Prediction and Disposition

10

The ability of mental health professionals (MHPs) to make accurate predictions of future violence has generated an enormous amount of research, theory, and debate since the early 1970s. Initially, most researchers were pessimistic about the ability of MHPs to predict violence. They based their view on a number of studies such as the one conducted by Kozol, Boucher, and Garofalo (1972), who found that predictions of future dangerousness, even if arrived at by a team of mental health experts, were highly inaccurate. In fact, Ennis and Litwack (1974) argued that MHPs have neither a special ability nor specialized training or techniques for making violence predictions. These authors concluded that the accuracy of MHPs’ predictions is not much better than that of laypeople.

Additional research, commonly known as the “Baxstrom” (Steadman and Cocozza, 1974) and the “Dixon” (Thornberry and Jacoby, 1979) studies, supported the earlier results. The investigators found that psychiatric patients who were confined for years because of a finding of dangerousness turned out not to be dangerous at all when they were eventually released. As a consequence of these and similar findings, Monahan (1981) concluded that MHPs were wrong twice as often as they were correct in their violence predictions and that, accordingly, significant changes in public policy needed to be made. However, further scrutiny uncovered a number of methodological weaknesses in what Monahan (1984) referred to as the “first generation” of research and thinking on this topic. Monahan stressed the need for a second generation of research, thought, and policy on violence prediction rather than just conducting additional studies to demonstrate the inaccuracy of MHPs in predicting future violence among mentally disordered individuals. He proposed the need to (1) study various methods of prediction, (2) use actuarial techniques that incorporate clinical information, (3) employ statistical tables on various situational factors (e.g., family, work, peer group environment), (4) study different populations of offenders, and (5) specify the time range of predictions.

Research and debate on the ability of MHPs to predict violence have continued (Menzies, Webster, McMain, Staley, and Scaglione, 1994; Monahan, 1996, 1997), with the current consensus being that some prognostic statements can, in fact, be made under some circumstances (Litwack and Schlesinger, 1987, 1999). Litwack and Schlesinger (1999) note that considerable strides have been made in research methodology, assessment of risk factors, development of actuarial risk assessment instruments, and guidelines for clinical assessment (Webster and Jackson, 1997). Although research findings and subsequent thinking have become more positive regarding the predictive ability of MHPs in comparison with the pessimistic view of the first generation of researchers, the criminal and civil justice systems have in fact never stopped relying on MHPs’ violence predictions. Many statutes have even required that an МНР offer such a prediction. Borum (1996) notes:

Despite a long-standing controversy about the ability of mental health professionals to predict violence, the courts continue to rely on them [MHPs] for advice on these issues [civil commitment and parole recommendations] and in many cases have imposed on them a legal duty to take action when they know or should know that a patient poses a risk of serious danger to others [Tarasoff v. Regents, 1976]. (p. 954)

Research and thinking on dangerousness predictions continue to evolve. A recent controversy resurrects an old issue regarding the superiority of clinical vs. actuarial methods, a question initially framed by Meehl (1954) in a much different context. The debate centers on whether MHPs should use clinical judgments or statistically grounded instruments in offering their violence predictions. Quinsey, Harris, Rice, and Cormier (1999) have proposed the total replacement of clinical assessments of dangerousness with actuarial methods: “What we are advising is not the addition of actuarial methods to existing practice, but rather the complete replacement of existing practice [clinical judgments] with actuarial methods” (p. 171). These authors argue that psychometric instruments such as the “Violence Risk Appraisal Guide” (Quinsey et al., 1999) are far superior to judgments made by individual practitioners or groups of practitioners.

Although a number of researchers have taken the general position that actuarial assessments are superior to clinical assessments (Douglas, Cox, and Webster, 1999; Gottfredson, 1987), other researchers do not share this view. For example, Litwack (2001) investigated the evidence on which the purported superiority of actuarial assessments was based and found this position to be seriously flawed.

It is frequently claimed, often as if it were hardly worth discussing, that actuarial assessments of dangerousness have proven to be superior to clinical assessments. However, the actual picture that emerges from the research is far more complex. In addition, apart from their relative merits vis-a-vis clinical assessments, far more is currently claimed regarding the utility of actuarial tools for assessing dangerousness than is merited by actual research findings.

(p- 411)

In fact, Litwack concluded that not only is there little empirical support for the contention that actuarial assessments of dangerousness are superior to clinical assessments, but in some circumstances clinical assessments may actually be superior to actuarial assessments. As a practical matter, the author believes (and we strongly agree) that not only will clinical assessments of violent behavior continue, but it is “hard to imagine that the day will ever come when actuarial assessments of dangerousness can properly and completely substitute for clinical assessments” (p. 437).

The research, thinking, and debate regarding the best way to predict future violence has continued. No one is able to predict future violence (or future behavior) of an individual with a high degree of accuracy, and no one can ever guarantee an individual will or will not act out in a violent way. Accordingly, risk assessment—which is not and should never be considered a prophecy of future behavior—is an analysis of various factors in a case that can contribute to future acting out. To help in risk assessment, various risk assessment instruments have been developed, some that are very quantitative in nature (e.g., Miller, 2006) and others (e.g., Douglas, Hart, Webster, and Belfrage, 2013) that use a structured professional judgment approach. The HCR-20 (Douglas et al., 2013) encompasses static factors (behaviors in the subject’s background that cannot change), as well as dynamic factors which can change (such as his response and compliance to treatment). An assessment of risk must take into account the nature, magnitude, probability, eminency, and frequency of violence. These factors, as well as response to psychotic symptoms, ability to mask or cover up symptoms, effectiveness of prior treatment, and openness with respect to discussing his history and various problems, are all important factors in predicting future violence. Predicting human behavior is exceptionally complex. There is no one method that can stand alone as being dispositive for predicting violence—all relevant factors must be taken into account. But in the final analysis, a clinical judgment by an experienced and informed mental health professional (МНР) has to be made.