2.11....9n17 -. .: ~ fixihkn...» 5 '13 “g, . :a :93; . (Javatkux. 1...... . .3. ... 2.2. .1 It R .3 Itib It > L a. SEN-gt. .a re {a . .L .l. , :Lr: . t: .. thr i . .. . ‘. I'IMHIHN. 2: Gus gnaw? w .. . 3 a... w. L . i . ,du. . x I: :8 .8: 1! A. .. . 3 x .. EX‘ )I.” t. i I... zit-3:75.39 :t t 3.10 ‘ A I 35%. M . . . ‘3. 3.. V 3...». u . 3| . TE)? . 2:... .. mmfi. iazhfiugitu: 1.4.9.. .. . £49 . .2. i dun i; xi an. n..- I? . {icgit I 11.1.. 3 21...: film. This is to certify that the dissertation entitled POLICE SUPERVISORS AND COMMUNITY POLICING: SUPPORTING OR SUBVERTING CHANGE? presented by Meghan Sarah Stroshine has been accepted towards fulfillment of the requirements for Ph.D. degree in criminal Justice Major prof ssor Date 2‘3 W 2,07.) 7/ MS U is an Affirmative Action/Equal Opportunity Institution 0- 12771 UBRARY Michigan State University PLACE IN RETURN Box to remove this checkout from your record. TO AVOID FINES return on or before date due. MAY BE RECALLED with earlier due date if requested. DATE DUE DATE DUE DATE DUE MAR 06 2004 D4170; 6/01 cJCIRC/DaleDue.p65-p.15 POLICE SUPERVISORS AND COMMUNITY POLICING: SUPPORTING OR SUBVERTING CHANGE? By Meghan. Sarah Stroshine A DISSERTATION Submitted to Michigan State University in partial fulfillment of the requirements For the degree of DOCTOR OF PHILOSOPHY Criminal Justice 2002 ABSTRACT FIRST-LINE POLICE SUPERVISORS AND COMMUNITY POLICING: SUPPORTING OR SUBVERTING CHANGE? By Meghan Sarah Stroshine According to many scholars, the support of first-line police supervisors is a critical component of the successful implementation and practice of community policing. Police supervisors are responsible for translating the broad mission and goals of a department into meaningful directives for subordinates, a task that is particularly difficult during a time of organizational change. Despite the important role police supervisors play in the ultimate success of community policing, few researchers have attempted to understand the reactions of police supervisors to community policing, nor have many tried to identify the factors that facilitate the support of community policing by these individuals. Of the research that does exist, the preponderance of past studies rely solely on attitudinal measures of support. These studies assume a link between attitudes and behavior; that the attitudinally supportive supervisor will also be the most likely to practice community policing. Prior research, however, has provided reason to question the link between attitudes and behavior. The current study sought to fill this gap in knowledge. Using data collected from a large—scale, multi-method study of policing in two departments practicing community policing (the Project on Policing Neighborhoods), this study measured the reactions of first-line police supervisors to community policing in behavioral terms (i.e., time spent engaged in community policing activities). Supervisors’ attitudes toward community policing were included as a key independent variable. A framework borrowed from the organizational psychology literature situated this study in the larger context of research on the link between attitudes and behavior, allowing not only for an assessment of the extent to which supervisors can “talk the talk” of community policing (i.e., attitudes), but also of the extent to which they also “walk the talk” (behaviors). The theoretical framework employed in this study also identified five categories of variables (i.e., demographic, personal characteristics, role states, work experiences, and group/leader relations) that were expected to influence the extent to which supervisors would engage in community policing. Ultimately, results did not support the hypotheses suggested by the theoretical framework used to guide this study. In the end, this research provided more information about the factors that influence supervisors’ attitudes toward community policing (originally conceptualized as an independent variable) than about their behavior. While this was not the initial goal of this research, these results add to the body of work in this area, and suggest areas of future inquiry. To my father, Robert W. Stroshine iv ACKNOWLEDGEMENTS This dissertation would not have been possible without the help and encouragement of many people. First and foremost, I would like to thank the members of my dissertation committee: Peter K. Manning, Stephen D. Mastrofski, J. Kevin Ford, Michael D. Reisi g, and Christina DeJong. I must extend particular thanks to the two men who served as the Chair of my committee at different times. Steve Mastrofski was my Chair until he left Michigan State University for George Mason University. Steve, I wanted you to be on my committee because I knew you would push me, challenge me, and even frustrate me, all in the name of improving me. You made my graduate career seem like a Sisyphean task at times, demanding more of me than I thought I was capable of, but you made me a much better scholar in the process. Thank you. Peter took over as Chair of my dissertation committee after Steve left. Pete, your friendship, generosity, and words of encouragement sustained me throughout my graduate career. In you, I had the ultimate support system. You believed in me more than I believed in myself, and that has made all the difference in the world. Not only did you contribute to my growth as an academic in countless ways, but you also made my graduate career an immensely enjoyable one. I will never forget our El Azteco seminars and (im)moderate nights out on the town. I would not have completed this dissertation without the help and support of many other people in my life. My dad provided me with the drive and confidence to pursue my dreams. Dad, you have been wonderful in encouraging me to know myself, my dreams, and then providing the emotional (and financial) support necessary for me to pursue those dreams. I hope that you look upon the success that Caryn and I are enjoying as a measuring stick of your success as a parent. We wouldn’t be where we are today without your love and support. My sister Carrie has often been my biggest fan and supporter, reminding me what I was doing and why (which often seemed the most elusive of answers). She has consistently demonstrated the meaning of unconditional love to me. In her, I have been fortunate to not only have a great sister, but a best friend as well. While I have been blessed with the family that God has given me, this section would not be complete without acknowledging the friends who have become the family I have chosen for myself. I wonder at times if I would be where I am today had it not been for Tom. Tom has been my rock over the years, providing me with a base of love, support, and confidence that made any goal seem possible. I will always be grateful to him for his nurturing and support. I was lucky enough to make some wonderful friends at MSU that made my time there as a graduate student not only bearable, but a time of my life I will always look back on with great fondness. I entered the program with Amanda L. Robinson, without whom I cannot imagine my graduate experience. The only way to deal with the stresses of graduate life is through humor and, occasionally, a few too many beers. In Amanda, I found the ultimate friend, comedienne, and drinking companion. It should not go unmentioned that I also found a great collaborator in Amanda. Chanaman rules! I met M. Kevin Gray the summer before moving to Michigan. He is one of the funniest people I have ever known, as well as one of the sweetest. I often think that vi Kevin and I are the same person in different bodies. We are eerily similar in things we find funny, entertaining, intriguing, and (especially) anxiety-inducing. I think this has made us particularly good at supporting each other over the years. Kevin and Amanda were pivotal in making my time at Michigan State a period of my life that will always hold a special place in my heart. It seems that in our five years at Michigan State we created a lifetime’s worth of memories. The in-jokes (e.g., “how’s 9, “ your bird?,” the “masta, get your f$%&in’ facts straight”) are endless. I can’t imagine what graduate life would have been like without either of them in it, nor would I want to. Finally, there are people who played a less direct, but still important, role in my completing this dissertation. Other friends on the first floor of Baker Hall at Michigan State—Sean and Tracy Varano, Jeff Cancino—always made work a place I enjoyed to be (so much so that not much work ever got done there). Despite the distance and miles between us, my friends Maria and Amy also been wonderful in the support and encouragement they have given me. My extended family has also been important. One of the reasons I chose to attend Michigan State was because of its close proximity to my many aunts, uncles, and cousins. I will forever be grateful for the time I was able to spend with them; I think this time afforded me the opportunity to be much closer to them than I would have been otherwise. I am especially grateful to my aunts, who helped me fill in many gaps about my family and the people I come from, ultimately helping me to get to know myself better in the process. Finally, I would be remiss not to mention people in Omaha, NE, who helped me through the final part of my graduate journey, (finally!) finishing this dissertation. I vii started my first job at the University of Nebraska at Omaha before completing my dissertation. My first few months in Omaha were consumed with trying to adjust to a new life, a new city, and a new job. Steve Wilmes and Denise Herz were very good friends to me during this difficult time. I will always be grateful to them for encouraging me to finish this dissertation once and for all, for welcoming me into their circle of friends, and for helping me to begin a new phase in my life. As this acknowledgement section demonstrates, while I am ultimately responsible for the completion of this dissertation and its contents, many people have played a role in who I am and where I am today. I am grateful for the opportunity to thank them in this acknowledgement section; I only hope I have done well in expressing these sentiments all along. viii TABLE OF CONTENTS LIST OF TABLES LIST OF FIGURES INTRODUCTION CHAPTER 1 POLICE SUPERVISION AND COMMUNITY POLICING Traditional and Community-Oriented Models of Policing Sources of Legitimacy Function of the Police Organizational Design Tactics and Technology Outcome Measures Supervision and Community Policing The Role of Police Supervisors Values of Police Supervisors Management Techniques Employed by Police Supervisors CHAPTER 2 THE REACTIONS OF POLICE SUPERVISORS TO POLICE REFORM: SUPPORT OR SABOTAGE? Quantitative Studies Quantitative Accounts Summary/Conclusion CHAPTER 3 FACTORS INFLUENCING SUPPORT FOR COMMUNITY POLICING: WHAT MATTERS? Training Participation Role Conflict Resources Emphasis on the Control Function of Supervisors Rewarding the Status Quo Opportunity Group Subculture Background Factors Summary/Conclusion CHAPTER 4 THEORETICAL FRAMEWORK & HYPOTHESES Problems with Previous Research on Supervisor Support for Community Policing xii xiv @0000 10 ll 13 14 14 18 19 21 22 31 34 37 37 38 4O 40 41 42 43 44 46 47 49 49 Prior Operationalizations of Support for Community Policing The Atheoretical Nature of Past Research Situating Supervisor Support for Community Policing in the Larger Context of Research on the Link between Attitudes and Behavior Theoretical Framework Literature on Organizational Commitment and Organizational Citizenship Behavior Antecedents of Organizational Commitment Consequences of Organizational Commitment Summary Reconceptualizing the Causal Ordering of Personal Characteristics, Role States, Job Characteristics, Work Experiences, Organizational Commitment and Organizational Citizenship Behavior Argument #1: Need to Account for Literature on Organizational Citizenship Behavior Argument #2: Accounting for the Causal Ordering of Attitudes Toward Community Policing, Role States, Job Characteristics, Work Experiences, Group/Leader Relations, and Community Policing Behaviors Applying the Organizational Commitment/Citizenship Behavior Framework to the Study of Supervisor Support for Community Policing A Conceptual Model of the Factors Affecting Police Supervisors’ Community Policing Behavior Hypotheses Model 1: Demographic Characteristics Model 2: Personal Characteristics Model 3: Role States Model 4: Work Experiences Model 5: Group/Leader Relations Conclusion CHAPTER 5 DATA AND METHODS Data Site Descriptions Description of the Data Sample Measures Dependent Variable Independent Variables Summary CHAPTER 6 STATISTICAL ANALYSES AND RESULTS .‘F. 49 51 52 54 6O 61 72 75 77 78 82 85 86 86 88 89 89 90 91 91 93 93 93 99 103 105 106 l 10 124 Collinearity Diagnostics Bivariate Analyses Multiple Regression Analyses Negative Binomial Regression Results Post Hoc Analyses Summary CHAPTER 7 DISCUSSION AND CONCLUSIONS Methodological Issues: What the Results Cannot Tell Us and Why Measurement Issues What the Analyses Do Tell Us: Substantive Matters The Practice of Community Policing by Police Supervisors Attitudes toward Community Policing Directions for Future Research Conclusion APPENDIX A CORRELATION MATRIX: DEPENDENT AND INDEPENDENT VARIABLES APPENDIX B TIIVHS ENGAGED IN COMMUNITY POLICING BY OBSERVATIONAL SESSION REFERENCES xi 126 132 134 134 138 143 145 145 145 150 150 152 157 160 162 165 168 Table 1. Table 2. Table 3. Table 4. Table 5. Table 6. Table 7. Table 8. Table 9. Table 10. Table l 1. Table 12. Table 13. Table 14. Table 15. Table 16. Table 17. LIST OF TABLES Summary of Extant Research A Comparison of Relationships between Predictor Variables in the Organizational Commitment and Organizational Citizenship Behavior Literatures Comparison of Variables Drawn from the Organizational Commitment, Citizenship Behavior, and Policing Literatures Sample of Indianapolis Police Supervisors by District, Shift, and Rank Sample of St. Petersburg Police Supervisors by District and Shift Attitudinal Measures of Support for Community Policing Demographic Characteristics Ability Index Survey Items Perceived Competence Survey Items Role Conflict Survey Items Reward Survey Items Participation Survey Items Opportunity Measures Group Cohesiveness Measures Leader Supportiveness Survey Items Zero-Order Correlation Coefficients and Tolerance Statistics: Demographic Variables Zero-Order Correlation Coefficients and Tolerance Statistics: Personal Characteristics xii 48 81 85 104 105 111 111-112 112-113 114-115 116-117 118 119 120 123 124 Table 18. Table 19. Table 20. Table 21. Table 22. Table 23. Table 24. Table 25. Table 26. Table 27. Table 28. Table 29. Zero-Order Correlation Coefficients and Tolerance Statistics: Role Conflict Variables Zero-Order Correlation Coefficients and Tolerance Statistics: Work Experience Variables Zero—Order Correlation Coefficients and Tolerance Statistics: Group/Leader Relations Variables Bivariate Correlates of Time Spent Engaged in Community Policing Negative Binomial Regression Results: Demographic Model Negative Binomial Regression Results: Personal Characteristics Model Negative Binomial Regression Results: Role Conflict Model Negative Binomial Regression Results: Work Experiences Model 129 130 131 133-134 135 136 136 137 Negative Binomial Regression Results: Group/Leader Relations Model Logistic Regression Results: Attitudes toward Community Policing Regressed on Role Conflict Variables Logistic Regression Results: Attitudes toward Community Policing Regressed on Work Experience Variables Attitudes toward Community Policing: Summary of Findings xiii 137 142 143 154 Figure 1. Figure 2. Figure 3. Figure 4. Figure 5. Figure 6. Figure 7. Figure 8. LIST OF FIGURES Antecedents and Consequences of Organizational Commitment and Organizational Citizenship Behavior Model Predicting Community Policing Behavior Among Police Supervisors Distribution of Time Spent on Community Policing Per Shift Breakdown of Supervisor Time Per Shift Number of Observational Sessions (Rides) Per Supervisor Percentage of Supervisors’ Shift by Task Percentage of Discretionary Time Spent Engaged in Community Policing Antecedents and Consequences of Attitudes toward Community Policing xiv 76 87 107 108 140 151 152 157 INTRODUCTION FIRST-LINE POLICE SUPERVISORS AND COMMUNITY POLICING: SUPPORTING OR SUBVERTING CHANGE? Community policing has been aptly described as a promise to change both the “means and ends” of policing (Goldstein, 1990; Mastrofski, 1991). The “ends” of policing are broadened beyond a narrow focus on “fighting crime” to incorporate quality of life issues, such as the reduction of disorder and the prevention of crime. The “means” also change substantially; a central feature of community policing involves engaging the public in the co-production of order (Kelling & Moore, 1988). Consequently, community policing represents no less than a large-scale organizational change (Zhao, Thurman, & Lovrich, 2000). Middle managers play a critical role in successful organizational change. Some scholars argue that the support of police supervisors’ is particularly important to the successful implementation of community policing (Alpert & Dunham, 1989; Bayley, 1994; Geller & Swanger, 1995; Goldstein, 1990; Skogan & Hartnett, 1997; Wycoff & Skogan, 1994); several studies have attributed the failure of community-oriented initiatives directly to supervisor sabotage (e.g., Sadd & Grinc, 1994; Sherman, 1975; Sherman, Milton, & Kelly, 1973). Supervisors are responsible for translating the broad mission and goals of an organization into meaningful directives for subordinates. In this way, first-line supervisors are the most salient representatives of the administration’s goals and policies (Lewin, 1943). Given the top-down nature of most reform, middle- In this paper, first-line supervrsors rs used interchangeably wrth middle-management and other similar terms; all terms reference individuals at the rank of sergeant or lieutenant who have responsibility for street supervision of subordinate officers. managers play a critical role in the change process. If managers resist change, the prognosis for successful organizational change is bleak. As Coyle-Shapiro (1999) commented: The dependency of change at one level on the level above creates a fragility in this linking-pin approach to change, which, if broken, could present obstacles to a smooth and organization-wide change. . . The support of first-line supervisors is crucial to effecting change at the level of employees (p. 4). Despite the importance of police supervisors to the successful implementation of community policing, very little is known about how supervisors perceive and react to this change.2 Of the research that does exist on the topic, most studies indicate that supervisors are “supportive” of community policing.3 This research, however, is characterized by significant flaws. First, most researchers have measured support for community policing by relying solely on attitudinal indicators of support.4 This research assumes that positive attitudes will translate into the practice of community policing, which may not be the case. Evidence of support for community policing ultimately comes in the actions and deeds of police, not their attitudes. It is not enough to know whether supervisors can “talk the talk” of community policing; we must determine whether they can (and will) “walk the talk.” This determination may only be made by conducting research which examines the extent to which supervisors engage in the practice of community policing. A second shortcoming of past research in this area is 2 A much larger body of literature exists regarding the reaction of police officers to community policing. 3 Here, I am referring to the quantitative (as opposed to qualitative) explorations of the topic. Qualitative studies almost overwhelmingly depict supervisors as resistant to community policing. This disparity, and the reasons for it, are explored in more depth in later chapters. 4 Behavioral indicators of support are considered superior to attitudinal measures because these measures provide firm evidence of support; ultimately, support is something demonstrated in actions and deeds, not feelings. While preferable. behavioral measures of support require observational research—a rarity in police research. that it has been largely atheoretical in nature. As a result, research on the topic more closely resembles a fishing expedition than a logical inquiry into the matter. Moreover, we are left with little understanding of the factors which may influence the extent to which supervisors engage in community policing. Situating the study of supervisor support for community policing in the larger context of the study of the link between attitudes and behavior will allow for a more definitive conclusion regarding the reaction of police supervisors to community policing. Current work in this area, relying on attitudinal definitions of support, has suggested that previous (19703) accounts of supervisor resistance to change overstate the case (Weisel & Eck, 2000). Such an assertion, however, is premature at best, and misleading at worst, until researchers conduct studies of this phenomenon using behavioral indicators of support. Ultimately, it is important whether supervisors will walk the walk of community policing, not merely talk the talk. Borrowing a theoretical framework drawn from work in the areas of organizational commitment and organizational citizenship behavior is helpful on several fronts. First, it provides a structure lacking in other examinations of supervisory reactions to community policing. This structure provides a logical set of variables that may influence supervisor support for community policing, and suggests several sets of testable hypotheses. Second, this framework is particularly well-suited for the study of the link between attitudes and behavior. This theoretical model acknowledges the importance of attitudes, while pointing to other—and perhaps more influential—sources of influence on behavior. In summary, while many scholars contend that police supervisors play a critical role in deciding the ultimate fate of community policing (e.g., Geller & Swanger, 1995; Sherman, 1975), very few studies have examined the reactions of police supervisors to community policing. Not only is this type of research rare, but the research that has been conducted is lacking. The extant research uses inappropriate—or certainly less than desirable—measures of support for community policing. Moreover, research in this area is characterized by barefoot empiricism rather than structured, theoretically grounded inquiry. These problems may be ameliorated by defining community policing as a behavioral phenomenon, and by specifically drawing on work conducted in the areas of organizational commitment and organizational citizenship behavior for a theoretical framework. The current study aims to fill the voids in knowledge surrounding supervisors and community policing by providing the answers to two research questions: (1) to what extent do supervisors support community policing, and (2) what factors significantly increase the likelihood of supervisory support of community policing? In providing answers to these research questions, the current study improves upon the extant research by employing a behavioral definition of support for community policing and by adopting a theoretical framework from the field of organizational psychology. The research questions are assessed using data collected during the Project on Policing Neighborhoods (POPN), a large-scale study of police officers and supervisors in Indianapolis, Indiana, and St. Petersburg, Florida. In the following sections, I provide the foundation for and describe the proposed study. In Chapter One, I compare and contrast the differences between traditional and community-oriented models of policing along several criteria. This discussion provides the backdrop for a consideration of the ways supervision is expected to change under a community-oriented model of policing, particularly with respect to the roles, values, and management techniques employed by police supervisors. In Chapter Two, I review the extant literature with respect to the reactions of police supervisors to community policing and prior reform initiatives (e.g., team policing). This discussion reveals that qualitative and quantitative studies have provided opposing views of police supervisors’ reactions to community policing. Specifically, qualitative studies, relying on behavioral definitions of support, tend to show that supervisors are overwhelmingly resistant to community policing, while positivistic studies (defining support for community policing as an attitudinal phenomenon) indicate that supervisors are generally supportive of this organizational change. This discussion has important implications for the current study, as it provides the rationale for not only defining the dependent variable in behavioral terms (i.e., the time supervisors spend engaged in community policing), but also because it identifies a key—and as yet untested— assumption that constitutes the main research question addressed by the current study: do positive attitudes toward community policing translate into community policing behaviors? Drawing on policing literature that has examined the “officer impact” of community policing, Chapter Three presents the current state of knowledge on factors that may play a significant role influencing the extent to which supervisors will engage in community policing. Several factors emerge in the literature, in particular: (1) training on community policing, (2) participation in the implementation process, (3) role conflict, (4) opportunity to engage in community policing, (5) group subculture, and (6) select personal demographics. Chapter Four provides the theoretical framework used to guide the current study, using literature drawn from organizational psychology on the topics of organizational commitment and organizational citizenship behavior. I argue that organizational commitment (i.e., support for organizational goals) may be considered largely analogous to attitudinal support for community policing, and organizational citizenship (i.e., pursuit of goals that benefit the organization) largely similar to the practice of community policing by police supervisors. Not only are organizational commitment and citizenship behavior similar to the support and practice of community policing from a conceptual standpoint, but the factors found to influence these outcomes are very similar to the factors identified in Chapter Three as significantly influencing supervisor support for community policing. Drawing on the literature on organizational commitment and citizenship behavior, I delineate a total of five models that will be tested using the data. The chapter concludes with a discussion of these models and hypotheses to be tested in the current study. Chapter Five provides a description of the Project on Policing Neighborhoods (POPN) study and the data collected during the course of this study that is used in this dissertation. This chapter also provides a detailed explication of the measurement of variables used in the current study, along with the presentation of some descriptive statistics on the dependent and independent variables. Chapter Six begins with a discussion of the data. First, I present the results of collinearity diagnostics performed on the independent variables. Second, I provide the results of bivariate (i.e., ANOVA) analyses performed to determine the nature of the relationships between the independent and dependent variables. This chapter concludes with the results of multivariate analyses (i.e., Negative Binomial and Ordered Probit) conducted on the data. Chapter Seven begins with a summary of the results presented in Chapter Six. These results are discussed from several vantage points. First, the results are assessed with regard to the hypotheses first presented in Chapter Four. Second, policy recommendations (where applicable) are offered. Finally, the results are discussed with regard to limitations of the current study, and by offering recommendations for future research in this area. CHAPTER ONE: POLICE SUPERVISION AND COMMUNITY POLICING In many community policing programs it is simply assumed that police officers can act in a “community orientation,” that police organizations can support this emerging style of policing, and that communities can differentiate community policing actions from those of traditional policing. Such assumptions produce the illusion that it is relatively easy for police agencies to convert from traditional to community policing. Nothing could be farther from the truth (Greene, 1998a, p. 142). In most police departments across the country, a traditional paradigm of policing has been replaced by a community policing model (Pelfrey, 1998). Among other changes, this shift represents transformations in the legitimacy and function of the police, the organizational design of police departments, the tactics and technologies employed by police, and the outcomes measured to establish “success” (Kelling & Moore, 1988). In this section, I contrast traditional and community-oriented models of policing with respect to these factors. After a discussion of these changes, I provide a consideration of the implications these changes have for police supervisors. Traditional and Community-Oriented Models of Policing Sources of Legitimacy Police acting according to a traditional paradigm of policing primarily derive their legitimacy from the law (Kelling & Moore, 1991), which “accord[s] to the police those special powers required to enforce the law and take action where peaceful means have not or may not be effective to secure good order (Bittner, 1970)” (Mastrofski, 1991, p. 61). Importantly, the needs and wants of the public only indirectly influence the police. While police operating under a traditional model of policing “ . . .understand that they are creatures of municipal governments, and are, to some degree, accountable to them and through them to the citizenry at large,” they nevertheless “cling to a strong sense of their own independence . . . they feel strongly a need for aloofness and authority to do their job” (Moore, 1998, p. 335). This differs considerably from police utilizing a community-oriented approach, where the demands of the citizenry constitute an explicit source of legitimacy. Although law remains the primary source of legitimacy, police also turn to the community for authorization (Kelling & Moore, 1988). Community policing is “based on local norms and values and individual needs” (Cordner, 1998, p. 48; see also Wilson & Kelling, 1982; but see Manning, 1984 for criticisms), and thus a “form of direct democracy of the community will” (Mastrofski, 1991, p. 61). Function of the Police The function of the police may be understood as the “professional domain” they have claimed as their own (Manning, 1997), and in this way constitutes an additional source of legitimacy (Mastrofski, 1991). For police in the reform era, the function of the police closely followed the use of the criminal law as the basic source of police legitimacy. The police function was that of “crime responsible professional work” (Manning, 1997, p. 92), using . . the criminal law to apprehend and deter offenders” (Kelling & Moore, 1988, p. 5). Community policing, on the other hand, is based in part on the tenet that police of the reform era too narrowly defined their role or mission. Consequently, police claim a far broader professional domain: the production and facilitation of order. While crime control remains important, police are also expected to engage in many activities usually shunned by traditional patrol officers, such as order maintenance, conflict resolution, crime prevention, fear reduction, problem-solving, and the provision of services (Goldstein, 1990; Kelling & Coles, 1997; Wilson & Kelling, 1982). Battling incivilities and disorder are now considered as important--if not more important--than the traditional crime fighting mandate of the police. Organizational Design The organizational design of reform-era police departments is the explicit manifestation of many of the assumptions of the scientific management theory of administration (e.g., Taylor, 1911). This theory makes several assumptions about man, the motivation of workers, the work performed, and the environment. Workers inherently dislike work and will try to avoid it; they prefer to be directed, wish to avoid responsibility, have relatively little ambition, and are motivated solely by economic incentives. Work is simple and may be broken down into routine tasks. In addition, the organization exists in a stable environment, and the organization is insulated from its influences (Taylor, 1911). These assumptions are evident in the organizational structure of reform-era police departments. For instance, assumptions about the “lazy” nature of workers are reflected in the unified command authority, strict hierarchies, and narrow spans of control characteristic of most police organizations; these structures ensure the effective monitoring, discipline, and control of subordinates (Moore & Stephens, 1991). Specialization and the division of labor expose assumptions about the motivation of workers and the nature of the work performed; “if tasks can be broken into components, 10 workers can become highly skilled in particular components and thus more efficient in carrying out their tasks” (Kelling & Moore, 1988, p. 5) While the paramilitaristic structure of police organizations serves both control and