THE“ ens srrY uamwes WNW“\HWHUHHHIUHH \mwm WI 3 1293 0103 37 1 This is to certify that the dissertation entitled The Antecedents and Consequences of Citizens' Attitudes Toward the Police presented by James Frank has been accepted towards fulfillment of the requirements for Interdisciplinaeregeein §chal Science with a concentration in Criminal Justice and Criminoiogy Mfiw Major pro(essor Date M 1;, x99: MS U i: an Affirmative Action/Equal Opportunity Institution 0-12771 LIBRARY MIChIQan State Unlverslty PLACE II RETURN BOX to roman this chodtodfrom your record. TO AVOID FINES return on or baton duo duo. DATE DUE DATE DUE DATE DUE THE ANTECEDENTS AND CONSEQUENCES OF CITIZENS’ ATTITUDES TOWARD THE POLICE BY James Frank A DISSERTATION Submitted to Michigan State University in partial fulfillment of the requirements for the degree of DOCTOR OF PHILOSOPHY School of Criminal Justice 1993 ABSTRACT THE ANTECEDENTS AND CONSEQUENCES OF CITIZENS’ ATTITUDES TOWARD THE POLICE BY James Frank During the past fifteen years a growing body of research has focused on the determinants of citizens’ attitudes toward the police. Considerably less attention has been directed at the consequences of citizen satisfaction, though one premise of the attitude research has been that satisfied consumers of police services become coproducers of police outputs. The present study examines each of these two issues; namely, those factors that explain citizen attitudes toward the police and the relationship between citizens’ attitudes toward the police and behaviors of the public that assist or support the police. The study also explores the impact that prior attitudes toward the police have on present attitudes and the causal linkages between variables in the citizens’ attitudes and coproduction models. As far as the determinants of citizens’ attitudes toward the police, the findings suggest that prior attitudes toward the police exert a strong influence on present attitudes. In addition, citizens’ perceptions of neighborhood conditions as problematic also had a substantial impact on attitudes, as community members who perceived neighborhood matters as problematic tended to hold less favorable beliefs than did individuals who did not see these matters as problems. Contrary to existing research, white respondents held less favorable general attitudes toward the police and were less likely than nonwhites to provide favorable evaluations of incident specific contacts with police officers. Finally, the individual level demographic variables were found to exert a greater influence than prior research has suggested. As to coproduction, attitudes do not appear to exert the direct positive influence that has been suggested. Furthermore, citizen willingness to coproduce was also influenced by citizen perceptions of neighborhood problems, in that, individuals who perceived issues as problematic were most likely to get involved in behaviors presumably to alleviate the problem. Apart from specific findings, the data suggest that theoretical models of citizen attitudes toward the police and the coproduction of police outputs should incorporate prior attitudes and citizen beliefs about neighborhood conditions. ACKNOWLEDGMENTS The completion of this dissertation was made possible through the assistance and cooperation of a number of people. First, I would like to thank my advisor, Dr. Timothy S. Bynum, for his assistance and support throughout this project and for not allowing me to delay any further in submitting a completed dissertation. Second, I am also grateful to Dr. Robert Worden for his beneficial criticism, his generous assistance, and for his willingness to share his time and expertise. Finally, I am also grateful to Dr. Stan Kaplowitz, Dr. Carol Weissert, and Dr. John Hudzik for their input into this project. I would also like to thank my family for all of the support they provided to me during the completion of the dissertation. I would especially like to thank my wife, Bonnie, for her support, patience, and encouragement to "get this thing done". Finally, I would also like to thank my children, Carrie, Brian, and Casey for putting up with me while I completed writing my dissertation. iv TABLE OF CONTENTS LIST OF TABLES LIST OF FIGURES CHAPTER 1 INTRODUCTION PURPOSE OF THE PROPOSED STUDY ENDNOTES CHAPTER 2 EXTANT RESEARCH ON CITIZENS’ ATTITUDES TOWARD THE POLICE OBJECTIVES OF PRIOR ATTITUDE RESEARCH CITIZENS’ ATTITUDES TOWARD THE POLICE: THE DEPENDENT VARIABLE Attitudes Toward Officer Behavior The Police Referent DETERMINANTS OF CITIZENS’ ATTITUDES: INDEPENDENT VARIABLES Respondents’ Demographic Characteristics Victimizations Neighborhood Context Experiences With the Police Comparative Assessments Objective Measures Citizen Perceptions of Officer Characteristics CONCLUSION ENDNOTES CHAPTER 3 LINKING ATTITUDES AND BEHAVIOR: CITIZEN ATTITUDES TOWARD THE POLICE AND THE COPRODUCTION OF POLICE OUTPUTS V xii 10 l3 l4 l6 l8 19 23 23 29 32 33 34 34 36 38 COPRODUCTION Citizen Role in the History of Policing Definition of Coproduction Classification of COproduction Activities Benefits of Coproduction LINKING ATTITUDES AND COPRODUCTION: LESSONS FROM THE SOCIAL PSYCHOLOGY LITERATURE The Attitude and Behavior Relationship Methodological Issues Conceptual Issues Conclusion DETERMINANTS OF COPRODUCTION Attitudes Neighborhood—Level Variables Collective Action Demographic Variables Conclusion CONCLUSION ENDNOTES 39 4O 43 43 47 49 50 52 62 66 66 67 7O 73 75 79 8O 82 CHAPTER 4 MODELING CITIZENS’ ATTITUDES AND COUNTERPRODUCTIVE BEHAVIORS: LINKAGES BETWEEN VARIABLES MODEL 1: ATTITUDES TOWARD THE POLICE Demographic Variables Neighborhood Context Variables Respondents’ Perceptions of Neighborhood Conditions Assessments of Response Time Evaluations of Experiences With the Police vi 85 88 89 9O 91 Prior Attitudes 91 Victimization 92 Attitudes Toward the Police 94 Hypotheses 94 Conclusion 96 MODEL 2: DETERMINANTS OF COPRODUCTION 96 Stake in the Neighborhood 98 Perceptions of Neighborhood Problems 99 Demographic Variables 100 Attitudes Toward the Police 101 Feasibility of Community Action 101 Likelihood Others Will Act 102 Coproduction 102 Hypotheses 102 Conclusion 103 ENDNOTES 105 CHAPTER 5 THE PRESENT STUDY: PROCEDURES AND METHODS 106 Sample Selection 106 Procedure 114 MODEL 1: DETERMINANTS OF CITIZEN ATTITUDES TOWARD THE POLICE 115 Dependent Variables Attitudes Toward the Police 115 Independent Variables Demographic Variables 119 Victimizations 129 Neighborhood Context Variables 130 vii Perceptions of Neighborhood Conditions 132 Experiences With the Police 138 Assessments of Response Time 142 Prior Attitudes Toward the Police 142 Analysis 144 MODEL 2: THE DETERMINANTS OF COPRODUCTION 145 Dependent Variables 146 Private Action 150 Collective Action 152 Drug Related Behaviors 154 All Behaviors 154 Independent Variables 155 Attitudes Toward the Police 155 Stake in the Neighborhood 158 Perceptions of Problems 163 Feasibility of Collective Action 165 Demographic Variable 167 Analysis 168 SUMMARY 169 ENDNOTES . 170 CHAPTER 6: CITIZENS’ ATTITUDES TOWARD THE POLICE: FINDINGS AND DISCUSSION 174 FINDINGS . 175 Global Attitudes Toward the Police 175 Direct Effects 175 Indirect and Total Effects 178 Drug Related Attitudes 184 Direct Effects 184 viii Indirect and Total Effects Order Maintenance Attitudes Direct Effects Indirect and Total Effects DISCUSSION CONCLUSION ENDNOTES CHAPTER 7 DETERMINANTS OF COPRODUCTION: FINDINGS AND DISCUSSION Correlations Path Analysis Private Behaviors Direct Effects Indirect and Total Effects Collective Behavior Direct Effects Indirect and Total Effects Drug Related Behaviors Direct Effects Indirect and Total Effects A11 Behaviors Direct Effects Indirect and Total Effects DISCUSSION CHAPTER 8 CONCLUSION LIST OF REFERENCES ix 186 192 192 194 199 208 209 210 210 214 214 214 216 219 219 222 225 225 228 231 231 233 233 244 258 TABLE TABLE TABLE TABLE TABLE TABLE TABLE TABLE TABLE TABLE TABLE TABLE TABLE TABLE TABLE TABLE LIST OF TABLES DETERMINANTS OF CITIZEN ATTITUDES TOWARD THE POLICE: DEPENDENT VARIABLES DETERMINANTS OF CITIZEN ATTITUDES TOWARD THE POLICE: INDEPENDENT VARIABLES ATTITUDE-BEHAVIOR REALTIONSHIP: DEPENDENT VARIABLES ATTITUDE-VEHAVIOR RELATIONSHIP: INDEPENDENT VARIABLES DIRECT EFFECTS OF ALL VARIABLES IN MODEL 1.1: GLOBAL ATTITUDES STANDARDIZED PATH COEFFICIENTS FOR MODEL 1.1: GLOBAL ATTITUDES DIRECT, INDIRECT, AND TOTAL EFFECTS OF MODEL VARIABLES ON GLOBAL ATTITUDES: MODEL 1.1 DIRECT EFFECTS OF ALL VARIABLES IN MODEL 1.2: DRUG ATTITUDES STANDARDIZED PATH COEFFICIENTS FOR MODEL 1.2: DRUG ATTITUDES DIRECT, INDIRECT, AND TOTAL EFFECTS OF MODEL VARIABLES ON DRUG ATTITUDES: MODEL 1.2 DIRECT EFFECTS OF ALL VARIABLES IN MODEL 1.3: ORDER ATTITUDES STANDARDIZED PATH COEFFICIENTS FOR MODEL 1.3: ORDER ATTITUDES DIRECT, INDIRECT, AND TOTAL EFFECTS OF MODEL VARIABLES ON ORDER MAINTENANCE ATTITUDES: MODEL 1.3 CORRELATIONS BETWEEN ATTITUDES AND BEHAVIORS DIRECT EFFECTS OF ALL VARIABLES IN MODEL 2.1: PRIVATE BEHAVIORS DIRECT, INDIRECT AND TOTAL EFFECTS FOR X 116 122 147 160 176 180 182 185 188 190 193 195 197 212 215 TABLE TABLE TABLE TABLE TABLE TABLE MODEL VARIABLES PRIVATE BEHAVIORS: MODEL 2.1 DIRECT EFFECTS OF ALL VARIABLES IN MODEL 2.2: COLLECTIVE BEHAVIORS DIRECT, INDIRECT AND TOTAL EFFECTS FOR MODEL VARIABLES COLLECTIVE BEHAVIORS: MODEL 2.2 DIRECT EFFECTS or ALL VARIABLES IN MODEL 2.3: DRUG BEHAVIORS DIRECT, INDIRECT AND TOTAL EFFECTS FOR MODEL VARIABLES DRUG RELATED BEHAVIORS: MODEL 2.3 DIRECT EFFECTS OF ALL VARIABLES IN MODEL 2.4: ALL BEHAVIORS DIRECT, INDIRECT AND TOTAL EFFECTS FOR MODEL VARIABLES ALL BEHAVIORS: MODEL 2.4 xi 218 221 224 226 230 232 235 Figure Figure Figure Figure Figure Figure Figure Figure LIST OF FIGURES Determinants of Attitudes Toward the Police Determinants of Coproduction Operationalization of Attitudes Toward The Police Models Operationalization of Coproduction Models Private Behaviors Collective Behaviors Drug Related Behaviors All Behaviors xii 86 97 120 156 217 223 229 234 CHAPTER 1 INTRODUCTION During the past two decades criminal justice researchers have become increasingly concerned with the attitudes of citizens towards the police (Brandl and Horvath, 1991; Percy, 1986; Erez, 1984; Brown and Coulter, 1983; Mastrofski, 1981; Decker, 1981; Scaglion and Condon, 1980; Koenig, 1980; Thomas and Hyman, 1977; Parks, 1976; Furstenberg and Wellford, 1973; Hahn, 1971; Boggs and Galiher, 1965). While this line of research has been guided by several factors, one significant premise has been that attitudes of the public toward the police are related to citizen willingness to engage in behaviors supportive of the police (Zamble and Annesley, 1987; Percy, 1986; Brown and Coulter, 1983; Scaglion and Condon, 1980; Koenig, 1980; Bell, 1979; Stipak, 1979; Hahn, 1971). These activities may range from the reporting of criminal behavior or providing police with information concerning suspicious behavior in the community to citizen willingness to participate in community activities (i.e., Block Clubs or Watches) that contribute to community safety and security. C} Most of the research concerning citizens’ attitudes and behaviors, and the relationship between these constructs, has focused on the determinants of public attitudes toward the police. While factors that explain public attitudes have been the subject of increased attention, very few 1 I studies have controlled for the impact of more than just a restricted number of explanatory variables. As such, there remain questions about whether existing research has both produced biased estimates and fully explained the determinants of public attitudes towards the police. In particular, three categories of explanatory variables have not received sufficient attention in existing research. First, the impact of respondents’ perceptions of local neighborhood conditions (e.g., fear of crime in the neighborhood, the quality of life) on attitudes toward the police has received only minimal attention (Stipak, 1979, Percy, 1986). In an attempt to assess the relationship between quality of life and attitudes toward the police researchers have primarily focused on crime rates in the city and have determined that these measures explain very little of the variance in citizens’ attitudes. In contrast, perceptions of neighborhood conditions may have a greater influence on attitudes because the involve information that may be more germane, especially if community members hold the police responsible for the existing conditions. A second area which has received only limited, though increasing, attention involves the influence that contact with the police has on citizens’ attitudes towards the police. While it has been suggested that contact with the police has a substantial impact on citizens’ perceptions of the police, the nature of this relationship has been subjected to only limited scrutiny (Skogan, 1991; Brandl and 2 Horvath, 1990; Mastrofski, 1981; Koenig, 1980, Dean, 1980). More specifically, extant research has not concerned itself with the qualitative dimensions of citizen experiences with the police and how these dimensions differentially impact on evaluations of the specific contact and more general attitudes toward the police. Third, the role played by prior attitudes toward the police on present beliefs has been ignored by extant research. It seems very possible that present attitudes are highly informed by prior attitudes. The failure of existing studies to include measures of these constructs is problematic since each category of variables appears to have substantive importance for a more complete explanation of citizens’ attitudes toward the police. In addition, previous studies that have explored the determinants of citizen attitudes towards the police have all utilized cross-sectional research designs. The use of data collected at a single point in time hinders the ability of researchers to specify the temporal order and examine the causal linkages that exist between both independent variables, and the independent and dependent variables. Furthermore, the utilization of this type of data has prevented empirical assessments of the impact that pre- existing global attitudes have on more recent general attitudes towards the police. Also problematic is the fact that the relationship between citizen attitudes and supportive behaviors has not 3 been subjected to empirical testing. Instead the attitude- behavior relationship has only been the subject of conjecture and untested hypotheses.1 Bell (1979) noted that the "manner in which police are perceived by citizens, to a great extent determines the quantity and quality of the cooperation police receive from the citizens" (pp. 196-197). Similarly, Stipak (1979) suggests that citizen attitudes deserve attention from future research because "widespread feelings of dissatisfaction with police may lower citizen cooperation with law enforcement personnel" (p. 49).2 Thomas and Hyman (1977) asserted that "positive attitudes provide potentially invaluable resources for the police" (p. 317). While several additional writers have more directly mentioned the attitude-behavior link (Brandl and Horvath, 1991; Decker, 1981; F “endall, 1974), others have implied that increased satisr. lion leads to public assistance in the performance of the police function (Skolnick and Bayley, 1988; Wyckoff, 1988; Goldstein, 1987; Zamble and Annesley, 1987; Scaglion and Condon, 1980). The implication is therefore that as citizens become more satisfied with the police, they will become coproducers of police outputs. In other words, citizens will engage in activities that help police make arrests, clear crimes, and maintain order in the community. Since, this is a job function normally performed by the police, when citizens engage in these behaviors they are coproducers. This hypothesized relationship between positive 4 attitudes and supportive behaviors is especially noteworthy because it is an integral assumption upon which several current police strategies are premised. Most notable of these programs is the recent movement towards community oriented policing. This program relies heavily upon improving relations between the police and community members, with the underlying assumption of the program being that satisfied consumers of police services will become the "eyes and ears" of the police (Goldstein, 1987). This transformation of a dissatisfied public to a more satisfied and cooperative one is premised upon two assumptions that form the framework of the philosophy behind community oriented policing. First, community members are perceived as’a resource to be mobilized (Weisburd and McElroy, 1988; Wyckoff, 1988; Goldstein, 1987) and, secondly, "community policing is supposed to be more satisfying to the public than traditional policing" (Skolnick and Bayley, 1988:26). Proponents of community oriented policing suggest that the mobilization of community resources is more likely to occur as citizens become more satisfied with police strategies. Ultimately, more satisfied individuals are expected to play an active role in assisting officers in the performance of their police duties. A related police strategy, problem oriented policing, also relies heavily upon community input and cooperation from neighborhood residents in solving community problems. 5 Specifically, community members are expected to help police identify issues of concern in their neighborhoods and, in addition, to contribute to the planning of solutions, as well as, to partake in actions that support the police (Cordner, 1988; Eck and Spelman, 1987). Again, it is anticipated that a more satisfied public will be more receptive to helping the police. While a great deal of interest in the policing community has focused on the implementation of these strategies, research has not empirically examined the underlying assumption of each policy that citizen attitudes have an impact on the performance of certain behaviors by the public. Furthermore, law enforcement administrators have recently begun to publicly acknowledge that the assistance of citizens would help them more effectively perform their job (Brown, 1989). This acknowledgement followed years during which the police stated that with more training and better technology they, alone, could perform the police role. Taken together, these recent developments indicate that a more complete explanation of attitudes toward the police, and a more thorough understanding of the relationship between citizens’ attitudes toward the police and the coproduction of police outputs, would contribute to our understanding of the role citizens might play in policing and what, if anything, police policies can do to encourage coproduction. PURPOSE OF THE PROPOSED STUDY This study analyzes the antecedents and consequences of citizen attitudes toward the police. To do so, two broad research questions are addressed. First, what factors determine or contribute to citizens’ attitudes toward the police? Second, are citizens who have more favorable/positive attitudes toward the police more likely to engage in behaviors that assist officers in the performance of the police function? In other words, what role do attitudes, in comparison to other relevant factors, play in explaining whether individuals will assist the police in the production of police outputs? The first question is directed at those factors that explain citizens’ attitudes toward the police. The second query involves the relationship between citizens’ attitudes and specific behaviors that assist the police in the performance of their job. The issue raised by the second question is a logical extension of extant research that has sought to explicate those variables that determine the attitude construct, though this matter has received considerably less attention. Answers to these questions should extend the substantive value of research concerned with the attitudes of citizens and, furthermore, furnish some indication whether the hypothesized relationship between attitudes and behaviors is correct. ENDNOTES 1. Two studies are often cited as supporting the assumption positive attitudes are related to citizen behaviors that are supportive of the police (Block, 1974; Hahn, 1971). However, an examination of these studies indicates that in neither study did the researchers have measures of actual behaviors that were engaged in by respondents. Instead, the respondents were asked to respond to a series of hypothetical situations by selecting a response option that they deemed appropriate. The measure used in these studies is therefore a measure of behavioral intentions. While behavioral intentions are believed to be significant predictors of subsequent behavior (Liska, 1984, 1974), there remains the possiblity that factors will intervene and constrain individuals from actually performing the intended behavior. 2. Many of the writers speak of the fact that citizen attitudes are related to citizen "cooperation" with the police. Admittedly, cooperation may include a number of possible forms of behaviors, one of which might include not acting in a hostile manner when confronted by the police. One could also easily argue that calling and reporting suspicious and/or criminal occurrences to the police would also be included within the types of behaviors contemplated by the term cooperation. CHAPTER 2 EXTANT RESEARCH ON CITIZENS’ ATTITUDES TOWARD THE POLICE During the past twenty years, an increasing body of research has sought to delineate the determinants of citizen attitudes toward the police. This research has naturally been driven by the research objectives of the individual conducting these studies. As objectives vary, so have the explanatory variables that have been examined. Over time, explanatory models have been derived which include either a greater number of explanatory variables, or at least different independent variables, with the inclusion of specific variables dependent on the objectives of a certain inquiry. At the same time, these studies have sought to explain a variety of citizen outlooks toward the police, again based on the objectives of the research endeavor. The result is a body of literature that has provided information concerning factors that influence a variety of perceptions that citizens possess concerning the police, without fully explaining the attitude construct. This chapter reviews existing research that pertains to citizen attitudes toward the police. It proceeds by first examining the research objectives of studies involving citizen attitudes toward the police. A discussion of the various dependent variables that have been utilized in attitudes toward the police research follows. The chapter 9 concludes with a review of the explanatory variables used in the existing attitudes toward the police literature. OBJECTIVES OF PRIOR ATTITUDE RESEARCH Research on citizens’ attitudes toward the police has been driven by three stated or implied objectives. An initial objective of many researchers was to explain the variance in citizens’ attitudes across social groups (Erez, 1984; Apple and O’Brien, 1983; Furstenberg and Wellford, 1973; Jacob, 1971; Hahn, 1971; Boggs and Galiher, 1965). As such, many of the studies following this line of inquiry portrayed respondents’ status characteristics as the substantively significant factors that influenced the public’s attitudes toward the police. In general, these studies determined that while citizens generally held favorable attitudes toward the police, there ls divergence in the extent to which members of certain social groups held positive attitudes. For example, blacks generally held less favorable attitudes than whites, young people possessed less favorable attitudes than older individuals, and males held less positive attitudes than females. However, while favorable attitudes toward the police varied with certain demographic characteristics, favorable attitudes were the norm rather than the exception. Thus, while whites were more supportive than blacks, blacks were still overwhelmingly supportive. More recently, researchers have justified exploring citizen attitudes toward the police on the basis that 10 citizens, as consumers of police services, are in a position to evaluate the effectiveness of police policies and procedures (Percy, 1986; Flanagan, 1985; Parks, 1984; Brudney and England, 1982; Percy, 1981; Parks et al., 1981; Whitaker, 1980; Percy 1978; Hindenlang, 1974). Citizen evaluations are perceived as especially valuable since they supplement the limited performance measures presently available to the police. Often surveys are designed to elicit feedback from the public regarding strengths and deficiencies in police behavior and thus enable the police agency to identify and adjust police policies and practices. The dependent variable in research following this line of inquiry often measures respondents’ evaluations of some performance domain of the police. Furthermore, as it became recognized that performance evaluations may be influenced by characteristics of the evaluators, their neighborhoods, and assessments of contact with the police, and in turn, become "subjective measures of outcomes" (Parks, 1984:119), the number of explanatory variables contained in models was greatly expanded beyond those in research focused solely on explaining variation across social groups. However, researchers disagree on the proper way to measure agency performance. Stipak (1979) believes that objective measures (e.g., inputs such as the number of police and/or number of arrests) are the proper indicators of agency performance. He contends that citizens are unaware of the actual performance of many government 11 services since they have little contact with government agencies. Relatedly, Percy (1980) claims that actual levels of service may have little influence on citizen evaluations of performance, unless they fall far below acceptable levels. Thus, requesting community members to assess areas of police performance with which they have no direct experience and, in general, have very little knowledge about, provides a suspect performance measure. This problem is compounded when the agency relies on these evaluations during the development and implementation of police policy. This situation has been described as the "misuse of a performance indicator" (Stipak, 1979). While these arguments appear quite plausible, it even more easily can be argued that while objective inputs are necessary for sufficient agency performance, they are no guarantee of better performance (Parks, 1984). For example, having a high ratio of officers to citizens (an objective input) may supply the resources for "better" performance, though it is no guarantee that "better or quality" performance will occur. How these officers are deployed, and how they act once deployed, will influence whether a high ratio of officers results in satisfactory performance. Thus, the use of subjective citizen surveys to assess the impact of police inputs and, in general, police performance appears justified. A final objective upon which citizen attitude research has been premised is that citizens’ attitudes toward the 12 police are theoretically and practically important factors that influence citizen willingness to engage in behaviors supportive of the police function (Brandl and Horvath, 1990; Zamble and Annesley, 1987; Percy, 1986; Coulter and Brown, 1983; Decker, 1981; Scaglion and Condon, 1980; Koenig, 1980; Bell, 1979; Stipak, 1979; Hahn, 1971). Attitudes, as evaluative internal states, are believed to predispose individuals to act in a manner consistent with the attitude (Ajzen and Fishbein, 1977; Schuman and Johnson, 1976; Liska, 1975). Based on this presumption, researchers have hypothesized that favorable attitudes toward the police would be related to citizen willingness to help the police whether individuals are engaged in consumption (i.e., recipients of police service) or production (i.e., helping police perform their job) roles. As was noted, this relationship has been the subject of considerable theorizing, but has not been subjected to rigorous testing. CITIZENS’ ATTITUDES TOWARD THE POLICE: THE DEPENDENT VARIABLE A review of studies commonly included within the "citizen attitudes toward the police" literature indicates that researchers have been concerned with a variety of outlooks that the public holds concerning the police. As such, the dependent variable (attitudes) has been conceived and operationalized in a variety of manners. Variance in the attitude measure is often due to the differing behaviors that are the subject of the measured attitude and/or the 13 object (police referent) of the attitude. While examinations of these various outlooks/beliefs adds to our understanding of "attitudes" people have concerning the police, it is also quite possible that factors that influence one type of attitude do not effect, or have a different effect, on other attitudes. This section discusses the different attitudes3 that have been utilized in existing research. Attention is directed at differentiating operationalizations of the attitude construct. The impact that selection of a specific conceptualization of the dependent variable may have on findings relative to the significance of explanatory variables is also addressed. Attitudes Toward Office; Behavior. Questions used to measure attitudes often have as their subject different dimensions of police behavior. First, a number of studies ask respondents to evaluate the quality of law enforcement, with some focusing on the overall quality of police performance (White and Menke, 1982; Koenig, 1980; Christenson and Taylor, 1983; Apple and Obrien, 1983; Thomas and Hyman, 1977; Scaglion and Condon, 1980; Stipak, 1979). For example, Apple and O’Brien (1983) asked respondents whether the "quality of police protection was exactly as you like it?". Koenig (1980) requested his subjects to decide whether "in general, police in your area are doing a very good, good, bad or very bad job?". In contrast, other researchers have questioned respondents about police 14 capabilities in dealing with specific conditions (Skogan, 1991; Percy, 1986; Erez, 1984). Skogan (1991) asked a series of questions in which respondents were to state "how good a job" police did in preventing crime, helping crime victims, and in keeping order on the streets. A second subject of attitude questions involves citizen perceptions of whether police act in an equitable manner. Questions within this category often ask respondents whether they believe police practices are equally applied to all groups of citizens, without concern for the characteristics of the involved parties (White and Menke, 1982: Thomas and Hyman, 1977; Hahn, 1971). Citizens have also been asked to assess whether the police are honest, fair, and courteous (White and Menke, 1982; Furstenberg and Wellford, 1973: Ennis, 1967; Reis, 1967). Finally, some studies ask citizens about their beliefs concerning the proper police role or a specific police policy (Zamble and Annesley, 1987; Flanagan, 1985; Homant, Kennedy, and Fleming, 1984). The dependent variable in these studies focuses on citizen attitudes about whether police should continue performing certain duties or whether officers should engage in other forms of conduct. Specifically, Flanagan (1985), requested respondents to state whether the police should respond to all calls or only calls identified as crimes -- a specific policy issue. The aforementioned attitude questions tap very different substantive dimensions of the attitude construct. 15 In general, researchers have failed to be concerned with the qualitative dimensions of the attitudes they have measured. This is unfortunate because different cognitive processes may be involved in responding to these questions. For instance, replies to questions concerning citizen satisfaction with services or quality of performance are likely informed by citizen expectations about the reasonable level of police performance. Respondents could therefore say that they are satisfied with poor performance becauSe they do not expect any better service. In contrast, questions that are intended to elicit responses pertaining to officer honesty or the equal application of police policies may be free from these comparative assessments. While many researchers have lumped these different operationalizations within the generic category of "attitudes toward the police", in reality they pertain to a variety of dimensions of the attitude construct. The Police Referent. Dependent variables in the citizens’ attitude research aIso differ as to police agency or the police officers that are subject of citizen attitudes. Variance in the specificity of the attitude object ranges from non—specific references to the police in general to questions concerning the behavior of an officer or officers with whom the person has had actual contact. More specifically, some respondents are only asked about the "police" without any reference to a certain agency (Zamble and Annesley, 1987; Christenson and Taylor, 1983; White and 16 Menke, 1982; Smith and Hawkins, 1973). Several studies ask respondents to reply in reference to the "local police" (Scaglion and Condon, 1980), with it explicitly noted that respondents’ are to focus on the behavior of officers within a specific department. Other researchers use questions that direct respondents’ attention to the police in her/his "area" (Koenig, 1980; Skogan, 1991), or in his or her "community" (Thomas and Hyman, 1977, Dean, 1980). Finally some questions ask citizens about the behavior of police in a particular type of police-citizen encounter (Brandl and Horvath, 1991; Percy, 1980; Poister and McDavid, 1978; Furstenberg and Wellford, 1973). While the intention is not to criticize the wording used in survey questions pertaining to the police referent, it might be improper to assume that the same information is cognitively accessed when formulating a response to each of these questions. For instance, replies to questions that refer to the police in general are likely to be premised upon attitudes about the police institution, along with perceptions of police departments an individual has had heard about or with which the individual has had contact. Responses to more specific questions about the police in the neighborhood are presumably based on information acquired during contact with police in the community or second hand information (i.e., vicarious contact) about the police in their neighborhood. Finally, the most specific questions, those that refer to the behavior of police officers in a 17 police-citizen contact situation, ask people to use information acquired in a specific encounter to formulate their responses. While similar factors may influence all these attitudes, such should not be automatically presumed. Researchers should therefore be concerned with the police referent when deriving models intended to explain citizen attitudes toward the police. DETERMINANTS OF CITIZENS’ ATTITUDES: INDEPENDENT VARIABLES Prior research has identified explanatory variables that are often included in models of citizen evaluations of police performance. These explanatory variables generally fall into seven categories: (1) respondents’ demographic characteristics (Erez, 1984; Coulter and Brown, 1983; Thomas and Hyman, 1977; Furstenberg and Wellford, 1973), (2) victimization experiences (Brandl and Horvath, 1991; Koenig, 1980; Poister and McDavid, 1978), (3) neighborhood level (Percy, 1986; Percy, 1981b; Stipak, 1979), (4) experiences with the police (Erez, 1984; Coulter and Brown, 1983; Mastrofski, 1981; Decker, 1981; Scaglion and Condon, 1980; Koenig, 1980; Parks, 1976; Boggs and Galiher, 1965), (5) comparative assessments of police work by respondents (Percy, 1986; Erez, 1984; Hahn, 1971), (6) objective performance measures (Percy, 1986; Coulter and Brown, 1983; Stipak, 1979), and (7) officer characteristics (White and Menke, 1982; Thomas and Hyman, 1977; Furstenburg and Wellford, 1973; Hahn, 1971; Ennis, 1967; Reis, 1967). 