mm, a p . 5.1.33. , . unzufikwmwg :: gfiummfi .L. “n d.’3 .i I“, v. .1 4.: \ )4 12.5.5}... :1... : {AILE I'll!!! 2.? I’lrvthllézl'lb I Izgrl‘ilyklirv‘llk I :‘I.‘ ”UL This is to certify that the dissertation entitled Nanetrtpolitan Social Disorganizatjm: A Miltflevel Aralysis of QJaJity of Life presented by - Jeffreyl‘fid‘aelCaijn - has been accepted towards fulfillment of the requirements for Doctor of W degree in SSC Interdiscipljrary Doctoral Progran with a Carentratim inCrindIalllstjce I Major professor Date “01-2009- MS U it an Affirmative Action/Equal Opportunity Institution 0-12771 LIBRARY Michigan State University PLACE IN RETURN BOX to remove this checkout from your record. TO AVOID FINES return on or before date‘due. MAY BE RECALLED with earlier due date if requested. DATE DUE DATE DUE DATE DUE B s s ‘1' v {Bl r SEP 1mm: ‘7 3 MRS l § Zing 6/01 cJCIRC/DataDuopes-p. 15 NONMETROPOLITAN SOCIAL DISORGANIZATION: A MULTILEVEL ANALYSIS OF QUALITY OF LIFE BY Jeffrey Michael Cancino A DISSERTATION Submitted to Michigan State University in partial fulfillment of the requirements for the degree of DOCTOR OF PHILOSOPHY School of Criminal Justice 2002 ABSTRACT NONMETROPOLITAN SOCIAL DISORGANIZATION: A MULTILEVEL ANAYSIS OF QUALITY OF LIFE BY Jeffrey Michael Cancino Guided by social disorganization theory and the emerging concept of collective efficacy, this study investigates variation in citizens' quality of life assessments in nonmetropolitan settings. Using survey data from 1,125 citizens nested in 31 residential units located in the State of Michigan, hierarchical linear modeling is used to examine the effects of structural antecedents, collective efficacy and crime on citizen-level quality of life outcomes. Results suggest that traditional social disorganization variables, such as economic disadvantage and residential instability, are not linked to quality of life. Collective efficacy, however, is inversely associated with citizen quality of life assessments. Citizens from residential units with higher concentrations of property crime report higher levels of perceived crime, fear, incivility, and risk of victimization. Dedicated to Brailee Madisen Cancino. Forgive me for missing the first two years of your life. Find a career that makes you happy. Dedicated to Brandi, you did a great job raising our daughter. Dedicated also to my loving and supporting parents whose educational dreams were realized through my accomplishments. I love you! iii ACKNOWLEDGEMENTS There are several people I would like to thank. First, my sincere gratitude is extended to Mike Reisig. Thank you for your mentoring. I would like to acknowledge David Carter for his support throughout my education at MSU. I am grateful for the academic insights that Peter Manning has provided over the years. I would like to thank Carol Weissert for her guidance. Lastly, I thank Kristy Holtfreter and Marcus Mizanin for their data collection. There are a few people who touched my life during my education at MSU. First, I would like to thank Beth Huebner for her help and advice during this dissertation. I wish you happiness. Second, I am appreciative and grateful for Meghan Sarah Stroshine, whose close friendship and support helped me through this project. But, most of all, thank you for the laughs and good times. Perhaps there is one person that I am truly indebted to, Sean Patrick Varano. Thanking you for everything would be equivalent to writing another dissertation. Thanks for your friendship, sacrifice and many laughs. I am going to miss you. iv TABLE OF CONTENTS LIST OF TABLES ....................................... viii LIST OF FIGURES ......................................... x CHAPTER 1 RESEARCH FOCUS .......................................... 1 The Nonmetropolitan Setting ........................... 2 The Nonmetropolitan-Metropolitan Continuum ............ 3 Defining Meaningful Patterns of Residential Location in the Urban Setting .................................. 5 Residential Units in Nonmetropolitan Areas ............ 7 Social Disorganization in the Nonmetropolitan Setting .............................................. 11 Research Objective ................................... 12 Organization of Dissertation ......................... 13 CHAPTER 2 THEORETICAL AND EMPIRICAL LITERATURE ................... 15 The Chicago School of Sociology: Pioneers in Social Disorganization Theory ................................ 16 The Chicago School Tradition: Three Eras ......... 17 Social Disorganization Theory .................... 20 Economic Status .................................. 22 Residential Mobility ............................. 23 Population Composition ........................... 23 Community Supervision and Control ................ 24 Summarizing Shaw and McKay ....................... 24 Contemporary Social Disorganization Theory ........... 26 Social Disorganization Theory: Beyond the Metropolis ....................................... 32 The Salience of Collective Efficacy .................. 36 Collective Efficacy .............................. 37 The Mediating Effect of Collective Efficacy ...... 4O Nonmetropolitan Social Organization .................. 44 Social Disorganization and Quality of Life: Making the Connection ........................................... 50 Quality of Life .................................. 50 CHAPTER 3 HYPOTHESES, DATA AND METHODS ........................... 55 Hypotheses ........................................... 55 Social Disorganization and Crime ................. 56 Social Disorganization and Collective Efficacy ... S7 Collective Efficacy, Crime and Quality of Life.... 58 Crime and Quality of Life ........................ 59 Social Disorganization and Quality of Life ....... 61 Data ................................................. 64 Community Survey ................................. 65 Census Data ...................................... 68 Official Crime Data .............................. 70 The Residential Unit ................................. 71 Constructing the Residential Unit ................ 72 Nonmetropolitan Residential Unit Variation ....... 73 Variables ............................................ 75 Citizen-Level Quality of Life Outcomes ........... 76 Social Disorganization Variables ................. 78 Collective Efficacy .............................. 84 Burglary Index ................................... 86 Citizen-Level Socio-demographic Variables ........ 87 Analytic Strategy .................................... 90 Rationale ........................................ 90 Analysis Procedure ............................... 91 Conclusion ....................................... 92 CHAPTER 4 FINDINGS ............................................... 93 Preliminary Statistics ............................... 93 Model Diagnostic Procedures ...................... 93 Citizen—Level Associations ................... 93 Residential Unit-Level Associations .......... 95 Discriminant Validity of Outcome Measures ........ 96 Citizen-Level ................................ 96 Residential Unit-Level ....................... 97 Correlates of Quality of Life at the Residential Unit—Level ....... . ................................ 97 Bivariate Associations Between Residential Unit Independent Variables ........................... 100 Hierarchical Linear Models .......................... 102 One-Way Analysis of Variance (ANOVA) Models ..... 103 Random Coefficient Models ....................... 105 Intercept-As-Outcome Models ..................... 111 Fixed Effects Hierarchical Models (Full Models).. 122 Discussion .......................................... 141 CHAPTER 5 SOCIAL CAPITAL AND POLICY IMPLICATIONS ................ 145 Defining Social Capital ............................. 146 Social Capital and Collective Efficacy .............. 147 Similarities .................................... 148 Distinctions .................................... 149 The Process of Social Capital ....................... 150 vi Economic Capital: Government Policy ............. 150 Cultural Capital: The Church, School and Police 152 Social Capital: The Family ...................... 154 Social Capital: The Community ................... 155 Policy Implications ................................. 159 Police Resources and Public Control ............. 160 Community Resources and Public Control .......... 163 Conclusion .......................................... 167 Future Research ..................................... 169 ENDNOTES .............................................. l 72 BIBLIOGRAPHY .......................................... 193 vii Table Table Table Table Table Table Table Table Table Table Table Table Table Table Table 10: 11: 12: 13: 14: 15: LIST OF TABLES Sample Statistics by Location ................. 66 Pooled Sample Characteristics ................. 68 Descriptive Statistics for Quality of Life Outcomes ................................. 78 Mean Percentage and Range of 1990 Census Items .................................. 81 Factor Pattern for Disorganization Variables ..................................... 83 Descriptive Statistics for Independent Variables ..................................... 89 Zero-order Correlation Coefficients between Citizen-Level Variables ............... 94 Zero-order Correlation Coefficients between Quality of Life Measures and Residential Unit-Level Measures ............... 99 Zero-order Correlation Coefficients between Residential Unit-Level Variables ..... 101 Decomposition of Variance and Residential Unit-Level Reliabilities of Quality of Life Measures ............................... 104 Random Coefficient Quality of Life Models 107 Variance of Quality of Life Models .......... 110 Intercept-as-Outcome Models for Perceived Crime ....................................... 114 Intercept-as-Outcome Models for Fear of Crime ....................................... 116 Intercept-as-Outcome Models for Perceived Incivility .................................. 119 viii Table Table Table Table Table Table Table 16: 17: 18: 19: 20: 21: 22: Intercept-as-Outcome Models for Risk of Victimization ............................... 121 Fixed Effects Hierarchical Models for Perceived Crime ............................. 124 Fixed Effects Hierarchical Models for Fear of Crime ....................................... 128 Fixed Effects Hierarchical Models for Perceived Incivility ........................ 133 Fixed Effects Hierarchical Models for Risk of Victimization ....................... 136 Summary of Hypotheses Between Residential Unit-Level Variables and Quality of Life Outcomes ................................... 139 Summary of Hypotheses Between Residential Unit-Level Variables ........................ 140 ix Figure Figure Figure Figure Figure Figure Figure LIST OF FIGURES Shaw and McKay's Classic Social Disorganization Theory ....................... 26 Sampson and Groves' Revised Theoretical Social Disorganization Model ........................ 32 Osgood and Chambers’ Theoretical Social Disorganization Model Extended to Nonmetropolitan Settings ..................... 35 Sampson et al.'s Collective Efficacy Model ... 41 Sampson and Raudenbush's Theoretical Collective Efficacy—Crime and Disorder Model ............ 43 The Theoretical Model to be Tested ........... 54 The Processes and Resources of Social Capital in Nonmetropolitan Settings ................. 158 CHAPTER ONE: RESEARCH FOCUS Urban sociological research has shown that quality of life varies across neighborhoods. Despite the wealth of research in the urban setting, little is known about the effects of ecological features on quality of life in more rural, less densely populated areas. For example, since the pioneering work of sociologists from the University of Chicago in the early part of the 20th century, large urban centers, such. as Chicago, Philadelphia, Boston, and. New York, have been the focus of criminological research. Empirical attempts tx> explore less “traditional settings, such as nonmetropolitan areas, are few and far between. Large metropolitan areas have experienced numerous economic and. social changes throughout .American. history. Many “big city” problems have been further complicated by shifts in rural populations (Madison, 1986). For example, urban areas experiencing an influx of rural citizens have been described as “disorganized" because of the breakdown in informal social controls that often follows migration to urban settings (Maccoby, Johnson, and Church, 1958). Progressive reformers of the late 19th and early 20th centuries focused on the negative consequences of rapid urban growth, but devoted little attention to rural areas (Madison, 1986:645). Urban areas were not alone in experiencing the pains of in-migration, and negative outcomes have also been noted in nonmetropolitan settings. For example, the exodus of rural residents can disrupt the social and economic fabric of nonmetropolitan areas (Albrecht, Albrecht, and Albrecht, 2000; Tickamyer and Duncan, 1990). To date, urban studies continue to far outnumber research conducted in less populated areas. As a result, the study of the factors affecting quality of life remains, in a relative sense, neglected. The Nonmetropolitan Setting Although the terms nonmetropolitan and rural areas are often used interchangeably, the former entails more variation in residential settings: from sparsely populated rural areas to small towns to more densely populated small city neighborhoods. Conceptualizing nonmetropolitan areas in a manner beyond simple rural classification is necessary because variation in social histories and economic patterns of deve10pment exist within the nonmetropolitan continuum. Subsequently, this reconceptualization will help provide a better understanding as to how variation across nonmetropolitan areas affects citizen quality of life. For example, research shows that citizens residing on rural farms report higher levels of fear of crime and risk of victimization than residents of small towns (Bankston, Jenkins, Thayer—Doyle, and. 'Phompson, 1987; see ralso Saltiel, Gilchrist, and Harvie, 1992). While it is necessary to identify variation in the nonmetropolitan setting, it is also informative to highlight the location of the nonmetropolitan conception on a larger continuum. The Nonmetropolitan-Metropolitan Continuum Nonmetropolitan and metropolitan areas can be conceptualized along a continuum (Sanders and Lewis, 1976:36). On one side of the continuum are nonmetropolitan areas. These areas are generally characterized as family- oriented, culturally homogenous, and spatially secluded. On the other side are metropolitan areas. These areas are stranger-oriented, culturally diverse, and spatially dense. Fisher (1982) found that most small—town social interaction is kinwbased, whereas urban social interaction is stranger- based. Research by Amato (1993) shows that urban residents assist and receive help from friends more than rural residents, and that urban residents live greater distances from relatives. Hofferth and Iceland (1998) found that familial ties tend to be stronger in rural areas. In combination, the evidence strongly suggests that social interaction varies across the nonmetropolitan- metropolitan continuum. Whether this variation influences citizen quality of life remains an open empirical question. Others question the validity of the nonmetropolitan— metropolitan continuum, and claim that exploring differences may be premature (Bell, 1992). Smith and Huff (1982) suggest nonmetropolitan and urban areas share many similarities. Among them include the prevalence of single mothers and pmwerty (Bachman, 1992; Laub, 1983; Smith and Huff, 1982). Critics of the nonmetropolitan-metropolitan continuum allege that economic and social characteristics reflect smaller gradations of variation rather than possessing entirely distinct features. While urban and rural researchers remain divided, recent evidence suggests that various forms of social and economic disadvantage are becoming increasingly common in sparsely populated rural areas. For example, poverty rates have increased disproportionately in rural areas (Albrecht et al., 2000). Do structural and social processes influence citizen quality of life in nonmetropolitan America? Beggs, Haines, and Hurlbert (1996) found that rural residents structure their social and personal ties differently according to location and environmental concerns. Beggs et a1. (1996:320) posit that “locality matters and, therefore, the more general call for taking seriously contextual effects in analyses of individual behavior” is necessary. Moreover, Belyea and Zingraff (1988:473) argue that urban-rural differences are explained “by behavior and attitudes determined by resident’s positions and interactional patterns within the local structures.” By exploring the effects of structural features and social processes in the nonmetropolitan setting, a better understanding of the causes and correlates of citizen quality of life may be revealed. Defining Meaningful Patterns of Residential Location in the Urban Setting Since the early ecological studies of the Chicago School, researchers have struggled to adequately define the term “community." For the most part, the term has been defined rather loosely; For example, Hollingshead (1948:145) notes that researchers have defined community in three different ways: 1) social aspects, such as group solidarity, cohesion, and social interaction; 2) geographic features, such as census tracts, block groups, roads, and businesses; and 3) socio-geographic characteristics where researchers attempt to identify both social and geographic components. Concerning the latter, Hillery (1955:11) argues that a definition of community consists of “persons in social interaction within a geographic area and having one or more common ties.” While the above characteristics provide some guidance, contemporary attempts to define “community" remain imprecise and must be approached with caution. Given the increased awareness among social scientists of the salience of macroeconomic forces (e.g., economic disinvestment) in determining within-community outcomes (see, for example, Wilson, 1987), the use of community as a unit of analysis has fallen out of favor because it fails to capture meaningful variation across smaller geographic units. Accordingly, researchers have turned to smaller units of analysis, such as “neighborhoods.” Burton, Price— Spratlen, and Spencer (1997:133) note that neighborhoods are conceptualized in different ways: neighborhood as geographic site, neighborhood as perception, neighborhood as social network, and neighborhood as culture. Given the many ways in which neighborhoods are conceptualized, it appears that this aggregate unit also suffers from a lack of precision. Despite the lack of precision (conceptual and empirical) in determining what constitutes “community” or “neighborhood,” debate persists over which unit of analysis is most useful in identifying salient social forces. For example, Massey (1996) claims that an individual's immediate surroundings are more likely to influence perception and behavior than the larger environment, such as census tract boundaries. In fact, many scholars have suggested that neighborhoods matter (Wilson, 1987; Elliot et al., 1996; Baba and Austin, 1989:179). The same cannot be said, however, for less densely populated geographic areas. Part of the problem lies in the fact that defining meaningful geographic units in nonmetropolitan areas is complicated by many factors. Residential Units in Nonmetropolitan Areas The term neighborhood is problematic in nonmetropolitan areas because it was developed by social scientists primarily interested in urban settings. Osgood and Chambers (2000:82) claim that “theory and research on crime and communities has almost exclusively defined communities as neighborhoods within large urban centers.” Others claim that neighborhoods may be an important unit of analysis irlani urban environment, but such aggregates are less applicable in rural areas (Darling and Steinberg, 1997). Thus, using the neighborhood as a unit of analysis in. nonmetropolitan. areas would likely result in several problems. The first problem is that nonmetropolitan areas vary from sparsely populated rural areas to small towns to more densely populated small cities. The neighborhood conception fails to take into account such variation. In addition, nonmetropolitan residents are likely to make the most of their extended surroundings more so than urban residents. Specifically, nonmetropolitan residents are more likely to travel farther to access social and physical resources due to open country characteristics (Darling and Steinberg, 1997:122). For example, residents in rural areas are more likely to live several miles from convenient stores, acquaintances, and other places of necessity (e.g., medical services), whereas resources and places of necessity are more abundant and in close proximity to residents in urban areas. For these reasons, rural residents are likely to make use of larger' geographic spaces out of necessity, while urban residents have the ability to choose among several resources available within walking distance. Another reason why the neighborhood conception is not suitable for the rural setting is that inner-city urban areas are economically homogenous (Wilson, 1987); however, nonmetropolitan areas encompass a wide range of social classes and incomes (Osgood and Chambers, 2000:109; Lynch and Cantor, 1992:345) . Given the problems associated with def ining “communi ty” and “ne ighborhood” in the nonmetropolitan setting, an alternative specification for aggregate residential patterns must be formulated. An aggregate unit that more accurately represents nonmetropolitan economic and social variation is a “residential unit.” Residential units encompass a larger body of land and include a smaller number of people when compared to urban neighborhoods. Residential units are unique because they capture rural structural and social variation that differ from urban “neighborhood clusters" (see Sampson, Morenoff, and Earls, 1999:638-639).l Although residential units are unique to nonmetropolitan areas, they possess many of the same characteristics of “community" and “neighborhood.” These characteristics are best understood along the nonmetropolitan-metropolitan continuum. For example, at the nonmetropolitan end, more rural residential units resemble “community” characteristics because they cover larger geographic areas. Moving toward the metropolitan end of the continuum residential units in small towns begin. to [possess characteristics similar to neighborhoods in that they represent smaller bodies of land and are more densely populated. Conceptually, then, residential units are similar with respect to traditional community' and. neighborhood. characteristics, but also overcome the weakness inherent in both traditional conceptualizations. Why do residential units matter? Residential units serve as unique units of analysis by providing an alternative aggregate measure that captures social and economic characteristics of nonmetropolitan areas. Because social and economic factors are hypothesized to be associated with citizen quality of life, residential units as an aggregate measure can help researchers to better understand the dynamics of citizen quality of life in these areas. The assumption here is that individuals respond to their social and economic surroundings. Since nonmetropolitan residents tend to structure social networks according to ecological concerns, citizen quality of life ultimately depends on the overall social organization of the community. 10 Social Disorganization in the Nonmetropolitan Setting Nonmetropolitan social and economic factors can be placed within the theoretical context of social disorganization. Social disorganization theory is based on the premise that low economic status, ethnic heterogeneity, and residential mobility lead tx: the disruption of community social organization, which in turn, leads to crime and delinquency (Shaw and McKay, 1942). More recently, social disorganization theory has been extended to include individual-level outcomes, such as legal cynicism (Sampson and Jeglum-Bartush, 1998). While many theories have been developed to explain the urban-crime phenomena, Laub (1983:192-194) suggests that most theories are germane with respect to nonmetropolitan areas. Supporting this claim, Osgood and Chambers (2000:85) argue that “social control has everything to do with general principles of social relations and nothing to do with urban versus rural setting.” This statement suggests that mechanisms of informal control are generalizable to nonmetropolitan settings. Osgood and Chambers (2000) tested the generalizability of social disorganization theory to nonmetropolitan areas. There is no reason to believe that community social disorganization ll should be strictly confined to urban areas. In addition, given recent out-migration of people from urban to suburban and less densely populated rural areas, testing social disorganization in communities of all sizes is warranted. Osgood and Chambers (2000:85) also contend that social disorganization is not uniquely urban when focusing on geographic locations, such as concentric zones. This uniquely urban approach has continued for too long in the criminological literature, and the time has come for social scientists to expand their research outside the urban-box and investigate less densely populated nonmetropolitan areas. Doing so is worthwhile because many of the structural features of urban life, such as poverty, are also found in nonmetropolitan areas. Research Objective Given these gradations of economic and social factors, a primary objective of this research is to assess the generalizability' of social disorganization. theory' to nonmetropolitan settings. Two research questions are proposed: First, do structural features (e.g., economic disadvantage) influence citizen quality of life in nonmetropolitan areas? Second, do social processes found 12 to influence crime and quality of life in urban areas have a similar influence in less densely populated areas? Again, the rationalization for this research is that structural and social characteristics are applicable to all types of environments. By assessing both social and structural variables simultaneously on a variety of citizen quality of life outcomes, this dissertation will provide a better understanding of the influence of ecological factors. Sampson and Wilson (1995:44) favor this community-level approach because “[r]esearch conducted at the individual level rarely questions whether obtained results might be spurious and confounded with community- level processes.” Therefore, the present research examines both aggregate and individual-level factors to investigate the link between social disorganization, crime, and perceived quality of life in nonmetropolitan areas. Organization of Dissertation The dissertation is organized into five chapters. Chapter Two provides an overview of the theoretical framework, as well as the extant research is this area. Chapter Three specifies the hypotheses to be tested, the methodological design, and identifies and defends the measurements of the independent and dependent variables 13 used in the analysis. Chapter Four reports the major research findings. The final chapter discusses the implications of the results in both theoretical and practical terms. 