DESIGN & ANALYSIS OF QUASI-EXPERIMENTS FOR CAUSAL INFERENCE

EDPSYCH 762

Classical experimental designs and their application to educational research, factorial treatment arrangements, confounding, repeated measures design, and related topics.

POLISCI/PUBAFFR 871

Compares the conceptual, statistical, and ethical issues of experimental, quasi-experimental and non-experimental designs for program evaluation. Definitions of outcomes, sample size issues, statistical biases in measuring causal effects of programs, and the reliability of findings will be emphasized using case studies selected from current public programs.

POPHLTH 805

Focuses on the use of viewpoints and design/analytical tools to render possible the estimation of causal effects in epidemiologic studies. Students learn about the rationale and use of study designs/analytic tools that build upon but are substantially different from the most common approaches used in epidemiologic research (experimental studies, case-control studies, and cohort studies).

ECON 707

Most empirical work in economics focuses on estimation of causal effects, including causal effects of an economic policy, a program, or a person's decision. Discussion of the basic causality issue and contemporary methods in econometrics for identifying and estimating causal effects.

PUBAFFR 281

Introduces the key conceptual and methodological tools used in public program evaluation, with an emphasis on understanding the forces that shape health and disease as well as various policy solutions. Introduces the Potential Outcomes Framework, also known as the Rubin Causal Model. Establishes the distinction between causation from correlation using counterfactual thinking. Explores a wide variety of experimental and quasi-experimental research methods used to estimate causal effects, including randomized experiments, regression, matching, instrumental variables, fixed effects, regression discontinuity, difference-in-differences, and synthetic control. Many of the causal inference methods that we discuss require statistical and computational training in order to implement. Focuses on the nontechnical conceptual, theoretical, and intuitive underpinnings of these methods that are most salient to policymakers.