Notes on the Clinical Assessment of Dangerousness in Offender Populations

David A. Cohen, M.A. Division of Forensic Psychiatry, Be’er Ya’akov Center for Mental Health P.O.B. 16, Ayalon, Prison, Ramlah, Israel

Mental health professionals are being called upon with increasing frequency to provide courts with dispositions regarding the dangerousness of violent and sex offenders. In certain cases, professionals are required to spontaneously report their impression that a given patient (whether inpatient or outpatient) is dangerous, and may be held liable for damages if they do not (Gage, 1990; Simon, 1990;Tarasoff v. Regents). The Judiciary demands, and expects to be provided with such assessments in parole hearings, sentencing and sexual psychopath hearings and insanity pleas Halleck, 1986; Pollock. 1990). Despite protests made by mental health professionals throughout the 1980’s and early 1990’s that dangerous behavior could not be predicted, recent literature suggests that if certain basic rules are followed clinicians can indeed accurately predict dangerousness in certain situations (Apperson, Mulvey & Lidz, 1993; Otto, 1994 Quinsey, 1995; Serin & Amos, 1995). This article reviews these rules, and suggests a method for the clinical evaluation of dangerousness using a semi-structured clinical interview. Several brief case histories will be provided to demonstrate the method’s utility.
Through the 1970’s and early 1980’s mental health professionals were interested primarily in the “prediction” of dangerousness in violent mental patients. These predictions were based on subjective clinical interviews which concentrated on one or two variables- usually type of offense and psychiatric diagnosis. Some authors were of the opinion that these predictions were valid some of the time and in the short term- maybe- (e.g. Monhan, 1981), while others were even less optimistic and stated flat out that psychiatrists could not predict dangerousness at all (e.g. Steadman & Cocozza, 1980). Research during this period (later called the “first generation”) consisted almost exclusively of natural history experiments, in which recidivism of individuals released against experts’ advice were assessed. Monhan (1984) suggested that “second generation thinking” should concentrate on the development of actuarial techniques, including the incorporation of clinical material into actuarial tables, studies that vary the factors used in making predictive decisions, to include situational items, and studies that vary the populations upon which predictive “technology” is brought to bear, including short term community studies (Monhan, 1984, p.13). Unfortunately, most of the “second generation” literature consisted of arguments over whether or not schizophrenia was related to violent behavior (e.g. Monhan, 1992, Rice & Harris, 1992), a question which in my mind it is still difficult to answer. Other second generation studies concentrated on the development of actuarial tables based on “static” variables such as type of crime, age and sex of victim, school records, age of onset of criminal behavior, etc. Such attempts did not fare much better than the first generation clinical trials (Hanson & Bussiere, 1996). Hanson & Bussiere, like Monhan before them, called for the inclusion of clinical “dynamic” factors in actuarial tables. I propose that “third generation” thinking must concentrate on the development of standards for the clinical prediction of dangerousness, based on a standardized conditional model, which is geared to the assessment of the relative probability of future violence in given situations. Such a model has been proposed by Mulvey & Lidz (1995), and is similar to the model which I have been developing in Israel and which is described here. The proposed model for “phenomenological assessment” is more limited in purpose than the situational model proposed by Mulvey & Lidz. In Israel we have no Tarassof type laws, and had no need for a model which would predict dangerousness in individuals who have never before exhibited violent behavior- something which Mulvey & Lidz’ model is designed to do. My model was developed to facilitate my work with the assessment of dangerous offenders for the Israel Prison Service and Parole Boards, and is only specific enough to assess dangerousness in individuals who have exhibited previous dangerous behavior- and who are willing to cooperate in a clinical interview which may span many hours and many sessions.

Theoretical Basis for Proposed Model

Prediction vs. Assessment

One of the first ‘breakthroughs’ in the development of a reliable method of prediction was the realization that the term ‘prediction’ itself was problematic (Monhan, 1988; McNiel & Binder, 1991). The accuracy of black and white dichotomous predictions of “dangerous” / “not dangerous” is hard to prove, and such predictions lump together patients who may need different therapeutic or legal approaches. Making relative assessments of probable dangerousness correct for clinicians’ errors which may be based on overconfidence or on the ignoring of base rates. The method also lends itself more to research, and emphasizes possible treatment courses by forcing the clinician to consider relative levels of dangerousness in different possible settings (Monhan, 1988; McNeil & Binder, 1991). It is also usefull in aiding the expert in making better reccomendations for parole, etc by identifying specific frameworks and restrictions which will guarantee that the released offender will not engage in further dangerous behavior.

