On the comparison of different tests for identification of a compound from its mass spectrum

It is shown that identification tests of different dimensions or dimensionless should not be evaluated (for their efficiency to identify molecules from their mass spectrum) by comparing the tests for a molecule itself (repeated measurements) with other molecules. This kind of tests must have similar dimensions (units). Another possibility is the comparison of tests on the basis of success of correct prediction for "unknown" molecules from a library of standards. (C) 2003 American Society for Mass Spectrometry.