A computer program for the statistical analysis of cancer inhibition/promotion experiments

The computer program discussed and presented in this paper is designed for the statistical analysis of data from cancer inhibition/promotion experiments that involve Type I censoring. The theoretical derivation of the method is described in Biometrics 43 (1987) 525-534 and Appl. Math. Lett. 1 (1988) 197-201. Turbo Pascal is used to compute maximum likelihood estimates of the parameters of interest: the mean number of induced tumors per animal, and the mean time to tumor detection. The program conducts a likelihood ratio test to determine if there is an overall experiment effect, and allows the user to select options to isolate the pair(s) of groups contributing to the experiments effect, to isolate group differences in terms of tumor number and growth rate, and to graph 95% confidence regions for the vector parameter of interest.