14 CHAPTER TWO: THEORETICAL AND EMPIRICAL LITERATURE The theoretical framework guiding this research is derived from the ecological perspective. The ecological perspective suggests that structural and social characteristics of a community or neighborhood influence an array of outcomes, such as crime and delinquency. Perhaps the most influential ecological perspective is social disorganization theory. Social disorganization theory posits that adverse structural changes (e.g., low economic status, residential mobility, and. ethnic heterogeneity) undermine social control, which in turn, fosters crime and delinquency in urban neighborhoods (ShaW' and. McKay, 1969). In recent years, however, social disorganization theory has been refined (Sampson and Groves, 1989), and used to explain crime in. nonurban settings (Osgood. and. Chambers, 2000). More recently developed models have emerged that identify specific neighborhood social mechanisms, such as collective efficacy, that mediate the effects of neighborhood disorganization (Sampson et al., 1997; Sampson et al., 1999; Sampson and Raudenbush, 1999). Put simply, social disorganization theory is well represented in the criminological literature. 15 From a larger perspective, neighborhood social disorganization has been linked to macrostructural antecedents, such as deindustrialization (Wilson, 1987; Sampson. and.‘Wilson, 1995). Deindustrialization. disrupts the economic stability (e.g., unemployment) and neighborhood social organization, which ultimately fosters crime. This chapter provides a review of influential ecological studies from the Chicago School, discusses the traditional dimensions of social disorganization theory, highlights. recent theoretical developments, discusses the role of collective efficacy, and articulates the link between social disorganization and quality of life. The Chicago School of Sociology: Pioneers in Social Disorganization Theory Much of the community ecology literature (e.g., locales with low economic status and changing population patterns) is influenced by the early work of the Department of Sociology at the university of Chicago. Park, Burgess, Shaw, and McKay are considered by many to be the pioneers of the Chicago School (Vold, Bernard, and Snipes, 1998:140- 149). During the early part of the Zou‘century, members of the Chicago School posited that social problems (e.g., crime and delinquency) were a product of the social 16 disruption influenced by ecological attributes of urban neighborhoods (i.e., concentric zones). What resulted from the attempt to better understand the impact of neighborhood conditions on various social problems, such as crime, was social disorganization theory. The following section provides a brief chronology of the early development of social disorganization theory. The Chicago School Tradition: Three Eras During the early part of the 20th century, Chicago School researchers viewed cities as living, growing, and constantly changing social systems. Successive waves of new immigrants from foreign countries and in-migration from rural areas were the impetus for change in urban communities. Change in the urban setting was believed to be linked to several aspects of social life, especially crime. The development of ecological scholarship occurred over a span of three eras (Hollingshead, 1948). The first era was termed the “normative-meliorative” because of its emphasis on improving and solving community problems (Hollingshead, 1948:137). The primary method of social inquiry during this era was to survey the community through fieldwork and first—hand contact with residents about ongoing social problems (Snodgrass , 1976 ) . 17 Researchers from this era were interested in alcoholism, poverty, crime, and economic exploitation. These problems were not distributed evenly, and were most common in slum areas where most research was conducted. The slums, however, were not the only areas of interest. Sociologists from this period also focused on country life because of decaying and malfunctioning institutions, such as the church and the family, and the migration of young people to the city (Hollingshed, 1948:138). The following era, which has been termed the “analytical” era, was essentially the birth of the Chicago School. Researchers from this period were interested in f ,- the history, development, population, and organization of urban communities (Hollingshed, 1948:140). Sociologists from the Chicago School studied “objectively the communal collectivity in terms of its relation to the larger society" (Hollingshed, 1948:137). This period produced more detailed and statistical research that centered on community norms and values. Robert E. Park and Ernest W. Burgess (1925) were among the first scholars to shed light on the effects of city growth. According to Hollingshed (1948:138-139), Park was influenced by Galpin's (1915) rural research2 by “bringing into focus his own thoughts 18 about the relationship between city growth and structure, institutional services, neighborhoods, and natural areas.” Due to significant social and economic change in large cities, urban areas became the focus of attention. The combined effort of Park and Burgess (1925) was one of the first studies examining the relationship between community ecology and crime in urban areas. Both Park and Burgess (1925) were interested in the processes of city' growth related to crime and delinquency. Their study consisted of dividing the city of Chicago into five concentric zones.3 Park and Burgess (1925) hypothesized that crime and delinquency would burgeon in Zone II (i . e . , the zone of transition) , which was most unstable (1 . e . , disorganized) and slum—like. Results showed that the zone of transition was responsible for much of the crime and delinquency in Chicago. Specially, Zone II displayed many characteristics of ineptitude, such as attrition of businesses and residents, deterioration in housing, low-income persons, and a breakdown in mechanisms of institutional social control (Reid, 2000:116-117). By the late 19203, rural and urban community research proliferated. The abundance of community research gradually shifted the focus of interest from norms and values to structural characteristics. For example, 19 sociologists were interested in the interrelations between people (e.g., social networks), social organizations (e.g., church and family), and social structures (e.g., the organizathn of both people and institutions)(Hollingshed, 1948). Thus, in 1930, the structural era emerged. During W this era, Clifford Shaw and Henry McKay were key figures. It was during this era that so/cmial disorganization theory emerged. As history shows, it appears that rural community research shaped the way Chicago School researchers examined urban communities. Suffice it to say, the Chicago School’s ecological perspective and Shaw and McKay's (1942) social disorganization theory has had a long shelf life. Social Disorganization Theory Shaw and McKay (1942) posited that juveniles were relatively normal and that delinquency was due to their physical and social environment. Guided by this view, Shaw and McKay (1969:xxxviii) analyzed Chicago neighborhood characteristics and official police and court records of juvenile delinquents. Shaw and McKay (1969) argued that -\ the capacity for social control in urban communities was much lower when compared to suburban and rural areas.‘ Much like Park and Burgess (1928), Shaw and McKay (1969:145) 20 found that certain neighborhoods in “transition” showed the highest delinquency rates in Chicago. The theoretical premise of Shaw and. McKay’s (1942) original formulation was that crime and delinquency was the result of weakened. community social organization. due to ’,,_ structural characteristics (e.g., low—economic status, residential mobility, and population composition). Shaw and McKay (1969) linked social disorganization (e.g., residential instability) to conditions endemic in the urban areas where newly' arriving‘ poor' were forced to settle. Because crime—producing factors were alleged to be inherent in neighborhoods where immigrants and the poor resided, one solution was to control these factors by developing means of informal social control, such. as supervising teenage peer groups. Shaw and McKay (1969) dismissed individualistic explanations of delinquency. Instead, they concluded that neighborhood-levelS attributes beEter explain aggregate patterns of crime (Shaw and McKay, 1969:320). Thus, social / disorganization is an ecological theory about places, not a theory of persons. Shaw and McKay's (1942) neighborhood level dimensions are important because over the last 75 years they have been a mainstay in criminological theory. The following subsections provide a detailed assessment of 21 the three structural dimensions most closely associated with social disorganization theory. Economic Status According to Shaw and McKay (1969:147-152), socially disorganized communities characterized by low economic status lack the necessary resources to effectively establish social control. Shaw and McKay (1969:4) posited that in order to resolve social problems in a given community, social organizations (e.g., churches and local neighborhood groups) must promote community awareness and intervention. A weak organizational base is the result of neighborhood residents failing to organize socially in terms of natural forces for effective social control (Vold et al., 1998:148). The low economic status-crime link suggests that communities characterized by low economic status should experience higher concentrations of crime. Overall, Shaw and McKay (1942) concluded that neighborhood socio-economic status does not have a direct relationship with crime, but instead effects mechanisms of social control, which then give rise to crime (Shaw and McKay, 1969:145). 22 Residential Mobility The second dimension associated with social disorganization theory is residential mobility. Residential mobility was conceptualized as the process whereby residents continuously move in and out of a given community (Shaw and McKay, 1969:147-149). Shaw and McKay (1969) posited that residential mobility would impede the development of social relations among residents . Therefore, the relationship between residential mobility and crime is that neighborhoods with high levels of population turnover (i.e., residential mobility) should experience higher rates of crime and delinquency because mobility disrupts social organization (Shaw and McKay, 1969:148). The inability' to establish social relations undercuts prevention and problem solving, which then promotes crime. Population Composition A third dimension of social disorganization theory is population composition. Shaw and McKay (1969:145) hypothesized that delinquency was associated with other social problems related to the migration of new residents of different values, norms, and beliefs into a neighborhood. The connection between population 23 composition and crime is based on the premise that a mix of racial and ethnic people disrupts the social equilibrium of neighborhood social control, which in turn, leads to crime and delinquency (Shaw and McKay, 1969:155). As a result, Shaw and McKay (1969:155) hypothesized crime and delinquency would be high in ethnically diverse neighborhoods. Community Supervision and Control According to Shaw and McKay (1969:176-185) residents of homogenous, stable, and affluent communities are more effective in controlling teenage deviance. Effective community supervision of teens is expected to act as a buffer by mediating the impact of social disorganization on delinquency. Their research showed that the ability of a community to supervise and control teenage peer groups who resided in disorganized communities was inversely related to crime and delinquency (Shaw and McKay (1969:176-186). Summarizing Shaw and.MCKay Neighborhood-level dimensions of social disorganization are connected in the sense that they work in combination and lead to neighborhood decline. For example, neighborhood residents who suffer from economic hardship and continuously move are less likely to 24 participate in the development of informal social control. The lack of resources fails to provide an economic commitment (e.g., owning the home) to the neighborhood. High residential turnover fails tx: permit sufficient time to establish common values with others in the neighborhood. In addition, neighborhoods that are racially and ethnically heterogeneous will be characterized by lower levels of social organization because of conflicting cultural values. The net result is that residents do not identify with the neighborhood, its appearance, or reputation and thereby lack common social bonds to effectively' build informal controls to deal with social problems (Vold et al., 1998:147). In the end, residents are left with a socially disorganized neighborhood where crime is more common. Figure 1 shows Shaw and McKay's theoretical model. 25 Figure 1: Shaw and McKay’s Classic Social Disorganization Theory Low economic status Inability of Residential _, 8 COMitY > Crime mobility to supervise and and control Delinquency teenage peer groups Population composition i Contemporary Social Disorganization Theory Some 75 years since its introduction, social disorganization theory is regularly discussed in the literature. A.long list of scholars have reformulated and refined Shaw and McKay's (1942) classic model (Bursik, 1988; Kornhauser, 1978; Sampson and Groves, 1989; Sampson et al., 1997; Sampson and Raudenbush, 1999), while others have tested it in nonurban settings (Kowalski and Duffield, 1990; Osgood and Chambers, 2000; Wilkinson, 1984). For the most part, Shaw and McKay’s (1942) original formulation is still intact. 26 While Shaw and McKay (1942) emphasized community disorganization characteristics--low economic status, residential mobility, and population composition--that undermined. community’ supervision. and fostered. crime, Kornhauser (1978) was the first to emphasize mechanisms of social organization in the community to explain crime. More specifically, she put forth the notion that community social control is nested in social disorganization. Kornhauser (1978:79) argued that institutional instability (e.g., a lack of churches) were key structural factors. As a result, social organization should be viewed as a control model. Put simply, social disorganization. weakens community social control (Kornhauser, 1978). Kornhauser’s argument is based on the systemic model. The systemic model is characterized as a system of controls involving locally based social networks, such as friends, family, and acquaintances that constitute the core social fabric of communities (Kasarda and Janowitz, 1974). According to Bursik and Grasmick (1993), these social networks (e.g., friends) represent different levels of control within communities. The relevance of the systemic model is that it addressed early criticism that the Chicago School researchers overemphasized dimensions of disorganization 27 (Sampson, 1995:556).6 Whyte (1934:75) argued that what looks like social disorganization from the outside is actually an intricate internal organization. He continued by suggesting that problems of the slums were the result of failed social organization that did not adapt to the changing structure of the community. Studies have found that integration of social ties, based on the systemic model, are important mediators between ecological influences and crime (Sampson, 1987; Sampson, 1988). Over the years, the systemic model has helped researchers better understand the community social disorganization-social process connection. Community social organization has been conceptualized as the ability of a community to realize the common values of its residents and maintain effective social controls (Kornhauser, 1978:120; Bursik, 1988; Sampson, 1988). ‘When a community's formal and voluntary organizations are weak, the ability of a community to defend against local problems (e.g., crime) is greatly reduced. When residents form local social ties their capacity for community social control is increased because they are better able to recognize strangers and likely to engage in guardianship behavior (Taylor et al., 1984:307; Skogan, 1986:216). 28 While different attributes of the systemic model have been shown to be inversely related to crime, researchers have become interested in contextual dimensions of community social organization likely to have the same affect (e.g., collective efficacy). In essence, the systemic model has (1) provided a starting point for measuring social processes and (2) allowed researchers to re-conceptualize social mechanisms of control at the aggregate level (Sampson and Lauritsen, 1994:58-59; see also Bursik, 1988). In a recent article by Markowitz et a1. (2001:311), Sampson and Groves (1989) are given credit as the first to examine the notion that neighborhood cohesion mediates the relationship between social disorganization and crime. While Sampson and Groves (1989:782) relied on the theoretical framework of Kornhauser (1978), who claimed that social disorganization was part and parcel to social organization, they also viewed social organization and disorganization as different ends of the same continuum. As they put it, “[s]ocial disorganization is clearly separable not only from the processes that may lead to it (e.g., poverty and residential mobility), but also from the degree of delinquent behavior that may result from it” (Sampson and Groves, 1989:778). 29 Sampson and Groves' (1989) study was unique because it examined various mechanisms of social organization, which was one of the criticisms of Shaw and McKay's (1969) model. Sampson and Groves' (1989) considered low economic status, residential mobility, and ethnic heterogeneity (traditionally known as population composition) as exogenous variables. Sampson anui Groves (1989:774) hypothesized that low economic status, residential mobility, ethnic heterogeneity (Shaw and McKay's original model) and family disruption led to community social disorganization, which in turn, increased crime and delinquency rates. They also viewed Shaw and McKay’s original dimension of supervised teenage peer groups as an endogenous variable. Results showed that “communities characterized by sparse friendship networks, unsupervised teenage peer groups, and low organizational participation had disproportionately high rates of crime and delinquency” (Sampson and Groves, 1989:799). It is interesting to note that social disorganization (e.g., low' socioeconomic status, residential stability, family disruption, heterogeneity) accounted for much of the effect on rates of burglary.7 Consistent with Sampson and Groves (1989), recent research shows that burglary is influenced by 30 several indicators of social disorganization (Rountree et al., 1994; Lynch and Cantor; 1992), such as single parent households (Smith and Jarjoura, 1989). Moreover, Sampson and Groves’ (1989) research also supports Shaw and. McKay's theoretical model. Figure 2 shows Sampson. and (Groves’ revised. social disorganization model. The italicized dimensions indicate changes to Shaw and McKay’s original model. 31 Figure 2: Sampson and Groves’ Revised Theoretical Social Disorganization Model l Low economic status Sparse local friendship Residential networks mobility Uhsu ervised Ethnic teenage peer I Crime heterogeneity — F groups and Delinquency Family disruption Low organizational ‘ participation urbanization Social Disorganization Theory: Beyond the Metropolis Despite the :numerous intellectual and. empirical contributions of the Chicago School, early sociologists focused primarily on urban areas. For example, Park and Burgess (1928; see also Shaw and. McKay, 1942) examined outer zones of the city including satellite towns and suburbs, but they did not examine structural characteristics in nonmetropolitan areas. For these reasons, Osgood and Chambers (2000:89) argue that a primary shortcoming of sociological research is that it focuses on 32 variation among neighborhoods within a single metropolitan area (see also Bursik, 1988). To address this shortcoming, Osgood and Chambers (2000:84) studied nonmetropolitan communities, such as small towns, that varied in their ability to realize common values and solve social problems (Osgood and Chambers, 2000:84). Their study included an analysis of arrest rates of juvenile violence in. 264 nonmetropolitan counties in four states. They reported that residential instability was associated with higher rates of rape, aggravated assault, weapons violation, and simple assaults (Osgood and Chambers, 2000:102). Ethnic heterogeneity was also significantly associated with higher rates of arrest for all violent offenses with the exception of homicide and simple assault. Family disruption (i.e., female-headed households) was significantly associated with higher rates of arrest for violent offenses other than homicide (Osgood and Chambers, 2000:103). Low economic status8 was found to be positively associated with rates of juvenile violence (Osgood and Chambers, 2000:87). Osgood and Chambers (2000) held some reservation regarding how the poverty hypothesis would actually behave because previous research indicates that neighborhood poverty works in combination with other indicators of 33 social disorganization, such as residential stability (Smith and Jarjoura, 1988; Shaw and McKay, 1969; Sampson, 1986).9 Results failed to support the low economic status hypothesis directly because populations of poorer nonmetropolitan communities may be more stable (Osgood and Chambers, 2000:106). This finding supports Shaw and McKay’s (1969) contention that poverty works in combination with other structural characteristics that influence crime. Overall, Osgood and Chambers found that variables commonly used in social disorganization research, such as residential instability, family disruption, and ethnic heterogeneity, were associated with higher rates of juvenile violence. Figure 3 shows their theoretical model. The italicized dimensions indicate changes to previous models of social disorganization theory. 34 Figure 3: Osgood and Chambers' Theoretical Social Disorganization Model Extended to Nonmetropolitan Settings Low economic status Residential fi instability Ethnic heterogeneity Family disruption Juvenile violence Population density' Nonmetropolitan counties T adjacent to metropolitan areas Osgood and Chambers’ (2000) study underscores the critical task. of testing existing theories in. different settings (Osgood. and. Chambers, 2000:108). However, one shortcoming is that Osgood and Chambers do not account for endogenous (or intervening) mechanisms of social organi zat ion . 1° To address this shortcoming , Sampson and colleagues Ihave closely' examined. social organization mechanisms, such as neighborhood collective efficacy, which 35 is hypothesized to reduce the effect of social disorganization on crime. The Salience of Collective Efficacy Neighborhood level research. is clearly' dominated. by studies that focus on structural predictors of social disorganization (e.g., poverty, residential stability, and ethnic heterogeneity) on crime. It is also necessary, however, to examine potential mediating11 effects of social mechanisms (Kornhauser, 1978:82; Cook et al., 1997:95-97; Taylor et al., 1984; Simcha-Fagan and Schwartz, 1986:695; Taub et al., 1980; Taylor et al., 1979; Greenburg et al., 1982). Here, the mediating social mechanism of interest is collective efficacy. The term mediating means that collective efficacy has the potential to reduce and/or reverse the effects of social disorganization on crime and quality of life. Sampson and colleagues have examined the emerging concept of collective efficacy and are heading the development of this body of research. The purpose of this section is to highlight the salience of collective efficacy, and discuss recent empirical research. 36 Collective Efficacy Self-efficacy serves as a starting point for understanding mechanisms of neighborhood social organization (e.g., collective efficacy). According to Bandura (1995:2) “self—efficacy theory addresses all of the sub-processes both at the individual and collective level.” At the collective level, efficacy has been operationalized as aggregated individual perceptions regarding a neighborhood's ability to produce positive outcomes for the common good (Bandura, 1995:33-38). Bandura's (1986, 1995) theory of self-efficacy provides a conceptual bridge between neighborhood structure and collective agency; In other words, structural characteristics (e.g., poverty) can disrupt neighborhood collective efficacy because it is embedded in a context that stratifies places of residence according to key social characteristics (Sampson et al., 1997:919). The theoretical premise of collective efficacy' is that community or neighborhood residents who actively engage, as a group, in the exercise of control (e.g., monitor) over behaviors in the neighborhood can reduce social problems (Sampson et al., 1997; Sampson et al., 1999; Sampson and Raudenbush, 1999). 37 Collective efficacy has been defined as task specific achievements, processes of active engagement, the exercise of control, and shared expectations among neighborhood residents to effectively maintain public order (Sampson et al., 1997; Sampson et al., 1999; Sampson and Raudenbush, 1999). Collective efficacy is comprised of two dimensions: social cohesion and informal social control. Social cohesion refers to neighborhood conditions whereby residents exhibit mutual trust and solidarity (Sampson et al., 1997:919), whereas informal social control is the general capacity of a group to regulate its members according to desired principles to realize collective goals (Sampson et al., 1997:918; Sampson and. Raudenbush, 1999:610). Sampson and Raudenbush (1999:611) add that social control should not be considered synonymous with repression or forced conformity; thus, “dimensions of social control are analytically separable not only from possible structural antecedents (e.g., poverty, instability) and effects (e.g., disorder, crime) but from the definition and operationalization of the units of analysis" (Sampson and Raudenbush, 1999:611). Collective efficacy places more weight on the notion of informal social control because it has the capacity to most effectively maintain public order 38 and control amid a wide range of social problems (e.g., violence, property crime, social and physical disorder). It is apparent that for collective efficacy to exist, social cohesion must precede informal social control. The level of effectiveness in informal social control depends on the level of social cohesion (Sampson and Raudenbush, 1999:612). In essence, collective efficacy begins with a certain degree of social capital. In other words, social capital has the capability to facilitate modes of action, such as collective efficacy. The social capital—collective efficacy connection will be discussed in Chapter 5. To summarize, collective efficacy is the linkage of mutual trust and the willingness to intervene for the common good (Sampson et a1 . , 1997 : 919; Sampson and Raudenbush, 1999:612) . Collective efficacy is seen in a variety of forms (e .g. , willingness to prevent and intervene in fights, voluntary community patrols) and is not limited to involving formal controls, such as the police. Nonetheless, the salience of collective efficacy lies with its ability to mediate the effect of social disorganization. 