functional purposes, the reality of police work often stands in stark contrast to the assumptions associated with a scientific approach to organizations. For one, police departments are not “closed systems,” but rather are significantly influenced by external forces (Kuykendall & Roberg, 1982). While reform-era police actively encouraged a “hierarchical, distant and authoritative” relationship with public (Manning, 1997), police administrators adopting a community-oriented approach hope to forge “equalitarian, intimate, and communal” relations with the citizenry (Manning, 1984, p. 222). To this end, scholars advocate two primary structural changes in a time of community policing: (1) decentralization and (2) the fixed, geographic assignment of officers (e.g., Goldstein, 1990). Decentralization is especially instrumental in modeling the formal organization of a department with the everyday realities of police work (Skogan, 1998). In addition, these changes are critical in increasing the responsiveness of officers and enabling them to be closer to the consumer, gaining the input necessary to define police priorities (e.g., Greene, 1989). Tactics and Technology The tactics and technology used by police departments operating under a traditional model reflect a concern with the mandate of controlling crime and preventing criminal victimization: random and preventive patrol, rapid response, and retrospective criminal investigation. (Alpert & Moore, 1997; Moore & Stephens, 1991). The technologies primarily used by police in a traditional paradigm of policing—the patrol 11 car, the two-way radio, the telephone, and weaponry—enabled the police to carry out (or at least present their work in terms of) their crime-fighting mandate. Under community policing, the reactive tactics associated with a traditional model of policing become only a small part of a larger box of tools that may be used to achieve and facilitate order. In addition to regular patrol and rapid response to emergency calls for service, community policing officers engage in foot patrol, community organizing and consultation, community meetings, information gathering, and “walk-and ride” or “stop, walk, and talk” programs (Kelling & Moore, 1988). These various tactics may be classified according the extent to which they facilitate what, according to Cordner (1998), are the three most important tactical elements of community policing: (1) positive interaction with citizens, (2) partnerships, and (3) problem-solving. Partnerships, often referred to as the cornerstone of community policing efforts, are based on the belief that “together, [the] police and public are more effective and more humane coproducers of safety and public order than are the police alone” (Skolnick & Bayley, 1998, p. 1). Strategies that encourage more positive interactions with citizens may lead to productive partnerships in the community, which in turn enable police to reduce crime and disorder in new and more effective ways. For example, the use of civil actions, city ordinances, and zoning regulations, which require the support and cooperation of other community agencies, constitute new and more effective ways of addressing disorder (Goldstein, 1996). Problem-solving--the third of Cordner’s (1998) tactical elements--reflects a shift in emphasis from the “incident” to the “problem” as the key unit of work (Alpert & Moore, 1997). Community policing officers are encouraged to look beyond the 12 immediate situation at hand (i.e., the incident) in an attempt to discover the underlying causes of community problems (Goldstein, 1990). Problem-solving involves the collection of data from many sources, including those inside as well as outside the department (e.g., individuals, public agencies, and private organizations) (Eck & Spelman, 1987). Technological advances enhance the ability of police to carry out their new role. Perhaps the most significant advance has been the introduction of computers to police departments. While the knowledge accumulated by a beat officer was once an individual property, it is now possible to collect, collate, analyze, store, and distribute this information throughout the department. In addition, specific software programs provide the police with unique opportunities to address crime and disorder. For example, crime- mapping software allows police to detect trends and patterns that assist in identifying “hot spots” upon which to focus their attention. Outcome Measures As with each of the criteria discussed thus far, the outcomes measured under traditional and community-oriented models of policing are closely tied to roles police are expected to perform. The most common measures of police officer performance under a traditional policing paradigm include the number of arrests made or citations issued, response times, clearance rates and the like. These measures speak to the effectiveness and efficiency of police officers with respect to their primary mission of fighting crime (Oettemeier & Wycoff, 1998). On the other hand, performance measures used by police departments practicing community policing reflect a broadened and enhanced police role. In addition to crime 13 control, performance measures include fear of crime, problem reduction (Kelling & Moore, 1988), measures of trust and confidence in the police, programs and resources allocated to strengthening police-community relationships (Alpert & Moore, 1997), having a sense of responsibility for an area and its people, collaborating with citizens to address crime and disorder (Oettemeier & Wycoff, 1998). As these examples illustrate, one unique challenge associated with performance measures in a time of community policing is “ . . . finding ways to express quality as a quantity, in other words, to make quality a countable commodity” (Trojanowicz & Bucqueroux ,1992). The question is not just whether officers are engaging in the prescribed behaviors, but whether they are exhibiting quality in their work. Supervision and Community Policing As the previous discussion explicates, community policing represents significant changes in police work. These changes have consequences for all in the police organization, but “inevitably, the changes in structure and style will affect line supervisors—sergeants—the most” (Spelman & Eck, 1996, p. 465). The shift to community policing requires “a major reorientation of police middle management,” specifically with regard to the roles, values, and management techniques of police supervisors (Goldstein, 1987, p. 13). It is according to these three criteria—roles, values, and management techniques—that I compare the nature of police supervision under traditional and community-oriented models of policing in the following paragraphs. The Role of Police Supervisors The roles of police supervisors in traditional and community-oriented models of policing are intimately tied to the roles subordinate officers are expected to play, as well 14 as the assumptions made about workers in each model. In a word, the role of the traditional police supervisor is “controller.” Supervisors are expected to monitor subordinate performance to control misbehavior and prevent abuses (Goldstein, 1977). Recall that the scientific management theory assumes that workers are inherently lazy and dislike work. Consequently, the primary responsibility of managers is to coerce, control, direct, and threaten workers so they will put forth effort toward organizational goals (Taylor, 1911)—in this case, crime control. A heavy reliance on sanctions for failure to comply with rules and regulations has led some to characterize supervision in such systems as negative (Weisburd, McElroy, & Hardyman, 1988). In a traditional model of policing, the role and expectations of police officers are highly specified by “voluminous, detailed rules” (Goldstein, 1993, p. 5). Police officers are expected to fight crime in accordance with the rules set forth by the administration; supervisors are expected to ensUre that officers carry out their mission efficiently and effectively. Consequently, supervisors are “directors who oversee workers who perform specific activities laid out in advance by management” (Kelling, Wasserman, & Williams, 1988, p. 4). Just as supervision under a traditional model of policing is closely associated with the roles performed by police officers, supervision in a time of community policing is likewise tied to the activities of subordinate officers—supervision involves a “different, more responsive attitude and managerial style that will stimulate, accommodate, and perpetuate desired behavioral changes that will occur as a result of redefining the officer’s role” (Oettmeier & Brown, 1991, p. 130). Police officers are no longer mere crime- fighters; they are alternately problem-solvers, community organizers, coordinators, 15 planners, and mediators. Furthermore, community policing is based on far different assumptions regarding the nature of workers than those associated with the professional era of policing. As Kelling & Moore (1988) stated, “the idea that workers have no legitimate, substantive interest in their work is untenable” (p. 11). Instead, officers may be enriched and motivated by jobs that provide them with autonomy, responsibility, and recognition. As a result of the changes in the roles of and assumptions about subordinate officers, police supervisors are expected to shed the role of “controller” in favor of a variety of other roles—-facilitator, coach, teacher, trainer, mentor, and guide (Geller & Swanger, 1995; Glensor & Peak, 2000; Goldstein, 1977; Kelling et al., 1988).5 It is through the utilization of such roles that managers may become “an important part of the support system for field operations” (Wycoff & Skogan, 1994, p.371). As Oettmeier and Brown (1991) described: a major portion of their role should be designed to support the officers’ attempts to identify citizen concerns, assist in mobilizing appropriate resources (or removing the impediments) to address those concerns, and assess the effects of the assistance provided (p. 131). Comparisons of supervision under traditional and community-oriented models have led some to suggest that supervision is greatly simplified under a traditional model of policing. As Goldstein (1990) described: The more routinized the work, the easier it is for the sergeant to check. The more restrictions placed on an officer . . . the easier it is to recognize situations that suggest wrongdoing. The more emphasis placed on rank and the symbols of the position, the easier it is for sergeants to rely on authority. . . to carry out their duties (p. 157). 5 If and when supervisors are expected to act in a “controlling” fashion, it is expected that they will accomplish this by a new system of values, supported by the organization's culture (Moore & Stephens, 1991). 16 Community policing requires that supervisors oversee the varied, innovative, and unpredictable problem-solving efforts of subordinates (Goldstein, 1990; Kelling & Bratton, 1993; Weisburd et al., 1988). This may be particularly difficult for supervisors socialized under the reform model of policing, as one of their “basic functions . . . has been to forestall creativity and innovation”(Kelling & Bratton, 1993, p. 9; emphasis added). Fostering creativity and innovation requires that supervisors relinquish the power and control so critical to their identity under a traditional model of policing (Van Maanen, 1983); community policing officers must be given greater autonomy and discretion in order to accomplish the goals associated with a new way of policing. Supervision may also become a more complex task in a time of community policing because the expectations of the administration are no longer clearly delineated. The expectation of control is relatively straightforward; on the other hand, what is means to be a facilitator, mentor, or coach is far more ambiguous. For example, at one time facilitation may mean marshaling the resources needed by subordinate officers; at another, it may simply mean brainstorming with officers to develop innovative responses to problems identified by community residents. Community policing involves slippery, nebulous terms and concepts whose meanings may change over time--a fact very disconcerting to supervisors used to doing things by the book. Finally, the responsibilities of supervisors under a community policing paradigm may greatly exceed those sergeants have in traditional models of policing (Kelling & Moore, 1988). As Spelman and Eck (1996) describe: Problem-solving puts a dual burden on supervisors. On the one hand, they must make many of the tough, operational decisions: setting priorities among different problems, facilitating communication and cooperation 17 with other divisions of the police department and outside agencies, and making sure their officers solve the problems they are assigned. On the other hand, sergeants must also provide leadership, encouraging creative analysis and response (pp.465-66). A chronic complaint of police supervisors is that they are consistently called upon to do more with less; in this respect, community policing certainly places increased demands on supervisors. Values of Police Supervisors The difference between supervision under professional and community-oriented models of policing may be conceived of as the distinction between what Bittner (1983) termed problems of “legality” and “workmanship.” The problem of legality—— characteristic of supervision under a traditional model of policing—places great emphasis on compliance with explicit rules. As long as officers “stay out of trouble,” supervisors are rather unconcerned with how their subordinates go about the job of crime-fighting. As Holdaway (1995) described, “supervision [is] about what an officer should not do or be caught doing, rather than what should be done competently” (p.73). In contrast, the problem of workmanship involves “the maintenance of minimally acceptable levels of knowledgeable, skilled, and judicious performance,” and more accurately describes supervision in a community policing atmosphere (Bittner, 1983, pp. 2-3). As Bayley (1991) described, “officers are judged according to their ability to achieve general objectives rather than simply avoiding the violation of rules” (p. 234). Supervisors are expected to nurture and promote the abilities of subordinate officers, and assist them in applying their knowledge and skills to the production of order. 18 Management Techniques Employed by Police Supervisors According to Moore and Stephens (1991), the role of traditional police managers is to find efficient means of achieving organizational goals. They do this through the traditional managerial functions of planning, organizing, coordinating, and controlling (Moore & Stephens, 1991). While first-line supervisors are primarily responsible for the managerial function of controlling, they also engage in planning, organizing, and coordinating. For instance, supervisors engage in a planning function by translating organizational goals into specific operational objectives for their subordinates. They also engage in organizing and coordinating by assigning personnel to different beats, scheduling shifts, and distributing resources. Community policing requires a new set of management techniques. While traditional police supervision is hierarchical and authoritarian, supervision in a time of community policing becomes more systematic than bureaucratic (Weisburd et al., 1988); middle managers must strongly support democratic ideals and management practices (Riechers & Roberg, 1990). Police supervisors may do this by adopting a variety of concepts and techniques from the private sector, such as participatory management, team- based work, management by objective, and quality circles, all of which are based on the assumption that empowering workers is not only the key to increasing commitment to organizational goals, but to providing officers with more enriching and satisfying jobs. As this chapter has illustrated, policing undergoes significant changes as a result of the shift to a community-oriented model of policing. The nature of police supervision in particular is substantially altered; police supervisors must adopt new roles, values, and management techniques in order to transition to a new way of doing their job. But are 19 supervisors ready for such change? Do they embrace these changes, or do they resent and resist them? Are there particular factors that make supervisors more or less likely to embrace or resist these changes? In order to provide some insight into these questions, I turn to a consideration of the literature that has examined the reactions of police personnel to community policing. 20 CHAPTER TWO: THE REACTIONS OF POLICE SUPERVISORS TO POLICE REFORM : SUPPORT OR SABOTAGE? The amount of research devoted to the examination of the impact community- oriented programs have on police personnel has greatly increased in the last ten years. Historically, “the focus of most evaluations has been on community impact (e.g., changes in citizens’ perceptions and fears, crime rates, levels of disorder), and not on ofiicer impact;” however, “as police organizations begin to realize that changes in police officers may be the first step toward the success or failure of community policing, such data have taken on new meaning” (Lurigio & Rosenbaum, 1994, p. 161; emphasis in original). A few issues merit comment before proceeding to an examination of the literature. First, while the amount of scholarly attention devoted to the reactions of police personnel to community policing has increased in recent years, studies that examine the experiences and reactions of middle managers to community policing remain rare (Geller & Swanger, 1995). Consequently, the studies included in this review examine the reactions of police personnel generally rather than police supervisors exclusively. However, since middle-management is thought to identify more closely with the rank- and-file than “the brass” (Van Maanen, 1983; 1984; Reuss—Ianni, 1983), this literature may very well constitute the best available evidence regarding the reactions of supervisors to community policing. Additionally, I did not restrict this literature review to the reactions of police strictly to community policing. Of the research that focuses exclusively on first-line supervisors, the majority of studies were conducted during the 21 team policing experiments of the 19703. As such, this literature review may be best thought of as a review of the reactions of all police personnel to police reform generally. Second, the studies I review are both quantitative and qualitative in nature, as both are important to a complete understanding of the reactions of police personnel to community policing. They are examined separately, however, as the different types of studies have used different Operationalizations of support for community policing and thus have produced different pictures of the reactions of police to reform movements. Quantitative Studies Many researchers have examined the reactions of police officers and supervisors to police reform using surveys as the primary form of data collection. One consequence of this data collection format is that support for community policing (or other reforms) is operationalized in attitudinal terms.6 Some studies measure attitudinal support for community policing explicitly by examining the attitudes of police personnel toward community policing generally or the specific programs of which they are a part. With other measures, such as increased job satisfaction, improved attitudes toward the public, or increased feelings of competence in performing community policing tasks, we may reasonably infer support (or a lack thereof) for community policing. Boydstun and Sherry (1975) were the first researchers to systematically explore the effects of innovative police practices on police personnel. Their research centered on San Diego’s Community Profile Development program, a predecessor of problem- solving in the department. This program was designed to provide a greater awareness 6 This is a necessity, as surveys are not conducive to the measurement of actual behavior. While policing researchers have occasionally used surveys to obtain self-reports of officer behavior, they have not (to this author’s knowledge) done so with regard to community policing activities. 22 and understanding of the communities served by officers and improving the response to area problems by developing new, proactive patrol strategies (Lurigio & Rosenbaum, 1994). Boydstun and Sherry (1975) used an experimental design to compare the experiences of a group of officers and sergeants randomly assigned to using the Community Profile Approach to a group of officers and sergeants working in the same beats at the same time but whom employed a more traditional approach to patrol. Surveys were administered to both groups of officers at the beginning, midpoint (6 months), and end of the project (12 months). Survey questions included perceptions of the police role, knowledge of patrol beats, and perceptions of the community. At the outset, both groups of officers were highly comparable. Over time, however, the responses of experimental and control officers diverged, demonstrating the positive impact of the program on experimental officers. For instance, experimental officers demonstrated an expanded understanding of the police role, increased knowledge of their beats, and attached a higher value to community. Moreover, although neither group demonstrated significant changes in job satisfaction ratings by the end of the project, experimental officers were more likely to report that they perceived their jobs as interesting and less likely to report that their jobs were frustrating (Boydstun & Sherry, 1975). Studies in the late 19703 and early 19803 centered on the impact of a variety of team policing programs on police personnel. The first study in this respect evaluated the effects of Cincinnati’s Community Sector Team Policing Program (COMSEC). COMSEC involved the permanent assignment of officers to an experimental 23 neighborhood (i.e., District 1). Each team of officers represented a self-sufficient and decentralized unit responsible for the delivery of all police services. A generalist role for police officers was encouraged; police officers were expected to perform both patrol and investigative functions (Schwartz & Clarren, 1977). The Police Foundation collected data from officers and middle managers (e.g., sergeants and lieutenants) at several points throughout the duration of the project by means of self-report surveys. The surveys included questions not only on job satisfaction, but on attitudes toward the community and their work as well. The survey responses of COMSEC officers and middle managers were compared to an equal number of officers and middle managers working outside the experimental district. During the initial stages of the evaluation, COMSEC officers reported several favorable results compared to their non-COMSEC counterparts. COMSEC officers reported experiencing a greater scope in job responsibilities and tasks, greater feelings of independence, and an increase in decision-making authority. However, there were few differences between COMSEC officers and traditional officers in terms of their relationship to the community. Moreoever, most positive changes that were noted in the early stages of the evaluation study disappeared by the end of the project. In the end, District I appeared little different than other districts in the city. Although specific differences in outcome variables by rank were not noted, Schwartz and Clarren (1977) reported that management played an important role in sabotaging the program. In particular, middle managers were unwilling to relinquish the control necessary to make COMSEC a success. Foot patrol was the next reform to sweep police agencies. In the early 19803, Flint, Michigan, undertook a large-scale study on the effects of their department’s 24 Neighborhood Foot Patrol Program. The program was conducted in 14 areas of the city. According to Trojanowicz (1983), foot patrol officers were expected to facilitate community organization; they were expected to empower community members to support and protect themselves. In 1985, Trojanowicz and Banas interviewed 64 patrol officers and 50 randomly selected motorized patrol officers in a posttest-only control group design. The groups were compared along several dimensions, including job satisfaction, perceptions of prospects for advancement, and attitudes toward work. The groups differed significantly in several ways. Foot patrol officers were more likely to feel that they were performing an important job, improving relations between the police and public, and doing a job the department viewed as important. They were also more enthusiastic about their jobs and more likely to believe that their work was increasing their chances of advancement in the police organization. Hayeslip and Cordner (1987) examined the effect of community-oriented patrol on officers in Baltimore County, Maryland. This project, referred to as the Citizen Oriented Police Enforcement project (COPE), emphasized close community contact and a reduction in fear (Cordner, 1986). This was achieved primarily through “stop, walk, and talk” and neighborhood canvassing tactics. The COPE project evolved in three stages. In the first stage, COPE looked little different than saturation patrol. The second stage of the project emphasized community contact and crime prevention. The final stage involved the implementation of problem-oriented policing. Surveys were administered to COPE and non-COPE officers at four different points in time. Baseline comparisons noted significant differences with respect to job satisfaction. More specifically, COPE officers were more satisfied at the outset, and 25 remained that way throughout the project (although satisfaction diminished somewhat over time). COPE officers also had more positive views toward the program at the outset and these attitudes remained constant through the study period. In comparison, control officers had moderate views toward the program initially, but these views declined over time. Hayeslip and Cordner (1987) conducted additional analyses to account for the possibility that officers’ characteristics (e.g., race, age, sex) were confounding the observed changes. The lack of significant findings in these analyses led the authors to conclude that “COPE participation caused the differences between COPE and control officer attitudes” (p. 115). Several studies focused on the impact of the Community Oriented Police Education program in Philadelphia, Pennsylvania (Greene, 1989; Greene & Decker, 1989). This police—community relations program was designed to encourage crime prevention activities, improved communication between police and citizens, and officers’ attachment to the communities they served (Greene, 1989). Greene and Decker’s (1989) study was specifically directed at determining the effects of COPE program involvement on officers. Data were collected via pre- and post-program surveys. Of the survey items that addressed job satisfaction, COPE officers demonstrated negative program effects: COPE officers were less satisfied with their prospects for job advancement, and were less likely to view their jobs as challenging or affording opportunities for self-directed work (Greene & Decker, 1989). McElroy, Cosgrove, and Sadd (1993) evaluated the impact of participating in New York’s Community Patrol Officer Program (CPOP). These researchers interviewed 51 program officers at two different points in time over a 6-month data collection period; 26 at the beginning of this study, all officers had been assigned to the program for an average of one year. The study assessed changes in T1 and T2 with respect to attitudes toward being a police officer, the CPOP program, the community, and the police department. Despite the positive hypothesized effects of CPOP participation, McElroy, et al. (1993) reported no changes in any of these areas. Based on the premise that participatory management may produce a shift in attitude from traditional to community policing (Wycoff & Skogan, 1994), two studies in the early 19903 examined the effect of a participatory management style on police personnel. In the first, Witte, Travis, and Langworthy (1990) administered surveys to over 150 officers in 14 police agencies located in southwestern Ohio. Witte et al. (1990) sought the answers to three research questions: (1) to what extent do police personnel support participatory management?, (2) to what extent is participatory management practiced?, and (3) do perceptions vary by rank?. While results suggested that respondents generally agreed that participatory management had tangible benefits for police organizations, few police personnel believed that their departments actually practiced participatory management. Morever, analyses failed to demonstrate differences in attitudes by rank. In a more in-depth study of this management technique, Wycoff and Skogan (1994) studied the effects of quality management on police personnel in Madison, Wisconsin. The quality management philosophy, “which called for decisions based on data and the input of employees in the decision-making process,” was implemented as a precursor to community policing in the department (p. 372). In 1989, the department formed the Experimental Police District (EPD) as a laboratory for police innovations. 