A three-year follow-up study of 49 psychiatric patients, representative of those found in long-stay care, tested whether behavioural, symptom and demographic variables as well as response-processing difficulties could predict levels of psychiatric care. The researchers were blind to the criteria for allocating patients to particular forms of care and the staff responsible for care had no access to information collected by the research team. Not only was it possible, using only these few factors, to predict the form of care, but one factor, response processing, was related to whether patients moved to more independent forms of psychiatric care.