Estimates of bias in a longitudinal coal study.

A large-scale longitudinal study of working coal miners is analyzed to determine biases due to studying only working miners and only volunteers. Reports of symptoms, pulmonary function indices, and chest radiographs were obtained from the first round of the study and comparisons were made of the rates and means among those who participated in the second round, those who had left the industry before the second round, and those who simply did not participate in the second round. Biases were present in nearly all variables studied, but the biases were fairly small.