Temperature control in assay of glycosylated hemoglobins.

1106 CLINICALCHEMISTRY,Vol. 26, No. 7, 1980 knowledge of the relative precision characteristics of the two compared methods (2-4). When such information is lacking, the best that can be done is probably to put A equal to 1. Unless there is an uncommonly large discrepancy in precision between the two methods, this will be more realistic in comparison studies than the assumption, implicit in the use of a leastsquares regression, that one of the methods has zero imprecision. For Schall et al.’s data set (1), with A 1 we find the Deming regression line to be