Multi-Variable Assessment

Another important “breakthrough” in dangerousness assessment came when clinicians realized that dangerousness (which can be loosely defined as the probability that one person will cause harm to another or to himself), like other human behavior, is a complex concept and cannot be “predicted” based on any single predictor variable. As early as 1976, Megargee (1976) warned that both personality and situational factors must be taken into account in the assessment of dangerousness, and even provided detailed instructions for decision making based on this idea. Monhan and others (Hepworth, 1985; Monhan, 1988; Monhan & Klassen, 1982, Nachshon, 1982; Steadman & Ribner, 1982) have also stressed the importance of situational factors. Today, it is accepted that violent behavior, and therefore dangerousness, is a complex behavior which, like other behaviors is determined by biological, psychological and sociological factors (Bender, 1991, Eron, 1994; Litwack, 1994; Pollock, 1990; Quinsey, 1995; Saddler & Huglus, 1992), and it goes without saying that all such factors must be assessed in order to produce an accountable assessment of dangerousness.

Projective Nature of Assessment

Another rather obvious rule in the assessment of dangerousness is that dangerousness cannot be predicted in the absence of an established pattern of violence (Litwack, 1994; Hall, 1987; Kozol, 1982; Pollock, 1990). Clinicians must be able to review not only the criminal records of the person being assessed, but also be able to understand the personality and situational factors which led to past violent behaviors, and be able to understand how these factors may influence future violent behaviors. In this case “pattern” entails at least two previous behaviors similar to the one being assessed. A one time offense does not necessarily constitute a pattern of behavior or a criminal lifestyle. Therefore it is important to obtain information from legal or social services, or even from family members regarding previous behaviors which may not have been reported, or which may have been reported but did not result in conviction or incarceration. When looking for a “pattern” of violence, one must elicit information on biological, psychological, and situational factors which may have served to either facilitate or inhibit violence in the past. The expert must be able to describe, with at least some accuract and conviction, both to himself and to the Court, not only why the offender did what he did, but why he did it when he did, why he didn’t do something else (e.g. if a rapist suffers from no identifiable paraphilia, why did he rape and not just beat his victim) and under what conditions he would be likely to do it again.

Situational Assessment

It is important to keep in mind that assessments of dangerousness are situational. For example, Israeli psychiatrists are often called upon to assess dangerousness in various familial or hospital settings. Clinical criminologists working in prisons are called upon to assess dangerousness of prisoners before furlough or before parole. Each assessment is made in accordance with a specific request from the court or the prison service. For example, when asked to assess dangerousness of a sex offender, the assessment will refer only to sexual dangerousness, not to other types of dangerous behavior. The conclusions of assessments made for behavior in a given situation may not be the same as those given in another. For example, a prisoner may be assessed as not dangerous for a 24 hour furlough from prison, during which he must sign in twice daily at the local police station, but he may indeed be found to be to dangerous to be granted parole.

Suggested Formula for the Assessment of Dangerousness

The four principles mentioned above, relative probability, multi- dimensional assessment, the need to establish prior patterns of dangerous behavior, and the situational specificity of the assessment form a basis for valid, reliable assessments of dangerousness, and the basic rules from which all others follow. Based on these rules, we can propose a general rule for assessments of dangerousness: The probability of future dangerous behavior is based on the similarity of the situation for which the patient is being evaluated to situations in which he was violent in the past. “Situation” is used as a generic term referring to biological, psychological and environmental factors. This rule can be expressed in the simple formula:


Where P(D2) is the probability of future dangerous behavior (which specific behavior varies from situation to situation and from offender to offender). The numbers 1 and 2 refer to the past (when the crime was committed) and the present and future respectively. B2, P2 & E2 are biological psychological and environmental factors valid at the time of the examination, and B1, P1 & E1 are these same factors at times in the past when dangerous behavior was observed. The greater the similarity between the past and the present, the closer P(D2) is to 0 and the greater the probability that the dangerous behavior will repeat itself.

