Why a Computerized Decision System does not always lead to the Correct DSM-Classification
According to Lars Bergman there are several reasons that keep computerized systems from producing the correct DSM-classification. The most important point might be that diagnosis in psychiatry is just to complex for a fully structured system to reach a good level of validity. Dynamic systems and appropriate statistics might be necessary. According to Dreyfus the intuition of the expert is an important part of the diagnostic process. This would also mean that the (paper version of the) SCID does not yield the expected validity. Other reasons are related to methodological problems of the study at hand: Most physicians used the computer version of the SCID for the first time. They also received only limited training and thus the computerized process might have interfered with their normal skills. Also the rule-following and logical step-by-step procedures designed for non-experts might have influenced validity.