Explaining the lung cancer versus radon exposure data for USA counties

Our study of lung cancer mortality rates, m, versus average radon levels in homes, r, for 1601 USA counties (Cohen 1995) found a very strong negative correlation. In a recent exchange (Cohen 1999), Sir Richard Doll suggested that these discrepancies may be due to lack of consideration of details on smoking habits, specifically 'the amount smoked, the proportion of ex-smokers who have stopped for different periods, the proportion of smokers who smoked cigarettes, cigars or pipes, and the age at which those smoking in different ways began'. The purpose of this letter is to explore these Doll suggestions. Our analysis is performed for males because the data are of better quality, but the results would be quite similar for females. Letter-to-the-editor