Stress in Policing : Research and Practice 9781846634772, 9781846634765

Citation preview

pijpsm cover (i).qxd

22/05/2007

11:18

Page 1

ISSN 1363-951X

Volume 30 Number 2 2007

Policing An International Journal of Police Strategies & Management

Stress in policing: research and practice Guest Editor: Professor Ronald Burke

www.emeraldinsight.com

Policing: An International Journal of Police Strategies & Management

ISSN 1363-951X Volume 30 Number 2 2007

Stress in policing: research and practice Guest Editor Professor Ronald Burke

Access this journal online _______________________________ 163 Editorial advisory board _________________________________ 164 Editorial ___________________________________________________ 165 Guest edtorial _____________________________________________ 166 Police stress: history, contributing factors, symptoms, and interventions Judith A. Waters and William Ussery _______________________________

169

Post-traumatic stress symptoms and cortisol patterns among police officers John M. Violanti, Michael Andrew, Cecil M. Burchfiel, Tara A. Hartley, Luenda E. Charles and Diane B. Miller ______________________________

189

Obesity and sleep: the Buffalo Police health study Luenda E. Charles, Cecil M. Burchfiel, Desta Fekedulegn, Michael E. Andrew, John M. Violanti and Bryan Vila __________________

203

Shift work and sleep: the Buffalo Police health study Luenda E. Charles, Cecil M. Burchfiel, Desta Fekedulegn, Bryan Vila, Tara A. Hartley, James Slaven, Anna Mnatsakanova and John M. Violanti

Access this journal electronically The current and past volumes of this journal are available at:

www.emeraldinsight.com/1363-951X.htm You can also search more than 150 additional Emerald journals in Emerald Management Xtra (www.emeraldinsight.com) See page following contents for full details of what your access includes.

215

CONTENTS

CONTENTS

Suicidal ideation among police officers in Norway Ronald J. Burke and Aslaug Mikkelsen ______________________________

228

continued

The impact of COMPSTAT on reported crime in Queensland Lorraine Mazerolle, Sacha Rombouts and James McBroom______________

237

Variation in police patrol practices: the precinct as a sub-organizational level of analysis Kimberly D. Hassell______________________________________________

257

Race and police reliance on suspicious non-verbal cues Richard R. Johnson ______________________________________________

277

Are causes of police stress global? Testing the effects of common police stressors on the Turkish National Police Hasan Buker and Filip Wiecko ____________________________________

291

Police officers’ assessment of operational situations R. Flin, Z. Pender, L. Wujec, V. Grant and E. Stewart _________________

310

Perspectives on policing _________________________________ 324 Policing on the web _______________________________________ 329 Book review _______________________________________________ 330

www.emeraldinsight.com/pijpsm.htm As a subscriber to this journal, you can benefit from instant, electronic access to this title via Emerald Management Xtra. Your access includes a variety of features that increase the value of your journal subscription.

How to access this journal electronically To benefit from electronic access to this journal, please contact [email protected] A set of login details will then be provided to you. Should you wish to access via IP, please provide these details in your e-mail. Once registration is completed, your institution will have instant access to all articles through the journal’s Table of Contents page at www.emeraldinsight.com/1363-951X.htm More information about the journal is also available at www.emeraldinsight.com/ pijpsm.htm Our liberal institution-wide licence allows everyone within your institution to access your journal electronically, making your subscription more cost-effective. Our web site has been designed to provide you with a comprehensive, simple system that needs only minimum administration. Access is available via IP authentication or username and password. Emerald online training services Visit www.emeraldinsight.com/training and take an Emerald online tour to help you get the most from your subscription.

Key features of Emerald electronic journals Automatic permission to make up to 25 copies of individual articles This facility can be used for training purposes, course notes, seminars etc. This only applies to articles of which Emerald owns copyright. For further details visit www.emeraldinsight.com/ copyright Online publishing and archiving As well as current volumes of the journal, you can also gain access to past volumes on the internet via Emerald Management Xtra. You can browse or search these databases for relevant articles. Key readings This feature provides abstracts of related articles chosen by the journal editor, selected to provide readers with current awareness of interesting articles from other publications in the field. Reference linking Direct links from the journal article references to abstracts of the most influential articles cited. Where possible, this link is to the full text of the article. E-mail an article Allows users to e-mail links to relevant and interesting articles to another computer for later use, reference or printing purposes. Structured abstracts Emerald structured abstracts provide consistent, clear and informative summaries of the content of the articles, allowing faster evaluation of papers.

Additional complimentary services available Your access includes a variety of features that add to the functionality and value of your journal subscription: Xtra resources and collections When you register your journal subscription online, you will gain access to Xtra resources for Librarians, Faculty, Authors, Researchers, Deans and Managers. In addition you can access Emerald Collections, which include case studies, book reviews, guru interviews and literature reviews. E-mail alert services These services allow you to be kept up to date with the latest additions to the journal via e-mail, as soon as new material enters the database. Further information about the services available can be found at www.emeraldinsight.com/alerts Emerald Research Connections An online meeting place for the research community where researchers present their own work and interests and seek other researchers for future projects. Register yourself or search our database of researchers at www.emeraldinsight.com/ connections

Choice of access Electronic access to this journal is available via a number of channels. Our web site www.emeraldinsight.com is the recommended means of electronic access, as it provides fully searchable and value added access to the complete content of the journal. However, you can also access and search the article content of this journal through the following journal delivery services: EBSCOHost Electronic Journals Service ejournals.ebsco.com Informatics J-Gate www.j-gate.informindia.co.in Ingenta www.ingenta.com Minerva Electronic Online Services www.minerva.at OCLC FirstSearch www.oclc.org/firstsearch SilverLinker www.ovid.com SwetsWise www.swetswise.com

Emerald Customer Support For customer support and technical help contact: E-mail [email protected] Web www.emeraldinsight.com/customercharter Tel +44 (0) 1274 785278 Fax +44 (0) 1274 785201

PIJPSM 30,2

164

EDITORIAL ADVISORY BOARD Dr Trevor Bennett Centre for Criminology, University of Glamorgan, UK

Professor Paul Mazerolle Key Centre for Ethics, Law, Justice and Governance, Griffith University, Australia

Inspector Patrick J. Devlin New York City Police Dept, New York, USA

Professor Mark H. Moore Kennedy School of Government, Harvard University, USA

Professor Graham Farrell Midlands Centre for Criminology and Criminal Justice, Loughborough University, UK John Firman International Association of Chiefs of Police Research Program Information, Alexandria, USA Dr James Frank Division of Criminal Justice, University of Cincinnati, USA Dr Lorie A. Fridell University of South Florida, USA

Policing: An International Journal of Police Strategies & Management Vol. 30 No. 2, 2007 p. 164 # Emerald Group Publishing Limited 1363-951X

Dr Kenneth Novak Department of Sociology, Criminal Justice & Criminology, University of Missouri-Kansas City, USA Dennis E. Nowicki Chief of Police, Law Enforcement Center, Charlotte, NC, USA Dr Christy A. Visher The Urban Institute, Washington DC, USA

Chief Joseph McNamara Hoover Institute, Stanford University, USA

Dr Sam Walker Criminal Justice-Annex, University of Nebraska at Omaha, USA

Professor Stephen D. Mastrofski Administration of Justice, George Mason University, USA

Dr David Weisburd Institute of Criminology, Jerusalem, Israel and University of Maryland, College Park, USA

Editorial

Editorial In this issue we are pleased to include a collection of papers exploring the topic of police officer stress. Professor Ronald Burke edited this special collection and we want to thank him for his efforts in bringing these papers to publication. In addition to these five papers, the issue contains five more articles, including a sixth paper exploring causes and correlates of police stress. The remaining articles address a several topics of importance to police management and operations. These include an evaluation of the impact of a COMPSTAT process on policing in Queensland, Australia, a assessment of the role of race in the development of suspicion by police officers, a discussion of police research using the precinct as the level of analysis, and officer assessment of operational situations. Each of these papers investigates what the police do, how they do it, and what may be the correlates of police practice. As usual, we believe this issue will be valuable to police administrators and scholars who are seeking improved understandings of the police and police practice. The issue also contains the “Perspectives on Policing” feature, which reviews and summarizes much of the latest published research on policing. A “Policing on the Web” review is also included. Finally, this issue contains a review of the recent book, Citizens, Cops, and Power. We encourage you to consider contributing to the journal as well through preparing reviews, submitting articles, or even simply suggesting topics for future issues. As always, we welcome your suggestions and opinions about how we can improve the journal. Please feel free to contact us, should you have questions, comments, or concerns. Also, consider volunteering to serve as a manuscript referee. For the journal to continue its tradition of high standards requires the participation of police practitioners and scholars in the important roles of author and reviewer.

165

Lawrence F. Travis III

Policing: An International Journal of Police Strategies & Management Vol. 30 No. 2, 2007 p. 165 q Emerald Group Publishing Limited 1363-951X

PIJPSM 30,2

166

Policing: An International Journal of Police Strategies & Management Vol. 30 No. 2, 2007 pp. 166-168 q Emerald Group Publishing Limited 1363-951X

Guest editorial I, as have many others, have undertaken various police studies over the past 20 years, very much influenced by the stressor-strain framework. These studies have contributed a great deal to our understanding of the work experience of police officers, their satisfactions and health, and as relevant prevention and treatment programs for those needing such help. Unfortunately, less attention has been given to the more positive character traits (courage, optimism) and the more enriching experiences (work engagement, helpfulness) of their jobs These also contribute to the psychological and physical well-being of police officers and their families. Although it has been suggested that suicide rates are higher in policing than in other occupations, the evidence presents mixed results. Ronald Burke and Aslaug Mikkelsen consider potential predictors of suicidal ideation in a large sample of Norwegian police officers. Only one officer in the sample had indicated an annual suicide attempt. Single police officers, officers reporting higher levels of both exhaustion and cynicism and, officers reporting less active coping and lower levels of social support indicated more suicidal ideation. Implications for prevention are discussed. Policing is a unique profession in several ways. Few others entrust individuals with responsibility for the safety and security of the general public at large, allocate powers of stop, search and seizure to them and, in some cases, equip them with lethal weapons. It is also a profession that occasionally places officers in situations with hostile individuals, and fortunately less frequently, requires them to discharge their weapons. There is a lot of mythology among the general public about police work. It is commonly believed that police work is among the most stressful and demanding occupations of any. The evidence supporting such a conclusion is mixed however. It has been shown that police officers are less burned out than the average occupation (see Schaufeli and Enzmann, 1998) In addition, studies have police stress have shown that the most common and negative aspects of policing are the result of bad police management and bureaucratic structures while stopping citizens, making arrests and discharging their weapons were reported as less stressful. Officers are trained in ways of dealing with the general public; it is harder to equip them in ways of dealing with a bad supervisor, potential bias in promotion processes, and courts that are soft on criminals. Other researchers have reported that the police profession is characterized by higher levels of marriage breakdown, spouse abuse, consumption of alcohol and suicide (Violanti, 1996). In addition, Kop et al. (1999) reported that levels of officer burnout were associated with more negative attitudes towards the general public, greater advocacy for the use of force in particular situations and greater force observed in actual dealings with the public. Preparation of this special issue was supported in part by the Schulich School of Business, York University. Louise Coutu managed the correspondence with contributors.

It seems important to better understand the antecedents and consequences of police stress given the important functions policing serves in any community, leaving moot the question of whether policing places officers under more, the same or less stress than members of other professions. The Toronto Star (Monday, December 4, 2006, p. A1) contained the following story headline “Toronto cops have pots: Survey”. It reported the results of a police sponsored health survey of Toronto police officers and concluded that too many officers were overweight, drank too much, smoked and had unhealthy eating habits. About 20 percent admitted they were obese. These health challenges were likely to have adverse effects on both job performance and health care costs. Shift work was identified as one possible cause of the poor eating habits and obesity. We are fortunate in having contributions in this collection that consider police officer lifestyles, sleeping patterns, and health. Given the role that policing plays in society, there is an inevitable tension in the police-public relationship; wariness by officers in their dealings with the public, police officers are more likely to “let their hair down” with other officers. We are fortunate in having several former and current police officers, some turned academic, among the contributors to this special issue. In addition, some of our contributors have devoted much of their personal and professional lives to understanding the policing experience and attempting to work with forces to improve the effectiveness and wellbeing of officers, and have, as a result, earned well-deserved reputations among the pre-eminent authorities in this field. They are a well-respected but difficult to sway audience. Contributors to this special issue present data from four countries having varied cultures, values and policing challenges: Norway and the USA. Norway reports relatively lower crime rates than does the USA. As a result the nature of policing and the challenges faced by police officers are likely to be slightly different in various countries. This collection attempts to capture this diversity. In addition, this collection emphasizes in a balanced way both research findings and practice, and the demands of the profession as well as the rewards. Overview of the contents Judith Waters and William Ussery, in a powerful way, highlight the stressors in an occupation at potential risk, serving as the launch pad for the articles that follow. They first review the history of police stress studies. They then describe prevention and treatment programs which unfortunately have not been sufficiently utilized because of the police culture. They conclude with a description of COP-2-COP. a confidential hotline for officers and their families staffed by retired officers and licensed professionals. Although work stress researchers are encouraged to include objective health indicators, few do. We are fortunate in having some contributions that do just that. John Violanti and his colleagues examined the association of post-traumatic stress disorder (PSTD) symptoms and salivary cortisol parameters. Police offers in the moderate and severe PSTD symptom categories had higher mean cortisol values. Events in police work may produce a constant hyper-vigilant state of arousal. Exposure to traumatic work events emphasize the need for more work on the impact of PSTD on police personnel.

Guest editorial

167

PIJPSM 30,2

168

Luenda Charles and her colleagues examined obesity and sleep problems among police officers. Previous research has suggested that police officers are more sleep deprived than the general public, that police officers are at least as obese as the general public, and that both sleep deprivation and obesity are associated with physical health problems. They found some support for the relationship of both sleep disorders and breathing problems and obesity in a randomly selected sample of 115 police officers from a single urban department. Most police officers (905) reported being tired upon awakening. The high incidence of tiredness among officers is a cause for concern however. In a related study, Luenda Charles and her colleagues use the same randomly selected sample of police officers in an examination of the effects of working the night shift on sleep disorders. They found, after controlling various personal and work situation factors, that police officers working nights slept fewer hours and had an increased prevalence of snoring. Finally, Ronald Burke and Aslaug Mikkelsen considered potential predictors of suicidal ideation among a large sample of Norwegian police officers. Although there is some evidence suggesting that suicide is a leading cause of death among police officers, only one officer indicated that he had attempted suicide. Single officers, officers indicating higher levels of burnout (exhaustion, cynicism) and officers engaging in less active coping and having less social support indicated more suicidal ideation. They conclude with some suggestions on suicide prevention. Ronald J. Burke Guest Editor References Kop, N., Euwema, M. and Schaufeli, W.B. (1999), “Burnout, job stress and violent behavior among Dutch police officers”, Work and Stress, Vol. 13, pp. 326-40. Schaufeli, W.S. and Enzmann, D. (1998), The Burnout Companion to Study and Practice: A Critical Analysis, Taylor & Francis, London. Violanti, J. (1996), Police Suicide: Epidemic in Blue, Charles Thomas, Springfield, IL.

The current issue and full text archive of this journal is available at www.emeraldinsight.com/1363-951X.htm

Police stress: history, contributing factors, symptoms, and interventions Judith A. Waters

Police stress: contributing factors 169

Psychology Department, Fairleigh Dickinson University, Madison, New Jersey, USA, and

William Ussery University of Medicine and Dentistry of New Jersey, Middlesex, New Jersey, USA Abstract Purpose – The purpose of this paper is to highlight the stressors involved in an occupation at potential risk – the profession of law enforcement. Design/methodology/approach – The paper reviews the history of police stress studies. It describes prevention and treatment programs that have unfortunately not been sufficiently utilized because of the police culture. Findings – The documented symptoms of stress include digestive orders, cardiovascular diseas, alcoholism, domestic violence, post-traumatic stress disorder, depression and suicide. While some police officers start their careers in excellent physical health, some retire early or even die from job-related stress disorders if the cumulative impact of stress exacts its toll. Originality/value – The paper offers a description of COP.2.COP a confidential hotline for officers and their families staffed by retired officers and licensed professionals. Keywords Police, Post-traumatic stress disorders, Stress, Law enforcement, Occupational therapy, United States of America Paper type Literature review

The nature of the “job” On Monday, August 7, 2006, Detective Kieran J. Shields and his partner, Detective Dave Thompson were dispatched to Taylor Street in Orange, New Jersey, to investigate a shooting (Kelley and Holl, 2006). At the scene, Detective Shields, 32, was shot to death, allegedly by a 19-year-old suspect. Despite the fact that Shields was wearing a bulletproof jacket, he was not protected. The shotgun blast penetrated both his neck and collarbone. Detective Shields, who came from a police family that included his own father, a retired officer from the Orange, New Jersey Police Department, had been recently promoted to detective after only four years on the force. He had already received several commendations and regularly volunteered his time to work with juvenile offenders. Detective Shields leaves a wife, two daughters (aged ten and seven) and an infant son. He grew up in Orange, developed his desire to be a police officer in Orange, joined the force and was promoted in Orange, and died in Orange. Detective Shields was murdered at a time when the number of murders in the State of New Jersey had recently risen by 7 percent to 418 victims. According to the latest edition of the New Jersey Uniform Crime Report, 12 percent of these homicides were related to gang activity (cited in Schewber and Holl, 2006). Being killed or injured

Policing: An International Journal of Police Strategies & Management Vol. 30 No. 2, 2007 pp. 169-188 q Emerald Group Publishing Limited 1363-951X DOI 10.1108/13639510710753199

PIJPSM 30,2

170

during the commission of a crime is one of the major challenges facing police officers and their families. In the August 9, 2006 edition of The New York Times, the Metro Briefing Section reported two deaths in a shooting in Queens, New York; the case of a man who died after a struggle with police officers in Pelham Manor, New York; the case of 12 people who were arrested on gang related charges; the case of a man who was killed and another was wounded in Brooklyn; and the case of a man who was arrested for racing his car at 130 mph with another man on the Long Island Expressway. These reports were the only ones found to be interesting to the newspaper. They did not include all the dangerous domestic violence cases or the myriad other calls that occur on a daily basis when the community reaches out for police protection. On the very same day that Detective Shields’ murder was reported (August 9, 2006), The New York Times also carried the story of two New Jersey Transit police officers who were suspended without pay for having sexual relations with a woman in their patrol car near Liberty Park in Jersey City on July 29, 2006. Although the two officers will not be charged with a sex crime since no coercion appears to have been involved, they will certainly face internal administration charges of “conduct unbecoming an officer”. The outcome could easily include the loss of their jobs. Every day police officers are faced with the challenges of the job, plus the opportunities far less than professional behavior. On August 10, 2006, a headline in the New York Times read “After Long Stress, Newsman in New Orleans Unravels” (Saulny, 2006). The story is about John McCusker, a photographer for Times-Picayune, who decided not to evacuate to Baton Rouge with the rest of the newspaper’s staff. Using a kayak and even swimming through the muddy waters and debris, he took pictures of the dramatic aftermath of Hurricane Katrina. He also personally experienced the misery and trauma that he photographed, having lost all his belongings, his family’s home, and the entire neighborhood where he had lived. The incident reported in the New York Times began as a traffic stop for erratic driving. However, when stopped, Mr McClusker backed up and apparently used his car as a weapon to pin an officer between two cars before he sped away, subsequently driving into other cars. He was finally stopped again. Both times he begged the officers to shoot him. He said he just wanted to die. According to James Arey, Commander of the Negotiation Team of the New Orleans Police Department’s Special Operations Division, McCusker made every effort to hurt the officers with his car. Arey pointed out that,“Our officers are well trained to recognize crises and attempts at suicide by cop” and that’s what this was (p. A21). The mass media image of police officers, outside of dramatic accounts of dangerous rescues and heroic responses to disasters, tends to focus on stories that will engage readers and sell newspapers. Consequently, stories of the bad cop, the brutal cop, the corrupt cop, the tainted authority figures may make more interesting reading than the good cop just doing his job (Waters et al., 1982). On the other hand, creating a superman image can make life even more difficult for the officer than it normally would be. In sum, police officers must be good psychologists at the same time they must secure the safety of the public and investigate the scene of a crime; address the needs of the victims, witnesses, and perpetrators; and face troubled individuals who may try to kill them or try to commit suicide. They must also be aware of the possibility that there are phantom assailants who are not immediately visible and they must still be able to keep their own reactions under control.

The events associated with Hurricane Katrina brought out both the best and the worst in the officers of the New Orleans Police Departments (CNN.com, September 13, 2005). In the days following the hurricane, officers performed acts of heroism in saving the victims of flooding, two officers committed suicide, and dozens (perhaps as many as 200) turned in their badges. The two officers who committed suicide were characterized as outstanding cops who used their own guns to take their lives. It should be noted that some of the officers who did not report for duty may have been trapped in their homes and were therefore unable to report to the command centers. In reality, large numbers of officers worked long hours, slept in their cars, went without normal sanitation facilities, and wore borrowed clothing because they had lost everything. History of police stress studies The study of stress, the identification of contributing factors and symptoms, and the development of prevention and treatment programs began in earnest in the mid to late twentieth century (Lindemann, 1944; Mantell, 1994; Maslach and Jackson, 1979; Maslach, 1982; Mitchell, 1983; Mitchell and Everly, 1993; Reese, 1987; Russell and Beigel, 1990; Waters et al., 1982). While, in those early years, some police agencies did establish employee assistance programs, fund conferences, conduct research, and establish prevention programs, the incidence of police stress continued to escalate and affect many officers and their families. Most officers, however, have not utilized the services that were available due to the strong cultural influences of the law enforcement profession. In order to address cultural impediments, two factors have to be taken into account; confidentiality and the competence of the counselors. In 2000, in New Jersey, COP-2-COP, the first confidential hotline for police officers and their families was established. It utilizes retired police officers trained in assessment and in crisis intervention techniques to answer the phones (Ussery and Waters, 2006). These volunteer counselors conduct interviews that can lead to referrals to licensed mental health professionals who have police experience. So far, the hotline staff has answered over 18,000 calls. COP-2-COP was expanded following the events of September 11, 2001, to respond to the needs of the survivors of the World Trade Center disaster. Retired Police Lieutenant William Ussery, Clinical Supervisor of COP-2-COp, also went to New Orleans to assist in the Critical Incident Stress Debriefing process after Hurricane Katrina. There is already sufficient evidence in the medical and mental health literature to demonstrate the relationships among various categories of life events, predisposing factors, and a broad spectrum of physical and psychological symptoms not only in populations-at-risk (e.g. law enforcement officers), but also in the general public. The symptoms that have been studied in police officers include poor job performance, increased accidents, sleep disturbances, marital discord, domestic violence, post-traumatic stress disorder, depression, suicide, alcohol and other drug abuse, ulcers and other digestive disorders, respiratory ailments, and cardiovascular disease. In the study of the consequences of the Oklahoma City Federal Building bombing (North et al., 1999), the researchers found elevated rates of post-raumatic stress disorder (PTSD) (American Psychiatric Association, 2000). The higher incidence of PTSD was associated more with some categories of survivors than others. These rates were often correlated with previous life experiences and the occupational tasks of

Police stress: contributing factors 171

PIJPSM 30,2

172

at-risk groups. There is no doubt that police officers, as an occupational category, are exposed to more acute and chronic life stressors than most other occupations and are vulnerable to the development of most, if not all, of these symptoms. The fact that police officers begin their careers in excellent physical health and retire early or die from job related stress disorders demonstrates the cost of continuous pressure and the need for ongoing emotional readjustment. While not everyone in a hazardous profession exhibits discernible symptoms of stress immediately following a traumatic incident, in the long run, the cumulative impact of stress exacts its toll. Of course, the cost associated with any stressful life event is a function of how each individual perceives that event. What may appear to be an exciting challenge to one person may seem like a threat to another. Also, since some people are capable of ventilating their feelings and discharging their emotions, they don’t suffer as much as others from stressful life events. Suppressed emotions are often a precursor to the development of stress related disorders (Weisinger, 1985). Obviously, the inherent nature of police work precludes the immediate discharge of emotions. It is certainly not appropriate behavior for a police officer who has been given the responsibility of maintaining stability in others to ventilate in public. Another way of deflecting the influence of stressful life events is to develop effective coping mechanisms. A few individuals who don’t appear to suffer from any symptoms include people who seem almost immune to stress. They may, in fact, be less sensitive to the human condition than people who do experience stress. Hurricane Katrina has already been identified as the worst natural disaster in the history of the USA. More than one million Americans were forced to leave their homes. Many of them will never return (American Psychological Association, 2006). According to The American Psychological Association (APA), the victims of natural disasters develop both psychological and physical health problems that are often mediated by the survivor’s culture as well as by social and historic factors. Culture can determine just which events will be defined as traumatic and which will be down graded. Thus, the survivor can feel shame as well as anxiety for becoming upset by an experience that his/her culture does not identify as a traumatic event. The responses to traumas fall into three stages (American Psychological Association, 2006). The first stage involves the immediate response, one that occurs in the direct aftermath of the disaster and coincides with emergency medical and humanitarian interventions. The next stage follows after the immediate emergency has passed. The third stage is comprised of the long-term responses that can emerge weeks, months, or even years after the actual event. It follows that psychological interventions must be matched to the stage as well as to the cultural background of the survivors. In the early 1970s, Karl Goodin, Chief of the Cincinnati Police Department, gave some of the first seminars to address the issues of police work and stress. He stated that police work was one of the most stressful occupations. He went on to note that many officers suffer from health problems including heart attacks, depression, and suicide in numbers that are much higher than individuals in the business world or government service and that career police officers die at younger ages than members of other occupational categories. “Police culture leads officers to believe that they are a special population that has superhuman abilities and no weaknesses” when actually they are particularly vulnerable due to their need for constant vigilance. Not only does

the individual officer deny his or her risk factors, but departments also ignore the problem. Within departments and academies, attention has always been paid to training for job related skills and to the need for up-to-date equipment. Less concern has been directed toward physical health and toward mental health and resiliency, proven tools for survival. While officer candidates have long been subjected to rigorous physical examinations, psychiatric screening has only been used recently. Moreover, current testing procedures are still underutilized and when they are used, the results are often misinterpreted. The study of police stress coincides with the development of applied psychology following World War II. With the growth of community psychology, it was recognized that environmental and cultural factors have a strong influence mental health, particularly responses to crises. Lindemann’s (1944) early study of the survivors of the Coconut Grove nightclub fire led the way for research on the impact of traumatic events on the development of a broad spectrum of symptoms. Law enforcement officers were identified as an occupation-at-risk due to their actual exposure to threatening situations and to their need for constant vigilance with respect to the “phantom assailant” behind the next door. Unfortunately, police culture interfered and still interferes with obtaining accurate information about suicides. Police officers tend not only to avoid discussions about suicide, but also to be highly resistant to any form of educational preventions programs or treatment for imminent acts of self destruction. The recognition of law enforcement as an occupation-at-risk for environmental, cultural, and personal factors did generate some prevention and treatment efforts such as Goodin’s lectures. Mental health professionals began to teach in training academies and do “road shows” going to conferences or to individual departments for each shift. They also worked with officers referred through employee assistance programs designed to address the issues of officers involved in shootings. In 1982, Waters, Irons, and Finkle published a scale that identified the major stressors in police work and in their private lives that were most troubling to officers (Waters et al., 1982). However, it was not until the end of the year 2000 that the first hotline was opened to respond to officers in need of assistance (Ussery and Waters, 2006). Following a series of police suicides in New Jersey from 1996 to 1998, a group of community leaders was able to lobby the State Legislature successfully to create a bill that would establish and fund a dedicated hotline for law enforcement officers and their families. Thus, COP-2-COP was operational before the terrorist attacks of September 11, 2001. The original mission of COP-2-Cop was to provide a 24-hour hotline for officers and their families throughout New Jersey. Both confidentiality and integrity in practice was and are basic components of the program. COP-2-COP provides clinical assessment and peer support by the staff, all of whom are retired police officers who have volunteered their time and have been trained in crisis intervention. The staff is also prepared to deliver Critical Incident Stress Management sessions to first responders in times of tragedy. When a client needs further guidance, he or she is referred to a licensed mental health professional, an individual previously screened for prior experience working with law enforcement personnel. The reputation of COP-2-COP has made it a model for other emergency service crisis intervention programs. The success of COP-2-COP has been attributed to the rapid rapport and the therapeutic alliance formed between the clients served and the counselors. COP-2-COP has received recognition from the New Jersey Governor’s Office, The New York City Police

Police stress: contributing factors 173

PIJPSM 30,2

174

Figure 1. Police stress flowchart

Department, The Port Authority Police Department of New York and New Jersey, The Federal Bureau of Investigation, The United States Secret Service, and The New York Times, as well as other publications. The police stress model In order to design functional prevention programs and/or treat officers-at-risk, it is important to understand the predisposing factors, the nature of the stressful life events experienced by officers and both the transient and long term responses to these events (see Figure 1). The predisposing factors begin with the biological status of the individual. While people outside of law enforcement are not usually required to pass comprehensive health examinations, police officers are given pre-employment “physicals”. Thus, they begin their careers as healthy people, only to develop a broad range of stress related disorders during their careers. The second biological factor, current state of health, refers to the influence that being a working officer has on one’s health. The longer an officer has been on the force, the more his resistance has been worn away. Psychological traits, states, and self expectations all influence the officer’s ability to solve problems and address job related challenges and family issues. If the officer subscribes to the cultural stereotype of law enforcement, he or she will want to be a strong authority figure, and a perceptive problem solver. For many victims of crime, the officer is also the first psychologist on the scene (Waters, 2002). During emergencies such as September 11, 2001, and Hurricane Katrina, the police, firefighters and EMTs were truly the “first responders” (Ussery and Waters, 2006). Many of the pressures experienced by officers are a product of environmental factors alone or a combination of psychological and environmental variables related to occupational issues. For example, one of the more obvious problems with a career in law enforcement is that it must inevitably involve shift work. Constant change of shifts

(on a weekly or bi-weekly basis) leads to serious health problems. Changing sleep patterns, digestive system circadian rhythms, and other bodily functions affects both physical and psychological well-being. The process of readjustment to shift change schedules exacts a toll on each officer. In addition, normal family life is disrupted when the officer must sleep during the day or be absent from special events (e.g. holidays and birthdays) that conflict with his or her tour of duty. For most officers, the roles and tasks associated with being a law enforcement officer can be debilitating. For each call, there is the potential for a violent confrontation with the ever-present possibility of being killed by an unknown assailant behind the next door. In responding to domestic violence calls (often 85 percent of an officer’s normal duties), the couple who have been fighting with each other may both turn on the officer. Domestic violence cases are particularly dangerous since one or both combatants may also have been drinking and/or using drugs. Police officers are always on duty. An officer once told me that going out to dinner with another officer or officers is like being taken to “The Last Supper” with all the officers sitting on one side of the table with their backs to the wall constantly surveying the room. The perpetual need for vigilance, even when off duty, also takes its toll on the officer’s level of resilience. Stressful life events include explosive, implosive and corrosive incidents. Police officers are faced with three types of stress: The explosive events (e.g. crimes in progress, terrorist situations such as September 11, 2001, and natural disasters) lead to acute and severe overt reactions, sometimes against other people. These reactions are often repressed in order for the officer to continue to fulfill his/her role of protecting the public. Long-term consequences follow when there is no intervention to reduce the impact of severe stress reactions. Some events have an implosive effect on the officer because of internal conflicts and the values that guided his/her choice of occupation. The inability to “make a difference”, the conflicts among family, and personal responsibilities, and job related considerations lead to the development of stress symptoms over a period of time. The daily tensions associated with police work have a corrosive effect, eroding confidence and wearing away at the individual’s level of hardiness and resiliency. The essential problem, especially with corrosive events, is that officers do not engage in the type of self protective behaviors that could reduce the price of living a high risk lifestyle. Not only does the individual ignore these long term pressures but the department may also trivialize the negative consequences of police work. The implication is that only the weak suffer stress related symptoms. Hiding one’s feelings becomes a badge of courage. Thus, the police culture combined with the reality of police work, even in the lowest crime rate areas, and the psychological factors that lead an individual to choose police work, all combine to create a situation that aggravates the inherent pressures of law enforcement. Stress responses can be transient or chronic. On any particular day, officers face the normal challenges of the job ranging from the boredom of watching traffic to the potential danger of a domestic violence call to the guaranteed risk of a “break and entry” in progress. The event may literally be the “last straw” in a long career of tension filled days or it may be a situation of such magnitude that no one could have predicted the scope of its impact (e.g. September 11, 2001). Some of the events are acute and only have a short term impact while others can become chronic or are chronic by nature. The resolution of a short-term, low-level event is often a return to the “status

Police stress: contributing factors 175

PIJPSM 30,2

176

quo”. Not everyone reacts the same way to the events associated with law enforcement or to those associated with other high risk occupations. However, we have sufficient data to assume that each officer and the officer’s family and supervisors should expect some sort of stress related response at one time or another, and do everything possible to address these situations with prevention programs and early interventions. Without some form of intervention, transient stress responses can develop into symptoms of physical and/or psychological pathology (e.g. suicidal ideation) requiring treatment. The resolution of even traumatic events can be psychological growth. When an individual is capable of coping with stressful events, he or she may actually develop new and effective stress reduction strategies and become stronger. There are many types of prevention and treatment programs. In order to address the problems of police stress, academies and various other agencies and departments offer courses while the candidates are still in training, schedule speakers to work with each shift, and plan conferences for family members. Police psychologists are available to work with officers in need of counseling and with their spouses and children. Some programs are funded by federal and state sources, and some by private foundations. There are also Alcoholic’s Anonymous meetings specifically scheduled for officers. Not all coping mechanisms are functional. For example, heavy alcohol consumption and substance abuse lead to more problems than they solve. Alan Leschner, former director of the National Institute on Drug Abuse, has stated on numerous occasions that there are two basic reasons for consuming alcohol or taking drugs, “to feel good and to feel better.” Consequently, in an effort to self medicate, the officer may use alcohol, a dysfunctional coping mechanism that is closely associated with poor job performance, domestic violence, and eventually suicide. Despite the negative consequences, as a coping mechanism alcohol is condoned by the law enforcement culture. Joseph Wambaugh, a former police officer, now a novelist, has described “choir practice”, the habit of gathering after a shift to drink and relax. Officers who retire sometimes even purchase police bars so that they can continue to see their peers. The problem is that alcohol consumption aggravates feelings of depression and decreases inhibitions. Alcoholism is related to marital discord, driving accidents, and medical disorders. The question remains about what can be done to prevent police suicide and to reduce occupational stress and dysfunctional behavior in officers. The single most important factor for anyone who is recovering from a crisis is reliance on a dependable support group. Hopefully, the network must have the knowledge and skills to give good advice. In any case, concerned listening is the prime requisite for a counselor or for a friend. We turn a deaf ear most of the time. As a society, we have become accustomed to asking people how they feel without paying the slightest attention to the answer. Working in a drug therapeutic community (milieu therapy for addicts) requires that you actually stop and listen when inquiring about someone else’s health. Failure to do so results in a “pull up”, a form of chastisement for failure to comply with the rules of the house. We need to extend the practice of listening to all our relationships. There are several reasons why police officers avoid contacting mental health professionals (Hackett and Violanti, 2003). First of all, there is a certain level of mistrust between officers and clinicians. Not only does the police culture militate against asking for help, but officers also question the competency of helpers outside of

the field to understand the pressures of law enforcement. Trusting others, even family members, is also seen as a sign of weakness. In addition, there is also a tendency to ignore such symptoms of depression as a decrease in energy, feelings of sadness or worry, and the sense of desperation that permeates one’s thought. Seeking help conflicts with the police image that is composed of individuals who are independent, competent, and trained to take care of dangerous situations, and to protect the public. Officers may also resist seeking help because they justifiably fear losing their jobs or suffering from other job actions. Specifically, police officers are seriously concerned about speaking to mental health professionals. They have legitimate cause for concern if they are later involved in court cases that address legal issues regarding their credibility. If an officer has seen a counselor and is later charged with using excessive force, the records can be subpoenaed. If an officer has filed a Workman’s Compensation claim and the records have been released, these documents may no longer be considered privileged. Officers may also be concerned that their fitness for duty may be called to question if it becomes known that they have seen a counselor. They may not receive a promotion or be considered for a special assignment if there are mental health issues in the way and there are state guidelines that specify that an officer must be free of mental or emotional problems or free from psychopathology to qualify for certain jobs. Thus, officers with stress related problems have a realistic basis for their fears of job action. For an officer who has lost his or her job for cause, there are few viable alternatives in the outside world that utilize the skills acquired in police training or on the job. Officers are also afraid of being prescribed medications by psychiatrists to treat their symptoms. Part of their concern is due to the potential side effects of some antidepressants, and part is based on their own attitudes towards addicts. Officers worry that they will feel out of control or become slow in responding to emergencies. In some cases, they must also report that they are taking medications. Police culture and the family Police officers are often “too proud and too fearful” to seek assistance when it would do the most good for the officer or for his family (Kirschman, 2000, p. 221). The best way to label such behaviors is to invoke Freud’s ubiquitous concept of denial. For example, many police officers deny the possibility that they could be so severely injured that they might need to retire early. Consequently, they rarely have any contingency plan for their future or their family’s security. They are often so deeply affected by the negative aspects of their work lives that the negativity permeates their off duty experiences. Their unexpressed hostility overflows into their home lives and is rarely reduced by positive activities. The officer’s children often bear the brunt of his or her criticism. Moreover, these attacks may be unrelated to the children’s actual behavior. Some police officers see their children’s independence as a challenge to their positions as authority figures and as marks of disrespect. Most police families live in constant fear. The spouses and children of police families worry about the danger that the officer faces every time he or she leaves the house. That growing fear is based on the fact that they don’t know what will happen when the officer goes to work; neither does the officer. Detective Shields did not know. On the morning of September 11, 2001, no one could have anticipated the consequences to the first responders or the other victims of international terrorism. Since those days, we have all become more anxious than we once were. It is a normal way of life now.

Police stress: contributing factors 177

PIJPSM 30,2

178

One would have to be very naı¨ve to believe that terrorism can ever be completely prevented. Kirschman (2000) has summed up the challenge of police work for the family by saying that there is more to life than the badge. One of the greatest risks of police work is that cops become so inflated, narcissistic, and self involved that they chance alienating their real families by over investing their time and energy in the work family, which all too frequently turns out to be fickle and unsupportive. She suggests that, “One of the greatest risks to families is that they surrender their own identity in favor of the officer. The women in this book, especially, have had a hard time focusing on themselves and later regretted losing that focus” (Kirschman, 2000, p. 247). Domestic violence Kirschman (2000) writes in her book, I love a Cop, that domestic violence is the leading cause of injuries to women in the 15-44 age group. It is estimated that double the number of women are raped by husbands or ex-husbands than by strangers. Over 90 percent of the assaults on women are actually a function of their efforts to escape domestic violence. What is equally distressing is that the children are watching. Data suggest that between three and ten million American children witness domestic violence in their own homes. Clearly, home is no safe haven. Research indicates that between 40 to 70 percent of the men who attack women also attack the children. Many women who will tolerate physical and sexual abuse to themselves only seek help when the children are at-risk. What is particularly disturbing is that a woman has more of a chance of being attacked by her domestic partner than a police officer has of being assaulted during his tour. All too often, the batterer is an officer. Kirschman (2000, p. 139) labels domestic abuse as “the best kept secret shame of policing”. Even if domestic violence only occurs at the same rate in police families as homes in the general public, some 60,000 to 180,000 families would be involved every year. We think that it occurs more often in law enforcement. One of the factors may be the divorce rate. Domestic violence actually occurs three times more frequently in divorce or separation situations than in intact households. Divorce almost appears to be a catching disease in police families. Another issue may be alcohol and drug abuse. Alcohol and or drugs are frequently involved in cases of domestic violence. However, according to Roberts (1996) alcohol is a correlative of violence, not the cause. Lowered inhibitions certainly affect the seriousness of the abuse. While cases of domestic violence occur across every segment of society and in every age category, uniformed patrol officers and narcotics officers are reported to have higher rates of domestic abuse than other officers (Kirschman, 2000). One of the problems is that fatigue seems to have a negative effect on self control. Officers working the night or swing shift and/or more than 50 hours per week, and those suffering from sleep deprivation and “burnout” are frequently involved in domestic violence. The question arises as to why individuals who are charged with the safety of the public could be “the most dangerous of all abusers and their spouses the most endangered” (Kirschman, 2000, p. 144). There are several additional critical factors involved. First of all, there is the presence of guns and the ability to use them. Police officers are also skilled in applying verbal and physical force when necessary. They know how the criminal justice system operates and they know the players. There is also the need for control and the pattern of hyper-vigilance plus the expectation that

people will comply with a police officer’s demands and show respect. The police follow a paramilitary model where orders are given and followed, sometimes without explanation. Many officers were in the armed forces. Finally, law enforcement, while undergoing some changes, is still predominantly a man’s world and emphasizes traditional gender roles, applauding action over efforts at communication. Disrespecting their authority undermines their sense of self esteem. Hanks (cited in Kirschman, 2000) postulates three categories of domestic abuse: Stress-related abuse, control related abuse, and special circumstances. In the case of stress-related abuse, the violence appears to follow a single crisis such as job loss or death of a family member. The violence is usually considered to be a serious problem by both partners. Both partners are also concerned that the children have been affected. The prognosis for behavioral charge in this situation is optimistic. Control-related abuse, on the other hand, is an ongoing problem and can assume many faces. The perpetrator rarely takes responsibility for the situation and proceeds to blame the victim. Furthermore, the abuser seems to have no empathy for the victim and is only calmed by abject obedience. The spouse develops symptoms of post-traumatic stress disorder and feels trapped in the situation. Unfortunately, unless the abuser does take responsibility for his/her behavior, therapy may actually do more harm than good. In all three of the categories, there are two necessities: The abusers must be able to control their actions and the targets must learn how to keep themselves and their children free from harm. Parents must also learn to recognize the signs of psychological problems in the children such as depression, anxiety, suicidal ideation, sleep disorders, psychosomatic symptoms, conduct disorders and poor school performances that can result from domestic violence. Treatment of a police officer involved in a case of domestic violence requires an investigation of charges of stalking, harassment, and physical abuse. Any officer identified either as a victim or perpetrator must be mandated to receive counseling (e.g. individual, couple, and family therapy as well as anger management sessions). The victims must be given protection. What is particularly interesting to note is the fact that more police officers have been killed while intervening in cases of domestic violence than in the “war on drugs”. Some of the very same officers trained in the procedures used to address domestic violence cases, actually go home and forget everything that they have learned. It is important to report cases of domestic violence since one of the warning signs for future violence is a past history of violent acts, stalking behavior, and threats (McMurry, 1995). In cases of domestic violence, the threats and abuse may spread from the home and neighborhood to the spouse’s job site because the perpetrator feels a need to control all aspects of the victim’s life. In fact, the spouse’s job itself is a source of stress to the perpetrator. The salary contributes to his or her financial independence and ability to leave the batterer at some time in the future. Consequently, the batterer is in danger of losing control over the situation. Suicide Even now, when we have already identified the sources of police stress and the dysfunctional responses that are associated with high risk occupations such as law enforcement, Reese (1987, p. xvi) points out that there remain “pockets of departmental and administrative resistance”. Supervisors are not always trained to recognize the

Police stress: contributing factors 179

PIJPSM 30,2

180

symptoms of stress, or if they are, they do not take appropriate action. Consequently, many of the ills associated with law enforcement such as alcohol abuse and marital conflict, fester until the results include physical illness, depression, domestic violence, and suicide. Although most departments provide officers with communications equipment, weapons, vehicles, and bullet resistant vests, Reese (1987, p. xix) notes that “we have not yet devised training programs that are capable of bullet-proofing the mind”. In general, police departments tend to deny the very existence of psychological factors. Actually, as we have noted, police officers are at greater risk for suicide than other professions. Due to misreporting, however, the present statistics underestimate the scope of the problem. What we do know is that the suicide rate for Federal Bureau of Investigation agents is 116 percent above the national rate (Field and Jones, 1999, cited in Hackett and Violanti, 2003). In recent years, we have witnessed the aftermath of several critical incidents of such magnitude that post-traumatic stress disorder and the potential for suicide among first responders comes as no surprise. The world has been exposed to natural and manmade disasters for which we were surprisingly unprepared (e.g. The Oklahoma City Federal Building bombing, the events of September 11, 2001, the Tsunami, Hurricane Katrina, and terrorist activities around the world). Any survivor, officer or civilian, who has suffered as a result of one of these events needs proper critical incident debriefing procedures and referrals to experienced mental health professionals. Training in suicide prevention is also necessary for all departments. In law enforcement, the prevention of suicide and other sequelae to traumatic events requires a strong support system. To the police officers, no one is better equipped to comprehend the pressures of law enforcement than another officer. Peer support programs are effective in terms of addressing many types of mental health problem. The concept of peer counseling is not a new idea. Law enforcement officers frequently gravitate toward more experienced colleagues who can serve as mentors in times of crisis. However, while law enforcement experience is critical and helps to establish rapport, training in the principles of crisis intervention and critical incident stress debriefing is also important. The training of peer counselors should be conducted by mental health professionals with a knowledge of counseling skills, crisis intervention theory and practices, early warning signs of acute or chronic stress, suicide lethality assessment, the facts of alcohol and other drug abuse, and the issues of confidentiality. The peer counselor’s mandate is to provide a safe, confidential climate for the client. It is also his or her responsibility to decide whether or not a referral to a licensed mental health professional should be the next step in the treatment process. Peer counselors should be selected based on several criteria, including trustworthiness, sensitivity to the issues of racial diversity and other cultural factors, and the ability to command respect and establish rapport with clients quickly. Motivation is not sufficient. No intervention activity is better than the personnel who implement its policies. The ripple effect of a suicide within a department spreads out to cover other officers, supervisors, members of the family, and friends. According to Mitchell (1983), the suicide of a peer is considered one of the most distressing incidents to face emergency service professionals. The question is always whether or not something could have been done to prevent the suicide. Could, for example, someone have recognized the

causes or precipitating events? Were there substance abuse problems or threats of job loss due to inappropriate behavior? Did the individual actually threaten to kill him or herself? Was there a recent noticeable change in behavior such as a sudden decrease in job performance? Have there been problems with co-workers such as angry outbursts or atypical withdrawals? Among the predictors of suicide (Hackett and Violanti, 2003) are prior attempts, a family history of suicide, a major relationship breakdown such as separation or divorce, an internal investigation or actual criminal charges, and the availability of weapons. The New York Post (July 12 2002) published the story of Daniel E. Steward, a 27-year-old New York City Fire Department medical technician, who hanged himself in the basement of his home on Long Island. He was deeply affected by the job of sifting through the rubble of “Ground Zero” searching for body parts. He left behind a distressing suicide note that described his heartache after removing body and body from the debris of the World Trade Center. Steward had been assigned to his duty for the two weeks immediately after September 11, 2001. However, he continued to search through the concrete and dust even on his days off until January of 2002. Although in the end, he only went to the site one day a week, the emotional trauma was already exacting a price. Despite the fact that “first responders” were being offered counseling in the months following 9/11, Steward only sought treatment on one occasion. Perhaps mandatory counseling for a longer period of time should have been utilized. There are, however, differing opinions on the value of coercion in requiring counseling (Waters, 2002). Steward, who was single, did not discuss his feelings or plans with any of his friends or co-workers. In order to depict the type of officers who call the COP-2-COP hotline and yet maintain confidentiality, we have compiled two fictional case studies. The first case study involves an officer whose normal assignment was patrol. He was among the first officers to the World Trade Center just after the first tower collapsed. In describing the event, he said to us: We were up near a building to avoid the falling bodies. It was terrible. I saw a man or maybe it was a woman, I don’t know which, split in half by a flagpole. I know now that they don’t die or pass out before they hit the ground. I could see them trying to break their falls with their arms and legs outstretched just before they hit the ground.

The officer manifested persistent symptoms of post-traumatic stress disorder. He avoided the debriefing sessions by taking vacation time right after 9/11. Later, in his discussions with a COP-2-COP counselor, he reported that he still had flashbacks, nightmares, and re-experiences the events surrounding 9/11. He said that he would never forget the smell of death coming from the debris that was filled with human remains. The phone counselor advised the officer to accept an individual referral to a licensed therapist. At first he refused, however, after several more calls over a period of three weeks, the officer recognized that he was actually feeling some relief from the phone calls alone and acquiesced to individual treatment. He spent six months in treatment which resulted in marked improvement. Although the officer made significant changes in his lifestyle and reduced the severity of his symptoms, he was made aware that symptoms do reappear and that he would be free to return to treatment or to call the hotline again. The hotline counselor still calls him periodically to provide support and to check on the officer’s stability. If alcohol consumption or drug abuse had been part of the officer’s symptom syndrome, the therapist would have

Police stress: contributing factors 181

PIJPSM 30,2

182

recommended either inpatient or outpatient substance abuse treatment. Treatment plans always reflect the needs of each caller and, when appropriate, may involve the families. The second typical case involves the wife of a law enforcement officer who called COP-2-COP in order to get help for herself, for husband, and for their children. She reported to the hotline counselor that she was afraid of her husband because of his angry threats and actual incidents of domestic violence. She was also concerned about the officer’s personal safety since he had alluded to suicide as his best solution on several occasions. For her own safety, the wife had already had a restraining order taken out against the officer. However, the officer attempted repeatedly to coerce her into dismissing the restraining order. He said that it would cost him his job and the loss of the family’s only source of income. The threat was very daunting since the wife was pregnant and worried about food and a place to live for herself and their two other children. The COP-2-COP staff informed the wife that if her husband did not violate the restraining order, he would not lose his job. The officer has, however, had his duty weapon removed and he has also been suspended from duty, due to the domestic violence complaint. He will be required, again by policy, to have a psychiatric evaluation and a “fit-for-duty” report. In all likelihood, the department will require that the officer have individual counseling sessions and attend a group anger management program. During the psychiatric evaluation, a history of alcohol and narcotics abuse was also uncovered. The substance abuse problem alone could jeopardize his job status. The next step should be some form of treatment. The exact nature would depend on the severity of the abuse. The COP-2-COP staff recognized the complexity of this police family’s problems, but they remained focused on the needs of the client. In this case, the client is the wife, and first issue is the safety of the woman and her children. Domestic violence cases must be taken seriously since they can escalate rapidly from verbal threats to homicide. Even after a divorce, husbands (more than wives) have sought retribution for imagined affronts. What is particularly interesting is that the perpetrators are so convinced about the justice of their claims and their subsequent acts of violence that they don’t even attempt to hide their identities or flee the scene (Waters et al., 2002). A personal sense of entitlement to punish their spouses or ex-spouses tends to override restraining orders and often results in murder and suicide. The staff of COP-2-COP utilizes appropriate facilities and programs to provide support, safe refuges, and legal and practical advice to battered spouses. It should be remembered that all police officers know the location of shelters which are often in close proximity to the police department. Domestic violence, murder, and suicide can be a product of burnout. Burnout, depression, and feelings of helplessness lead some officers to contemplate self-destructive acts while other officers develop hostility particularly towards the criminal element in their communities, especially when racism is involved. Attitudes towards police officers range from praise and hero worship to almost joyous attempts to find flaws and dishonest behavior on the part of all officers. The families suffer the fall-out. Not only do the officers’ schedules interfere with having a normal family life but the disruption of their circadian rhythms leads to temper tantrums and lack of patience. In an effort to improve the predictability of a family life and to enable officers to have some weekend time with their families, a number of

departments have gone to 12-hour shifts and three- or four-day work weeks. In some cases, however, the plan backfired. The officers with predictable free time, have taken second jobs!

Police stress: contributing factors

Divorce Police divorce is so prevalent that in some departments the rate ranges from 50 to 80 percent. With the rates in the general public on the rise, the gap is closing. Marriages that take place after the officer is already on the job last longer than those begun before the individual joins the force due to lack of false expectations.

183

Hurricane Katrina and family conflicts The COP-2-COP staff assisted the Southern Law Enforcement Foundation in providing mental health support to police officers affected by both Hurricanes Katrina and Rita. The devastation of New Orleans was more than most people could have imagined or was depicted by the media. The disruption of regular police functions was a metaphor for the breakdown of society. The actual critical incident occurred when the hurricane had passed. The police were relieved from their regular shifts to stand by in the city. Most of them found rooms in vacant hotels or with friends or went to their own homes, if that was still possible. They had been told to return at 4:00 p.m. Unfortunately, some officers found themselves flooded out or marooned before returning to work. The stories of rapidly rising waters and the need to protect themselves from distraught citizens, were overpowering. One officer swam with her month-old baby to “Headquarters” which was then at the Crystal Palace, the highest ground in the district. There were no communications since police radios, cell phones, and landlines were out. The same forces of nature, which prevented public transportation and communication, destroyed the chain of leadership in the department. There were reports of numerous police desertions which attracted widespread media coverage and public disdain. Many of these officers in the flood zones went home to protect their loved ones. They were torn between two oaths, “to protect and serve” and “to love, honor, and obey.” By now the public is aware that there were at least two suicides of officers in New Orleans. In one case, a respected officer with 32 years in the force, reportedly, was so distraught over the desertions and disorder that he took his own life. This is only one example of the extent of the disequilibrium during the attempt to restore calm that was further complicated by the lack of available mental health resources in the effected areas. Coping strategies Coping strategies can be categorized according to the source of the responsibility: The individual officer, or the department (Waters et al., 1982, p. 25). Individual coping strategies include: . the development of a dependable support system; . improved communication skills; . a means of ventilating feelings appropriately; . a regular exercise program with a minimal time expenditure of 30 minutes a session;

PIJPSM 30,2

.

.

.

184

. . . .

a diet that contains elements necessary for optimal functioning and excludes elements that have negative values (e.g. a high fat diet); the development of other activities that provide for recreation, change of focus, and positive feedback; regular vacations; muscle relaxation exercises; meditation; the use of biofeedback; and participation in self-help groups.

Departmental strategies include: . realistic job-related training in police functions; . open communication channels between officers and supervisors; . opportunities for meaningful input into departmental decisions whenever possible; . reassessment of shift hours; . constructive feedback on job performance; . workshops on dealing with marital conflicts, good parenting procedures, and preparation for retirement, etc.; . training for supervisors in good management techniques; . opportunities for “debriefing” sessions at the end of the shift; and . stress management training. Notes: Since not all strategies will fit the lifestyles of all officers. Each individual must select those techniques that are appropriate to his/her needs. More than one strategy is necessary to deal with the stresses of contemporary life. The COP-2-COP program The COP-2-COP program utilizes several components of successful crisis intervention strategies applied to other populations. For example, the stages in Robert’s Seven Step Crisis Intervention Model are very similar to the COP-2-COP guidelines (Roberts, 1996). It should be noted, however, that many of the steps occur either simultaneously or are repeated during the process. For example, the first and second steps are difficult to separate: (1) plan and conduct a crisis assessment (including measures of lethality); (2) establish rapport and a therapeutic relationship; (3) identify the caller’s major problems including the precipitating events; (4) deal with feelings. Be an active listener and validate the caller’s emotions; (5) generate and explore alternative coping strategies and skills; (6) develop and formulate an appropriate action plan; and (7) establish a follow-up plan and agreement.