27 Innovations had to demonstrate success at this location before being implemented on a department-wide basis. Although officers throughout the department received training on quality policing, it was in the EPD that this approach was given the greatest emphasis and support. Data were collected at three points in time across a 3-year period. Survey items inquired about officers’ job satisfaction, perceptions of working conditions, and attitudes toward community policing or problem—solving. The responses of EPD officers were compared to those of non-EPD officers. Although there were no significant differences between EPD and non-EPD officers with respect to attitudes toward community policing or problem-solving, there were several positive changes with regard to job satisfaction and perceptions of working conditions. In particular, EPD officers experienced significant increases in satisfaction with the kind of work performed, the organization, supervision, job growth potential, and strength of task identity (Wycoff & Skogan, 1994). Lurigio and Skogan (1994; 1998) studied the effects of community policing on officers’ attitudes and perceptions in Chicago, Illinois. In 1993, the Chicago Police Department implemented the Chicago Alternative Policing Strategy (CAPS) in 5 prototype districts. Lurigio and Skogan’s (1994) first study of CAPS was conducted in 1993, prior to the implementation of CAPS. Surveys were disseminated to line personnel at the conclusion of orientation training. These surveys solicited information on job satisfaction, participatory management, orientation to community policing, and attitudes toward CAPS. In analyses comparing the responses of first-line supervisors to line personnel, Lurigio and Skogan (1994) found that attitudes toward community policing differed significantly by rank. Higher ranking officers exhibited more support for 28 community policing, as evidenced by greater agreement with statements such as “police officers should work with citizens to try and solve problems on their beat” (Lurigio & Skogan, 2000, p. 253). In a later study, Lurigio and Skogan (1998) distributed surveys to two groups of officers: 1,169 CAPS officers and 335 patrol officers assigned to other districts throughout the city. Data were collected at two points in time--before the start of the program and 2 years later when the program was implemented on a city-wide basis. Lurigio and Skogan’s (1998) research demonstrated that the program had significant and positive effects on participants; CAPS officers had more positive attitudes toward the program, had more favorable attitudes toward their dealings with citizens, and were more satisfied with their jobs on a number of dimensions. The research of Piquero et al. (1998) reports the effects of the Philadelphia Housing Authority Police Department’s (PHAPD) community policing/problem-solving program, located in the 11th Street Corridor in north Philadelphia. The objectives of this program included promoting better relationships between police and citizens, increased crime reduction and prevention activities, reduced victimization, and increased community cohesion (p. 101). Administrators anticipated that the program would also have positive benefits for police personnel. To evaluate this aspect of the program, the surveys were administered at two points to two groups of officers: officers participating in the experimental districts and control officers working elsewhere. The survey items asked questions regarding the nature of daily police work, job satisfaction, perceptions of the community, community problems, and role expectations. Only one factor—the 29 nature of daily police work—emerged as significant in the second wave analyses. More specifically, control officers were significantly more active than experimental officers. Lewis, Rosenberg, and Sigler (1999) compared the acceptance of community policing among police officers and command staff (i.e., those at the rank of sergeant and above) in the Racine, Wisconsin, police department. Although community policing was practiced solely by officers and management staff assigned to a specialized sub-unit, data in this study were collected from all sworn personnel (N=167). The dependent variable, attitude toward community policing, was operationalized as support for community policing and problem-solving activities. The researchers reported that command staff had the most favorable attitudes toward community policing as indicated by mean values; however, the researchers did not employ tests of statistical significance so it is not possible to conclude with confidence that managers in this study were any more or less likely to support community policing than officers of lower ranks. On the whole, most quantitative studies appear to demonstrate positive effects of community policing on police personnel; they tend to be more satisfied, feel more empowered, perceive their job as more interesting, more rewarding, and less frustrating (Lurigio & Rosenbaum, 1994; Weisel & Eck, 2000; Wycoff, 1988; Wycoff & Skogan, 1994). Of the ten evaluation studies reviewed here, six reported positive effects or changes owing, at least in part, to participation in a community-oriented program. Three studies reported no changes; only one study indicated negative results (i.e., Greene & Decker, 1989). While quantitative studies appear to support Weisel and Eck’s (2000) contention that a solid core of line officers feel favorably about community policing, qualitative 30 accounts paint a much different picture of the reactions of officers to community-oriented programs. Although fewer in number, qualitative accounts provide what is sorely lacking in most quantitative studies of reactions to community policing: explicit attention to middle managers. As such, I provide a review of these studies next. Qualitative Accounts Qualitative accounts of support for community policing or other reforms are based on in-depth case studies of police departments and the community-oriented programs of which they are a part. Unlike quantitative studies, in which the method of data collection restricts the measurement of support for community policing to an attitudinal phenomenon, qualitative studies use several data collection techniques, and thus are able to account for a variety of reactions—both attitudinal and behavioral. The earliest descriptions of the responses of police supervisors to community policing are found in anecdotal reports of the team policing experiments in the 19703. By all accounts, middle managers played a powerful role in contributing to the demise of team policing in cities throughout the country (Sherman, 1973; Sherman, 1975; Sherman, Milton, & Kelly, 1973). According to Sherman (1975), middle managers committed three “sins” in their subversion of team policing: (1) sins of omission, (2) sins of commission, and (3) active disagreement openly expressed to their superiors (p. 364). Sins of omission involved the failure of police supervisors to acknowledge and deal with the conflicts and problems that arose as team policing was adopted. For instance, rather than dealing with the turf battles and power struggles that erupted during the initial phases of team policing, police supervisors often chose to ignore them. More explicit were sins of commission. As Sherman (1975) describes, “the most effective torpedo was 31 simply bad-mouthing: sending the word out through the grapevine that the team idea was no good” (p.368). Lastly, supervisors involved in the team policing experiments often openly expressed their opposition to and hostility toward team policing, although this primarily happened only when the bone of contention involved policy matters the police supervisors could not circumvent without attracting the attention of superior officers (Sherman, 1975). In perhaps one of the best known field studies of community policing, Sadd and Grinc (1994), working under the auspices of the Vera Institute of Justice, conducted an evaluation of Innovative Neighborhood Oriented Policing (INOP) in eight sites across the country: Tempe, AZ; Louisville, KY; Hayward, CA; Houston, TX; New York, NY; Norfolk, VA; Portland, OR, and Prince George’s County, MD. In some locations, INOP represented a “first attempt” at community policing; in others, INOP was a small component of previously established, citywide community policing efforts. The focus of INOP also varied considerably by site. While in most sites INOP represented attempts to 9 apply community policing tactics to “drug demand reduction,’ in other communities, INOP was synonymous with the provision of community-based services and information (p. 27). Despite differences in the operationalization and implementation of INOP across sites, the reactions of police personnel to these initiatives were remarkably uniform: “the level of enthusiasm for community policing among patrol officers was weak, at best” (Sadd & Grinc, 1994, p. 35). The reaction of police supervisors was even more misanthropic. Supervisors, who felt that community policing was being forced upon 32 them with little or no input to the process, reacted to efforts with “considerable” and “substantial” resistance, effectively sabotaging INOP efforts in several cities. Wilkinson and Rosenbaum (1994) conducted an evaluation of Neighborhood- Oriented Policing (NOP) in Aurora and Joliet, Illinois. In both locations, the goals of NOP included improving the quality of life, the reduction of fear and criminal activity, and increased public confidence in the police. In addition, both departments sought to apply community policing tactics to gang and drug activity. The reactions of police personnel across sites were markedly different. In Joliet, while officers appeared supportive of NOP, “the supervisors’ lack of interest in community policing was apparent” (p. 119). In fact, management resistance was so substantial that supervisors from a bureau outside patrol (i.e., Research and Planning) were brought in to see the program through its first year. In Aurora, nearly the opposite situation occurred; while officers were opposed to NOP, supervisors demonstrated support. In this case, Wilkinson and Rosenbaum (1994) reported that one supervisor was able to salvage the program by establishing an informal culture conducive to change. Greene, Bergman, and McLaughlin (1994) reported on the reactions of police personnel to community policing in Philadelphia, Pennsylvania. Community policing in this city was aimed at the accomplishment of three objectives: (1) an enhanced police- citizen partnership, (2) a managerial culture focusing on group—decision making, and (3) a rebuilding of the internal “thinking system” of the department to sustain long-term adaptation and growth (Greene et al., 1994, p. 96). Due to the integral role police managers were to play in the attainment of these goals, much of the training was directed at personnel holding supervisory positions. In addition, managers’ ideas and opinions 33 were solicited throughout the implementation period. The end result was a “core element of police managers to spearhead change” (p. 106).7 Perhaps most remarkable is that this base of support was established despite less—than-enthusiastic attitudes on the part of line personnel. Summary/Conclusion As the preceding discussion illustrates, there is widely divergent evidence regarding officers’ attitudes toward community policing. Relying on quantitative or qualitative studies to the exclusion of the other produces a very different picture of the reactions of police personnel to community policing. In general, quantitative studies demonstrate a substantial amount of support for community policing among police personnel; qualitative accounts lead us to the opposite conclusion. This appraisal changes, however, when one exclusively examines the results of findings as they pertain to police supervisors. Three quantitative studies examined the reactions of police supervisors (Lewis et al., 1999; Lurigio & Skogan, 1994; Witte et al., 1990). Of these studies, only one (Lurigio & Skogan, 1994) demonstrated that supervisors were more likely than patrol 7 The reactions of supervisors provided in the above accounts closely resemble a classification scheme offered by Goldstein (1990). Goldstein predicted that personnel might be classified as supporters. pacifists and resisters or saboteurs. depending on the extent to which they supported community policing. Supporters and pacifists would vary in the degree to which they committed (attitudinally) to community policing, but both would engage in the behaviors sought by the administration. On the other hand, resisters or saboteurs would be vociferous in their opposition to community policing and refuse act according to the wishes of police administrators. Although Goldstein's (1990) typology was more or less a prediction, a supervisor located in Weisel and Eck's (2000) research described the attitudinal and behavioral reactions of police personnel to community policing in a remarkably similar manner, although he used flavors to demarcate different responses. Chocolate personnel “need to be sold upon the merits of organizational change” (p. 270); they need to understand the logic of change before committing to it. Once convinced, however, they are staunch supporters of the concept. Vanilla personnel are compliant; they may not agree with community policing, but they are willing to follow direction and carry out the tasks sought by police administrators. Finally, strawberry personnel pose significant problems for administrators. These persons are vehemently opposed to the proposed change, outspoken in their beliefs, and no amount of convincing will change their minds. 34 officers to support community policing. Of the four qualitative accounts reviewed here (Greene et al., 1994; Sadd & Grinc, 1994; Sherman, 1975; Wilkinson & Rosenbaum, 1994), the majority of studies reported negative reactions to community policing on the part of police supervisors (Sadd & Grinc, 1994; Sherman, 1975; Wilkinson & Rosenbaum, 1994). What is to be made of these disparate findings? Which picture of supervisors’ reactions to community policing is more accurate? While it may be that neither type of study yields a completely accurate depiction of supervisors’ reactions to community policing, qualitative accounts are likely closer to the “truth.” Quantitative studies, which by necessity rely on attitudinal indicators of support for community policing, implicitly assume that positive attitudinal reactions to community policing will translate into the behavioral practice of community policing. A significant amount of literature on the matter has found only weak to moderate support for a direct attitude-behavior link (Riksheim & Chermak, 1993; Sherman, 1980), thus this assumption is tenuous at best. On the other hand, qualitative accounts provide some behavioral indication of the reactions of police to innovative programs. While these accounts do not provide a criticism-proof measure of behavior,8 they nevertheless acknowledge that support is a phenomenon with both attitudinal and behavioral components. This measurement issue is particularly important concerning community policing. It is not enough to know that a supervisor can or will “talk the talk” of community policing; we must also know if he/she can or will (and under what conditions) “walk the walk” of community policing. In the Qualitative accounts typically do not differentiate between attitudes and behaviors in their accounts of support for or resistance to reform movements. Instead, support is treated as an amalgamation of attitudes and behavior. final analysis, it is the behavior of police supervisors that is most important. This is an issue that will be explored in further depth in the Chapters 4 and 5, when the framework guiding my research and the measurements of my dependent variables are delineated. For now, I turn to a discussion of the conditions or factors that are identified in the policing literature as determinants of support for community policing. 36 CHAPTER THREE: FACTORS INFLUENCING SUPPORT FOR COMMUNITY POLICING: WHAT MATTERS? As Wilkinson and Rosenbaum (1994) noted, “The fundamental question about community policing. . . is not ‘should it be implemented?’—the concept is already extremely popular with policymakers—but rather ‘How should it be implemented?”’ (p. 110). We know from research that “changes . . . having normative sponsorship within the department . . . are the most likely and quickest to occur” (Greene, 1998b, p. 153). In addition, we know that police supervisors play a critical role in the successful implementation of community policing (Alpert & Dunham, 1989; Bayley, 1994; Geller & Swanger, 1995; Goldstein, 1990; Sherman, 1975; Skogan & Hartnett, 1997; Wycoff & Skogan, 1994). What is still necessary is to come to an understanding of the factors that influence whether middle managers will work to support or subvert community policing. In this section, I provide a summarization of the factors previous research has identified as playing a significant role in determining the reactions of police personnel9 to community policing. Training Scholars have noted that many line personnel are reluctant to embrace community policing because they have little understanding of the concept (e.g., Goldstein, 1990; Kelling & Moore, 1988, Sadd & Grinc, 1994). In Sadd and Grinc’s (1994) research, lack of support “could be traced in every instance to inadequate efforts on the part of police 9 Given the scarcity of research that examines the reactions of police supervisors to community policing, this literature review relies on the broader literature that includes police officers’ support of community policing. 37 administrators to communicate the philosophy, goals and tactics of community policing to their officers” (p.36). In other words, lack of support for community policing was the direct result of a lack of training. Williams and Sloan (1990) also acknowledged the importance of training in laying the groundwork for successful implementation. Education and training are clearly important from a practical standpoint. Community policing, with its new roles and functions, requires officers to learn new ways of doing their jobs. In this way, training is an invaluable conduit of the administration’s expectations. While training has practical value, the resources and personnel devoted to it also convey important symbolic messages. For one, training may engender greater trust between lower ranks and the administration, preventing the assumption “that management is imposing (again) something that is not in their best interests” (Rosenbaum & Lurigio, 1998, p. 205). In addition, training also communicates in no uncertain terms that community policing is going to be real rather than a passing fad—a problem noted in some research (e. g., Sadd & Grinc, 1994). Participation The literature suggests that police supervisors may oppose community policing because they perceive it as a top-down directive from the administration. Sadd and Grinc's (1994) research of community policing initiatives in eight cities showed that “common to all sites was the perception that community policing was being ‘shoved down the throats’ of patrol officers without their input into the process” (p. 40). In this way, the “the transition to community policing. . . is exacerbating the historical 38 antagonism between the two ‘cultures’ (i.e., management and line officer) of policing”(Sadd & Grinc, 1994, p. 40). Directives that impact patrol work are always to be viewed with caution and suspicion because members of the brass “don’t really know what happens on the street” (Skogan & Hartnett, 1997, p. 73); however, police officers are particularly antagonistic when they have not been consulted about changes that directly influence their jobs. Organizational theorists postulate that participation greatly increases the likelihood of commitment to organizational policy (e.g., Shanahan, 1978). Research in policing has supported the validity of this assertion; Wilkinson and Rosenbaum’s (1994) study of community policing in Joliet and Aurora, Illinois, demonstrated that participation was a key factor in whether officers would support or resist community policing. Officers in Aurora were not allowed input into the planning process; rather, officers were expected to carry out administrative directives passed down the chain of command. In the best cases, a lack of participation and input led to significant feelings of resentment; in the worst cases, officers withdrew and refused to practice community policing. In contrast, participatory management principles, which encourage officer input in decision-making processes, were followed at the Joliet site. These officers were far more tolerant of change and ultimately implementation was far more successful. In short, the evidence suggests that when employees are “given a voice” in the planning process, they are far more likely to be tolerant of, or receptive to, change; they have a “stake” in the outcome of those policies and procedures (Witte et al., 1990, p.4). 39 Role Conflict Role conflict may be defined as “conflict created by incompatible role expectations” where “. . . complying with one set of role-related pressures hinders or prevents compliance with a different set of role-related pressures” (Gordon, 1996, p. 376). Role conflict has been noted as a problem by many with relation to community policing, but perhaps none so eloquently as Nowicki (1998) in describing the “mass of contradictions” and “mixed messages” community policing poses for officers and managers (p. 265). Role conflict may be engendered in several ways: by failing to provide supervisors with the resources necessary to perform a new role, by demanding accountability for activities associated with a traditional model of policing, and by rewarding behavior concordant with a reactive mode of policing. Each of these unique contributors to role conflict is discussed below, as their effect may have deleterious consequences for the successful implementation of community policing. As Geller and Swanger (1995) asserted, “faced with conflicting role demands (e. g., maintain status quo vs. become an agent of change), middle managers often choose the familiar” (p. 33). Resources In a time of community policing, patrol supervisors must assume responsibility for freeing up as much time as possible for their subordinates to engage in problem- solving. In order to do this, however, other department systems, resources, and policies must be in place (Webster & Connors, 1996).'0 Failure to ensure that police supervisors are allocated the resources necessary to engage in community policing can create cases of ’0 For example, DeJong, Mastrofski, & Parks (2000) found that district staffing levels (an important resource) increased the likelihood that patrol officers would engage in problem—solving. 40 role conflict that would not exist otherwise. This point is illustrated in the following examples. In recounting the experiences of Houston and Birmingham’s community policing efforts, Skogan (1994) referred to the “911 problem;” the conflict between the need to respond to calls for service and the need to be freed from the radio (p. 176). When this war is waged, calls for service will almost always win the battle. Since community policing is viewed as a “discretionary add-on” to the core responsibility of fighting crime and “because it is seen as ‘soft,’ and aimed more at community and public relations than at crime control, it is often delayed and resisted when crime and workloads are on the rise” (Kennedy, 1993, p. 2). This problem is particularly salient for supervisors responsible for units that are expected to engage in both reactive policing and community policing. This was the case in Wilkinson and Rosenbaum's (1994) study of NOP in Aurora and Joliet, Illinois. Officers in Aurora (who remained responsible for responding to calls for service while also instructed to engage in community policing) claimed that their supervisors placed too much emphasis on responding to calls, implying that problem-solving was their secondary responsibility. Emphasis on the Control Function of Supervisors Recounting the reasons for the failure of team policing in the 19703, Sherman (1975) spoke of supervisors’ experiences of role conflict. In particular, he noted that administrators were guilty of supporting “the continued definition of mid-management’s function solely in terms of control, and not in any terms of support” (p.370; emphasis in original). One way of continuing to define the managerial role in terms of control is to limit the power of supervisors to change the way work is performed. For instance, in 41 order to provide the structure necessary for subordinates to engage in community policing, supervisors must be given the power to do new things, such as construct work teams, change an officer’s partner or shift in response to perceived needs, mandate training, and make a variety of procedural changes to make the attainment of goals go more smoothly (Geller & Swanger, 1995). When the administration fails to do this, “senior managers. . . put middle managers in untenable positions by holding them to account for defects in police performance, but declining to delegate available power to correct for those defects” (Geller & Swanger, 1995, p. 28). This quote also points to a related problem: a “zero-tolerance” for mistakes (Geller & Swanger, 1995). “Officers feel that failure will not be tolerated” (Nowicki, 1998, p. 267), and supervisors fear that they will continue to be held responsible for the failures and mistakes of employees. When this is true, supervisors are encouraged to continue their control function. Community policing requires innovation, creativity, and sometimes, mistakes. Yet if mistakes are not tolerated by the administration, then middle managers may feel compelled to continue to operate in a controlling rather than facilitative manner. Rewarding the Status Quo Role conflict may also be aggravated or intensified by evaluation systems that continue to represent the “old” ways of doing things (Glensor & Peak, 2000). Although community policing must be evaluated and rewarded, most departments continue to judge officers by traditional standards (Goldstein, 1987). This is a clear way to send mixed messages. “Where significant mixed messages are sent about whether community 42 policing really matters, the status quo is likely to be king, and the middle managers will be his loyal subjects” (Geller & Swanger, 1995, p. 33). There is an inherent conflict between traditional performance evaluation systems and the new police role expected under community policing.'1 If traditional, crime control behaviors are rewarded in a department practicing community policing, then administrators are guilty of committing the “folly of rewarding A, while hoping for B” (Kerr, 1975). Administrators may desire community policing, but they send the clear message that traditional policing is what will be rewarded. Officers in the LAPD, when questioned about ways to improve the shift to community policing, consistently referred to the importance of new evaluation systems (Greene, 1998b). Evaluation systems must link community policing to the way police organizations evaluate, promote, and reward employees. “In a community policing context, performance evaluations do far more than simply evaluate performance; they serve as important vehicles for. . . conveying organizational expectations and rewarding behavior concordant with [a] broadened police role” (Chandek, 1999, p. 10). Opportunity Very few studies have sought to quantify the amount of time police officers spend engaging in community-oriented activities (cf DeJong, Mastrofski, & Parks, 2000; Frank, Brandi, & Watkins, 1997; Parks et al., 1999). Of the studies that have done this, only the study of DeJong et al. (2000) examined the factors that might increase or decrease the likelihood of this type of behavior. DeJong et al. (2000) examined a variety of organizational factors that were thought to influence whether police officers would 1 . . . . . 1 Although this drscussron centers on formal evaluation systems. the conclusions drawn from the literature are equally applicable to informal rewards. 43 engage in problem-solving. One variable that exerted a significant effect on time spent in community policing was opportunity; officers who worked in areas beset by greater distress spent significantly more time problem-solving per shift than officers responsible for less aggrieved beats. Group Subculture Supervisors are commonly described as the “men in the middle” (Trojanowicz, 1980). This places supervisors in a unique position, caught between line officers and administrators—whom Reuss-Ianni (1983) respectively referred to as the “street” and “management” cultures of policing. Subordinate officers are the “raw material” of police supervisors (Van Maanen, 1983); supervisors are intimately tied to the men and women they supervise. Many of the day-to-day realities of supervision make it more likely that first—line supervisors will identify more closely with the men and women they supervise (Reuss—Ianni, 1983; Van Maanen, 1983; 1984). There may be instances, however, when supervisors will identify more closely or associate with members of “the brass.”’2 This being the case, it is important to consider the implications 3 supervisor’s identification with subordinates or superiors has for whether the supervisor will choose to engage in community policing. Supervisors who more closely identify with the men and women under their command may be more likely to resist or sabotage community policing efforts. Individuals in the subordinate rank of police officer place great emphasis on the “crime- fighting” role of the police (Greene, 1998b). Indeed, studies that have examined police ’2 For example, this might be the case among supervisors who are interested in advancing in the organization (Brown, 1981). officers’ attitudes have demonstrated that officers view themselves primarily with respect to this role (e.g., Worden, 1995). As Moore and Stephens (1991) elucidated: Crime control is the purpose that attracts the greatest enthusiasm and commitment from the police themselves. Many officers join police departments to become members of the ‘thin blue line’ that protects decent people from predatory criminals. . . .As a result, culture forms in the police department that sees crime fighting as not only the most important function, but the only honorable one (p.30). When crime fighting is viewed as the only honorable police function, community policing “gets labeled as ‘social work,’ the job of ‘empty holster guys,’ and not of ‘real police officers’”(Skogan & Hartnett, 1997, p. 12). Officers faced with the choice of engaging in crime control activities (e.g., making arrests, random preventive patrol) and community policing will choose those activities that “have traditionally high status in policing”—those associated with a professional model of policing (Weisburd & McElroy, l991,p.100) Supervisors may be particularly enmeshed in the street culture of policing and thus dissident; most of these men and women were recruited, selected, trained, and socialized under a professional model of policing. As Moore and Stephens (1991) articulated: The traditional police manager. . . is committed to an orthodoxy which includes ‘not making waves; and which excludes innovation. This commitment has served him or her well because they have advanced through the ranks by following the orthodoxy (p. 108). The success of police supervisors under a traditional model of policing thus gives them a strong stake in maintaining the status quo (Roberg, 1994). On the other hand, supervisors may appear to identify more closely with members of the administration in the department. These individuals may actually embrace the 45 values and goals of the administration or merely engage in the prescribed organizational behaviors to win the approval of members of the administration. In either case, we might expect a greater identification with the administration (and thus more support of community policing) among first-line supervisors in specific cases; for example, among more educated supervisors or among those who have an interest in promotion (Carter, Sapp, & Stephens, 1989; Skolnick & Bayley, 1988, Williams & Sloan, 1990). Background Factors The background characteristics of police officers or supervisors tend to have little power as overall explanations of police behavior (Riksheim & Chermak, 1993; Sherman, 1980); however, they are essential to include in any model as controls, as officers’ characteristics may attenuate the effects of other variables. For example, several researchers have speculated that age, tenure, and education are important considerations in the implementation of community policing, with older, more experienced, and less educated officers likely to be the most resistant to change (Carter et al., 1989; Skolnick & Bayley, 1988; Williams & Sloan, 1990; cf Skogan & Hartnett, 1997; Weisel & Eck, 2000). It is most likely that these variables do not wield direct effects on support of community policing, but rather influence their identification with subordinate officers, cynicism regarding past change, and the like, exerting indirect effects on support of community policing. Race and sex are also important control variables, as studies have demonstrated that they may influence levels of support of community policing. In particular, females and minorities tend to be more supportive of community policing than their white, male counterparts (Skogan & Hartnett, 1997). Miller (1999) has postulated that this is the 46 result of being excluded from the dominant (i.e., male, white) culture of policing. Participating in community policing (e.g., by applying for neighborhood policing officer positions) allow these otherwise marginalized groups to gain status, prestige, and a sense of belonging in the police organization. Finally, another key factor may be interest in promotion. Wishing to advance in the organization may compel officers (who may not have otherwise) to become “yes men.” Summary/Conclusion The body of research reviewed in this chapter points to several factors that may increase the likelihood that supervisors will support community policing: training, participation in the planning and implementation process, opportunity, and identification with upper management. This literature also identifies factors that make the acceptance and practice of community policing less likely: role conflict or identification with subordinate officers. A summary of these factors, and their relationships with support for community policing, is presented below. While this literature provides us with some limited direction, it is also difficult to make clear sense of these findings. What do these factors represent? Can they be classified in any meaningful way? Is there a grouping of factors that can be expected to exert the greatest influence on whether supervisors will support or subvert community policing? The current study attempts to provide some insight into these issues by stepping outside the policing literature. Drawing on research in other disciplines, more often guided by well-established frameworks and models, allows for a more logical ordering of the variables described here. Additionally, a theoretical framework will allow me to make hypotheses regarding the relationships among variables. The framework that 47 will be used to guide my inquiry, as well as a description of the data I will use in my research, is provided in the following chapter. Table 1. Summary of Extant Research Relationship to Support for Factor Definition Community Policing Training Understanding of community + policing goals, programs Participation Participation in the planning/ + implementation process Role Conflict Conflict created by incompatible role expectations 0 Resources + o Emphasis on the “Control” __ Function of the Police - Rewarding the Status Quo — Opportunity Opportunity to engage in + community policing Group Subculture Identification with “street” versus “management” cultures 0 Identification with Subordinate Officers Background 0 Race (minority) + Characteristics 0 Age _ 0 Sex (female) ‘1' 0 Tenure -— 0 Promotion Ambition + 48 CHAPTER FOUR: THEORETICAL FRAMEWORK & HYPOTHESES Thus far, I have provided a review of the extant policing literature as it pertains to supervisor support for community policing. The policing literature on this topic is lacking in two main ways. First, most studies have operationalized “support” for community policing as an attitudinal phenomenon. As a result, there are conflicting accounts regarding the reaction of police supervisors to police reform. Second, prior research on supervisors and community policing has been atheoretical in nature. As a result, most research examining the factors that influence support for community policing resembles a fishing expedition rather than a tight, logical, or orderly empirical inquiry. In this chapter, I outline the problems these issues have posed for drawing any conclusions regarding how supervisors react to community policing, and how research in the area of organizational psychology—specifically, research on organizational commitment and organizational psychology—can be used to remedy these problems. Problems with Previous Research on Supervisor Support for Community Policing Prior Operationalizations of Support for Community Policing Most studies of the reactions of police personnel to community policing have relied on attitudinal indicators of support for community policing.13 By operationalizing support for community policing in this manner, researchers have treated attitudes and behavior as essentially synonymous entities. Researchers concluding that a strong base of attitudinal support for community policing exists consequently have suggested that rampant accounts of officer and/or supervisor resistance to community policing are 13 Common indicators of attitudinal support are the levels of agreement with such statements as “law enforcement is the most important responsibility of the police” or “the police and the public should work together to prevent and control crime.” 