The Assessment of Potential Threat and Future Prediction of Violence: A Second Look


Joseph A. Davis 1


Publication Release: Davis, J. A. (2001). The Assessment of Potential Threat: A Second Look. Journal of Police and Criminal Psychology. vol. 1, no. 1, pages 1-16. Published through the Society of Police and Criminal Psychology, San Marcos, Texas. Southwest Texas State University. Submitted January 2001 for peer review. Accepted April, 2001.


Keywords: threat assessment, dangerousness, prediction, violence, duty to warn,


Tarasoff, case management, assault.


Abstract:  The assessment of dangerousness is not a diagnosis, but rather a clinical impression based on the individual's past history of violence and many other factors. "With regard to the mental state, it has been noted that “the essence of dangerousness appears to be a paucity of concern for others'" (Roth, 1974). As a subjective opinion, predicting violence and future dangerousness bears the weight of much scrutiny and due diligence. Threat assessment evaluations have inherent social and legal ramifications, and the responsibility must be accepted with the knowledge that accuracy may not always be achieved. This paper discusses the various aspects of predicting future violence and dangerousness and the caveats that come with such tasks.

1.    The author, Joseph A. Davis is on the part-time faculty in the College of Sciences, Department of Psychology, California State University at San Diego and adjunct visiting faculty, USC – School of Public Policy, Center for the Administration of Justice  in Los Angeles. He is a senior partner with the TAP Group, Inc., the Threat Assessment and Prevention Group in San Diego, Los Angeles and Long Beach, California and a senior member of the Stalking Strike Force and Stalking Case Assessment Team (SCAT) with the San Diego District Attorney’s Office.. Over the past 17 years, Dr. Davis’s research and teaching interests have focused on public safety, clinical-forensic psychology, and traumatology. He is the editor of the book published by CRC Press (July 2001) titled Stalking Crimes and Victim Protection: Prevention, Intervention, Threat Assessment and Case Management. A subject matter expert in abnormal psychology, psychopathology, personality psychology, trauma psychology and psychology-law, Dr. Davis’ applied work focuses on violence in the workplace, threat assessment and risk management, trauma prevention, critical incident stress debriefing and intervention, PTSD, and the crime of stalking.

2.    Special Acknowledgement: Acknowledgements are provided to my former graduate student in psychology-law, Sandra Wallace, J.D. for her valuable assistance and support in providing a literature research for this article and subject matter.

Clinical Prediction of Violent Behavior

Analogous to the inexactitude by which the human brain may be understood, the compelling task of predicting violent behavior or dangerousness cannot be accomplished with empirical certainty. Behavior, in general, may be defined as an individual's overt response to his/her environment. Dangerousness is loosely defined as the potential to commit a physical act of violence upon another person. Since behavior itself is based upon unquantifiable variables (both external and internal), the capability of predicting behavior, violent or otherwise, is necessarily limited. Similarly, the term dangerousness used by clinicians in diverse settings is, in fact, a vague word for which a definition has not been codified or standardized for uniform interpretation.

            Violence, as a human predicament, continues to permeate our society. In the past few decades, victims and law enforcement professionals have increasingly demanded answers to the cause and deterrence of violence. While often chiding scholars of the mind (psychologists and psychiatrists) for their imprecise science, those involved in criminal justice and mental health nonetheless pass the responsibility of predicting dangerousness to those same practitioners.

            Clinicians, understandably, find themselves in a quandary. They must consider the consequences of relying on inconclusive data against the weight of professional ethics and duty. Clearly, more research and understanding of the violent mind and its predilections are imperative, as are legislatively enacted standards for optimum assessment by the clinician. This analysis shall explore the clinician's role in the prediction of violence, the concept of duty, as well as discuss the current roadmap to competency in this endeavor.

The Role of the Evaluator in Threat Assessment

Notwithstanding the imprecise methodology involved, the concept of dangerousness is well accepted in legal contexts. A determination of dangerousness constitutes the legal grounds by which an individual's liberties may be temporarily or permanently deprived. A security specialist, executive protection professional, behavioral scientist, criminologist, psychologist or psychiatrist may be called upon to render such a finding in any one of many areas in which he/she has expertise.

