Research on Student Threat Assessment

Studies Show How Schools Can
Safely Deal With Student Threats of Violence

Although both the FBI and Secret Service reports made a compelling case for student threat assessment, schools had no experience with this approach and there were many questions concerning the practical procedures that should be followed. In response, researchers at the University of Virginia developed a set of guidelines for school administrators to use in responding to a reported student threat of violence. Threat assessment teams are trained in a six-hour workshop that prepares them to use a 145-page threat assessment manual (Cornell & Sheras, 2006).

The Virginia model of threat assessment is an approach to violence prevention that emphasizes early attention to problems such as bullying, teasing, and other forms of student conflict before they escalate into violent behavior. School staff members are encouraged to adopt a flexible, problem-solving approach, as distinguished from a more punitive, zero tolerance approach to student misbehavior. As a result of this training, the model is intended to generate broader changes in the nature of staff-student interactions around disciplinary matters and to encourage a more positive school climate in which students feel treated with fairness and respect. Consistent with this goal, a pre-post survey study of 351 school staff members who completed the Virginia workshop found that participants became less anxious about the possibility of a school homicide, more willing to use threat assessment methods to help students resolve conflicts, and less inclined to use a zero tolerance approach (Allen, Cornell, Lorek, & Sheras, 2008). Similar effects were found for principals, psychologists, counselors, social workers, and law enforcement officers.

The Virginia guidelines follow a seven-step decision-tree. In brief, the first three steps constitute a triage process in which the team leader (a school administrator such as the principal or assistant principal) investigates a reported threat and determines whether the threat can be readily resolved as a transient threat that is not a serious threat. Examples of transient threats are jokes or statements made in anger that are expressions of feeling or figures of speech rather than expressions of a genuine intent to harm someone.

Any threat that cannot be clearly identified and resolved as transient is treated as a substantive threat. Substantive threats always require protective action to prevent the threat from being carried out. The remaining four steps guide the team through more extensive assessment and response based on the seriousness of the threat. In the most serious cases, the team conducts a safety evaluation that includes both a law enforcement investigation and a mental health assessment of the student. The culmination of the threat assessment is the development of a safety plan that is designed to address the problem or conflict underlying the threat and prevent the act of violence from taking place. For both transient and substantive threats, there is an emphasis on helping students to resolve conflicts and minimizing the use of zero-tolerance suspensions as a disciplinary response.

First study. The Virginia threat assessment guidelines were field-tested in 35 public schools encompassing an enrollment of more than 16,000 students in grades K-12 (Cornell et al., 2004). School-based teams evaluated 188 student threats that involved threats to hit, stab, shoot, or harm someone in some other way. Most of the threats (70%) were resolved as transient threats and the remaining 30% were substantive threats that required more extensive assessment and protective action. The threat assessment teams placed special emphasis on understanding the context and meaning of the threat, and developing a plan to address the underlying conflict or problem that stimulated the student to resort to threatening behavior. Use of this problem-solving approach meant that relatively few students received long-term suspensions or expulsions from school. Only three students were expelled from school, although half of the students (94) received short-term suspensions (typically 1-3 days). Notably, follow-up interviews with the school principals found no cases in which the threats were carried out. This study was published in School Psychology Review.

Second study. A second study enlarged the sample from the first study in order to identify more cases of students receiving special education services (Kaplan & Cornell, 2005). This study found that students in special education (120 cases) made disproportionately more threats, as well as more severe threats, than peers in general education (136 cases). Nevertheless, use of school suspension was consistent for students in special and general education, suggesting that the use of threat assessment did not lead to increased suspension rates among students in special education. This study was published in Behavioral Disorders.

Third study. A third study examined the Virginia threat assessment model in Memphis City Schools, one of the nation’s largest school districts, serving a largely disadvantaged urban population in which 75% of students are eligible for a free or reduced price meal (Strong & Cornell, 2008). This study examined outcomes for 209 cases referred to a centralized threat assessment team because the principal deemed them to merit long term suspension. At least 110 of the cases involved explicit threats to shoot, stab, or kill someone, as well as 99 other threats to attack someone, commit a sexual assault, burn down or blow up the school, etc. Approximately 38% of the students were receiving special education services (compared to a 12% baseline) and 71% had been academically retained at least one year. This study found that all of the student threats were resolved without any detected act of violence. Almost all students were able to return to their school or an alternative school placement, with only five students receiving long-term suspensions without school services. Plans to assist each student included modifications to special education plans, the provision of academic and behavioral support services, and referrals to community-based mental health services. After the threat assessment, the number of disciplinary office referrals for these students declined by approximately 55% through the remainder of the school year. This study was published in Behavioral Disorders.

Fourth study. A major limitation to the previous studies was the absence of a comparison group of schools. Our fourth study examined the use of the Virginia model in the statewide population of Virginia high schools (Cornell, Sheras, Gregory, & Fan, in press). The 95 high schools using the Virginia model (according to principal report on a state-mandated audit of school safety conditions) were compared to 131 schools using a locally developed threat assessment model and 54 schools not using a threat assessment approach. This was a retrospective comparison using data on student victimization and perceptions of school climate that were available from the Virginia High School Safety Study, a separate study of high school safety conditions funded by the U.S. Department of Justice (Cornell & Gregory, 2008). This study did not collect case data on student threats, so schools were compared on the basis of more general outcomes that could be expected from the adoption of a threat assessment approach, based on an anonymous survey of randomly selected students from each high school. This threat assessment study found that students in schools using the Virginia model reported less bullying, greater willingness to seek help for bullying and threats of violence, and more positive perceptions of the school climate than students in either of the other two groups of schools, with an overall multivariate effect size of h2 = 0.15, which is a medium sized effect (Cornell, Sheras, Gregory, & Fan). In addition, schools using the Virginia guidelines had an average of 10 long-term suspensions compared to 15 for the other two groups (p < .05, d = .30). Group differences could not be attributed to school size, minority composition or socio-economic status of the student body, neighborhood violent crime, or the extent of security measures in the schools, which were statistically controlled.

Currently, we are initiating a randomized controlled study of threat assessment. School systems interested in participating in this study can contact Dewey Cornell for more information.