Up to now, clinical studies only succeeded in differentiating between great categories of psychotropic drugs, but failed to prove finer differences of effects within these categories of substances. Two points of the testing-method are discussed: 1. problems which arise when rating pathological behaviour and 2. problems of sampling psychiatric patients. A great part of symptoms that clinicians and psychologists used to consider as relevant proved to be extremely rare. Total scores cannot be taken as a measure of the therapeutic effect, because they don't express adequately the degree of severity of the illness before and after treatment, and there is a symptom that is independent from the observer, i.e. the frequency of the symptom in different clinical pictures. The frequency is an inverse ratio to the specificity of the symptom. It is then argued that even in clinical studies, it would be possible to choose among the variety of descriptive symptoms those which fulfil requirements of the probabilistic test-model of Rasch and to take only those symptoms to characterize the degree of severity of psychic disturbance in trials with psychotropic drugs. Conclusions are then drawn from a study including three groups of physicians (specialists for internal diseases, psychiatrists and general practitioners): failure to differentiate between placebo and a Minor-Tranquilizer was not due to the inefficiency of the drug, but ought to be attributed to the lack of sharpness of the observations made by untrained judges. A significant difference between placebo and the Minor-Tranquilizer was yet found, but only in the group of psychiatrists. The comparison of the first 13 and the last 13 cases in the two remaining groups, however, reveals a learning process in the course of the study. The main problems of sampling are discussed, i.e.: the loss of information as a consequence of taking the mean in a group of psychiatric patients, the role of biological rhythms, which was hitherto insufficiently considered, and finally it is demonstrated in connection with two selected cases of depressive patients that enormous difference of psychophysiological responsiveness can be hidden behind very similar clinical pictures. It is pointed out that the existing research strategy is adjusted to great samples, which were composed on the basis of behavioral characteristics, and that it failed to differentiate subtle effects of psychotropic drugs. Only experiments involving a much greater display, which take into account all aspects of observation of the selected single cases and longitudinal studies can answer the question which is the right medicine for the right patient. Psychophysiological and biochemical methods have here priority over other methods.
