June 14-17
Instructors: Randall Calvert and Andrew Martin (both of Washington University in St. Louis)

The Foundations seminar aims to provide tools with which students can develop their own statistical models to test predictions derived from formal theories. Although most participants in the Summer Institute know the basics of rational choice theory and statistical analysis, it will be helpful to construct those techniques in a manner that facilitates statistical testing of rational choice models.

Standard introductions to rational choice theory often imply that such theories generate only point predictions, unsuitable for testing. The Foundations seminar presents important varieties of rational choice models, specifically spatial voting models and non-cooperative game theory, in a form that emphasizes the techniques by which these models can be used to generate testable implications through comparative statics analysis and the analysis of parameter variations across a population.

Likewise, basic courses in statistical methods often give scant coverage to techniques especially suited to the testing of predictions generated from formal models. Accordingly, Foundations covers techniques fundamental to the Summer Institute's advanced seminars: maximum likelihood estimation, Bayesian inference, model specification for comparative statics predictions, model comparison, and simulation. It also introduces software to be used in the subsequent advanced seminars.

June 18-19
Instructor: Mark Fey (University of Rochester)

Much of the political science literature suffers from a disconnect between theory and the statistical techniques used to test or analyze theory. During this module, we will examine methods for explicity linking theory and statistical analysis, especially in a strategic context. Topic covered in this section will include random utility models (RUM), selection models, quantal response equilibrium (QRE), and structural econometric models of strategic interaction, including signaling.

For over fifty years, the Euclidean spatial voting model has informed a scholarship on voting and decision-making in legislatures, courts and other deliberative bodies. It is no exaggeration to say that a great deal about what is known about voting in parliaments relies on the logic of the spatial model. The public opinion and political behavior literature also suggests that the spatial voting model is an excellent characterization of voter decision-making in mass publics. The spatial model has also spawned a tremendous amount of theoretical development, ranging from instability results in multi-dimensional settings (e.g., McKelvey 1979, Schofield 1977) to theories about “gridlock” in the U.S. Congress (e.g., Krehbiel 1998).

This course seeks to better connect intense theoretical interest in spatial voting models with data analysis. How does one analyze the recorded votes of a deliberative body (“roll call” data) and estimate ideal points? How does one take a battery of issue questions on a survey and summarize the underlying issue space? Given estimates of ideal points in such a space, how does one use them in other models? How does one go about computing equilibrium behavior from spatial models? In this course, we present methodological tools that will allow students to (a) operationalize the spatial model in their own research; (b) use the spatial model in other statistical models of behavior; and (c) use computational approaches to compute equilibrium predictions of various sorts of formal models.

June 24-26, 28
Instructors: Rick Wilson (Rice University), Stephen Haptonstahl (University of California-Davis)

As in other sciences, the development of rigorous, deductive theory leads to hypotheses that demand testing. Laboratory experiments are intended to provide the most controlled tests of these hypotheses. In political science, laboratory experiments play a critical role since so many of the theories presume knowledge of preferences which are typically unknown in the field. Political science experiments typically guarantee subjects differing financial rewards for different outcomes, thereby inducing known preference rankings over those outcomes. Typically, different treatments will be identical except for one key theoretical variable, which may be the institutional voting rule, information conditions, communication possibilities, or preferences. With random assignment of subjects to treatments, differences across treatments may be confidently attributed to the treatment variable.

Topics covered will include voting experiments, public good experiments, tests of non-cooperative bargaining theory, experiments on information, and recent innovations in political science experimentation. We will discuss links between theory and experiment, experimental design, the role of pilot experiments, experimental technique, data gathering and data analysis. Students will have a chance to participate both as subjects and as observers, and will be asked to design an experiment. Readings will be reports of experiments in political science and economics journals.

Professor Rick Wilson of Rice University is the primary instructor of this course. In addition to historical research on institutions of American politics, Rick has done some of the most innovative and powerful experiments in the discipline. He has designed and administered experiments on strategic voting, agendas and agenda costs, and repeated prisoners' dilemma games, coordination games, and ultimatum games, among others. He has received numerous National Science Foundation grants in support of his experimental research.

Professor Stephen Haptonstahl of UC Davis is a recent graduate of Washington University. A former EITM participant, Steve uses experiments, formal theory, Bayesian statistics and computational modeling to understand how institutional structure affects the flow and aggregation of information in bureaucracies, legislatures, and courts.

June 29-July 1
Instructor: Jason Roberts (University of North Carolina)
Guest Instructor: Alan Wiseman (Ohio State University)

In recent years, theoretical advances concerning legislative institutions, legislative parties, and the individual behavior of legislators have generated many methodological challenges. A central concern of the field is the development of appropriate tests of theories about the choice of formal and informal institutions. Closely related are theories of individual behavior in a strategic context. The seminar is designed to give intensive consideration of several problems at the intersection of theory and method.

The instructor for the seminar is Jason Roberts. Professor Roberts is an assistant professor specializing in American political institutions, with an emphasis on the U.S. Congress. He earned his B.S. in Political Science from the University of North Alabama (1998), his M.A. in Political Science from Purdue University (2000), and his Ph.D. from Washington University in St. Louis (2005). Prior to joining the faculty at UNC, Professor Roberts was an assistant professor of Political Science and Law at the University of Minnesota. His research interests include parties and procedures in the U.S. Congress and congressional elections. He is currently working on a project that explores the role of ballot type on the competitiveness of congressional elections in the United States.

June 29-July 1
Instructor: Robert W. Walker (Washington University in St. Louis)

Formal models have become more common in the political economy of international relations, but a notable disconnect remains between the theoretical and empirical worlds. In this seminar, we will discuss the unique challenges posed by data that are observed across both space and time through the lens of political economy models in international trade, finance, and conflict. We will construct, examine, and/or replicate dynamic models and models of transitions, measurement models, multiple imputation and other advanced techniques in cutting-edge applications of cross-national research using R.

Professor Robert Walker is an Assistant Professor in the Department of Political Science and the Center for Applied Statistics at Washington University in St. Louis specializing in international relations, international political economy, and political methodology.  He is a 2005 Ph. D. in political science (studying under Randy Stone and Curt Signorino) from the University of Rochester and came to Washington University after stints in the Department of Political Science at Texas A & M University, the Department of Government at Dartmouth College and the Institute for Quantitative Social Science at Harvard University.  He teaches courses in international relations and political methodology in the Department of Political Science and courses in applied statistics and econometrics for the Center for Applied Statistics.