What’s in a Word? Prediction

By Al Murdach, LCSW, (retired)

These columns explore some of the meanings of words that today have a significant impact on the social work profession.  Our word for this column is prediction.

Social Workers are often called upon to make predictions about the outcomes of their work with clients and patients.  Often these predictions (usually called “prognosis” in medical jargon) are routine and not especially important.  However, occasionally these forecasts can be significant and involve such serious issues as client dangerousness, suicidality, mental competence, ability to function independently, potential for spousal abuse, use of harmful substances, ability to respond to treatment, etc.  Given the importance of these concerns, it would appear that the process of clinical prediction needs more attention in the social work literature, where it is now passed over summarily.

Numerous studies of clinical prediction in health and mental health settings have been done since the 1950’s, especially by psychologists. However, the results of these studies are not especially encouraging.  In general, they show that clinician’s predictions of future problematic behavior are sometimes correct but are also often erroneous or misleading.  There are several factors causing these results.

Social Constraints—The social context within which practitioners operate can have a profound effect on their ability to predict client behavior accurately.  Constraints such as lack of necessary information, administrative pressures, fear of legal reprisal,lack of time for necessary research, and fatigue or overwork can all lead to errors in forecasting client progress.  Add to this the uncertainty that pervades the lives of both clients and practitioners and the resulting mix can constitute a morass of informational buzz that defies clarity of any kind.

Psychological Constraints—The practitioner’s own mental biases can also endanger the capacity to predict correctly.  The practitioner’s past experiences may both enhance and hinder their abilities to correctly observe and interpret client behavior.  Correlations that appear in client behavior may sometimes be illusory and guided more by the practitioner’s eagerness to see a desired result than an accurate representation of actual behavior.

Faulty Work Habits—Some professional work habits, once seen as praiseworthy, may actually inhibit the practitioner’s ability to make accurate predictions about their clients.  The time honored reliance on individual assessment interviews, for example, has been proven in numerous studies to lead to erroneous conclusions about the future performance of clients. The same is true for psychological testing.  The same is also true for the professional habit of collecting and relying on extensive family history to make predictions about future client behavior.  None of these work standards has been shown to increase accuracy in clinical prediction.

So what are clinicians to do when they are asked or required to make predictions?  A variety of authors suggest the following cautions.  First, be aware that some future events will be“unique” for certain clients and can never be predicted.  Second, don’t ignore “base rates” or the incidence of certain behaviors in the client demographic in which your particular client belongs. Third, try to make predictions for the short term, say the next few days or weeks.  The longer the term the more tenuous predictions become. Finally, and most importantly, know as well as possible what the past behaviors of your client have been, especially in the immediate past leading up to the present current difficulties.  A variety of  authors have stressed that the data repeatedly show that prior behavior provides the most powerful indicator of future behavior.

In general, most authors agree that professional prediction about future client behavior will continue to be difficult  for practitioners because they, as “experts”, will increasingly be pressed to provide the public with ever more accurate forecasts about events over which they often have little control.