Outcomes were simulated to be unrelated to the exposure (bottom plot, showing false positive rates on the y-axis) or causally influenced by the exposure (top plot, showing true positive rates on the y-axis) with varying degrees of measurement imprecision applied to the exposure variable (x axis).