\ ' ‘HHAq\" i! i) Michiga _3t‘ate Universrty This is to certify that the thesis entitled THE IMPACT OF OFFICER CHARACTERISTICS ON POLICE RESPONSIVENESS IN SERVICE ENCOUNTERS presented by MICHAEL T. ROSSLER has been accepted towards fulfillment of the requirements for the MS. degree in Criminal Justice CA/j (/V Major Professor’s Signature {/17 /)o Date MSU is an Affirmative Action/Equal Opportunity Employer PLACE IN RETURN BOX to remove this checkout from your record. TO AVOID FINES return on or before date due. MAY BE RECALLED with earlier due date if requested. DATE DUE DATE DUE DATE DUE 5108 KzlProilAccatPrelelRCIDateDue.indd THE IMPACT OF OFFICER CHARACTERISTICS ON POLICE RESPONSIVENESS IN SERVICE ENCOUNTERS By Michael T. Rossler A THESIS Submitted to Michigan State University in partial fulfillment of the requirements for the degree of MASTER OF SCIENCE Criminal Justice 2010 ABSTRACT THE IMPACT OF OFFICER CHARACTERISTICS ON POLICE RESPONSIVENESS IN SERVICE ENCOUNTERS By Michael T. Rossler This thesis examines the impact of officer characteristics (i.e., officer gender, education, race, experience, and assignment) on police responsiveness in service encounters (e.g., compliance with requests for officers to file a report, act on behalf of the citizen with an agency, provide physical assistance, or provide information). Data are drawn from observational research conducted in Indianapolis, Indiana, in 1996, and St. Petersburg, Florida, in 1997. Previous literature on responsiveness has not included service encounters, and previous research on service has not included responsiveness to requests. The results of this analysis indicate that officers are quite responsive, fulfilling 63.3 percent of all service requests. Further, multiple factors were modeled to determine their impact on officer responsiveness (i.e., citizen characteristics and situational attenuators). Multivariate analyses indicate that nonwhite officers and officers in a community policing assignment are both less likely to fulfill a citizen’s request than are white officers or officers in a traditional patrol assignment. Additionally, the most influential factor was citizen disrespect. Potential implications of this research, along with directions for future research and policy implications are discussed. ACKNOWLEDGEMENTS I owe a great debt of gratitude to a number of people who made this possible. First, I would like to thank the members of my thesis committee (Dr. Timothy Bynum, Dr. Christina DeJong, and Dr. William Terrill) for their direction in this research endeavor. In particular, I would like to thank Dr. Terrill, not only for providing a wonderful policing foundation through coursework, but also for his patience through countless revisions, and always taking the time to explain relevant details and answer a host of questions. I would also like to thank Jason Rydberg for his assistance in the early stages of my thesis, and for offering useful advice on how to proceed with various obstacles. In addition, I would also like to thank my family for their continued support in not only through my academic career, but in all aspects of life. iii Table of Contents LIST OF TABLES ................................................................................... iv CHAPTER I: INTRODUCTION .................................................................. 1 CHAPTER II: LITERATURE REVIEW .......................................................... 3 Officer Performance and Accountability ................................................. 4 Legitimacy .................................................................................... 8 Officer Behavior in Service Encounters ................................................ 10 Officer Characteristics and Satisfaction in Service Encounter ....................... 13 Officer Characteristics and Responsiveness ............................................ 15 Officer Characteristics and Service ...................................................... 16 Limitations .................................................................................. 17 Proposed Thesis Research Questions ..................................................... 18 CHAPTER III: METHODS ....................................................................... 20 Study Sites .................................................................................. 22 Data .......................................................................................... 22 Observational Data ................................................................... . ...... 20 Interview Data .............................................................................. 24 Variable Descriptions ...................................................................... 24 Data Analysis ............................................................................... 28 Validity, Reliability, and Generalizability .............................................. 28 CHAPTER IV: RESULTS ......................................................................... 31 Descriptive Statistics ....................................................................... 33 Bivariate Analysis .......................................................................... 35 Multivariate Analysis ...................................................................... 39 CHAPTER V: SECONDARY ANALYSIS .................................................... 48 Descriptive Statistics ....................................................................... 48 Bivariate Analysis .......................................................................... 49 Multivariate Analysis ...................................................................... 51 CHAPTER VI: DISCUSSION ................................................................... 54 Overview of Findings ..................................................................... 54 Further Questions for Consideration .................................................... 56 Limitations of the Present Inquiry ....................................................... 58 Directions for Future Research ........................................................... 59 Policy Implications ........................................................................ 61 APPENDICES ....................................................................................... 62 REFERENCES ...................................................................................... 63 iv List of Tables Table 4.]: Citizen Role ............................................................................. 32 Table 4.2: Citizens’ Requests for Service and the Police Response ......................... 33 Table 4.3: Descriptive Statistics for Officer Characteristics ................................. 34 Table 4.4: Descriptive Statistics for Control Variables ......................................... 35 Table 4.5: Bivariate Distribution of Officer Responsiveness by Officer Gender .......... 36 Table 4.6: Bivariate Distribution of Officer Responsiveness by Officer Education ....... 37 Table 4.7: Bivariate Distribution of Officer Responsiveness by Officer Race ............. 37 Table 4.8: Bivariate Distribution of Officer Responsiveness by Officer Assignment. . ...38 Table 4.9: Effects of Independent and Control Variables on Police Compliance with Citizen Requests for Service ....................................................................... 44 Table 4.10 Effects of Independent and Control Variables on Police Compliance with Citizen Requests for Service ....................................................................... 43 Table 5.1 Distribution of Explanation Offered by Officers ................................... 46 Table 5.2 Bivariate Distribution of Refusal Explanation by Officer Gender ............... 48 Table 5.3 Bivariate Distribution of Refusal Explanation by Officer Education. . . . . . ...49 Table 5.4 Bivariate Distribution of Refusal Explanation by Officer Race .................. 50 Table 5.5 Bivariate Distribution of Refusal Explanation by Officer Assignment .......... 50 Table 5.6 Logistic Regression Officer Characteristics, Controls by Explanation .......... 