Adaptive Mixture Modelling Metropolis Methods for Bayesian Analysis of Non-linear State-Space Models.

We describe a strategy for Markov chain Monte Carlo analysis of non-linear, non-Gaussian state-space models involving batch analysis for inference on dynamic, latent state variables and fixed model parameters. The key innovation is a Metropolis-Hastings method for the time series of state variables based on sequential approximation of filtering and smoothing densities using normal mixtures. These mixtures are propagated through the non-linearities using an accurate, local mixture approximation method, and we use a regenerating procedure to deal with potential degeneracy of mixture components. This provides accurate, direct approximations to sequential filtering and retrospective smoothing distributions, and hence a useful construction of global Metropolis proposal distributions for simulation of posteriors for the set of states. This analysis is embedded within a Gibbs sampler to include uncertain fixed parameters. We give an example motivated by an application in systems biology. Supplemental materials provide an example based on a stochastic volatility model as well as MATLAB code.

Essays on State-Building and Sectarian Violence

\abstract

This dissertation seeks to explain the role of governmental and non-governmental actors in increasing/reducing the emergence of intergroup conflict after war, when group differences have been a salient aspect of group mobilization. This question emerges from several interrelated branches of scholarship on self-enforcing institutions and power-sharing arrangements, group fragmentation and demographic change, collective mobilization for collectively-targeted violence, and conflict termination and the post-conflict quality of peace. This question is investigated through quantitative analyses performed at the sub-national, national, and cross-national level on the effect of elite competition on the likelihood of violence committed on the basis of group difference after war. These quantitative analyses are each accompanied by qualitative, case study analyses drawn from the American Reconstruction South, Iraq, and Cote d'Ivoire that illustrate and clarify the mechanisms evaluated through quantitative analysis.

Shared findings suggest the correlation of reduced political competition with the increased likelihood of violence committed on the basis of group difference. Separate findings shed light on how covariates related to control over rent extraction and armed forces, decentralization, and citizenship can lead to a reduction in violence. However, these same quantitative analyses and case study analysis suggest that the control of the state can be perceived as a threat after the end of conflict. Further, together these findings suggest the political nature of violence committed on the basis of group difference as opposed to ethnic identity or resource scarcity alone.

Together, these combined analyses shed light on how and why political identities are formed and mobilized for the purpose of committing political violence after war. In this sense, they shed light on the factors that constrain post-conflict violence in deeply divided societies, and contribute to relevant academic, policy, and normative questions.