18 Respondents’ Demographic Characteristics. The literature on citizen evaluations has highlighted the relationship between individual level variables and attitudes toward the police. The race of the citizen has received the most attention from researchers. A fairly consistent finding of this line of research is that nonwhites (principally African-Americans) are normally less satisfied with police services than whites and, thus, hold less favorable attitudes toward the police (Parks, 1984; Scaglion and Condon, 1980; Percy, 1980; Smith and Hawkins, 1973; Furstenberg and Wellford, 1973). The attitudes of nonwhites are often thought to result from two factors: minority meml s are more likely to have negative contact with the police and/or hold more negative attitudes toward governmental authority (Skogan, 1978). In addition, Flanagan (1985) suggests that racial groups may hold varying expectations about the role police perform in society. While not specifically addressed in his research, one could surmise that these expectations would be related to varying attitudes toward the police. Though less satisfied with police performance than whites, nonwhites still voice generally positive attitudes when global measures of performance are utilized. Existing research has also found that younger individuals often possess less positive attitudes toward the police than older citizens (Apple and O’Brien, 1983; Scaglion and Condon, 1980; Smith and Hawkins, 1973; Boggs 19 and Galiher, 1965). One explanation for this finding is that as people get older they tend to believe that the police play a more legitimate role in protecting the status quo, though this may be changing as younger generations come to possess more conservative beliefs. In addition, younger people are also believed to have a lower stake in conformity. Finally, younger individuals are more likely to have hostile (antagonistic) contact with the police (Wellford, 1973). As such, younger individuals may possess less favorable attitudes toward the police than older individuals. Findings concerning the influence of gender on attitudes have been m. 1. Apple ' O’Brien (1983) found that females voice more positive evaluations of police than do males (see also Thomas and Hyman, 1977). However, Boggs and Galiher (1965) found in their sample of African-American respondents that gender was not a statistically significant predictor of attitudes. It has been suggested that females may have more favorable attitudes toward the police than males because they generally have contact with the police that is less antagonistic than contact that typically occurs between the males and the police.2 The income level of respondents has also been found to be related to attitudes toward the police. The majority of research has determined that income is positively related to attitudes. Specifically, as income increases citizen attitudes toward the police are believed to become more 20 favorable (Erez, 1984; Apple and O’Brien, 1983; Hindenlang, 1974). It has been suggested that higher income individuals view police as performing an important role in protecting their position within society. More affluent individuals may therefore be more supportive of police activities that they see as related to this role. Also, higher income people are less likely to have had (or heard about) negative or antagonistic experiences with the police. As such, these people may also express more favorable global attitudes toward the police than less fortunate individuals. At the same time, wealthier people may also expect more from the police concerning specific types of performance that involve the protection of property and personal security. As such, their expectations about the activities police perform may vary from those of people in a different financial situation. If this assumption is correct, then income may only be positively related to attitudes about performance domains where performance is consistent with the person’s expectations. The relationships between race, sex, and/or income and citizens’ attitudes may not be as straightforward as has been suggested. Several studies have suggested that there might be interactions between these variables. For instance, there may be an interaction between race and income. Boggs and Galiher (1965) found that the relationship between race and income for African-Americans was curvilinear. The lowest status respondents (street 21 people) in his study had the least favorable attitudes, while low status (though not as low as the street people) household residents possessed more favorable opinions. However, higher status household residents held attitudes that were more positive than the street respondents, though less favorable than the lower class household respondents. Boggs and Galiher (1965) believed that these findings resulted because street respondents and higher status African-Americans each experienced more negatively evaluated contacts (street people were stopped and searched while higher status individuals were subjected to car searches that were perceived by respondents as unjustified) with the police than did low status household respondents. This research suggests that income and race combine to influence police contact and citizen evaluations of the police-citizen experience. An interaction between race and sex may also occur. Researchers have suggested that members of minority groups3 are more likely than others to request police services involving a range of situations (Cumming, Cumming, and Edell, 1965). As such, female minority respondents are likely to have more contact with the police than white females. If prior research is correct that minority group members possess less positive genral attitudes toward the police than whites, these general attitudes may influence evaluations of the officer conduct during the encounter. This situation may be compounded by the nature of the 22 police performance requested during an experience with the police, especially if nonwhite females are utilizing the police to carry out a variety of service related tasks. Specifically, the behavior requested may not be viewed by officers as involving "real police work" and is therefore less likely to be carried out with the same vigor as other police activities. Furthermore, there is likely to be less consensus between police administrators, officers, and the public concerning the performance preferred in these situations. As such, police performance may be within the range of options available to the officer(s), but may not conform to the expectations of the involved citizen resulting in a negative evaluation of the police officer. Thus, it appears that the relationship between gender, race, and attitudes toward the police may not be as straightforward as with other variables. Victimizations. Research has also found that being a crime victim influences citizens’ attitudes toward the police (Koenig, 1980; Smith and Hawkins, 1973). Furthermore, it has been suggested that a series of Victimizations may have a more pronounced negative impact on evaluations than does a single victimization (Coulter and Brown, 1983; Poister and McDavid, 1978). Presumably this results because victims blame the police for violations of their person and/or property. Neighborhood Context. Neighborhood level variables are also considered important factors that influence citizen 23 attitudes toward the police. The neighborhood context in which people find themselves has been determined to ‘ influence evaluations of government provided services (Christenson and Taylor, 1983; Apple and O’Brien, 1983). Neighborhood level variables fall into two general categories. Variables descriptive of the socioeconomic characteristics of neighborhood comprise the first category. The second category includes citizens’ perceptions of neighborhood conditions. Most of the attention directed at neighborhood level conditions has focused on demographic characteristics of neighborhoods. Two variables, percent of residents who are nonwhite and mean neighborhood income, have been included in prior studies (Flanagan, 1985; Apple and O’Brien, 1983; Parks, 1981; Brudney and England, 1982). These two measures are often included (Flanagan, 1985; Apple and O’Brien, 1983; Jacob, 1971) because the neighborhood context in which a person resides provides the opportunity for neighborhood experiences and interactions. Attitudes toward police service may be influenced by the norms and expectations of relevant reference groups (Brudney and England, 1982; Christenson and Taylor, 1983), that are exchanged during interactions between group members. Over time people may internalize and adopt the norms of their reference groups (Tajfel, 1982; Blumer, 1958), which in turn influences their attitudes toward the police. Since African-Americans generally voice less positive 24 evaluations of police performance than Anglo-Americans, the presumption is that the attitudes of individuals (white and nonwhite) residing in predominantly nonwhite communities will be negatively influenced by the neighborhood context. Specifically, Apple and O’Brien (1983) found that an increase in the number of nonwhites in the community negatively influenced the attitudes of African-Americans in the neighborhood. Namely, the attitudes of these individuals were less positive than the attitudes of individuals living in communities that contained fewer nonwhite residents. These authors contend that this finding resulted because individuals in these neighborhoods had more opportunity to interact with individuals who possessed less positive attitudes toward the police. If racial composition has the effect suggested by this research (Apple and O’Brien, 1983), one would also anticipate that the attitudes of African-Americans residing in white neighborhoods may be more positive than the attitudes of African—Americans who live in predominantly minority areas. As to Anglo—Americans, Apple and O’Brien (1983) found that their attitudes were also influenced by neighborhood racial composition. However, they suggested that the attitudes of Anglo-Americans may be affected indirectly through perceptions of community safety. Apple and O’Brien found that whites residing in communities with a substantial minority population were less likely to feel safe in their neighborhood, with concern over personal safety being 25 responsible for the less positive attitudes toward the police voiced by these individuals. As such, these people are believed to hold less favorable attitudes than individuals with similar racial characteristics that live in predominantly white communities. Citizen attitudes are also believed to be influenced by the economic context of the neighborhood (mean neighborhood income) in which the person resides. Similar to the manner in which neighborhood racial composition influences attitudes, the economic status of the local community is believed to provide citizens with chances to interact with fellow community residents who are of similar economic status as themselves. If economic status has the effect that prior research suggests, then neighbors are likely to share similar attitudes toward the police. For instance, the attitudes of citizens who reside in economically depressed areas may be influenced through contact with other members of the community who are more likely to hold unfavorable views of the police, since economically disadvantaged individuals generally have less favorable attitudes toward the police than wealthier people (Erez, 1984; Apple and O’Brien, 1983; Hindenlang, 1974). Similarly, individuals residing in wealthier communities are presumed to be more likely to have contact with individuals that possess more favorable general attitudes toward the police. For these reasons, it is generally assumed that as mean neighborhood income decreases so will satisfaction with 26 police services. Stipak (1979) noted that respondents’ "attitudes toward general neighborhood conditions may influence levels of satisfaction with the police" (p. 54). Unfortunately, there is only limited research that has addressed this second category of neighborhood level variables. Instead, existing research has focused primarily on citizen perceptions crime rates or objective measures of reported crime as a proxy for neighborhood conditions. While criminal occurrences are admittedly representative of a neighborhood condition, so are a range of other local matters (i.e., people hanging out on the street, prostitution, vacant home, etc.) which may each influence attitudes people possess concerning the job police are doing regulating behavior within the community. The failure to include measures of this construct is unfortunate for several reasons. First, immediate neighborhood conditions should be salient to respondents. Beliefs about local community conditions are therefore more likely to be cognitively accessed (Fazio and Williams, 1986), which in turn, may cause them to influence attitudes toward the police . Secondly, the impact of these factors may be especially important where the problem is one citizens consider within the function and control of the police (Percy, 1986; Percy, 1981; Stipak, 1979). A short example will highlight how neighborhood conditions may influence citizen attitudes toward the police. Beliefs about neighborhood matters such as local 27 drug sales, the use of illegal drugs, vagrancy, and prostitution within the community are matters which have the potential to influence the quality of life of community residents. These conditions may also spawn other related concerns among citizens. Community residents may become fearful of being victimized by either a participant in these activities or an offender who does not actually partake in these activities but who preys on participants (Cohen and Felson, 1979). In addition, the quality of life of area residents may also be influenced by street crimes and increases in disorder that often accompany these types of activities. While prior research has examined the influence of fear of crime on attitudes toward the police (Zamble and Annesley, 1987; Thomas and Hyman, 1977), concern with only this issue fails to capture the effect of neighborhood conditions that do not produce victimization worries. Many problems may only influence citizen perceptions of the job police are doing controlling matters which influence more general aesthetic conditions in the neighborhood. What makes these conditions applicable to a study of attitudes toward the police is that the regulation of each of these activities and/or conditions generally falls within the mandate of the local police. Thus, a local police department’s failure to control, or at least sufficiently manage, these conditions may impact on citizens’ attitudes toward the police. The incorporation of measures relative to this construct should permit a more complete 28 understanding of the impact that a range of neighborhood level variables have on attitudes. Experiences flipp the Police. Experiential measures involve encounters between citizens and the police. In general, citizen experiences with the police have been shown to greatly influence their evaluations of overall quality of police performance (Zamble and Annesley, 1987; Parks, 1984; Scaglion and Condon, 1980; Rusinko, Johnson and Hornung, 1978; Furstenberg and Wellford, 1973). Several researchers suggest that experiences should have more predictive power than demographic variables (Parks, 1984; Koenig, 1980; Winfree and Griffiths, 1971), since without contact people have much less information on which to base their opinions (Coulter and Brown, 1983). There are three dimensions of police-citizen contact that are integral to an assessment of citizen attitudes toward the police. These dimensions may be used to differentiate types of experiences from one another (Skogan, 1991; Dean, 1980; Scaglion and Condon, 1980). The first two dimensions, why the contact occurs and by whom it is initiated are both important. Types of police and citizen experiences that may occur are of infinite variety. Prior research (Dean, 1980) has utilized four types of encounters, namely, police responses to (1) calls for assistance, (2) for information, or (3) to report a crime, and (4) street stops initiated for the purpose of questioning and/or to enforce the law. In the first two situations, the contact ' 29 is initiated by the citizen to secure something from the police, whereas in the last situation (stops and questioning) the police are acting against the person involved in the contact. In the third situation (a victimization experience), the police are responding to a situation that the individual has not voluntarily placed her/his self within, and in fact, may hold the police responsible for the victimization that precipitated the encounter. The third dimension, citizen assessments of their experience with the police, appears to be the most critical dimension when relating experiences to citizen attitudes toward the police (Percy, 1986; Scaglion and Condon, 1980). An assumption of existing research has been that personal evaluations of specific instances of police behavior influence more general attitudes a person may possess towards the police. The existence or nonexistence of characteristics associated with each of these dimensions of police encounters makes the contacts not only qualitatively different from one another, but different in ways that may affect attitudes toward the police. While citizens’ assessments of officer behavior during police-citizen encounters are believed to influence more general attitudes toward the police, the magnitude of the impact often varies depending upon whether the contact is positively or negatively evaluated (Erez, 1984, Jacobs, 1971). Specifically, negatively rated contacts with the 30 police are more likely to influence attitudes than are positively evaluated encounters with the police (Dean, 1980: Poister and McDavid, 1978; Parks, 1976; Jacob, 1971). In fact, Jacob (1971) noted that positively evaluated contacts have very few consequences for citizens’ attitudes, while negative evaluations deflate overall attitudes toward the police. This might stem from the fact that citizens normally possess positive attitudes, unless they have some reason to believe the police are not competent, fair, or honest. Thus, the valence of the evaluation interacts with the person’s prior attitude. However, merely because an experience was positively or negatively rated should not lead one to assume that all similarly evaluated experiences (all positive or negative evaluations across types of occurrences) will equally enhance or decrease attitudes toward police performance. The first two dimensions of citizen encounters, the reason the contact occurred and whether the contact was citizen or police initiated, are also important to consider (Skogan, 1991; Erez, 1984; Mastrofski, 1981; Scaglion and Condon, 1980). For example, being stopped and questioned by the police are police initiated encounters that normally involve an adversarial relationship between the citizen and the officer. In addition, the relationship is often an antagonistic one, in that there is a negative affective component to the interaction. A positive evaluation of this type of experience is likely to have at most a minimal 31 impact on attitudes toward the police as the evaluation interacts with the type of contact in affecting global attitudes. In fact, the nature of the interaction may outweigh the positive evaluation of the officer’s performance. In contrast, requests for information or assistance entail citizen initiated contact. Since officers are not taking action against involved parties, these types of interactions are less likely to be characterized as antagonistic, though they probably still involve an adversarial component. Positive evaluations of requests for assistance or information are more likely to have a greater impact than positive evaluations of stop situations. Comparative Assessments. Extant research has also determined that citizen expectations concerning certain facets of police work should also be included in models that attempt to explain public attitudes concerning police performance (Percy, 1986; Erez, 1984; Coulter and Brown, 1983; Hahn, 1971). Citizens apparently make comparative assessments between what they expect and what actually occurs which, in turn, influence their overall attitudes toward police work. These expectations have been determined to be more predictive of attitudes than actual performance measures. This has most commonly been operationalized with expectations of response time. In these studies, citizens were asked to state whether the police response time was 32 faster, slower, or about the same as expected. This strategy permitted an assessment of response time in terms of whether it was congruent with the individual’s expectations. These response time assessments were found to be more important than measures of actual response time (Percy, 1986; Parks, 1984; Percy, 1978) when explaining citizen evaluations of police performance. Thus, perceptions of performance have the potential to influence attitudes to a greater degree than do more objective measures of actual performance. Objective Measures. As previously noted, there are disagreements in the literature concerning the proper manner in which to operationalize objective measures and whether objective measures even have an impact on citizens’ attitudes toward the police (Percy, 1986; Coulter and Brown, 1983; Parks, 1981; Stipak, 1979). Objective measures that have been utilized in prior models of determinants of citizens’ attitudes have included the number of officers, ratio of officers to citizens, crime levels, and actual response time. In general, these measures have been found to explain very little of the variance in citizen attitudes. The objective measures used in these analyses are normally citywide aggregate indicators. Since citywide measures may not be indicative of neighborhood conditions they may not be salient to individual respondents. This may account for the limited impact these measures have had on individual respondents (Parks, 1984; Coulter and Brown, 33 1983; Stipak, 1979). Neighborhood crime data may have more of an impact than citywide crime levels. However, citizen perceptions of crime levels as evidenced by community fears of becoming a crime victim, irrespective of the amount of actual crime in the neighborhood, may be an even more significant predictor of citizen attitudes toward the police. Citizen Perceptions pf Officer Characteristics Several studies have suggested that citizen perceptions of officer honesty, trustworthiness, and respect for citizens influence global attitudes toward the police (White and Menke, 1982; Ennis, 1967; Reis 1967; Hindenlang, 1974). Specifically, individuals who believe that police officers are dishonest and likely to take bribes are not likely to possess favorable global attitudes pertaining to police performance. Likewise, beliefs by community members that police act in a disrespectful manner towards citizens have a negative influence on overall attitudes toward the police. It is also quite possible that negative global attitudes may lead people to believe that individual officers are dishonest. CONCLUSION Most researchers who have examined factors that determine citizen attitudes toward the police have been concerned with only a limited set of hypotheses and have not Sought to fully delineate those factors that explain attitudes. While respondents’ demographic characteristics 34 have been included in almost all models, the influence of neighborhood conditions on attitudes has received only scant attention. At the same time, the relationship between experiences with the police and citizen attitudes has not been fully explored. Furthermore, no one has attempted to examine the influence that pre-existing attitudes may have on more recent measures of the same attitudes. Finally, only a few studies have attempted to empirically assess the relationship among explanatory variables and their direct and indirect influences on attitudes toward the police (Parks, 1984; Scaglion and Condon, 1980). An analysis of this type would provide a more complete understanding of role these factors play in explaining attitudes. Therefore, even though there have been consistent findings as to the impact of several predictors, explanations of citizen attitudes toward the police begs for further inquiry through the inclusion of variables that compose a more complete model and specify the relationship among variables. As Percy (1986) noted "previous studies have erred by using oversimplified models of influences on citizen satisfaction. Researchers must examine a full set of influences on citizen perceptions of service performance" (p. 81) to provide a more complete understanding of the factors that explain citizens’ attitudes toward the police and to isolate the effects of any one of these factors. 35 ENDNOTES 1. A variety of outlooks that citizens have toward the police have been addressed in prior research (e.g., how "satisfied", "how good a job" are police doing). These studies are often included within the "attitude" research. While studies that assess satisfaction and performance evaluations are often grouped together this should not imply that these attitudes are synonymous. 2. Females are more likely than males to have contact that involves requests for a variety of services. These types of contact, therefore, do not involve situations where the police are acting against the involved party. 3. Most research involving minority groups has been involved either African-Americans or Spanish speaking Americans. When the phrase "members of minority groups" is utilized in the text it refers to these two groups. 36 CHAPTER 3 LINKING ATTITUDES AND BEHAVIOR: CITIZEN ATTITUDES TOWARD THE POLICE AND THE COPRODUCTION OF POLICE OUTPUTS Empirical research which has been directed at the attitudes of citizens toward the police has focused almost solely on the determinants of citizen attitudes. Considerably less attention has been directed at the consequences of favorable attitudes; namely, what behaviors flow from satisfied consumers of police services. In fact, most existing research has used citizen satisfaction as the dependent variable and has presumed or implied that the manner in which police are perceived by citizens determines the nature and extent of supportive behavior police receive from citizens. Coproduction involves individuals and/or groups acting outside of their regular roles to contribute to the production of goods and services they may also consume (Parks et al., 1981). It occurs when both regular (the police) and consumer (citizens) producers act to transform inputs into goods and/or services (Percy, 1981). The citizen component of the service delivery process is "coproduction". When citizens engage in behaviors that are intended to support and assist the police in the production of police outputs they are coproducing. Presently, there exists a growing body of criminal justice and public administration literature that discusses 37 the coproduction of police outputs. However, this research has shown only limited concern for the role of citizens’ attitudes in influencing decisions of community members to help police perform their function in society. Instead, this research has developed a framework for defining and categorizing coproductive behaviors. There does exist several bodies of knowledge, outside of more traditional criminal justice research, that provide information concerning the factors that influence citizen coproduction of police outputs and the role that citizen attitudes toward the police may play in this process. Specifically, existing community action research has been concerned with neighborhood level factors that influence the decision to participate, or not participate in behaviors that might improve the quality of life in an area. The second line of research involves the literature on collective action. This research suggests that citizens are not necessarily inclined to join groups and partake in the activities of groups that provide "public goods". The third body of literature involves extant social psychology research that has been directed at the empirical analysis of the attitude and behavior relationship. The intention of this chapter is to bring together these three areas of research. The first portion of the chapter is devoted to an examination of the coproduction or behavioral component of the present study. A brief history of the law enforcement role played by lay citizens in 38 provided. Next is a discussion of the definition of coproduction, along with a classification scheme for coproduction behaviors. Finally, research addressing the determinants of coproduction is reviewed. The second portion of the chapter is devoted to the social psychology research that has examined the nature and measurement of attitudes, and the relationship between attitudes and behaviors. In this portion attention is initially directed at conceptual and methodological issues surrounding the measurement of attitudes and behaviors. The chapter concludes with a discussion of those factors that may constrain or enhance the attitude-behavior relationship. COPRODUCTION During the past decade municipalities have suffered from shrinking resources. As a result, the finances necessary to maintain municipal services at present levels have been lacking. The result has often been a cutback in local service delivery budgets at a time when production costs are increasing (Wilson, 1981; Rich, 1981; Parks et al., 1981). Citizen coproduction of police outputs has been suggested as a means to maintain service levels or to augment production by departments facing decreasing personnel levels. Furthermore, public service requires the involvement of citizen consumers if police are to be effective (Percy, 1981). Citizen involvement may take a variety of forms. Skogan and Antunes (1979) note that police rely on citizens 39 to provide them with information since knowledge of crime occurrences is most often possessed by community residents. Wilson and Kelling (1982) suggest that citizens can coproduce by showing concern for neighborhood conditions. Neighborhood cleanups may indicate that the community "cares" and is willing to exert some control over problematic situations that develop within their neighborhood. In this manner, citizens may discourage crime, making performance by the police easier and possibly, more effective. Citizen Role I he History pf Policing. A review of the history of policing in the United States indicates that citizen participation in policing is a role that is not new to community members. However, the exact nature of the role played by lay people has varied as policing evolved in America. Much of this change in role occurred along two dimensions. The first dimension pertains to whether citizen performance of the police role was obligatory or performed under voluntary conditions. The second dimension consists of whether community members were substitutes for a formal police force or provided support to an existing government sanctioned police agency. For instance, during colonial times citizens were obligated to protect their communities under the "watch and ward" system (Walker, 1983; Johnson, 1981). In this capacity, citizens performed the police role and were the actual producers of what are now considered police outputs. 4O In the mid to late 1800’s, as cities began to emerge and grow, they encountered a host of social problems that were not susceptible to control under the existing watch and ward systems. In addition, citizens began to demand that police services, along with a host of other municipal services, be provided by local governments. This development resulted in the emphasizing of formal government sanctioned police departments as primary service providers and the underemphasis of the role of citizens as producers (Johnson, 1981). This trend, the underemphasis of citizen contributions to the police process, continued during the period between the 1880’s to the mid 1960’s. Several specific factors during this time period moved citizens further away from performance of the police role. First, in late 1800’s and early 1900’s many police departments were plagued by political corruption as local political leaders controlled "police policies, department organization, and personnel selection" (Johnson, 1981:105). In response to this condition, police reformers sought to change policing by making it a profession. In order to achieve this goal the reformers believed that it was necessary for police agencies to be free from outside interference into departmental affairs (Walker, 1992; Johnson, 1981). This interference included not only that exerted by local political leaders, but also input provided by ordinary citizens. Second, separation of the police from the community was 41 also enhanced by technological developments. The emergence of the police cruiser resulted in reduced interaction between patrol officers and community members. The rise of crime labs and training academies led departments to believe that they could perform the police function without the help of citizens. Thus, the emergence of professionalization and the call for autonomy, combined with technological changes terminated the participation of many citizens in police related activities. During the past twenty-five to thirty years the role of citizens in the production of community safety and security has been revived. Much of this change has occurred because police have begun to acknowledge that even with expanded resources they are unable to cope with increases in disorder and crime. In addition, academic research has determined that citizens are situated in a position that permits them to control much of the information on which performance of the police function depends (Skogan and Antunes, 1979). . Currently, efforts are underway to again establish the relationship between the community and the police. These efforts are most evident in the emergence of police strategies commonly termed "community oriented policing" and "problem oriented policing" (Skolnick and Bayley, 1988; Eck and Spelman, 1987; Goldstein, 1987). Though qualitatively different, each approach relies heavily on continuous citizen input into the police process. Citizen input may entail the provision of information to the police or the 42 definition of "problems" within the community, along with a number of other behaviors. The importance of these strategies for the present study is that they involve a return of citizens to a role in the provision of police functions -- a role they can now perform to support an existing police agency. Definition pf Coproduction As noted, coproduction involves individuals and/or groups acting outside of their regular roles to contribute to the production of goods and services they may also consume (Parks et al., 1981). When citizens engage in behaviors that assist the police in the production of police outputs they are coproducing (Percy, 1981). It is therefore a process that binds together the hired producer and the consumer (Wilson, 1981) to achieve a function normally considered within the occupational role of the producer (the police). Recent research attention has suggested that citizens are not just passive recipients of police services. Citizens are viewed as possessing both the resources and capabilities to join in the production process (Wycoff, 1988). By engaging in certain activities community members may become active participants in the production of police outputs (Skolnick and Bayley, 1988; Weisburd and McElroy, 1988; Greenberg et al., 1983; Brudney and England, 1982: Wilson, 1981; Rich, 1981; Percy, 1981; Whitaker, 1980; Percy, 1978). Classification pf Coproduction Activities. Behaviors 43 that are intended to assist the police can take a variety of forms. Citizen conduct may range from reporting information about suspicious activities to engaging in anti-crime marches. These behaviors can be classified into two categories; namely, private action or collective behaviors (Sharp, 1984; Percy, 1981; Schneider and Eagle, 1975). There are two methods that can be used to determine whether a specific behavior is placed within the private action or collective behavior category. The first method focuses on the who benefits (i.e., the target of the information) from the coproduction activity, the individual alone or the collectivity. If the information reported to the police or to a neighborhood group pertains to conditions involving the person reporting, then it would be private action. Alternatively, if the focus of the information involves neighborhood conditions, then it would be classified as collective behavior. The second method focuses on the behavior of the party engaged in the coproduction of police outputs. Specifically, this categorization scheme looks at whether the person acts alone or interacts with other community members while engaging in behaviors supportive of the police (Sharp, 1984). Sharp suggests that citizen-initiated contacting (voice) of public officials is an individualistic mode of participation, irrespective of the target of the information provided. On the other hand, working with others in the neighborhood would be "communal participation" 44 or collective behavior. Utilizing Sharp’s (1984) criteria, calls to neighborhood organizations would be classified as collective behaviors notwithstanding the target of the information, while direct calls to the police about neighborhood conditions by an individual would be private action. Sharp’s (1984) categorization scheme is used in the present study. Private and collective actions take a variety of forms. As noted, private action entails behaviors that are engaged in by individual community members acting alone. One form of private action involves providing information to the police. Conduct subject to regulation by the police often occurs out on the street or in public facilities and out of sight of police officers. Details concerning an incident are therefore often within the control of the victim or other members of the public who may have witnessed the incident. For police to gain access to the information, and to effectively carry out their mandate, citizens must act as conduits for the information (Skolnick and Bayley, 1988; Weisburd and McElroy, 1988; Farrell, 1988; Skogan and Antenunes, 1979). For example, if criminal activity occurs outside the view of officers, citizens must first report the occurrence to the police before they may even become involved. Whether a suspect is arrested probably depends on whether citizens provide information concerning the occurrence and a possible offender. This information may include general biographical information, descriptive 45 information, and/or knowledge concerning the whereabouts of a suspect. Citizens who provide the police with these types of information help in the performance of the police role and contribute to the production of police outputs. Providing police with information need not only concern suspect-oriented information. Citizens can assist police by reporting suspicious circumstances they observe in their neighborhoods (Goldstein, 1987; Percy, 1981; Washnis, 1976). For instance, citizens may call the police to report unusual activity around a neighbor’s home that they suspect is drug related traffic. Community members may also report to the police a suspicious person seen wandering around a neighbor’s home. Information of this nature may permit police to intervene prior to the commission of an offense or to disrupt ongoing criminal activity. At the same time, citizens may also call the police to report more general neighborhood conditions, such as, the fact that power lines or tree limbs are down and pose hazards. A second form of private behavior involves activities that are intended to makes one’s home more secure (Percy, 1981; Washnis, 1976). Though not of concern in the proposed study, these behaviors might include actions such as putting outside lights on a home, protective bars on the windows, and/or more secure locks on possible entrances. By engaging in these activities citizens reduce the risk that they will be victimized; a benefit to not only the citizen but also the police.’ In these instances community members are 46 coproducing through crime prevention activities. In contrast to private actions, citizens may also engage in neighborhood level collective behaviors. These forms of conduct involve community members working together in some form of joint action (Sharp, 1984; Percy, 1981). Collective activities may range from providing information about a crime or suspicious occurrence to a community organization to community based actions focused on a specific neighborhood problem. Typical behaviors that comprise this category include participation in citizen patrols, block watches or block clubs, anti-crime marches, safety awareness programs, neighborhood clean-up campaigns, and victimization prevention programs (Skolnick and Bayley, 1988; Cordner, 1988; Goldstein, 1987; Rosenbaum, 1987; Lavrakos and Herz, 1982; Washnis, 1976). Using Sharp’s (1984) definitions these behaviors are distinguishable from private actions because they involve citizens working together, though they also generally focus on neighborhood issues of safety and security. Irrespective of whether the conduct is private or joint action, the involved citizens are contributing to the performance of the police function. At the same time, these actions may improve the quality of life in the community. Finally, individuals may "feel better" about themselves knowing that they are contributing to the general welfare of their neighborhood. Benefits 9; Coproduction. Several benefits may be 47 derived from coproduction. As has been noted, citizens may contribute to the production of police outputs and outcomes and make their communities safer (Ostrom, Parks, Whitaker and Percy, 1979; Ostrom, 1973). Specifically, they may help police make arrests and clear crimes by providing information that deals with specific criminal incidents. People may also provide information that allows police to disrupt activity before it becomes illegal behavior. Residents may also engage in target hardening activities (i.e., home security measures) that make the commission of criminal activity more difficult. Community members may also coproduce by showing concern for and taking action to remedy untended neighborhood conditions. Coproduction, therefore, has the potential to improve the quality of life in the neighborhoods of coproducers (Rosenbaum, 1987; Wilson and Kelling, 1982; Lavrakos and Herz, 1982; Percy, 1981) and, at the same time make performance by police departments easier and more effective. Communities are safer and performance of the police role easier for two specific reasons. First, self protective measures (i.e., target hardening) reduce the opportunities for being victimized, or at least make the successful commission of certain crimes more difficult. Second, individual and community actions that involve obtaining and providing information to the police increase the risk of apprehension to offenders. 48 LINKING ATTITUDES AND COPRODUCTION: LESSONS FROM THE SOCIAL PSYCHOLOGY LITERATURE A number of researchers who have attempted to explain citizen attitudes toward the police have assumed that attitudes are significant predictors of whether community members become coproducers of police outputs. Unfortunately, these same individuals have not been concerned with methodological and conceptual issues that may influence this supposed relationship. Instead, they have blindly asserted that attitudes and behaviors are related, without concern for the fact that each construct represents complex phenomenom. A substantial amount of social psychology research has been directed at the attitude and behavior constructs and the empirical analysis of the relationship between attitudes and behavior. This social psychology literature provides insights that criminal justice researchers have unfortunately ignored. This is especially true for researchers who have hypothesized that there is a direct causal relationship between citizen attitudes toward the police and the willingness of community members to engage in behaviors that may contribute to the coproduction of police outputs. This section first discusses the social psychology research that has examined the nature and measurement of attitudes, and the relationship between attitudes and behaviors. In this section attention is initially directed at why there is a presumed causal relationship between 49 attitudes and behaviors. Next is a discussion of measurement issues that are believed to enhance the attitude-behavior relationship. The final portion of this section of the chapter proceeds with a discussion of those factors that may constrain or enhance the attitude-behavior relationship. In this portion of the chapter, explanatory variables in addition to attitudes are examined. The Attitude and Behavior Relationship One significant concern of the discipline of social psychology has been the prediction of behavior, with an area of research within this field of study focusing on the influence of attitudes on behavior. Implicit in this research has been the theoretical assumption that attitudes determine, reflect or, at least, correlate substantially with behaviors (Schuman and Johnson, 1976; Wicker, 1969). For example, people that have a favorable attitude towards an object are expected to perform favorable behaviors toward the same object. At the same time, people who possess unfavorable attitudes toward an item are likewise assumed to act unfavorably toward the object. Reviews of the early (prior to the 1960’s) investigations exploring the congruence between attitudes and behaviors have indicated that the these two variables are "related to an extent that ranges from small to moderate in degree" (Schuman and Johnson, 1976: 178). Wicker (1969) in concluding his review of attitude-behavior studies was even less positive in his perception of the value of 50 attitudes for predicting behavior. He noted that, "taken as a whole, these studies suggest that it is considerably more likely that attitudes will be unrelated or only slightly related to overt behaviors than that attitudes will be closely related to actions". (p. 65) Thus while attitudes may predispose people to certain forms of behavior these comprehensive reviews of empirical studies indicate that the possession of a certain set of beliefs is no guarantee that conduct congruent with the attitude will result (also, see Sherman and Fazio, 1983; Weigel and Newman, 1976; Ajzen and Fishbein, 1977). Not willing to abandon the premise that attitudes are important predictors of behavior, researchers subsequent to the publication of these articles attempted to account for the inconsistent findings in the existing attitude-behavior literature. One line of inquiry that evolved suggested that the inconsistent findings in the attitude-behavior research occurred because of methodological problems. These methodological issues pertained to the improper operationalization and measurement of the attitude and behavior constructs. A second, and related, position that emerged in the social psychology literature was that attitudes are only one of many factors that determine behavior. To properly explain behavior researchers believed it was necessary to specify those variables, along with the underlying attitude, that cause variation in behavior (Ajzen and Fishbein, 1977; Liska, 1974). As such, most research since the initial reports of low or nonsignificant relations 51 between attitudes and behavior have incorporated other variables that might constrain and/or enhance the relationship into their explanatory models. Still, recent research confirms that attitudes and behaviors are not always congruent.1 This ongoing debate over whether attitudes are sufficient predictors of behaviors has been termed the "consistency controversy" by Liska (1975). While there remains a question over the exact role of attitudes in determining behavior, there is a belief that the inconsistent findings that have resulted in attitude-behavior research have partially occurred because of methodological and conceptual concerns. Methodological Issues. Methodological issues fall into two general categories: (1) the attitude and behavior constructs and (2) general measurement error. These characteristics of attitudes and the measurement of attitudes toward the police are first examined. Next, a discussion of relationships between attitude and behavior measures is presented. A number of definitions exist for attitudes. These definitions have several common characteristics. First, attitudes are believed to contain an evaluative component. As such, they involve either positive or negative feelings that a person possesses (Petty and Cacioppo, 1981; Ajzen and Fishbein, 1977; Liska, 1974). Second, these feelings are directed at some "social object" (e.g., a person, issue, or object), commonly referred to in the literature as the 52 attitude object (Liska, 1974). In discussing these first two characteristics Alwin (1973:258) noted, "attitudes are thought of as a relationship between the individual and objects which have both direction and strength". A third feature of attitude definitions is the assumption that attitudes predispose people to act in a certain manner in conformity with their attitude (Ajzen and Fishbein, 1977; Weigel and Newman, 1976; Sherman and Fazio, 1983). This third feature is responsible for the aforementioned assumption, by criminal justice researchers and lay people, that attitudes and behaviors are causally related. As noted, this assumption has not always been confirmed as an empirical reality. It is the second characteristic of attitudes, the object at which the attitude is directed, that is of primary concern in research that empirically examines the relationship between attitudes and behaviors. This is noted for several reasons. First, the attitude object directs the focus of the attitude being measured. Namely, the attitude being explored in the research concerns a person’s evaluation and feelings towards a specific item designated by the attitude object. SeCond, the attitude object must be theoretically related to the behavior that the attitude is alleged to influence. That is, attitude measures should only be expected to "predict behaviors that are appropriate to the attitude under consideration" (Wiegle and Newman, 1976: 795). If the attitude object is not relevant to the 53 target behavior, it would be improper to anticipate that the attitude measure would predict the behavior under analysis. For instance, Wiegel and Newman (1976) measured respondents’ attitudes about environmental issues in an attempt to see if they were predictive of behaviors involved with recycling and a neighborhood clean-up campaign. In this instance, attitudes about the environment are theoretically related to behaviors favorable to environmental clean-up. Third, the attitude object influences the level of specificity of the attitude measure. For instance, in the question "how satisfied are you with the police", the explicit attitude object is the institution of policing, though individuals probably utilize a specific department as their reference point. In contrast, a question that states "how satisfied are you with the job the local police are doing controlling the sale of drugs in your neighborhood", focuses the attention of the respondent on a more specific attitude object; namely, the local police department’s regulation of drug sales in the respondent’s neighborhood. Concern with the specificity of the attitude (and behavior) measure is paramount since it has been suggested that the degree of specificity in attitude and behavior questions should be similar to enhance the attitude and behavior relationship (Lord, Lepper, and Mackie, 1984: Liska, 1975; Fishbein and Ajzen, 1975). Ajzen and Fishbein (1977) noted that there should be correspondence (similarity) between measures of the attitudinal predictors 54 and behavioral criteria as to the act, the target at which the action is directed, the context in which the act is performed, and the time of performance. They (Ajzen and Fishbein, 1977) noted in their review of existing research that when there was a lack of correspondence between the operationalizations of these constructs, correlations between measured attitudes and behaviors were not statistically significant. In contrast, when there was correspondence statistically significant relationships between the measured attitudes and behaviors were observed. More specifically, if general attitudes are measured, then the behavioral criterion should be equally general if the objective is to maximize the correlation (Liska, 1975: Liska, 1974). Similarly, if a specific attitude is measured, then a specific behavior intention (Ajzen and Fishbein, 1977) or a specific overt behavior (Lord, Lepper, and Mackie, 1984: Schuman and Johnson, 1976, Schwartz, 1978) should also be measured. Again, the objective is to ensure correspondence between the measures. For example, asking someone about their attitudes toward the police taps a very general (global) attitude towards the police and may not be related to a specific coproductive behavior such as reporting drug related conduct to the police in your neighborhood. The general attitude may not contain an affective component concerning the local police and/or, more likely, the respondent’s attitudes relative to the job police are doing in his/her immediate residential area 55 involving the sale and use of illegal drugs. If such is the case, then the attitude would not be expected to predict such behaviors. Studies that have attempted to predict specific behaviors toward organizations and institutions based on general attitudes respondents possess towards these same institutions have often determined that the two variables are not statistically related (see Schuman and Johnson, 1976 for a review of these studies). The social psychology research concerning the need for correspondence has generally been ignored by research examining citizens’ attitudes toward the police. However, limited attention has focused on the specificity of attitude measures (White and Menke, 1982) though it is grounded within a framework provided by two political scientists. According to Easton (1965) and Dennis (1976), attitudes can be classified as being either diffuse or specific based on the object of the measured attitude. Diffuse attitudes refer to general attitudes an individual has toward an institution or the ideological foundation of the institution (Dennis, 1976). A question requesting a diffuse attitude regarding the police ("how satisfied are you with the police") has as its attitude object the institution of policing. White and Menke (1982:226) suggest that "values such as order, justice, and fairness in procedure" form the police institution ideology, and thus provide the information on which this attitude is premised. Whether this framework is directly transferable to 56 police research is open for debate. Easton (1965) and Dennis (1976) were concerned with the office of the President. This position transcends the entire country. The object of any attitude question concerning the presidency is this single office. Public perceptions of the power and credibility of the President attach to a specific Officeholder and the office at any given time (Neustadt, 1980). In contrast, there is no true national police force that may be the object of a general attitude question. When responding to a question about the police, the person’s attention is most likely directed towards the behavior of officers within a specific department, or several law enforcement agencies, that the individual has had contact with or heard about. Thus, there may not be a truly diffuse attitude towards the police as claimed in prior research, though one could easily argue that some attitudes toward the police are more genral than others. In contrast, the attitude object of a specific attitude is a particular individual within the institution (Dennis, 1976). Questions which tap a specific attitude require the respondent to refer to the performance of a particular person and render an evaluation of the quality of the individual’s performance. Asking a respondent to state whether he/she was satisfied with the way an officer handled an encounter involving the person would be an example of a question intended to elicit a specific attitude. In this situation, the object of the requested attitude is the 57 performance by a specific officer at a designated time. Easton (1965) and Dennis (1976) therefore specify an attitude dichotomy by characterizing attitudes as being either diffuse or specific. However, this dichotomy does not account for attitudinal questions which request citizens to refer to a more specific attitude object than the institution of policing, yet one not as specific as the behavior of a certain role incumbent. For instance, asking respondents about the police in their neighborhood directs their attention to behaviors by the police within their community, the general quality of policing in their neighborhood, and a specific group of officers that work within a defined geographical area. As such, this question elicits an attitude that does not fit within either the diffuse or specific category. Similarly, an attitudinal question that requests individuals to state their evaluations of police performance in their neighborhood involving the sale and use of drugs is more specific as to the attitude object than the'prior example. Still, this query would not be representative of a specific attitude as defined by Easton and Dennis. For purposes of the present study, attitudes toward the police are perceived as composing a continuum with diffuse and specific attitudes as the anchor points. The first example above would fall on this continuum closer to the diffuse end, while the second would fall between the first example and the specific attitude anchor point. Specificity 58 of the attitude object is an extremely important issue because the attitude object must be theoretically related to the behavior under analysis for there to be a realistic possibility of finding a statistical relationship. A few examples are provided to more fully explain how correspondence and theoretical relevance between attitude and behavior measures are related. These same examples illustrate how the issue of correspondence might surface in the present study. As noted, a question that asks respondents to state "how satisfied" they are with the police, measures very general attitudes toward the institution of policing (White and Menke, 1978). This attitude measure may be so diffuse that it does not measure attitudes toward neighborhood police. Thus, one would not expect to find that the magnitude of the relationship between this attitude and behaviors supportive of local law enforcement is very great. In other words, the attitude may be so general that it does not correlate well with specific behaviors, such as reporting information about suspected criminal activity in the community. Merely because an individual has a positive overall attitude towards the police does not necessarily imply that he/she will act to support or assist the actions of his/her local police department. To avoid this problem a "global" attitude measure may be used. For instance, citizens may be asked how satisfied they are with the "police in their neighborhood". This type 59 of question focuses the attention of the respondents on local law enforcement personnel, though it is global in the sense that it does not call attention to specific activities engaged in by officers of the local police force. One would anticipate that the relationship between this attitude and behaviors supportive of local police would be greater than in the previous example. Questions that elicit public attitudes relative to the job police are doing controlling a certain form of behavior (i.e., drug use or neighborhood disorder) focus on a more specific attitude object than do the prior attitude/satisfaction measures. In this instance, the attitude object, the regulation of drug use by neighborhood police, is more specific than in either of the above situations. If the behavioral measure included reporting drug related behavior and /or engaging in marches against places where drugs are sold/used in the neighborhood, there would be substantive correspondence between the attitude and behavior items. Based on the assumptions of extant attitudes toward the police studies and existing social psychology research, one might reasonably expect that the attitudes and behaviors in this hypothetical case to be even more highly correlated than in any of the prior examples. Admittedly, it is not always possible to ensure correspondence between the attitude and behavior measures. If such can not be guarded against, the utilization of a 60 multi-item behavioral scale is suggested. Wiegel and Newman (1976) determined that this approach was most appropriate when examining the relationship between general attitudes and specific behaviors, since people who hold equally favorable general attitudes regarding an object may vary in their actions towards the attitude object (see also Schuman and Johnson, 1976). Their findings indicated that correlations between attitudes and a comprehensive behavioral scale were substantially higher (and presumably more accurate) than correlations with separate behaviors (Wiegel and Newman, 1976). A final example will illustrate how the use of a multi- item behavioral scale may increase the magnitude of statistical relationships without jeopardizing the integrity of involved measures of the constructs. Citizen attitudes relative to the job police are doing regulating the sale and use of drugs may be related to citizen behaviors involving the reporting of information concerning observed drug use and sales in the neighborhood. People with favorable perceptions of the job police are doing combating drugs may provide details of drug activity to the police. At the same time, some members of the public may decide to engage in alternative strategies that ultimately benefit police control of drug related behavior. Citizens may decide to organize a protest march in front of a known or suspected crack house. An individual may personally confront users and dealers." One would expect that individuals with less 61 favorable attitudes toward the police would engage in these last two types of conduct. The failure to construct a behavioral scale that taps alternative behavioral options such as these, may result in a finding that the attitude and a single behavior are unrelated, while the same attitude is predictive of other equally substantial contributions to police performance. Conceptual Issues. Another concern with prior citizen attitude research revolves around the assumption that there is a direct causal link between favorable attitudes toward the police and the provision of public support. While this assumption may be correct, there is substantial social psychological literature that suggests otherwise. Indeed, a substantial amount of research has attempted to systematically incorporate other variables, in addition to the measured attitude, which may enhance and\or constrain the extent and direction of the attitude and behavior relationship (Liska, 1984; Ajzen and Fishbein, 1977; Schuman and Johnson, 1976; Liska, 1974). One category of variables that has received attention in the social psychology literature is situational factors. Situational factors may include social norms (Fishbein and Ajzen, 1976), the norms of a person’s reference groups, or at least the respondent’s perception of what the reference group favors or expects (Andrews and Kandel, 1979), and the fact that the conduct must be performed in the public domain where it is might be seen by others (Liska, 1975; Warner and 62 DeFleur, 1969). Situational factors may influence (i.e. they may constrain or enhance) the relationship between attitudes and behaviors (Sherman and Fazio, 1983; Ajzen and Fishbein, 1977; Fishbein and Ajzen, 1976; Schuman and Johnson, 1976; Liska, 1975). Situational factors could increase consistency between satisfaction and coproduction, though in many situations they might also be expected to constrain the relationship. If the situational factor (e.g., reference group support) is congruent with the attitude then one would anticipate that the effect would be to increase consistency between the attitude and behavior (Liska, 1974). For example, if a respondent holds a favorable attitude towards the police and the attitudes of the person’s reference group favors providing support to the police, then the situational factor and the attitude are congruent. One would, therefore, anticipate that the relationship between the respondent’s attitude and private coproduction would be strengthened. If a person holds a favorable global attitude towards the police but she believes that most of her peers (the reference group) do not trust the police since they think police officers are involved with much of the criminal activity in the community, the person has several options available to her if the decision is made to attempt to remedy a problem in the area. The person may turn to a local community organization and act in conjunction with fellow community residents. In this instance, the 63 relationship between attitudes and collective behaviors would be enhanced. However if no community organization exists and the individual is afraid of reprisals from her peers if she cooperates with the police she may refuse to intervene. In this example the attitude and the situational factor are inconsistent and the relationship between the measured attitude and behaviors supportive of the police is likely to be constrained. Community social norms may have similar effects (Rosenbaum, 1986). In neighborhoods where there is not consensus about issues such as the need to cooperate with one another to solve problems, people are less likely to engage in collective action. Similarly, where there is not agreement about whether certain neighborhood issues are problematic people may not act to remedy the problem. Finally, even assuming that there is limited consensus about neighborhood problems, people within the area may not agree about the means to be utilized to remedy or control the problems. In each situation social norms may reduce the likelihood of private or collective behaviors supportive of the police. If the existing attitude towards the police research is correct, then citizen action intended to help the police produce outputs is even more unlikely if the person holds a negative attitude towards the police and encounters one of the above situational factors. Liska (1984) and Warner and DeFleur (1969) contend that the location where the behavior is to take place influences 64 the community members willingness to engage in the behavior. In the present study conduct may occur either within a person’s home or out in the public domain. Private behaviors allow the actors to retain their anonymity (i.e., call the police hotline using your own phone), while public actions require the individual to sacrifice some of their anonymity. Public actions may associate the parties with the police or a group attempting to curtail behaviors that others in the community support. As such, it is suggested that private actions may be more likely to be engaged in than public behaviors, especially if general social norms do not favor citizen behavior that assists performance of the police role. It must also be remembered that there are costs involved in affirmatively supporting the police. Costs may entail time, effort, and being ostracized from a social group. These costs may discourage people with positive dispositions towards the police from getting involved. Finally, extant research on attitudes toward the police research also fails to account for attitudes other than satisfaction with the police that may be responsible for supportive behavior. Attitudes toward specific community issues may determine whether a person gets involved in the production process. Similarly, a respondent’s beliefs about the efficacy of collective self help may influence her/his willingness to get involved in group actions (Rich, 1980). Finally, attitudes about the future quality of life in the 65 community may also impact on one’s willingness to engage in behaviors that may improve community conditions (Greenberg et al., 1983; Lavrakos and Herz, 1982; Percy, 1981b; Washnis, 1976). By assuming that attitudes toward the police are the sole attitudes responsible for overt actions, researchers provide a simplistic approach to a complex decision. In fact, focusing on this single attitude may cause people to ignore a number of attitudes relevant to the performance of the examined behavior. Coproductive behaviors may result from a number of attitudes which might include satisfaction with the police. Conclusion. Unfortunately, this body of social psychology research has been largely ignored by individuals who have suggested that satisfied consumers of police services are more likely to engage in coproductive behaviors than are unsatisfied individuals. The present study incorporates many of the aforementioned suggestions, along with findings discussed in the next section concerning the determinants of coproduction. It is believed that such will provide a more complete understanding of the nature of the relationship between citizen attitudes toward the police and pubic willingness to help police perform the law enforcement role. DETERMINANTS OF COPRODUCTION There appears to be agreement in the literature that citizen coproduction of police outputs can only be 66 beneficial to police performance. However, less consensus exists concerning those factors that cause individuals to move from being solely consumers of police services to also acting as coproducers. This section focuses on factors existing research has suggested may be related to coproduction. Attitudes. As previously noted, some of the researchers that have examined the determinants of public attitudes toward the police contend that more satisfied consumers are more likely, than less satisfied consumers, to engage in behaviors that contribute to police performance. Thus, they hypothesize that there is a positive relationship between the two variables. Contrary to the above . suggestion, it also seems quite plausible that the relationship between satisfaction and coproduction is not as straightforward as has previously been hypothesized. In fact, the two constructs may not even be positively related to one another. The contention that attitudes and behaviors may not be related to the extent suggested is premised on the belief that the reasons why someone holds a positive or negative attitude may influence the relationship between attitudes and behaviors. Citizens may be either satisfied or dissatisfied for a variety of reasons. Furthermore, all satisfied consumers of police services probably do not possess their stated attitude for identical reasons. For instance, some people may hold positive attitudes toward the 67 police because they believe that local law enforcement personnel are effectively controlling disorder within their community. Other individuals, may base their positive attitudes on the belief that police are not very effective in controlling crime and disorder, though at the same time they think the police are doing everything they possibly can under the circumstances. In this situation, the positive attitudes result because the police department is meeting the expectations of this group of citizens. On the other hand, citizens may hold negative attitudes toward the police for a variety of specific reasons. Police constituents may believe that the police department should implement different operational strategies, provide more foot patrol officers, be more receptive to community input, and/or concentrate on different crimes. Specific concerns may therefore influence more general attitudes. Finally, citizen attitudes may be the result of long standing distrust of the police and may not be grounded in any specific concerns about operational strategy or department effectiveness. This list of reasons why citizens may possess certain attitudes is not intended to but exhaustive, but is noted because citizen attitudes have underlying dimensions that may influence willingness to engage in coproductive activities. Several scenarios are provided to highlight how reasons why someone possesses a specific attitude may influence the relationship between attitudes and behaviors. First, 68 unsatisfied citizens may be just as likely as satisfied community members to coproduce. If citizen dissatisfaction results because of beliefs that present police operational strategies are unable to cope with problems of disorder and crime, then citizens may decide to help the police by directly providing information to officers. In this situation citizens are dissatisfied with the manner in which the police are operating, though they still believe that the police may be effective if they change tactics and/or receive help from citizens. However, if individuals are dissatisfied with the police and do not believe that the agency or officers are receptive to input from individual community members, these people may turn to community groups and provide the group with information. Finally, community members who do not have confidence in the ability of the police to control crime may assume responsibility for crime prevention themselves. As such, they may engage in other forms of collective action, such as, citizen patrols or marches directed at disrupting specific types of criminal activity. In each of these instances, less than favorable attitudes toward the police may be related to behaviors that are supportive of local law enforcement. Admittedly, it is also possible that less satisfied residents may decide not to act privately or in conjunction with other community residents. Second, satisfied individuals may decide to not participate in activities supportive of the police. They 69 may choose not to act because they believe the police are doing a satisfactory job and do not need citizen help. If citizen satisfaction is premised on the belief that the police are doing all they can, or better than expected, satisfied consumers may decide that it is not worthwhile to provide the police with information. Furthermore, if the citizen believes that collective action is not likely to be effective, then the person may also choose not to select this option. Alternatively, satisfied community members may perform the range of activities that some researchers have suggested. Third, factors may intervene and constrain the relationship between the attitude and the decision to engage in behaviors supportive of the police. For instance, citizens holding positive attitudes toward the police may believe that others will act and that there is therefore no reason for them to get personally involved. Also, opinions about the efficacy of citizen action may influence the decision to participate in the policing process. In each case there is reason to believe that the relationship is not as direct as has been suggested. Neighborhood-Level Variables. In addition to attitudes toward the police, a number of neighborhood level variables may influence citizen willingness to partake in behaviors supportive of the police. In fact, the neighborhood context in which attitudes arise may mediate the relationship between satisfaction and coproduction (Lowery and Lyons, 70 1989; Sharp, 1984). Having a stake in one’s neighborhood is believed to influence a person’s willingness to engage in behaviors supportive of the police (Greenberg et al., 1983; Lavrakos and Herz, 1982; Percy, 1981; Rich, 1981; Washnis, 1976). ‘People who own their home, and thus have a stake in the community, have an investment in the neighborhood and may have "greater motivation to adopt constructive behaviors" (Lowery and Lyons, 1986:333; Haeberle, 1987; Rosenbaum, 1987; Cox, 1982). Residents who intend to remain in the neighborhood for a period of time also have a stake in the community. These people may be more willing to engage in behaviors that improve the quality of life in the area than people who see the locale as a place of temporary residence and plan to "exit" (Lowery and Lyons, 1986; Sharp, 1984: Hirschman, 1970). Relatedly, a person’s belief that certain conditions within the neighborhood are problems may influence her/his willingness to engage in behaviors intended to alleviate or diminish the impact of these problems. These conditions may involve non-police matters (e.g., lack of adequate housing or high unemployment) or police issues (e.g., visible drug use or high rates of property destruction). Lavrakos and Herz (1982) note that neighborhood problems must reach a certain threshold before they become a salient factor in the decision to engage in coproductive activities (and probably before they influence attitudes toward the police). 