39 The Mediating Effect of Collective Efficacy Sampson et al. (1997) examined the effects of collective efficacy on violent victimization and homicide rates. This study came out of the Project on Human Development in Chicago Neighborhoods (PHDCN’).1‘2 Sampson et al. (1997:919; see also Sampson and Raudenbush, 1999:613) hypothesized that neighborhoods will also vary in their capacity for collective efficacy. Sampson et al. (1997) used three concepts of social disorganization (e.g., concentrated disadvantage, immigrant concentration, and residential stability). Results from their analysis showed that collective efficacy mediated the association between concentrated disadvantage and violence. Sampson.efi: al. (1997) also found that collective efficacy mediated the association between residential instability and crime. Overall, collective efficacy at the neighborhood level plays a vital role in the reduction of violence. Figure 4 shows Sampson et al.’s conceptualization of collective efficacy in relation to social disorganization, crime and victimization. 40 Figure 4: Sampson et al.'s Collective Efficacy Model ET I Concentrated disadvantage Crime Collective Residential H efficacy D and stability Victhization Immigrant concentration T l Similar to the research of Sampson et al. (1997), Sampson and Raudensbush (1999) also assessed the mediating effects of collective efficacy;13 Here, they shed new light on whether crime is a cause or consequence of observed disorder by specifying an alternative explanation to the “broken windows" thesis (Wilson and Kelling, 1982). Sampson and Raudenbush (1999:608) argue that disorder is not a direct cause of crime; instead, disorder is crime itself. They reason“ that disorder and crime are both the products of weakened social controls and structural antecedents (Sampson and Raudenbush, 1999:626). The structural constraints are considered exogenous variables, *while collective efficacy, disorder and crime are considered. endogenous variables (Sampson. and. Raudenbush, 1999:634). Collective efficacy is hypothesized to inhibit neighborhood disorder . 15 41 Results from their analysis showed that structural characteristics (particularly concentrated disadvantage and mixed land use) were strongly associated with physical and social disorder16 (Sampson. and. IRaudenbush, 1999:637). Collective efficacy predicted lower observed disorder after controlling not only for socio-demographic and land ‘use variables, but for perceived disorder and prior rates of predatory crime as well. Public disorder and predatory crimes were related in similar ways to disadvantage and collective efficacy. These findings led Sampson and Raudenbush (1999:637) to conclude that public disorder and most predatory' crimes share similar theoretical features and are explained by concentrated disadvantage and low collective efficacy. Figure 5 shows Sampson and Raudenbush's theoretical model with two additional social disorganization variables italicized. 42 Figure 5: Sampson and Raudenbush’s Theoretical Collective Efficacy-Crime and Disorder Model l Concentrated disadvantage Residential stability Crime Immigrant Collective concentration .__> efficacy ___, and Population Disorder density Mixed land-use l The empirical evidence clearly shows that collective efficacy plays a key role in reducing crime, disorder, and controlling'lbehavioru 1Placing' collective efficacy' under empirical scrutiny shows that it can mediate the effect of social disorganization. If so, then, neighborhood collective efficacy, perhaps, is likely to influence citizens' quality of life assessments. Despite the mounting evidence suggesting that collective efficacy matters, it has yet to be tested in nonmetropilitan settings. 43 Nonmetropolitan Social Organization To date, only a few studies have examined social disorganization :uu the nonmetropolitan. setting}7 It is customary' to assume that nonmetropolitan areas are characterized In! greater solidarity, social cohesion, security, and law abiding residents (Sorokin et al., 1930). However, there are mixed findings whether' mechanisms (of social organization in more rural areas are linked to social problems and quality of life. For example, some claim that close—knit, well integrated nonmetropolitan communities have the ability" to control social problems (Kowalski and Duffield, 1990; Freudenburg, 1986). Others argue, however, that the lack of social relationships in terms of distance from others may increase social problems (Wilkinson, 1984) and decrease quality of life (Saltiel et al., 1992; Bankston et al., 1987). There are two assumptions regarding the link between rurality and quality of life (Kowalski and Duffield, 1990; Saltiel et al., 1992). Saltiel et al. (1992:543) state these assumptions in relation to nonmetropolitan locations: One [assumption] is the view of rural areas as close-knit, well—integrated communities. This is the image of the small town where neighbors provide protection . The second [assumption] is more characteristic: of agriculture areas, where the way of life associated with farming is marked 44 by large distances and lesser reliance on others. The fragmented. social organization. increases sensitivity to risks of victimization. Therefore, nonmetropolitan areas may reflect wide variation in social organization that must be considered beyond traditional urban specifications. It seems that in coder to accurately examine the effects of social disorganization on crime and quality of life, we need to understand how social organizatirui operates in |.90|) that included the following 1990 census items: percent rural and population density. To assess the variance of ruralness across the :nonmetropolitan. continuum, several descriptive statistics were evaluated. Variance measures the average squared deviations of scores around the mean; however, the unit of measurement of the variable influences the variance and can be difficult to interpret. Ruralness had a minimum value of -2.05 and maximum value of .90, with a range of 2.95. The minimum 'value (-2.05) represents less rural areas, such as small town (St. Johns) and small city neighborhoods (Traverse City), while the maximum value 73 (.90) reflects more rural geographic units (Grand Traverse County). The range is an unstable measure of variance because it takes only two values into account and can be greatly changed by a single value (Miller and Whitehead, 1996:64). Thus, to establish that ruralness varied in an acceptable fashion, a histogram was assessed. Because ruralness was a weighted factor score, it had a mean of 0.00 and standard deviation of 1.00. More important is the distribution of ruralness across the 31 residential units. The values for ruralness (Skewness = -.72; Kurtosis = -.93) indicate a negatively skewed, leptokurtic distribution. Kurtosis measures the degree of peakedness. When closely examined, the distribution indicates two residential units occupying the value of -2.05 and one residential unit occupying the value of -1.75 in the extreme left end of the tail. Moving toward the center of the distribution, the next ruralness values of interest are -.50 and -.25. Here, the number of residential units remains consistent between 2 and 3, respectively. There are total of 14 residential units evenly distributed between —2.00 and -.25 (i.e., the smaller values) at the less rural end of the continuum. On the opposite end of the distribution, positive values indicate more rural areas. There is a sharp peak at value 74 .75 (leptokurtic: distribution) corresponding' *with. 10 residential units. The last value to the extreme right shows six residential units at the ruralness value of .90. Thus, there are total of 17 residential units at the most rural end of the distribution. To summarize, the geographic display of the data show an empirical pattern suggesting that as we move toward the right of the distribution, represented by larger ruralness values, the number of residential units sharply increases to reflect residential units that are larger in size and more sparsely populated (e.g., Grand Traverse County). This pattern suggests that enough variation exists to achieve the research objective at hand. The next section highlights citizen- and residential unit-level variables. Variables The following section has three objectives: (1) to operationalize citizen-level quality of life outcomes; (2) to operationalize residential unit-level predictors; and, (3) to operationalize citizen-level socio-demographic controls. 75 Ci tizen-Level Quality of Life Outcomes The quality of life variables28 consist of four terminal-dependent endogenous citizen-level measures: (1) perceived crime, (2) fear of crime, (3) perceived risk of victimization, and (4) perceived incivility. Perceived crime reflects an individual’s subjective judgment whether crime is decreasing or increasing (see Skogan, 1999). Perceived crime is a single survey item asking whether respondents believe that crime has been getting worse in their neighborhood over the past couple of years (1 = strongly disagree to 5 = strongly agree). Fear of crime is a four-item additive scale. Respondents were asked to indicate their level of fear according to the following: (1) being robbed by someone who has a gun or knife; (2) someone breaking into your house to steal things; (3) someone stealing your car; and (4) someone attacking you physically. Responses for each item ranged from 1 = least fearful to 10 = most fearful. The inter-item correlation ranged from .58 to .86 (Cronbach's alpha = .89). Perceived risk of victimization focuses on the concern for safety and the potential for harm. Risk levels are high when individuals’ feel that something could happen to them (Skogan, 1999). Risk of victimization is measured 76 using a four-item additive scale that asked individuals how safe they would feel being: (1) out alone in their neighborhood at night; (2) home alone at night; (3) out alone in their neighborhood during the day; and (4) home alone during the day. Responses for the four items ranged from 1 = very safe to 5 = very unsafe. The inter-item correlation ranged from .57 to .95 (Cronbach’s alpha = .91). Perceived incivility is measured as a five-item additive scale. Respondents were asked about specific problems in their neighborhood. It is important to note that nonmetropolitan settings are unlikely to experience problems with gangs when compared with urban areas (Ball, 2001) . To reduce measurement error and adjust for differences in nometropolitan settings, perceived incivility taps into less severe measures of social disorder (i.e., noisy neighbors, public drinking, and drug dealing) and physical decay (i.e., litter and run down buildings). Each of the incivility items originally featured a three-point scale (1 = no problem, 2 = a problem, and 3 = serious problem). The inter-item correlation ranged from .25 to .36 (Cronbach’s alpha = .67). 77 Overall, the four dependent variables were positively skewed. To normalize the distributions, the natural log was taken for each dependent variable. Descriptive statistics of these variables are presented in Table 3. The: next subsectirul operationalizes five independent residential unit-level variables. Table 3: Descriptive Statistics for Quality of Life Outcomes (N = 1,125) Variable Mean SD Min. Max. Perceived crime .57 .55 .00 1.61 Fear of crime 2.52 .72 1.39 3.69 Perceived incivility 1.70 .17 1.61 2.71 Risk of victimization 1.74 .40 1.39 3.00 Social Disorganization variables Given the literature reviewed thus far, three social disorganization factors consistently emerge: economic status, residential mobility, and population composition. Over the years, these dimensions have been modified. In many respects, such modifications correspond with the changing social and economic landscapes of contemporary urban neighborhoods. Economic status is no longer used and has been replaced by socioeconomic status (Sampson, 1986; Kornhauser, 1978), concentrated disadvantage (Sampson et 78 al., 1997), and economic disadvantage (Reisig and Holtfreter, 2000). Economic status, or some derivative thereof, has often been measured according to percent unemployed” jpublic .assistance, female-headed. families, poverty, and the like (see also Taylor, 1999:75). One dimension that had maintained its original form is residential mobility (e.g., residential stability [Sampson et al., 1997; 1999; Sampson and Groves, 1989]; residential instability [Peterson, Krivo, and Harris, 2000:37]; residential attrition [Krannich et al., 1989]). Residential mobility has commonly been measured as percent of persons living in the same house for less than five years and the percent of owner-occupied homes (Sampson et al., 1997:920; Taylor and Covington, 1993; Kornhauser, 1978). Recently, Peterson et al. (2000:37) operationalized residential instability' using’ residential mobility (percentage of residents five years of age and older living in different dwellings in the past five years), rental occupancy (percentage of occupied housing units that are renter-occupied), and vacancy rate (percentage of all housing units that are vacant). Population composition (to use Shaw and McKay’s terminology) is rarely used, and has been replaced by ethnic heterogeneity. From an operational standpoint, 79 ethnic heterogeneity is measured by race and ethnic (e.g., percent Hispanic and African-American) variation in a neighborhood (Taylor, 1999:75) . In a study conducted by Rountree et al. (1994:90), ethnic heterogeneity was operationalized as the product of the percentage of nonwhite residents (including black, Asian, Native American, and Hispanic) and white residents. Measured in this way, ethnic heterogeneity varied from complete homogeneity (racial and ethnic composition representing all white or all ethnic) to perfect heterogeneity (racial and ethnic composition representing half white and half nonwhite). Given the pooled sample characteristics in Table 2, the descriptive statistics for race show almost complete homogeneity (largely white populations). Accordingly, the current study excludes ethnic heterogeneity from the analysis. In the present research, social disorganization is represented by three variables: economic disadvantage, residential instability, and economic affluence. These variables are considered exogenous due to their direct and indirect effects on crime (Sampson and Groves, 1989; Simcha-Fagan and Schwartz, 1986:687; Sampson et al., 1997; Sampson and Raudenbush, 1999). In other words, these variables are the “triggering events” that set in motion 80 the loss (or increase) of quality of life. Several 1990 census items were included in the analysis. Table 4 provides the mean percentage and range of census items used in the analysis. Table 4: Mean Percentage and Range of 1990 Census Items Variable Mean Ran e N Lowest Highest % Poor 9.50 3.27 28.73 31 % Female-headed w/ children 9.23 1.22 27.80 31 % Unemployed 4.52 .00 9.46 31 % Rented occupied homesa 25.96 3.50 56.42 31 % Household income $75k+ 6.22 .69 31.69 31 % College education 17.63 3.27 50.48 31 % Professional/Managerial 17.17 2.66 33.79 31 work aPercent lived in residential unit less than 6 years is excluded because it is a survey item aggregated to the residential unit-level. The social disorganization. variables were extracted using principal components analysis with varimax rotation. Land at al. (1990; see also Welsh et al., 1999:88) recommend factor analyzing census measures to reduce multicollinearity and reduce the number of underlying dimensions. Varimax rotation maximizes the variance of the 81 pattern matrix and is orthogonal (i.e., the resulting factors are uncorrelated) (Kim and Mueller, 1978:55-58). Varimax rotation was chosen because it is commonly used and yields factors that are uncorrelated. Table 5 shows the loading pattern from the factor analysis. Two factors with eigenvalues greater than 1.0 were extracted. Factor 1 indicates a pattern consistent with opposite ends of the socioeconomic continuum. Recent scholarship argues for separating the upper tail of the socioeconomic distribution from the lower tail (Sampson et al., 1999:637). Given the loading pattern for Factor 1 and previous empirical support for separating the distribution, two variables were constructed: economic disadvantage (i.e., the lower-end of the socioeconomic distribution) and economic affluence (i.e., the upper-end of the socioeconomic distribution). 82 Table 5: Factor Pattern for Disorganization Variables (N = 31). Variables Factor 1 Factor 2 % Poor -.639 .246 % Female-headed -.586 .474 w/ children % Unemployed -.841 -.296 % Rented homes -.286 .735 % Less than 6 .136 .794 years in neighborhood % Household .798 -.162 income $75k+ % College .902 -.126 % Professional/ .924 -.006 Managerial work Note: Factor loadings greater than .50 are underlined. Economic disadvantage is measured as a weighted factor regression score (eigenvalue = 1.89; factor loadings = > .70) that included the following 1990 census items: percent poor, percent female-headed families with children, and percent labor force unemployed. Economic affluence is also measured as a weighted factor regression score (eigenvalue = 2.64; factor loadings = > .90) that included the following 1990 census items: percent households with income $75,000 and higher, percent college education bachelors and 83 higher, and percent adults in professional/managerial occupations. The second principal component, residential instability; is measured using one aggregated community survey item and one census item. The survey item asked respondents “How long have you lived in your current neighborhood?” “his item was aggregated to reflect the percentage of residents who had lived in the residential unit less than 6 years. The census item measured percentage of renter occupied homes. The weighted factor regression score was relatively healthy with an eigenvalue = 1.24 and factor loadings = > .70. Collective Efficacy Collective efficacy is a six-item additive scale aggregated to the residential-unit level. Collective efficacy is made up of two conceptually distinct, yet related variables: social cohesion and informal social control. Social cohesion gauged the level of mutual trust and solidarity among neighbors and is represented by a three—item scale. Residents were asked to indicate their level of agreement with the following statements: 1) this is a close-knit neighborhood, 2) people in the neighborhood can be trusted, and 3) people in the neighborhood generally 84 don’t get along with each other (reversed scored). Each item originally featured a five point Likert-type scale. Informal social control is also represented by a three-item! additive' scale. Informal social control measures the likelihood that neighbors would intervene for the common good in the following situations: 1) children spray painting graffiti on a local building, 2) children showing disrespect to an adult, and 3) a fight breaking out in front of their house. Responses were coded using a five point Likert-type scale. Responses to all six items *were aggregated. to the residential unit-level and summed to create collective efficacy. The inter-item correlations for collective efficacy ranged from .17 to .69 (Ckonbach’s alpha = .79). While collective efficacy has traditionally been used in urban research, it has yet to be tested in nonmetropolitan settings. Using it required minor adjustments. For example, collective efficacy in its original form asks residents about children. hanging out on street corners, budget cuts with respect to the local fire station, and the like (Sampson et al., 1997:919-920); but these scenarios are less likely to occur in nonmetropolitan areas. Thus, to reduce measurement error, some of the original social cohesion and informal social control items were not used. 85 Burglary Index While previous social disorganization research has modeled burglary as the terminal dependent variable (Smith and Jarjoura, 1989; Rountree et al., 1994; Lynch and Cantor, 1992), this research carries the analysis one step further by examining the influence of property crime on perceived quality of life. The crime of burglary is relevant for several reasons. First, property crime (e.g., burglary) is far more prevalent than person crime (e.g., homicide) in nonmetropolitan areas (Taylor and Shumaker, 1990:621; also see Osgood and Chambers, 2000). Second, operationalizing narrowly defined classes of crimes, such as burglary, increases the internal homogeneity of crime categories and thereby reduces measurement error (Lynch and Cantor, 1992:342; also see Bellair, 2000:146; Welsh et al., 1999:91). Lastly, Rountree et al. (1994:389) argue that burglary is particularly well-suited when conducting community-level studies because “measures of community context are relevant only to crimes at or near the home.” Burglary index is considered an endogenous variable and is a multisource measure of property crime across residential units. Burglary index is a weighted factor regression score measure that includes incidents of police— recorded burglary per 100 residents (burglary rate), 86 aggregated self-report survey victimization (percent victim of burglary in past 6 months), and aggregated perceptions of burglary (percent reported that burglary was a problem/serious problem) (eigenvalue = 1.74; factor loadings = > .70). Citizen-Level Socio-demographic variables Socio-demographic variables are included to control for survey response bias and spurious effects (Sampson et al., 1997:921; Sampson et al., 1999:640; Raudenbush. and Sampson, 1999:133). Because the sample is racially homogenous (i.e., largely white), race is measured using a single dummy variable: minority (1 = minority, 0 Caucasian). Age (respondent’s age in years) and male (1 = male, 0 = female) are included because research suggests that elderly citizens and females are more likely to experience higher levels of fear (LaGrange and Ferraro, 1989; Lawton and Yaffe, 1980; Ortega and Myles, 1987; Clarke, Ekblom, Hough, and. Mayhew, 1985), while younger persons are less fearful (Stephens, 1999:62; Garofalo and Laub, 1978; DuBow et al., 1979). Married (1 = married, 0 = otherwise) is included because single individuals have higher rates of victimization due to lack of guardianship (Smith. and. Jarjoura, 1989:621; Lynch. and. Cantor, 1992). 87 Socio-economic status is a weighted factor score that includes education, family income, and occupational prestige (eigenvalue = 1.82; factor loadings = > .70). The last two controls are homeowner (1 = homeowner, 0 = otherwise) and years in neighborhood (# of years lived in current neighborhood). Descriptive statistics for citizen- and residential unit-level variables used in this analysis are shown in Table 6. 88 Table 6: Descriptive Statistics for Independent Variables‘11 (N = 1,125) Variable Mean SD Min. Max. Independent Variables Citizen-level predictors Minority .03 .17 .00 1.00 Male .63 .48 .00 1.00 SES .00 1.00 -2.68 2.63 Age 54.19 15.30 19.00 98.00 Married .69 .46 .00 1.00 Ownhome .89 .31 .00 1.00 Years in neighborhood 14.73 13.25 1.00 84.00 Residential unit-level predictorsb Economic disadvantage .00 1.00 -1.37 3.44 Residential instability .00 1.00 -l.74 2.25 Economic affluence .00 1.00 -1.30 3.37 Collective efficacy 21.55 .84 19.54 22.88 Burglary index .00 1.00 -1.34 2.80 aTotal sample size is 1,125 citizens and. 31 residential units. bDescriptive statistics for residential units are based on residential unit as the unit of analysis. The next section highlights the analytic strategy used for this research. It provides a review of the advantages 89 of hierarchical linear modeling (HLM) over other multivariate techniques, and outlines the four-step modeling procedure used to conduct the analyses. Analytic Strategy Rationale Hierarchical linear modeling (HLM) has many advantages over traditional multivariate techniques. For example, ordinary least-squares regression would be limited with these data because it would ignore the aggregate clustering of individuals. In contrast, HLM takes advantage of these natural clusters by simultaneously' modeling citizen- and aggregate—level models (Bryk and Raudenbush, 1992:xiv), thus decreasing the probability of committing‘ a Type I error (Kreft and. DeLeeuw, 1998:10). The rationale for using HLM is that it serves as the most effective way to model citizen—level outcomes using both individual- and residential unit-level independent variables. To perform a reliable multilevel analysis, 10 cases within 10 aggregates must be present (Mok and Flynn, 1998:413). For this research, the average number of respondents within each residential unit is 36.29. In short, it appears as though HLM can be used here to regress quality' of life outcomes on individual- and residential 9O unit-level variables. HLM resolves many of the problems that multivariate and other multi-level models encounterfi29 yet is still based on traditional OLS regression assumptions of independence, normality, and variance (Bryk and Raudenbush, 1992:15). Analysis Procedure A four-stage modeling procedure will be used for each quality of life outcome. The first model, the One—Way ANOVA" provides descriptive statistics detailing the appropriateness of the data for HLM techniques, such as reliability estimates. The random coefficient model, which is the second step in the process, provides a first look at the effects of citizen-level predictors on the outcome measure. This model also indicates whether any of the citizen—level slopes vary across aggregate units. Where such variation is found, it can be modeled as a function of residential-unit characteristics (Rountree an: al., 1994:396). The third step (or the means-as-outcomes model) helps determine whether residential-unit variables influence quality of life outcomes. Finally, the citizen and residential unit-level models are combined into a single hierarchical model. 91 Conclusion This chapter' has provided several hypotheses to Ibe tested with respect to social disorganization and quality of life in nonmetropolitan settings. The chapter also outlined how social disorganization, perceived quality of life, and other variables were operationalized. The chapter highlighted how the data were gathered, how residential units were constructed, and why HLM is appropriate when using data with a nested structure. The next chapter discusses the findings. 92 CHAPTER FOUR: F INDINGS In this chapter, the major research findings are presented. Before doing so, however, several diagnostic procedures were performed to help increase confidence in the findings. For example, bivariate correlations and OLS regression models helped address issues concerning multi- collinearity and discriminant validity. Along the way, some initial hypotheses testing were conducted. After doing so, the analysis proceeds with the estimation of a series of hierarchical linear models. Preliminary Statistics Model Diagnostic Procedures Citizen-Level Associations. Zero-order correlations between the citizen-level independent variables were explored to help determine whether multi-collinearity was present. Table 7 presents bivariate correlations between citizen-level variables used in the analyses. The results indicated that multi-collinearity was not a problem because the Pearson correlation values did not exceed .80 (Menard, 1995:66) . In fact, the highest correlation was observed between years in neighborhood and age (.46) , while the weakest correlation was between age and male (- . 01) . The 93 mean inter-correlation for all of the variables presented in Table 7 was .24. Table 7: Zero-order' Correlation. Coefficients between Citizen-Level Variables Measure (2) (3) (4) (5) (6) (7) (1) Minority -.04 -.14* 13* -.03 -.05 .06 (2) Male —-- 20* - 01 41* .10* .05 (3) sssa --- --- -.20* 26* 20* -.21* (4) Age --- --- --- -.08* 14* .46* (5) Married -—- --- --- -—- .27* .02 (6) Ownhome --- —-- --- -—- --- .16* (7) Years in Neigh. --- --- --- -—- --- --- aWeighted factor score *p < .01 (two tailed tests) Because relying solely on bivariate analysis to investigate the existence of multi-collinearity is viewed as unsatisfactory (Berry and Feldman, 1985:43), OLS regression diagnostics (not shown) provided additional support that multi-collinearity was not a problem. When perceived crime was regressed on citizen-level variables, the tolerance statistics for each independent variable was greater than .70. According to Menard (1995:66) a 94 tolerance statistic less than .20 is cause for concern. In sum, the tolerance statistics revealed no evidence of multi-collinearity among the citizen-level variables used in the multivariate models. Residential Unit-Level Associations. Bivariate results from the independent residential unit-level variables showed no signs of multi-collinearity. The mean inter—item correlation between all of the residential unit— level variables was .32 (see Table 9). However, as expected, an inverse association was observed between economic affluence and economic disadvantage (-.68). It is important to note that concerns regarding multi- collinearity are not a problem given that the correlation between. economic affluence and. economic disadvantage: did not exceed .80. Based on Menard’s (1995) tolerance guidelines, no evidence of mmlti-collinearity was detected when perceived crime was regressed on residential unit-level variables. However, the tolerance statistic revealed that economic disadvantage (.52) and economic affluence (.48) might be of some concern. Despite the comparatively low tolerance statistics, it was decided to keep these variables in their original form because it is theoretically salient to separate the upper—tail of the socioeconomic distribution 95 from the lower-tail (Sampson et al . , 1999 : 637) . Overall , while the multivariate model showed a modest degree of multi-collinearity, there was no evidence suggesting a problematic model specification. Discriminant Validity of Outcome Measures Citizen-Level. A second set of zero—order correlations was estimated to determine the magnitude of the association between the quality of life outcome measures at the citizen-level. These bivariate results (not shown) helped determine whether the quality of life variables used in the analyses possessed discriminant validity. Meeting the criterion of discriminant validity shows that “your measure of a concept is different from measures of similar but distinct concepts” (Maxfield and Babbie, 1998:110). In other words, how do we know that perceived crime is empirically distinct from risk of victimization? The weakest correlation was observed between fear of crime and perceived incivility (.16), while the strongest correlation was between perceived incivility and perceived crime (.40). Although the correlations were statistically significant, the magnitudes did not exceed .80. Accordingly, it is safe to conclude that the quality 96 of life measures at the citizen-level possessed discriminant validity. Residential Unit-Level. Prior ecological research has emphasized the need to establish discriminant validity at the aggregate level (Sampson et al., 1997:922-923; Sampson et al., 1999:642—643; Cook et al., 1997:97). The bivariate results (not shown) for the outcome measures at the residential-unit level revealed some concern regarding the correlation between perceived crime and risk of victimization (.75) and risk of victimization and fear of crime (.71). Despite the high values, the Pearson correlations were less than .80. Therefore, results support the discriminant validity of the scales. Fear of crime and. perceived incivility (.18) showed the ‘weakest correlation. Correlates of Quality of Life at the Residential unit-Level Table 8 presents zero-order correlations between the residential unit-level and outcome variables. Collective efficacy was inversely related to three of the four quality of life outcomes. Hence, three hypotheses were supported. For' example, as collective efficacy' decreases, perceived crime significantly increases (H4a). A similar relationship was observed between collective efficacy and 97 perceived incivility (H4c). 1n: addition, the relationship between collective efficacy and risk of victimization was significant and in the expected direction (H4d). The only observed relationship that failed to reach statistical significance was between collective efficacy and fear of crime (H4b); nevertheless, the sign was in the expected direction. Overall, strong to moderate negative relationships were observed between collective efficacy and perceived crime (- . 63) , perceived incivility (- . 62) , and risk of victimization (- . 51) ; yet, a weak negative relationship was observed for fear of crime (- .27) . These results suggest that citizens living in residential units characterized by low levels of collective efficacy are more likely to rate their quality of life in more negative terms . 98 Table 8: Zero-Order' Correlation. Coefficients. between Quality of life Measures and Residential Unit-Level Measures Perceived Fear of Perceived Risk of Measure crime crime incivility victimization Economic disadvantage“ .16 .26 .23 .29 Residential instability“ .05 .11 .07 .16 Economic affluence“ -.29 -.09 -.30 -.23 Collective efficacy -.63** —.27 -.62** -.51** Burglary index“ .79** .38* .71** .64** “Weighted factor score *p < .05, **p < .01 (two-tailed tests) Results also indicated positive and significant correlations between. burglary' index. and. quality’ of life outcomes. The relationship between burglary index and perceived crime (.79; HSa), perceived incivility (.71; H5c), and risk of victimization (.64; H5d) are strong and in the hypothesized direction. Fear of crime is correlated less strongly with burglary index (.38; HSb), yet remains statistically significant. These findings suggest that citizens living in residential units with higher rates of 99 burglary are more likely to perceive that crime, incivility, risk, and fear are problematic. The weakest relationships were observed between social disorganization indicators and quality of life outcomes. In other words, the results do not support the hypotheses regarding the link between economic disadvantage (H6a through H6d), residential instability (H7a through H7d), economic affluence (H8a through H8d) and quality of life outcomes. Although these coefficients failed to reach statistical significance, the relationships were in the hypothesized direction. Focus here is on the magnitude of the correlations as opposed to statistical significance because the latter depends on sample size. Nevertheless, it is worth repeating that both collective efficacy and burglary index significantly influenced the quality of life outcomes. Bivariate Associations Between Residential Unit Independent Variables Table 9 examines the zero-order correlations between residential unit-level variables included in the analysis. These initial tests indicated that collective efficacy and burglary index were correlated (-.57). Among the zero- order correlations, the association between collective efficacy and burglary index was the only hypothesis (H3a) 100 posited to reach statistical significance. This finding reaffirms pervious research showing that collective efficacy and crime are inversely related (Sampson et al., 1997; Sampson et al., 1999). Table 9: Zero-Order' Correlation. Coefficients Ibetween Residential Unit-Level Variables Measure (2) (3) (4) (5) (1) Economic disadvantage“ .22 -.68* -.15 .14 (2) Residential instability“ --- -.22 -.12 .24 (3) Economic affluence“ --- --- .32 -.17 (4) Collective efficacy --- --- --- -.57* (5) Burglary index“ --- --- --- --- “Weighted factor score * p < .01 (two—tailed tests) The correlation between economic affluence and collective efficacy (.32), although weak and insignificant, was in the hypothesized direction (H2c). The directional accuracy of this finding is consistent with prior research that suggests economically affluent residential areas are more likely to experience higher levels of collective efficacy (Sampson et al., 1999). In addition, the 101 hypotheses that economic disadvantage (H2a) and residential instability (H2b) would be inversely related to collective efficacy were in the expected direction; however, neither coefficient achieved statistical significance. The relationships between structural variables, such as economic disadvantage (Hla), residential instability (Hlb), and economic affluence (ch) and burglary index were in the hypothesized direction, but failed to reach statistical significance. Hierarchical Linear Models Using a four-step HLM procedure (Bryk and Raudenbush, 1992), the analyses began with preliminary ANOVA models. The One-way ANOVA models helped determine the amount of variation in the outcomes within and between residential units, as well as provide reliability estimates30 for outcome measures at the aggregate-level. Next, random coefficient models were estimated. The random coefficient models examined citizen-level predictors on the outcome measure. These models indicated whether any of the citizen-level slopes varied across aggregate units. Where such variation was found, it was modeled as a function of residential-unit characteristics (Rountree: et ral., 1994). Third, intercept-as—outcome models, which include the 102 residential unit-level 'variables, were estimated. These models helped determine whether residential-unit variables influenced quality of life outcomes. The final step entailed estimating HLM models where the outcome measures were regressed on both citizen- and residential unit-level variables simultaneously. One-way Analysis of Variance (ANOVA) Mbdels Table 10 presents HLM decomposition of variance components and residential unit-level reliabilities for quality of life outcomes. The residential unit reliability for perceived crime (.73) and perceived incivility (.76) are quite high. This means that parameter variance was reliably captured at the residential unit-level. Put differently, high estimates (>.70) indicate that residential unit differences can be modeled with a high degree of precision. 103 Table 10: Decomposition of Variance and Residential Unit- Level Reliabilities of Quality of Life Measures Variance Perceived Fear Perceived Risk of Components crime of crime incivility victimization Within-RU variance(o“) .28 .50 .03 .15 Between-RU variance(nm) .02 .01 .00 .00 Intraclass correlation .07 .03 .09 .01 RU reliability .73 .50 .76 .33 Note: N = 1,125 citizens nested in 31 residential units. The residential unit reliability for fear of crime is lower at .50. Although the ability to detect residential unit differences is somewhat thwarted, the value remains within the bounds of acceptability (see Duncan and Raudenbush, 1999). Previous research (Sampson and Jeglum- Bartusch, 1998:796) has considered. similar 'values (e.g., .54) acceptable. The reliability estimate for risk. of victimization is low at .33, indicating that it may be difficult to model. Table 10 also presents intraclass correlation (ICC) values.”' The ICC (p==tm,/ o“+-tmfl indicates the percentage of the scale's variance between residential units, with the remainder apportioned to random error and citizen-level 104 variation.32 In general, the intraclass correlations for the four scales ranged between 1% and 9%. An ICC of 1% should be met with caution. Duncan and Raudenbush (1999:10) advise caution when interpreting small ICC, as effect sizes commonly viewed as large translates into small proportions of variance in individual outcomes. However, the ICC for perceived incivility was .09. This means, for example, that approximately 9% of the variation in perceived inc ivility was between residential units . Results indicate that sufficient variation exists in the outcomes to estimate residential unit—level models (see Duncan and Raudenbush, 1999). Overall, the primary results support proceeding with more complex modeling, but caution should be exercised when interpreting the results for the risk of victimization models. Random Coefficient Mbdels Prior to modeling residential unit-level effects, random-coefficient models were estimated. Justification for doing so was twofold. First, the models examine the association between citizen-level variables and the outcomes in a multivariate context. Second, the models help determine whether any of the citizen-level slopes vary significantly across residential units. If citizen-level 105 slopes vary across residential units, then aggregate variation can be modeled using contextual variables (Rountree et al., 1994). All citizen-level variables were group-mean centered in the following analyses. Table 11 addressed whether citizen-level variables were related to the outcomes. Table 11 presents four quality of life random coefficient models. Minority and ownhome were set as fixed due to a lack of variation across residential units. The coefficients for perceived crime showed that, on average, males and old residents were more likely to report problems associated with crime. Prior research indicates that elderly are more likely to perceive crime as problematic (Garofalo and Laub, 1978), and males are less likely to perceive crime as a problem (LaGrange and Ferraro, 1989; Lawton and Yaffe, 1980). Ownhome showed a significant inverse relationship *with.jperceived. crime. In other words, citizens who do not own their home are more likely to perceive problems concerning crime. The citizen- level variables explained 5% of the within RU variance in the perceived crime model. 106 Amumou ooafimuro3uv Ho. v as. .mo. v a. .uouonucousm no uouuo unaccoum .oucowufiuuoou 107 consuucucsums: one moucEwunm .oDm nachos scausaue> mo xosd on one uOqu as you ueonQBO use xuwuocwz "ouoz am so as am om-oanoaz oochHme oucmaum> loo.v Loo.c noo.v noo.v mo. oo. om.- oo.- Hm. oo. oo.H oo. onmz GH .mHW Aao.c Amo.c iso.c Aso.c ao.a- 50.- Hm.H- mo.- mo.H so. «oo.m- mH.- osonc3o Amo.v xHo.c Amo.v Aao.v oo. mo. oH.H- Ho.- mo.- mo.- om.o- Ho. ooauumz loo.v loo.v Loo.c loo.v so.- oo.- .mm.m- oo.- «ism.mn Ho.- t.mm.o- Ho. oma xHo.c xHo.v Amo.v Aao.c ttoo.m- oo.- am. oo. mo.- oo.- HH.- oo.- mmm Amo.v lHo.c loo.v Amo.v tram.m- NH.- Ho.H- Ho.- tioo.v- am.- «¢H.m 0H. mam: loo.v Amo.c AHH.V Aso.c mo.- oo.- mm.a oo. mm.H as. am. oo. soauoca: “Ho.v Aso.v Amo.v Amo.c tram.mma as.H 41mm.sos Hb.H ttmm.mm Hm.m igbs.oa om. odoouoooH Oflumuuu Q OfiucHuu Q Osbourn Q Ofiumnuu Q GOHDMNAEHDOH> >uflafl>wocfl mEHHO oEflHO mo xmflm um>aooumm mo chm ou>woouom AmNH.H u zc maooo: mesa no soaamso womanhooooo soocmm “as manna The fear of crime model revealed that male and age were inversely and significantly related to fear of crime. These findings are inconsistent with previous research that suggests elderly and females are more likely to experience higher levels of fear (LaGrange and Ferraro, 1989; Lawton and Yaffe, 1980; Ortega and Myles, 1987; Clarke, Ekblom, Hough, and Mayhew, 1985), while younger persons are less fearful (Stephens, 1999:62; Garofalo and Laub, 1978; DuBow et al., 1979). The citizen-level variables accounted for 7% of the within RU variance in the fear of crime model. Perceived incivility was inversely related to age. More specifically, younger citizens are likely to perceive various forms of social disorder and. physical decay' as problematic in their immediate surroundings. This finding is consistent 'with. Ball (2001) who found. that in. rural Maine, youth perceived disorder and decay more of a problem than adults. The model explained 8% of the within RU variance associated with perceived incivility. In the final model, male and SES were inversely and significantly related to risk of victimization. These findings suggest females perceive higher levels of risk when compared to men. In addition, citizens with lower SES are more likely to report higher levels of risk. One explanation for this relationship is that monetary 108 constraints prohibit citizens from. installing locks, alarms, or other security devices that help provide a sense of security and safety (Taylor and Schumaker, 1990). Table 12 addressed whether any of the citizen-level slopes varied significantly with quality of life outcomes across residential units. The models for perceived crime, fear of crime, and risk of victimization revealed that none of the citizen-level variables varied across residential units. According to Walsh et al. (1999:98), this means that the relationships between citizen—level variables and quality of life outcomes were similar across the 31 residential units. However, the perceived incivility model revealed that age and years in neighborhood varied significantly. Because variation was found to exist, age and years in neighborhood were modeled as a function of residential-unit variables. To examine this variation, two multi-level interaction models were estimated. 109 0H. mm.ov mo. mm.m¢ no. mm.m¢ om. mm.mm DQmZ .mHN ma. mm.mm mm. mm.mm om.A m¢.¢N HH. oo.o¢ Umflhhmz om.A mH.mm oo. oa.mo om. om.om Hm. oo.mm 0mm om.A mm.ma om.A hm.mm mm. mm.Hm om.A mo.¢m mmm we. ¢H.om om.A om.mm ma. mo.bm mm. mm.Nm OHME moam>nm ax mdam>im. «x moam>um «x wch>um ax GOADmNHEflu0fl> >uaaa>a0dw mEfiHo oEfluO mo xmflm um>fluoumm mo scum uo>fiooumm AmNH.H u zv maouoz sued mo quHMSO mo moccflum> "ma oHQMB 110 In the first interaction model, the slope for age was modeled as a function of economic disadvantage, residential instability, and collective efficacy (results not shown). Here, the effect of age persisted. Additionally, a cross- level association between age and residential instability (t-ratio = -3.68) and collective efficacy (t-ratio = -3.31) was revealed. These findings suggest that older citizens reside in more stable residential units with lower levels of collective efficacy. In the second interaction model, the slope for years in neighborhood was modeled as a function of economic disadvantage, residential instability, and collective efficacy. Results (not shown) revealed no multi-level interactions. Intercept-As-Outcome Mbdels Having estimated the regression equations for citizen- level variables to explore possible variation across residential units, a series of intercept—as-outcome models were examined. for' purposes of hypotheses testing; 'The intercept-as-outcome models helped determine whether residential unit-level variables influenced quality of life outcomes. 131 these models, however, citizen—level variables were not included, so these models represent lenient tests. Here, intercepts for quality of life 111 variables were modeled as a function of residential-unit contextual characteristics. The residential unit-level variables were centered around the grand mean. The residential-unit models were specified with random error terms. The error term represents the variability that remains after the residential unit-level variables have been entered into the model. The following tables present two models for the four quality of life variables. Two separate models were estimated due to the sample size (N=31) at the residential unit-level. A small sample restricts statistical power. The common rule of thumb is that one needs at least 10 observations for each predictor at the aggregate-level (Byrk and Raudenbush, 1992:211). Despite these limitations, Bryk and Raudenbush (1992:198) claim “implausible results arising from units with small sample size are not a problem because the estimation methods are robust.” Model 1 in Tables 13 through 20 included economic disadvantage, residential instability, and collective efficacy. Model 2 also included collective efficacy; however, economic disadvantage and residential instability were replaced with economic affluence and crime. Economic disadvantage and economic affluence are separated in each 112 model because of collinearity concerns. These two models are assessed throughout the remainder of the analyses. Table 13 presents results for perceived crime. In Model 1, the hypothesis that economic disadvantage would have a direct positive effect on perceived crime (H6a) was not supported. Ihi other words, economic disadvantage did not appear to influence perceived crime. This finding was inconsistent with prior urban research that revealed aggregate-level economic indicators, such as economic disadvantage, significantly affects perceptions of crime (Aneshenel and Sucoff, 1996; Campbell, 1981; Robert, 1998; MacIntyre, MacIver, and Sooman, 1993; Sooman and MacIntyre, 1995). Moreover, this finding does not support Skogan’s (1990:75) argument that neighborhoods characterized by economic deprivation foster subjective responses that crime is prevalent, imminent, and thereby leads to further notions of personal vulnerability (Skogan, 1990:75). It is important to note that Skogan’s, and much of the prior research pertains to urban neighborhoods. Therefore, many of the inconsistent findings are attributed to the research settings (i.e., urban versus nonmetropolitan settings). The results also revealed that residential instability was not positively associated with perceived crime (H7a). However, the observed relationship between collective 113 efficacy and perceived crime (H4a) was inverse in nature and statistically significant. Stated differently, residential units with low levels of collective efficacy are more likely to be inhabited by residents who perceive crime as problematic. Table 13: Intercept-as—Outcome Models for Perceived Crime Model 1 Model 2 Coefficient Coefficient Variable (SE) t-ratio (SE) t-ratio Intercept .58 24.38** .58 33.14** (.02) (.02) Economic disadvantage .02 .98 -- -- (.02) Residential instability -.00 -.25 -- -- (.03) Collective efficacy —.13 -5.22** -.04 -1.85* (.02) (.02) Economic affluence -- -- -.02 -1.42 (.02) Burglary index -- —— .11 5.80** (.02) *p < .10, **p < .05 (two-tailed tests) 114 In Model 2, the relationship between collective efficacy and perceived crime (H4a) was significant and in the hypothesized direction. The results failed to support the hypothesis that economic affluence would have a direct negative effect on perceived crime (H8a) because the coefficient, although in the expected direction, did not achieve statistical significance. Lastly, the hypothesis that crime would have a direct positive effect on perceived crime (H5a) was supported. In other words, as crime increased at the residential-unit level, perceived crime also increased. By comparing Models 1 and 2 above, the evidence suggested that the inclusion of burglary index resulted in a diminished association between collective efficacy and perceived crime (from -.13 to -.04). Table 14 presents results for fear of crime. In Model 1, the results failed to confirm the hypotheses that economic disadvantage (H6b) and residential instability (H7b) would have a direct positive effect on fear of crime. Hence, the two structural predictors in Model 1 had no bearing on fear of crime. Collective efficacy, on the other hand, was significant and inversely related to fear of crime (H4b) . In other words, fear of crime was higher among citizens living in residential units with low levels of collective efficacy. This finding is consistent with 115 the social control model, which posits that a breakdown in the development and enforcement of local norms for social behavior is the major determinant of fear (Greenburg et al., 1985; Lewis and Salem, 1981; Podolefsky and DuBow, 1980). Table 14: Intercept-as-Outcome Models for Fear of Crime Model 1 Model 2 Coefficient Coefficient Variable (SE) t-ratio (SE) t—ratio Intercept 2.51 82.70*** 2.51 91.81*** (.03) (.03) Economic disadvantage .03 .78 -- -- (.04) Residential instability .00 .01 -- -— (.03) Collective efficacy -.05 -1.63* -.01 -.31 (.03) (.04) Economic affluence -- -- .00 .04 (.03) Burglary index -— —- .06 3.14** (.02) *p < .10, **p < .05, ***p < .01 (two-tailed tests) 116 In. Model 2, the observed relationship between collective efficacy and fear of crime (H4b) was in the expected direction, but not statistically significant. The diminished influence of collective efficacy appears to be a result of the inclusion of burglary indexi Here, the hypothesized relationship between burglary index and fear of crime (H5b) was in the expected direction. Lewis and Salem (1981) argue that actual crime rates are the basic cause of fear and other emotional reactions to crime. The hypothesis that economic affluence would have a direct negative effect on fear of crime (H8b) was not supported. Table 15 presents the results for perceived incivility. In Model 1, the findings failed to support the hypotheses that economic disadvantage would have a direct positive effect on perceived incivility (H6c). This finding is inconsistent with Sampson and Raudenbush (1999:637) who found that concentrated disadvantage was positive and significantly associated with physical and social disorder . These inconsistencies reflect different research settings (i.e., urban versus nonmetropolitan settings). The hypothesis that residential instability would have a direct positive effect on perceived incivility (H7c) was not supported. 