Topics in Political Methodology

Catalog description

This course examines a range of statistical models widely used in political science that generalize from linear normal regression. PL SC 504 Topics in Political Methodology (3) This is an elective course in statistical methods designed to meet the particular needs of students in the political science Ph. D. curriculum. PL SC 504 is tailored to focus on the specific issues that arise in the types of data found in political science applications. Students are expected to have completed the three required foundational courses in political methodology or their equivalents. This course examines a range of regression-like models widely used in empirical political science. Its core focus is on maximum likelihood estimation of models for various kinds of limited-dependent and qualitative response variables. Specific models covered are widely used in political science today, including binary logit and probit, multinomial logit and probit, ordered logit and probit, and Poisson regression models. Additional topics include models for time-to-event (survival) data, panel data and time-series cross-sectional analysis, item response theory, multi-level models, and methods for causal inference using observational data. Students will apply these models in a series of homework assignments, a replication project, and a final exam. Empirical political scientists must have familiarity with these models; these techniques represent a minimal level of statistical competence necessary for those seeking to do advanced quantitative analysis in the political science. The material in this course is technical, but students will be given an intuitive rationale for each model. Weekly homework assignments will be based on data from published research in political science.