71 Naturally, a person who does not see neighborhood conditions as problematic would probably not get involved in helping police through community action or individual behavior. The relationship between neighborhood level variables and citizen behaviors also may not be as direct as some presume. Several researchers contend that merely because a person intends to remain a community resident does not automatically imply that the individual will become an active participant in the coproduction process (Lyons and Lowery, 1986; Sharp, 1984; Hirschman, 1970). Also, just because a person perceives neighborhood conditions as problematic does not mean the resident will act to remedy the situation. A citizen who has decided not to leave (i.e., not exit) may choose one of three avenues to pursue (Lyons and Lowery, 1989; Lowery and Lyons, 1986; Sharp, 1984). First, the community member may decide to participate in community behavior and/or individual action in response to the neighborhood problem, especially if the individual has a stake in the area. This type of participation has been labeled the "voice" option (Hirschman, 1970). The second option involves individuals who decide to remain despite dissatisfaction with neighborhood conditions in the hope that conditions will improve. This alternative has been termed the "loyalty" option (Lowery and Lyons, 1986; Sharp, 1984). A "loyalist" may eventually decide to become involved in coproduction activities. Finally, a person may remain in the 72 neighborhood, not expect improvement in community conditions, and not partake in actions intended to correct problematic conditions (see the discussion of "neglect" by Lowery and Lyons [1986] and "non loyalists" by Sharp [1984]). Collective Action. Existing research on collective behavior suggests that most people are not joiners of community groups. Furthermore, most people never participate in community based activities intended to combat neighborhood crime (Rosenbaum, 1987). Some writers suggest that these findings are not surprising, since not all neighborhoods have community groups, nor are they suffering from substantial enough problems to cause people to join neighborhood groups (Lavrakos and Herz, 1982). However, even when presented with the opportunity to join existing organized community organizations most people do not partake in formal community group activities (Gallup, 1982). Thus it should not be expected that a large number of respondents in the present study have participated in collective behaviors, though it remains possible that those with more positive attitudes toward the police may also be those individuals that have engaged in forms of collective coproduction. Unwillingness to become involved in community based police related activities may occur for several reasons. Individuals may not join in these activities because of factors that constrain the ability of people to participate: 73 factors such as a lack of time, employment that conflicts with the time most community activities take place, and having young children at home, which interfere with the ability to participate in some forms of community behavior. Citizens may also choose not to expend the resources (time and effort) necessary to get involved. The fact that community groups provide "public goods" may be an additional reason why people do not join community groups (Olson, 1971). Namely, public goods are benefits that are indivisible in that if the benefits are provided to anyone they must be provided to everyone, irrespective of whether the beneficiary participated in the production of the benefit. Thus individuals may decide not to get involved in community activities because they can receive the same benefits (public goods) without having to absorb the costs (e.g., time, effort) associated with participation (Rich, 1980; Olson, 1971). In other words, individuals may see no advantage to involvement in private or collective activities when benefits obtained from such conduct will become available to all neighborhood residents whether they participated in the process or not. Public unwillingness to get involved may be exacerbated by the fact that identification with the police or a community activity may have negative repercussions for the participant. Public opinion concerning a neighborhood matter likely ranges from approval to disapproval. For instance, drug sales involve a number of willing buyers and 74 sellers. Loitering on streets or in vacant lots involves willing participants. Calling the police or participating in neighborhood activities intended to remedy or curtail these perceived problems may lead to the person being identified as a "snitch" or more violent reprisals by people who are not supportive of the person’s contacting of the police. As such, a person may decide not to get involved when she/he can acquire the same benefits if others act without having to subject her/his self (and possibly family) to these potentially negative repercussions. ngoqraphic Variables. Research findings relative to the impact of demographic variables on the decision to coproduce are mixed. Demographic variables have been found to be related to citizen coproduction, though the impact of status characteristics varies depending on whether the coproductive behaviors involve private or public action. One general finding of existing research is that neighborhood based collective crime prevention is more likely to be engaged in by certain types of individuals and in some types of neighborhoods (Rosenbaum, 1986; Greenberg et al., 1983; Lavrakos and Herz, 1982; Washnis, 1976). More specifically, Haeberle (1987) noted that as income, education, and occupational status of neighborhood residents increase so does the likelihood that someone will participate in collective activity. Thus, collective activity is often believed to be more common in communities 75 comprised of middle to upper class individuals. In these neighborhoods there is likely to be more consensus about matters that are problematic and the possible remedies that may be engaged in to alleviate the problem. Furthermore, these areas are often organized to a greater degree than lower status communities, making collective action a more viable option available to residents. However, while the typical description of participants in collective activities are as noted above, several studies have indicated that the relationship between status characteristics of participants and collective action are not always as suggested. For instance, Rich (1981) noted that since upper class individuals have the means to purchase private police services they may be less likely to personally act in a coproductive capacity. The purchase of private services is a choice that is not available to lower class residents. Thus, residents in upper class neighborhoods may not personally get involved in these types of coproductive activities. Several additional studies also question the veracity of statements concerning the "typical" participant in collective behavior. Lavarakas and Herz (1982) noted that "race is the characteristic that shows the most striking differential patterns between participators and nonparticipators" (p. 491). In their study, minority groups (African-Americans and Mexican-Americans) were over represented as participants, while Anglo Americans were much 76 less likely to participate in community based anti-crime activities (also see Sharp, 1984). This result was alleged to occur for several reasons; minorities have an activist tradition, there is a greater likelihood that they live in high risk neighborhoods, and these individuals often are less confident in the abilities of the police (Lavrakas and Herz, 1982). In combination, these factors may lead certain groups of people who want to help police to turn to collective action or individual conduct instead of private behaviors. In addition, members of minority groups are often more attached to their neighborhoods and neighborhood groups than formal governmental organizations (Haeberle, 1987). Distrust of local government and a lack of faith in its willingness to invest in solutions to the problems of the less powerful may be reasons why members of minority groups turn to neighborhood based groups. Additionally, neighborhood groups, composed of members with similar ethnic and/or socioeconomic status, may provide a more supportive atmosphere in which to interact. Most people feel more comfortable associating with members of their own group (Tajfel, 1982). Finally, within such community groups there may be greater agreement as to the problems and solutions that need to be and can be implemented successfully in the area. These assumptions do not conclusively indicate that participants are likely to be of a specific socioeconomic 77 status, though they do question the stereotypical characteristics that are often applied to participants in collective anti-crime activity. Based on the aforementioned, it could be argued that the findings only indicate that members of minority groups in upper class neighborhoods are more likely than whites in these same neighborhoods to participate in community groups. Thus race is a more important characteristic than is social class. However, other research also suggests that participation is most likely a reaction to deteriorating community conditions that occur in urban environments (Rosenbaum, 1987). Research findings concerning the impact of age on collective coproduction indicate that it may be curvilinear. Participation in collective activities appears to increase until age fifty and then decrease. Specifically, the elderly (over age 65) comprise the age bracket least likely to become involved in collective action. Individuals between the ages of thirty and fifty-five represent the group most likely to engage in neighborhood anti-crime activities. Members of the eighteen to thirty-four age group are less likely than people within the next bracket to engage in coproductive behaviors, while at the same time they are more likely than the elderly to partake in collective behaviors. The eighteen to thirty-five age bracket represents the portion of people’s lives when many individuals have a significant vested interest in the community since they are often purchasing homes and raising 78 families (Haeberle, 1987: Greenberg et al., 1983; Lavrakos and Herz, 1982; Washnis, 1976). This investment in the community is believed to be a factor that acts as a catalyst for both private and collective behaviors. Gender and coproduction has received only limited attention in the community action literature, and even less mention in the private action research. In general, men are more likely than women to participate in public actions (Lavrakas and Herz, 1982). However, participation rates vary depending on the involved activities. In activities that conform to traditional male roles, such as neighborhood escorts and citizen patrols, men are more likely than women to be participants. When participation is measured in terms of attendance at anti-crime community meetings or whistle- stop programs, the disparity in participation rates is greatly reduced. Conclusion. Research on coproduction falls into three areas. One area involves research on attitudes of citizens towards the police. The second area includes community action research which has been concerned with neighborhood level factors that influence the decision to participate, or not participate in behaviors that might improve the quality of life in an area. The third line of research involves the literature on collective action. This study brings together these areas of research by examining the contribution of attitudes, neighborhood level measures, and the problems associated with collective action to more fully explain 79 coproduction by citizens. CONCLUSION One of the emphases of the preceding chapters is that our present body of knowledge concerning the relationship between citizen attitudes toward the police and the coproduction of police outputs has been informed by researchers pursuing several distinct lines of inquiry. It appears that the work of researchers following one line of inquiry has often ignored, or at a minimum, been unconcerned with the findings of individuals pursuing one of the other paradigms. While the focus of one’s study naturally determines the issues and existing research of interest, there is such an overlap between these areas that the failure to incorporate other existing relevant research is unfortunate. More specifically, our present understanding of the determinants of citizen attitudes toward the police and their connection between attitudes and citizen behaviors has been uninformed by the social psychology literature. At the same time, social scientists focusing on citizen participation in community anti-crime efforts have generally ignored the existing attitude research and its assumption that citizens with favorable attitudes toward the police are more likely than others to engage in conduct that supports performance of the police role. Finally, while the body of literature concerning collective anti-crime efforts is constantly increasing very little empirical attention has ' 80 been directed at the correlates of private coproduction. 81 ENDNOTES 1. See Worden (1989) for an example where attitudes of police officers were found to have only limited explanatory value. 82 CHAPTER 4 MODELING CITIZENS’ ATTITUDES AND COPRODUCTIVE BEHAVIORS: LINKAGES BETWEEN VARIABLES Most empirical assessments of citizens’ attitudes toward the police have proposed a series of hypotheses that are premised on the existence of a direct unidirectional relationship between predictors and the dependent variable of interest (Skogan, 1991; Flanagan, 1985; Erez, 1984; Apple and O’Brien, 1983; White and Menke, 1982; Koenig, 1980: Bell, 1979; Thomas and Hyman, 1977; Black, 1974; Smith and Hawkins, 1973; Jacob, 1971; Boggs and Galiher, 1965). At the same time, the coproduction literature has also followed a similar strategy and explored the direct relationship between explanatory variables and coproduction (Haeberle, 1987; Greenberg et al., 1983; Lavarakas and Herz, 1982; Percy, 1979; Washnis, 1976; Eagle and Schneider, 1975). Using correlation and ordinary least squares regression techniques, researchers following both research paradigms have focused on the relationships between selected variables. Two studies run counter to this general trend. Parks (1984) developed a causal model that explored the linkages between objective and subjective determinants of both assessments of police response time and more general citizen attitudes toward the police. Scaglion and Condon (1980), using a number of causal models, empirically examined the direct and indirect effects of demographic variables on 83 citizen evaluations of their experiences with police officers. They also assessed the influence of demographic variables and citizen evaluations of personal police-citizen interactions on general levels of citizen satisfaction with the police. Each study made substantial contributions to our understanding of the relationships that exist between variables contained in each of their predictive models. Furthermore, the two research endeavors represent the only empirical assessments of some of the untested assumptions that have been offered relative to the indirect influences that may be exerted by variables that are often contained in existing studies.1 This chapter is devoted to the development of conceptual models of the determinants of citizens’ attitudes toward the police and of the factors that influence the public’s participation in coproductive behaviors. The models have been derived from previous research and specify the causal influences a number of variables may directly and indirectly exert on both attitudes and behavior. Model building is "intended to increase our understanding of social phenomena in a greater fashion than simply correlating independent and dependent variables" (Asher, 1983: 9). Increased understanding is achieved in several ways. First, model building requires specification of not only the relationships between the independent and dependent variables in each model, but also make explicit the relationships among explanatory variables. Second, 84 models specify causal relationships between constructs. Accordingly, the models portray the linkages between variables and emphasize the fact that many of the variables used in prior studies may have direct and indirect effects on citizens’ attitudes and/or their behaviors. Third, model building also requires that the researcher account for the temporal ordering of variables in the predictive model. Thus, panel data are ideally suited for the estimation of models. While correlations provide valuable information about the relationships among variables, when this statistical technique is used with cross-sectional data the correlations provide only limited information about the causal relationships between these same variables (Asher, 1983). In contrast, panel data when used to estimate predictive models provide more powerful information concerning causality. MODEL 1: ATTITUDES TOWARD THE POLICE The first model portrays the determinants of citizens’ attitudes toward the police. This section of the chapter contains the definitions of constructs substantively related to the formation of citizens’ attitudes toward the police. The temporal ordering of variables and the indirect ‘ relationships ignored in most existing research are also discussed in this section. Demographic Variables. In the model, (see Figure 4.1) the construct demographic variables refers to the race, age, 85 .ii . [if I 21%., ,......-a-- i ..... :..-.:_ 22588365 | ‘\\\\ 'ol.alllp'.||l!llll 08:. uncommom .\ .6 EuEmmomm< oozed 2: wu5wtw> 2:3 muocotocxm 1 039836800 .0 acoumzqgm :. L‘ a 2+: . a “63:; no «a; / 335.2; “38:60 6858582 ESE—Eco cooncoanwfiz .6 acozaoocom ooflaom may mcumsoe moosuwuu< mo mucmcflsuouoo H.v musaflm 86 educational level, household income, and gender of respondents. The impact of demographic variables on citizens’ attitudes may occur in a variety of fashions. Demographic variables may have a direct impact on attitudes toward the police. They may also have indirect effects through citizen perceptions of neighborhood conditions, evaluations of police behavior during police-citizen interactions, and citizen assessments of response time. As Parks (1984) noted, Vdifferent persons may respond differently to the same phenomenon" (p. 121). I The model, depicts demographic variables as having a direct influence on citizens’ evaluations of the police. Existing research has concluded that demographic variables influence both the nature and type of encounters people have with the police. Specifically, studies have found that, in comparison to Anglo Americans, African-Americans and Spanish-speaking Americans were more likely to have police initiated interactions, that the experiences/interactions often involved a threat of sanctions being applied by the police, and that these experiences were generally more intrusive in that they quite often included searches of the involved citizens (Skogan, 1991; Erez, 1984; Jacob, 1971: Hahn, 1971). Furthermore, officer behavior during these encounters was often perceived by citizens as "abusive" (Erez, 1984; Hahn, 1971). Demographic variables thus appear to influence each of the several dimensions of police- citizen experiences. In turn, the socioeconomic status of 87 respondents may "work indirectly through evaluations of service to affect general satisfaction" (Scaglion and Condon, 1980: 490) with local law enforcement. Demographic variables are also portrayed as directly influencing citizen assessments of response time. Percy (1986) and Flanagan (1985) noted that race especially influences expectations of service quality. In addition, Block (1974) claimed that citizen perceptions of police response was conditioned by an individual’s socioeconomic characteristics. The influence of demographic variables on attitudes may therefore occur indirectly through assessments of response time. Demographic variables might also directly impact on becoming a crime victim and respondents’ perceptions of neighborhood problems. For instance, lower class individuals are often subjected to higher victimization rates than upper class individuals (Erez, 1984; Apple and O’Brien, 1983). Individuals with certain status characteristics may also reside in communities that are afflicted by problems that either do not exist, or are not as apparent, in other communities. In each case the impact of demographic variables on attitudes is portrayed as indirect. Neighborhood Context Vapiables. Neighborhood context variables encompass neighborhood income levels and the percent of residents within the neighborhood that are nonwhite. These variables therefore affect the type of 88 people that a person will have contact with at least around their home. Attitudes toward government services are believed to have their origins in the expectations of reference groups that are communicated to and internalized during interactions with group members (Percy, 1986; Christenson and Taylor, 1983; Apple and O’Brien, 1983; Brudney and England, 1982). It is in this manner that neighborhood context directly influences attitudes toward the police. The context in which people reside, and presumably spend a substantial portion of their time, also may directly influence the types of problems they encounter and the victimization fears they possess. For instance, vacant homes, prostitution, and people hanging out in empty lots or on the street are conditions that are probably more common in certain types of neighborhoods than others. Citizen perceptions of these neighborhood conditions as problematic, or not, therefore appear to vary depending on the neighborhood context in which people find themselves. Thus, in addition to the direct impact that neighborhood context may have on attitudes, this construct may also have indirect effects through citizen perceptions of neighborhood conditions. Respondents’ Perceptions 9: Neighborhood Conditions. This construct refers to one segment of the social context. Problems may range from loitering to fear of being victimized within the community and, as noted, the salience 89 of a specific problem should vary by neighborhood context. Christenson and Taylor (1983) suggest that the neighborhood context associated with service delivery directly influences evaluations of the service provider. Accordingly, the Model (see Figure 4.1) indicates that citizen perceptions of neighborhood problems have a direct influence on police related attitudes possessed by community members (Percy, 1986, 1981; Stipak, 1979). Neighborhood problems may be especially influential if the specific matter is one that falls within the purview of the police. Citizen beliefs that neighborhood problems fall within the jurisdiction of the police, would likely deflate (i.e., make them less favorable) attitudes. Likewise, a belief that neighborhood conditions are either not problematic or not the fault of the police would beneficially influence (i.e., make them more favorable) attitudes toward the police. Assessments pf Response Time. This construct refers to citizens’ perceptions of how quickly the police respond when contacted. Percy (1986, 1978) has documented a process during which citizens compare the time it takes the police to respond with expectations they possess about the anticipated conduct. If the police respond faster than anticipated, it has a beneficial impact on citizen attitudes. If the police respond slower than expected, the citizen assessment reduces citizen attitudes. In accordance with Percy’s findings, assessments of response time are 90 portrayed in Figure 4.1 as directly influencing general attitudes toward the police and citizen evaluations of police behavior during experiences with the police. Evaluatione e; Experiences Wipe ppe Police. Evaluations of experiences with the police refers to citizen judgements concerning the behavior of officers during incident specific contacts. These evaluations refer to incidents that occurred during a six month period prior to measurement of citizens’ more general attitudes. Citizen evaluations are portrayed in Figure 4.1 as directly influencing more general attitudes these same members of the public possess towards the police (Skogan, 1991; Percy, 1986; Mastrofski, 1981; Dean, 1980; Scaglion and Condon, 1980). During encounters with the police, members of the public are able to acquire information about the behavior of local police officers. This previously acquired personal information, which has been cognitively processed and stored by the individual, is likely to be retrieved when formulating responses to questions that tap more general attitudes toward the police, especially since this may be the only first-hand information these people possess. Prior Attitudes. In the Model, ATT(t) refers to existing attitudes toward the police.2 Most, if not all, empirical assessments of citizens’ attitudes toward the police fail to account for the impact of pre-existing global attitudes. Barring an unforeseen, intervening event that 91 radically alters someone’s perception of the police, future attitudes may be the same as prior attitudes. This would be especially true if attitudes are fairly stable over time. Figure 4.1 shows pre-existing attitudes as having both direct and indirect effects on police related attitudes of the public. The impact of pre-existing attitudes may be indirect through a person’s evaluation of officer behavior during contact with the police. The cognitive processes associated with stereotyping help explain this assumption. Information received from any source (e.g., during a police- citizen contact) pertaining to a member of a stereotyped group (the police) is processed in accordance with the receiver’s existing stereotypes (Hamilton, 1981). Existing stereotypes may even bias the processing of information so that the information being processed appears to reinforce the existing belief system (Hamilton and Rose, 1981; Snyder, Tanke, and Berscheid, 1977). Thus, an individual may see officer conduct as conforming to pre-existing attitudes the person has concerning the police; an individual with a positive attitude towards the police is more likely to positively evaluate police performance during a specific encounter, while one with a negative attitude is more likely to be dissatisfied with police conduct during such an experience. Victimizations. The construct victimization encompasses the number of times an individual has been victimized by specific illegal conduct. This construct is 92 portrayed in the Model (see Figure 4.1) as influencing respondents’ perceptions of neighborhood problems and evaluations of police performance during the victimization experience, in addition to overall attitudes toward the police. The hypothesized impacts are premised upon several factors. First, someone who has been victimized within his/her neighborhood may fear further similar experiences. Second, a crime victim may believe that certain community conditions are responsible for his/her victimization experience. The victim may now see the condition as problematic. For instance, someone who is assaulted by teenagers in front of a vacant home, may come to believe that the presence of vacant buildings in which people may loiter are problems to which the police should be paying closer attention. In the above situations, victimization experiences may indirectly influence attitudes toward the police by first influencing perceptions of neighborhood problems. Third, many individuals hold the police responsible for maintaining their own personal safety and security. This belief may cause the victimization experience to have an effect on citizens’ ratings of the performance of officers that respond pursuant to the criminal occurrence (Brandl and Horvath, 1991). Thus, an additional indirect effect may occur. This same belief may also cause a victimization to have a direct influence on attitudes. A person who suffers 93 a victimization experience, and especially a personal victimization (Koenig, 1980; Smith and Hawkins, 1973), may transfer the experience to subsequent evaluations of the overall quality of community law enforcement personnel. All other things being equal, respondents who believe that the police are responsible for her/his victimization likely possess less than favorable attitudes toward the police. Attitudes Toward The Police. The dependent variable (ATT t+1) in the Model encompasses citizens’ attitudes toward the police. Attitudes may take several forms. They may include attitudes toward a specified law enforcement agency and/or more specific attitudes relative to a police department’s performance of certain police functions. Hypotheses. The hypotheses to be examined relative to Model 1 are... Hypothesis 1. Respondent’s race influences prior attitudes towards the police, present attitudes towards the police, and evaluations of experiences with the police. White respondents are more likely than nonwhite individuals to have more favorable attitudes towards the police on each of these attitude measures. Hypothesis 2. Respondent’s race also affects citizen perceptions of neighborhood conditions, and in turn attitudes towards the police. Hypothesis 3. Older individuals are more likely than younger persons to have more favorable attitudes toward the police. Hypothesis 4. Age influences prior attitudes towards the police and evaluations of experiences with the police. Older respondents are more likely than younger individuals to have more favorable attitudes towards the police on each of these attitude measures, which in 94 turn, positively influence present attitudes towards the police. Hypothesis 5. Respondents in higher income neighborhoods are more likely than respondents from lower income neighborhoods to possess favorable attitudes towards the police. Hypothesis 6. Income influences prior attitudes towards the police and evaluations of experiences with the police. Wealthier respondents are more likely than less wealthy individuals to have more favorable attitudes towards the police on each of these attitude measures, which in turn, each positively influence present attitudes towards the police. Hypothesis 7. Females are more likely than males to hold more favorable attitudes toward the police. Hypothesis 8. Victimization experiences affect perceptions of response time and evaluations of experiences with the police in a negative fashion. Hypothesis 9. Individuals who have been victimized are less likely than people who have not had such experiences to hold favorable attitudes toward the police. Hypothesis 10. Prior attitudes influence both present attitudes and evaluations of experiences with the police in a positive direction. As prior attitudes increase, evaluations of experiences increase and so do present attitudes. Hypothesis 11. Negative evaluations of the performance of police in specific experiences have a negative influence on attitudes towards the police, while positive evaluations of the performance of police have a positive influence on attitudes towards the police. Hypothesis 12. Individuals who believe that the police responded faster than expected are more likely than individuals who saw the police response as slower than expected to hold more favorable attitudes towards the police, whether present attitudes or perceptions of experiences with the police. Hypothesis 13. Individuals who see neighborhood situations as problematic are less likely than individuals who do not hold such opinions to have favorable attitudes toward the police. Hypothesis 14. Neighborhood context variables influence attitudes towards the police. As the percent of 95 nonwhite residents increases, attitudes toward the police decrease. At the same time, as neighborhood income increases, attitudes toward the police increase. Conclusion The model as portrayed in Figure 4.1 provides a description of the formation of citizen attitudes that is more complex than predictive models contained in most prior empirical assessments of citizens’ attitudes. Not only does the model incorporate existing attitude research, but it also illustrates that many constructs previously perceived as having only direct influences may also contribute significantly by influencing other variables that ultimately influence the formation of public attitudes toward the police. Furthermore, the Model provides a role for prior attitudes of citizens, a variable that has been conspicuously absent in existing studies. MODEL 2: DETERMINANTS OF COPRODUCTION An expanding number of studies have addressed the issue of citizen coproduction of police outputs. While much of this literature has focused on defining what coproduction entails, several studies have attempted to empirically assess the correlates of coproduction, while others have only theorized about the factors that explain coproduction. The model depicted in Figure 4.2 specifies the relationships among a number of variables mentioned in existing studies. At the same time, the model also incorporates the premise of citizen attitude towards the police research that citizens’ attitudes are related to citizen willingness to 96 : 3. toga—Lem coop—62:36: .\ 2: E 3.5m \\ _ 9:638; .o \\ mcozcoouon. cosmos—5560 .6 5:568: :ofluuscoumou no wucmcflauwuoo m.v ousoflm E 2.3:? 3365; ofiaaconoD 97 engage in behaviors that assist the police. If citizen behavior occurred in a vacuum, unaffected by the environment, one might expect attitudes to explain a substantial portion of coproductive behaviors. However, such is not thg'case. Instead, citizen behavior occurs within an environment composed of factors that may enhance or constrain the relationship between attitudes and behavior. The objective of this model is to specify the roles these factors play in determining citizen coproduction of police outputs. SLQEQ ip pee Neighborhood. In its most general sense, this construct refers to respondents’ attachment to their community. As people become more attached, whether through ownership of their residence or income generating property within the immediate area, they acquire more of a stake in the neighborhood. Likewise, people who have decided that they will remain in the community for a period of time would be more attached to the neighborhood than parties who see the area as solely a temporary residence. According to the model, the influence of stake in the neighborhood on coproduction may occur in one of two ways. First, stake in the neighborhood may indirectly influence citizen behaviors by influencing perceptions of neighborhood problems. For example, people who have a stake in the neighborhood are probably more sensitive to neighborhood matters than individuals who do not have such an interest. As such, they may view the matter as problematic and become 98 involved in remedying the situation. Second, stake in the neighborhood may directly influence coproduction. This hypothesized relationship is premised on the assumption that individuals with a stake in the neighborhood have an interest in improving the quality of life in the community.3 This interest may be based upon economic (e.g., to maintain property values) or psychological (e.g., feel safer while raising a family) factors. Furthermore, people with a stake in the community are probably focused on both the immediate and future welfare of the community. When these factors associated with stake in the neighborhood are combined (interest and focus of residents), they may be the catalyst for citizen activities that are intended to have a positive impact on neighborhood safety and security. Whether these actions actually have such impacts is another question. Perceptions 9: Neighborhood Problems. This construct refers to citizen perceptions about whether certain neighborhood conditions are problematic. As noted in the previous discussion concerning Model 1 (Figure 4.1), and as can be seen in Figure 4.2, citizens’ beliefs about neighborhood problems are portrayed as influencing the attitudes of community members towards the police. In addition, perceptions of neighborhood problems are believed to indirectly influence coproduction through these same attitudes. At the same time, perceptions of neighborhood problems 99 are portrayed as exerting a direct influence on the willingness of people to engage in coproductive behaviors. The existence of neighborhood problems, or at least the perception that problems exist, that require citizen efforts to remedy them may be a necessary condition before people become involved in the coproduction process. The failure of residents to perceive conditions as problematic would presumably lead people to inaction since there would not be a specific target towards which citizen behaviors might be directed. Demographic Variables. In this model, as in the previous model, demographic variables include the race, gender, household income, educational level, and age of respondents. The influence of demographic variables on coproductive behaviors may come about in two ways. First, demographic variables may directly influence decisions to participate in behaviors supportive of the police. For example, existing research has suggested that status characteristics are associated with participation in neighborhood-based anti-crime efforts (Haeberle, 1987; Lavrakas and Herz, 1982). Specifically, many researchers believe that lower status individuals are less likely to engage in coproduction. Second, and even more likely, demographic variables influence coproduction indirectly through citizens’ attitudes toward the police. As previously noted status characteristics have been found to influence citizen 100 attitudes toward the police. Members of the public who hold negative attitudes toward the police may turn to their local community group and act in a collective capacity. If no such group is in existence, the person may decide to either act alone or not act at all. Attitudes Toward the Police. In conformity with the assumptions of prior research on citizens’ attitudes toward the police, existing attitudes are depicted in Figure 4.2 as directly influencing behaviors that support the police. In addition, the model suggests that these attitudes influence subsequent attitudes. This assumption is discussed in greater detail with reference to the first model. Feasibility 9; Community Action. The feasibility of neighborhood action refers, first, to whether residents are of the opinion that actions by citizens are effective in remedying problems. Citizens who do not believe that actions by community members are effective in remedying neighborhood conditions would likely decide not to subject themselves to the costs associated with behaviors perceived as without value. One would anticipate that the feasibility of community action would directly influence decisions concerning whether to engage in the coproduction of police outputs. Whether community residents believe they live in a neighborhood where people help each other might also shape their perceptions of the feasibility of neighborhood action. People who believe that others are not likely to help remedy a situation may turn to private actions, while collective 101 behaviors may be supported by opposite beliefs. Likelihood Others Wiii Age. This construct refers to citizen views on whether other residents will act to remedy conditions within the neighborhood. Prior research has suggested that this construct has a direct influence on decisions to act (Olson, 1971). Specifically, citizens are less likely to act if they believe that they can share in the benefits of the behavior of others. Since coproduction produces a public good, individuals may be more likely to decide not to partake in certain behaviors if they perceive that their neighbors will act to remedy the problematic condition in their residential area. Thus, the public goods literature suggests a hypothesis that is contrary to that discussed in the preceding paragraph on the feasibility of collective action. Coproduction. In the model, coproduction encompasses the behaviors of citizens that assist the police in the production of agency outputs. These may be performed by citizens as individuals or collectively in conjunction with existing community groups. As can be seen in Figure 4.2, coproduction may be influenced by, or result from, all of the aforementioned constructs. Hypotheses The hypotheses for this portion of the study are .... Hypothesis 1. Favorable attitudes will be positively and significantly related to citizen behaviors. The magnitude of the relationships should be greatest between the individual behaviors and the more specific attitudes. 