A large data base will be compiled from all data available on radon levels in U.S. homes. It is expected that this will give reasonably accurate mean radon levels, r, in about 2000 U.S. counties. These will then be correlated with lung cancer rates, m, and numerous socioeconomic variables for these counties to test the predictions of linear-no threshold theory for the dependence of m on r. Basic problems and limitations of this type of analysis, and of ecological studies in general will be investigated using a variety of models. Effects of random errors in these data will be explored. A large case-control study of the radon-lung cancer relationship will be completed. When conclusions are reached on the measured dose-response relationship between radon and lung cancer, explanations for, and consquences of, this relationship will be explored.
