Intervention Summary - Virginia Student Threat Assessment Guidelines

Documents Reviewed

The documents below were reviewed for Quality of Research. The research point of contact can provide information regarding the studies reviewed and the availability of additional materials, including those from more recent studies that may have been conducted.

Study 1

Cornell, D., Sheras, P., Gregory, A., & Fan, X. (2009). A retrospective study of school safety conditions in high schools using the Virginia Threat Assessment Guidelines versus alternative approaches. School Psychology Quarterly, 24(2), 119-129.

Study 2

Cornell, D. G., Gregory, A., & Fan, X. (2011). Reductions in long-term suspensions following adoption of the Virginia Student Threat Assessment Guidelines. NASSP Bulletin, 95(3), 175-194.

Study 3

Cornell, D. G., Allen, K., & Fan, X. (2012). A randomized controlled study of the Virginia Student Threat Assessment Guidelines in kindergarten through grade 12. School Psychology Review, 41(1), 100-115. (NOTE: At the time of the NREPP review, the manuscript of this article had been submitted for publication but not yet accepted.)

Supplementary Materials

Allen, K., Cornell, D., Lorek, E., & Sheras, P. (2008). Response of school personnel to student threat assessment training. School Effectiveness and School Improvement, 19(3), 319-332.

Bandyopadhyay, S., Cornell, D. G., & Konold, T. R. (2009). Validity of three school climate scales to assess bullying, aggressive attitudes, and help seeking. School Psychology Review, 38(3), 338-355.

Cornell, D. (2011). School Climate Bullying Survey: Descriptive report. Charlottesville: University of Virginia.

Cornell, D., & Gregory, A. (2008). Virginia High School Safety Study: Descriptive report of survey results from ninth grade students and teachers. Charlottesville: University of Virginia.

Cornell, D. G., Sheras, P. L., Kaplan, S., McConville, D., Douglass, J., Elkon, A., et al. (2004). Guidelines for student threat assessment: Field-test findings. School Psychology Review, 33(4), 527-546.

Gottfredson, G. D., Gottfredson, D. C., Payne, A. A., & Gottfredson, N. C. (2005). School climate predictors of school disorder: Results from a national study of delinquency prevention in schools. Journal of Research in Crime and Delinquency, 42, 412-444.

Threat Assessment and Response Protocol (adapted from the Guidelines for Responding to Student Threats of Violence, by Cornell and Sheras, 2006)

Outcomes

Outcome 1: Long-term school suspensions and expulsions
Description of MeasuresLong-term school suspensions and expulsions were measured using the following:
  • Data retrieved from the public archival database of the Virginia Department of Education (VDOE). The VDOE database captures State-mandated electronic reports that include 113 categories of disciplinary infractions. These reports, which are filed yearly by all public school principals, also indicate whether the infraction resulted in a student's long-term suspension or expulsion.
  • Information retrieved from principal-completed school records of student threat incidents. The records indicate whether or not students identified as having made a threat of violence received a long-term suspension or expulsion. The record of each student threat incident also includes a description of the incident and the steps taken by the school in response to the threat.
Key FindingsA retrospective cross-sectional study was conducted during the 2006-07 school year with 9th-grade students in 95 public high schools that used V-STAG, 131 public high schools that used a locally developed threat assessment program, and 54 public high schools that had no threat assessment program. The average number of long-term school suspensions (i.e., denial of school attendance for more than 5 days) and expulsions (i.e., denial of school attendance for at least 1 year) per school was lower for all students in high schools that used V-STAG than it was in high schools that used a locally developed threat assessment program (10.50 vs. 15.71; p = .05) and in high schools that had no threat assessment program (10.50 vs. 15.28; p = .05), after controlling for total student enrollment, proportion of minority students, proportion of students eligible for reduced-price meals, annual number of neighborhood violent crimes, number of school resource officers employed at the school, and student perceptions of school security. Small effect sizes were associated with the difference between high schools that used V-STAG and high schools that used a locally developed threat assessment program (Cohen's d = 0.30) and the difference between high schools that used V-STAG and high schools that had no threat assessment program (Cohen's d = 0.30).

