A Multilevel Bayesian Framework to Analyze Climate-Fueled Migration and Conflict
Abstract
Do climate conditions and extreme events fuel conflict and migration? This question is of growing interest given increasingly dire climate change projections. It is commonly addressed in causal studies that leverage natural experiments by using a multivariate linear regression model with fixed effects. We show that in the climate-migration-conflict nexus, the features of the data generating process and the implicit prediction motivation can lead to a substantial departure from the assumptions of the typical linear reduced form model,challenging the reliability of inferences. We propose a unifying hierarchical Bayesian framework for inferences from the same natural experiments, and describe its benefits for internal and external validity and for analyzing the heterogeneity in response to climate. Using a conflict dataset representative of the literature, we illustrate the misleading results that can ensue from the typical approach and the advantages of the hierarchical Bayesian framework.
- Publication:
-
AGU Fall Meeting Abstracts
- Pub Date:
- December 2022
- Bibcode:
- 2022AGUFMGC32D..07P