117 While these two structural hypotheses were not supported, the hypothesized relationship between collective efficacy and perceived incivility (H4c) was supported in Model 1. Stated differently, citizens living in residential units with lower levels collective efficacy perceived higher levels of incivility. This is consistent with existing research conducted in the urban setting (Sampson and Raudenbush, 1999). 118 Table 15: Intercept-as-Outcome Models for Perceived Incivility Model 1 Model 2 Coefficient Coefficient Variable (SE) t-ratio (SE) t-ratio Intercept 1.71 216.78* 1.71 283.85* (.01) (.01) Economic disadvantage .01 1.07 -- -- (.01) Residential instability -.00 -.17 —- -- (.01) Collective efficacy -.04 -4.43* -.02 -1.49 (.01) (.01) Economic affluence —- -- -.01 -1.41 (.01) Burglary index -- -- .03 3.94* (.01) *p < .01 (two-tailed tests) While collective efficacy was significant in Model 1, it failed to reach statistical significance in. Model 2 (i.e., p < .10). The hypothesized relationship between economic affluence and perceived incivility (H8c) was not supported. Results did, however, confirm the hypothesis that burglary index would have a direct positive effect on 119 perceived incivility (H5c). This finding supports the crime-incivility connection (Hunter, 1978; Wilson and Kelling, 1982; Taylor, 1996). Overall, both models suggest that perceived incivilities are the result of crime and, to a lesser extent, lower levels of collective efficacy. Table 16 presents the intercept-as-outcome results for risk of victimization. In Model 1, the hypothesis that economic disadvantage would have direct positive effect on risk of victimization (H6d) was not supported. The hypothesized association between residential instability and risk of victimization (H7d) was also not supported. Collective efficacy' and. risk. of 'victimization. ‘were inversely and significantly related (H4d). In. essence, citizens living in residential units with lower levels of collective efficacy reported higher assessments of risk. 120 Table 16: Intercept-as-Outcome Models for Risk of Victimization Model 1 Model 2 Coefficient Coefficient Variable (SE) t-ratio (SE) t-ratio Intercept 1.74 151.93* 1.74 160.82* (.01) (.01) Economic disadvantage .02 1.50 -— -- (.01) Residential instability .00 .10 -- -- (.02) Collective efficacy -.05 -4.11* -.02 -1.65 (.01) (.01) Economic affluence -- -- -.00 -.53 (.01) Burglary index -- -- .04 4.05* (.01) *p < .01 (two-tailed tests) In Model 2, the association between collective efficacy and risk of victimization was not statistically significant. While in the expected direction, the hypothesized relationship between economic affluence and risk of victimization (H8d) was not confirmed. However, the hypothesized. relationship ibetween. burglary' index; and 121 risk of victimization (H5d) was statistically significant. In other words, as rates of burglary increase, risk of victimization increase. Overall, the residential unit-level results reveal several important findings. First, structural predictors appear not to influence citizen quality of life assessments in nonmetropolitan settings. Second, residential units with lower levels of collective efficacy are likely to be inhabited by citizens who report lower levels of quality life. Third, when economic affluence and crime are added to the model (i.e., Model 2), the influence of collective efficacy is diminished. In short, Model 2 suggests that crime not only attenuates collective efficacy, but also significantly influences perceived crime, fear of crime, perceived incivility, and risk of victimization. Fixed Effects Hierarchical Models (Full Models) The main question addressed in this study is, once individual correlates are controlled, what is the effect of social disorganization and collective efficacy on citizen- level quality of life assessments? To address this question, two full hierarchical models were estimated for each quality of life outcome (8 models total). As 122 previously mentioned, two models were estimated due to sample size (N=31) (Bryk and Raudenbush, 1992:211). Table 17 pmesents fixed effects hierarchal models for perceived crime. In Model 1, the Chi-square statistic (x2 = 70.78, p < .01) indicated that citizen perceptions of crime varied across residential units. controlling for mdnority and four other individual-level predictors, respondents who were male, young, and renters reported higher levels of perceived crime. At the residential unit-level, the results showed that citizens residing in residential units characterized by higher levels of collective efficacy reported significantly lower levels of perceived crime (H4a). This result suggested a contextual effect of residential unit collective efficacy' on. perceived. crime. The findings from Model 1 also indicated that the hypothesized relationships between economic disadvantage (H6a), residential instability' (H7a) and. perceived. crime were not supported. Model 1 accounted for 48% of the explained variance between residential units, while the sociodemographic variables explained 3% of the variation at the citizen-level. 123 Table 17: Fixed Effects Hierarchical Models for Perceived Crime Model 1 Model 2 b t—ratio t-ratio Variables (SE) (SE) Intercept .58 24.38*** .58 33.15*** (.02) (.02) Citizen-Level (N=1,125) Minority .06 .80 .06 .80 (.07) (.07) [.02] [.02] Male .09 2.68** .09 2.68** (.03) (.03) [.08] [.08] SES .00 -.23 .00 -.23 (.01) (.01) {-.01] {-.01] Age -.01 -4 24*** -.01 -4.24*** (.00) (.00) [-.14] {-.14] Married -.01 -.13 -.01 -.13 (.04) (.04) [.00] [.00] Ownhome -.14 -2.04** -.14 -2.04** (.07) (.07) [-.08] {-.08] Years in neighborhood .00 1.13 .00 1.13 (.00) (.00) [.05] [.05] Residential Unit-Level (N=31) Economic disadvantage .02 .97 -- -- (.02) [.03] Residential instability -.01 —.25 -- ~— (.03) {-.01} Collective efficacy -.13 -5.22*** -.04 -1.85**t (.02) (.02) {-.63] {-.20] 124 Table 17: Continued Economic affluence -- -- -.02 -1.42 (.01) {—.12] Burglary index -— -— .11 5.80*** (.02) [.64] x“ 70.78*** 37.27* Variance Explained (Percentages) Within-residential unit 3 3 Between-residential unit 48 86 Note: Standardized coefficient in brackets *p < .10, **p < .05, ***p < .01 (two-tailed tests) tOne-tailed test In Model 2, the Chi-square statistic (12 = 37.27, p < .10) also indicated that perceived crime differed across residential units. Citizen—level correlates behaved in a manner similar to Model 1. What is more, collective efficacy was found to be inversely and significantly related to perceived crime (H4a) . When compared to Model 1, however, the magnitude of collective efficacy was considerably weaker (from -.63 to -.20). So, what variable is responsible for such reduction in magnitude? The answer, burglary index. .As hypothesized, a: strong relationship was observed between burglary index and perceived crime (H5a). 125 In fact, after controlling for collective efficacy, economic affluence, and a multitude of citizen-level correlates, burglary index was the strongest determinant of perceived crime. It is important to note that in both models economic disadvantage (H6a) and economic affluence (H8a) are in the hypothesized direction, but not statistically significant. The results from Model 2 revealed that by including burglary index and economic affluence, and excluding other structural predictors (e.g., economic disadvantage and residential instability), the amount of explained between-residential unit variation nearly doubled from 48% to 86%. Clearly, the inclusion of burglary index significantly improved the model. In sum, the evidence indicates that collective efficacy and burglary index are the most important aggregate variables. In contrast, the social disorganization predictors do not significantly influence perceived crime in the nonmetropolitan setting. Table 18 presents fixed effects hierarchal models for fear of crime. In Model 1, the Chi-square statistic (x2 = 57.81, p < .01) indicated that fear of crime differed across residential units. At the citizen—level, fear of crime was significantly higher among females and younger citizens. After controlling for seven citizen-level variables, 126 collective efficacy was the only residential unit-level predictor to reach statistical significance (H4b). In essence, fear of crime was lower among citizens in residential units with comparatively higher levels of collective efficacy. Economic disadvantage (H6b) and residential instability (H7b) were not associated with fear of crimmu Model 1 accounted for 55% of the explained variance between residential units, while the sociodemographic variables explained 4% of the variance at the citizen—level. 127 Table 18: Fixed Effects Hierarchical Models for Fear of Crime Model 1 Model 2 b t-ratio b t-ratio Variables (SE) (SE) Intercept 2.51 82.94*** 2.51 92.13*** (.03) (.03) Citizen-Level (N = 1,125) Minority .16 1.49 .16 1.48 (.11) (.11) [.04] [.04] Male -.24 -4.60*** -.24 -4.60*** (.05) (.05) [-.16] {-.16] SES .00 .10 .00 .07 (.02) (.02) [.00] [.00] Age -.00 —3.24** -.00 -3.18** (.00) (.00) [-.ll] {-.11] Married —.02 -.42 -.02 -.42 (.05) (.05) {-.01] {—.01} Ownhome .07 1.04 .08 1.07 (.07) (.07) {-.03] [.03] Years in neighborhood -.00 -.26 -.00 -.40 (.00) (.00) [- 01] {-.02} Residential Unit-Level (N=31) Economic disadvantage .03 .79 —- -- (.04) l 16] Residential instability .00 .01 -- -- (.03) [.00] Collective efficacy -.05 —1.61*T —.01 -.28 (.03) (.04) {-.22} {-.04] 128 Table 18: Continued Economic affluence -- -- .00 .03 (.03) [.01] Burglary index -- —- .07 3.10** (.02) [.37] x2 57.84*** 50.98** Variance Explained (Percentages) Within-residential unit 4 4 Between-residential unit 55 26 Note: Standardized coefficient in brackets *p < .10, **p < .05, ***p < .01 (two-tailed tests) tOne—tailed test In Model 2, the Chi—square statistic (x2 = 50.98, p < .05) indicated that levels of fear differed across residential units. Fear of crime was also significantly higher among females and younger citizens. Collective efficacy was not significantly correlated with fear of crime (H4b), but was in the hypothesized direction. Unlike Model 1, which explained over 50%, Model 2 explained a little more than a quarter of the variance. The citizen- level variables accounted for 4% of the within-residential unit variance. According to Lewis and Salem (1986), fear of crime is a consequence of the erosion of social control. Fear and mistrust may breakdown people’s ability to form mutually 129 supportive bonds to help each other deal with the threats in the community, which further reduces collective efficacy. By inhibiting efforts of goodwill, crime prevention efforts are reduced. And this may be the case here because burglary index, as hypothesized, was positively and significantly associated with fear of crime (H5b) . Skogan (1991:45) highlighted the reciprocal feedback effects that crime and its consequences cause. If people shun their neighbors out of fear of crime, fewer opportunities exist for the development of collective efficacy. Weakening of informal social control fuels more crime. Crime is corrosive because it undermines trust among neighbors (Skogan, 1989) . Baba and Austin (1986) found that burglary victimization had an impact on residents’ perception of urban neighborhood fear of crime. It appears that regardless of setting (i.e., urban versus nonmetropolitan), burglary influences fear of crime. Another possible explanation for fear of crime is that nonmetropolitan areas are characteristic of family and kin- based ties; local crimes are likely to increase levels of fear in other residents if they hear about events through local social contacts. Skogan and Maxfield (1981) found exaggeration of fear of crime related to neighbors talking 130 about crime and participating in victimization prevention groups. This indirect viCtimization model posits that local social ties amplify the impact of the crime event, which then increases fear levels (Covington and Taylor, 1993; “Taylor and Hale, 1986) . In short, high crime rates and high levels of fear lead to the atomization of the community (Gates and Rohe, 1987). Regardless of the urban- nonmetropolitan setting, the combination of crime and social interactions among citizens residing in residential units governed by close-ties are likely to influence fear. Table 19 presents fixed effects hierarchal models for perceived incivility. In Model 1, the Chi-square statistic (x2 = 73.58, p < .01) indicated that perceived incivility differed across residential units. At the citizen-level, younger citizens were significantly more likely to report perceived incivilities. At the residential unit-level, after controlling for seven citizen-level variables, collective efficacy was inversely and significantly related to perceived incivility (H4c). Stated differently, citizens residing in residential units characterized by higher levels of collective efficacy reported significantly lower levels of perceived incivility. Findings for economic disadvantage and residential instability did not support the specified hypotheses (H6c and H7c). Model 1 131 accounted for approximately 50% of the between and 2% of the within residential unit variation. 132 Table 19: Fixed Effects Hierarchical Models for Perceived Incivility Model 1 Model 2 b t-ratio b t-ratio Variables (SE) (SE) Intercept 1.71 216.67*** 1.71 283.10*** (.01) (.01) Citizen-Level (N=1,125) Minority .04 1.49 .16 1.48 (.03) (.11) [.04] [.04] Male -.01 - 94 -.01 - 94 (.01) (.01) {-.02] {-.02] SES .00 .24 .00 .24 (.01) (.01) [.01] [.01] Age -.00 -2.62* -.00 -2.61* (.00) (.00) {-.09] {-.09] Married -.01 -1.23 -.01 —1.23 (.01) (.01) {-.03] {-.03] Ownhome -.03 -1.48 -.03 -1.48 (.02) (.02) [-.OS] {-.05] Years in neighborhood -.00 -.04 -.00 -.04 (.00) (.00) {-.00] {-.00] Residential Unit-Level (N=31) Economic disadvantage .01 1.07 -- -- (.01) [.17] Residential instability -.00 —.17 -- -- (.01) {-.03] Collective efficacy -.04 -4.43*** -.02 -1.42*T (.01) (.01) [-.57] {-.28} 133 Table 19: Continued Economic affluence -- -- -.01 -1.49 (.01) {-.12] Burglary index —- -- .03 3.93*** (.01) [.51] x? 73.58*** 47.69*** Variance Explained (Percentages) Within-residential unit 2 2 Between-residential unit 49 28 Note: Standardized coefficient in brackets *p < .10, **p < .05, ***p < .01 (two-tailed tests) tOne-tailed test In Model 2, the Chi-square statistic (x2 = 47.69, p < .01) revealed that perceived incivility differed across residential units. Consistent with Model 1, younger citizens also reported higher perceptions of incivility in Model 2. The hypothesis that collective efficacy would negatively effect perceived incivility was supported (H4c). However, the strength of the estimate was weaker (-.28) when compared to Model 1 (-.57). In Model 2, burglary index was positively and significantly linked to perceived incivility (H5c). This finding supports the crime-perceived incivility hypothesis. According to Skogan (1990), neighborhoods that are orderly, clean, and safe; houses, apartments, and buildings that are 134 well-maintained; and. residents *who rare respectful toward one another and of each other’s property, are likely to experience less incivilities. Although the relationship between economic affluence and perceived incivility was not statistically significant (H8c), it was in the expected direction. Model 2 accounted for 28% of the between and 2% of the within residential unit variance. Table 20 presents fixed effects hierarchal models for risk of victimization. Unlike previous Chi-square statistics, which showed quality of life outcomes to differ across residential units, the Chi-square statistic (30.65) for Model 1 did not achieve statistical significance. In other words, risk of victimization did not differ across residential units. Controlling for minority and four other citizen-level predictors, respondents who were female and of lower socioeconomic status reported higher perceptions of risk. This finding is consistent with previous research suggesting that females and the poor are more likely to report risk. of ‘victimization. (Newman, 1972; Sampson. and Wooldredge, 1987; Hough, 1987; McDowell, Loftin, and Wersima, 1989). 135 Table 20: Fixed Effects Hierarchical Models for Risk of Victimization Model 1 Model 2 b t-ratio b t-ratio Variables (SE) (SE) Intercept 1.74 151.98*** 1.74 160.91*** (.01) (.01) Citizen-Level (N=1,125) Minority -.00 -.03 -.00 -.03 (.08) (.08) {-.00] [-.00] Male -.12 —5.98*** -.12 -5.98*** (.02) (.02) [-.14] [-.14] SES —.04 -3.95*** -.04 -3.95*** (.01) (.01) {-.01] {-.01] Age -.00 -.79 -.00 -.79 (.00) (.00) {-.03] {-.03} Married .02 .94 .02 .94 (.02) (.02) [.02] [.02] Ownhome -.08 -1.85 -.08 -1.85 (.04) (.04) [-.06] [-.06] Years in neighborhood .00 .61 .00 .61 (.00) (.00) [.02] [.02] Residential Unit-Level (N=31) Economic disadvantage .02 1.50 —- -- (.01) [.25] Residential instability .00 .11 -- ~- (.02) [.03] Collective efficacy -.05 -4.10*** -.02 -1.66*’r (.01) (.01) {-.53] {-.21] 136 Table 20: Continued Economic affluence -- -- -.00 -.53 (.01) {-.06] Burglary index -- -- .04 4.05*** (.01) [.51] x“ 30.65 24.53 Variance Explained (Percentages) Within—residential unit 3 4 Between-residential unit 68 86 Note: Standardized coefficient in brackets *p < .10, **p < .05, ***p < .01 (two-tailed tests) tOne-tailed test Adjusting for citizen-level variables, the hypothesized relationship between collective efficacy and risk of victimization (H4d) was supported. Economic disadvantage (H6d) and residential instability (H7d), however, were not supported, but in the expected direction. At the residential unit-level, Model 1 accounted for 68% of the ‘variance. Am. the citizen-level, less 'variance *was explained (i.e., 3%). Like Model 1, the Chi-square statistic (24.53) for Model 2 also revealed that risk of victimization did not differ across residential units. The results for male and SES persisted in Model 2. Collective efficacy (H4d) and burglary index (H5d) were significant and in the 137 hypothesized directions; yet, the magnitude of estimate for collective efficacy' diminished. in. Model 2 from -.53 to -.21. At the residential unit-level, Model 1 accounted for 86% of the variance. At the citizen—level, less variance was explained (i.e., 3%). Table 21 presents a summary of the hypotheses tested between residential unit-level variables and quality of life outcomes. As Table 21 indicates, the general pattern of findings partially supported the collective efficacy- quality of life associations. In addition, the relationship between. burglary index. and. quality' of life outcomes were also supported. 138 Table 21: Summary of Hypotheses Between Residential Unit-Level Variables and Quality of life Outcomes Perceived Fear Perceived Risk of crime of crime incivility victimization Economic disadvantage + + + + Residential instability + + + + Economic affluence - — - - Collective efficacy * - * * Burglary index * * * * * = significant at .10 level and in the expected direction. -“+” or “-” = not significant at .10 level but in. the expected direction. Table 22 presents a summary of hypotheses tested between residential unit-level variables. The first set of hypotheses shows that structural variables (e.g., economic disadvantage, residential instability, and economic affluence) were not supported. The relationship between collective efficacy and burglary index was the only hypothesis to be statistically significant among the residential unit-level variables presented in Table 22. 139 Table 22: Summary of Hypotheses Between Residential Unit- Level Variables. Collective efficacy Burglary index Economic disadvantage - + Residential instability - + Economic affluence + - Burglary index * * = significant at .01 level and in the expected direction “+” or “-” = not significant at .01 level but in the expected direction. Overall, the fixed effects hierarchical models revealed four' general patterns. First, structural variables (i.e., economic disadvantage, residential instability, and economic affluence) were not significantly correlated with quality of life outcomes. Second, collective efficacy33 reduced concerns over perceived crime, perceived incivility, fear of crime, and risk of victimization. Third, collective efficacy was attenuated by crime“ (i . e . , Model 2) . Fourth, burglary index was positively associated with all of the quality of life outcomes. In sum, the observed finding at the residential unit-level indicates that citizen quality of life assessments are influenced by contextual factors, but not entirely as hypothesized. These patterns and deviations are discussed in greater detail below. 140 Discussion While social disorganization theory offers powerful constructs for explaining crime and delinquency, it does not seem to generalize well with regard to citizen quality of life assessments in more rural, less densely populated areas. In this study, social disorganization. exogenous variables (i.e., economic disadvantage, residential instability, and economic affluence) did not set in motion reductions in quality of life. Rather, evidence revealed that residential unit crime rates were directly responsible for loss in quality of life. No claim is made that social disorganization theory has been examined in its entirety. However, the results do not support the primary contention of this research: social disorganization would significantly influence quality of life in nonmetropolitan settings. The lack of significance in the social disorganization predictors may possibly stem from the fact that little social disorganization existed in nonmetropolitan areas to begin with. Despite these limitations, the findings provided support for the secondary contention that collective efficacy reduces negative perceptions of crime, fear, incivility, and risk of victimization. After controlling 141 for citizen and residential unit-level predictors, citizens residing in residential units characterized by higher levels of collective efficacy rated their quality of life in more positive terms. The unique contribution of this researdh is that it has explored the influence of collective efficacy in nonmetropolitan settings. Collective efficacy and burglary rates revealed strong contextual effects, and thus play' an important role in determining citizen. quality' of life assessments in nonmetropolitan settings. While Osgood and. Chambers (2000) found that social disorganization per se (e.g., residential instability and ethnic heterogeneity) generalized well to nonmetropolitan Florida, Georgia, South Carolina, and Nebraska, the results presented here do not allow such a generalization. If one pays close attention to the states in Osgood and Chambers’ (2000) study, they are representative of the South. The rural literature has found that the South, compared with other regions (e.g., North and Midwest), is likely to experience greater structural inequality (Tickamyer and Duncan, 1990). Future replications of Osgood and Chambers' (2000) work in other regions throughout the United States may report findings similar to those highlighted above. 142 Do structural features of social disorganization influence citizen quality of life in nonmetropolitan settings? Social disorganization structural predictors do not seem to influence citizen quality of life assessments across nonmetropolitan residential units. iDoes collective efficacy, found to influence crime and quality of life in urban areas, have a similar influence in less densely populated areas? Collective efficacy directly enhances quality of life. However, crime indirectly reduces collective efficacy and directly leads to lower levels of quality of life. Not until more nonmetropolitan settings are examined can definitive statements be made about the applicability of social disorganization to all settings. Consistent with Osgood and Chambers (2000), however, testing and expanding social disorganization theory outside the urban box is warranted. While structural antecedents are important, researchers must also examine important social mechanisms that have the ability to undermine structural forces on crime and quality of life. Therefore, the critical task is to develop research projects that test the mediating capability of collective efficacy in socially disorganized nonmetropolitan settings. Given the evidence reported throughout this dissertation, an interesting question remains to be 143 addressed: how can citizens in the local community increase collective efficacy and reduce crime? One possible answer worth exploring is building social capital (Coleman, 1988a). The utility of social capital in terms of crime reduction and prevention is the topic of the next chapter. 144 CHAPTER FIVE: SOCIAL CAPITAL AND POLICY IMPLICATIONS Over the last six decades, social disorganization theory has primarily been tested in the urban setting. The present research. sought to assess social disorganization theory in nonmetropolitan areas. In addition, one of the more important social process concepts that has emerged in the ecological literature (i.e., collective efficacy) was gauged. Along the way, three major findings were revealed. First, structural social disorganization predictors did not influence citizen quality of life. Second, collective efficacy was associated with higher citizen quality of life evaluations. Third, rates of burglary were inversely related to levels of collective efficacy, and also negatively associated quality of life. While research suggests that structural characteristics influence aggregate rates of crime and disorder (Shaw and McKay, 1942; Sampson and Groves, 1989; Sampson et al., 1997; Sampson and Raudenbush, 1999), structural characteristics did not influence quality of life in the nonmetropolitan settings observed here. One plausible explanation for these null findings concerns social capital. 