52 Chapter I : Introduction Policing encompasses a wide variety of tasks, which makes quantifying officer performance a difficult task (Fyfe, 1993). Traditional measures of police work have included arrest (Smith and Visher, 1981; Greenberg, et al., 1979), response time (Spehnan and Brown, 1981), and clearance rates (Chaiken et al., 1996). While these measures have useful attributes, they are inadequate in addressing the majority of time and services that police officers contribute during the course of their duties (Skolnick and Fyfe, 1993). Good policing is broader in scope than traditional performance measures indicate, and it is important to consider an officer’s response to needs expressed by citizens during the course of their interactions (Fy‘fe, 1993). An often overlooked alternative measure of police performance is responsive service. Officer behavior during a police-citizen encounter can significantly influence public satisfaction with the police (Reisig and Parks, 2002). Citizens may actually be more pleased by the effort of an officer to resolve an issue than with the desired outcome (Mastrofski, 1999). Guyot (1991) finds that officers are often judged more by their actions than actual task accomplishment, particularly in cases involving personal harm. Mastrofski (1999) tends to agree that police activity is often judged by the public not through traditional measures of effectiveness, but through the officer’s ability to demonstrate responsiveness to the citizen’s concern. In addition, responsive service builds connections between the police and the community and increases the legitimacy of the police when implemented (Mastrofski et al., 1996; Mastrofski, 1999). Examination of responsive police action has been forwarded by Mastrofski et al. (2000) in their work on officer compliance with requests to control citizens. In particular, they found that various officer characteristics such as gender, experience, and assignment influenced the extent to which officers are willing to comply with citizen requests. The findings indicate that officer based differences do exist, but are complicated by other factors (such as an officer’s willingness to control a citizen), rather than simply the officer’s willingness to comply with citizen requests. This thesis will seek to further study officer based determinants of responsive service. Using data from an observational study conducted in two medium-sized cities, this inquiry will examine the impact of officer characteristics on providing service (i.e., filing a report, acting on the citizen’s behalf with a government/official agency or private organization, providing physical assistance, or providing information on how to deal with a problem) through officer compliance with citizen requests. Analysis of the impact that officer characteristics have on these “softer” measures of police responsiveness has not been made available in the current body of literature. This document will begin with a review of the literature to highlight the complexities of police accountability measures, police legitimacy, and officer behavior in service related encounters. The literature review will provide an empirical overview of the impact of officer characteristics (i.e., education, experience, gender, race, community policing proclivity) on service related behaviors. Additionally, the thesis will highlight a conceptual framework, proposed research questions, and a description of the methodology. Following this, descriptive statistics, bivariate distributions, and multivariate analyses will be presented for the dependent variable, along with a subsequent chapter examining the effect of the independent and control variables on willingness to provide an explanation for refusing a request. Chapter II: Literature Review Historically, the police mandate has undergone extensive changes moving from the political era, to the reform era, to the community era. Throughout these eras the role of a police officer has been defined by the decisions made by policy makers and police chiefs. The desired outcomes of police activities have changed concurrent with the mandates (Kelling and Moore, 1988) creating a history of various performance measures. Measures of police performance under the professional model are generally identified in I the literature as reported crime rates, overall arrests, clearance rates, and response times (Alpert and Moore, 1993). These four measures have become ingrained in policing, politics, and the media as the standards by which to evaluate effective policing on an individual and organizational level. Ostensibly, research has continued to evaluate what actions or outcomes represent quality policing throughout political changes and new mandates. Some scholars believe that professional model measures fail to capture the true essence of policing and are not a good representation of the ways in which officers use their time (Fyfe, 1993; Parks et al., 1999). With respect to the most recent era, community policing research has focused on crime prevention and control, citizen satisfaction, and overall quality of life (Reisig and Parks, 2002). These measures of police performance are necessary, but they simply reflect the outcomes of a community policing intervention primafacie. Use of coercive force to solve human issues has been the presiding framework for understanding the police function (Bittner, 1970). Influential law enforcement leaders have attempted to place police solely in a crime fighting role, while reformers have sought to define departmental successes as responding to the needs of valued community members. Quantifying the quality and effectiveness of police work has been plagued by the unclear role of officers across communities with differing demographics that require different services and responses. In order to effectively evaluate the quality of police services, some researchers have offered rather unrealistic policies indicating that the police should focus solely on fighting crime while leaving service and order maintenance functions to other organizations (Manning, 1971). A more realistic approach is offered by Mastrofski (1988), whereby officers could be evaluated on a number of less traditional criteria such as the ability to use verbal coercion, protect citizen rights, and impart a sense of fairness into their low-visibility informal decisions. Further along this spectrum are researchers (Guyot, 1991; Muir, 1977) who view officer empathy as a defining trait in determining effective police work, and believe that human interaction compose a larger portion of officer performance. Officer Performance and Accountabilig Quantifying police productivity is a difficult task that originated in the early 18008 with the advent of municipal police departments (Monkkonen, 1981). These early measures reflected the general duties of police work at this incipient stage. Early police measurements sometimes included the number of lost children or drunken citizens returned home (Monkkonen, 1981). Police would serve various functions in this time period including shooting stray dogs, completing boiler inspections, and even participating in census collection (Monkkonen, 1981). Eventually, police departments began to remove themselves from these duties, and measurements of officer performance began shifting to accommodate the change. By the late 18005 and early 19003 there was a grong awareness concerning police corruption and inefficiency (Maguire and Uchida, 2000). As a result, organizations such as the Bureau of Municipal Statistics in New York began collection and analysis of police data to investigate and combat corruption. While the data and administrative suggestions made by the Bureau had little effect on reforming the department, their actions did establish a practice of collecting accountability measures such as arrests, clearance rates, response times, and even crime rates. Further reform efforts that sought to professionalize policing cemented the common measures of crime currently in existence. Systematic data collection of crime rates began with the advent of the Uniform Crime Reports (UCR) in the 19303, which served the purposes of providing information on crime to improve management and operational strategies and to provide empirical evidence that could be used to refute public belief of crime waves. Despite criticisms fi'om scholars that the crime rate is an extremely flawed measurement of police performance (Beattie, 1960; Decker, 1977; Kitsuse and Cicourel, 1963; Maltz, 1977; Robison, 1966; Seidman and Couzens, 1974; Walker, 1992; Wolfgang, 1963), politicians still use these figures to reward or attack police organizations (Maguire and Uchida, 2000). In addition to criticisms about the reliability of measures such as the crime rate, other traditional measures such as arrest and response time are limited by the scope of the police mandate that they actually encompass (Maguire and Uchida, 2000). Professional measures focus primarily on the law enforcement function of policing, and do not provide a holistic picture of what makes an effective police officer (Klockars, 1999). Movement away from crime and arrest related data that can be controlled by departmental directives is an important first step in accurately measuring police performance (Klockars, 1999). Measurement of