102 Hypothesis 2. Scores on the attitudes toward the police questions will be more highly correlated with scores on the behavioral indexes than with performance or nonperformance of each of the separate behaviors from which the indexes were derived. Hypothesis 3. Having a stake in one’s neighborhood influences a person’s perception of neighborhood problems and coproduction. Individuals who have a stake in their neighborhood are more likely to be involved in collective behaviors and behaviors involving the police. Stake in the neighborhood is also positively related to perceptions of neighborhood conditions. Hypothesis 4. Citizens who see neighborhood matters as problems will be more likely to get involved in helping police through community action or individual behavior, than will individuals who do not perceive these matters as problematic. Hypothesis 5. Individuals who expect to live in the neighborhood a year from now will be more likely to engage in coproduction than will individuals who expect to move. Hypothesis 6. People who perceived neighbors as supportive (versus go their own way) will be more likely to engage in collective actions to remedy problems in their neighborhood. Hypothesis 7. Citizens who believe that neighbors would report observed activity to the police will be less likely to act on their own. This should apply to both collective action and individual behavior. Hypothesis 8. Nonwhite respondents are more likely to engage in collective self-help than private action. Hypothesis 9. Older individuals (people over fifty) are less likely to get involved in collective and private action than are younger individuals. Hypothesis 10. Income and education are positively related to attitudes and coproduction behaviors. As income and educational levels increase, attitudes towards the police become more favorable and increases in coproduction activities occur. Conclusion The model as portrayed in Figure 4.2 provides a description of the effect that attitudes and other variables may have on citizen willingness to engage in 103 the coproduction of police outputs. The model represents an attempt to bring together the literature on coproduction, collective behavior, and community participation. In addition, the model suggests that many of the constructs previously perceived as having only direct influences may also contribute significantly by influencing other variables that ultimately influence coproduction. 104 ENDNOTES 1. For instance Apple and O’Brien (1983) suggest that neighborhood composition influences a person’s perception of neighborhood conditions though they do not test this assumption. Likewise, Christenson and Taylor (1983) note that the social context influences evaluations without specifying how such occurs. 2. These attitudes were measured six months prior to the dependent (attitude) variable in the model. 3. All people within a neighborhood have at least some minimal stake in the community. For example, a person who rents living quarters within the community and expects to live there only a short period of time is probably concerned about matters that impinge on her/his immediate personal safety. However, this same person has only a limited stake in the community since it is much easier for this person to exit form the area since her/his attachment is limited. A person who owns her residence has a greater attachment since it also includes an economic interest. 105 CHAPTER 5 THE PRESENT STUDY: PROCEDURES AND METHODS Data for the present study were collected as part of a larger project that focused on narcotics enforcement in four parts of the city of Detroit. A three-wave telephone survey1 was conducted to collect data concerning community members’ perceptions of neighborhood conditions, their participation in community activities, attitudes toward the police, residents’ participation in behaviors supportive of the police (directly or indirectly by contributing to community safety), and data concerning respondents’ demographic characteristics. This chapter proceeds in the following manner. First, the methodology used to collect the data is addressed. The second section of the chapter discusses operationalization of the variables used in the citizens’ attitudes toward the police models and the statistical techniques utilized in estimating the models. The final section follows a similar format except that all discussions pertain to the coproduction component of the study. Sample Selection The larger project involved enhanced narcotics enforcement in selected areas of the city of Detroit. These areas were selected because had sufficient drug-related activity to warrant the use of enhanced enforcement. Since a portion of the larger project focused on citizen perceptions of this police strategy in the selected areas, a 106 cluster sampling procedure was utilized to ensure that all blocks within each of the selected neighborhoods were represented. In order to achieve this objective, households were randomly selected from each blockface in each of the four selected areas. This was accomplished by using a reverse phone directory and dividing each street to correspond to available addresses on each block. Three households were then randomly selected on each block, with two of these households acting as alternates. If there was drug enforcement activity on the block during the nine months prior to beginning the project, the selection procedure was repeated producing an additional household on these blocks. This process resulted in the selection of another 186 households.2 Since the project involved a panel design, individuals who participated in the first wave were retained for inclusion in the second wave, as were second wave participants for the third wave. This requirement was effectuated in the following manner. Each respondent was advised, at the conclusion of the first-wave interview, that her/his name (or an alias) was needed in case the interviewer’s supervisor wished to contact the person later to clarify collected information. If the respondent was the first person contacted on the blockface then that person and the two selected alternates were retained in the sample for the next wave. However, if the respondent was not the first 107 person contacted on the block, and was thus an alternate, additional households (that became alternates) were selected for the next interview period to replace households that were not successfully contacted. In the end, three households were selected from each blockface, with two of the households representing alternates.3 The interviews were conducted by Survey Research Center in the Social Science Research Bureau at Michigan State University. Prior to the start-up of the interview process for each wave, interviewers were required to attend a meeting to discuss the survey instrument and the larger study. At these training sessions the interviewers were advised, by myself, of the nature and objectives of the drug enforcement project, the types of data to be collected during the telephone survey portion of the project, the importance of the survey information in relation to the achievement of certain project objectives, the personally sensitive (to respondents) nature of some of the information to be collected, and the issues that may surface during the interview process. Interviewers were also provided the opportunity to ask questions about the project and provide input concerning question wording. Prior to each wave, the interviewers conducted practice interviews under the supervision of staff members of the Social Science Research Bureau on their Computer Assisted Telephone Interview (CATI) system. Wave 1 interviews were conducted during February, 1990. The second wave interviews were completed in October, 108 1990, while the final wave conducted in April of 1991. The interviewer, after explaining the general purpose of the study,4 requested to speak with someone eighteen years or older who had resided in the home for at least six months. If someone meeting these characteristics was not home (i.e., no answer or no one 18 or older home), a return call was scheduled. This process was repeated until a refusal was received or numerous unsuccessful contacts had been attempted. As a general rule, eight to ten call-backs per household were attempted before a decision was made whether to replace a household with an alternate. Telephone interviews were attempted between the hours of 9:30 A.M. and 9:00 P.M. during the week. Interviews were also conducted on Saturdays and Sundays during the afternoon and early evening. While this interview schedule was utilized because it accommodated the work schedules of employees of the Social Science Research Bureau, it also was expected to achieve another sampling objective. By attempting interviews throughout the week and on weekends it was anticipated that contact could be achieved with members of households who worked traditional hours (i.e., 8:00 A.M. to 5:00 P.M. on Monday through Friday). Since interviewing during weekdays only has the potential to result in a sample that includes an over-representation of females and older individuals (Lavrakas, 1987:18), it was hoped that by contacting households during periods when males were more likely to be in the home to answer the telephone this 109 potential sample bias could be minimized. There were also several socioeconomic characteristics associated with the jurisdiction under study that suggested that the sample would be composed of a representative sample of community residents. First, the city of Detroit suffers from an above average unemployment rate. As such, potential respondents of both sexes are less likely to be at work and, at the same time, more likely to be in the home at a time when contact with the household is attempted. Second, Detroit’s employment history includes traditional "blue— collar" manufacturing jobs that involve shift work. While many males may work day shifts, there was also reason to believe that male and female household members involved in this type of employment would be in the home at various times throughout the day, and not only after 5:00 P.M.. In light of these factors it was possible that the community sample would not include an over representation of females, even though a more complex sample selection process was not utilized (Kish, 1967). Telephone surveys have both weaknesses and strengths. The major weakness of this survey methodology is that it excludes households, and the individuals therein, if they either do not have a telephone or have an unlisted telephone number. Whether this introduces some type of response bias is unknown. However, given that ninety-four percent of the people in the United States have phones (Babbie, 1983), then the only group one for sure knows has been excluded is the 110 homeless, a group removed from most sampling methods. As to those parties with unlisted telephone numbers, the issue is more complicated. The number of unlisted telephone numbers may be quite large in metropolitan areas (Dillman, 1978). In fact, information provided by the Detroit telephone company indicated that thirty-eight percent of the people with telephones had unlisted numbers. Research by Brunner and Brunner (1971) noted that there were differences between people who have listed and unlisted numbers. They found that individuals most likely to have unlisted numbers were often less educated, younger, more likely to be divorced, and more likely to be employed in blue collar jobs. If one applies the findings of Brunner and Brunner (1971) to the present endeavor, they indicate that blue collar and lower status individuals were most likely to be excluded from the potential sample by the failure to contact people with unlisted numbers. The only way to circumvent problems of unlisted numbers when using telephone surveys is through random-digit dialing. While this method would likely pick up unlisted numbers, a question remains as to whether respondents with unlisted numbers would contribute information that differs from that elicited from people with listed numbers. This is noted because the reasons why people have unlisted numbers -- people fearful of being burglarized (or victimized through other activity), individuals who do not want to be bothered by telephone solicitors or harassed by 111 obscene calls -- suggest that these individuals do not want to be bothered by unsolicited phone calls. When coupled with the sensitive nature of the data being collected, one would anticipate a high refusal rate with people having unlisted telephone numbers. This defect would appear to be more detrimental than other response bias problems. In addition, random digit dialing would be incompatible with the cluster sampling procedure being utilized. People who had recently moved into the neighborhood were probably not listed in the telephone book, even though the reverse directory is published at six month intervals. Since the survey instrument was intended to collect data concerning community members’ perceptions of neighborhood issues and the quality of life in the area, participation in the survey was restricted to household residents who had resided in the same location for at least six months. Thus, the fact that persons who recently moved into the neighborhood were excluded, both by survey decision rules and by not being in the phone book, was not viewed as problematic. While homelessness, unlisted telephone numbers, and transience are three factors that reduce the potential households (and ultimately respondents) available for inclusion in the sample, there is no reason to believe that the sampling process introduced substantial non-response bias. In addition, other methods for selecting the sample would probably not have reduced any bias related to these 112 three factors. Specifically, methods which rely on selecting respondents in a household based on some formula (i.e., most recent birthday, etc.) would not lead to the inclusion of homeless people or people with unlisted numbers. There are also several advantages to telephone surveys. Specifically, telephone surveys provide a certain amount of anonymity. This would seem especially significant for the present endeavor in that respondents were questioned about the activities of police, drug sellers and buyers, and their participation in activities supportive of the police - all sensitive topics. Since the respondents were not required to face the interviewer and were assured that their responses were not going to be disclosed to anyone, respondents did not have to be concerned with being identified with the interviewer. The completion rate of phone surveys should therefore be higher than if face to face interviews had been conducted, especially in inner- cities (Kalton, 1983:66).5 Furthermore, since the identity of respondents was protected, as were their responses, the participants might be more honest in providing what may in some instances be socially disapproved responses (Babbie, 1983:235-236). Lavrakas (1987:18) contends that respondents provide reasonably accurate information when questioned in telephone surveys since it is much easier to refuse to participate, than to waste time attempting to trick the interviewer. As 113 evidenced by the length of the interviews (average length between 20 and 25 minutes for each wave), many respondents apparently were very willing to respond to the questions asked. Thus it appears that the use of telephone interviews may have facilitated participation. Procedure The present study involves three interrelated research issues. The first issue pertains to the determinants of citizen attitudes toward the police. The second issue entails an examination of factors which are related to, and influence, citizen willingness to engage in the coproduction of police outputs, with emphasis on the influence of citizen attitudes toward the police on the decision to become a coproducer. The final research issue involves an attempt to disentangle the causal relationship between attitudes and behaviors; namely do prior attitudes influence subsequent attitudes and/or do attitudes toward the police influence behaviors. This section discusses the methodology utilized in the present study to analyze these three issues. The chapter proceeds by first examining the operationalization of variables contained in Model 1 (see Figure 5.1), the distribution of responses for applicable questions, and the statistical techniques to be utilized. This process is repeated during the discussion of variables contained in the second model. 114 MODEL 1: DETERMINANTS OF CITIZEN ATTITUDES TOWARD THE POLICE The dependent variables to be examined in this portion of the study are identified in Table 5.1. The independent variables are displayed in Table 5.2. These two tables also contain the coding schemes, frequencies, means, and standard deviations, where applicable, for each variable. Dependent Variables Attitudes Ieward the Police. In the present study, the determinants of three attitudes toward the police are examined; namely, global attitudes (Model 1.1), drug-related attitudes (Model 1.2), and attitudes toward the job police are doing maintaining order (Model 1.3). All data used to operationalize the three dependent variables were collected during the administration of the final wave interview. The dependent variables were operationalized through three separate questions. The attitude object varies in each of the inquiries. The first question, a global measure of citizen satisfaction with the police is utilized in Model 1.1. Specifically, respondents were first advised to think about the "police in their neighborhood" and were then asked, "How satisfied are you with the police?". In response, individuals were to state whether they were "very satisfied, somewhat satisfied, somewhat dissatisfied, or very dissatisfied" with the police. As can be seen in Table 5.1, almost two-thirds (64.4%) of the respondents voiced generally favorable attitudes toward neighborhood police, with 24.4 percent being "very satisfied" and 40.0 percent of 115 TABLE 5.1 DETERMINANTS OF CITIZEN ATTITUDES TOWARDS THE POLICE DEPENDENT VARIABLES (ATTITUDES) VARIABLE VALUES N % IN GENERAL, HOW SATISFIED Very dissatisfied (1) 59 14.8 ARE YOU WITH THE POLICE? Somewhat dissatis. (2) 80 20.5 Somewhat satisfied (3) 156 40.0 Very satisfied (4) 95 24.4 HOW GOOD A JOB ARE THE Poor job (1) 98 26.7 POLICE DOING CONTROLLING Fair job (2) 112 28.1 THE STREET SALE AND USE OF Good job (3) 98 24.6 ILLEGAL DRUGS IN YOUR Very good job (4) 59 16.1 NEIGHBORHOOD? HOW GOOD A JOB ARE THE Poor job (1) 77 20.1 POLICE DOING TO KEEP Fair job (2) 133 34.6 ORDER ON THE STREETS AND Good job (3) 104 27.1 SIDEWALKS IN YOUR Very good job (4) 70 18.2 NEIGHBORHOOD? 116 the respondents "somewhat satisfied". At the same time, one-fifth (20.5%) of the subjects stated that they were "somewhat dissatisfied" and almost fifteen percent (14.8%) were "very dissatisfied". In general, these responses represent less positive attitudes than have been recorded in prior research which has found that most people are satisfied with the police (Gallup, 1982; Decker, 1980). The dependent variables in the other two models explored citizen evaluations of agency performance of more specific police functions. The attitude object is more specific in these models than in the prior one. In Model 1.2 respondents were requested to evaluate the job "police are doing controlling the street sale and use of illegal drugs in your neighborhood?" The dependent variable in Model 1.3 was operationalized by asking respondents about "how good a job the police are doing to keep order on the streets and sidewalks" in the neighborhood. After each of these two queries the following response options were read to the party being interviewed: "very good, good, fair, poor, or no opinion". When compared to the global attitudes of respondents, the distribution of responses to the more specific questions indicate that panel members hold less positive attitudes toward police performance of these two police roles. Specifically, over half of the respondents selected the two least favorable options, "fair" (30.5%) or "poor" (26.7%), when asked about the job police are doing controlling the 117 sale and use of drugs in the neighborhood. The least common response was that police were doing a "very good" job (16.1%). A similar pattern was observed with citizen perceptions of the job police were doing maintaining order on the streets and sidewalks (see Table 5.1). Namely, a slight majority of respondents believed police behavior was "fair" (34.6%) or "poor" (20.1%), while the least common response was again "very good" (18.2%).6 Thus, respondents apparently perceive that the police are doing a slightly better job maintaining order than regulating the sale of drugs, though in neither situation are their attitudes collectively as favorable as their overall attitudes.7 Several explanations may be offered why citizens’ global attitudes are more positive than perceptions about police performance of more specific functions. One explanation for the generally positive global attitudes toward the police, in light of the less than positive more specific attitudes, is that respondents are satisfied with police performance of job duties that were not measured and these attitudes are the basis for the more positive global attitudes. A second, and more plausible, explanation is that people generally hold benign global attitudes that contain a positive affective component not present in perceptions about police performance of designated functions. Citizens’ global attitudes are likely to be premised on a variety of factors, such as beliefs about the 118 institution of policing and the importance of the police in society, in addition to perceptions of actual performance. In contrast, when responding to questions about "how good a job the police are doing" performing specific roles, the respondent’s attention is directed to perceptions of task related conduct or perceptions of specific problems in the community. In addition, the tasks involve conduct that is problematic and which police often have trouble policing. As such, it should not be surprising in neighborhoods such as these that perceptions of the more specific features of police performance are less favorable than global attitudes. The benefit is that the different distributions of responses across the three attitudes suggests the need to examine whether each has a different underlying attitude structure or whether they are similar. Independent Variabiee In all but one instance, operationalizations of the independent variables discussed below are utilized in all three models (see Figure 5.1). The only exception involves operationalization of the neighborhood conditions construct. As will be more fully discussed in this section, this variable was operationalized in a manner so that it would contain only those conditions related to the dependent attitude variable. Demographic Variables. Data on all of the respondents’ characteristics were obtained during Wave 2 of the telephone survey. "Age" was collected by requesting respondents to 119 / //I 0:2 .2 IIIIII. ._.m_wm232 aoemz / ,\\\I mm....mw< ./ .852 I. \ 85.. III) \ \ .55.. . bm_mw 3/ \\ \\ 9.7.5.0055 ~_ . MS... zsotzmm /// \\ \f» p \mOmA: .Oz Iit\.\ E . \\_\ ....< .256. . / i ZO_._.<:mm “IIIIII . NBO< _ m0< any/9:02 02.202 #20me :Oz whim—9mm:— maDOzO “.0 him— Eo...5m .:oz .5 6...... Mambo: cofluodcoumou mo COwumNflHmcoflumquo ~.m ousmflm 156 are doing maintaining order in the neighborhood was deleted from this portion of the study. This attitude was not used to explain coproduction for two reasons. First, the behavioral options involve citizen conduct that would be theoretically relevant to the two attitude questions that remained in the model. The failure to include citizens’ perceptions of the job police are doing maintaining order on the streets would not jeopardize the level of specificity between the attitude and behavior constructs. Second, the correlation between these two questions (.598) indicates that they are moderately to highly related. Thus, inclusion of this attitude, for both statistical and theoretical reasons would add very little to the analysis. Since these attitudes were measured six months prior to the coproductive behaviors, they are properly time-ordered for the model. The remaining two attitudes toward the police questions are used for several reasons. The global satisfaction/attitude measure taps very general attitudes of community members towards local law enforcement. These attitudes may be premised on very general beliefs about the personal characteristics of police, their demeanor towards citizens, general beliefs about police authority, and/or specific behavior attributed to the police. Thus, this attitude may not be specific enough to influence coproduction behaviors. At the same time, though, all of the listed behaviors might be perceived by respondents as important to an assessment of the overall quality of their 157 local police officers. If so, then this general attitude may influence the decision to engage in one or more of the listed activities. Furthermore, police reformers presume that increasing general attitudes toward the police will lead to higher levels of participation. In contrast, the attitude object in the more specific attitude measure pertains to a specific type of police performance. Furthermore, the more specific attitude measure (drug control) concerns matters residents of these neighborhoods consider a serious problem and they pertain to Offensive conduct that community members can attempt to control through coproductive behaviors. For example, respondents indicated that drug use is a serious problem in the neighborhoods. There are presently a variety of behaviors (i.e., call the drug hotline, call the police, organize a neighborhood march, etc.,) that community members may engage in to help police control the sale and use of illegal drugs. These behaviors involve the same object (control of drugs) as this attitude. If prior attitudes toward the police research is correct, then these attitudes are more likely to be related to aforementioned conduct than the general/global attitude question. Spefie ip pee Neighborhood. Whether respondents have a stake in their neighborhood was operationalized through two variables. First, respondents were asked whether they rented or owned their home. As can be seen Table 5.4, this variable was coded as a "rent - own" dichotomy. Slightly 158 over one-fourth (26.1%) of the respondents indicated that they rent, while the remainder said that they either own or were in the process of purchasing their home. If respondents owned their homes, the assumption was that they were more attached to the community than if they rented. The second variable tapped the mobility of residents and their attachment to the community. Specifically, respondents were asked to state "how likely is it that you will still be living in this neighborhood a year from now?" The response options included "very likely, somewhat likely, somewhat unlikely, and very unlikely". Over one-half (55.2%) of the sample noted that it was very likely that they would be residing in the neighborhood for at least another year. An additional 18.3 percent of the panel members said it was likely they would remain in the community during the coming year, while the same percent of people said it was unlikely they would stay in the neighborhood. At first glance these figures may appear to indicate that most of the residents look favorably upon their neighborhoods and have voluntarily decided to stay put. Also, they may be surprising since the neighborhoods suffer from numerous economic, social, and crime problems. However, there are several reasons that may explain why so many respondents anticipate remaining in the neighborhood. First, homes in the study sites are very inexpensive. Second, there is only limited demand for many of the homes 159 TABLE 5.4 ATTITUDES-BEHAVIOR RELATIONSHIP: INDEPENDENT VARIABLES VARIABLE VALUES N % Prior Attitudee IN GENERAL, HOW SATISFIED Very dissatisfied (1) 67 17.4 ARE YOU WITH THE POLICE? Somewhat dissatis. (2) 81 21.0 Somewhat satisfied (3) 158 41.0 Very satisfied (4) 79 20.5 HOW GOOD A JOB ARE THE Poor job (1) 93 26.7 POLICE DOING CONTROLLING Fair job (2) 133 38.2 THE STREET SALE AND USE OF Good job (3) 74 21.3 ILLEGAL DRUGS IN YOUR Very good job (4) 48 13.8 NEIGHBORHOOD? HOW GOOD A JOB ARE THE Poor job (1) 79 20.8 POLICE DOING TO KEEP Fair job (2) 140 36.8 ORDER ON THE STREETS AND Good job (3) 103 27.1 SIDEWALKS IN YOUR Very good job (4) 58 15.3 NEIGHBORHOOD? Stake ip Community RESIDENCE Rent (1) 104 26.1 Own (2) 294 73.9 HOW LIKELY IS IT THAT Very unlikely (1) 71 18.3 YOU WILL LIVE IN THE Unlikely (2) 32 8.2 NEIGHBORHOOD A YR. FROM Likely (3) 71 18.3 NOW Very likely (4) 214 55.2 Perceptions e: Problems KIDS HANGING OUT ON THE No Problem (1) 181 45.6 STREETS MAKING NOISE, Small Problem (2) 130 32.7 STARTING FIGHTS Big Problem (3) 86 21.7 PEOPLE HANGING AROUND ON No Problem (1) 190 48.1 THE STREET, OR IN PARKS OR Small Problem (2) 107 27.1 VACANT LOTS Big Problem (3) 98 24.8 PROSTITUTION No Problem (1) 248 67.4 Small Problem (2) 66 17.9 Big Problem (3) 54 14.7 160 TABLE 5.4 (cont’d) VARIABLE VALUES N % Perceptions pi Problems PEOPLE SELLING DRUGS IN No Problem (1) 126 34.3 THE NEIGHBORHOOD Small Problem (2) 106 28.9 Big Problem (3) 135 36.8 TOO MUCH TRAFFIC ON THE No Problem (1) 218 54.9 STREET Small Problem (2) 98 24.7 Big Problem (3) 81 20.4 VACANT OR ABANDONED No Problem (1) 140 35.4 HOMES IN NEIGHBORHOOD Small Problem (2) 122 30.8 Big Problem (3) 134 33.8 LITTER AND TRASH ON THE No Problem (1) 130 32.7 STREETS AND SIDEWALKS Small Problem (2) 154 38.8 Big Problem (3) 113 28.5 PEOPLE DRINKING ON THE No Problem (1) 220 56.1 STREET Small Problem (2) 101 25.4 Big Problem (3) 71 18.1 Collective Action HOW MUCH OF AN EFFECT CAN No effect (1) 40 10.3 COMMUNITY RESIDENTS HAVE Little effect (2) 77 19.7 OVER CONDITIONS IN THE Some effect (3) 123 31.5 NEIGHBORHOOD A large effect (4) 150 38.5 HOW MUCH OF AN EFFECT CAN No effect (1) 21 5.8 COMMUNITY RESIDENTS HAVE Little effect (2) 32 8.8 OVER CONDITIONS IN THE Some effect (3) 156 43.0 NEIGHBORHOOD A large effect (4) 154 42.4 IS NEIGHBORHOOD MOSTLY ONE Help each other(0) 192 50.7 WHERE PEOPLE GO OWN WAY OR Go own way (1) 187 49.3 HELP EACH OTHER? HOW LIKELY IS IT THAT Very unlikely (1) 47 12.0 NEIGHBOR WOULD CALL POLICE Unlikely (2) 39 9.8 TO REPORT SUSPICIOUS Likely (3) 117 29.4 ACTIVITY? Very likely (4) 189 47.5 161 TABLE 5.4 (cont’d) VARIABLE VALUES N % Collective Action HOW LIKELY IS IT THAT Very unlikely (l) 58 14.6 NEIGHBOR WOULD CALL POLICE Unlikely (2) 42 10.6 TO REPORT DRUG RELATED Likely (3) 89 22.4 ACTIVITY? Very likely (4) 188 47.2 HOW LIKELY IS IT THAT Very unlikely (1) 32 8.0 NEIGHBOR WOULD CALL POLICE Unlikely (2) 24 6.0 TO REPORT ILLEGAL Likely (3) 76 19.1 ACTIVITY? Very likely (4) 254 63.8 162 in the involved neighborhoods thus residential property sales are not necessarily a feasible option. Third, the price of homes in the suburbs to which most people would want to flee may be beyond the means of most residents. For these reasons it is very likely that most people feel trapped in the neighborhoods and therefore expect to remain in them by necessity.20 These factors pose several possible implications for the present study. Namely, people who feel trapped in their neighborhood may not have the same stake in their immediate community as do people who have voluntarily decided to reside in a specific location. Also, even though many of these individuals have decided to not "exit", they may also not intend to "voice" concerns and instead may plan to remain with the hope that conditions improve (Lyons and Lowery, 1986: Sharp, 1984). Thus, the impact of this measure on coproduction may be muted. Eerceptione 9: Problems. The neighborhood problems construct was operationalized using a scale similar to that used to operationalize neighborhood problems in Model 1.3. The only difference was that the question pertaining to drug use in the area around respondents’ residences was included. The eight questions were: 1. Kids hanging out on the streets making noise, starting fights, or just bothering people. (Problem) 2. Prostitution. (Problem) 163 3. Vacant or abandoned homes in the neighborhood. (Problem) 4. Litter and trash on the streets and sidewalks. (Problem) 5. People drinking on the streets. (Problem) 6. People hanging around on the street, or in parks or vacant lots. (Problem) 7. Too much traffic on the streets. (Problem) 8. People selling drugs in the neighborhood. (Problem) Principal components factor analysis indicated that these questions tap a single dimension, in that one significant factor was produced with an eigenvalue of 4.132. All of the remaining factors had eigenvalues below .799. These eight items were therefore used to create a neighborhood problems scale (Cronbach’s Standardized Alpha = .856). Scale scores were then computed using the factor scale coefficients. Scores ranged from 5.72 to 17.17, with low scores representative of perceptions that the listed conditions are less problematic, while higher are representative of the view that the conditions are problems. The neighborhood fear questions were not used in this portion of the study. This decision was based on the finding that people fearful of crime are often consumed by their fear and do not get involved in remedying the underlying problem (Rosenbaum, 1987). In addition, the importance of this construct is based on the feeling that people who see conditions in the neighborhood as problematic are more likely to engage in conduct that is intended to alleviate the condition than those who do not see 164 neighborhood matters as problems. Feasibility e; Collective Action. Explanatory variables that tap respondents’ perceptions of the feasibility of collective action within their neighborhood were also included. This construct was operationalized through three variables which focus on different dimensions of collective action. The first variable involves citizens’ perceptions of the efficacy of individual or group action. In the model explaining private action (Model 2.1) responses to the following question was used: "how much of an effect can community residents have over conditions in their neighborhood", with closed-ended response options including "a large effect, some effect, a little effect, or no effect". As Table 5.4 displays, the modal response (38.5% of the respondents) was that residents can have a "large effect", with "some effect" the second most common response (31.5%). Only thirty percent of the sample noted that community residents could have "little" or "no effect". Responses to a slightly different question were used with the collective action model (Model 2.2). This question focused on beliefs about the effectiveness of community groups. Respondents were asked "How much of an effect can community groups have over conditions in their neighborhood?" The same response options, as above, were used with this question. Replies to this query indicate that sample expressed a greater belief in the potential effectiveness of collective action than individual actions. 165 Namely, almost eighty-five percent of the sample said that community groups could have either "some" or a "large" effect. The remainder of the sample was split fairly evenly between the remaining two categories. The second variable tapped perceptions of neighborhood cohesiveness. Interviewees were asked to state whether they believed that neighbors generally "helped each other or went their own way". Responses were almost evenly split between the two categories (192 people responded that people help each other, while 187 individuals said people go their own way). The presumption was that people who perceived neighbors as supportive would be more likely to engage in private and, especially, collective action. The third variable operationalizes a dimension of Olson’s (1970) public goods argument. Namely, he noted that individuals do not engage in collective action (and probably any action) if they think they can receive the same benefits through the less costly option of inaction. This variable was constructed using three questions that examined whether respondents believed that their neighbors would call the police to report suspicious or illegal behavior. Respondents were requested to state whether it was "very likely, likely, unlikely, or very unlikely" that a resident on their block would call the police in each of the following three situations: 1. To report suspicious activity. 2. To report drug-related activity. 166 3. To report a crime. The frequency distributions for these three questions appear in Table 5.4. In response to two of the questions almost fifty percent of the respondents indicated that it was very likely that neighbors would report "suspicious activity" (47.5%) and "drug related activity" (47.2%). Considerably more than half (63.8%) of the sample noted that it was very likely that fellow residents would report "illegal activity". These three items were factor analyzed to determine whether they tap a single dimension -- perceptions that neighbors will report. Results of principal components analysis confirmed that these items do tap a single dimension. One significant factor was produced with an eigenvalue of 2.369, with the remaining factors all having eigenvalues under .361. The three items had factor loadings between .869 and .903. A three-item scale (Cronbach’s Standardized Alpha = .866) was computed using the factor scale coefficients with factor scale scores ranging from 2.37 to 9.47. Low scale scores represent respondents that had less of a belief that neighbors would report the incidents, while higher scores represented individuals who had more of a belief in the likelihood that others would report.21 Qemoqraphic Variable. The final category is composed of individual and neighborhood level demographic variables. The measures used are almost the same as those utilized in 167 the first portion of the study. The individual level variables include age, race, household income, sex, and educational level of the respondent. The neighborhood level variables encompass percent nonwhite and neighborhood income levels (see Table 2). These variables are included because several people have suggested that demographic variables are related to citizen willingness to participate in private and collective behavior (Haeberle, 1987; Sharp, 1984; Lavrakas and Herz, 1982; Rich, 1980). The only variable that was operationalized differently for this portion of the study was "age". In response to beliefs that the impact of age on collective behaviors is curvilinear, age was decomposed into two measures. The first measure, AGEl, operationalize the age bracket during which participation increases. People between eighteen and fifty years old received a value representative of their chronological age. Individuals over age fifty were coded as 0. In contrast, for the second variable (AGE2) respondents under fifty were assigned a value of 0, while the person’s actual age was used for people over fifty. Analysis Favorable attitudes toward the police are presumed to be positively related to coproductive behaviors. Citizens’ attitudes, along with a number of other factors, are also portrayed as determinants of behaviors supportive of local law enforcement. In order to test these assumptions a two step process was utilized. First, correlations between the 168 attitude measures and each behavioral measure were computed. Correlations were also used to assess the relationships between the attitudes and the behavioral scales. For instance, the attitude toward police regulation of drug sales and use was correlated with the individual drug related behaviors. Next, the relationship between this same attitude and the drug behavioral scale was examined. The process was repeated for each attitude measure. Second, the four models were examined using a path analytic technique. Since the dependent variable in each equation is a continuous behavioral scale, ordinary least squares regression techniques are appropriate. This permits an exploration of the direct and indirect effects of the variables contained in each of the models. SUMMARY The present study examines the determinants of citizen attitudes toward the police and the relationship between citizen attitudes and citizen willingness to engage in certain behaviors that assist the police. In addition, this endeavor assesses the independent impact of attitudes on coproduction by also including a number of factors that existing research indicates are also substantively important. As such, the findings provide insight into whether the assumptions of prior research and recent police policy initiatives are correct concerning the relationship between satisfaction and citizen coproduction. 