The Use of Parametric vs. Nonparametric Tests in the Statistical Evaluation of Rating Scales

In psychiatric studies, treatment efficacy is usually measured by rating scales. These scales have ordinal (rank) level and the statistical evaluation of the scale scores should be performed with nonparametric rather than parametric tests. In recent years, nonparametric statistical procedures for repeated measures have been developed for the evaluation of clinical trials in psychiatry. To assess the frequency of application of nonparametric tests in psychiatric trials, an analysis was performed on all treatment studies with panic disorder patients (DSM-III/III-R) that could be traced in the literature. This survey revealed that nonparametric tests were used in only 16.7 % of all studies.