The aims of the research program are to develop statistical techniques that will be useful in the planning and analysis of environmental and early detection studies. Specific objectives in the environmental area include investigation of the efficiency losses resulting from inexact measurements of exposure data, and methods for analyzing environmental studies where there is limited information, at the individual level, for all but those persons that develop the disease in question. The former project can lead to more efficient designs of environmental studies by allowing the varying costs of different degrees of accuracy in obtaining exposure data to be balanced with the corresponding efficiencies of the statistical methods for testing for associations. The second project explores the potential of relatively low-cost environmental studies that only require detailed information on disease cases, and utilize more readily available aggregate data for the rest of the study population. Specific objectives for early detection studies include the elucidation of a parallelism in structure with carcinogenicity experiments, improved estimators of mean sojourn time in preclinical states, and relaxation of the commonly used assumption that disease-free duration is independent of duration in the preclinical state. The long-term aims of this component of the research plan are to develop an improved understanding of the results of screening programs for the early detection of disease.