Kanel suggests, “Although the ABC model of crisis intervention has a three stage approach, in an actual interview the components of one stage could be used at any time” (Kanel, 2007, p. 70). Sometimes during the interview process, the caseworker needs to be flexible and recognize the needs of the client and apply appropriate, timely interventions. Since the COP-2-COP hotline staff is comprised of retired law enforcement officers helping other officers, establishing rapport and developing a therapeutic relationship is a comparatively simple task. Both the phone staff and the callers have similar backgrounds and occupational experiences, speak the same language, and are concerned with many of the same problems. The assessment guidelines used by the volunteers require answers to the following questions: (1) the nature of the presenting problem; (2) the severity of the situation and the length of time that the problem has existed; (3) the actual impact on the officers ability to function on the job and at home; (4) the precipitants of the situation or the immediate cause of the problem; (5) any past history or current substance abuse; (6) other relevant aspects of past history; (7) previous inpatient and/or outpatient psychiatric history; and (8) current medical problems and medical history. In another section of the initial interview, callers are asked about suicidal or homicidal ideations, whether or not they have access to their duty weapons, or any other firearms or lethal methods. The volunteers, as well as the clinicians who serve as referrals, have been trained and certified as trainers by John Violanti using a suicide prevention tool, “QPR.” Written by Dr Paul G. Quinett in an instructional booklet entitled Suicide, the Forever Decision. The guidelines require that the staff: . question anyone reporting suicidal tendencies or depression symptoms; . persuade the person to seek further assistance; and . refer the person to a mental health professional. Throughout the assessment process, the staff expresses concern for the caller’s feelings. They not only listen to what is being said and to the important issues that are being avoided, they also explore past coping strategies trying to separate the functional from the dysfunctional. During the assessment process, the staff members are careful not to interject their own suggestions. As any experienced therapist knows, clients frequently seek solutions to their problems from the counselor. Why should they do all the work necessary to solve a problem when a short cut is available. It is essential, however, that the callers develop their own solutions with guidance. The caller’s symptoms of depression often permeate the interview. When callers are asked what they do for entertainment, they may respond with descriptions of what they used to do. They may also report on their alcohol use at this point or discuss family problems including marital discord and admit to acts of even domestic violence. The assessment process helps the staff to assist callers in developing a viable action plan. It is critical that each caller participates in the development of the plan so that a sense of personal control is reestablished and cognitive functioning is improved. The

Police stress: contributing factors 185

PIJPSM 30,2

186

staff members try to set up parameters so that the callers are not overwhelmed by the enormity of their problems. The staff members also follow up on problem resolution by contacting the caller every ten days until they are convinced that the callers are stable. If necessary, they will call more frequently. The callback component is valued by both the clients and the staff. The clients appreciate the continuing concern and the staff members are rewarded by the clients’ positive comments. In order to build a list of competent and experienced licensed mental health professionals for referrals, 200 police departments in New Jersey were surveyed for the names of therapists who had been utilized in the past with success. The survey yielded 150 providers with expertise working with the law enforcement community. Even these professionals were given training on such topics as the police personality, post-traumatic stress disorder, and critical incident stress debriefing. Following the events of September 11, 2001, other hotlines were put in place for survivors and for all the emergency service personnel affected by the disasters. The firefighters, police officers, and EMS teams were viewed as victims of a crime. Family members were also welcome to call. Summary The image of police officers is that they are action oriented, problem solvers who are in control of their own emotions. They are supposed to be strong, resilient, and, of course, stoic. To be stoic means that they must remain unaffected by the violent and vicious behaviors that they encounter every day of the week. Feelings are addressed by repressing them. The price of readjustment ranges from simple irritability to heart attacks and suicide. The only way to address the problems is through screening, training, on going prevention programs, and early interventions and treatment based on programs such as the COP-2-COP model. References American Psychological Association (2006), “APA’s response to international and national crises: addressing diverse needs, 2005 Annual Report of the APA Policy and Planning Board”, American Psychologist, Vol. 61 No. 5, pp. 513-21. American Psychiatric Association (2000), Diagnostic and Statistical Manual of Mental Disorders: DSM-IV TR, Author, Washington, DC. Hackett, D.P. and Violanti, J.M. (2003), Police Suicide: Tactics for Prevention, Charles C. Thomas, Springfield, IL. Kanel, K. (2007), A Guide to Crisis Intervention, 3rd ed., Thomson, Belmont, CA. Kelley, T. and Holl, J. (2006), “Suspect held in officer’s shootings in Orange”, The New York Times, pp. B1-B6. Kirschman, E. (2000), I Love a Cop: What Police Families Need to Know, The Guilford Press, New York, NY. Lindemann, E. (1944), “Symptomatology and management of acute grief”, Journal of Neurons and Mental Disease, Vol. 181 No. 11, pp. 709-10. Mantell, M.R. (1994), “Ticking bombs”, Psychology Today, January/February, pp. 20-1. Maslach, C. and Jackson, S. (1979), “Burned out cops and their families”, Psychology Today, Vol. 12 No. 12, pp. 58-62. Maslach, C. (1982), Burnout: The Cost of Caring, Prentice Hall, New York, NY.

McMurry, K. (1995), “Workplace violence: can it be prevented?”, Trial, Vol. 31 No. 12, pp. 10-12. Mitchell, J.T. (1983), “When disaster strikes: the critical incident stress debriefing process”, Journal of Emergency Medical Services, Vol. 8 No. 1, pp. 36-9. Mitchell, J. and Everly, G. (1993), Critical Incident Stress Debriefing: An Operations Manual for the Prevention of Traumatic Stress Among Emergency Services and Disaster Workers, Chevron Publishing, Ellicott City, MD. North, C.S., Nixon, S.J., Shariat, S., Mallonee, S., McMillen, J.C., Spitznagel, E.L. and Smith, E.M. (1999), “Psychiatric disorders among survivors of the Oklahoma City bombing”, Journal of the American Medical Association, Vol. 282, pp. 755-62. Reese, J. (1987), A History of Police Psychological Services, US Government Printing Office, Washington, DC. Roberts, A.R. (1996), “Introduction: myths and realities regarding battered women”, in Roberts, A.R. (Ed.), Helping Battered Women: New Perspectives and Remedies, Oxford University Press, New York, NY, pp. 3-12. Russell, H.E. and Beigel, A. (1990), Understanding Human Behavior for Effective Police Work, Basic Books, New York, NY. Saulny, S. (2006), “After long stress, newsman in New Orleans unravels”, The New York Times, p. A21. Schewber, N. and Holl, J. (2006), “Officials say slain detective and murder suspect crossed paths before”, The New York Times, pp. B1-B6. Waters, J. (2002), “Moving forward from September 11th. A stress/crisis/trauma model”, Brief Therapy and Crisis Intervention, Vol. 2 No. 1. Waters, J., Irons, N. and Finkle, E. (1982), “The police stress inventory: a comparison of events affecting officers and supervisors in rural and urban areas”, Police Stress, Vol. 5 No. 1, pp. 18-25. Waters, J.A., Lynn, R.I. and Morgan, K.I. (2002), “Workplace violence: prevention and intervention, theory and practice”, in Rapp-Paglicci, L.A., Roberts, A.R. and Wodarski, J.S. (Eds), Handbook of Violence, John Wiley & Sons, Inc., New York, NY, pp. 378-413. Ussery, W.J. and Waters, J.A. (2006), COP-2-COP hotlines: programs to address the needs of first responders and their families. Brief Treatment and Crisis Intervention. Weisinger, H. (1985), The Anger Workout Book, Quill, New York, NY.

Further reading American Psychological Association (2001), “Response from APA [To Herbert et al., “Primum non nocere.”]”, APA Monitor on Psychology, Vol. 32 No. 10, pp. 4-8. Black, J. (2000), “Personality testing and police selection. Utility of the “Big Fire”, New Zealand Journal of Psychology, Vol. 29 No. 2, p. 24, available at: www.questia.com (accessed August 24, 2004). Dohrenwend, B.S. (1978), “Social stress and community psychology”, American Journal of Community Psychology, Vol. 6 No. 1, p. 2. Everly, G.S., Lating, J.M. and Mitchell, J.T. (2000), “Innovations in group crisis intervention: critical incident stress debriefing (CISD) and critical incident stress management (CISM)”, in Roberts, A.R. (Ed.), Crisis Intervention Handbook: Assessment Treatment and Research, Oxford University Press, New York, NY, pp. 77-97. Goode, E. (2001), “Some therapists fear services could backfire”, New York Times, p. L21.

Police stress: contributing factors 187

PIJPSM 30,2

188

Herbert, J.D., Lilienfeld, S., Kline, J., Montomery, R., Lohr, J., Brandsma, L., Meadows, E., Jacobs, W.S., Goldstein, N., Gist, R., McNally, R.J., Acierno, R., Harris, M., Devilly, G.J., Bryant, R., Eisman, H.D., Kleinknecht, R., Rosen, G.M. and Foa, E. (2001), ““Primum non nocere” [Letter to the editors]”, APA Monitor on Psychology, Vol. 32 No. 10, p. 4. von der Kolk, B.A., Weisaeth, L. and von der Hart, O. (1996), “History of trauma in psychiatry”, in von der Kolk, B.A., McFarlane, A.C. and Weisaeth, L. (Eds), Traumatic Stress: The Effects of Overwhelming Experience on Mind, Body, and Society, The Guilford Press, New York, NY. Violanti, J.M. (1991), “Post trauma vulnerability: a proposed model”, in Reese, J., Horn, J. and Dunning, C. (Eds), Critical Incidents in Policing, Government Printing Office, Washington, DC, pp. 365-72. Violanti, J.M. (1996), Police suicide: Epidemic in Blue, Charles C. Thomas, Springfield, IL. Waters, J. and Finn, E. (1995), “Handling client crises effectively on the telephone”, in Roberts, A.R. (Ed.), Crisis intervention and Time-Limited Cognitive Treatment, Sage, Thousand Oaks, CA, pp. 251-89. Corresponding author Judith A. Waters can be contacted at: [email protected]

To purchase reprints of this article please e-mail: [email protected] Or visit our web site for further details: www.emeraldinsight.com/reprints

The current issue and full text archive of this journal is available at www.emeraldinsight.com/1363-951X.htm

Post-traumatic stress symptoms and cortisol patterns among police officers John M. Violanti

Post-traumatic stress symptoms

189

Department of Social and Preventive Medicine, School of Public Health and Health Professions, State University of NY at Buffalo, Buffalo, New York, USA and

Michael Andrew, Cecil M. Burchfiel, Tara A. Hartley, Luenda E. Charles and Diane B. Miller Health Effects Laboratory Division, National Institute for Occupational Safety and Health (NIOSH), Centers for Disease Control and Prevention, Morgantown, West Virginia, USA Abstract Purpose – The purpose of the present study is to examine associations between post-traumatic stress disorder (PTSD) symptoms and salivary cortisol parameters. Design/methodology/approach – PTSD symptoms and cortisol responses were measured in a random sample of 100 police officers. The impact of event scale (IES) categorized into subclinical, mild, moderate and severe levels was employed to measure PTSD symptoms. Cortisol was analyzed from saliva samples over a period of three days and included an awakening response, high protein lunch challenge, whole day (diurnal), and a dexamethasone suppression test (DST). Findings – Officers in moderate and severe PTSD symptom categories had higher mean awakening cortisol values. A significant sample-time by PTSD interaction ( p ¼ 0.008) was found for awakening cortisol responses. Officers in the severe PTSD symptom category showed a blunted response to the cortisol protein meal challenge compared to those in lower PTSD categories. Diurnal cortisol levels suggested an increasing trend across subclinical to severe PTSD categories respectively ( p ¼ 0.15 test for trend). DST ratios were lower in moderate and severe PTSD symptom categories (6.86 and 8.03 respectively) than in the subclinical and mild categories (9.32 and 10.43 respectively). Research limitations/implications – The sample was not representative of all police in the USA. These results suggest that associations between psychological trauma symptoms and dysregulation of cortisol patterns may exist and could possibly affect future health outcomes in police officers. Practical implications – Exposure to trauma and disaster events emphasizes the need to further investigate the health impact of PTSD on police personnel as well as other first responder groups. Originality/value – This article will not only be of interest to those in the police service but to the general public. The present study may serve to provide a guide for larger police population investigations on PTSD and physiological impact. Keywords Post-traumatic stress disorders, Police, Occupational health and safety, United States of America Paper type Research paper

This work was supported by the National Institute for Occupational Safety and Health (NIOSH), contract no. HELD01B0088. The findings and conclusions in this report are those of the authors and do not necessarily represent the views of the National Institute for Occupational Safety and Health.

Policing: An International Journal of Police Strategies & Management Vol. 30 No. 2, 2007 pp. 189-202 q Emerald Group Publishing Limited 1363-951X DOI 10.1108/13639510710753207

PIJPSM 30,2

190

Introduction PTSD and police work Selye (1984) recognized police work as a stressful occupation. Police work has been described as “civilian combat”, and police officers face the almost certain possibility of exposure to traumatic events in their work (Violanti and Paton, 1999). Examples of traumatic occurrences include shootings, physical assault, witnessing violence and familial abuse, handling dead bodies, and disaster scenes such as 9/11 or Hurricane Katrina in New Orleans (Paton and Smith, 1996; Paton and Violanti, 1997; Paton et al. (2004). Multiple exposures to such trauma increases the risk of psychological disorders such as post-traumatic stress disorder (PTSD) and depression (Stephens et al., 1999). Even officers whose PTSD symptoms are in remission may experience a reactivation of symptoms from earlier job related traumas (Wilczak, 2002; Violanti, 2002). The Buffalo Cardio-Metabolic Occupational Police Stress (BCOPS) pilot study was conducted to establish a baseline for the first known population-based study designed to identify biomarkers of stress and psychosocial factors in the high stress occupation of police work (see Violanti et al., 2006 for a description of the full methodology). The present paper describes findings from the BCOPS pilot study concerning the association between reported PTSD symptoms and the salivary cortisol response patterns observed: at awakening; . after a high protein meal challenge; . over the course of a day; and . after taking a low dosage (0.5 mg) of dexamethasone at bedtime-a dexamethasone suppression test (DST). Methods Police sample The Buffalo, New York Police Department, an urban police force of 934 officers at the time of sampling, was the selected police site. A random sample (n ¼ 100) was generated in the department using a computer-generated random number table. No specific inclusion criteria were used for the study other than the participant should be a sworn police officer and willing to participate in the study. Officers with missing data (n ¼ 8) were excluded from analysis. One officer had missing PTSD data, and seven had incomplete cortisol samples. A final sample of 92 officers was included for analysis. All phases, testing, and reports of the study were approved by the State University of New York at Buffalo Internal Review Board and the National Institute for Occupational Safety and Health Human Subjects Review Board. The Center for Preventive Medicine, State University of New York at Buffalo, School of Public Health and Health Professions, Buffalo, NY, served as the data collection site. Measures PTSD symptomatology The impact of event scale (IES) was used to measure psychological symptoms of PTSD (Horowitz et al. (1979)). The IES was categorized based on a previously established formulation (Beaton et al., 1999; Corneil et al., 1999). Categorization was based on the mean IES score of a norm group of diagnosed PTSD patients admitted for treatment (M ¼ 35.3; SD ¼ 17.2) (Horowitz et al. (1979)). A 0.50 standard deviation from the mean

defined trauma stress levels, since that variation limit best distinguished those with mild and those with moderate or severe reported trauma stress symptoms. Consistent with other studies, the following levels of PTSD symptomatology were employed: 0-8 (subclinical); 9-25 (mild); 26-43 (moderate); and . 44 (severe). The prevalence of elevated PTSD symptoms was defined using both the moderate and severe categories (IES score $ 26), consistent with several other studies (Chang et al., 2003; Corneil et al., 1999; Langeveld et al., 2004; Williams et al., 2002). Salivary cortisol Cortisol has long been considered a biomarker for both acute and chronic stress (McEwen, 2004). Although cortisol can feasibly be measured in blood, urine and saliva, its measurement in saliva has come to be preferred because the cortisol present is unbound thus providing the level of biologically active hormone and the small amounts present in saliva can be easily detected and quantified by immunoassay. Officers were provided with Salivettes (Sarstedt, USA), a commercially available collection device consisting of a dental roll and a centrifuge tube, for the collection of saliva samples; at the designated collection time the dental roll is removed from the centrifuge tube and placed in the mouth for approximately two minutes allowing for saturation of the roll. In the lab, the tube is centrifuged to provide a non-viscous saliva sample for assay; centrifuged samples are maintained at 2 20oC until assayed for cortisol by a commercially available chemiluminescence immunoassay (IBL, Hamburg, Germany) at the Technical University of Dresden (Dressendorfer et al., 1992). Cortisol was measured in saliva under the following conditions or challenges for the purposes of the present study: . A high protein meal cortisol challenge was performed at the clinic site, which involved collection of a baseline saliva sample immediately followed by the ingestion of a high protein shake consisting of commercially available protein and carbohydrate powder mixed in water to provide 55 grams of protein to serve as a convenient substitute for a high protein meal; four additional samples were collected at 15 minute intervals beginning 15 minutes after the shake was finished (Rosmond and Bjorntorp, 1998, 2000); . Officers were provided with eight additional Salivettes and a 0.5 mg dexamethasone tablet for off-clinic site testing and collection of saliva over the next two days to determine cortisol levels at awakening, over the course of the day (i.e. diurnal), and following the dexamethasone tablet at bedtime for the DST (Table I). Statistical analysis The cortisol data distributions, between subjects at each time point in the study, were examined and found to be approximately lognormal. Because of this the data were log-transformed to normalize the distribution prior to the application of parametric statistical methods (e.g. ANCOVA). Small numbers of apparent outliers in salivary cortisol data may also exert undue influence on analytical results (Neylan et al., 2005). In the present study a few cortisol measurements fell clearly outside the expected range of the assay. We applied several approaches to deal with extreme outliers including the following:

Post-traumatic stress symptoms

191

PIJPSM 30,2

192

Table I. Timing of salivary cortisol sampling, BCOPS pilot study

Sample no.

Day

Approximate time

Sample characteristics Baseline (prior to protein lunchtime challenge) 15 minutes after lunchtime challenge 30 minutes after lunchtime challenge 45 minutes after lunchtime challenge 60 minutes after lunchtime challenge First Awakening sample 15 minutes after awakening 30 minutes after awakening 45 minutes after awakening Immediately before eating midday meal Immediately before eating evening meal Before bedtime and before taking dexamethasone tablet Post dexamethasone awakening sample

1 2 3 4 5 6 7 8 9 10 11 12

1 1 1 1 1 2 2 2 2 2 2 2

11:10 am 11:20 – 12:30 11:20 – 12:30 11:20 – 12:30 11:20 – 12:30 Awakening Awakening Awakening Awakening Lunchtime Dinnertime Bedtime

13

3

Awakening

. .

.

repeated assays to confirm outlier status; used high performance liquid chromatography (HPLC) to examine samples from participants with repeated outlying cortisol values (e.g. greater than 200 nmol/L) to confirm these high values were not due to authentic cortisol (such examination resulted in the elimination of data from three participants); and consistent with previous studies of this type (Neylan et al., 2005) converted any remaining cortisol value greater than 2.3 standard deviations above the mean to a missing case.

In order to perform this last step we computed the mean plus 2.3 standard deviations for log cortisol for all time points during the diurnal cortisol sampling. We then back-transformed these to the concentration scale and used the largest value (152 nmol/L) as our cut point for setting individual cortisol values to missing. This led to a total of seven cortisol values (among four participants) being set to missing. ANCOVA was used to analyze associations between cortisol levels and PTSD symptom categories. Unadjusted and adjusted means and 95 percent confidence intervals were calculated using log scale data and then back-transformed into concentration scale for convenience of results interpretation. Waking and high protein meal data were analyzed using mixed models to account for the correlation between repeated measurements within individual participants (SAS Proc Mixed (SAS/STAT Users Guide, Version 8, Cary, NC: SAS Institute Inc., 1999). Preliminary analyses determined the optimal covariance structure for inclusion into mixed models to be compound symmetric within PTSD groups, allowing covariance and variance parameters to vary between PTSD groups. A two-way PTSD group by cortisol sample-time interaction model was calculated for both the waking and lunch responses in order to determine whether or not the responses were parallel between PTSD groups. Diurnal AUC was calculated using the trapezoidal rule for numerical integration (Pruessner et al., 2003). Other variables were assessed with simple one-way ANCOVAs for differences in PTSD group means. In addition to one-way tests for equality among means across the PTSD groups we also calculated a one degree of freedom test for linear trend across the four PTSD

groups using a linear contrast. This approach has more power than the statistical test for equality among the group means and tests the null hypothesis of a linear trend across the groups. This test was reported where it was reasonable to assume the possibility of a linear trend in the data. Regression analyses using the IES score as a continuous predictor for the various cortisol parameters (results not shown) led to p-values of similar magnitude to those obtained using the linear contrast approach. The DST was reported as the ratio of the first waking cortisol level, without dexamethasone, to the post dexamethasone first waking value because the ratio corresponds to a difference when data are log transformed, allowing for valid statistical comparisons and straightforward interpretation of suppression as a fold change in cortisol levels after administration of dexamethasone. In other words, the ratio in concentration scale is analogous to a pre-post difference in the log scale. This quantity is a natural result of: . applying the log-transform to normalize the two cortisol variables; . calculating the pre-minus-post dexamethasone difference in the log scale; . performing the ANCOVA to obtain the adjusted mean difference and its confidence limits; and . applying the reverse log transform to the adjusted mean differences and confidence limits. Sample distributions of cortisol levels were summarized using means (nmol/Liter) and 95 percent confidence intervals (CI) across levels of PTSD symptom severity. PTSD and cortisol associations were adjusted for age, gender, smoking and alcohol consumption. Smoking, alcohol intake, and gender were included as dummy coded variables representing the categories listed in Table II. Age was included in the models as a continuous variable. Results Of the 100 randomly selected officers invited to participate, PTSD data were available for 92 officers and of these up to 75 officers had complete information also for cortisol and the covariates included in the analysis. Although differences were not statistically significant, demographic characteristics of the participants (Table III) revealed the following: officers with severe PTSD symptoms were slightly older (46.2 years), had a higher BMI (31.4 kg/m2) and were slightly more likely to report higher alcohol consumption (22.2 percent) compared to those with less severe PTSD symptoms. Table II provides mean values of cortisol parameters across PTSD symptom categories adjusted for age, gender, smoking, and alcohol use. There were little differences between unadjusted and adjusted results. In participants with subclinical and mild categories of PTSD symptoms, cortisol levels increased after the high protein meal (Figure 1) and declined during the four subsequent 15-minute intervals in the expected fashion following an acute challenge. Participants with moderate and severe levels of PTSD symptoms exhibited a somewhat different pattern. Those with moderate PTSD symptoms had increases after the meal, but these cortisol levels remained high during the timed period instead of returning to a baseline level. Officers with severe PTSD symptoms showed a delayed response to the protein meal challenge

Post-traumatic stress symptoms

193

17

20 20 19 21

20 20 20 20

(4.95, 17.58)

9.32

23

24 28 22 29

24 24 24 24 22

0.05 8.91 (6.22, 12.78) 1.40 (0.82, 2.39) (3.79, 5.91) 4.73 (2 4.04, 2 2.52) 2 3.28

(6.06, (6.83, (5.40, (4.82,

7.03) 9.21) 9.13) 6.21) 6.58)

24 23 24 24

0.03

(3.54, (4.60, (4.60, (3.10, (3.31,

14.05) 15.83) 12.53) 11.17)

9.22 10.40 8.23 7.34

4.99 6.51 6.48 4.39 4.67

N

(4.17, 8.03) (5.12, 9.86) (5.46, 10.52) (4.52, 8.70) (3.71, 7.23) 0.06

10.43

(6.10, 17.84)

(4.82, 12.85) (5.13, 13.86) (8.04, 21.42) (6.53, 17.40) 0.22 9.53 (6.85, 13.25) 1.45 (0.92, 2.28) 6.08 (4.94, 7.48) 2 2.81 (2 3.46, 2 2.15)

7.87 8.43 13.40 10.66

5.79 7.10 7.58 6.27 5.18

10

14 17 10 17

14 13 12 14

15 14 15 13 14

(2.91, 8.69) (3.43, 10.42) (4.43, 13.24) (4.07, 12.55) (4.06, 12.32) 0.36

6.86

(3.09, 15.21)

(8.85, 25.46) (9.03, 26.82) (7.62,23.56) (4.81, 13.83) 0.17 12.29 (8.10, 18.67) 2.78 (1.60, 4.85) 6.44 (4.81, 8.64) 2 1.90 (2 2.70, 2 1.09)

15.01 15.56 13.12 8.15

5.03 5.98 7.66 7.14 7.07

7

8 8 6 8

8 8 8 8

9 9 9 9 9

N

8.03

11.50 1.28 6.49 2 3.58

7.75 10.86 10.94 13.16

4.75 4.95 4.63 3.18 6.47

(3.06, 21.07)

(5.60, 10.75) (7.84, 15.04) (7.90, 15.16) (9.50, 18.24) 0.03 (6.46, 20.47) (0.55, 2.98) (4.36, 9.67) (2 4.81, 2 2.35)

(1.86, 12.10) (1.94, 12.62) (1.82, 11.81) (1.25, 8.11) (2.54, 16.49) 0.08

Severe Mean (95% C.I.)

0.82

0.61 0.21 0.25 0.05

0.19 0.42 0.42 0.11

0.92 0.85 0.71 0.23 0.60

p-value *

0.62

0.32 0.81 0.15 0.99

na

na

Ptrenda

obtained from linear contrast test for trend; b-values are in thousands – AUCg: Area Under Curve Ground

Notes: Results adjusted for age, sex, smoking, and drinking beverages containing ethanol. *p-value for the difference between any of the means; a Ptrend: p-value

Awakening (nmol/l) 1st waking 15 min 30 min 45 min p-value * Mean awakening (nmol/l) Bedtime (nmol/l) Diurnal AUCg,(nmol/l)b Diurnal Slope x 1000 Dexamethasone Suppression Test Pre/Post ratio

23 22 23 22 23

Lunch challenge (nmol/l) Baseline 15 min 30 min 45 min 60 min p-value *

Table II. Risk-factor adjusted mean cortisol parameters by PTSD symptom severity N

PTSD symptom severity Mild Moderate Mean (95% C.I.) N Mean (95% C.I.)

194

Salivary cortisol parameters

Sub-clinical Mean (95% C.I.)

PIJPSM 30,2

Variable Age (years) BMI (kg/m2) Education level (%) Less than 12 Years High School/GED College , 4 years College 4 þ years Cigarette smoking (%) Never smoker Former smoker Current smoker Alcohol intake (%) None Less than 1 drink/week 1 – 7 drinks/week 8 or more drinks/week Years of police service Impact of events score Intrusive score Avoidance score Men (%) Ethnicity (%) European-American African-American Hispanic-American

Subclinical (N ¼ 25) 43.5 29.3

(8.1) (4.4)

PTSD symptom severity Mild Moderate (N ¼ 36) (N ¼ 21) 43.0 27.8

0.0 12.0 28.0 60.0

0.0 16.7

(8.0) (4.7)

43.3 28.0

(8.9) (4.9)

Severe (N ¼ 10) 46.2 31.4

25.0

0.0 14.3 33.3 52.4

10.0 10.0 40.0 40.0

30.4 52.2 17.4

46.9 31.2 21.9

50.0 38.9 11.1

44.4 44.4 11.1

13.0 30.4 47.8 8.7 15.4 3.4 1.4 2.0 68.0

31.2 15.6 34.4 18.8 13.8 18.5 8.7 9.8 52.8

22.2 27.8 44.4 5.6 14.2 33.8 15.7 18.1 57.1

11.1 33.3 33.3 22.2 14.1 47.7 24.6 23.1 50.0

(7.2) (12.0)

Post-traumatic stress symptoms p-valuea 0.35 0.36 0.74

195

0.75

0.59

76.0 20.0 4.0

(9.1) (3.1) (2.2) (2.2)

83.3 13.9 3.8

(9.7) (5.1) (4.9) (4.5)

61.9 28.6 9.5

(8.4) (4.7) (4.4) (3.9)

(9.1) (3.2) (4.7) (3.0)

0.75 NA NA NA 0.64 0.57

70.0 20.0 10.0

Notes: PTSD ¼ Post-traumatic stress disorder; BMI ¼ body mass index; BP ¼ blood pressure. Values are means with standard deviations in parentheses or percentages. a For continuous variables the p-values are from tests for linear trend across PTSD severity; for categorical variables the p-values are from Fisher’s exact tests of independence between the rows and columns of the frequency table

Table III. Physiologic, lifestyle and demographic characteristics by PTSD symptom severity

Figure 1. Lunch cortisol response by PTSD severity model estimates, covariate adjusted

PIJPSM 30,2

196

with little increase in cortisol levels after the meal and a sharp increase at the 60-minute level. In general, participants with subclinical and mild PTSD symptoms displayed the expected cortisol pattern at awakening (Figure 2) with low initial values followed by a rise and fall. However, those officers with severe PTSD symptoms showed a pattern in which cortisol levels increased over the collection period and remained elevated. Those with moderate symptoms showed a pattern that was elevated at waking and then decreased; cortisol levels remained higher for a longer period of time before decreasing than did those with less severe PTSD symptoms. The first awakening mean cortisol level did not vary significantly across PTSD categories, although officers in the moderate category tended to have higher values. There was a statistically significant difference between first awakening and the 45-minute cortisol samples in the severe PTSD symptom category ( p ¼ 0.03) indicating a meaningful rise in cortisol levels after awakening without the expected decrease. The group by sample-time interaction effect was close to significance ( p ¼ 0.12) for the lunchtime challenge (Figure 1), testing the hypothesis that cortisol patterns across time are parallel between PTSD groups. Cortisol responses to the lunchtime challenge in severe PTSD categories were noticeably flatter than those in less severe categories, and moderate responses tended to stay elevated compared to the other categories. Similarly, a statistically significant interaction ( p ¼ 0.008) was found for the awakening cortisol response (Figure 2).

PTSD symptoms and the diurnal cortisol response In general, cortisol levels decreased across the day but the diurnal cortisol response (Table II) suggested an increasing trend in cortisol levels across subclinical to severe PTSD symptom categories respectively ( p ¼ 0.15 test for trend). The total amount of daily cortisol secretion as reflected by the diurnal AUCg was highest in the severe PTSD symptom category. Those in the severe PTSD category not only had the highest daily secretion but also a steeper decline in cortisol as they moved toward the end of the day. This was also reflected in mean bedtime cortisol levels, which were lowest in those with severe PTSD symptoms.

Figure 2. Waking cortisol response by PTSD severity model estimates, covariate adjusted

PTSD symptoms and the DST Cortisol suppression ratios, the ratio of the first awakening cortisol level on day two and the post-dexamethasone awakening cortisol on the morning of day three (Table II), appeared to be slightly lower in those reporting moderate or severe PTSD symptoms (6.86 and 8.03 respectively) than those reporting subclinical or mild symptoms (9.32 and 10.43 respectively).

Post-traumatic stress symptoms

197 Discussion The strengths of this study included the availability of standardized cortisol samples adjusted for other risk factors, the use of a standardized protocol, and high response rates and cooperation. Cortisol assessment over a three-day period provided a reasonable sampling period, with a 92 percent compliance for complete saliva samples and the IES. The present study may serve to provide a guide for larger police population investigations on PTSD and physiological impact. Our results suggest hypothalamus-pituitary-adrenal (HPA) dysregulation based on associations and interactions among reported PTSD symptoms and cortisol patterns found across cortisol parameters. Following the high protein lunchtime challenge, levels in officers with more PTSD symptoms did not respond in an expected pattern, that is, cortisol levels did not rise due to the high protein challenge and then return to a base level. This result differed from that of a previous study which found that cortisol levels in healthy individuals generally rose two to four times higher than baseline subsequent to a high protein meal challenge (Rosmond and Bjorntorp, 2000). In moderate PTSD categories, cortisol levels rose among officers and remained high during the 45-minute time span instead of returning to a basal level. Similar patterns were noted in the cortisol awakening measure. As such, our data suggest an inappropriate response of the HPA axis to several standard challenges for the moderate and severe PTSD group. Cortisol patterns found in this study suggest “allostatic load”, where failure to initiate or shut off mediators under conditions of chronic or acute stress leads to HPA dysregulation (McEwen and Seeman, 1999; McEwen, 1998, 2000, 2004). Other difficulties associated with traumatic stress exposure include HPA overreactivity, exaggerated startle response, sleep disruption, and nightmares. Events in police work may bring about a hypervigilance state, where officers experience an alteration in physiology, despite the absence of threat, and maintain a constant state of arousal (Yehuda, 2002, 2004; Yehuda et al., 2004). Chronically traumatized people are hypervigilant, anxious, aggressive, and have no determinable baseline of psychological calm even after exposure to threatening situations has long ended (Yehuda, 2004). The residual impact of trauma may extend beyond a police career and include symptoms of acute generalized anxiety, worry, and depression. These symptoms contribute to a level of morbidity even after police officer retirement (Violanti and Paton, 1999). Our data would fit an allostatic load model in that those exhibiting higher levels of PTSD are under higher allostatic load and therefore their HPA axis is not operating in a healthy manner. Our diurnal cortisol data also provides support for the idea that the HPA axis is not functioning efficiently in the officers displaying increased PTSD symptoms. Covariate-adjusted diurnal (whole day) cortisol levels indicated a nearly significant ( p ¼ 0.15) increasing trend as PTSD increased in severity, suggesting that officers

PIJPSM 30,2

198

with increased PTSD symptoms may be experiencing increased activation of the HPA axis throughout the entire day. Whole day cortisol secretion levels in officers with severe PTSD symptoms did not appear to return to basal levels as readily as those who experienced lower PTSD symptom levels. Other investigations have also noted aberrant functioning of the HPA axis in those suffering from PTSD or displaying PTSD symptoms. Yehuda and McFarlane (1997) found that physiologic responses fail to return to the baseline, pre-trauma state in PTSD patients. In patients with long-term PTSD, findings suggest that the stress response is chronic and that the biological response is different from responses in normal subjects and patients with major depressive disorders (Yehuda, 2002; Otte et al., 2005). Previous studies have reported results contrary to the present findings. Yehuda (2004) suggested that 24-hour cortisol levels were actually lower in PTSD patients. Our findings suggested that cortisol is higher, not lower, through the diurnal cycle in those who report high PTSD symptoms. Others have also suggested that greater suppression of cortisol in PTSD patients as measured by DST occurs as a result of enhanced negative feedback processes (McEwen, 2004; Yehuda, 2001, 2004). We found that DST suppression ratios were less in those with higher PTSD symptoms. There are several issues that may account for differences among cortisol-PTSD studies. The first is measurement. Our cortisol sample measure was obtained from saliva instead of urine or blood samples, allowing for an efficient immunoassay of unbound cortisol. The DST was expressed somewhat differently than other studies, calculating the ratio of the first waking cortisol level to the post dexamethasone first waking value. We viewed a ratio approach as having well known statistical properties that can be more clearly interpreted than other analytical techniques. We measured cortisol responses across levels of PTSD, a somewhat different approach than measuring PTSD versus non PTSD samples. This approach may be useful in helping to determine thresholds of physiological response. Although our measure of PTSD involved reported symptoms and not actual diagnoses, the IES is based on norms from diagnosed PTSD cases. It is acknowledged that IES captures intrusive and avoidance symptomatology, which renders usefulness as a measure for post-traumatic stress (Weiss and Marmar, 1997). Both the intrusion and avoidance scales of the IES have displayed acceptable reliability (Chronbach alpha of 0.79 and 0.82, respectively), and a split-half reliability for the whole scale of 0.86 (56). The IES has also displayed the ability to discriminate between varieties of traumatized groups from non-traumatized groups (Weiss et al., 1997). A second factor in study differences involves expected normal cortisol patterns. Stone et al. (2001) for example, found that a proportion of profiles in a normal healthy population are flat (51 percent had typical cycles, 17 percent flat cycles and 34 percent inconsistent cycles). Individual differences are large, and day-to-day stability is only modest. Curves vary with time, place and mode of awakening. The shape of a cortisol curve may be indicative of a long-term response to chronic stress or a variant within the normal range (Stone et al., 2001). The use of different populations is a third factor. Many studies on PTSD and cortisol rely on other populations who have experienced specific instances of trauma such as rape or auto accidents (Resnick et al., 1993; Delahanty et al., 2000) while police exposure to trauma may be multifaceted. The police have not yet been fully studied in terms of biological responses to trauma and little is known of the biological effects of

either direct or the threat of direct exposure to trauma among police officers. Of note is recent work by Ramon et al. (2004) on the Dutch police who found a reduced volume of the hippocampus in officers with PTSD. Reduced hippocampal volumes have also been found with other PTSD populations as well as with many disorders and conditions including major depression and aging (Miller and O’Callaghan, 2005). There may be additional modifying factors in this study which impacted PTSD and cortisol associations. Only a small number of officers (31 percent) in our sample reported moderate or severe PTSD symptoms while the remainder experienced subclinical or mild symptoms. However, officers who did report moderate or severe PTSD symptoms had mean IES scores of 33.8 (4.7 SD) and 47.7 (3.2 SD) respectively, considerably higher than other groups reported in the literature (Beaton et al., 1999). Although, we do not have history of trauma exposure in our sample, prior trauma and intensity of the response to a traumatic event may increase levels of PTSD (Schnurr et al., 2002). Stress reactions other than PTSD may mediate the relationship between trauma exposure and biological anomalies. Examples are comorbidity with other disorders, life events, chronic daily work stresses, or a direct result of experiences such as military combat (Friedman and Schnurr, 1995). Research on PTSD and major depression has indicated that PTSD comorbid with major depression is much more similar to PTSD alone and distinctly different from major depression alone (Schnurr et al., 2002). Other research has suggested that PTSD may play an indirect rather than a direct role in HPA activation (Schnurr et al., 2002). Exposure to traumatic work events and major disaster events emphasizes the need to further investigate the impact of PTSD on police personnel as well as other first responder groups. Recent large scale events such as 9/11 and Hurricane Katrina present police officers with situations where causation, scale, distribution and complexity contribute to levels of trauma risk by presenting demands that exceed the level of resources generally available to deal with them (Gidron, 2002; Galea et al., 2002). Terrorism, for example, has increased trauma risk in more fundamental ways. For example, officers must accommodate the legacy (e.g. increased levels of fear in the community, changes in security precautions, perceiving the world as increasingly threatening) of terrorism into their thinking. In short, the range of circumstances that today affect police officers’ vulnerability to PTSD has increased dramatically and should be the subject of additional scientific investigation (Flin and Arbuthnot, 2002). References Beaton, R., Murphy, S., Johnson, C., Pike, K. and Corniel, W. (1999), “Coping responses and post-traumatic stress symptomatology in urban fire service personnel”, Journal of Traumatic Stress, Vol. 12, pp. 293-308. Chang, C., Lee, L., Connor, K.M., Davidson, J.R.T., Jeffries, K. and Lai, T. (2003), “Post-traumatic distress and coping strategies among rescue workers after an earthquake”, The Journal of Nervous and Mental Disease, Vol. 191, pp. 391-8. Corneil, D.W., Beaton, R., Murphy, S., Johnson, C. and Pike, K. (1999), “Exposure to traumatic incidents and prevalence of post-traumatic stress disorder symptomatology in urban firefighter in two countries”, Journal of Occupational Health Psychology, Vol. 4, pp. 131-41. Delahanty, D.L., Raimonde, A.I. and Spoonsier, E. (2000), “Initial post-traumatic urinary cortisol levels predict subsequent PTSD symptoms in motor vehicle accident victims”, Biological Psychiatry, Vol. 48, pp. 940-7.

Post-traumatic stress symptoms

199

PIJPSM 30,2

200

Dressendorfer, R.A., Kirschbaum, C., Rohde, W., Stahl, F. and Strasburger, C.J. (1992), “Synthesis of a cortisol-biotin conjugate and evaluation as a tracer in an immunoassay for salivary cortisol measurement”, Journal of Steroid Biochemial And Molecular Biology, Vol. 43, pp. 683-92. Friedman, M.J. and Schnurr, P.P. (1995), “The relationship between trauma, PTSD, and physical health”, in Friedman, M.J., Charney, D.S. and Deutch, A.Y. (Eds), Neurobiological and Clinical Consequences of Stress: From Normal Adaptation to PTSD, Lippincott-Raven, Philadelphia, PA, pp. 507-24. Flin, R. and Arbuthnot, K. (Eds.) (2002), Incident Command: Tales from the Hot Seat, Ashgate, Aldershot. Galea, S., Ahern, J. and Resnick, H. (2002), “Psychological sequelae of the September 11 terrorist attacks in New York City”, New England Journal of Medicine, Vol. 346, pp. 982-7. Gidron, Y. (2002), “Post-traumatic stress disorder after terrorist attacks: a review”, Journal of Nervous and Mental Disorder, Vol. 190, pp. 118-21. Horowitz, M.J., Wilner, N. and Alvarez, W. (1979), “Impact of event scale: a measure of subjective stress”, Psychosomatic Medicine, Vol. 41, pp. 209-18. Langeveld, N.E., Grootenhuis, M.A., Voute, P.A. and de Haan, R.J. (2004), “Post-traumatic stress symptoms in adult survivors of childhood cancer”, Pediatric Blood and Cancer, Vol. 42, pp. 604-10. McEwen, B.S. (1998), “Protective and damaging effects of stress mediators”, New England Journal of Medicine, Vol. 338, pp. 171-9. McEwen, B.S. (2000), “Allostasis and allostatic load: implications for neuropsychopharmacology”, Neuropsychopharmacology, Vol. 22, pp. 108-24. McEwen, B.S. (2004), “Allostasis and allostatic overload and relevance to the pathophysiology of psychiatric disorder”, Annals of NY Academy of Science, Vol. 1032, pp. 1-7. McEwen, B.S. and Seeman, T. (1999), “Protective and damaging effects of mediators of stress: elaborating and testing the concepts of allostasis and allostatic load”, Annals of NY Academy of Science, Vol. 896, pp. 30-47. Miller, D.B. and O’Callaghan, J.P. (2005), “Aging, stress and the hippocampus”, Ageing Research Reviews, Vol. 4, pp. 123-40. Neylan, T.C., Brunet, A., Pole, N., Best, S.R., Metzler, T.J., Yehuda, R. and Marmar, C.R. (2005), “PTSD symptoms predict waking salivary cortisol levels in police officers”, Psychoneuroendocrinology, Vol. 30, pp. 373-81. Otte, C., Lenoci, M., Metzler, T., Yehuda, R., Marmar, C. and Neylan, T. (2005), “Hypothalamic-Pituitary-Adrenal axis and sleep in post-traumatic stress disorder”, Neuropsychopharmacology, Vol. 30, pp. 1173-80. Paton, D. and Smith, L.M. (1996), “Psychological trauma in critical occupations: methodological and assessment strategies”, in Paton, D. and Violanti, J.M. (Eds), Traumatic Stress in Critical Occupations: Recognition, Consequences, and Treatment, Charles C. Thomas, Springfield. IL, pp. 125-39. Paton, D. and Violanti, J.M. (1997), “Long term exposure to traumatic demands in police officers: behavioral addiction and its management”, in Haberman, G. (Ed.), Looking Back, Moving Forward: Fifty years of New Zealand Psychology, Psychological Society, Wellington, NZ. Paton, D., Violanti, J.M., Dunning, C. and Smith, L. (2004), Managing Traumatic Stress Risk: A Proactive Approach, Charles C. Thomas, Springfield, IL, pp. 5-25. Pruessner, J.C., Kirschbaum, C., Meinlschmid, G. and Hellhammer, D.H. (2003), “Two formulas for computation of the area under the curve represent measures of total hormone

concentration versus time-dependent change”, Psychoneuroendocrinology, Vol. 28, pp. 916-31. Ramon, J.L., Lindauer, E.-J. V., Jalink, M., Olff, M., Carlier, I., Majoie, C.B.L.M., Den Heeten, G.J. and Gersons, B.P.R. (2004), “Smaller hippocampal volume in Dutch police officers with post-traumatic stress disorder”, Biological Psychiatry, Vol. 56, pp. 356-63. Resnick, H.S., Kilpatrick, D.F. and Dansky, B.S. (1993), “Prevalence of civilian trauma and post-traumatic stress disorder in a representational national sample of women”, Consultant Clinical Psychology, Vol. 6l, pp. 984-91. Rosmond, R. and Bjorntorp, P. (1998), “Endocrine and metabolic aberrations in men with abdominal obesity in relation to anxio-depressive infirmity”, Metabolism: Clinical and Experimental, Vol. 47, pp. 1187-93. Rosmond, R. and Bjorntorp, P. (2000), “The hypothalamic-pituitary-adrenal axis activity as a predictor of cardiovascular disease, type 2 diabetes and stroke”, Journal of Internal Medicine, Vol. 247, pp. 88-197. Schnurr, P.P., Friedman, M.J. and Bernardy, N.C. (2002), “Research on post-traumatic stress disorder: epidemiology, pathophysiology, and assessment”, Psychiatric Practice, Vol. 58, pp. 877-89. Selye, H. (1984), “Police stress”, Police Stress Magazine, pp. 2-5. Stephens, C., Long, N. and Flett, R. (1999), “Vulnerability to psychological disorder: previous trauma in police recruits”, in Violanti, J.M. and Paton, D. (Eds), Police Trauma: Psychological Aftermath of Civilian Combat, Charles C. Thomas, Springfield, IL, pp. 65-74. Stone, A.A., Schwartz, J.E., Smyth, J., Kirschbaum, C., Cohen, S., Hellhammer, D. and Grossman, S. (2001), “Individual differences in the diurnal cycle of salivary free cortisol: a replication of flattened cycles for some individuals”, Psychoneuroendocrinology, Vol. 26, pp. 295-303. Violanti, J.M. (2002), “Trauma stress in police work”, in Paton, D. and Violanti, J.M. (Eds), Traumatic Stress in Critical Occupations: Recognition, Consequences and Treatment, Charles C.Thomas, Springfield, IL, pp. 137-43. Violanti, J.M. and Paton, D. (1999), Police Trauma: Psychological Aftermath of Civilian Combat, Charles C. Thomas, Springfield, IL, pp. 1-5. Violanti, J.M., Burchfiel, C.M., Miller, D.B., Andrew, M.E., Dorn, J., Wactawski-Wende, J., Beighley, C.M., Pierino, K., Joseph, P.N., Vena, J.E., Sharp, D.S. and Trevisan, M. (2006), “The Buffalo Cardio-Metabolic Occupational Police Stress (BCOPS) pilot study: methods and participant characteristics”, Annals of Epidemiology, Vol. 16, pp. 48-156. Weiss, D. (1997), “The impact of event scale – revised”, in Wilson, J. and Keane, T. (Eds), Assessing Psychological Trauma and PTSD, Guilford, New York, NY. Wilczak, C. (2002), Three Perspectives on Trauma from New York City Police Officers, Div.17 Newsletter 2002,Vol. 3, Division of Counseling Psychology of the American Psychological Association, Washington, DC, pp. 13-15. Williams, W.H., Evans, J.J., Wilson, B.A. and Needham, P. (2002), “Prevalence of post-traumatic stress disorder symptoms after severe traumatic brain injury in a representative community sample”, Brain Injury, Vol. 16, pp. 673-9. Yehuda, R. (2001), “Biology of post-traumatic stress disorder”, Journal of Clinical Psychology, Vol. 62, pp. 41-6. Yehuda, R. (2002), “Post traumatic stress disorder”, New England Journal of Medicine, Vol. 346, pp. 108-14. Yehuda, R. (2004), “Risk and resilience in post-traumatic stress disorder”, Journal of Clinical Psychiatry, Vol. 65, Suppl 1, pp. 29-36.

Post-traumatic stress symptoms

201

PIJPSM 30,2

202

Yehuda, R. and McFarlane, A.C. (1997), Psychobiology of Post-Traumatic Stress Disorder,Vol. 821, New York Academy of Sciences, New York, NY, pp. 1-50. Yehuda, R., Golier, J.A., Halligan, S.L., Meaney and M., Bierer (2004), “The ACTH response to dexamethasone in PTSD”, American Journal of Psychology, Vol. 61, pp. 1397-403. Further reading Blanchard, E.B., Kolb, L.C. and Prins, A. (1998), “Psycho-physiological responses in the diagnosis of post-traumatic stress disorder in Vietnam veterans”, Journal of Nervous and Mental Disorder, Vol. 179, pp. 99-103. Blanchard, E.B., Hickling, E.J., Taylor, A.E., Loos, W.R. and Gerardi, R.J. (1994), “The psychophysiology of motor vehicle accident related post-traumatic stress disorder”, Behavior Therapy, Vol. 25, pp. 453-67. Bremner, D. and Vermetten, E. (2004), “Neuroanatomical changes associated with pharmacotherapy in post-traumatic stress disorder”, in Yehuda, R. and McEwen, B. (Eds), Biobehavioral Stress Response, New York Academy of Sciences, New York, NY, pp. 154-7. Corneil, D.W. (1995), “Traumatic stress and organizational strain in the fire service”, in Murphy, L.R., Hurrell, J.J. Jr, Sauter, S.L. and Keita, G.P. (Eds), Job Stress Interventions, American Psychological Association, Washington, DC. Friedman, M.J. and McEwen, B.S. (2004), “Post-traumatic stress disorder, allostatic load, and medical illness”, in Schnurr, P.P. and Green, B.L. (Eds), Trauma and Health: Physical Health Consequences of Exposure to Extreme Stress, American Psychological Association, Washington, DC, pp. 157-88. McEwen, B.S. and Stellar, E. (1993), “Stress and the individual: mechanism leading to disease”, Archives of Internal Medicine, Vol. 153, pp. 2093-101. McEwen, B.S. and Wingfield, J.C. (2003), “The concept of allostasis in biology and biomedicine”, Hormone Behavior, Vol. 43, pp. 2-15. Sundin, E.C. and Horowtiz, M.J. (2002), “Impact of event scale: psychometric properties”, British Journal of Psychiatry, Vol. 180, pp. 205-9. Sundin, E.C. and Horowitz, M.J. (2003), “Horowitz’s impact of events scale: evaluation of 20 years of use”, Psychosomatic Medicine, Vol. 65, pp. 870-6. van der Kolk, B.A. (1996), “Stress versus traumatic stress: from acute homeostatic reactions to chronic psychopathology”, in van der Kolk, B.A., McFarlane, A.C. and Weisaeth, L. (Eds), Traumatic Stress: The Effects of Overwhelming Experience on Mind, Body and Society, Guilford Press, New York, NY, pp. 77-101. Yehuda, R. (2002), “Clinical evidence of biologic findings in PTSD”, Psychiatric Quarterly, Vol. 73, pp. 123-33. Corresponding author John M. Violanti can be contacted at: [email protected]

To purchase reprints of this article please e-mail: [email protected] Or visit our web site for further details: www.emeraldinsight.com/reprints

The current issue and full text archive of this journal is available at www.emeraldinsight.com/1363-951X.htm

Obesity and sleep: the Buffalo Police health study Luenda E. Charles, Cecil M. Burchfiel, Desta Fekedulegn and Michael E. Andrew

Obesity and sleep

203

Biostatistics and Epidemiology Branch, Health Effects Laboratory Division, National Institute for Occupational Safety and Health, Centers for Disease Control and Prevention, Morgantown, West Virginia, USA

John M. Violanti School of Public Health and Health Professions, Department of Social and Preventive Medicine, State University of New York at Buffalo (JMV), Buffalo, New York, USA, and

Bryan Vila Department of Political Science, Washington State University, Spokane, Washington, USA Abstract Purpose – Thus study aims to look at the prevalence of obesity and its association with sleep problems among police officers. Design/methodology/value – The authors conducted a cross-sectional study of the relationship between obesity and sleep disorders among 110 randomly selected police officers from the Buffalo, New York, Police Department in 1999. Participants, who ranged in age from 26 to 61 years (mean ^ SD ¼ 39.5 ^ 7.5), responded to sleep related questions and had anthropometric measurements taken. Findings – Results show that several measures of obesity were significantly associated with sleep-disordered breathing in police officers, but not with other sleep problems. Originality/value – A major strength of the study was that it was conducted in a cooperative and motivated study population. It was possible to assess a wide range of anthropometric measurements, including many that are important but are rarely used to measure obesity in epidemiologic studies such as abdominal height, neck circumference, and neck-to-height ratio. In addition, the assessment of the anthropometric indices was performed by trained clinic staff using standardized procedures. Keywords Obesity, Police, Personal health, United States of America Paper type Research paper

Introduction Obesity and sleep problems are becoming major public health problems in industrialized countries (Stein and Colditz, 2004; National Sleep Foundation, 2005). Several studies have shown that there is a positive association between obesity and The authors would like to thank Dr Chunlin Dong for her assistance with dataset preparation and Dr Girija Syamlal for her insightful comments on this manuscript. This work was supported by the National Institute for Occupational Safety and Health (NIOSH), contract No. HELD01B0088. The findings and conclusions in this article are those of the authors and do not necessarily represent the views of the National Institute for Occupational Safety and Health.

Policing: An International Journal of Police Strategies & Management Vol. 30 No. 2, 2007 pp. 203-214 q Emerald Group Publishing Limited 1363-951X DOI 10.1108/13639510710753216

PIJPSM 30,2

204

various sleep problems (Sharma et al., 2004; Dancey et al., 2003; Young et al., 2005). Spiegel et al. (1999) reported that chronic short sleep duration and/or poor sleep quality may increase the risk of obesity, and that short sleep duration may increase the risk of Type 2 diabetes. In a prospective cohort study of young adults, sleep duration was a strong and significant predictor of weight gain (Hasler et al., 2004). There was a monotonic trend toward lower weight gain among those with longer sleep duration and there were significant cross-sectional associations between short sleep duration and obesity in persons younger than 35 years. Gangwisch et al. (2005) conducted a study using the National Health and Nutrition Examination Survey I (NHANES I) Follow-up studies to assess the relationship between sleep duration, obesity, and weight gain. Their results showed that sleep duration was inversely related to obesity among the younger age group. Among persons 32 to 49 years, those who got two to four hours of sleep per night at baseline gained the most weight over the follow-up period, while those who got ten or more hours of sleep gained the least weight. Results from the National Sleep Foundation survey conducted in 2005 on American adults also found that duration of sleep was inversely related to obesity (National Sleep Foundation, 2005). The occurrence of certain sleep problems also may increase the risk of morbidity and mortality. Decreased sleep time has been associated with increased risk of hypertension (Gangwisch et al., 2006) and reduced life expectancy (Yousaf and Sedgwick, 1996), and daytime sleepiness has been associated with impaired cognition (Foley et al., 2001). A cohort study by Yaggi et al. (2005) found that obstructive sleep apnea significantly increased the risk of stroke or death. The seriousness of sleep apnea is underscored by the fact that only a small proportion of persons afflicted with the condition is diagnosed clinically (Young et al., 1997). Fatigue due to moderate sleep loss affects performance similar to moderate alcohol intoxication (Dawson and Reid, 1997; Williamson and Feyer, 2000). A vicious cycle can be created whereby poor sleep quality increases the risk for a variety of chronic disorders and, in turn, these disorders further decrease the quality of sleep. Police officers experience poorer sleep quality and fewer hours of sleep than the general public (Neylan et al., 2002; Vila, 2006). Although we do not know the extent to which obesity affects police officers, anecdotal evidence suggests that they are at least as likely to be obese as the general public (Morioka and Brown, 1970). The main objective of this study is to investigate cross-sectional associations between anthropometric indices of obesity and sleep problems for persons in this occupational group. Methods Study population and design This study utilized a cross-sectional design. The Center for Preventive Medicine, State University of New York (SUNY) at Buffalo, School of Public Health and Health Professions, Buffalo, NY, served as the data collection site (Violanti et al., 2006). Informed consent was obtained from all participants and this study was approved by the State University of New York at Buffalo’s Internal Review Board and the National Institute for Occupational Safety and Health Human Subjects Review Board. In 1999, a random sample of 115 police officers was selected from 934 officers employed by the Buffalo Police Department. The participation rate was 100 percent – five officers who

did not have information on several of the anthropometric variables were excluded from the study resulting in a sample size of 110, 44 females and 66 males. Assessment of sleep The outcomes of interest were sleep quality and sleep quantity. The self-reported questionnaire on sleep elicited information on behaviors or characteristics while sleeping and on daytime sleepiness. Table I provides a complete list of questions used. No questions were asked about insomnia. Several of these questions are validated in instruments such as the Pittsburgh Sleep Quality Index questionnaire (Buysse et al., 1989) and the sleep apnea survey used by Maislin et al. (1995). The sleep questions were evaluated on a six-point scale that included “not sure/don’t know/not applicable”, “strongly disagree/never”, disagree/, 1 per week, somewhat agree/1-2 times per week, agree/3-4 times per week, and “strongly agree/5-7 times per week”. All sleep problems were analyzed as yes vs.. never, or as “3-7 times per week,” “ , 1-2 times per week,” and “never.” Participants were also asked how many hours they slept each night during the previous five weekdays (i.e. Sunday through Thursday) and during the previous weekends (i.e. Friday and Saturday nights). We combined the hours of sleep reported daily for each period to give total hours of sleep per 24-hour period. Sleep duration was categorized as 0-4.5, 5-6.5, and . ¼ seven hours. Assessment of anthropometric measurements for obesity Body mass index (BMI) was calculated as weight in kilograms divided by height in meters squared. Waist circumference, hip circumference, and neck circumference were measured and rounded to the nearest 0.5 centimeter. Abdominal height (in centimeters) was measured three times and the average value was used in analyses. Neck-to-height ratio, waist-to-hip ratio, and waist-to-height ratio were also calculated. Taller people are expected to have larger neck and waist circumferences, so the inclusion of neck-to-height, waist-to-hip, and waist-to-height ratios in this study was necessary to account for body height and hip size. This standardization also served to minimize the impact of normal phenotypic differences between men and women. BMI is not the ideal measure of obesity since larger muscle mass may also increase this index. Therefore, several other indices (mentioned above) were used to provide a more accurate assessment of obesity. Assessment of covariates Participants were given a self-administered questionnaire to provide information on demographic characteristics, lifestyle behaviors, medical history, and sleep quality and quantity. Police officers reported their highest level of education and their marital status. Participants indicated the number of years employed as a police officer and their present rank in the police force (e.g. police officer, lieutenant, detective, etc.). Participants were asked how often they consumed alcoholic beverages: 12 oz. can or bottle of beer, one medium glass of wine, and one shot of liquor. The total number of drinks per month (of each type) was summed and then divided by four to give the approximate total number of drinks consumed per week. Participants reported their smoking status as current, former, or never. The Center for Epidemiological Studies Depression scale (CES-D) was used to measure psychological distress and depression

Obesity and sleep

205

Table I. Prevalence of sleep problems among police officers by sex 72

100

13 52 45 6.1

28 89

82 99

110

57 74 8 16

All Yes

74 89 79 94

Note: *p-value for difference between women and men ,0.05. Source: Charles et al. (1999)

At night, my sleep disturbs my bed partner’s sleep I am told I snore in my sleep I am told I stop breathing in my sleep I suddenly wake up gasping for breath during the night I have or have been told that I have restless legs I feel tired upon awakening and want to go back to sleep I am very sleepy during the daytime and struggle to stay awake Hours of sleep per 24-hour period during the previous 7 days 0-4.5 5-6.5 $7 Mean and SD

Total

11.8 47.3 40.9 1.7

72.0

34.2 89.9

77.0 83.2 10.1 17.0

%

44

38

32 36

21 29 28 36

Total

7 19 18 6.0

33

11 33

15 22 2 7

Women Yes

15.9 43.2 40.9 1.9

86.8

34.4 91.7

71.4 75.9 7.1 19.4

%

66

62

50 63

53 60 51 58

Total

206

Sleep problems

6 33 27 6.2

39

17 56

42 52 6 9

Men Yes

9.1 50.0 40.9 1.6

62.9 *

34.0 88.9

79.3 86.7 11.8 15.5

%

PIJPSM 30,2

in the participants (Radloff, 1977). The CES-D, a 20-item scale with responses ranging from “rarely” to “most of the time”, has been widely used and has good reliability for measuring symptoms of depression and stress. Information was obtained on physical activity during the previous seven days. Participants also reported the degree (hours per week, hours per weekend) and intensity (moderate, hard, very hard) of three types of physical activity (occupational, household and sports). The data were then used to create a total physical activity score that was computed by summing the intensities of the three types (i.e. occupational, household and sports) of physical activity performed during the weekday and the weekend. Intensity score was computed as the product of number of hours and intensity of physical activity where moderate, hard and very hard were assigned intensities of 1, 2 and 3 respectively. Statistical methods Analytical methods included univariate measures, t-tests, Pearson’s correlation, analyses of variance and covariance. The prevalence of each sleep problem was determined for the population. The association between duration of sleep and obesity was assessed using Pearson’s correlation and analysis of variance. Analyses of variance and covariance were used to obtain the mean values of covariates by sleep quality problems, categorized as never (0 times a week), low (, 1 – 2 times a week), and high (3-7 times a week). The p-values for linear trend across these categories were based on orthogonal contrast coefficients. The following variables were evaluated for their potential role as effect modifiers and/or confounders: age, physical activity, smoking status, alcohol intake, sex, depression, and shift work status. All analyses were conducted using the SAS system, version 9.1 (SAS Institute, 2002). Results Demographic and anthropometric characteristics of the study population are presented in Table II. The participants ranged in age from 26 to 61 years and women and men had similar average ages (39 years). As expected, mean levels of all anthropometric measures were significantly larger in men compared to women. The mean years of police service reported by men was 14.8 years compared to 10.2 years for women. Highest level of education attained was similar between the sexes but a higher proportion of women (70.5 percent) held the rank of police officer than men (59.1 percent). Overall, the prevalence of reported current smoking was low (19.3 percent), but it was somewhat higher, though not statistically significant, among women (25.0 percent) than among men (15.4 percent). The majority (62.7 percent) of the officers were married. Table I provides the prevalence of sleep problems by sex. The two sleep problems with the highest prevalence were tiredness upon awakening (89.9 percent) and snoring (83.2 percent). “Stop breathing during sleep” had the lowest reported prevalence (10.1 percent). The percentage of women who reported daytime sleepiness (86.8 percent) was significantly higher than men who reported this problem (62.9 percent). Similar proportions of women and men reported that they slept $ seven hours per 24-hour period during the previous seven days. Mean values of the anthropometric measures for three sleep-disordered breathing problems are presented in Table III. The other sleep problems were not associated with

Obesity and sleep

207

PIJPSM 30,2

208

Table II. Characteristicsa of the study population

All Range

N

Women Mean (SD)

N

Men Mean

26.0 – 61.0 1.0 – 36.0 117.0 – 273.4 157.0 – 196.5 19.5 – 39.8 64.5 – 126.0 0.67 – 0.99 0.36 – 0.69 83.0 – 133.3 14.3 – 32.5 31.3 – 48.5 0.19 – 0.30 0.0 – 17.3 0.0 – 38.0 0.0 – 120.0

44 44 44 44 44 44 44 44 44 44 40 40 44 42 44

39.1 10.2 160.9 166.7 26.3 80.5 0.77 0.48 104.2 19.0 35.8 0.21 2.2 8.2 11.3

(6.3) (6.6) (28.8) (5.9) (4.5) (10.1) (0.06) (0.06) (11.2) (2.9) (3.6) (0.02) (3.3) (8.2) (13.6)

66 66 66 66 66 66 66 66 66 65 65 65 66 62 66

39.8 14.8 205.6 179.0 29.1 97.1 0.90 0.54 108.5 21.9 41.5 0.23 3.3 6.9 11.2

Education (%): , ¼ High school/GED ,4 yrs college . ¼ 4 yrs college

19.1 54.6 26.4

7 24 13

15.9 54.6 29.6

14 36 16

21.2 54.6 24.2

Rank (%): Police officer Sergeant/lieutenant/captain Detective Other

63.6 16.4 16.4 3.6

31 6 4 3

70.5 13.6 9.1 6.8

39 12 14 1

59.1 18.2 21.2 1.5

Smoking status (%): Current Former Never

19.3 32.1 48.6

11 18 15

25.0 40.9 34.1

10 17 38

15.4 26.2 58.5

Marital status (%): Single Married Divorced

23.6 62.7 13.6

13 21 10

29.6 47.7 22.7

13 48 5

19.7 * 72.7 * 7.6 *

Shift work (%): Day/afternoon shift Night shift

82.1 17.9

36 4

90.0 10.0

51 15

77.3 22.7

Covariates Age (years) Years of service Weight (lbs) Height (inches) BMI (Kg/m2) Waist circumference (cm) Waist-to-hip circumference ratio Waist-to-height circumference ratio Hip circumference (to nearest 0.5 cm) Abdominal height (cm) Neck circumference (cm) Neck- to-height ratio Alcohol – number of drinks/wk Depression score (CES-D) Physical activity score

(SD) (8.2) (9.6) * (31.0) * (6.9) * (3.9) * (10.3) * (0.06) * (0.06) * (8.0) * (2.8) * (2.9) * (0.02) * (4.0) (5.7) (18.4)

Notes: *p-value for difference between women and men ,0.05. Source: Charles et al. (1999)

any of the anthropometric variables and so were not included in this table. Except for waist-to-hip ratio, all anthropometric variables were significantly associated with snoring before and after adjustment for age, physical activity, smoking, alcohol intake, sex, depression, and shift-work. In addition, positive stepwise trends (although non-significant) were observed between five anthropometric measures and sleep apnea after full risk factor adjustment. BMI, hip circumference, abdominal height, neck

84.9 90.2 99.0 19.0 20.6 22.8 0.82 0.85 0.88

15 50 24 15 50 23 15 50 24

13 48 24

13 48 24 0.22 0.22 0.24

38.0 38.9 42.0

0.49 0.52 0.56

102.8 106.2 112.3

15 50 24

15 50 24

25.9 27.5 30.9

(1.0) 25.4 (0.6) 26.8 (0.8) 29.6 ,0.001 (2.2) 100.5 (1.2) 105.6 (1.8) 109.9 0.001 (3.0) 85.2 (1.7) 89.2 (2.4) 95.6 ,0.001 (0.7) 18.7 (0.4) 20.1 (0.6) 21.7 ,0.001 (0.02) 0.85 (0.01) 0.85 (0.02) 0.87 0.038 (0.01) 0.50 (0.01) 0.52 (0.01) 0.55 ,0.001 (1.1) 36.8 (0.6) 37.8 (0.8) 40.3 0.004 (0.006) 0.21 (0.003) 0.22 (0.004) 0.23 0.004

(1.2) (0.7) (1.0) 0.002 (2.6) (1.6) (2.2) 0.003 (2.8) (1.7) (2.3) 0.002 (0.8) (0.5) (0.7) 0.002 (0.02) (0.01) (0.01) 0.225 (0.02) (0.01) (0.01) 0.006 (1.0) (0.6) (0.8) 0.002 (0.010) (0.003) (0.004) 0.010

Snoring Model 1 Model 2 Mean (SE) Mean (SE)

15 50 24

Frequency of sleep problems N

Never ,1 – 2 times/wk 3 – 7 times/wk Ptrend Never ,1 – 2 times/wk 3 – 7 times/wk Ptrend Never ,1 – 2 times/wk 3 – 7 times/wk Ptrend Never ,1 – 2 times/wk 3 – 7 times/wk Ptrend Never , 1 – 2 times/wk 3 – 7 times/wk Ptrend Never ,1 – 2 times/wk 3 – 7 times/wk Ptrend Never ,1 – 2 times/wk 3 – 7 times/wk Ptrend Never ,1 – 2 times/wk 3 – 7 times/wk Ptrend

a

67 4 4

67 4 4

71 4 4

71 4 4

70 4 4

71 4 4

71 4 4

71 4 4

N

a

0.22 0.23 0.24

39.4 403 42.0

0.52 0.52 0.53

0.85 0.88 0.83

20.9 21.1 22.6

91.6 93.2 92.7

107.8 105.4 112.1

28.1 28.3 30.8

(0.5) 27.7 (2.2) 28.6 (2.2) 29.9 0.235 (1.1) 107.0 (4.6) 106.6 (4.6) 111.0 0.366 (1.5) 91.6 (6.5) 93.4 (6.5) 92.7 0.868 (0.4) 20.5 (1.6) 21.0 (1.6) 22.1 0.283 (0.01) 0.85 (0.04) 0.87 (0.04) 0.84 0.665 (0.01) 0.53 (0.03) 0.52 (0.03) 0.54 0.847 (0.5) 38.4 (2.0) 39.7 (2.0) 40.3 0.206 (0.003) 0.22 (0.011) 0.23 (0.011) 0.23 0.156

(0.7) (2.0) (2.2) 0.312 (1.5) (4.5) (4.7) 0.391 (1.8) (5.3) (5.6) 0.845 (0.5) (1.4) (1.5) 0.254 (0.01) (0.03) (0.03) 0.622 (0.01) (0.03) (0.03) 0.844 (0.6) (1.7) (1.8) 0.275 (0.004) (0.010) (0.011) 0.316

Stop breathing Model 1 Model 2 Mean (SE) Mean (SE)

74 13 3

74 13 3

78 13 3

78 13 3

77 13 3

78 13 3

78 13 3

78 13 3

N

a

0.22 0.23 0.25

39.2 40.3 43.3

0.52 0.52 0.57

0.85 0.84 0.86

20.6 20.6 23.8

90.6 91.1 99.2

106.8 107.8 115.9

27.8 29.0 32.0

(0.5) (1.2) (2.5) 0.108 (1.1) (2.6) (5.4) 0.105 (1.4) (3.6) (7.5) 0.265 (0.4) (0.9) (1.8) 0.091 (0.01) (0.02) (0.05) 0.764 (0.01) (0.02) (0.04) 0.200 (0.5) (1.2) (2.4) 0.096 (0.003) (0.006) (0.013) 0.057 0.22 0.24 0.25

38.5 41.0 43.2

0.52 0.53 0.58

0.85 0.87 0.87

20.2 20.5 24.3

89.8 92.8 101.0

104.9 106.5 116.7

27.1 28.9 32.3

(0.6) (1.4) (2.5) 0.036 (1.4) (3.1) (5.5) 0.036 (1.6) (3.4) (6.1) 0.068 (0.4) (0.9) (1.6) 0.014 (0.01) (0.02) (0.03) 0.606 (0.01) (0.02) (0.04) 0.089 (0.5) (1.1) (2.0) 0.023 (0.003) (0.007) (0.013) 0.042

Gasping for breath Model 1 Model 2 Mean (SE) Mean (SE)

Notes: aThe sample sizes are for model 1. Model 1: Adjusted for age; Model 2: Adjusted for age, physical activity, smoking, alcohol intake, sex, depression, and shift-work. Source: Charles et al. (1999)

Beck-to-height ratio

Neck circumference

Waist-to-height ratio

Waist-to-hip ratio

Abdominal height

Waist

Hip circumference

BMI

Anthropometric obesity measurements

Obesity and sleep

209

Table III. Adjusted mean values of anthropometric obesity measurements by sleep problems

PIJPSM 30,2

210

circumference, and neck-to-height ratio were significantly associated with reports of waking up gasping for breath after adjustment for age, physical activity, smoking, alcohol intake, sex, depression, and shift-work. In almost every case, persons who reported experiencing a sleep-disordered breathing problem three to seven times per week had the largest mean anthropometric measurement. An inverse trend was observed between duration of sleep and mean waist circumference: 0-4.5 hrs (92.1 cm), 5-6.5 hrs (90.3 cm), and $ seven hrs (89.2 cm), but the trend was not statistically significant ( p ¼ 0.381). Additional analyses were performed to determine which of the anthropometric measures best predicted the association with snoring (data not shown). Logistic regression analyses, using various methods (e.g. stepwise, forward, backward, and anthropometric measures in the models simultaneously and separately), identified abdominal height as the best predictor of snoring. In addition, the prevalence of snoring was obtained for each quartile of the anthropometric measures and the difference between the first and fourth quartiles was calculated. A total of 100 percent of persons in the fourth quartile of abdominal height reported snoring, and the interquartile difference in snoring prevalence was also the largest for abdominal height (31.6 percent). Discussion The results of this study showed statistically significant associations between all anthropometric measures (except waist-to-hip ratio) and snoring. Abdominal height was consistently identified as the best predictor of snoring prevalence. Significant trends were also observed between some of the anthropometric measures and waking up gasping for breath. Positive trends were observed between some of the other anthropometric measures and “stop breathing during sleep”, but these were not statistically significant, probably due to the small sample sizes. Our results are in concordance with previous studies. Young et al. (2005) estimated that about 17 percent of American adults aged 30-69 years have mild or worse sleep-disordered breathing and that in 41 percent of these persons, the sleep-disordered breathing is attributable to having a BMI . ¼ 25 kg/m2. They also estimated that about 5.7 percent of these adults have moderate or worse sleep-disordered breathing and that the sleep-disordered breathing is attributable to excess weight in 58 percent of these adults. In a cross-sectional study, Khoo et al. (2004) identified obesity and neck circumference, among several other factors, as risk factors for snoring and sleep-disordered breathing in an Asian population. In a large cohort study, patients with obstructive sleep apnea syndrome were found to be more obese (Yaggi et al., 2005). In the present study, persons reporting sleep problems three to seven times per week had a mean BMI of approximately 30 kg/m2, which is indicative of obesity. Anatomical reasons for sleep-disordered breathing problems include large tonsils, a long pharyngeal airway, and the collapsibility potential of the airway (Schellenberg et al., 2000; Malhotra et al., 2002). We found no significant association between any of the anthropometric measures and duration of sleep – although these results may have been partially due to small sample size. These results contradict some previous findings (Hasler et al., 2004). Results from the National Sleep Foundation survey found that those who were considered obese were more likely than those who were underweight or average weight

to sleep less than six hours per night on weekdays (18 percent vs. 11 percent) and to be at risk for sleep apnea (57 percent vs. 10 percent) (National Sleep Foundation, 2005). In a prospective study of 1,001 patients from four primary care practices in Virginia, total sleep time decreased as BMI increased, except in the extremely obese group (Vorona et al., 2005). This association was found even after patients with sleep disorders were excluded. Physiology may play an important role in the obesity-sleep association. In a clinical trial, sleep restriction was observed to increase appetite, especially for calorie-dense foods with high carbohydrate content, and reduce secretion of the anorexigenic hormone leptin (Spiegel et al., 2004). Among 1024 participants of the Wisconsin Sleep Cohort Study, serum ghrelin, leptin, adiponectin, insulin, and glucose were significantly correlated with BMI (Taheri et al., 2004). There was a significant increasing trend in leptin with increasing average nightly sleep duration, and a significant decreasing trend in ghrelin with increasing total sleep time. Leptin suppresses appetite while ghrelin stimulates appetite (Taheri et al., 2004). It is possible that a vicious cycle is created when sleep loss results in decreased leptin production causing an increase in appetite which results in obesity. Then obesity may increase the risk of sleep-disordered breathing problems which would exacerbate the sleep-deprived condition and decrease leptin levels. These potential pathways could be explored in future prospective studies. Depression may also play an integral role in the association between obesity and sleep. Several studies have shown that obese persons tend to experience more symptoms of depression than persons of normal weight (Kress et al., 2006; Herva et al., 2006; Simon et al., 2006) and that depression can also result in obesity (Goodman and Whitaker, 2002). There is also evidence to support the fact that depressed persons have more disturbed sleep than those who are not depressed (Kaneita et al., 2006). Disturbed sleep affects levels of melatonin, a substance that is primarily secreted by the pineal gland and is produced during the night to promote sleep (Pandi-Perumal et al., 2006). In turn, low levels of melatonin are associated with depression (Pandi-Perumal et al., 2006), which creates or further exacerbates any existing sleep disorders and obesity. There are limitations to this study. The information on sleep quality was obtained by subjective reporting with no objective means of validating the responses and the possibility exists that sleep problems may have been overstated or understated. Such bias would likely have been non-differential, with a tendency to weaken any association that may have existed. The use of self-reported instead of measured sleep duration is less likely to be problematic because the congruency between self-reported sleep duration is likely to be good with those obtained through actigraphic monitoring. Due to the cross-sectional design of the study, we are unable to determine the temporal nature of the associations between sleep problems and obesity. Another limitation is the relatively small sample size, which prevented stratification of the models by sex. A major strength of the study was that it was conducted in a cooperative and motivated study population. We were able to assess a wide range of anthropometric measurements, including many that are important but are rarely used to measure obesity in epidemiologic studies such as abdominal height, neck circumference, and neck-to-height ratio. In addition, the assessment of the anthropometric indices was performed by trained clinic staff using standardized procedures. Although we did not

Obesity and sleep

211

PIJPSM 30,2

212

use a standardized sleep questionnaire, several of the sleep questions used in this study are also used in validated questionnaires. Finally, we were able to assess the potential effects of confounding. In conclusion, results show that several measures of obesity were significantly associated with sleep-disordered breathing problems in this sample of police officers. Previous studies have shown that severe sleep-disordered breathing is associated with an increased risk of cardiac arrhythmias, stroke, and death (Mehra et al., 2006; Yaggi et al., 2005) and that police officers are disproportionately at risk for such health problems (Violanti et al., 1998). Future studies investigating this question are recommended and would be enhanced by using a prospective design and optimal indicators of obesity.

References Charles, L.E., Burchfiel, C.M., Fekedulegn, D., Andrew, M.E., Violanti, J.M. and Vila, B. (1999), Buffalo Police Health Study, 1999. Buysse, D.J., Reynolds, C.F., Monk, T.H., Berman, S.R. and Kupfer, D.J. (1989), “The Pittsburgh Sleep Quality Index: a new instrument for psychiatric practice and research”, Psychiatry Research, Vol. 28 No. 2, pp. 193-213. Dancey, D.R., Hanly, P.J., Soong, C., Lee, B., Shepard, J. Jr and Hoffstein, V. (2003), “Gender differences in sleep apnea: the role of neck circumference”, Chest, Vol. 123 No. 5, pp. 1544-50. Dawson, D. and Reid, K. (1997), “Fatigue, alcohol and performance impairment”, Nature, Vol. 388 No. 6639, p. 235. Foley, D., Monjan, A., Masaki, K., Ross, W., Havlik, R., White, L. and Launer, L. (2001), “Daytime sleepiness is associated with 3-year incident dementia and cognitive decline in older Japanese-American men”, Journal of the American Geriatrics Society, Vol. 49 No. 12, pp. 1628-32. Gangwisch, J.E., Malaspina, D., Boden-Albala, B. and Heymsfield, S.B. (2005), “Inadequate sleep as a risk factor for obesity: analyses of the NHANES I”, Sleep, Vol. 28 No. 10, pp. 1289-96. Gangwisch, J.E., Heymsfield, S.B., Boden-Albala, B., Buijs, R.M., Kreier, F., Pickering, T.G., Rundle, A.G., Zammit, G.K. and Malaspina, D. (2006), “Short sleep duration as a risk factor for hypertension: analyses of the First National Health and Nutrition Examination Survey”, Hypertension, Vol. 47 No. 15, pp. 833-9. Goodman, E. and Whitaker, R.C. (2002), “A prospective study of the role of depression in the development and persistence of adolescent obesity”, Pediatrics, Vol. 110 No. 3, pp. 497-504. Hasler, G., Buysse, D.J., Klaghofer, R., Gamma, A., Ajdacic, V., Eich, D., Rossler, W. and Angst, J. (2004), “The association between short sleep duration and obesity in young adults: a 13-year prospective study”, Sleep, Vol. 27 No. 4, pp. 661-6. Herva, A., Laitinen, J., Miettunen, J., Veijola, J., Karvonen, J.T., Laksy, K. and Joukamaa, M. (2006), “Obesity and depression: results from the longitudinal Northern Finland 1966 Birth Cohort Study”, International Journal of Obesity (Lond), Vol. 30 No. 3, pp. 520-7. Kaneita, Y., Ohida, T., Uchiyama, M., Takemura, S., Kawahara, K., Yokoyama, E., Miyake, T., Hara, S., Suzuki, K. and Fujita, T. (2006), “The relationship between depression and sleep disturbance: a Japanese nationwide general population survey”, The Journal of Clinical Psychiatry, Vol. 67 No. 2, pp. 196-203.

Khoo, S.M., Tan, W.C., Ng, T.P. and Ho, C.H. (2004), “Risk factors associated with habitual snoring and sleep-disordered breathing in a multi-ethnic Asian population: a population-based study”, Respiratory Medicine, Vol. 98 No. 6, pp. 557-66. Kress, A.M., Peterson, M.R. and Hartzell, M.C. (2006), “Association between obesity and depressive symptoms among US Military active duty service personnel, 2002”, Journal of Osychosomatic Research, Vol. 60 No. 3, pp. 263-71. Maislin, G., Pack, A.I., Kribbs, N.B., Smith, P.L., Schwartz, A.R., Kline, L.R., Schwab, R.J. and Dinges, D.F. (1995), “A survey screen for prediction of apnea”, Sleep, Vol. 18 No. 3, pp. 158-66. Malhotra, A., Huang, Y., Fogel, R.B., Pillar, G., Edwards, J.K., Kikinis, R., Loring, S.H. and White, D.P. (2002), “The male predisposition to pharyngeal collapse: importance of airway length”, American Journal of Respiratory and Critical Care Medicine, Vol. 166 No. 10, pp. 1388-95. Mehra, R., Benjamin, E.J., Shabar, E., Gottlieb, D.J., Nawabit, R., Kirchner, H.L., Sahadevan, J. and Redline, S. (2006), “Association of nocturnal arrhythmias with sleep-disordered breathing: the Sleep Heart Health Study”, American Journal of Respiratory and Critical Care Medicine, Vol. 173 No. 8, pp. 910-6. Morioka, H.M. and Brown, M.L. (1970), “Incidence of obesity and overweight among Holulu police and firemen”, Public Health Reports, Vol. 85 No. 5, pp. 433-9. National Sleep Foundation (2005), 2004 Sleep in America Poll, National Sleep Foundation, Washington, DC. Neylan, T.C., Metzler, T.J., Best, S.R., Weiss, D.S., Fagan, J.A., Liberman, A., Rogers, C., Vedantham, K., Brunet, A., Lipsey, T.L. and Marmar, C.R. (2002), “Critical incident exposure and sleep quality in police officers”, Psychosomatic Medicine, Vol. 64 No. 2, pp. 345-52. Pandi-Perumal, S.R., Srinivasan, V., Maestroni, G.J., Cardinali, D.P., Poeggeler, B. and Hardeland, R. (2006), “Melatonin: nature’s most versatile biological signal?”, The FEBS Journal, Vol. 273 No. 13, pp. 2813-38. Radloff, L.S. (1977), “The CES-D scale: a self-report depression scale for research in the general population”, Applied Psychological Measurement, Vol. 1, pp. 385-401. SAS Institute (2002), SAS User’s Guide: Statistics, Version 9.1, SAS Institute, Inc., Cary, NC. Schellenberg, J.B., Maislin, G. and Schwab, R.J. (2000), “Physical findings and the risk of obstructive sleep apnea: the importance of oropharyngeal structures”, American Journal of Respiratory and Critical Care Medicine, Vol. 162 No. 2, pp. 740-8. Sharma, S.K., Kurian, S., Malik, V., Mohan, A., Banga, A., Pandey, R.M., Handa, K.K. and Mukhopadhyay, S. (2004), “A stepped approach for prediction of obstructive sleep apnea in overtly asymptomatic obese subjects: a hospital based study”, Sleep Medicine, Vol. 5 No. 4, pp. 351-7. Simon, G.E., Von Korff, M., Saunders, K., Miglioretti, D.L., Crane, P.K., Van Belle, G. and Kessler, R.C. (2006), “Association between obesity and psychiatric disorders in the US adult population”, Archives of General Psychiatry, Vol. 63 No. 7, pp. 824-30. Spiegel, K., Leproult, R. and Van Cauter, E. (1999), “Impact of sleep debt on metabolic and endocrine function”, Lancet, Vol. 354 No. 9188, pp. 1435-9. Spiegel, K., Tasali, E., Penev, P. and Van Cauter, E. (2004), “Brief communication: sleep curtailment in healthy young men is associated with decreased leptin levels, elevated ghrelin levels, and increased hunger and appetite”, Annals of Internal Medicine, Vol. 141 No. 11, pp. 846-50.

Obesity and sleep

213

PIJPSM 30,2

214

Stein, C.J. and Colditz, G.A. (2004), “The epidemic of obesity”, The Journal of Clinical Endocrilogy and Metabolism, Vol. 89 No. 6, pp. 2522-5. Taheri, S., Lin, L., Austin, D., Young, T. and Migt, E. (2004), “Short sleep duration is associated with reduced leptin, elevated ghrelin, and increased body mass index”, PLoS Medicine, Vol. 1 No. 3, pp. 210-7. Vila, B. (2006), “Impact of long work hours on police officers and the communities they serve”, American Journal of Industrial Medicine, Vol. 49 No. 11, pp. 972-80. Violanti, J.M., Vena, J.E. and Petralia, S. (1998), “Mortality of a police cohort: 1950-1990”, American Journal of Industrial Medicine, Vol. 33 No. 4, pp. 366-73. Violanti, J.M., Burchfiel, C.M., Miller, D.B., Andrew, M.E., Dorn, J., Wactawski-wende, J., Beighley, C.M., Pieri, K., Joseph, P.N., Vena, J.E., Sharp, D.S. and Trevisan, M. (2006), “The Buffalo Cardio-Metabolic Occupational Police Stress (BCOPS), pilot study: methods and participant characteristics”, Annals of Epidemiology, Vol. 16 No. 2, pp. 148-56. Vorona, R.D., Winn, M.P., Babineau, T.W., Eng, B.P., Feldman, H.R. and Ware, C. (2005), “Overweight and obese patients in a primary care population report less sleep than patients with a rmal body mass index”, Archives of Internal Medicine, Vol. 165 No. 1, pp. 25-30. Williamson, A.M. and Feyer, A.M. (2000), “Moderate sleep deprivation produces impairments in cognitive and motor performance equivalent to legally prescribed levels of alcohol intoxication”, Occupational and Environmental Medicine, Vol. 57 No. 10, pp. 649-55. Yaggi, H.K., Concato, J., Kernan, W.N., Lichtman, J.H., Brass, L.M. and Mohsenin, V. (2005), “Obstructive sleep apnea as a risk factor for stroke and death”, The New England Journal of Medicine, Vol. 353 No. 19, pp. 2034-41. Young, T., Peppard, P.E. and Taheri, S. (2005), “Excess weight and sleep-disordered breathing”, Journal of Applied Physiology, Vol. 99 No. 4, pp. 1592-9. Young, T., Evans, L., Finn, L. and Palta, M. (1997), “Estimation of the clinically diagnosed proportion of sleep apnea syndrome in middle-aged men and women”, Sleep, Vol. 20 No. 9, pp. 705-6. Yousaf, F. and Sedgwick, P. (1996), “Sleep disorders”, British Journal of Hospital Medicine, Vol. 55 No. 6, pp. 353-8. Further reading Kripke, D.F., Garfinkel, L., Wingard, D.L., Klauber, M.R. and Marler, M.R. (2002), “Mortality associated with sleep duration and insomnia”, Archives of General Psychiatry, Vol. 59 No. 2, pp. 131-6. Corresponding author Luenda E. Charles can be contacted at: [email protected]

To purchase reprints of this article please e-mail: [email protected] Or visit our web site for further details: www.emeraldinsight.com/reprints

The current issue and full text archive of this journal is available at www.emeraldinsight.com/1363-951X.htm

Shift work and sleep: the Buffalo Police health study

Shift work and sleep

Luenda E. Charles, Cecil M. Burchfiel and Desta Fekedulegn Biostatistics and Epidemiology Branch, Health Effects Laboratory Division, National Institute for Occupational Safety and Health, Centers for Disease Control and Prevention, Morgantown, West Virginia, USA

215

Bryan Vila Department of Political Science, Washington State University, Spokane, Washington, USA

Tara A. Hartley, James Slaven and Anna Mnatsakanova Biostatistics and Epidemiology Branch, Health Effects Laboratory Division, National Institute for Occupational Safety and Health, Centers for Disease Control and Prevention, Morgantown, West Virginia, USA, and

John M. Violanti School of Public Health and Health Professions, Department of Social and Preventive Medicine, State University of New York at Buffalo (JMV), Buffalo, New York, USA Abstract Purpose – Working on the night shift is a potential source of occupational stress and has been associated with sleep disorders. The purpose of this paper is to investigate the association between shift work and sleep problems among police officers from Buffalo, New York. Design/methodology/approach – Randomly selected officers (n ¼ 111) responded to questions on sleep quality and quantity. Shift work data were obtained from daily payroll records from 1994 to the exam date (1999-2000). Prevalence ratios (PR) were obtained using Poisson regression models that examined associations of shift work with sleep quality and quantity. Findings – Among police officers, night shift work was significantly and independently associated with snoring and decreased sleep duration. Originality/value – Although the sleep questions were similar to those used in validated sleep questionnaires, a major strength of this study was the availability of daily work history data on all officers for up to five years prior to the current examination. Keywords Shift work, Police, Stress, United States of America Paper type Research paper

Introduction Research shows that up to 20 percent of workers are not able to tolerate working at night (Scott and Ladou, 1990). This is due to the influence of the circadian rhythms, the The authors would like to thank Dr Chunlin Dong for her assistance with dataset preparation. The findings and conclusions in this article are those of the authors and do not necessarily represent the views of the National Institute for Occupational Safety and Health.

Policing: An International Journal of Police Strategies & Management Vol. 30 No. 2, 2007 pp. 215-227 q Emerald Group Publishing Limited 1363-951X DOI 10.1108/13639510710753225

PIJPSM 30,2

216

disruption of which may lead to dysregulation of physiological mechanisms (Minors and Waterhouse, 1990). Working on schedules other than the regular day shift is a potential source of occupational stress (Brugere et al., 1997) and the consequences include various problems such as increased risk of injuries at work (Smith et al., 1994; Monk, 1990), gastrointestinal symptoms (Caruso et al., 2004), disruption of family and social lives (Colligan and Rosa, 1990), and mental disorders (Scott et al., 1997; Shields, 2002). One of the most common problems related to shift work is a decrease in the quality and quantity of sleep. Certain sleep disorders may increase the risk of morbidity and mortality. For example, daytime sleepiness has been associated with impaired cognition (Foley et al., 2001) and decreased sleep time has been linked with increased risk of hypertension (Gangwisch et al., 2006), increased risk of Type 2 diabetes (Yaggi et al., 2006), and reduced life expectancy (Yousaf and Sedgwick, 1996). A cohort study by Yaggi et al. found that obstructive sleep apnea significantly increased the risk of stroke or death (Yaggi et al., 2005). Snoring is associated with excessive daytime sleepiness, headaches, and reduced ability to concentrate and learn new tasks (Ulfberg et al., 1996a; 1996b). Poor sleep quality influences the secretion of pro-inflammatory substances and stress hormones in the body (Vgontzas et al., 2004; Backhaus et al., 2004). Fatigue due to moderate sleep loss affects performance similar to moderate alcohol intoxication (Williamson and Feyer, 2000; Dawson and Reid, 1997). Shift work may affect the quality of police officers’ sleep, potentially threatening their health, safety, and performance (Gabarino et al., 2002; Vila, 2006; Vila et al., 2002). The primary objective of this study was to examine the potential association of shift work with sleep quality and quantity among police officers in Buffalo, New York. Methods Study population The Buffalo New York Police Department, a mid-sized urban police force of 934 officers at the time of sampling, was the selected site for this study. The Center for Preventive Medicine, School of Public Health and Health Professions, State University of New York at Buffalo, served as the data collection site (Violanti et al., 2006). In 1999, a random sample of 115 police officers was selected from all officers in the department using a computer-generated random sample table. No specific exclusion criteria were used. All of these officers agreed to participate in the study. Informed consent was obtained from all participants. This study was approved by the State University of New York at Buffalo Internal Review Board and the National Institute for Occupational Safety and Health Human Subjects Review Board. Four officers who did not have sufficient work history data were excluded from the study resulting in a sample size of 111. Assessment of shift work Shift work data were obtained from city of Buffalo payroll records, which provided a day-by-day account of shift history for each officer in the present study from 1994 to the date of exam at baseline (1999-2000). The time that participants started their shift was used to classify the participants into one of the following three shifts: day shift, if the start time was between 0400 and 1159; afternoon shift, if the start time was between 1200 and 1959; and night shift, if the start time was between 2000 and 0359. The total

number of hours worked during the day, afternoon, and night shift were computed for each participant. Taking into account the length of time a participant was working (from first date of work to date of exam at baseline), the computed hours were standardized on a weekly basis (i.e. hours worked per week). In addition, the percentage of total hours worked on each shift was calculated. Two types of dichotomous shift work variables were created. In the first method, participants were classified as shift workers if the combined proportion of hours worked on the afternoon and night shifts was at least 50 percent. In the second method, participants were classified as shift workers if the proportion of hours worked on the night shift was at least 50 percent.

Shift work and sleep

217

Assessment of sleep The outcomes were sleep quality and sleep quantity. The questions on sleep quality are shown in Table I. Several of these questions have been validated in instruments such as the Pittsburgh Sleep Quality Index questionnaire (PSQI) (Buysse et al., 1989) and the sleep apnea screen used by Maislin et al. (1995). These questions were evaluated on a six-point frequency scale and the response choices were “not sure/don’t know/not applicable”, “strongly disagree/never”, “disagree/less than once per week”, “somewhat agree/1 –2 times per week”, “agree/3 – 4 times per week”, and “strongly agree/5 –7 times per week”. All sleep problems were re-categorized and analyzed as: a) “yes” versus “never”; and b) “3-7 times per week”, “ , 1-2 times per week”, and “never”. Participants were also asked how many hours, on average, they slept each 24-hour

Sleep conditions

N

At night, my sleep disturbs my bed partner’s sleep 75 I am told I snore in my sleep 90 I am told I stop breathing in my sleep 79 I suddenly wake up gasping for breath during the night 95 I have or have been told that I have restless legs 81 I feel tired upon awakening and want to go back to sleep 99 I am very sleepy during the daytime and struggle to stay awake 100 Hours of sleep per 24-hour period during the previous week * 111 0 – 4.5 5 – 6.5 $7

All workers No. of cases %

Night shift workers N No. of cases %

Day and afternoon shift workers N No. of cases %

57 75

76.0 14 83.3 17

10 17

71.4 61 100.0 73

47 58

77.1 79.5

9

11.4 15

1

6.7 64

8

12.5

17

17.9 18

1

5.6 77

16

20.8

28

34.6 17

7

41.2 64

21

32.8

89

89.9 18

16

88.9 81

73

90.1

72

72.0 18

12

66.7 82

60

73.2

13 53 45

11.7 47.8 40.5

1 15 4

5.0 75.0 20.0

12 38 41

13.2 41.8 45.1

20

91

Note: *p-values for difference between night and day/afternoon shift workers , 0.05 Source: Charles et al. (1999)

Table I. Prevalence of sleep problems and duration among police officers by shift work

PIJPSM 30,2

period during the previous five weekdays (i.e. Sunday through Thursday) and during the previous weekend (i.e. Friday and Saturday nights). We averaged the hours of sleep reported for the two periods to give the average total hours of sleep per 24-hour period during the previous seven days. Due to the small sample size in some groups, sleep duration was re-categorized as 0 –4, 5 – 6, and $ seven hours for analysis of variance (ANOVA) and as , 7 and $ seven hours for Poisson regression analyses.

218 Assessment of covariates Participants completed questionnaires (self- and interviewer-administered) to provide information on demographic characteristics, lifestyle behaviors and medical history. Police officers reported their highest level of education that ranged from “less than 12 years of school” to “graduate degree”. Participants indicated the number of years employed as a police officer and their present rank in the police force (e.g. police patrol officer, lieutenant, detective, etc.). Participants were asked how often they consumed alcoholic beverages where one drink consisted of a 12 oz can or bottle of beer, one medium glass of wine, or one shot of liquor. The total number of drinks per month (of each type) was summed and then divided by four to give the approximate total number of drinks consumed per week. Participants reported their smoking status as current, former, or never. Body mass index (BMI) was calculated as kilogram per meter squared. The Center for Epidemiologic Studies Depression scale (CES-D) was used to measure depression in the participants (Radloff, 1977). The CES-D is a 20-item test, which has good reliability for measuring symptoms of depression and stress (Chronbach alpha of 0.85) (Radloff, 1977). Respondents rated items on a four-point scale based on the frequency of symptom occurrence in the previous seven days: 0 (rarely or none of the time, less than 1 day), 1 (some or little of the time, 1 –2 days), 2 (occasionally or a moderate amount of time, 3 – 4 days), and 3 (most of the time, 5 – 7 days). Four of the 20 items were designed to identify positive symptoms and hence were reverse-coded. The CES-D score is the sum of the scores from these 20 items and range from 0 to 60. Respondents with scores of 6 – 15 are unlikely to be clinically depressed, scores of 16– 21 indicate mild to moderate depression, and scores of 22 or greater are associated with major depression (Radloff, 1977). Respondents reported the duration (hours per week, hours per weekend) and intensity (moderate, hard, very hard) of three types of physical activity (occupational, household, and sports) that they engaged in during the previous seven days. A total physical activity score was then computed by summing the intensities of the three types (i.e. occupational, household, and sports) of physical activity performed during the weekday and the weekend. Intensity score was computed as the product of number of hours and intensity of physical activity, where moderate, hard, and very hard levels were assigned intensities of 1, 2, and 3, respectively. Statistical methods Descriptive statistics were used to characterize the study population. ANOVA models were used to examine trends in mean values of selected covariates across ordinal categories of sleep problems. Orthogonal contrast coefficients were used to compute the p-values assessing linear trend. Prevalence of each sleep problem was compared across shift work categories. Poisson regression analyses relating shift work to each sleep problem were performed and estimates of prevalence ratios and the

corresponding 95 percent confidence intervals (CIs) were computed (Zou, 2004). Unadjusted and multivariable-adjusted prevalence ratios were also estimated. Prevalence ratios were used instead of odds ratios because they are the preferred effect measure for use in epidemiologic studies when the outcome event is common (i.e. . 10 percent prevalence) (Zou, 2004; Checkoway et al., 2004). Effect modification of the association between shift work and sleep problems was assessed for gender, age, smoking status, alcohol consumption, education, physical activity, and BMI. All analyses were conducted using the SAS system, version 9.1 (SAS Institute, 2001).

Shift work and sleep

219

Results Participants in this study, 63.1 percent of who were men, ranged in age from 26 to 61 years old (mean ^ S.D. ¼ 39.4 ^ 7.5 years) (Table II). A total of 53 officers (47.8 percent) worked the day shift and 20 officers (18 percent) worked the night shift.

Covariates

N

All officers Mean ^ SD

Night shift workers (n ¼ 20) N Mean ^ SD

Day and afternoon shift workers (n ¼ 91) N Mean ^ SD

111 111 107 111 105 111

39.4 ^ 7.5 13.5 ^ 8.7 27.9 ^ 4.4 2.9 ^ 3.7 7.2 ^ 6.8 11.3 ^ 16.6

20 20 19 20 16 20

36.4 ^ 5.6 8.0 ^ 6.2 28.3 ^ 4.3 2.0 ^ 2.4 5.9 ^ 3.9 15.2 ^ 28.6

91 91 88 91 89 91

40.1 ^ 7.8 * 14.8 ^ 8.7 * 27.8 ^ 4.4 3.1 ^ 3.9 7.5 ^ 7.2 10.5 ^ 12.7

Gender: Women Men

41 70

36.9 63.1

4 16

20.0 80.0

37 54

40.7 59.3

Education (%): # High school/GED College , four years College $ four years

21 59 31

18.9 53.2 27.9

3 13 4

15.0 65.0 20.0

18 46 27

19.8 50.6 29.7

Rank (%): Police officer Sergeant/lieutenant/captain Detective Other

73 18 19 1

65.8 16.2 17.1 0.9

17 3 0 0

85.0 15.0 0.0 0.0

56 15 19 1

61.5 16.5 20.9 1.1

Smoking status (%): Current Former Never

22 35 53

20.0 31.8 48.2

7 5 7

36.8 26.3 36.8

15 30 46

16.5 33.0 50.5

Marital status (%): Single Married Divorced

25 70 15

22.5 63.1 14.4

7 12 1

35.0 60.0 5.0

18 58 15

19.8 63.7 16.5

Age (years) Years of service BMI (kg/m2) Alcohol (number of drinks/wk) Depression score (CES-D) Physical activity score

Note: *p-values for difference between night and day/afternoon shift workers , 0.05 Source: Charles et al. (1999)

Table II. Characteristics of the study population and mean values and standard deviation (SD) of selected covariates by shift work

PIJPSM 30,2

220

Compared to day and afternoon shift workers, night shift workers were significantly more likely to be younger (36.4 vs. 40.1 years; p ¼ 0.047) and to have fewer years of service (8.0 vs. 14.8 years; p ¼ 0.001). Women were twice as likely to work the day and afternoon shift (40.7 percent) as the night shift (20.0 percent). The majority of officers had completed at least some college education (81 percent), and 65.8 percent of the officers reported their rank as police patrol officer. Although not statistically significant, current smoking was more prevalent in night shift workers (36.8 percent) than in day and afternoon workers (16.5 percent). The prevalence of sleep problems varied widely (Table I). The majority of officers reported feeling tired upon awakening (89.9 percent) and snoring (83.3 percent). In contrast, only nine officers (11.4 percent) reported having sleep apnea (stop breathing during sleep) and 17 officers (17.9 percent) said that they wake up gasping for breath during the night. This prevalence is based on the subset of officers who answered these specific questions. Only one of the officers who reported sleep apnea problems worked the night shift. Overall, 45 officers (40.5 percent) reported getting $ seven hours of sleep in a 24-hour period during the previous week and only four (20.0 percent) of these were night shift workers. The mean values of selected covariates were obtained for three levels of sleep problems (data not shown). No association between years of service and sleep problems was observed. Overall, mean CES-D score (depression) increased with increasing frequency of most sleep problems; the associations were significant for restless leg syndrome ( p for trend ¼ 0.024) and tiredness upon awakening ( p for trend ¼ 0.015). BMI was positively and significantly associated with snoring ( p for trend , 0.001). A positive trend also was observed between BMI and waking up gasping for breath but the association was not statistically significant. Although none of the trends for physical activity scores were statistically significant, lower scores were more likely to be seen in officers with more frequent sleep problems. An inverse trend between physical activity and tiredness upon awakening was evident but the association was only borderline significant ( p for trend ¼ 0.059). Univariate and multivariate models of the association between night shift work and sleep problems are shown in Table III. Night shift work was significantly associated with snoring. The prevalence of snoring was 26 percent higher in night shift workers compared to the other workers (PR ¼ 1.26; 95 percent C.I. ¼ 1.12 – 1.41). Adjustment for years of service (PR ¼ 1.21; 95 percent C.I. ¼ 1.06 – 1.38) and other risk factors (depression, BMI, physical activity, and gender) did not alter this significant association appreciably (PR ¼ 1.16; 95 percent C.I. ¼ 1.00 – 1.33). Among officers with BMI , 30 kg/m2, the prevalence of snoring was 22 percent higher among night shift workers compared to other workers (PR ¼ 1.22; 95 percent C.I. ¼ 1.02-1.46) after adjustment for years of service, depression, and physical activity. The association between shift work and snoring among officers with BMI $ 30 kg/m2 was not statistically significant. BMI was a significant effect modifier in the association between shift work and snoring but not with other sleep problems. After risk factor adjustment, the prevalence of short sleep duration (, ¼ 6.5 hours) per 24-hour period during the previous week was 44 percent higher among night shift workers compared to other workers (PR ¼ 1.44; 95 percent C.I. ¼ 1.00 – 2.06; p ¼ 0.047). Age was omitted from the final models since it was strongly correlated with years of service.

Sleep quality/quantity At night, my sleep disturbs my bed partner’s sleep I am told I snore in my sleep I am told I stop breathing in my sleep I suddenly wake up gasping for breath during the night I have or have been told that I have restless legs I feel tired upon awakening and want to go back to sleep I am very sleepy during the daytime and struggle to stay awake Hours of sleep per 24-hour period during the previous week (, 7 hr vs. $ 7 hr)

Model 1 Model 2 Model 3 PR 95% C.I. PR 95% C.I. PR 95% C.I.

Shift work and sleep

0.93 0.65-1.33 0.99 0.68-1.45 0.94 0.62-1.43 1.26 1.12-1.41 1.21 1.06-1.38 1.16 1.00-1.33 0.53 0.07-3.95 0.67 0.08-5.68 0.59 0.02-14.3 0.27 0.04-1.89 0.25 0.03-1.89 0.27 0.03-2.67 1.26 0.64-2.45 0.91 0.46-1.80 1.02 0.53-1.95

221

0.99 0.83-1.18 0.97 0.82-1.16 0.97 0.81-1.16 0.91 0.64-1.30 0.83 0.58-1.19 0.90 0.62-1.32 1.46 1.09-1.94 1.35 0.99-1.85 1.44 1.00-2.07

Notes: PR=prevalence ratio; C.I.=confidence interval. Results obtained from Poisson regression models; Model 1 – unadjusted; Model 2 – adjusted for years of service; Model 3 – adjusted for years of service, depression, BMI, physical activity, and gender. Source: Charles et al. (1999)

The association between shift work and sleep was stratified by gender, age, alcohol consumption, smoking status, education, and physical activity to determine whether these factors might modify the association between shift work and sleep (data not shown). Among women, the prevalence of restless leg syndrome was 134 percent higher among night shift workers compared to workers on other shifts (PR ¼ 2.34; 95 percent C.I. ¼ 1.05-5.24). After adjustment for years of service, depression, BMI, and physical activity, the prevalence among women was still elevated but no longer statistically significant (PR ¼ 1.99; 95 percent C.I. ¼ 0.69-5.78). No association was observed among men. Among officers who had a high school diploma or less, the prevalence of daytime sleepiness was 37 percent higher among night shift workers compared to other workers after risk factor adjustment (PR ¼ 1.37; 95 percent C.I. ¼ 1.01-1.86). Gender, smoking status, education, and physical activity were not found to be significant effect modifiers. Discussion This study was one of the few to investigate shift work related sleep problems among police officers. Unlike the one previous study of police officer sleep problems that compared day, evening and midnight shift officers (Vila et al., 2002), our study compared the sleep problems of officers who worked night shifts with those who worked both day and afternoon shifts. We think this dichotomy is more appropriate because several factors make nightshift work much more likely to have an adverse effect on sleep quality than either day or afternoon shifts. First, night shifts are more disruptive to the circadian system because workers must frequently rotate between diurnal and nocturnal sleep patterns to accommodate family and social activities (Monk, 2000; Sack et al., 1992). This is not the case with afternoon shift workers whose later bedtime hours are seldom a problem biologically (Monk, 2000). In fact, they may even get more sleep than day shift workers (Monk, 2000). Second, the pattern of melatonin secretion and its synchronization with sleep is drastically altered in night

Table III. Association of sleep quality and quantity in police officers who work night shifts compared to those who work day and afternoon shifts

PIJPSM 30,2

222

shift workers (Sack et al., 1992). Melatonin, a substance that is primarily secreted by the pineal gland, is produced during the night and promotes sleep (Pandi-Perumal et al., 2006). Thus, night shift workers may face a greater frequency of sleep disruption compared to evening shift workers. In a random sample of police officers in the city of Buffalo, New York, we found a statistically significant independent association between snoring and night vs. day/afternoon shift workers. There was a 16 percent greater prevalence of snoring among night shift workers compared to workers on day and afternoon shifts. Snoring is an important indicator of sleep problems that may be more readily noticeable by subjects than other measures associated with apnea. We also found a 44 percent higher prevalence of less than seven hours of sleep among night shift workers compared with other workers. Our study did not reveal significant associations for night shift work with any of the other sleep problems. We were unable to find any previous studies reporting an association between night shiftwork and snoring. Furthermore, the epidemiologic literature on shift work and sleep problems shows inconsistent results. Several studies, conducted among various occupational groups, found no associations between shiftwork and sleep problems (Portela et al., 2004; Klawe et al., 2005; Hui et al., 2002), while others reported significantly more sleep problems for workers on the night and/or afternoon shifts compared to the day shifts (Burch et al., 2005; Harma et al., 1998). These apparently contradictory results may be caused by differing definitions of shift work (e.g. night shift, night plus afternoon shift, rotating shifts); sleep assessment instruments (e.g. polysomnography, the Sleep & Health Questionnaire (SHQ), the Epworth Sleepiness Scale (ESS), a global Sleep Disorder Score (SDS)); and sleep questions as well as the wide range of sample sizes employed. Garbarino et al. found no association between shift work and sleep when sleep was assessed by the ESS score, but found a significantly higher prevalence of sleep disorders among shift workers when the global SDS was used (Garbarino et al., 2002). We expected to discover even more significant associations between night shift work and sleep problems than snoring and the 44 percent increased prevalence of sleeping , seven hours per night because night work generally is characterized by increased subjective and objective sleepiness. One reason for night shift sleepiness is that an individual is exposed to work at the nadir (low point) of the well-established circadian pattern (Akerstedt, 2003). For the night shift worker, the period of maximum alertness (late afternoon) will strongly interfere with sleep, whereas the nadir (early morning) will similarly strongly promote sleep during work hours. Investigating associations with shift work presents unique problems because officers who remain on the night shift often tend to be younger and/or better able to adequately tolerate the demands and stresses of working nights (Burch et al., 2005). This especially tends to be the case in departments where officers with more job seniority are given precedence in work-shift selection so that older officers, who tend to tolerate night shift work less well and take longer to recover from sleep disruption, are better able to migrate to shifts that they tolerate better (Vila et al., 2002). Those who are not as easily able to adapt would have to change to other shifts but may still experience sleep problems since the effects of working at night may persist for some time after the shift change (Dumont et al., 1997). Burch et al. found no differences in the prevalence of fatigue or mental symptoms between workers on the night and day shifts suggesting

some degree of adaptation among night workers participating in their study (Burch et al., 2005). In addition, it has been shown that satisfaction with the shift schedule influences shift workers’ ability to cope with their schedule (Axelsson et al., 2004). Information regarding satisfaction with night shift was not collected and so we were unable to assess its influence on our results. In our study, the association between shift work and sleep problems may have been under-estimated. We found a high prevalence of tiredness upon awakening. We could not definitively determine the prevalence of sleep apnea for several reasons. Symptomatic obstructive sleep apnea includes symptoms such as snoring and/or temporary cessation of breathing during sleep, with the usual consequence being hypersomnolence during waking hours (Victor, 1999). We obtained a wide range of responses to each of these questions. The very low response rate to “stop breathing during sleep” may be due to the fact that a number of individuals, especially those without a bed partner, are not aware that they have this problem. Although we collected information on marital status, we did not ask about bed partners nor did we survey the spouses or partners of the participants. In addition, we were unable to assess asymptomatic sleep apnea, which is more common than the symptomatic type (Victor, 1999). In annual surveys of randomly selected American adults, the National Sleep Foundation (NSF) found that increasing numbers of American adults are experiencing sleep problems on a regular basis (National Sleep Foundation, 2005). Of the NSF respondents, 57 percent reported sleeping more than seven hours per night during the workweek as compared to only 45 percent of police officers in our study. Several covariates were significantly associated with sleep problems. For example, BMI showed a significantly positive trend with snoring, and was significantly and positively associated with restless leg syndrome and tiredness upon awakening. Results from previous studies are consistent with this association between BMI and snoring (Yaggi et al., 2005; Lindberg et al., 1998). Our study also found that lower physical activity scores were more likely to be seen in officers with more frequent sleep problems although none of the trends were statistically significant. Previous studies have provided evidence that regular physical activity has a protective effect on sleep-disordered breathing (Peppard and Young, 2004; Marchesini et al., 2004). There are several limitations in this study. The sample size of 111 is relatively small and we cannot rule out misclassification of sleep problems in this study. The information on sleep was obtained by subjective reporting with no objective means of validating the responses and so the possibility exists that sleep problems may have been overstated or understated. Such bias would likely have been non-differential, with a tendency to weaken any association that may have existed. Also, officers were asked to report the duration of sleep only during the previous seven days and this number may not have been representative of their true average sleep duration. In addition, the sleep of shift workers tends to be fragmented and if they did not accurately tabulate their hours of sleep at various times during the 24-hour period, a differential bias would have exaggerated the association between shift work and sleep duration. However, the use of self-reported sleep duration has shown good congruency with those obtained through actigraphic monitoring (Lockley et al., 1999; Kushida et al., 2001). Another limitation pertains to the cross-sectional design of the study which precludes inferences regarding causality. We were not able to investigate the influence of chronotype (i.e. morningness or eveningness) on the shift work-sleep association. Chronotype affects

Shift work and sleep

223

PIJPSM 30,2

224

sleep quality and quantity and does so in a sex-specific manner in young adults (Mongrain et al., 2005). A major strength of the study was the availability of daily work history data on all officers for up to five years prior to the current examination. The work history data included hours of overtime as well as regular work, and this objective measure removed any possibility of information bias. The study was conducted on a stratified random sample of officers with a high (100 percent) response rate. In addition, we were able to assess potential effect modification and confounding on a fairly wide range of sleep problems. Finally, our sleep questions were similar to those used in validated sleep questionnaires. In summary, this study found that night shift work was independently and significantly associated with snoring and short sleep duration. Additional findings indicated that BMI may also be a risk factor for snoring and that physical activity may be protective for several sleep problems. In addition to the many health problems that are associated with snoring (and insufficient sleep), snoring also affects the quality of life of the snorer’s bed partner (Ulfberg et al., 2000). Interventions, such as a worksite exercise program (Atlantis et al., 2006), to alleviate the sleep problems associated with night shift work will reduce the associated health problems and safety issues among police officers and will also improve the quality of life for their bed partners, thus improving the lives of a larger proportion of the population. Future studies investigating the association between shift work and sleep problems may be strengthened by increasing the sample size, utilizing more objective methods to assess sleep quality, and by using a prospective study design. References ˚ kerstedt, T. (2003), “Shift work and disturbed sleep/wakefulness”, Occupational Medicine, A Vol. 53 No. 2, pp. 89-94. Atlantis, E., Chow, C-M., Kirby, A. and Singh, M.A.F. (2006), “Worksite intervention effects on sleep quality: a randomized controlled trial”, Journal of Occupational Health Psychology, Vol. 11 No. 4, pp. 291-304. Axelsson, J., Akerstedt, T., Kecklund, G. and Lowden, A. (2004), “Tolerance to shift work – how does it relate to sleep and wakefulness?”, International Archives of Occupational and Environmental Health, Vol. 77 No. 2, pp. 121-9. Backhaus, J., Junghanns, K. and Hohagen, F. (2004), “Sleep disturbances are correlated with decreased morning awakening salivary cortisol”, Psychoneuroendocrinology, Vol. 29 No. 9, pp. 1184-91. Brugere, D., Barrit, J., Butat, C., Cosset, M. and Volkoff, S. (1997), “Shift work, age, and health: an epidemiologic investigation”, International Journal of Occupational and Environmental Health, Vol. 3 No. 2, pp. S15-S19. Burch, J.B., Yost, M.G., Johnson, W. and Allen, E. (2005), “Sleep, and shift work adaptation”, Journal of Occupational and Environmental Medicine, Vol. 47 No. 9, pp. 893-901. Buysse, D.J., Reynolds, C.F. III, Monk, T.H., Berman, S.R. and Kupfer, D.J. (1989), “The Pittsburgh Sleep Quality Index: a new instrument for psychiatric practice and research”, Psychiatry Research, Vol. 28 No. 2, pp. 193-213. Caruso, C.C., Lusk, S.L. and Gillespie, B.W. (2004), “Relationship of work schedules to gastrointestinal diagnosis, symptoms, and medication use in auto factory workers”, American Journal of Industrial Medicine, Vol. 46 No. 6, pp. 586-98.

