Political Methodology Research

Markov Models for Multiple States (in .pdf format)

My 2005 PolMeth poster displays work on Markov models for cross-sections of ordered and nominal time series. With applications to human rights abuses and exchange rate regimes, I details some of the relevant considerations in the specification of time series models for discrete phenomena.

Picking a Fight: Disputes, Deterrence, and War-Time Alliances (w/ Curt Signorino)

We specify and test a game theoretic model of triadic interstate conflict. Scholars of international conflict have elaborated a number of explanations for interstate dispute escalation, reciprocation, and diffusion, but have yet to consider these explanations in an interrelated fashion. We assess the impacts of issues, nuclear and conventional military capabilities, similarity of interests, and domestic political regimes on the decisions to escalate, reciprocate, and join a militarized interstate dispute using an iterated version of Signorino's statistical models for extensive form games. The results suggest that potential joiners are of critical importance in understanding the escalation and reciprocation of interstate disputes.

Exploring the Micro Dynamics of Political Participation: Unpacking Trends in Political Participation through Time (in .pdf format)

This paper, joint with Valeria Sinclair-Chapman and Fred Harris explores the micro-level relationships between a twelve item battery of political participation acts from 1973 to 1994. We pay particular attention to differences between blacks and whites and to changes in the relationship between particular acts and latent participation.

Poster Summary (in .pdf format)

This project unifies a series of papers examining the determinants of international human rights, the details of which are listed below. The central arguments of these papers combine critical theoretical analysis with more appropriate estimation techniques for the analysis of international human rights.

Statistical Models for Policy Substitutes

I analyze research design and estimation techniques for policy substitutability with explicit attention to the role of strategy. Foreign policy substitutability requires an examination of the factors that link choices to the broader themes that particular actions represent. This analysis leads to a simultaneous equations framework with a specific view toward testing hypotheses of substitution, complementarity, and no relation among distinct policy choices. I analyze issues concerning the estimation of models in this class analytically and in a series of Monte Carlo analyses of the appropriateness of various techniques. I then extend the theoretical and empirical models to analyze agent-specific substitution patterns in a study of uses of force and economic sanctions by the United States in the post-World War II period.

Some Unpleasant Reduced-Form Arithmetic

This paper analyzes some unpleasant reduced form arithmetic that arises from theoretical substitution or complementarity. First, we construct a theoretical and empirical model of substitutable choices before defining a latent utility specification of substitution and complementarity. We then reduce the class of sensible models to a single specification using a condition similar to Heckman's "Principal Assumption". We then show that models of single policy choices or multivariate probit models that do not properly account for the structural implications of substitutability and complementarity are actually reduced form estimates, thus the interpretation of estimated parameters for these models is only correct in the case of true instruments. In short, we cannot derive inferences about which we are confident without a serious theoretical defense of the empirical specification. We conclude with suggestions for overcoming these problems/

Previously Published Research

You Must Remember This: Dealing with Long Memory in Political Analyses

This paper contributes to the analysis of long-memory time series in political science. In particular, after discussing the underlying issues in generating fractional integration, we investigate techniques for the analysis of fractionally integrated political time series. In particular, we demonstrate an increased probability of spurious relationships with I(d) variables and further demonstrate the strong likelihood that many commonly used political time series are fractionally-integrated. Our simulation results suggest the value of ARFIMA and fractionally-differenced regression based time series models.