Category
Causal Graphs & Reasoning
11 free guides in this category.
2 guides
Backdoor, Frontdoor & IV
Backdoor Criterion
A rule for identifying which variables to control for when estimating cause-and-effect relationships from observational data, helping us separate genuine causal effects from misleading correlations.
Read guideFrontdoor Criterion
A method for identifying causal effects in some causal graphs with unobserved confounders, by using an observable mediator variable along the causal path.
Read guide3 guides
DAG Foundations & Basics
d-Separation
A graphical criterion for reading conditional independence from causal diagrams, revealing which variables influence each other and how information flows through a network.
Read guideDAG Terminology
The core vocabulary for directed acyclic graphs: nodes, edges, paths, ancestors, descendants, and topological order—the building blocks for modeling cause, dependency, and sequence.
Read guideCausal Markov Condition
A fundamental rule stating that knowing an event's direct causes makes its more distant causes irrelevant for prediction.
Read guide4 guides
Do‑Calculus & Identification
Adjustment Set Selection
The process of identifying which variables to control for to isolate causal relationships in complex systems.
Read guideInterventions and the Do-Operator
A tool for distinguishing between seeing a relationship (correlation) and predicting the effect of making a change (causation).
Read guideThe G-Formula
A mathematical formula that identifies causal effects from observational data by adjusting for confounders, enabling estimation of what would happen under different treatment scenarios.
Read guideConditioning on a Collider
The practice of analyzing a subset of data based on a shared effect of a cause and an outcome, which can create spurious associations.
Read guide1 guide
Mediation & Moderation
1 guide