Key points are not available for this paper at this time.
A FREQUENTLY used procedure in epidemiology and other disciplines involves the comparison of disease rates in two populations, communities, or groups.While epidemiologists desire, of course, to know the true population rates, they must, of necessity, be content with observed rates, often derived from samples of the populations.These observed rates may dif- fer from the true rates of the samples as a re- sult of error introduced by misclassification according to disease status.In other words, some persons with the disease will be erroneously classified by the study procedure as being with- out the disease and some persons without the disease will be classified, in error, as having the disease.As well described by Rubin and co- workers (1), the apparent difference in sample rates is related to the true difference in sample rates by the formula (P'1-P'2) = (P1-P2) (1-q-0) 1 where:
Gullen et al. (Mon,) studied this question.