police activities outside of arrest, response time, and clearance rates has become even more important in the age of community policing (Maguire and Uchida, 2000). Response to lacking information on community policing and the tasks officers engage in through community policing activity has become an important topic in both police accountability and community policing circles. Maguire and Uchida (2000) suggest that the lack of a readily definable “bottom line” in policing coupled with an obsession to intensively measure the crime related features of arrest and crime rate ignore the essential officer-citizen interaction in encounters. In support of this argument they cite Mastrofski’s (1999) six elements of “policing for people” (i.e., attentiveness, reliability, responsive service, competence, manners, and fairness) as activities that should be measured to compare police organizations over place and time. These six elements reflect the nature of citizen expectations beyond law enforcement and order maintenance functions. Essentially citizens expect the police to arrive in a timely fashion, know the proper way to address a problem, provide service, be polite, and act in a fair manner. While some of these elements are quite hard to quantify (e.g., fairness) others can be more easily identified and quantified through observable police actions (i.e., responsiveness to citizen requests). Using inadequate measures of police accountability may have more problematic consequences than not being able to identify the most skilled officers or effective police organizations. Macro-level concerns about using crime rates as a yardstick for police performance have been highlighted by Manning (1971), who believes the practice of evaluating police departments based on the local crime rate can lead to reduced satisfaction among the populace. Essentially, because police activity has little impact on the crime rate, claiming that police activity will substantially lower the crime rate may provide a situation where citizens will never be satisfied with the ability of police departments to achieve the mandate they have established. The inability of police organizations to control dynamic outcomes such as the crime rate can influence the perceptions of citizens. To illustrate, Reisig and Chandek (2001) tested the effect that expectancy disconfirmation, a differing level of performance in practice than the level anticipated by the citizen, has on a citizen’s specific and global satisfaction with the police. Researchers surveyed citizens who had a recent encounter with police in either a breaking and entering call for service (voluntary) or a traffic stop (involuntary). It was established that disparity between expectations and actual performance was a moderate to strong predictor of citizen satisfaction with the specific encounter. While disconfirmation was not found to be a significant predictor of global satisfaction with the police, officer behavior in these encounters did serve as a predictor of global satisfaction. These findings provide a context for the proposed inquiry, as Reisig and Chandek (2001) indicate that officers and the citizenry must explore the common expectations that they have for officer behavior during an encounter. Using police-citizen encounters as a framework for evaluating police performance is viewed as a good method of tapping into alternative measures of accountability (Maguire and Uchida, 2000; Mastrofski, 1981). Responsiveness, for example, in the form of complying with citizen requests, is a distinct measure of quality in police behavior during encounters. Once alternative dimensions of policing like responsiveness can be accurately measured, police organizations may have a device to present police effort and effectiveness to the general public (Alpert and Moore, 1993; Ottemeier and Wycoff, 1994). The general progression from this expression of responsiveness would be that the public could have a greater understanding of the overall services that the police provide, which may improve the legitimacy of police actions. Legitimacy Confidence in the police is an essential element of officers being able to complete their tasks effectively. Police rely on the populace to alert them to violations of the criminal law, assist in investigations by providing information, and to serve as witnesses in court. A study by F urstenburg and Wellford (1972) used telephone surveys of all citizens who called the police over a one month period to inquire about the performance of officers responding to calls for service. Citizens indicated a much higher satisfaction rate with the police when officers took the time to explain their activities and the actions they would take in handling a citizen’s complaint. When the police followed up on a complaint to explain progress, this activity also improved citizen satisfaction. Those satisfied with calls for service indicated they would call the police again in a similar situation. On the whole, Furstenberg and Wellford (1972) indicate that police behaviors during an encounter can increase public confidence in the police. Effort on the part of police has also been linked to citizen satisfaction that citizens feel with those services. Relying on a telephone survey of Chicago residents, Skogan (2005) distinguished between citizen-initiated and officer-initiated encounters to examine several predictors of citizen satisfaction. Officer helpfulness was determined to be the most important predictor of satisfaction with citizen initiated encounters, followed by politeness, attentiveness, and response time. Satisfaction with officer initiated encounters was best predicted by officer fairness and politeness. As such, the relationship between officer behavior and citizen satisfaction can have serious implications for the legitimacy of the police. The evidence presented thus far makes a compelling case for the impact of officer behavior on citizen satisfaction with police services, but this does not directly translate to improved officer behavior enhancing the legitimacy of the police. Fortunately, this connection has received a large amount of empirical support (Lind and Tyler, 1988; Reisig and Lloyd, 2009; Tyler, 1990; Tyler, 1997; Tyler, 2005). A unique opportunity to test the relationship between police actions and legitimacy in pre and post 9/11 New York City was conducted by Sunshine and Tyler (2003) through the use of mail (pre) and telephone (post) surveys. Given that police community relations were less emphasized in pre 9/11 New York than post 9/11, an opportunity to examine a potential impact was presented. The research determined that perceived legitimacy of the police was related to citizen cooperation and compliance with the police. Citizens who perceived the police as legitimate were also more willing to empower the police by giving them more tools to complete their mandates. Police legitimacy in this study was most strongly related to the perception of officer fairness, although the actual activities in which police engaged were not specified. An interesting note by the authors is that process policing was not directly related to cooperation, compliance, or empowerment. Essentially, the actions taken by the public to help the police (report crime, offer information to the police) flows through legitimacy (sense of obligation to obey the police). Police behavior, as construed by the authors, is an antecedent to police legitimacy. This indicates a link between the way police officers act and the public’s willingness to comply with requests. Testing of the impact that police actions have on the perceived legitimacy of the police has also occurred more recently abroad. Reisig and Lloyd (2009) examined the relationship between police legitimacy and citizen cooperation in J arnaica, a country where the Jamaican Constabulary Force (J CF) is experiencing low levels of perceived legitimacy. Contributing factors are a high rate of violent crime within Jamaica, and a perception the J CF is impolite and disrespectful to citizens. In fact, some citizens have turned to local community leaders (i.e., Dons) for protection. Amid this crisis, the J CF has an extremely difficult time obtaining information or cooperation from citizens. Surveying the local high schools, Reisig and Lloyd (2009) found that procedural considerations (i.e., treating people with respect, taking the time to listen, taking peoples’ needs into consideration) predicted police legitimacy (accepting the decisions of the police) and citizen cooperation (i.e., reporting crimes, reporting suspicious activity, and helping the police find a suspect). Recent research on procedural justice has indicated that not only is the process by which the police handle citizen encounters important from a human rights philosophical standpoint, but it has tangible effects that benefit police agencies willing to put forth the effort (Reisig and Lloyd, 2009). From this standpoint, it is important to move beyond indications that “softer” officer behaviors are important, and determine which officers actually engage in important process-oriented policing. Part of this movement should be determining which officers are willing to comply with citizen requests. 