A 3-year study was conducted with 23 public high schools that used V-STAG and 26 public high schools that had no threat assessment program. Data were retrieved from the VDOE archival database for the school years before (2006-07) and after (2008-09) V-STAG training and implementation in the intervention schools. The total number of long-term suspensions (i.e., denial of school attendance for more than 10 days) and expulsions (i.e., denial of school attendance for at least 1 year) by high school for the 2006-07 (baseline) and 2008-09 (follow-up) school years was converted to a rate on the basis of the fall student enrollment for the corresponding school year. Relative to baseline, the long-term school suspension and expulsion rate at follow-up in high schools that used V-STAG was lower than that in high schools that had no threat assessment program (8.2 at baseline and 3.9 at follow-up per 1,000 students in high schools that used V-STAG vs. 10.9 at baseline and 10.9 at follow-up per 1,000 students in high schools that had no threat assessment program; p < >

A randomized controlled study was conducted during the 2008-09 school year with students in 40 public schools (26 elementary, 8 middle, and 6 high schools in a school division) that were assigned to the V-STAG condition or the control condition (i.e., a 1-year delay before receiving V-STAG). Of the students who made threats of violence at school during the study, those in schools assigned to the control condition were almost 3 times as likely as students in schools assigned to the V-STAG condition were to receive a long-term suspension (i.e., denial of school attendance for more than 10 days) or expulsion (i.e., denial of school attendance for at least 1 year) (p = .01). This group difference was associated with a medium effect size (odds ratio = 2.86).