49 erroneous (e.g., Weisel & Eck, 2000). A direct link between attitudes and behavior is presumed; the implicit assumption is that the attitudinally supportive police officer or supervisor is more likely to engage in community policing behaviors. This is not necessarily true. There can be a wide gap between the attitudes and beliefs one holds and the behavior one exhibits; attitudes and behavior can be strongly correlated, weakly correlated, or not correlated at all (Fishbein, 1966). Describing the behavior of police managers, Kuykendall (1985) differentiated between management philosophy (what one believes) and management application (what one actually does). While the ideal may be when there is congruence between philosophy and application, this is not always possible, nor even feasible. Swanson, Territo, and Taylor (1998), for example, found that even if supervisors support a philosophy like participatory management, they are still more likely to rely on traditional means of managing in practice. Kuykendall and Roberg (1982) argued that police supervisors often feel compelled to revert to traditional management methods with some employees, such as rookies or the complaint-prone, who are not capable or prepared for participation. With respect to the study of supervisor support for community policing, there may be several reasons that a supervisor’s attitudinal support for community policing does not translate into the behavioral practice of community policing. Some supervisors may feel positively about community policing but be unable or unwilling to practice it due to conflicting demands or limited resources; others may engage in community policing, but not be sold upon its merits. In short, attitudinal support for community policing does not necessarily translate into the practice of community policing. While it might be that 50 supervisors who feel positively about community policing are more likely to engage in community policing, this has not yet been established by research. At this juncture, we can only conclude that there may be a baseline of attitudinal support for community policing among police. It remains premature and misleading to conclude that accounts of supervisor resistance to community policing are inaccurate; this conclusion may be reached only after research is conducted that relies on behavioral indicators of support for community policing. Ultimately, the proof of support for community policing will come in the deeds and actions of police, not their attitudes. In order to overcome the problem of past operationalizations of support for community policing, researchers should use behavioral indicators of support for community policing. This could be accomplished, for example, by measuring the amount of time a supervisors spends engaged in community policing. Given that most previous studies presume a link between police officers’ attitudes toward community policing and their actual engagement in it, researchers could also include supervisors’ attitudes toward community policing as a key independent variable of interest. In this way, the study of supervisor support for community policing could be situated in the larger context of research on the link between attitudes and behavior. The Atheoretical Nature of Past Research In addition to using inappropriate operationalizations of support for community policing (or making inappropriate generalizations regarding officers’ reactions to community policing based on attitudinal measures of support), past research is also problematic because previous researchers have tended to adopt a kitchen sink approach to their examination of the factors that influence support for community policing. The 51 result is a laundry list or hodge-podge of variables that may (or may not) have an effect on whether supervisors will work to support or subvert community policing. Without a theoretical framework to guide research in this area, researchers are left with little in the way of a logical or cohesive set of explanators upon which to focus their energies and little in the way of an understanding of the relationships among these variables. In order to remedy this situation, theory is needed. Work conducted in organizational psychology—specifically in the areas of organizational commitment (a job attitude) and organizational citizenship behavior (a type of job performance)—provides a viable framework for understanding the relationship between supervisor attitudes toward community policing (3 job attitude) and the practice of community policing by supervisor (a type of job performance). Situating Supervisor Support for Community Policing in the Larger Context of Research on the Link between Attitudes and Behavior Research in many disciplines concerns the relationship between how individuals feel about a particular thing, object, or issue and how individuals actually act with respect to that thing, object, or issue. In the policing literature, several studies have examined the nexus between police officer attitudes and behaviors (Friedrich, 1977; Worden, 1993; Riksheim & Chermak, 1993; Sherman, 1980; Worden, 1989). On the whole, this research suggests that the link between police attitudes and behavior is questionable at best (for reviews of this literature, see Riksheim & Chermak, 1993; Sherman, 1980). The failure to find a relationship between attitudes and behavior is not the strict domain of police researchers; work in other disciplines also often fails to reveal what are presumed to be logical or intuitive relationships between attitudes and behavior. As one sociologist concluded, “no matter what one’s theoretical orientation may be, he has no reason to 52 expect to find congruence between attitudes and actions and every reason to expect to find discrepancies between them” (Deutscher, 1966, p. 247). Given that attitudes are generally poor predictors of behaviors no matter what the attitude or behavior under examination, what conclusions may be drawn? Is it safe to say that attitudes exert no effect on behaviors, or that the effect is so weak, that research in this area is futile? Should one simply assume that a supervisor’s attitudes toward community policing have no or little effect on his/her actual behavior, and search elsewhere for viable predictors of community policing behavior? Work in the area of attitude theory suggests that attitudes may not be predictive of behaviors in previous studies for at least three reasons: (1) there is a lack of congruence or specificity between the attitude and behavior measures, (2) the attitude measure is poor in quality (i.e., unreliable), or (3) there is a failure to identify external or situational factors that may be influencing the behavior under examination (Fishbein & Ajzen, 1975; Zanna, Olson, & Fazio, 1980). In response, researchers may focus more in—depth attention on: (1) ensuring congruence between the attitude and behavior under examination; (2) attempting to produce a superior attitude construct, definition, or measurement; or (3) conducting research that seeks to identify external factors that may serve to moderate the link between attitudes and measurement (Fishbein & Ajzen, 1974). Researchers who have adopted this third approach tend to believe that attitudes are only one possible set of influences on behavior; the key to understanding the often weak or nonexistent relationship between attitudes and behavior is to identify other factors—typically situational in nature—which are thought to exert a more direct effect on behavior. 53 In studies examining the relationship between police attitudes and behavior, the class of situational variables upon which attention is most often focused includes features of the police-citizen encounter. This is largely because the outcome variables examined in this literature are the consequence of interactions with citizens (e.g., arrest, use of force). Not only is this level of analysis (i.e., the encounter) inappropriate for the study of supervisor support for community policing, but so too are the variables that are explored, such as the number of bystanders, location, or citizen demeanor. Fortunately, this is an area in which theory in the field of organizational psychology provides some direction. Theoretical Framework The organizational psychology literature provides guidance with respect to a relevant class of situational variables that may affect supervisors’ performance on the job (i.e., the extent to which they engage in community policing). Broadly speaking, organizational psychology concerns itself with the behavior of organizational actors. Central issues in the study of organizational behavior include job performance, workers’ commitment to the organization, and the how to elicit the cooperation of employees in the accomplishment of its goals (Walsh & Tseng, 1998). One common set of variables examined in this literature includes characteristics of the work environment that may exert an influence on individuals’ behavior (e.g., performance). Work conducted in organizational psychology in the areas of organizational commitment and organizational citizenship behavior provides a framework for understanding the factors that influence on-the-job performance. 54 Organizational commitment is a term used to represent a set of attitudes that align the worker's effort with the goals of the organization (Hodson, 1991). Most commonly, organizational commitment is conceptualized as an attitudinal constructl4 defined in part as “a strong belief in and acceptance of the organization’s goals and values” and a “willingness to exert considerable effort on behalf of the organization” (Mowday, Porter, & Steers, 1982, p. 27). A plethora of studies have been conducted on the antecedents and consequences of organizational commitment. As a work-related attitude, one of the consequences of organizational commitment that has been examined is job performance. One type of job performance that has received a great deal of research attention in relation to organizational commitment is organizational citizenship behavior. In order to understand organizational citizenship behavior, one must make the distinction between in-role and extra-role performance. In-role performance on the job is that which is traditionally expected of employees and which is rewarded explicitly by the organization. Extra-role performance, on the other hand, is discretionary behavior (MacKenzie, Podsakoff, & Aheame, 1998). Organizational citizenship behavior is an example of discretionary extra-role performance in that it is ". . . not directly or explicitly recognized by the formal reward system, and . . . in the aggregate promotes the effective functioning of the organization” (Organ, 1988, p.4). This type of behavior connotes extra effort on the part of workers (Hodson, 1999) and has the accumulative effect of increasing the " While defined primarily in terms of attitudes, behaviors are also implied in most definitions of commitment; the committed individual not only believes in and accepts the organization’s goals (i.e., an attitude), but is also willing to exert considerable effort toward the accomplishment of those goals (i.e., a behavioral intention) (Fishbein, 1967). Steers (1977) makes the distinction between active and passive commitment. When organizational commitment is viewed as a positive affective response to the environment and a behavioral intention concerning future performance, this is an instance of active commitment (because behavior is implied). One may also be passively committed, however, in which case he/she is attitudinally committed to the organization, but these feelings do not translate into behavioral intentions. 55 effectiveness and productivity in an organization. These behaviors are considered extra- role because organizations are not able to anticipate in totality through formal job descriptions all the possible behaviors that are needed to achieve organizational goals (George & Brief, 1992). The concept of organizational citizenship behavior may be particularly relevant for persons in supervisory positions. Organ (1988), for instance, suggested that organizational citizenship behavior would be relatively more important for supervisors than the rank-and-file for at least three reasons: (1) employees are expected to contribute more to the organization as they move up the hierarchy (Organ, 1988; Borman & Motowidlo, 1993), (2) persons in supervisory positions are expected to act as role models and leaders given their highly visible positions (MacKenzie, Podsakoff, & Paine, 1999), and (3) organizational citizenship behaviors may help persons in supervisory positions become more effective. In other words, “ . . . managers are expected to make broader contributions to the organization than their subordinates . . . regardless of whether these expectations are explicitly stated by the managers’ superiors” (MacKenzie, Podsakoff, & Paine, 1999, p. 400). Like studies in the area of organizational commitment, a wealth of research has concerned itself with the antecedents and consequences of organizational citizenship behavior. Several researchers have suggested that one of the possible consequences of organizational commitment is organizational citizenship behavior (Scholl, 1981; Wiener, 1982; Williams & Anderson, 1991); recent studies support the link between these constructs (MacKenzie, Podsakoff, & Ahearne, 1998; Mathieu & Zajac, 1990; O’Reilly & Chatman, 1986; Schappe, 1998). The relationship between these two constructs, 56 however, is said to be weak at best. As Mathieu and Zajac (1990) concluded, “the present findings suggest that commitment has relatively little direct influence on performance in most instances” (p. 184). Instead, the relationship between organizational commitment and performance (i.e., organizational citizenship behavior) is likely moderated by a number of other factors (e. g., personal characteristics, work conditions). In several ways, a supervisor’s attitudes toward community policing and his/her practice of community policing may be considered roughly analogous to the concepts organizational commitment and organizational citizenship behavior, respectively.ls Community policing is being sold, at least in part, as the “future” of policing. Given the emphasis placed on community policing in most departments across the country,16 positive attitudes toward community policing may be considered a form of organizational commitment. By supporting community policing as a viable mode of policing, positive attitudes toward community policing represent a “strong belief in and acceptance of the organization’s goals and values” (Mowday, Porter, & Steers, 1982, p. 27). Positive attitudes toward community policing and organizational commitment may also be considered similar in that each is thought to influence a particular type of job performance. Attitudinal support for community policing (i.e., “winning the hearts and minds of police officers and supervisors”) is thought to impact the extent to which police personnel engage in community policing (Lurigio & Skogan, 2000), since the ’5 Certainly, these concepts are not entirely equivalent. In the following section, I provide a more in-depth discussion of organizational commitment and organizational citizenship behavior, and the ways these might and might not be considered analogous to attitudes toward community policing and the behavioral practice of community policing by police supervisors. 16 Recent research indicates that the majority of police departments have implemented some type of community-oriented program (e.g., foot patrol, block watch) in the last three years (Zhao, Thurman, & Lovrich, 2000), and over 90% of Americans are being served by community policing officers (US. 57 “supportive” supervisor will be more likely to engage in behaviors that further the goals and objectives of the police organization. The practice of community policing may be understood as a type of job performance, in that performance includes “actions or behaviors that are relevant to the organization’s goals and that can be scaled (measured) in terms of each individual’s proficiency (that is, level of contribution)” (Campbell et al., 1993). Community policing behaviors may be considered akin to organizational citizenship behaviors in two ways. First, each is a behavior that furthers the goals and improves the functioning of the organization. If a supervisor were to engage in community policing behaviors in a police department in which community policing was practiced, he/she would be furthering the goals of that department. The second way in which community policing behaviors and organizational citizenship behaviors are similar is that neither are explicitly recognized or rewarded by the organization. While community policing has become the dominant paradigm in policing (Cordner, 1998), with few exceptions, evaluation systems in policing have failed to keep pace with this change in organizational strategy (Glensor & Peak, 2000). Instead, evaluation systems continue to maintain an emphasis on the crime- fighting activities of police. Performance evaluations for supervisors may be particularly outdated, focusing almost exclusively on the monitoring and control aspect of police supervision, rather than the facilitating or mentoring role expected of them under a philosophy of community policing. As such, the practice of community policing among supervisory personnel, or their encouragement of such behavior in subordinate officers, continues to go unnoted, unrecognized, and otherwise unevaluated—makin g such behavior discretionary on the part of the police supervisor. Department of Justice, 1999). 58 Not only can organizational commitment and organizational citizenship behavior be considered analogous to attitudes toward and the practice of community policing, respectively, but the relationships expected among these constructs may also be similar. In each case, the former constructs (i.e., organizational commitment and attitudes toward community policing) are attitudinal constructs. The latter constructs (i.e., organizational citizenship behavior and the behavioral practice of community policing) may be considered the behavioral manifestations of underlying attitudes. In other words, organizational citizenship behavior and the behavioral practice of community policing can be considered visible “proof” of the extent to which an employee has embraced the goals of the organization. The extent to which one is willing to exert extra effort on the organization’s behalf becomes known through organizational citizenship behavior (or the practice of community policing). The relationship between organizational commitment and organizational citizenship behavior may also be considered analogous attitudes toward community policing and the behavioral practice of community policing in that each set of related constructs is expected to be only weakly related, emphasizing the importance of examining other factors that might influence this outcome. The preceding discussion has established that there are several ways in which the concepts of organizational commitment and organizational citizenship may be considered analogous to attitudes toward community policing and the behavioral practice of community policing, respectively. With these similarities in mind, I turn to a more in—depth consideration of the literature on organizational commitment and organizational citizenship behavior. 59 Literature on Organizational Commitment and Organizational Citizenship Behavior Interest in organizational commitment emerged with the identification of this construct “as an important variable for understanding the work-relevant behavior of organizational members” (Angle & Perry, 1983, pp.123-124). Organizational theorists originally conceptualized organizational commitment as a unidimensional construct (Buchanan, 1974; Mowday, Steers, & Porter, 1979; Porter et al., 1974). Organizational commitment was said to be characterized by three factors: (1) a strong belief in and acceptance of the organization’s values, (2) a willingness to exert considerable effort on behalf of the organization, and (3) a strong desire to maintain membership in the organization (Porter et al., 1974). Subsequent analyses later served to demonstrate the multi-dimensional nature of this concept. Angle and Perry (1981) were the first to delineate different types of organizational commitment by differentiating between value commitment and commitment to stay. Value commitment was defined as a positive affective response to the organization; this response involved an acceptance of the organization’s goals and the implied willingness to exert extraordinary effort on the organization’s behalf. Commitment to stay, on the other hand, represented an individual’s willingness to remain with the organization (rather than pursue alternative job opportunities). This type of commitment, derived from Becker’s (1960) notion of “side-bets” (Somers, 1995), was equated with the cost/benefit ratio of leaving the organization. Value commitment and commitment to stay were later termed affective and continuance commitment, respectively, by Meyer and Allen (1984). In 1990, Allen and Meyer identified yet a third component of organizational commitment, normative commitment. Normative commitment is an individual’s sense of obligation to an 60 organization (Lahiry, 1994), or a loyalty to the organization and its activities (Wiener, 1982). In sum, then, researchers have identified three types of organizational commitment: (1) affective, (2) continuance, and (3) normative (Meyer & Allen, 1991). Of these three types of organizational commitment, attitudes toward community policing can be best thought of as related to a particular type of organizational commitment; in particular, affective commitment. Affective (or value) commitment is the type of commitment traditionally defined as “an individual’s attitude toward the organization, consisting of a strong belief in, and acceptance of, an organization’s goals” and a “willingness to exert considerable effort on behalf of the organization” (Mowday, Porter, & Steers, 1982, p. 27). Affective commitment is also the type of commitment that is most directly the result of an individual’s work conditions. Other types of commitment, such as continuance commitment, may be the result of factors outside the control of the organization. For example, whether one chooses to remain with or leave an organization may be only partially related to an individual’s work conditions. Value commitment, on the other hand, is likely far more influenced by the work environment of an employee. Value commitment may be considered similar to March and Simon’s (1958) decision to produce, a decision explicitly about the work environment - does one choose to exert effort on the organization’s behalf or not? Antecedents of Organizational Commitment In the literature on commitment and its relation to behavior, researchers have examined several classes of variables that are thought to “cause” organizational commitment. In the most general sense, the antecedents of organizational commitment can be categorized as personal or situational in nature (Angle & Lawson, 1993). The 61 classification of variables as either personal or situational can be thought of as two different “models” of organizational commitment (Angle & Perry, 1983; Steers, 1977). The examination of personal variables is predicated on the view of organizational commitment as member-based (Angle & Perry, 1983). That is, organizational commitment is the result of factors inherent to the individual. Commitment results from the history and personal attributes a member brings to the organization (e.g., age, tenure, education) or from the member’s interactions with the organization. As Angle and Perry (1983, p. 127) described, “the member-based model focuses on what the member brings to the organization as well as what he or she does there” (emphasis in original; see also Porter & Miles, 1973). The second model of organizational commitment examines the influence of situational (i.e., organizational) variables. This model is based on the view that it is “the organization that is considered to be the initiator of actions that lead ultimately to an increase in the member’s organizational commitment” (Angle & Perry, 1983, p.124). In this model, the employee and organization are involved in an exchange relationship, a relationship that is based (at least in part) on a norm of reciprocity. When an organization treats its employees well and meets the employees’ needs and expectations, commitment becomes more likely. Overall, studies that have compared the efficacy of these two models in explaining organizational commitment have demonstrated the superiority of the organization-based model. Angle & Perry ( 1983), for instance, found that the member- based model accounted for 11% of the variance in organizational commitment while the 62 organization-based model accounted for 58% of the variance in the dependent variable.'7 These findings led Angle and Perry (1983) to conclude: commitment to an organization comes about through a process of reciprocation between organization and member. In essence, the organization provides the individual valued resources and, in exchange, the individual provides not only skills and energies but also his or her commitment to the organization (p.140). In a similar vein, Steers (1977) compared the explanatory power of individual- based and organization-based models in predicting organizational commitment. Both sets of antecedents were significantly correlated with organizational commitment, “emphasizing the diverse sources of factors affecting employee commitment in organizations” (p.51). Consistent with expectations, however, organizational influences (i.e., work experiences) were the most closely associated with levels of organizational commitment, explaining more variance in commitment than personal characteristics. In the paragraphs that follow, I review the literature with respect to these two general categories of organizational commitment antecedents. Personal Characteristics Personal characteristics are factors unique to the individual that may impact on levels of organizational commitment. This category of predictor variables is comprised of demographic characteristics, individual characteristics (e.g., Meyer, Irving, & Allen, 1998), and role states (Mathieu & Zajac, 1990). Demographic characteristics are variables such as age, sex, and gender. Individual characteristics, also known as individual difference variables, include such factors as need for achievement, need for ’7 Angle and Perry (1983) also combined variables from the individual and organization models to create a third model. The adjusted R2 for the third combined model was .600, indicating that the inclusion of individual variables only slightly increased the amount of variance explained. 63 autonomy, and need for affiliation.’8 Role states have been defined as perceptions that result from the work environment that influence affective responses such as organizational commitment (Mathieu & Zajac, 1990, p. 180). Several studies have examined the influence of personal characteristics, those “which define the individual,” on organizational commitment (Steers, 1977, p. 47). Despite the small magnitude of the effect personal characteristics have on organizational commitment, several studies have demonstrated the influence of specific personal characteristic variables; these variables and the findings pertaining to them are reviewed below. Demographic Characteristics. Age. Meyer and Allen (1984) suggest workers’ levels of organizational commitment increases with age. As workers get older, they have invested more time with the organization, tipping the scales in organization’s favor; the costs associated with leaving the organization outweigh the benefits of leaving at this stage in the individual’s career. Not only have workers invested more in their organization, but employment opportunities have concurrently decreased as well. The result is a positive correlation with organizational commitment. Tenure. A meta-analysis of nearly 200 studies on organizational commitment by Mathieu & Zajac (1990) differentiated between organizational and position tenure. Both types of tenure tend to increase commitment (Allen & Meyer, 1993; Angle & Perry, 1981; Mowday, Porter, & Steers, 1982), although Mathieu and Zajac’s (1990) analysis suggested that the magnitude of this correlation is small. Moreover, while organizational tenure showed a stronger relationship to commitment in 8 . . . . . . . . . . . ’ Indrvrdual difference variables Will not be consrdered m thrs paper, as my primary interest lies in the effect(s) of the work environment on behavior. 