Causal linkages of human migration flow networks: A regional analysis

Migration is one of human’s most drastic adaptation strategies against unfavorable conditions. With flows from and to origins and destinations, migration data are necessarily network data. Embedded within network data is interdependency among data points (flows) that renders some traditional statistical analyses, including causal inference techniques, inappropriate. To address this issue, we have developed a novel analysis, combining causal inference techniques with quadratic assignment procedure (QAP) to infer causal relationships from network data and applied it to the datasets that include migration flows and their potential drivers – these include socioeconomic, political, and environmental factors (e.g., flood and drought). We implemented this analysis for the African region data. The preliminary results are reported; the limitations and future work are discussed. We anticipate that this novel method will be applicable to a wide variety of network data in other fields.