Abstract |
A Monte Carlo study was conducted to investigate the estimated standard errors of the estimate and the 95 percent confidence interval estimates associated with the trimmed Spearman-Karber (SK) estimators of the ED50 and the logistic model maximum likelihood estimator (MLE). The simulated binary response bioassay experiments had widely spaced doses with 5, 10, or 20 subjects per dose. For data following a logistic tolerance distribution, the trimmed SK confidence intervals were nearly as accurate as the logistic MLE intervals. For heavy-tailed tolerance distributions, the trimmed SK confidence intervals were more accurate than those based on the logistic MLE. |