cases of the same category on the same day and location, they are added up.
– Direction of fire’: in, out, in/out; incidents with artillery are coded separately. N/A (with or without artillery).
– If a case includes ‘in’ and ‘N/A’, i’t is coded as ‘in’. – The more clear information overrules.
– If various types of weapons are reported in a case (SALW, tanks, machine guns) and artillery, it’s coded as artillery. This rule applies to in, out and in/out.
– If OSCE reports that the red team has incoming machine gun fire, it is assumed that the blue team is shooting. Such a situation would be considered (and coded) as a violation done by the blue team.
– It is not coded if a incident was very intense or just a short fire. The only difference made is if there was artillery or not.
Data from OSCE reports might be a good sample for measuring dynamics at the frontline. It’s experimental, feedback welcome. Maybe coding is too rough or bias in data too big.
In any case, it’s a colored summary of OSCE reports.
Spreadsheet for replication or further research: