Inference about comparative precision in linear structural relationships (Q579804)

We study point and confidence region estimators of the precisions of instruments, and of the ratios of such precisions to the precision of a standard instrument, in situations where a nonreplicated linear structural model describes the measurements simultaneously obtained from several instruments. Some drawbacks of maximum likelihood estimators of the precision ratios are noted. A union-intersection test for the null hypothesis \(H_ 0\) that the standard instrument is at least as precise as the other instruments is proposed for situations where the relative precision of the standard instrument is known.