The Four Winds Blog

Predicting Violence in Adolescents


By David Pogge, PhD.
Predicting violence is a challenge for all mental health professionals, and one that becomes particularly acute in the aftermath of widely publicized tragedies like those at Columbine, Sandy Hook, or Stoneman Douglas High School. These shocking events often precipitate requests by parents and school officials to determine whether or not a specific teen poses a threat to their school or community. Unfortunately, it is almost impossible to predict these sorts of events. Although widely publicized, these mass killings are so rare that it is almost impossible for prediction to improve on the Base Rate. Base Rate is the rate at which an event normally occurs. If something happens 50% of the time, then anything that predicts with more than 50% accuracy will improve our prediction of the event. However, if something occurs 99% of the time, then a predictor would have to be more than 99% accurate to improve our prediction. For example, if 99% of the time it does not rain in your town, then you will be right 99% of the time if you predict that it will not rain today. 

When events are extremely common, you will almost always be right if you predict that they will happen, and when they are extremely rare you will almost always be right if you predict that they won’t. Given the millions of students that attend thousands of schools every day without committing mass killings, you will almost always be correct to predict that any particular student will not commit a mass killing on any particular day. There is nothing that has yet been identified that is so powerful a predictor that it improves on the base rate. It is easy to look back and say “We should have known” but in reality almost everything we might have used to predict these tragedies was far more likely to be wrong than to be right.

Most violence by adolescents does not involve mass killings and does not take place during the school day. There are more common forms of violence that disrupt the lives of adolescent victims and perpetrators that can be predicted with some degree of accuracy by trained clinicians. Research shows that clinical judgment can improve on the base rate when predicting more conventional forms of adolescent violence if clinicians focus on the relevant information and weigh it appropriately.

The factors that predict risk for violence in adolescents fall into three major categories:  Historical factors include past violence, particularly if it started early in life; history of delinquency and rule-breaking by the teen or his/her parents; violence at home including physical abuse; disrupted relationships with caregivers in early childhood; history of self-harm and suicide attempts; and poor school achievement.  Contextual factors include delinquent peers, rejection by peers, poor coping resources and lack of social supports, absent or inadequate parents, and a disorganized, impoverished, or crime-ridden neighborhood.  Clinical factors include negative attitudes towards rules and authorities, a pattern of impulsivity and risk-taking, poor self-control, substance use, anger control problems, low empathy, lack of remorse, serious rule breaking, lack of adult supervision, and a low commitment to school. By itself, none of these necessarily predicts violence, but the more that are true and the more severe they are, the greater the likelihood that an adolescent will act in a violent manner.

While these can be assessed informally by a competent clinician, the Structured Assessment of Violence Risk in Youth (SAVRY) is a clinical judgment tool developed by Borum, Bartel, and Forth (2006) that is designed to help the clinician focus on and weigh these factors when forming a judgment about violence risk. This tool not only insures systematic analysis of the relevant factors, but documents the fact that the clinician’s judgment was thorough, systematic, and consistent. However, even without the use of this specific instrument, a careful review of the scientifically validated predictors of risk for violence in adolescents can significantly improve clinical judgment and help predict the kind of violence that can be predicted, while understanding the problem of the Base Rate can help the clinician avoid trying to predict those rare, anomalous, and dramatic events that cannot be predicted.

To Make
A Referral, call:
1-914-763-8151 or
24 Hours A Day,
7 Days A Week

Four Winds Blog/Newsletter Articles


January 2019

February 2019

April 2019

May 2019

June 2019

July 2019

August 2019

September 2019