U.S. flag

An official website of the United States government, Department of Justice.

NCJRS Virtual Library

The Virtual Library houses over 235,000 criminal justice resources, including all known OJP works.
Click here to search the NCJRS Virtual Library

Predictive Modeling in Hostage and Barricade Incidents

NCJ Number
Criminal Justice and Behavior: An International Journal Volume: 35 Issue: 9 Dated: September 2008 Pages: 1136-1155
Patrick Michaud; Michel St-Yves; Jean-Pierre Guay
Date Published
September 2008
20 pages
This study provides empirical information to improve decisionmaking in hostage and barricade situations.
The study demonstrates that, through statistical analysis, it is possible to use a series of static and dynamic individual and situational factors to predict the sequence of events of critical incidents effectively. The results of this study highlight the importance of considering the origin of the emergency call and being barricaded in one's home in the assessment of perpetrator's propensity for violence; the fact that time is not always on the police's side during critical incidents; and the relatively minor importance of hostage taking in the general comprehension of these high-risk events. The objective of this work was to present an integrated and formalized vision of current conceptualizations of critical incidents to optimize decisionmaking. To this end, logistic regression and recursive partitioning models were presented and compared. In all, 18 distinct static and dynamic individual and situational factors were used to predict various outputs during this type of situation. The study examined 534 hostage and barricade incidents that occurred in the Province of Quebec, excluding the city of Montreal, between 1990 and 2004 that were managed by the Province’s specialized crisis intervention structure. The criteria for the unit’s response were: hostages had been taken; gunshots had been fired; or the lives of the perpetrator or others were in imminent danger. An analytical checklist was devised and employed to quantitatively analyze data contained in post-incident reports. A total of 509 reports met considerations for analysis from the original population of reports. Tables, figures, note, and references


No download available