Forensic psychology: Part 4: Chapter 15: Assessing offender populations

by James R.P. Ogloff, J.D., Ph. D., R Psych. ^{Footnote 1} 

At the heart of any risk assessment process lies one question - will this person engage in dangerous behaviour upon release? This question is deceptively simple because, as we will review here, neither "What is dangerousness?" nor "How can we determine whether a person will be dangerous?" are particularly straight forward questions.

Without engaging in the academic debate about what dangerousness really is, we simply note that the use of the term "dangerousness" here relates both to physically violent behaviour directed at oneself or others and to threats of physically violent or intimidating behaviour directed toward oneself or others. This definition is necessarily broad to include the scope of physically and verbally aggressive behaviour that may be sufficient to warrant criminal intervention.

Many scholars and practitioners have noted that, when discussing the prediction of dangerousness, it is critical to specify exactly what is being predicted. When we use the term "danger," we really mean that someone is at risk of harm. By contrast, the term "dangerousness" is used to describe some act, situation or person that places people at risk of harm.

As we will emphasize in this article, simply describing an individual as "dangerous" - whether it is done by a mental health professional or a layperson - is far too vague to be of any value to the Parole Board. Indeed, very few, if any, individuals are "dangerous" to all others across all situations. Therefore, a recurring theme in this article will be to emphasize how important it is to evaluate the particular factors (internal and external) that precipitate an individual's dangerous or violent behaviour.

Also, it is important to note that the focus here is on the assessment of dangerousness specifically, as opposed to risk generally. You will notice that some of the material will overlap to some extent with the information you have heard, and will hear, about risk assessment generally. Risk assessment is a broad term that is used to identify the range of factors that may jeopardize an inmate's reintegration into society (e.g., employment possibilities, support systems, proclivity to commit further crimes, dangerousness, etc.). Thus, dangerousness may be viewed as simply one of the factors that may place an inmate at risk of failing on parole.

Given the above information, we will emphasize that, when making predictions of an individual's degree of dangerousness, you must consider detailed information concerning 1) the nature of the dangerous acts to which a person may be disposed; and 2) the risk conditions or factors that might dispose the person to act in a dangerous manner that places others at risk for harm.

As the above discussion makes very clear, the prediction of dangerousness is a particularly difficult, complicated and controversial chore. Empirical research generally shows that mental health professionals and others tend to over-predict violence. As a result, there are five caveats - or warnings - that we would like to discuss first. After these caveats have been introduced, we can then move on to briefly summarizing the literature that has evaluated our ability to predict dangerousness and providing you with concrete information about the risk conditions or factors that may be most effective in determining whether an individual will engage in dangerous behaviour that will place others at risk of harm.

i) Parameters of the prediction
It is never enough to simply mention that a person is dangerous. Rather, it is essential to specify, with as much detail as possible, the parameters within which an individual may be dangerous, including the type of dangerous or threatening behaviour, the type of victim and the time frame within which the dangerous behaviour may occur. Thus, instead of saying "this person is dangerous," we should think of dangerous prediction in more probabilistic terms. For example:

ii) Sources of information
It is important to consider the validity of the information upon which a prediction of dangerousness is being made. It is best to have corroborated evidence and carefully documented incidents rather than to merely rely on generalizations made by people or self-reports made by the offender.

On one hand, it is all too easy for reports of violence or violent episodes to become embellished. For example, in some work I was doing, I came upon the file of a mentally ill inmate who had been found unfit to stand trial after being charged with assault with a weapon. Of course, such a charge conjures up all sorts of heinous visions. After investigating the details of the incident leading to the arrest, however, I learned that, in fact, the man had thrown a butter knife at a home care worker. While this behaviour could still be labelled violent or dangerous, it is qualitatively different than attacking someone with a hunting knife or gun.

On the other hand, if dangerous incidents are not carefully documented, they may be dismissed far too readily. An example of this occurred when a simple note in a patient's chart indicated that the patient had thrown a staff member against another staff member, resulting in both of them being "hurt." Upon further investigation, I learned that, in fact, the patient had rather forcefully attacked and thrown one staff member into another, resulting in both staff members being injured seriously enough to warrant workers' compensation benefits. Thus, this incident could easily have been overlooked because of the lack of careful documentation.

