visage €g$§£t ?‘ ( ru- e,‘ _ .& . 3? mam e 9%“ i£é~W¢ 5535:: 5:“, u} 1‘15“ J.‘ _v ' 3‘33: f t. ‘5‘, gang. . b \..u’ .4“. m a: V -=e ‘v- ’- .. ,55’4’?‘ .‘unl v NW. “gum." M: “a: .W 1‘" Eight” ‘7‘” 755%}er "$.31; '21: MICHIGAN STATE UNIVERSUTY LIBRARIES“ lllllll ll l l Ll llllllll \lllll ll ll‘ \ 3 1293 00759 728 LIBRARY Michigan State University This is to certify that the thesis entitled REACTIONS TO CRIME: THE RELATIONSHIP BETWEEN ATTITUDINAL AND BEHAVIORAL RESPONSES TO CRIME presented by Talbert J. Cottey has been accepted towards fulfillment of the requirements for Master of Sciencedegree in Criminal Justice fay/79m Major professor 0-7639 MS U is an Affirmative Action/Equal Opportunity Institution - —_.-—-—_—_.——-—u— .— 4 . .—-_4— v44 PLACE IN RETURN BOX to remove thIe checkout from your record. TO AVOID FINES return on or before due due. DATE DUE ‘ DATE DUE DATE DUE .' [TB 0 531%» ' ‘ "FL—A 3 r - I. ‘ \I i_ l w \ ‘ MSU Is An Affirmdive AdIoNEquel OpportunIty Institution ammo-9.1 REACTIONS TO CRIME: TEE RELATIONSHIP BETWEEN ATTITUDINAL AND BEHAVIORAL RESPONSES TO CRIME BY Talbert J. Cottey A THESIS Submitted to Michigan State University in partial fulfillment of the requirements for the degree of MASTER OF SCIENCE School of Criminal Justice 1989 . i) (1) '5' t (00019 ABSTRACT REACTIONS TO CRIME: THE RELATIONSHIP BETWEEN ATTITUDINAL AND BEHAVIORAL RESPONSES TO CRIME BY Talbert J. Cottey This study investigates the relationship between fear of crime and crime prevention action. Although much research has investigated what people feel and do in response to the threat of crime, there is an unfortunate dearth of comprehensive research examining the relationship between attitudinal and behavioral reactions to crime. In order to clarify this issue, this research posits two explanatory models. A personal crime versus property crime distinction is drawn between models. Variables in these models include crime prevention actions, attitudes about crime, individual characteristics, and residential characteristics. Path analysis was used to assess the direct and indirect effects of all variables. Gender, and income were the strongest predictors of attitudinal and behavioral reactions to crime. As if in a vacuum, reactions to crime were relatively unaffected by past experiences, the environment, or prevailing social conditions. Findings indicated that attitudes about crime have a strong and consistent relationship with crime prevention behavior. ACKNOWLEDGEMENTS I would like to express my sincere gratitude to Dr. Timothy Bynum for the guidance and wealth of knowledge he has provided. I am in debt to Dr. Vincent Hoffman for his many hours of valuable advice, to Dr. Robert Worden for his statistical expertise, and to Dr. Phyllis Kayten for her assistance while in Washington, DC. I would like also to thank Fred Giroux and Peter Belmore for their help in preparation of this manuscript. I am in debt to Sue for her love and patience. I wish lastly to thank my parents for instilling in me the value of education and providing nothing less than complete support for my academic endeavors. iii Section LIST OF LIST OF CHAPTER CHAPTER CHAPTER CHAPTER TABLE OF CONTENTS TABI‘ES O O O O O O O O O O O O O FIGIIRES O I O O O O O O O O O O I 1: INTRODUCTION . . . . . . . . 2: LITERATURE REVIEW . . . . . Introduction . . . . . . . . Attitudinal Reactions to Crime Measuring Fear . . . . . . . . Measuring Behavioral Reactions to Crime and Preventive Behavior . . Towards an Attitude-Behavior Relat' Development of Research Questions Research Questions . . . . . Subsidiary Questions . . . . Summary . . . . . . . . . . . . . 3: METHODS . . . . . . . . Introduction . . . . . . . Data Set . . . . . . . . . Neighborhood Selection Sampling . . . . . . . Data Source . . . . . Measurement of Variables . . Crime Prevention Behaviors Attitudes About Crime . . Individual Characteristics Residential Characteristics Analysis . . . . . . . . . . . . 4: RESULTS . . . . . . . . . . Introduction . . .‘. . . . Results from Cross Tabulations Fear for Person . . . . . Action to Protect Person . Concern for Property . . . Action to Protect Property . 00003-000000 oooogo H o o o o E (D o o o 0 Results from the Zero-Order Correlation Matrix 0 O O O O O O I O O 0 iv Page vi vii CHAPTER LIST OF Results from Path Analysis . Path Analysis for Action to Protect Person ...... Direct and Indirect Effects on Action to Protect Person . Path Analysis for Action to Protect Property . . . . . Direct and Indirect Effects on to Protect Property Overview of Path Analyses . . 