You are here: Home / Data Sets / COW War Data, 1816 - 2007 (v4.0) / Dyadic Inter-State War Dataset

Dyadic Inter-State War Dataset

The dyadic war dataset is built and fully compatible with the dyadic MID dataset (Maoz et al. 2019). It provides dyadic records of interstate wars over the period of 1816-2010, with a dyad-year unit of analysis. The dataset contains several innovations and changes compared to the war or MID datasets. Specifically,

  • It applies the definition of war and war participation to a dyadic setting. A dyadic war is defined as a series of sustained and lethal battles between the armed forces of two states that is a part or the whole of an interstate war.
  • For a dyadic war to be present both states must meet one or both of the following conditions: (1) the state must suffer 100 or more battle-related deaths in the battles with the other state, or (2) the state must deploy 1,000 or more troops in battles against the other state.
  • War outbreak and war termination is defined in terms of battles, not in terms of declarations of war. A dyadic war starts on the start date of the first battle between the armed forces of the states making up the dyad. Accordingly, a dyadic ends on the last day of the last battle between the armed forces of the dyad.
  • Each war is identified both in terms of its own dyadic code, and in terms of the dyadic MID of which it is a part.
  • The data have been cleaned and each dyad that was part of a multilateral war was researched separately to determine whether it met the conditions stated above. The data also reflect a systematic comparison to the COW war data and another interstate war dataset (Reiter, Stam, and Horowitz 2014).
  • To the extent possible, fatality data were broken down by individual state and dyad. However, in many cases involving a state that fights several opponents in the context of a multilateral war, such a breakdown is not possible. We provide “best” estimates of this breakdown.
  • The data are fully compatible with the dyadic MID dataset. The dyadic war dataset is a proper subset of the dyadic MID dataset: every war in the dyadic MID dataset is part of a dyadic MID (but not vice versa).
  • The outcome of the war is adjusted for the specific dyad. The winner in a dyadic war may have a different designation than the winning side in the COW war dataset.
  • We assign specific roles to joiners in multilateral war: a joiner on the target side in a multilateral war may be the initiator against a state on the initiator’s side in the multilateral war.


Please cite: Zeev Maoz, Paul L. Johnson, Jasper Kaplan, Fiona Ogunkoya, and Aaron Shreve 2019. The Dyadic Militarized Interstate Disputes (MIDs) Dataset Version 3.0: Logic, Characteristics, and Comparisons to Alternative Datasets, Journal of Conflict Resolution (forthcoming).