Fig. 5From: Bayesian identification of structural coefficients in causal models and the causal false-positive risk of confounders and colliders in linear Markovian modelsDirected acyclic graph for the structural causal model which describes the effect of treatment X on outcome Y under the collider variable C. UX, UY, UW and UC denote unobservable latent variables. The arrow connecting W and X is assumed to be directed from W to X and not from X to W, either due to subject-domain knowledge or temporal informationBack to article page