It is always risky to rely exclusively on an inmate's self-report of previous behaviour. There are simply many reasons for the inmate to distort the nature or extent of that behaviour. Therefore, it is important to consider evidence that may serve to confirm or disconfirm any information obtained by self-report (see Mulvey &Lidz, 1993). Such a process is effective in ensuring that the assessor is not being manipulated by a deceptive inmate (Rogers, 1988).

iii) Myths surrounding the prediction of dangerousness: Illusory correlations
Many myths surround the prediction of dangerousness. Most are based on what we call "illusory correlations," that is, relationships that we believe may exist between dangerousness and some potential risk factor, when no actual relationship has been supported by research. For example, it is not uncommon for people to believe that mentally ill people are more likely to be violent and dangerous than non-mentally ill people. However, there is little evidence that such a strong relationship reliably exists.

Illusory correlations occur when follow-up information is obtained unsystematically. Indeed, we tend to remember those events or situations when we were right (i.e., we just knew that a certain person was dangerous and would reoffend) rather than those cases when our suspicions were not confirmed (i.e., when a person who we thought was dangerous does not reoffend and return to prison). It is important for Parole Board members to know that they should not rely on relationships between risk factors and dangerousness that they may believe are valid — based on their own experience — without having some empirical evidence that the relationships reliably exist.

Rather than reviewing some or all of the prediction myths that exist, a later section of this article will instead delineate the factors that have been found to be validly related to dangerousness.

iv) Low base rates for violence and dangerousness
Although it is sometimes difficult to believe, the actual rates for violent behaviour among people in society is very, very low. Even among former inmates, the rates are quite low. For example, the prevalence of violent recidivism among inmates released on parole varies from 10% to just over 50%, depending on the source of information and the sample. The fact that base rates for violence and violent recidivism are quite low contributes to the difficulty of predicting which offenders will be dangerous if paroled and which will not be.

Indeed, the more frequently a behaviour occurs in society, the easier it is to predict that behaviour. Take coffee drinking for example. Since so many people drink coffee, if you are asked to predict whether any given individual drinks coffee and your prediction is "yes," you have a high probability of being correct. However, if a behaviour occurs less frequently, the probability of correctly predicting the occurrence of that behaviour is low. For example, the murder rate in Canada is approximately 2 per 100,000; therefore, it is exceedingly difficult, impossible really, to predict who will be murdered and who will commit murder.

The Corrections and Conditional Release Act (CCRA) makes use of base rate information. For example, the accelerated parole review provisions (ss. 125-126) pertain to those offenders who have a low base rate for violence prior to incarceration (i.e., a first-time federal offender whose offence is not violent), while the detention provisions (ss. 129-132) apply to offenders with a higher base rate of violence (e.g., offenders who are serving a sentence for a violent offence, or for a serious drug offence).

v) Cultural differences
Cultural differences between the people making predictions of dangerousness and the inmates themselves may contribute to errors in determining who is actually dangerous. For example, a highly verbal white, middle-class person may react to stressful situations very differently than a person who has difficulty speaking English (or French) and is not comfortable in the setting.

Theoretically, there is no empirically proven, effective way for anyone to determine with any degree of accuracy who will and who will not recidivate in a violent way. This is particularly true for long-term predictions. Nonetheless, the CCRA requires Parole Board members to make such determinations. In this section, I will note some of the research that has investigated the prediction of dangerousness.

i) Studies investigating the prediction of dangerousness
In the 1960s, researchers began to focus their attention on the extent to which mental health professionals could accurately predict dangerousness. The "first generation" of research considered the ability of mental health professionals to accurately predict whether forensic patients, psychiatric patients or inmates would engage in dangerous behaviour upon release. In a well-known review of this literature, Professor John Monahan concluded that: psychiatrists and psychologists are accurate in no more than one of three predictions of violent behaviour over a several-year period among institutionalized populations (Monahan, 1981, pp. 47, 49).

Professor Monahan's damning review spawned a ((second generation" of research that has been characterized by a more systematic approach to the research (e.g., using carefully defined measures of violence), a more careful use of predictor variables and more focus on short-term predictions of dangerousness. The results of these "second generation" studies are briefly reviewed in this section. Readers are referred to the reference section for more extensive reviews of the empirical literature on the prediction of dangerousness.

When mental health professionals (and others) make dangerousness predictions, they do so based on either clinical judgment or some actuarial instrument. The predictions based on clinical judgment are known as "clinical predictions," while those based on instruments that assess predictor variables are called "actuarial predictions." Some attention has been paid in the literature to the relative utility of these two approaches in the prediction of dangerousness. This research is reviewed below.