10 Officer Behavior in Service Encounters Difficulty often arises when scholars attempt to determine which police actions qualify as service. Wilson states that service results from “client” requests that occur upon a call for service to which the police respond freely to the preferences of individual citizens (Wilson, 1967: 5). Because police services generally result in no perceived direct cost to individual taxpayers, they may be inclined to request services frequently. As such, it becomes the duty of individual officers to ration off services by deciding which requests to fulfill and which to deny. Officers are thereby the responsible gatekeepers of fulfilling needs expressed by citizens, and the reasons citizen requests are fulfilled or denied remains scarcely examined. Responsiveness, or giving clients what they want, has become the modern definition for the rationing of these services. Mastrofski (1999) believes that officers should fulfill citizen requests when possible, and provide a good-faith effort or explanation when these tasks cannot be completed. Policing should go beyond the rules outlined by bureaucracy. Even if the police do not have the ability to accomplish a law enforcement or order maintenance task, they should still be available to attend to citizens’ needs and “pick up the pieces” following a traumatic event (Mastrofski, 1999; 2) Alpert and colleagues (2001) agree that police performance should be addressed on this measure, featuring officer responsiveness as a measure of quality rather than productivity. In fact, quality is described as “conformance to customer needs, a fundamental component of community policing wherein the customer or consumer is the community” (Alpert et al., 2001: 81). An expression of need by citizens can be conceptualized by a direct request of an officer for some form of service. 11 Extensive examinations of the impact officer characteristics have on officer behavior are present throughout the literature (Sherman, 1980). Research focusing on experience (Bayley and Garofalo, 1989; Mastrofski et al., 2000), education (Finckenauer, 1975; Paoline and Terrill, 2007), race (Terrill and Mastrofski, 2002; Brown and Frank, 2006), and gender (Paoline and Terrill, 2004; Sherman, L.J., 1975) have given some insight into the impacts of officer characteristics on traditionally measured behaviors. However, with respect to service related behavior, Riksheim and Chermak (1993) found that only 17 studies prior to 1980 focused on the relationship between individual level variables and service behaviors. Gender, height, and attitudes were found to have no relationship to service related behaviors, while race and length of service provided mixed findings about the effect that these variables have on officer behaviors. Research between the Sherman (1980) and Riksheim and Chermak (1993) reviews found that only five studies have focused on individual level characteristics and officer behavior. The five studies focused solely on attitudes, and indicated that no relationships exist between attitude and officer behavior. Non-traditional evaluations of policing have been scantily covered following Mastrofski’s (1988) implication that a more holistic assessment of police work should commence. In one such study, using officer determinations of co-workers, Bayley and Garofalo (1989) studied the relationship between colleagues’ perception of skill and the ability of officers to minimize violent citizen encounters. In terms of individual level characteristics, skilled officers were identified as being more experienced (having on average 18 months more experience than less skilled officers). The objective of the Bayley and Garofalo (1989) study was not directed at service related behaviors, but 12 useful tangential findings were provided. In addition to the primary research question, officer sympathy, fi'iendly advice, and advice about legal remedies were also considered. Officers who were determined to be exceptionally skilled by their co-workers were significantly more likely to engage in these three actions than were officers considered to be average. Skilled officers were also more likely to provide a conclusion during an exit rather than absolving themselves by claiming “there was nothing the police could do” (Bayley and Garofalo, 1989; 17). It is important that police express effort in citizen encounters (Skogan, 2005), which is demonstrated more frequently by experienced oflicers (Bayley and Garofalo, 1989). Oflicer Characteristics and Satisfaction in Service Encounters The paucity of available data on “policing for people” (Mastrofski, 1999) provides for a situation where proxy measures must be used in order to examine differences in officer characteristics, particularly in pre-community policing era research. While Sherman (1980) indicated that only age and education provided directional service in terms of officer service behavior, using citizen satisfaction with officer responses can illustrate tenuous relationships for the purpose of hypothetical development. An early example of this involves retrospective examination of police-citizen encounters. Using telephone surveys, Carlson and Sutton (1979) inquired about citizens’ recent experience with an officer from that department (citizens were identified using police logs). Citizens were most satisfied when officers demonstrated rapport, involvement, and success. Further analysis indicated that officer education was positively related to citizen satisfaction, while oflicer experience was not related to citizen satisfaction. Findings indicate that concentrating on the principles of “policing for people” (Mastrofski, 1999) 13 can improve citizen satisfaction with police services, and that some officer characteristics (education) may influence these police behaviors. Officer conversational style, which was examined in response to more calls for service being handled directly by telephone, has been supported as a predictor of citizen satisfaction in police citizen encounters. Using a newly implemented Telephone Response Unit (TRU), Glauser and Tullar (1985) randomly sampled five incoming phone calls per day, providing 60 tapes over a two week study period. People contacting the agency, a large metropolitan police department in the Southeastern United States, through TRU whose calls were selected were called back a short time later, and an appointment was scheduled for the researcher to administer the questionnaire. Citizens were found to be more satisfied with officers who provided more extensive information and gave specific directions on procedural matters. Less satisfaction was found in encounters where the officers questioned citizens or acted in a domineering fashion. Dorothy Guyot (1991) has long supported police strategies that serve citizens as customers, and non-traditional measures of police work that exhibit officer effort and attentiveness toward the concerns of citizens. A crude measure of officer attentiveness is the amount of time officers spend at a particular call. Guyot (1991: 58) relates this to faulty management directives that “pull officers away from work so that they can resume looking for and waiting for work.” Termed “nuisance calls” by police organizations (Alpert and Moore, 1993; 121) officers have been shown to reduce time spent on calls for assistance that were not considered criminal or representing an emergency situation at the direction of supervision (Sparrow et al., 1990). Certainly, crime related or emergency calls should take priority over calls for non-emergency assistance (Guyot, 1991); 14 however, officers should not forgo an opportimity to interact with and assist a citizen to go in-service and carry out a simple patrol function (Guyot, 1991). The impact of this directive on citizen satisfaction has been reinforced empirically by Coupe and Griffiths (1999) in their study of satisfaction with police response to a burglary and the amount of time officers spent on scene. Satisfied citizens had an officer on scene for an average of 40 minutes while dissatisfied citizens had an officer on scene for an average of 27 minutes. The rate of dissatisfaction grew substantially higher when the officers remained at a call for less than ten minutes. A common explanation for officers spending less than ten minutes at a call is that little or no evidence is available (Coupe and Griffiths, 1999). Citizens were also more satisfied when they received updates from officers, furthering support of effort and attentiveness as the source of citizen satisfaction with their encounter. While Coupe and Griffiths (1999) provide a detailed description of high quality behaviors, they do not provide information about the types of officers most likely to engage in these behaviors. Further development of service quality and officer characteristics may be inferred from Brown’s (2007) examination of differing service firnctions and the willingness of the citizen in the interaction. Brown (2007) used a municipal services survey to examine citizen satisfaction with voluntary (use of city parks), client (calls regarding refuse disposal), and captive (traffic stop) interactions with city services. Generally, citizens were more satisfied when they had a positive experience, and when the interaction was more voluntary. This adds to the argument that proactive stops may result in lower satisfaction. 