169 ENDNOTES 1. Most of the data analyses in this study use data collected as part of the second and third waves. A panel consisting of participants in all three waves was not utilized because it would have entailed a loss of 150 respondents. 2. The four selected areas included 577 street blocks. On 186 of these blocks (32.2%) drug-related police activity occurred during the nine months prior to start-up of the drug enforcement project. Therefore, 186 additional respondents were selected using the same process previously described, with one addition. Households already selected were deleted so that each remaining household listed in the directory had an equal opportunity to be selected. 3. For some blockfaces, it was not possible to secure two alternates due to the fact that a number of streets in the city of Detroit consist of only a few habitable residences with the remainder Of the structures vacant or, in some cases, the lots vacant. In addition, unlisted phone numbers likely contributed to a lack of available households for the sample in some locations. 4. When respondents first participated in the survey process they were told that the purpose of the survey "is to find out about neighborhood problems in the city of Detroit" and that "your answers will help us to learn more about your neighborhood". During the recontact process in subsequent interviews, respondents were told that the present survey was similar to the prior one about neighborhood problems and that "we are concerned how neighborhoods have changed" since the last interview. 5. During the same time as the present study was being conducted, the Vera Institute was conducting a study in New York city that also focused on drug enforcement activities. In their project urban ethnographers attempted to conduct in-person interviews of city residents. In their study, a response rate of only approximately 20% to 25% was achieved. In contrast in the present study there were response rates of 76.8%, 74.9%, and 73.8% in Wave 1, Wave 2 and Wave 3, respectively. 6. Respondents also could respond that they were "did not know" or "had no opinion" concerning the attitude measures, though these options were not read to the interviewee. At the same time, respondents could also refuse to answer each of the three attitude questions. Since the data coding format made it impossible to differentiate why someone did not provide a usable answer, these individuals were deleted from the analyses. For instance, someone could have said no Opinion and hold a favorable or unfavorable attitude and 170 just not want to voice it. Eight of the panel respondents did not voice a usable response to the global attitude question, 31 panel members failed to provide a usable response to the drug control attitude, while fourteen people failed to respond to the maintenance of order attitude. 7. T—tests of statistical significance indicated that there were statistically significant differences between global attitudes and the other two specific outlooks toward the police; namely, global v. drug related attitudes -- t = 8.52; p = .000; global v. order maintenance -- t = 7.38: p = .000. Admittedly, t-test are only supposed to be used with interval level data. They are utilized here only for confirming what appears, fairly obvious, from examining the actual distribution of the data. 8. On the actual survey instruments the list of matters included, in addition to those utilized herein; people drinking on the street; people hanging out; traffic on the street; police stopping of people; and loud parties. These areas were not listed by respondents as problematic and thus were not used in the present study. 9. The survey instrument listed several additional victimization situations citizens may have worried about. These situations include someone will try to involve your child or a family member in using drugs, in selling drugs, and someone will harm you or a family member if you call the police. 10. Prior to calculating the factor scale scores a decision was made to substitute mean scores for missing data. This was done by calculating the mean for each question separately and then inserting the mean if a respondent failed to answer a specific question. As can be derived from the frequencies presented in Table 2, substitutions for missing data were necessary in the following "problem questions": prostitution and people selling drugs (31 people each question), vacant home and litter or trash on street (2 people each question), and kids hanging out on street (1 person). As to the "victimization situations" missing data appeared in the following amounts: break into home (1 person), attack or beat you up (4 people), rob or steal (3 people), harass you (47 people). This method was viewed as preferable over one that required substituting mean scale scores since it permitted use of responses to those questions that were answered and were included in the scales. 11. A similar procedure as was discussed in Note 10 was utilized when constructing this three-item scale. 12. It is impossible to account for all possible types of encounters that citizens will have with the police, though 171 3 the four included herein are likely to be representative of the dominant forms of encounters. In addition, these four types have been utilized in prior research on the police (Dean, 1980). 13. To properly time order the variables in the model it was necessary to exclude experiences that were prior to the first attitude measure. Therefore, only experiences that occurred within the last six months were included in the analysis. This permitted the information to be collected during Wave 3. 14. Very few people failed to provide a usable answer to these questions. Specifically, only 6 respondents to information satisfaction questions, 1 person in reference to assistance from the police, and 3 people in victimization experiences said they were "neither satisfied nor dissatisfied" or "had no opinion". 15. A third option would have involved adding an additional eight variables so that each experience could be included (i.e., two positive experiences Of each type and two negative experiences). It was believed that the addition of these eight additional variables would not contribute sufficiently to the analysis, especially since only a limited number of people were involved in more than one similarly evaluated situation. 16. Analyses were conducted using both coding schemes and no significant differences were Observed. 17. When individuals who noted that had not "seen or heard" about the listed activities are excluded from the analyses, very little changes. 18. If all of the questions were perceived by respondents as pertaining to only drug related activities, then you would expect there to be a large number of respondents who answered that they reported all of the activities if they reported one of them. This was not the case. 19. In all of the target locations there were existing community groups that were involved in organizing community residents. These groups were normally associated with a local religious institution, a community tenant group, a business group, and/or were an umbrella group for a specific neighborhood. While many of these groups were involved in providing a range of social services to community members, many were also active in organizing local anti-crime efforts. According to information secured from community knowledgeables, many residents turned to local community group when it came to providing information. 20. During the in-person interviews with community 172 knowledgeables many residents voiced the aforementioned reasons. Most people noted that they were "trapped" by the availability of cheap housing and limited incomes. 21. In situations where there was missing data the mean values for each question were utilized. In the questions concerning suspicious activity this was necessary in six situations, in the drug question twenty-one people failed to respond with a usable response, and in the question referring to reporting of illegal activity twelve substitutions were necessary. For a further discussion of how this was done, see note 10. 173 CHAPTER 6 CITIZENS’ ATTITUDES TOWARD THE POLICE: FINDINGS AND DISCUSSION This chapter contains a discussion of the findings relative to the determinants of citizens’ attitudes toward the police. The results of the data analysis are presented in the following order: Model 1.1 involving citizens’ global attitudes is first addressed; second, Model 1.2 pertaining to citizens’ perceptions of the job police are doing controlling the sale and use of drugs is examined; third, respondents’ perceptions of the job police are doing maintaining order on the streets and sidewalks in their neighborhood (Model 1.3) is discussed last. All three of the models were estimated using ordinary least squares regression techniques. 1 The data are presented in a manner that focuses on the direct, indirect, and total effects of each explanatory variable on citizens’ attitudes toward the police. The direct effects are the standardized regression coefficients and appear in a number of tables in the chapterz. The indirect effects are the products of the coefficients along each of the paths from a particular variable through intervening variables to attitudes toward the police (ATT t+1). The total effect of each variable was calculated by summing the direct and indirect effects for each measure (see Parks, 1984: 125). This technique was utilized in an attempt to determine whether previous research had possibly 174 misstated the influence of variables by accounting only for direct effects. In addition, path analysis permits an exploration of the linkages among variables. FINDINGS Global Attitudes Toward The Police Direct Effects. The direct influences of the model variables on global attitudes toward the police are presented in Table 6.1. Based upon the coding scheme, a positive regression coefficient indicates that an increase in the value of an independent variable is positively associated with an increase in the person’s attitude towards the police. A negative coefficient is therefore, related to a decrease in an individual’s outlook concerning the police in her/his neighborhood. When the direct effects are examined, the findings are supportive of the belief that citizen evaluations of experiences with the police are better predictors of attitudes than demographic variables. Four of the experiences were found to be statistically significant. Three of these encounters (negative information, positive assistance, and negative assistance contacts) involved situations where the experience was initiated by the citizen. Furthermore, the coefficients for these three encounters are all in the expected direction. Namely, negative evaluations tended to decrease global attitudes, while the positive evaluation increased this attitude. 175 TABLE 6.1 DIRECT EFFECT OF ALL VARIABLES IN MODEL 1.1: GLOBAL ATTITUDES Variable B Beta T Sig T Demographic Age .003 .060 1.250 .210 Gender -.029 -.014 -.316 .752 Income -.089 -.098 -2.033 .042 Education .050 .061 1.259 .208 Race .034 .017 .334 .738 Neighborhood Context Ngh. Inc .108 .062 1.397 .163 Ngh. Race .001 .028 .580 .562 Perceptione QT Ngh. Conditione Ngh. Prob -.159 -.264 -4.823 .000 Ngh. Fear -.016 -.028 -.515 .606 Prior Attitude Att (t) .424 .425 9.181 .000 Comparative Assessments ASS. Resp Time .686 .135 2.86 .004 Victimization Victim .032 .021 .460 .645 Evaluations 9; Experiences Pos. Info .049 .034 .813 .416 Pos. Assist .307 .168 4.010 .000 Pos. Stop -.011 -.005 -.110 .912 Pos. Vict -.401 -.114 -2.573 .010 Neg. Vict -.105 -.018 -.400 .689 Neg. Stop .075 -.026 .614 .539 Neg. Assist -.248 -.103 -2.400 .017 Neg. Info -.147 -.102 -2.285 .023 Reported Significance Levels are for 176 two-tailed tests In the fourth statistically significant experience, a positive evaluation of officer(s) behavior subsequent to a victimization experience, the coefficient was negative. While this was not anticipated, it is also not overly surprising. It is quite possible that other dimensions of the encounter besides the responding officer’s behavior influenced the general attitude. For instance, the respondent may hold the local police responsible for the victimization. In addition, the fact of being victimized may be accessed, with less emphasis on what the officer did, when responding to the global attitude question. In contrast, the only demographic variable that exerted a significant influence on the dependent variable was respondents’ family income. The data indicate that as income increases people tend to hold less favorable global attitudes toward the police. The remainder of the demographic variables do not appear to have strong substantive direct influences. Three additional variables in the model exerted strong direct influences on citizens’ attitudes. As can be seen in Table 6.1, the coefficient for respondents’ comparative assessments of the police response time was significant and positive. This result confirms the findings of earlier studies which determined that a police response believed to be faster than expected has a positive influence on citizens’ attitudes (Parks, 1984; Percy, 1980). The remaining two variables, perceptions of 177 neighborhood conditions and prior global attitudes, that were found to be statistically significant are factors that are not usually included in research concerning citizens’ attitudes towards the police. Not only were these two measures significant predictors of global attitudes, but their standardized coefficients indicate that they exert fairly strong direct effects. Specifically, the standardized coefficient for the prior global attitude measure was larger than the coefficient of any of the other variables. It therefore, appears that respondents’ present attitudes are highly informed by pre-existing attitudes. The findings relative to citizen perceptions of neighborhood conditions indicate that this variable also exerts strong direct influences on global attitudes toward the police. As anticipated the influence was negative. Namely, respondents who viewed neighborhood conditions as problematic, were more likely to hold less positive attitudes toward the police. Indirect epg Tepei Effects. The examination Of the indirect effects of variables in the model indicates that focusing solely on the direct links between certain variables and the dependent variable understates the influence that several factors have on attitudes toward the police. This is especially true for the impact of certain demographic variables. The path coefficients for all variables in the model are contained in Table 6.2, while Table 6.3 displays the direct, indirect, and total effects 178 of all variables contained in the model. The importance of examining the combination of the direct and indirect influences is quite evident with respondent’s race. Race has a somewhat weak positive direct effect on global attitudes. However, race exerts a number of significant indirect effects on attitudes through other variables in the model. Race is hypothesized to have a direct effect on perceptions of neighborhood problems, prior global attitudes toward the police, fear of Victimizations, assessments of police response time and several evaluations of police performance in specific encounters. These variables, in turn, are portrayed as influencing global attitudes toward the police. When one computes the total effect this variable has on attitudes toward the police, the total effect has a coefficient (-.188), which is considerably larger than the variable’s direct effect. Most of the indirect effects of this variable are through the prior attitude and perceptions of neighborhood problem variables. More specifically, whites were found to hold less favorable prior attitudes than nonwhites. In addition, white respondents were more likely to see neighborhood conditions as problematic, which in turn, has a negative influence on attitudes toward the police. A similar pattern is observed for the variable age. Respondent’s age has a moderate direct effect on global attitudes toward the police. However, this variable has 179 TABLE 6.2 STANDARDIZED PATH COEFFICIENTS FOR MODEL 1.1: GLOBAL ATTITUDES Model Ngh Ngh Prior Asses Num Variables Prob Fear Att Resp. Vict Age -.258*** -.070 .247*** .017 -.121** Gender -.025 .125** .040 -.049 -.072 Income -.188*** -.164*** -.019 -.039 -.018 Education -.072 -.160*** .085 -.036 .082 Race .202*** .201*** -.160** -.095* .050 Ngh Race -.054 -.096* -.007 -.023 Ngh Income -.150*** -.014 .115** -.116** Prior Attitude -.079 Num Victim .241*** * p < .05 (one—tailed test) ** p < .05 (two-tailed test) *** p < .01 (two tailed test) 180 .umou ooaaou ozuc Ho. v a .44 Aumou ooaaeuuosuc mo. v a 44 Lowe» ooaaeuuococ mo. v a 4 ssagmm.lsssamm. OEHB .Qmmm save. «aroma. Efluofl> Esz ocsufl smmm.l mho. RRHNH.I «*«ova.l Sec. Nmo. Ado. vmo.l 99¢ .Mm smoa. s«maa. mmo.l vmo. omo. woo.l mmo. «ssowa.l 00mm HHO.I mac.l mmo.l Goo. hmo.l eve. NHO.I sarmaa. mumuscm omo.l veo.l amo.l «*«MGH.I mmo. mwo.l smoa. mmo. OEOOGH hmo. Hmo.l *«mma.l hmo.l hao.l 8*«OMN.I mmo.l HHO. HODCOU rmma.l cmo.l «shNH.I mmo.l meo.l *«swma.l mmo.l mmo. 00¢ OmcH wmmd moum uofl> u0fl> moum Gums OmcH moanmflum> mmz moz moz mwz mom mom mom mom H0602 .o.u:ooc m.o mum Epz upsufl 4mmH.I mmo.l HNo.I moo.l woo. moo.l «moa. gmo. uu< .Hm «med. «toma. OHO.I mmo. muo. «tNHH.I Nmo. «NHH.I 000% cao.| moo.I eco.| coo. mno.u Hoo. evo.l 444mwan oumoscm mao.l mwo.l Ono.I astoma.l mmo. wmo.l 00H. Ono. OEOOCH mmo. mvo.l trame.I Hoo.l moo. *44HHN.I ovo.l moo. Moccmo soma.l mmo.l «stoma.l ammo.l Hmo.l tamma.l Nmo.l Nmo. 004 OOCH Hmm< moum uofi> upw> doum fimm< OucH moanmwum> qoz omz moz moz mom mom mom mom H060: .o.u:ooc m.o mamas 189 TABLE 6.6 DIRECT, INDIRECT AND TOTAL EFFECTS OF MODEL VARIABLES ON DRUG ATTITUDES: MODEL 1.2 VARIABLE DIRECT INDIRECT TOTAL Demographics Age —.035 .177 .142 Gender -.022 -.019 -.041 Inc -.074 -.064 -.138 Ed -.021 .053 .032 Race -.082 -.105 -.187 Neighborhood Context Ngh. Inc -.024 .087 .063 Ngh. Race -.004 -.037 -.041 Perceptions pi Neighborhood Conditions Ngh. Prob -.322 -.322 Ngh. Fear -.026 -.026 Prior Att. Att(t) .397 .025 .422 Comparative Assessmente Ass.Resp Time -.024 -.021 -.045 Victimizations Victim -.027 -.013 -.040 Evaluatione _§ Experiences Pos Info .102 .102 Pos Assist -.O32 -.032 Pos Stop -.029 -.029 Pos Vict .93 .093 Neg Vict -.025 -.025 Neg Stop -.055 -.055 Neg Assist -.020 -.020 Neg Info -.109 -.109 190 The indirect effect of age through this measure is also quite substantial (.051). Examination of the total effects of respondent’s race reveals that this variable also contributes in an indirect manner to citizens’ perceptions of the job police are doing regulating the sale and use of drugs. The indirect effects are principally through the same two variables as with age, perceptions of neighborhood problems and prior attitudes. For instance, white respondents were more likely than nonwhites to see drugs as a problem in the community, with perceptions of drug problems exerting a negative impact on evaluations of police performance pertaining to this type of illegal behavior. With reference to prior attitudes, whites were also more likely to possess less favorable existing perceptions than were nonwhites. This fact resulted in an additional negative indirect effect of race on the dependent variable. Two general conclusions may be drawn concerning the total effects of the demographic variables. First, three of the five demographic variables have indirect effects that exceed their direct effects. The exceptions are gender and income where the indirect effects approach, but do not equal, the direct influences. Second, when you compare the total effects of the demographic variables to those observed for evaluations of the eight incident specific experiences with the police, the effects of respondent’s age, race and income each exceed the individual effects of experiences 191 with the police. Order Maintenance Attitudes Direct Effects. The direct influence of variables in Model 1.3 are presented in Table 6.7. Generally, the findings indicate that three of the constructs exert statistically significant effects on citizens’ perceptions of the job police are doing maintaining order on the streets and sidewalks in the respondent’s neighborhood. These three constructs are the same as those for the prior models; namely, prior attitudes, perceptions of neighborhood problems, and evaluations of experiences with the police. More specifically, respondents’ attitudes as measured in Wave 2 were found to exert a strong positive influence on the Wave 3 attitude measure. As in the two previous models, the coefficient for this variable exceeds that of any of the other variables included in the model. Again, it appears that present attitudes are highly informed by pre-existing outlooks concerning this same attitude. A strong direct effeCt was also observed for the neighborhood problems measure. The influence of this variable on citizen views of police performance was as expected. Respondents who perceived the neighborhood matters as creating disorder were most likely to negatively evaluate police performance involving these same items. The influence of this variable, as in the prior models, was extremely significant. Finally, three of the police-citizen experiences were 192 DIRECT TABLE 6.7 EFFECT OF ALL VARIABLES IN MODEL 1.3: ORDER ATTITUDES Variable B Beta T Sig T Qemographice Age .002 .034 .897 .370 Gender -.050 -.020 -.514 .607 Income -.039 -.044 -.876 .381 Education -.003 .001 .013 .989 Race -.005 -.001 -.055 .956 Neighborhood Context Ngh. Inc. .091 .042 1.140 .255 Ngh. Race .001 .043 .792 .428 Perceptions pi Ngh. Conditions Ngh. Prob -.154 -.409 -6.867 .000 Ngh. Fear .046 -.069 -1.312 .190 Prior Attitude Attitude (t) .378 .360 7.465 .000 Comparative Assessment Ass. Resp Time -.041 -.008 -.169 .866 Victimizations Victim -.004 -.007 -.007 .952 Evaluation pi Experienq_e Pos. Info .154 .105 2.427 .015 Pos. Assist .259 .135 3.266 .001 Pos. Stop -.117 -.048 -1.152 .250 Pos. Vict -.158 -.047 -1.004 .316 Neg. Vict -.210 -.031 -.782 .434 Neg. Stop —.033 -.011 .260 .795 Neg. Assist -.224 -.088 -2.032 .043 Neg. Info -.026 -.016 -.400 .689 Reported Significance Levels are for 193 two-tailed tests also statistically significant. Positive evaluations of information and assistance encounters and negative assessments of assistance contacts were each not only significant, but also in the expected directions. The positive evaluations of police performance during the two specific instances had the effect of increasing attitudes towards the police, while the negatively evaluated situation decreased attitudes. As can be seen in Table 6.7 the direct effects of the remaining variables were negligible. None of the standardized coefficients for the demographic variables exceeded .044, which was slightly higher than either of the neighborhood context variables. The other experiences with the police also had coefficients that were below .048. Indirect and Total Effecte. The standardized coefficients for the model’s paths are presented in Table 6.8. Table 6.9 provides the indirect and total effects. Focusing only on the direct links in this model, also understates the influence that several of the variables have on citizens’ attitudes. Again, this is especially true for the demographic variables. An example of this combination of direct and indirect effects can be seen by focusing on the influence of respondent’s race. This variable has a very weak direct influence on the dependent variable. However, this variable does exert a number of substantial indirect influences on perceptions of police performance through other variables in 194 TABLE 6.8 STANDARDIZED PATH COEFFICIENTS FOR MODEL 1.3: ORDER ATTITUDES Model Ngh Ngh Prior Asses Num Variables Prob Fear Att Resp. Vict Age -.227*** -.105 .206*** -.005 -.121** Gender .031 .126** .126** -.054 -.072 Income -.l76*** -.168*** .027 .041 -.018 Education -.082 -.157*** .103* -.046 .082 Race .164*** .183*** -.137** -.085 .050 Ngh Race -.065 -.097* .020 -.023 Ngh Income -.125** -.019 .092* -.116** Prior Attitude .006 Num Victim .242*** * p < .05 (one-tailed test) ** p < .05 (two-tailed test) *** p < .01 (two tailed test) 195 A0000 ooaaou ozuc ao. v a 444 Loom» ooaaouuosuc mo. v a 44 .0000 ooaamuaococ mo. v a 4 44444m.u444-m. osae .amom va. 444mma. Eau0a> Esz, mpsua «mom.l mvo.l moo.| OOO.I mac. nmo.| omo. moo.l uuc .um «moa. «aoma. mmo.| Ono. nmo. 4mmO.I mmo. «4waa.l 00mm mmc. vao.l mmo.| pmo. vpo.l vvo. sao.l 444NON. 0umospm NOO... mnO.I 5N0... ttOva.l ONO. FOO... moa. ONO. 0EOOCH avO. mmo.l h.zsava... Ono... ONO. «*«aam.l th.I maO. H0pc0w amO... hmO.I «stoma... omO... OmO.I asscaa.l th.I mnO. 06¢ Omca aumc moum u0a> u0a> moum amm< OHCH m0anmapm> 002 002 O02 002 mom mom mom mom a0coz Ao.p:oo. m.o mamas 196 TABLE 6.9 DIRECT, INDIRECT, AND TOTAL EFFECTS OF MODEL VARIABLES ON ORDER MAINTENANCE ATTITUDES: MODEL 1.3 VARIABLE DIRECT INDIRECT TOTAL Demographice Age .034 .173 .207 Gender -.020 .071 .051 Inc -.044 +.121 .077 Ed -.001 +.107 .106 Race .001 -.124 -.123 Neighborhood Context Ngh. Inc .042 .085 .127 Ngh. Race .043 .040 .083 Perceptions QT fleighborhood Conditione Ngh. Prob -.409 -.409 Ngh. Fear -.069 -.069 Prior Att. Att(t) .360 .019 .379 Comparative Assessments Ass.Resp Time -.008 -.005 -.013 Victimization Victim -.007 .021 .014 Evaluations QT Experieneee Pos. Info .105 .105 Pos. Assist .135 .135 Pos. Stop -.048 -.048 Pos. Vict -.047 -.047 Neg. Vict -.031 -.031 Neg. Stop -.011 -.011 Neg. Assist -.088 -.088 Neg. Info -.016 -.016 197 the model. Race has a strong direct effect on neighborhood conditions, perceptions of crime fears, the prior attitude measure, and four of the evaluations of citizen experiences with the police. When the indirect effects of race are calculated, this variable exerts most of its indirect influence through neighborhood problems and prior attitudes. For instance, the direction of the relationship between race and perceptions of neighborhood disorder was positive. Therefore, whites were more likely than nonwhites to view the listed situations as problematic, with neighborhood order problems exerting a strong negative influence on attitudes about local law enforcement. Whites also were more likely than nonwhites to hold less favorable pre- existing attitudes toward the police. Pre-existing perceptions of police performance, as noted exerted a strong influence of the dependent variable. Therefore, the indirect effect of race on attitudes was quite substantial (.049) . Several other indirect effects of race are also worthy of mention. First, race has a direct effect on victimization fears and an indirect effect on the dependent variable through this item (-.013). Second, race has a significant direct effect on four of the police-citizen experiences. In two positively evaluated situations (stops and information requests) whites were less likely to positively evaluate the police performance. In the two 198 situations that were negatively evaluated (assistance and information encounters) whites were more likely than nonwhites to provide negative evaluations. Similar combinations of the indirect and direct influences can also be Observed for several other demographic variables. Age exerts a weak direct influence on attitudes, though the model indicates that this variable also had a number of indirect influences. These indirect influences are principally through neighborhood problems (.093), and prior attitudes (.074). As portrayed in Table 6.9 respondent’s income also exerts substantial indirect effects, with most of the indirect influence occurring through perceptions of neighborhood problems (.072). The indirect effect Of education followed a similar pattern. Specifically, this variable indirectly influenced the dependent variable through prior attitudes (.037) and perceptions of neighborhood problems (.034). As noted, these indirect effects occurred because of the substantial direct effects of the demographic variables on perceptions of problems, prior attitudes, and selected experiences with the pOlice (see Table 6.8). DISCUSSION When the findings of the three models are compared several consistent findings appear. The findings in some instances confirm prior research, while in others they contradict the findings of existing studies. Finally, several of the findings suggest considerations for 199 additional research. A consistent finding across all three models is that public perceptions of neighborhood matters as problems reduces citizens’ positive attitudes towards the police. In other words, citizens that saw the measured neighborhood matters as problems, were more likely to voice less positive attitudes towards the police than were individuals who did not see these matters as problems. This finding was consistent across all three models and the standardized coefficient in each model indicated that the influence of this variable was strong. Citizens appear to hold the police responsible for not controlling community conditions before they reach some threshold where they are viewed as problems. This finding was anticipated. Unfortunately, prior research has not consistently addressed citizen perceptions of neighborhood issues when explaining attitudes toward the police. Several researchers have asked respondents about their perceptions of crime in their neighborhoods (Percy, 1986) and about neighborhood safety (Apple and O’Brien, 1983), though they have not assessed concerns about a range of neighborhood conditions which according to the findings herein appear to influence attitudes towards the police. In light of the findings of Stipak (1979), especially, that objective crime measures have only minimal influence on attitudes the present findings take on added importance. It seems plausible that citizens’ perceptions of neighborhood conditions provide 200 salient information on which people base their attitudes toward the police. Neighborhood conditions are more immediate and personal than crime data and information concerning these matters is more likely to be cognitively accessed by respondents. At the same time, it should be noted that respondents’ worries about becoming a crime victim in their neighborhood had only minimal impact on attitudes. In all three models the direction of this variable’s coefficients were as expected. People who were more worried about being victimized were more likely to hold less positive attitudes. Another finding that was fairly consistent across the models was that the impact of several demographic variables was greatly increased when indirect effects were also taken into account. Respondents’ age had a fairly large (when compared to other variables in the models) positive influence on attitudes towards the police. This confirms the findings of prior research that has normally found that older individuals possess more favorable attitudes towards the police than younger individuals. In the present study, age had substantial indirect effects through perceptions of neighborhood problems. Generally, older individuals voiced less concern about neighborhood conditions than did younger individuals. This might have resulted because these individuals are more concerned about more pressing immediate concerns, such as, personal income and physical health. Also, these people may 201 be less likely to actually encounter the listed conditions, especially if they are elderly and spend substantial time inside. In contrast, younger individuals who are likely to out on the street may perceive the listed conditions (e.g., people loitering, drinking on the street, vacant homes, etc.) as threatening to their own security and their families. The variable race also was determined to have substantial effects on attitudes in each model. The standardized coefficients indicate that whites had less favorable attitudes toward the police than did nonwhites. This principally occurred through the fact that whites saw neighborhood conditions as problematic, which in turn, had the effect of deflating attitudes towards the police in each model. Whites were also more likely to hold less positive prior attitudes than did nonwhites. Race was also determined to influence respondents’ evaluations of experiences with the police (see Table 6.2, Table 6.5, and Table 6.9). In those situations where race was found to have a statistically significant influence on the evaluation, whites were more likely to provide a negative evaluation and less likely to provide a positive assessment of officer performance during the contact. Furthermore, this result occurred when the effects of prior attitudes on the evaluations were controlled. These findings are contrary to prior research that has fairly conSistently found that whites have more positive 202 attitudes than do nonwhites. At the same time, the observed impact of race on attitudes should not be that surprising. It should be remembered that in the study site whites comprise a substantial minority of the population and the police force is primarily composed of African—Americans. Thus, whites may be perceived as merely voicing the opinion that is commonly attributed to minority groups in existing research. Also, many of the white residents may be individuals who lived in these neighborhoods during periods when their neighborhoods were stable middle class communities. As such, they may now see conditions in the community deteriorating and hold the police partially (or more fully) responsible for the changing conditions. If true, then the observed attitudes are not counter intuitive. The impact of income on citizens’ attitudes varied across the models. With global and order maintenance attitudes the impact of family income was positive. In contrast, the influence of income on drug related performance perceptions was negative. In each instance, the variable’s indirect effects were substantial through perceptions of neighborhood conditions. This may have resulted because wealthier individuals saw drugs as having a greater impact on their lives than other neighborhood conditions. For instance, these individuals may believe that substantial drug problems in the study sites are responsible for decreased property values. Also, they may see drug related activity as restricting their ability to 203 freely move about in their community. The effect of gender on attitudes followed a similar pattern as income. Namely, the total influence of gender (being female) was positive in the global and order models, while it was negative in the drug model, though the total effect of this variable was weak. The findings concerning income and gender suggest that respondents View drug related police behavior differently than the police conduct in the other models. At the same time, community members also appear to have different concerns about drug activities in the community than they have with other neighborhood conditions. If the direction of the coefficients for the indirect effects are compared across the three models, one finds that there is consistency in the order and global attitude models. In contrast, many of the variables in the drug model exert opposite influences on the dependent variable. This might occur because drugs pose a threat that is more salient to many residents. Also, the attitude measure refers-to a specific problem that police are to regulate, and furthermore, a community activity that many people see as problematic. The other attitude measures refer to either no specific performance (global) or one that is less clearly defined (order maintenance). The impact of the neighborhood context variables was mixed. Neighborhood race exerted a weak positive impact in two of the three models (global and order). The effect of 204 this variable was consistent with expectations and also was consistent with the findings concerning race. Namely, as percentage of nonwhites in the neighborhood increased, so did attitudes toward the police in these models. Since nonwhites generally held more favorable attitudes than whites this finding is congruent with the assumption that community context influences the attitudes of people in the neighborhood. However, the findings concerning the influence this variable has on perceptions of performance regarding drugs is not consistent. Finally, the attitude models provide important information concerning the influence that pre-existing attitudes have on subsequent outlooks toward the police. Prior research has primarily been concerned with the relationship between perceptions of officer performance in incident specific situations and more general attitudes toward the police (Skogan, 1991; Dean, 1980, Scaglion and Condon, 1980). The presumption is usually that more specific incidents influence more general attitudes. The present study provides insight into not only this relationship, but also the influence that prior general attitudes have on perceptions of incident specific situations. In addition, the study also looks at the role played by prior general attitudes on citizen attitudes about police performance in specific situations. First, pre-existing attitudes were found to exert substantial influence on present attitudes in all three 205 models. This construct has not been included in prior models attempting to explain citizen attitudes toward the police. The present study suggests that the inclusion of such a measure is necessary at a minimum as a control so that the individual impact of other variables in the model may be more properly assessed. When the direct effect of the variables in the models are computed without including pre-existing attitudes several differences appear. The amount of explained variance in each model is decreased. Specifically, in the global attitude model the adjusted R2 goes from .45 to .31, in the drug related performance model it decreases from .46 to .34, and in the order maintenance model from .43 to .32. Also, the importance of evaluations of experiences with the police is increased when prior attitudes are not included. In the global attitude model, all of the same experiences are significant though the coefficients are larger. Also, negative evaluations of experiences approach significance (.065). In the other two models additional incident specific encounters become significant. In the model of citizen perceptions of drug performance by the police positively evaluated victimization experiences are now statistically significant, while positive evaluations of stops become significant in the model of citizen attitudes concerning police maintenance of order on the streets and sidewalks in the neighborhood. Also, as with the global attitude model, the standardized coefficients of variables 206 are often of greater magnitude. Thus, the failure to include prior attitudes appears to inflate the significance of incident specific evaluations of police performance. The impact of pre-existing general attitudes on specific evaluations was also addressed in the study (se Table 6.2, Table 6.5, and Table 6.8). Due to the coding of variables, one would anticipate that the direction of the effect would be positive with the positively evaluated situations, and negative with the negatively evaluated encounters. For instance, positive general attitudes likely contain favorable information about the police that is used to evaluate specific police performance. As a result one would expect people with positive attitudes to similarly evaluate specific performances, and to be less likely to negatively evaluate these situations, assuming the general attitude is transferable to the specific situation. This is confirmed in many of the situations measured in the models, and in all of the experiences where the influence of prior attitudes was found to be statistically significant. Preexisting stereotypes may bias the processing of the information acquired during the experience so that it conforms to the preexisting attitude. It also appears that the influence of prior attitudes varies by type of experience and the influence was greatest when global attitudes are used as the measure for prior attitudes. This seems reasonable since previous drug related attitudes may not be transferable to specific 207 situations, whereas global attitudes are by their nature more general. The present study also addressed the effect that specific evaluations have on general attitudes. The influence of these experiences varied across type of encounter. One general trend was that all but two of the situations that were statistically significant, involved situations where the citizen initiated the encounter with the police. These are situations where individuals likely expect police to be more supportive and less authoritarian. The behavior of officers during these encounters appears to influence overall attitudes of the public concerning the police. CONCLUSION Several of the variables exert substantial total effects that are not accounted for in models that only concern themselves with the direct effect of variables. This was most pronounced with the demographic and neighborhood context variables. In addition, models that fail to include pre-existing general attitudes may overstate the influence of other variables in their explanations of citizen attitudes. Finally, neighborhood conditions appear to exert substantial effects on citizen attitudes as they appear to hold police responsible for the existence of these conditions. 