THE PROBLEM OF RANDOMIZATION IN PSYCHOTHERAPEUTIC RESEARCH

In a research project comparing inpatient and outpatient psychotherape utic treatment for patients suffering from anxiety disorders, 132 subj ects were allocated to both treatments by means of randomisation. Beca use of the patients' voluntary agreement or refusal of the proposed tr eatment, the subsamples are markedly affected in their homogeneity. It seems that more disturbed patients agreed to inpatient psychotherapy, whereas less disturbed patients refused that form of treatment. Bias toward or against a given form of treatment turned out to be the most potent predictor in multivariate analysis. The influence of this disto rtion of the samples is discussed with regard to the generalisation of the results. Proposed countermeasures have their shortcomings, and in the end it seems that randomisation should be avoided in psychotherap eutic research.