SINGLE
Good Single-Stage Designs for Logistic Regression When Parameter Values are Uncertain
SINGLE finds "good'" single-stage designs for conducting
dose-response experiments in which a logistic or log-log models the
probability of response as a function of dose. A prior distribution
describes the initial uncertainty as to values of the model
parameters; choice of prior includes normal, log-normal, and uniform
distributions. A point prior distribution can also be used in which
case parameter values are known. A "good" design minimizes the
variance of a user chosen function of parameters of the model, e.g.
the dose yielding a specified probability of response.
Contact: Barry W. Brown
Language: Fortran 77
