Share Description
In this course, you will :
- Using potential outcomes, define causal effects.
- Explain the distinction between association and causation.
- Use causal graphs to express assumptions.
- Use a variety of causal inference methods (e.g. matching, instrumental variables, inverse probability of treatment weighting).
- Determine which causal assumptions are required for each statistical method.
Syllabus :
1. Confounding and Directed Acyclic Graphs (DAGs)
- Confounding
- Causal graphs
- Relationship between DAGs and probability distributions
- Paths and associations
- Conditional independence (d-separation)
- Confounding revisited
- Backdoor path criterion
- Disjunctive cause criterion
2. Matching and Propensity Scores
- Observational studies
- Overview of matching
- Matching directly on confounders
- Greedy (nearest-neighbor) matching
- Optimal matching
- Assessing balance
- Analyzing data after matching
- Sensitivity analysis
- Data example in R
- Propensity scores
- Propensity score matching
- Propensity score matching in R
3. Inverse Probability of Treatment Weighting (IPTW)
- Intuition for Inverse Probability of Treatment Weighting (IPTW)
- More intuition for IPTW estimation
- Marginal structural models
- IPTW estimation
- Assessing balance
- Distribution of weights
- Remedies for large weights
- Doubly robust estimators
- Data example in R
4. Instrumental Variables Methods
- Introduction to instrumental variables
- Randomized trials with noncompliance
- Compliance classes
- Assumptions
- Causal effect identification and estimation
- IVs in observational studies
- Two stage least squares
- Weak instruments
- IV analysis in R