5: DISCUSSION AND CONCLUSION Introduction . . . . . . . The Attitude-Behavior Nexus Individual Characteristics Income . . . . . . . . Age and Gender . . . . Victimization . . . . Residential Characteristics Neighborhood Crime Rates Social Interaction . . . Conclusion . . . . . . . . . REFERENCES . . . . . . . . . Action 49 51 53 55 57 57 60 6O 60 64 64 66 67 69 69 71 73 76 LIST OF TABLES number Page 1. Amount of Fear for Person by Individual and Residential Characteristics . . . . . . . . . . . 41 2. Amount of Action Taken to Protect Person by Individual, Characteristics, Fear, and Residential Characteristics . . . . . . . . . . . 43 3. Amount of Concern for Property by Individual and Residential Characteristics . . . . . . . . . 45 4. Amount of Action Taken to Protect Property By Individual Characteristics, Concern, and Residential Characteristics . . . . . . . . . . . 47 5. Zero-Order Correlation Matrix for Individual, Attitudinal, Residential, and Behavioral variables 0 I O O O O O O O O O O O O O O O O O O 50 6. Direct and Indirect Effects on Action to Protect Person . . . . . . . . . . . . . . . . . 54 7. Direct and Indirect Effects on Action to Protect Property . . . . . . . . . . . . . . . . 54 vi LIST OF FIGURES Number Page 1. A Model for Attitudinal and Behavioral Reactions to Personal Crime . . . . . . . . . . . 32 2. A Model for Attitudinal and Behavioral Reactions to Property Crime . . . . . . . . . . . 32 3. Path Analysis for Attitudinal and Behavioral Reactions to Personal Crime . . . . . . . . . . . 52 4. Path Analysis for Attitudinal and Behavioral Reactions to Property Crime . . . . . . . . . . . 56 vii . , nlfi’lfi ‘ . l i! a: v .1 1 W - .- r; In. .3. CHAPTER 1 INTRODUCTION People react to the threat of crime by developing attitudes which may lead to engaging in preventive behavior. These responses are usually characterized in terms of fear of crime and crime prevention action. Although there appears to be a logical connection between how people feel about crime and what they do in response to it, previous research has failed to provide evidence supporting such a relationship. Attitudinal and behavioral reactions to crime reflect an important dimension about the extensive impact of crime. Specifically, these responses do not necessarily come from direct experience with crime, but from how people interpret the general threat of crime in an individual, social, and environmental context. In fact, persons who are least likely to encounter crime typically report greater levels of fear and are more likely to engage in crime prevention action (Skogan, 1981; Baumer, 1985). Hence, crime not only adversely affects the victim, but also may impact the lives of potential victims by bringing about attitudinal and behavioral change. ‘11 Unfortunately, substantial cost is associated with this change. Fear becomes an emotional stressor that can strip one's sense of security. Crime prevention actions are also costly. Installing locks, burglar alarms, or other security devices requires a substantial financial investment that has no guarantee of return. In addition to monetary costs, there are also intangible losses resulting from reactions to crime. Many behaviors, such as staying home at night or avoiding public transportation, restrict mobility and force isolation. In sum, these behaviors may reduce the quality of life. Although the negative consequences of attitudinal and behavioral reactions to crime are recognized, the origin of these responses remains vague. The bulk of previous research on reactions to crime is targeted towards the attitudinal component fear, with less emphasis on crime prevention behavior. Additionally, there is a failure to examine a possible relationship between how people feel about crime in terms of fear, and what people do to stop crime in the form of crime prevention behavior. Studies employ an array of conceptualizations and measurements of fear. Correspondingly, researchers use various techniques for conceptualizing and measuring preventive behaviors. These various approaches depict attitudinal and behavioral reactions to crime in an assortment of ways, ranging from single to multidimensional concepts. Yet the validity and practicality of many of these practices has never been tested, thus weakening the value of these studies' findings. Furthermore, these