Fig. 3
From: Assessing causal treatment effect estimation when using large observational datasets
Comparison of adjusted linear model and unadjusted 2SLS IV model estimates from 200 simulated data sets for a sample size of 20,000 and no unmeasured confounding; Different strengths of IV α1 = 0.1 (weak), α1 = 0.3 (moderate), α1 = 0.5 (strong) are given across the x-axis; dashed blue horizontal line is the true causal treatment effect β1 = 3.0, error bars are 95% confidence intervals for the treatment effect estimates for each simulated data set. Black error bars indicate a significant effect estimate, red error bars indicate a non-significant effect estimate