Theoretically, the factors B, P & E may be divided into facilitating and inhibiting factors. Facilitating factors for various types of violent behavior include mental illness (Monhan, 1992), Organic disorders, (Doerr & Carlin, 1991; Hales & Yudofsky, 1987, Nachshon, 1990), inborn sexual tendencies (Byne & Parsons, 1993), object relations (Kernberg, 1992), especially borderline pathology (Hart, Dutton & Newlove, 1993; Dutton, 1994), psychopathy or antisocial personality disorder, (Serin & Amos, 1995; Tiihonen & Hakola, 1994) history of childhood abuse (Luntz & Widom, 1994; McCormack, et al., 1992), and familial discord or family situation (Barnett, Fagan & Booker, 1991; Freund, Watson & Dickey, 1992). Inhibiting factors include controls such as familial support, probation, past or continuing medication or psychotherapy, etc. (Hepworth, 1985; Reckless, 1967). However, all these factors must be assessed within the context mentioned above. If it is found that a patient was never violent in the past despite serious borderline pathology and an unstable marital relationship, there is no reason to assume he will be violent in the future. If it is found that a patient has committed several sex crimes in the past despite the apparent lack of sexual pathology, despite the fact that he has a happy marriage, and despite the fact that he was on probation, there is every reason to suspect that he will commit similar actions in the future, even if all these supposedly countervailing factors continue to be present.

Although it may seem that it is impossible to elicit all the information suggested above, it is actually relatively easy. Court records are readily available to clinicians (at least in Israel, where the author practices), and community based social workers can provide excellent information on past behavior. A structural clinical interview (Kernberg, 1981, 1984) designed to assess borderline pathology and object relations is an excellent framework for assessing such dangerousness. The traditional structural interview starts by asking the patient how he feels, what his complaints are and why he has come for treatment. The forensic interview should start by explaining the purpose of the examination, the patient’s rights regarding the examination and possible outcome. This brief encounter usually elicits useful information not only regarding the patient’s ability to give informed consent to the examination, but also regarding his basic level of object relations. For example, in certain patients my explanation that the patient has the right to appeal if he does not like my opinion may elicit a blatant paranoid reaction and panic, such as “What- have you been talking to my wife? What (expletive) has she tried to tell you about me?” or it may elicit a more manipulative, psychopathic reaction such as “but you’re such a nice person- and I can tell that your religious- why would you write anything bad about me?”. Such opening statement also elicits information about the patient’s attitude towards his crime. For example, when I inform sex offenders that they are being examined because parole boards require mental health assessment of all sex offenders, replies range anywhere from the prognostically grave “well you can forget it, because I didn’t do anything” to the more hopeful “yes it was a terrible crime which I have discussed may times with my social worker”. These responses indicate the direction of the interview, in that together with objective information obtained from outside sources they provide bases for clarification and confrontation which are integral parts of the structural interview.

Clinical Examples

The following examples will illustrate the above points:

1. Rafi, convicted for the repeated beating of his wife and children, was interviewed before parole in late 1992. He was calm and polite during the interview. He pointed out that he had admitted his crimes in court on the advice of his lawyer but that he was really innocent, and repeatedly characterized himself as a model father and husband. When it was pointed out that this did not exactly coincide with the crimes which he admitted having committed, he responded, in an excited, loud voice, that my comments proved that I didn’t know anything, that he gave everything for his family, who richly deserved the beatings he handed out. This reaction started a train of speech in which clear narcissistic traits, grandiosity, primitive object relations and rage in reaction to criticism became evident. However, it also became apparent during the interview that Rafi had been undergoing an acute family crisis, which led to episodic alcohol use and subsequent unemployment, immediately prior to and during the time he administered the beatings. There was no knowledge of family violence before the onset of the crisis. During the 12 months of his imprisonment the crisis had been resolved, he had not used alcohol or been violent during his furloughs from prison, had exhibited ability to work up to 10 hours a day on regular basis, and had been promised work as a taxi driver at a taxi station near his home. Despite his personality disorder, he was recommended for parole, which was granted. However, he continued to drink, and his wife threatened to leave him. He was re-arrested several months later and is still in prison. 2. Vladimir, a chronic alcoholic in prison for beating his wife, was interviewed in late 1992, also for parole. During the interview he stressed that he could not understand what the fuss was about, and why anyone would want to stop his early release. He explained that “God gave us two hands and one mouth” and that therefore hands were more useful in “negotiations” with his wife than was his mouth. His wife richly deserved what she got, and would “get more” unless she changed. He denied ever using alcohol. Vladimir was diagnosed as suffering from a narcissistic personality disorder (malignant narcissism- Kernberg, 1992). He was granted parole despite my negative recommendation. Five months later he was again arrested for beating his wife while intoxicated. He has since been released, after serving a 23 month sentence, and was re-arrested several months later. As of Aug. 1996 he is still in prison. 3. Albert, a recidivist sex offender diagnosed as suffering from Schizophrenia and an unspecified paraphilia, underwent extensive treatment during his five year incarceration for rape. During that time there were no acute psychotic symptoms, but there was serious cognitive, affective and functional impairment. He received cognitive therapy (Perris, 1989), was allowed to work outside the prison (under supervision) and was granted home furloughs of up to 96 hours. He had limited but adequate (considering his impairment) insight into his crimes, was able to function at a low level in a regular prison ward, and travel to and from home unaccompanied. However, he could only do this while receiving high doses of medication on a regular basis. The prison psychiatrist and the author recommended parole only after a social worker certified that Albert had a part time job and a place to live with a family member, and after the psychiatric clinic in his home town guaranteed in writing that they would oversee his continued medication. He has been out of prison for six months, and is to our knowledge still receiving medication, and has not committed any further crimes.


