Fig. 3
Scenario A: Increase in biases of treatment effects in the model adjusting for IV (X3As) compared to those in the model without adjusting for IV under the presence of unmeasured confounder (X1). Bias of treatment effects was calculated by subtracting true values of marginal expectations obtained in a large (N = 10,000,000) sample from g-computation estimates across the 10,000 datasets in each situation. Each panel shows the magnitude of increase in biases when the g-computation algorithm additionally included IV (X3As) which is associated with T1 and T2 (odds ratio = 2.0, 5.0, or 10.0) under the presence of an unmeasured confounder (X1). Biases of [Y1, 1] – E [Y0, 0] (the right column) were multiplied by − 1 to provide intuitive information on the gap from the true estimates because the true estimates of [Y1, 1] – E [Y0, 0] were negative