15 Oflicer Characteristics and Responsiveness Perhaps the most useful study of responsive service and individual level characteristics is Mastrofski et al.’s (2000) examination of officer compliance with citizen requests to control other citizens. Using data from the Project on Policing Neighborhoods (POPN) the researchers examined the value of a wide variety of officer and situational characteristics. Situational characteristics (legal considerations, requester intoxicated, etc.) were the most useful in predicting the likelihood that when a citizen requested that the officer advise or persuade, warn or threaten, force someone to leave the scene, or arrest another citizen that the officer would take the requested action. In terms of individual level variables, Mastrofski and colleagues (2000) found that female officers were much less likely to comply with the citizen requests to control other citizens at the scene. The authors attributed this phenomenon to the female officers’ reluctance to use coercion rather than a reluctance to fulfill requests for service. More experienced officers were also less inclined to fulfill the requests for control of another citizen. This was attributed to the wisdom or cynicism of older officers and/or a possible unwillingness to engage in time consuming work. An officer’s proclivity toward community policing had an effect in the expected direction, as these officers were more inclined to fulfill the requests of the citizens. Developing the actual responsiveness of officers and impact of individual level characteristics in a service capacity is limited by the focus on responding to demand for control. While controlling citizens is an integral portion of police work (Bittner, 1970), differences in the rate at which officers use coercion (Paoline and Terrill, 2007) may affect the rates at which they comply with requests that require police action of a coercive nature. 16 Oflicer Characteristics and Service A theme in the current research following Sherman’s (1980) examination is a decrease in research directed at service related officer behavior (Riksheim and Chermak, 1993). However, in a recent study, Sun et al., (2008) examined officer differences with respect to coercive (arrest, interrogation, search, threaten to use force, restraint, use of handcuff, pain compliance, incapacitation methods, and draw/discharge weapon) and non-coercive behaviors (providing physical assistance and information on officers’ own initiative, providing physical assistance and information requested by citizens, filing an incident report, acting on citizens' behalf, advising citizens to sign complaint, to use legal process, to seek help of other agencies, to seek help of family or friends, to help another), and found significant situational, individual, and neighborhood characteristics using data from the Project on Policing Neighborhoods. In terms of officer characteristics, only the officer unit had a significant impact while all other variables were controlled. Surprisingly, community policing officers were less likely to engage in non-coercive activities than were their patrol counterparts. While this finding provides little support for officer level characteristics in non-coercive behavior, the scale used by Sun et al. (2008) does not distinguish between officer initiated behaviors or officer responsiveness to citizen requests, the true test of Mastrofski’s (1999) element of responsiveness as a characteristic of quality policing. While measuring police productivity in non-coercive encounters remains important, focusing on direct requests is a better representation of the quality of police responses in cases where the citizen has directly expressed a need. 17 Limitations While Mastrofski (1981) generally supports survey research as a method by which to evaluate officer performance during police citizen encounters, social observation data provides a unique opportunity to use rich data in evaluating officer performance in response to citizen requests for service. Social observation data provides a less biased perspective and catalogues the events as they occurred as opposed to relying on an involved party’s perception. The POPN source currently available allows for detailed examination of any differences in officer responsiveness to citizen requests. In addition, prior research to date is limited in two primary ways. First, while recent research has made strides toward examining differences in officer proclivity toward service behavior (i.e., Sun et al., 2008), the scales used in this research do not directly examine the officer’s willingness to comply with citizen requests. More specifically, the dependent variables in the Sun et al. study included service (non- coercive) activities that officers engaged in through their own initiative (i.e., proactive service). Additionally Sun and colleagues’ (2008) thirteen item scale included whether officer’s offered comfort in the encounter. While this scale is an adequate measure of officer non-coercive productivity, greater refinement is necessary to accurately examine officer responsiveness (i.e., restricting cases to those where a citizen makes a specific request). Second, responsiveness can be accomplished by officers in situations where they cannot or will not comply with a request if they provide a reason for the refusal. Mastrofski (1999: 2) indicates that officers “can be responsive even when they deny a 18 citizen’s request by explaining the denial.” Thus, measuring and assessing responsiveness is broader than just simply complying or not complying with requests. Proposed Thesis Research Questions Given the limitations identified in the prior literature on police responsiveness, the proposed study involves three primary research questions: 1. What is the frequency with which citizens request service during encounters with the police? 2. How often do the police fulfill citizen requests for service? 3. What characteristics predict officer responsiveness to citizen requests for service? -l.‘.‘hl [IA '5 - . 5” .'( After examining these three primary questions, a secondary analysis will be conducted based on those cases where officers do not comply with citizen requests by asking a fourth research question: 4. What characteristics predict officer willingness to provide an explanation for refirsing to comply with a citizen request for service? 