Required number of simulations to constrain the true-positive rates (top panels) and false-positive rates (bottom panels). Subsamples of size ﹩{N}_{\mathrm{sim}}﹩ are drawn from a sample of 100,000 simulations, and true- and false-positive rates are calculated for each subsample. One can observe high variability for subsamples with low ﹩{N}_{\mathrm{sim}}﹩, but as ﹩{N}_{\mathrm{sim}}﹩ increases the calculated quantities converge to a given value. These simulations were from a scenario with ΔA = 0.4 and N_EEC = 25. Left and right columns represent different threshold values for the Bayes factor required for detection of a trend.