Authors: Gavin de Becker,Thomas A. Taylor,Jeff Marquart
The criteria above should also be applied to cards which arrive with flowers, telephone messages, statements made during visits, or any other type of communication from members of the general public.
Appendix 6
An introduction to Threat Assessment and Management, a confidential White Paper Report by Gavin de Becker, circa 1980
This confidential White Paper Report from Gavin de Becker & Associates addresses some of the issues to be considered when managing situations involving mentally ill persons who focus upon public figures. This information is presented in general terms and does not constitute a recommendation on how to manage a specific situation.
There are some aspects of threat assessment and management (TAM), which are unique to the type of case being evaluated. For instance, the dynamics of cases involving media figures, public officials and others who do not have a personal relationship with the pursuer are not the same as those where there is a personal relationship.
To understand this, we must look at the current (circa 1980) body of knowledge in the area of behavioral predictions. We must also realize that statistics are not nearly so valuable here as they might be in other research because there are so few incidents to study. Accordingly, it has been important to widen the base to include near attacks, actions in furtherance of an attack, stalking, threats to attack, lying-in-wait, inappropriate or bizarre communications, and inappropriate encounters. It is, of course, a sample of widely different situations and people, and since there are few binding theories that can encompass all of them, predictions should be limited to areas where success is likely.
Dangerousness is not a permanent state of being, or an attribute of a person; dangerousness is situational. Further, we are not predicting dangerousness on it's own. Rather, we are discussing the likelihood of escalation or violence between the person being pursued and the person under assessment. We are not guessing what would happen if you put Person "A" in a room with Person "B", because the elements of circumstance, opportunity, intent and chance are difficult to predict, and because dangerousness itself is the sum total of many variables. A few are:
I avoid references to a "violent disposition" or nature. Even though this characteristic might well be relevant, its absence is not a guarantee of so-called "non-dangerousness." I also avoid the term "potentially dangerous," because this can fairly be applied to nearly everyone. Most of us possess that potential, and faced with certain circumstances, it might be acted on. Accordingly, one must be careful not to confuse the potential to do harm with the intent to do harm.
In his excellent monograph, Predicting Violent Behavior, Dr. John Monahan offers an instructive analogy on prediction in general:
If asked to predict in which direction this monograph would fall, if it were let go, the reader could technically state only that every other solid object he or she has let go of in the past has (eventually) fallen down rather than risen up or remained suspended. What allows for the prediction that this object, if released in the future, will also fall down is that we possess a theory -- gravity -- that can plausibly let us generalize from the past class of cases to the current individual case. This theory also allows us to set boundary conditions on this prediction, so we know that, if the monograph were let go in space, outside the force of the earth's gravity, it would not fall but would remain stationary. The catch, of course, is that we understand gravity much better than we understand violence and tend simply to assume that whatever conditions operated to produce violence in the past will also do so in the future. This may often be a plausible assumption, but there are exceptions.
Many cases involving public figure attackers, stalkers, and workplace violence are exceptions.
Monahan noted there is a widespread tendency to base predictions on an assessment of past behavior. Mental health practitioners and the criminal justice system put great weight on this factor: "If there is one finding that overshadows all others in prediction, it is that the probability of future crime increases with each prior criminal act" (Monahan 1981). While this is doubtlessly true in the context of traditional criminal violence, in the limited context of public figure attack, for instance, this is not a reliable predictor. I do not base this solely on the fact that most public figure attackers or those who commit violence in the workplace did not have a history of criminal behavior. Rather, I question the grouping of violent behaviors that are not actually similar to each other. For example, there are outbursts of rage where one strikes someone, turns over furniture, breaks windows, etc. That is clearly violent, but contrast it with acts that, while they might have violent results, are not physically violent. Is placing a bomb violent? For that matter, is shooting at a public figure violent in the same sense that fighting is violent? Shooting at someone can be a calm and calculated act. The results are violent, but the action is not. So one must weigh the type of violence to determine its relevance to an assessment. There is frenetic, tantrum-like violence; and there is calculated, controlled violence, just as there is rage, and there is anger. A history of violence might demonstrate how a person deals with aggression or conflict. For that reason, a history of violent behavior might well be relevant to an assessment.
However, the absence of violent behavior in one's past might be absolutely irrelevant to the assessment.
In considering this, recall that many persons who perform media crimes, such as the assassin and the sniper on the school grounds, were later described as "timid and unassuming," "quiet," "a good neighbor," etc. These men are not often bar-room brawlers, who vent their aggression. Rather, they are men who dwell on their aggression, and instead of "letting off steam," they contain it.
There are also some important distinctions between dangerous behavior and violent behavior. Dr. Monahan discusses this issue:
At first glance, the definition of "violent" or "dangerous" behavior appears to be a straightforward one about which only academics could quibble. Being shot, stabbed, or punched is violent. The meaning of a sign reading "dangerous" placed behind a gasoline truck or near a patch of thin ice is likewise clear. Yet as soon as one goes beyond these obvious examples, problems arise. While Sarbin (1967) cogently distinguishes between violence and dangerousness ("Violence denotes action; danger denotes a relationship"), virtually all others hold the terms synonymous. Some define violence to include only injury or death to persons, while others include the destruction of property. Violent thoughts are considered dangerous by some "because patients with fears and fantasies of violence sometimes act them out" (Ervin and Lyon 1969).
