Here are some slides and accompanying publications on using DAGs in practice. These slides cover about six lecture hours, with exercises.
Most of Dr. Elwert’s courses start with some version of these slides on the first day and then progress to more advanced topics, such as time-varying treatments, instrumental variables, and causal mediation analysis.
The first pdf shows a typical two-day course. The second presentation briefly reviews basic causal and counterfactual concepts. The third presentation presents essential DAG terminology and principles, including d-separation, testable implications, causal sources of association, nonparametric identification, backdoor criterion, and the difference between confounding and selection. The fourth presentation highlights the importance of endogenous selection and gives numerous real examples.
- Typical Short Course
- Introduction to Causality and Counterfactuals
- DAGs: Elements, Interpretation, Identification
- Endogenous Selection Bias: Lots of Examples
The slides draw mostly from two papers, which also include additional material. If you use the slides, please cite these papers.
- Elwert, Felix. 2013. “Graphical Causal Models.” Pp. 245-273 in S. Morgan (ed.), Handbook of Causal Analysis for Social Research. New York: Sage Publications. [pdf]
- Elwert, Felix, and Christopher Winship. Forthcoming. “Endogenous Selection Bias.” [pdf]
Audience: Dr. Elwert’s courses mostly serve graduate students and quantitative researchers in the behavioral and biomedical sciences. Familiarity with regression analysis and basic probability at the level of an applied regression sequence is helpful. Of late, I see an increasing number of applied statisticians and folks from industry (market research and analysts) in my courses, which I really enjoy.