64 general, position tenure was more strongly associated with attitudinal commitment to the organization. Sex. Some studies have revealed a negative relationship between sex and organizational commitment. Recent research (Dodd-McCue & Wright, 1996) suggests that findings regarding sex and commitment remain inconclusive. For example, Mathieu and Zajac (1990) found a small but positive relationship between gender and commitment. Researchers have suggested that women are more committed than men because of the additional barriers they must overcome in order to gain organizational membership (Grusky, 1966). Race. Similar suggestions have been made regarding the relationship between race and organizational commitment. Like their female counterparts, minorities face greater obstacles in gaining membership to organizations. The difficult process of entry into organizations tends to have a more profound effect on attachment to the organization, resulting in greater levels of organizational commitment among persons of Color. Education. Research has documented an inverse relationship between education and organizational commitment, with more educated individuals less likely to be committed to their organizations (Mathieu & Zajac, 1990). Mowday, Porter, and Steers (1982) suggested that more educated individuals may have greater expectations of the organization. When the organization fails to live up to these expectations, individuals are less likely to develop attachment to the organization. Additionally, more educated individuals have a greater number of job opportunities, 65 making it less likely that they would become too committed to any one organization (Mathieu & Zajac, 1990). Ability. Mathieu and Zajac’s (1990) study showed a positive relationship between ability and (affective) organizational commitment. In their interpretation of this relationship, Stevens, Beyer, and Trice (1978) proposed that highly skilled individuals are more valued by organizations. Persons with greater ability are most likely subject to greater rewards and opportunities in the organization, which in turn increases their commitment. Perceived Competence. Perceived competence exhibits a large positive correlation with attitudinal organizational commitment (Mathieu & Zajac, 1990). In explaining this finding, researchers have suggested that individuals become more committed when they perceive the organization as providing for their growth and achievement needs (Morris & Sherman, 1981). Although few studies have explored the effects of ability on organizational commitment, those that have indicate that these variables are positively correlated (Mathieu & Zajac, 1990). Role States. Role states are attitudes or perceptions that result from the work environment (Mathieu & Zajac, 1990; Mowday, Porter, & Steers, 1982). Mathieu and Zajac’s (1990) meta-analysis indicated that three role states were related to organizational commitment: role conflict, role ambiguity, and role overload. Role conflict may be defined as “logically incompatible demands made upon an individual by two or more persons” (Kahn, 1973, p. 5). Research has documented an inverse relationship between role conflict and organizational commitment; as role conflict increases, commitment levels decrease (Brown & Peterson, 1993; MacKenzie, Podsakoff, 66 & Aheame, 1998). Role ambiguity “captures. . . unclear expectations” (Mayer & Schoorman, 1998, p. 20) or a discrepancy between the information needed to perform a role adequately and the information possessed (Kahn, 1973). Several studies have documented a negative relationship between role ambiguity and organizational commitment (Gregersen & Black, 1991; MacKenzie, Podsakoff, & Aheame, 1998; Morris & Koch, 1979; Morris & Sherman, 1981; Schechter, 1985; Welsh & LaVan, 1981). In a similar vein, other researchers have found a positive relationship between commitment and role clarity (Domstein & Matalon, 1989; Mayer & Schoorrnan, 1998). Finally, role strain, which may be thought of as the requirements of the role in comparison to the resources necessary to effectively carry out that role, has been shown in several studies to possess a negative relationship with organizational commitment (Mathieu & Zajac, 1990). Situational Characteristics The situational characteristics discussed in this literature almost exclusively focus on characteristics of the work environment of the organizational member, as these factors are those thought to most directly influence work-related attitudes and behaviors. Researchers have come up with a variety of means of classifying situational characteristics into meaningful categories (Harrison & Hubbard, 1996). For example, some researchers have compartmentalized situational characteristics into two smaller subcategories: (1) job characteristics and (2) work experiences (Angle & Perry, 1983; Steers, 1977). Job characteristics are characteristics of the individual’s job such as the autonomy, variety, opportunities for social interaction, and amount of feedback provided on the job (Hackman & Oldham, 1975, 1976). Work experiences, on the other hand, 67 provide an indication of the quality and nature of the individual’s work experiences. Another category of situational characteristics used by researchers is group-leader relations (Mathieu & Zajac, 1990). Group-leader relations are characterizations of the working relationships an individual shares with his/her co-workers and supervisors. These four categories of situational characteristics, the variables that comprise them, and results from studies are presented below. Job Characteristics. Job characteristics have been examined with respect to organizational commitment based on the premise that individuals with more enriching and rewarding jobs will be more committed to the organizations in which they work (Steers, 1977). As Eby et al. (1999) proposed, working in a job “with motivating potential,” as indicated by certain job characteristics (i.e., skill variety, task significance, identity, autonomy), increases motivation. In turn, motivation has the effect of increasing affective commitment to the organization (p.465). Most often, researchers have drawn upon Hackman and Oldham’s (1975, 1976) job characteristics model in order to assess the influence of job characteristics on organizational commitment, evaluating jobs on the basis of their autonomy, skill variety, task significance, and autonomy. The job characteristics that have been demonstrated to significantly correlate with organizational commitment are discussed below. Autonomy. Studies have revealed a small positive relationship between autonomy and organizational commitment (Eby et al., 1999; Mathieu & Zajac, 1990). Mathieu and Zajac (1990) suggested that persons with more autonomy find their jobs more challenging and enriching, thereby affecting levels of organizational commitment in a positive fashion. Alternatively, workers who experience more 68 autonomy on the job may feel more empowered, which translates into more positive feelings toward the organization, manifesting itself in higher levels of organizational commitment (Eby et al, 1999). Skill Variety. Mathieu and Zajac’s (1990) meta-analysis indicated a medium-sized positive relationship between skill variety and organizational commitment (cf Eby et al., 1999). Jobs with greater skill variety may indicate more skilled employees, who are more likely to exhibit commitment to the organization (Mathieu & Zajac, 1990; Stevens, Beyer, & Trice, 1978). It is also possible that persons with jobs requiring a variety of skills find the jobs more challenging and enriching, thereby affecting levels of organizational commitment in the same fashion as autonomy. Work Experiences. The inclusion of variables capturing work experiences is based on the belief that it is not simply characteristics of the job that may impart attachment to the organization, but more generalized experiences within the organizational setting as well. This school of thought emanated largely from Buchanan (1974), who argued that commitment is largely a consequence of work experiences. Steers’ (1977) work highlighted the importance of distinguishing between job characteristics and work experiences: work experiences were more closely related to commitment than were job characteristics. Several work experience variables have been shown to correlate significantly with organizational commitment; the findings as they pertain to these variables are discussed below. Participation. Perhaps the work experience variable demonstrating the most robust relationship with organizational commitment is participation. Several studies have revealed a significant and positive relationship between participation and 69 commitment levels (Domstein & Matalon, 1989; Eby et al., 1999; Fields & Thacker, 1992; Leana, Ahlbrandt, & Murrell, 1992; Mathieu & Hamel, 1989; Mayer & Schoorman, 1998; Rhodes & Steers, 1981; Schechter, 1985). For example, Leana, Ahlbrandt, and Murrell (1992) found that involvement in problem-solving teams increased organizational commitment. Domstein and Matalon (1989) and Mathieu and Hamel (1989) found decentralization of decision-making to be related to commitment, with individuals engaging in this type of decision-making exhibiting higher levels of organizational commitment. Rewards. Many researchers have conducted research on the way perceptions of fairness and equity may influence organizational commitment. These studies have shown that affective commitment significantly increases when the workplace is perceived as a fair and equitable one (Whitener & Walz, 1993; Williams & Hazer, 1986). Explicit rewards provided by the organization to the individual are one way of promoting a view of the workplace as fair (Eby et al., 1999). As Mueller et al. (1994) described, fairness is the “degree to which rewards and punishments are related to performance inputs into the organization” (p.186). When one is rewarded for exerting effort on behalf of the organization, commitment becomes more likely (Mueller et al., 1994). Group-Leader Relations. This category of variables encapsulates various means of tapping into the quality of relationships an employee shares with the persons with whom he/she works. One possesses relationships of both a vertical (i.e., equal) and horizontal (i.e., subordinate or superordinate) in the workplace. Assessing the quality of both types of relationships is important, as “[c]o-worker relations are an important part of 70 the ‘social climate’ at work (Moos, 1986:14) and provide a setting in which workers experience meaning and identity” (Hodson, 1999, p. 463). Group Cohesiveness. Group cohesion has been a popular focus of research on organizational commitment. One common means of assessing group cohesion is the study of value congruence, as this . . is an important dimension of an individual’s fit with the work environment”(Adkins, Ravlin, & Meglino, 1996, p. 439). Randall and Cote (1991) examined the relationship between workgroup attachment, defined as an individual’s identification and sense of cohesiveness with fellow organizational members, and organizational commitment. This study was based on the premise that the social bond is an important influence on organizational commitment and that the work group acts as a major socializing force in an organization. In line with social involvement theory, Randall and Cote (1991) hypothesized that employees who were committed to the work group to which they belonged would be more likely to be committed to the organization as well (Kanter, 1968). Results of this study demonstrated support for the positive nature of this relationship, as have other studies (e.g., Cohen, 2000; Mueller et al., 1994; Steers, 1977; Stone & Porter, 1975; Welsch & LaVan, 1981). While most work demonstrates a positive relationship between workgroup cohesiveness and organizational commitment, other work suggests this relationship is not as clear-cut as one might think. Howell and Dorfman (1981), for example, found a negative correlation between group cohesiveness and organizational commitment. This work suggests that it may be the nature of the values around which the group is coalesced that is important, not simply the level of group cohesiveness. For instance, if a group is bound together by a distrust of management, such beliefs are likely to translate into less, 71 not more, commitment. On the other hand, if the workgroup holds similar (and positive) values such as teamwork, such a supportive environment may translate to higher levels of organizational commitment. In this respect, it is important to take into consideration the nature of the values under examination in addition to the level of workgroup cohesion. Supervisory Feedback. Eby et al. (1999) found a direct and positive relationship between supervisory feedback and organizational commitment. Supervisory feedback also influenced organizational commitment indirectly through motivation. Eby et al. (1999) suggested that the importance of feedback derives from its importance in reassuring individuals that they are valuable contributors to the organization. As they stated, “without feedback, even the most challenging and meaningful jobs may fail to foster commitment because an individual’s role in the overall functioning of the organization is unclear” (p. 474). Steers (1977) also examined the relationship between feedback provided on the job and commitment, finding a small but statistically significant relationship. Consequences of Organizational Commitment The great majority of studies examining the consequences of organizational commitment have adopted what Chelte and Tausky (1987) termed the “organizational behavior approach,” which is based on the belief that commitment may be inferred “not only from the opinions and beliefs of organizational participants but also from their actions” (p.554). Thus, research of this type is aimed at examining the logical behavioral consequences of organizational commitment. In research examining the outcomes of organizational commitment, researchers have examined a variety of dependent variables, including turnover (Cohen, 2000; Eby et al., 1999; Steers, 1977), intent to stay (Aryee & 72 Heng, 1990; Cohen, 1993), performance (Angle & Perry, 1983; Aryee & Heng, 1990; Mayer & Schoorman, 1992; Meyer et al., 1989; Schechter, 1985; Steers, 1977), job satisfaction (Schechter, 1985), absenteeism (Blau, 1986; Cohen, 2000; Steers, 1977), and tardiness (Blau, 1986). Researchers have also examined the link between organizational commitment and extra-role or organizational citizenship behavior (Aryee & Heng, 1990; MacKenzie, Podsakoff, & Aheame, 1998; Mathieu & Zajac, 1990; Mayer & Schoorman, 1992; Meyer & Allen, 1991; O’Reilly & Chatman, 1986; Organ & Ryan, 1995; Schappe, 1998; Schechter, 1985; Wagner & Rush, 2000). Before examining the relationship between organizational commitment and organizational citizenship behavior, it is useful specify what is meant by this term. The work of Katz (1964) is often cited as laying the foundation for the study of organizational citizenship behavior. Katz (1964) argued that three behaviors are critical to the effective functioning of organizations: (1) people must be induced to enter and remain with the organization, (2) employees must carry out specific role requirements in a dependable manner, and (3) there must be innovative and spontaneous activity that goes beyond role prescriptions (p.132). Organizations need innovative and spontaneous activity that goes beyond role prescriptions because it is not possible to delineate all actions necessary for effective organizational functioning in formal job descriptions (George & Brief, 1992). In this sense, behaviors falling in this category may be considered a type of “extra-role” performance (Organ & Ryan, 1995). After Katz (1964), researchers became interested in “contextual performance,” which was also based on the distinction between in-role and extra-role performance 73 (Borman & Motowidlo, 1993; Motowidlo & Van Scotter, 1994). “Contextual performance” (i.e, extra-role performance) is differentiated from “task performance” (i.e., in-role performance) in that the former involves “doing things that are not main task functions but are importance because they shape the organizational and social ‘context’ that supports task activities” (Organ & Ryan, 1995, p. 776). Finally, we come to what is now called organizational citizenship behavior, commonly defined as “individual behavior that is discretionary, not directly or explicitly recognized by the formal reward system, and that in the aggregate promotes the effective functioning of the organization” (Organ, 1988, p. 4). As this definition implies, the distinction between in- and extra-role performance is particularly salient to organizational citizenship behavior. Like organizational commitment, researchers have identified organizational citizenship behavior as a multi-dimensional construct. In general, most research relies on a two-dimension conceptualization of organizational citizenship behavior as articulated by Smith, Organ, & Near (1983) (Organ & Ryan, 1995). Here, organizational citizenship behavior encompasses two factors: altruism and compliance. Altruism involves providing aid to a specific person in need of assistance, typically a co-worker. Compliance encompasses behaviors of a more impersonal nature (e.g., exemplary attendance, use of work time) that contribute to overall organizational functioning. More recent studies recognize three additional organizational citizenship behavior factors identified by Organ (1988): courtesy, sportsmanship, and civic virtue. Courtesy involves gestures taken to prevent problems of co-workers. Sportsmanship has been defined as a willingness to sacrifice without complaint. Finally, civic virtue is involvement in matters that pertain to the governance of the organization. The behavioral practice of community 74 policing is most analogous to compliance, where the extra effort exerted by the employee contributes to the overall functioning of the organization, but is rather general in nature. Returning to the relationship between organizational citizenship behavior and organizational commitment, a variety of studies have been conducted examining antecedents of organizational citizenship behavior. One such antecedent is organizational commitment. Results of research investigating the organizational commitment- organizational citizenship behavior link demonstrate a significant and positive relationship between these constructs (Aryee & Heng, 1990; MacKenzie, Podsakoff, & Aheame, 1998; Mathieu & Zajac, 1990; Mayer & Schoorman, 1992; Meyer & Allen, 1991; O’Reilly & Chatman, 1986; Organ & Ryan, 1995; Schappe, 1998; Schechter, 1985; Wagner & Rush, 2000). Summary Figure 1 presents a summary of the relationships discussed in the preceding discussion on the antecedents and consequences of organizational commitment. This figure, drawn in part from Mathieu and Zajac (1990, p. 174), graphically depicts the antecedents and consequences of organizational commitment as indicated by a review of the literature. The straight lines between personal and situational categories of variables and organizational commitment indicate direct relationships between these factors and commitment. In addition, the directional signs (+/--) placed next to individual variables depict the nature (i.e., direction) of relationships expected between personal and situational variables and organizational commitment as well as organizational commitment and organizational citizenship behavior. 75 3 338$ mOgémm m—ImZmNF—LU AP~ZD§OU a<>>Oh a§—:< I 2235 20m leasm 23. TV 8:209:00 33885— 3 5:5 gala 1383.— TV soEnE< cone—=05 TV 05:3. .30 TV Paco“. :oEmom TV cannon—om 3 SEE: 3 ”sea 3 uw< mag cinnamon—on mnemgconsm 8:.5 38:2 SEE—om wag—ca— mzaafifieu 952—5.:— Eee—Z .N 2sz 87 model. These models regress the practice of community policing by supervisors on demographic characteristics, personal characteristics, role states, work experiences, and group-leader relation variables. Model 1: Demographic Characteristics Research in the area of organizational psychology indicates the following variables are positively related to organizational commitment: age, position tenure, organizational tenure, gender (females), and race (minorities). Education is expected to be negatively related to organizational commitment (Mathieu & Zajac, 1990). The policing literature suggests similar hypotheses with the exception of age, tenure, and education; in this literature, age and tenure are expected to decrease the likelihood of community policing behaviors (Skolnick & Bayley, 1988; Williams & Sloan, 1990) while more educated officers are expected to be more likely to engage in community policing (Carter, Sapp, & Stephens, 1989).24 A final variable, promotion ambition, is not explicitly identified in either the policing or organizational literatures. This variable should nevertheless be included in a study of supervisor support for community policing because, at least intuitively, one would expect that supervisors who wish to advance in the organization will be more likely to pursue organizationally defined goals (i.e., community policing). Based on the policing and organizational psychology literatures, it is possible to offer hypotheses about the relationships between personal characteristics and supervisor support for community policing. Hypothesis 1: Positive relationships are expected between gender (female), race (minority), promotion ambition, education, and the practice of community policing by police supervisors. Negative relationships are predicted between age, position and organization tenure, and the 24 Where the organizational psychology and policing literatures offer competing hypotheses regarding the influence of particular variables, hypotheses are offered based on the policing literature. 88 community policing behaviors of supervisors. Model 2: Personal Characteristics The organizational psychology literature suggests that ability and perceived competence are variables that significantly increase the likelihood of organizational commitment (Mathieu & Zajac, 1990) and organizational citizenship behavior (MacKenzie, Podsakoff, & Aheame, 1998; Maurer & Tarulli, 1994; Noe & Wilk, 1991). While these variables are not explicitly recognized in the policing literature as exerting a significant impact upon the practice of community policing, they may be conceptualized as similar to two variables that are mentioned in the policing literature: knowledge and training (Sadd & Grinc, 1994; Williams & Sloan, 1990). Knowledge and training may be conceived of as providing ability and instilling perceived competence; without the requisite knowledge or understanding of community policing, supervisors may not perceive themselves as possessing the ability or competence to perform these behaviors effectively. Hypothesis 2: Positive relationships are expected between self assessments of knowledge of and training in community policing and the practice of community policing by police supervisors. Model 3: Role States Researchers have investigated the impact of three role states on organizational commitment: role ambiguity, role conflict, and role overload (Mowday et al., 1982). In these studies, all three perceptions of one’s work environment exhibit a negative relationship to organizational commitment (Mathieu & Zajac, 1990). Policing scholars have also identified role conflict as influencing supervisor support for community policing (Sherman, 1973, 1975). The extant literature suggests the following hypothesis: 89 Hypothesis 3: A negative relationship is expected between role conflict and the extent to which supervisors engage in the practice of community policing. Model 4: Work Experiences The organizational psychology literature suggests that the category of work experiences may be expected to exert the strongest effect, comparatively speaking, on organizational commitment (Angle & Perry, 1983; Steers, 1977). The literature in organizational psychology points to participation in decision-making as a significant predictor of organizational commitment (Domstein & Matalon, 1989; Eby et al., 1999; Fields & Thacker, 1992; Harrison & Hubbard, 1996; Leana, Ahlbrandt, & Murrell, 1992; Mathieu & Hamel, 1989; Mathieu & Zajac, 1990; Mayer & Schoorman, 1998; Rhodes & Steers, 1981; Schechter, 1985) and organizational citizenship behavior (Coyle-Shapiro, Kessler, & Purcell, 1999; Hodson, 1999; Moorman, Niehoff, & Organ, 1993; Organ & Ryan, 1995; Thibaut & Walker, 1975; Van Yperen, van den Berg, & Willering, 1999). Perceptions of equitable rewards are also important in predicting levels of organizational commitment (Whitener & Walz, 1993; Williams & Hazer, 1986) and organizational citizenship behavior (Farh, Podsakoff, & Organ, 1990; Moorman, 1991; Neihoff & Moorman, 1993). Researchers in the area of policing have identified these same variables (Glensor & Peak, 2000; Sherman, 1975; Wilkinson & Rosenbaum, l994)—as well as opportunity (DeJong, Mastrofski, & Parks, 2001)—as significant predictors of support for community policing. These findings suggest the following hypothesis: Hypothesis 4: Positive relationships are expected between participation, opportunity, rewards, and the practice of community policing by police supervisors. 9O Model 5: Group-Leader Relations Group-leader relations, such as group cohesiveness and supportive leadership, are widely cited as exerting significant effects on commitment (Cohen, 2000; Glisson & Durick, 1988; Mathieu & Zajac, 1990; Mueller et al., 1994; Randall, 1993; Steers, 1977; Stone & Porter, 1975; Welsch & LaVan, 1981) and organizational citizenship behavior (Farh, Podsakoff, & Organ, 1990; Maurer & Tarulli, 1994; Podsakoff, MacKenzie, & Bommer, 1996; Schnake, Dumler, & Cochran, 1993; Settoon, Bennett, & Liden, 1996; Smith, Organ, & Near, 1983; Vaaneren, van den Berg, & Willering, 1999; Wayne, Shore, & Liden, 1997). The policing literature also suggests that these factors may influence the behavior of police supervisors. As the “men in the middle” (Trojanowicz, 1980), supervisors may be influenced by their subordinate and superior officers (Reuss- Ianni, 1983), as well as their fellow supervisors. The organizational psychology literature suggests, however, that attitudes and behavior might be most directly influenced by the individuals with whom one most regularly interacts (Jackson, 1992). Police supervisors arguably have the greatest contact with their subordinate officers. Consequently, this group may be expected to exert the greatest influence on their behavior. These research findings lend support to the following hypothesis: Hypothesis 5: Positive relationships are expected between group cohesion (to the extent that the group is coalesced around values supportive of community policing), supervisory feedback/support, and the practice of community policing by police supervisors. Conclusion This chapter has outlined the ways in which work in the field of organizational psychology—specifically the study of organizational commitment and organizational citizenship behavior—can be used to overcome the problematic nature of previous 91 inquiries into the reaction of police supervisors to community policing. For one, operationalizing supervisor support for community policing as a behavioral construct allows for Situating this study within the broader context of the relation between attitudes and behavior. Second, a framework borrowed from the literature in organizational psychology, namely research conducted in the areas of organizational commitment and organizational citizenship, provides a logical set of variables upon which to focus in the quest to understand the community policing behaviors of police supervisors. 92 CHAPTER FIVE: DATA AND METHODS Data The current study is a secondary analysis of data originally collected for the N [J- funded Project on Policing Neighborhoods (POPN). This study, conducted in the Indianapolis (IN) and St. Petersburg (FL) Police Departments (during the summers of 1996 and 1997, respectively), involved multiple data collection methods. The primary method of data collection was the Systematic Social Observation (880) of police officers and supervisors. In addition to observations, structured interviews were conducted with the majority of police officers and field supervisors. Unstructured, open-ended interviews were also conducted with field supervisors during observational sessions, and with district commanders and other police personnel in each department. Collectively, these data sources provide insight into police behavior during a time of community policing. Site Descriptions Site #1: Indianapolis, Indiana The jurisdiction of the Indianapolis Police Department (IPD) is referred to as the Police Services District, a portion of Indianapolis-Marion county for which the department is responsible.25 At the time of this study, the IPD served a population of approximately 375,000.26 The UCR Index Crime was 100 per 1,000 residents and 37 per 25 The remainder of the Consolidated City-County of Indianapolis is served by the Marion County Sheriff’s Department and other smaller city and town departments (Parks et al., 1999). 26 In 1995, the year prior to the POPN study, the population served by the Indianapolis Police Department was estimated to be 377.723 (Parks et al.. 1999). 93 officer (Parks et al., 1999). In the years 1996-1997, the department employed just over 1,000 sworn officers, or approximately 2.7 officers per 1,000 residents. Nearly half (N=492) of the department’s sworn officers were assigned to a patrol function. The sworn force was 84% male and 79% white (Parks et al., 1999). The geographic responsibility for the Indianapolis Police Department was divided among four patrol districts: North, West, East, and South. Within each district, officers were assigned to one of five shifts: Day (5:00 am to 2:00 pm), Day Tact (9:00 am to 5:00 pm), Middle (1:00 pm to 9:00 pm), Late Tact (7:00 pm to 3:00 am), and Late (10:00 pm to 6 am). These shifts were staggered so that adjoining shifts overlapped during times of peak service needs. Officers’ and supervisors’ work schedules were determined by their assignment to one of three work schedules with rotating days off (referred to as A, B, or C “letter days”). Thus, work squads were demarcated by district, shift, and work schedule. A squad was comprised of the officers and supervisors assigned to the same district, shift, and letter day. Each patrol district in the [PD was housed in a separate building and headed by its own Deputy Chief. The Deputy Chiefs were vested with considerable autonomy and latitude in determining the day-to-day operation of the districts. This resulted in variety among the districts, particularly with regard to community policing: the organization and practice of community policing varied considerably by district. In the West district, community policing was engaged in primarily by eight officers and one sergeant assigned to the Day Tact shift. Community policing in this district tended toward an aggressive order maintenance approach. The squad would work in unison to identify areas in need of attention, typically areas beset by street-level 94 narcotics activity, prostitution, and gambling. These problems were dealt with by aggressive enforcement tactics such as drug interdiction and stings. The organization of community policing in the North district was similar to that of the West, with community policing officers (and a supervisor) assigned to the Day Tact Shift. Unlike the West, however, the focus of community policing in the North was more on “community building” (Mastrofski, Worden, & Snipes, 1995) than aggressive order maintenance. Officers were encouraged to interact with community members toward the end of forging more cooperative relationships with members of the public. The organization and focus of community policing in the East District can also be differentiated from that in other districts. In this district, community policing officers were assigned to the Crime Impact Unit. While assigned to a specific unit, officers were not required to work a set schedule; instead, officers could flex their schedules according to the problem(s) they were working on at the time. Community policing in this district is best conceptualized as having a problem—solving emphasis, with officers encouraged to use the SARA model of problem-oriented policing to identify and resolve problems (Eek & Spelman, 1987). No community policing officers were assigned to the South District. While the Deputy Chief in this district devoted a portion of the day tact shift to conducting community policing activities, this was essentially in name only—there was no overarching plan to community policing in this district (Shepard, 1999). Finally, a note on supervision in Indianapolis: it is questionable whether true supervisor-subordinate relationships existed in the IPD. First, the IPD did not have a formal evaluation mechanism in place; consequently one means of determining the 95 supervisor-subordinate relationship (i.e., by whom evaluates whom) is nonexistent. Second, there was considerable variation in the way supervisors working the same district, shift, and letter day shared supervisory responsibilities. Some supervisors split responsibility for a shift geographically (e.g., splitting the district by geographical boundaries), while others split supervisory responsibility by task (e.g., one supervisor would remain in the station performing administrative tasks while the other(s) would assume responsibility for the street supervision of officers). This variation in supervisory practices, as well as the fact that the [PD had no formal evaluation mechanism in place, makes it very difficult to establish one-to-one supervisor-subordinate relationships. Instead, supervision in the IPD might best be considered a shared responsibility (i.e., as in an organizational work group) among supervisors assigned to the same district, shift, and work schedule. Site #2: St. Petersburg, Florida St. Petersburg, Florida, located in southern Pinellas County, was inhabited by approximately 240,000 residents at the time of the POPN study. While St. Petersburg has a smaller population, its UCR Index Crime Rate rivaled that of Indianapolis, with 99 per 1,000 residents and 47 per officer (Parks et al., 1999). The violent crime rate in St. Petersburg (per 100,000 population) was also well above that of the national average: 2,254.4 versus 716.0 (Bureau of Justice Statistics, 1995). In the summer of 1997, the SPPD employed just over 500 sworn officers, or approximately 2.1 officers per 1,000 residents. Over half (N=283) of the department’s sworn officers were assigned to patrol. The sworn force was 78% white and 87% male (Parks et al., 1999). SPPD deployed patrol officers using four shifts: Day (7 am to 3 96 pm), Evening (3 pm to 11 pm or 4 pm to 12 am), 4th Relief (7 pm to 3 am), and Midnight (11 pm to7 am or 12 am to 8 am). The department divided geographic responsibility for patrol among three districts (North, South, and West), although they were all housed at department headquarters. Each district was responsible for a “zone” of the department’s 48 Community Policing Areas (CPAs).27 A zone was comprised of three sectors, with each sector representing a conglomeration of CPAs. The North was responsible for the 403, 503, and 605 sectors (the 403-603 zone), the South was responsible for the 108, 205, and 308 sectors (the 105- 305 zone), and the West was responsible for the 705, 80s, and 90s sectors (703-905 zone). At the time of this study, the SPPD had 63 community policing officers (CPOs). At least one CPO was assigned to each of the department’s CPAs, with some CPAs assigned more than one CPO. Unlike their 9ll-responder counterparts, CPOs were not assigned to a particular shift; they were allowed to adapt their schedules to the needs of the CPAs for which they were responsible. In St. Petersburg, the organization and practice of community policing was far more uniform than in Indianapolis. The permanent, geographic deployment of officers was designed to facilitate community-building by making “community policing officers the main liaison with neighborhood organizations” (Parks et al., 1999, p. 489). This tactic also encouraged a problem-solving approach to crime and disorder. Community policing officers, working in conjunction with community members, were able to identify areas/problems in need of attention and work together to take corrective courses of action. 27 In St. Petersburg, CPAs are analogous to the concept of a patrol beat; it is the smallest unit of geographic responsibility. 97 Supervision in St. Petersburg reflected “a compromise between geographic deployment and the strict squads philosophy” (Shepard, 1999, p. 91). Sergeants were responsible for all patrol officers working in the sector for which they were assigned responsibility. For example, a supervisor in the North (responsible for CPAs in the 405 through 605) would be assigned responsibility for either the 403, 503, or 603 sector (and all patrol officers assigned to CPAs within that sector). Sergeants were also given responsibility for one or more CPAs (and the CPOs working in those areas) within their sector. Supervisors were expected to perform both administrative and street supervision functions. The formal organization of supervision facilitated this; supervisors were scheduled to work 3 “temporal” (24 hours) and 2 “flex” shifts (16 hours) per week. On temporal days, sergeants were responsible for the street supervision of all officers assigned to their district and shift (not only the officers assigned to the sector for which they were directly responsible). Sergeants typically conducted roll call on these days as well. On flex shifts, supervisors were expected to engage in community policing (e.g., by working with the CPO(s) assigned to their CPA(s) on problem-solving activities) and administrative tasks in the office (e.g., payroll, scheduling).28 In contrast to Indianapolis, the SPPD had a formal evaluation system in place. The responsibility for evaluating the officers assigned to a sergeant’s sector (and CPA(s)) helped reinforce formal supervisor-subordinate relationships at this site. While this may appear to make supervision in St. Petersburg more straightforward than in Indianapolis, 28 While these were the formal expectations of police supervisors, flex days were often used for other purposes. For example, during the summer we observed supervisors, they often used flex days to attend mandatory training sessions. In addition, since summer is a popular time for vacations, sergeants often used their flex days to cover for other sergeants (assigned to temporal duty) who were on vacation. 