In assessing and managing safety hazards, we must also have a clear definition of what constitutes "harm." Harms are not limited to the obvious, such as property damage, assault, battery, stabbing, shooting, etc. We cannot overlook the harm caused by anxiety and fear. There is harm in invading privacy, and in such actions as stalking, trespass, lying-in-wait, etc. I raise this because we are not seeking to avoid, avert or manage only those situations that are traditionally considered dangerous.
Specifically directed or aimed dangerousness is not a permanent state of being. For a situation to be dangerous, there must be sinister intent (or at least disregard for another's safety), there must be the means to deliver harm, there must be a victim who is vulnerable and accessible, and there must be a set of circumstances that provide advantage to the attacker. Many of these things are measurable, but such abstract factors as inclination, predisposition, and intention take special skill and understanding of the Pursuer in order to make an effective judgment.
A hateful man, with weapons expertise and military experience, who has expressed sinister intent toward a specified target might well be considered dangerous. This determination might be less valid if he is confined to a wheelchair and incarcerated for life in a distant prison. He might well have a violent predisposition and a sinister intent, but he does not have the opportunity and ability to perform an attack, and cannot at this time, act on opportunities that might develop. That can change, but so can the assessment change. There has been a lot of research into the prediction of dangerousness that indicated it is not possible to reliably predict dangerousness no matter how much relevant information is available or who is making the judgment. This includes research conducted in controlled environments such as mental hospitals and jails, research conducted under the direction of Congressional Committees, Presidential Commissions, the Secret Service, the Institute of Medicine, etc. I don't necessarily disagree with these results in the context of their application for most law enforcement and mental health professionals. This consensus, however, is not entirely relevant to identifying, classifying and assessing in order to enhance safety.
Most studies on this subject have sought to predict dangerousness from a statistical point of view among focus populations (prisons, jails, mental hospitals, society at large). Some work has involved predicting dangerousness among individuals who distinguished themselves by making threats (including suicide threats). Our work differs substantially because it measures the likelihood of escalation or violence between particular identified individuals. Only part of this measurement involves factors related to the sinister intent of the pursuer. Further, most of the studies are directed toward actions that might be taken once dangerousness is predicted, i.e., civil commitment, extended incarceration, denial of bail or parole, arrest (on threat charges, for example), detainment for mental health observation, and other forms of intervention. Our overall goal is to reduce risk and hazard. Controlling the conduct of the person under assessment might be only a part of most management plans, however it is the primary part of probation conditions, parole supervision, detainment, arrest, commitment, and the other interventions traditionally applied by the criminal justice system and the mental health community.
When intrusive interventions might ultimately result from an assessment of dangerousness, it is clear that the incidence of predictions that wrongly identify people as "dangerous" must be extremely low in order to be acceptable in our society. (These erroneous predictions are called False Positives. Those predictions that fail to predict dangerousness where it is present are called False Negatives.) Failing to predict dangerousness where it exists can have (and has had) tremendous negative impact on the targets of violence, and on society at large. This is a given. However, what we are addressing here is the impact upon those persons determined dangerous who would not have performed a violent act in the absence of intervention. Continuing along these lines, consider this example:
Assume that a method of predicting dangerousness will correctly identify 85% of the people who will later engage in dangerous behavior. Therefore, this prediction technique has a false negative rate of 15% (i.e., the technique will incorrectly predict that 15% will not engage in dangerous behavior). The technique also correctly identifies 90% of those who will not behave dangerously (i.e., the method will have a false positive rate of 10%).
This technique is then applied to a random sample of 100,000. Since the base rate for violent crime in the U. S. is 187 per 100,000, the test will correctly identify 85% or 159 of the 187 dangerous persons. Twenty eight false negatives would also be made (i.e., 28 persons would be erroneously predicted not to engage in dangerous behavior).
Because of the 10% false positive rate, the test would erroneously predict 9,981 of these persons will commit acts of violence. When the false positives above are added to the 28 false negatives, it is obvious that the test will make 10,009 errors. If no prediction had been made, there could have been only 187 errors (the number of persons who act violently out of 100,000).
Having presented these thoughts, it is easy to see that when the dangerousness assessment might result in highly intrusive forms of intervention, including denial of freedom, even a method with a high success rate is not good enough. In the criminal justice system, one is presumed innocent until proven guilty, and cannot be punished for acts not yet committed, no matter how persuasive the evidence. Intrusive intervention based upon a prediction of future behavior violates the most fundamental rights in a democratic society; punishment for past acts, not detention for the future acts (Monahan, 1981). Nearly all research in this field has been judged in the light of these facts, yet there are other applications of behavioral science predictions which are practical and proper.
Research into the behavioral sciences, particularly toward early identification of dangerousness, has been limited by fears of the potential abuses of such knowledge. Still, there are those who believe that these issues should be considered from a societal point of view, as well as from the individual's point of view. The target has as much right to not be victimized as the person being assessed has to retain his civil rights.
Believing that dangerousness is usually far too situational to accurately predict on its own, I nonetheless feel that effective assessment and management is possible and practical in our work, within the constraints of a free society. There are several reasons: