On the Maximum Likelihood Estimator of the Emax Model

It is known in the literature that maximum likelihood esti- mation of the Emax model parameters often encounters computational problems. Our contribution provides a new understanding and control of all the experimental situations. In particular, exact MLE for a three-point experimental design is shown, and we identify the two scenarios where the MLE fails. We show that the D-optimal design generally yields a favorable response for the existence of the Maximum Likelihood Estima- tion (MLE).