Fig. 3From: A comparison of methods to estimate the survivor average causal effect in the presence of missing data: a simulation studyEstimates from the simulation study. Estimated using 10,000 observations simulated 1200 times for each scenario. The odds ratio effect of the unmeasured variable (U) on the outcome (Y), ORUY, was set to 0.5 in (a) and to 2 in (b). The black line represents the true exposure effect (on the log odds ratio scale) of − 0.51. ORUZ is the odds ratio of the unmeasured variable, U, on survival, ZBack to article page