The long range objectives of this proposal are to develop and evaluate statistical techniques which are useful in the design and analysis of human studies arising in cancer research. Specific topics for the current funding period include: (1) distribution-free methods for comparing treatment groups with respect to repeated measures data in randomized clinical trials; (2) methods for detecting and controlling for overdispersion in censored survival data; and (3) estimator of odds ratios in case-control studies which incorporate distributional assumptions concerning exposure.