Although these cases do not prove or disprove the validity of the method outlined above, they serve to illustrate the interaction between biological psychological and environmental factors and their relationship to past and future dangerous behavior. By adopting theoretically sound, accountable, methods of assessing dangerousness, clinicians can overcome both problems of loss of reliability inherent in totally subjective clinical evaluations and loss of sensitivity inherent in actuarial approaches. The method outlined here basically consists of projecting an individual’s patterns of past violence into the future, something which, if done properly, leaves little room for subjectivity or for bias, but concentrates wholly on the case at hand. Factors commonly known to be associated with violence, and which appear on actuarial tables (or would, if they existed) provide the clinician with a field of areas to be covered in the examination, but do not necessarily determine the final conclusions of the resultant disposition. It may be that the above ideas seem mundane or obvious, or simply an assemblage of old ideas. However, the clear statement of a theoretically based clinical method for decision making is the first step in the development of a truly universal, accountable methodology for assessing dangerousness in all types of offenders.


Apperson, L.J., Mulvey, E.P., & Lidz, C.W. (1993). “Short-term clinical prediction of assaultive behavior: Artifacts of research methods”. American Journal of Psychiatry, 150, 1374-1379.

Barefoot v. Estelle, 463 U.S. 880. (1983).

Barnett, O.W., Fagan, R.W., & Booker, J.M. (1992). “Hostility and stress as mediators of aggression in violent men”. Journal of Family Violence, 6, 217-241.

Bender, E.P. (1991). “Looking ahead: New psychiatry, old values”. American Journal of Psychiatry, 148, 1123-1129.

Byne, W., & Parsons, B. (1993). “Human sexual orientation: The biologic theories reappraised”. Archives of General Psychiatry, 50, 228-239.

Doerr, H.O., & Carlin, A.S. (1991). Forensic Neuropsychology: Legal and Scientific Bases. N.Y.: Guilford.

Dutton, D.G. (1994). “Behavioral and affective correlates of Borderline personality organization in wife assaulters”. International Journal of Law & Psychiatry, 17, 265-277.

Eron, L.D. (1994). “Theories of aggression: From drives to cognitions”. In: L.R. Huesmann (ed.) Aggressive Behavior: Current Perspectives. N.Y.: Plenum. Pp. 3-11.

Freund, K., Watson, R., & Dickey, R. (1992). “Sex offenses against female children perpetrated by men who are not pedophiles”. Journal of Sex Research, 29, 409-423.

Gage, B.C. (1990). “The duty to protect in inpatient psychiatry”. In: J.C. Beck, ed. Confidentiality Versus the Duty to Protect. Washington, D.C.: APA Press. Pp. 55-76.

Hales, R.E. & Yudofsky, S.C. (1987). American Psychiatric Press Textbook of Neuropsychiatry. Washington, D.C.: APA Press.

Halleck, S.L. (1986). The Mentally Disordered Offender. Rockville, MD: NIMH.

Hall, H.V. (1987). Violence Prediction. Guidelines for the Forensic Practitioner. Springfield, IL: Thomas.

Hanson, R.K. & Bussiere, M.T. (May, 1996) “sex offender risk predictions” Forum on Corrections Research.

Hart, S.D., Dutton, D.G., & Newlove, T. (1993). “The prevalence of personality disorder among wife assaulters”. Journal of Personality Disorders, 7, 329-341.

Hepworth, D. (1985). “Dangerousness and the mental health tribune”. In: D.P. Farrington & J. Gunn (eds.) Aggression and Dangerousness. N.Y.: Wiley. Pp. 155-183.

Kernberg, O.F. (1981). “Structural interviewing”. Psychiatric Clinics of North America, 4, 169-195.

Kernberg, O.F. (1984) . Severe Personality Disorders. New Haven CT.: Yale University Press.

Kernberg. O.F. (1992). Aggression in Personality Disorders and Perversions. New Haven CT.: Yale University Press.

Kozol, H.L. (1982). “Dangerousness in society and law”. University of Toronto Law Review, 13, 241-267.

