| Publication |
2014.
|
| Summary/Abstract |
This essay reviews diffusion as studied in large-N civil war research. In doing so, a number of pitfalls and lacunae are identified. First, the definition of diffusion as a process-whereby internal conflict in one location alters the probability of internal conflict erupting in another location at a later point in time-entails a number of difficulties for empirical modeling. Researching such a process involves an attempt to study a phenomenon that, in essence, is unobservable. It also creates difficulties in identifying relevant units of analysis, because the process involves at least two units. Second, diffusion is customarily identified based on correlations within a spatial and temporal proximity. Classifying it in this way risks simultaneously over- and underestimating cases of diffusion, which in turn generates uncertainty regarding the main determinants of diffusion. With these observations in mind, this essay ends with a word of caution for policymakers, with relevance extending beyond diffusion of civil war.
|