DAGs, Confounding, and Conditioning

Bayesian

DAGs

Rethinking

Notes on Week 5 of Statistical Rethinking (2023 edition): why adding post-treatment variables skews causal estimates.

Published

August 21, 2026

Code

library(dagitty)

dag <- dagitty('dag {
  Treatment -> Outcome
  Confounder -> Treatment
  Confounder -> Outcome
  Treatment -> PostTreatment -> Outcome
}')

plot(dag)

Plot coordinates for graph not supplied! Generating coordinates, see ?coordinates for how to set your own.