Studies Measuring OutcomeStudy 1, Study 2, Study 3
Study DesignsExperimental, Quasi-experimental
Quality of Research Rating3.1 (0.0-4.0 scale)
Outcome 2: Alternative school placement
Description of MeasuresAlternative school placement was measured using information retrieved from principal-completed school records of student threat incidents. The records indicate whether or not students identified as having made a threat of violence were placed in an alternative school setting. The record of each student threat incident also includes a description of the incident and the steps taken by the school in response to the threat.
Key FindingsA randomized controlled study was conducted during the 2008-09 school year with students in 40 public schools (26 elementary, 8 middle, and 6 high schools in a school division) that were assigned to the V-STAG condition or the control condition (i.e., a 1-year delay before receiving V-STAG). Of the students who made threats of violence at school during the study, those in schools assigned to the control condition were more than 7 times as likely as students in schools assigned to the V-STAG condition were to receive an alternative school placement (p = .01). This group difference was associated with a large effect size (odds ratio = 7.69).
Studies Measuring OutcomeStudy 3
Study DesignsExperimental
Quality of Research Rating2.8 (0.0-4.0 scale)
Outcome 3: Bullying infractions in school
Description of MeasuresBullying infractions in school were measured using the following:
  • The Bullying Climate and Bullying Victimization subscales of the School Climate Bullying Survey, a 45-item self-report instrument used to assess the extent and nature of bullying problems in school. The Bullying Climate subscale consists of 7 items that describe the extent of teasing and bullying that student respondents observe taking place at school. Using a 4-point scale ranging from 1 (strongly disagree) to 4 (strongly agree), students rate their level of agreement with each item (e.g., "Students here often get teased about their clothing or physical appearance," "Bullying is a problem at this school"). Response scores are summed (some items are reversed scored), and total scores range from 7 to 28, with higher scores reflecting more observed bullying and teasing at school. The Bullying Victimization subscale consists of 4 items that describe the extent of bullying that student respondents experienced. Using a 4-point scale ranging from 1 (never) to 4 (several times per week), students rate each item to indicate the frequency of bullying experienced in the past month. Response scores are summed, and total scores range from 4 to 16, with higher scores indicating more frequent bullying victimization in school. For both subscales, bullying was defined as "the use of one's strength or status to injure, threaten, or humiliate another person. Bullying can be physical, verbal, or social. It is not bullying when two students of about the same strength argue or fight."
  • A criminal victimization index composed of 7 items from the student version of the Gottfredson Effective School Battery. The items in the index describe forms of criminal victimization, and using a response of "true" or "false," students rate each item to indicate whether the forms of criminal victimization were personally experienced in school over the past year (e.g., "damage to personal property exceeding $10," "received obscene remarks or gestures from a student," "had a weapon pulled on me"). "True" responses are summed, and total scores range from 0 to 7, with higher scores indicating more forms of criminal victimization experienced in school.
  • Three bullying infraction categories (assaults of other students, threats of other students, and bullying of other students) selected from the VDOE public archival database, which captures State-mandated electronic reports that include 113 categories of disciplinary infractions. These reports are filed yearly by all public school principals.
Key FindingsA retrospective cross-sectional study was conducted during the 2006-07 school year with 9th-grade students in 95 public high schools that used V-STAG, 131 public high schools that used a locally developed threat assessment program, and 54 public high schools that had no threat assessment program. Findings from this study included the following:
  • Compared with 9th graders in high schools that used a locally developed threat assessment program, 9th graders in high schools that used V-STAG had less of a bullying school climate (16.48 vs. 16.83; p = .05), less bullying victimization (1.21 vs. 1.40; p = .05), and less criminal victimization (1.27 vs. 1.41; p = .05), after controlling for total student enrollment, proportion of minority students, proportion of students eligible for reduced-price meals, annual number of neighborhood violent crimes, number of school resource officers employed at the school, and student perceptions of school security. These group differences were associated with small effect sizes (Cohen's d = 0.35, 0.38, and 0.40, respectively).
  • Compared with 9th graders in high schools that had no threat assessment program, 9th graders in high schools that used V-STAG had less of a bullying school climate (16.48 vs. 16.96; p = .05), after controlling for total student enrollment, proportion of minority students, proportion of students eligible for reduced-price meals, annual number of neighborhood violent crimes, number of school resource officers employed at the school, and student perceptions of school security. This group difference was associated with a small effect size (Cohen's d = 0.45).
A 3-year study was conducted with 23 public high schools that used V-STAG and 26 public high schools that had no threat assessment program. Data were retrieved from the VDOE archival database for the school years before (2006-07) and after (2008-09) V-STAG training and implementation in the intervention schools. The total number of bullying infractions by high school for the 2006-07 (baseline) and 2008-09 (follow-up) school years was converted to a rate on the basis of the fall student enrollment for the corresponding school year. Relative to baseline, the bullying infraction rate at follow-up in high schools that used V-STAG was lower than that in high schools that had no threat assessment program (5.3 at baseline and 1.1 at follow-up per 1,000 students for schools that used V-STAG vs. 1.2 at baseline and 2.0 at follow-up per 1,000 students for high schools that had no threat assessment program; p < >
Studies Measuring OutcomeStudy 1, Study 2
Study DesignsQuasi-experimental
Quality of Research Rating2.9 (0.0-4.0 scale)
Outcome 4: Supportive school climate
Description of MeasuresSupportive school climate was measured using the following:
  • The Learning Environment subscale of the California Healthy Kids Survey, a self-report survey of youth resiliency, protective factors, and risk behaviors. The Learning Environment subscale consists of 8 items regarding the following question: "How much do you agree that adults in this school…?" Using a 5-point scale ranging from 1 (strongly disagree) to 5 (strongly agree), students indicate their agreement with each item (e.g., "really care about all students," "want all students to do their best," "believe that every student can be a success"). Response scores are summed, and total scores range from 8 to 40, with higher scores indicating a more supportive school learning environment.
  • The Help Seeking domain subscale of the School Climate Bullying Survey, a 45-item self-report instrument used to assess the extent and nature of bullying problems in school. The Help Seeking domain subscale consists of 8 items that assess a student respondent's willingness to seek help from teachers or school staff for bullying and threats of violence. Using a 4-point scale ranging from 1 (somewhat disagree) to 4 (strongly agree), students indicate their agreement with each item (e.g., "If another student was bullying me, I would tell one of the teachers or staff at school," "There are adults at this school I could turn to if I had a personal problem"). Response scores are summed, and total scores range from 8 to 32, with higher scores indicating greater willingness to seek help from teachers and school staff.
Key FindingsA retrospective cross-sectional study was conducted during the 2006-07 school year with 9th-grade students in 95 public high schools that used V-STAG, 131 public high schools that used a locally developed threat assessment program, and 54 public high schools that had no threat assessment program. Findings from this study included the following:
  • Compared with 9th graders in high schools that used a locally developed threat assessment program, 9th graders in high schools that used V-STAG had a more supportive school climate in terms of a more supportive learning environment (27.75 vs. 27.08 on the Learning Environment subscale; p = .05) and greater student willingness to seek help from teachers and school staff (22.58 vs. 22.14 on the Help Seeking subscale; p = .05), after controlling for total student enrollment, proportion of minority students, proportion of students eligible for reduced-price meals, annual number of neighborhood violent crimes, number of school resource officers employed at the school, and student perceptions of school security. These group differences were associated with small effect sizes (Cohen's d = 0.31 and 0.27, respectively).
  • Compared with 9th graders in high schools that had no threat assessment program, 9th graders in high schools that used V-STAG had a more supportive school climate in terms of a more supportive school learning environment (27.75 vs. 26.79 on the Learning Environment subscale; p = .05) and greater student willingness to seek help from teachers and school staff (22.58 vs. 21.87 on the Help Seeking subscale; p = .05), after controlling for total student enrollment, proportion of minority students, proportion of students eligible for reduced-price meals, annual number of neighborhood violent crimes, number of school resource officers employed at the school, and student perceptions of school security. These group differences were associated with small effect sizes (Cohen's d = 0.42 and 0.40, respectively).
Studies Measuring OutcomeStudy 1
Study DesignsQuasi-experimental
Quality of Research Rating2.9 (0.0-4.0 scale)
Outcome 5: School counseling support and parental involvement
Description of MeasuresSchool counseling support and parental involvement were measured using threat assessment team documentation forms and information retrieved from principal-completed school records of student threat incidents. The records indicate whether or not students identified as having made a threat of violence were provided with some form of mental health counseling service appropriate to their needs and whether or not a school conference was held with a student's parents. The record of each student threat incident also includes a description of the incident and the steps taken by the school in response to the threat.
Key FindingsA randomized controlled study was conducted during the 2008-09 school year with students in 40 public schools (26 elementary, 8 middle, and 6 high schools in one school division) that were assigned to the V-STAG condition or the control condition (i.e., a 1-year delay before receiving V-STAG training and implementation). Of the students who made threats of violence at school during the study, those in schools assigned to the V-STAG condition were almost 4 times as likely as students in schools assigned to the control condition were to receive mental health counseling services (p = .01) and approximately 2.5 times as likely as students in schools assigned to the control condition were to have a school-parent conference (p = .01). These group differences were associated with medium (odds ratio = 3.98) and small (odds ratio = 2.57) effect sizes, respectively.
Studies Measuring OutcomeStudy 3
Study DesignsExperimental
Quality of Research Rating2.5 (0.0-4.0 scale)

