Seminar by Jie Jian

Survival in networks: Cox model with latent blocks

In political science, survival analysis holds the key to answering various intriguing questions. We seek to understand when two countries establish diplomatic relations, the duration of existing commitments between nations, the longevity of wars, and the sustainability of peace between rivals. However, when these events are the interactions (edges) among countries (nodes) on a network, they may not be independent even given the covariates. For instance, when the United States recognized the People's Republic of China in the 1970s, it led to a wave of other countries, particularly those with close relations to the United States, also establishing diplomatic relations with People's Republic of China. We incorporate latent community structures into the Cox proportional hazards model when analyzing the formation of international relations. In this presentation, we will provide a brief overview of survival analysis in political science and the concept of community detection, and explore how the integration of these two ideas yields novel insights into international relations.