Charles, L.E., Burchfiel, C.M., Fekedulegn, D., Andrew, M.E., Violanti, J.M. and Vila, B. (1999), Buffalo Police Health Study, 1999. Checkoway, H., Pearce, N. and Kriebel, D. (2004), “Research methods in occupational epidemiology”, in Kelsey, L., Marmot, M.G. and Stolley, P.D. (Eds), Monographs in Epidemiology and Biostatistics,Vol. 34, Oxford University Press, New York, NY, pp. 211-46. Colligan, M.J. and Rosa, R.R. (1990), “Shift work effect on social and family life”, Occupational Medicine, Vol. 5 No. 2, pp. 315-22. Dawson, D. and Reid, K. (1997), “Fatigue, alcohol and performance impairment”, Nature, Vol. 388 No. 6639, p. 235. Dumont, M., Montplaisir, J. and Infante-Rivard, C. (1997), “Sleep quality of former night-shift workers”, International Journal of Occupational and Environmental Health, Vol. 3 No. 2, pp. S10-S14. Foley, D., Monjan, A., Masaki, K., Ross, W., Havlik, R., White, L. and Launer, L. (2001), “Daytime sleepiness is associated with 3-year incident dementia and cognitive decline in older Japanese-American men”, Journal of the American Geriatrics Society, Vol. 49 No. 12, pp. 1628-32. Gangwisch, J.E., Heymsfield, S.B., Boden-Albala, B., Buijs, R.M., Kreier, F., Pickering, T.G., Rundle, A.G., Zammit, G.K. and Malaspina, D. (2006), “Short sleep duration as a risk factor for hypertension: analyses of the first National Health and Nutrition Examination Survey”, Hypertension, Vol. 47 No. 5, pp. 833-9. Garbarino, S., Nobili, L., Beelke, M., Balestra, V., Cordelli, A. and Ferillo, F. (2002), “Sleep disorders and daytime sleepiness in state police shift workers”, Archives of Environmental Health, Vol. 57 No. 2, pp. 167-73. Harma, M., Tenkanen, L., Sjoblom, T., Alikoski, T. and Heinsalmi, P. (1998), “Combined effects of shift work and life-style on the prevalence of insomnia, sleep deprivation and daytime sleepiness”, Scandinavian Journal of Work, Vol. 24 No. 4, pp. 300-7. Hui, D.S., Chan, J.K., Ko, F.W., Choy, D.K., Li, T.S., Chan, A.T., Wong, K.K. and Lai, C.K. (2002), “Prevalence of snoring and sleep-disordered breathing in a group of commercial bus drivers in Hong Kong”, Internal Medicine Journal, Vol. 32 No. 4, pp. 149-57. Klawe, J.J., Laudencka, A., Miskowiec, I. and Tafil-Klawe, M. (2005), “Occurrence of obstructive sleep apnea in a group of shift worked police officers”, Journal of Physiology and Pharmacology, Vol. 56 No. 4, pp. 115-7. Kushida, C.A., Chang, A., Gadkary, C., Guilleminault, C., Carrillo, O. and Dement, W.C. (2001), “Comparison of actigraphic, polysomnographic, and subjective assessment of sleep parameters in sleep-disordered patients”, Sleep Medicine, Vol. 2 No. 5, pp. 389-96. Lindberg, E., Taube, A., Janson, C., Gislason, T., Svardsudd, K. and Boman, G. (1998), “A 10-year follow-up of snoring in men”, Chest, Vol. 114 No. 4, pp. 1048-55. Lockley, S.W., Skene, D.J. and Arendt, J. (1999), “Comparison between subjective and actigraphic measurement of sleep and sleep rhythms”, Journal of Sleep Research, Vol. 8 No. 3, pp. 175-83. Maislin, G., Pack, A.I., Kribbs, N.B., Smith, P.L., Schwartz, A.R., Kline, L.R., Schwab, R.J. and Dinges, D.F. (1995), “A survey screen for prediction of apnea”, Sleep, Vol. 18 No. 3, pp. 158-66. Marchesini, G., Pontiroli, A., Salvioli, G., Novi, R.F., Vitacolonna, E., Taboga, C., Ciccarone, A.M. and Grossi, E. and QUOVADIS Study Group (2004), “Snoring, hypertension and Type 2 diabetes in obesity. Protection by physical activity”, Journal of Endocrinological Investigation, Vol. 27 No. 2, pp. 150-7.

Shift work and sleep

225

PIJPSM 30,2

226

Minors, D.S. and Waterhouse, J.M. (1990), “Circadian rhythms in general”, Occupational Medicine, Vol. 5 No. 2, pp. 165-82. Mongrain, V., Carrier, J. and Dumont, M. (2005), “Chronotype and sex effects on sleep architecture and quantitative sleep EEG in healthy young adults”, Sleep, Vol. 28 No. 7, pp. 819-27. Monk, T.H. (1990), “Shiftworker performance”, Occupational Medicine, Vol. 5 No. 2, pp. 183-98. Monk, T.H. (2000), “What can the chronobiologist do to help the shift worker?”, Journal of Biological Rhythms, Vol. 15 No. 2, pp. 86-94. National Sleep Foundation (2005), National Sleep Foundation Sleep in America poll, National Sleep Foundation, Washington DC. Peppard, P.E. and Young, T. (2004), “Exercise and sleep-disordered breathing: an association independent of body habitus”, Sleep, Vol. 27 No. 3, pp. 480-4. Pandi-Perumal, S.R., Srinivasan, V. and Maestroni, G.J. (2006), “Melatonin; nature’s most versatile biological signal?”, The FEBS Journal, Vol. 273 No. 13, pp. 2813-38. Portela, L.F., Rotenberg, L. and Waissmann, W. (2004), “Self-reported health and sleep complaints among nursing personnel working under 12 h night and day shifts”, Chronobiology International, Vol. 21 No. 6, pp. 859-70. Radloff, L.S. (1977), “The C.E.S.D. scale: a self-report depression scale for research in the general population”, the CESD, Vol. 1, pp. 385-401. SAS Institute (2001), SAS User’s Guide, Version 9.1, SAS Institute, Cary, NC. Sack, R.L., Blood, M.L. and Lewy, A.J. (1992), “Melatonin rhythms in night shift workers”, Sleep, Vol. 15 No. 5, pp. 434-41. Scott, A.J. and Ladou, J. (1990), “Shiftwork: effects on sleep and health with recommendations for medical surveillance and screening”, Occupational Medicine, Vol. 5 No. 2, pp. 273-99. Scott, A.J., Monk, T.H. and Brink, L.L. (1997), “Shiftwork as a risk factor for depression: a pilot study”, International Journal of Occupational and Environmental Health, Vol. 3 No. 2, pp. S2-S9. Shields, M. (2002), “Shift work and health”, Health Reports, Vol. 13 No. 4, pp. 11-31. Smith, L., Folkard, S. and Poole, C.J. (1994), “Increased injuries on the night shift”, Lancet, Vol. 344 No. 8930, pp. 1137-9. Ulfberg, J., Carter, N., Talback, M. and Edling, C. (1996a), “Headache, snoring and sleep apnea”, Journal of Neurology, Vol. 243 No. 9, pp. 621-5. Ulfberg, J., Carter, N., Talback, M. and Edling, C. (1996b), “Excessive daytime sleepiness at work and subjective work performance in the general population and among heavy snorers and patients with obstructive sleep apnea”, Chest, Vol. 110 No. 3, pp. 659-63. Ulfberg, J., Carter, N., Talback, M. and Edling, C. (2000), “Adverse health effects among women living with heavy snorers”, Health Care for Women International, Vol. 21 No. 2, pp. 81-90. Vgontzas, A.N., Zoumakis, E., Bixler, E.O., Lin, H.M., Follett, H., Kales, A. and Chrousos, G.P. (2004), “Adverse effects of modest sleep restriction on sleepiness, performance, and inflammatory cytokines”, The Journal of Clinical Endocrinology and Metabolism, Vol. 89 No. 5, pp. 2119-26. Victor, L.D. (1999), “Obstructive sleep apnea”, American Family Physician, Vol. 60 No. 8, pp. 2279-86. Vila, B. (2006), “Impact of long work hours on police officers and the communities they serve”, American Journal of Industrial Medicine, Vol. 49 No. 11, pp. 972-80.

Vila, B., Morrison, G.B. and Kenney, D.J. (2002), “Improving shift schedule and work-hour policies and practices to increase police officer health, safety and performance”, Police Quarterly, Vol. 5 No. 1, pp. 4-24. Violanti, J.M., Burchfiel, C.M., Miller, D.B., Andrew, M.E., Dorn, J., Wactawski-wende, J., Beighley, C.M., Pierino, K., Joseph, P.N., Vena, J.E., Sharp, D.S. and Trevisan, M. (2006), “The Buffalo Cardio-Metabolic Occupational Police Stress (BCOPS), pilot study: methods and participant characteristics”, Annals of Epidemiology, Vol. 16 No. 2, pp. 148-56. Williamson, A.M. and Feyer, A.M. (2000), “Moderate sleep deprivation produces impairments in cognitive and motor performance equivalent to legally prescribed levels of alcohol intoxication”, Occupational and Environmental Medicine, Vol. 57 No. 10, pp. 649-55. Yaggi, H.K., Araujo, A.B. and McKinlay, J.B. (2006), “Sleep duration as a risk factor for the development of type 2 diabetes”, Diabetes Care, Vol. 29 No. 3, pp. 657-61. Yaggi, H.K., Concato, J., Kernan, W.N., Lichtman, J.H., Brass, L.M. and Mohsenin, V. (2005), “Obstructive sleep apnea as a risk factor for stroke and death”, The New England Journal of Medicine, Vol. 353 No. 19, pp. 2034-41. Yousaf, F. and Sedgwick, P. (1996), “Sleep disorders”, British Journal of Hospital Medicine, Vol. 55 No. 6, pp. 353-8. Zou, G. (2004), “A modified Poisson regression approach to prospective studies with binary data”, American Journal of Epidemiology, Vol. 159 No. 7, pp. 702-6. Corresponding author Luenda E. Charles can be contacted at: [email protected]

To purchase reprints of this article please e-mail: [email protected] Or visit our web site for further details: www.emeraldinsight.com/reprints

Shift work and sleep

227

The current issue and full text archive of this journal is available at www.emeraldinsight.com/1363-951X.htm

PIJPSM 30,2

Suicidal ideation among police officers in Norway Ronald J. Burke York University, Toronto, Canada, and

228

Aslaug Mikkelsen Stavanger University College, Stavanger, Norway Abstract Purpose – The purpose of this study is to examine potential predictors of suicidal ideation among a large sample of Norwegian police officers. Some have suggested that suicide is a leading cause of death among police officers. Design/methodology/approach – Data were collected using anonymously completed questionnaires from 766 officers, a 60 percent response rate most measures included were commonly used by other researchers. Predictors included personal demographics, work situation characteristics, job demands, burnout components, work outcomes and coping responses. Logistic regression analysis was used as the prevalence of suicidal ideation was strongly skewed; most police officers indicated no suicidal ideation. Two criterion groups were created; police officers indicating no suicidal ideation (n ¼ 495) and police officers indicating some suicidal ideation (n ¼ 124). Findings – Single police officers, officers reporting higher levels of both exhaustion and cynicism (burnout components), and officers engaging in less active coping and reporting lower levels of social support indicated more suicidal ideation. Research limitations/implications – Use of self-report data raises the possibility of response set tendencies. Practical implications – Organizations can undertake efforts to prevent potential suicide of their members. It appears that reducing levels of burnout, increasing social support, and highlighting the benefits of active coping represent useful starting points. Originality/value – This study contributes to our understanding of suicidal ideation among police officers. Keywords Suicide, Police, Stress, Norway Paper type Research paper

Policing: An International Journal of Police Strategies & Management Vol. 30 No. 2, 2007 pp. 228-236 q Emerald Group Publishing Limited 1363-951X DOI 10.1108/13639510710753234

It is believed that suicide is a leading cause of death among police officers (Janik and Kravitz, 1994) and that policing, as an occupational group is dangerous, stressful and associated with high levels of suicide (Hill and Clawson, 1988; Violanti, 1995a, b; Mohandie and Hatcher, 1999). Methodological problems limit the usefulness of many of the studies concluding that suicide rates are higher among police officers than in the general population (Hem et al., 2001). These limitations include small sample sizes, different time periods covered, a failure to consider the type of police work being done and regional variations in the general suicide rate. In addition, it is difficult to specify an appropriate comparison group. Hem et al. (2001) conducted a systematic review of the published literature and conclude that the data on rates of police suicide compared to other Preparation of this manuscript was supported in part by the Rogaland Institute, Norway and the School of Business, York University. The authors acknowledge the support of the Police Union in conducting the study and collecting the data. Lisa Fiksenbaum assisted with data analysis.

groups are inconclusive. They advocate continued but improved research on this topic as well as systemic studies on attempted suicide and suicidal ideation among police. It is still important to better understand suicide, attempted suicide and suicidal ideation among police officers (Baker and Baker, 1996; Violanti, 1995a). As Violanti (1996) has observed, the sometimes violent nature of police work and the attitudes needed to protect officers from such violence affects their psychological makeup, firearms are readily available and the rigid nature of police organizations and the police role reduces flexibility. This study considers the role of various predictors of suicidal ideation among a large sample of police officers in Norway. Suicidal ideation is likely to be associated with suicide attempts (Weismann et al., 1999; Vandivorte and Locke, 1979). Most previous studies have been undertaken in North America and Europe (Abdollahi, 2002). It has been suggested that models of policing differ in various countries, with Norway reflecting an approach to policing that is different from the prevailing North American approach. This study examines predictors of suicidal ideation among police officers. A research model was developed based on the police suicide literature (Rudd, 1990) to both guide the selection of variables to be included in the study and inform data analysis (see Figure 1). In this model, six blocks of variables were considered as predictors of suicidal ideation among police officers. The blocks of predictors are shown in the order in which they were considered in the analysis. The first block of predictors consisted of personal demographic measures (e.g. gender, marital status, age). It has been shown that males, single officers, and older

Suicidal ideation among police in Norway 229

Figure 1. Research model

PIJPSM 30,2

230

officers are more likely to contemplate and attempt suicide. The second block of predictors included work situation characteristics (e.g. size of department, shift schedule). It has been found that officers regularly working shifts are more likely to contemplate suicide (Cantor et al., 1995). The third block of predictors included a variety of job demands (e.g. emotional, cognitive), job demands found to be an antecedent of suicidal ideation. The fourth block of predictors included three burnout components, these (e.g. exhaustion, cynicism) reflect work attitudes likely to be associated with suicidal ideation (see Schaufeli and Enzmann, 1998). The next block of predictors, work outcomes, included work responses (e.g. job satisfaction, organizational commitment) which if low would increase suicidal ideation. The last block of predictors, coping resources, included measures likely to lessen the likelihood of suicide ideation (e.g. active coping, social support). Janik and Kravitz (1994) suggest that the stress endured by police officers leads to unhealthy coping responses reflected in suicide. Method Respondents Table I shows the demographic characteristics of the police sample (n ¼ 766). Most respondents were male (84 percent), married (82 percent), had children (88 percent), held constable positions (62 percent), worked in urban areas (73 percent), worked in large departments (100 or more, 36 percent) worked between 36– 39 hours per week (86 percent), worked five or less hours of overtime per week (75 percent), held fairly long police tenure (21 years or more, 39 percent) and were born in 1960 or before (42 percent). Procedure Data were collected from 766 police officers using anonymously completed questionnaires, a 60 percent response rate: Questionnaires were mailed by the Police union and returned to an independent research institute. Some measures were translated from English to Norwegian for this study while other measures had already been translated into Norwegian by others (e.g. Maslach Burnout Inventory). Measures Personal demographic and work situation characteristics. A number of personal demographic and work situation characteristics (e.g. age, marital status, organization size, police tenure) were measured by single items typical of those used by others. . Work demands. A number of work demands were measured by scales from the Copenhagen Psychosocial Questionnaire (COPSOC) developed by Kristensen and Borg (2001). . Quantitative demands was measured by a seven-item scale (a ¼ 0.83). Respondents indicated their agreement with each item on a five point scale (5 ¼ Always, 3 ¼ Sometimes, 1 ¼ Never/hardly ever). A sample item was “Do you have to work very fast?” . Cognitive demands was assessed by an eight-item scale (a ¼ 0.79). One item was “Does your work require you to make difficult decisions?”

Gender: Male Female Year of birth: 1950 or earlier 1951-1960 1961-1970 1971 and later Education: 13 years or less 14-16 17 or more Location: Rural Urban Size of department: 1-19 20-49 50-99 100 or more Shift work: Regular Sometimes Around the clock Marital status: Single Married Partner in policing: Yes No Parental status: Children Childless Organizational level: Manager Supervisor Constable Police tenure: 10 years or less 11-20 21-30 31 years or more Hours worked: 35 or less 36-39 40 or more Overtime hours: 0-5 6 or more

N

%

640 125

83.7 16.3

93 214 257 167

12.7 29.3 35.2 22.8

135 464 159

12.8 61.2 11.0

209 557

27.3 72.7

194 168 119 272

25.8 27.3 15.8 36.1

167 266 329

72.0 35.0 43.0

134 631

17.5 82.5

95 663

12.5 87.5

602 147

80.4 19.6

20 252 451

2.8 34.9 62.4

239 220 232 67

32.0 28.8 30.4 8.8

66 628 37

9.0 85.9 5.1

402 164

75.4 24.6

Suicidal ideation among police in Norway 231

Table I. Demographic characteristics of sample

PIJPSM 30,2

.

.

.

232 .

.

.

.

Emotional demands was measured by a three-item scale (a ¼ 0.80). An item was “Does your work put you in emotionally disturbing situations?” Demands for hiding emotions was assessed by two items (a ¼ 0.59). One item was “Does your work require that you hide your feelings?” Sensorial demands was measured by a five-item scale (a ¼ 0.58). An item was “Does your work require a great deal of concentration?” Responsibility was assessed by a three-item scale (a ¼ 0.58). One item was “Could it injure other people if you make mistakes in your work?” Role clarity was measured by a four-item scale anchored by (5) to a large extent, (3) Somewhat, and (1) to a very small extent. One item was “Do you know exactly which areas are your responsibility?” Role conflict was assessed by a four-item scale (a ¼ 0.71). An item was “Are contradictory demands placed on you at work? Leadership quality was measured by an eight-item scale (a ¼ 0.94). A sample item was “To what extent would you say that your immediate superiors appreciate the staff and show consideration for the individual?”

Burnout Three burnout components were measured by the Maslach Burnout Inventory – General Survey (MBI-GS) developed by Schaufeli et al. (1996). (1) Exhaustion was measured by five items (a ¼ 0.86). One item was “I feel emotionally drained from my work”. (2) Cynicism was measured by five items (a ¼ 0.78). One item was “I have become less enthusiastic about my work”. (3) Professional efficacy was measured by six items (a ¼ 0.80). One item was “At my work, I feel confident that I am effective at getting things done”. Respondents indicated how frequently they experienced each item on seven-point frequency scale (0 ¼ never, 6 ¼ always). Work outcomes Three work outcomes were included, all from the CQPSOQ developed by Kristensen and Borg (2001). (1) Job satisfaction was measured by seven items (a ¼ 0.75). Respondents indicated how satisfied they were with each item on a four-point scale: (4) Very satisfied, (1) Highly unsatisfied. One item was “How satisfied are you with your usual take home pay? (2) Organizational commitment was measured by four items (a ¼ 0.58). An item was “Do you enjoy telling others about your place of work?” Responses were made on a five-point scale (5, to a large extent, 1 ¼ to a very small extent) (3) Sense of community was measured by three items (a ¼ 0.84). One item was “Do you feel part of a community at your place of work?” Coping resources Three coping resources were considered.

(1) Active coping was measured by a 12-item scale (a ¼ 0.40) developed by Finset et al. (2002). Response were made on a five-point agreement scale (5 ¼ completely agree, 1 ¼ Disagree completely). One item was “I make an active effort to find a solution to my problems”. (2) Social support was measured by a four-item scale (a ¼ 0.79) from the COPSOQ (Kristensen and Borg, 2001). An item was “How often do you get help and support from your colleagues?” Responses were made on a five-point scale (5 ¼ to a large extent, 1 ¼ to a very small extent). (3) Social relations was assessed by a two-item scale (a ¼ 0.58) developed by Kristensen and Borg (2001). An item was “Is it possible for you to talk to your colleagues while you are working?” Responses were made on a five-point extent scale. Suicidal ideation Suicidal ideation was assessed by a four-item measure (a ¼ 0.84) developed by Paykel et al. (1974) and used later by others (see Tyssen et al., 2001) The items were: (1) Have you ever felt that life was not worth living? (2) Have you ever wished you were dead?” (3) “Have you ever thought of taking your own life even if you would not really do it?” (4) “Have you ever reached the point where you seriously considered taking your life, or perhaps made plans how you would go about doing it?” A fifth item then asked “Have you tried to take your own life in the past year?” (yes/no). This question was not used in the analysis since so few police officers had attempted suicide (n ¼ 1 male). Results Analysis strategy Logistic Regression analysis was used as the prevalence of suicidal ideation was strongly skewed; most police officers indicated no suicidal ideation. Two criterion groups were created: police officers indicating no suicidal ideation (N ¼ 495) and police officers indicating some suicidal ideation (N ¼ 124). Predictors of suicidal ideation Table II shows the results of logistic regression analysis in which six blocks of variables served as predictors of suicidal ideation (see Figure 1). The following comments are offered in summary. First, personal demographics accounted for a significant amount of variance on suicidal ideation; single police officers indicated higher levels of suicidal ideation than did married police officers. Second, work demands showed a significant increment in explained variance suicidal ideation; police officers reporting greater emotional demands also reported more suicidal ideation. Third, burnout components accounted for a significant amount of explained variance; police officers reporting higher levels of exhaustion and police officers reporting higher levels of cynicism and exhaustion and lower levels of efficacy indicated more suicidal ideation. Fourth, coping resources accounted for a significant

Suicidal ideation among police in Norway 233

PIJPSM 30,2

234

Table II. Predictors of suicidal ideation

Suicidal ideation (N ¼ 630) Personal demographics Marital status ( 2 0.14) Work situation characteristics Job demands Emotions Burnout Cynicism Exhaustion Efficacy Work outcomes Coping and social resources Active coping Percentage correct 0 Non0 0= Non0 =

R2

P

0.04 0.08

0.01 NS

0.19

0.001

0.28 0.29

0.001 NS

95.2 35.5 83.2 495 124

increment in explained variance on suicidal ideation; police officers utilizing more active coping indicated less suicidal ideation. Fifth, work situation characteristics, and work outcomes failed to account for significant increments in explained variance. Sixth, burnout components accounted for the largest increment in explained variance. Seventh, all predictor variables considered accounted for moderate amounts of explained variance in suicidal ideation (R 2 ¼ 0.10) suggesting the existence of other important but unmeasured variables. Discussion Our findings are consistent in some ways with previous conclusions and suggest useful diagnostic possibilities for possibly reducing police suicide. Previous research has shown that single police officers indicate higher levels of suicidal ideation, that burnout components are similarly associated with greater suicidal ideation and that coping responses are associated with less suicidal ideation. Somewhat surprisingly, few job demands were related to levels of suicidal ideation. It was expected that police officers reporting greater job demands would indicate higher levels of suicidal ideation. It appears that job-related feelings as reflected in the burnout components are a better indication of the experience of work as a police officer than are levels of job demands. The pattern of findings suggest three practical implications for police management. First, monitoring levels of burnout would serve as a potential early-warning signal. Second, undertaking efforts to increase levels of social support in the workplace may also have dividends (Loo, 1999, 2001). Third, including information on active coping in police officer training and development would also be potentially useful. Monitoring levels of exhaustion and cynicism through both informal observations serves as a useful early warning device. In addition peer support has proved useful in prevention. Presenting material on violence turned inward in orientation and training

programs for beginning and more senior officers would make the departmental community more responsible for early detection and prevention. Employee assistance programs have been found to be helpful in non-police settings. Police officers are often reluctant to use these services for fear that if found that this would reflect badly on them. In addition providing assistance to individual officers following a traumatic event or providing help to departments following the suicide of an officer through interdepartmental communications and debriefing has been advocated (Loo, 2001). Special attention also needs to be devoted to police officers recommended for or already receiving psychiatric treatment (Loo, 1986). What other factors might be included to more completely predict levels of suicidal ideation? Several possibilities exist including an inventory of traumatic work events (e.g. police shooting), the inclusion of non-work, of a traumatic nature (e.g. divorce) measures of stable personality characteristics (e.g. neuroticism) and indicators of psychological health (e.g. depression). Limitations of the study Some limitations of the study should be acknowledged to put the results in a larger context. First, all data were collected using self-reports raising the possibility of common method variance. Second, some of the measures had internal consistency reliabilities below the generally accepted level of 0.70. Third, it is not clear the extent to which these findings generalize to police officers in other countries. References Abdollahi, M.K. (2002), “Understanding police stress research”, Journal of Forensic Psychology Practice, Vol. 2, pp. 1-24. Baker, T.E. and Baker, J.P. (1996), “Presenting police suicide”, FBI Law Enforcement Bulletin, Vol. 65, pp. 24-7. Cantor, H.C., Tyman, R. and Slater, P.J. (1995), “A historical survey of police suicide in Queensland, Australia, 1843-1992”, Suicide and Life-Threatening Behavior, Vol. 25, pp. 499-507. Finset, A., Steine, S., Haugli, L., Steen, E. and Laerum, E. (2002), “The brief approach/avoidance coping questionnaire: development and validation”, Psychology, Health & Medicine, Vol. 7, pp. 75-85. Hem, E., Berg, A.M. and Ekeberg, O. (2001), “Suicide in police – a critical review”, Suicide and Life-Threatening Behavior, Vol. 31, pp. 224-33. Hill, K.Q. and Clawson, M. (1988), “The health hazards of ‘street level’ bureaucracy: mortality among the police”, Journal of Police Science and Administration, Vol. 16, pp. 243-8. Janik, J. and Kravitz, H.M. (1994), “Linking work and domestic problems with police suicide”, Suicide and Life-Threatening Behavior, Vol. 24, pp. 267-74. Kristensen, T.S. and Borg, V. (2001), The Copenhagen Psychosocial Questionnaire (COPSOQ), National Institute of Occupational Health, Copenhagen. Loo, R. (1986), “Suicide among police in a federal force”, Suicide and Life-threatening Behavior, Vol. 16, pp. 379-88. Loo, R. (1999), “Police suicide: the ultimate stress reaction”, in Violanti, J.M. and Patton, D. (Eds), Police Trauma: Psychological Aftermath of Civilian Combat, C.C. Thomas, Sprinfield, IL, pp. 241-84.

Suicidal ideation among police in Norway 235

PIJPSM 30,2

236

Loo, R. (2001), “Effective postvention for police suicide”, Australian Journal of Disaster and Trauma Studies, Vol. 5, pp. 132-9. Mohandie, K. and Hatcher, C. (1999), “Suicide and violence risk in law enforcement. Practical guidelines for risk assessment, presentation, and intervention”, Behavioral Sciences and the Law, Vol. 17, pp. 357-76. Paykel, E.S., Myers, J.K., Lindenthal, J.J. and Tanner, J. (1974), “Suicidal feelings in the general population: a prevalence study”, British Journal of Psychiatry, Vol. 124, pp. 460-9. Rudd, M.D. (1990), “An integrative model of suicidal ideation”, Suicide and Life Threatening Behavior, Vol. 24, pp. 174-83. Schaufeli, W.B. and Enzmann, D. (1998), The Burnout Companion to Study and Practice: A Critical Analysis, Taylor & Francis, London. Schaufeli, W.B., Leiter, M.P., Maslach, C. and Jackson, S.E. (1996), The Maslach Burnout Inventory – General Survey Test Manual, Consulting Psychologists Press, Palo Alto, CA. Tyssen, R., Vaglum, P., Gronvold, N.T. and Ekeberg, O. (2001), “Suicidal ideation among medical students and young physicians: a nationwide and prospective study of prevalence and predictors”, Journal of Affective Disorders, Vol. 64, pp. 69-79. Vandivort, D.S. and Locke, B.Z. (1979), “Suicide ideation: its relation to depression, suicide and suicide attempt”, Suicide and Life Threatening Behavior, Vol. 9, pp. 205-18. Violanti, J.M. (1995a), “The mystery within understanding police suicide”, FBI Law Enforcement Bulletin, Vol. 4, pp. 19-23. Violanti, J.M. (1995b), “Trends in police suicide”, Psychological Reports, Vol. 77, pp. 688-90. Violanti, J.M. (1996), Police Suicide: Epidemic in Blue, C.C. Thomas, Springield, IL. Weissman, M.M., Bland, R.C., Canino, G.J., Greenwold, S., Hwu, H.G., Joyce, P.R., Karam, E.G., Lee, C.K., Lellouch, J., Lepine, J.P., Newman, S.C., Rubio-Stipec, M., Wells, J.E., Wichramaratne, P.J., Wittchen, H.U. and Yeh, E.K. (1999), “Prevalence of suicide ideation and suicide attempts in nine countries”, Psychology and Medicine, Vol. 29, pp. 9-17.

To purchase reprints of this article please e-mail: [email protected] Or visit our web site for further details: www.emeraldinsight.com/reprints

The current issue and full text archive of this journal is available at www.emeraldinsight.com/1363-951X.htm

The impact of COMPSTAT on reported crime in Queensland

The impact of COMPSTAT

Lorraine Mazerolle and Sacha Rombouts School of Criminology and Criminal Justice, Griffith University, Mt Gravatt, Australia, and

James McBroom School of Environmental Studies, Griffith University, Mt Gravatt, Australia

237 Received 9 August 2006 Revised 25 October 2006 Accepted 6 December 2006

Abstract Purpose – The purpose of this paper is to evaluate the impact of Queensland Police Service’s version of COMPSTAT, known as “Operational Performance Reviews” (OPRs), on reported crime. Design/methodology/approach – The paper employed interrupted time series analytic techniques to examine the impact of OPRs on various categories of reported crime in Queensland. The analyses assessed the extent to which OPRs were associated with crime reductions across the 29 police districts in Queensland. Findings – The introduction of OPRs was found to be associated with a significant decrease in the total number of reported offences in Queensland. OPRs were found to have their strongest effect on reported unlawful entries into dwellings and unlawful entries into other properties. The overall finding was that OPRs appeared to be a cost-effective approach to controlling crime, resulting in an overall saving of AUD$1,162,175. Practical implications – The current study suggests that COMPSTAT generally – and Queensland’s OPRs in particular – can be an effective police management mechanism, which results in observable reductions in reported crimes. Moreover, OPRs appear to be most effective at reducing property-related crime. Modifications of this performance management process may help to generalize its effects to other types of crime. Originality/value – This study was the first to provide a systematic examination of the impact of OPRs on a relatively large number of offence categories. Keywords Police, Performance management systems, Crimes, Performance management, Time series analysis, Australia Paper type Research paper

This project was supported by grant number 22/03-04 from the Criminology Research Council. The authors wish to acknowledge support from Toni Makkai of the Australian Institute of Criminology as well as research assistance from David Soole and Pu Lin at various times throughout this project. They are also deeply indebted to the members of the Queensland Police Service for their support throughout this project. In particular, the authors are appreciative of the cooperation provided to us by members of the Operational Performance Review Unit, under the leadership of Superintendent Peter Savage, and personnel in the Information Resources Centre for providing the data, particularly Lucia La Spina and Peter Conroy. Many thanks also to Dr Ann Scott for working with us to get the project off-the-ground. The authors are particularly appreciative to the Commissioner of Police, Bob Atkinson for his openness and support throughout our research efforts to evaluate the impact of Queensland’s version of COMPSTAT called OPRs in Queensland. The views expressed are the responsibility of the authors and are not necessarily those of the Criminology Research Council. Address all correspondence to Professor Lorraine Mazerolle, School of Criminology and Criminal Justice, Griffith University, Mt Gravatt campus, Nathan, QLD 4111.

Policing: An International Journal of Police Strategies & Management Vol. 30 No. 2, 2007 pp. 237-256 q Emerald Group Publishing Limited 1363-951X DOI 10.1108/13639510710753243

PIJPSM 30,2

238

Introduction In 1994 the New York City Police Commissioner, William Bratton, implemented a new, strategic approach to crime control and order maintenance that he coined “COMPSTAT”[1]. COMPSTAT is a management strategy that is designed to reduce, prevent and control crime. It involves executive police officers (Commissioners, Chiefs, Deputy Commissioners) meeting with precinct or district commanders in a high-pressure forum on a regular basis (see Henry, 2002; McDonald, 2002; Silverman, 1999). Overall, COMPSTAT can be viewed as a four-step process including accurate and timely intelligence, rapid deployment, effective tactics, and follow-up and assessment (Walsh and Vito, 2004). During COMPSTAT meetings the participants review district crime statistics, discuss emerging crime problems, identify strategic approaches for controlling crime problems, and discuss progress made during the previous period. Police Executives then set, in consultation with the District Commanders, a series of goals and objectives that must be achieved prior to the next COMPSTAT meeting. In 1996 COMPSTAT won a Harvard University “Innovations in Government Award”. Many proponents have claimed that COMPSTAT made an important contribution to the significant crime drop in New York City during the 1990s (see Bratton, 1997, 1998; Dodenhoff, 1996; Safir, 1998). Popular literature and common opinion among many in the police community is that COMPSTAT can be credited with impressive reductions in crime and improvements in neighbourhood quality of life (Bouza, 1997; Gurwitt, 1998; Kelling and Sousa, 2001; Maple, 1999; Remnick, 1997; Witkin, 1998). Indeed, Silverman and O’Connell (1997) consider COMPSTAT to be the “linchpin strategy” that binds together other policing tactics such as zero tolerance, problem-oriented policing, order maintenance policing and police efforts that seek to reduce crime and improve quality of life. Others are not so convinced, arguing that COMPSTAT was not the primary factor leading to reduced crime (e.g. Brereton, 1999; Dixon, 1998; Eck and Maguire, 2000) failing to deliver real, innovative reform (see Weisburd et al., 2006). Despite the arguments of COMPSTAT effectiveness, the idea of COMPSTAT captured the imagination of police departments throughout the world. Since the first appearance of COMPSTAT in the USA during the mid-1990s, police department executives, police ministers and police agency representatives from Australia, Canada and the UK have travelled to New York City to review the COMPSTAT approach. Australian police representatives brought the idea of COMPSTAT back to Australia, and made some changes to suit our local conditions (see ABC Lateline, 4 June 1998; Brereton, 1999). The Australian COMPSTAT versions are variously referred to as Operational Performance Reviews (Queensland), Operations Crime Reviews (New South Wales), Corporate Management Group Performance Reviews (Tasmania), Organisational Performance Reviews (Western Australia), COMPSTAT (Victoria) and Performance Outcome Reviews (South Australia). Maas (1998) suggests that this wide diffusion of COMPSTAT across democratic countries in recent years is testament to the faith that police put in the COMPSTAT process for reducing crime problems. The claim, however, that COMPSTAT and Australian versions of COMPSTAT can reduce crime remains conjecture. Some critics of COMPSTAT doubt the likelihood that COMPSTAT was in fact the “New York Miracle” that many have claimed (see Brereton, 1999; Dixon, 1998; Grabosky, 1999).

Weisburd et al. (2003), 2006) suggest that COMPSTAT should be interpreted more as an effort to maintain and reinforce the “bureaucratic” or “paramilitary” model of police organization (that had been under attack by scholars for most of the last two decades) than as a catalyst to crime reduction or as a vehicle to reform models of policing (see also Magers, 2004). It is clear that the lack of empirical research that examines the contribution of COMPSTAT to crime reduction confounds these debates (but see Chilvers and Weatherburn, 2004; Eck and Maguire, 2000 as important exceptions). Thus, one important, yet unanswered, question is how much credit (if any) can COMPSTAT (and Australian versions of COMPSTAT) be given for declines in crime in cities where it has been implemented? Our paper evaluates the impact of Queensland Police Service’s version of COMPSTAT known as “Operational Performance Reviews” (OPRs). Our study examines the impact of OPRs on reported crime in Queensland. We begin this paper with a brief synopsis of the background literature that informs our research. In the second part of our paper, we describe our data and data extraction and aggregation methods. The third section examines the statewide impact of introducing OPRs in Queensland in 2001. We use ARIMA interrupted time series analysis to assess and isolate the direct impact OPRs had on different categories of crime across the state. We conclude with a discussion of the policy implications. Background literature In the late 1960s and 1970s many police departments in democratic societies were heavily committed to the professional or “reform model” of policing (see Kelling and Moore, 1988). The police drew their legitimacy and authority through the law. By “professionalising” their management practices, they emphasised crime control as the central function of police, they centralised their organisational structures and emphasised crime investigative, rapid response and preventive patrolling activities. By the mid-1980s, however, every major police strategy to prevent or control crime had been “unmasked” by scientific research (see Bayley, 1994; Weisburd and Eck, 2004) and scholars began to challenge the fundamental premise of whether the police could have a significant impact on crime (see Gottfredson and Hirshi, 1990; Greenwood et al., 1978; Levine, 1975). In short, while the police had long considered their core role as effective “crime fighters,” the scientific evidence seemed to suggest otherwise. Despite the reluctance of the police to admit that they were largely ineffective during the 1970s and 1980s in reducing crime, it is clear that policing has undergone significant change during the 1990s and now into the twenty-first century. Police departments around the world have changed their emphasis from almost exclusive focus on reacting to crimes after they have been committed to being more effective in controlling crime and embracing crime prevention and problem-oriented policing as central to their mission. In this transformation, the police have become more consultative with community members and stakeholders and have re-engineered their police organisations to become more operationally accountable (see McDonald, 2002) and they have adopted a variety of new approaches to policing (Bayley, 1994; Eck and Spelman, 1987; Goldstein, 1990; Kelling and Moore, 1988; Mazerolle and Ransley, 2006; Skogan and Hartnett, 1997).

The impact of COMPSTAT

239

PIJPSM 30,2

240

There is now a growing body of scientific evidence to suggest that the police can indeed be effective at reducing crime problems (see Bayley, 1998, Eck and Maguire, 2000; Mazerolle et al., 2007; Sherman et al., 1997; Weisburd and Eck, 2004). One of the major trends in police management that has swept across democratic societies since the mid-1990s has been the adoption of COMPSTAT and COMPSTAT-like programs. COMPSTAT originated in New York City in 1994 under then Police Commissioner William Bratton. At the core of the approach are four crime reduction principles: (1) accurate and timely intelligence about crime made available at all levels in the organization; (2) selection of the most effective tactics for specific problems; (3) rapid deployment of people and resources to implement those tactics; and (4) “relentless” follow-up and assessment to learn what happened and make subsequent tactical adjustments as necessary (Bratton, 1998; Maple, 1999; McDonald, 2002; Silverman, 1999). David Weisburd and his colleagues from the Police Foundation conducted a survey across a stratified sample of American police agencies with municipal policing responsibilities to gauge the uptake and modifications made to the COMPSTAT approach (Weisburd et al., 2001). A total of six key elements emerged as central features of COMPSTAT programs across the USA: mission clarification; internal accountability; geographic organization of command; organizational flexibility; data driven problem identification and assessment; and innovative problem solving. The innovation of COMPSTAT, we suggest, lies more in the convergence of these six elements rather than implementation of any one (or more) of these elements in isolation. The widespread popularity and appeal of COMPSTAT is perhaps unprecedented in recent police history. Mark Moore (1997, p. 67) states: [. . .] Commissioner Bratton’s bold statement – reacceptance of responsibility for controlling crime – was a very important moment in leadership of the criminal justice system.

Journalists such as Dodenhoff (1996), Witkin (1998) and Remnick (1997) seem convinced that COMPSTAT led to reductions in crime in New York City; police leaders such as Bill Bratton as the “author” of COMPSTAT (Bratton, 1997, 1998), Howard Safir (1998), and Tony Bouza (1997) hail COMPSTAT as the answer to contemporary crime problems; Bratton’s right-hand man and arguably the person who drove most of COMPSTAT in NYC in the early days, Jack Maple (1999), was certain COMPSTAT was the answer to rising crime problems in NYC; and other New York City based authors such as Vince Henry (2002), Phyllis McDonald (2002), Eli Silverman (1999) as well as Bratton’s long term confidant George Kelling (see Kelling and Sousa, 2001) argue strongly that COMPSTAT led to reductions in crime not just in New York City but also in other cities around the world. Others are not so convinced. John Eck and Ed Maguire (2000, pp. 234-235) carefully examined New York City’s homicide data and concluded that: [. . .] the implementation of COMPSTAT in New York in 1994 cannot be credited independently with the decline in homicides in that city. . .[and] that other changes in New York City’s policing practices implemented around the same time as Compstat (e.g zero tolerance policing) cannot [also] not be given credit for the decline. [Moreover] the diffusion of

[ . . . ] the COMPSTAT process to other cities throughout the United States came too late to have produced the national decline in homicides.

Similarly, Rosenfeld et al. (2005) recently concluded that evidence could not be found to support an impact of COMPSTAT on homicide trends. In Australia, David Dixon (1998) and David Brereton (1999) are equally sceptical of the ability of COMPSTAT to reduce crime. Dixon (1998) points to the relationship between the decline in the crack cocaine epidemic and the falling murder rate. Brereton (1999), p. 8) examines both violent crime and burglaries across five major cities in the USA and similarly concludes that there is “little evidence to support claims for New York exceptionalism”. Australian researchers, Chilvers and Weatherburn (2004), recently evaluated the impact of “Operation and Crime Review (OCR) Panels”: New South Wales’ version of COMPSTAT. Employing time series analysis, they found a significant reduction in robberies, break-and-enters and motor vehicle thefts attributable to the introduction of OCR panels. The OCRs had the strongest effect on break-and-enter offences (Chilvers and Weatherburn, 2004). Against this backdrop of debate surrounding the effectiveness of COMPSTAT to reduce crime, our evaluation of OPRs in Queensland sought to measure the impact of OPRs on reported crime across the state of Queensland. We specifically examine the impact of OPRs on the five priority areas identified by the State of Queensland’s Commissioner of Police. Our central research questions are: What impact (if any) did the introduction of OPRs in Queensland have on reported crime? What impact (if any) did the introduction of OPRs have on specific categories of crime, particularly those categories of crime identified as priority issues from the outset (i.e. public safety including reducing offences against the person, domestic violence and traffic safety offences, reducing property crime including unlawful entry offences and unlawful use of motor vehicles? Research site The State of Queensland in Australia covers 666,876 square miles and has a population of just over 3.5 million. Queensland is serviced by just one state police department (Queensland Police Service) that has nearly 9,000 sworn officers. Queensland’s version of COMPSTAT, known as Operational Performance Reviews (or OPRs for short), encapsulate many of the key elements of COMPSTAT. Chief Commissioner Atkinson initiated the Operational Performance Review (OPR) process in early 2001 and sought to closely integrate the Queensland Police Service (QPS) version of problem-oriented policing that they call Problem-Oriented and Partnership Policing (POPP) within the OPR process. The Commissioner articulated that the OPRs should focus (at least in the initial implementation period[2]) on five priorities. These included: public safety (reducing offences against the person, domestic violence and traffic safety offences), reducing property crime (including unlawful entry offences and unlawful use of motor vehicles), reducing and better handling of calls for service; better handling of major planned and unplanned events; and gaining a better understanding of unique district issues. These unique district issues could range from drug problems, to drinking in public issues to whatever crime and quality of life problems that seemed to plague each individual district. OPR forums are conducted at the Police District level of aggregation. OPR meetings in Queensland are chaired by Commissioner Atkinson and assisted by both Deputy

The impact of COMPSTAT

241

PIJPSM 30,2

242

Chief Executives (Operations and Resource Management). Assistant Commissioners from the Operations Support Command and State Operations Support Command also attend the OPR forums. District Officers attend the OPR meetings accompanied by their respective regional Assistant Commissioner, the Chief Superintendent from the region, as well as expert support personnel (such as regional intelligence coordinators, regional crime coordinators and traffic coordinators). The meetings run for about two hours and all presentations of data are made in PowerPoint format. In short, OPRs in Queensland captured the six key elements that define COMPSTAT programs across the USA (mission clarification, internal accountability, geographic organization of command, organizational flexibility, data driven problem identification and assessment, and innovative problem solving). Queensland Police Service (QPS) was already geographically organized into districts (that served as the operational level of implementation for OPRs). But OPRs clarified the key goals for the QPS, brought about unprecedented district-level accountability, enabled districts to trial new organizational arrangements (such as shift rostering and personnel allocations), institutionalized the importance of what QPS calls Problem and Partnership Oriented Policing (PPOP), and elevated the role of crime analysts within the organization. Data Our research uses monthly reported offence data for the State of Queensland in Australia[3]. The data provided by the Queensland Police Service (QPS) were gathered in the form of monthly counts of offences (“Crisp Codes”) for each Police Division (N ¼ 295) in Queensland from July 1995 to June 2004. In total, there were 108 observation points: 73 pre-OPR monthly counts and 35 post-OPR monthly counts. The data for each of the 295 police divisions were assigned to the appropriate police district (N ¼ 29). We use the State of Queensland as our unit of inquiry. While OPRs were introduced to different police regions at different points in time, the date chosen for the statewide analysis (August, 2001) was the earliest date of implementation. We use the earliest date as our point of “interruption” because all District Commanders were informed of the innovation at this time. We expect that the communication and extensive consultation about the introduction of the OPRs would have created what is known as an “announcement effect” (Smith et al., 2002) such that District Commanders would have altered their management and operational strategies at this time. Once the data were obtained from the QPS the 60 crisp codes for different offences were collapsed to create 13 different crime types. These offence categories were checked with QPS researchers to ensure their conceptual validity. A subset of the data for different offence types and time periods was independently checked to assess the accuracy of the police department data extraction and their aggregation processes. Analytic approach Our state-wide impact evaluation of OPRs used a quasi-experimental design to assess the magnitude and direction of the intervention on reported crime incidents. Time series analysis is considered to be a strong quasi-experimental design (Cook and Campbell, 1979) and is a popular method used to examine the impact of legal interventions (see for example, White et al., 2003).