145 This chapter has two objectives. The first half of the chapter systematically identifies sources of social capital at various levels that likely influence citizen quality of life. I do this by identifying: (1) macro-level factors (e.g., federal policies) that have historically influenced the development of nonmetropolitan communities, (2) societal institutions (e.g., church, school, police) that have traditionally provided social control, (3) the family and its role in providing supervision and community development, and (4) characteristics of the community, such. as norms and values, that play a role in solving problems through informal social control. The latter half of the chapter, which has a policy- orientation, focuses on crime intervention, prevention, the enhancement of community social organization, and cooperation between citizens and criminal justice agencies. Using the social capital framework allows alternative social resources to be identified to improve citizen quality of life and reduce crime. Defining Social capital Although social capital is defined in many ways (Portes, 1998; Sandefur' and. ILaumann, 1998), it is conceptualized here as resources produced through 146 relationships (Coleman, 1990:304). More specifically, social capital is a social good embodied in the structure of social networks (Coleman, 1990). Social capital can manifest itself in a variety of institutions, such as school, police, and family. Accordingly, social capital is also found in the larger aggregates in which these institutions are embedded, such as communities. Social Capital and Collective Efficacy This section seeks to advance the discussion of social capital one-step further by linking it with collective efficacyi35 Identifying sources of social capital can provide a better understanding concerning the development of collective efficacy, as well as reduce levels of social disorganization. For example, Sampson (1995:199) has previously identified the social disorganization-social capital connection; he argues that the lack of social capital is one of the primary features of socially disorganized communities (see also Cbleman, 1988a; Sampson 1992; Putnam, 1993). Sampson (1995) suggests that the theoretical task is 11) highlight characteristics of communities that produce social capital among families and children. 147 Coleman (1988:98) argues that social capital is a form of social organization. Mechanisms of social organization are viewed as a control model, and social disorganization adversely affects levels of community social control (Kornhauser, 1978; see also Sampson and Groves, 1989). Therefore, the logic is that if social disorganization is not present or effectively combated, then social capital (i.e., social organization) is more likely to facilitate levels of collective efficacy. What is the link between collective efficacy and social capital? While collective efficacy and social capital share similar characteristics, they do possess analytically distinct qualities. These distinctions are apparent in terms of resources and modes of action. I begin by highlighting the similarities. Similarities. First, collective efficacy is associated with dimensions of social cohesion, trust, and informal social control (Sampson et al., 1997; Sampson et al., 1999). Social capital is also associated with social cohesion, trust (Wall et a1. 1998:303-304; Coleman, 1988a:101; Putnam, 1995:67,73), and intervention that benefits a group or community (Wall et al., 1998:304; Portes, 1998:6,12; Coleman, 1988a:100; Putnam, 1995:67,73). Second, collective efficacy has been defined as task 148 specific achievements, processes of active engagement, the exercise of control, and shared expectations among neighborhood residents to effectively maintain public order (Sampson et al., 1997; Sampson et al., 1999; Sampson and Raudenbush, 1999) . Social capital is broadly defined as a resource derived from social networks of shared l \ expectations and obligations that are activated for the purpose of enhancing social mobility, economic growth, political prominence, and community vitality (Wall et al., 1998:304-308) . Distinctions. Despite these similarities, Sampson et al. (1999:634-635) argue that social capital is distinct from collective efficacy because the latter involves “active engagement” in the exercise of control, whereas the former is a process in terms of “resource potential.” A more efficient way to view these distinctions is from a structural re source—agency perspective . According to Sampson et al. (1999:635) social capital is a commodity of resources (e.g., voluntary associations) embedded in the structure that may or may not proceed in active engagement. Social capital, then, has the potential to facilitate (or hinder) social action (Wacquant, 1998:26; see also Coleman, 1988b:100) . 149 In contrast, Bandura (1986:449-452) views collective efficacy as an important element in sustaining members’ commitment to their cause, it is an assessment of their agency. Hence, collective efficacy is considered the “agency;” it is a modus of operandi for action. Agency refers to the willingness of individuals to control and affect their environment by taking action. Collective efficacy is a locus of social control whereby citizens develop their capacity to serve as effective causal agents in their day-to-day lives. Collective efficacy, then, to some degree depends on social capital. The Process of Social Capital To understand the accumulation process of social capital in nonmetropolitan settings, we must take a step back and see how economic and cultural capital influences the development of family and community. Doing so will help us to better understand collective efficacy. Economic Capital: Government Policy Economic capital is a primary step toward generating social capital. Fox (1995) claims that social capital is co-produced between federal/state actors and local groups.“‘5 An early example of economic capital was President Theodore Roosevelt’s 1908 Commission on Country Life, which called 150 A /_ national attention to improve rural life. Roosevelt’s Commission concluded that the “major sources of problems of rural people were lack of organization, failure of rural social institutions, and inadequate infrastructures” (Summers, 1986:348). Six years later, Congress passed the Smith-Lever Act of 1914 aimed at better planning, more happiness, and improved education for rural Americans (Hooks and Flinn, 1981). In 1925, the Purnell Act authorized economic and sociological investigation for development and improvement of rural institutions, home, and life (Sanderson, 1927). By the 19303, rural America was experiencing a mass exodus of residents to major urban centers. According to Madison (1986), the out-migration of residents negatively affected the economic base in rural areas. As a result, the federal government continued to provide economic capital to rural American. Later, in 1961 Congress passed the Area Redevelopment Act (ARD) that provided capital to problem areas in the form of low-cost loans to industry, and loans and grants to communities for upgrading infrastructural support to attract new industry (Summers, 1986:366).37 The literature suggests that governmental action has greatly aided American rural communities.“8 The history of community social organization portrays vertical integration 151 (i.e., federal intervention) as a significant factor in the development of American rural communities. In short, it is my contention that economic policy influences the formation of social capital in nonmetropolitan areas. Cultural Capital: The Church, School and Police The church and school are institutional structures where resources can be found. Ini other words, the church and school consist of ties where individuals can draw upon resources by virtue of membership (Wacquant, 1998:28). In the 19003, the church and school served as a way to socialize and train the young, and instill a sense of community, stability, and tradition for adults . On one hand, the church was viewed as the seminal formal institution. Loomis and Davidson (1939:28) argue that the church had priority over all other types of social agencies. Loomis (1939:2) claims that :hi nonmetropolitan America, it was customary for residents to visit new arrivals and invite them to church. On the other hand, Wacquant (1998) argues that no organization better exemplified institutional influence than the school. However, Madison (1986) argues that quickly as churches and schools became the centerpiece for nonmetropolitan life, the out-migration of nonmetropolitan 152 residents to urban cities in seek of economic and social opportunities led to their demise. According to Madison (1986:645), “Progressive Era reformers concentrated their prodigious energies and talents on the receiving end of the rural—to—urban populations shift.” The police provide yet another source of social capital in nonmetropolitan .America. Police-citizen relations have traditionally been defined according to close relationships whereby citizens exhibit more confidence in the police, when compared to urban residents. Unlike urban areas, where police provide little physical security, legal protection, and services (Kotlowitz, 1991), it is quite possible that citizens rely less on police because nonmetropolitan areas are governed by traditional family and kin ties that react with non-legal sanctions (Weisheit et al., 1994). In other words, citizens are more likely to intervene to solve social problems.39 Sampson (1995:209) argues that the social capital model can. be extended to agents of criminal justice. In addition, Bursik and Grasmick (1993) highlight the importance of public control (e.g., police services) that helps sustain community organization and crime control. 153 Social Capital: The Family For the most part, family and kin informally handle many types of social conflict, and the nuclear family is an important source of social capital (Coleman, 1988a). Coleman (1988a:111) reasoned, however, that capital deficits will develop if healthy relations between children and parents are not maintained. Because social relations in nonmetropolitan areas are likely to be governed by family and kin networks, available social capital is likely higher when compared to urban areas. Hofferth and Iceland (1998) found that families living in rural areas are more likely to exchange exclusively with kin than families living in urban areas. Since families are sometimes isolated from. the community, larger social networks are needed to promote social action. The resourcefulness of the family to facilitate social action. was hindered. by rural-urban. migration. during the mid-19003 (Madison, 1986), and. additional resources 'were needed to foster informal social controls . According to Taylor (1927), the community appeared most suitable to facilitate social action given characteristics, such as group orientation and non-competing institutions and associations. Thus, the community was viewed as a place 154 where social relations could produce resources of social capital that would promote collective community action. Social capital: The COmmunity The social capital literature has extended the concept from an individual resource to a feature of communities (Coleman 1988a:113; Portes, 1998; Wall et al., 1998). Sampson et al. (1999:634) suggest that “sources of social capital tied to local community context are analytically distinct from the more proximate family processes and relationships observed inside the homey” The two concepts are distinct because social capital, as a resource, has the potential to be converted into modes of action (i.e., agency) by residents in the community. According to Wall et al. (1998:311-312) communities that are well governed and moving ahead economically usually are richer in social capital, whereas more economically disadvantaged communities usually lack social capital (Wall et al., 1998:311). It is often customary to assume that nonmetropolitan areas are characterized by greater solidarity, social cohesion, and law-abiding residents (Sorokin et al., 1930). Coleman (1988a:104) argues that community’ norms and effective sanctions are more likely to be in place due to communal traditions. 155 Coleman (1988a:104-105) claims that norms and effective sanctions can inhibit crime by making it possible to walk freely outside at night and enable old persons to leave their homes without fear for their safety. Another reason that nonmetropolitan communities may experience higher levels of social capital is because most nonmetropolitan areas are racially homogeneous. Research suggests that variation in race and ethnicity (i.e., ethnic heterogeneity) is likely to produce different norms of behavior (Shaw and McKay, 1942). Different community cultures and value systems are linked to differential rates of crime (Short, 1990:11-12; Luckenbill and Doyle, 1989) and leads to the formation and transmission. of deviant subcultures (Kornhauser, 1978:75). In. essence, cultural heterogeneity impedes communication and obstructs the pursuit of common cultural values (Bursik, 1988; Sampson, 1988).“° Figure 7 shows that social capital is best viewed as a process and resource of social control. Social capital can flow from different entities, both formal and informal, and at different levels (e.g., family and community). The process begins with economic capital in the form of state capital. Formal institutions receive economic capital. The police, for example, can convert this economic capital 156 into social capital. The transformation of social capital by' formal institutions 3provides families ‘with. resources, support, and the means of social control against wayward behavior. Bursik and Grasmick (1993) highlight the importance of public control, which they define as the capacity of local community organizations to obtain extra- local resources (such as police) that help sustain neighborhood organization and crime control. In other words, by working together, citizens and police, have the capacity to convert social capital into collective efficacy. 157 Figure 7: The Processes and Resources of Social Capital in Nonmetropolitan Settings Resource — Economic Capital: Government Policy Cultural Capital: Church, School, and Police Social Capital: The Family Social Capital and Collective Efficacy: The Community Agency 158 Social capital appears to be most effective at the community—level in terms of informal social control. However, since crime inhibits the development of collective efficacy and reduces quality' of life, the next section addresses public policy related to crime. I focus on police initiatives and more important community policy aimed at securing alternative resources of social capital. Policy Implications While much of the social capital literature presented thus far may indicate that nonmetropolitan areas are resourceful and possess characteristics of social organization, this does not mean that these areas are crime free. The second question asked in this chapter stems from the findings regarding crime in Chapter Four. If social capital is likely to be more abundant in nonmetropolitan settings, why did citizens report higher quality of life assessments when residing in residential units characterized by higher levels of collective efficacy? This pattern of findings reported in Chapter Four suggests that public agencies should take a more active role at reducing crime, given that citizens are unable to build collective efficacy at levels that can reduce the negative impact of crime. As previously mentioned, social 159 capital is conceptualized here as a source of social control. Therefore, social capital. is ‘useful for advocating crime related policies from a police and community perspective. After all, social capital has the potential to facilitate crime prevention strategies and collective efficacy. Police Resources and Public Control While the community theme is important, I first argue that criminal justice agents are key players in reducing crime, and. increased. ;programmatic collaboration. ‘with community residents is needed. Several studies have emphasized that citizens in the community can address problems of crime, disorder, and fear of crime by securing ties to public officials and the police (Kelling and Coles, 1996; Medoff and Sklar, 1994; Podolfsky and Dubow, 1980; Rabrenovic, 1996; Rooney, 1995; Skogan and Hartnett, 1997). As mentioned earlier, however, many of the infractions in nonmetropolitan. areas are Ihandled. informally' whereby residents respond with non-legal sanctions (Weisheit et al., 1994).41 Assuming that the community does, in fact, take more proactive measures in crime prevention than police, any increase in police initiated crime prevention efforts would 160 generate resources of social capital and enrich police- citizen relations. Perhaps it is time for nonmetropolitan police to take a more active role in fighting crime and develop meaningful crime-related programs so that citizen quality of life can be improved. Police-citizen cooperation, for example, can generate new resources of social capital that can be used to make possible collective efficacy. For instance, police can take part in town hall meetings that inform and collect information from residents about community concerns. In addition, town hall meeting can inform residents about crime prevention or “target hardening” (Newman, 1972; Jeffrey, 1977), such as installing deadbolt locks and alarms (Taylor and Schumaker, 1990). More complex police policy includes identifying “hot spots” of crime, such as burglary. This approach is consistent with social disorganization theory that identifies crime prone places and communities rather than people (Sampson, 1995; Brantingham and Brantingham, 1999) .“2 Once identified, citizens can take part in community—based task forces aimed at monitoring hot spots (Kelling and Coles, 1996). Like many police departments, citizen participation in police programs alleviates many of the 161 problems of understaffed departments. In turn, this exchange of services helps to produce social capital. As Figure 7 indicated, it is important for police to promote vertical integration with local communities for the purpose of securing extra-local resources. “When. local organizations are unstable and isolated, and when the vertical links of community institutions to the outside are weak, the capacity' of a community to defend its local interests is weakened” (Sampson, 1995:214). This can be accomplished. by' citizens in the community‘ securing resources of jpublic control from. local governments (Sampson, 1995; Bursik and Grasmick, 1993; Velez, 2001). Building police-citizen relations can have additional benefits. For example, Skogan (1990) found that in Newark police-citizen ties led to a decrease in fear of crime. In addition, such policies call for citizens to request resources from those with political decision-making responsibilities. Bursik (1989) found that political decision-making is salient for neighborhood levels of crime and victimization. Velez (2001:840) recently noted “residents must establish ties to city elites in order to influence political decisions that affect their neighborhoods, including their level of crime.” 162 According to Bursik and Grasmick (1993:17-18) public control refers to the capacity of community social networks to solicit and secure external resources by establishing ties to local government such as the police. In addition, Sampson (1995:214) refers to public control as the “ability to secure public and private goods and services that are allocated by groups and agencies located outside of the neighborhood.” The idea is that such goods and services reflect resources obtained from outside the community that can be used inside the community for the purpose of social control. Although police have an important role, residents must take advantage of all the opportunities of social capital and collaborate with criminal justice agencies to better facilitate collective efficacy. Community Resources and Public Control While law enforcement strategies are helpful, they remain too simplistic. Although employing police strategies may temporarily reduce crime across nonmetropolitan. Michigan. areas, I argue for’ policy’ that generates resources of social capital that are likely to help facilitate collective efficacy. Community social control is most effective when citizens regulate the behavior of residents and visitors (Bursik and Grasmick, 163 1993). How is this function best carried out? By building social capital and. providing resources to community members. Community strategies can have long-term effects in developing various resources of social capital. Thus, policies should focus on community mobilization that emphasizes strategies to capture the attention of police and political constituents. While these approaches may be carried out informally in nonmetropolitan areas, they are nonetheless important. According to Rabrenovic (1996), in New York, citizens from a local community captured the attention of city officials by having lunch at the Mayor’s office door to protest that their community was unsafe due to crime. Once citizens have captured the attention of public officials, resources are more likely to follow. In sum, neighborhoods with strong ties to public officials and the police are able to secure resources that effectively diminished victimization risks (Velez, 2001:855). More common community mobilization strategies include citizen patrol associations (Yin, 1977) and voluntary block associations (Perkins et al., 1990), which. provide additional resources that facilitate modes of action above and beyond police patrols. Perkins et al. (1990:90) argues that block associations may' help reduce crime and fear 164 because (1) residents are likely to share similar concerns, (2) participation rates are higher than any other community program, and (3) small face—to-face crime prevention should work better than larger organizations. According to Greenburg (1983) , crime prevention associations are more successful in homogenous areas . Given the socio- demographic homogeneity characteristics of nonmetropolitan Michigan areas, community-level strategies may be more successful in creating social capital and facilitating action to address crime. Sanders and Lewis (1976) claim that it is a truism in nonmetropolitan community life that voluntary organizations provide the mechanisms for carrying out many communal goals.“3 Community mobilization might also reinforce the idea that community residents can be relied upon to maintain public order. Lynch and Cantor (1992) argue that burglary is a function of guardianship and suggest that neighborly watchfulness can affect the risk of burglary (Lynch and Cantor, 1992:356). In this case, community members should encourage residents to take part in neighboring, which in turn, can facilitate neighborhood watches and surveillance (e.g., citizen patrols). Neighboring is characterized by social interaction among neighbors, such as talking and gathering socially. 165 Research has indicated that neighborhoods with high levels of neighboring were associated with lower rates of crime (Bellair, 1997, 2000; Warner and Rountree, 1997). It is important to note that more rural, less densely populated areas (e.g., Grand Traverse County) are likely to face some difficulty at accomplishing this neighboring task, but small towns (St. Johns) and small cities (Traverse City) may have fewer problems doing so. Because burglary usually occurs when no one is home and at night, citizens should watch the homes of others and take action when needed. Community watchfulness does not have to involve formally organized crime prevention programs to be effective. Podolefsky (1983) found that informal social control appeared strongest and most effective in neighborhoods without much organized crime prevention activity. Bankston et al. (1987) concluded that perceptions of fear and victimization with respect to burglary in more rural areas was due to the lack of neighborly behavior (e.g., house watching) and other forms of social monitoring within the community. In sum, crime has the potential to limit access to resources (i.e., social capital) that are necessary' for facilitating collective efficacy.“ By increasing social capital, there is reason to believe that citizens can fully 166 develop their capacities for mobilizing resources aimed at collective efficacy. Developing' collective: efficacy requires citizens to have access and be afforded the opportunity to experience resources (e.g., access to police and government officials) of social capital. Conclusion Drawing on the concept of social capital, this chapter has identified various institutions (e.g., police and community) that are likely to facilitate collective efficacy in nonmetropolitan areas. .Accordingly, nonmetropolitan areas demonstrate evidence of residential unit structural continuity (i.e., the lack of social disorganization) likely to generate social capital (Coleman, 1990). However, community social control cannot remain effective given the negative influence of crime, and as a result, alternative resources are needed to reduce crime and enhance citizen quality of life. Because social capital is lodged in structure (Sampson and Raudenbush, 1999:635), it is considered a limited good (Wall et a1, 1998:311).45 In other words, if the structure of the community becomes socially disorganized (i.e., loses its continuity), then social capital may wane. When resources of social capital cannot be found or have been 167 depleted, Coleman (1988b) advocates manipulating the social structure to generate social capital and bring about social change. One way of manipulating the social structure is to attract the attention of public service and government officials and secure alternative resources of social capital that can facilitate and reinforce collective efficacy. The empirical evidence presented throughout this dissertation has shown that nonmetropolitan residents’ concerns with quality of life are not merely reflections of romantic visions, but rather are rooted in measurable effects of collective efficacy and crime. Consequently, this research reaffirms theoretically and empirically that collective efficacy holds promise as an aggregate-level attribute (Sampson et al., 1997; Sampson et al., 1999); but also highlights the negative impact of crime on collective efficacy and quality of life. Similar to collective efficacy, Portes (1998:21) believes that the greatest theoretical promise of social capital is to define it as a structural property of large aggregates. Nonetheless, a central theme of this research is that crime, regardless of where it occurs (e.g., nonmetropolitan settings), remains to be a community wide problem, which in turn, affects quality of life assessments. 168 Because citizens are less likely' to jparticipate in collective efficacy due, in part, to crime there are greater chances they will remain confined to their homes and avoid social interactions and networking among neighbors aimed at building social capital. Coleman (1988b:98) argues that social capital, based on the structure of relations among persons facilitates action, “making possible the achievement of certain ends that in its absence would not be possible” (see also Coleman 1990:300). In short, social capital has the potential to bring about greater control over crime. Future Research This dissertation has extended research on communities (e.g. collective efficacy and social disorganization), crime, and quality of life beyond the urban box. However, the research is limited because there was little evidence of socially disorganized residential units. Therefore, making broad. generalizations that social disorganization theory has applications to communities of all sizes (Osgood and Chambers, 2000) is cautioned. Nevertheless, social disorganization is an appropriate starting point for developing criminological theories specific to rural settings (Osgood and Chambers, 2000:108). 