We analyze how the determinants of the civil conflict in Colombia at the municipal level respond to changes in the empirical strategy. We estimate several econometric models that vary in three dimensions: 1) the data set, 2) the measure of conflict, and 3) the estimation method. We find substantial differences in the signs and statistical significance of the marginal effects in response to using a different data set, changing the definition of conflict, and employing different empirical methodologies.