Litwack, T.R. (1994). “Assessments of dangerousness: Legal, research, and clinical developments”. Administration and-Policy-in-Mental-Health, 21, 361-377.

Luntz, M.A., & Widom, C.S. (1994). “Antisocial personality disorder in abused and neglected children grown up”. American Journal of Psychiatry, 151, 670-674.

McCormack, A., Rokous, F.E., Hazelwood, R.R., Burgess, A. (1992). “An exploration of incest in the childhood development of serial rapists”. Journal of Family Violence, 7, 219-228.

McNiel, D.E., & Binder, R.E. (1991). “Clinical assessment of the risk of violence among psychiatric inpatients”. American Journal of Psychiatry, 148, 1317-1321.

Megargee, E.I. (1976). “The prediction of dangerous behavior”. Criminal Justice & Behavior, 3, 3-21.

Monhan, J. (1992). “Mental disorders & violent behavior: Perceptions & evidence”. American Psychologist, 47, 511-521.

Monhan, J. (1988). “Risk assessment of violence among the mentally disordered: Generating useful knowledge”. International Journal of Law & Psychiatry, 11, 249-257.

Monhan, J. (1984). “The prediction of violent behavior: Toward a second generation of theory and policy”. American Journal of Psychiatry, 141, 10-15.

Monhan, J. (1981). The Clinical Prediction of Violent Behavior. Washington, D.C.: GPO.

Monhan, J., & Klassen, D. (1982). “Situational approaches to understanding and predicting individual violent behavior”. In: M. Wolfgang & N. Weiner (eds.), Criminal Violence. Beverly Hills: Sage. Pp. 292-319. Mulvey, E.P., & Lidz C.W. (1995). “Conditional Prediction: A model for dangerousness to others in a new era”. International Journal of Psychiatry & Law, 18, 129-143.

Nachshon, I. (1982). “Toward biosocial approaches in criminology”. Journal of Social Biological Structures, 5, 1-9.

Nachshon, I. (1990). “Neurological bases of crime, psychopathy, and aggression”. In: L. Ellis & R.E. Hoffman (eds.) Crime in Biological social and Moral Contexts. N.Y.: Basic Books.

Otto, R.K. (1994). “On the ability of mental health professionals to ‘predict dangerousness’: A commentary on interpretations of the ‘dangerousness literature'”. Law & Psychology Review, 18, 43-68.

Perris, C. (1989). Cognitive Therapy with Schizophrenic Patients. N.Y.: Guilford Press.

Pollock, N.L. (1990). “Accounting for predictions of dangerousness”. International Journal of Psychiatry and Law, 13, 207-215.

Quinsey, V.L. (1995). “The prediction & explanation of criminal violence”. International Journal of Psychiatry & Law, 18, 117-127.

Reckless, W.C. (1967). The Crime Problem. N.Y.: Appelton, Century, Crofts.

Rice, M.E., & Harris, G.T. (1992). “A comparison of criminal recidivism among schizophrenic and non- schizophrenic offenders”. International Journal of Law & Psychiatry, 15, 397-408.

Saddler, J.Z., & Huglus, Y.F. (1992). “Clinical problem solving and the biopsychosocial model”. American Journal of Psychiatry, 149, 1315-1323.

Simon, R.I. (1990). “The duty to protect in private practice”. In: J.C. Beck, ed. Confidentiality Versus the Duty to Protect. Washington, D.C.: APA Press. Pp. 23-42.

Steadman, H.J. & Cocozza, J.J. (1980). “The prediction of dangerousness- Baxtrom: A case study”. In: G. Cooke, ed. The Role of the Forensic Psychologist. Springfield, Il.: Thomas. Pp. 204-215.

Serin, R.C., & Amos, N.L. (1995). “The role of psychopathy in the assessment of dangerousness”. International Journal of Psychiatry & Law, 18, 231-238.

Steadman, H.J., & Ribner, S. (1982). “Life stress & violence among ex-mental patients.” Social Science in Medicine, 16, 1641-1647.

Tarasoff v. Regents of U. of California, 551 P2d 344, 17 Cal3d 425 (Cal Sup Ct 1976).

Tiihonen, J., & Hakola, P. (1994). “Psychiatric disorders and homicide recidivism”. American Journal of Psychiatry, 151, 436-438.

This article was prepared by the team at Criminal Profiling. This website began way back in 1999 - over 15 years of the latest Criminology news and updates. If you've found this article of interest, please do share and comment! We love all your views and opinions. If you have a story to tell, please do let us know.