Fig. 1From: Quantifying causal effects from observed data using quasi-interventionThree types of causal graphs. A confounding factor in the data (a), X can exert an indirect effect on Y through an intermediary variable (b); an instrumental variable is found for replacement studies (c). Ux and Uy are exogenous variables, representing any location or random effect that can affect the relationship between endogenous variablesBack to article page