Mental Health and Related Settings                                                                               

Commitment to mental health facilities is, in most cases, predicated on a dangerousness factor (i.e., whether a person may be dangerous to himself or others). Likewise, discharges and passes must meet some criteria for potential violent behavior. The clinician's worst nightmare is, of course, to err on the side of liberality, and release a potentially dangerous individual to the community who does thereafter commit a violent act.

One well‑known case involved John Hinckley's requests for unsupervised leaves from St. Elizabeth’s (where he had been interned subsequent to acquittal by reason of insanity for the attempted assassination of then‑president Reagan). Dr. Glenn Miller of that institution determined that Hinckley was sufficiently responsible to merit the leaves. Failing to account for Hinckley's earlier correspondences with convicted killers, and other suspicious activities, Dr. Miller was pressured by public outcry to cancel Hinckley's unescorted leaves.

In an earlier California case, Phillip Jablonski was labelled potentially dangerous by a Loma Linda Veterans Administration Medical Center (VAMC) psychiatrist. The VAMC psychiatrist, however, found no legal basis for committing Jablonski who, in fact, refused hospitalization. A short time thereafter, Mr. Jablonski murdered his girlfriend who was preparing to move out of their apartment.

These and countless other less well‑known cases illustrate the growing need to accurately predict dangerousness and apply a standard for dealing with potentially violent persons. "Yet several influential studies conducted in the early 1970s indicate that mental health professionals might as well consult a medium or draw straws when it comes to predicting violent behavior" (Bower, 1984). One might ask whether a trustworthy predictive assessment model can be actually achieved due to the complex nature of the conduct being examined.

Prison and Related Correctional Environments

Predicting dangerousness of the criminal, as opposed to those persons suffering a mental impairment, appears to be an equally difficult and elusive task. Many correctional personnel, criminologists, psychologists and psychiatrist forsake caution in hopes of rehabilitating, heretofore, violent offenders. In these special instances regarding the correctional population, low recidivism may be viewed as tantamount to rehabilitation. One problem, of course, is the unreliability of recidivism measurement. Even if accurate data were available and assuming a low rate of recidivism, the total number of resultant violent crimes should serve to nullify justification for early paroles, work releases and furloughs.

The Patuxant Research Model

Maryland Patuxent Institution became notorious in the late 1980s for its rehabilitation through psychotherapy. Two cases in particular showcase the lack of skill employed by clinicians in predicting dangerousness when granting early releases and paroles. In one case, James Stavarakas, in jail for rape, fled a work furlough site and committed another rape. In another, a released murderer by the name of Willie Horton terrorized a Maryland couple. Subsequent to these incidents, Patuxent modified its liberal release programs; however, an increased ability to accurately predict future violence was not evidenced.

Academic and Research Methodology Environments

Although no unequivocal standard for predicting dangerousness has emerged, a common thread does exist. The literature supports the consensus that criminologists, psychiatrists, psychologists and clinical social workers look for repeated or recent violent behavior, or verbal threats. A February 1984 article of the American Journal of Psychiatry reported that psychiatrists in a particular survey considered patient hostility, agitation, previous assaultiveness and suspiciousness. Beyond these general observations, at least two comprehensive approaches have been developed in order to achieve more accuracy.

An Assessment Model of Predicting Future Violence

Through considerable research and experience, John Monahan, Ph.D., a professor of psychology and law at the University of Virginia who is affiliated with the Institute of Law, Psychiatry and Public Policy located on the grounds of Blue Ridge Memorial Hospital, devised a questionnaire designed to allow clinicians and threat evaluators a broader context from which to evaluate potential threats of violence and future dangerousness (Monahan, 1985).

Clarification of the task serves as the cornerstone of Monahan's model. Although seemingly elementary, the professional must discern whether one is really being asked to predict dangerousness (i.e., is it an issue in the case?). Dr. Robert Sadoff, writing for the APA Psychiatric News (1987) stated that 11 psychiatrists could not predict future violent behavior and that courts should not call upon them to do so" (Vatz, 1989).

In spite of this pronouncement, such determinations must be made frequently in matters of parole, probation, pardons, length of sentence, civil commitments and in numerous other contexts. Monahan urges a circumspect approach to prediction of violence.

