Rating scales in psychopharmacology. Statistical aspects.

The scaling problem of measurement of clinical data in psychiatry should be differed from, although connected with, the statistical problem of statistical significance testing in psychopharmacological trials. The scaling levels (ordinal versus interval) of a rating scale as well as the statistical tests (non-parametric versus parametric) are different in Phase II and Phase III trials. Sufficient statistics of improvement curves and of recovery rates are discussed.