Predictive Modeling in Hostage and Barricade Incidents

• Author: Patrick Michaud,Michel St-Yves,Jean-Pierre Guay
• Published date: 01 September 2008
• Date: 01 September 2008
• DOI: http://doi.org/10.1177/0093854808319955
• Subject Matter: Articles

1136

PREDICTIVE MODELING IN HOSTAGE

AND BARRICADE INCIDENTS

PATRICK MICHAUD

University of Montreal

MICHEL ST-YVES

Sûreté du Québec

JEAN-PIERRE GUAY

University of Montreal

The study examined 534 hostage and barricade incidents. Its objective was to present an integrated and formalized vision of

current conceptualizations of critical incidents to optimize decision making. To this end,logistic regression and recursive par-

titioning models are 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. Results 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.

Keywords: critical incidents; hostage situations; barricade situations; crisis negotiation; police decision making; predictive

modeling

Since its very beginnings, the study of hostage and barricade incidents has itself been

hostage to an intuitive and empirically fragmentary conceptual framework. This has not

only limited innovative thought and the development of new ideas but also handicapped

police decision making. In fact, the predictive abilities of law enforcement personnel man-

aging critical incidents have been impaired by both the absence of a formal, integrated

vision of the factors that contribute to the conceptualization of these events and the current

exclusive reliance on the experience, judgement, and intuition of practitioners working in

the field. This lack of decision-making guidance is particularly problematic in the dynamic

context of hostage and barricade incidents, which are characterized by a high risk of vio-

lence, high-stakes outcomes, time pressure, and limited information. Consequently, there

seems to be a need for empirical assistance that can support police decision makers in their

efforts to peacefully resolve such complex and volatile situations. This aid can be ensured

by quantitatively modeling and subsequently predicting perpetrators’ behaviors during crit-

ical incidents.

CRIMINAL JUSTICE AND BEHAVIOR, Vol. 35 No. 9, September 2008 1136-1155

DOI: 10.1177/0093854808319955

© 2008 International Association for Correctional and Forensic Psychology

AUTHORS’NOTE: An earlier draft of this article was presented at the 2007 annual meeting of the American

Society of Criminology in Atlanta, Georgia. Please note that the content of this article does not necessarily

reflect the views and opinions of the Sûreté du Québec. Correspondence concerning this article should be

addressed to Patrick Michaud, School of Criminology, University of Montreal, C.P. 6128 succursale centre-

ville, Montreal, Quebec, Canada, H3C 3J7; e-mail: patrick.michaud@umontreal.ca.

Michaud et al. / PREDICTIVE MODELING IN CRITICAL INCIDENTS 1137

THE PERPETUAL RISK OF VIOLENT BEHAVIOR

The prediction of imminent violent behavior is of crucial importance to decision makers

managing hostage and barricade incidents, mainly because of the omnipresent risk of violence.

There are several reasons for this. First, the instigators of these crisis situations are disorga-

nized, exhibit impaired judgement, and have, especially in the first 45 minutes of the incidents,

a limited ability to fully appreciate the consequences of their actions (Noesner & Dolan, 1992).

Second, the consumption of intoxicants—observed in two thirds of perpetrators (Michaud, St-

Yves, & Guay, 2008)—stimulates irritability and aggressiveness, which accentuates this impair-

ment in judgement and increases the propensity to violence (Fagan, 1990, 1993; Murphy &

Wetzel, 1990; Parker & Auerhahn, 1998). Third, the perpetrators are generally unprepared for

the arrival of the police, as the emergency call to authorities is usually made by a witness or

someone close to them (Michaud et al., 2008). The arrival of the police is therefore an addi-

tional emotional burden and a potential source of frustration. Fourth, the recognition by the

perpetrators of the only two probable—but not initially expected—outcomes of the incident

(surrender, with eventual criminal conviction, or death) is surely a significant cause of dissat-

isfaction that can promote violence. Finally, perpetrators may be driven to violence by nego-

tiators’ incessant attempts to establish verbal contact, by their realization of their precarious

position (besieged, with a limited range of movement), or by their fear of being overcome by

force. Police experience indicates that the potential for violence is higher when the perpetrator

is barricaded in his or her home (Flood & Dalfonzo, 2005) or when he or she disposes of a

considerable arsenal, such as multiple weapons and ample ammunition (Strentz, 2005).

PREDICTING VIOLENCE IN THE ABSENCE OF VALID AND RELIABLE TOOLS

According to Meehl (1954), there are two approaches to risk assessment: application of

predictive actuarial rules based on statistical probability and clinical prognostics based on

practical experience. In critical incidents, each of these approaches has limitations that sig-

nificantly alter its effectiveness in predicting violence.

The generic actuarial instruments currently used to predict violence are ill suited to the

operational realities of hostage and barricade incidents. First, none of them have been val-

idated for the populations encountered in such situations. Second, although the goal in

hostage and barricade incidents is the prediction of imminent violent behavior (i.e., vio-

lence in the first few minutes and hours of the incident), the majority of these instruments

focus on the prediction of violence over the short (days, weeks) or long (months, years)

term. Examples of the former are the Broset Violence Checklist (Almvik, Woods, &

Rasmussen, 2000) and the Short-Term Assessment of Risk and Treatability (Webster,

Martin, Brink, Nicholls, & Middleton, 2004), whereas examples of the latter are Hare’s

Psychopathy Checklist (Hare, 1991, 2003), the Historical, Clinical and Risk Management

20-item scale (Webster, Douglas, Eaves, & Hart, 1997), and the Violence Risk Assessment

Guide (Quinsey, Harris, Rice, & Cormier, 1998). Third, this type of high-risk situation is

associated with decisional urgency, a context for which existing actuarial instruments have

not been developed. For example, some checklists may take considerable time to complete

or may require information that may not be available at the time of assessment. Both of

these requirements may lead to delays in assessing the risk.