Clinical prediction of dangerousness
Several studies have been conducted to determine the extent to which mental health professionals are able to accurately classify dangerous and non-dangerous inmates or patients. The details of each study are not particularly important here, so the correct classification rates of several of these studies are present in tabular form below. ^{Footnote 2}  As these figures show, the ((second generation" studies indicate that mental health professionals are able to predict dangerousness better than if the predictions were simply made by chance. The high false positive and false negative rates are still troublesome, however.

Actuarial prediction of dangerousness
In contrast to the clinical prediction studies, the actuarial prediction of dangerousness involves the identification of potential predictor variables that may be useful in determining who will — or will not — be dangerous following release or discharge. Once again, the results from these studies are summarized in tabular form. ^{Footnote 3} 

As the research shows, actuarial predictors of dangerousness tend to be more accurate overall (i.e., fewer false positives and false negatives) than clinical predictions (see Hall, 1988, for a review). Therefore, it is advisable to be extremely cautious in accepting any individual's clinical appraisal of another person's potential dangerousness.

Next, we turn to a brief review of some of the instruments that are currently being used to assist in the prediction of dangerousness.

ii) Tools for predicting dangerousness
Ever since researchers and clinicians began to systematically consider the question of whether an individual will be dangerous at some point in the future, people have attempted to develop instruments to aid in this task. Unfortunately, for the most part, the instruments that have been developed to date show little promise.

It may be worth noting explicitly here that general personality scales, such as the Minnesota Multiphasic Personality Inventory (and the MMPI-2) and the California Psychological Inventory, and subjective measures such as the Rorschach Ink Blots, generally have not proven useful in predicting future dangerousness.

iii) Measures used to predict violent recidivism
The Statistical Information on Recidivism (SIR) scale
There are two important limitations to the SIR Scale as a predictor of violent recidivism. First, it was designed to predict recidivism generally, not violent recidivism. Indeed, as Nuffield herself acknowledges, the SIR Scale is not a predictor of violence. Second, as Nuffield also acknowledges, the SIR Scale has been found to be unsuccessful in predicting recidivism among aboriginal people and women. Therefore, caution should be taken in applying it to other identifiable subgroups.

The Psychopathy Checklist (PCL) and Psychopathy Checklist - Revised (PCL-R)
Beginning in the early 1980s, Dr. Robert Hare, several of his colleagues and scores of other researchers have performed several studies investigating the relationship between psychopathy and institutional and criminal behaviour, treatment efficacy and recidivism (see Hare and Hart, 1993, for a review). Hart, Kropp and Hare ( 1988) administered the PCL to 231 male inmates prior to their conditional release from prison. As Appendix A illustrates, the PCL was able to rather reliably predict the outcome over time. These results have been supported by others (e.g., Harris, Rice & Cormier, 1991; Serin, 1991; Serin, Peters & Barbaree, 1990).

Interestingly, to take one of the studies as an example, even though the PCL was more accurate than any other measure in the prediction of violent recidivism, the rate of accuracy was still 78% for psychopaths, who had a high rate of recidivism (77%) - almost four times higher than the recidivism rate for other releases (21 %). In a recent study, Dr. Ralph Serin compared several instruments that have been found to be useful in predicting recidivism (e.g., the Salient Factors scale used in U.S. parole decisions, the SIR Scale, the PCL, etc.). All of the measures predicted general recidivism well above chance, however, the only measure that successfully predicted violent recidivism was the PCL.

iv) The limitations of measures used to predict violent recidivism
Regardless of the effectiveness of any prediction instrument, no prediction instrument can ever be completely accurate in determining whether a particular individual will recidivate. The reason for this is rather straightforward. Prediction instruments for recidivism are based on the extent to which factors successfully predict who will or will not recidivate across a sample of offenders. Just as with any sample, there will be many individual differences across people. Some offenders who do not meet the "high risk" profile criteria will recidivate, while some of those who do meet the "high risk" profile criteria will not recidivate. Therefore, it is always important for decision makers to consider each case separately, on its own merits, prior to making a decision about whether the offender may recidivate violently.

