Can Mental Health Professionals Predict Violent Behavior in Schools and Agencies?

Not surprisingly, violence has been on my mind lately. And so when I reached the Violence Risk Assessment section of the Clinical Interviewing text revision, I decided to cut and paste it here. It doesn’t immediately answer the question of whether mental health professionals can predict violence and so if you’re impatient and prefer to stop reading now, the answer to that question is, more or less, “No.”

Assessment and Prediction of Violence and Dangerousness

During an assessment interview, John had the following exchange with a 16-year-old client.

John: I hear you’ve been pretty mad at your shop teacher.

Client: I totally hate Mr. Smith. He’s a jerk. He puts us down just to make us feel bad. He deserves to be punished.

John: You sound a little pissed off at him.

Client: We get along fine some days.

John: What do you mean when you say he “deserves to be punished”?

Client: I believe in revenge. Really, I feel sorry for him. But if I kill him, I’ll be doing him a favor. It would end his miserable life and stop him from making other people feel like shit.

John: So you’ve thought about killing him?

Client: I’ve thought about walking up behind him and slitting his throat.

John: How often have you thought about that?

Client: Just about every day. Whenever he talks shit in class.

John: And exactly what images go through your mind?

Client: I just slip up behind him while he’s talking with Cassie [fellow student] and then slit his throat with a welding rod. Then I see blood gushing out of his neck and Cassie starts screaming. But the world will be a better place without his sorry ass tormenting everybody.

John: Then what happens?

Client: Then I guess they’ll just take me away, but things will be better.

John: Where will they take you?

Client: To jail. But I’ll get sympathy because everyone knows what a dick he is.

During an initial interview or ongoing therapy, clients may describe aggressive thoughts and images. Some clients, as in the preceding example, will be concise about their thoughts, feelings, and images. Others will be less clear. Still others will be evasive and will avoid telling you anything about violent thoughts or intentions.

Assessing for violence potential is similar to assessing for suicide potential; it’s a stressful responsibility and predicting violence is extremely difficult. However, similar to suicide assessment, we still have a legal and ethical responsibility to conduct violence or dangerousness assessments that meet professional standards.

Over the years, there have been arguments about how to most accurately predict violence (Hilton, Harris, & Rice, 2006). Essentially, there are three perspectives.

1. Some researchers contend that actuarial prediction based on specific, predetermined statistical risk factors is consistently the most accurate procedure (Quinsey, Harris, Rice, & Cormier, 2006).

2. Some clinicians believe that because actuarial variables are dimensional and interactive with individual and situational characteristics, prediction based on the clinician’s experience and intuition is most accurate (Cooke, 2012).

3. Others take a moderate position, believing that combining actuarial and clinical approaches is best (Campbell, French, & Gendreau, 2009).

Researchers have consistently reported that actuarial approaches to violence prediction are more accurate than clinical judgment (Monahan, 2013). However, actuarial violence prediction is not without its flaws (Szmukler, 2012; Tardiff & Hughes, 2011).

Narrowing in on Particular Violent Behaviors

Researchers who investigate actuarial assessment protocols have reported that different violent behaviors are associated with unique predictor variables. Below, we provide three examples of violence predictors for three different specific violent behaviors or populations. The goal is to sensitize you to different violent behavior patterns.

Fire-setting. Fire-setting is a particular dangerous behavior that may or may not be associated with interpersonal violence. Nonetheless, depending on your work setting and the clinical population you serve, you may find yourself in a situation in which you need to decide whether to warn a family or potential victim about possible fire-setting behavior.

Mackay and colleagues (2006) reported on specific behaviors included on a fire-setting prediction assessment. They identified the following variables—in decreasing order—as predictive of fire setting:

  •  Younger age at the time of the first fire-setting behavior.
  • A higher total number of fire-setting offenses.
  • Lower IQ.
  • Additional criminal activities associated with the index (initial) fire.
  • An offender acting alone in setting the initial fire.
  • A lower offender’s aggression score. (Interestingly, offenders with higher aggression scores were more likely to be violent, but less likely to set fires.)

We focus first on fire setting here because fire-setting predictors illustrate a general violence-prediction principle. Past violence is a reasonably good predictor of future violence only with regard to specific past and future violence. For example, future fire-setting potential is best predicted by past fire-setting behavior. Similarly, future physical aggression is best predicted by past physical aggression. But a history of physical aggression is not a good predictor of fire setting.

Homicide Among Young Men. Loeber and associates (2005) conducted a large-scale landmark study of homicide among young men living in Pittsburgh. This study is notable because it was both prospective and comprehensive; the authors tracked 63 risk factor (predictor) variables in 1,517 inner-city youth. Obviously, even this large-scale study is limited in scope, and technically the results cannot be generalized beyond inner-city Pittsburgh youth. Nevertheless, the outcome data are interesting and lend insight into risk factors that might contribute to homicidal violence in other populations.
Results from the study indicated that violent offenders scored significantly higher than nonviolent offenders on 49 of 63 risk factors across domains associated with child, family, school, and demographic risk factors. The range and nature of these predictors were daunting. The authors reported:

. . . predictors included factors evident early in life, such as the mother’s cigarette or alcohol use during pregnancy, onset of delinquency prior to 10 years of age, physical aggression, cruelty, and callous/unemotional behavior. In addition, cognitive factors, such as having low expectations of being caught, predicted violence. Poor and unstable child-rearing factors contributed to the prediction of violence, including two or more caretaker changes prior to 10 years of age, physical punishment, poor supervision, and poor communication. Undesirable or delinquent peer behavior, based either on parent report or self-report, predicted violence. Poor school performance and truancy were also among the predictors of violence. Finally, demographic factors indicative of family disadvantage (low family SES, welfare, teenage motherhood) and residence in a disadvantaged neighborhood also predicted violence. Among the proximal correlates associated with violence were weapon carrying, weapon use, gang membership, drug selling, and persistent drug use. (p. 1084)

Homicidal violence was best predicted by a subset of general violence predictor variables. Specifically, homicide was predicted by “the presence or absence of nine significant risk factors:

• Screening risk score
• Positive attitude to substance use
• Conduct disorder
• Carrying a weapon
• Gang fight
• Selling hard drugs
• Peer delinquency
• Being held back in school
• Family on welfare (p. 1086).

In particular, boys who had at least four of these nine risk factors were 14 times more likely to have a future homicide conviction than violent offenders with a risk score less than four.

Violence and schizophrenia. In and of itself, a diagnosis of schizophrenia doesn’t confer increased violence risk. Instead, research indicates there are specific symptoms—when seen among individuals diagnosed with schizophrenia—associated with increased risk. These symptoms include severe manifestations of:

  • Hallucinations
  • Delusions
  • Excitement
  • Thinking disturbances. (Fresán, Apiquian, & Nicolini, 2006)

This research suggests that clinicians should be especially concerned about violence when clients diagnosed with schizophrenia have acute increases in the intensity and frequency of their psychotic symptoms.

Research versus Practice

For a short guide to predicting violence, see a previous post:




Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s