Matching and Weighting for Causal Inference
A 4-Day Remote Seminar Taught by
Stephen Vaisey, Ph.D.
To see a sample of the course materials, click here.
This course offers an in-depth introduction to matching and weighting methods using the R package. Matching and weighting are quasi-experimental techniques for estimating causal effects from observational data using the potential outcomes or counterfactual framework. They are often (but not always) based on propensity scores. These techniques are now widely used in the social sciences, health sciences, management and public policy.
Starting July 27, we are offering this seminar as a 4-day synchronous*, remote workshop. Each day will consist of a 3-hour live lecture held via the free video-conferencing software Zoom. You are encouraged to join the lecture live, but will have the opportunity to view the recorded session later in the day if you are unable to attend at the scheduled time.
Each lecture session will conclude with a hands-on exercise reviewing the content covered, to be completed on your own. An additional lab session will be held Tuesday and Thursday afternoons, where you can review the exercise results with the instructor and ask any questions.
*We understand that scheduling is difficult during this unpredictable time. If you prefer, you may take all or part of the course asynchronously. The video recordings will be made available within 24 hours of each session and will be accessible for four weeks after the seminar, meaning that you will get all of the class content and discussions even if you cannot participate synchronously.
Closed captioning is available for all live and recorded sessions.
MORE DETAILS ABOUT THE COURSE CONTENT
Researchers use matching and weighting to identify the causal effect of a treatment on an outcome — such as the effect of a college education on earnings, the effect of divorce on child outcomes, or the effect of a training program on employee productivity — when assignment to the treatment is not random. A major advantage of these techniques over standard regression methods is that they can easily produce different estimates of causal effects for subjects who are likely to receive the treatment and for those who are unlikely to receive it, a distinction that is especially important for policy work.
This seminar will guide you from simple exact matching to recent developments like coarsened exact matching, entropy balancing, and matching frontier techniques that show how effects vary across the full range of possible match quality. We will also show how to integrate matching with regression to create “doubly robust” estimates of causal effects. You will get practical experience by working through exercises from the social and health sciences.
Though the seminar will focus on hands-on understanding, we will also use causal graphs (directed acyclic graphs or DAGs) to look more deeply into the assumptions required to achieve unbiased estimates. You will learn how these graphs can be used in your own research.
This is a hands-on course.
The instructor will use R with RStudio to demonstrate the techniques. To participate in the hands-on exercises, you are strongly encouraged to have your computer available with R and RStudio already installed. You need only a basic familiarity with R to get the full value of the course.
If you’d like to take this course but are concerned that you don’t know enough R, there are excellent on-line resources for learning the basics. Here are our recommendations.
WHO SHOULD Register?
This course is for anyone who wants to learn to apply matching and weighting to observational data to improve their causal inferences. You should have a basic foundation in linear and logistic regression.
1. Theoretical background
a. the experimental ideal
b. the potential outcomes framework
c. defining different treatment effects
d. conditional independence assumption
2. Exact Matching
a. the goal of exact matching
b. implementation in MatchIt
c. getting treatment effect estimates with WLS
d. feasible estimates
3. Propensity-score methods
a. theory and estimation
b. propensity-score stratification in MatchIt
c. propensity score matching in MatchIt
d. assessing covariate balance
– mean differences
– Kolmogorov-Smirnov distances
– using cobalt for plotting balance diagnostics
e. matching variations: replacement, calipers, etc.
f. bootstrapping for standard errors
g. propensity-score weighting
– IPTW using WeightIt
– covariate balancing propensity scores using WeightIt
4. Non-parametric methods
a. Mahalanobis distance matching in MatchIt
b. entropy balancing in WeightIt
5. Parametric regression with preprocessed data
a. double robustness property
b. estimation with MatchIt/WeightIt + lm
6. Advanced topics (briefly as time permits)
a. multinomial treatments in WeightIt
b. continuous treatments in WeightIt
c. synthetic control method (difference-in-differences + weighting)
“I really enjoyed Dr. Vaisey’s course. He has a knack for great and concise explanations. I appreciate how he focused on the meat of the content and made sure he conveyed the most salient information. I feel like I got a lot out of the 3 days. It was the springboard I needed to take the next leap on my own.”
Paul Bernhard, Department of Veterans Affairs
“I recommend this stimulating course. Lectures were very easy to follow and Stephen Vaisey was a brilliant teacher. The course had high-quality content.”
Jaana Minkkine, Tampere University
“This course was superb for getting a feel of what propensity models are and how to use them across many different applications. There are a lot of variations covered and the caveats of all are thoroughly discussed. Steve is a fantastic, enthusiastic, and engaging speaker. I would definitely recommend.”
Emily Bucholz, Boston Children’s Hospital
“Having worked in various research capacities following my Ph.D. training, this course provided the most robust instruction on causal inference and matching/weighting methodology that I’ve experienced to date. Dr. Vaisey provides practical applications for both parametric and non-parametric techniques, all supported by well thought out code and visualizations. I unequivocally recommend this course to anyone researching or working in a field involving causal inference analyses or applications.”
Lauren Johns, Independence Blue Cross
“Stephen is an excellent teacher, and this course greatly helped fill in some fundamental gaps in my understanding. I feel like I have a much stronger foundation for causal inference research after completing this course.”
Justin Williams, University of California, Los Angeles