In the final part of this article, we will highlight and briefly discuss those factors that have been found to be at least somewhat effective in helping predict dangerous behaviour.

v) Predictor variables/factors of dangerous behaviour
As the information reviewed so far shows, there are a number of important cautions that must be kept in mind when attempting to predict dangerousness. The prediction of dangerousness is fraught with difficulty, largely because of the low base rate of dangerousness and violence among the general population and among inmates released from prison. In addition, both clinical and actuarial approaches to the prediction of dangerousness have proven to be less than perfect, often resulting in unacceptable levels of false positives and false negatives. Finally, even the efficacy of the more promising predictive measures, such as the PCL, is limited by false positives and false negatives.

Generally speaking, actuarial instruments clearly have been proven to be more accurate in predicting recidivism than clinical judgments alone. However, as Serin and Barbaree (1993) point out, "with the high proportion of violent offenders in the federal corrections system, simply relying on a past history of violence to make judgments about future violence would result in an unacceptably high false-positive error rate" (p. 22).

Indeed, although the CCRA specifically requires Parole Board members to decide whether "the offender, if released, is likely to commit an offence involving violence" (Accelerated Parole Reviews, s.126[2]), ^{Footnote 4}  the measures and instruments developed to date simply predict who will fail or succeed on parole generally - not who will fail as a result of committing an additional violent offence.

This sad state of affairs might lead one to throw up one's arms and say, "it can't be done!" However, statutes and practical needs require courts, parole boards and review boards to make decisions that often hinge on the extent to which the decision maker believes that the individual will be dangerous to others when (and if) released. In its preliminary report of April 1993, the Solicitor General's Working Group on High Risk Offenders concluded that:

Thus, in the final, and arguably most important section of this article, we turn to a delineation of some of the variables that may be useful in determining whether any given individual will be "dangerous" at some point in the future and under some circumstances. Although some of the factors discussed undoubtedly overlap with items from the SIR Scale and the PCL, specific consideration of the factors identified below should help Parole Board embers determine the likelihood that an offender will recidivate violently if released.

vi) The HCR-20
Some of the research reviewed and noted in this article has actually served a most important purpose - that being the identification of a variety of variables that have been found to reliably predict dangerous and violent behaviour by some researchers, for some individuals, under some circumstances. In an ongoing project, Professor Christopher Webster has been developing a scheme for the assessment of dangerousness and risk (Webster & Eaves, 1993). This scheme is known as the HCR 20. Appendix B sets out the overall organization of the HCR 20.

The HCR-20 is really an acronym for "Historical," "Clinical" and "Risk" variables that may be useful in assessing dangerousness and violence. The historical predictors include those variables that are "fixed" or "static." In this sense, they may be considered variables that are based on an inmate's past or historical situations and functioning. The clinical variables are those factors that are current and are based on some up-to-date assessment of relevant clinical material obtained from health and mental health professionals. Finally, the risk variables are intended to aid decision makers in determining how well an individual may manage risk in different contexts upon release. Thus, the HCR-20 combines actuarial, clinical and risk management factors.

We should emphasize that because the HCR-20 is an untested scheme, its value at this early stage lies not in its proven reliability, but in its sound theoretical and conceptual construction. It is based on the principles first espoused by Monahan (1981) that have really driven much of the work in the prediction of dangerousness over the past decade.

To this extent, the HCR-20 combines actuarial, clinical and risk management factors that address historical, immediate and future considerations for the individual being assessed. Thus, we present the elements of the HCR-20 more for their conceptual appeal, and for the extent to which the elements are supported by empirical research, than for the proven reliability of the scheme as a whole.

Each of the three phases of the HCR-20, and the variables that comprise them, will be discussed briefly below. For a detailed description of the HCR-20, see Webster and Eaves (1993).

Phase I: Historical/static/actuarial predictors
(1) Previous violence
The best known fact concerning the prediction of violence is simply that one of the best predictors of future violent behaviour is a history of past violent behaviour (Monahan, 1981; Webster & Eaves, 1993). Unfortunately, given the large percentage of federal inmates with some history of violent offences or violent behaviour, this fact is not particularly helpful to Parole Board members.

Nonetheless, the probability of a person engaging in future violent behaviour increases with each prior incident of such behaviour. Therefore, the stronger the evidence of a history of violent behaviour, the more likely the offender will engage in further dangerous behaviour - and be unsuccessful on parole.

(2) Age
Age relates to the prediction of future violence in two general ways. First, there appears to be an inverse relationship between age and future violence. That is, the younger the offender, the more likely that offender is to engage in violent behaviour in the future.

