When will updated MID data be available?

The coding for MID 5 is ongoing. We anticipate that the updated data will be available in late summer 2020.

What is the difference between a Militarized Interstate Incident (MII) and a MID?

An incident is a single threat, show, or use of force. Incidents that involve the same participants and the same issue are grouped into the same MID, unless there is a gap of at least six months between them.

What is the difference between originators and joiners?

Originators are involved in a MID on the first day. Joiners join sometime after the first day. Only multilateral MIDs can have joiners. The Side A and B codings are less meaningful for joiners because a state that joins Side A after the first day cannot actually be considered to have initiated the MID. Therefore, some people prefer to drop joiners in studies of MID initiation (see recommendations by Bennett and Stam).

What is the difference between Side A and Side B?

Side A consists of the state that took the first militarized action that initiated the MID and any other states that joined the MID on the same side as that state. Side B is the state that was targeted in the action that initiated the MID and any other states that joined the MID on that state’s side.

Many research designs treat Side A as the “aggressor.” Although this may be a reasonable assumption in the aggregate, it is not true in all cases for several reasons. First, the first state to take military action is not always the state responsible for instigating the dispute. Second, as noted above, the Side A/B coding is less meaningful for joiners. Third, in the case of a mutual clash (HiAct coding 17), Side A is assigned arbitrarily.

Is there incident-level data available prior to 1993?

Sort of. The Escalatory Incidents Dataset (MID v.2.1EE) is hosted by the Correlates of War Project. As the name implies, this dataset contains only incidents that escalated the hostility level of a MID. The rules used for deciding which incidents to include in this dataset are not necessarily intuitive, so it is important to read the codebook carefully before using it.

How can I obtain dyadic MID data?

Please see the first entry on our Related Datasets page.

Isn’t the MID dataset full of minor disputes, such as fishing disputes, that we should not actually care about?

The MID dataset does contain many minor disputes, but we consider this to be a feature rather than a bug. Including minor disputes enables us to study how disputes escalate as well as get a better sense of low-level military tensions that indicate strained relations between countries. Regarding the controversy over “fishing boat MIDs” specifically, the existence of these MIDs often indicates the presence of a salient territorial dispute between states, in which one state has chosen to signal its resolve by harassing or impounding private vessels from another state. Users who are primarily interested in more serious disputes have several options, such as dropping disputes below a certain threshold for hostility, fatalities, or duration.

Does the MID dataset contain errors?

Yes, like any data set, the MID data contain errors. Early iterations of the MID project did not keep detailed records of the source material used to code disputes, due to limited storage capacity and different norms regarding best practices. This makes it difficult to verify the accuracy of some coding decisions. We have recently conducted an analysis to verify some of these earlier decisions and believe that any remaining “noise” in the data is unlikely to affect most results. For our response to suggested corrections by Gibler, Miller, and Little, please see our Related Datasets page and our ISQ research note.

Why does the MID dataset contain so few threats?

We have a very high bar for coding threats. A threat is only codable if both the military action to be taken and the specific conditions under which it would be taken are stated explicitly. For an alternative coding procedure, see the Militarized Compellent Threat Data.