Study Populations

The following populations were identified in the studies reviewed for Quality of Research.

StudyAgeGenderRace/Ethnicity
Study 113-17 (Adolescent)51% Male49% Female63% White23% Black or African American5% Hispanic or Latino5% Race/ethnicity unspecified3% Asian1% American Indian or Alaska Native
Study 213-17 (Adolescent)18-25 (Young adult)50% Female50% Male53% White19% Black or African American15% Hispanic or Latino11% Asian2% Race/ethnicity unspecified
Study 36-12 (Childhood)13-17 (Adolescent)73% Male27% Female73% Black or African American24% White3% Hispanic or Latino

Quality of Research Ratings by Criteria (0.0-4.0 scale)

External reviewers independently evaluate the Quality of Research for an intervention's reported results using six criteria:

For more information about these criteria and the meaning of the ratings, see Quality of Research.

OutcomeReliabilityof MeasuresValidityof MeasuresFidelityMissingData/AttritionConfoundingVariablesDataAnalysisOverallRating
1: Long-term school suspensions and expulsions3.53.21.74.02.73.53.1
2: Alternative school placement1.53.02.04.02.53.52.8
3: Bullying infractions in school3.13.31.53.52.53.52.9
4: Supportive school climate3.33.51.83.32.83.02.9
5: School counseling support and parental involvement0.52.52.04.02.53.52.5

Study Strengths

Sample reliability was provided for the student self-report scales on bullying, help-seeking behavior, and the learning environment, all of which are known in the field and have good psychometric properties established by independent investigators. In considering test-retest reliability, there was an assessment of the stability of the number (or proportion) of long-term school suspensions issued in each school across multiple years, resulting in a report of high intraclass correlation. Criterion validity was high for the long-term school suspension and expulsion data, which came from either the VDOE public archival database or the standard school records completed by school principals. In all three studies, a detailed manual was used to train school threat assessment teams in the use of V-STAG. In one study, a compliance score for intervention fidelity was constructed from measurable items based on the manual. One study used random assignment by school type to control for many potential confounding variables as well as sophisticated hierarchical linear modeling to statistically control for the clustering of students within schools. All three studies included effect sizes for the group differences in outcome findings.

Study Weaknesses

There was a lack of reliability information on principals' and school staffs' use of similar criteria for measuring alternative school placement outcomes. There was concern about interrater reliability across school staff for the school counseling support and parental involvement outcome measure, which relied on threat documentation forms that had been completed by the threat assessment teams of all participating schools; some teams did not complete these forms. Many bullying incidents may not have come to the attention of school staff and, consequently, may not have been reported. Intervention fidelity was not measured in the two quasi-experimental studies. Two studies lacked random assignment, and one of these studies (the retrospective study) did not offer information on potential preimplementation baseline differences between conditions.

http://www.nrepp.samhsa.gov/ViewIntervention.aspx?id=263