Diffusion, co-evolution and strategic interdependence in comparative and international politics: New spatial econometric and event history approaches.

摘要

Interdependence is ubiquitous across theories of democratization. For example, the level of democracy in one country might be dependent on its level in other countries; the timing of democratization might be related to the survival of existing democracies. In contrast, much of the empirical literature on democratization has modeled the level of democracy and the timing of regime transitions as if all the observational units and events were independent.;The essays in this thesis explore three sources of interdependence in the study of democratization: the first concerns the causal connection between the emergence and collapse of democracy; the second is the diffusion of political regimes across countries; The third the reinforcement and local convergence of political regimes across countries and over time. Although each type of interdependence raises a unique set of methodological challenges, the emergence of "feedback loops defines the common mathematical characteristics of these difficulties. A feedback loop is formed when a change in an outcome (e.g., the level of democracy) influences the outcomes of other units, which in turn comes back to affect the outcome that experienced the original shock. Both the inter-event (e.g., the emergence and breakdown of democracies) and inter-unit (e.g., diffusion and reinforcement) dependencies generate these recursive flows of effects across observational units.;In this thesis, I develop two systems of equations (SEQ) models to account for the three sources of interdependence. In all the models presented here, I take a so-called "substantive" approach, rather than a "nuisance" approach, in order to model the theoretically-informed structure of dependence. In Essay 1, I develop a multivariate event history model in order to incorporate the two-way causal relationships between the emergence and breakdown of democracies. In Essay 2, I develop a multivariate event history model for data with right- censored observations, building on the model developed in Essay 1. Essay 3 introduces a new spatial econometric model, which estimates the existence and the strength of both regime diffusion and regime reinforcement, using time-series cross-sectional data of democracy levels.