GOVT 6039 - [Advanced Topics in Quantitative Empirical Methodology]

Fall. Next Offered: 2019-2020. 4 credits. Letter grades only.

Permission of instructor required.

B. Corrigan.

We will gain experience with advanced statistical methods for social scientists–those incorporating complex statistical assumptions as well as those that relate to causality. Students will apply these methods in a series of labs with a heavy emphasis on computation, with the option wherever feasible to work with R, Stata, or (under self-direction) in some other preferred modeling environment. The first half of the course details maximum likelihood and Bayesian estimation frameworks as flexible strategies for tackling any data generating process. Two labs examine ordinal/multinomial, unrestricted count, and duration data, and working with a “custom” model. The third lab explores how to use a multi-level model to represent nonsphericity and heterogeneity. The second half of the course introduces a graphical grammar for representing causal dependence, with three additional labs oriented towards IV, matching/weighting, and multi-stage causal effects estimation.