Partition as a Solution to Ethnic War: An Empirical Critique of the Theoretical Literature

Partitioning a state after civil war ens has been debated by political scientists and policy makers. It is believed that partition is the most favorable solution to a civil war because the fighting groups will not have tol ive together in the future, hence probability of violence will be eliminated. In this article, Sambanis rejects this argument and empirically proves that partitioning is not a better solution for civil war torn state than other alternative solutions. 
Sambanis focuses on the cases of partition that took place after a civil war, not partitions that took place peacefully. He tests three hypotheses: partitions facilitate postwar democracy, partitions prevent civil war recurrence, partitions reduce low-level violence after the war. He compiled a cross-sectional data set that includes all civil wars since 1944. His data set includes 125 civil wars of which 21 ended with partition. 
He finds that “as ethnic heterogeneity increases, the probability of a partition decreases significantly”. For Sambanis this finding suggests that it is very difficult for extremely diverse societies to coordinate and win a civil war. He finds a positive correlation between the size of ethnic group and probability of partition. This is to say that if the fighting ethnic group is large, then they are more likely to overcome coordination problem and defend their terrirtory. He also finds positive and significant correlation between level of violence and probability of partition. This suggests that more violent conflicts are more likely to end with partition. Another interesting finding is that identity wars are more likely to end with partition than ideological wars. 
With regard to the relationship between partition and postwar democracy, he finds support for this hypothesis. However, he warns us that this effect may not be the result of democratization but due to declining democracy during the prewar period. Moreover, Sambanis notes that it is very difficult to collect data about democracy level of new states that are not internationally recognized. Therefore, we must be cautious about the findings about democracy.  
His findings show that partition does not prevent war recurrence. His findings imply that “nonethnic partitions are more stable and peaceful than ethnic partitions”. He does not find strong evidence to support the argument that partition prevents low-level violence after civil war. 
In conclusion, this is a decent test of arguments proposed by advocates of partition. Sambanis shows that partition is not as clean and easy solution to civil wars as it is thought to be.