Beyond the pivotal question of whether a prediction is being requested, one must then ask for what purpose is the determination being sought. Apparently, confusion exists at this preliminary juncture. For instance, many judges surveyed could not articulate the reasons for their requests for mental health examinations. Clinicians and other related examiners offer dangerousness predictions where one is often not requested and, further, fail to respond to issues of competency and responsibility which are the presenting legal issue at hand. Clearly, it is preferable not to undertake the task unless it is specified.

Prior to proceeding on a specific request, the clinician must then pursue an introspective analysis pertaining to his/her own competency and ethics. Not only must the clinician be confident of his/her understanding of relevant literature, he/she must consider whether a prediction would tend to promote social policy better left to the judiciary or legislature.

A clinician must thoroughly investigate the circumstances bringing to issue an individual's potential violent behavior. All parties to the incident should be questioned so as to eliminate possible inconsistencies. Such factors as provocation, for instance, might tend to dissuade against future violence.

As discussed earlier, obtaining an individual's complete history of previous violent behavior is crucial in determining dangerousness. Monahan advises that, in particular, one should note the frequency as well as whether the pattern seems to be escalating or decreasing in nature. Violence may be further analyzed according to type and location (i.e., domestic abuse, school or bar fights, arson, etc.).

Demographic Variables and Base Rate Behavior

"Statistically, the greatest accuracy is achieved by designating the smallest  

number of persons as likely to commit future violence" (Roth, 1974). Demographics provide a reliable and generally accessible guide for determining dangerousness based on statistical data, allowing the clinician to include or exclude a person from the confines of a group. Demographic characteristics to be considered are age, sex, I.Q., social class, education, race, residential and/or employment stability, and history of substance abuse. A clinician who understands the data will know that age, for instance, will affect one's determination. Data shows that violence peaks in the late teens and early 20’s; if the person being evaluated is 40, 50, or 60, this part of the analysis (in itself) would argue against future violence.

Comparing the base rate of violence of individuals in the same circumstance as the subject is another crucial device. In many cases the base rate may be obtained from published, material (i.e., hospital records, police reports, etc.). When base rate information is not available, "we must use our heads" (Monahan, 1985). It is often necessary to extrapolate by asking why the base rate of violence of a group similar to the subject should be higher than that of the population at large (Davis, Siota and Stewart, 1999).

Recognizing Psychological Stress and Accumulated Stressors in Assessment

Stress may be defined as an individual's responses and coping mechanisms relative to the demands of his/her environment. More stress occurs as the demands ratio increases. Monahan utilizes Dr. Raymond Novaco's two‑prong approach to further define stress (i.e., appraisal and expectations). Appraisal refers to cognitive interpretation of an event. Violent‑prone persons will often view a seemingly innocuous event as provocative or intentional. Similarly, expectations of such an individual may result in anger or other inappropriate emotional arousal when an outcome does not meet those expectations.

Three primary categories which should concern the clinician are family stressors, peer group stressors, and employment stressors. The clinician should carefully investigate appraisal and expectation patterns evolving from these areas, such as relationship frustrations, friendship disruptions, unemployment, and the like. One reason for working with a complete history of the person to be assessed is so that the clinician may compare prior events in which the subject has reacted violently to the event precipitating the evaluation. A complete history should include the primary areas of possible stressing contexts (Davis et al, 1999).

Likely Victims and Means to Victimize

When assessing a person’s future violent inclination, it is important to consider the potential victim pool. Typically, vulnerable groups such as the elderly or children may be potential targets in many violent crimes. Women are likely targets for crimes such as rape or other forms of sexual assault. A review of the subject's history may suggest a narrower group or, perhaps, identifiable targeted persons who are at risk.

The clinician needs to be aware of warning signs in the form of fantasy, threats or anger directed at a named person. An assessment of dangerousness coupled with foreseeability poses a genuine area of concern for the clinician.

Whether the subject may, in fact, carry out a threat or act upon the violent impetus depends upon the means available to facilitate the deed. Survivalist or martial arts training, military background, including expertise with guns, knives or bombs should be carefully noted.

Given the guide by Monahan (1985) as a reasonable approach to assessing dangerousness, the clinician having accepted this responsibility must thoroughly review the relevant background material for accuracy. See Table 1.