differing methodologies have produced conflicting findings and have also left many questions unanswered. Because fear of crime may affect people's lives independent of actual experiences with crime, an important question is what factors contribute to fear. In general, findings indicate that gender, age, and income are strong predictors of fear (Baumer, 1985). But after more than two decades of research, there is little evidence suggesting that experiences with crime are related to levels of fear. Victimization is consistently found to be a weak predictor of fear (Biderman et al, 1967; Black and Long, 1973; Fowler and Mangione, 1974). Even though research on this issue is voluminous, there is no conclusive resolution as to why experiences with crime do not increase levels of fear. Scholars also voice uncertainty about the effect of social interaction on fear. Much research suggests that social networking or bonding eliminates crime-related anxiety, enhances feelings of security, and provides the means for informal control (Jacobs, 1961: Gubrium, 1974: Henig & Maxfield, 1978). However, recent studies have also indicated that social interaction vicariously magnifies experience with crime and serves as a mechanism for generating greater levels of fear (Skogan, 1977). Hence, .\ DR 56‘ I \p F “on b :0. UN- MU t 1 .13“. as RA... “V6... §-. 3‘. the precise influence that social interaction has on fear is unknown. As with attitudinal reactions to crime, the primary motivator of behavioral reactions is the threat of crime. However, explanation of the effect of crime on behavior is fundamentally divided. Some criminologists envision crime as a shared threat that increases community solidarity and binds people together for collective action (Durkheim, 1933: Cohen, 1966) while others suggest crime separates the community and precipitates the need for individual action (Conklin, 1975; Skogan, 1981). Modern criminological study appears to support a view of crime as being a catalyst for individual action (Skogan, 1981). An underlying assumption of much study is that fear also tends to encourage people to take crime prevention action. Yet researchers have virtually disregarded the relationship between crime-related attitudes and preventive behavior. This void in research, however, may be justified. In sociological research, scholars have traditionally questioned the validity of attitude-behavior relationships. But in a methodological review of attitude-behavior studies, Schuman and Johnson (1976) provide insight to the study of attitude-behavior relationships by suggesting how such research can produce valid results. These findings have important implications for studying a relationship between attitudes of fear and crime prevention behavior. 1.. ‘ 2'"- Haul 6.. ks Most important is that Schuman and Johnson encourage measuring attitudinal and behavioral variables at the same level of specificity. Where the behavior is a single specific act, this means that the attitudinal measure should be specific also, and closely congruent with the act (Schuman & Johnson, 1976, p. 171). Recent research on fear and preventive behavior employs a distinction made according to whether attitudes and behaviors are related to either personal or property crime. By using this distinction and adhering to the additional guidelines of Schuman and Johnson, this research suggests that the relationship between fear and crime prevention action can be successfully examined. This study investigates the relationship between attitudinal and behavioral responses to crime. For explanatory purpose, two models of crime prevention are presented. Although both models provide explanation for attitudinal and behavioral reactions to crime, a personal- property crime distinction is drawn between the two models. The first model provides explanation for attitudinal and behavioral reactions to personal crime. The second model furnishes explanation for attitudinal and behavioral reactions associated with property crime. Since the genesis of fear and crime prevention action is still not clear, individual and residential variables are incorporated into these models. Multiple regression is used to reveal the predictive value of situational and attitudinal variables in the taking of crime prevention action. Path analysis is then used to assess the relationship between attitudinal and behavioral reactions to crime and to determine the