Various methods of interval estimation of the median effective dose

The bootstrap, a resampling approach to statistical inference, has been recommended as an alternative to the more traditional parametric approaches. Assuming a logistic dose-response curve, a number of studies have investigated the percentage coverage probabilities of various methods of interval estimation of the median effective dose (ED50). The several methods have been proposed for constructing confidence intervals using the bootstrap. This article compares three methods of interval estimation of the ED50 based upon bootstrap resampling with four large sample parametric approaches.