Second, the younger the offender was when he or she first engaged in violent behaviour, the more likely that person is to be - and to continue to be - a violent offender. Of course, the difficulty with considering age as a predictor of future violence is that most offenders are quite young. Therefore, particular attention should be paid to the offender's historical pattern of violent behaviour. Generally speaking, a pattern of frequent, very violent behaviour may be more predictive of future violence than occasional violent behaviour.

(3) Lifestyle stability
Offenders who have more solid employment histories, family ties and social support systems tend to be at lower risk for parole violations. However, it is less clear that stable lifestyles are directly linked to lower risk for violent behaviour.

(4) Alcohol or drug abuse
There is a general relationship between substance abuse and violent behaviour. When attempting to predict future violent or dangerous behaviour, it is especially important to consider the extent to which alcohol or drug abuse relates to the offender's history of violence. Violence itself is difficult enough to control. It is not hard to imagine how difficult it is for offenders to control both violent tendencies and substance abuse problems.

(5) Mental disorder
We do not have time to recount the long debate over whether — and to what extent — mental disorder is predictive of future violence. For our purposes, it is important to note that, in recent years, the research appears to show that mental disorder may be somewhat predictive of an increased risk for violence. However, it is also important to emphasize that, while personality disorders in general and psychopathy in particular seem to be related to an increase in the risk of violence, other mental disorders (such as schizophrenia) may actually be inversely related to violence. Suffice it to say that most mentally ill people are not at risk for violence and that, at best, only a weak relationship exists between mental disorder and violence. Therefore, when trying to determine whether an individual with a history of mental disorder may be violent in the future, one should consider any prior relationship between violence and mental illness, as well as the social supports that are in place for community placement.

(6) Psychopathy
As noted above, the PCL seems somewhat promising for predicting who may be at risk for violence on parole. This is not particularly surprising, given that many of the items on the PCL overlap with the factors we know to be related to increased risk for dangerous behaviour. Nonetheless, people who score "high" on psychopathy, as defined by the PCL, are at significant risk for violent recidivism.

(7) School maladjustment
Two factors concerning an individual's adjustment in elementary school are related to an increased risk for violent behaviour: 1) academic ability and school accomplishments; and 2) classroom conduct and general adjustment to school. Unfortunately, these factors will be difficult to assess accurately, especially that of behaviour in elementary school. Nonetheless, these factors have been found to be strongly related to violent failure on parole.

(8) Personality disorder
As noted previously, personality disorders have been found to relate to an increased risk for violent behaviour. This is especially true for antisocial personality disorder (psychopathy) and borderline personality disorders.

(9) Prior release failure
Given that the best predictor of future violent behaviour is a history of previous violent behaviour, it should not be terribly surprising to realize that failure on prior conditional release — especially violent failure — is a reliable predictor of failure on future conditional release.

(10) Parental separation
Related to a stable lifestyle and school adjustment, offenders who were separated from their parents prior to the age of 16 have been found to be at an increased risk for future violent behaviour. This is only true in cases where the individual was separated from parents for reasons other than death.

Phase 2: Clinical/present variables
(11) Insight
In the HCR-20, Webster and Eaves (1993) relate this factor to an individual's insight into his or her mental disorder. Here, this factor relates more generally to the offender's acceptance of responsibility for his or her criminal behaviour and the extent to which the offender appreciates the significance of the criminal behaviour. Therefore, it is important that the offender not try to place the blame for his or her criminal behaviour on society, the criminal justice system or other external causes of behaviour.

(12) Attitudes
An attempt should be made to determine whether the offender's attitudes are pervasively pro- or anti-social. Attention should be paid to the extent to which the offender is hostile, impulsive, guarded, inconsistent, irresponsible, aggressive, etc. Because we are concerned primarily with risk for violent behaviour here, it will be important to consider the offender's attitudes toward prior violent behaviour. Is the offender sorrowful and regretful? Or, does the offender minimize and deny the severity and quality of his or her acts?

(13) Symptoms
In contrast to Webster and Eaves' (1993) focus on psychiatric symptomatology, we are concerned here with "symptoms" of unusually violent or aggressive behaviour or ideation. Has the offender engaged in particularly heinous behaviour? Does the offender have violent fantasies? Is the offender's violence related in any way to some cognitive or perceptual deficiencies?

(14) Emotional stability
Emotional instability appears to be a clinical risk factor for violence. Therefore, it is important to assess the extent to which the offender is emotionally stable or secure. Furthermore, given the importance of relating risk factors back to the offender's history of violent behaviour, particular attention must be paid to the extent to which a pattern of emotional instability is associated with the offender's previous violent behaviour.

