Abstract
The fundamental approach of Malani and Van Ryzin {JASA, 1988) is utilized to develop a nonparametric test for dose-related trend for use with tumorigenicity data from animal bioassays that include interim sacrifices. By constraining the age-specific tumor incidence rate to be nonnegative, the proposed procedure improves both the size and power of the Malani-Van Ryzin test, A real-data example and a small Monte Carlo simulation study illustrate the improvements achieved by the new test.