98 three issues muddy the waters. First, since CPOs were not assigned to a particular work shift, the direct supervision of CPOs cannot be assumed. Second, sergeants were rarely in the field on flex days; instead, their subordinates were supervised by one of the other two supervisors assigned to the same district and shift. Consequently, although officers likely had the greatest amount of contact with their “official” supervisor, officers likely viewed any supervisor assigned to their district and shift as a supervisor as well. Finally, at times, patrol officers were assigned to fill in for sergeants (who were on vacation, sick leave, or otherwise unavailable). These patrol officers were designated as “acting” temporal sergeants for a shift. Patrol officers accorded this special function had already passed the sergeant's exam but had not yet been promoted. Acting sergeants were vested with the same authority and responsibilities as any other sergeants. Description of the Data Systematic Social Observation (SSO) of Police Supervisors. The POPN project involved the systematic observation of police supervisors at both research sites. Fieldwork in Indianapolis began in June, 1996 and concluded in August, 1996. Trained observers accompanied individuals with street supervision responsibilities29 during their normally scheduled shifts. In Indianapolis, the sampling of police supervisors was based on supervision as it was experienced in each of the study’s 12 study beats3°--thus 29 In Indianapolis, sergeants and lieutenants were responsible for the street supervision of officers. In St. Petersburg, sergeants were primarily responsibly for street supervision, although this activity was also performed by lieutenants on select shifts and patrol officers who occasionally covered sergeants' shifts. 30 The POPN study involved the observation of officers and supervisors assigned to 12 of the department’s 50 beats distributed across the four patrol districts. These beats were selected on the basis of spatial and temporal sampling. Beats were selected on the basis of an index of socio-economic distress (i.e., the sum of the percentage of female-headed families, percentage of unemployment among the adult population, and percentage of the population that was below the 50% poverty level). with the sample over- 99 sampling was not based on particular supervisors, but rather the study beat and shift. The original sampling plan involved conducting 5 supervisor rides for each study beat in a district (one for each shift). Since the number of study beats was not uniform among districts, it was expected that there would be more supervisor rides in some districts than in others.3| Additionally, one observer was assigned to conducting rides with police lieutenants (one ride per shift per district). In the end, changes to work schedules, the unavailability of supervisors, and other similar, unexpected happenings led to several alterations of this plan. The number of Indianapolis supervisors observed ultimately was fifty-eight supervisors (39 sergeants and 19 lieutenants; 78% of all IPD supervisors). These supervisors were observed over the course of 87 rides, for a total of over 600 hours of observation. In St. Petersburg, the original sampling plan for supervisors was similar to that of Indianapolis; the sampling of supervisors was done on the basis of supervision as it was experienced by officers in each of the 12 study CPAs in the city.32 The sampling plan was also modified to incorporate the nature of supervision in St. Petersburg (with sergeants engaging in temporal and flex duties), so that sergeants would be observed engaging in both types of required supervisory duties. Lieutenants in St. Petersburg were only observed during watch commander shifts, at which time they engaged in street biased toward areas marked by greater socio-economic distress to capture greater variation in service demands and conditions (Mastrofski et al., 1998, p. l). 31 The East district had four study beats (20 supervisor rides), the North and West districts had three study beats (15 supervisor rides each), and the South district had two study beats (lO supervisor rides). 32 The study beats in St. Petersburg (12 of the department's 48 CPAs) were selected to match the Indianapolis distribution of beats in terms of socio-economic distress, with an oversampling of areas beset by greater levels of distress (and police activity). These twelve CPAs were distributed across the department’s three patrol districts. 100 supervision responsibilities; consequently, only four lieutenants were observed in St. Petersburg.33 The final sample of supervisors in St. Petersburg also reflects an aspect of supervision described earlier: that occasionally patrol officers filled in for sergeants in a role referred to by the SPPD as an “acting sergeant.” In the end, 28 supervisors (25 patrol sergeants, 4 patrol lieutenants, 8 patrol officers “acting” as sergeants; 96% of all SPPD supervisors) were observed over the course of 72 observational sessions, providing a total of over 300 hours of supervisor observation. During the observation of supervisors in Indianapolis and St. Petersburg, observers took notes on the behavior of police supervisors, as well as the officers and citizens with whom they interacted. After observational sessions, observers used these notes to provide detailed narrative accounts of the rides. Observers converted their notes into coded data using observation instruments designed specifically for the project.34 The supervisor observation instruments consisted of five forms: the ride form, activity form, encounter form, officer form, and citizen form.35 One ride form was completed for every observation of a supervisor. Using this form, observers coded information such as the site, district, rank, and shift of the supervisor. The activity form was used for events that were not classified as “encounters” with other police or citizens (behaviors that typically were performed alone). This form captured information such as 33 With this exception, the position of lieutenant in St. Petersburg was strictly administrative. 34 These observation instruments were pilot tested prior to the POPN study by observers riding with police in department including Richmond, VA; Albany, NY; and Flint, MI. Additionally, the experience gleaned after the first summer of data collection in Indianapolis led to several additions to the forms prior to data collection in St. Petersburg. Research assistants on the POPN project later went back and coded this additional information from the Indianapolis supervisor ride narratives. 3S . . . . Consequently, observers collected data at five levels of analysrs: the ride. acttvrty. encounter, officer, and citizen levels of analysis. 101 the type of activity in which the supervisor was engaged, the length of the activity, the type of problem at which the activity was directed, and whether the activity was part of a long-term plan or project. The encounter form was used to code information about one of the following types of encounters: (1) any communication (verbal or otherwise) between the supervisor and an officer or citizen that lasts more than one minute, (2) any time there was physical force between the supervisor and another officer or citizen, (3) any time there were three or more verbal exchanges between the supervisor and another officer or citizen (each party speaking alternatively three times), or (4) any time the supervisor issued a command or direction to one or more other officers (Mastrofski et al., 1996). The encounter form was used to code information such as the length of the encounter, other participants in the encounter (i.e., officers, citizens, or both), the type of problem at which the encounter was directed, and the type of decisions that were made during the encounter. When encounters involved an officer, the ofiicer form was used to code certain characteristics of the officer, such as rank and age, as well as the nature of the interaction between the supervisor and officer. When encounters involved a citizen, the observer coded a citizen form, capturing information such as the age, race, sex, income, and demeanor of the citizen. Structured Interviews of Supervisors. Structured interviews were conducted with police sergeants and lieutenants at both research sites. Sixty-nine of the 74 (93%) police supervisors in Indianapolis were interviewed; all 37 SPPD police supervisors were interviewed. The interviews were designed to collect information on a variety of topics, such as supervisors’ beliefs about proper police roles, goals, and priorities; important 102 functions of police supervisors; supervisors’ perceptions of district management’s priorities, and attitudes toward community policing. Demographic information (e.g., race, sex, age) and background characteristics (e.g., education, tenure) were also obtained. The current study will use both data sources in its measurement of variables and statistical analyses. The measurement of variables, and information on the statistical analyses that will be performed, are presented in subsequent sections of this chapter. Sample In defining my sample, several considerations were taken into account. First, my study was intended to provide information on first-line police supervisors, the “men in the middle” who are charged with the responsibility of translating broad organizational missions and goals into meaningful directives for their subordinates. In order to ensure that I provided information on this group of individuals, I considered the rank-related responsibilities of supervisors at the two research sites. In Indianapolis, sergeants and lieutenants were responsible for street supervision and therefore encouraging or facilitating the practice of community policing among subordinates (N = 74). This was not the case in St. Petersburg, where the responsibilities of lieutenants were far more administrative in nature. Since lieutenants at this location were arguably more administration- than subordinate-oriented, they were excluded from my sample (N = 4). In addition, observations were conducted with patrol officers “acting” as sergeants at the St. Petersburg research site. These individuals were also excluded from my sample, as I did not believe they would accurately represent the attitudes, experiences and views of first-line supervisors (N=8). After excluding the lieutenants and “acting” sergeants from 103 the St. Petersburg site, I was left with a total of 25 supervisors from the St. Petersburg research site. Table 4. Sample of Indianapolis Police Supervisors by District, Shift, and Rank North East South West Total District District District District Day Shift 2 (Sgt.) 3 (Sgt.) l (Sgt.) 1 (Sgt.) 7 (Sgts.) l (Lt.) 1 (Lt.) l (Lt.) 1 (Lt.) 4 (Lts.) Day Tact 2 (Sgt.) 2 (Sgt.) 1 (Sgt.) 2 (Sgt.) 7 (Sgts.) Shift 1 (Lt.) l (Lt.) 1 (Lt.) 1 (Lt.) 4 (Lts.) Middle Shift 2 (Sgt.) 3 (Sgt.) 2 (Sgt.) 3 (Sgt.) 10 (Sgts.) l (Lt.) 1 (Lt.) l (Lt.) O (Lt.) 3 (Lts.) Late Tact 2 (Sgt.) 2 (Sgt.) 2 (Sgt.) 1 (Sgt.) 7 (Sgts.) Shift 0 (Lt.) 1 (Lt.) l (Lt.) 1 (Lt.) 3 (Lts.) Late Shift 1 (Sgt.) 2 (Sgt.) 2 (Sgt.) 3 (Sgt.) 8 (Sgts.) l (Lt.) 1 (Lt.) O (Lt.) l (Lt.) 2 (Lts.) Total 9 (Sgts.) 12 (Sgts.) 8 (Sgts.) 10 (Sgts.) 39 (Sgts.) 4 (Lts.) 5 (Lts.) 4 (Lt.s) 4 (Lts.) l7 (Lts.) A second consideration in creating my sample involved the data sources available for each supervisor. In order to examine demographic characteristics, personal characteristics, role states, work experiences, and group—leader relation variables, both data sources (i.e., observational, interview) were necessary. Using both data sources, however, required that supervisors were interviewed and observed, something that did not always happen. In Indianapolis, 56 supervisors (39 sergeants and 17 lieutenants) were interviewed and observed. In St. Petersburg (after excluding the 12 supervisors who were either lieutenants or patrol officers “acting” as sergeants), 25 supervisors were both interviewed and observed. This provided a total of 81 supervisors for whom all 104 three data sources were available. Tables 4 and 5 provide a breakdown of the observed supervisors in Indianapolis and St. Petersburg by district, rank, and shift. Table 5. Sample of St. Petersburg Police Supervisors by District and Shift South North West Total District District District Day Shift 2 (Sgts.) 3 (Sgts.) 4 (Sgts.) 9 (Sgts.) Evening Shift 3 (38’9“) 3 (Sgts.) 1 (Sgts.) 7 (Sgts.) Night Shift 3 (Sgts.) 3 (Sgts.) 3 (Sgts.) 9 (Sgts.) Total 8 (Sgts.) 9 (Sgts.) 8 (Sgts.) 25 (Sgts.) Measures Dependent Variable Supervisor Support for Community Policing This study measured supervisor support of community policing in behavioral terms. As DeJong et al. (2000) articulated, “given that personnel time comprises most of the police budget, tracking time allocated to problem solving is a logical way to reveal the amount of effort of this sort actually made” (pp.7-8). The measurement of time spent in community policing used in this study was very similar to the measure of “problem- oriented policing” used by DeJong et a1. (2000) in their analysis of POPN data. These researchers classified events as illustrative of problem-oriented policing when an activity or encounter met one of the following criteria: (1) Was this activity part of a long-term plan or project to deal with a problem? (2) Did the police try to determine the nature, extent, or causes of the problem? (3) Were the police trying to prevent the occurrence or recurrence of a problem? 105 (4) Did the activity involve communicating with representatives of citizen organizations or representatives of other service-providing organizations? When events met these pre~specified criteria, the length of the event was calculated by subtracting the start time from the end time of the event. Next, DeJong et al. (2000) summed all event-based time spent in engaging in community policing over an observation period. My measurement of time spent in community policing, however, differs in two respects from that used by DeJong et al. (2000). First, the measure used in this study represents the average amount of time supervisors spent in engaging in community policing per shift. Since some supervisors were observed for more than (or less) than one shift, this measure was standardized by total time observed. Second, because supervisors may monitor or encourage community—oriented behaviors in their subordinates in addition to engaging in community-oriented activities and encounters of their own volition, my measurement of time spent engaging in community policing also included the time supervisors spent in encounters with officers where one or more of the above criteria were met. A histogram depicting the distribution of cases for this measure is presented in Figure 3. As this chart shows, the range of time spent in community policing per shift ranged from a low of 0 to a high of 42.5 percent. As a group, supervisors spent an average of approximately 12% of a shift engaged in community policing (SD. = 9.08). At first glance, this may not appear to be a large percentage of time. However, it is possible to compare the average amount of time spent on community policing per shift with other tasks supervisors regularly perform. The following pie chart (Figure 4) breaks 106 down the average amount of time supervisors spent on different tasks during a shift. As this figure demonstrates, supervisors spent the majority of their time engaged in general patrol (28%) and in encounters with the public (30%). The percentage of time spent engaging in community policing (11.7%) would be the next largest amount of time spent in any activity. This category is not displayed in the chart, however, as it overlaps with the other categories presented. In other words, time supervisors spent on tasks such as “information gathering” or “administrative tasks” may also be classified as time spent engaged in community policing. In sum, the average amount of time supervisors spent engaged in community policing tasks is somewhat substantial relative to other activities in which the supervisor might engage. Figure 3. Distribution of Time Spent on Community Policing Per Shift Std. Dev = 9.08 Mean = 11.7 N = 81.00 % of community policing per shift 107 Figure 4. Breakdown of Supervisor Time Per Shift Encounters w/ HJinc 30% General Patrol 28% 0% Personal t" ‘ 10% ' , . ‘1, Problemdirected i “ 5°/o Adrrinistrative . l 14% ‘ l lnlorrration l Gathering 2% While I have described the measurement of my dependent variable, it is difficult to get a “feel” for what this measure might represent in terms of supervisor behavior. In order to provide additional context, I provide some examples of behaviors that would be captured using this coding scheme. Community Policing Behavior Example l: 5136 (W/M, 40s) met with c137 (W/F, 305), who worked for the city’s code enforcement unit. S1 and Cl had been meeting on a weekly basis for a little over a month regarding an abandoned house located in S 1 ’5 district. S1 was working with C1 to have the house torn down. The abandoned house had become a haven for drug dealers and users. Sl reported that neighbors were complaining about the amount of drug activity going on at the house, and that his officers had made several arrests there in recent weeks. S1 and Cl spent the meeting discussing the different ways that S1 36 “S 1" stands for the supervisor under observation. 37 "Cl" refers to the citizen with whom the supervisor is interacting. 108 and his officers could document the problem and work together to have the house torn down. Community Policing Behavior Example 2: S1 (W/F, 40s) explained that she would be spending most of the shift engaged in a prostitution sting. Sl informed me that officers were allowed to work overtime to conduct prostitution stings from time to time in a particular part of town (this was funded by Weed and Seed monies). S1 acted as the “prostitute” in these stings when she was on duty and the officers were conducting their stings. The problem area was a part of her district centrally located to several factories. Community Policing Behavior Example 3: SI (WM, 503) met with Cl (W/M, 40s), the manager of a large grocery store. C1 was having trouble keeping homeless and vagrant persons from congregating outside his store entrance, where they would ask patrons for extra change. S1 told me that we were dropping in on C1 to see if a plan that one of his officers had devised to deal with the problem had been successful. According to S1, C1 had called the police numerous times about the loiterers bothering his customers. For some time, the police would respond and order the loiterers to leave. While the loiterers would leave for a short amount of time, they always returned. The officer had come up with a creative way of trying to deal with the problem. The officer had informed C1 to call him on his cell phone when vagrants were there in sufficient number. The officer would respond and issue the men (and/or women) “no trespass” warnings. If the person was there again after being issued a “no trespass” warning, he or she could be arrested for trespassing. Cl greeted Sl warmly, obviously familiar with him. Sl asked Cl how things had been going with loiterers outside his store. C1 responded that the community policing officer’s plan had been enormously effective in driving away the loiterers. Cl stated that it only took the officer coming out once or twice before the loiterers “got the hint” and probably moved elsewhere. 51 and C1 chatted casually for awhile before Sl heard a call on the radio and told C1 he would have to leave. 109 Independent Variables Supervisors ’Attitudinal Support for Community Policing This study measures supervisors’ attitudinal support for community policing using the Indianapolis and St. Petersburg supervisor interview data. Attitudes toward Community Policing is an additive index created by summing supervisors’ responses to two questions about their conceptions of the police role. Police supervisors were provided with a list of seven patrol officer goals: (1) handling calls for service to their assigned area; (2) making arrests and issuing citations; (3) reducing the number of repeat calls for service to the same address; (4) seizing drugs, guns, and other contraband; (5) reducing the level of public disorders; (6) getting the public involved in improving the neighborhood; and (7) reducing the public’s fear of crime. Supervisors were asked to select two items from the list that they considered the “most important” goals for patrol officers to accomplish. They were also asked to select the two tasks they considered the “least important” for officers to accomplish. The remaining items were to remain unmarked, indicating that they felt the tasks were neither the most nor the least important goals for patrol officers to accomplish. These items were then coded so that “1 = most important” and “0 = other (least important or neither most nor least important)” To create an Attitudes toward Community Policing Index, the answers to the two goals (i.e., reducing the number of repeat calls for service to the same address and getting the public involved in improving the neighborhood) were summed. Descriptive statistics for these survey items are presented in Table 6. After summing these items, the scale ranged in value from 0-2. The mean of the scale was 0.54 (SD. = .67) indicating that 110 most supervisors only selected only one of the survey items as the “most important” goal of patrol officers. Table 6. Attitudinal Measures of Support for Community Policing Survey Item Value N % Reducing the number 0f repeat 0 = Otherwise 60 74.1 calls for servrce to 1 = Most Important 21 25.9 the same address Getting the public involved in 0 = Otherwise 58 71.6 improving the l = Most Important 23 28.4 neighborhood Demographic Characteristics The current study includes several demographic characteristics. These variables include sex (male/female), race, age, tenure, education, and promotion ambition. A specification of the measurement of these independent variables is provided in Table 7. Table 7. Demographic Characteristics Variable Coding N % Sex 0 - male ?: f2: 1 — female ' Race 0 - white 71 87.7 1 — black 10 123 Age # of years Mean: 44.09 SD: 7.99 Range: 31-70 Tenure # of years on force Mean: 18.96 S. D.: 7.01 Range: 9-41 111 Table 7 (cont’d). Education 0 - Otherwise 47 58.0 1 — Bachelor’s degree or greater 34 420 Importance of 0 - very unimportant 17 18.5 Promotion 1 - somewhat unimportant 14 173 2 - somewhat important 31 38.3 3 - very important 21 25.9 Personal Characteristics A review of the organizational psychology literature suggested that ability and perceived competence significantly influence organizational commitment and job performance (and thus perhaps support for community policing). The POPN interview did not contain questions about perceived ability per se, but supervisors were asked to provide an indication of: (a) the amount of training received on community policing topics and (b) the level of knowledge they believed they had on those topics. These measures may be used as proxies for ability and perceived competence, respectively. Table 8. Ability Index Survey Items Survey Item Value N % Community policing concepts 0 _ none 3 3'7 and principles 1 ‘ < 1 day 6 7.4 2 — 1-2 days 35 43.2 3 —- 3-5 days 23 28.4 4 - > 5 days 14 17.3 Code enforcement and use of O — none 40 49.4 civil regulations 1 __ < 1 day 32 39.5 2 — 1-2 days 7 8.6 3 — 3-5 days 2 2.5 4 - > 5 days 0 O 112 Table 8 (cont’d). Using data to analyze O " none 35 43.2 neighborhood problems 1 ' < 1 day 33 40,7 2-1-2days 9 11_1 3 — 3-5 days 3 3.7 4 - > 5 days 1 1.2 Mediation 0 - none 53 65.4 1 - < 1 day 15 18.5 2- 1-2 days 11 13.6 3 — 3-5 days 2 2.5 4 - > 5 days 0 0 Organizing community groups 0 - none 60 74-1 l - < 1 day 17 21.0 2 - 1-2 days 2 2. 5 3 — 3-5 days 0 0 4 - > 5 days 2 2_ 5 During the POPN interviews, police supervisors were asked to indicate the amount of training they had received on five community policing topics in the past three years: (1) community policing concepts and principles, (2) code enforcement and use of civil regulations, (3) using data to analyze neighborhood problems, (4) mediation, and (5) organizing community groups. Supervisors could respond with the following options: “0 = none,” “1 = less than one day,” “2 = 1-2 days,” “3 = 3-5 days,” and “4 = more than 5 ’9 days. Descriptive statistics for these survey items are presented in Table 8. These items were summed to create an Ability Index.38 This the index ranges in value from 0- 38 Principal components factor analysis these items were measuring the same underlying construct (Eigenvalue: 2.15; item-to-item correlation coefficient range: .09 -.58). The index is correlated with the factor scale above .95. The index is used because it is more readily interpretable and statistically indistinguishable from the correlated factor scale. 113 20. The mean of the scale is 4.8 (SD. = 2.72), indicating that most supervisors averaged less than a day of training on community policing-related topics.39 The alpha coefficient for the scale is 0.65. Supervisors were also asked to provide self-assessments of their knowledge in the same topic areas outlined above (i.e., community policing concepts and principles, code enforcement and use of civil regulations, using data to analyze neighborhood problems, mediation, and organizing community groups). Response options included “0 = not very knowledgeable,” “1 = fairly knowledgeable,” and “2 = very knowledgeable.” Table 9 provides descriptive statistics for these items. A Perceived Competence Index was created by summing four40 questions regarding the level of knowledge each supervisor reporting having regarding community policing topics. This index ranges in value from 0-10 and has a mean of 3.58 (SD. = 1.71). This value indicates that, on average, supervisors perceived themselves as being “fairly knowledgeable” on topics related to community policing.41 The reliability coefficient for this index is 0.62. Table 9. Perceived Competence Survey Items Survey Item Value N % Community policing 0 = not very knowledgeable 6 7.4 concepts and principles 1 = fairly knowledgeable 46 56.8 2 = very knowledgeable 29 35.8 39 Dividing the mean (4.8) by 5 (the number of measures used to create the scale) provides a value of 0.96, which corresponds approximately with less than one day of training (or a value of l). 40 Principal components factor analysis revealed that only four of the five knowledge measures were measuring the same construct. Factor loadings indicated that knowledge of code enforcement should be dropped from the index. After doing so, the Eigenvalue was 1.89 and the item-to-item correlation coefficient range was .10-.44. Again, the index was correlated with the factor scale above .95 and is thus used in its place. 41 Dividing the mean (3.6) by 4 (the number of measures used to create the scale) results in a value of 0.90. This value corresponds approximately with fairly knowledgeable (or a value of l on the individual item scale). 114 Table 9 (cont’d). Code enforcement and use 0 = not very knowledgeable 30 37.0 ofcivil regulations 1 = fairly knowledgeable 48 59.3 2 = very knowledgeable 3 3.7 Using data to analyze 0 = not very knowledgeable 24 29.6 neighborhood problems 1 = fairly knowledgeable 46 56.8 2 = very knowledgeable 11 13.6 Mediation 0 = not very knowledgeable 18 22,2 1 = fairly knowledgeable 51 63.0 2 = very knowledgeable 12 14.8 Organizing community 0 = not very knowledgeable 45 55.6 groups _ 1 = fairly knowledgeable 29 35.8 2 = very knowledgeable 7 8.6 Role States The literatures in both organizational psychology and policing point to the importance of role states to the study of organizational commitment and organizational citizenship behavior, and thus perhaps the practice of community policing. In particular, role conflict and role overload are thought to decrease the likelihood of positive outcomes, such as the practice of community policing. There are several means by which role conflict or overload may be engendered. As several researchers have recounted, supervisors may experience role conflict when they held accountable for crime control while simultaneously expected to encourage problem-solving among subordinate officers (Kennedy, 1993; Sherman, 1975). Role overload may result when supervisors are expected to incorporate the role demands related to the practice of community policing. or the new type of management style expected of them under community policing, while 115 remaining responsibility for all of tasks expected of them under a traditional model of policing. Several survey questions were used to assess the extent to which supervisors might be experiencing role conflict or role overload. These items are presented in Table 10. Supervisors were asked the extent to which their Deputy Chiefs or District Management generally held them responsible for crime reduction in their area. 9’ 66 Supervisors were able to respond by stating “hardly any or none, some,” or “a lot.” Supervisors who felt greater pressure to reduce crime might experience more role conflict than supervisors who did not experience such demands. Being held accountable for crime reduction alone, however, would not translate into role conflict. Instead, this might be dependent on the job the department had done in providing the resources necessary to engage in community policing. Supervisors were asked three questions about the job their department had done: (1) providing the time necessary to engage in community policing, (2) providing the information necessary to engage in community policing, and (3) distributing the workload fairly between 91 l-responders and CPOs. Supervisors were offered the following response categories: “poor,” “fair,” “good,” or “excellent.” Few supervisors responded that their departments had done an “excellent” job in these areas; consequently, the latter two response categories (i.e., good and excellent) were collapsed into one. Table 10. Role Conflict Survey Items Survey Item Value N % Job IPD/SPPD has done distributing 0 — poor 37 45.7 the workload fairly between CPOs l — fair 26 34.6 and 911 responders 2 — good /excellent 16 19.8 116 Table 10 (cont’d). 0 - poor 31 38.3 Job IPD/SPPD has done providing 1 — fair 23 28.4 enough time for problem-solving 2 — good /excellent 27 33.3 Job IPD/SPPD has done in providing 0 — poor 15 18.5 the information necessary for 1 — fair 32 39.5 problem-solving 2 - good /excellent 34 42.0 Extent to which District Chief holds 0 ‘ “0‘ at a“ 8 9'9 supervisor accountable for crime 1 — some 53 65-4 reduction 2 - a lot 20 24.7 Influence over whether subordinates 0 -— otherwise 19 23.5 are allowed to go out-of-service for 1 _ a lot 62 76.5 problem-solving Finally, role conflict may also be dependent upon the extent to which supervisors are given the power or authority to manage their resources. This was measured by using a survey question that asked supervisors were asked how much influence they had over whether officers were allowed to go out of service for problem-solving: “hardly any or ’1 ‘6 none, some,” or “a lot.” These response categories were collapsed so that “ 0 = otherwise” and “1 = a lot.”42 Work Experiences Rewards. Conventional wisdom and a wealth of psychological research suggest that managers should reward the behavior they desire in workers (e.g., Kerr, 1975). Applied to the current context, we would expect police supervisors to engage in the behaviors that are rewarded by upper management. Supervisors were asked two 42 Only one supervisor responded that he/she had hardly any or no influence over allowing officers to go out of service in order to problem-solve. 117 questions during the POPN interview regarding the extent to which community policing behaviors were rewarded by department administrators. First, supervisors were asked about the likelihood that district management would recognize good performance. 9’ 6‘ Response options were “very unlikely, unlikely,” “likely,” and “very likely.” Second, supervisors were asked to evaluate the job their department had done rewarding officers who do a good job problem-solving. Responses were on a 4-point ordinal scale ranging from “poor” to “excellent.” These categories were later collapsed to a 3-point scale where “0 = poor,” “1 = fair,” and “2 = good/excellent.”43 The distribution of supervisors’ responses to these questions is depicted in Table 11. Table 11. Reward Survey Items Survey Item Value N % Likelihood that District O—very unlikely/unlikely 29 35.8 Management will recognize good 1 — likely 29 35.8 performance 2 — very likely 23 28.4 Job IPD/SPPD has done in O - poor/ fair 58 71.6 rewarding ofi‘icers who do a good 1 - good/excellent 23 28.4 job problem-solving Participation. Studies have consistently demonstrated the importance of participation in getting supervisory personnel to buy into the change process (Sadd & Grinc, 1994; Shanahan, 1978; Wilkinson & Rosenbaum, 1994). Supervisors were asked two questions that tap into their level of participation in the day-to-day operations of the police department. First, supervisors were asked how much influence they had over department policies about patrol operations: “none,” “hardly any,” or “some.” Few 43 Very few supervisors (N=3) claimed that their departments had done an “excellent” job rewarding officers for problem-solving. As such, these cases were collapsed into the “good” category. 