169 Future research in nonmetropolitan settings is warranted” First, research should closely' examine ‘more socially disorganized areas. As Tickamyer and Duncan (1990) note, socially disorganized (e.g., poverty) nonmetropolitan communities are likely to be found in the South (see also Osgood and Chambers, 2000). By including socially’ disorganized communities, we can. gain. a Ibetter sense of the effect of collective efficacy and its ability (or inability) to reduce crime and enhance quality of life. Second, nonmetropolitan communities in the South and Southwest are more likely to have a wide variation of racial and ethnic differences compared to the Midwest. In the Southwest, for example, Hispanics are a growing population with interesting immigration patterns, are often considered .the working poor, and possess unique cultural lifestyles (Martinez, 1996). This ethnic variation (i.e., ethnic heterogeneity) increases the chances of testing Shaw and McKay’s (1942) social disorganization model. Third, future research should examine community-level characteristics based on a continuum that includes urban, suburban, small-town rural, and rural farming communities. This continuum will provide a wide variation of community characteristics that are likely to produce different outcomes. Lastly, researchers should investigate whether 170 communities with more social capital are likely to facilitate collective efficacy while controlling for other contextual and individual-level variables. Overall, the present research has provided a rare look at contextual features of the community that are likely (unlikely) to influence quality of life in nonmetropolitan Michigan areas. However, to truly acquire a sense of the contextual-level relationships at work, researchers must consistently use criminological theories of crime to guide research in these settings. In addition, practitioners must devote special attention to cuime-related issues and identify alternative resources of social capital to reduce crime and enhance quality of life assessments. 171 ENDNOTES Sanders and Ensminger (1940) were the first to apply the cluster method. The neighborhood cluster method was used in “areas where local neighborhood identification was strong and neighborhoods, rather than individuals, could be clustered around one of the predesigned community centers” (Sanders and Ensminger, 1940:3). Smith (1941:391) described neighborhoods as “small clusters of families...they are the next group beyond the family to have social significance.” Recently, Sampson and Jeglum-Bartusch (1998:783) conceptualized neighborhood as an “ecological sub-section of a larger community—a collection of both people and institutions occupying a spatially defined area that is conditioned by a set of ecological, sociodemographic, and often political forces.” Elliot et al. (1996:390) conceptually described. neighborhood. as “a transactional setting that influences individual behavior and development both directly and indirectly.” They argue that the neighborhood is indicative of a multi dimensional cluster of traits whereby neighborhoods are changing structurally and individually. In other words, 172 neighborhoods are not static entities; they are part of the political, economic, and cultural context in which they are located. According to Baba and Austin (1989:768), neighborhoods are part of a complex ecological system involving interaction and subjective evaluations. Despite the Chicago Schools’ interest in urban cities, research did not exclusively focus on these areas. Like Galpin’s influential rural research in America, Thomas and Znaniecki (1927), colleagues from the Chicago School, first developed the theoretical concepts of community solidarity and disorganization in a study on the effects of migration and industrialization on rural communities of Polish peasants. Research produced during this era would later greatly influence others. Meanwhile, researchers at Columbia University were also conducting rural community research. Rural researchers focused on “the disorganization effects of communications on the small community, and showed how people and institutions made successive adjustment 3 to the expanding urban world” (Hollingshed, 1948:138). 173 Zone I was conceptualized as the central business district characterized by overall economic and social stability, while Zone II was conceptualized as the area immediately around Zone I characterized by economic and social instability (Vold et al., 1998:143). As one starts to move outward from the city, neighborhoods experience low levels of social disorganization and high levels of community social control. Shaw and McKay adopted a neighborhood-level approach (see also Wilson, 1987; Sampson and Groves, 1989; Rountree, Land, and Miethe, 1994, for contemporary neighborhood level theory). This approach. places emphasis on neighborhood level variables that are expected to effect individual outcomes over and above individual level variables (Rountree et al., 1994:389). The community-level perspective is unique in that it moves away from a simple kinds of people analysis to a focus on contextual characteristics that influence various outcomes (Sampson and Wilson, 1995:54). The link between the systemic model (Kasarda and Janowitz, 1974) and social disorganization theory is 174 that both presuppose ecological influences, which undermine formal and informal ties likely to control local community problems. For example, communities that experience high resident attrition and turnover (e.g., residential mobility) are less likely to develop meaningful and effective social networks due, in part, to residents’ short stay in the community (Sampson, 1988). Therefore, the effectiveness of social control depends on frequent contact and density' that binds residents together' as. a social community (Bursik and Grasmick, 1993:4). Communities characterized by extensive friendship networks, high organizational participation, and effective control of teenage peer groups had lower than average rates of burglary (Sampson and Groves, 1989:790). Low economic status was measured according to the proportion of persons living below the poverty line and unemployment rate (Osgood and Chambers, 2000:95). The connection between economic status and crime and delinquency is consistent with pervious research in that communities with low economic status lack resources, have greater residential instability, and are more likely to attract immigrants. These 175 10. 11. 12. factors, working in combination, impede the development of informal social control and thereby lead to crime and delinquency (Bursik and Grasmick, 1993:39). The closest dimension of social organization is family disruption, but Sampson and Groves would argue that this variable is an exogenous variable and more appropriately represents a structural antecedent. Cook et al. (1997:95) argue that “social organization has continued to play a mediating role, so that a set of neighborhood attributes is assumed to instantiate social disorganization, which then affects individual outcomes.” In a related study, Sampson et al. (1999) use the PHDCN data to assess mechanisms of social organization of 8,500 individuals across 342 neighborhoods. Sampson et al. (1999) examined whether neighborhood social organization mediates the effect of structural antecedents on control of children. The authors tap three new sources of social organization: intergenerational closure, reciprocated exchange, and child-centered social control (a hybrid of collective efficacy). They viewed mechanisms of social organization (e.g., 176 13. child—centered social control [collective efficacy], intergenerational closure, and reciprocated exchange) as endogenous variables, whereas structural antecedents (e.g., concentrated disadvantage, residential instability, concentrated affluence) were considered exogenous. Findings showed that residential stability and concentrated affluence are strong predictors of intergenerational closure and reciprocal exchange (Sampson et al., 1999:633). Concentrated disadvantage, in turn, was found to be associated with lower expectations for shared child control. More important, results for spatial proximity showed that neighborhoods with higher levels of intergenerational closure, reciprocal exchange of information, and shared willingness to intervene on behalf of children are more likely to influence adjacent neighborhoods with respect to building effective means of social organization mechanisms (Sampson et al., 1999:657). Their study included the systematic social observation (880) of 23,000 street segments and a survey of 3,500 residents in 196 Chicago neighborhoods. 177 14. 15. Sampson and Raudenbush (1999:609) provide further reasoning for the importance of examining disorder rather than crime alone: [Wlhile both crime and disorder reflect common origins, crime may be less relevant for understanding processes such as population abandonment and perceived incivility of urban life [and thus] propose that disorder is the more visually proximate or immediate neighborhood cause of theoretical interest, even if it is not a direct cause of further crime. As a result, collective efficacy is relevant to explaining incidents of crime and disorder (i.e., collective efficacy can prevent disorder). Physical disorder was operationalized according to the presence or absence of items such as cigars, garbage, and litter, while social disorder was operationalized by the presence or absence of items such as adults loitering, drinking alcohol in public, and public intoxication (Sampson and Raudenbush, 1999:618). The authors write that items on the disorder scales “bear a conceptual affinity with concurrent “crime" in the sense of violation” (Sampson and Raudenbush, 1999:618) . Predatory crime was measured from respondents’ report whether they 178 16. 17. had. experienced in the last six; months a 'violent victimization or a household burglary or theft victimization. Sampson and Raudenbush (1999) used systematic social observation data (SSO) to build these measures. The little research that has been conducted in the nonmetropolitan setting is comparative in nature. Belyea and Zingraff (1988) explored the relationship between fear of crime in urban and rural locations. Using a random sample of 3,109 respondents in North Carolina, residents were asked about levels of fear and anxiety concerning crime. Predictors of fear included residential location (e.g. rural, town < 2,500, town 2,5000-9,999, city 10,000-24,999, city 25,000—49,999), age, gender, race, education, income, victimization, crime perception, crime rate, and seriousness of crime. The multivariate analysis showed that rural residents have as significantly lower level of fear than their urban counterparts after controlling for gender, race, age, education, and income. One possible explanation is that residents in rural areas tend to help one another, whereas in urban locations, relationships are stranger-based and individuals are less helpful 179 (Amato, 1993). Kennedy' and. Krahn (1984) examined fear* of crime of rural residents who» migrated. to urban areas. Using survey data of 736 residents from two large western Canadian cities, the study estimated the effects of size of the community of origin, length of residence, and other demographic variables on fear of crime. The results showed that “for new arrivals in a city, size of place of origin has a substantial effect on fear of crime, but this effect is shortlived...the larger the community of origin, the safer the current big city resident feels" (Kennedy' and. Krahn, 1984:257). .After controlling for community of origin, women felt less safe than men. As expected, men originating from rural areas felt less safe than men from urban areas (Kennedy and Krahn, 1984:257-258). Due to the strong familial-based social ties commonly found in rural areas, people who had recently arrived to the city were trusting of others. Trust among those coming to the city from rural areas ultimately led to the fear of crime. However, once acclimated to the urban way of life, rural residents adjusted to stranger-based relations, which resulted in a decrease in fear of crime. 180 18. 19. 20. 21. Rurality is measured as percent of a county’s total population composed of rural nonfarm residents and the percent of rural farm residents. Density of acquaintanceship was operationalized as the average proportion of people in the community who do not know others in the community (Freudenburg, 1986:30). Consistent with Wilkinson, Beggs et al. (1996:316) found that nonmetropolitan residents have higher proportions of familial-neighbor based relations and do not depend on others when forming social networks. Hence, it appears that social networks in rural areas are based on familial relations. Using phone surveys from 415 Utah nonmetropolitan residents aged 18 and over, Stinner et al. (1990) examined whether community size effects the linear development of systemic models more than community attachment. Stinner et al. (1990) estimated the effect of community size (i.e., villages, cities, and nonmetropolitan areas), demographic characteristics (e.g., duration of residence, SES, homeownership), and three concepts of community attachment (e.g., involvement, amity, and sentiment). The multivariate results showed support for the systemic model 181 22. 23. 24. (Kasarda and Janowitz, 1974) suggesting that residential tenure and social interaction based on close-knit meaningful relationships matter more than community size (Stinner et al., 1990:497). Bachman (1992) found that elderly who reside in nonmetropolitan areas have a greater chance of becoming victims of household crimes (burglary, larceny, motor theft) than do elderly metropolitan residents. Rountree et al. (1994:387) state that “an important motivation for multilevel analysis is the potential for progress towards the goal of theoretical integration in criminology...this type of integration places causal significance on both large-scale forces and individual-level adaptations” (see also, Lynch and Cantor, 1992; Taylor, 1996; Sampson and Wilson, 1995) . The study of how aggregate properties influence individual outcomes is an important step toward integrating theory and research across individual and aggregate units (Smith and Jarjoura, 1989:624). St. John’s police officials were interested in learning about county residents. As a result, residents who lived outside St. John’s city limits 182 25. 26. were surveyed, but excluded from this analysis. Excluding non-city residents is an analysis issue, not a sample quality issue. While property crimes were exclusively used for this research project, 1998 UCR crime data were examined in each of the sample sites. In St. Johns: 2 aggravated assaults, 51 vagrancies, 7 disorderly conduct, 28 larcenies, 27 liquor violations, and 35 non-aggravated assaults. In Grand Traverse County: 65 aggravated assaults, 1600 vagrancies, 438 disorderly conduct, 803 larcenies, 126 liquor violations, and 548 non—aggravated assaults. In Traverse City: 26 aggravated assaults, 209 vagrancies, 167 disorderly conduct, 420 larcenies, 98 liquor violations, and 252 non-aggravated assaults. Because nonmetropolitan areas are characteristic of large geographic areas, Darling and Steinburg (1997:121) claim that automobile transportation may alter the “boundedness" in rural areas. Thus, the selection of social ties (e.g., friendship groups) and sense of community as a result of transportation patterns lead to variation in rural, suburban, and urban geographical boundaries (Darling and Steinberg, 1997:121). 183 27. 28. Homogeneity and independence assume that individuals in the same group are closer or more similar than individuals in different groups (Bryk and Raudenbush, 1992:xiv). Moreover, “[i]ndividuals are all independent; group components are independent between groups but perfectly correlated within groups...some groups might be more homogenous than other groups, which means that the variance of the group components differ” (Bryk and Raudenbush, 1992:xiv). Overall, these dependent quality of life variables, according to Skogan (1999:47), tap unique dimensions of the crime phenomena. First, perceived crime assesses an individual’s perception of crime in the neighborhood context (Skogan, 1999) and reflects beliefs about crime levels or trends (Gates and Rohe, 1987:427). Individual assessment of crime is often accurate, while perceptions of fear and victimization are inaccurate (Skogan et al., 1981). Moreover, researchers have found that perceived crime can be independent of fear of crime (Furstenburg, 1971; Hartnagel, 1979). Second, fear of crime is distinct from victimization because fear is more widespread than victimization (Covington and Taylor, 1991; Skogan, 1999; Dubow et al., 1979). Researchers have 184 29. regularly asked individuals to rate their fear levels in addition to assessing perceptions of becoming a victim of certain crimes (Skogan, 1999). Lastly, risk of victimization is distinct from fear of crime because it represents a more cognizant assessment of the likelihood of victimization (LaGrange and Ferraro, 1989). Two studies, in particular, represent excellent examples of various multilevel shortcomings with respect to insufficient cases. Geis and Ross (1998) used 1995 Community, Crime and Health data that surveyed 2,482 adults in Chicago aged 18-92 and census tract data to examine neighborhood effects of disorder on individual outcomes of perceived powerlessness within small city and rural locations. A shortcoming of their methodological approach employed OLS estimates rather than HLM. The use of OLS was a consequence of insufficient cases required to perform a reliable multilevel analysis. For example , approximately two- thirds of the census tracts contained only one respondent. While using OLS, perhaps, was appropriate under the circumstances of their study, by' not using’ HLM they' could. not generate reliable estimates of the variance within 185 30. 31. 32. tracts. Like Geis and Ross, Robert (1998) utilized a similar multilevel model and used OLS regression using SUDAAN software. Her multilevel study examined whether community level predictors such as percent receiving public aid and percent unemployed effected individual outcomes of diseases, disability, and subjective health over and above individual level predictors of age, race, sex, assets, and education. SUDAAN served to adjust for standard error coefficients using a Taylor series linearization method rather than HLM because the survey sampled only a few respondents within each census community. Reliability is a function of sample size in each of the RUs and the proportion of the total variance that is between RUs relative to the amount that is within RUs (Sampson et al., 1999:642) The ICC is computed as the ratio of the amount of variation between RUs relative to the amount of variation within them, plus measurement error, plus any statistical interactions between neighborhood and individual difference attribute (Cook et al., 1997:107). Note that much of the variation associated with quality of life (e.g., fear of crime .50) was due to 186 33. differences among citizens within the same RU. However, research has shown that small variance between aggregate units resulted in a large effect associated with differences between aggregates (Sampson. and. Jeglum-Bartusch, 1998:794-799; Sampson et al., 1999:641—642). The multivariate models empirically revealed that collective efficacy has the capability to reduce quality of life assessments, however, it is important to note that such models are “under-specified." In other words, there exists the possibility that other variables, not included in the model, influence citizen quality of life assessments, such as local friendship networks, organizational participation, and other social processes variables (Sampson and Groves, 1989; see also Kornhauser, 1978; Kasarda and Janowitz, 1974). However, additional variables are not of theoretical interest. Instead, the theoretical interest emphasized the salience of social cohesion and informal social control through collective efficacyu Moreover, variables such as local friendship networks pertain to social cohesion and fail to account for informal social. 187 34. 35. To address the concern about the attenuation of the effects of collective efficacy when burglary index was entered into the mixed model, I estimated additional hierarchical models for each of the quality of life outcomes. Ihi this model, I controlled for citizen—level variables and included only burglary index at the residential unit-level. Next, I compared the mixed model with the crime model estimates to determine the degree of attenuation. For perceived crime, the coefficient for burglary index when compared to the mixed model was reduced from .14 to .11. The burglary index coefficient for fear of crime (from .08 to .07), perceived incivility (from .04 to .03), and risk of victimization (from .05 to .04) revealed minor reductions. Overall, evidence suggests some degree of attenuation. Several studies have drawn on the concept of social capital to explain the effects of negative social capital (e.g., exclusion of outsiders) and its negative consequences (e.g., inadequate education and poor police service) in urban neighborhoods (Portes, 1998; Wacquant, 1998; Hagan and Coleman, 2001; Laub and Sampson; 1993; Sampson and Lauritsen, 1994; Sampson, 1995). 188 36. 37. 38. O’Conner (1973) associated social capital with state and federal expenditures. Coleman (1988a; 1988b) argues that economic rationality is inextricably tied to social capital. The ARD set the stage for additional governmental initiatives, such as the 1964 Economic opportunity Act, the 1965 Appalachian Regional Commission, and the Economic development Act of 1965, aimed at economic development in rural areas experiencing high rates of poverty and unemployment (Summers, 1986) . The development of economic capital continued, and in 1972 the Rural Development Act and the Rural Development Policy Act of 1980 (Summers, 1986) allocated more money to rural America. For example, Madison (1986) argues that economic development, improved educational systems, and low crime rates are the result of government intervention. Wacquant (1998) has noted that federal policies (e.g., economic capital) are responsible for promoting services (e.g., physical safety, education, welfare, healthcare, etc.), which in turn, translate into improved social capital (see also Sampson, 1995). 189 39. 40. Weisheit et al’s. (1994) study, which included qualitative interviews with 46 rural and 28 rural police chiefs, suggested. that 'modern. (community policing draws heavily on ideas and practices that have long been traditions in nonmetropolitan areas. It is relatively common for rural police officers to know citizens personally, have frequent face to face contact with them, and engage in a variety of problem-solving activities that fall outside law enforcement (Weisheit et al., 1994:550). Research suggests that nonmetropolitan police tend to go above and. beyond crime fighting and. offer a 'variety' of services, which in turn, encourages citizen participation in problem solving (Flanagan, 1985; Decker, 1979; Gibbons, 1972; Meagher, 1985). The co- production perspective involving police-community partnership toward crime prevention (Skogan, 1989), can lead to social capital. The logic is that relations among persons, especially criminal justice personnel, can facilitate action. The evidence presented in this dissertation showed that Michigan residential units were comprised of 97% Caucasian. Because of this racial homogeneity, community residents are likely to share similar norms 190 41. 42. 43. 44. and values, thereby making it easier to convert social capital into collective efficacy. Moreover, studies show that nonmetropolitan citizens are initiators in helping police solve social problems (Brandle en: al., 1994; Decker, 1979; Meagher, 1985). However, this approach may be difficult given the lack of technology and sophistication in collection of crime data in nonmetropolitan areas. An informed community that mobilizes (e.g., citizen patrols) for the purpose of providing additional surveillance may help build collective efficacy and lead to the formation of new social ties (i.e., social capital) by increasing local awareness regarding crime. By providing local awareness and accurate information about crime, citizens can gain a sense of what is going on and make sensible decisions about participating in social control activities (Rohe, 1985; Rohe and Gates, 1985). In other words, the social structural origins of crime limit social networking (i.e., social capital) because citizens are apprehensive about quality of life, which ultimately discourages getting involved in community-wide participation. 191 45. Wall et al. (1998:311) argues that “social capital, as an aspect of social organization—trust, norms, and networks—persists in the long run and re-asserts it self under suitable circumstances.” In other words, social capital can increase with use and diminish with disuse, allowing for either virtuous or vicious cycles (Wall et al., 1998). Nevertheless, “working together is easier in a community blessed with a substantial stock of social capital" (Putnam, 1993:35-36). 192 B I BL I OGRAPHY Albrecht, D. E., Albrecht, C. M., & Albrecht, S. L. 2000. Poverty in nonmetropolitan America: Impacts of industrial, employment, and family structure variables. Rural Sociology 65:87-103. Amato, P. R. 1993. Urban-rural differences in helping Friends and family members. Social Psychology Quarterly 50:249-262. Aneshenal, C. S., & Sucoff, C. A. 1996. The neighborhood context of adolescent mental health. Journal of Health and Social Behavior 37:293-310. Baba, Y., & Austin, D. M. 1989. Neighborhood environmental satisfaction, victimization, and social participation as determinants of perceived neighborhood safety. Environment and Behavior 21:763- 780. Bachman, R. 1992. Crime in nonmetropolitan America: A National accounting of trends, incidence rates, and idiosyncratic vulnerabilities. Rural Sociology 57:546-560. Ball, C. 2001. Rural perceptions of crime. Journal of Contemporary Criminal Justice 1:37-48. Bandura, A. 1986. Social Foundations of Thought and Action. Englewood Cliffs, NJ: Prentice-Hall. Bandura, A. 1995. Self-efficacy in Changing Societies. Stanford University: Cambridge University Press. Bankston, W. B., Jenkins, Q. A. L., Thayer-Doyle, C. L., & Thompson, C. Y. 1987. Fear of criminal victimization and residential location: The influence of perceived risk. Rural Sociology 52:98-107. Beggs, J. J., Haines, V. A., Hurlbert, J. S. 1996. Revisiting the rural—urban contrast: Personal networks in nonmetropolitan and metropolitan settings. Rural Sociology 61:306-325. 193 Bell, M. M. 1992. The fruit of difference: The rural- urban continuum as a system of identity. Rural Sociology 57:65-82. Bellair, P. E. 1997. Social interaction and community crime. Examining the importance of neighbor networks. Criminology 35:677-701. 