Visual Guides for Causal Inference
A collection of visual guides designed to help applied scientists learn causal inference.
This repository is licensed under a CC BY 4.0. This means you are free to share or adapt these materials however you see fit as long as you provide attribution (Kat Hoffman).
Current guides and educational illustrations include:
- Targeted Maximum Likelihood Estimation (TMLE), a doubly robust semiparametric estimation method commonly used for causal inference. The guide shows the steps for estimating the mean difference in outcomes, adjusted for confounders, for a binary outcome and binary treatment. A full tutorial with
Rcode is available on my blog.
- Superlearning (also known as stacking), an ensemble learning method recommended to use with TMLE. A full tutorial with
Rcode is available on my blog.
- G-computation (also called the parametric g-formula), an estimation method for causal inference which involves estimating outcome regressions under substitutions to the data.
- Inverse Probability Weighting an estimation method for causal inference which involves estimating treatment regressions (propensity scores) and reweighting the observed outcomes.
- Causal inference intervention types with examples using U.S. air pollution data and a corresponding blog post
- Identification vs. Estimation in casual inference.
Causal Inference Comics
I've also recently been playing around with comics for causal inference concepts. Here's a few so far:
References
Laan, M. J., & Rose, S. (2011). Targeted learning: Causal inference for observational and experimental data. New York: Springer.
Licensing
Visual Guides for Causal Inference by Kat Hoffman is licensed under CC BY 4.0