[Table 1]


Screening Questions for the Threat Assessment Evaluator


1.         What events precipitated the question of the person's potential for violence being raised, and in what context did these events take place?

2.         What are the person's relevant demographic characteristics?

3.         What is the person's history of violent behavior?

4.         What is the base rate of violent behavior among individuals of this person's background? 

5.         What are the sources of perceived stress in the person's background and current situation?

6.         What cognitive and affective factors indicate that the person may be predisposed to maladaptively cope with stress in threatening, violent acting out manner?

7.         What cognitive and affective person may be predisposed to nonviolent manner?

8.         How similar are the contexts violent coping mechanisms in which the person likely will factors indicate that the cope with stress in a in which the person has used the past to the contexts In function in the future?

9.         In particular, who are the likely (targeted) victims of the person's violent behavior? How available are they?

10.       What means does the person possess to commit acts of violence?

11.       Am I giving a balanced consideration to the factors indicating the absence of violent behaviors, as well as to the factors indicating its occurrence?

12.       Am I giving adequate attention to what I estimate as the base rate of violent behavior among persons similarly situated to the person being examined?

13.       Other significant clinical or forensic variables involved in this case?

SOURCE: Excerpted from Monahan, J. (1985) as published in Ewing, C. P. (ed.) (1985), Forensic Psychology, Psychiatry and the Law, Professional Resources, Florida.






The Assessment of Risk Potential

Primarily an approach to predicting dangerousness in those persons with mental

disorders, risk assessment is an effort to more effectively balance individual and societal rights. Since the public has long viewed the mentally disordered as unpredictable and violent prone, clinicians using the broad criteria of imminence, seriousness and frequency have been able to commit on those grounds. With those numbers constantly increasing (nationally, over 300,000 in 1980), mental and public health professionals seek a more effective means of balancing the rights of all involved.

This is seen as a step away from the legal concept of dangerousness to the decision‑making concept of risk. It is a shift from the one‑time prediction to a focus on the continuing management and treatment of the mentally ill. In theory, risk assessment is an approach whereby minimum standards of restriction of the patient will be based on a more reliable statistical foundation.

MacArthur Foundation on Mental Health and Law

In 1989, the John D. and Catherine T. MacArthur Foundation's Research Network on Mental Health and the Law comprised of 12 people from the areas of law, psychiatry, criminology, psychology and sociology studied and recommended the risk assessment method. Recognizing Monahan's conclusion that his and other previous methods were inadequate due to marginally available demographics, overly restrictive patient samples and lack of research coordination, the MacArthur Research Network sought to develop markers of risk based on focused research of a broader sampling.

Five important factors to be tested are as follows: amount and type of social support available; impulsiveness; reactions to provocation; empathy; and nature of delusions and hallucinations, if any. The Network relied on existing tests to assess certain responses, such as anger. Other research results deemed to be reliable were incorporated (e.g., those measuring impulsivity). The continuum aspect of the risk assessment approach considers the changeable nature of risk levels and the importance of risk management as an integral feature.

Ethical Considerations in Threat Assessment

As alluded to earlier, a clinician might decline participating in the determination process on issues which lend themselves to policy making. Cases requiring the clinician to recommend length of retention, for instance, should be relegated to decision makers where possible. Similarly, since dangerousness cannot be measured objectively, requests for the clinician to establish levels of violence necessary for commitment should be declined or such aspects severed from the assessment.

Another area of potential ethical conflict for the clinician lies in matters of confidentiality. The duty of confidentiality is expressed as an ethical mandate that requires the professional, i.e., physician, etc., to hold as sacrosanct all secrets divulged to him/her by the patient.

There are occasions in which the clinician is compelled to breach confidences in order to protect the patient or others. The American Psychiatric Association (APA) provides for an expected and condoned breach in the following situations: without intervention the individual will probably commit murder or suicide; the individual who is responsible for many others (such as an airline pilot) shows serious judgment impairment; in cases of dangerous or contagious disease; and firearm or knife wounds.

Naturally, arbitrary breaches would discourage confidences from being divulged and defeat the entire therapy process. "It is clearly recognized that the very practice of psychiatry vitally depends upon the reputation in the community that the psychiatrist will not tell" (Slovenko, 1960). However, in predicting dangerousness the conflict suggests, and perhaps dictates, resolution as set forth in Tarasoff v.  Regents of University of California (1976) 131 Cal. Rptr. 14.

