A Note on Measurement Errors and Detecting Real Differences

Abstract

Abstract Two groups, N 1 = α1 + b 1 and N 2 = α2+ b 2, are compared in terms of their respective morbidity rates, p 1 and p 2, in the presence of errors of misclassification (α1 and α2 are sure components; b 1 and b 2 are potential error terms). The four extreme situations for the compositions of N 1 and N 2 are considered. Rate p 1 is assumed to apply to α1 and b 1; p 2 is assumed to apply to α1 and b 2. If there is a difference in rates for the error-free situation, then the observed difference is smaller in each of the error situations. Further, measurement errors reduce the chance of detecting a real difference.