Applications of statistics in clinical chemistry. A critical evaluation of regression lines.

When two chemical determination methods are compared, a regression line is commonly used to elucidate constant and proportional errors. The classical procedure for the calculation of such a line is compared to the fairly unknown procedures described by Deming and by Bartlett. It is shown that the reliability of the classical procedure is restricted to only some cases. Deming's procedure is more generally applicable and also has the advantage, that estimates are obtained for the random errors in both chemical determination methods. Bartlett's procedure is not recommended.