The Landmark Case of Tarasoff and Beyond

This action revolves around the murder of a girl by an individual who had previously informed his therapist of his intention to kill her. Mr. Podar had been an outpatient at Cowell Memorial Hospital at the University of California at Berkeley when he informed Dr. Moore of his intentions toward an unnamed but identifiable girl, that being, Tatiana Tarasoff.

Tarasoff’s parents filed suit in the Superior Court of Alameda (California) alleging wrongful death due to negligence on the part of Dr. Moore and Cowell Memorial Hospital for failing to warn their daughter in advance of Mr. Podar’s malicious intentions. The police who had temporarily detained the man were also named defendants, but the case against them was later dismissed.

The complaint stated that when a therapist determines that a patient poses a potential violent threat to a known person or persons, a professional duty arises to use reasonable care in protecting the potential victim from the perpetrator. Liability was predicated on two grounds: failure to warn plaintiffs of the impending danger, and defendant's failure to effect Mr. Podar's confinement.

Issues of Prevention and Duty to Warn

Numerous government and civil code sections (e.g., Healing Arts and Institutions §46, Negligence §9, Law Enforcement Officers §17, etc.) contain language imposing a duty of law enforcement professionals and medical practitioners to warn a foreseeable victim of impending harm or danger. The concept of reasonable care is a basic principle upon which standards of professional conduct may be measured. It derived from a common law duty to act according to the standards imposed by one's profession, specifically, a duty to warn someone with whom a special relationship existed. That common law precept has continuously been regenerated through considerable case law.

In Tarasoff case, that relationship was clearly established as between Mr. Podar and Dr. Moore. However, the duty was not limited when only the therapist/patient relationship exists. In fact,

“there now seems to be sufficient authority to support the conclusion that by entering into a doctor-patient relationship the therapist becomes sufficiently involved to assume some responsibility for the safety, not only of the patient himself, but also of any third person whom the doctor knows to be threatened by the patient” (Fleming & Maximov, 1974).

Physicians, hospitals, and all those in the medical sciences or healing arts, as are other public institutions and professionals, are charged with a public interest. More often than not, these moral duties are perceived as legal duties as well.

Beyond Tarasoff on Dangerousness Prediction and Public Policy

            The Supreme Court reversed the lower court's ruling in favor of defendants and ruled instead that: When a therapist determines, or pursuant to the standards of his profession should determine, that his patient presents a serious danger of violence to another, he incurs an obligation to use reasonable care to protect the intended victim against such danger" (Tarasoff, 1976).

This case illustrates the trend to impose the additional duty of controlling the conduct of the potentially violent person. This is, as well, an extension of the common law concept. Courts are increasingly inclined to expand the definition of special relationship so that the two‑part duty prevails. This development, of course, greatly increases the responsibility when predicting dangerousness in that the clinician may have to consider the affirmative duty to warn and, ultimately, to control the conduct of another.

Loss of Personal Liberties

             Finally, the clinician must assess the serious consequences of predicting dangerousness as it may impact the deprivation of an individual's liberty. As mental health professionals become aware of the ever‑increasing responsibility of predicting dangerousness, including the expanded duties attached thereto, the risk of greater numbers being civilly committed is genuine. "Treatment in exchange for liberty is the logic" (Goldman, 1984).

             During therapy, patients are urged to verbalize their inner feelings and frustrations. "Since a frequent goal of treatment is to encourage the patient to discharge suppressed feelings, including aggression and even anger, therapy often involves a period of increased instability immediately preceding a breakthrough" (Fleming & Maximov, 1974). Thus, it is the very nature of privilege and confidentiality that may serve to support a finding of dangerousness, resulting in a person's involuntary confinement. Rights of the one in most cases do not prevail against rights of the many. "The protective privilege ends where the public peril begins" (Tarasoff, 1976). However, clinicians must carefully consider the effects of long‑term confinement versus the right to liberty before effecting an indeterminate deprivation.

