Fig. 3

Simulations of the interim analysis. A The number of correct interim decisions—where an interim was stopped where there was no significant finding or continued for effect sizes of 0 and 0.6. Error bars indicate 95% confidence intervals generated by bootstrapping the simulation results sampling 100 results 1000 times. B The change in mean % correct interim decisions as more subjects are added to interim. The interim number 15 was chosen as the relative benefit of adding more decreases above this point. C The effect of number of subjects on overall interim pass rate. This figure displays the % of interims that would go on to recruit a full study from populations of fixed effect sizes