How can micro-level research on the dynamics of civil war be improved? This chapter from Order, Conflict, Violence analyses micro-level studies of civil war, identifying a mismatch between their micro-level empirical focus and their macro-level conceptual and theoretical focus. This mismatch leads to difficulties that introduce bias: problematic proxies resulting from concept conflation, observational equivalence, endogeneity, overaggregated variables, and the omission of significant variables. Engaging with cases, careful and detailed collection of fine-grained data, and thorough theorisation are therefore needed.
A new research programme has emerged on the microdynamics of civil war. This calls for the systematic collection of data at the subnational level and its sophisticated analysis. Micro-level analysis offers the possibility of improving data quality, testing microfoundations and causal mechanisms, maximising the fit between concepts and data, and controlling for extraneous variables.
However, weaknesses in studies on the microdynamics of civil war need to be addressed. The research literature reveals five recurrent flaws specifically relating to micro-level civil war studies. Examples of these problems are found by analysing three micro-level studies on the civil war in Nepal.
- Problematic proxies are evident where the idea of war/violence is conflated with that of violence. For example, using fatality numbers as an indication of the intensity of conflict is erroneous, as a lack of violence does not always indicate a lack of hostilities.
- Observational equivalence occurs where poverty is correlated to high fatalities. This correlation may mask other explanations of high fatalities that relate to social opportunity or geographic considerations.
- Endogeneity is a significant problem given the difficulty of comparing statistics before, during, and after the conflict. Additionally, conflict itself negatively affects data collection and accurate measurement.
- Overaggregation takes the form of coding issues at the unit level of analysis and overaggregated numbers related to violence.
- Omitted variable bias occurs where the geographic variable of territorial control is omitted.
Policymakers must understand how these flaws impact on the findings of micro-level studies. Researchers should, for example, strive to correct these flaws through:
- Collecting disaggregated data despite the fact that it is time consuming and costly. Disaggregated data can prevent researchers from omitting the crucial control variable; conversations with study participants can reveal unaccounted for, but significant factors.
- Preventing the significant omission of geographic concerns by including a variable for territorial control. When it comes to coding territorial control there is no easy alternative to either direct and careful data collection using all available sources, or prior coding by the insurgents or counterinsurgents themselves.
