Introducing the Categorically Disaggregated Conflict (CDC) dataset

Conflict researchers have increasingly stressed the importance of distinguishing between different categories of civil conflict, such as ethnic vs non-ethnic. However, the data on conflict categories has remained limited. This paper introduces the Categorically Disaggregated Conflict (CDC) dataset, which categorizes conflicts based on the two most commonly used distinctions, ethnic-vs-non-ethnic and governmental-vs-territorial, resulting in four conflict categories: ethnic governmental, ethnic territorial, non-ethnic governmental and non-ethnic territorial. While not the first of its kind, the CDC contains a number of novel features. Aside from its unique conceptualization of ethnic conflict, the CDC provides coding of the key component variables (language, religion and "race"), allowing users to re-code ethnic/non-ethnic conflicts into several alternative lists (e.g. religious/non-religious). Furthermore, the CDC provides detailed descriptions documenting coding choices for every single conflict, allowing users to track individual coding decisions. To demonstrate the value of the CDC, this paper replicates a recent study by Cederman, Gelditsch and Buhaug, based on the ACD2EPR—the only extant alternative to the CDC. The findings of the replication analysis challenge some of the key conclusions of the original study, substantiating the need for alternative categorically disaggregated datasets.