The aim of the time series approach was to isolate and evaluate the direct impact of the implementation of OPRs on reported offences in Queensland. To accomplish this task, SPSS ARIMA interrupted time series analysis was used to analyse the effect of OPRs on reported offences over time. This consisted of two main stages: (1) model-building to identify and control for the patterns occurring in the frequency of offences[4]; and (2) impact assessment to determine whether changes in reported offences were associated with the implementation of OPRs (see Mazerolle et al., 2007 for details of the analytic approach). Time series forecasting was used to provide a graphical display of the effectiveness of OPRs. Once a model for a particular crime type had been developed based on the pre-OPR time series, this pre-OPR model was used to obtain predicted values corresponding to the post-OPR time period. When plotted against the actual reported offences in the post-intervention period it enabled a visual comparison of “what would have likely happened” versus “what actually happened” in terms of reported offences[5]. As a scoping exercise, the monetary “savings” of reduced crime due to OPRs were calculated to contextualise the results in terms of their practical significance. Mayhew (2003) provides estimates of the average cost of an individual type of offence in Australia, taking into account both tangible and intangible costs. These figures were adapted for the current study to derive an estimate of the total cost savings (in Australian dollars) in crime occurring the post-OPR period. In order to calculate the “savings” attributable to OPRs the costs for an average crime incident were multiplied by the OPR coefficient produced in the time series analysis. “Savings” were produced for eight individual crime types for which costs were available. Overall cost-effectiveness of OPRs was evaluated by deriving the ratio of the cost of OPRs to the costs of various crimes[6]. Results Total reported offences Our most important analysis focused on the impact of OPRs on the total amount of reported offences across the State of Queensland. Inspection of the pre-OPR time series observations revealed an increasing linear trend as well as a seasonal component with peaks at around January to April (summer) of each year. This suggested that the pre-OPR time series observations required both regular and seasonal differencing. Examination of the ACF/PACF pattern after regular and seasonal differencing revealed negative ACF spikes at lag (1) and lag (12) indicating both regular and seasonal moving average components. There was also a large positive ACF spike at lag (11). The PACF plot revealed a higher-order moving average component (q ¼ 2) due to the largest spike occurring at lag (2). An ARIMA (0, 1, 2) (0, 1, 1)12 model for pre-OPR time series observations was specified and all components of this model were significant. The ACF/PACF plot of the residuals of this model revealed that the residuals appeared to approximate “white noise”. None of the residuals were significant nor exceeded the standard error limits[7]. Once the pre-OPR time series observations had been adequately modelled, the intervention component was then added to the model. All components of this model

The impact of COMPSTAT

243

PIJPSM 30,2

244

were significant. Inspection of the residuals for this model revealed that all fell approximately within two standard error limits and were non-significant. The residuals were plotted against time and again revealed the spike at July 1997. Table I presents the parameter estimates for total reported offences. As Table I shows, once the series was regularly and seasonally differenced and regular and seasonal moving average components were modelled adequately, the introduction of OPRs was found to be associated with a significant decrease in the total number of reported offences in Queensland. This is an important policy finding. What our time series results show is that when reported crime data (1995 to 2004) for the entire State of Queensland are examined and all seasonal, trend, random “noise” (i.e. ad hoc changes in police practice, application of laws, in and out-migration patterns, statewide, ad-hoc crisis etc.) are factored into the model, then the OPRs introduced by the police department in early 2001 led to a direct and statistically significant decline in crime. To visualise the amount of crime reduced directly by the OPRs, we used time series forecasting techniques to plot the predicted crime rate assuming OPRs had not been introduced against the actual crime rate post OPR implementation. Our time series forecasting models were used to obtain predicted values for the post-OPR time series observations based upon the pre-OPR ARIMA model. These values were then plotted against the observed reported offences for the post-OPR time series observations (see Figure 1). Note that the solid line represents the predicted values of total reported offences estimated from the pre-OPR time series while the dashed line represents the actual total reported offences after implementation of OPRs. As Figure 1 shows, had OPRs not been introduced across the State of Queensland in early 2001, we would have predicted a steady, seasonally-based increase in crime. Indeed, by June 2004, the actual number of reported offences was 8,495. Without the introduction of OPRs, we could have expected to see the number of reported offences to be around 11,700, ceteris paribus. This is a saving of about 3,200 crimes that can be attributable to the introduction of OPRs in Queensland. Parameter Pre-intervention: MA1 MA2 SMA1 Post-intervention: MA1 MA2 SMA1 OPR

Table I. Parameter estimates for total reported offences

Model-fitting information: Akaike’s information criterion Schwarz’s Bayesian criterion Likelihood ratio test Residual variance Standard error

Estimate (SE)

T-ratio

P-value

0.53 0.38 0.84

(0.15) (0.14) (0.39)

3.56 2.68 2.13

0.00 0.01 0.04

0.40 0.26 0.93 2972.32

(0.10) (0.10) (0.36) (278.86)

4.01 2.65 2.57 23.49

0.00 0.01 0.01 0.00 1,397.83 1,408.04 2694.91 109,222.58 330.49

The impact of COMPSTAT

245

Figure 1. Forecasted versus actual total reported offences in Queensland

The question of whether OPRs are a cost-effective intervention may be addressed by weighing the costs of OPRs against the savings attributable to reduced crime (see Mayhew, 2003). The cost of OPRs for 2004/2005 included salaries amounting to $476,500 and travel expenses ($40,000) while the equipment costs (e.g. computers, projectors) amount to around $62,000. This brings the total cost of OPRs since their implementation to approximately $1,611,500. When we weighed these operational costs of OPRs against the savings of reduced crime ($2,773,675), the introduction of OPRs appears to have been cost-effective, resulting in an overall saving of $1,162,175. In the following sections, we build a story as to what types of offences have largely contributed to the decline in crime attributable to the introduction of the OPRs in Queensland. We focus on 13 offences that represent the primary offences that the Police Department identified as priority offences. Table II summarizes the results of the ARIMA models. Dangerous driving offences The pre-OPR time series observations for dangerous driving offences did not exhibit a linear or seasonal trend while the ACF/PACF pattern revealed negative spikes at ACF lag (1) and lag (13). This was indicative of an autoregressive component. The pre-OPR observations were best described by an AR(2) model as all residuals were non-significant and within standard error limits. Once the OPR was added to the model, a third-order autoregressive component was needed to make the residuals

PIJPSM 30,2

246

Table II. ARIMA time series models for 13 crime types

Crime type

Pre-OPR model

Homicide 2 related offences Dangerous driving Unlawful use of motor vehicles Serious assaults Common assaults Sexual offences Armed robbery Unarmed robbery Ill treatment of children Menacing person offences Unlawful entry into dwellings Unlawful entry into other premises Breach DV orders

(0,0,0) (2,0,0) (2,2,0)(1,1,0) (0,1,1)(0,1,1) (0,1,1)(0,1,1) (2,1,0)(1,1,0) (2,1,0) (2,0,0) (2,1,0) (3,1,0)(1,1,0) (0,1,1)(1,1,0) (1,1,0) (0,1,1)(0,1,0)

12 12 12 12

12 12 12

Post-OPR model

OPR coefficient

(0,0,0) (3,0,0) (0,2,1)(1,1,0) 12 (0,1,1)(1,1,0) 12 (0,1,1)(0,1,1) 12 (2,1,0)(1,1,0) 12 (2,1,0) (2,0,0) Log(2,1,0) (3,1,0)(1,1,0) 12 (0,1,2)(1,1,0) 12 (2,1,0) (0,1,1)(1,1,0) 12

20.286 3.04 228.56 234.11 235.98 2117.84 210.97 28.31 0.25 32.04 2435.6 * * 2274 * 8.81

Notes: p , 0.05; * * p , 0.01

non-significant. Intervention analysis using an ARIMA (3,1,0) model revealed that OPRs did not have a significant impact on reported dangerous driving offences. Unlawful use of motor vehicle offences Inspection of the pre-OPR time series observations revealed the presence of a quadratic trend with reported offences initially decreasing and then increasing from 1999 onwards. There was also evidence of seasonality with peaks in January to April of each year. Inspection of the ACF/PACF pattern after regular (d ¼ 2) and seasonal differencing displayed both regular and seasonal autoregressive processes. An ARIMA (2,2,0)(1,1,0)12 was specified and the residuals of this model were all non-significant and within their standard error limits. The introduction of the post-OPR observations required that the model be changed to a moving average process, resulting in an ARIMA (0,2,1)(1,1,0) 12 model specification. While the residuals of this model approximated “white noise”, there appeared to be non-stationarity of variance, which no transformation could remedy. Overall, the results suggested that OPRs were associated with a non-significant decrease in unlawful use of motor vehicle offences. Nonetheless, a reduction of 28.56 motor vehicle offences translates into a “cost saving” of about $186,675. Serious assaults The pre-OPR time series observations for serious assaults revealed an increasing linear trend and seasonality in the form of peaks in October to January of each year. After regular and seasonal differencing, the ACF/PACF pattern revealed large negative spikes at ACF lag (1) and lag (12), indicating both regular and seasonal moving average components. After model estimation, both components were significant and all but one residual (lag (14) did not exceed standard error limits. Once the OPR was added to the model, the seasonal moving average component was non-significant and close to the bounds of invertibility. A seasonal autoregressive component was instead added to the model. All residuals were non-significant except for lag (24). A higher order seasonal autoregressive component (AR2) was therefore added to the model; however, this produced little change so the most parsimonious model was retained – an ARIMA

(0,1,1)(1,1,0)12 model. The results showed that serious assaults exhibited a non-significant decrease associated with the introduction of OPRs. Common assaults The pre-OPR time series observations for reported common assaults displayed a linear trend as well as seasonal variation (peaks in November to March of each year). There also appeared to be an unusual drop in around May, 1997. After regular and seasonal differencing, the ACF/PACF pattern revealed negative spikes at ACF lag (1) and lag (12) suggesting regular and seasonal moving average components. These components emerged as significant during parameter estimation and all residuals of the model were non-significant and within standard error limits. These same components remained significant with the introduction of OPRs into the ARIMA (0,1,1)(0,1,1)12 model and all residuals remained non-significant. The results demonstrated a non-significant decrease of common assaults associated with the implementation of OPRs. While both serious and common assaults did not appear to be significantly affected by OPRs, the “savings” in non-significant decreases of assaults amounted to $120,600. Sexual offences Inspection of the raw time series revealed a slightly increasing linear trend and a seasonal pattern with peaks in October of each year. After regular and seasonal differencing, the ACF/PACF pattern revealed a negative spike at lag (1) and the largest PACF negative spike at lag (2), suggesting an autoregressive process. An ARIMA (2,1,0)(1,1,0)12 model provided the best fit to the pre-OPR time series data with the residuals being non-significant and within their standard error limits. The results demonstrate that OPRs were associated with a non-significant decrease in reported sexual offences. From a practical perspective, however, this non-significant reduction resulted in a “saving” of $295,000. Armed robbery ARIMA intervention analyses were conducted separately on armed and unarmed robbery offences reported in Queensland. The pre-OPR time series displayed a linear trend and a possible lower outlier (December, 1996). After regular differencing, the ACF/PACF pattern revealed a negative spike at ACF lag (1) with the largest PACF spike at lag (2). A second-order autoregressive component was therefore specified. The estimation of the pre-OPR model found both autoregressive components to be significant, however, the residuals of this model revealed a pattern of increasing variance over time (i.e. variance non-stationarity). The outlier was therefore replaced with the mean of the preceding and following month. An ARIMA (2,1,0) model was fit to the pre-OPR time series and the residuals of this model approximated “white noise”. However, the residuals still appeared to display non-stationarity of variance due primarily to a lower outlier in the post-OPR time series observations (September 2003). The outlier was replaced and the resulting residuals displayed stationarity of variance. When the OPR was added to this model, the AR1 and AR2 components remained significant, however, the OPR coefficient was non-significant. The residuals of this model approximated white noise while the results demonstrated a non-significant decrease of armed robberies associated with implementation of OPRs.

The impact of COMPSTAT

247

PIJPSM 30,2

248

Unarmed robbery The pre-OPR time series observations revealed a slight decreasing linear trend that did not require differencing. The ACF/PACF pattern showed a positive spike at ACF lag(1), suggesting an autoregressive component. An ARIMA (2,0,0) model was the best fit to the pre-OPR data with the residuals being non-significant and within their standard error limits. This model was also found to be an appropriate fit when the OPR was added. The results demonstrated that the introduction of OPRs were associated with a non-significant decrease in unarmed robberies. Overall, the decreases in both armed and unarmed robberies resulted in a “saving” of $68,400.

Ill treatment of children offences The pre-OPR time series observations revealed a slight increasing linear trend and a possible outlier (September, 1996). The ACF/PACF pattern revealed negative spikes at ACF lag(1) and lag(13) with the largest PACF spike at lag(2), suggesting a higher-order autoregressive component. An ARIMA (2,1,0) model appeared to be the best fit for the pre-OPR time series observations, with all residuals being non-significant and only one exceeding the standard error limits. However, when the OPR was added to this model, the residuals appeared non-stationary in their variance. This was primarily due to the higher number of offences from October 2003, onwards compared to the rest of the series. Transformation of the data improved the residuals somewhat; therefore the results from the transformed data are presented. We note that in 2004 there were major changes throughout the State of Queensland in handling child safety issues. This was primarily due to the Crime and Misconduct Commission’s Protecting Children report (Crime and Misconduct Commission, 2004) that resulted in a number of recommendations for reform of the child safety system. In 2004/2005, $91,312,107 was provided by the State Government to implement the entirety of this report’s recommendations. Since our impact evaluation of the OPRs did not include the period of time when these major policy and organisational changes were taking place, we feel confident to assume that our assessment of the impact of the OPRs was not confounded by these child safety changes in Queensland.

Menacing person offences Our category of “menacing persons” included being armed so as to cause fear, stalking, and other offences against the person apart from physical and sexual assaults. Inspection of the pre-OPR time series observations revealed an increasing linear trend. Additionally, there appeared to be a seasonal trend in the form of peaks occurring in November to March of each year. After regular and seasonal differencing, the ACF/PACF pattern displayed negative spikes at ACF lag (1), (7), (12), (15), and (20) while large PACF spikes were present at lag (2) and lag (12). This pattern was suggestive of both regular and seasonal autoregression. An ARIMA (3,1,0)(1,1,0)12 was the best fit for the pre-OPR time series observations. All residuals with the exception of lag (15) were non-significant and within their standard error limits. The results showed that the introduction of OPRs was associated with a non-significant increase in menacing person offences.

Unlawful entry – dwelling The pre-OPR time series observations did not display a linear trend but did display seasonality in the form of peaks in January to April of each year. After seasonal differencing, the ACF/PACF pattern revealed a negative spike at ACF lag(1) and positive spikes at lags (11) and (12). This pattern was indicative of a regular moving average component and seasonal autoregression. An ARIMA (0,1,1)(1,1,0) 12 was specified and its coefficients were significant. The residuals of this model were all non-significant and within their standard error limits and no variance non-stationarity was present. When the OPR was added to the model, an additional moving average component was needed. As Table II shows, the introduction of OPRs was associated with a statistically significant decrease in unlawful entries into dwellings. This finding is consistent with Chilvers and Weatherburn’s (2004) evaluation of the New South Wales equivalent of OPRs in Queensland and suggests that a large proportion of the decline in crime in Queensland is attributable to more effective policing via the OPR process for dealing with unlawful entry problems. Figure 2 below shows the forecasted versus actual reported unlawful entries. As Figure 2 shows, had OPRs not been introduced across the State of Queensland in early 2001, we would have likely seen a steady, seasonally-based increase in unlawful entries. Indeed, by June 2004, the actual number of unlawful entries was 2,429. Without the introduction of OPRs, we could have expected to see the number of unlawful entries

The impact of COMPSTAT

249

Figure 2. Forecasted versus actual reported unlawful entry of dwellings

PIJPSM 30,2

250

to be almost double the actual figure, ceteris paribus. Based on our analysis, we find that the introduction of OPRs resulted in a “savings” of about $870,000 in the state crime cost of residential burglaries. Unlawful entry – other premises As with the other crime categories we conducted a time series analysis of the impact of OPRs on unlawful entry of “other premises” (including shops and other commercial or industrial buildings). Inspection of the pre-OPR time series observations did not appear to reveal a linear or seasonal trend. The ACF/PACF pattern was suggestive of an autoregressive component with positive spikes at ACF lag (1), (2), (3), and a negative spike at ACF lag (16). The largest PACF spike occurred at lag (1). An ARIMA (1,1,0) model was specified and the residuals of this model were non-significant and within their standard error limits. The addition of the post-OPR observations required the model to be adjusted to an ARIMA (2,1,0). All residuals were non-significant and within their standard error limits. As Table II shows, the introduction of OPRs was associated with a statistically significant decrease in reported unlawful entry offences into other premises (e.g. commercial establishments). We estimate the statewide crime “savings” that can be directly attributable to the introduction of OPRs to be about $1,233,000. That is, if the OPRs had not been introduced, then the cost of crime to Queensland would have been $1.2 million more in crimes of unlawful entry to commercial/business premises. We note, however, that there was a decline in unlawful entries throughout most of the states in Australia, starting at about the same time that the OPRs were introduced into Queensland (see Mazerolle et al., 2007). This is a major confounding factor that we cannot directly answer in the present analysis. Notwithstanding the Australia-wide trends in this category of crime that somewhat taint the strong findings that we show in our ARIMA analysis, we suggest that more research needs to be undertaken to understand the drop in crime in Australia, within the context of Australian states introducing COMPSTAT-like initiatives. Breach domestic violence (DV) orders The pre-OPR time series observations revealed an increasing linear trend and seasonality, with peaks in October to January of each year. After regular and seasonal differencing, the ACF/PACF pattern displayed a negative spike at ACF lag (1), indicative of a moving average process. An ARIMA (0,1,1)(0,1,0) 12 model for the pre-OPR time series possessed residuals that were non-significant and within their standard error limits. The model including the post-OPR observations required specification of a seasonal autoregressive component for its residuals to approximate “white noise”. The results showed that the introduction of OPRs was associated with a non-significant increase in reported breaches of domestic violence orders. We expect this insignificant increase was largely due to efforts within the police department to emphasize the importance of charging people for breaches of DV orders (Superintendent Savage, personal communication, 11 May 2005). Conclusions From the outset, we assumed that there could be four possible OPR outcomes: First, reported offences may have been increasing prior to OPRs and then continued to

increase despite the introduction of OPRs. In this scenario, we would conclude that OPRs were ineffective in making any type of impact on crime in Queensland. A second outcome could have been that reported offences may have been increasing and then stabilised in the post-OPR time period (i.e. a levelling off effect which would not emerge as a significant reduction). In this scenario, we would conclude that the OPRs were successful in harnessing the trend towards a rise in crime, but failed turn the trend around. A third outcome might have involved reported offences being either stable or slightly increasing in the period prior to the introduction of OPRs and then exhibiting a non-significant decrease associated with the introduction of OPRs. In this case, we would have concluded that the OPRs were a contributor to the crime reduction, yet not especially important. Indeed, in this scenario other factors (such as demographic characteristics, income, and employment) were likely to be the more important factors contributing to the crime reductions than OPRs. This is the line of argument described by Eck and Maguire (2000). The fourth possible outcome that we looked for in our statewide examination of the impact of OPRs in Queensland was when reported offences had been initially increasing and then exhibited a significant decrease associated with OPRs. In this scenario, we would conclude that the introduction of OPRs was an important factor contributing to the crime drop across Queensland. We suggest that the last three types of outcomes are indicative of “success”, regardless of statistical significance, particularly considering that population growth across the State of Queensland in the last three years follows a steadily increasing pattern from around 58,000 people per year to 81,000 people from 2003 to 2004 (Queensland Government, 2005). Our analysis reported in this paper used ARIMA interrupted time series analysis to assess the impact of OPRs on reported offences in Queensland. Using crime data spanning a ten-year period (1995 to 2004 inclusive), our results showed that the introduction of OPRs contributed to a statistically significant reduction in total reported offences. We found that the introduction of OPRs prevented around 972 crimes, saving the state approximately $2,773,675[8]. Specifically, OPRs were associated with a significant reduction in unlawful entry offences (unlawful entries into dwellings and other properties). Serious assaults, common assaults, sexual offences, armed robberies, unarmed robberies, and unlawful use of motor vehicle offences all exhibited non-significant decreases associated with the implementation of OPRs. Given these results, we suggest that OPRs in Queensland were a catalyst to important crime reductions in Queensland. The design of OPRs in Queensland captured the six key elements that define COMPSTAT programs across the United States (mission clarification, internal accountability, geographic organization of command, organizational flexibility, data driven problem identification and assessment, and innovative problem solving). Prior to the introduction of OPRs, Queensland Police Service (QPS) had been geographically organized into districts (that then served as the operational level of implementation for OPRs) and had launched Problem and Partnership Oriented Policing (PPOP). The introduction of OPRs brought about unprecedented district-level accountability and enabled districts to trial new organizational arrangements (such as shift rostering and personnel allocations) with the blessing of the QPS leadership, in a controlled and documented environment. OPRs helped to institutionalize the importance of the QPS version of problem-oriented policing (Problem and Partnership Oriented Policing) and provided a forum for the

The impact of COMPSTAT

251

PIJPSM 30,2

252

Police Commissioner to educate and reiterate the importance of PPOP. The reliance on crime analysts to prepare and present graphs and maps in the OPR forums elevated the role of crime analysts within the organization in a way that would not have been possible without the OPR forums. Overall, our results point to how COMPSTAT (and in the Queensland case OPRs) are indeed a significant factor that leads to crime reduction most likely as a result of the way the process brings together key reform elements (such as POP, crime analysis etc) within police organizations. We note, however, some important limitations of our research findings. When we examined the Queensland crime trends in context to the rest of Australia, we note that Australia reported drops in unlawful entries from about 2001 onwards, confounding somewhat our OPR impact evaluation for category of crime. There are many possible explanations for the drop of this category of crime: Perhaps the heroin drought in Australia reduced the rate of unlawful entries? Perhaps the economic factors in Australia around the same time that COMPSTAT like programs were introduced were, in fact, more important predictors of the drop in unlawful entry than COMPSTAT (see Blumstein and Wallman, 2000). Overall, we suggest that our results show that OPRs in Queensland are an important factor in reducing crime yet it was beyond the scope of this current research to rule out (or confirm) the importance of other explanations that have accompanied crime drops elsewhere in the world. As such, we suggest that there is a need for careful, state-by-state analysis of the introduction of COMPSTAT-like programs in the context of assessing economic, political and social trends in order to understand the role of COMPSTAT in contributing to the drop in crime in Australia. Notes 1. The term “COMPSTAT” is shorthand for “computer driven crime statistics”. COMPSTAT is, however, much more than police examining crime statistics. For an excellent review of the COMPSTAT model see Bratton (1998). 2. The OPRs were initially implemented in August, 2001. Human Resource Management, Financial Management and Professional Standards and Ethical Practice were included in the reviews from February/March, 2002. Corporate level commands, including State Crime Operations Command and Operations Support Command were introduced in 2003 and in 2004 the Administration Division and Information Management Division were included. 3. We note the limitations of using reported crime offence data in criminological research. Low reporting rates for some categories of crime, inaccuracies in the data records (e.g. incorrect crime offence dates, non-existent locations, incorrect crime category codes, changes over time in reporting and coding practices etc) as well as administrative errors all comprise potential limitations with the QPS crime offence data. Nonetheless, the crime offence data used in our research comprise the most scrutinised data used by the police in Queensland and are considered, at least in the State of Queensland, as being the “best” indicator of crime available. 4. Unit root testing (the Dickey-Fuller test) and examination of the raw time series plots were used to determine if serial and/or seasonal trends were present and consequently whether a given offence type required differencing. 5. It should be noted that forecasting is not possible if the time series has been differenced. In this situation, the forecasting process simply plots a constant mean for the pre-OPR time series. 6. The cost savings were only calculated for those offences that demonstrated a decrease in the post-OPR time period. Unlawful use of motor vehicles: A decrease of 2 28.5 multiplied by

$6,550 per offence ¼ $186,675. Assaults were calculated together as the costs of assaults were presented in Mayhew as a whole: Serious assaults ( 2 34) þ common assaults ( 2 33) ¼ 2 67 multiplied by $1,800 per offence ¼ $120,000. Sex offences: A decrease of 2 118 multiplied by $2,500 per offence ¼ $295,000. Armed Robbery ( 2 11) þ Unarmed robbery ( 2 8) ¼ 2 19 multiplied by $3600 ¼ $68,400. B/E from dwellings ( 2 435) multiplied by $2000 ¼ $870,000. B/E other premises ( 2 274) multiplied by $4,500 ¼ $1,233,000. All of these figures added together make a grand total of $2,773,675. 7. The residuals were plotted against time and revealed a possible outlier (July 1997). This raw time series observation corresponding to July 1997 was replaced and the analysis was performed with the replaced value. However, this made little difference to the results and therefore the original July 1997 value was retained for impact assessment. 8. This figure is a conservative estimate given that the information on the costs of some types of crimes was unavailable.

References Bayley, D.H. (1994), Police for the Future, Oxford University Press, New York, NY. Bayley, D.H. (1998), What Works in Policing, Oxford University Press, New York, NY. Blumstein, A. and Wallman, J. (Eds.) (2000), The Crime Drop in America, Cambridge University Press, New York, NY. Bouza, T. (1997), “NYPD blues – good, lucky, or both?”, Law Enforcement News, Vol. 31 No. 8, p.. 10. Bratton, W. (1997), “Crime is down in New York City: blame the police”, in Dennis, N. (Ed.), Zero. Tolerance: Policing a Free Society, 2nd ed., Institute of Economic Affairs (Health and Welfare Unit), London, pp. 29-43. Bratton, W. (1998), Turnaround: How America’s Top Cop Reversed the Crime Epidemic, Random House, New York, NY. Brereton, D. (1999), “Zero tolerance and the NYPD: has it worked there and will it work here?”, paper presented at the Australia Institute of Criminology Conference, Mapping Boundaries of Australia’s Criminal Justice System, Canberra 22-23 March 1999. Chilvers, M. and Weatherburn, D. (2004), “The New South Wales Compstat process: its impact on crime”, Australian and New Zealand Journal of Criminology, Vol. 37 No. 1, pp. 22-48. Cook, T.D. and Campbell, D.T. (1979), Quasi-Experimentation: Design and Analysis Issues for Field Settings, Houghton Mifflin, Boston, MA. Crime and Misconduct Commission (2004), Protecting Children: An Inquiry into Abuse of Children in Foster Care, Crime and Misconduct Commission, Brisbane. Dixon, D. (1998), “Broken windows, zero tolerance and the New York miracle”, Current Issues in Criminal Justice, Vol. 10 No. 1, pp. 96-106. Dodenhoff, P. (1996), “LEN salutes its 1996 People of the Year, the NYPD and its COMPSTAT process”, Law Enforcement News XXII, Vol. 458, pp. 1-5. Eck, J.E. and Maguire, E.R. (2000), “Have changes in policing reduced violent crime? An assessment of the evidence”, in Blumstein, A. and Wallman, J. (Eds), The Crime Drop in America, Cambridge University Press, Cambridge, pp. 207-65. Eck, J.E. and Spelman, W. (1987), Solving Problems: Problem-Oriented Policing in Newport News, Police Executive Research Forum, Washington, DC. Goldstein, H. (1990), Problem-Oriented Policing, McGraw-Hill, New York, NY.

The impact of COMPSTAT

253

PIJPSM 30,2

254

Gottfredson, M. and Hirshi, T. (1990), A General Theory of Crime, Stanford University Press, Stanford, CA. Grabosky, P. (1999), Zero Tolerance Policing, Trends and Issues,Vol. 102, Australian Institute of Criminology, Canberra. Greenwood, P., Chaiken, J. and Petersilia, J. (1978), The Criminal Investigation Process, Heath, Lexington, MA. Gurwitt, R. (1998), “The comeback of the cops”, Governing, January, pp. 14-19. Henry, V. (2002), The Compstat Paradigm: Management Accountability in Policing, Business and the Public Sector, Looseleaf Law Publications, New York, NY. Kelling, G.L. and Moore, M.H. (1988), “From political to reform to community: the evolving strategy of police”, in Greene, J.R. and Mastrofski, S.D. (Eds), Community Policing: Rhetoric or Reality, Praeger, New York, NY, pp. 3-25. Kelling, G.L. and Sousa, W.H. (2001), Do Police Matter? An Analysis of the Impact of New York City’s Police Reforms, Center for Civic Innovation at the Manhattan Institute, New York, NY. Levine, J. (1975), “The ineffectiveness of adding police to prevent crime”, Public Policy, Vol. 23 No. 4, pp. 523-45. Maas, P. (1998), “What we’re learning from New York City”, Parade, May 10, pp. 4-6. Magers, J.S. (2004), “COMPSTAT: a new paradigm for policing or a repudiation of community policing?”, Journal of Contemporary Criminal Justice, Vol. 20 No. 1, pp. 70-9. Maple, J. (1999), The Crime Fighter: Putting the Bad Guys Out of Business, Doubleday, New York, NY. Mayhew, P. (2003), Counting the Costs of Crime in Australia: Technical Report, Australian Institute of Criminology, Canberra. Mazerolle, L. and Ransley, J. (2006), Third Party Policing, Cambridge University Press, Cambridge. Mazerolle, L., Soole, D. and Rombouts, S. (2007), “Drug law enforcement: a systematic review”, Police Quarterly (forthcoming). McDonald, P.P. (2002), Managing Police Operations: Implementing the NYPD Crime Control Model Using COMPSTAT, Wadsworth Publishing, Belmont, CA. Moore, M.H. (1997), “The legitimation of criminal justice policies and practices”, National Institute of Justice, Perspectives on Crime and Justice: 1996-1997 Lecture Series, US Department of Justice, Washington, DC. Queensland Government (2005), Queensland Government Population Growth – Highlights and Trends, Department of Local Government, Planning and Recreation, Brisbane. Remnick, D. (1997), “The Crime Buster”, The New Yorker, February 24 and March 3, pp. 94-109. Rosenfeld, R., Fornango, R. and Baumer, E. (2005), “Did ceasefire, compstat, and exile reduce homicide?”, Criminology and Public Policy, Vol. 4 No. 3, pp. 419-50. Safir, H. (1998), The COMPSTAT Process, New York City Police Department, Office of Management Analysis and Planning, New York, NY. Sherman, L.W., Gottfredson, D., MacKenzie, D., Eck, J., Reuter, P. and Bushway, S. (1997), Preventing Crime: What Works, What Doesn’t, What’s Promising?, US Department of Justice, National Institute of Justice, Washington, DC. Silverman, E.B. (1999), NYPD Battles Crime: Innovative Strategies in Policing, Northeastern University Press, Boston, MA.

Silverman, E.B. and O’Connell, P.E. (1997), “Revolutionizing the police: fighting crime in New York City”, Security Journal, Vol. 9, pp. 101-4. Skogan, W. and Hartnett, S. (1997), Community Policing, Chicago Style, Oxford University Press, New York, NY. Smith, M.J., Clarke, R.V. and Pease, K. (2002), “Anticipatory benefit in crime prevention”, in Tilley, N. (Ed.), Analysis for Crime Prevention, Criminal Justice Press, Monsey, NY. Walsh, W.F. and Vito, G.F. (2004), “The meaning of Compstat: analysis and response”, Journal of Contemporary Criminal Justice, Vol. 20 No. 1, pp. 51-69. Weisburd, D. and Eck, J. (2004), “What can police do to reduce crime, disorder and fear”, Annals of the American Academy of Political and Social Science, Vol. 593, pp. 42-65. Weisburd, D., Mastrofski, S., McNally, A.M. and Greenspan, R. (2001), COMPSTAT and Organizational Change: Findings from a National Survey, The Police Foundation, Washington, DC. Weisburd, D., Mastrofski, S., McNally, A.M., Greenspan, R. and Willis, J. (2003), “Reforming to preserve: COMPSTAT and strategic problem-solving in American policing”, Criminology and Public Policy, Vol. 2, pp. 421-56. Weisburd, D.S., Willis, M.J. and Greenspan, R. (2006), “Changing everything so that everything can remain the same: COMPSTAT and American policing”, in Weisburd, D. and Braga, A.A. (Eds), Police Innovation: Contrasting Perspectives, Cambridge Studies in Criminology, Cambridge, pp. 284-304. White, M.D., Fyfe, J.J., Campbell, S.P. and Goldkamp, J.S. (2003), “The police role in preventing homicide: considering the impact of problem-oriented policing on the prevalence of murder”, Journal of Research in Crime and Delinquency, Vol. 40, pp. 194-225. Witkin, G. (1998), “The crime bust”, US News and World Report, May 25, pp. 28-36.

Further reading Glass, G.V. (1997), “Interrupted time-series quasi-experiments”, in Jaeger, R.M. (Ed.), Complementary Methods for Research in Education, 2nd ed., American Educational Research Association, Washington, DC, pp. 589-609. Griffiths, E. and Chavez, J.M. (2004), “Communities, street guns and homicide trajectories in Chicago, 1980-1995: merging methods for examining homicide trends across space and time”, Criminology, Vol. 42 No. 4, pp. 941-78. Kubrin, C.E. and Herting, J.R. (2003), “Neighbourhood correlates of homicide trends: an analysis using growth-curve modeling”, Sociological Quarterly, Vol. 44 No. 3, pp. 329-50. Littell, R.C., Milliken, G.A., Stroup, W.W. and Wolfinger, R.D. (1996), SAS System for Mixed Models, SAS Institute Inc., Cary, NC. McDowall, D., McCleary, R., Meidinger, E.E. and Hay, R.A. (1980), Interrupted Time Series Analysis; Quantitative Applications in the Social Sciences, Sage, Beverly Hills, CA. Prenzler, T. and King, M. (2002), The Role of Private Investigators and Commercial Agents in Law Enforcement,Vol. 234, Australian Institute of Criminology, Canberra. Quigley, B.M., Leonard, K.E. and Collins, R.L. (2003), “Characteristics of violent bars and bar patrons”, Journal of Studies on Alcohol, Vol. 64, pp. 765-72. Sampson, R.J., Raudenbush, S.W. and Earls, F. (1997), “Neighbourhoods and violent crime: a multilevel study of collective efficacy”, Science, Vol. 277 No. 5328, pp. 918-24.

The impact of COMPSTAT

255

PIJPSM 30,2

256

Shaw, C. and McKay, H. (1942), Juvenile Delinquency and Urban Areas: A Study of Rates and Delinquents in Relation to Differential Characteristics of Local Communities in American Cities, The University of Chicago Press, Chicago, IL. Shearing, C. and Stenning, P. (1987), Private Policing, Sage, Beverley Hills, CA. Skogan, W. (1990), Disorder and Decline Crime and Spiral to Decay in American Cities, Free Press, New York, NY. Tabachnick, B.G. and Fidell, L.S. (2001), Using Multivariate Statistics, 4th ed., Allyn and Bacon, Boston, MA. Weatherburn, D., Lind, B. and Ku, S. (1999), “Hotbeds of crime? Crime and public housing in urban Sydney”, Crime and Delinquency, Vol. 45, pp. 256-71. Weisburd, D., Bushway, S., Lum, C. and Yang, S.M. (2004), “Trajectories of crime at places: a longitudinal study of street segments in the city of Seattle”, Criminology, Vol. 42 No. 2, pp. 283-322. Corresponding author Lorraine Mazerolle can be contacted at: [email protected]

To purchase reprints of this article please e-mail: [email protected] Or visit our web site for further details: www.emeraldinsight.com/reprints

The current issue and full text archive of this journal is available at www.emeraldinsight.com/1363-951X.htm

Variation in police patrol practices The precinct as a sub-organizational level of analysis Kimberly D. Hassell Department of Criminal Justice, University of Wisconsin-Milwaukee, Milwaukee, Wisconsin, USA

Variation in police patrol practices 257 Received 26 August 2006 Revised 13 November 2006 Accepted 16 January 2007

Abstract Purpose – The purpose of this study is to examine whether patrol officers believe that police practices vary by precinct assignment and whether the precinct acts as a sub-organizational level of analysis in police organizations. Design/methodology/approach – The research design is a multi-method ethnographic case study including observational fieldwork, both structured and unstructured interviews and official departmental documents. Findings – The study finds that patrol officers overwhelmingly agree that police practices vary at the precinct level of analysis. This variation is perceived to be caused by: individual officer temperament/personality and level of experience,; culture, nature and expectations of the clientele/citizens; nature of calls for service, higher call loads and officer safety concerns; and command and precinct rules/norms. The study also finds that the precinct is a viable and important level of analysis within police organizations. Research limitations/implications – This study highlights the need to examine variation in police behavior within organizations at the precinct level of analysis. Practical implications – The findings from this study have considerable practical implications because the findings indicate that to understand police patrol practices, police practitioners must investigate variation in the informal structures/cultures of police organizations at the sub-organizational level of the precinct. Originality/value – This paper is valuable because police researchers have investigated police behavior at the individual, situational, neighborhood, organizational and legal levels of analyses but have largely overlooked the intersection of these levels of analyses: the precinct. Keywords Police, Policing, United States of America Paper type Research paper

Introduction One of the most elusive concepts for police researchers is discretion. Studies have documented that many decisions made by the police are largely informal in nature, not guided by formal standards or legal norms, but, instead, instituted ad hoc to “get the job done” (National Research Council, 2004). Uncovering patterns in police discretion has been a major focus of research in the last three decades. To this end, researchers have initiated extensive investigation into uncovering what factors influence these informal decisions and, consequently, police patrol practices. Much insight on police organizational behavior has come from observational studies; observational studies have uncovered patterns in police practices, offering a clarity that attitudinal analyses

Policing: An International Journal of Police Strategies & Management Vol. 30 No. 2, 2007 pp. 257-276 q Emerald Group Publishing Limited 1363-951X DOI 10.1108/13639510710753252

PIJPSM 30,2

258

cannot capture. Police observational studies have largely investigated police behavior at five distinct levels of analysis: individual, situational, neighborhood, organizational and legal (Sherman, 1980; Ricksheim and Chermak, 1993). The individual level of analysis focuses on the characteristics of officers that might influence police patrol practices. The situational level of analysis centers on factors specific to a certain situation/encounter and suspect/victim/complainant relationships. The neighborhood level of analysis focuses on certain aspects of neighborhoods, such as the degree of heterogeneity and social disorder. The organizational level of analysis focuses on differences between subcomponents of organizations or organizations as a whole. Finally, the legal level of analysis concerns the role of procedural and substantive law on police patrol practices (Sherman, 1980; Ricksheim and Chermak, 1993). Taken together, these studies indicate that legal factors associated with each situation/encounter are the strongest predictors of police patrol practices (National Research Council, 2004). The research on the relationship between extralegal factors – race, sex, and socioeconomic status – and police patrol practices is less consistent. Most research indicates that police behavior is influenced by extra-legal factors but studies have not produced patterns of findings consistent enough to produce valid and generalizable conclusions (National Research Council, 2004). While observational studies conducted from the 1960s through the 1990s have enlightened the study of law enforcement, there remain unclear patterns in the findings. While the situational level of analysis has been heralded as the most influential in uncovering the relationship between extra-legal factors and police behavior, this study illustrates that we have missed an important point of intersection between all five levels of analysis: the precinct. The present study extends current assessments of police behavior by focusing on the precinct as a sub-organizational level of analysis. Using negotiated order theory as the foundation for analysis, this study demonstrates how the external environment forces police organizations to adapt their formal and informal structures. The formal and informal structures of police organizations are mutually reinforcing and to the extent that factors external to the organization vary, so will the structure of the organization (Strauss, 1978). Variable structural arrangements within an organization produce variable cultures. Officers assigned to each precinct are expected to interpret and negotiate the structural components at the precinct-level to produce distinct approaches in police practices (Klinger, 1997; Hassell, 2004, 2006). As environmental factors within the precinct vary, so will the structural arrangements causing variation in the manner in which officers negotiate order (Strauss, 1993; Klinger, 1997). In that respect, this study addresses the question: Are there differences in police patrol practices at the precinct level of analysis? .

Literature review Policing in the USA is a franchise consisting of anywhere between 17,000 and 21,000 agencies at multiple political levels: federal, state, county, municipality and township (National Research Council, 2004). The policing franchise, consequently, is extensive, politically and socially entrenched and geographically anchored (Klinger, 1997). Police organizations respond to the political, geographic and social entrenchment. They do this by first subdividing police work into geographically smaller, more manageable assignments. These assignments are typically referred to as beats or districts

(hereinafter referred to as districts). Within any one jurisdiction, there are several districts. In smaller jurisdictions, as Klinger (1997) explains, all patrol districts are managed by a single command structure: the police organization. The police organization will operate out of one administrative command and organize police patrol responsibility territorially by districts. As sizes of jurisdictions increase, Klinger (1997) argues, the number of districts extends the administrative capacity of a single management structure. As a result, in these larger areas, patrol districts are grouped to form separate territorial units with separate administrative/managerial structures, usually referred to as precincts. These precincts are jointly connected by the agency’s superordinate administrative/managerial structure. Larger jurisdictions, therefore, coordinate control through the geographic assignment of responsibility by first organizing into precincts and then, within that organizational framework, into patrol districts (Klinger, 1997). In his 1997 article “Negotiating order in patrol work”, Klinger explains how ecological communities are created by the manner in which police patrol work is carried out within the three geographic associations of patrol work (municipality, precinct and districts). In larger agencies, officers are deployed as single units to particular districts, but typically cross-district boundaries within their precinct to assist other officers, and so forth. In this manner, the district does not represent an organizational boundary, but merely a division of labor for a particular shift (Klinger, 1997). Precincts and jurisdictional boundaries are more impenetrable. These boundaries act as organizational borders that, by their very nature, create distinct and relatively independent communities within the infrastructure of the police organization (Klinger, 1997; Hawley, 1950, 1986). Most police officers, for example, interact with others in the same precincts; officers attend roll calls and other informal meetings (e.g. writing reports, checking e-mail) at the precinct. These communities that are formed at the precinct-level have distinct environments in which they function and adapt (Klinger, 1997). These distinctions, it is argued, produce a unique culture among police patrol officers that would vary in accordance with environmental features distinct to each precinct. Past research supports the idea that the precinct is the natural point of intersection between the five general levels of analyses – individual, situational, neighborhood, organizational and legal – used to study police behavior (Klinger, 1997; Kane, 2003). As the research on police patrol practices indicates, variables at all levels of analyses have been shown to have an effect on police patrol practices (Sherman, 1980; Ricksheim and Chermak, 1993; National Research Council, 2004). Legal factors, such as seriousness of the offense and strength of evidence, have all been found to affect police patrol practices (Black, 1971; Black and Reiss, 1970; Friedrich, 1977; Sykes and Clark, 1975; Lundman et al., 1978; Sherman, 1980; Mastrofski et al., 1995; Mastrofski et al., 2000). The legal environment in which the police organization is entrenched is also significant in explaining police behavior. As previously explained, each jurisdiction has autonomy and authority to establish legal codes. In this way, legal factors would manufacture variation in police patrol practices in as much as legal codes vary from jurisdiction to jurisdiction. Organizational factors such as organizational ideology (e.g. traditional vs. community oriented policing) and size have been shown to be significant predictors

Variation in police patrol practices 259

PIJPSM 30,2

260

Figure 1. Levels of analysis in studying police patrol practices

of police patrol practices (Terrill, 2001; Eitle et al., 2005). Responsibility for geographically larger jurisdictions causes organizations to expand, out of necessity, and, as a result, the organization must adapt the structural arrangements accordingly. As the organization expands, the formal structure of the organization changes from that of a smaller agency with administrative control falling under one single administrative/managerial structure. Instead, larger organizations are subdivided into several administrative/managerial command structures within one superordinate administrative command (Klinger, 1997). This structural component causes changes in police patrol practices in that each command structure in each precinct produces a distinct culture. Neighborhood factors have also been shown to affect police patrol practices. The levels of social disadvantage, disorder and crime, as well as racial/ethnic heterogeneity, have all been determined to be significant variables in explaining variation in police patrol practices (Smith, 1984; Sampson and Lauritson, 1997; Smith et al., 1984; National Research Council, 2004). Situational factors such as ascribed characteristics as well as suspect resistance and disrespect/demeanor, mental illness, supervisory presence and officer safety issues have demonstrated effects on police patrol practices (Black and Reiss, 1970; Lundman et al., 1978; Friedrich, 1977; Black and Reiss, 1970; Black, 1971; Lundman, 1974; Sykes and Clark, 1975; Smith and Visher, 1981; Smith et al., 1984; Worden and Sheppard, 1996; Swatt, 2002; Hassell, 2006). Furthermore, the same ascribed characteristics, as well as level of experience/length of service, have been shown to affect policing at the individual level of analysis (Friedrich, 1977; Fyfe, 1979; Terrill, 2001) (Figure 1). Previous research on police behavior has revealed that each level of analysis – individual, situational, neighborhood, organizational and legal – is important on its own. Factors at each level of analysis have been demonstrated to have a significant effect on police patrol practices. These separate levels of analyses come together at the precinct to form an important sub-organizational level of analysis. The structural arrangements of police organizations affect the informal structure, or the “way things are done around here”. These structural components (formal and informal) cause changes in police patrol practices in each precinct as each precinct reacts to different structural characteristics as well as micro-political influences in its immediate external environment (Klinger, 1997). These changes produce a distinct culture at the precinct-level, which in turn produces variation in police patrol practices between precincts. Using negotiated order theory, this study tests this assertion.

Negotiated order theory Negotiated order theory rests on the assumption that formal and informal structures are intertwined through interaction; as a result, both structure and interaction play a major role in the negotiation of social order (Hogelucht and Geist, 1997, p. 2; Strauss, 1978). These negotiations create new meaning of what constitutes social order. Negotiated order theory, therefore, enables us “to recognize the significance of individual actors who are in association with one another” (Hogelucht and Geist, 1997, p. 3; Maines and Charlton, 1985, p. 302). Strauss (1978) outlined four primary propositions of negotiated order theory. First, social order, or the status quo, is negotiated order. Strauss (1978, p. ix) observed that social orders, even those that may be repressive in nature, would not exist without some form of negotiation. This point is particularly salient for police patrol officers because patrol work is largely unsupervised (directly) and may not always fall within the parameters of formal policies and procedures (Klinger, 1997). Second, Strauss (1978) emphasized the role of structure, both formal and informal, in the negotiation of social order. Negotiations follow lines of communication, as organized through formal structural components (Fine, 1984). Negotiations, therefore, are structured. As Strauss et al. (1985, pp. 5-6); see also Fine (1984, p. 241) explained: The negotiated order on any given day could be conceived of as the sum total of the organization’s rule and policies, along with whatever agreements, understandings, pacts, contracts and other working arrangements currently obtained. These include agreements at every level of organization, of every clique and coalitions, and include covert as well as overt agreements.

Formal and informal structures are both important and, consequently, are reinforcing in that negotiations are dynamic and adaptive. The third proposition is that negotiations are renewed, revised and reconstituted over time (Fine, 1984; Strauss, 1993; Strauss et al., 1975; Smith et al., 1984). The structural contexts within which negotiations occur, then, is represented as fluid and temporary (Reed, 1985). Fourth, and a related point, structural changes in the organization require a revisit of the negotiated order (Fine, 1984). We would expect, consequently, that the negotiation of order, in the context of law enforcement, would be adaptive and dynamic as well. Additionally, organizational negotiations are not shielded from external influence. Organizational negotiations will be affected by political, social and cultural forces within which the organization functions (Crank, 2003; Katz, 2001). Based on the four propositions of negotiated order theory, this study investigates whether there are precinct-level differences in police cultures and, concomitantly, police patrol practices in a large, municipal Mid-western police department. More specifically, the two following hypotheses are tested: H1. Police patrol officers will perceive that similar situations are handled differently in each precinct. H2. Police patrol officers will perceive that the precinct-level differences are attributed to factors at five levels of analyses: individual, situational, neighborhood, organizational and legal.

Variation in police patrol practices 261

PIJPSM 30,2

262

Methodology Research site Midwestern Municipality (a pseudonym) is in the heart of the Midwestern United States. According to the 2000 Census, the city population in 2002 was approximately 400,000 persons. The racial breakdown of the city is predominantly white (78.4 percent), with African Americans (13.3 percent) and Hispanics (6.1 percent) comprising the two largest minority groups[1]. The racial composition of sworn officers in MMPD is 82 percent white, 11 percent African American, 5 percent Hispanic and 2 percent “other”. The Midwestern Municipality Police Department (hereinafter referred to as MMPD) is nationally accredited by the Commission on Accreditation for Law Enforcement Agencies and employs approximately 760 sworn police officers. MMPD is organized into four precincts, with separate and distinct precinct stations[2]. Upper administration is centralized in one main headquarters as is the investigation bureau. All special units are housed in the same headquarters location. Further, booking procedures are handled at the headquarters; all reports and evidence are forwarded to the main headquarters for processing and archival. The precincts are organized geographically as Northwest Precinct (NW), Northeast Precinct (NE), Southeast Precinct (SE) and Southwest Precinct (SW). Each precinct is organized into two geographic command areas that are further subdivided into nine individual districts. Three patrol crews are assigned to each shift, for a total of nine crews per precinct. The patrol shifts for each precinct are organized into three standard shifts: (1) A shift – overnight shift; (2) B shift – day-time shift; and (3) C shift – afternoon shift. On each shift, one crew is permanently assigned to one geographic area; the second crew is permanently assigned to the other geographic area; and the third crew is a relief crew and works two days in each geographic area. Officers are not permanently assigned to individual districts; these assignments are provided during roll call. The patrol workweek is established as a four-day workweek (four days on and two days off). This means that on any given shift, there will be two crews working in each precinct. All roll calls are handled in the respective precincts for all shifts. Each precinct has a separate precinct captain. In addition to the precinct captain, each shift in each precinct has a lieutenant, for a total of three lieutenants for each precinct. Each crew, additionally, is assigned its own sergeant, resulting in nine sergeants assigned to each precinct. Each sergeant is assigned a crew of patrol officers, which varies by precinct and shift. Official records obtained from MMPD reveal that the largest crews are assigned to the afternoon shifts in the NE precinct while the smallest crews are assigned to the morning shifts in the SW precinct. Research design The research design is a multi-method ethnographic case study including observational fieldwork, both structured and unstructured interviews, and official documents obtained from MMPD. During field observation, the author accompanied patrol officers systematically recording the following data: the nature of police-citizen

interaction (officer-initiated or citizen-initiated); the reason for the interaction; the start time of interaction; the response time (time elapsed between time of call and actual arrival); the number of officers involved; the disposition of interaction; and the end time of the interaction. All observational notes were recorded with pen and paper. Field observations spanned a six-month time period, from July 2003 through November 2003. Participation by the patrol officers was voluntary. During roll calls, the sergeant would request volunteers who were willing to participate in the research. Most patrol officers were aware of the study as the observer met with officers during roll calls at each shift in each precinct prior to the start of data collection. The observation period totaled 434 hours of participant observation. During field observation, unstructured interviews were conducted. These interviews involved asking patrol officers to explain their decision-making process and to discuss their feelings/thoughts about situations involving discretionary decisions. Overall, 76 unstructured interviews were completed with 76 officers across all four precincts. Since the study is a cross-precinct analysis, there was a need for standardized assessment to make valid comparisons across precinct assignments. An 11-page questionnaire was designed to uncover police patrol officers’ perceptions of precinct-level factors in order to investigate precinct-level variation. A total of 72 structured interviews (based on the questionnaire) were completed by 18 officers in each of the four precincts[3]. A coding sheet was developed based on the questionnaire responses and data were analyzed both qualitatively and quantitatively. Several questions from the questionnaire were used in this analysis: demographic variables documented officer sex, race, age, years of police experience and years of service with MMPD. Officers were asked about persistent crime/deviance problems in their precincts as well as the number of calls for service. Each officer was also asked whether they believe that police patrol practices vary by precinct. That is, officers were asked whether there were precinct-level differences in police patrol practices produced by precinct-level factors. Officers were then asked to explain their responses (i.e. what are the precinct-level factors that produce precinct-level variation in police patrol practices? Why do you think there is precinct-level variation in police practices?). Participants were also asked whether officers handle similar situations differently based on precinct assignment; all respondents were then required to explain their responses. These responses were recorded with pen and paper and later transcribed into Microsoft Access for query analyses and SPSS for statistical analyses. Field observation notes also documented visible precinct-level variation in police patrol practices and findings from the analysis are included in this examination. Official agency documents, which included precinct-level statistical reports, were collected in an effort to cross-reference data collected during the interviews. For example, precinct-level statistical reports documented the types/nature of crimes as well as the magnitude of their occurrence in each precinct. Precinct-level statistical reports also documented population density. Prior evaluation reports provided a basis for contextual/historical analysis, which informed on bidding processes, geographic restructuring, relevant policies and command structure changes. Additionally, police union newsletters and departmental reports informed the analysis; the newsletters corroborate the cultural tone of the organization, as it was recorded by the observer.

Variation in police patrol practices 263

PIJPSM 30,2

264

Sample Field observation was conducted with 76 patrol officers. Overall, 72 structured interviews[4] were completed during field observation (18 completed questionnaires in each of the four precincts)[5]. A total of 73 percent of the participating officers were male while 27 percent were female. The overall composition of sworn officers in MMPD is 80 percent male and 20 percent female. Overall, 87 percent of the sample identified themselves as white, while 6 percent identified themselves as African American, 4 percent as Hispanic and 3 percent as other. These figures compare to the departmental racial/ethnic population of sworn officers: 82 percent white, 11 percent African American, 5 percent Hispanic and 2 percent as another racial/ethnic identity. The mean age of the participating officers is 34 (range of 23-53; standard deviation of seven). Officers participating in this study ranged in their tenure at MMPD from one to 22 years, with a mean experience level of seven-and-a-half years (standard deviation of six). Precinct descriptions Southwest precinct. The SW precinct is one of the two largest (geographically) precincts with the fewest number of patrol officers assigned. It is approximately three times the geographic size of the smallest precinct. The precinct is a mix of residential and business, including most of the city’s apartment complexes, three hospitals, the local university and the largest shopping mall in the city. The cluster of apartment complexes within this precinct represents the densest part of the city. According to MMPD statistical reports, the precinct has the lowest crime level with most calls for traffic accidents, traffic hazard issues and intrusion alarms. The residents of this precinct are primarily middle- to upper-income non-Hispanic whites; the residents of this precinct are the wealthiest in the city. The housing market ranges from rental properties to the most expensive homes in the city. Data from unstructured interviews with command staff indicate that many of the officers assigned to the precinct are the most experienced officers in the department. The sergeants in this precinct are also the most experienced with the most amount of time on the job. According to the SW precinct commander, the policing style within the precinct is solely reactive; proactive policing is discouraged. The primary patrol responsibility within the precinct is to answer the calls for service. Northwest precinct. The NW precinct is the second largest precinct (geographically) with the second fewest numbers of officers assigned (second to SW). The precinct is almost three times the geographic size of the smallest precinct. The precinct is a mix of residential and business, including two shopping malls, apartment complexes, and one hospital. In addition, the precinct has an apartment complex that is part of the scattered-site subsidized housing project. According to MMPD statistical reports, the precinct has the second lowest crime level; most calls are for traffic accidents, intrusion alarms and civil disturbances. The demographics of the precinct are mixed. The precinct is composed of two areas: the 10-area and the 20-area. The 10-area is mostly middle- to upper-income Whites while the 20-area is composed of a blend of middle-income Whites, lower-income African Americans, and impoverished Sudanese immigrants. Data from unstructured interviews with command staff indicate that many of the officers assigned to the precinct range from mid-level experienced officers (five to seven years of service) to newer officers. The sergeants vary from mid-level

experienced to newly appointed sergeants. According to the NW precinct commander, the policing style within the precinct is primarily reactive with a minor emphasis on proactive policing. Patrol responsibilities within the precinct include answering calls for service. Northeast precinct. The NE precinct is one of the two smallest precincts (geographically) and has the highest number of officers assigned. The precinct is primarily residential. According to MMPD statistical reports, the NE precinct has the highest crime level within the city with calls for service for civil and domestic disturbances, violent street-level crime and drug-related activities. On the first ride along in the precinct, for example, there were five armed disturbance calls within the first 15 minutes. It is also known among the patrol officers as the most dangerous precinct, with respect to police officer safety: the last four police officers killed in the line of duty were assigned to the NE precinct. The precinct is predominately comprised of African American residents with relatively few white persons residing within the precinct. The NE precinct is also the most economically deprived in the city. Additionally, according to officers and newspaper accounts, there is a long history of racial conflict between residents of the precinct and the police department. Data from unstructured interviews with command staff indicate that the officers assigned to the precinct are a mix of mid-level experienced officers (five to seven years of service) and newer officers. The sergeants vary from mid-level experienced sergeants to newly promoted sergeants. According to the NE precinct commander, the policing style is both reactive and proactive, with a heavy emphasis on proactive policing. The patrol responsibilities within the precinct are answering calls for service and proactive law enforcement activities. Southeast precinct. The SE precinct is the second smallest precinct geographically and has the second highest number of officers assigned (second to NE). The precinct is a mix of residential and business, including the downtown area, which is the newly renovated economic hub of the city. According to MMPD statistical reports, the precinct has the second highest crime level with most calls for service for civil and domestic disturbances, as well as gang and drug related activities. The demographics of the precinct are mixed. The precinct is comprised of two distinct areas: the 50-area and the 60-area. The 50-area is a mix of middle- to upper-income white residents who live within the downtown area and lower-income white and Hispanic residents. The 60-area is comprised primarily of lower-income Hispanic residents. Data from unstructured interviews with command staff indicate that the officers assigned to the precinct range from mid-level experienced (five to seven years of service) officers to newer officers. The SE sergeants vary from mid-level experienced to newly appointed sergeants. According to the precinct commander, the policing style within the precinct is both reactive and proactive, with a heavy emphasis on proactive policing in the 50-area. Patrol responsibilities within the precinct include answering calls for service and proactive law enforcement activities. Findings One of Klinger’s principal arguments is that police patrol practices vary within an organization at the precinct-level of analysis. Using triangulated data, the study assesses whether there are precinct-level differences in police behavior. Although participant observation was utilized to observe patrol practices, officers were also

Variation in police patrol practices 265

PIJPSM 30,2

266

asked during interviews whether or not they perceived differences in police patrol practices at the precinct-level. Officers were also asked whether they believe that similar situations are handled differently in each precinct. Across all precincts, 93 percent (66 of 71)[6] of the officers stated that similar situations are handled differently in each precinct; that there is, in fact, an organized precinct-level effect on police patrol practices. In SW, 15 out of 18 officers reported that there is precinct variation in police patrol practices. In both NW and NE, 17 out of 18 officers perceive variation while every officer (N ¼ 17) in SE stated that there is precinct-level variation in police patrol practices. Officers reported several reasons why situations would be handled differently depending upon the precinct in which it occurred: Individual temperament/personality (25 out of 71 officers), level of experience and newness to job (19 out of 71 officers), culture of clientele/citizen expectations (13 out of 71 officers), nature of calls/higher call loads/officer safety (20 out of 71 officers) and command and precinct rules/norms (9 out of 71 officers). Many officers did not restrict their explanations to one reason for precinct-level variation but mentioned two to three reasons for differences in police patrol practices. Additionally, data from field observations corroborated these same patterns. Individual temperament/personality and level of experience – individual level of analysis During interviews, many officers stated that individual temperament/personality and level of experience were important factors in explaining precinct-level variation in police patrol practices. The variation occurs at the precinct level of analysis because, as officers frequently mentioned, most younger and newer officers bid to work in NE and SE because of the higher call load and the nature of calls that occur in these high-crime precincts. For those younger and newer officers who prefer to work in the western part of the city, seniority limits their choices. An officer assigned to SW explained it this way: Older officers are not as aggressive as younger officers. The time on the job affects how officers handle things. The older officers want to work in SW and NW and the younger officers cannot bid out of the east. So younger officers work in NE and SE.