2000. Informal surveillance and street crime: A complex relationship. Criminology 38:137-170. Belyea, M. J., & Zingraff, M. T. 1988. Fear of Crime and Residential Location. Rural Sociology 53:473-486. Berry, W. D., & Feldman, S. 1985. Multiple Regression in Practice. Newbury Park: Sage. Blalock, H. M. 1984. Contextual-effect models: Theoretical and methodological issues. Annual Review of Sociology 10:353-372. Boardman, J. D., & Robert, S. A. 2000. Neighborhood socioeconomic status and perceptions of self-efficacy. Sociological Perspectives 43: 117-136. Brandl, S. G. 1994. Global and specific attitudes toward the police: Disentangling the relationship. Justice Quarterly 1:121-134. Brantingham, P. L., & Brantingham, P. J. 1999. A theoretical model of crime hot spot generation. Studies on Crime and Crime Prevention 1:7-26. Bryk, A. S., & Raudenbush, S. W. 1992. Hierarchical Linear Models: Applications and Data Analysis Methods. Newbury Park: Sage Publications. Bursik, R. J., Jr. 1988. Social disorganization and theories of crime and delinquency: Problems and prospects. Criminology 26:519-552. 1989. Political decision-making and ecological models of delinquency: Conflict and consensus, pp. 65- 83 in Theoretical Integration in the Study of Deviance and Crime, Messner, S. F., Krohn, M. D., & Liska, A. E. (eds.). Albany: State University of New York Press. 194 Bursik, R. J. Jr., & Grasmick, H. G. 1993. Neighborhoods and Crime. New York: Lexington Books. Burton, L. M., Price-Spratlen, T., Spencer, M. B. 1997. On ways of thinking about measuring neighborhoods: Implications for studying context and development outcomes for children, pp. 132-144 in Neighborhood Poverty: Policy Implications in Studying Neighborhoods, Brooks-Gunn, J., Duncan, G. J., & Aber L. J. (eds.). New York: Russell Sage Foundation. Campbell, A. 1981. The Sense of Well-being in America. New York: McGraw Hill. Clark, R., Ekblom, P., Hough, M., & Mayhew, P. 1985. Elderly victims of crime and exposure to risk. Journal of Criminal Justice, 24:1-9. Clemente, F., & Kleiman, M. B. 1977. Fear of crime in the United States: A multivariate analysis. Social Forces 56:519-531. Coleman, J. 1988a. Social capital in the creation of human capital. American Journal of Sociology 94:S95- 120. 1988b. The creation and destruction of social capital: Implications for the Law? Notre Dame Journal of Law, Ethics, and Public Policy 3:375-404. Coleman, J. 1990. Foundations of Social Theory. Cambridge, Mass: Harvard University Press. Cook, T. D., Shagle, S. C., & Degirmencioglu, S. M. 1997. Capturing social process for testing mediational models of neighborhood effects, pp. 94-119 in Neighborhood Poverty: Policy Implications in Studying Neighborhoods, Brooks-Gunn, J., Duncan, G. J., & Aber L. J. (eds.). New York: Russell Sage Foundation.In Brooks-Gunn Conklin, J. E. 1975. The Impact of Crime. New York: Macmillan. Covington, J., & Taylor, R. B. 1991. Fear of crime in urban residential neighborhoods. The Sociological Quarterly 32:231-249. 195 Darling, N., & Steinberg, L. 1997. Community influence on adolescent achievement and deviance, pp.120-131 in Neighborhood Poverty: Policy Implications in Studying Neighborhoods, Brooks-Gunn, J., Duncan, G. J., & Aber L. J. (eds.). New York: Russell Sage Foundation.In Brooks-Gunn Decker, S. 1979. The rural county sheriff: An issue in social control. Criminal Justice Review 4:97-111. Dillman, D. A. 1978. Mail and Telephone Surveys: The Total Design Method. New York: John Wiley & Sons. Dubow, F., McCabe, E., & Kaplan, G. 1979. Reactions to crime: A critical review of the literature. Washington, DC: US Government Printing Office. Duncan, G, & Raudenbush, S. 1999. Assessing the effects of context in studies of child and youth development. Educational Psychologist 34:29-41. Elliot, D. S., Wilson, W. J., Huizinga, D., Sampson, R. J., Elliot, A., & Rankin, B. 1996. The effects of neighborhood disadvantage on adolescent development. The Journal of Research in Crime and Delinquency 33:389-405. Erskine, H. 1974. The polls: Fear of violence and crime. Public Opinion Quarterly 38:131-145. Ferraro, K. F., & LaGrange, R. I. 1987. The measurement of fear of crime. Sociological Inquiry 57:70-101 Fischer, C. S. 1982. To Dwell among Friends: Personal Networks in Town and City. Chicago: University of Chicago Press. Flanagan, T. J. 1985. Consumer perspectives on police operational strategy. Journal of Police Science and Administration 13:10-21. 196 Fox, J. 1995. How does civil society thicken? The political construction of social capital in rural Mexico. “Paper presented to Government Action, Social Capital Formation and Third World Development, a conference sponsored by the Economic Development Working Group, American Academy of Arts and Sciences, Cambridge, Massachusetts, May 5-6. Fruedenburg, W. R. 1986. The density of acquaintanceship: An overlooked variable in community research? American Journal of Sociology 92:27-63. Furstenburg, F. F., Jr. 1971. Public reactions to crime in the streets. American Scholars 40:601-610. Galpin, C. J. 1915. Social Anatomy of an Agricultural Community. University of Wisconsin Agricultural Experiment, Research Bulletin No. 34, Madison, Wisconsin. Garofalo, J. 1979. Victimization and the fear of crime. Journal of Research and Crime and Delinquency 16:80- 97. Garofalo, J., & Laub, J. 1978. The fear of crime: The broadening of our perspective. Victimology 3:242-253. Gates, L. B., & Rohe, W. M. 1987. Fear and reactions to crime a revised model. Urban Affairs Quarterly 22:425-453. Geis, K. J., & Ross, C. E. 1998. A new look at urban alienation: the effect of neighborhood disorder on perceived powerlessness. Social Psychology Quarterly 61:232-246. Gibbons, D. C. 1972. Crime in the hinterland. Criminology 10:177-191. Gordon, M. T., & Riger, S. 1980. Fear and avoidance: A link between attitudes and behavior. Victimology 4:395-402. Greenburg, S. 1983. External solutions to neighborhood- based problems: The case of community crime prevention. Paper presented at the Annual Meeting of the Law and Society Association, June. 197 Greenburg, S., & Rohe, W. M. 1984. Neighborhood design and crime. Journal of the American Planning Association 50:48-61. Greenburg, S., Rohe, W. M., & Williams, J. R. 1982. Safe and secure neighborhoods: Physical characteristics and informal territorial control in high and low crime neighborhoods. Washington, DC: U.S. Department of Justice, Community Crime Prevention Division, National Institute of Justice. Greenburg, S., Rohe, W. M., & Williams, J. R. 1985. Informal social control and crime prevention at the neighborhood level: Synthesis and assessment of the research. Washington, DC: U.S. Department of Justice, Community Crime Prevention Division, National Institute of Justice. Hagan, J. & Coleman, J. P. 2001. Returning captives of the American war on drugs: Issues of community and family reentry. Crime and Delinquency 3:352-367. Hartnagel, T. 1979. The perception and fear of crime: Implications for neighborhood cohesion, social activity and community affect. Social Forces 58:176- 190. Hillery, G. A., Jr. 1955. Definitions of community: Areas of Agreement. Rural Sociology 20:111-123. Hindelang, M. J., Gottfredson, M. R., & Garofalo, J. 1978. Victims of Personal Crime: An Empirical Foundation for a Theory of Personal Victimization. Cambridge: Ballinger Press. Hofferth, S. L., & Iceland, J. 1998. Social capital in rural and urban communities. Rural Sociology 63:574- 598. Hollingshead, A. B. 1948. Community research: Development and present conditions. American Sociological Review 13:135-156. Hooks, G. M., & Flinn, W. L. 1981. The country life commission and early rural sociology. The Rural Sociologist 1:95-100. 198 Hough, M. 1987. Offenders choice of targets: Findings from the victims’ survey. Journal of Quantitative Criminology 3:355-369. Hunter, A. 1974. Community change: A stochastic analysis of Chicago’s local communities, 1930-1960. American Journal of Sociology 79:923-947. Hunter, A. 1978. Symbols of incivility. Paper presented at the Annual Meeting of the American Society of Criminology. Dallas, TX. Jeffrey, C. R. 1977. Crime Prevention Through Environmental Design. Beverly Hills, CA: Sage. Kasarda, J., & Janowitz, M. 1974. Community attachment in mass society. American Sociological Review 39:328- 339. Kelling, G. L., & Coles, C. M. 1996. Fixing Broken Windows: Reducing Order and Reducing Crime in Our Communities. New York: Touchstone. Kennedy, L. W., & Krahn, H. 1984. Rural-urban origin and fear of crime: The case for “Rural Baggage." Rural Sociology 49:247-260. Kim, 0. J., & Mueller, C. W. 1978. Introduction To Factor Analysis: What It Is and How To Do It. Newbury Park: Sage Publications. Kotlowitz, A. 1991. There Are No Children Here. New York: Doubleday. Kornhauser, R. 1978. Social Sources of Delinquency: An Appraisal of Analytical Models. Chicago, IL: University of Chicago Press. Kowalski, G. S., & Duffield, D. 1990. The impact of the rural population component on homicide rates in the United States: A county-level analysis. Rural Sociology 55:76-90. Krannich, R. S., Berry, E. H., & Greider, T. 1989. Fear of crime and rapidly changing rural communities: A longitudinal analysis. Rural Sociology 54:195-212. 199 Kreft, I., & De Leewu, J. 1998. Introducing Multilevel Modeling. Thousand Oaks, CA: Sage Publications. LaGrange, R. L., & Ferraro, K. F. 1989. Assessing age and gender differences in perceived risk and fear of crime. Criminology 27:697-719. LaGrange, R. L., Ferraro, K. F., & Supancic, M. 1992. Perceived risk and fear of crime: Role of social and physical incivilities. Journal of Research in Crime and Delinquency 29:311-334. Land, K. C., McCall, P. L., Cohen, L. E. 1990. Structural covariates of homicide rates: Are there any invariances across time and social space. American Journal of Sociology 95:922-963. Laub, J. H. 1983. Patterns of offending in urban and rural areas. Journal of Criminal Justice 11:129-142. Laub, J. H. & Sampson, R. J. 1993. Turning points in the life course: Why change matters to the study of crime. Criminology 31:301-25. Lavrakas, P. J., Normoyle, J., Skogan, W. G., Herz, E. J., & Lewis, D. A. 1980. Factors related to citizen involvement in personal, household, and neighborhood anti-crime measures: An executive summary. Evanston, IL: Northwestern University Citizen Participation and Community Crime Prevention Project, Center for Urban Affairs. Lawton, M. P., & Yaffe, S. 1980. Victimization and fear of crime in elderly public housing tenants. Journal of Gerontology 35:768-779. Lewis, D. A., & Riger, S. 1986. Crime and Stress: On the internalization of a social problem, pp. 26-53 in Redefining Social Problems, Seidman, E, & Rappaport, J. (eds.). New York: Plenum Press. Lewis, D. A., & Salem, G. 1981. Community crime prevention: An analysis of a developing strategy. Crime and Delinquency 27:405-421. 200 Lewis, D. A., & Salem, G. 1986. Fear of Crime: Incivility and the Production of a Social Problem. New Brunswick, NJ: Transaction. Loomis, C. P. 1939. Informal social participation in the planned rural communities. Sociometry 2:1-37. Loomis, C. P., & Davidson, D. M. 1939. Social agencies in the planned rural communities. Sociometry 2:24-42. Luckenbill, D., & Doyle, D. 1989. Structural position and violence: Developing a cultural explanation. Criminology 27:419-436. Lynch, J. P., & Cantor, D. 1992. Ecological and behavioral influences on property victimization at home: Implications for opportunity theory. Journal of Research in Crime and Delinquency 29:355-362. Maccoby, E. E., Johnson, J. P, & Church, R. M. 1958. Community integration and the social control of juvenile delinquency. Journal of Social Science 14:38-51. Macintyre, S., MacIver, S., & Sooman, A. 1993. Area, class and health: Should we be focusing on places of people? Journal of Social Policy 22:213-234. Madison, J. H. 1986. Reformers and the rural church, 1900-1950. The Journal of American Histopy 73:645- 668. Markowitz, F. E., Bellair, P. E., Liska, A. E., & Liu, Jianhong. 2001. Extending social disorganization theory: Modeling the relationships between, cohesion, disorder, and fear. Criminology 39:293-319. Martinez, R. Jr. 1996. Latinos and lethal violence: The impact of poverty and inequality. Social Problems 2:131-146. Massey, D. S. 1996. The age of extremes: Concentrated affluence and Poverty in the Twenty-First-Century. Demography 33:995-412. 201 Maxfield, M. G., & Babbie, E. (1998). Research Methods for Criminal Justice and Criminology. New York: Wadsworth. Medoff, P, & Sklar, H. 1994. Streets of Hope: The Fall and Rise of an Urban Neighborhood. Boston, Mass: South End Press. Menard, S. 1995. Applied Logistic Regression Analysis. Thousand Oaks, CA: Sage Publications. McDowell, D., Loftin, C., & Wersima, B. 1989. Multi-level risk factors for burglary victimization. Paper presented at the Annual Meetings of the American Society of Criminology, Reno, NV, Novenmber. Meagher, M. S. 1985. Police patrol styles: How pervasive is community variation? Journal of Police Science and Administration 13:36-45. Michalos, A. C., & Zumbo, B. D. 2000. Criminal victimization and the quality of life. Social Indicators Research 50:245-295. Miller, L. S., & Whitehead, J. T. 1996. Introduction to Criminal Justice: Research and Statistics. Ohio: Anderson Publishing. Mok, M., & Flynn, M. 1998. Effects of catholic school culture on students’ achievement in the higher school certificate examination: A multilevel path analysis. Educational Psychology 18:409-432. Newman, 0. 1972. Defensible Space. New York: Macmillan. Norris, F. H., & Kaniasty, K. 1991. The psychological experience of crime: A test of the mediating role of beliefs in explaining the distress of victims. Journal of Social and Clinical Psychology 10:239-261. O'Conner, J. 1973. The Fiscal Crisis of the State. New York: St. Martin’s Press. Ortega, S. T., & Myles, J. L. 1987. Race and gender effects of fear of crime: An interactive model with age. Criminology 25:133-152. 202 Osgood, D. W., & Chambers, J. M. 2000. Social disorganization outside the metropolis: An analysis of rural youth violence. Criminology 38:81-116. Park, R. E., & Burgess, E. W. 1925. The City. Chicago: University of Chicago Press. Park, R. E., Burgess, E. W., & McKenzie, R. D. 1928. The City. Chicago: The University of Chicago Press. Parker, K. D., & Ray, M. C. 1990. Fear of crime: An Assessment of related factors. Social Spectrum 10:29- 40. Perkins, D. D., Florin, P., Rich, R. C., Wandersman, A., & Chavis, D. M. 1990. Participation and the social and physical environment of residential blocks: Crime and community context. American Journal of Community Psychology 1:83-115. Peterson, R. D., Krivo, L. J., & Harris, M. A. 2000. Disadvantage and neighborhood violent crime: Do local institutions matter? The Journal of Research in Crime and Delinquency 37:32-63. Podolsfsky, A. 1983. Case Studies in Community Crime Prevention. Springfield, IL: Charles C. Thomas. Podolsfsky, A, & Dubow, F. 1980. Strategies for Community Crime Prevention: Collective Responses to Crime in Urban America. Reactions to Crime Project, center for Urban Affairs, Northwestern University, Springfield, Ill.: C.C. Thomas. Portes, A. 1998. Social capital: Its origins and applications in modern sociology. Annual Review of Sociology 24:1-24. Putnam, R. 1993. The prospectus community: Social capital and public life. American Progpect, Spring:35-42. 1995. Bowling alone: America’s declining social capital. Journal of Democracy 1:65-78. Rabrenovic, G. 1996. CommunityyBuilders: A Tale of Neighborhood Mobilization in Two Cities. Philadelphia: Temple University Press. 203 Reid, 8. T. 2000. Crime and Criminology. New York: McGraw Hill. Reisig, M. D., & Holtfreter, K. L. 2000. Participation in rural crime prevention: A multilevel analysis. Presented at the 37fl‘Annual Meeting of the Academy of Criminal Justice Sciences, New Orleans, LA. Reiss, A. J., Jr. 1973. Monitoring the Quality of Criminal Justice Systems, pp. 388-403 in The Human Meaning of Social Change A. Campbell and P. Converse (eds.). New York: Russell Sage. Reppetto, A. 1974. Residential Crime. Cambridge, MA: Ballinger. Robert, S. A. 1998. Community-level socioeconomic status effects on adult health. Journal of Health and Social Behavior 39:18-37. Rohe, W. M. 1985. Urban planning and mental health. Prevention in Human Services 4:79-110. Rohe, W. M., & Gates, L. B. 1985. Planningywith Neighborhoods. Chapel Hill: University of North Carolina Press. Rooney, J. 1995. Organizing the South Bronx. Albany: State University of New York Press. Ross, C. E. 1993. Fear of victimization and health. Journal of Quantitative Criminology 9:159-175. Ross, C. E. 2000. Neighborhood disadvantage and adult depression. Journal of Health and Social Behavior 41:177-187. Ross, C. E., & Mirowsky, J. 1999. Disorder and decay: The concept and measurement of perceived neighborhood disorder. Urban Affairs Review 34:412—432. Rountree, P. W., & Land, K. C. 1996. Perceived risk versus fear of crime: Empirical evidence of conceptually distinct reactions in survey data. Social Forces 74:1353-1376. 204 Rountree, P. W., Land, K. C., & Miethe, T. D. 1994. Macro-micro integration in the study of victimization: A hierarchical logistic model analysis across Seattle neighborhoods. Criminology 32:387-405. Saltiel, J., Gilchrist, J., & Harvie, R. 1992. Concern about crime among Montana farmers and ranchers. Rural Sociology 57:535-545. Sampson, R. J. 1986. Effects of socioeconomic context on official reaction to juvenile delinquency. American Sociological Review 51:876-885. 1988. Local friendship ties and community attachment in mass society: A multilevel systemic model. American Sociological Review 53:655-779. 1992. Family management and child development: Insights from social disorganization theory, pp.25 in Advances in Criminological Theory: Facts, Frameworks, and Forecasts, Vol. 3, McCord, J. (ed.). New Brunswick, NJ: Transaction. 1995. The Community, pp. 193-216 in Crime, Wilson, J. Q., & Petersilia, J. (eds.). ICS Press: San Francisco. Sampson, R. J., & Groves, W. B. 1989. Community structure and crime: Testing social-disorganization theory. American Journal of Sociology 94:774-802. Sampson, R. J., & Jeglum-Bartusch, D. 1998. Legal cynicism and (subcluture?) tolerance of deviance: The neighborhood context of racial differences. Law and Society Review 32:777-804. Sampson, R. J., & Lauritsen, J. L. 1993. Violent victimization and offending: Individual-situational- ,and community-level risk factors, pp.1-114 in Understanding and Preventing Violence, Vol. 3, Reiss, A. J. Jr., & Roth, J. A. (eds.). Washington, D. C.: National Academy Press. 205 Sampson, R. J., Morenoff, J. D., & Earls, F. 1999. Beyond social capital: Spatial dynamics of collective efficacy for children. American Sociological Review 64:633-660. Sampson, R. J., & Raudenbush, S. W. 1999. Systemic social observation of public spaces: A new look as disorder in urban neighborhoods. American Journal of Sociology 105:603-651. Sampson, R. J., Raudenbush, S. W., Earls, F. 1997. Neighborhood and violent crime: A Multilevel study of collective efficacy. Science 277:918-924. Sampson, R. J., & Wilson, W. J. 1995. Toward a theory of race, crime, and urban inequality, pp.37-54 in Crime and Inequality, Hagan, J., & Peterson, R. D. (eds.). Stanford, CA: Stanford University Press. Sampson, R. J., & Wooldredge, J. D. 1987. Linking the micro-and macro-level dimensions of lifestyle-routine activity and opportunity models of predatory victimization. Journal of Quantitative Criminology 3:371-393. Sandefur, R., & Laumann, E. O. 1998. A paradigm for social capital. Rationality and Society 10:481-501. Sanders, I. T., & Ensminger, D. 1940. Alabama rural communities: A study of Chilton county. Montevallo, Alabama: Daily Press. Sanders, I. T., & Lewis, G. F. 1976. Rural community studies in the United States: A Decade in review. Annual Review of Sociology 2:35-53. Sanderson, D. 1927. Scientific research in rural sociology. The American Journal of Sociology 2:177- 193. Shaw, C. R., & McKay H. D. 1942. Juvenile Delinquency and Urban Areas: A Study of Rates of Delinquency in Relation to Differential Characteristics of Local Communities in American Cities. Chicago: University of Chicago Press. 206 Shaw, C. R., & McKay H. D. 1969. Juvenile Delinquency and Urban Areas: A Study of Rates of Delinquency in Relation to Differential Characteristics of Local Communities in American Cities. Chicago: University of Chicago Press. Short, J. F., Jr. 1990. Delinquency and Society. Englewood Cliffs, NJ: Prentice Hall. Silver, E. Extending social disorganization theory: A multilevel approach to the study of violence among persons with mental illness. Criminology 4:1043-1074. Simcha-Fagan, 0., & Schwartz, J. E. 1986. Neighborhood and delinquency: An assessment of contextual effects. Criminology 24:667-703. Skogan, W. 1990. Disorder and Decline: Crime and the Spiral of Decay in American Cities. New York: Free Press. 1991. Disorder and Decline. New York: Free Press. 1986. Fear of crime and neighborhood change, pp. 203-299 in Communities and Crime, in Reiss, A. J., Jr., & Tonry, M., (eds.). Chicago: University of Chicago Press. 1989. Communities, crime and neighborhood organization. Crime and Delinquency 3:437-457. 1999. Measuring what matters: Crime, disorder, and fear. Measuring What Matters: Proceedings From the Policing Research Institute Meetings, pp. 37-54. Washington, DC: National Institute of Justice. Skogan, W. G., & Hartnett, S. M. 1997. Community Policing, Chicago Style. New York: Oxford University Press. Skogan, W. G., & Maxfield, M. G. 1981. Coping With Crime: Individual and Neighborhood Reactions. Beverly Hills: Sage Publications. 207 Smith, T. L. 1941. The role of the community in American rural life. Journal of Education Sociology 14:387- 400. Smith, B. L., & Huff, C. R. 1982. Crime in the country: The vulnerability and victimization of rural citizens. Journal of Criminal Justice 10:271-282. Smith, D. A., & Jarjoura, G. R. 1989. Household characteristics, neighborhood composition and victimization risk. Social Forces 68:621-640. Snodgrass, J. 1976. Clifford R. Shaw and Henry D. McKay: Chicago Criminologists. The British Journal of Criminology 16:1-19. Sooman, A., & Macintyre, S. 1995. Health and perceptions of the local environment in socially contrasting Neighborhoods in Glasgow. Health and Place 1:15-26. Sorokin, P., Zimmerman, C. C., & Galpin, C. J. 1930. A Systemic Source Book in Rural Sociology. Minneapolis, MN: University of Minnesota Press. Stephens, D. W. 1999. Measuring What Matters: Proceedings From the Policing Research Institute Meetings, pp. 55- 56. Washington, DC: National Institute of Justice. Stinner, W. F., Loon, M. V., Chung, She-Woong, & Byun, Y. 1990. Community size, individual social position, and community attachment. Rural Sociology 55:494-521. Summers, G. F. 1986. Rural community development. Annual Review of Sociology 12:347-371. Suttles, G. 1986. The Social Order of the Slum. Chicago: University of Chicago Press. Taub, R. G., Taylor, D., & Dunham, J. D. 1980. Crime, ear of crime and the deterioration of urban neighborhoods. Chicago: National Opinion Research Center. Taub, R. G., Taylor, D., & Dunham, J. D. 1984. Paths of Neighborhood Change: Race and Crime in Urban America. Chicago: University of Chicago Press. 208 Taylor, C. C. 1927. Research in rural sociology. America Journal of Sociology 2:211-221. Taylor, R. B. 1996. Neighborhood responses to disorder and local attachments: The systemic model of attachment, social disorganization, and neighborhood use value. Sociological Forum 11:41-57. Taylor, R. B. 1999. The incivilities thesis: Theory, measurement, and policy. Measuring What Matters: Proceedings From the Policing Research Institute Meetings, pp. 65-88. Washington, DC: National Institute of Justice. Taylor, R. B., & Covington, J. 1993. Community structural change and fear of crime. Social Problems 40:375-395. Taylor, R. B., Gottfredson, S. D., & Brower, S. 1984. Block crime and fear: Defensible space, local social ties, and territorial functioning. Journal of Research in Crime and Delinquency 21:303-331. Taylor, R. B., Gottredson, S., Brower, S., Drain, W., & Dockett, K. 1979. Toward a residential based model of community crime prevention: Urban territoriality, social networks and design. Baltimore, MD: Center for Metropolitan Planning and Research, Johns Hopkins University. Taylor, R. B., & Hale, M. M. 1986. Testing alternative models of fear of crime. Journal of Criminal Law and Criminology 77:151-189. Taylor, R. B., & Perkins, D. D. 1994. The impact of block level fear and citizen participation on changes in anxiety and depression: A stress and coping framework. Unpublished manuscript. Taylor, R. B., & Shumaker, S. 1990. Local crime as a natural hazard: Implications for understanding the relationship between disorder and fear. American Journal of Community Psychology 18:619-641. Thomas, W. I., & Znaniecki, F. 1927. The Polish Peasant in Europe and America. New York: Dover. 209 Tickamyer, A. R., & Duncan, C. M. 1990. Poverty and opportunity structure in rural America. Annual Review of Sociology 16:67-86. Velez, M. B. 2001. The role of public social control in urban neighborhoods: A multilevel analysis of victimization risk. Criminology 4:837-863. Vold, G. B., Bernard, T. J., & Snipes J. B. 1998. Theoretical Criminology. New York: Oxford University Press. Wacquant, L. J. D. 1998. Negative social capital: State breakdown and social destitution in America’s urban core. Netherland Journal of Housing and the Built Environment 13:25-40. Wall, E., Ferrazzi, G., Schryer, F. 1998. Getting-the goods on social capital. Rural Sociology 2:300-322. Warner, J. R. 1981. Rural crime, rural criminals, rural delinquents: Past research and future directions, pp. 35-44 in Juvenile Justice in Rural America, Jankovic, J. (ed.). Washington, DC: Law Enforcement Assistance Administration. Warner, B. D., & Rountree, P. W. 1997. Local social ties in a community and crime model: Questioning the systemic nature of informal social control. Social Problems 44:520-536. Weisheit, R. A, Wall, E. L., & Falcone, D. N. 1994. Community policing in small town and rural America. Crime and Delinquency 4:549-567. Welsh, W. N., Greene, J. R., & Jenkins, P. H. 1999. School disorder: The influence of individual, institutional, and community factors. Criminology 37:73-115. Whyte, W. F. 1943. Street Corner Sociepy: The Social Structure of an Italian Slum. Chicago: University of Chicago Press. Wilkinson, K. P. 1984. Rurality and patterns of social disruption. Rural Sociology 49:23-36. 210 Wilson, W. J. 1987. The Truly Disadvantaged: The Inner City, the Underclass, and Public Policy. The University of Chicago Press: London. Wilson, J. Q., & Kelling, G. 1982. Broken windows. Atlantic Monthly 211:29-38. Yin, R. 1977. Citizen Patrol Projects. Wshington, D.C.: U.S. Department of Justice. 211