Many practice situations require that social workers make predictions about the future behavior of their clients. Returning children to the care of natural parents, screening mentally ill people for admission to psychiatric treatment facilities, referring high-risk elderly patients for aftercare services, assessing an individual's chances of committing suicide--all these and many other activities require that practitioners make some type of prognosis or forecast about their clients' future progress. In addition, practitioners find themselves subject to growing pressures to provide expert opinions about the future course of events in their clients' lives. This situation has arisen because the need for professional opinions, especially those regarding individuals whom society considers as particularly vulnerable (for example, children and elderly people) or threatening (for example, individuals classified as dangerous), is becoming a public and collective (Bosk, 1989, p. 461). Error is a problem of special concern in clinical prediction, but despite the growing importance of this topic (Gambrill, 1990), social work practice texts typically give it scant notice. Unlike authors in such fields as psychiatry, nursing, and psychology, social work clinicians typically have neglected to consider the details of the prediction process in clinical work because of their focus on the decision-making activity of the client rather than the clinician. Current events, however, are forcing a reevaluation of this position as community concern, liability issues, and legal and administrative pressure force professionals to pay increased attention to understanding the prediction process and the ways it can be improved (Besharov, 1985; Gambrill, 1990). Prediction is a common feature of clinical reasoning and constitutes an important feature of the clinical inference process. According to Wiggins (1973), prediction is generally composed of the following four steps: 1. data collection, during which available information is assembled and reviewed 2. identification of variables, during which meaningful cues and patterns are noted 3. formulation of a hypothesis, during which clinical inferences are made based on the variables that have been identified 4. forecasting, during which hypotheses are used to generate probabilities about future trends and results. Although these steps are not followed in every case, they reflect commonly described features of clinical prediction. In form they resemble the process of reasoning generally associated with the physical sciences (Gill, 1986). This process is also used in social work practice. Case Study A social worker in a psychiatric hospital was asked by ward staff to assist a 32-year-old female patient in arranging her air travel to another state in preparation for her discharge. The social worker contacted the airline about travel information and was told that as the social worker, she would be required by airline regulations to sign a statement that, in the worker's professional judgment, the patient was psychiatrically stable and sufficiently in control so as to pose no risk to others and that the patient did not require any specialized behavioral management during her flight. To honor this request, the social worker met with the patient to evaluate her condition, discussed her progress with staff, and quickly reviewed her hospital records. While interacting with the patient and staff, the worker searched for patterns or cues that would indicate the presence of problems that would preclude air travel by the patient. Because none were found, the social worker felt that she could confidently predict that the patient was fully capable of making the flight. The worker then completed and signed the statement required by the airline company. While driving the patient to the airport, however, the worker suddenly felt that she had made an error. …