Terrorism and land use in agriculture: The case of Boko Haram in Nigeria - a replication attempt of the paper by Adelaja & George (2019)

This replication study tests the repeatability of the study conducted by Adelaja and George (2019a) on the relation between Boko Haram related events of armed conflict and agricultural land use decisions. For the computational reproduction of the original results, we used the same datasets and the same methodological approach. Yet, our findings substantially deviate from the results in the original study. To test the robustness of our own results, we applied comprehensive robustness checks, such as using a different measure for armed conflict, including an additional survey wave of the GHS panel data, and applying two different timewise merging procedures. The results of the robustness checks were very sensitive to changes in the underlying data. Hence, we question the internal validity of the study's findings and argue that there is most likely a more general mismatch between the data, the methodology and the research questions. The bottom-line of this study is not that there are computable differences between Adelaja and George (2019a) and our estimations, but rather that caution should be taken in deriving policy implications from Adelaja and George (2019a) as their results lack internal validity.