(15) Treatability
This factor relates to the extent to which offenders respond to past and present attempts at rehabilitation. The offender's ability to try to use correctional programs to overcome the problems raised by his or her various risk factors (e.g., substance abuse, attitudes) should be carefully assessed, given the importance of those risk factors.

Phase 3:Risk/future factors
(16) Plan feasibility
If an offender has previously engaged in violent behaviour and nothing has changed in his or her life, there is no reason to believe that the offender will not engage in similar behaviour in the future. Therefore, careful attention must be paid to the offender's discharge plan. Is it suitable, safe, realistic — and tailored to meet the offender's specific needs (e.g., substance abuse, emotional instability, etc.)? Without careful planning designed to reduce risk factors and increase support systems, the offender's discharge plan may not be sufficient to adequately decrease the likelihood of failure on parole.

(17) Access
Following from the information in point 16 above, offenders' risk for violence upon release will increase to the extent that they have access to the people or substances that make them at risk for violence. Will the offender have access to possible victims? Weapons? Accomplices? Will the offender be at risk for substance abuse again — has the offender made plans to control his or her abuse problems?

(18) Support
It is extremely important that offenders have a support system in place that will help them decrease their risk for violence. The support system may include, but is not limited to, a partner, family and law-abiding friends. In addition, it is important for the offender to have access to the professional support needed to survive (e.g., self-help groups, psychological care, employment counselling).

(19) Compliance
All the planning and support systems in the world will not be enough if the offender does not comply with the supports or aids that have been set in place for him or her. How can you tell whether an offender will be compliant with the discharge plans that have been arranged? This is almost as difficult to answer as the basic question of whether the offender will engage in violent behaviour. Is the offender apparently sincere about the plans that have been arranged? Has the offender worked to change his or her behaviour and taken advantage of available programs while incarcerated? Is the offender's discharge plan destined to fail (i.e., is it unrealistic, does it place the offender at possible risk for factors that may trigger violence)?

(20) Stress
Stress factors increase an already violent person's risk for violent behaviour. While some stress factors are relatively easy to predict (e.g., employment, availability of needed services), others are difficult or impossible to predict (e.g., death of a support person). Nonetheless, a careful consideration of the relationship between stress and violence, as well as an accounting of the likely stress factors in the offender's life, will help further determine the offender's risk for violence on parole.

Scoring
The information from the HCR-20 is provided not as a psychometric tool, but as a guide to the factors that have been found to be reliable predictors of dangerousness and violence. Therefore, rather than trying to calculate an overall score, one should consider the extent to which the factors described pertain to a specific inmate. In so doing, it is important to realize that not all items are equally important for each inmate. For example, school maladjustment, or the fact that an inmate's parents were separated, may not be particularly useful information in some cases. In other cases, however, such information may be useful to understand the overall pattern of an inmate's emotional and behavioural problems.

i) General comments and the definition of dangerousness

• The prediction of dangerousness is a difficult task.
• The term "danger" means that someone is at risk for harm. "Dangerousness" is used to describe some act, situation or person that places people at risk for harm.
• The prediction of dangerousness requires the consideration of detailed information concerning: I) the nature of the dangerous acts to which a person may be disposed; and 2) the risk conditions or factors that might dispose the person to act in a dangerous manner that places others at risk for harm.

ii) Caveats concerning the prediction of dangerousness
Rather than mentioning that a person is, or is not, dangerous, it is essential to specify, with as much detail as possible, the parameters within which an individual may be dangerous, including the type of dangerous or threatening behaviour, the type of victim, the time frame within which the dangerous behaviour may occur, etc.

• It is important to consider the validity of the information upon which a prediction of dangerousness is being made.
• Because many myths surround the prediction of dangerousness, particular care must be taken to refrain from relying on relationships between risk factors and dangerousness without having good empirical support that the relationship reliably exists.
• Base rates of violent parole failures are very important considerations for the prediction of dangerousness. The lower the base rate of violent behaviour, the more difficult the prediction of future dangerousness.
• Cultural differences between the people making predictions of dangerousness and the inmates themselves may contribute to errors in determining who actually is dangerous. Similarly, some measures or instruments that have been developed to predict recidivism are not valid with some minority populations.

iii) Measures used to predict violent recidivism
No single psychometric instrument has been developed that reliably and accurately predicts violent recidivism or parole failure. However, the PCL shows some promise in this area.

iv) Predictor variables/factors of dangerous behaviour

• In order to make predictions of future dangerous behaviour as accurate as possible, special attention must be paid to several historical, clinical and risk management factors.
• When considering the factors related to dangerous behaviour, one must also carefully consider the extent to which the factor relates to the particular offender's violent behaviour and the extent to which the risk factor has been ameliorated or alleviated.

