Two Arizona prison inmates violently overpower corrections officers and take two guards captive, barricading themselves in a guard tower and brandishing weapons. During 15 days of negotiations, the inmates use the hostages to keep tactical teams at bay.¹

Data-Driven Prediction of Outcomes in Hostage and Barricade Incidents

A couple of comparisons may put the tools’ high accuracy rates into perspective. Statistically, the correlations between the tools’ scores and crisis incident outcomes are about as strong as those observed between standardized test scores and academic performance in college and graduate school. Further, they appear capable of predicting outcomes in crisis incidents almost as well as mammograms predict breast cancer.⁴

This level of accuracy enables law enforcement to assess a crisis and, based on statistical analyses of thousands of cases across the United States, assert with reasonable confidence, “An incident with these features tends to involve violence after onset X% of the time and successfully resolve by negotiation or surrender Y% of the time.”

Finding Success

CNU’s innovative tools resulted from a collaborative effort that began in 2012, when a special agent invited a forensic psychologist to help teach law enforcement to assess violence risk in crisis incidents. The forensic psychologist off-handedly remarked that CNU could use the thousands of incident reports contained in the FBI’s Hostage Barricade Database System (HOBAS) to improve the prediction of outcomes in crises. Immediately, the agent set the project in motion.

Three teammates joined them — another FBI agent who teaches graduate-level statistics, an FBI victim services psychologist who teaches criminology, and a data analyst who works in the private sector. The Critical Incident Response Group (CIRG) facilitated the first meeting at the FBI’s Denver office in 2014 and, along with an interagency partner, offered ongoing, steady support until the tools were finalized in 2020.

A technical write-up of the project was recently published in an international peer-reviewed journal.⁵ The article has already generated positive responses from law enforcement and other professionals outside the United States.

In addition, CNU introduced a 45-minute instruction block to teach law enforcement officers about the tools during its annual Crisis Negotiations Coordinator Conference (CNC). Attendees provided overwhelmingly positive feedback on the tools’ practical usefulness, and they contributed to CNU’s growing understanding of the tools’ statistical properties. A sample of CNC attendees provided preliminary evidence that, when presented with a complex historical scenario, law enforcement officers can independently agree on how each item should be scored about 90% of the time (see table).

For illustrative purposes, CNU has also applied the two tools retrospectively to three crisis incidents that are well-known to crisis negotiators and commonly used as case studies in training scenarios.

A) 2013 Midland City, Alabama, underground bunker hostage incident, briefly discussed at the beginning of the article

B) 1999 St. Martin Parish, Louisiana, prison takeover by eight Cuban inmates, which resolved following six days of negotiation⁶

C) 1996 Honolulu Seal Masters of Hawaii hostage incident, a six-hour standoff that resolved via tactical action⁷

Tool 1 correctly indicated the risk for violence after onset was substantially above average for each incident. Tool 2 indicated the likelihood of resolution by negotiation or surrender was below average for incidents A and C and as average for incident B.

Conclusion

Although current research indicates both tools can be reliably scored and are reasonably accurate in predicting outcomes, incident commanders are discouraged from using them as substitutes for professional judgments or continuous assessments. Rather, the tools should be used as aids to make informed, defensible, and data-driven decisions in hostage and barricade incidents.

Additional research will help users better understand the tools’ accuracy, usefulness, strengths, and limitations. The FBI’s Crisis Negotiation Unit intends to continue to conduct research on the tools, offer training on them, and solicit ongoing feedback from local and state law enforcement partners.