Issues of Public Policy

In light of the landmark Tarasoff decision, the pursuit of refining the research methods and accuracy in predicting dangerousness may become a double-edged sword. As a result of the term, dangerous, left deliberately undefined in statutes, "... psychiatric judgments are read into law, the psychotherapist's determinations rarely being challenged in court" (Dershowitz, 1968). In the same vein, it is evident that criminologists, psychologists and psychiatrists are increasingly heeding society's expectations of duty to the detriment of client interests.


As the essence of therapists evaluations become more and more aligned with law enforcement goals and objectives, the tendency to “over predict” results in a significant threat to a client’s or patient's constitutional rights. In this sense treatment is synonymous with punishment, and those patients hoping for a cure or crying for help may be unduly compromised.

It is, obviously, a much simpler task to identify conflicting interests than to resolve those interests equitably. It is resolution, however, that must be attempted. Some theories have emerged in recent years.

The concept of informed consent arose in the 1972 landmark case of Cobbs v. Grant. This case involved a doctor/patient relationship, the reference to therapist/patient included as a logical expansion. Such informed consent would mandate evidentiary exclusions similar to Miranda warnings. Applying the concept to the dangerousness issue, clinicians could implement some countervailing safeguards to the confidential nature of therapy (i.e., if informed consent is not given by the patient, the confidences cannot be divulged).

             Insofar as accuracy of prediction, it is abundantly clear that more comprehensive

empirical studies must be undertaken to insure that statistical data is complete and current. Further safeguards might suggest that a second clinical impression be obtained prior to recommending confinement. It is imperative that the courts and legislature assume more responsibility by framing less vague standards by which the therapist may assess future violent behavior.

                 "Deference [to therapists] . . . should not be tolerated, for upon these predictions turn the interests, even the constitutional rights, of the patient, as well as the interests of the victim. Perhaps including life itself" (Flaming & Maximov, 1974).

                 Conversely, therapists should be well acquainted with prevailing and developing law in the area of dangerousness prediction and, as well, become attuned to well‑founded dogma in related fields for possible analogy and practical application. Only if all components in the dangerousness issue join forces to effect a fair and standardized structure can protection for the competing and equally important rights and interests be attained.




            Bower, B. (1984, June 9). Predicting Dangerousness: Future Imperfect. Science News. 365-367.

Davis, J., Siota, R. L., & Stewart, L. M. (1999). Future Prediction of

Dangerousness and Violent Behavior: Psychological Indicators and Considerations for

Conducting an Assessment of Potential Threat. Canadian Journal of Clinical Medicine,

Edmonton, Alberta, vol. 6, no. 3, 52, 59.

            Decisions of the Supreme Court. [On‑line]. Found onsite at this location:

            Dershowitz, A. (1968, Feb.‑Mar.). Psychiatry in the Legal Process: A Knife That

Cuts Both Ways, Trial, 4, 29‑31.

            Fleming, J. G. & Maximov, B. (1974). The Patient or His Victim The Therapist's

Dilemma. California Law Review. 62, 1025‑1068.

            Goldman, H. H. (1984). Forensic Psychiatry. Textbook of General Psychiatry.

Lang Medical Publishing. 887‑898.

            Hyde, M. O. & Forsyth, E. H., (1991). The Violent Mind. New York: Franklin


            Monahan, J. (1985). Evaluating Potentially Violent Persons. Psychology,

Psychiatry and the Law. In Charles P. Ewing (Editor). Professional Resource Exchange,

Boca Raton, 9‑39.

            People v. Burnick. (1975). 14 C. 3d 306‑337.

            Roth, A. (1974). What Can We Predict. Task Force Report, American Psychiatric

Association on Dangerousness. 1-6. U. S. Government Printing.

            Slovenko. (1960). Psychiatry and a Second Look at the Medical Privilege, Wayne

Law Review. 6, 175.

             Steadman, H. J. (1991). From Dangerousness to Risk Assessment: Implications

for Appropriate Research Strategies. Criminality of Mentally Disordered Persons. 39‑49.

             Tarasoff v. Regents of University of California. (1976). 17 C.3d. 425‑464.

             Vatz, R. E. & Weinberg, L. S. (1989, November). The Arrogance of Predicting

Dangerousness. USA Today, 62-63.

Wallace, S. (1995). Prediction of Violence. Unpublished research, San Diego; Author.