19 Chapter III: Methods Examination of the proposed inquiry will be provided by the Project on Policing Neighborhoods (POPN). The general impetus for POPN data collection was to allow for researchers to describe policing operations on a daily basis. Two mid-sized cities, Indianapolis, Indiana (Summer 1996) and St. Petersburg, Florida (Summer 1997) served as the study cites for this research (Mastrofski et al., 2000). Data collection focused on citizen encounters, and used systematic observation of officers to catalog behavior in these interactions. Officers were also interviewed to provide information about their attitudes and backgrounds. Study Sites Study sites for POPN were evaluated and selected based on several factors. Indianapolis and St. Petersburg were both willing to host research within their departments for the one year study period. Additionally, these cities have diverse demographic and socio-economic populations (Paoline and Terrill, 2007). Community policing initiatives had also been implemented in these departments, prompting an examination of differences between community policing officers and organizations. Indianapolis was the more populous city during the examination, and in turn employed more full time police officers. Considering the cities’ similar crime rates, the lower number of sworn police officers in St. Petersburg resulted in a greater work responsibility for their officers. The city of Indianapolis had higher rates of unemployment, minority population, residents in poverty areas, and female headed households (Terrill, 2001). Indianapolis and St. Petersburg also had several differences that made them compelling departments for study. In terms of the specific number of officers, 20 Indianapolis employed 416 firll time officers while St. Petersburg employed 246 during the study period. The departments’ community policing initiatives also differed in commitment and strategies. St. Petersburg committed 60 officers (23%) to community policing, while Indianapolis committed only 25 officers (6%). Community policing in St. Petersburg originated as a distinct program two years before Indianapolis began its program. Deployment of officers in St. Petersburg was on a geographic basis, in an attempt to build a sense of officer responsibility for the problems occurring in their area (Terrill, 2001). Indianapolis focused more on directed patrol in which officers attempted to proactively address low level problems and quality of life issues. Personal interaction, a more common conception of community policing, was also a primary focus of Indianapolis on top of the more aggressive strategies (Paoline, Myers, and Worden, 2000) Indianapolis and St. Petersburg also differ in measures of traditional professionalism (Shernock, 1992). In terms of the disparity between officers and the general population, Indianapolis officers were more educated than the typical citizen. Thirty-six percent of officers in Indianapolis had a four year degree, while only 26 percent of the citizens also had a four year degree (US Census Bureau, 2000). In St. Petersburg, the margin was much closer with 26 percent of officers and 23 percent of the general population having a four year degree (Terrill, 2001; US Census Bureau, 2000). On another measure of professionalism, Indianapolis officers also required more hours of training (1,392) than St. Petersburg (Terrill, 2001). While training focuses primarily on traditional measures (Traut et al., 2000), officer training may have implications for non- traditional measures as well. 21 932 This inquiry will, as previously stated, employ two separate aspects of the POPN data. Systematic observation of police officer’s in their field assignments and the personal interviews with those officers to catalog their backgrounds and attitudes. The objective field observations of patrol officers is the most important part of this proposed study, as this information is integral to understanding police citizen encounters. Observational Data Observational data on officers were collected through a technique called Systematic Social Observation (SSO) (Mastrofski et al., 1998) as observer participant (Mastrofski et al., 2009). Observer-participants do not make an effort to keep a distance from the individuals they are observing (in this case patrol officers and citizens). Additionally, researchers acting as observer-participants do not try to actively participate in the interactions between officers and citizens (Babbie, 1995). Before the observers began field work, they were required to take a one semester course (at Michigan State University or SUNY Albany) in $80. Observers also participated in ride-alongs at local departments for additional development of research skills (Terrill and Mastrofski, 2002). Physical data collection occurred on patrol through a matched sample of beats in each city. Observers, as is directed by $80, took notes on officer activities and police citizen encounters, providing specific details about the citizens. Encounters between officers and citizens were described as any “face-to-face communication that took place between officers and citizens that took over one minute, involved more than three verbal exchanges between officer and citizen, or involved significant physical contact between the officer and citizen” (Terrill, 2001: 50). Observers transcribed their notes at the end of 22 each day’s field observation, per POPN protocol. Officers were assured that the information observed and transcribed'would remain confidential, and were permitted to read the notes taken on their rides, but were not permitted to read the notes taken on rides with other officers (Parks et al., 1999). Beats were sampled for observation from each city to permit comparisons. Twelve beats were selected from each city, Indianapolis having 50 total beats and St. Petersburg having 48 total beats. An anticipated higher probability of police citizen encounters provided the rationale for POPN researchers to sample data from these more active beats, where a larger number of encounters were likely to occur. Research directors also selected the beats based on a socioeconomic index of the neighborhood conditions. Factors considered included the percent of families with a female head of household, employed adults, and the population living below 50 percent of the poverty level (Terrill, 2001). Indianapolis registered as the more distressed city with socioeconomic index scores for beats ranging between 4 and 76, with a median score of 36. St. Petersburg, the less distressed of the two cities featured an index range of 4 to 103, with a median score of 15. Using these scores, beats were selected from the 2nd, 3rd, and 4th quartiles of the Indianapolis scores, and St. Petersburg beat scores were then selected to match beats within Indianapolis (Terrill, 2001). Sampling of rides within these beats called for observation of every shift in the selected beats. Rides also included observations from the general patrol officers and community policing officers, and included rides on both high activity and lower activity days. In order to maximize the number of encounters observed, the project directors oversampled the busier shifts generally occurring Thursday through Saturday. 23 When field observations were completed, Indianapolis observers had collected data on 194 patrol officers and St. Petersburg observers had completed ride-alongs with 128 patrol officers. Indianapolis and St. Petersburg researchers also observed 48 and 37 supervising officers respectively. During 2,800 hours of observation, Indianapolis police engaged in 6,485 citizen encounters, while the St. Petersburg police were observed for 2,900 hours and were witnessed in 5,500 citizen encounters (Paoline and Terrill, 2007). Interview Data Interviews were completed by personnel who were trained and employed solely for completing officer interviews. These researchers conducted interviews with the observed officers. Officers were questioned for about 25 minutes using survey forms containing questions about officer’s background characteristics and attitudes. Considering the proposed inquiry seeks to specifically compare officer’s willingness to comply with citizen requests for service, background characteristics such as experience, education, race, gender, as well as attitudes toward community policing will be of particular interest to this study. The in-person interviews received an excellent response rate, with 95 percent in Indianapolis and 97 percent responding in St. Petersburg. Variable Descriptions Dependent Variables This analysis will examine three dependent outcome measures of officer responsiveness to citizen requests by drawing on the following four specific questions captured by the POPN trained observers: (1) Did the citizen ask police for information on how to deal with a problem? (2) Did the citizen ask police for physical assistance for self 24 or others? (3) Did the citizen ask the police to file a report? (4) Did the citizen ask police to act on the citizen’s behalf with a government official/agency, or private organization? Based on the approach applied by Mastrofski et al. (2000) these measures will be examined as a dichotomous variable (1 = officer complied, 0 = officer did not comply, officer promised to/partially complied) and as an ordinal variable (0=did not comply, l=promised to/partially complied, 2=fully complied). Placing the variables into two forms of analysis will allow for an examination of which officers are more likely to show an inclination to provide responsive service. Additionally, a third dependent variable will examine officer differences in offering explanations of officer denials of citizen requests for service. This variable will also be dichotomous (1 = officer provided an explanation for denying request, 0 = officer did not provide an explanation for denying request). As noted, this third dependent variable has been identified by Mastrofski (1999) as an integral aspect of responsiveness. Separation fiom the other two variables seems appropriate, as