Figure 3.

Variations of the directed acyclic graph (DAG) in Figure 2a: differences in causation by exposures. a) Neither $X_1$ nor $X_2$ causes Y. Bias amplification still occurs. $r_{12}$ is the bivariate correlation coefficient for $X_1$ and $X_2$. b) Both exposures cause the outcome. Here there is no optimal solution: Both single-exposure and mutually adjusted regression coefficients are biased.