Another officer in NE reported that older officers bid out west while younger officers gravitate toward the east, which causes differences in police patrol practices. This same officer said, “In SW, guys don’t like traffic stops. In NE, officers are fairly aggressive, making traffic stops and backing people up. In SW, they have the retired on duty cops.” An officer assigned to NW commented, “Younger, more aggressive officers are drawn toward SE and NE because it’s a high crime area, more people to arrest.” Another officer assigned to NW said, “Officers with a higher threshold for excitement and stress are drawn to NE.” One officer, assigned to NW, put it curtly, “Newer officers make more traffic stops and do officer-initiated stuff. Older officers answer their radio calls and want to fit in dinner.” In fact, officers also told stories of other officers who were forced to bid to the western precincts because of their “overly-aggressive nature.” As an officer assigned to SW explained, “I know officers who were in NE and were forced to bid on another precinct. The officer will slow down but will return to aggressive policing. They will do the same thing out here.”

Culture and nature of the clientele/citizen expectations – situational and neighborhood levels of analyses Observational and interview data also captured strong inter-precinct variation in culture and nature of the clientele. As one officer assigned to SE said: There is not much variation in attitude. There is variation in the people we deal with, variation in the clientele. You handle calls the same but you encounter different attitudes toward the police, which make it more difficult to handle calls. It is difficult to get compliance from people who hate the police. In NE, you find that people teach their children to dislike the police. This is based on what people hear and experience. Dealing with the police is always a negative contact because either you are victimized, ticketed or arrested. [It is] always a difficult and negative experience, which creates more tension between the public and police.

Another officer assigned to SE said that precinct differences “are largely based on nationality of citizens”. As this officer explained, in the SE precinct, the population consists of Hispanics, white persons, Native Americans, Sudanese and African Americans. Most residents are Hispanic. In NE, there is a concentration of African Americans and white persons. Most residents are African American. In NW, according to officers, there is an equal mix of African Americans (in 20 area) and white persons (in 10 area). Finally, SW is comprised mostly of white persons. An officer assigned to NW echoed these officers, saying: The clientele is different in each precinct. Pockets of crime are all over. In NE, there are more anti-police people, more dangerous situations for officers. In SE, there is the language barrier. A simple situation can escalate because of the language barrier and gang problems. In NW, there is a big influx [of immigrants] on the 20-side. It is like working NE. The 10-side [of NW precinct] is like the 70-area [SW precinct]. In SW, there are more traffic violators and DUIs and some officers like to work DUIs.

Citizen culture and expectations of police, as officers explain, are dramatic influences on police patrol practices in each precinct. Many officers hold similar perceptions – regardless of the precinct to which they are assigned. Officers assigned to SW stated: There are a variety of people and cultures and racial/ethnic groups. They have different kinds of relationships with police and different expectations from police. There are times in NE when you have to tell somebody to shut up. Different expectations; different everything. In NE, officers are more curt, ruder with citizens based on the way people are treating the officer. Down in those areas, officers have a heightened awareness. When people are vulgar at you in SW, you blow it off. When it happens constantly, it affects you in NE. When a NE officer comes out to SW, you see the difference. The officer jumps down their [citizens] throats. I think, “you’re gonna get a complaint”. . . .There are higher expectations for police out here in SW.

Officers assigned to NE also explained how culture and expectations impact police patrol practices: Ethnicity of different precinct makes a difference. It is not a racist thing. It’s what you are used to working. In SE, you have more Hispanics and in NE you have more blacks. On disturbance calls, black females are loud and emotional and it is a lot different than a call in SW. The Sudanese don’t like female police officers. Different cultures shape policing practices. Black females don’t respond to politeness. You have to be assertive and

Variation in police patrol practices 267

PIJPSM 30,2

authoritative. The Hispanic culture don’t think females have authority so they don’t like female officers.

Officers assigned to SE also pointed to the manner in which culture and expectations cause police patrol practices to vary by precinct. An officer working in SE said:

268

Perception of the police varies by precinct. In SW, people believe they are above the law and name drop. There is a language barrier in SE. On calls with Hispanics, you have to start with the guy. In Mexican families, the guy is in charge. This makes it more difficult for female officers.

Finally, officers assigned to NW also relayed that cultures and citizen expectations are important factors in understanding precinct variation in police patrol practices. In fact, one officer working in NW stated that his/her policing style changes based on whether he/she is working the 10- or 20-side of the NW precinct. This officer explained that things happen at a faster pace in NE, SE and the 20-area (in NW). (S)he said (s)he adapts his/her policing style in between the two areas: “I am more laid back on the 10-side.” Other officers assigned to NW explained how the culture and climate of the clientele impacts police practices: SW is the precinct that doesn’t have a lot going on. They don’t want you to look too aggressive and proactive because it makes them look worse.

Another officer, assigned to NW but who has worked in SE, put it this way: There is a higher expectation and a difference in the people themselves, reflective of how much these people don’t have a life. How the smallest things annoy them; nosey little neighbors who have no idea how bad it is in other parts of the city who think it is the worst thing in the world for a neighbor to have a car for two months that hasn’t been plated. [In SE] you have people who are concerned that the school children going to [public] elementary who have to worry about their kid getting run over, prostitution, drugs and the guns that the drug dealers have. They have real crime to worry about.

This officer explained why he/she bids and transfers between the NW and SE precincts: It’s a little bit of a mental break when I switch. Everybody has an issue, a problem that they are dealing with and everybody thinks their problem is the most serious. It gives me a break from one type of policing, one type of area. And I think if I were living in their area, I would think it was a problem too.

Furthermore, officers explained that local community groups and local politics affect policing practices differently in each precinct. As one officer assigned to NW stated, “Environment shapes funding in each precinct, which will change policing.” This sentiment was repeated by an officer assigned to SW: “It’s a little bit of politics. You have to tell the community what they want to hear instead of what they should hear. You have to tell the public what they want to hear. Yeah, you are ’BS-ing’ a little bit.” Officers also discussed how citizen disrespect varies by precinct assignment, particularly between the SW, the precinct with the lowest crime rate and NE, the precinct with the highest crime rate. As an officer assigned to SE explained, “In SW, you get attitudes of the wealthy. “I pay your salary. Why are you giving me a ticket? How dare you.’ I hate working SW because of those attitudes.” At the other end of the spectrum is the NE precinct. An officer assigned to SE described the climate in NE as

“lots of animosity toward police. It’s almost like everybody up there is taught to hate cops.” Officers in NE complain that the clientele perceive officers as racist. During participant observation, salutations (in the form of hand waves, smiles, or nods) were never observed in NE. In its place was swearing and barrages of “Bad boys, bad boys, whatcha gonna do”. An officer in SW explained the racial tension in NE saying, “If you do aggressive policing and you are in a minority neighborhood, you will get tagged a heavy-handed cop . . . . Race relations are so poor. It never used to be like that.” As officers explain, the “racist” label does not only apply to white officers. In fact, an officer working SE said, “Black citizens get on black officers a lot harder: Uncle Tom this and Uncle Tom that. They treat black officers much worse than white officers.” An officer assigned to SE differentiated the climate in SE from NE saying, The difference between NE and SE is the volume and drama. Black people are more verbal, more dramatic. In SE, they don’t play that so much. In NE, they are playing for the crowd . . . When you talk to people in NE, you talk to people differently to get the same reaction . . . I would rather deal with ten Mexicans than one black because of the mouth. Hispanics are more courteous to police.

One officer in SE said s(he) feels like there is more community support in SE than NE: “In NE, they [clientele] would call me and then yell at me and tell me how worthless I am and how bad the police are.” Nature of calls, higher call loads and officer safety – situational and neighborhood level of analysis Field observation and interviews also uncovered distinct variation in the nature of the calls for service in each precinct. The major variation occurred between the eastern and western precincts, particularly NE and SW. An officer assigned to SE explained, “Cops in NE don’t have a lot of time. Citizen contacts aren’t friendly. In SW, it is the opposite.” An officer in NW further elucidates on these variations, “Each precinct has a different element of crime so we have to handle the crime differently. If you have a lot of misfits smoking dope or doing gang activity, you are going to crack down on a bunch of kids standing on a corner.” During observation, all participating officers were asked to describe the most common crime, disorder or neighborhood problems in their precinct. Officers were also asked to discuss the nature of most of their calls for service. There were inter-precinct variations in officers’ perceptions of crime and disorder in their assigned precincts. Officers in the NW, NE and SE precincts were more likely than those in the SW precinct to believe that civil/neighborhood disturbances and domestic disturbances were the primary problems and generated the most calls for service. In contrast, almost two-thirds of the officers assigned to the SW precinct, but relatively few of the officers assigned to the other three precincts, mentioned traffic accidents as an important source of their calls for service. NE and SE precinct officers were more likely than officers in the two western precincts to believe that assaults and drug-related activity were more common. The nature of the calls for service, as well as the perceived crime problems, varies significantly between the eastern and western precincts. In the western precincts, no officers indicated that street-level crimes (e.g. drug-related activity, prostitution, gang-related activity, intoxicated persons calls) occur regularly; this is in direct

Variation in police patrol practices 269

PIJPSM 30,2

270

contrast to officers’ reports in the eastern precincts. Officers in the western precincts report that most calls are what can be classified as service-related calls, such as civil/neighborhood disturbances (e.g. couples arguing over property ownership, arguments between neighbors), traffic accident reports and residential and business alarm calls. Officers in the eastern precincts perceive more of their calls to be street-level crime problems, such as gang-related activity, drug-related activity, and fight disturbances/assaults, in addition to handling civil/neighborhood disturbances and domestic disturbances/domestic violence calls. In the SE precinct, however, there also appears to be more attention given to order-maintenance type calls, such as intoxicated person calls, prostitution and drug-related activity. Official agency data that was collected during interviews and field observation support the officers’ perceptions. The nature of the calls for service and the call load results in variation in officer-initiated activity and report writing. At an initial meeting, each precinct captain was asked to explain the priorities in their precincts. The captains from the eastern precincts explained that their officers primarily respond to calls for service but that officers are encouraged to be proactive (stop and frisks, traffic stops, etc.). Conversely, the captains from the western precincts explained that service-oriented activities, in addition to responding to calls for service, were stressed as important to their officers. Officers also commented on the variation in officer-initiated activity among the eastern and western precincts; an officer from NW explained, “[there are] more traffic stops in east . . . .” Officers also commented on variation in reports; an officer from NW said, “NE guys don’t do as many reports as we do. They try to get out of reports, especially on domestic cases. Repetitive calls. . .you know they won’t change.” Another officer from NW said, “. . .the NE officers hate doing reports. . .One time they broke up an attempted burglary. The guy left and the NE officers never did a report of anything. The NW officer would have done an information report. They [NE officers] are on to the next thing.” Several officers also conveyed that officer-safety issues affect police patrol practices. An officer assigned to SW explained: “Officer safety issues affect how officers in each precinct police. NE officers might blow up faster than officers in SW in order to avoid a crowd gathering, which is an officer safety issue.” Another officer assigned to the SW precinct said: SW is laid back. Nothing ever happens. On an alarm call, I’ll think I’ve been here before. No big deal. In another precinct where there are more robberies and burglaries, they will be a little more cautious. They won’t cancel their back-up like some officers do out here.

This sentiment was repeated several times: In NW, [the focus on] officer safety is zero. They are complacent and have a “ho-hum attitude.’ NE officers are more aggressive. NE officers take charge of situations. NW officers would just stand there (Officer assigned to NE). I was a lot more aware of officer safety when I worked in NE and SE. I was pulling weapons off people and had more people resisting arrest. I had to react to the problems they have (Officer assigned to SW).

In NE, the second you get out of a car and you are investigating suspicious people, you handcuff them. I did that here [SW] and the boy said he was offended to be handcuffed. It is more safety-oriented in NE (Officer assigned to SW).

The concerns regarding danger and the need for heightened officers safety that were common in NE, SE and the 20-area in NW were not generally found among officers in SW. Officers were observed calling off backup several times. As one officer in NW said, “it’s a macho thing. They [officers assigned to SW] don’t wait for backup before responding. They cancel the call. It’s a constant thing.” Command and precinct rules/norms – organizational level of analysis During interviews, officers also discussed the relationship between command and precinct rules/norms and police patrol practices. Officers mentioned that command officers are concerned primarily with the public’s perception of crime. Crime is higher, and the nature of crime more serious, in the eastern precincts; therefore, the command staff are more permissive of aggressive police practices. As an officer explained, “. . .expectations of command. . .in NE and SE, you handle the call and move on. You do less reports. In SW and NW, you do a more thorough investigation and do more reports.” Priorities also vary. In MMPD, one service provided to citizens is Mayor’s complaints. If a citizen is concerned over a minor problem (such as cars parked for long periods of time, cars without proper tags, speeding, etc.), they may call the Mayor’s line and police will respond. As officers explain, command attention to officer responses to Mayor’s complaints varies by precinct. An officer explained, “In NW, the lieutenant was very serious about mayor’s complaints. You had to check it several times. In SE, if you check it one time, you are good.” Further, officers commented that command’s pet peeves are handled more diligently. Discussion This study highlights several factors that explain why the precinct, as a level of analysis, is a cogent point of intersection between the individual, situational, neighborhood, organizational and legal levels of analysis. At the individual level, this study informs that newer and more aggressive officers choose to bid to work in higher crime precincts. Officers who want to engage in more aggressive law enforcement practices gravitate toward the precincts that offer them those opportunities, which results in concentrations of vigorous officers assigned to higher crime precincts. An analysis of both observational and interview data revealed that the contextual variation between precincts is quite pronounced, which also impacts both the situational and neighborhood levels of analysis. At the situational and neighborhood levels of analyses, the nature of the calls for service and the related officer-safety concerns impacts both the culture of the officers in each precinct and the manner in which their work is performed. Officers working in the NE and SE precincts handle more serious calls for service and, consequently, work in a more dangerous atmosphere, where situations and encounters can become immediately explosive. The last four officer fatalities in Midwestern Municipality occurred in the NE precinct; the most recent officer fatality occurred during the data collection period for this study. The officers assigned to the SW and NW precincts – the lower crime precincts – do not share the acute awareness of impending danger as officers assigned to the NE and SE precincts – the higher crime precincts.

Variation in police patrol practices 271

PIJPSM 30,2

272

Recent immigration patterns have also affected the landscape of police patrol work. Studies have begun to document the challenging nature of interactions between English-speaking police officers and Spanish-speaking populations (Herbst, 2003). These language barriers produce variations in police patrol practices in Hispanic communities. Further, as the data indicate, situations intrinsic to the Hispanic culture impacts police patrol practices in predominately Hispanic neighborhoods. These situations are expected to become more dramatic as the United States Census Bureau estimates that by the year 2050, Hispanics are expected to represent 47 percent of the population in the USA (Holmes, 1996; Herbst, 2003). Historical accounts have documented that African-American cultures have long-standing perceptions of racial inequity in police practices (Barlow and Barlow, 2000). These perceptions have figuratively, and literally, put the police and African-Americans at odds, in some places, which impacts police patrol practices in African American neighborhoods. Further intensifying the cultural divide is the natural pattern of racial and ethnic segregation in most American cities. The 2002 Census report indicates that racial and ethnic residential segregation is still a major issue in American municipalities and is higher among African Americans across all other racial and ethnic groups. Hispanics are the next most highly segregated ethnic population. Residential segregation pattern is a pressing matter for law enforcement because cities, particularly in the North and Midwest, have become increasingly racially isolated (Wilson, 1987). These patterns of racial segregation impact police patrol practices at the precinct level of analysis because precinct boundaries, in Midwestern Municipality, almost fall directly in line with racial and ethnic residential segregation patterns. At the organizational level of analysis, administrative ideologies have a dramatic effect on precinct-level patrol practices (Engel, 2001). Each precinct has its own command structure. In the current study, officers stated that command officers institute policies and make decisions based on the public perception of crime; in addition, officers perceived that individual sergeants, lieutenants and captains have certain pet-issues that result in differential police practices across precincts. Patrol officers also stated that precinct commanders differ in what issues they regard as serious and the magnitude of officers’ responses to these issues. Additionally, at the legal level of analysis, laws specific to each municipality will produce variation in police practices at the precinct level of analysis. In Midwestern Municipality, if, during a traffic stop, a motorist cannot produce proof of automobile insurance, the motorist is issued a street-release arrest citation. At this point, officers can search the automobile and the motorist (search incident to an arrest). This municipal ordinance affects law enforcement differently at the precinct level of analysis because motorists are less likely to have automobile insurance in poorer neighborhoods. The poorer neighborhoods fall in the NE and SE precincts, which are the two high-crime precincts. Officers are more aggressive in higher crime precincts and, in turn, initiate a greater number of traffic stops in these precincts. Further, philosophical changes in the institution of policing produce variation in police command ideologies and, as a result, police patrol practices. The institution of policing has been engulfed in reform since its inception. The current philosophical shift toward the community policing ideology has changed the nature of police-citizen interactions, as well as citizen expectations of the police (Kelling and Moore, 1988). In

attempts to maintain and, in some cases, recapture legitimacy for the institution of policing, police reformers have advocated citizen satisfaction as one of the primary performance measures of police effectiveness (Greene, 2000; Barlow and Barlow, 2000). The denotation of citizen satisfaction as a measure of police effectiveness, however, is subjective. What denotes citizen satisfaction diverges based on myriad factors external to the organization: varying racial and ethnic cultures, the nature of community problems and the expressed needs and expectations of the clientele. As a result, police departments have been encouraged to assuage political interest groups and citizen outcry by adapting the delivery of police services. These adaptations occur at the precinct level of analysis where precinct commanders strive to meet the expectations of their clientele. As the nature of the clientele and the nature of crime and deviance vary between precinct boundaries, so will the provision of police services between precincts. Implications of the study The findings from this study have considerable implications for police practitioners and researchers. First and foremost, to understand police patrol practices, police practitioners and researchers must investigate the informal structure of police organizations at the sub-organizational level of the precinct. It is clear that the informal structure and, as a result, police patrol practices vary considerably within a police organization at the precinct level of analysis. Additionally, police practitioners interested in affecting change within a police organization must be aware that effective implementation will take different shapes depending on the informal structures within each precinct. The implication for police organizations is that in designing and implementing strategic plans or other goal-building strategies, police administrators must consider the different needs of each precinct. Some policies enacted department-wide may in fact be more or less effective depending on the informal structures of each precinct. Future research should continue to delve into the organizational processes that occur at the sub-organizational level of the precinct. Notes 1. All references to race/ethnicity will follow the guidelines set forth by the United States Census Bureau. 2. The Law Enforcement Management and Administrative Statistics reports that the national average number of separate and distinct precincts in large, municipal police organizations is four, which strengthens the external validity of the findings in the current study. 3. Formal organizational structure centers on two dimensions: structural complexity, which describes the way the organization slots, places, organizes or locates its work and workers; and structural control, which describes mechanisms by which an organization achieves coordination and control along its work and its workers (Maguire, 2003; Hall, 1991). Researchers studying informal organization structure, as it is referred here, seek to explain organizational behavior through examination of the interrelationships among organizational members, those who are being controlled and producing the organizational outputs. Informal organizational structure refers loosely to the culture (or cultures) of an organization (Thompson and Luthans, 1990, p. 319; Ott, 1989; Schein, 1985). 4. Statistical power analysis (SPA) was used to determine the appropriate sample size. SPA takes advantage of the relationship among four variables involved in statistics inference: sample size, significance criterion, population effect size and statistical power (Cohen, 1992; Weisberg et al., 1996). In this particular study, because this is the first empirical deductive

Variation in police patrol practices 273

PIJPSM 30,2

274

assessment of precinct-level variation, it is expected that data analyses would produce a large effect between each research group, or each precinct (Cohen, 1992). Additionally, in criminal justice research, the standard alpha, or significance level, is 0.05. As Cohen (1992, p. 158) reports, to compare mean responses at a large effect size across four groups using an alpha level of 0.05, the appropriate sample size is 18 cases (in each group). For a multivariate analysis, using the same effect size and alpha level, with nine independent variables, the sample size would need to be at least 50 cases. Furthermore, in determining the appropriate sample size it is important to keep the number of cases in each group similar. As Weisberg et al. (1996, p. 350) explain, “. . .studies in which the sizes of the groups examined are relatively similar are more powerful than those in which the sizes of the groups are markedly different.” For these reasons, the total sample for the quantitative analysis in the current study is 72 officer responses, or 18 officer responses in each precinct covering all patrol shifts. 5. Structured interviews were not completed with four participating officers; therefore, only 72 structured interviews were conducted and included in the analysis. 6. The analysis is restricted to 71 participating officers because one officer assigned to the SE precinct did not answer that particular question. References Barlow, D.E. and Barlow, M.H. (2000), Police in a Multicultural Society: An American Story, Waveland Press, Prospect Heights, IL. Black, D. (1971), “The social organization of arrest”, Stanford Law Review, Vol. 23, pp. 1087-111. Black, D. and Reiss, A.J. (1970), “Police control of juveniles”, American Sociological Review, Vol. 35, pp. 63-77. Cohen, J. (1992), “A power primer”, Psychological Bulletin, Vol. 112, pp. 155-9. Crank, J. (2003), “Institutional theory of police: a review of the state of the art”, Policing: An International Journal of Police Strategies & Management, Vol. 26 No. 2, pp. 186-207. Eitle, D., Stolzenberg, L. and D’Alessio, S.J. (2005), “Police organizational factors, the racial composition of the police, and the probability of arrest”, Justice Quarterly, Vol. 22 No. 1, pp. 30-57. Engel, R.S. (2001), “Supervisory styles of patrol sergeants and lieutenants”, Journal of Criminal Justice, Vol. 29 No. 4, pp. 341-55. Fine, G.A. (1984), “Negotiated orders and organizational cultures”, in Turner, R.H. and Short, J.F. (Eds), Annual Review of Sociology, Annual Reviews, Palo Alto, CA. Friedrich, R.J. (1977), “The impact of organizational, individual and situational factors on police behavior”, unpublished doctoral dissertation, University of Michigan, Ann Arbor, MI. Fyfe, J.J. (1979), “Administrative interventions on police shooting discretion: an empirical examination”, Journal of Criminal Justice, Vol. 7, pp. 303-23. Greene, J. (2000), “Community policing in america: changing the nature, structure, and function of the police”, Policies, Processes, and Decisions of the Criminal Justice System, Criminal Justice 2000, No. 3, pp. 299-370. Hall, R.H. (1991), Organizations: Structure, Processes and Outcomes, Prentice Hall, Englewood Cliffs, NJ. Hassell, K.D. (2004), “A cross-precinct analysis of police patrol practices and the negotiation of order among patrol” unpublished doctoral dissertation, University of Michigan. Hassell, K.D. (2006), Police Organizational Cultures and Patrol Practices, LFB Scholarly Publishing, New York, NY.

Hawley, A. (1950), Human Ecology: A Theory of Community Structure, Ronald Press, New York, NY. Hawley, A. (1986), Human Ecology: A Theoretical Essay, University of Chicago Press, Chicago, IL. Herbst, L.E. (2003), “The impact of new immigration patterns on the provision of police services in Midwestern communities”, unpublished dissertation, University of Nebraska, Omaha, NB. Hogelucht, K.S.B. and Geist, P. (1997), “Discipline in the classroom: communicative strategies for negotiating order”, Western Journal of Communication, Vol. 61, pp. 1-34. Holmes, S.A. (1996), “Census sees a profound shift in the US”, New York Times, March 17, p. A48. Kane, R.J. (2003), “Social control in the metropolis: a community-level examination of the minority group-threat hypothesis”, Justice Quarterly, Vol. 20 No. 2, pp. 265-95. Katz, C.M. (2001), “Establishment of a police gang unit: an examination of organization and environmental factors”, Criminology, Vol. 39 No. 1, pp. 37-73. Kelling, G.L. and Moore, M.H. (1988), The Evolving Strategy of Policing, Perspectives on Policing, National Institute of Justice, Washington, DC. Klinger, D. (1997), “Negotiating order in patrol work: an ecological theory of police response to deviance”, Criminology, Vol. 35 No. 2, pp. 277-306. Lundman, R.J. (1974), “Routine police arrest practices: a commonwealth perspective”, Social Problems, Vol. 32, pp. 127-41. Lundman, R.J., Sykes, R.E. and Clark, J.P. (1978), “Police control of juveniles: a replication”, Journal of Research in Crime and Delinquency, Vol. 15, pp. 74-91. Maguire, E.R. (2002), Organizational Structure in American Police Agencies: Context, Complexity and Control, State University of New York Press, Albany, NY. Maines, D.R. and Charlton, J.C. (1985), “The negotiated order approach to the analysis of social organizations”, Studies in Symbolic Interaction, Vol. 8, pp. 217-308. Mastrofski, S., Snipes, J.B., Parks, R.B. and Maxwell, C.D. (2000), “The helping hand of the law: police control of citizens on request”, Criminology, Vol. 38, pp. 307-42. Mastrofski, S.D., Worden, R.E. and Snipes, J.B. (1995), “Law enforcement in a time of community policing”, Criminology, Vol. 33 No. 4, pp. 539-63. National Research Council (2004), Fairness and Effectiveness in Policing: The Evidence, The National Academies Press, Washington, DC. Ott, J.S. (1989), The Organizational Cultural Perspective, Dorsey Press, Belmont, CA. Reed, M. (1985), Redirections in Organisational Analysis, Tavistock Publications, London. Ricksheim, E.C. and Chermak, S.M. (1993), “Causes of police behavior”, Vol. 21, pp. 353-82. Sampson, R.J. and Lauritsen, J.L. (1997), “Racial and ethnic disparities in crime and criminal justice in the United States”, in Tonry, M. (Ed.), From Ethnicity, Crime, and Immigration: Comparative and Cross-National Perspectives, University of Chicago Press, Chicago, IL, pp. 311-74. Schein, E.H. (1985), Organizational Culture and Leadership: A Dynamic View, Jossey-Bass, San Francisco, CA. Sherman, L. (1980), “The causes of police behavior: the current state of quantitative research”, Journal of Research in Crime and Delinqeuncy, pp. 69-100. Smith, D.A. (1984), “The organizational context of legal control”, Criminology, Vol. 22, pp. 19-38.

Variation in police patrol practices 275

PIJPSM 30,2

276

Smith, D.A. and Visher, C.A. (1981), “Street-level justice: situational determinants of police arrest decisions”, Social Problems, Vol. 29, pp. 167-77. Smith, D.A., Visher, C.A. and Davidson, L.A. (1984), “Equity and discretionary justice: the influence of race on police arrest decisions”, Journal of Criminal Law and Criminology, Vol. 75, pp. 234-49. Strauss, A. (1978), Negotiations: Varieties, Contexts, Processes and Social Order, Jossey-Bass, San Francisco, CA. Strauss, A. (1993), Continual Permutations of Action, Aldine de Gruyter, New York, NY. Strauss, A., Fagerhaugh, S., Suczek, B. and Wiener, C. (1985), The Social Organization of Medical Work, University of Chicago Press, Chicago, IL. Swatt, M.L. (2002), “Demeanor and arrest revisited: reconsidering the direct effect of demeanor”, Journal of Crime & Justice, Vol. 25 No. 1, pp. 23-39. Sykes, R.E. and Clark, J.P. (1975), “Theory of deference exchange in police-civilian encounters”, American Journal of Sociology, Vol. 81 No. 3, pp. 584-600. Terrill, W. (2001), Police Coercion: Application of the Force Continuum, LFB Scholarly Publishing, New York, NY. Thompson, K.R. and Luthans, F. (1990), “Organizational culture: a behavioral perspective”, in Schneider, E. (Ed.), Organizational Climate and Culture, Jossey-Bass, San Francisco, CA, pp. 319-44. Weisberg, H.F., Krosnick, J.A. and Bowen, B.D. (1996), An Introduction to Survey Research, Polling, and Data Analysis, 3rd ed., Sage, Thousand Oaks, CA. Wilson, W.J. (1987), Truly Disadvantaged: The Inner City, the Underclass, and Public Policy, University of Chicago Press, Chicago, IL. Worden, R.E. and Shepard, R.L. (1996), “Demeanor, crime, and police behavior: a reexamination of the police services study data”, Criminology, Vol. 34 No. 1, pp. 83-105. Further reading Black, D. (1980), The Manners and Customs of the Police, Academic Press, New York, NY. Lundman, R.J. (1974), “Domestic police-citizen encounters”, Journal of Police Science and Administration, Vol. 2 No. 1, pp. 22-7. Reuss-Ianni, E. (1993), Two Cultures of Policing, Transaction, New Brunswick, NJ. United States Bureau of the Census (2002), Racial and Ethnic Residential Segregation in the United States, 1980–2000, US Government Printing Office, Washington, DC. Visher, C.A. (1983), “Gender, police arrest decisions, and notions of chivalry”, Criminology, Vol. 21, pp. 5-28. Corresponding author Kimberly D. Hassell can be contacted at: [email protected]

To purchase reprints of this article please e-mail: [email protected] Or visit our web site for further details: www.emeraldinsight.com/reprints

The current issue and full text archive of this journal is available at www.emeraldinsight.com/1363-951X.htm

Race and police reliance on suspicious non-verbal cues

Race and police reliance

Richard R. Johnson Department of Criminal Justice, Washburn University, Topeka, Kansas, USA

277

Received 5 June 2006 Revised 25 October 2006 Purpose – This study seeks to evaluate the accuracy of non-verbal behaviors in differentiating Accepted 30 November 2006

Abstract

between criminals and innocent citizens in real-life police-citizen encounters, and evaluate the impact of race as a confounding influence on the display of these non-verbal behaviors. Design/methodology/approach – This study involved the frame-by-frame analysis of 240 videotaped interactions between citizens and police officers to determine the frequency with which citizens of different races and differing roles (offender versus non-offender) display specific non-verbal behaviors officers are trained to believe are indicators of suspicion. Findings – The findings suggest that the non-verbal cues of frequent speech disruptions, frequent or inappropriate smiles, the avoidance of eye contact, and increased hand gestures are poor indicators of criminal involvement and are strongly influenced by the race/ethnicity of the individual. Research limitations/implications – The videotaped police-citizen interactions on which this study was based were from a “reality television” show, thus preventing a random sample. Practical implications – This study would be useful to police officers who rely on the use of non-verbal cues, training personnel who instruct in the areas of non-verbal communication, and researchers who examine the potential causes of racial bias in the actions of police officers. Originality/value – While earlier studies have investigated this issue with laboratory experiments under artificial conditions, this study provides a test of the influence of race on non-verbal behavior during real life official police-citizen interactions in the field. Keywords Non-verbal communications, Police, Race, African Americans, Hispanics, United States of America Paper type Research paper

Introduction Police officers are often confronted with situations in which they have to make quick judgments about the motives and roles of people based on little information or previous contact. In handling disputes and crimes they must quickly assess who is a victim and who is an offender. One-tool officers use to help them disambiguate the roles and motives of the citizens they encounter is non-verbal cues. Survey research has revealed that police officers (Akehurst et al., 1996; Stromwall and Granhag, 2003; Vrij and Semin, 1996), correctional officers (Lakhani and Taylor, 2003; Vrij and Semin, 1996), parole officers (Porter et al., 2000), prosecutors (Stromwall and Granhag, 2003), and judges (Stromwall and Granhag, 2003) often perceive that frequent speech disruptions (such as stutters and long pauses), frequent or inappropriate smiles, the avoidance of eye contact, and increased hand gestures are interpreted as non-verbal indicators of nervousness related to involvement in crime, or that a person is trying to be deceptive. Yet how effective are these non-verbal behaviors at helping officers differentiate those engaged in criminal behavior from those who are not? Also, what influence do cultural differences play in the frequency with which these behaviors are displayed?

Policing: An International Journal of Police Strategies & Management Vol. 30 No. 2, 2007 pp. 277-290 q Emerald Group Publishing Limited 1363-951X DOI 10.1108/13639510710753261

PIJPSM 30,2

278

Research exists that suggests a person’s non-verbal behaviors vary significantly from one ethnic group to another (Vrij et al., 1992; Winkel and Vrij, 1990), perhaps making reliance on non-verbal cues racially or ethnically biased. The present study attempts to evaluate the accuracy of these non-verbal behaviors in differentiating between criminals and innocent citizens in real-life police-citizen encounters. The study also seeks to evaluate the impact of race or ethnicity as a confounding influence on the display of these non-verbal behaviors. Specifically, this study seeks to answer three questions. First, are differences in the non-verbal behaviors of speech disruptions, smiles, eye contact, and hand gestures significantly correlated with involvement in criminal activity? Second, are differences in the non-verbal behaviors of speech disruptions, smiles, eye contact, and hand gestures significantly correlated with race? Third, which characteristic (crime involvement or race) has a stronger relationship to these non-verbal behaviors? Developing suspicion The courts in the USA have acknowledged that the police may interpret the situations they encounter using all of the clues at their disposal. The circumstances, statements, and behaviors of the citizens they encounter may be evaluated for signs of suspiciousness, deception, or danger. In its decision in the case of Terry v. Ohio (1968), the US Supreme Court ruled that when a police officer observed unusual behavior by a citizen that caused the officer to suspect a crime was about to occur, the officer was permitted to briefly detain, and conduct a limited search, of the citizen. The Court ruled that officers might base their suspicions of a citizen’s conduct on their professional experience and past training. Similarly, in Carroll v. United States (1925), the Court ruled that searches of vehicles are permissible without a search warrant in situations where an officer had developed probable cause to believe that illegal activities were occurring. The development of probable cause involves facts and circumstances that would lead a reasonable person to believe that a crime is occurring or about to occur. These facts and circumstances may at times also include the interpretation of citizen behavior. One method used by police officers to detect the suspicious or criminal activities is non-verbal behavior. Surveys of patrol officers and detectives have revealed that many rely upon non-verbal behavioral cues to interpret citizen honesty. Akehurst et al. (1996) surveyed 60 police officers regarding their beliefs about suspicious behavioral cues. The majority of these officers agreed that when people are trying to be deceptive they tend to increase their involuntary speech disruptions and movements of the eyes, mouth, and hands. Specifically, these officers believed that when people are involved in criminal activities they increasingly stutter or take pauses while speaking; smile frequently or inappropriately; avoid eye contact with the person with whom they are speaking; and make frequent hand and arm gestures. The greater degree to which these behaviors occur, the more suspicious the officers stated they would be of the person they were encountering (Akehurst et al., 1996). Vrij and Semin (1996) surveyed 50 police officers (patrol officers and detectives) finding similar perceptions of suspicious non-verbal cues. These officers also interpreted increases in the frequency of smiles, speech disruptions, eye contact avoidance, and hand and arm gestures as suspicious and likely to indicate deception on the part of the speaker. Stromwall and Granhag (2003) surveyed 102 police officers and again found that they held identical

views about the deceptive meanings of frequent speech disruptions, smiles, gaze aversion, and hand gestures. The perception of a non-verbal behavioral indicator has been reinforced by formal police training. The most widely accepted interview and interrogation training provided to the police in the USA, the Reid Method of Interviewing and Interrogations, incorporates many techniques to determine the truthfulness of interviewee statements (Blair and Kooi, 2004). As part of this training, the Reid Method teaches officers to observe for non-verbal indicators of stress or deception, including fidgeting; shifting of body posture; frequent leg and foot movements; frequent hand gestures; avoiding eye contact when speaking; fast or stuttered speech; frequent pauses while speaking; excessive or inappropriate smiling; and profuse sweating (Blair and Kooi, 2004; Brown, 2001; Inbau et al., 2001). Another source of police training about non-verbal cues comes from highway drug interdiction training. In response to the problem of drug transportation by motor vehicle in the USA, a number of police agencies have trained their patrol officers in methods of detecting suspicious or criminal behavior while conducting traffic stops (Remsberg, 1997). These drug interdiction training programs often teach officers that frequent or inappropriate smiles, frequent speech disruptions, avoidance of eye contact, and increasingly frequent arm or leg movements are signs of nervousness and deception (Connors and Nugent, 1990; Remsberg, 1997). Social psychological research has provided support for the belief that such non-verbal behaviors are accurate clues to detecting deceptive activity. Miller et al. (1983) and deTurck and Miller (1990) observed interviews with college student test subjects who were instructed to make both truthful and deceptive statements in a laboratory environment. The students observed were found to exhibit significant increases in smiles, speech disruptions, gaze aversion, and body fidgeting while being deceptive. Fugita et al. (1980) used police experts trained in detecting deception to interview male college students who had been instructed to give both true and untrue information. While focusing on both the verbal and non-verbal behaviors of the students the interviewers were able to detect the deceptive and truthful statements at a rate better than chance. Horvath et al. (1994) used four experienced police investigators trained in the Reid Method to review videotaped interviews of college students being both truthful and deceptive. Relying on both verbal and non-verbal behaviors the four investigators detected truthful statements and lies at a rate significantly greater than chance. It is important to note, however, that all of these experiments took place in an artificial laboratory environment and involved college students in low-stakes deceptive activity. None of the lies told dealt with serious matters and the test subjects were not in danger of serious legal repercussions if their lies had been discovered. These experiments also lacked the stress involved with being the suspect of a crime, or the trepidation of facing a real police officer in an official contact in a street environment. Vrij et al. (2001) have questioned whether people in a low stress/low risk laboratory environment would behave in the same way as they would when confronted with a real official police interrogation or contact on the street. In fact, a number of intervening factors have been found to influence the frequency with which non-verbal behaviors are displayed. Stress (deTurck and Miller, 1985), the complexity of the lie being told (Vrij and Heaven, 1999), the opportunity to rehearse the lie (Miller et al., 1983), the interpersonal space between the interviewer and interviewee (Winkel et al., 1988), and

Race and police reliance

279

PIJPSM 30,2

280

personal awareness of these non-verbal cues (deTurck and Miller, 1990) have all been found to significantly influence the degree to which an individual displays these non-verbal indicators of deceit. Furthermore, race and ethnicity are a substantial influence on the baseline at which these suspicious non-verbal behaviors are displayed. Winkel and Vrij (1990) have drawn attention to the fact that the majority of experiments supporting the accuracy of non-verbal cues for detecting suspicious and deceptive behaviors have involved primarily Caucasian test subjects and have failed to control for the influence racial or cultural differences may pose. For example, Fugita et al. (1974) observed interviews involving Caucasian and African-American college students in a laboratory environment. They noted that compared to Caucasian students, African-American students engaged in less eye contact with the interviewer and exhibited more speech pauses. LaFrance and Mayo (1976), Smith (1983), and Ickes (1984) each observed non-verbal communication in conversational dyads of people of different races. All three of these studies found that, compared to Caucasians, African-Americans were significantly less likely to look at the other party and were significantly more likely to smile during their conversations. Winkel and Vrij (1990) conducted videotaped mock police interrogations of Caucasian students and students of Surinamese decent. The videotapes were analyzed for race differences in non-verbal behaviors and the results indicated that the test subjects from Surinam engaged in significantly more gaze aversion than the Caucasians. Vrij et al. (1992) replicated this finding with a new sample confirming that, compared to Caucasians, South Americans demonstrate significantly more gaze aversion, pauses in speech, and variety in voice pitch. These findings suggest that when compared to whites, non-whites display higher levels of these suspicious non-verbal behaviors as a baseline, and police officers are at risk of considering this normal behavior as suspicious. The present study seeks to determine the influence of both race and involvement in criminal behavior on the frequency of displaying the suspicious non-verbal cues of smiles, speech disruptions, gaze aversion, and hand gestures during real-life police-citizen encounters. Specifically, the present study seeks to test three hypotheses. First, do the frequency of smiles, speech disruptions, gaze aversion, and hand gestures during official police-citizen contacts differ significantly between those who are actively engaged in a crime and those who are not? Second, do the frequency of smiles, speech disruptions, gaze aversion, and hand gestures made during official police-citizen contacts differ significantly by the citizen’s race? Third, if race and involvement in crime are accurate predictors of these non-verbal cues, which is the stronger predictor? Methodology The ideal way to test these hypotheses would be through viewing real police-citizen encounters, preferably through the medium of videotape so that the citizen behaviors can be reviewed and closely analyzed, frame by frame. In order to collect a significant sample of high quality, videotaped interactions between citizens and police officers during real life official police contacts, the present study used video footage from the reality television show Cops. This popular television show involves camera crews who ride along with police officers from various major US cities. The camera crews film the

interactions of the officers with citizens as they unfold in real life. No actors or narration are involved in the television show. While the editing process of the show selects only police-citizen interactions that the producers deem would be of the greatest entertainment value to their viewers, such as those that involve arrests or violence (Hallett, 1996; Hallett and Powell, 1995), the interactions aired do involve real police-citizen contacts as they actually occurred. During the years 2002 and 2003, the researcher recorded 480 episodes of this television show on videotape from the Court TV and FOX Network channels. Each episode was reviewed for verbal interactions between police officers and citizens in which the citizen was visible from the front and from the waist up for at least 60 seconds of video footage. Viewing the citizen from this angle and for at least this duration allowed accurate observation of the citizens’ displays of non-verbal behaviors. Review of the recorded video footage identified 452 police-citizen interactions that met the above criteria. Of the 452 interactions identified, 223 involved a police interaction with what appeared to be an “innocent” citizen, based on the information shown during the video segment. These “innocent” citizens were not considered criminal suspects by the officers in the segment, but rather were either a crime victim, witness, or a recipient of some sort of non-crime related police service (e.g. parent of a lost child, person whose car was on fire, etc.). The remaining 229 interactions involved persons who were suspected of committing a crime. In 217 of these 229 interactions, evidence was present during the video segment to demonstrate that the suspect had committed a crime (but not necessarily the specific crime for which they were charged). Because arrest may not be an accurate indicator of criminality, cases were only included when evidence the suspect committed an offense was presented in the video segment. The evidence of guilt shown in these clips consisted of such things as witness statements, physical evidence, admissions of guilt, and/or the commission of the actual crime was caught on tape by the film crew. If such types of evidence were not presented the case was then excluded from the analysis. Due to the evidence presented to the viewer it could be determined with relative confidence that these 217 police-citizen interactions involved an interaction with a criminal offender who had just completed, or was in the process of committing, a crime. It is also important to note that all of the criminal suspect clips that were considered occurred before the suspect was placed in handcuffs and before an official arrest decision had been made by the officers, demonstrating the behavior that would have been observed by the officer while he/she was formulating an arrest decision. From this sampling frame of 440 police-citizen interactions (223 involving non-offending citizens and 217 involving suspected offenders), a stratified sample of cases was randomly selected based on two descriptive criteria – race and involvement in crime. Randomly selected from this sampling frame were interactions involving 80 African-Americans (40 non-offenders and 40 suspected offenders), 80 Caucasians (40 non-offenders and 40 suspected offenders), and 80 Hispanics (40 non-offenders and 40 suspected offenders). Because very few of the interactions in the sampling frame involved people of races other than these, only these three racial groups were included. This created a sample of 240 cases, evenly divided among the three races being analyzed and the two categories of involvement in crime.

Race and police reliance

281

PIJPSM 30,2

282

Table I provides the descriptive data on the citizens in the sample and reveals that the sample is not homogeneous in sex or age. It was decided, however, that there would be no need to control for sex or age since none of the previous literature reviewed had identified significant differences in the display of these non-verbal cues of suspicion by sex or age in adults. For every case in the sample the researcher reviewed each videotaped police-citizen interaction multiple times and measured four categories of non-verbal behavior that the literature has shown are important in police determinations of suspiciousness or deception. The number of smiles, the number of speech disruptions (defined as “ahs”, “ums” or stutters), the number of seconds of continuous eye contact, and the number of individual hand movements the citizen made while speaking with the officer, were recorded. Each of these four categories was then divided by the total number of seconds of footage in which the citizen was visible from the front and from the waist up, in order to determine the citizen’s average rate per second for each non-verbal behavior. While the minimum number of seconds of continuous clear footage for being included in the sampling frame was 60 seconds, the mean number of seconds of footage was 96, with a standard deviation of 42 seconds. After the rate per second figure was determined, it was then multiplied by 60 seconds in order to determine the citizen’s rate per minute for each non-verbal behavior – smiles, speech disruptions, seconds of eye contact, and hand gestures. Findings Table II reveals the means and other descriptive data for each of the four non-verbal behaviors under evaluation, displayed by the six-predictor variable conditions (three races £ 2 crime involvement categories). An initial review of this data for the entire sample, and each subcategory within the sample, suggests a large degree of individual

Table I. Sample descriptives

N

%

Sex: Male Female

148 92

61.7 38.3

Age: Teens 20’s 30’s 40’s 50’s 60’s 70’s

21 87 68 39 15 7 3

8.8 36.3 28.3 16.3 6.3 2.9 1.2

Race: African American Caucasian Hispanic

80 80 80

33.3 33.3 33.3

Committed crime: Yes No

120 120

50.0 50.0

Mean

SD

Min.

Max.

Race and police reliance

Smiles per minute: African American/no crime African American/crime Caucasian/no crime Caucasian/crime Hispanic/no crime Hispanic/crime Entire sample

1.98 1.30 1.03 0.53 0.24 0.00 0.77

3.45 1.55 1.32 0.96 0.46 0.00 1.87

0.00 0.00 0.00 0.00 0.00 0.00 0.00

13.34 10.69 3.87 3.48 1.54 0.00 3.87

283

Speech disturbances per minute: African American/no crime African American/crime Caucasian/no crime Caucasian/crime Hispanic/no crime Hispanic /crime Entire sample

2.41 1.75 2.55 2.62 3.39 2.62 2.27

3.45 2.09 2.60 1.74 3.35 2.37 2.42

0.00 0.00 0.64 0.56 0.41 0.85 0.00

7.50 10.69 9.68 6.49 12.00 9.35 13.34

Seconds of eye contact per minute: African American/no crime African American/crime Caucasian/no crime Caucasian/crime Hispanic/no crime Hispanic/crime Entire sample

32.80 36.16 45.60 47.70 35.85 29.70 34.94

13.62 10.97 16.64 6.55 10.58 12.81 15.05

7.00 8.40 0.86 4.34 7.03 0.00 0.00

49.41 45.00 24.55 36.30 52.00 41.00 52.00

Hand gestures per minute: African American/no crime African American/crime Caucasian/no crime Caucasian/crime Hispanic/no crime Hispanic/crime Entire sample

28.39 23.98 7.89 17.43 22.14 31.41 23.68

14.41 12.95 6.64 8.98 13.43 11.54 14.59

0.00 0.00 0.00 31.00 23.00 13.43 0.00

58.46 56.00 58.00 56.00 57.00 53.33 58.46

variation across cases. It appears from the standard deviations and ranges in Table II that there is tremendous variation between individuals within each sub-category of race and crime involvement in the display of the non-verbal behaviors being evaluated. For example, among the African-Americans not involved in crime, smiles had a range of more than 13 per minute, speech disruptions had greater than seven per minute, hand gestures had almost 58 per minute, and seconds of eye contact had a range of 49 seconds per minute. Similarly, large variations were also found among all the other subgroups. Vrij et al. (1997) have argued that because significant variation exists between individuals in how frequently they normally display certain non-verbal behaviors, without knowledge of each person’s baseline behavior it is extremely difficult to correctly identify whether or not they have increased or decreased non-verbal behaviors due to deceptive behavior. Based on the degree of individual differences

Table II. Comparisons of sample means

PIJPSM 30,2

284

within each subcategory in the descriptive statistics found in Table II, it appears that there is merit to this argument. Based on the police officer perceptions of suspicious non-verbal behaviors found in the literature (Akehurst et al., 1996; Stromwall and Granhag, 2003; Vrij and Semin, 1996), one would expect that the subjects who had engaged in criminal activity would display more smiles, speech disruptions, gaze aversion, and hand gestures than those who had not committed a crime. An initial comparison of the subcategory means by race in Table II found little support for these beliefs. Among Caucasians, the subjects who had committed criminal activity actually displayed fewer smiles and less gaze aversion, but more speech disruptions and hand gestures than those who had not committed a crime. Therefore, the behavior of the Caucasians in the sample only partially fit the perceptions held by police officers about suspicious non-verbal cues. Among African-Americans, the subjects who had committed criminal activity again displayed fewer smiles, speech disruptions, gaze aversion, and hand gestures than those who were not suspected of a crime. These findings for African-Americans are all in the exact opposite direction predicted by police officer beliefs and training about suspicious non-verbal behavior. The Hispanic subjects also displayed some non-verbal behavior differences that conflicted with police officer beliefs and training. The Hispanic subjects who had committed a crime again displayed fewer smiles and speech disruptions, but did display more gaze aversion and hand gestures than Hispanics who had not committed a crime. These raw differences of means begin to call into question the effectiveness of police beliefs in suspicious non-verbal behaviors. This is especially the case with regard to African-Americans. While the Caucasians in the sample displayed some of the behavior differences anticipated between criminal suspects and innocent citizens, the African-Americans displayed behavioral differences in the opposite directions anticipated for all four non-verbal behaviors measured. The Hispanic subjects also displayed a mixture of anticipated and unanticipated behavior differences. This initial comparison suggests that there are racial and ethnic differences in the display of the non-verbal cues believed to be indicative of deception and crime involvement. Comparisons of the means in Table II across race categories for the criminal offenders and the innocent citizens also reveals race differences in non-verbal behaviors unrelated to involvement in criminal activity. For example, among only the subjects who had not engaged in crime, the African-American subjects displayed more smiles, gaze aversion, and hand gestures than the Caucasian subjects. Likewise, among non-criminal subjects, Hispanics displayed more speech disruptions, gaze aversion, and hand gestures than Caucasians. Based on the beliefs held by officers and the training they receive, it is possible that the non-criminal African-Americans and Hispanics would appear more suspicious than Caucasians, simply due to these natural behavior differences. Nevertheless, the raw differences displayed in Table II may have been the result of chance and sampling error and therefore must be analyzed for the statistical significance of their differences. The means of the various subgroups were analyzed for true statistical differences through a two-way, multiple analysis of variance (MANOVA) test; the results of which are displayed in Table III. With regard to the non-verbal behavior of smiling, race was the only significant predictor of differences

Sum of squares Smiles per minute: main effects Race Suspected of a crime Interaction effect Total explained variation

80.356 1.901 7.991 90.248

Speech disruptions per minute: main effects Race 15.747 Suspected of a crime 22.632 Interaction effect 7.605 Total explained variation 1143.953

F

E-squared

14.121 * * * 0.668 1.404 6.344 * * *

0.1063 0.0025 0.0106 0.1194

7.874 4.823 * 0.810 4.786

Race and police reliance

285

0.0069 0.0198 0.0067 0.0402

Seconds of eye contact per minute: main effects Race 9057.013 Suspected of a crime 1049.933 Interaction effect 226.439 Total explained variation 10333.385

42.921 * * * 9.951 * * 1.073 19.588 * * *

0.2586 0.0230 0.0065 0.2951

Hand gestures per minute: main effects Race 9150.472 Suspected of a crime 1378.850 Interaction effect 1439.264 Total explained variation 11968.587

35.227 * * * 10.616 * * 5.541 * * 18.430 * * *

0.2160 0.0326 0.0340 0.2825

Notes: Significance levels: * p , 0.05; * * p , 0.01; * * * p , 0.001

across sub-categories. Whether or not the person had committed a crime was not a significant predictor, suggesting that the non-verbal cue of smiling was not a valid clue to engagement in criminal behavior but result of racial/ethnic differences. The E-square proportional reduction in error measure revealed that almost 11 percent of the individual variation in smiling behavior was accounted for by the subject’s race. Less than 1 percent was explained by involvement in crime, and even this could have been by chance due to the fact this variable was not statistically significant. In the case of speech disruptions, this non-verbal behavior was significantly different across criminal offenders and innocent citizens. However, involvement in crime only explained approximately 2 percent. Also, as was seen in Table II, the differences in speech disruptions related to crime involvement differed in direction by race. While Caucasian criminals appeared to display slightly more speech disruptions than non-criminal Caucasians, the pattern was reversed for African-Americans and Hispanics, with criminal suspects displaying fewer speech disruptions than non-offending citizens. For the non-verbal behavior of eye contact, both race and involvement in crime were significant predictors, and together explained almost 30 percent of the variation in this behavior. However, the E-squared values revealed that the individual’s race explained about 26 percent of the variation and involvement in crime only accounted for about 3 percent of the variation. Also, as revealed in Table II, the Caucasians and African-Americans actually engaged in more eye contact when involved in crime than when innocent – the opposite of what police training and officer beliefs would lead one to expect.