208 ENDNOTES 1. Due to concerns voiced concerning the use Of least squares regression with ordinal level dependent variables, the impact of all variables hypothesized to have a direct effect on the dependent variable in each model was also computed using an ordinal probit technique. Without exception, the results for each of the three models were quite similar. Specifically, all significant variables in the regression equations were also significant in the probit analyses. Also, the direction of all coefficients were the same. The remainder of the information obtained in the probits (e.g., likelihood estimates, etc.) is not directly comparable. Since the concern was only with whether specific variables remained significant, the fact that least squares regression tends to introduce bias concerning the estimates of coefficients was minimized. 2. The direct effects are presented in tabular form for these three mo 209 CHAPTER 7 DETERMINANTS OF COPRODUCTION: FINDINGS AND DISCUSSION This chapter focuses on the determinants of coproduction, the behavioral component Of the study. Since a central purpose of this study was to assess the hypothesized relationship between citizens’ attitudes toward the police and coproduction, attention is first directed at the relationship between measures of these two constructs. Next, an empirical assessment of the coproduction models using path analytic techniques is provided. The chapter concludes with a discussion of the findings in light of the hypotheses contained in Chapter 4. Correlatioge Extant research contends that citizens who possess favorable attitudes toward the police are more likely to engage in behaviors that contribute to police performance than are individuals who have less favorable attitudes (Brandl and Horvath, 1991; Zamble and Annesley, 1987: Goldstein, 1987; Scaglion and Condon, 1980; Bell, 1979: Stipak, 1979; Thomas and Hyman, 1977). One way to test this relationship is to correlate the three attitude measures (global attitudes, order maintenance attitudes, and attitudes toward police performance of drug related activities) with the individual behavioral measures. A second manner in which to examine this supposed relationship is to correlate the attitude responses with the behavioral 210 scales that were developed for each model. This is suggested because individual respondents may decide to engage in a range of behaviors that are theoretically related to the attitude. It was therefore hypothesized that favorable attitudes (Wave 2) would be positively related to citizen behaviors (Wave 3). Additionally, scores on the attitudes toward the police questions were expected to be more highly correlated with scores on the behavioral scales than with performance or nonperformance of each of the separate behaviors from which the indexes were derived. The attitude-behavior correlations are presented in Table 7.1. The data do not support the assumption that favorable attitudes would be positively related to the measured behaviors. Only six (18%) of the correlations involving single act behaviors were statistically significant. Three of the six behaviors involved reporting information about a crack house to the police, while a fourth private behavior concerned reporting illegal activity to the police. All four of theses relationships were negative with less positive attitudes related to coproductive behaviors. The remaining two statistically significant relationships involved participation in community activity. As to these two situations, the correlation coefficients indicate that participation is positively related to citizens’ attitudes concerning two specific areas of police performance. In these situations, participation in community group 211 TABLE 7.1 CORRELATIONS BETWEEN ATTITUDES AND BEHAVIORS Attitudes Behaviors General Order Drugs Collective Self Help Community Meeting .019 .039 .001 Block Meetings .067 .096 .114 Participation Activities .123 .161 .179 Report Drugs To Comm -.011 -.074 -.054 Report Crack To Comm .002 -.024 -.067 Report Illegal To Comm -.013 .033 -.056 Report Suspicious To Comm .067 .089 .063 Behavioral Scale .064 .084 .059 Private Behavior Report Drugs To P01 -.053 -.096 -.095 Report Crack To Pol -.195* —.176* -.207* Report Illegal To Pol -.122 -.155* -.136 Report Suspicious To P01 .018 -.071 .001 Behavioral Scale -.125 -.171 -.153* Drug Behaviors1 Behavioral Scale -.103 -.138* -.156* Private and Collective Behaviors1 Behavioral Scale -.010 -.016 -.027 * = .01 significance Individual behaviors composing these scales are reported in the collective self help and individual behavior categories. 212 activities was positively related to order and drug performance attitudes. When all of the correlations involving individual behaviors are examined two additional features are evident. First, when the direction of the relationships is examined, one finds that in sixteen of the thirty-three (48.5%) correlations there is a negative relationship. This is most pronounced with private behaviors where nine of the twelve (75%) are inverse relationships. Second, when the magnitude of the correlations is examined all but ten (30.3%) of the correlation coefficients are below the .10 level. Furthermore, of these ten correlations only six of them are above .150. As such, the observed relationships are generally weak (see Table 7.1). The correlations between the attitudes and behavioral scales provided mixed results. In general, the magnitude of the relationships between the behavioral scales and the citizens’ attitudes were greater than the magnitude of the relationships between the same attitudes and individual behaviors. As can be seen in Table 7.1, this is most evident in the situations involving private and drug behaviors where all but one of the single act correlation coefficients are less than the scale coefficients. Unfortunately, the coefficients indicate that there is an inverse relationship between the behavioral scale and each attitude. Furthermore, the observed relationships are weak to moderate and range from .103 to .171. Finally, for three 213 of the four types of conduct the correlations between the behavioral scale and the general attitude were weaker than those involving the same scale and the two more specific attitudes. Path Analveie. All of the coproduction models were estimated using path analysis. This techniques was utilized so that the direct and indirect effects of variables in the models could be examined. In this portion of the chapter, Model 2.1 involving private behaviors is first addressed, and is followed by discussions involving collective behaviors (Model 2.2), citizens’ drug related conduct (Model 2.3), and finally, a model that includes all of the measured behaviors (Model 2.4). The discussions pertaining to each model briefly examine those variables that had substantial direct and indirect effects on the four categories of citizen behavior. Several variables receive special attention in these discussions since they are.responsible for explaining more of the variance in the dependent variable than any of the other measures. Following the individual examinations of the four models a compariSon of the results across the behavioral categories is presented. Private Behaviors Direct Effects. As can be seen in Table 7.2 the direct effect of attitudes toward the police on private action by community members is minimal. In addition, the coefficients 214 TABLE 7.2 DIRECT EFFECT OF ALL VARIABLES IN MODEL 2.1: PRIVATE BEHAVIORS Variable B Beta T Sig T Demographic Agel .008 .160 1.207 .228 Age2 .001 .043 .298 .765 Race .359 .173 2.471 .014 Gender .049 .023 .396 .692 Income .008 .010 .147 .883 Education .039 .048 .726 .468 Ngh. Inc .018 .010 .174 .862 Ngh. Race -.001 -.028 -.409 .682 gpege ip Neighborhood Rent/Own -.058 -.056 -.982 .326 1 Year -.026 -.038 -.629 .530 Neighborhood Conditions Problems .097 .323 4.700 .000 Collective Action Help Each Other .098 .050 .743 .458 Others Will Report -.036 -.091 -1.339 .181 Eff. Residents -.006 -.009 -.146 .884 Attitudes Global -.003 -.002 -.003 .997 Drug -.032 -.033 -.460 .646 Reported significance levels are for two-tailed test 215 for the attitude variables were insignificant and in a direction opposite to that hypothesized. Due to these factors, one could say, with caution, that a decrease in the favorableness of a person’ attitude is related citizen willingness to engage in the measured behaviors. Several other variables in the model were found to influence private behaviors. Mostnotably, perceptions of neighborhood problems had a substantial direct effect. The coefficient for perceptions of problems indicates that as perceptions of neighborhood matters as problems increase, so does private action. This confirms what was hypothesized. Respondents’ race was also found to have a direct positive influence on private action. Whites were more likely than nonwhites to engage in private behaviors. This finding confirms what some have noted; namely, that nonwhites are often less likely than whites to call the police directly, and instead rely more on local community groups if they are going to participate in coproduction. The results concerning the collective action variables were mixed. The effect of each variable on private action was weak, with the belief that neighbors would call the police exerting the strongest influence. As Table 7.2 indicates, increases in the belief that others will act, tends to decrease behaviors by individual citizens. Indirect and Total Effects. The indirect effects of applicable variables are generally minimal (see Table 7.3 and Figure 7.1). This is especially true for the 216 III-ulna... .II. lliil ~.> _ MS... 2302.25. an“: the-32.3.3. -=. V .— o 7.... 2.6.3.3.. a... 4 .. . \\ 7.... 2.3.5.5.: a... v ._ . . . .47 east“. :2. \ x/ 7. 302.: .02 mos. ZO_._.MSQQ 0pm>flhm a.> muzmam 217 TABLE 7.3 DIRECT, INDIRECT AND TOTAL EFFECTS FOR MODEL VARIABLES PRIVATE BEHAVIORS: MODEL 2.1 Variable Direct Indirect Total Demographic Agel .160 -.002 .158 Age2 .043 -.004 .039 Race .173 .000 .173 Gender .023 -.001 .022 Income .010 .002 .012 Education .048 .000 .048 Ngh. Inc .010 -.001 .009 Ngh. Race -.028 -.003 i -.031 Stake i Ngh Rent/Own -.056 -.004 -.060 1 Yr. Live -.038 .080 .042 NQD. Conditions Problems .323 .014 .337 Collective Action Eff. Resid. -.009 -.009 Others -.091 -.091 Report Help Each .050 .050 Attitudes Global -.002 -.002 Drug -.033 -.033 218 demographic variables which are hypothesized to have indirect effects through the two attitude measures. The weak direct effects of these two variables (global attitudes and perceptions of drug related performance) cause the indirect effects of variable operating through them to be reduced. For instance even though both age variables have significant direct effects on global attitudes, attitudes do not substantially influence citizen coproduction. Thus, the effect of age on coproduction through the intervening variable is limited. The single variable that does indirectly contribute to coproduction is one of the "stake in the neighborhood" measures. Beliefs by respondents’ that they will reside in the neighborhood for at least one more year has a strong positive direct effect on perceptions of community problems, and in turn, a moderate indirect effect on private action through this same variable. In accord with the relationship hypothesized, attachment to the community, as operationalized here, was positively related to community problems and to willingness to participate in private action. In fact, the indirect effect of this variable through neighborhood conditions exceeds its direct effect on coproduction. Collective Behavior Direct Effects. The findings concerning the direct influence of the model variables on collective action are provided in Table 7.4. The findings confirm the previously 219 presented correlations between attitudes and coproduction measures. Namely, neither citizens’ global attitudes nor drug related attitudes were significant predictors of citizen involvement in community level coproduction. Furthermore, the direction of the regression coefficients were not positive as anticipated. Thus, it appears that factors besides the measured attitudes explain variation in coproduction. Two of the demographic variables exert statistically significant influences on willingness to engage in collective behaviors. Specifically, the coefficient for individuals over age fifty (AGE2) was significant and positive; as age increases, so does participation. This finding was contrary to the relationship hypothesized. Normally, it is believed that as people get older they are less likely to be involved in collective action. It should also be noted that AGE1 was also positive, though not significant. Respondent’s educational level was also found to be highly significant. In addition, the direction of the education coefficient was positive. This finding supports prior research (Rosenbaum, 1987; Haeberle, 1987) which suggested that education is positively related to participation in neighborhood based anti-crime activities. Apart from the demographic variables, several of the collective action variables were significant. As anticipated, perceptions of a supportive environment (i.e., 220 TABLE 7.4 DIRECT EFFECTS OF ALL VARIABLES IN MODEL 2.2: COLLECTIVE BEHAVIORS Variable B Beta T Sig T Demographic Agel .024 .136 1.018 .309 Age2 .023 .235 1.632 .103 Race .741 -.111 -1.553 .121 Gender .220 .031 .545 .586 Income .011 .004 .059 .952 Education .456 .173 2.616 .009 Ngh. Race .016 .101 1.491 .137 Ngh. Inc .242 .043 .716 .474 WEED. Rent/Own .211 -.063 -1.080 .278 1 Year .090 .041 .668 .504 ngy Conditions Problems .106 .110 1.593 .112 Collective Action Eff. Groups .304 .106 1.781 .076 Neighbors Report -.130 -.102 -1.481 .139 Help Each Other .796 .125 1.859 .0641 Attitudes Global -.213 -.067 -.917 .359 Drugs -.007 -.002 -.034 .973 221 residents help each other) exerts a positive effect on participation in collective action. Similarly, a second variable that tapped perceptions of the effectiveness of community groups was also significant. The coefficient for this variable indicates that individuals who believe in the efficacy Of community groups, are more inclined to participate in coproduction activities with fellow residents. Surprisingly, respondents’ beliefs that others will report was not significant, though the coefficient approached the accepted significance level. Furthermore, the direction of the coefficient was negative as the collective action literature would suggest. Namely an increase in beliefs that others will participate, decreases the likelihood that a respondent will personally act. Indirect and Total Effects. When the indirect effects are computed, there is very little change observed (see Table 7.5 and Figure 7.2). This is primarily due to the weak direct effect of the attitude variables that were hypothesized to intervene between the demographic variables and variables operationalizing stake in the community. The largest indirect effect (which was still small) appeared with one of the variables representing attachment to the community (i.e., will you live here one year from now), and this effect was principally through perceptions of neighborhood conditions. In other words, people who intend to reside in the community for at least another year, are 222 a , a» . .5... 226229. .3. .52....5. .c. 4 a . 7.». ..u=c..c.s. a... v .. 4 .v...— 23:39.8:- 63. v c. a 3.33%. - moé . .. . 20.2056... . . .. -. 02:25.. .:awuuummmmmmmmwmrx .. .. zsozso .IIHWIHIIIIIiiiiiIIIiii . . . . ~.o_...:..._._e:. HEHIIIIIIIIII 3.3 I." I. Tit, III 00<~=._02 -_ .8 . / oz.:oz . 2.. 5.0.5.2 :02 .5020 5:200 .._0 .._.._..._ #20;st ._ .02 mp0a>mn0m 0>au00aaoo m.> musmam 223 TABLE 7.5 DIRECT, INDIRECT AND TOTAL EFFECTS FOR MODEL VARIABLES COLLECTIVE BEHAVIORS: MODEL 2.2 Variable Direct Indirect Total Demographic Agel .136 -.016 .120 Age2 .235 -.024 .211 Race .111 .006 .117 Gender .031 -.002 .029 Income .004 .006 .010 Education .173 9.003 .170 Ngh. Inc .043 -.005 .038 Ngh. Race .101 .001 .102 Stake i Ngh Rent/Own -.063 -.003 -.060 1 Yr. Live .041 .032 .073 ngy Conditions Problems .110 ' .024 .134 Collective Action Eff. Groups .106 .106 Others -.102 ' -.102 Report Help Each .125 .159 Attitudes Global -.067 -.067 Drug -.002 -.002 224 more likely to see local conditions as problematic, which in turn, causes an increase in participation in the range of provided collective action options. The perceptions of neighborhood problems variable also had indirect effects on the dependent variable, with the indirect effects primarily through the global attitude measure. As can be seen in Table 7.5 "Age2" exerts the strongest total influence on collective action. Namely, an increase in age for people over fifty is related to an increase in collective action. As noted, this finding does not support prior studies which have suggested that collective action decreases for the elderly. ngg Related Behaviors Direct Effects. Table 7.6 portrays the direct effects of variables contained in the model depicting citizen participation in anti-drug efforts. These behaviors included telephone calls to the police or a community group to report suspected drug activity. As previously mentioned drug problems were substantial in all of the study sites. Therefore, to not know about any such activity would appear to be the result of a conscious decision by a citizen to not get involved. In this model attitudes toward the job police are doing controlling the sale and use of illegal drugs had a statistically significant effect on drug related coproduction. The direction of the coefficient indicates that the effect was negative. Thus, a decrease in citizens’ 225 TABLE 7.6 DIRECT EFFECT OF ALL VARIABLES IN MODEL 2.3: DRUG BEHAVIORS Variable B Beta T Sig T Demographic Agel .014 .288 2.088 .037 Age2 .006 .222 1.491 .137 Race .186 .098 1.341 .180 Gender .027 .014 .231 .817 Income -.012 -.016 -.230 .818 Education .078 .105 1.529 .127 Ngh. Inc .031 .019 .316 .752 Ngh. Race .002 .005 .072 .942 geege ip Neighborhood Rent/Own -.066 -.070 -1.171 .242 1 Year -.010 -.016 -.264 .791 Neighborhood Conditions Problems .106 .162 2.490 .013 Collective Action Help Each Other .081 .048 .652 .515 Others Report -.017 -.048 -.682 .495 Eff. Groups .041 .051 .774 .439 Eff. Residents .019 .030 .450 .653 Attitudes Global -.025 -.028 -.379 .704 Drug -.125 -.137 -1.852 .065 Reported significance levels are for two-tailed test 226 ratings of police performance is related to an increase in participation. The direction of the effect was not as hypothesized. Still, the findings seem reasonable. A person not fully satisfied with police performance may help the police because they do not believe that the police are able to control the problem themselves. As in the prior models, citizen perceptions of community issues as problems (here drugs) also exerted a strong positive effect on the measured behaviors. Specifically, as perceptions of community drug behavior worsened, citizen participation in anti-drug behavior increased. This activity was presumably intended to remedy a problem believed to exert a negative impact on quality of life in the community. Several of the demographic variables also had substantial direct effects on coproduction as operationalized in this model. Most notable is the effect of the AGE1 variable. As can be seen in Table 7.6, this variable (age of individuals eighteen to fifty) had a strong positive effect. The other age variable (AGE2, individuals over age 50) also had a strong positive effect, though not of the magnitude of the measure pertaining to younger individuals. Participation, therefore, appears to increase across all ages, though the increase is greater as individuals approach age fifty. The coefficients for race and educational level suggest that these factors have direct and positive 227 influences on participation. Though neither coefficient was statistically significant according to traditional standards, educational level approached the significance level using a one-tailed test. Indirect and Total Effects. Figure 7.3 reveals that several of the demographic variables had substantial effects on the attitude variables. However, due to the weak direct effect of global attitudes on drug coproduction, the indirect effects of these individual level measures through global attitudes was limited. Of the demographic variables, the largest indirect effects (though still minimal) were observed for the age variables primarily through attitudes toward police performance of drug enforcement activities. Perceptions of community problems also indirectly influenced coproduction through the two attitude measures. The indirect effect of perceptions of community problems through drug attitudes was substantial (.041) in comparison to the indirect effects of other model variables. More specifically, as perceptions of drug problems increased, attitudes toward police behavior decreased, which in turn, increased participation. Generally, the same substantive pattern was observed for the indirect effect of neighborhood drug problems through global attitudes, though the indirect influence on coproduction was much weaker (.007). 228 as . MS... 2.6?sz / -¢U- z—C—.§— cc. V a. 0. out. out-.....CU- n... V o- o .b- 5.. do. be". 7.8. —r.:£—..3.€ £0. 0 o. o . .2084. . \ ..I.I--.I..I. \ \. noes—52 - ... 62.. 68.. .. 5n... \ \\ ......< 0.55 \ ... ee\ \ \ . ... 3.. \. .\ \ .\ \\ \ .\\ ‘\ u \Il \ IIIIIIIIIIII .. ...—05. . 0 III 2. III: ..S\ ......< 42004.0 4“../ 20.5.5000... i--. ill/4X \ \III. I//.I/ / 02.25. iii 3.. - - HHHWIIII ..IIIII-......I. III I msozmo 2 4.. ...merIyHlI........WIII.I.I_:. - ----....II ,..-...II- “III Sc. - N 00< ~.o.><:am ....HIIHHHHHH.I.II.-- ---. I .Ii --.II-i.I-......ii- ‘1§//../I///a.lll.f!ti-l ... .I. I.II.II.- . . II... I,..I. . .i Jill. .— C28. . . .— mo< II/ .I. III...IIII I .I...I.fllai . I/I III/I ...co. ..... / //-. III- III 85.52 5:. III! . 02.102 p.29. :62 / 3.2.5.82 9596 ...o ......n. .50....Dm .:02 ...0 ......— mp0a>m20m pmumamm Opuc m.e whomam 229 TABLE 7.7 DIRECT, INDIRECT AND TOTAL EFFECTS FOR MODEL VARIABLES DRUG RELATED BEHAVIORS: MODEL 2.3 Variable Direct Indirect Total Demographic Agel .288 -.021 .267 A962 .222 .014 .236 Race .098 .006 .104 Gender .014 -.016 .002 Income -.016 .008 .008 Education .105 .001 .106 Ngh. Inc .019 -.011 .008 Ngh. Race .005 -.015 -.010 Rent/Own -.O7O -.009 -.079 1 Yr. Live -.016 .009 -.007 th. Conditions Problems .162 .048 .210 Collective Action Eff. Groups .051 .051 Eff. Resid. .030 .030 Others -.048 -.048 Report Help Each .048 .048 Attitudes Global -.028 -.028 Drug -.137 -.137 230 _;; Behaviors Direct Effects. The direct effects of variables in this model on willingness to engage in the all of the measured citizen behaviors are displayed in Table 7.8. Several of the results are similar to those observed for the other models, yet there also several differences. These differences may be attributable to the range of types of activities citizens may engage in and the forms of community conduct to which the behaviors are directed. Two of the collective action variables were statistically significant and in the expected directions. An increase in community beliefs that fellow residents will report observed activity to the police, decreases participation by the individual holding such an outlook. This finding is in accord with the collective action research that claims that people will not get involved in activities if they can receive the same benefits through inaction, a less costly choice. Furthermore, respondents who believed that other residents were supportive, in that they help each other, were more likely to engage in coproduction (see Table 7.8). As noted, there are also several similarities between the results for this model and those previously discussed. First, perceptions of community conditions as problematic exerted a strong positive direct effect. Second, the coefficients for respondents’ educational level indicate that this variable also had a positive influence on 231 TABLE 7.8 DIRECT EFFECTS OF ALL VARIABLES IN MODEL 2.4: ALL BEHAVIORS Variable B Beta T Sig T Demographic Agel .026 .220 1.63 .105 Age2 .016 .230 1.58 .114 Race .064 .014 .198 .843 Gender .176 .038 .642 .521 Income -.006 -.003 .005 .996 Education .267 .151 2.258 .024 Ngh. Race .006 .061 .893 .372 Ngh. Inc .188 .050 .819 .413 §L§K§ 12 Egg; Rent/Own -.154 -.068 -1.162 .246 1 Year .037 .025 .411 .681 Ngh. Conditions Problems .159 .244 3.485 .000 Collective Action Eff. Groups .161 .084 1.297 .195 Eff. Residents .055 .036 .545 .586 Neighbors Report -.107 -.126 -1.795 .073 Help Each Other .684 .159 2.333 .020 Attitudes Global -.090 -.042 -.576 .565 Drugs -.100 -.046 -.635 .526 Reported significance levels are for two tailed test 232 coproduction. Third, the two age variables each approached statistical significance (one-tailed test) and were positive. Finally, the standardized coefficients for the two attitude measures were negative and indicate that they each have a weak influence on coproduction. Indirect and Tgta; Effects. Figure 7.4 displays the standardized regression coefficients for all paths in the model. As can be seen in Table 7.9 the indirect effects of variables were generally consistent with those observed for the prior models. Namely, most of indirect effects are through either perceptions of neighborhood problems and/or one of the variables used to operationalize stake in the neighborhood. Most notably, the indirect effects for residential stability (will live there one year from now) were almost three times the variable’s direct effects (.069 to .025). Most of this variables indirect effects (.061) were through perceptions of neighborhood conditions. At the same time, neighborhood problems, through the two attitude variables indirectly influenced coproduction (.033). Very minimal indirect effects were observed for the other variables in the model. DISCUSSION A major objective of this portion of the study was to assess the relationship between attitudes toward the police and citizen coproduction of police outputs. The data lead to the conclusion that attitudes toward the police are not highly correlated with coproductive behaviors. Furthermore, 233 .I ~; _ m5..— .nc. 28.5.93. .0. v .. . 7.3 2.55.?! 9.. v .— . .0... 28.5.2.6 no. v .. . nmo. .Imflflfl IIlliIlii'I'l \ \zsofizsm ...Ovn \ 6 p \\ flog—:02 ......I \X..\:\.. ..\z.. \\\ \/ // ...: /E \ 5:\ III t< oomo \.\ mos. \ A \ \ \\.II|II icy “. III I a. <:uo ..-/HIIIIIIHIHII I I. II III IIIIIIIIIWII... . 4 III. . .IIIII .IXI. one. i. ~ Mao’s 4.7!. /III . I 85:62 c. .... III/III. I]. {o 6.. o, /// EBB. :oz //.,. /_,/// 22693:. / 02:62 ...O “Em. LDOEO 3:200 .EOnEDm ...Oz :0 “Em. muofl>mnmm HH< v.6 whooflm 234 TABLE 7.9 DIRECT, INDIRECT AND TOTAL EFFECTS FOR MODEL VARIABLES ALL BEHAVIORS: MODEL 2.4 Variable Direct Indirect Total Demographic Agel .220 -.013 .207 Age2 .230 -.020 .210 Race .014 .003 .017 Gender .038 -.004 .034 Income .004 .006 .010 Education .151 -.007 .144 Ngh. Inc .050 -.004 .046 Ngh. Race .061 -.004 .057 Rent/Own -.068 -.006 -.074 1 Yr. Live .025 .068 .091 Ngh. Conditions Problems .244 .033 .277 Collegtiyg Action Eff. Groups .084 .084 Eff. Resid. .036 .036 Others -.126 -.126 Report Help Each .159 .159 Attitudes Global -.042 -.042 Drug -.046 -.046 235 attitudes are in most cases not statistically significant determinants of citizen behaviors that assist the police. As such, it appears that the relationship between citizen outlooks on the police and citizen conduct is not as direct as prior research has suggested. The fairly consistent finding of negative relationships between attitudes and behaviors in the models and many of the correlations may result because of several factors. Citizens with less favorable attitudes toward the police may coproduce because they believe that present police behavior is ineffective in producing community safety and security and that their local law enforcement agency needs input from community members. Alternatively, respondents with more favorable attitudes may believe that the police do not need their assistance since they are already performing satisfactorily. Citizens with positive attitudes towards the police may also feel that there is no need to get involved in community activities expected to support the police role. However, the positive correlations in the community self-help category indicate this may not be the case. More specifically, it appears that people with favorable attitudes do attend community and block club meetings, and participate in community level anti-crime activities. At the same time, they are less likely to report information to their neighborhood organizations. It is possible that actual group activities, where police and citizen 236 cooperation is stressed lead, to these more favorable attitudes. Alternatively, people with more favorable attitudes may see group activities as a way to solicit additional help for the police. The findings of the correlation analysis were generally supported by the results of the regression equations. In most all of the situations, attitudes towards the police were not significant predictors of the measured behaviors and were negative. This was especially true for the general satisfaction measure. However, citizens’ attitudes concerning the regulation of drug-related behavior was a significant determinant of citizen reporting to the police. This finding may have resulted because drug use is a serious problem in all of the areas surveyed. Support for this position may be found in the strong positive influence that perceptions of problems have on coproduction. Situations that people believe are serious and which the police are not seen as doing a credible job controlling/regulating, are factors that appear to influence citizen behavior. As to drug use, it is a behavior that is disruptive of respondents’ lives, citizens did not state overly favorable attitudes about police behavior relative to this problem and, furthermore, community members are able to use an anonymous hotline when contacting the police. Unfortunately, the citizen attitude in this situation was negatively related to citizen behavior. While one would not suggest that the police perform poorly to stimulate 237 citizen activity, one should not anticipate that increases in levels of citizen satisfaction will necessarily lead to increases in citizen behavior. As to the consistently weak relationships, they may have occurred because of problems with the specificity of the attitude measures. First, it could be argued that the attitude measures are so diffuse that they do not measure perceptions citizens have of the police in their neighborhoods. As such, the impact of the attitude measures on behaviors would not be expected to be great. However, it should be noted that respondents were requested to evaluate the behavior of the police in "their/your neighborhood" and two questions involved salient attitude objects. Second, some might claim that the attitudes and behaviors are not substantively related. As such, the attitude should not be expected to predict the behaviors included in the behavioral scales. Some support for this position may be found with the drug model (Model 2.3). This was the only model where the attitude object and the behaviors were on the same level of specificity (drug attitudes and specified drug behaviors), and also the only situation where the direct influence of an attitude measure was statistically significant. However, if one examines the behavioral scales they contain several alternative types of conduct in which the individual may engage and a wide range of illegal or suspicious conduct that may be the object of the coproductive behavior. To assert that general 238 attitudes toward local police performance is not substantively related to the listed behaviors does not seem correct. The path models also highlight the fact that other factors do a more satisfactory job explaining behaviors than do attitudes. Several variables in the model had a consistent relationship with the behavioral scales. Most notably, citizen perceptions of neighborhood conditions as troublesome consistently influenced citizen willingness to get involved. When the respondent felt that there were problems that probably needed attention, the likelihood of action individually or through collective behavior was enhanced. Possibly, people perceived these matters as impediments to improved living conditions and acted to lessen the impact of the problem. Naturally, if a condition was not problematic then no action would be necessary. Thus, citizen behaviors are probably responsive to conditions within their neighborhood context. Several of the demographic variables exhibited some consistency. Respondents’ educational level influenced the decision to partake in collective activity, drug related coproduction, and was also related to the scale that included all behaviors. It should be kept in mind that all of these scales had collective behaviors included within them and that the coefficient was largest with the collective action scale. In contrast, education only exerted a weak influence on private behaviors. People with 239 more education may be able to acquire a better understanding of community problems and available remedies. In addition, they may be more focused on remedies that have the potential to provide future benefits and improve the overall quality of life in the community. The findings concerning older individuals were generally contrary to what was expected. Basically participation in coproduction did not decrease for people over fifty. In fact, in the collective action situations this variable exerted a strong positive influence on behavior. The fact that some of the collective behaviors could be performed from within one’s home (e.g., calling your community group with information or to report something) may have caused this finding to be contrary to existing research findings. The impact of AGE2 (i.e., respondents over fifty years old) on drug related behaviors was even greater than with collective action. This resulted because of the indirect effects of this variable through attitudes concerning the job police were doing controlling drugs. The stake in the community variables provided mixed results. Whether the respondent rented or owned her/his home exerted a consistently weak influence that was often negative. The other variable that operationalized stake in the community, whether the person expected to reside in the community one year from now, had more of an impact on behaviors, though the effect was still limited. This 240 measure (resident stability) did have an indirect effect on behavior through perceptions of neighborhood problems. Namely, people who expected to remain in the community were more likely to see specified matters within the neighborhood as requiring attention. When these two variables are considered together, they tend to lend support for what has been termed the "neglect" (Lowery and Lyons, 1986) or "non loyalist" (Sharp, 1984) position. This position refers to residents who do not exit from the community, but at the same time, do not expect improvement in community conditions, and do not intend to participate in coproduction activities. It is important to keep in mind that many of the people in the study sites are "trapped’ in their community in the sense that they are financially unable to go elsewhere. Thus, they may be "in", but not apart of the community, and therefore may not be "committed" to improving life in the neighborhood. The findings concerning the collective action variables were inconclusive. Citizen beliefs that neighbors would report the observed behaviors had a negative influence (generally weak, except with "all behaviors") on coproduction. Existing research on collective behavior (Olson, 1971) supports this finding. Citizens often will not expend resources when they can receive the same benefits without having to become involved. Thus, there is no need to partake in community affairs when benefits obtained by the group will become available to all neighborhood 241 residents. Similarly, there is no need to act and become identified with a situation when you can acquire the same benefits if others act. Beliefs in the efficacy of community groups and individual residents had varying effects that were weak but positive as anticipated. Finally, in interpreting these findings it is necessary to remember that individuals are usually not joiners of groups or participants in anti-crime activity. Besides the public goods argument, there are situational factors that were not accounted for in the models that influence citizen conduct. After one of the drug reporting questions respondents were asked to state several reasons why they did not provide information. Individuals noted that they did not report known activities because they feared reprisals (18.5%) if they acted, believed that other community members would report (16.4%), had no proof that the behavior conformed to their beliefs (15.8%), and many (14%) felt that the police will not or could not do anything about the situation. Citizens also noted that they did not attend community or block meetings because they had to work (48%) or had other conflicts (25%). These factors would act to constrain the relationship between attitudes and behaviors. Several factors suggest that the findings should be taken with caution. First, as previously noted, there may be problems with the specificity of the attitude measures. Second, citizens could choose to engage in behaviors not part of the indexes; such as, target hardening activities. 