Causal Inference glossary

Clear, one-line definitions of the Causal Inference terms used across the OgbonLab textbooks. Each entry links to the interactive sections where the idea is taught.

ate

Average treatment effect: E[Y(1) − Y(0)] across the population, in potential-outcomes notation.

att

Average treatment effect on the treated: E[Y(1) − Y(0) | T = 1]; the effect for those who actually received treatment.

backdoor criterion

A set Z satisfies the backdoor criterion if it blocks all non-causal paths from T to Y without blocking causal ones.

confounding

A common cause of treatment and outcome that biases naive comparisons; addressed by adjustment, IV, or randomisation.

See: Confounding and the DAG toolkit, Capstone 2, an observational study with confounding

dag

Directed acyclic graph encoding causal relationships; arrows show direct effects, and d-separation reads off conditional independence.

difference-in-differences

Compares before/after changes between treated and control groups; identifies causal effect under parallel trends.

See: Difference-in-differences and synthetic controls

do-calculus

Pearl's three rules for transforming interventional expressions P(Y | do(X)) into observational quantities via a DAG.

instrumental variable

A variable that affects the outcome only through the treatment; identifies causal effects in the presence of confounding.

See: Instrumental variables

itt

Intention-to-treat: analyse subjects by assigned group regardless of compliance; preserves randomisation but dilutes effects.

propensity score

e(X) = P(T = 1 | X); balancing it across groups removes confounding from measured covariates in observational data.

See: Propensity scores, matching, and IPTW

rct

Randomised controlled trial: subjects are randomly assigned to treatment or control, making T independent of confounders.

regression discontinuity

Exploits a sharp assignment rule at a cutoff to compare units just above and just below for local causal effect.

See: Regression discontinuity