Table III. MANOVA analysis by race and involvement in crime

PIJPSM 30,2

286

Finally, with regard to hand gestures displayed, race and involvement in crime were both significant predictors, and an interaction effect between these two variables was also revealed. Combined, all three influences explain more than 28 percent of the individual variation in this non-verbal behavior. However, the individual E-square values again indicated that race was by far the greatest predictor in the model, by itself explaining almost 22 percent of the individual variation. Also, while involvement in crime did explain about 3 percent of the variation, the directional differences between criminals and non-criminals was again in the opposite direction anticipated for African-Americans. While Caucasian and Hispanic criminal suspects displayed slightly more hand gestures than non-criminals of these races, African-American criminal suspects demonstrated slightly fewer hand gestures than non-criminals of this race. Discussion The previous literature suggested that police officers and other criminal justice professionals widely perceive that increases in smiles, speech disruptions, hand gestures, and avoidance of eye contact suggest deceptive or criminal behavior (Akehurst et al., 1996; Stromwall and Granhag, 2003; Vrij and Semin, 1996). The previous literature has also demonstrated that these beliefs are reinforced by police investigative training in the USA (Blair and Kooi, 2004; Inbau et al., 2001; Remsberg, 1997). The analysis here attempted to test the validity of these beliefs under field conditions involving official police-citizen contacts between officers and either criminal offenders or non-offenders. Based on the analysis here there is little evidence to support the effectiveness of these non-verbal cues in differentiating between criminals and non-criminals under the actual field conditions of official police-citizen interactions on the street. Although not tested directly, individual variation across individuals within each sub-category was great. Previously,Vrij et al. (1997) had suggested that individuals significantly vary in their baseline frequencies for displaying the non-verbal behaviors reviewed here, causing Vrij to question the ability to detect changes in the frequencies of these behaviors unless one first is familiar with the individual’s baseline behavior. When police officers encounter citizens during traffic stops, or at the scene of calls, it is unlikely that they are intimately familiar with the baseline non-verbal behaviors of many of the citizens they contact. Therefore, individual differences between persons in how frequently they naturally display non-verbal behaviors may be a confounding influence that makes it difficult or impossible for a police officer to accurately detect the deceptive non-verbal cues of a stranger. Race also appeared to be a confounding variable of considerable importance. Race was a significant predictor of individual differences in three of the four non-verbal cues evaluated, and when race was significant it explained between seven and 13 times more variation than did involvement in crime. Specifically, African-Americans who were not involved in crime generally displayed much higher levels of these suspicious non-verbal cues than did Caucasians, even Caucasians who were involved in criminal activity. Likewise, the Hispanics not involved in crime also displayed more speech disruptions, gaze aversion, and hand gestures than both categories of Caucasians. This finding is consistent with the previous research by Winkel and Vrij (1990) and Vrij et al. (1992) which found that police officers viewed the non-verbal behaviors of

non-Caucasians more suspiciously than the non-verbal behaviors of Caucasians in mock interrogations. This suggests that if police officers are looking for elevated levels of the non-verbal behaviors analyzed here to help them differentiate who is, and is not, involved in crime, they are far more likely to find these elevated levels in law-abiding African-Americans and Hispanics than they are to find them in Caucasians involved in crime. Therefore, the reliance on these non-verbal cues may be resulting in increased levels of undue suspicion by the police toward law-abiding African-American and Hispanic citizens. Not only was the influence of involvement in crime on these non-verbal behaviors very minor, but when it was significant, the behavior differences were often in the opposite directions anticipated by police officers. While police officer beliefs and formal police training suggests that people will display more frequent smiles when attempting to cover up criminal behavior, the findings here suggest that for all three race groups the offenders smiled less frequently than the innocent citizens. Officers anticipate more speech disruptions from guilty parties but in the case of the African-American and Hispanic subjects in this study, the offenders made fewer speech disruptions than the non-offenders. Officers expect criminals to avoid eye contact more frequently, but in the findings here for African-Americans and Caucasians the non-offenders avoided eye contact more frequently than the offenders. Finally, police officers expect guilty citizens will make more frequent hand gestures, yet in this study African-American offenders made fewer hand gestures than African-American non-offenders. Therefore, even after controlling for the significant baseline differences of the three race groups, the non-verbal cues officers use appear extremely faulty and inconsistent with real behavior under field conditions. Another important point is that the models evaluated here never explained more than 30 percent of the variation across individuals. While 30 percent is an important amount of variation to explain, even after accounting for sampling and measurement error it does indicate that a substantial majority of the variation was unexplained by race or involvement in crime. The literature has offered a number of other possible confounding influences that could have impacted these non-verbal behaviors. As was mentioned earlier, Vrij et al. (1997) identified the significance of individual baseline differences. Other confounding influences identified have been the complexity of the lie being told, the opportunity to rehearse the lie, the interpersonal space between the parties who are speaking, personal awareness about – and control of – suspicious non-verbal cues, and stress (deTurck and Miller, 1985; Miller et al., 1983; Vrij and Heaven, 1999; Winkel et al., 1988). These have all been found to significantly influence the degree to which an individual displays these non-verbal indicators of suspicion. It is impossible to determine to what extent these factors influenced the non-verbal behaviors displayed in this study. Likewise, it would probably be equally as difficult for a police officer to determine what factors are truly having the most influence on a citizen’s non-verbal behaviors. It is also important to remember that the previous evaluations of the effectiveness of these non-verbal cues have specifically related to how well they revealed deceptive statements. However, police officers have been trained to use these non-verbal indicators more broadly, as a way to differentiate between criminal and non-criminal citizens (Connors and Nugent, 1990; Remsberg, 1997). It is impossible to determine if all of the criminal suspects in the sample were making deceptive statements, or if all of the

Race and police reliance

287

PIJPSM 30,2

288

non-suspects were being entirely truthful. It is possible that the innocent citizens, while not engaged in crime, were being deceptive in some way. Therefore, truly deceptive statements, regardless of criminal involvement, may also have been an intervening variable to be considered as part of the unexplained variance. Conclusion Due to the inherently difficult and dangerous nature of police work it is understandable that police officers would use all of the information at their disposal to interpret the people and circumstances that they encounter. Nevertheless, it is crucial, for both the fair treatment of citizens and the safety of officers, that the information officers use to interpret situations be as accurate as possible. In the present study, observations of real-life police-citizen interactions found the display of the “suspicious” non-verbal behaviors of frequent smiles, speech disruptions, gaze aversion, and frequent hand gestures to be ineffective and erroneous indicators of involvement in crime. The present study supported the previous literature in suggesting that individuals vary greatly in the frequency with which they display the non-verbal behaviors that police officers and other criminal justice professionals perceive as indicators of criminal suspiciousness or deception. It also suggested that the difference in non-verbal behavior between criminals and non-criminals is minor and often inconsistent with the differences expected based on police officer training and beliefs. Finally, it also supported the previous literature suggesting that the citizen’s race poses a significant influence on the frequency with which these non-verbal behaviors are displayed, with race being a far stronger influence than crime involvement. Based on these findings, and the previous literature in this area, it is difficult to believe that under the true field conditions of an official police-citizen interaction, a police officer could accurately detect deceptive or criminal behavior through the use of these four non-verbal cues. Therefore, police agencies should reconsider the training of officers to detect suspicious behavior through smiles, speech disruptions, avoidance of eye contact, and hand gestures. As long as the risk exists that police officers are being unwittingly taught to interpret the normal non-verbal behaviors of African-Americans and Hispanics as suspicious, perhaps this training should be discontinued entirely. Since the literature has suggested that a large number of police officers, corrections officers, prosecutors, and judges also rely on these same non-verbal cues to help them make determinations about clients in the criminal justice system, efforts should be made to educate these professionals about the influence of race and culture on non-verbal behaviors. Finally, the influence of non-verbal behaviors on officer decision-making during police-citizen interactions needs further study in order to determine to what extent officers rely on these non-verbal behaviors, as opposed to other forms of information, when in the field. References Akehurst, L., Kohnken, G., Vrij, A. and Bull, R. (1996), “Lay persons’ and police officers’ beliefs regarding deceptive behavior”, Applied Cognitive Psychology, Vol. 10 No. 4, pp. 461-71. Blair, J. and Kooi, B. (2004), “The gap between training and research in the detection of deception”, International Journal of Police Science and Management, Vol. 6 No. 2, pp. 77-83. Brown, M.F. (2001), Criminal Investigation: Law and Practice, 2nd ed., Butterworth-Heinemann, Boston, MA.

Carroll v. United States (1925), “267 US 132”. Connors, E.F. and Nugent, H. (1990), Street-Level Narcotics Enforcement, US Bureau of Justice Assistance, Washington, DC. deTurck, M. and Miller, G. (1985), “Deception and arousal: isolating the behavioral correlates of deception”, Human Communication Research, Vol. 12 No. 2, pp. 181-201. deTurck, M. and Miller, G. (1990), “Training observers to detect deception: effects of self-monitoring and rehearsal”, Human Communication Research, Vol. 16 No. 4, pp. 603-20. Fugita, S., Hogrebe, M. and Wexley, K. (1980), “Perceptions of deception: perceived expertise in detecting deception, successfulness of deception and non-verbal cues”, Personality & Social Psychology Bulletin, Vol. 6 No. 4, pp. 637-43. Fugita, S., Wexley, K. and Hillery, J. (1974), “Black-white differences in non-verbal behavior in an interview setting”, Journal of Applied Social Psychology, Vol. 4 No. 4, pp. 343-50. Hallett, M. (1996), “Police victimization from law and order: gun and roses on COPS”, The Criminologist, Vol. 21 No. 1, pp. 1-10. Hallett, M. and Powell, D. (1995), “Backstage with COPS: the dramaturgical reification of police subculture in American crime info-tainment”, American Journal of Police, Vol. 14 No. 1, pp. 101-29. Horvath, F., Jayne, B. and Buckley, J. (1994), “Differentiation of truthful and deceptive criminal suspects in behavior analysis interviews”, Journal of Forensic Sciences, Vol. 39 No. 3, pp. 793-807. Ickes, W. (1984), “Compositions in black and white: determinants of interaction in interracial dyads”, Journal of Personality & Social Psychology, Vol. 47 No. 2, pp. 330-41. Inbau, F.E., Reid, J.E., Buckley, J.B. and Jayne, B.P. (2001), Criminal Interrogation and Confessions, Williams and Wilkins, Baltimore, MD. LaFrance, M. and Mayo, C. (1976), “Racial differences in gaze behavior during conversations: two systematic observational studies”, Journal of Personality & Social Psychology, Vol. 33 No. 5, pp. 547-52. Lakhani, M. and Taylor, R. (2003), “Beliefs about the cues to deception in high- and low-stake situations”, Psychology, Crime & Law, Vol. 9 No. 4, pp. 357-69. Miller, G., deTurck, M. and Kalbfleisch, P. (1983), “Self-monitoring, rehearsal, and deceptive communication”, Human Communication Research, Vol. 10 No. 1, pp. 97-117. Porter, S., Woodworth, M. and Birt, A. (2000), “Truth, lies, and videotape: an investigation of the ability of federal parole officers to detect deception”, Law and Human Behavior, Vol. 24 No. 6, pp. 643-58. Remsberg, C. (1997), Tactics for Criminal Patrol: Vehicle Stops, Drug Discovery, and Officer Survival, Calibre Press, Northbrook, IL. Smith, A. (1983), “Non-verbal communication among black female dyads: an assessment of intimacy, gender, and race”, Journal of Social Issues, Vol. 39 No. 3, pp. 55-67. Stromwall, L. and Granhag, P. (2003), “How to detect deception? Arresting the beliefs of police officers, prosecutors and judges”, Psychology Crime & Law, Vol. 9 No. 1, pp. 19-36. Terry v. Ohio (1968), “392 U.S. 1”. Vrij, A., Akehurst, L. and Morris, P. (1997), “Individual differences in hand movements during deception”, Journal of Non-verbal Behavior, Vol. 21 No. 2, pp. 87-102. Vrij, A., Dragt, A. and Koppelaar, L. (1992), “Interviews with ethnic interviewees: non-verbal communication errors in impression formation”, Journal of Community & Applied Social Psychology, Vol. 2 No. 3, pp. 199-208.

Race and police reliance

289

PIJPSM 30,2

290

Vrij, A., Edward, K. and Bull, R. (2001), “People’s insight into their own behavior and speech content while lying”, British Journal of Psychology, Vol. 92 No. 2, pp. 373-89. Vrij, A. and Heaven, S. (1999), “Vocal and verbal indicators of deception as a function of lie complexity”, Psychology, Crime & Law, Vol. 5 No. 3, pp. 203-15. Vrij, A. and Semin, G. (1996), “Lie experts’ beliefs about non-verbal indicators of deception”, Journal of Non-verbal Behavior, Vol. 20 No. 1, pp. 65-80. Winkel, F., Koppeleaar, L. and Vrij, A. (1988), “Creating suspects in police-citizen encounters: two studies on personal space and being suspect”, Social Behavior, Vol. 3 No. 3, pp. 307-18. Winkel, F. and Vrij, A. (1990), “Interaction and impression formation in a cross-cultural dyad: frequency and meaning of culturally determined gaze behavior in a police interview setting”, Social Behavior, Vol. 5 No. 5, pp. 335-50. Further reading Vrij, A. and Mann, S. (2001), “Telling and detecting lies in a high-stake situation: the case of a convicted murderer”, Applied Cognitive Psychology, Vol. 15 No. 2, pp. 187-203.

To purchase reprints of this article please e-mail: [email protected] Or visit our web site for further details: www.emeraldinsight.com/reprints

The current issue and full text archive of this journal is available at www.emeraldinsight.com/1363-951X.htm

Are causes of police stress global? Testing the effects of common police stressors on the Turkish National Police Hasan Buker Minot State University/Turkish Institute of Police Studies, Minot, North Dakota, USA, and

Filip Wiecko

Are causes of police stress global? 291 Received 9 February 2006 Revised 17 May 2006 Accepted 13 November 2006

Washington State University, Washington, USA Abstract Purpose – This study aims to assess the effects of commonly examined police stressors’ on the members of a developing country’s centralized police department: Turkish National Police (TNP). Design/methodology/approach – This study is based on a data collected through a self-administered survey among the members of the TNP during the summer of 2005 (n ¼ 812). Using multivariate level OLS regression models, predicting effects of commonly examined police stressors on the participants’ stress levels are analyzed. Findings are evaluated in comparison to existing literature about police stress. Findings – This study indicates that organizational issues are the most important causes of stress in policing. Besides, it was found that several police stressors, as found for local police departments, might not be having the same effects for larger, centralized police departments. Practical implications – Modern policing can be a less stressful job if the police organizations take necessary steps towards applying modern management techniques at both macro and micro levels. Demographic differences, danger at work, or workload should not be counted as predictors of stress in policing without a through consideration of organizational matters. Originality/value – This is the first study empirically and systematically assessing the issue of stress among the members of the TNP. In addition, it is one of the rare studies published in English regarding the issue of police stress in a developing country. Keywords Police, Stress, Turkey Paper type Research paper

Recent studies on occupational stress agree that policing contemporary society is stressful (see Davidson and Veno, 1978; Petrone and Reiser, 1985; Reese, 1986; Violanti, 1985; Webb and Smith, 1980). In general, exposure to danger and violence, exercising discretion under critical circumstances, and other specific characteristics of police work, (i.e. excessive and irregular working hours) along with police organizations themselves, are factors comprising the phenomenon of “police stress” (Conroy and Hess, 1992; Crank and Caldero, 1991; Violanti and Aron, 1994). Societies’ expectations of police are also changing with the advent of community oriented policing in recent years. Police are expected to place the duty of serving society in front of protecting it. The authors would like to thank to Dr Nicholas Lovrich and the Department of Governmental Studies and Services of Washington State University, Dr Solomon Zhao, Dr Travis Pratt and the reviewers who provided valuable guidance and feedback during the development of this study.

Policing: An International Journal of Police Strategies & Management Vol. 30 No. 2, 2007 pp. 291-309 q Emerald Group Publishing Limited 1363-951X DOI 10.1108/13639510710753270

PIJPSM 30,2

292

This change in policing may be another important factor contributing to police stress (Aron, 1992; Zhao et al., 2003a). Recent studies exploring police stress focus on determining specific stressors and their effects on stress (Lawrence, 1984; Malloy and Mays, 1984). Several stressors, such as organizational characteristics, jurisdiction characteristics, and personal differences, are identified as specific causes of occupational stress in policing. These studies generally examine police organizations in the USA. Most of these studies focus on local organizations and have thus far omitted state and federal law enforcement agencies (see Newman and Rucker-Reed, 2004). Only a limited number of studies published in English examine the variety of contributors to police stress in countries outside of the USA (see Berg et al., 2005; Davidson and Veno, 1979; Coman and Evans, 1991; Lee, 2002; Stansfield, 1996; Zukauskas et al., 2001). Yet, there is a lack of information addressing whether generally accepted stressors have the same effects in other nations’ police organizations as they do in local police organizations in the USA. In this study, commonly accepted police stressors are examined in the context of the Turkish National Police (TNP), a centralized and democratic police organization. The TNP performs law enforcement tasks regarded as normal in a democratic country, such as detection and prevention of crime in the interest of public safety, and serving the members of the community in several areas of the daily life (Skolnick, 1999). Studying the effects of general police stressors identified in current literature in the context of the TNP stands as an important matter for the general policing studies because Turkey is a developing country. This development process has brought the Turkish society under a social change and strain along with the process of the European Union membership (Jacoby and Mann, 2004). Bennet (1997, p. 296) stated the importance of examining the administrative issues of policing in a developing society as follows: With democratization and human rights rapidly becoming international imperatives, knowledge about how to increase receptivity to change in traditional policies and operations, reduce costly turnover in personnel, and increase the incidence of cooperative behavior is necessary. Most police forces in the developing world operate with very limited financial and personnel resources, so they must understand how to use those resources most effectively and efficiently. At the same time, developing nations . . . are experiencing dramatic increases in crime, and particularly crimes of violence, that challenge their established ways of operating.

Bennet’s statements regarding developing countries are valid for Turkey as well. Experiencing dramatic increases in crime, more opportunities for people to react to the numerous social problems in democratic ways (legal demonstrations, etc.) have produced significant challenges for the Turkish police (TNP, 2005). With this changing social context in mind, studying the effects of generally accepted stress-causing factors in American and other countries’ police forces on members of the TNP presents a valuable opportunity. For those interested in international policing, the results of this study provide important insight for evaluating the soundness of factors commonly accepted as being the causes of stress in police departments. This study, in addition, is an important step for the TNP because no prior research, empirically evaluating the level of stress among its members, exists[1]. Existing studies on police stress in Turkey provide a cursory examination and do not provide empirical measure of stress levels among the members of the TNP (Yesilorman, 2003). The following study provides an analysis of data from the TNP along with a presentation of findings. The final section discusses these findings in comparison to

the existing literature and provides practical and theoretical implications for further studies in policing. Causes of police stress In order to operationalize causes of stress in police work (i.e. danger, violence, bureaucracy, administration, etc.), some studies specify four basic factors driving stress in police organizations: the nature of police work, the public, the criminal justice system, and the characteristics of the organization (Reese, 1986; Violanti and Aron, 1994). Other studies based on officers’ perceptions, group these factors in two categories; the stressors stemming from organizational factors and those rooted in the nature of police work (Aron, 1992; Crank and Caldero, 1991; Martelli et al., 1989; Morash et al., 2006; Spielberger et al., 1981; Swanson et al., 1998; Violanti and Aron, 1995). The first group has emerged over the latter one as being more effective in determining the level of stress among police officers (see Violanti and Aron, 1995; Morash et al., 2006). Stressors rooted in organizations typically relate to administrative policies and attitudes that are bothersome to officers (Violanti and Aron, 1993, Violanti and Aron, 1995). Commonly, these stressors are supervision (management) practices, promotion practices, and work conditions (Aron, 1992; Brown and Campbell, 1990; Storch and Panzarelle, 1996). More specifically, these stressors include: . lack of participation in decisions affecting daily organizational practices; . lack of administrative support; . a punishment centered philosophy or unfair discipline (Comman and Evans, 1991; Ellison, 2004; Kroes, 1986; Reiser, 1974); and . alienation among police officers resulting from the quasi-militaristic nature of the organization (Golembiewski and Kim, 1991). Brooks and Piquero (1998) report department size is an important predictor of how these organizational practices affect stress. Officers feel the effects of these stressors less in small departments than in large urban departments. Second to the organizational causes, findings suggest the nature of police work is an important contributor to the presence of stressors. Police work, by its nature, is dangerous, requires working in shifts, receives public apathy, is mostly boring, and requires exposure to misery and death. Taken together, these factors are potential predictors of stress in policing (see Aron, 1992; Ellison, 2004; Spielberger et al., 1981; Violanti and Aron, 1993). On the other hand, the status of the nature of police work, as a stressor has been problematic. Storch and Panzarelle (1996), and Kroes and Gould (1982), for instance, note police officers rarely mention danger, violence, or human misery as sources of stress in their work environment, as long as questionnaires do not specifically ask about the existence or the extent of these factors. Threats of physical harm and exposure to violence are not common for many small-sized police departments. Police officers, therefore, might be describing their work as being more dangerous than it actually is (Cullen et al., 1983). Some scholars argue officers and departments might be exaggerating the issue of stress in police work. Golembievski and Kim (1991), for instance, report no significant stress symptoms based on a standardized measure, although they find a large number of stressors commonly accepted as the sources of stress. Brown and Campbell (1994,

Are causes of police stress global? 293

PIJPSM 30,2

294

p. 5) state “there remains the possibility that some aspects of the ‘police stress’ issue might be socially constructed for certain political purposes rather than reflecting true concerns for health and welfare”. Supporting this idea, Ellison (2004, p. 50) argues these political purposes “might include malingering in order to claim or exaggerate occupationally related ill-health, the use of stress as a vehicle for vested interests, including those of the unions and a stress bandwagon”. She adds, “stress is one of the current fads and departments that wish to be seen as innovative may wish to be seen as promoting remedial measures” (Ellison, 2004, p. 50). General demographics, such as age, gender, race, marital status, experience in the department, rank, assignment type, being commissioned or non-commissioned, have also been included in most police stress-related studies in order to assess their direct and indirect effects (see Aron, 1992; Crank and Caldero, 1991; Newman and Rucker-Reed, 2004; He et al., 2002; Violanti and Aron, 1995; Zhao et al., 2003b). Being a racial, ethnic and gender minority, for instance, is reported to be a possible stressor for police officers (Kanter, 1977; Krimmel and Gormley, 2003). These demographic variables, however, are insignificant determinants of stress for police officers when considered together with organizational characteristics and other stressors (Morash et al., 2006; Zhao et al., 2003b). As provided above, the literature on stress in policing provides extensive information regarding what causes police stress. Most of the current studies on this topic examine the issue on local police agencies in the USA or other nations. One important goal of this research is to test the validity of these factors on a centralized police department. Method Data The data used for this study comes from a survey administered among the members of the TNP in the summer of 2005. In order to conduct the survey, the responsible unit of the department’s headquarters was informed about the research goals, target population, and survey questions. Anonymity and volunteer participation rules of the survey were stressed during the approval process. Upon the approval of the department officials, the survey was delivered to the target jurisdictions. Since the department is a national, centralized unit, the survey was conducted in seven different jurisdictions in different geographical regions of the country. Each of these jurisdictions has different populations, crime rates, number of officer, and growth rates. In each of these jurisdictions, the personnel were informed about the survey’s intent and coverage. In addition, volunteer participation and the anonymity of the survey results were emphasized. Civilian officers, police officers and mid-level supervisors were invited to participate in the survey. The researchers administered the surveys in specially reserved rooms for those who voluntarily agreed to participate. Again, participants were assured of the anonymity of their survey responses and their right to leave the room without completing the survey at any time. From 1,015 survey forms, 812 were filled and returned appropriately, constituting an 80 percent response rate. This survey is a revised replication of a survey conducted by Washington State University, Division of Governmental Studies and Services for the Spokane Police Department (Zhao et al., 2003a). It was necessary to remove some items from the original survey because of translation problems from English to Turkish. The final version of the survey was checked by three different social scientists in the subject

country other than the researcher himself to assure the survey questions are identical to the original version. The survey was tested on 25 practice participants asked to indicate any vague points they discovered. Dependent variable Defining the main dependent variable of this study, stress, has been problematic. Previous studies consider stress as a type of illness imposed on a person from an outside source, causing physical or psychological discomfort to a certain extent (Brown and Campbell, 1990). Sager and Wilson (1995), however, report there are 15 different definitions of stress in the literature, making it a difficult construct to study. Following Jex and Beehr’s (1991, p. 312) suggestion to use of the term stress as referring “to the entire field of inquiry rather than any particular variable or phenomenon”, recent police stress studies (Ellison, 2004; Jaramillo et al., 2004), and industrial and organizational psychology literature refer to stress as a field of inquiry rather than a particular construct. This study, as well, follows this recent trend by using the term stress as a “field of inquiry that investigates the relationship between job stressors and strains” (Jaramillo et al., 2004, p. 323). Stress levels of the participants are measured with the Brief Symptom Inventory (BSI) scale, which is a brief form of the Symptom Check List 90 (Derogatis and Melisaratos, 1983). Psychometric validity of the BSI has been tested and sustained in several empirical studies (see Derogatis and Savitz, 1999). Several studies on police stress use this scale’s original and modified forms (see Aaron, 2000; He et al., 2002; Zhao et al., 2003b) In this study, we use a summed scale for stress based on the BSI. Some studies focus on specific proxies for stress such as somatization, depression, or anxiety. The rationale behind this strategy lies in the fact these three proxies constitute three quarters of the psychiatric conditions seen in the US health care system (Derogatis and DellaPietra, 1994). In the same context, several studies (see He et al., 2002; Zhao et al., 2002) base their examinations of police stress on these symptoms. Furthermore, some studies treat these three proxies as separate dependent variables, but in the case of our data, the correlations between the three are so high (r . 0.7) they are functionally the same; making three separate analyses redundant. Instead, we treat stress as a summed scale of 43 of 49 items in the BSI. In total, the BSI provides 49 symptoms of stress. Each of these symptoms is rated on a five-point scale ranging from 0 (never) to four (extremely). A factor analysis was conduct on the 49 measures and results indicated 43 variables, which loaded on one construct (Eigen value ¼ 14.194) with high factor loading (above 0.4). Six variables, however, cross-loaded on three other components and were removed from the summed scale. The 43 items used are highly reliable (alpha ¼ 0.949). Independent variables Control variables. We include individuals’ demographic characteristics in order to control for effects of several other factors causing stress in policing. Typical control variables such as, gender, rank, marital status, education status, and years in service are included in the analysis to account for the effects these variables have on the other explanatory variables. In addition, the analysis covers jurisdiction characteristics to test whether other explanatory variables act differently among these demographics as suggested by several studies on centralized national police setting (see Brooks and Piquero, 1998). In

Are causes of police stress global? 295

PIJPSM 30,2

296

total, seven jurisdictions of the TNP, of different sizes and different geographical regions of the country, are included in the data. Based on the information about several demographics of jurisdictions, such as number of crimes, number of officers, population, and growth rate, two composite measures are used to assess jurisdiction effects: crime/population rate (per 1,000 people) and crime/officer rate (per 100 officers). The first measure is associated with the total population of the jurisdiction and reflects the department size, as well as the social status within jurisdiction in terms of crime. The intent of the latter measure is to reflect the workload within the jurisdiction. Officer number reflects the number of sworn police officers in the department as of 2004. The number of crimes refers to the number of cases the police department officially dealt with in 2004[2]. Finally, the jurisdiction population reflects the number of residents living within the boundaries of the actual jurisdiction. All descriptive analyses are displayed in Table I. Explanatory variables. Job motivation: This survey measures job motivation with Hackman and Oldham’s (1980) Job Diagnostic Survey (JDS). They find employees’ perceptions about their work are associated with their job satisfaction and commitment. This scale taps into some dimensions of work characteristics also identified as being important factors in organization-related causes of the occupational stress in policing (see Morash et al., 2006; Zhao et al., 2002). These characteristics are the meaningfulness of work (employees’ perceiving the work as worthwhile and important), and the knowledge about the results of the work (employees’ perception about the feedback they are given about their effects of their work). We assess Meaningfulness with the perceptions of skill variety, task significance, and task identity in work. Finally, knowledge reflects perceptions regarding the adequacy of feedback coming from superiors or colleagues regarding work efforts. In summary, the four dimensions reflecting the motivation level of officers work are: skill variety (a ¼ 0.65), task significance (a ¼ 0.58), task identity (a ¼ 0.56), and feedback (a ¼ 0.67)[3]. Note that the alphas in some cases are not as large as one would desire. Hackman and Oldham (1980) themselves state that the job characteristics, as measured by JDS, are not independent of one another. When a job rates high on some characteristics, it also tends to be high on some other dimensions. They state this issue as a limitation to their measurement, and recommend approaching the results cautiously. In our study, these composite dimensions are included, because previous research has consistently included and accepted them as valid measures of job motivation levels (see Barr et al., 1978 for a review, and Zhao et al., 2002; Zhao et al., 2003 for recent examples). Job satisfaction: There are several types of departmental policies mentioned as potential stressors in policing (Violanti and Aron, 1993, 1995; Zhao et al., 2002). Among these factors are supervision methods and promotion practices. Studies report job content itself is as an important variable in determining stress among police officers (Aron, 1992; Brown and Campbell, 1990; Storch and Panzarella, 1996). The data determines the participants’ satisfaction levels on these issues based on Smith et al. (1969) job satisfaction scale. In the analyses, we use three dimensions of job satisfaction in order to explain changes in the participants’ levels of clinical stress: . satisfaction with job content (a ¼ 0.68); . satisfaction with supervisor (a ¼ 0.72); and . satisfaction with co-workers (a ¼ 0.80).

73.5 22.9 2.7 0.9 38.1 22 22.3 11.8 5.8 3.0 85 8.6 3.3 43.5 33.7 19.8 3.0

309 179 181 96 47 24 690 70 27 353 247 161 24

86.1 13.8

%

597 186 22 7

699 112

Gender: Male (0) Female (1)

Marital status: Married (1) Single (2) Engaged (3) Romantic relation (4) Years in the TNP: 5 and below (1) 6-10 (2) 11-15 (3) 16-20 (4) 20 and above (5) Rank: Civilian officer (1) Line police officer (2) Sergeant (3) Lieutenant (4) Education level: High school (1) Associates Degree (2) Bachelor’s Degree (3) Graduate Degree (4)

N

Variable

1.82

2.13

2.25

1.31

0.15

Mean

0.85

0.49

1.2

0.56

0.46

SD

Min

Max

Pop.a

N. offb

N. cric

Cri/popd

Cri/offe

(continued)

Gro. ratef

Are causes of police stress global? 297

Table I. Descriptive statistics

Table I. 222 104 273 28 67 63 55

27.3 12.8 33.6 3.4 8.3 7.8 6.8

%

1.58 1.59 1.23 1.44 1.62 10.35 15.23 14.63 1.71 4.0 0.74 0.84 0.61

23.33 33.5 35.98 5.36 18.2 0.86 1.18 0.59

1.83

SD

4.7 4.08 5.59 3.42 4.36

3.3

Mean

0.00 0.00 0.00

7

0

0.00 0.00 0.00

1 1 1 1 1

Min

4.00 4.00 4.00

24

7

45 54 54

7 7 7 7 7

Max 2,897 10,033 603 82 196 93 55

Pop.a 17,727 27,247 1,133 165 645 153 121

N. offb 28,643 201,892 8,639 886 6,820 827 521

N. cric 9.8 20.12 14.32 10.74 34.73 8.87 9.45

Cri/popd

165.8 740.9 762.49 536 1,057.3 540.52 430.58

Cri/offe

22.15 29.47 44.13 28.49 20.73 46.27 58.76

Gro. ratef

Notes: a £ 1000; bNumber of officers; cNumber of crime; dCrime number/population £ 1000; eNumber of Crime/number of officer £ 100; fIncrease in the population between 1990-200 as %

Jurisdictions: J-1 J-2 J-3 J-4 J-5 J-6 J-7 Job dimensions: Skill variety Task identity Task significance Autonomy Feedback Job satisfaction: Satisfaction with work Sat. with supervisors Sat. with co-workers Commitment to TNP: Lowest – 0 Highest – 7 Bureaucracy index: Stress symptoms: Depression Anxiety Somatization

N

298

Variable

PIJPSM 30,2

Organizational bureaucracy: Presence of bureaucracy within particular departments can be another explanatory factor in predicting police stress (see Reese, 1986; Violanti and Aron, 1994; Zhao et al., 2002). In this study, bureaucracy is assessed with a composite measure of five items (a ¼ 0.79). These items assess the respondents’ perceptions about the presence of several problems within the department, such as too much red tape, inadequate management, etc. The choices range from 1 – no problem to 3 – serious problem. The composite measure of bureaucracy, similar to the Zhao et al.’s (2002) measure, is a summed scale based on five (see Appendix for all items). These typical characteristics of bureaucracy in police organizations are identified in several studies (see Brown and Campbell, 1994; Coman and Evans, 1991). The measure of organizational bureaucracy is meant to reflect several organizational problems as developed in the Zhao et al.’s (2002) study.

Analytic strategy This study does not directly assess employees’ personal perceptions of how different stressors affect their personal stress levels. Instead, a clinical scale measures stress levels while several characteristics of police work, organization, and work related experiences provided by the participants themselves are examined. Analyses in this study, therefore, seek to explain how stress levels relate to, and are predicted by the nature of police work, organization, and personnel. Assessing police work stressors in this manner has some advantages over asking participants directly about their perceptions of how multiple stressors in their occupational life influence their levels of stress. As mentioned before, when one asks police officers about their perceptions of potential stressors, they may exaggerate the effects of occupational stress factors. (Brown and Campbell, 1994; Cullen et al., 1983; Ellison, 2004; Golembievski and Kim, 1991; Kroes and Gould, 1982; Storch and Panzarelle, 1996). The strategy used in this paper assesses the effects of stressors indirectly. Instead of asking participants directly about their satisfaction with supervisors, several indirect items are used to measure their level of satisfaction. In some parts of the survey, such as perceptions of work-related problems, participants are asked direct questions, but these problems are not identified as possible stressors. This characteristic of the study may also be a limitation. Since the majority of studies ask respondents about their perceptions of several stressors directly, the results of this study should be approached with caution when making comparisons with previous studies. Yet, we believe this method of assessing stress yields reliable outcomes. The data used in the analyses is also limited in terms of assessing all possible stressors as discussed in the literature. There is no information in this data, for instance, regarding the availability of the coping mechanisms or the level of exposure to danger at work. Measures similar to the level of exposure to danger at work, such as crime rate and crime/officer rate, however, have been included. In order to understand the relationship between several variables and stress levels, multivariate analyses are used. In two different OLS models, stress is regressed with control and explanatory variables. The first model tests the effects of demographic variables on stress, while the second model includes all independent variables. Since demographic characteristics can be significant predictors of by themselves and tend to washout once more robust models are constructed, the two-model approach is used in this study. The findings from these analyses are discussed in detail below.

Are causes of police stress global? 299

PIJPSM 30,2

300

Findings This study reflects an observation of 811 officers of the TNP from seven different jurisdictions (cities). Most of the participants are male (86.1 percent), married (73.5 percent), line police officers (85 percent) with an experience between six and 15 years in the TNP. The descriptive statistics of the participants and variables are displayed in the Table I. One-way correlations between all variables are in Table II. In order to assess the effects of several predicting variables on the stress score we run two OLS regression models. These models assess the effects of demographic characteristics, jurisdiction characteristics, job characteristics, job satisfaction, organizational bureaucracy, and commitment to the department in two separate models. All demographic variables are included in the first model as a separate group to assess how they behave when they are alone in a model and together with other variables. While both model 1 and model 2 are statistically significant, model 2 provides a relatively high explanatory power (R 2 0.19) when compared to model 1 (R 20.005). The results from the OLS regression analyses are presented in Table III. In both models, control variables show no significant effect. Effects of other explanatory variables do not change across the demographic characteristics of the participants. Dimensions of job satisfaction are strong and significant predictors of stress. Our analyses indicate that officers, who are more satisfied with their work, supervisors, and fellow officers, are less prone to experience stress. Among the job satisfaction measures, satisfaction with supervisor is the strongest predictor. Satisfaction with work itself is as strong as satisfaction with supervisor. Although not as strong of a predictor as the preceding two variables, Satisfaction with co-workers, is an important measure predicting the level of stress. Job motivation measures, as defined by the meaningfulness and responsibility factors, show only minor effects on stress. It is possible that the job satisfaction measures are simply better predictors of stress because they tap into motivation measures as well. It follows logic that officers who are motivated and find their job meaningful are also officers generally satisfied with their work. When they are assessed with job satisfaction measures, however, their effect disappears because of this correlation between motivation and satisfaction measures. The bureaucracy index measure, which covers several problematic issues within the department, such as excessive workload, inadequate staff, inadequately specific policies/procedures, inadequate supervision/direction and too much “red tape” within the department, is found to be strongly related to the individual level of stress in the department. These problems are identified as common causes of stress in police departments. Based on this finding, the greater the rate of bureaucracy, the higher the level of stress for the officers working in the department will be. Our analysis indicates this index is the second strongest significant predictor of stress other than job satisfaction measures. Jurisdiction characteristics offer some interesting results as well. Neither the crime/population rate, meant to reflect dangerousness of a jurisdiction for police officers, nor the crime/officer rate, meant to reflect workload for police officers, are found to be significant predictors of stress among police officers. This result is surprisingly contrary to what has been found in many other studies. The possible explanations for this particular finding and other findings are discussed.

Gender Rank Education Marital status Years in TNP Skill variety Task identity Task significance Feedback Satisfaction work Satisfaction promotion Satisfaction supervisor Satisfaction co-worker Crime/pop rate Crime/officer rate Summed stress scale Bureaucracy index

1

2

0.01

2 0.05

2 0.07* 2 0.09

0.00 2 0.09* 0.02** 2 0.2** 0.01 2 0.02

0.1**

0.00

1 2 0.12** 1 0.021 0.41** ** 0.15 0.06 2 0.10** 2 0.01 2 0.13** 0.12** 2 0.08 0.16** ** 2 0.11 0.11** 2 0.04 0.08* 2 0.07 0.14** 0.01 0.27**

Notes: *p # 0.05; **p # 0.01

17

14 15 16

13

12

1 2 3 4 5 6 7 8 9 10 11

0.09

2 0.08* 2 0.17** 0.01

0.00

0.01

1 0.15** 2 0.17** 2 0.02 0.02 0.01 2 0.01 2 0.04 0.16**

3

5

2 0.12

0.09* 0.05 0.03

2 0.05

2 0.23

0.27** 0.21**

0.17**

1 0.33** 0.53** 0.43** 0.27**

7

0.26**

1 0.39** 0.57** 0.51** 0.49** 0.09**

6

0.17**

0.3**

1 0.49** 0.41** 0.13**

8

0.22**

0.34**

1 0.49** 0.21**

9

2 0.06

0.03

2 0.14** 0.01

2 0.08*

2 0.15** 0.08* 2 0.02 0.07 0.05 2 0.05 0.08* 0.02 0.06 0.07 ** ** ** 2 0.06 2 0.11 2 0.13 2 0.18 2 0.18**

0.05

0.09**

1 2 0.44 1 2 0.27** 0.31** ** 2 0.12 0.13** ** 2 0.19 0.19** 2 0.17** 0.21** 2 0.17** 0.22** * 0.08 2 0.05

4

0.15**

0.3**

1

11

0.33**

1

12

1

13

14

15

2 16**

20.14** 20.17** 20.18**

0.07*

20.02

0.02 0.01 0.01 20.06 1 0.04 0.03 0.07* 20.00 0.75** 1 ** ** ** ** 20.28 20.18 20.32 20.24 20.04** 20.06**

0.32**

0.45**

1 0.3**

10

20.017

1

16

1

17

Are causes of police stress global? 301

Table II. Correlation matrix

PIJPSM 30,2

302

Dependent variable: summed stress index Model 1 Model 2

Independent variables Control variables: Years employed by TNP Sex (1male, 0 female) Marital status (1 married, 0 not married) Education status Current rank

2 0.899/0.914 2 1.727/2.900 2 1.323/2.566 0.400/1.298 2 3.028/2.214

Bureaucracy measure: Bureaucracy summed index

Table III. OLS regression analysis

(20.040) (20.020) (20.021) (0.012) (20.054)

0.010/0.865 21.448/2.695 1.019/2.380 20.210/1.193 1.072/2.092 0.916/0.235

(2 0.000) (2 0.018) (0.016) (2 0.006) (2 0.001) (0.131) * * *

Job meaningfulness and responsibility measures: Skill variety Task significance Task identity Feedback

1.687/0.784 22.219/0.925 0.774/0.692 20.152/0.745

(0.095) (2 0.098) (0.044) (2 0.009)

Job satisfaction measures: Satisfaction with work index Satisfaction with supervisor index Satisfaction with coworkers index

20.485/0.115 20.367/0.069 20.227/0.067

(2 0.180) * * * (2 0.200) * * * (2 0.119) * * *

Jurisdiction characteristics: Crime to population rate Crime to officer rate Constant Model F R2

20.397/0.207 0.005/0.005 55.153 * * * 12.306 * * * 0.188

(2 0.097) (0.047)

42.043 * * * 0.851 0.005

Notes: Coefficients are presented as slope estimate/standard error (standardized coefficients are in parenthesis). *=p , 0.05; * *=p , 0.01; * * *=p , 0.001

Discussion The goal of this paper is to empirically test the effects of generally defined police stressors in a different context; in this case the TNP. Using data representing a national sample of police officers working for the TNP, from bottom to middle ranks, this research provides important findings for the future police stress studies. More specifically, our findings have mixed implications for police stress literature. Consistent with the existing body of literature, our findings indicate that the most important stressors are rooted in the structure and the management practices of the police departments. First, our analysis indicates that the existence of bureaucracy is an important predictor of stress among police officers, which is consistent with findings of several earlier studies (see Brown and Campbell, 1994; Coman and Evans, 1991; Zhao et al., 2002). As used in this study, the bureaucracy index covers a group of organizational issues, such as excessive workload, too much red type, inadequate personnel (see Appendix A for full items covered in this index). Second, dimensions of job satisfaction are drastically affecting the level of stress in a negative direction, which again is consistent with the findings of earlier studies (see Aron, 1992; Brown and Campbell, 1990; Storch and Panzerella, 1996; Zhao et al., 2003). Officers, who are more satisfied with their supervisors, co-workers, and work itself, are less likely to feel

stress in their occupational life. These findings validate the existing argument in police stress literature, which states the most important stressors are rooted in the organizational environment. Our analyses, however, find work characteristics, in most dimensions, do not have significant effects when job satisfaction dimensions are controlled. This is slightly counter to what Zhao et al. (2002) find while using the same measures, but not controlling for job satisfaction levels. They find autonomy and feedback are related to stress. Our findings, however, indicated only skill variety at work is associated with our overall stress measure. This finding suggests officers’ satisfaction with several dimensions of their job (i.e. work itself, supervisors, and co-workers) negates the negative effects of work characteristics on stress levels. Although the duties officers perform can have several negative aspects, they might end up with less stress if they are highly satisfied with their job. As mentioned earlier, endemic characteristics of police work, such as exposure to danger, violence and heavy workload are identified as important stressors in policing. Several scholars raise concerns about the validity of these stressors. They contend police stress issues might be exaggerated by the officers, or their departments (see Kroes and Gould, 1982; Storch and Panzaralla, 1996). They argue, police officers rarely mention danger, violence, or human misery as a source of stress, as long as questionnaires do not specifically ask about these factors. Threats of physical harm and exposure to violence are not common for many small-sized police departments. Police officers might be describing their work as being more dangerous than it actually is (Cullen et al., 1983). Our research design intends to overcome this bias by assessing the effects of natural stress indicators rather than contextual stress measures. As mentioned before, two important jurisdiction characteristics are used; one assesses the crime rate in the jurisdiction while the other assesses the workload with the crime/officer rate. When these measures are included in the regression models, the results support the ideas mentioned above regarding the stressing effects of danger, heavy workload and similar natural characteristics of policing. We find that officers’ stress rates are not related to workload in their jurisdictions. The crime rates, likewise, are negatively related to two symptoms of stress. Crank (1997), and Crank and Caldero’s (1999) extensive studies on police culture and ethics provide important insight to explaining this situation. In these studies, authors argue being involved in regular police activities (e.g. arrest, pursuit, crime scene processing, traffic stops) does not constitute a negative situation in the daily lives of police officers. These situations, indeed, increase the satisfaction officers get from their occupation. Lack of opportunities to realize what is considered as “real policing” among officers likely causes several dissatisfactions (Crank, 1997; Crank and Caldero, 1999). Assessing the effects of jurisdiction characteristics is also of importance in this context. The TNP is a centralized department that operates in several jurisdictions across the country and is based on uniform laws, practices, policies, and mentalities. Indeed, officers in these jurisdictions are recruited and trained based on a standard criterion and curricula. In this setting, we find that jurisdiction size does not have an effect. Our analyses find that in larger departments, officers are experiencing less stress, which is contrary to Brooks and Piquero’s (1998) findings, which state that the department size is positively associated with officers’ stress levels. Their findings, however, are based on US municipal departments. In our case, and similar other studies based on national police departments, although the jurisdictions are different, there is one unique department, the TNP in this case. It can be stated conclusively that

Are causes of police stress global? 303

PIJPSM 30,2

304

department size will not have the same effect across jurisdictions of a national police department as it does in independent municipal departments. Winfree and Taylor (2004) similarly find that across different-sized jurisdictions of New Zealand Police Organization, which is also a national police department, several officer attitudes, including job stress, do not change significantly. We agree with their argument, which states this is due to the unified structure of the department, and the similar backgrounds of officers. Although several demographic characteristics of officers such as gender, marital status, and education (see Aron, 1992; Crank and Caldero, 1991; He et al., 2002; Laufersweller-Dwyer and Dwyer, 2000; Newman and Rucker-Reed, 2004) were found to be associated with officers’ level of stress, our analyses indicate that these characteristics, when assessed with other organizational predictors, do not create a significant difference in experiencing stress. In short, participants of our study are experiencing similar types and levels of stress regardless of their gender, experience, rank, marital status and education level. This finding is consistent with what Morash et al. (2006) found in their recent study for American police departments. Policy implications from our study are not restricted to the TNP. We find the nature of police work per se is not a significant source of stress. What makes policing a stressful job is better understood within the macro and micro level implications and relations within the department. Administrators should put more emphasis on modern management methods that provide a relaxed environment for officers, and in turn, make them less stressed. More specifically, in any department, regardless of size, possible danger at work, demographic differences, and negative work characteristics, officers might be more or less stressed. This has many things to do with how officers perceive their intimate and greater work environment and less to do with the nature of police occupation, personal differences, or specific types of work characteristics. From a comparative perspective, our study supports the general argument in current police stress literature, which states the most important stressors in policing are rooted in administrative policies. When several studies conducted in other countries (e.g. Lithuania (Zukauskas et al., 2001), Norway (Berg et al., 2005), Korea (Lee, 2002), England (Kirkcaldy et al., 1998), Australia (Comman and Evans, 1991; Davidson and Veno, 1979), New Zealand (Winfree and Taylor, 2004), and Singapore (Lim and Teo, 1998)) are taken into account, it is possible to argue that unprofessional management, organizational bureaucracy, and a negative environment at the workplace are strong predictors of police stress around the world. It can also be argued that the negative nature of police work, which is commonly attributed to higher levels of stress among officers, is only valid when combined with unprofessional management in police departments. These negative aspects of police work, however, do not create stress for officers when several administrative policies and camaraderie in the workplace increases officers’ job satisfaction levels. Our study lacks some important information, especially regarding the role of coping mechanisms. Our findings and arguments should be interpreted accordingly. Future studies should examine the role of administrative policies on the personnel’s stress levels, on both macro and micro levels, in several democratic police organizations across the globe, with consideration for internal and external coping mechanisms. With these studies, policing literature will have a more comprehensive explanation for the question we pose as the title of this study “are causes of police stress global?”

Notes 1. Our review of the published studies did not yield any result regarding the stress measures of the TNP members. 2. This number does not reflect all activities that the department conducted, such as answering service calls. Rather, it is the number of cases, where there is a process of complaint regarding a criminal activity, an arrest or similar official procedures. 3. The autonomy measure from original instrument was removed because of a low reliability rate (0.33) References Aron, F. (1992), “An analysis of sources of police stress”, Master’s thesis, Russell Sage Collage, Albany, NY. Aaron, J.D. (2000), “Stress and coping in police officers”, Police Quarterly, Vol. 3, pp. 438-50. Barr, S.H., Brief, A.P. and Aldag, R.J. (1978), Measurement of Perceived Task Characteristics, College of Business Administration, University of Iowa, Ames, IA. Berg, A.M., Hem, E., Lau, B., Haseth, K. and Ekeberg, O. (2005), “Stress in the Norwegian police service”, Occupational Medicine, Vol. 55, pp. 113-20. Bennet, R.R. (1997), “Job satisfaction among police constables: a comparative study in three developing nations”, Justice Quarterly, Vol. 14, pp. 295-323. Brooks, L.W. and Piquero, N.L. (1998), “Police stress: does department size matter”, Policing: An International Journal of Police Strategies & Management, Vol. 21, pp. 600-15. Brown, J.A. and Campbell, E.A. (1990), “Sources of occupational stressing police”, Work and Stress, Vol. 4, pp. 305-18. Brown, J.A. and Campbell, E.A. (1994), Stress and Policing: Sources and Strategies, John Wiley, New York, NY. Comman, G. and Evans, B. (1991), “Stressors facing Australian police in the 1990s”, Police Studies, Vol. 14, pp. 153-65. Conroy, D.L. and Hess, K.M. (1992), Officers at Risk: How to Identify and Cope with Stress, Custom Publishing Co, Placerville, CA. Crank, J.P. (1997), Understanding Police Culture, Anderson Publishing, Cincinnati, OH. Crank, J.P. and Caldero, M. (1999), Police Ethics: The Corruption of Noble Cause, Anderson Publishing, Cincinnati, OH. Crank, J.P. and Caldero, M. (1991), “The production of occupational stress in medium-sized police agencies: a survey of line officers in eight municipal departments”, Journal of Criminal Justice, Vol. 19, pp. 339-49. Cullen, F.T., Link, B.T., Travis, L.F. and Lemming, T. (1983), “Paradox in policing: a note on perceptions of dangers”, Journal of Police Science and Administration, Vol. 11, pp. 457-62. Davidson, M.J. and Veno, A. (1979), “Police stress in Australia: a current perspective”, Australian and New Zealand Journal of Criminology, Vol. 12, pp. 153-61. Davidson, M.J. and Veno, A. (1978), “Police stress: a multicultural, interdisciplinary review and perspective”, Abstracts on Police Science, Vol. 6, pp. 187-99. Derogatis, L. and Melisaratos, N. (1983), “The brief symptom inventory: an introductory report”, Psychological Medicine, Vol. 13, pp. 595-605. Derogatis, L. and Savitz, K. (1999), “The SCL-90-R, brief symptom inventory and matching clinical rating scales”, in Maruish, M. (Ed.), The Use Psychological Testing for Treatment,

Are causes of police stress global? 305

PIJPSM 30,2

306

Planning and Outcomes Assessment, Lawrence Erlbaum Associates Inc., Mahwah, NJ, pp. 1-43. Derogatis, L. and DellaPietra, L. (1994), “Psychological tests in screening psychiatric disorder”, in Maruish, M. (Ed.), The Use Psychological Testing for Treatment, Planning and Outcomes Assessment, Lawrence Erlbaum Associates Inc., Mahwah, NJ, pp. 34-47. Ellison, K. (2004), Stress and the Police Officer, Charles C. Thomas, Springfield, IL. Golembiewski, R. and Kim, B. (1991), “Burnout in police work: stressors, strain, and the phase model”, Police Studies, Vol. 14, pp. 74-80. Hackman, J.R.. and Oldham, G.R. (1980), Work Redesign, Addison-Wesley, Reading, MA. He, N., Zhao, J. and Archbold, C.A. (2002), “Gender and police stress: the convergent and divergent impact of work environment, work-family conflict, and stress coping mechanisms of female and male police officers”, Policing: An International Journal of Police Strategies & Management, Vol. 25, pp. 687-708. Jacoby, T. and and Mann, M. (2004), Social Power and the Turkish State, Frank Cass, New York, NY. Jaramillo, F., Nixon, R. and Sams, D. (2004), “The effect of law enforcement stress on organizational commitment”, Policing: An International Journal of Police Strategies & Management, Vol. 28, pp. 321-36. Jex, S.M. and Beehr, T.A. (1991), “Emerging theoretical and methodological issues in the study of work-related stress”, Research in Personnel and Human Resources Management, Vol. 9, pp. 311-65. Kanter, R. (1977), Men and Women of the Corporation, Basic Books, New York, NY. Kirkcaldy, B., Siefen, G. Eysenck, M. and Furnham, A. (1998), “Gender, anxiety and self image”, Personality and Individual Differences, Vol. 24 No. 5, pp. 677-684. Krimmel, J.T. and Gormley, P.E. (2003), “Tokenism and job satisfaction for policewomen”, American Journal of Criminal Justice, Vol. 28, pp. 73-88. Kroes, W. (1986), Society’s Victim: The Police Officer, Charles C. Thomas, Springfield, IL. Kroes, W.H. and Gould, S. (1982), “Job stress in policemen: an empirical study”, Police Stress, Vol. 1, pp. 9-10. Laufersweler-Dwyer, D.L. and Dwyer, R.G. (2000), “Profiling those impacted by organizational stressors at the macro, intermediate and micro levels of several police agencies”, The Justice Professional, Vol. 12, pp. 443-69. Lawrence, R.A. (1984), “Police stress and personality factors: a conceptual model”, Journal of Criminal Justice, Vol. 12, pp. 247-63. Lee, S. (2002), “A study of Korean police sergeants’ stress”, International Journal of Comparative and Applied Criminal Justice, Vol. 26, pp. 85-99. Lim, V.K. and Teo, T.S. (1998), “Effects of individual characteristics on police officers’ work related attitudes”, Journal of Managerial Psychology, Vol. 13, pp. 334-42. Malloy, T.E. and Mays, G.L. (1984), “The police stress hypothesis: a critical evaluation”, Criminal Justice and Behavior, Vol. 11, pp. 197-224. Martelli, T., Waters, L. and Martelli, J. (1989), “The police stress survey: reliability and relation to job satisfaction and organizational commitment”, Psychological Reports, Vol. 64, pp. 267-73. Morash, M., Haarr, R. and Kwak, D. (2006), “Multilevel influences on police stress”, Journal of Contemporary Criminal Justice, Vol. 22, pp. 26-43. Newman, D.W. and Rucker-Reed, M.L. (2004), “Police stress, state-trait anxiety, and stressors among US Marshals”, Journal of Criminal Justice, Vol. 32, pp. 631-41.

Petrone, S. and Reiser, M. (1985), “A home visit program for stressed police officers”, The Police Chief, Vol. 52, pp. 36-7. Reiser, M. (1974), “Some organizational stressors on police officers”, Journal of Police Science and Administration, Vol. 2, pp. 156-9. Reese, J. (1986), “Policing the violent society: the American experience”, Stress Medicine, Vol. 2, pp. 233-40. Sager, J.K. and Wilson, P.H. (1994), “Clarification of the meaning of job stress in the context of salesforce research”, Journal of Personal Selling and Sales Management, Vol. 15, pp. 51-64. Smith, P.C., Kendall, L.M. and Hullin, C.L. (1969), The Measurement of Satisfaction in Work and Retirement, Rand McNally, Chicago, IL. Spielberger, C., Westberry, L., Grier, K. and Greefield, G. (1981), The Police Stress Survey: Sources of Stressing Law Enforcement, Human resources Institute, Tampa, FL. Skolnick, J.H. (1999), Ideas in American Policing: On Democratic Policing, Police Foundation, Washington, DC. Stansfield, R.T. (1996), Issues in Policing: A Canadian Perspective, Thompson Educational Publishing, Toronto. Storch, J.E. and Panzarelle, R. (1996), “Police stress: State-trait anxiety in relation to occupational and personal stressors”, Journal of Criminal Justice, Vol. 24, pp. 99-107. Swanson, C.R., Territo, L. and Taylor, R.W. (1998), Stress and Police Personnel. Police Administration: Structures, Processes and Behavior, 4th ed., Prentice Hall, Englewood Cliffs, NJ. Turkish National Police (2005), Crime Statistics, available at: www.egm.gov.tr/akkm/ index_dosyalar/istatistik.htm (accessed 30 September 2006). Violanti, J. (1985), “The police stress process”, Journal of Police Science and Administration. Violanti, J.M. and Aron, F. (1995), “Police stressors: variations in perception among police personnel”, Journal of Criminal Justice, Vol. 23, pp. 287-94. Violanti, J.M. and Aron, F. (1994), “Ranking police stressors”, Psychological Reports, Vol. 75, pp. 824-30. Violanti, J.M. and Aron, F. (1993), “Sources of police stressors, job attitudes, and psychological distress”, Psychological Reports, Vol. 72, pp. 899-904. Yesilorman, M. (2003), “Stress on police officers and stress management”, Journal of Police Studies, Vol. 5, pp. 91-120. Webb, S.D. and Smith, D.L. (1980), “Police stress: AA conceptual overview”, Journal of Criminal Justice, Vol. 8, pp. 251-7. Winfree, L.T. and Taylor, T.J. (2004), “Rural, small town, and metropolitan police in New Zealand: Differential outlooks on policing within a unified police organization”, Policing: An international Journal of Police Strategies & Management, Vol. 27, pp. 241-63. Zukauskas, G., Dapsys, K., Jasmontatite, E. and Susinskas, J. (2001), “Some psychological problems of police officers in Lithuania”, Policing: An International Journal of Police Strategies & Management, Vol. 24, pp. 301-9. Zhao, J., Lovrich, N.P., Gaffney, M.J., Smith, M. and Spokane, P.D. (2003), Commissioned Employee Survey: General Findings, Washington State University, Division of Departmental Studies and Services, Pullman, WA. Zhao, J., He, N.P., Lovrich, N. and Cancino, J. (2003), “Marital status and police occupational stress”, Journal of Crime & Justice, Vol. 26, pp. 23-46.