coverage of officer explanation of refusal will employ only cases where officer’s denied a citizen’s request. Independent Variables The primary variables for examination are officer background characteristics. Five major variables (i.e., gender, race, education, experience, and patrol assignment) can be identified as potential predictors of whether officers will comply with requests for service. First, gender differences in compliance with citizen requests have been demonstrated in terms of requests for control of another citizen (Mastrofski et al., 2000), with male officers more likely to comply. Second, race is also an important factor to consider when examining officer behavior. Mastrofski and colleagues hypothesized a 25 reduction in officer compliance based on mixed race encounters (white officer/nonwhite citizen, nonwhite officer/white citizen), although no significant difference was found. Third, education is also a compelling attribute of officers that could have an impact on their willingness to comply with citizen requests. While some believe that a college education improves police work by sensitizing officers (Finkenauer, 1975), others anticipate that a disparity between well educated officers and a poorly educated public results in a disconnect that would make more educated officers less likely to comply (Mastrofski et al., 2000). This variable can be expressed dichotomously, by examining officers with and without college degrees. Fourth, years of police experience may also impact the way that officers respond to citizens. Mastrofski and colleagues (2000) have found that experience reduces the willingness of officers to comply, either out of cynicism or wisdom. This variable is continuous, as officers in the POPN study ranged in experience from 0-31 years of service (Mastrofski et al., 2000). Finally, officers in the POPN study also varied in their assignments, with some engaged in traditional patrol and 911 response functions while some were assigned solely to community policing duties (Parks et al., 1999). These assignments are easily categorized as a dichotomous variable, and community policing officers are hypothesized to be more willing to comply with citizen requests for service. Control Variables Certain control variables are needed in the analytic model due to their stated relevance in previous studies (Mastrofski et al., 1996; Mastrofski et al., 2000; Sun et al., 2008). Two sets of variables (i.e., citizen and situational) are needed in the analysis in order to control for alternative explanations of officer behavior. 26 A common explanation of officer behavior has been the characteristics of the citizen who the officer is encountering. Previous literature reviews (Sherman, 1980; Riksheim and Chermak, 1993) indicate that citizen characteristics have been shown to impact officer behavior across studies in the United States. Citizen demographic characteristics collected during POPN include age, race, gender, and socioeconomic status. These citizen data are limited by the method of collection, where the trained observers collected these data via their own observations, rather than through interviews as the officer data were collected. The citizen’s demeanor, alcohol involvement, and evidence of a legal violation will all be included in the analysis as well. These variables are important because officers tend to have less regard for citizens who question their authority, and undesirable citizens who are drunk or may have committed a crime (Muir, 1977; Schafer and Mastrofski, 2005). A group of variables that describe the setting are also entered in to the model. A variable of particular concern will be the number of officers present at the scene, and the number of citizens present at the scene. In lieu of a specific effect indicated in the literature, it seems logical that officers would be less inclined to provide citizen service in front of fellow officers, as this may make the complying officer appear soft in front of their colleagues. The lack of research as to how the impact of the number of citizens present at an encounter would affect service makes determining results difficult. However, an officer who feels uncomfortable or intimidated with large groups may be less inclined to comply with requests. In addition, whether the encounter was initiated by the officer or the citizen (i.e., proactivity) should also be considered. It is posited that proactive officers would be more likely to comply with citizen requests, as their increased 27 activity level would increase willingness to serve citizens using discretionary time (Mastrofski et al., 2000). This variable can have an impact on officer and citizen behavior, as has been evidenced by Brown (2007). Study site is also an important consideration, as organizational differences and officer strategies may have an impact on an officer’s choice to return to service rather than invest time in a service encounter. In regard to study site, it is hypothesized that officers from St. Petersburg will be more willing to comply with citizen requests, as their policing style is less aggressive and emphasizes a problem solving approach. For a depiction of all variables and their hypothesized effect, see Appendix A. Data Analysis The dependent variables in this inquiry are best addressed through logistic regression and ordered logit. Logistic regression is useful when the dependent variable is categorical, but ordered logit allows the dependent variable to be measured by non- compliance, partial compliance/promised compliance and full compliance (Liao, 1994; Long, 1997; McKelvey and Zavoina, 1975). Using these two forms of analysis provides greater detail into the extent to which officers with citizen requests for service, and if not which officers are likely to offer a rationale. Validity, Reliability, and Generalizability Systematic social observation (S80) is the method used to collect the integral pieces of data for this inquiry. This method of field research utilizing trained observers generally produces valid data. Field observation and SSO in particular, allows researchers to capture events or phenomena that may be more difficult to glean fiom official records or survey data. Reactivity, or a change in the behavior of the observed 28 due to the presence of a researcher, introduces a threat to validity. This concern is common to field observation (Mastrofski and Parks, 1990) and generally if affected, officer behavior will return to normal after a short period of time (Westley, 1970). To address the potential pitfalls of reactivity POPN protocol used two measures of officer reactivity. In the first measure, ride level reactivity, the observer made an assessment of the officer’s reaction to the observer’s presence over the course of the shift. The second measure indicated the degree to which an observer believed that the officer altered their ’behavior during a specific encounter due to the observer’s presence (Spano, 2003). Overall, observers indicated only 0.5 percent of all observations may have been altered due to the researcher’s presence (Terrill, 2001 ). Upon deeper analysis of the reactivity issues, Spano (2003) found that signs of reactivity were generally associated with an officer’s concern for the observer’s safety, particularly if the observer was female. In order to minimize threats to validity occurring from reactivity, project directors promised the confidentiality of officers and allowed them to read the notes collected by their observers (Mastrofski et al., 1998). In terms of generalization, the POPN observational data is limited due the selection of beats within Indianapolis and St. Petersburg. Beat selection focused on areas particularly affected by economic distress, and may not be representative of the more advantaged beats within the two study cities. However, the POPN data do have advantages over other observational data on officer encounters (Paoline et al., 2000). Previous works have focused on much larger mimicipalities with large patrol staff and more bureaucratic management structures, making generalization to medium sized cities or rural areas more problematic. St. Petersburg and Indianapolis are more similar to a 29 larger portion of police organizations nationwide, making POPN data more apt for generalization than larger cities used in previous studies (Terrill, 2001). 