direct and indirect effects of all variables. CHAPTER 2 LITERATURE REVIEW Introduction This study examines the relationship between attitudes about crime and crime prevention in an urban environment. Although research has extensively examined fear and crime prevention behavior over the past 20 years, there is still little consensus as to how to conceptualize and measure these reactions to crime. Review of literature sheds light on correlates, previous conceptualizations, and measurements of reactions to crime. Many of these earlier approaches, however, are inconclusive. Therefore, this review also discusses implications for studying attitudinal and behavioral reactions to crime by drawing a personal crime versus a property crime distinction. This review further addresses an interrelationship between fear and action. There are two potential areas of literature that examine responses to crime. The first looks at how people feel about crime. These attitudinal studies usually describe feelings about crime in terms of fear. The second ares of literature examines what people do about the threat of crime. These behavioral responses are understood as crime-prevention actions. It is reasonable to assume a simple relationship between attitudinal and behavioral reactions to crime. Yet there remains an unfortunate dearth of comprehensive research examining the relationship between fear and crime-prevention action. Attitudinal Reactions to Crime Past studies have considered many factors that may contribute to the fear of crime. For example, some of the earliest research on fear and crime revealed an inverse relationship between income and fear (Biderman et al., 1967; Ennis, 1967; Clemente & Kleinman, 1977). Lower income is also strongly associated with higher levels of victimization. Possibly this is because income is a powerful determinant of living conditions. Individuals with less money tend to live in dangerous areas (Baumer, 1985). An important criticism of expressions of fear is that it is an irrational response (Skogan, 1981). This argument stems from the apparent paradox surrounding fear. Persons possessing the greatest level of fear are the least likely to suffer victimization. This situation is particularly true for women and the elderly. Women and the elderly have significantly lower victimization rates than other groups, but consistently report higher levels of fear (Biderman et al., 1967; Conklin, 1975; Garofalo, 1977a; Clemente & Kleinman, 1977; Riger, 1981; Baumer, 1985). The fact that 'I'CZI voli res; uses indi the women and the elderly report higher levels of fear does not imply that measurements of fear are invalid or that these persons are necessarily irrational. Both groups may possibly see themselves as being more vulnerable (Stinchcombe et al., 1978), and/or realize that the impact of victimization may be more severe for them than for others. Skogan and Maxfield (1981) present a cognitive/ volitional model that views fear of crime as a rational response to a perceived threat of harm. This model of fear uses indicators to describe individual characteristics indicative of vulnerability to crime, the threat present in the local environment, and the knowledge that people have of proximate criminal events (Baumer, 1985). Experiences of victimization, high levels of reported crime, and a lack of social interaction are all thought to contribute to the threat and knowledge of criminal activity. Some of the more surprising and controversial findings on fear involve the impact of victimization. Conceptually, fear of crime is often associated with experiences of victimization. Research, however, has consistently found a weak relationship between victimization and fear. Early findings by Biderman et al. (1967), Black and Long (1973), and Fowler-Mangione (1974) did not report a significant difference between victims and nonvictims in their fear of crime. Nor did Dubow et al. (1979) find much evidence 10 indicating that past victimization generates greater fear. Even studies that consider severity and amount of victimization fail to establish anything but a weak association between victimization and fear (Hindelang, Gottfredson & Garofalo, 1978). Sparks, Glenn, and Dodd (1977) even found a negative relationship between fear and victimization by robbery and assault. Research has elicited rigorous debate as to why a weak relationship between victimization and fear has been found repeatedly. Some