The Ethnic Power Relations Data: A Critique Brenton D. Peterson *† March 8, 2016 Abstract The Ethnic Power Relations (EPR) dataset Cederman, Wimmer and Min (2010); Wim- mer, Cederman and Min (2009a) provides researchers broad cross-country data on eth- nic groups’ access to political power within their states. Since its inception, the data have been used extensively in the literature on intrastate wars, but they have also found use in studies of ethnic politics, elections and a variety of other topics. I argue that the EPR data is flawed for three primary reasons. First, ethnic groups in the data are often aggregated in ways that would be unrecognizable to country experts and group members themselves. Second, politically irrelevant ethnic groups are ex- cluded from the data, but are often as politicized as those included. Third, groups’ access to political power is often coded in ways that demonstrably contradict the cod- ing rules themselves and do not match the empirical reality of ethnic power relations over time. I discuss these critiques in the context of the Kenyan data included in the EPR dataset, providing evidence for my claims from research in linguistics, history and anthropology, electoral results, new data on the ethnic background of Kenya’s Cabinet ministers since independence, original survey data, and experimental behav- ioral measures of ethnic identification. The errors and lack of conceptual clarity in the EPR have significant consequences for studies using the data, at times introducing systematic biases and clouding our understanding of ethnic relations. * Ph.D. Candidate, Department of Politics, University of Virginia, Charlottesville, VA 22902. † I owe a great deal of thanks to my research assistants, Adan Galma Dabasso, Donald Kariuki, Maina Kinyua, Samuel Nyabwari and Umuro Robah, who endured my endless questions and implemented my surveys masterfully. Thanks to Mike Poznansky for his comments and Charles Hornsby for generously sharing his data.