30 Chapter IV: Results The following chapter contains the statistics, analyses, and findings for the current research questions. First, descriptive statistics (i.e., frequencies) are used to present the distribution of the dependent variable (e.g., compliance with service requests), the independent variables (e. g., officer gender, education, race, experience, and assignment), and the control variables (e.g., citizen race, gender, age, socioeconomic status, demeanor, drug/alcohol influence; number of officers present at the encounter, number of citizens present at the encounter, and whether the encounter was proactive). Second, bivariate distributions of the dependent measure by applicable officer characteristics (experience excluded in bivariate analysis) will be expressed. Finally, this section will employ binary logistic regression for the dependent variable measured dichotomously, and ordered logit for the ordinal level analysis. Descriptive Statistics Considering the conceptual framework of this study (i.e., compliance with service requests) the present analysis will include citizens identified by POPN observers as victims, disputants, service recipients, helpless persons, third parties, and witnesses]. As shown in Table 4.1, the majority of citizens making a request for service from the police are identified as victims (59.8 percent), followed by service recipients (13.6 percent), witnesses (9.8 percent), third parties (8.7 percent), disputants (7.1 percent), and helpless persons (1.1 percent). Citizens identified by POPN observers as suspects, quasi-police, non-police service providers, friends, and occupational acquaintances are not included in this analysis. 31 Table 4.1 Citizen Role Citizen Role N Percent Victim 622 59.8 Service Recipient 142 13 .6 Witness 101 9.7 Third Party 91 8.7 Disputant 74 7.1 Helpless Person 11 1.1 Total 1 ,041 100.0 The first two questions offered in this inquiry are the frequency with which citizens request services, and the fi'equency with which officers comply with these requests. To help address these questions, the descriptive statistics for officer compliance with citizen requests are depicted in Table 4.2. As shown, the base rate for full compliance across all requests is 63.3 percent. Citizens were more likely to request that officers provide information than any other service related request, followed by requests that an officer file a report, provide physical help for the requester or another citizen, and that an officer act on their behalf with an official or agency. Officers were most likely to fully comply with requests for information (79.4 percent), followed by requests for physical help (65.2 percent), requests that an officer act with an agency (58.5 percent) and requests that the officer file a report (40.5 percent). The total number of citizens in the dataset is 1,041 , indicating several citizens made multiple requests of the officer at a single encounter, as the total number of requests was 1,222. 32 Table 4.2 Citizens’ Requests for Service and the Police Response % Promised or # of % Not . % Completely Type Of Reques‘ Requests fulfilled Pm’any firlfilled Fulfilled File a Report 402 19.2 40.3 40.5 Act with an Agency 41 22.0 19.5 58.5 Provide Information 549 2.9 17.7 79.4 Provide Physical 230 15.2 19.6 65.2 Help Total 1222 11.2 25.5 63.3 A deeper examination indicates that multiple requests were made in 16.2 percent of cases. In the majority of these cases, officers chose to fulfill or deny all requests made by a citizen, and a split result (complying with at least one request while also not complying with at least one request) occurred in only 18.9 percent of cases where multiple requests were made (3.1 percent of all cases). Due to this complication, and considering the conceptual fi'amework where officer effort is a primary consideration, officers will be credited with their greatest degree of effort in the encounter, and officers in compliance with at least one request will be considered willing to comply with a citizen’s request. Table 4.3 shows the descriptive statistics for officer characteristics. Overall, 83 percent of officers were male, 38 percent had a four year degree, 25 percent were non- white, and 55 percent were in a community policing assignment. Additionally, the average officer has 8.11 years of experience. 33 Table 4.3 Descriptive Statistics: Officer Characteristics (N =l,04l) Variable Range Mean Std. Dev. Male* 0-1 .83 .380 Four Year Degree 0-1 .38 .485 Nonwhite 0-1 .25 .43 1 Experience“ 0-31 8.1 1 6.596 Community Officer 0-1 .55 .498 *35 missing cases **36 missing cases Depicted in Table 4.4 are descriptive statistics for the control variables. This includes both citizen characteristics and encounter characteristics. For the citizen variables, slightly less than half of citizens in these encounters were nonwhite (48 percent) and male (45 percent). Additionally, slightly more than half of the citizens involved in request encounters were of middle wealth or above. Few of the citizens were disrespectful (5 percent) or visibly under the influence of a substance (5 percent). Age is defined in terms of need, as was the method used by Mastrofski and colleagues (2000) to include requesters coded as either young or elderly. The reference category (0) includes citizens whom the perceived need would be low (i.e., 13-59 years old) and the study category (1) were citizens where the perceived need would be higher (i.e., preschool to 12 years and 60 years or older). Findings indicate that about 8 percent of citizens fall into a category that could be considered high need. In terms of the encounter, 10 percent were initiated by the officer. This is not unexpected considering that these are encounters in which the citizen is making a request of the officer. Additionally, on average 1.73 officers were present at each encounter. It should also be noted that 52 percent of encounters involved a single officer. The number of citizens present displays a wide range, fiom 1-120, with an average of 4.10. Despite 34 this wide range, 75 percent of all encounters had 4 or less citizens present, and 95 percent of encounters had 11 or fewer citizens present. Table 4.4 Descriptive Statistics for Control Variables Variable Range Mean Std. Dev. Citizen Variables Nonwhite 0-1 .48 .500 Male 0-1 .45 .498 Age 0-1 .08 .272 Middle Wealth Plus 0-1 .56 .496 Disrespectful 0-1 .05 .220 Drug/Alcohol Influence 0-1 .05 .210 Encounter Variables Proactive Encounter 0-1 .10 .300 Number of Officers* 1-11 1.73 1.144 Ntunber of Citizens* l-120 4.10 7.192 *denotes 2 missing cases Bivariate Statistics The third research question posed in this inquiry seeks to examine relationships between officer characteristics and willingness to comply with requests for service. This portion of the analysis is the first step in examining their potential relationships, through presentation of the data in joint distribution tables and use of the chi-square test of independence. Chi-square tests the relationships between officer gender, education, race, and assignment and the dependent variable (i.e., compliance with requests for officers to file a report, act on behalf of the citizen with an agency, provide physical assistance, or provide information). For this analysis, the chi-square test of independence is useful because it allows testing of the null hypothesis that officer characteristics and compliance with requests for service are distributed independently (Bachman and Paternoster, 2004). An indication of significance in a chi-square test means that the null hypothesis is rejected, and that it is likely that the variables are related in some manner. An additional 35 statistic provided within the contingency tables is gamma (7). Gamma is generally used when the variables of interest are measured in an ordinal manner. This statistic indicates the degree to which error in predicting the dependent measure is reduced by knowledge of the independent measure (Bachman and Paternoster, 2004). Table 4.5 displays the bivariate distribution of officer responsiveness by officer gender. As shown, officer gender and responsiveness are distributed independently (x2=2.45, p=.29), and the null hypothesis is assumed. This indicates that when measured as an ordinal variable, there is no relationship between officer gender and willingness to comply with requests for service. Table 4.5 Bivariate Distribution of Officer Responsiveness by Officer Gender (N =1,006) Officer Action Female 11 (%) Male n (%) Did Not Comply 16 (9.1) 76 (9.2) Promised or Partially Complied 49 (27.8) 186 (22.4) Fully Complied ll 1 (63.1) 568 (68.4) Total 176 (100) 830 (100) x2=2.454, p=.293, y=.098 The bivariate distribution of officer responsiveness by education is depicted in Table 4.6. Officer education and the willingness to comply with a citizen’s request is independently distributed (x2=2.90, p=.23), indicating that the null hypothesis should be accepted. This means that no relationship is shown between officer education and willingness to comply with a citizen’s request for service. 36 Table 4.6 Bivariate Distribution of Officer Responsiveness by Officer Education (N =1,04 1)