242 Third, the results may not be generalizable to other communities. Still, the results suggest that assumptions about the presumed relationship between attitudes and citizen actions deserve further attention. 243 CHAPTER 8 CONCLUSION This study attempted to achieve two unique objectives. First, it sought to more fully explain citizen attitudes toward the police. In order to increase our knowledge concerning the determinants of citizens’ attitudes, a causal model was proposed that suggests the temporal ordering of variables, the need to account for both the direct and indirect effects of variables, and the importance of including prior attitudes in models that explain present attitudes toward the police. Second, the study also empirically examined the relationship between citizens’ attitudes toward the police and the public’s willingness to engage in the coproduction of police outputs. This was accomplished within the framework of a model that included factors that may enhance or constrain this attitude-behavior relationship (i.e., perceptions of neighborhood conditions, efficacy of residents and groups, etc.). Unfortunately, these factors are often ignored by criminal justice researchers who hypothesize that a direct relationship exists between attitudes toward the police and coproduction. At the same time, it is necessary to acknowledge that the present research suffers from several general limitations. First, the sample was restricted to residents living within four selected areas within the city of 244 Detroit. As such, the findings presented herein may not be generalizable to residents in other neighborhoods within the city and/or residents in other municipalities. However, apart from questions concerning the representativeness of the sample (see Chapter 5), there are other features of the study which suggest that the findings are generalizable. While one can not state definitively that specific law enforcement concerns in these areas are similar to other cities, it is likely that many large police departments are presently saddled with reduced budgets and increasing problems of crime and disorder. In addition, residential neighborhoods in urban areas are undergoing changes in the composition of their populations and many are suffering from economic and related social problems. Thus, the context within which the police department in this study found itself and the factors which were determined to influence citizen attitudes towards local law enforcement are probably comparable to other municipalities. As such, the constructs that had the most explanatory power (perceptions of neighborhood conditions, demographic variables, prior general attitudes, and evaluations of experiences with the police) would likely retain their theoretical importance in future models. For instance, citizen perceptions of neighborhood problems were found to influence both attitudes toward the police and citizen willingness to engage in behaviors that assist or support the police. This construct tapped whether 245 community conditions were perceived by community members as problems. As such, people perceiving the same conditions may have varying perspectives on whether the conditions are problematic. In addition, the importance of the construct is not constrained by the conditions that comprised the "perceptions of problems scale". Even residents of more affluent (and "nicer") neighborhoods are likely to perceive certain conditions within their residential areas as problematic, though the nature of the problems may vary from those in the selected study sites. One would expect that citizen perceptions of neighborhood conditions as problems in these more affluent communities would likely have similar effects on attitudes toward the police as those evidenced herein. A second limitation pertains to the study’s inability to fully examine the relationship between attitudes and behaviors. While one objective of the present study was to examine the relationship between attitudes and coproduction, attention was principally focused on the influence of attitudes toward the police on behavior (coproduction), and not the effect of behavior on citizens’ attitudes. This strategy was premised on prior research which has hypothesized a direct unidirectional link between these two constructs. In addition, recent police reformers have suggested that increases in favorable general attitudes toward the police are related to coproduction (Wyckoff, 1988; Goldstein, 1987). This study was unable to provide 246 information concerning the influence that citizens’ behaviors have on attitudes toward the police. This is unfortunate because it seems quite possible that behavior may be an important determinant of attitudes toward the police. For instance, coproduction may influence perceptions of neighborhood conditions as well as attitudes towards the police. Specifically, citizen behavior may have a beneficial impact on neighborhood conditions as community members act to remedy neighborhood problems. If so, findings from the present study indicate that citizens’ general attitudes toward the police may become more favorable. Also, coproduction, especially if engaged in jointly with the police, may increase citizen appreciation of the complexities of police work which may also influence attitudes toward police performance. A proper test of this relationship was not possible in the present study because sufficient behavioral data from Wave 2 was not available. Third, the coproduction models might be enhanced if data were available that accounted for the number of times individuals engaged in each of the measured behaviors. This would permit the dependent variable in these models to account for additional behaviors by individuals who select a preferred form of coproduction and then engage in that behavior on a repeated basis. Operationalizing this variable in such a manner would likely increase the number of behaviors in which some citizens engaged. However, there is nothing that would indicate that this 247 would necessarily change the results, though this remains a possiblity. For example, if measuring the dependent variable in this manner increases everyone’s measured behaviors then no differences in the findings would likely result. If the number of coproductive behaviors of people with less than favorable attitudes toward the police are increased, then a stronger negative relationship between attitudes and behavior may be found. Finally, if changing the operationalization of this variable only increases the number of behaviors of citizens with favorable attitudes then conclusions concerning the influence of attitudes on coproduction would differ from those reported. Still, one could easily argue that an individual who decided to get involved in helping the police would have engaged in a number of the available options, since most did not involve any increase in effort or other additional costs. Fourth, an additional limitation may be the failure to collect information concerning respondents’ attitudes relative to the efficacy of the police. Many respondents (14%) noted that they did not provide information to the police because they believed that the police could not, or would not, do anything with the information. These beliefs about the effectiveness of the police constrain a person’s willingness to engage in certain forms of coproduction. Specifically, a person having these beliefs may not think that her/his efforts will lead to beneficial behavior by local law enforcement officers. 248 In spite of these limitations, the findings contribute to the existing knowledge concerning the determinants of citizens’ attitudes as well as the relationship between attitudes and the coproduction of police outputs. As to the determinants citizens’ attitudes toward the police first, findings concerning the influence of race on attitudes are of substantive significance. An examination of the distribution of responses indicates that white respondents reported less positive attitudes toward the police than prior research has attributed to this group, while the responses of nonwhites were more consistent with prior research. Overall, the general attitudes of white respondents toward the police were less favorable than the attitudes of nonwhites. This finding was contrary to existing research which has consistently found that nonwhites have less favorable general attitudes about police performance than whites. In addition, this finding was confirmed by the study’s inquiry into the influence of respondents’ race on perceptions of police behavior during citizen experiences with the police. Specifically, whites were less likely to positively evaluate police behavior than were nonwhite respondents involved in these encounters. At the same time, whites were more likely to negatively evaluate officer conduct in these experiences with the police. While some may suggest that this finding resulted due to factors associated with the study site, there remains 249 another more plausible explanation. Namely, changes in the population of urban centers, changes in the racial composition of municipal governments, and the composition of police forces so that nonwhites hold a greater proportion of these positions, have resulted in whites becoming members of the minority group in these situations. What the present study’s findings may evidence is the assumption of attitudes by white respondents that were generally attributed to minority groups members. Replication of these findings in other large cities is naturally recommended. However, if confirmed, this finding may be suggestive of changing attitudes that this "new" white minority possesses towards urban police forces. Furthermore, police administrators need to recognize that this group may hold less favorable attitudes than previously assumed. The present data also indicate that prior research may have overstated the influence that citizen evaluations of experiences with the police have on general attitudes towards local law enforcement agencies and officers. The effect of evaluations of police-citizen encounters may have been overstated because existing research failed to control for the influence of prior general attitudes on present general attitudes. This study suggests that present attitudes are informed by the existing belief system of individuals. Furthermore, general attitudes were found to influence citizen evaluations of police behavior during police-citizen interactions. The full extent of this 250 influence may have been suppressed by the fact that many respondents did not have personal contacts with the police: thus, there was only limited variation in the experience variables. While extant research has been primarily concerned with the role of specific attitudes (i.e., evaluations of citizen experiences with the police) on more general attitudes (Percy, 1986; Dean, 1980; Jacob, 1971), the present study suggests that influence of prior general attitudes on present general attitudes is of substantive and statistical significance. Prior research appears to have also overstated the influence that experiences with the police exert on attitudes in comparison to other theoretically justified variables contained in prior models (Skogan, 1991; Percy, 1986: Scaglion and Condon, 1980). This finding likely resulted because most existing research utilized statistical techniques which prohibited examination of the indirect effects of included variables. Namely, when the total (indirect and direct) effect of several of the demographic variables (especially as to race, income, and education) are computed, these variables influence citizens’ attitudes toward the police to an extent equal to, or greater, than evaluations of experiences with the police. Relatedly, citizen expectations concerning police performance during assistance or information situations are apparently different than with victimization and stop encounters. Evaluations of these informal, but official, 251 citizen initiated contacts were found to exert statistically significant influences on more general attitudes toward the police, while the influence of evaluations of stop and Victimizations influences were minimal. It is quite possible that in these encounters citizens expect a less authoritarian and more supportive response from the police than in the other two situations. When the police behavior did not mirror these expectations, citizens negatively evaluated officer conduct which tended to decrease more general attitudes toward the police. A consistent finding across all three models was that citizens who viewed neighborhood matters as problems, were less likely to voice positive attitudes toward the police than were individuals who did not see these matters as problems. Unfortunately, prior research has not consistently addressed citizen perceptions of neighborhood issues when explaining attitudes toward the police. Several researchers have asked respondents about their perceptions of selected conditions (i.e., crime in their neighborhoods and neighborhood safety), though they have not assessed citizen views about a range of neighborhood conditions. In light of the findings of Stipak (1979) especially, that objective crime measures have only minimal influence on attitudes, findings relative to the impact of neighborhood conditions take on added importance. Neighborhood conditions are more immediate and personal than crime data and information concerning these matters is more likely to 252 be cognitively accessed by respondents. Citizens’ perceptions of neighborhood conditions provide salient information that seems to inform respondents’ attitudes toward the police. Several findings concerning the behavioral component of the present study are also worthy of further attention. Most notable was the finding that attitudes toward the police do not appear to have the influence on coproduction that prior research has hypothesized. In fact, an effect that is opposite of that hypothesized may result. Namely, people with less favorable attitudes may be more likely to engage in coproduction than individuals with favorable attitudes. Quite possibly community members see their role as helping police perform only when the department is unable to perform satisfactorily on its own. Several reasons exist why citizens who have favorable attitudes toward the police may have been less likely to coproduce. Individuals may hold favorable attitudes because they think the police are performing effectively on their own and that there is, therefore, no need to offer help. At the same time, citizens may voice favorable attitudes because police performance is meeting their expectations, whether it is at an objectively high level or not. To these people, engaging in behaviors supportive of the police may be perceived as a costly endeavor that will not provide commensurate benefits. This is especially true if these people believe that their local department is already 253 at its optimal level of performance. Naturally, one would not suggest that departments behave in a manner to reduce positive attitudes in order to stimulate coproduction. In addition, the present study also indicates that citizen perceptions of community problems play a substantial role in explaining not only attitudes, but also coproduction. An increase in citizens’ perceptions of neighborhood conditions as troublesome were found to have had a substantial negative influence on citizens’ attitudes toward the police. The data also indicate that when conditions in a community reach a certain threshold, where they are perceived as problems, people become more likely to get involved in actions intended to reduce the impact of the condition on their own lives or the community. In other words, when conditions get so bad that they are especially problematic people are more likely to act. However, when conditions improve, as determined by perceptions of community members, residents may cease to engage in coproduction activities. The impact of neighborhood problems on both attitudes and coproduction should be of special concern to police administrators. This is noted because many neighborhood problems are not susceptible to control or even regulation by the police. However, the present study seems to suggest that as far as citizen attitudes are concerned the public expects the police to do more about many of their concerns. Motivating citizens to engage in coproduction is 254 further compounded by the fact that as attitudes toward the police become more favorable citizens become less likely to coproduce. Thus, improving attitudes may deprive the police of a resource necessary to provide continued performance at a level to maintain favorable attitudes. In contrast, individuals with less favorable attitudes were more likely to coproduce. Finally, findings from the present study concerning the relationship between attitudes and coproduction do not support the central presumption upon which community policing is based. Proponents of this policing strategy believe that community oriented policing will lead to more favorable attitudes towards the police. It is then surmised that favorable general attitudes toward the police will cause community members to engage in behaviors that support or assist the police in the production of outputs. Three factors discussed in the present study question this presumption. First, the belief that putting more officers on the street, where they will likely have non-adversarial contacts with the public, will by itself increase attitudes toward the police may not be correct. Citizens may or may not approach officers on the street and/or be willing to talk to officers when contact is initiated by the officer. In addition, the assumption that non-adversarial contact alone (assuming that such occurs) will improve attitudes fails to recognize the complexity of the attitude construct. It 255 ignores the history behind some unfavorable attitudes and the fact that people enter these encounters with existing attitudes that influence the processing of information acquired during these contacts. Furthermore, the influence that perceptions of neighborhood conditions exert on attitudes is not accounted for in this assumption. Contact without behavior that reduces the impact of neighborhood conditions on residents is not likely to result in more favorable attitudes. Second, the data suggest that increasing public attitudes toward the police may not have the impact on citizen willingness to coproduce that is presumed by supporters of community policing. The data lead to the conclusion that increases in general attitudes may not be related to higher levels of coproduction. In fact, as noted, increases in levels of satisfaction may actually decrease levels of citizen participation. Third, the presumption that there is a direct causal link between attitudes and coproduction ignores other factors that might influence citizen willingness to coproduce. As previously mentioned, citizen perceptions of neighborhood conditions exert a substantial effect on the decision to coproduce. In addition, the extent to which the community is organized may impact on the decision to coproduce. Well organized communities probably provide more opportunities for citizens to act in a collective fashion. Organized communities might also foster a belief 256 among residents that private actions are also worthwhile. The existence of a community leader committed to citizen involvement in the police process also be a catalyst for coproduction. Finally, it has also been suggested that several features associated with collective action influence citizen willingness to get involved. Most citizens do not act when they can receive the same benefits through inaction. All of these factors suggest that improving attitudes of the public will not be an easy task for police agencies. At the same time, inspiring the public to become active participants in the police process will also be difficult. 257 LI ST OF REFERENCES LIST OF REFERENCES Aldrich, J. and F. Nelson (1984). Linear Probability, Logit. and Probit Models. Beverly Hills: Sage. Ajzen, I and M. Fishbein (1977). Attitude-behavior relations: A Theoretical analysis and review of empirical research. Psychological Bulletin, 84(5): 888-918. Andrews, K. H. and D. B. Kandel (1979). Attitudes and behavior: A specification of the contingent contingency hypothesis. American Sociological Review, 44:298-310. Apple, N. and D. O’Brien (1983). Neighborhood racial composition and residents’ evaluation of police 11:76-84. Asher, H. (1983). Causal Modelingl 2nd Edition. Beverly Hills: Sage. Babbie, E. (1983). The Practice 9; Sogial Research, 3rd Edition. Belmont, Ca.: Wadsworth Publishing Co. Bell, D. J. (1979). Police and public opinion. Journal 9; Police Science and Administratio , 7(2):196-205. Block, R. (1974). Why notify the police: The victim’s decision to notify the police of an assault. Criminolo , 11(4):555-569. Blumer, H. (1958). Race prejudice as a sense of group position. Pacific Sociological Beviey, 1: 3-7. Boggs, S.L. and S. F. Galliher (1965) Evaluating the police: A comparison of black street and household respondents. Social Problems , 13:393-406. Borgida, E. B. and B. Howard-Pitney (1983). Personal involvement and the robustness of personal salience effects. Journal 93 Personality agg Social Psychology, 45: 460-470. Brandl, S. and F. Horvath (1991). Crime victim evaluation of police investigative performance. Journal 2; Criminal Justice, 19:293-305. Brown, K. and P. B. Coulter (1983). Subjective and objective measures of police performance. Public Administration Review, 43:50-58. 258 Brown, L. P. (1989). "Community Policing: A Practical Guide for Police Officials," Perspectives pp Policing, Vol. 12, Washington, D. C.: U.S. Government Printing Office. Brudney, J. L. and R. E. England (1982). Urban policy making and subjective service evaluations: Are they compatible. Public Administration Review, 42:127-135 Brunner, J. A. and G. A. Brunner (1971). Are voluntarily unlisted telephone subscribers really different? Journal pf Marketing Research, 8:121-124. Christenson, J. and G. S. Taylor (1983). The socially constructed and situational context for assessment of public services. Social Sciences Quarterly, 63: 264-274. Cohen, L. E. and M. Felson (1979). Social change and crime rate trends: A routine activity approach. American Sociological Review, 44: 588-608. Cordner, G. (1988). "A Problem-Oriented Approach to Community-Oriented Policing", in Jack R. Greene and Stephen Mastrofski (eds.) Community Policing: Rhetoric 9; Reality, (Westport:Praeger), pp. 135-152. Cox, K. (1982). Housing tenure and neighborhood activism. Urban Affairs Quarterly, 18: 107-129. Cumming, E., I. Cumming, and L. Edell (1965). Policeman as philosopher, guide and friend. Social Problems, 12: 276- 288. Dean, D. (1980). Citizen ratings of the police: the difference police contact makes. Law and Policy Quarterly, 2:445-471. Decker, S. (1981). Citizen attitudes toward the police: A review of past findings and suggestions for future policy. Journal pf Police Science and Adm' istration, 9(1):80-87. Dennis, J. (1976). Who supports the presidency. Society, 13:48-53. Dillman, D. (1978). Mail and Telephone Surveys. New York: John Wiley and Sons. Easton, D. (1965). A Framework for Political Analysis. Englewood Cliffs: Prentice Hall. Eck, J. E. and W. Spelman (1987). Problem Solving: Problem Oriented Policing ip Newport News. Washington, D. C.: Police Executive Research Forum. 259 Ennis, P. (1967). Criminal Victimization in the United States: A Report pf a National Survey. Chicago: National Opinion Research Center. Erez, E. (1984). Self-defined desert and citizen’s assessment of the police. Journal pf Criminal Law and Criminology, 75(4):1276-1299. Farrell, M. J. (1988). "The Development of the Community Patrol Officer Program: Community Oriented Policing in the New York City Police Department," in Jack R. Greene and Stephen Mastrofski (eds.), Community Policing: Rhetoric 9; Reality (Westport: Praeger), pp. 73-88. Fazio, R.H. and C.J. Williams (1986). Attitude accessibility as a moderator of the attitude perception and attitude-behavior relations: An investigation of the 1984 presidential election. Journal pf Personality app Social Psychology, 51:505-514. Fishbein, M. and I. Ajzen (1975). BeliefI Attitude, Intention and Behavior: An Introduction To Theory and Research. Reading, Mass.: Addison-Westley. Fishbein, M. and I. Ajzen (1976). Misconceptions about the fishbein model: reflections on a study by songer-nocks. Journal pf Experimental Social Psychology, 12: 579-584. Fitzgerald, M. F. and R. F. Durant (1980). Citizen evaluations and urban management: Service delivery in an era of protest. Public Administration Review, 40: 585-594 Flanagan, T. J. (1985). Consumer perspectives on police operational strategy. Journal pf Pplice Science and Administration, 13(1):10-21. Furstenberg, F.and C. F. Wellford (1973). Calling the police: the evaluation of police service. Law and Society Review, 8:393-406. Gallup, G. (1982). The Gallup Report £200. Princeton, N.J.: Gallup Poll Goldstein, H. (1987). Toward community-oriented policing: potential, basic requirements, and threshold questions. Crime and Delingpency, 33:6-30. Greenberg, S., W. Rohe, and J. Williams (1983). Neighborhood conditions and community crime control. Community Action, 1(5):39-42. Haeberle, S. A. (1987). Neighborhood identity and citizen participation. Administration and Sppiety 19(2): 178-196. 260 Hahn, H, (1971). Ghetto assessments of police protection authority. Law and Society Review, 6(Nov.):183—194. Hindenlang, Michael J. (1974). Public opinion regarding crime, criminal justice and related topics. Journal pf Research ip Crime and Delingpency, July: 101-116 Hirschman, A. (1970). ExitI Voice. and Loyalty. Cambridge: Harvard University Press. Jacob, H. (1971). Black and white perceptions of justice in the city. Law and Society Review, 5:69-89. Johnson, D. R. (1981). American Law Enforcement: A History. St. Louis: Forum Press. Kalton, G. (1983). Introduction pp Survey Sampling. Beverly Hills: Sage. Kish, L. (1965). Survey Sampling. New York: John Wiley and Sons. Koenig, D. J. (1980). The effect of crime victimization and judicial or police contacts on public attitudes toward the local police. Journal pf Police Science epg Administration, 8:243-249. Krosnick, J. A. and H. Schuman (1988). Attitude intensity, importance, and certainty and susceptibility to response effects. Journal p; Personality epg Social Psychology, 54(6): 940-952. Kuykendall, J. (1974). Styles of community policing. Criminology, 12(2):229-240. Lavrakas, P. J. (1987). Telephone Survey Methods: Sampling, Selection, and Supervision. Beverly Hills, Ca.: Sage Publishing Co. Lavrakos, P. J. and E. Herz (1982). Citizen participation in neighborhood crime prevention. C imin lo , 20:479- 498. Liska, A. E. (1984). A critical examination of the Fishbein/Ajzen attitude-behavior model. Social Psychology Quarterly, 47: 61-74. ----------- (ed.) (1975). The Consistency Controversy: Readings pp the Impact pf Appippge pp Behavior. New York: Academic Press. ----------- (1974). Emergent issues in the attitude- behavior consistency controversy. Apepigep Spgipipgipei Review, 39:261-272. 261 Lord, C. G., M. R. Lepper, and D. Mackie (1984). Attitude prototypes as determinants of attitude-behavior consistency. Journal pf Personality epg Social Psychology, 46(6):1254-1266. Lowery, D. and W. E. Lyons (1989). The impact of jurisdictional boundaries: an individual-level test of the tiebout model. Journal p; Politics, 51(1):?3-97. Lyons, W. E. and D. Lowery (1986). The organization of political space and citizen responses to dissatisfaction in urban communities: an integrative model. Tpe Journal p; Politics, 48:321-346. Mastrofski, S. (1981). Surveying clients to assess police performance. Evaluation Review, 5:397-408. McKelvey, R. D. and W. Zavoina (1975) A statistical model for the analysis of ordinal level dependent variables. Journal p; Mathematical Sociolo , 103-120. Neustadt, R. (1980). Presidential Power. New York: Wiley. Olson, M. (1971). The Logic pf Collective Action: Public Goods and the Theory pf Groups. Cambridge, Mass: Harvard University Press. Ostrom, E. (1973). On the Meaning and Measurement of Output and Efficiency in the Provision of Urban Police Services. Journal pf Criminal Justice, 1:93-112. Ostrom, E., R. B. Parks, G. P. Whitaker, and S. L. Percy (1979) "The Public Service Production Process: A Framework for Analyzing Police Services", in Ralph Baker and Fred A. Meyer, Jr. (eds.) Evaluatipg Aiternative Law- Enforcement Policies (Lexington: D.C. Heath, 1979), pp. 65-73. Palmer, D. and Carlson (1984). Problems with regression 13:64-81. Parks, R. B. (1984). Linking objective and subjective measures of performance. Public Administration Review, 44:118-127. -------------- (1982). Citizen surveys for police performance assessments: some issues in their use. The Urban Interest, Spring:17-26. 262 -------------- (1981). "Police response to victimization effects on citizen attitudes and perceptions". In W. G. Skogan (ed.) Sample Surveys pr rpe Victims pr Crime, (Cambridge, Ma.:Ballinger), pp. 89-104. Parks, R. B., P.Baker, L. Kiser, R. Oakerson, E. Ostrom, V. Ostrom, S. L. Percy, M. Vandivort, G. Whitaker, and R. Wilson (1981). Consumers of public services: Some economic and institutional considerations. Policy Studies Journal, 9(7):1001-1011. Percy, S. L. (1986). In defense of citizen evaluations as performance measures. Urban Affairs Quarrerly, 22:66-83. ------------ (1981). Citigen Coproduction pr Community Safety and Security. Unpublished dissertation Indiana University. ------------ (1930). Response time and citizen evaluation of 8:75-86. ------------ (1978). Conceptualizing and measuring citizen co-production of community safety. Policy Stugies Journal, 7:486-492. Petty, R. and J. T. Cacioppo (1981). Aptitudes and Persuasion: Classic and Contemporary Approaches, Dubuque, IA: William Brown Co. Poister, T. H. and J. McDavid (1978). Victims’ evaluations of police performance. Journal pr Criminal Justice, 6:133- 149. Reis, A. (1967). srugies ip Crime apd Leg Enforcement ip Major Metropolirep Areae, Voi. 1y §egripp 2. Washington, D.C.: Government Printing Office. Rich, R. (1981). The roles of neighborhoods: Organization in urban service delivery. Urban Affairs Papers, 1:81-93. -------- (1980). A political-economy approach to the study of neighborhood organizations. American Jpprpei pf Political Science, 24(4):559-592. Rosenbaum, D. P. (1987). The theory and research behind neighborhood watch: Is it a sound fear and crime reduction strategy. Crime and Deiipgpepcy, 33(1):103-134. Sample, J. and R. Warland (1973). Attitude and prediction of behavior. Social Forcee 51: 292-304. Scaglion, R. and R. G. Condon (1980). Determinants of attitudes toward city police. Criminology, 17:485-494 263 Schneider, A. L. and J. Eagle (1975). The Effectiveness pr Citizen Participation in Crime Prevention: A Random Outlaw Model. (Eugene. Oregon: Oregon Research Institute). Schuman, H. and M. P. Johnson (1976). "Attitudes and behavior." In A. Inkeles, J. Coleman and N. Smelser (eds.), Annual Review pr Sociolo , Vol. 2. Palo Alto, CA pp. 161-207. Schuman, H. and S. Presser (1981). Questions and Answers ip Attitude Surveys. New York: Academic Press. Sharp, E. (1984). Exit, voice, and loyalty in the context of local government problems. Western Poiitical Quarterly, 37(1):67-83. ‘ Sherman, S. and R. Fazio (1983). Parallels between attitudes and traits as predictors of behavior. Journal pr Personality, 51(3):309-345. Skogan, W. and G. E. Antunes (1979). Information, apprehension and deterrence: exploring the limits of police productivity. Journal pf Criminal Justice, 7:217-241. Skogan, W. G. (1978). Citizen satisfaction with police services: Individual and contextual effects. Police Studies Journal, (special issue):469-479. ------------- (1991). The impact of routine encounters with the police. Paper presented at the Annual Meeting of the American Society of Criminology. Skolnick, J. H. and D. H. Bayley (1988)."Theme and Variation in Community Policing," in Michael Tonry and Norval Morris (eds.), Crime epg Justice, Volume 10 (Chicago: University of Chicago Press, 1988), pp. 1-37. Smith, P. E. and R. O. Hawkins (1973). Victimization, types of citizen-police contacts and attitudes toward the police. Law and Sociery Peview, 8:135-152. Stipak, B. (1979). Citizen satisfaction with urban services: Potential misuse as a performance indicator. Public Admipistretion Review, 39:46-52. Tajfel, H. (1982). Social psychology of intergroup relations. Annual Review pr Psychology, 33: 1-39. Thomas, C. W. and J. Hyman (1977). Perceptions of crime, fear of victimization and public perceptions of police performance. Journal pr Poiice Science epg Admipistration, 5(3):305-317. 264 Walker, S. (1992). The Police in America: An Introduction. New York: McGraw Hill. Warner, L. G. and M. L. DeFleur (1969). Attitude as an interactional concept: social constraint and social distance as intervening variables between attitudes and action. American Sociological Review, 34:153-169. Washnis, G. J. (1976). Citizen Involvement ip Crime Prevention. Lexington, Ma.: Lexington Books. Weigel, R.H. and L.S. Newman (1976). Increasing attitude behavior correspondence by broadening the scope of the behavioral measure. Journal pr Personaiity epg Social P§ycholoqv, 33:793-802. Wellford, C. (1973). Age composition and the increase of recorded crime. Criminology, 11:61-70. Weisburd, D and J. E. McElroy (1988). "Enacting the CPO Role: Findings From the New York City Pilot Program in Community Policing," in Jack Greene and Stephen Mastrofski (eds.), Community Policing: Rhetoric pr Reality, (Westport: Praeger), pp.89-101. Whitaker, G. P. (1980). Citizen participation in service delivery. Public Administration Review, (May/June):240- 246. White, M. and B. A. Menke (1978). A critical analysis of surveys on public opinions toward police agencies. Journal pr Police Science epg Administration, 6:204-218 Wilson, J. Q. and G. L. Kelling (1982). The police and neighborhood safety: broken windows. Atlantic Monthly, March:29-38. Wilson, R. K. (1981). Citizen coproduction as a mode of participation. Urban Affairs Papers, 1:37-49. Winship, C. and R. D. Mare (1984) Regression models with ordinal variables. American Sociological Review, 49:512- 525. Wycoff, M. A. (1988). "The Benefits of Community Policing: Evidence and Conjecture," in Jack R. Greene and Stephen Mastrofski (eds.), Community Policing: Rhetoric pr Reality (Westport: Praeger), pp. 103-120 Zamble, E. and P. Annesley (1987), Some determinants of public attitudes toward the police. Journal pr Police Science and Administratio , 15(4):285-290. 265 Zanna, M. P., J. W. Olson, and R. H. Fazio (1980). Attitude-behavior consistency: an individual difference perspective. Journal pr Personality epg Social Psychology, 38:432-440. 266