ABSTRACT
The accurate assessment of exposure in epidemiologic studies can have an important impact on the results of these studies. Of course, if exposure assessment in the diseased and nondiseased study subjects differs, a bias is induced. Equally inaccurate assessment of exposure in the two groups results in equal misclassification. As pointed out by Bross [1], this results in a loss of power but does not affect the size of the statistical test of association between disease and exposure. As a consequence, important disease exposure associations can be overlooked.
