journal article

Assessing Risk Assessment: A Baltic Test

by Michael Johns

Nationalism and Ethnic Politics
vol. 8, iss. 1
2002

Is it possible to accurately gauge the potential of ethnorebellion by minority groups? This paper analyses the process through which two risk assessment models classify the Russian minorities in Latvia and Estonia. After briefly tracing the histories of the two groups, the risk assessment models in Gurr's Peoples Versus States (2000) and Tellis et al's Anticipating Ethnic Conflict (1997) are then examined. The two models use very different criteria for their forecasts, with the Gurr model using the Minorities at Risk dataset, and the Tellis et al model relying on qualitative analysis. This piece attempts to identify where the potential pitfalls in the Gurr model lie, and provides an improved ranking. After examining the data it becomes clear the importance of accurate coding, and the dangers in using a small number of dichotomous variables. The Tellis et al model is used as a check on the Gurr model. The paper concludes with the understanding that risk assessment is by nature inaccurate, but it is a useful tool that needs to be refined.