“…After β 0 , β 1 , and σ are obtained by a simple regression analyses, POD(a) is calculated as the probability thatâ is greater than a given threshold,â th , on the premise that four requirements, linearity of the parameters, uniform variance, uncorrelated observations, and normal errors [2], are fulfilled. Whereas recent studies have proposed the modifications of Equation (1) so that developed POD can evaluate the probability of false alarm [12] or the asymptotic property in actual inspections [13,14], Equation (1) accompanied by the four requirements, linearity of the parameters, uniform variance, uncorrelated observations, and normal errors, has been used as one of the most important backgrounds to construct POD on the basis of statistical analyses since its development in the 1980s. This study points out two issues concerning the conventional method stated above.…”