Selection of optimal dose.

Dose and dilution levels for quantal response bioassays and dilution assays are selected rather arbitrary. Some guidelines for designing experiments for binary data arising in such fields as bioassay studies, dilution series assays, and dose-response curves are provided. It is shown that for estimation purposes, the study should be designed in such a way so that the experimental doses lie around a critical interval. Failure to do so many produce serious bias in the Maximum Likelihood estimates. This critical interval, however, depends on the unknown parameters. With the help of extensive computer simulations for the logit model ways of overcoming this difficulty are outlined.