Are causes of police stress global? 307

PIJPSM 30,2

308

Zhao, J., He, N.P. and Lovrich, N. (2002), “Predicting five dimensions of police officer stress: looking more deeply into organizational settings for sources of police stress”, Police Quarterly, Vol. 5, pp. 43-62. Further reading Bittner, E. (1970), The Function of Police in Modern Society, Government Printing Office, Washington, DC. Hale, W., Cockran, C. and Hedgepeth, B. (1984), “Norms for the elderly on the brief symptom inventory”, Journal of Consulting and Clinical Psychology, Vol. 52, pp. 321-2. Turkish National Police Organization Act (1937). TNP Health Department (2000), Department Stress and Coping Mechanisms, TNP Printing Office, Ankara. Appendix. Stress measures In the past week how often did you experience the following problems? 0 – Never, 1 – A little bit, 2– Moderately, 3 – Quite a bit, 4– Extremely Symptoms . Faintness or dizziness. . Someone controls thoughts. . Others are to blame. . Trouble remembering. . Pains in heart or chest. . Afraid in open places. . Thoughts of ending life. . People cannot be trusted. . Poor appetite. . Scared for no reason. . Temper outbursts. . Lonely when with people. . Feeling blocked. . Feeling lonely. . Feeling blue. . No interest in things. . Feeling fearful. . Feelings easily hurt. . People dislike you. . Feeling inferior. . Nausea or upset stomach. . Watched or talked about. . Trouble falling asleep. . Afraid to travel.

. . . . . . . . . . . . . . . . . . .

Trouble getting breath. Hot or cold spells. Avoid certain things. Mind going blank. Punished for sins. Hopeless about future. Trouble concentrating. Tense or keyed up. Thoughts of death. Urges to harm someone. Urges to break things. Uneasy in crowds. Spells of terror panic. Frequent arguments. Nervous when alone. Feel worthlessness. People take advantage. Feelings of guilt. Something wrong w mind.

Organizational bureaucracy Please write the number that most accurately describes the extent of these factors are a problem in your work. 1 – No problem, 2 – A problem, 3 – Serious problem, 0 – Uncertain . Excessive workload. . Inadequate staff. . Inadequately specific policies/procedures. . Inadequate supervision/direction. . Too much “red tape” within the department.

To purchase reprints of this article please e-mail: [email protected] Or visit our web site for further details: www.emeraldinsight.com/reprints

Are causes of police stress global? 309

The current issue and full text archive of this journal is available at www.emeraldinsight.com/1363-951X.htm

PIJPSM 30,2

Police officers’ assessment of operational situations R. Flin, Z. Pender, L. Wujec and V. Grant

310

School of Psychology, University of Aberdeen, Aberdeen, UK, and

E. Stewart Grampian Police, Aberdeen, Aberdeen, UK

Received 21 January 2006 Revised 6 June 2006 Accepted 25 October 2006

Abstract Purpose – The aim of this study is to identify the dimensions used by police officers in Scotland to discriminate and categorise operational situations. Design/methodology/approach – This article was based on two studies: study 1 was a card-sorting task: 23 officers categorised a set of 19 typical police operational situations, analysed using multi-dimensional scaling; study 2 was a judgement task: 112 police officers rated 20 situations (19 from study1) in terms of amount of time to make a decision, risk to self, risk to others, familiarity and stress. Frequency data and correlations were calculated. Findings – Study 1 results showed that the two main dimensions used to discriminate between situations were “familiarity” and “risk to the officer”. Study 2 found that most situations required a decision in less than three minutes. Rank and experience were related to familiarity but not to the other judgements. The situations requiring the fastest decisions were also judged to be of higher risk and more unfamiliar. Risks to self and to others were highly correlated, and higher risk situations were judged as more stressful. Research limitations/implications – This preliminary study shows that knowledge elicitation techniques can be used to improve our understanding of police officers’ knowledge and cognitive skills for operational policing. Practical implications – Good situational judgements form the basis of effective operational decision-making. Understanding how officers “read” situations and how this develops with expertise can provide valuable information for police training and critical incident management. Originality/value – Police officers’ cognitive skills have rarely been studied in relation to routine operational policing. This study is a first attempt to examine situational judgement skills. Keywords Situation analysis, Decision making, Policing, Skills, Scotland Paper type Research paper

Introduction We approached All Saints Road. . . This was a mistake because . . . we were suddenly confronted by a crowd of about 50 people who started attacking the car with missiles. I had been too preoccupied thinking how I was going to deal with the situation and what I was going to say to any of the “street leaders” to notice the route Sergeant Hole was taking (Moore, 2002, p. 74 ret. Chief Superintendent, Metropolitan Police, London, discussing the Notting Hill riot). Policing: An International Journal of Police Strategies & Management Vol. 30 No. 2, 2007 pp. 310-323 q Emerald Group Publishing Limited 1363-951X DOI 10.1108/13639510710753289

[. . .] the command perspective, central to which is the ability to attend to critical aspects of an unfolding situation – which implies that other aspects are ignored or quickly dismissed as less important – make sense of what’s going on . . . (Sarna, 2002, p. 40: ret. Captain of Police, Oakland, California).

Policing in Scotland is similar to the style employed in many other jurisdictions across the world but there are some notable differences. Traditional Scottish policing is relatively low on numbers, low on power and high on accountability. Policing is undertaken with public consent, which does not mean acquiescence but a broad tolerance indicating a satisfaction with helping and enforcement roles. This is epitomised by the single constable, close to his or her community, patrolling a beat (area) with the consent of the general public, armed only with lawful powers and use of discretion. Police officers in Scotland do not routinely carry firearms and the police use of firearms is a highly specialised sphere of operations conducted only by authorised officers who have been specially selected and trained. Consequently, police officers must be able to rapidly assess a wide range of situations in order to decide on an appropriate response (course of action). This applies to constables and sergeants who may be first on the scene, as well as to more senior officers who will be called to larger incidents. Very little research has been conducted into police decision-making, despite its critical role in effective operational policing (Moore, 2002; Stewart and Flin, 1996) and well publicised cases where erroneous decisions have been taken by police officers (see Gladwell, 2005). In other professions requiring rapid decision making in risky environments, most notably aviation (Appendix 1) (Jensen, 1995; Orasanu and Fischer, 1997) and military (Cohen and Freeman, 1997), there has been significant research to understand and train operational thinking skills. This paper examines how police officers in Scotland assess situations they are likely to encounter in both routine and less common events. It draws on research into naturalistic decision making (NDM) (Flin et al., 1997; Zsambok and Klein, 1997) which studies how experienced workers make decisions in real world environments. Of particular interest is how police officers rapidly assess and categorise new situations. Operational situations encountered by police officers can be classified into different types (Loree, 1995). But with the exception of studies of authorised firearms officers (Vrij et al., 1995), little is known about how they make these assessments and whether they use some kind of situational estimation and categorisation process. There have been some attempts to categorise the main types of events that police officers encounter. For example, Sykes et al. (1983) examined 3,000 civilian encounters with an American police sample and categorised these into 21 types (e.g. domestic; assault; prowler; burglary). This list was then given to 100 officers who were asked to rate the 21 situation types on dimensions such as: “rush” vs. “take it easy”; “dangerous” vs. “safe”). Using factor analysis they found that there were five clusters of situations - hazardous (e.g. assault), annoying (e.g. neighbour trouble), necessary (e.g. prowler), boring (e.g. men drinking) and adventurous (e.g. car chase). They also reported that the initial descriptions by the police dispatcher of the nature of the incident did not influence officers’ decisions about the nature of the incident, upon arriving at the scene. An officer’s behaviour is a response to the actual processes occurring in the situation, after he gets there, not a function of prior subjective expectation of situations. Officers assert their own definitions of a situation rather than uncritically accepting that of a complainant or police dispatcher (Sykes and Brent, 1983, p. 48).

If routine operational situations are not diagnosed and managed properly by the police in the opening phase, then they can rapidly escalate into a major incident (c.f. Rodney

Assessment of operational situations 311

PIJPSM 30,2

312

King riots in Los Angeles (Cannon, 1998); Hillsborough football stadium disaster in England (Taylor, 1990)). The exploratory studies reported below were designed to examine situation assessment in police officers by knowledge elicitation methods, previously used to study airline pilots’ decision making. Operational decision making across professions can be conceptualised as a two-stage process – the first requires an assessment of the current situation (What is the problem?), on the basis of this judgement, the second stage is determining the appropriate reaction (What to do in response?) (Crichton and Flin, 2002; Orasanu and Fischer, 1997). This paper concentrates on situation assessment – the first stage of the decision making process. One significant development from the naturalistic decision making approach has been the shift of research emphasis from the moment of choice (i.e. choosing what to do) to the initial situation assessment stage (Lipshitz et al., 2001). This is a component of a broader cognitive skill, called situation awareness, that encompasses information gathering, situation assessment and projection of future states (Banbury and Tremblay, 2004; Endsley and Garland, 2000) – crucial in safety-critical tasks, such as operating a control room or flying an aircraft and equally relevant to operational policing tasks. Aviation studies show that poor situation assessment is a more common cause of accidents rather than poor choice of action based on an adequate diagnosis (Orasanu et al., 1993). Aviation and policing may look like entirely unrelated work domains. However, from a psychological perspective, the thinking skills required by professionals in high-risk environments are remarkably similar. Police officers on patrol are working in uncertain, time pressured, high-risk settings and they must be able to make rapid assessments of the situations encountered to take appropriate action. Similarly, for airline pilots, situational assessment and good judgement are significant skills contributing to efficient and safe job performance. So knowledge elicitation techniques that have been used to study cognitive processes in high-risk domains (e.g. aviation) could be adapted to study police officers’ situation assessment (See Crandall et al. (2006) for a review of these methods). Orasanu and her colleagues at NASA conducted a series of studies examining the situation assessment stage of pilots’ decision making. When pilots were asked to categorise in-flight situations requiring a decision, using a card-sorting task, the results showed that they identified risk, time pressure, situational ambiguity and response determinancy as relevant dimensions (Fischer et al., 1995). There were also some differences found in relation to experience: Captains were sensitive to risk and time pressure, while First officers tended to emphasise the response aspect of decision-making. This type of study can indicate the ways in which professionals are structuring their knowledge of previous situations they have encountered, and reveals how they may categorise new situations. Orasanu and Fischer (1997, p. 352) also found that situational categorisation determined the decision method that pilots would use. Their key situational estimates were: . What is the problem? . How much time do I have? . What is the level of risk? On the basis of this calculation of available time and risk, the pilots would either employ rapid decision making (e.g. rule based response selection) or slower analytic,

option comparison. This card sorting method has also been used to examine the mental categorisation of decisions taken in nuclear power plant emergencies (Crichton et al., 2005) and situations handled by air traffic controllers (Niessen et al., 1999). Given the dearth of previous research into police officers’ decision making, these methods were adapted to examine their situational categorisation.

Assessment of operational situations

Study 1 The first study assessed how police officers categorised typical operational situations. It was hypothesised that police officers would discriminate between typical events based on available time and level of risk, as shown in Orasanu and Fisher’s (1997) model of pilots’ situation assessment for decision making.

313

Method Materials. In Fischer et al.’s (1995) study, they obtained the descriptions of in-flight situations requiring pilots to make decisions from the Aviation Safety Reporting System (ASRS) database. This is a confidential, non-jeopardy incident reporting system for airline pilots, which will release de-identified information on reported situations for research purposes No similar incident database for routine policing operations was publicly available in the UK. Therefore, a set of 19 typical operational scenarios, ranging in complexity and familiarity (in the same type of format as Fischer et al.’s scenarios) was generated for use with Scottish police officers. The scenarios were developed by a Chief Inspector responsible for operational policing in one of Scotland’s cities. He had 16 years of police service including a broad range of operational experience in city and rural environments including serious crime investigation, management of firearms incidents and management of public disorder situations. The scenarios were based upon actual events that were typical of situations faced by police officers in his force in their day-to-day duties but they also included some situations that were “beyond the regular” day to day officer experience (but were realistic having been adapted from events that had recently occurred). The situations were intended to represent a typical message that would be received from control, or from a more senior officer via the radio, or from a member of the public, when the officer arrived on-scene (Appendix 2 for the set). For example: . You are in an isolated house on the outskirts of the town. A man has doused himself in petrol and is threatening to set fire to himself. There is apparently no-one else involved. . You are on duty at 10 a.m. on a Monday morning. You have been instructed to attend at the Clydesdale Bank [one of Scotland’s high street banks] in town. On arrival, you are told that shots have been fired. Procedure. Following the card sorting procedure advised by Rugg and McGeorge (1997), each officer was given a set of 19 cards each showing one of the situations. The instructions were to sort the cards into groups, on the basis of one dimension (theme) at a time. They were then asked to give and explain a label for the categorisation (theme) they had used (e.g. situations involving weapons or no weapons, situations of three different levels of difficulty, situations that take low, medium or high numbers of officers). Participants could produce as many sorts as they wished. Only the first sort was used for analysis.

PIJPSM 30,2

314

Multidimensional scaling (Kruskal and Wish, 1978) is a statistical technique that can be used with sort data. It displays the structure of distance-like data as a geometrical picture and was used to analyse the data to map the perceived similarities between the situations, and to identify the dimensions apparently underlying the respondents’ grouping patterns. Analysis of the matrix produced by the card sorting talk was conducted using the ALSCAL method (Takane et al., 1977) on SPSS 6 for Windows. Participants. A total of 23 police officers from one Scottish police force (Force A: policing an urban and rural environment) took part in the study. There were 20 constables (mean experience ten years) and three sergeants (mean 18 years experience). Results. The average number of sorts produced was 2.6, range 1-5; the average number of groups of situations produced was 5.6. From the multidimensional scaling, two dimensions were found, accounting for 93 per cent of the variance (RSQ ¼ 0.93). From an examination of the situations, and the sorting labels given by participants, two dimensions were identified as underlying the officers’ grouping of situations. The first dimension was labelled “preplanning/familiarity” with familiar incidents such as those related to football matches and public order situations at one end of the dimension and more unusual/unexpected at the other (e.g. involving firearms and violent crimes). The second dimension “risk to the individual officer” ranged from situations with little risk to the officer (e.g. missing person, New Age travellers) to more dangerous events such as firearms, dealing with suicides, assaults and violent crime. Discussion. The police officers appeared to differentiate situations in terms of their familiarity associated with preplanning (e.g. football match crowds are a regular event) and in terms of the level of personal risk to which they were exposed. Unlike the pilots (Fischer et al., 1995), they did not appear to use available time as a dimension to discriminate between situations for decision making. This difference may be because pilots are always in a time-limited situation once the aircraft is in flight due to the fixed quantity of fuel onboard and this may account for their particular attention to the temporal characteristics of the problems they encounter. But police officers do not have unlimited time to make decisions and the next study used a different technique with the same set of situations, to explore whether they did discriminate between situations with respect to available time and level of risk. Exposure to occupational stress is well documented for police officers (Brown and Campbell, 1994) and stress affects situational awareness (Endsley, 1995) but little is known about how stress is related to police officers’ situation assessments. To determine if judgements of risk and time are linked to expected stress, this variable was also included in the following study. Operational events also vary in terms of frequency and respondents were also asked to rate situations in terms of familiarity to determine how this influenced stress, risk and time judgements. Study 2 Police officers’ judgements relating to time and risk (own or other) in relation to the same situations were assessed to determine whether these were influenced by familiarity, stress, experience or rank. The following predictions were made:

.

. .

Judgements of time, risk and stress would be related, i.e higher risk situations would need faster decisions, and these would also be perceived as more stressful. Ratings of risk to self and to others would be related. Situations that were more familiar would be rated as having more time, lower risk and lower stress (Lower risk situations being more common).

Method Procedure. It was decided after discussion with experienced police officers that a self-completion questionnaire using simple decision scenarios would be the most efficient method of data collection rather than individual interviews, given demands on police time. The questionnaire was first given to five Constables who were asked to complete it and return it to the University, noting any points of ambiguity. No changes were made to the questionnaire on the basis of their feedback. The survey was conducted with two Scottish police forces A & B. (Force A had been involved in study 1). In the first phase of data collection, 90 questionnaires were distributed to Force A, 34 were returned, 32 usable (35 per cent usable response rate). In addition, the five questionnaires completed in the pilot study were included, giving a sample size n ¼ 37. A second sample was collected a year later, in Force A, 50 questionnaires were distributed to a different division (35 responded, 70 per cent) and neighbouring Force B (of similar size and also policing urban and rural environments) also distributed 50 questionnaires (40 responded, 80 per cent). In total, 112 questionnaires were used for the analyses. Participants. The 112 respondents were 77 Constables, 24 Sergeants and 9 Inspectors (two officers did not give their rank). Levels of experience were: up to 5 yrs –17 per cent; 6 – 10 yrs – 21 per cent; 11 – 15 yrs – 22 per cent; over 16 yrs– 40 per cent. Materials. The questionnaire consisted of the same set of 19 scenarios from study 1, plus a new terrorist situation (no. 20) written by a Superintendent. For each scenario, five questions were asked. The instructions were as follows: You are asked to read through each scenario, and try to imagine, to the best of your ability, that it is a real incident, which requires a real decision. Your judgement is required for the following factors: (1) Amount of time available. Does the incident you are faced with require an immediate response – i.e. how long do you have to make a decision? (2) Risk to self. When dealing with this incident, how much risk is there to your own personal safety? (3) Risk to others. When dealing with this incident, how much is the safety of the public at risk? (4) Familiarity of the situation. How familiar is this type of situation compared to others that you have previously dealt with? (5) Level of stress experienced when dealing with incident. When dealing with this situation, how stressed do you feel? Are you worried or do you feel pressurised?

Assessment of operational situations 315

PIJPSM 30,2

316

These were set as rating scales with four options (confirmed as appropriate response options by a Chief Inspector) as follows: (1) Perceived time available: . 0 –3 minutes; . 5 minutes; . 15 minutes; . and 30 minutes. (2) Risk to self: . no risk of harm to personal safety; . little risk, approach with caution; . risk to personal safety, dangerous; and . high risk, very dangerous. (3) Risk to others: . no risk of harm to others; . little risk, approach with caution; . risk of safety to others, dangerous situation; and . high risk, very dangerous. (4) Familiarity of situation: . very familiar, it or a similar incident occurs regularly; . frequent occurrence, have dealt with it/similar incident before, more than once; . this situation/similar incident, has occurred once before; and . unfamiliar, never dealt with it/similar incident before (5) Level of stress experienced when dealing with this situation: . no stress felt; . feel slightly stressed; . feel considerable stress; and . feel very stressed. Respondents were also asked to give their rank and to indicate years of service (0 – 5; 6 – 10; 11 –15; over 16 years). Results There were a range of responses given for each rating, modal scores for the group are shown in Table I. In terms of familiarity, most scenarios were rated from unfamiliar to frequent in relation to prior experience. Levels of risk to self and others varied from low to very dangerous, although few were of no risk. Most situations (80 per cent) were thought to require a decision within 0 – 3 minutes. The majority of situations (75 per cent) were judged to cause only slight levels of stress for the responding officers.

Familiarity Unfamiliar More than once More than once Unfamiliar Unfamiliar Unfamiliar Unfamiliar Unfamiliar Unfamiliar Occurred once before Occurred once before Frequent Occurred once before Occurred once before Unfamiliar Unfamiliar Frequent Frequent Frequent Unfamiliar

Scenario

1. Bank robbery 2. Serious assault 3. Football/street disturbance 4. Gas leak 5. Football/overcrowding 6. Chemical spill/appartment 7. Football/disturbance 8. Chemical spill/road tanker 9. Attempted suicide by fire 10. Domestic violence 11. Dual suicide 12. Breaking and entering 13. Assault at arrest 14. Trespassing 15. Stabbing/drug related 16. House fire/people trapped 17. Capture of wanted suspect 18. Missing elderly person 19. Pub lock-in/drug related 20. Terrorist threat

0-3 mins 0-3 mins 0-3 mins 0-3 mins 0-3 mins 5 mins 0-3 mins 0-3 mins 0-3 mins 0-3 mins 0-3 mins 0-3 mins 0-3 mins 30 mins 0-3 mins 0-3 mins 30 mins 0-3 mins 15 mins 0-3 mins

Time Very dangerous Dangerous Dangerous Dangerous Little risk Little risk Little risk Dangerous Dangerous Dangerous Little risk Dangerous Dangerous Little risk Very dangerous Very dangerous Little risk No risk Little risk Dangerous

Risk to self Very dangerous Dangerous Dangerous Dangerous Dangerous Dangerous Dangerous Very dangerous Very dangerous Very dangerous Little risk Little risk Dangerous Little risk Very dangerous Very dangerous Little risk No risk Little risk Dangerous

Risk to others

Considerable stress Slightly stressed Slightly stressed Slightly stressed Slightly stressed Slightly stressed Slightly stressed Slightly stressed Considerable stress Considerable stress Slightly stressed Slightly stressed Slightly stressed Slightly stressed Considerable stress Considerable stress Slightly stressed No stress Slightly stressed Slightly stressed

Stress

Assessment of operational situations 317

Table I. Modal scores

PIJPSM 30,2

318

Using analysis of variance to examine group differences, the effects of rank and experience were only shown on the familiarity scores, (F ¼ 9.69 (2,109) p , 0.01; F ¼ 11.9, df (3, 109) p , 0.01). Inspectors reported more familiarity with the situations than constables. In relation to years of police service, officers with up to five years experience reported lower familiarity with the situations than officers with over 11 years experience; those with 6 – 10 years had lower familiarity than the 16 years plus group. Rank and experience had no effect on ratings of time, risk, or stress. Therefore, the following analysis was based on the combined data set. To examine whether judgements of time and risk were linked, and whether they were associated with ratings of familiarity or stress, the mean scores on these variables over the 20 scenarios were correlated, see Table II. The pattern of correlations shows that situations judged as requiring fast decisions were also thought to be of higher risk to the officer, and tended to be more unfamiliar (e.g. armed robbery; people trapped in fires). Risks to self and to others were significantly correlated and the higher both these risk judgements, the more stress would be experienced.

Discussion The results of the situation ratings show that for a range of typical Scottish police incidents, officers are required to make decisions very quickly, most officers judged that in 80 per cent of the events that a decision would have to be made within three minutes. It may be that the shortest category of “time” (three minutes) was too long and that officers could have discriminated further between situations requiring almost immediate decisions and those permitting a few minutes thinking time. One study of fire brigade officers estimated that at incidents they made a decision every three minutes on average (Burke and Hendry, 1997). Most police officers rated 50 per cent of the situations as unfamiliar, which seemed slightly surprising, as the scenarios had been written to reflect fairly common incidents for this region. This result might have been related to the nature of the response task and it may be that they were judging whether they had encountered this actual situation before, as opposed to the general type of situation. In terms of risk, 60 per cent of the scenarios were usually judged as dangerous or very dangerous and this probably does reflect the level of hazard to which Scottish police officers are regularly exposed. Only 25 per cent of the situations were thought likely to cause considerable stress to the attending officer.

Table II. Correlation between situation judgement variables

Time Risk to self Risk to others Familiarity

Risk to self

Risk to others

Familiarity

Time

20.197 *

2 0.133 0.662 * *

20.220 * 0.013 0.016

2 0.056 0.467 * * 0.432 * * 0.133

Notes: * p , 0.05, * * p , 0.01

The results do suggest that police officers’ time and risk judgements are in fact related (at least in relation to direct risks for the officer). Increased experience and seniority did not influence the ratings of time, stress or risk. General discussion The two studies represent an attempt to understand police officers’ situational judgements when they attend operational incidents in Scotland. While the scenarios used were designed for this jurisdiction, the findings should have applicability for police forces dealing with a similar range of incidents. It would appear that police officers do discriminate between events in terms of familiarity, available time to make decisions and also in terms of level of risk to self and others. (Some situations present equivalent risks for the officer and the public). As others have suggested (Sykes and Brent, 1983) police officers appear to use their experience to categorise the situations they encounter by type. Time and risk dimensions may be super-ordinate categories. Certainly if some kind of mental pattern matching process is used to recognise the type of situation faced and to rapidly retrieve options for dealing with it, then this kind of categorisation in memory would be required. It has been argued (Crego and Harris, 2002; Stewart and Flin, 1996) that police officers probably use rapid, intuitive decision making strategies, such as recognition-primed decision making (Klein, 1998) when dealing with operational incidents. Certainly the typical characteristics of many police operations (dynamic, high risk, noisy) would not provide conditions conducive to engaging in systematic, analytical comparison of alternative responses. These situational judgement tasks do not provide any information on decision style but the finding that decisions need to be taken within three minutes in many hazardous situations, does suggest that fast recognition plus rule based strategies are being used. The “rules” may well be a mixture of formal procedures (e.g. radio for backup or begin containment), as well as more informal, personal “rules” (if this situation happens, then I always make that response). Certainly for major incidents, the first responding officer’s initial situation assessment and decisions can be key to subsequent command and control: The failure to address critical incident management needs in the early minutes of a rapidly unfolding incident may create problems, such as blocked ingress and egress, no control over volunteers, duplication of effort, and poor utilization of on-scene resources, later on (Sarna, 2002, p. 39).

There were no differences in the ratings relating to expertise or rank, suggesting that judgement patterns were independent of seniority. However, no attempt was made to examine the influence of particular expertise (e.g. authorised firearms officers, traffic specialists) on situational judgements. The scenarios employed in the study presented minimal information and richer descriptions could have provided more cues that might have revealed different effects of experience. More sophisticated methods of cognitive task analysis (Crandall et al., 2006) could be used to analyse police officers’ stored knowledge of situations and how this influences their decision making, such as interviews or computer-based tasks. Possibilities for future research into police patrol officers’ decision making could examine the effects of time pressure on cue selection. Which information do they focus on when they sense that they need to make a response quickly? Do they have strategies

Assessment of operational situations 319

PIJPSM 30,2

320

for “buying time”? Experienced professionals do this better than novices who also have a tendency to react too quickly when under pressure. In operational policing, officers are likely to have to make multiple decisions in complex scenarios (if, when and how to intervene, whether to call for back up, how much force to use etc). It would be useful to explore how they prioritise and schedule these actions and how they make trade-offs between competing goals. In other professions, where situation assessment and decision-making are regarded as safety-critical skills, they are studied and addressed explicitly in training (Flin and Maran, 2004). The kind of scenarios and rating dimensions presented here could be adapted to suit the operational demands of any particular police force. They could also be used as part of a group discussion exercise with police officers on situation assessment and initial decisions (see Crichton et al., 2002 for a description of Tactical Decision Games used with prison offers). As police officers are being called upon to deal with increasingly risky and difficult operational situations, their judgement and decision making expertise become ever more important. Understanding how they “read” situations and use their stored knowledge of events to help them interpret them and decide which actions to take is a subject worthy of police time and investigation. References Banbury, S. and Tremblay, S. (Eds.) (2004), A Cognitive Approach to Situation Awareness: Theory and Application, Ashgate, Aldershot. Brown, J. and Campbell, E. (1994), Stress and Policing, Wiley, Chichester. Burke, E. and Hendry, C. (1997), “Decision making on the London incident ground”, Journal of Managerial Psychology, Vol. 12, pp. 40-7. Cannon, L. (1998), Official Negligence. How Rodney King and the Riots Changed Los Angeles and the LAPD, Times Books, New York, NY. Cohen, M. and Freeman, J. (1997), “Improving critical thinking”, in Flin, R., Salas, E., Strub, M. and Martin, L. (Eds), Decision Making Under Stress, Ashgate, Aldershot. Crandall, B., Klein, G. and Hoffman, R. (2006), Working Minds. A Practitioner’s Guide to Cognitive Task Analysis, Bradford Books, Cambridge, MA. Crichton, M. and Flin, R. (2002), “Command decision making”, in Flin, R. and Arbuthnot, K. (Eds), Incident Command. Tales from the Hot Seat, Ashgate, Aldershot. Crichton, M., Flin, R. and McGeorge, P. (2005), “Decision making by on-scene commanders in nuclear emergencies”, Cognition, Technology & Work, Vol. 7, pp. 156-66. Crichton, M., Flin, R. and Rattray, W. (2002), “Training decision makers – tactical decision games”, Journal of Contingencies and Crisis Management, Vol. 8, pp. 208-17. Crego, J. and Harris, C. (2002), “Training decision making by team based simulation”, in Flin, R. and Arbuthnot, K. (Eds), Incident Command. Tales from the Hot Seat, Ashgate, Aldershot. Endsley, M. (1995), “Toward a theory of situation awareness in dynamic systems”, Human Factors, Vol. 37, pp. 32-64. Endsley, M. and Garland, D. (Eds.) (2000), Situation Awareness, LEA, Mahwah, NJ. Fischer, U., Orasanu, J. and Wich, M. (1995) in Jensen, D. (Ed.), “Expert pilots’ perceptions of problem situations”, In Proceedings of the 8th International Symposium on Aviation Psychology, Ohio State University, Columbus, OH. Flin, R. and Maran, N. (2004), “Identifying and training non-technical skills in acute medicine”, Quality and Safety in Health Care, Vol. 13, Suppl 1, pp. i180-i184.

Flin, R., Salas, E., Strub, M. and Martin, L. (Eds.) (1997), Decision Making Under Stress, Ashgate, Aldershot. Gladwell, M. (2005), Blink. The Power of Thinking without Thinking, Allen Lane, New York, NY. Jensen, R. (1995), Pilot Judgment and Crew Resource Management, Avebury Aviation, Aldershot. Klein, G. (1998), Sources of Power. How People Make Decisions, MIT Press, Cambridge, MA. Kruskal, J. and Wish, M. (1978), Multidimensional Scaling, Sage, Beverley Hills, CA. Lipshitz, R., Klein, G., Orasanu, J. and Salas, E. (2001), “Taking stock of naturalistic decision making”, Journal of Behavioral Decision Making, Vol. 14, pp. 331-52. Loree, D. (1995), Violent Incidents, Canadian Police Research Centre. Royal Canadian Mounted Police, Delta, Technical Report-05-98E, April. Moore, T. (2002), “Police commander – the Notting Hill Riot”, in Flin, R. and Arbuthnot, K. (Eds), Incident Command. Tales from the Hot Seat, Ashgate, Aldershot. Niessen, C., Eyferth, K. and Bierwagens, T. (1999), “Modelling cognitive processes of experienced air traffic controllers”, Ergonomics, Vol. 42, pp. 1507-20. Orasanu, J., Dismukes, K. and Fischer, U. (1993), “Decision errors in the cockpit”. In Proceedings of the Human Factors and Ergonomics Society, 37th Annual meeting, Seattle, October, HFES, San Diego, CA. Orasanu, J. and Fischer, U. (1997), “Finding decisions in natural environments: the view from the cockpit”, in Zsambok, C. and Klein, G. (Eds), Naturalistic Decision Making, LEA, Mahwah, NJ. Rugg, M. and McGeorge, P. (1997), “The sorting techniques: a tutorial paper on card sorts, picture sorts and item sorts”, Expert Systems, Vol. 14, pp. 80-93. Sarna, P. (2002), “Managing the spike: the command perspective in critical incidents”, in Flin, R. and Arbuthnot, K. (Eds), Incident Command. Tales from the Hot Seat, Ashgate, Aldershot. Stewart, E. and Flin, R. (1996), “Taking action”, Policing Today, Vol. 2 No. 4, pp. 14-17. Sykes, R., Brent, E. and Policing, A. (1983), Social Behaviourist Perspective, Rutgers University Press, New Brunswick, NJ. Takane, Y., Young, F. and Leeuw, J. (1977), “Nonmetric individual differences multidimensional scaling: an alternating least squares method with optimal scaling features”, Psychometrika, Vol. 42, pp. 7-67. Taylor, P. (1990), The Hillborough Stadium Disaster, Final Report, HMSO, London. Vrij, A., van der Steer, J. and Koppelaar, L. (1995), “The effects of physical effort on police officers’ perception and aggression in simulated shooting incidents”, Psychology, Crime & Law, Vol. 1, pp. 301-8. Zsambok, C. and Klein, G. (Eds.) (1997), Naturalistic Decision Making, LEA, Mahwah, NJ.

Further reading Fischer, U., Davison, J. and Orasanu, J. (2003) in Jensen, R. (Ed.), “What makes flight situations risky? Examining commercial and general aviation pilots’ concepts of risk”, In Proceedings of the 12th International Symposium on Aviation Psychology Ohio State University, Columbus, OH. Orasanu, J. (1993), “Decision making in the cockpit”, in Wiener, E., Kanki, B. and Helmreich, R. (Eds), Cockpit Resource Management, Academic Press, San Diego, CA.

Assessment of operational situations 321

PIJPSM 30,2

Appendix 1. Sample aviation decision scenario from Fischer et al. (1995) A medium large transport is cleared for takeoff on runway 13C. Power is advanced and takeoff roll commences with the crew setting takeoff power. At approximately 100kts, a crew member notices a light twin engine taxing at a fairly high speed toward 13C. At the same time, he hears the tower controller making repeated unsuccessful attempts to contact the small plane.

322

Appendix 2. Policing scenarios (1) You are on duty at 10 a.m. on a Monday morning. You have been instructed to attend at the Clydesdale Bank [one of Scotland’s high street banks], in town, following a report of an armed robbery. On arrival you are told that shots have been fired. (2) It is 2 a.m. and you are instructed to attend a disturbance outside a guesthouse. On arrival, you are advised that a person has been stabbed and another slashed by one of four youths, who were last seen going into an identified house. (3) It is match day after a home game against Rangers [major Scottish football team], which they have won. Reports come in that the buses carrying the away fans are being ambushed. A group of males are throwing missiles [rocks, bottles, coins], as the buses travel away along a busy street. You are on duty in that street. (4) It is 2 p.m. on a damp Tuesday afternoon, you are instructed to attend at a gas leak at a four-flatted property [A block of flats containing four aprtments]. The leak appears to be coming from the common room landing on the ground floor and all the residents are elderly and bed-ridden. There is no reply at the upstairs flat but the resident is believed to be at home. (5) You are on duty at the football ground turnstiles on a match day. At 3 p.m. the ground is almost empty, however there is a sudden flood of people from nearby pubs trying to get into the ground. There are also families with young children entering. Crushing is reported. (6) You are instructed to attend at a 16-floored multi-storey block of flats where an unknown clear liquid has been dripping into the basement. On arrival you are met by the fire brigade, the fluid is found to be giving off fumes of some sort. It is 7pm and raining heavily outside. The block has numerous elderly residents. (7) You are on duty on a match day, when a report comes in from the local bus company to attend to a disturbance on a double-decker bus. When you arrive, there is a great deal of over-crowding and non-payment of fares. The driver does not wish to move. More people are trying to get on. (8) You are on duty when there is a report of a road accident on a busy street. A tanker containing hazardous chemicals has over turned and the contents are leaking down the hill towards a residential area. (9) You are in an isolated house on the outskirts of the town. A man has doused himself in petrol and is threatening to set fire to himself. There is apparently no one else involved. (10) You attend a domestic dispute. A young girl, who tells you her parents are fighting in the living room, invites you into the house. You enter the room and witness the male punch the female. He also has a knife in his hand. (11) You are on mobile patrol alone and are sent to a neighbour’s concern call. Having ascertained that there is a definite cause for concern for a 40-year-old disabled male occupier, you force entry into the flat. Inside you find the male occupier in bed with his similarly aged girl friend. Both have serious slash wounds to the wrists, his wounds being more severe. A large kitchen knife lies at her side of the bed and a suicide note signed by the male lies at the other. He is unconscious whilst she is semi-conscious.

(12) You are on duty at 1 a.m., checking the rear of premises, when you encounter two males breaking into a shop. (13) You are on duty when you try to arrest a suspect for a serious assault. The suspect and his two accomplices barricade themselves in a top floor flat in a three-storey tenement. When you try to approach the building, the three start throwing items of furniture out of the window to prevent you from gaining entry. (14) You are called to attend a convoy of New Age Travellers in a small village. When you stop them, some of them try to turn off the road and attempt to camp in a nearby field. [New Age Travellers were mobile anarchist/environmental protestors who were commonplace in Scotland at this time]. (15) You are on duty when you are instructed to attend a stabbing within a flat in a run down area of the town. On your arrival there is a male with a serious stab wound. There are six males in the flat, all drunk or high on drugs. All are being very aggressive. There is also a young male child crying, being held by another male in the bedroom. (16) You are instructed to attend at a house fire. Upon arrival, you see two people trapped at an upstairs window. The fire brigade have not arrived yet. (17) You are instructed to attend at a nightclub on Thursday night at 12.30 a.m. The club is full. There are reports that a known felon is inside the club. There is a warrant for his arrest. (18) It is 10 p.m. on a December evening and an elderly lady has been reported missing from her daughter’s home. She suffers from Alzheimer’s disease and requires medication. She is poorly clothed with only her nightgown on. (19) You are on foot patrol within a housing estate at the edge of town at 11.55 p.m. A local resident whom you know well approaches you. He advises you that a drinking party is going on in a nearby estate pub; drugs are also apparently on sale. You know the licensee to be away on holiday and that the premises are only licensed until 11 p.m. (20) You are called to the offices of an American oil company. The office manager asks you in to his office and hands you a paper that has Arabic style writing on it. He explains to you that he removed this from an envelope sent to the office today and that the envelope also contained a grey coloured powder. He wants to show you the powder.

Corresponding author Professor Rhona Flin can be contacted at: [email protected]

To purchase reprints of this article please e-mail: [email protected] Or visit our web site for further details: www.emeraldinsight.com/reprints

Assessment of operational situations 323

PIJPSM 30,2

324

Policing: An International Journal of Police Strategies & Management Vol. 30 No. 2, 2007 pp. 324-328 q Emerald Group Publishing Limited 1363-951X

Perspectives on policing Reducing Homicide through a “Lever-Pulling” Strategy Edmund McGarrell, Steven Chermak, Jeremy Wilson and Nicholas Corsaro Justice Quarterly, Vol. 23 No. 2, 2006 pp. 214-231 McGarrell et al. (2006) examined a problem oriented policing (POP) strategy designed to reduce firearm-related homicides in Indianapolis, IN. The Indianapolis Violence Reduction Partnership (IVRP) was modeled after other successful POP approaches aimed at deterring firearm-related violence. Specifically, researchers implemented a multi-pronged, multi-agency initiative similar to those used in Boston and Minneapolis. Project Ceasefire was part of a larger project known as the “Boston Gun Project” (McGarrell et al., 2006). This initiative targeted youthful gang members known to have committed multiple violent offenses. Through a deterrence-based approach, Project Ceasefire successfully reduced homicides and other firearm-related violence by approximately 60 percent (McGarrell et al., 2006). This strategy was replicated in Minneapolis, MN and produced similar results. Indianapolis provided a third context in which programmatic elements of Project Ceasefire were evaluated. Using a problem solving approach, researchers identified the nexus of Indianapolis’ firearm-related violence. Official police data indicated that repeat offenders were involved in the majority of homicides (McGarrell et al., 2006). Consequently, IVRP enlisted help from federal, state, and local agencies as well as community members to address the city’s firearm-related violence; consequently, this multi-agency task force tailored a response to Indianapolis’ gun violence problem. Much like Boston’s experience, law enforcement agencies in Indianapolis relied on focused deterrence as the lynchpin of their strategy. That is, high-risk offenders were targeted with “lever-pulling” meetings where they were informed they were being watched and any violation of the law would be met with the most severe sanction possible (McGarrell et al., 2006). The approach did not rely solely on the threat of heavy-handed criminal sanctions. High-risk offenders involved in “lever-pulling” meetings were also informed about various social service programs available to them through agencies participating in IVRP (McGarrell et al., 2006). McGarrell et al. (2006) used 54 months of homicide data covering the period January 1, 1997 to June 30, 2001. They employed an ARIMA model to test for significant differences in the number of monthly homicides after IVRP was implemented. After running the appropriate diagnostics to ensure that no autocorrelation and/or lag functions were present, their model suggests that IVRP had an immediate and statistically significant impact on homicides in Indianapolis. More specifically, McGarrell et al. (2006) report IVRP resulted in an initial 34.3 percent reduction in monthly homicides and assert the effect was sustained over the duration

of their evaluation. These findings are congruent with those from the evaluations of Boston and Minneapolis. To rule out the possibility that there was a national or regional downward trend in homicides during the study period, McGarrell et al. (2006) estimated an ARIMA model using six additional cities comparable to Indianapolis. Their results indicated that a slight downward trend was occurring during the study period, but none of the control sites experienced the statistically significant decline in homicides that Indianapolis did. Based on the results of three evaluations, it appears that “lever-pulling” or targeting high-risk, chronic offenders is an effective method to reduce gun related violence in general and homicides more specifically. Although promising results have been reported across three different sites, McGarrell et al. (2006) note that programmatic elements of Project Ceasefire need to be replicated in other cities to ensure that these initiatives are truly the driving force behind homicide reductions. Charles F. Klahm IV University of Cincinnati, USA

Understanding and Preventing Gang Violence: Problem Analysis and Response Development in Lowell, Massachusetts Anthony Braga, Jack McDevitt and Glenn Pierce Police Quarterly, Vol. 9 No. 1, 2006 pp. 20-4 Braga et al. (2006) provide a preliminary evaluation of a problem oriented policing (POP) program undertaken in Lowell, MA. The strategy was designed to accomplish two objectives: identify patterns in gun related violence and homicides, and determine whether firearm violence and particularly homicides were a result gang related activities. Following problem-oriented approaches that have produced successful results in Boston and Minneapolis, researchers began to investigate firearm-related crimes in Lowell. Official data sources from various agencies were used to gauge the extent of firearm-related violence. Their inquiry led them to conclude that Lowell had experienced an increase in firearm-related violence between 2000 and 2002 and, moreover, this violence was concentrated in disadvantaged neighborhoods and committed disproportionately by young, minority males (Braga et al., 2006). Upon further investigation, they found that a majority of offenders and victims were repeatedly involved in criminal activity. Over 80 percent of offenders and over 45 percent of victims had at least one prior on their record and a substantial number were on active probation (Braga et al., 2006). Having identified patterns in firearm-related violence and basic demographic characteristics of offenders and victims, researchers next investigated whether gangs were responsible for Lowell’s increase in gun crimes. Offenders and victims involved in firearm-related incidents were matched against the Lowell Police Department’s (LPD) gang member database to determine if they were associated with a gang (Braga et al., 2006). If there was no match in the database, researchers conducted focus groups with

Perspectives on policing

325

PIJPSM 30,2

326

LPD Investigative Services Division and the Gang Unit to assist with identifying known gang members. Based on their examination, it was determined that over 70 percent of homicides and 35 percent of gun assaults were committed by offenders considered to be associated with a gang (Braga et al., 2006). The initiative in Lowell used a successful programmatic feature from Boston’s Project Ceasefire known as “lever-pulling”. More specifically, high-risk offenders were targeted and informed they were being watched and that any violation of the law would be met with the most severe sanction possible (Braga et al., 2006). Lowell, MA represents a unique scenario in that one of the primary gangs is Asian. The structure and operating techniques of Asian gangs differs from more traditional gangs. Braga et al. (2006) report that elderly members control the actions of younger members and, in order to reduce associated firearm violence, LPD was better served to pressure the illegal gambling halls operated by these elderly members. LPD believed they would be more successful at indirectly controlling gun violence in the Asian community by threatening the primary enterprise of Asian gang elders. Although gang elders were not often involved in violent criminal acts themselves, LPD threatened to close down their gambling rings if younger Asian gang members associated with them committed violent crimes (Braga et al., 2006). The authors only report results from preliminary simple analyses. Examining nine months prior to implementing their problem-oriented approach and nine months after, researchers found that Lowell’s homicide rate was reduced by 50 percent and gun violence in general declined by 24 percent (Braga et al., 2006). Rigorous statistical analysis is necessary to rule out other possible explanations for the sharp decline in firearm-related violence. Notwithstanding this caveat, Braga et al.’s study provides a detailed framework for researchers to follow when implementing a problem-oriented policing approach. This preliminary evaluation demonstrated how problem-oriented responses might have to be augmented as more information is learned, as was the case with their response to Asian gangs. Although a complete evaluation of the Lowell, MA project has yet to be published, in light of other research findings the preliminary results reported here suggest that “lever-pulling” continues to be a successful law enforcement intervention. Charles F. Klahm IV University of Cincinnati, USA

Race and Officer Decision Making: Examining Differences in Arrest Outcomes between Black and White Officers Robert Brown and James Frank Justice Quarterly, Vol. 23, No. 1, 2006 pp. 96-126 Brown and Frank (2006) contribute to the overall understanding of officer decision-making by presenting results from an observational study of Cincinnati Police Department officers. The data used in their analysis was collected between April 1997 and April 1998, during which observers witnessed 614 police-suspect encounters.

In order to determine possible differences in arrest decision-making, their analysis was restricted to observations of police-citizen encounters where the citizen involved was considered to be a suspect at some point during the contact (Brown and Frank, 2006). Their primary research question addressed the possibility that influences on officer decision making vary by officer race. That is, are white and black officers influenced by the same factors when it comes to arrest decisions? The authors originally intended to estimate and report Hierarchical Linear Models (HLM) to assess the nature of officer decision making. However, due to the small number of cases nested within each geographical area, concerns about the stability of the coefficients needed to be addressed. Brown and Frank (2006) thus estimated and reported a series of logistic regression models in order to assess influences of officer race on arrest decisions. Initially a model including all police-suspect encounters was estimated, producing results consistent with prior research. Specifically, the authors found that officers were more likely to arrest younger, non-white, male suspects and that quality of evidence, suspect demeanor and whether the suspect was intoxicated influenced arrest decisions (Brown and Frank, 2006). Although officer race was not included in the reported model, Brown and Frank (2006) footnote that white officers were more likely to make an arrest than black officers when they estimated an additional model (not reported) including all police-suspect encounters. As previously mentioned, the nexus of this article was to determine if certain factors differentially affect black and white officers’ decision-making processes. Consequently, Brown and Frank (2006) partitioned the data and estimated two additional models, one for black officers and one for white officers. Contrary to conventional wisdom, their analysis revealed that exogenous variables in the models differentially influenced black and white officers’ arrest decisions. When arrest decisions were examined separately, certain factors influenced black officers’ decision making but produced no such effect on white officers’ decisions to arrest. For example, black officers’ decisions were influenced by length of service, suspect race[1], number of bystanders and whether a crime was committed in the presence of officers, but none of these factors affected arrest decisions for white officers (Brown and Frank, 2006). Conversely, the authors found that white officers were influenced by the quantity of evidence and being dispatched to the scene whereas black officers were not. While these differences between officers are important, it should be noted that only two correlates exhibited statistically different effects on officer decision making. Despite the fact that white officers were more likely to make an arrest in general, black officers were more likely to arrest black suspects yet less likely to arrest white suspects than their counterparts (Brown and Frank, 2006). The only other factor that differentially influenced black and white officers significantly was years of experience. The longer black officers worked with CPD the less likely they were to make an arrest. This was not the case for white officers. The findings presented by Brown and Frank (2006) diverge from previous study results and suggest officer race does not have a null effect on arrest decisions. At least in the sample they examined, white and black officers were differentially influenced by certain factors surrounding the police-suspect encounters. Although their findings could be context specific and most previous studies indicate these factors have little

Perspectives on policing

327

PIJPSM 30,2

influence over officer behavior, future research should consider these correlates to determine the true relevance of officer level characteristics on decision making. Charles F. Klahm IV University of Cincinnati, USA

328

Note 1. There is a discrepancy between the results reported in Table 2 on page 117 and the discussion relating to influences on black officers behavior in the “Discussion” section on page 119. Personal correspondence with the authors confirmed that reported coefficients and significance indicators in Table 2 are correct. In the “Discussion” section the authors report: “In our Black officer model arrest outcomes were influenced by length of service, suspect gender, whether a crime was committed in their presence, and by the number of citizen bystanders witnessing the encounter, none of which significantly influenced the behavior of White officers in the study” (pg. 119). Accordingly, it would be interpreted that black officers were more likely to arrest male suspects than white officers. However, Table 2 indicates that white officers were also more likely to arrest males than females thus creating the contradiction. The passage on page 119 should have indicated that black and white officers were differently influenced by suspect race, not gender. Therefore, the correct interpretation should be that black officers were more likely to arrest minority suspects than white officers.

Policing on the web Crimes Against Children Research Center Crime committed against children is a subject that has received a great deal of media attention in recent years. With the development of the internet, fears and concerns about crimes targeting children increased dramatically. However, until recently, there have been few scholarly attempts to study crimes committed via the internet. Of the recent efforts to study internet crimes, one stands out, both in its comprehensiveness and in the availability of the findings. The Crimes against Children Research Center (CCRC), created in 1998, at the University of New Hampshire has conducted two national surveys of online youth, the results of both are available via their web site (www.unh.edu/ccrc/). The first Youth Internet Safety Survey was launched in fall of 1999 and continued through the spring of 2000. The survey involved a phone survey of a nationally representative sample of America’s youth between the ages of 10 and 17 years. The second Youth Internet Survey was conducted five years later. The results of the two surveys are available on the CCRC’s web site, including publications concerning such topics as the prevalence of youth harassment online, the characteristics of online harassment considered distressing by online youth, children’s exposure to online pornography, the impact of internet use on children’s mental well-being, risk assessments and many others. In addition to the published findings concerning the Youth Internet Safety Survey, the CCRC web site also offers links to resources concerning child victimization, such as tip-lines, contact information for children’s services agencies, links to relevant web sites and reports from various other studies with the CCRC has been, or is currently, involved. With many police departments making the decision to form special internet investigation units, join task forces or to be a valuable resource for their community, there is certainly a demand for information concerning internet crime. For departments seeking such information, especially information concerning internet crimes against children, the CCRC web site will prove to be a valuable, if not essential, resource.

Policing on the web

329

J.W. Carter II University of Cincinnati, Cincinnati, USA

Policing: An International Journal of Police Strategies & Management Vol. 30 No. 2, 2007 p. 329 q Emerald Group Publishing Limited 1363-951X

PIJPSM 30,2

330

Policing: An International Journal of Police Strategies & Management Vol. 30 No. 2, 2007 pp. 330-332 q Emerald Group Publishing Limited 1363-951X

Book review Citizens, Cops, and Power Steve Herbert University of Chicago Press Chicago, IL Review DOI 10.1108/13639510710753298 With the growing body of literature on community policing, one does not often see so superlative a piece of work that combines the study of political theory and policing to explore the relevance of community policing as both a philosophy and a practice. Steve Herbert utilizes a theoretical treatment to analyze a qualitative case study conducted in West Seattle to examine the role of community in local governance. In this exploration, Herbert seeks to engage the reader in asking two pivotal questions. First, should “community” be a legitimate and effective political actor with the ability to engage in community policing? Second, if the ability exists, how should the community interact with agencies of the state, such as the police? To this point, these two provocative questions have eluded significant and consistent consideration. Herbert’s begins this exploration by first discussing the terrain of community, melding both the notion of humans as not only social creatures but also as taking on the Aristotelian typology of being political creatures as well. The introductory piece goes on to further define community in three aspects, community as recovered (or thick version I), community as discovered (or thick version II), and the thin version of community where citizens are viewed as the means to an end versus being ends in themselves. Since community policing partly relies upon civic engagement and democracy at the local level, this first chapter provides this discussion on community in an effort to frame the question of whether neighborhoods truly are communities. In the end, Herbert discovers that the community he is studying does not necessarily meet the standard for thick or thin and that the analysis of the political capacity of the community deserves further attention. In the following chapter, Herbert deals specifically with the question of political capacity. As he notes, community policing epitomizes notions of neoliberalism with the devolution of power to the local level in an effort to increase self-determination. Therefore, communities and neighborhoods must have some level of political capacity. However, as Herbert finds, the political capability of a community largely relies upon its economic health. As a result, disadvantaged neighborhoods will have the propensity to suffer the greatest and will not prosper from a policing philosophy that has increased expectations of communal governance. Herbert further delves into the issue examining the relationship between the police and the community, how the police structure their work, and their ultimate quest for legitimacy. The community policing philosophy stresses the need for the police to engage in co-production with the community, solicit input from local groups, and subsequently respond to that input. Yet, as Herbert finds, the police tend to separate themselves because they are members of a politically vulnerable institution that can readily and easily endure pain from ill-motivated activists and public officials. This

separation from the public is an important mechanism of protection for the police. Therefore, the police construct themselves as a separate and powerful social group. Furthermore, Herbert finds that the police also construct community through their bureaucratic routines. As such, the state generates the community by defining what a community actually is. In addition, as Herbert has noted, traditionally there is a resistance from the police to engage in community policing. Herbert engages the reader in a discussion of how the police have constructed a reality in which traditional police action is valorized over community action. Subsequently, the police view the community as a mere information provider and a tool at their disposal, not a partner in co-production. As a result, this study finds that the police further separate themselves through this cultural construct because it views traditional police work as being masculine, adventurous, and full of machismo. A truly equal partnership of co-production would involve a feminization of police work that would erode this self constructed paradigm. Toward the latter part of the book, Herbert begins to discuss some of the findings of his qualitative study in regards to the relations between police and community. The findings indicate that in West Seattle the residents demonstrated a strong acceptance of the police and displayed and interest in working as closely as possible with officers. Furthermore, the residents viewed the police as being indispensable in making their neighborhoods safe and livable. However, Herbert also found that many citizens felt that the police’s impulse towards authority is too often over emphasized and utilized unnecessarily. In this study, the community also expressed frustration with the police construction of the community. Although the community expressed an understanding for the police construction, residents frequently complained about how their complaints were distorted, ignored, or lost forever in the bureaucratic process. In short, Herbert provides this evidence to help the reader understand why it so difficult to find a way that the police-community relationship, as an example of the state-society relationship, can further collective and cooperative problem solving. In the final chapter, Herbert concludes that the evidence points towards skepticism about the promise of community policing and other efforts at increased local self-governance. Herbert states that the community cannot be expected to be a vigorous force for political action. Additionally, the community cannot relate in a simple manner to the police or any other component of the state. Therefore, Herbert ends by exclaiming that community is an ineffective tool to increase the political capacity of urban neighborhoods and improve citizen oversight of state agencies. Herbert has done a superlative job at looking at the complex relations, both political and social, that construct the manner in which the community and the agents of the state operate. Herbert deserves praise for tackling the effectiveness of community policing from the unique angle of political theory and local governance. Yet, after reading this piece, one wonders if there is any other way from which this topic could be viewed. This book will undoubtedly be of interest to a wide variety of readers. While the book seeks to discuss the relationship between community and the police in carrying out crime control, it will also be of interest to those interested in civic engagement and local governance at the neighborhood levels. Herbert’s case study of a city that has adopted community policing provides us with some evidence that perhaps we should rethink this strategy. First, our communities may not have the political capacity or the desire to participate in efforts of local

Book review

331

PIJPSM 30,2

332

governance. Second, are police the most effective agents of the state to carry out this mission, given Herbert’s findings? Additionally, the point should not be lost that Herbert identifies that political and economic actions at the macro levels of the government and social system undoubtedly have impacts on the neighborhoods at the micro levels. Subsequently, these small, localized groups will have little impact, if any, on the larger dynamics of the system. Therefore, it may be unfair for us to place this locus of responsibility upon the community. As a comment, and not a criticism, it would be interesting to explore this issue by examining a diverse set of communities composed of various backgrounds and housing types. Although the case study of West Seattle provides us with some intriguing and important findings, the results to this point are undetermined as to their universal application. Do these findings hold true within and across communities? This is work for future scholars, but Steve Herbert has taken us in a significant leap forward. In all, Herbert has built upon his previous work and his future works are much anticipated. Matthew A. Jones Portland State University, Portland, Oregon, USA