PUBL 611: Causal Inference

Tuesdays 4:30-7 pm 

Instructor Jane Arnold Lincove

jlincove@umbc.edu 

OVERVIEW AND COURSE OBJECTIVES

A central goal of evaluation and policy research is estimating the causal effect of an intervention, program, or policy change on behaviors and outcomes. Often a randomized experiment can provide the best causal evidence. However, in many policy contexts experiments are infeasible or unethical, and alternative research designs and analytic strategies can be employed to address causal questions.

This course provides instruction and practice in research designs and analytic strategies that are commonly employed to address causal questions in evaluation and policy research such as propensity score matching, instrumental variables, regression discontinuity, difference-in-differences, and others. We will consider the underlying logic of each method, how and when to apply them to a policy research question, model assumptions and how they can be tested, and data requirements and limitations. Students will learn from illustrative examples and hands-on experience using data and Stata statistical software.

LEARNING OBJECTIVES 

Upon completion of this course, students are expected to be able to: 

·         assess the appropriateness of different research designs and analytic strategies

·         apply different research designs and analytic strategies to estimate specific causal relationships 

·         test the validity of assumptions upon which different research designs and analytic strategies rest

·         interpret results

·         communicate methods and results to a policy audience

PREREQUISITES

Students should have prior experience with regression analysis and statistical software (preferably Stata). PUBL 600 (Research Methods) and PUBL 604 (Statistical Analysis) or similar coursework are required.

PUBL 604:  Statistical Analysis

(taught by Professor Lincove on Wednesdays at 7:10pm)

An introduction to the concepts and methods of descriptive and inferential statistical techniques.  Bi-variate and multi-variate statistical techniques are discussed.  (Also listed as SOCY 604.)

Prerequisites:  PUBL 600 or SOCY 600 or equivalent, college algebra and consent of instructor.