118 supervisors (N=3) responded that they had no influence over patrol matters, so this category was collapsed with “hardly any.” Second, supervisors were asked to rate the importance of providing input on department policy. Supervisors could indicate that this was one of the most, least, or neither most important nor least important of their supervisory functions. This variable was dichotomized so that “l 2 most important” and “0 = otherwise.” Table 12. Participation Survey Items Survey Item Value N % Influence over patrol matters 0 — none/hardly any 51 63'0 1 30 37.0 — some Providing input on O - otherwise 73 90.1 department policy 1 - most Important 8 9,9 Opportunity. In a study that seeks to predict supervisor behavior (specifically, community policing activities), it is important to control for the opportunity to do so. In research examining the behavior of police officers during the course of the POPN project, researchers have employed a beat-level measure of socioeconomic distress (e.g., Parks et al., 1999)44 as an indicator of opportunity (DeJong et al., 2000). Another important indicator of opportunity is the supervisor’s shift. The opportunity to engage in community policing is arguably going to be greater during daylight shifts than evening or night shifts (when most businesses are open and people are out and about). Two survey items (each measuring the shifts of Indianapolis and St. Petersburg shifts) were re-coded to create a new shift variable with three categories: “0 = night shift,” “I = In these studies, the socioeconomic distress of a beat is measured by summing the percentage of unemployed residents in the work force, the percentage of residents living below 50% of the poverty level, and the percentage of single female-headed households. 119 aftemoon/evening shift,” and “3 = day Shifty,“ Descriptive information on this variable is presented in Table 13. A second means of assessing “opportunity” to engage in community policing was to examine the workload of the district and shift to which the supervisor is assigned. Some shifts receive more calls for service or are staffed by fewer officers, leading to less free or uncommitted time to devote to community policing. In this study, district and shift workload was measured by examining the average number of calls for service relative to the average number of officers assigned to a shift. Dispatch data and staffing information were obtained from each department. Using data on the calls for service over the observation periods, it was possible to determine the average number of calls responded to in each district and on each shift. It was also possible to measure the average number of officers that were working in each district and shift. The average number of calls for service was then divided by the average number of officers, providing an estimate of the average number of calls responded to by officers on any given shift. Table 13. Opportunity Measures Survey Item Value N % Supervisor Shift 0 - night 30 37-0 1 - evening 20 24.7 2 -— day 31 38.3 Mean: 2.98 District and Shift SD: 1. 40 workload Range: 1.32-6.05 45 While this variable might be most accurately deemed to be nominal in nature, I have ordered the shifts in an ordinal fashion based on the assumption that the day shift presents the greatest opportunity to engage in community policing and the night shift presents the least opportunity to engage in community policing. 120 Group/Leader Relations Group Cohesiveness. The organizational psychology literature suggests that an important group relations variable is the congruence of values between an individual and his/her co-workers (Adkins et al., 1996). When value congruence exists, there can be beneficial organizational effects, such as increased productivity and effectiveness. This assumes, however, that the value congruence is such that it supports organizational goals. As discussed in the last chapter, it is also possible that value congruence can exist among co-workers but to the detriment of the organization. Thus, any measure of value congruence must also take into account the nature (or direction) of the attitudes held by a group. The position of supervisors in the police hierarchy suggests that they may be influenced by the extent to which their values coalesce with three different groups: (1) their subordinate officers, (2) their co-workers (i.e., fellow supervisors), and (3) administrators. Given the three groups that may exert an influence on supervisor behavior, I created three value congruence measures, measuring “value congruence” as the degree to which a supervisor’s attitudinal support for community policing (as defined earlier in this Chapter) was consistent with that of his/her subordinate officers, fellow supervisors, and superior officers. To create a measure of value congruence between a supervisor and his/her subordinates, supervisors were matched with the officers working the same district and shift. Information on the officers’ districts and shifts were determined by using data from the Officer Survey. Next, an Attitudes toward Community Policing Index (described earlier in this chapter) was created for each officer. The scores of all the subordinate 121 officers working the same district and shift were then averaged, and the supervisor’s Attitude toward Community Policing Index score was then subtracted from this value. The resulting values could be negative (indicating that the supervisor’s Attitude Index was more supportive of community policing, zero (indicating value congruence between the supervisor and his/her subordinates), or positive (indicating that the supervisor was less supportive of community policing). These values were then collapsed to create a dichotomous variable where “1 = value congruence (no difference in Index scores of supervisor and officers)” and “0 = other (supervisor more or less supportive of community policing than subordinates)” Creating a value congruence variable for fellow supervisors followed the same general steps as those for subordinate officers. Supervisors were matched with other supervisors working the same district and shift. The Attitudes toward Community Policing Index scores of the other supervisors were averaged, and then the supervisor’s Index score was subtracted from this value. Again, the resulting values were collapsed to create a dichotomous variable assessing whether value congruence was present or not. Creating value congruence scores for supervisors and their administrators was different than the process outlined above. Unlike patrol officers and supervisors, administrators were not surveyed during the course of the POPN study. Instead, supervisors were asked for their opinions of their district management’s support for community policing. Supervisors were asked to provide assessments of their district management’s attitudes using the same set of questions used to create the Attitudes toward Community Policing Index. The value congruence score created for superior officers thus reflects a supervisor’s opinion, rather than “fact.” 122 Table 14. Group Cohesiveness Measures Variable Coding N % Value Congruence with 0 — Other 49 62.0 Administration 1 — Value congruence 30 38.0 Value Congruence with Fellow O — Other 69 87.3 Supervisors l — Value congruence 10 12.7 Value Congruence with Officers 0 — Other 37 46.8 1 -— Value congruence 42 53.2 Descriptive information on the three value congruence measures provided in Table 14 shows that supervisors appeared to experience the greatest value congruence with subordinate officers. This is not entirely surprising, given that these men and women are those with whom the supervisor spends the majority of his/her time. On the other hand, it was somewhat surprising that supervisors experienced the least amount of value congruence with fellow supervisors. Leader Supportiveness. The literature on organizational psychology suggests that leader supportiveness can play a key role in encouraging behavior that is in line with organizational goals (Eby et al., 1999; Steers, 1977). Supervisors were asked two questions during the course of the POPN interviews that were used leader supportiveness. First, supervisors were asked how often they ask for and receive 9’ 6‘ ,9 6‘ resources. Original response options included “never, seldom, sometimes,” “usually,” and “always.” The distribution of responses at either end of the scale, however, necessitated that the variable be recoded so that “0 = never/seldom,” “1 = 123 sometimes,” and “2 = usually/always.”46 Supervisors were also asked how often their decisions about patrol operations are supported by the higher-ups in the department. ’9 ‘6 H 6‘ Supervisors could respond by stating “seldom, sometimes, usually,” or “always.” The categories of this variable were collapsed due to the low number of supervisors who reported that their decisions were “always” supported. Descriptive statistics for these measures are presented in Table 15. Table 15. Leader Supportiveness Survey Items Survey Item Value N % How often supervisor asks for 0 — never/rarely 18 22.5 resources and receives thema 1 _ sometimes 36 44.4 2 - usually/always 27 33.3 0 — seldom 19 23.5 How often decisions about _ patrol operations are supported 1 ‘ sometimes 62 76-5 by the administration 3 Valid percentages reported. Summary This chapter has provided a detailed description of data and methods that were employed in the current study. This chapter began by describing the Project on Policing Neighborhoods (POPN) study, including descriptions of the two research sites and the different types of data that were collected. Next, I described the sampling procedures used in this research. With this background in place, I provided a delineation of the construction of the dependent and independent variables. Supervisors’ support for community policing was defined in behavioral terms and assessed using observational data. In particular, supervisors’ support for community policing was measured as the 46 Only one respondent claimed that s/he never asked for or received resources and two claimed that they always asked for and received resources. 124 average percent of a shift that they spent engaged in community policing behaviors. Survey items and official data were used to construct variables that the theoretical framework used in this study suggests will significantly the likelihood that supervisors will engage in community policing. In the following chapter, I provide the results of the statistical analyses performed on these data. 125 CHAPTER SIX: STATISTICAL ANALYSES AND RESULTS This chapter presents the results of statistical analyses conducted on the data as measured in the preceding chapter. This chapter begins with a summary of the collinearity diagnostics performed on the data to ensure that multicollinearity was not a problem among the independent variables. Results of bivariate analyses (ANOVA) are then presented, followed by the results produced by multivariate regression analyses. In each of these set of analyses, the results are discussed according to each of the five models that were tested. To review, these models were comprised of the attitudes toward community policing index and the following variables: (1) demographic characteristics, (2) personal characteristics, (3) role conflict variables, (4) work experience variables, and (5) group/leader relations variables. Collinearity Diagnostics Prior to conducting bivariate and regression analyses on the data, correlational analyses were performed and tolerance statistics and variance inflation factors (VIFs) were computed to ensure that there were no problems with multicollinearity among the independent variables in each model. One can conclude that multicollinearity is not a problem when the inter-correlation values among independent variables do not exceed .80 (Menard, 1995, p. 66). Examining the correlations among independent variables alone, however, does not ensure freedom from problems with multicollinearity (Berry and Feldman, 1985, p.43). 126 Additional and more conservative tests involve the calculation of tolerance statistics and variance inflation factors (VIFs).47 Tolerance statistics represent the percentage of variance in a particular predictor variable that is not explained by the other independent variables. The variance inflation factor is reciprocal of the tolerance statistic, and provides an indication of how much multicollinearity has increased the standard error of the regression coefficient. The tolerance statistic can range in value from 0 to 1, with lower values (<.20) indicating a problem with multicollinearity (Menard, 1995, p.66). Bivariate analyses and other collinearity diagnostics were performed for the first model, comprised of supervisor demographic characteristics and their attitudes toward community policing. Table 16 presents the correlation matrix for these independent variables, as well as the tolerance statistics. A problem with collinearity among the variables of age and tenure is indicated by both the correlation of these items (.93) and their tolerance statistics (.14 and .13, respectively). For regression analyses, it was necessary to drop one of the variables to correct for this problem. I elected to drop tenure, as the literature more often discusses the importance of age in determining work- related attitudes and openness to change. The second model of variables consists of indicators of perceived knowledge, perceived ability, and attitudes toward community policing. The correlation matrix and tolerance statistics for these variables presented in Table 17 do not indicate problems 47 Tolerance statistics and Variance Inflation Factors (VIFs) are obtained by conducting a linear regression analysis in SPSS. While linear regression is not appropriate given the nature of the dependent variable in this study, Menard (1995) notes that . . the functional form of the model for the dependent variable is irrelevant to the estimation of collinearity (p. 66).” In other words, the concern with multicollinearity is among the independent variables, not the relationship of the independent variables with the dependent variable. 127 with multicollinearity. The greatest inter-correlation value between variables (between perceived ability and perceived competence) is within acceptable limits at .43, and the tolerance statistics are all above the .20 cutoff suggested by Menard (1995). Table 16. Zero-Order Correlation Coefficients and Tolerance Statistics: Demographic Variables Variable 1 2 3 4 5 6 7 Tolerance Statistic Female 1 1.0 .84 Minority 2 .16 1.0 .92 Age 3 -.23* -.05 1.0 .14 Tenure 4 -.27* -.07 .93** 1.0 .13 Education 5 -.07 -.17 -.32** -.33** 1.0 .77 Promotion Ambition 6 -.22* -.01 -.04 -.07 .30** 1.0 .84 Attitudes toward CP 7 -.06 .13 .15 .06 -.03 .15 1.0 .91 * p (.05 **p < .01 *** p < .00] 128 Table 17. Zero-Order Correlation Coefficients and Tolerance Statistics: Personal Characteristics Variable 1 2 3 Tolerance Statistic Training Index 1 1.0 .80 Knowledge Index 2 .43** 1.0 .80 Attitudes toward Community 3 .07 -.09 1.0 .98 Policing * p <.05 **p < .01 *** p < .001 Table 18. Zero-Order Correlation Coefficients and Tolerance Statistics: Role Conflict Variables Variable 1 2 3 4 5 6 Tolerance Statistic Job IPD/SPPD has done 1 1.0 .55 distributing the workload fairly between CPOs and 911 responders Job IPD/SPPD has done 2 .63** 1.0 .53 providing enough time for problem-solving Job IPD/SPPD has done in 3 .30* .31** 1.0 .78 providing the information necessary for problem-solving Extent to which District Chief 4 26* ‘14 '09 1'0 .93 holds supervisor accountable for crime reduction Influence over whether 5 .16 .07 .14 .09 1.0 .94 subordinates are allowed to go out-of-service for problem- solving Attitudes toward Community 6 .05 .23* -.22* -.04 .11 1.0 .82 Policing *p < .05 **p < .01 ***p < .00] 129 The third model of independent variables is comprised of indicators of role conflict and attitudes toward community policing (see Table 18). Again, inter-correlation values of less than .80 and tolerance statistics greater than .20 indicated that multicollinearity was not a problem among the independent variables. Table 19. Zero-Order Correlation Coefficients and Tolerance Statistics: Work Experience Variables Variable 1 2 3 4 5 6 7 Tolerance Statistic Likelihood that District 1 1.0 .71 Management will recognize good performance Job IPD/SPPD has done 2 .13 1.0 .88 in rewarding officers who do a good job problem- solving Influence over department 3 .04 .08 1.0 .96 policies about patrol operations Providing input on department policy DayShift 5 .34** .02 -.1o -.15 1.0 .83 District and Shift 6 -40** .20 .02 -.02 -.20 1.0 .70 Workload Attitudes toward 7 29** -.Ol .04 .12 .08 -38** 1.0 .82 Community Policing *p < .05 **p < .0] *** p < .001 The fourth model under examination consisted of work experience variables. Table 19 presents the zero-order correlations among the work experience variables in this study. The greatest correlation, between the likelihood that district management would recognize good performance and the workload on a supervisor’s district and shift (.40), 130 was within acceptable limits. The highest tolerance statistic (district and shift workload; .70) was also far greater than the .20 level that would cause concern about collinearity. Table 20. Zero-Order Correlation Coefficients and Tolerance Statistics: Group/Leader Relations Variables Variable 1 2 3 4 5 6 Tolerance Statistic Value congruence 1 1.0 .95 with peers Value congruence 2 -.14 1.0 .95 with Dist. Mgt. Value congruence 3 -.10 .16 1.0 .39 with officers How often 4 -.01 .01 -.01 1.0 .85 supervisor asks for and receives resources How often decisions are supported by the administration Attitudes toward 6 -.03 .13 .78 ** -.03 -.07 1.0 .40 Community Policing *p < .05 “p < .01 ***p < .001 The final model was comprised of value congruence and leader supportiveness variables. Table 20 presents the zero-order correlations among these variables. The greatest correlation (between value congruence with subordinates officers and attitudes toward community policing) was .78, falling just shy of the .80 cutoff suggested by Menard (1995). The highest tolerance statistic for both variables, however, were greater than the .20 level. The more conservative nature of this test indicates that multicollinearity is not a problem among these variables. 131 Bivariate Analyses Table 21 presents the correlations between the dependent variable in this study (time spent engaged in community policing) and the independent variables.48 As depicted in Table 21, supervisors’ attitudes toward community policing were not predictive of the time they spent engaged in community policing behaviors. At least at the bivariate level, then, there is reason to question the (direct) link between attitudes and behavior. None of the demographic characteristics were significantly correlated with the dependent variable. This is not entirely surprising, as the literature suggests that these variables would only be weakly related to behavior, if at all. Both personal characteristic variables were significantly related to time spent engaged in community policing. Supervisors who reported having received more training on community policing topics (i.e., community policing concepts and principles, code enforcement and use of civil regulations, using data to analyze neighborhood problems, mediation, and organizing community groups) or who rated themselves as more knowledgeable in these areas were significantly more likely to engage in community policing. One role conflict variable, the extent to which a supervisor perceives District Management as holding them accountable for crime reduction, was positively correlated with the outcome variable. When supervisors felt that they were held accountable for crime reduction, they were more likely to engage in community policing over the course of their shift. Finally, value congruence with District Management was significantly and positively related to community policing; supervisors who perceived that their attitudes toward community 48 While this section focuses on the relationships between the independent and dependent variables, Appendix A presents a correlation matrix depicting the relationship between all variables in this study. 132 policing were closely aligned with those of management were more likely to spend time engaged in community policing during the course of a shift. All told, the bivariate analyses revealed just four significant relationships between the independent and dependent variables. Simple bivariate analyses, however, may mask the existence of relationships that might exist once the effects of other variables are partialled out. In the next section, I provide the results of a series of multiple regression analyses performed on the data. Table 21. Bivariate Correlates of Time Spent Engaged in Community Policing % Time Spent Engaged in Community Policing £ 2 Attitudes toward Community Policing .066 .566 Demographic Characteristics Female .005 .969 Minority -.036 .751 Age -. 100 .380 Education -.094 .412 Promotion Ambition -.096 .401 Personal Characteristics Training Index .270 .016 Knowledge Index .305 .006 Role Conflict Variables Job IPD/SPPD has done distributing the workload fairly .087 .444 between CPOs and 911 responders Job IPD/SPPD has done providing enough time for -.108 .344 problem-solving Job IPD/SPPD has done in providing the information —.023 .841 necessary for problem—solving 133 Table 21 (cont’d). Extent to which District Chief holds supervisor accountable .224 .047 for crime reduction Influence over whether subordinates are allowed to go out .135 .234 of service for problem-solving Work Experience Variables Likelihood that District Management will recognize good -.051 .654 performance Job IPD/SPPD has done in rewarding officers who do a .124 .275 good job problem-solving Influence over department policies about patrol operations -. 148 .193 Providing input on department policy -.087 .445 Day shift -.081 .477 District and shift workload .218 .054 Group/Leader Relations Variables Value congruence with officers .066 .560 Value congruence with fellow supervisors .023 .844 Value congruence with District Management .304 .006 How often supervisors asks for and receives resources .019 .870 How often decisions are supported by the administration -.104 .356 Multiple Regression Analyses Negative Binomial Regression Results Figure 3 (see Chapter 5) provided the distribution of the dependent variable in this study. As the histogram demonstrated, the distribution of time supervisors spent engaged in community policing made the use of a linear regression model such as Ordinary Least Squares (OLS) inappropriate. Instead, the distribution of the variable suggested that either a Poisson or Negative Binomial regression model would provide a better fit to the data. There are several ways of determining which regression model is a better fit to one’s data. The defining characteristic of the Poisson model is that “the conditional mean of the outcome is equal to the conditional variance” (Long, 1997, p.218). This limitation of the Poisson model makes its use inappropriate for most dependent variables in 134 criminal justice, as was the case in the current study. The distribution of time spent engaged in community policing did not meet the assumptions of a Poisson distribution, as the variance for this variable (82.3) greatly exceeded its mean (11.7). The conclusion that the Poisson model was inappropriate for use in the current study was also made after examining the alpha values produced for Negative Binomial regression analyses. Alpha values are a parameter of overdispersion. When alpha equals zero, there is equidispersion; in other words, the central assumption of the Poisson model is not violated and therefore Poisson is the more appropriate model. On the other hand, when the alpha values are statistically significant, it is an indication that the dependent variable is overdispersed, meaning that the Poisson assumption is violated and making Negative Binomial more appropriate for the data (Long, 1997). The alpha values for all five regression models in the current study were significant. Based on these criteria (i.e., the variance exceeded the mean and an overdispersed dependent variable), the Negative Binomial model was deemed to be the appropriate regression model to use. The results of these analyses are presented in Tables 22-26. Table 22. Negative Binomial Regression Results: Demographic Model Variable 2 fl 2 Constant 2.86 1.22 .019 Gender -.05 .32 .884 Age -.01 .21 .745 Race -.08 .36 .818 Education -.14 .31 .645 Promotion Ambition .01 .12 .967 Attitudes toward CP .41 . 15 .724 Log-Likelihood - 279.83 Chi-Squared (df) 288.92(1) .000 Alpha .71 . 15 .000 135 Table 23. Negative Binomial Regression Results: Personal Characteristics Model Variable _B_ fl _Ij Constant 1.83 .38 .000 Training Index .05 .06 .362 Knowledge Index .08 .08 .302 Attitudes toward Community .03 .10 ' .784 Policing “ Log-Likelihood -277.54 Chi-Squared (df) 255.85(1) .000 Alpha .65 .137 .000 Table 24. Negative Binomial Results: Role Conflict Model Variable g _S_._I.L E Constant 2.03 .45 .000 Job IPD/SPPD has done distributing the workload fairly between CPOs and 911 .21 .23 .360 responders Job IPD/SPPD has done providing enough -.23 .21 .276 time for problem-solving Job IPD/SPPD has done in providing the .01 .19 .979 information necessary for problem-solving Extent to which District Chief holds .21 .24 .382 supervisor accountable for crime reduction Influence over whether subordinates are allowed to go out-of-service for problem- .12 .27 .654 solving Attitudes toward Community Policing .58 .11 .606 Log-likelihood -277.73 Chi-Squared (df) 252.43(l) .000 Alpha .66 .13 .000 136 Table 25. Negative Binomial Regression Results: Work Experiences Model Variable B S.E. P Constant 1.91 .52 .000 Likelihood that District Management will recognize .11 .20 .579 good performance Job IPD/SPPD has done in rewarding officers who .06 .32 .838 do a good job problem-solving Influence over department policies about patrol -.16 .27 .550 operations Providing input on department policy -.25 .41 .552 District and shift workload .12 .13 .355 Day shift -.08 .27 .756 Attitudes toward Community Policing .08 .11 .445 Log-Likelihood -278.42 Chi-Squared (df) 264.85(1) .000 Alpha .67 .13 .000 Table 26. Negative Binomial Regression Results: Group/Leader Relations Model Variable J S.E. P Constant 2.44 .34 .000 Value Congruence with Fellow Supervisors -.33 .36 .346 Value Congruence with District Management .26 .30 .383 Value Congruence with Subordinates Officers .01 .36 .969 How often supervisor asks for resources and -.03 .18 .861 receives them How often decisions about patrol operations are -.08 .36 .834 supported by the administration Attitudes toward Community Policing .02 .14 .896 Log-Likelihood -397.80 Chi-Squared (df) 3137(6) .000 Alpha .67 .13 .000 Tables 22-26 present the results of the regression analyses performed on the demographic characteristics, personal characteristics, role conflict, work experiences, and group/leader relations models. For all of these models, the Negative Binomial regression model proved to be a good fit to the data, as indicated by significant Chi-Square values. The Chi—Square value indicates whether the hypothesis that all regression coefficients 137 except the intercept have a value of zero is supported by the data (Aldrich & Nelson, 1984, p.55). Despite significant models, none of the independent variables significantly predicted the amount of time spent engaged in community policing. In other words, none of the independent variables examined improved my ability to predict the amount of time supervisors spent engaged in community policing above and beyond that which would have been obtained by random chance (i.e., relying solely on the mean). Post Hoc Analyses A series of post hoc analyses were performed in order to determine if alternative coding or analytic strategies would produce difference results than those reported in Tables 22-26. The first set of analyses entailed the use of a different measurement of the dependent variable. In these analyses, rather than using the average percentage of time supervisors spent engaged in community policing per shift, the dependent variable was measured as the total amount of time supervisors spent engaged in community policing. In order to control for the different lengths of time that supervisors were observed, another independent variable (total time observed) was then added to each of the five models. Results of these analyses (not shown here) produced the same results (i.e., significant models, but no significant variables) as the original Negative Binomial regression analyses. The second set of analyses I performed also entailed an alternative coding of the dependent variable. Assuming that the reason that the regression analyses were failing to yield any significant variables was the distribution of my dependent variable, I chose to collapse my dependent variable into three categories: low efiort (0-6% per shift spent engaged in community policing), medium eflon (6-17% per shift spent engaged in 138 community policing), and high effort (more than 17% of shift engaged in community policing)” Using this coding of the dependent variable, five Ordered Probit regression models were run. Of all analyses conducted, only the “Personal Characteristics” model produced a significant finding. Here, training was significantly related to supervisors’ level of effort expended on community policing. Officers who reported having a greater amount of training on community policing topics (e.g., mediation, organizing community groups, and crime data analysis) were significantly (p < .05) more likely to engage in community policing than were officers who reported receiving less training in these areas. With this one exception, results (not shown here) were largely concordant with those produced using Negative Binomial regression equations. The failure to find any significant findings (except for training in the Ordered Probit analyses) led me to question the reliability of the dependent variable. A reliable measure is one “where the same data would have been collected each time in repeated observations of the same phenomenon” (Maxfield & Babbie, 1998, p. G4). According to additional analyses, the reliability of the dependent variable in this study is problematic for at least two reasons. First, as shown in Figure 5, 54% (N=44) of supervisors were observed only once during the course of the POPN project. For these supervisors, the reliability of the dependent variable is unknown, as repeated observations were not made. It is not possible to say with any certainty whether the percentage of time they spent engaging in 49 This decision was based on the distribution of cases among these three categories; approximately 1/3 of the cases fell into each category. 139 community policing was typical or an aberration, the result of some special circumstance(s) that enabled or discouraged them from engaging in this behavior. Figure 5. Number of Observational Sessions (Rides) Per Supervisor 2+ Rides g 46% , 1 Ride (N=37) 54% (N=44) A second reason that the reliability of the dependent variable is questionable is demonstrated by a closer examination of the supervisors who were observed multiple times (see Appendix B). Among this group of supervisors, there is great variation in the percentage of time spent engaging in community policing by observation session. In the most glaring example, one supervisor (#65) spent only 3% of his time engaged in community policing on one ride, and over half (54%) of his time on this behavior during another ride. For most other supervisors observed multiple times, there was great variability from one observational session to another. A final set of analyses performed on the data confirmed that the dependent variable was a very unstable and unreliable estimate of supervisors’ support for community policing, and ultimately to blame for the lack of findings reported earlier in 140 this chapter. In this set of analyses, the Attitudes toward Community Policing Index50 5 ' in other was treated as the dependent variable in four logistic regression equations; words, attitudes toward community policing were regressed against demographic, personal characteristics, role conflict, work experience, and group/leader relations variables. Of the five models tested, only three models were yielded significant predictors of supervisors’ attitudes toward community policing.52 The results of these models are presented in Tables 27-28. As depicted in Table 27, two role conflict variables significantly predicted supervisors’ attitudes toward community policing. Supervisors who believed that their department had done a good job in providing enough time for problem-solving were significantly more likely to have favorable attitudes toward community policing. In fact, supervisors who believed their department had done a good job in this respect were almost 3 times as likely to exhibit attitudinal support for community policing than were supervisors who believed their department had not done a good job providing adequate time to engage in problem-solving. Surprisingly, results also show that supervisors who believed that their department had done a good job providing the information necessary 50 In the original coding scheme used for the Attitudes toward Community Policing Index, very few supervisors (N=8) indicated that both “reducing the number of repeat calls for service to the same address" and “getting the public involved in improving the neighborhood” were the most important functions of the police (a value of “2” according to the original coding scheme). For these analyses, the original values of “l” and “2” were collapsed, so that a value of “0” indicated that the supervisor rated neither “reducing the number of repeat calls for service to the same address” nor “getting the public involved in improving the neighborhood” were the most important functions of the police and a value of “1” indicated that the supervisor selected at least one of these items as the most important function of the police. 