[Methodological and statistical aspects of therapy evaluation exemplified by the Berlin psychotherapy study]

The answer to the question concerning the efficacy of psychotherapy is increasingly sought metaanalytically. For this purpose almost exclusively investigations are taken into consideration which have at least one control group. Using realistic data, the present paper refers to some basic considerations for the investigation and evaluation of therapeutic changes, which are however not always satisfyingly solved in metaanalytical studies. Investigations, therefore, especially of naturalistic samples, such as may be found in the context of outpatient or inpatient routine care, have a high clinical relevance for the evaluation of therapeutic changes even if they do not satisfy experimental requirements/standards. However, they do supplement experimental studies where the important aspect of external validity is concerned.