Militarized interstate disputes
) are conflicts between states
that do not involve a full-scale war
. These include any conflicts in which one or more states threaten, display, or use force against one or more other states. They can vary in intensity from threats of force to actual combat short of war.
A MID is composed of a sequence of related militarized incidents, all but the first being an outgrowth of or response to a previous militarized incident.
An initiator of a war need not necessarily be the same as the initiator of a preceding MID, since a MID can be started by a show of force, whereas the initiator of a war begins the actual combat.
Under this definition, over 2400 MIDs have been identified from 1816 to 2014 in the Correlates of War
Some studies have characterized the dataset as flawed. A 2012 study found that the dataset "often coded incorrectly" the outcomes of disputes where threats of force were issued, with implications for research on the democratic peace
and audience costs
A 2017 study found that the coding in the dataset was deeply flawed with significant effects on the findings of studies that relied on the dataset:
After strictly applying MID coding rules, we recommend dropping 251 cases (or over 10% of the dataset), as either we were unable to find a militarized incident in the historical record or the dispute appeared elsewhere in the data. We found evidence linking 75 disputes to other cases, and we could not identify 19 cases in the historical record. Among the remaining disputes, we recommend major changes (changes in dispute year, fatality level, and participants) in 234 disputes and minor changes in 1,009 disputes. Though we identified several systematic problems with the original coding effort, we also find that these problems do not affect current understandings of what predicts the onset of interstate conflict. However, estimates in our replications of three recent studies of dispute escalation, dispute duration, and dispute reciprocation all witness substantial changes when using corrected data—to the point of reversing previous conclusions in some cases.