Chapter 15 Appendix A, B

Convit, A., Jaeger, J., Lin, S.P., Meisner, M. (St.. J. Volavka. (1988). "Predicting assaultiveness in psychiatric inpatients: A pilot study," Hospital and Community Psychiatry, 39, 429-434.

Hall, G.C.N. (1988). "Criminal behaviour as a function of clinical and actuarial variables in a sex offender population," Journal of Consulting and Clinical Psychology, 56, 773-775.

Hare, RD. & S.D. Hart. (1993). "Psychopathy, mental disorder and crime," in S. Hodgins (Ed.), Mental Disorder and Crime. Newbury Park, CA: Sage.

Hart, S.D., Kropp, P.R. & RD. Hare. (1988). "Performance of male psychopaths following conditional release from prison," Journal of Consulting and Clinical Psychology, 56, 227-232.

Harris, G.T., Rice, M.E. & C.A. Cormier. (1991). "Psychopathy and violent recidivism," Law and Human Behaviour, 15, 625-637.

Janofsly, J.S., Spears, S. & D.N. Neubauer. (1988). "Psychiatrists' accuracy in predicting violent behaviour on an inpatient unit," Hospital and Community Psychiatry, 39, 1090-1094.

Klassen, D. & W.A. O'Connor. (1988a). "Predicting violence in schizophrenic and nonschizophrenic patients: A prospective study," Journal of Community Psychology, 16, 217-227.

Klassen, D. & W.A. O'Connor. (1988b). "A prospective study of predictors of violence in adult male mental health admissions," Law and Human Behaviour, 12, 143-158.

Klassen, D. & W.A. O'Connor. (1990). "Assessing the risk of violence in released mental patients: A cross-validation study," Psychological Assessment: A Journal of Consulting and Clinical Psychology, 1, 75-81.

Melton, G.B., Petrila, J., Poythress, N.G. & C. Slobogin. (1987). Psychological Evaluations for the Courts: A Handbook for Mental Health Professionals. New York: The Guilford Press.

Monahan, J. (1981). The Clinical Prediction of Violent Behaviour. Rockville, MD: National Institute of Mental Health.

Monahan, J. (1991). "Mental disorder and violent behaviour: Perceptions and evidence," American Psychologist, 47, 511-521.

Mulvey, E.P. & C.W. Lidz. ( 1993). "Measuring patient violence in dangerousness research," Law and Human Behaviour, 17, 277-288.

Otto, R.K. ( 1992). "Prediction of dangerous behaviour: A review and analysis of "second generation" research," Forensic Reports, 5, 103-133.

Rofman, E.S., Askinazi, C. & E. Fant. (1980). "The prediction of dangerous behaviour in emergency civil commitment," American Journal of Psychiatry, 137, 1061-1064.

Rogers, R. (Ed.). (1988). Clinical Assessment of Malingering and Deception. New York: Guilford.

Sepejak, D., Menzies, R.J., Webster, C.D. & F.A.S. Jensen. (1983). Bulletin of the American Academy of Psychiatry and Law, 11, 171-181.

Serin, R. (1991). "Psychopathy and violence in criminals," Journal of Interpersonal Violence, 6, 423-431.

Serin, R. & H.E. Barbaree. (1993). "Decision issues in risk assessment," Forum on Corrections Research, 5(2), 22-25.

Serin, R., Peters, R.D. & H.E. Barbaree. (1990). "Predictors of psychopathy and release outcome in a criminal population," Psychological Assessment: A Journal of Consulting and Clinical Psychology, 2, 419-422.

Webster, C.D. & D. Eaves. (1993). The HCR-20: A Scheme for the Assessment of Dangerousness and Risk. Mental Health, Law, and Policy Institute, Department of Psychology, Simon Fraser University.

Working Group on High Risk Offenders. (1993). Report of Preliminary Recommendations by the Working Group on High Risk Offenders, Ottawa: Solicitor General of Canada.