51 An analysis including group/leader relations variables was not conducted, due to multicollinearity between the dependent variable (Attitudes toward Community Policing Index) and the value congruence variables. The problem with multicollinearity is not entirely surprising, given that the Index was used to construct the value congruence variables. 52 None of the variables in the demographic or personal characteristics models significantly predicted supervisors’ attitudinal support of community policing. 141 for problem-solving were significantly less likely to possess favorable attitudes toward community policing. Table 27. Logistic Regression Results: Attitudes toward Community Policing Regressed on Role Conflict Variables Variable Coefficient Std. Error Log- Odds Ratio Constant -.99 .80 Job IPD/SPPD has done distributing the .10 .47 1.11 workload fairly between CPOs and 911 responders Job IPD/SPPD has done providing enough 1.05 ** .44 2.85 time for problem-solving Job IPD/SPPD has done in providing the -1.17 ** .42 .31 information necessary for problem-solving Extent to which District Chief holds '10 '45 1'1 1 supervisor accountable for crime reduction Influence over whether subordinates are 1.27 .66 3.57 allowed to go out-of—service for problem- solving Log-likelihood -111.28 Chi-Squared (df) 19.37 (5) ** * p < .05 ** p < .01 *** p < .001 Table 28 shows that a supervisor’s workload was predictive of his/her attitudes toward community policing. The greater the workload (as measured by the average number of calls responded to per shift), the less likely it was that the supervisor would express favorable attitudes toward community policing. While significant, the impact of this variable was not necessarily that great; the log-odds ratio for this variable indicates that busier workloads only reduced the likelihood of favorable attitudes toward community policing by 54%. I42 Table 28. Logistic Regression Results: Attitudes toward Community Policing Regressed on Work Experience Variables Variable Coefficient Std. Error Log- Odds Ratio Constant Likelihood that District Management will .32 .35 1.38 recognize good performance Job IPD/SPPD has done in rewarding .71 .64 2.03 officers who do a good job problem-solving Influence over department policies about .35 .53 1.43 patrol operations Providing input on department policy -.05 .83 .95 District and shift workload -.78 ** .26 .46 Day shift -.48 .58 .62 Log-Likelihood -111.28 Chi-flared (df) 18.12 (6) ** * p < .05 ** p < .01 *** p < .001 Summary This chapter presented results of bivariate and multivariate analyses. Bivariate analyses revealed only 4 out of 24 independent variables were significantly correlated with the average percent of time a supervisor spent engaging in community policing during a shift. With one exception (i.e., training) the 15 different regression equations using alternative codings of the dependent variable failed to produce any significant results. The lack of findings yielded by these analyses led me to perform several post hoc analyses on the data in order to determine the source(s) of the underlying problem. An examination of the variation in time spent engaging in community policing by observational session and analyses where attitudes toward community policing were regressed against demographic, personal, role conflict, work experience, and group/leader 143 relations variables demonstrated that the dependent variable used in this study was a highly unstable and unreliable estimate of supervisor’s support for community policing. In the following chapter, I try to make sense of these findings, discussing various methodological issues raised by the current analyses. I conclude the dissertation with a discussion of the implications of my findings, from both a theoretical and practical standpoint. 144 CHAPTER SEVEN: DISCUSSION AND CONCLUSIONS The purpose of this study was two-fold: (l) to gauge the extent to which supervisors “supported” community policing (measured as the average amount of time a supervisor engaged in community policing per shift) and (2) to determine the factors that significantly influenced the extent to which supervisors would engage in community policing. Traditionally, the final chapter of a dissertation would provide a summarization of the results produced in light of these two research questions, followed by a discussion of the policy implications that followed from the results, methodological issues encountered in the research, and suggestions for future research. The results presented in Chapter Six make such an outline inappropriate. Rather than follow the outline traditionally used, this chapter is organized into three major sections: (1) methodological issues (i.e., what the results cannot tell us and why), (2) substantive issues (i.e., what the results can tell us), and (3) directions for future research in this area. Methodological Issues: What the Results Cannot Tell Us and Why Measurement Issues This dissertation was designed to provide the answers to two research questions: (1) to what extent do supervisors support community policing as demonstrated in their behavior, and (2) what factors that significantly influenced the extent to which supervisors would engage in community policing. As highlighted in Chapter Six, one of the major obstacles in providing an answer to these questions was the measurement of the dependent variable in this study. This was not the only measurement issue encountered. however. One of the main drawbacks of secondary data analysis is that the researcher is, 145 by necessity, limited to data as they were originally collected (Maxfield & Babbie, 1998). In particular, the measurement of the two main constructs in this study—time spent engaged in community policing and attitudes toward community policing—were less than ideal. In the following paragraphs, I discuss some of the measurement issues related to these variables. As a means of measuring support for community policing in behavioral terms, the dependent variable was a calculation of the amount of time supervisors spent engaged in community policing. Ultimately, this measurement was problematic due to the great variation in time spent engaged in community policing by observational session. More than half (54%; N = 44) of the supervisors in this study were observed only once. For these supervisors, it was not possible to determine whether the amount of time spent engaged in community policing was typical or atypical of their normal behavior. An examination of the variation among the supervisors in this study that were observed on more than one occasion (46%; N = 37), however, suggests that we should not place much stock in the reliability of this variable. As demonstrated in Appendix B, supervisors did not tend to engage in community policing in a consistent fashion. The inconsistent manner in which supervisors engaged in community policing suggests that this activity is highly context-dependent. Although the data did not allow me to explore this matter in further depth, knowledge of the departments in this study suggests that there was at least one key organizational factor that exerted a significant influence on when supervisors would engage in community policing. In St. Petersburg, supervisors had “temporal” and “flex” duties. On days when supervisors had “temporal” duties, they were responsible for holding roll call, monitoring the behavior of officers on 146 the street, and making themselves available to officers in the event this was necessary. “Flex” days were specifically reserved for administrative duties, training, and assisting CPOs under their command with community policing projects. For supervisors at this site, it is quite likely that whether the supervisor engaged in community policing when he/she was observed was a matter of whether he/she was on a “temporal” or “flex” day. A second measurement issue related to the dependent variable in this study is the selection of the criteria that were used to identify behaviors that were considered to represent community policing. Human behavior can be complex, subtle, and nuanced. For these reasons, it is often not easily amenable to quantification. Recall that supervisor time was classified as “community policing” when it met four criteria: the encounter or activity was (1) part of a long-term plan or project to deal with a problem, (2) the police tried to determine the nature, extent, or causes of the problem during the encounter or activity, (3) the police were trying to prevent the occurrence or recurrence of a problem, or (4) the encounter or activity involved communicating with representatives of citizen organizations or representatives of other service-providing organizations. While these criteria have been used by previous researchers (DeJong, et al., 2000), one could argue that this results in a narrow and rather artificial measurement of community policing. Not only is this a rather artificial measure, but it may also not necessarily have been the best measure to use for supervisors. DeJong et al. (2000) used these criteria to provide an estimate of the amount of time that police ofi‘icers spent engaged in problem- solving activities. The problem in applying this operationalization to the study of supervisors is that many behaviors that might be considered supportive of community policing may not have been captured using this scheme. As outlined in Chapter One, 147 community policing means an entirely new role for supervisors. Rather than simply monitoring or controlling subordinate officers, supervisors are expected to mentor, coach, and act as facilitators (e.g., Goldstein, 1990; Skogan & Hartnett, 1997). Insofar as subordinate officers are concerned, the coding scheme used in this research only captures the time that a supervisor was present when a subordinate officer was actually engaging in community policing. It was not able to account for behaviors such as “coaching” or “mentoring” that would constitute engaging in community policing for supervisors. Another drawback of the criteria used to provide a measure of “community policing” is that the resulting definition of community policing that did not necessarily match the reality of what community policing meant in the two study departments. Community policing remains an amorphous term, with very little consensus regarding what does (and does not) comprise a community policing program. As Cordner (1997) stated, “community policing remains many things to many people” (p.452). While using a uniform definition of community policing was necessary, it also ignored the reality that the two study sites (and even districts within study sites) were operating under very different definitions of community policing. The organization and practice of community policing in St. Petersburg was uniform across the three districts and designed to facilitate community building and a problem-solving approach to crime and disorder (Parks et al., 1999). Most likely, the criteria used to define community policing in this study would have captured much of the community policing behavior engaged in by supervisors in St. Petersburg. In Indianapolis, however, the definitions of community policing differed considerably by district. For instance, community policing in the West District of the Indianapolis Police 148 Department heavily emphasized aggressive law enforcement tactics. It is less likely that the coding scheme would have captured the time supervisors in the West District of the Indianapolis Police Department spent engaged in aggressive law enforcement (e.g., running gun/drug interdiction), unless the supervisor made it clear to the observer that this behavior was part of a long-term plan or project to deal with the problem. As the preceding discussion illustrates, the dependent variable used in this research was problematic in a number of respects. The other key construct—attitudes toward community policing—might also be subject to criticism. Attitudes toward community policing was measured as an index comprised of two survey items, capturing the importance police supervisors placed on different police goals (i.e., reducing the number of repeat calls for service to the same address and getting the public involved in improving the neighborhood). These items were two of a list of seven possible goals. From this list, supervisors were required to pick two items that were the most important and two that were the least important. The remaining items were left unchecked, indicating that they were neither the most nor least important. Due to this coding scheme, it was only possible to use two items to measure attitudes toward community policing. It is certainly questionable whether supervisors’ responses to two survey items adequately tap into a complex construct such as attitudes. Moreover, it is questionable whether these particular survey items provide an adequate representation of the principles or core meaning of community policing. Rather than representing a supervisor’s attitudes toward community policing, this measure might more accurately be viewed as a supervisor’s orientation toward the police role. Police continue to strongly identify with the image of their role as that of “crime-fighter” (e.g., 149 Manning, 1997; Worden, 1995). Not only do police officers, but so too does the public. demanding accountability for crime reduction. Under these circumstances, it is questionable whether supervisors would have ranked reducing the number of repeat calls for service to the same address or getting the public involved in improving the neighborhood as the most important goal of the police. What the Analyses Do Tell Us: Substantive Matters The Practice of Community Policing by Police Supervisors While the previous section acknowledged the problematic nature of the definition of time spent engaged in community policing, the results nevertheless provide us with some indication of what supervisors do on an average shift, and a rough estimate of the extent to which supervisors engage in community policing. Figure 6 provides a breakdown of the time the supervisors in this study spent on different tasks. On average, supervisors spent approximately 12% of a shift engaged in community policing behaviors. At first glance, this may not appear to be a substantial amount of time. Compared to the way supervisors spend the rest of their shift, however, this should not be considered a particularly low level of effort. Perhaps most striking is the time spent engaged in community policing relative to time spent engaged in other tasks—such as administrative duties (14%), general patrol (28%), and en route (11%). The time spent engaged in general patrol and en route may be viewed as a significant part of a supervisor’s duties. It is in these ways that most supervisors “keep tabs” on their subordinates. Most often, supervisors engage in general patrol in order to be available to their subordinates. They spend this time monitoring the radio and the activities of their officers. Most, but not all, of supervisors’ en route time is spent going 150 to scenes at which they have been requested by subordinate officers or which they are required to attend by policy. Administrative duties are another significant portion of middle managers’ duties. Combined, these three supervisory duties account for over half (53%)of a supervisor’s time. Figure 6. Percentage of Supervisors’ Shift by Task Encounters w/ F’trblic 30.4 Meetings w/ public 0-2 Personal 9.8 Administrative 13-7 Information Gathering Ptoblem-directed En Route * General Patrol II 10.6 28 10 15 20 25 30 35 Taking this as a rough estimate of a supervisor’s committed time, supervisors are left with less than ‘/2 a shift to spend in a discretionary manner. This means that supervisors, on average, are spending one-fifth of their discretionary time engaged in community policing (see Figure 7). In this light, the extent to which supervisors are engaging in policing appears rather substantial, particularly when one considers that the practice of community policing is largely a discretionary activity (particularly on the part of police supervisors). At best, it is often considered an “add-on” to the already numerous responsibilities and roles expected of the police. This is particularly true of 151 supervisors, whose main responsibility, despite the shift to a community-oriented model of policing, remains monitoring and controlling subordinate officers’ behavior. Figure 7. Percentage of Discretionary Time Spent Engaged in Community Policing a Community Policingl 1. Other A Attitudes Toward Community Policing Policing scholars have long advocated that attitudinal change is the first step to ensuring the successful implementation of community policing (Goldstein, 1990; Skogan & Hartnett, 1997; Lurigio & Rosenbaum, 1994; Lurigio & Skogan, 2000). As Lurigio and Rosenbaum (1994) suggested, failing to win the hearts and minds of police personnel “ . is to risk program failure due to apathy, frustration, resentment, perceived inequality, fear of change, and other factors that mitigate against the successful implementation of community policing” (p.147). According to this argument, attitudinal change is viewed as a necessary precursor to behavioral change, and only the first of many steps necessary to achieve long-lasting organizational change. While this has long stood as the conventional wisdom on the matter, I argued in the introductory chapters of this dissertation that research examining the link between attitudes and behavior suggested that there was little reason to believe that attitudes would translate into behavior. If this were true, then many police administrators might be being misled about the best way to go about making the shift from a crime-control model of policing to one that is community-oriented in nature. Consequently, I made the 152 argument that previous researchers had inappropriately operationalized support for community policing as an attitudinal variable. Instead, I made the case that support is much more about action than belief, and that ultimately it wasn’t important whether supervisors could “talk the talk” of community policing, but whether they would “walk the talk” of community policing. The theoretical framework employed in this study also reflected this conviction; the framework allowed me to situate this study in the larger context of research on the link between attitudes and behavior, where attitudes were expected to play a minimal role in influencing behavior. Instead, other factors (demographic, personal characteristics, role states, work experiences, and group/leader relation variables) were expected to play a more direct and meaningful role in predicting the extent to which supervisors would engage in community policing. Several analyses presented in Chapter Six of this dissertation provided reason to question the reliability of the dependent measure in this research. As a result, this dissertation is not able to shed light on of the questions central to this dissertation research, particularly whether attitudes toward community policing predict the extent to which supervisors will engage in community policing. Let’s suppose, however, that policing scholars are right, and that the best means of ensuring the successful implementation of community policing is by achieving buy-in among the officers and supervisors who are responsible for carrying out this mode of policing. Until future research is able to tell us whether attitudes toward community policing do, in fact, translate into on-the-job behavior, it is worthwhile to know what factors significantly impact on attitudes, and the analyses presented in Chapter Six shed 153 some light on this matter. Table 29 summarizes the results of bivariate and multivariate analyses using attitudes toward community policing as a dependent variable. As this table shows, five variables were significantly correlated with supervisors’ attitudes toward community policing. Most of these variables were positively correlated with attitudes. When supervisors felt that they had influence over whether their officers were allowed to go out of service for problem-solving, that District Management did a good job recognizing good performance, and when their attitudes were similar to those of their fellow supervisors and District Management, they were significantly more likely to hold favorable attitudes toward community policing. The relationship between workload and attitudes was an inverse one; supervisors assigned to shifts with fewer calls for service were more likely to express positive views about community policing. Table 29. Attitudes toward Community Policing: Summary of Findings Variable Relationship with Attitudes toward Community Policing_ Bivariate Analyses Influence over whether subordinates are allowed to go + out of service for problem-solving Likelihood that District Management will recognize + good performance District and shift workload - Value congruence with fellow supervisors + Value congruence with District Management + Multivariate Analyses + Job IPD/SPPD has done providing enough time for problem-solving Job IPD/SPPD has done in providing the information necessary for problem-solving District and shift workload Table 29 also depicts the results of Logistic regression analyses using attitudes toward community policing as the dependent variable. Again, workload exerted a significant influence on the likelihood that supervisors would hold positive attitudes 154 toward community policing. Supervisors’ perceptions of the job that their departments had done in providing enough time or the information necessary for problem-solving also significantly predicted supervisors’ attitudes toward community policing. When supervisors perceived that their department provided enough time for problem—solving, they were more likely to feel positively about community policing. This makes a certain amount of sense, as one of the major obstacles to community policing is providing officers with the time necessary (i.e., time when officers are freed from the radio) to engage in community policing (Kennedy, 1993). On the other hand, supervisors who believed that their departments had done a good job in providing the information necessary for community policing were less likely to express favorable views of community policing. The literature on community policing suggests that officers and supervisors often feel that community policing is a reform being shoved down their throats (Sadd & Grinc, 1994); perhaps this relationship reflects this sentiment. The results of analyses conducted with attitudes toward community policing as the dependent variable provide some support for the theoretical framework presented in Chapter Four. In that chapter, I presented a model depicting the antecedents and consequences of organizational commitment, which I argued was analogous to attitudes toward community policing (see Figure l). Ultimately, I proposed a different model (see Figure 2) to reflect the dependent variable used in this study. The problems with the reliability of the dependent variable do not allow for an assessment of the applicability of this model to the study of supervisors and support for community policing. Of the results in this dissertation that we can be most confident in (those conducted with attitudes 155 toward community policing as the dependent variable), however, it appears that the data provided a good fit to the model presented in Figure 1. A slightly modified version of this model53 is presented in Figure 8. Role conflict and work experience variables were significantly related to attitudes toward community policing in this study (see Table 29). No relationships were revealed between demographic, personal, or group/leader relations variables. This should not be interpreted as definitive proof that these relationships fail to exist, however; without a sample of sufficient size that allows for the regression of attitudes toward community policing against all five categories of variables simultaneously can such a conclusion be made. Directions for Future Research More research is needed in order to come both to an understanding of the factors that shape and explain supervisory behavior in general, and the factors that facilitate or discourage supervisors to support community policing in particular. While important from a practical standpoint, this research is also theoretically important. First, from an organizational standpoint, if we seek to theoretically understand the ways that organizations behave, then we must understand the behavior of individual actors within the organization. Without this information, our understanding of organizational behavior is limited and incomplete. A great deal of research has been dedicated to patrol personnel, as well as those at the apex of the police hierarchy. Unfortunately, the same attention has not been paid to the “men in the middle.” This is a critical gap in knowledge; “a full understanding of leadership behavior requires that it be 53 Figure I has been modified to reflect the categories of variables used in the policing literature and in this research. rather than the categories of variables found in the organizational psychology literature. 156 9:80:22 6032:: 58 eofimofioa»: 8:823: 0:: lllll ”802... At xomncoow b83595 Ali 30538500 95.5 TV woo—:30 20m 1111 TV 8550 29m 33:. o . hegemony T133525 2o: 8.3m 9.3— ¢ 4 3 A+v mvsaom “Ogém—m hfiWfi—Mflu A+V :ouufiogm lllll mouaoflflxlm—r .3 91:036.— LI Any—356% . 3. >9 A+v :EDSEOU aghht< OUCDaUn—EOU 8>mDUHUm A+V 95:3. % > 3 555. TV 52338 3 3 2552 3:23:50 voZoBom A+V oEEom 3 page. Badge: 3 ow< 83:83:30 Bane—39330 cinemas—on 3:83.— nofiutouuuugb 85.9.8330 Began Bianca 8.0533309 fie§3§=< wig—en— 3E=EEeU 2:38 8:333. a: 8955.32.00 c5“ 3:ocooou:< .w man—mg 157 studied as a dependent variable as well as an independent variable” (Farris & Lim, 1972: 214). Second, it is assumed in organizations that there is a link between the abstract value statements of the administration and behavioral practices at the service delivery level. This is particularly true of police organizations, which possess structural manifestations of this assumption-such as a hierarchy and chain of command. If we truly seek to understand the behavior of police organizations, we need to learn is how the translation of organizational values and objectives occurs at the level of middle management. Research that provides more insight into middle managers and the role they play in facilitating or subverting change is important theoretically and practically. Theoretically, it will help improve our understanding of the “men in the middle;” practically, such research may play an integral role in informing police administrators how to better ensure the success of community policing. What is the next logical step, based on the results presented here? If I were to design a study to address the problems encountered in this research, what would it look like? The data that used in this study were limited in several ways with respect to the current inquiry. The reliability of the dependent variable in this study could be addressed in a sampling plan that ensured that supervisors would be observed in multiple sessions. Perhaps the most significant limitation of the data was that support for community policing—representing a change in attitudes and behavior away from a traditional model of policing—is a topic that might be best explored over time. Change is not an instantaneous, dichotomous (no change/change) phenomenon. Instead, change is a 158 process, involving several stages or steps.54 In general, systematic observation does not lend itself well to the study of long-term phenomena such as change. The POPN data (and supervision data in particular) do not allow for such an examination. The project was designed to provide a rich, descriptive source of information about police work generally, not for studying patterns of behavior over time. Moreover, most supervisors in the POPN study were observed only once, necessarily precluding the ability to make any statements or draw any conclusions about patterns of behavior (or behavioral change). A project better suited to studying the central research questions in this research would thus involve the observation of supervisors over an extended period of time, or involve observations occurring at different intervals over an extended follow-up period. In this study, I was unable to conduct many of the analyses I would have liked to, because of my sample size. In particular, the small sample size precluded my ability to simultaneously investigate the effects of demographic, personal, role conflict, work experience, and group/leader relations variables. I was also unable to examine whether there was an interaction between attitudes toward community policing and various characteristics to determine whether it was the combination of these factors—rather than each individually—that might exert an influence on supervisor behavior. A better study, then, would have a sample of sufficiently larger size. Finally, while analyses using attitudes toward community policing yielded several significant results, a study designed from the “ground up” would involve a different 54 For example, the Transtheoretical Model of Behavior Change (used extensively in areas such as smoking cessation. weight loss, and addiction) suggests that change occurs in distinct stages occurring over time (Prochaska & DiClemente, 1982, 1984). In this model attitudinal change precedes behavioral change. Behavioral change is the final step in this model; in order for “true” change to occur. the desired or “goal” behavior must be maintained for a minimum of 6 months. 159 measure of this phenomenon. The improvements suggested in this section would allow for a full test of the model presented in Figure 8. Conclusion Whether the personnel within these police agencies had a clear understanding of the goals and objectives of “community policing” has been much less documented or understood. Whether personnel were prepared adequately for the “new” role implied of community policing is equally less clear. Whether the organizational systems within any particular police agency could support a shift from traditional to community-based policing has also been generally overlooked in most of these studies (Greene 1998a, p. 147; emphases in original). The road to community policing is often a rocky one. This may be particularly true for police supervisors, who are provided little in the way of direction by police administrators. This is a particularly precarious situation given that many scholars contend that police supervisors play a critical role in the ultimate success or failure of community policing. Despite the criticality of police supervisors to the successful implementation of community policing, researchers have failed to make supervisors, their reactions to community policing, and the reasons behind their reactions the explicit focus of empirical research. By conducting the research outlined in this dissertation, I attempted to fill some of the voids in knowledge by providing the answers to two important research questions: (1) to what extent do supervisors support community policing, and (2) what factors significantly increase the likelihood of supervisory support of community policing? Unfortunately, methodological problems prevented me from fully addressing these research questions. While I was not able to address these questions, I was able use the theoretical framework to examine these research questions using an alternative 160 dependent variable. Results of analyses using attitudes toward community policing demonstrated that this framework has potential for helping us understanding the factors that influence attitudes toward community policing, and how these attitudes relate to the practice of community policing. It is hoped that future research in this area, perhaps such as that outlined above, may shed additional light on this matter. 161 v0.0 000 w _ .0 M00 00.0 00.0 N00 \- ~ .0 00.0- 0— .0 00.0 00.0 3 .0 * 0N0 004 00.0 m 0 .0 0_ .0 v0.0- 3 .0 .1. 0M0 V0.0- : .0 3 .0- M00 2 .0 0~ .0 00.0- 2 .0 2 .0 00.— 50.0 00.0 2.0 w0.0 :10 V0.0 w0.0 0N0 M00 W00- m0.0- de 0N0 .2. mm.0 * wad ** m.v.0 004 2.0 V0.0 00.0- ©~.0 0_.0- ** 0M0- ... 0N0 N00 050 :0 * mmd 00.0 ** ~00 ** 3H0 2.0 2.0- 0~.0 004 no.0- 00.0- 00.0- 00.0- 00 .0- w ~ .0- _0.0- 00.0- 00.0 w0.0- NN.0 m0.0- m0.0 0~ .0 no.0- V0.0- 00.0- ** _ m0 004 m * 00.0- :00- 00.0- M00- 00.0 0~ .0 vmd- Ng .0- M00- V00- N00- m0.0- 00 .0- 3 .0- m _ .0 _ _.0 00.0- 3.0- 50.0- 004 v 3 .0 v0.0- 00.0 2 .0 _0.0 00.0 NN.0 0N0 00.0 00.0- M00 mmd 3 .0 N 0 .0 00 .0 00.0- 00.0 N00- : 0 .0- 00 .0 004 m ** ** ** 0 _ .0 0N0 >00 5.0 0N.0 mmd- NWO w 0 .0- N00- 3.0 00.0- w0.0 m _ .0 0_ .0 N00- 50.0 NN.0 M00- 0m.0- mm.0- m0.0- 004 N ** 0N Nm0h< NN A<>m~Q a A<>mm0 0N A<>mn~0 a Em aw H> a mamm0mm 2 mmflm<3ma mg .357: 3 ACME-«d m _ Emmwu-DO Nu QmmZOmZH : mOmmmqu—r 0m MMOZVMDE 0 m—Sam0flmm w @950va h 0023;: 0 BOEOwE m m0m1500 v meZMO m m0