Using Network Analysis to Explore the Politics of Disasters in Urban Nepal

• In Event: Hazards, Disasters, and Their Aftermath

Abstract

When exploring latent dimensions and clusters using data-driven methods, it is critical to be precise about the nature of that which one is exploring and the assumptions of a chosen method. I argue that network-based community detection is a method worth considering when one does not want to make strong assumptions of underlying DGPs. In this type of analysis, the data are conceptualized as a network in which variables are nodes and the ties between them are a user-defined measure of distance. This method presents a middle ground between the two poles of the ‘black boxes’ of many cluster analysis implementations and the strong assumptions of latent factor or class analyses and data reduction techniques on the other. I explore this technique through a Monte Carlo simulation study and with novel household-level survey data (n = 1800) collected in urban Nepal and a theory of disaster management and reconstruction as public good in the opinions of Nepali urban residents. This paper thus offers network-based community detection as a methodological contribution, the understanding of disaster management and post-disaster reconstruction as a public good as seen in urban Nepal as a theoretical contribution, and a novel dataset of opinion data as an empirical contribution.

Author

• Shana Scogin, University of Pennsylvania