POWER:

The main weakness of residential radon studies

J. Lembcke, J. Conrady, K. Martin

Abstract: Two main mistakes are frequent in testing a hypothesis in epidemiology. One may reject the null hypothesis when it is true (Type I error, expressed as significance), or one may fail to reject the null hypothesis when it is false (Type II error, expressed as power).Significance and power are the most important statistical criteria for the quality of epidemiological studies. Power estimates are primarily needed in planning a study to calculate the required sample size related to an estimated risk. However, there is little information in case a study did not meet the postulated power and the consequences of such mistake. This results in studies regarding the lung cancer risk due to indoor radon which are not qualified regarding their power. The paper addresses the quality of recent radon studies, and the minimum standards for "good epidemiology" regarding power and significance.

Submitted to Health Physics, 2000