suggest that the confounding variables of age and sex are responsible. However, when these groups were examined more carefully, there was little evidence indicating a difference in levels of fear between victims and nonvictims (Hindelang et al., 1978). Agnew (1985) argues that the influence of victimization on fear is determined by how the victim interprets the experience. At the center of his argument is the suggestion that victims employ neutralization techniques--i.e., denial of injury or denial of vulnerability--to mediate the impact of victimization. Even though Agnew's findings were convincing, his study has not been replicated. The influence of victimization, in terms of longevity, is also not known. Skogan (1986) states that the effects of victimization may deteriorate over time. Yet there is very little longitudinal research addressing this issue. (n ff '1 f) ’4. J- r' *4. r" 11 Another explanation advanced is that fear is the product of multiple factors. The quantitative consistency of research demonstrates a weak relationship between victimization and fear. Personal experience with crime then is clearly only one determinant of fear (Skogan, 1986). Studies have also considered the relationship between official rates of crime and fear. The use of official crime rates has two purposes. Foremost, crime rates are thought to measure the threat that proximate crime had on fear. Secondarily, local crime rates are used to measure information about crime disseminated to residents. It is suggested that uniform crime reports underestimate the level of crime (O'Brien, 1985). Underestimation comes from the failure of citizens to report incidents of crime, organizational pressures by police agencies to get crime rates down, and the failure of police to follow through on complaints. Hence, questions of validity are raised if one is attempting to measure the influence of real levels of crime on fear. However, uniform crime report data are found to lead to much more accurate estimates of the amount of crime that people are aware of than does victimization data (Lewis & Maxfield, 1980). This is because people are more likely to hear about crime in their city through news reports of uniform crime report figures than any other source of crime data. L: ‘1 t}. 5L 5e [I ’ 12 As with any use of official crime rates, however, several problems exist. Official reports of crime are not always readily available to the public. If reported crime rates are released, they are subject to the distortion and manipulation of mass communications, such as television and newspapers. It is, therefore, difficult for residents to obtain accurate information about the relative danger of their neighborhood (Baumer, 1978). In addition to the aforementioned problems, some findings indicate that the relationship between objective crime rates and fear is not direct. Lewis and Maxfield (1980), report that neighborhood-level crime rates and perceptions of incivility--i.e., conditions indicating that community social control is weak--interact to increase fear. Social networks, in that they provide a sense of community, are also associated with levels of fear (Hunter & Baumer, 1982; Skogan & Maxfield, 1981). Social interaction with neighbors provides emotional support for dealing with the anxieties induced by crime (Gubrium, 1974). Through this emotional bonding, strong social networks enhance feelings of security (Henig & Maxfield, 1978). Social bonding or interaction, as it is sometimes referred to, may contribute to levels of informal control. In W. Jacobs (1961) states that the elimination of crime can come through an informal control system established by the residents of a ta! 9? 'l .0. S; r i It! DJ 13 neighborhood. She further suggests that peace can be maintained by an intricate and almost unconscious network of voluntary controls and standards among the people themselves. By establishing this social network, residents can ensure their safety from crime and the fear of crime. Reflective of Jacob's view is Newman's (1973) concept of defensible space. This model for residential environments inhibits crime by creating the physical expression of a social fabric that defends itself, allows for easy recognition of intruders, and provides for the security of families, neighbors, and friends. The views of Jacobs and Newman portray social interaction as the cornerstone of informal control. Although neighborhood bonds may facilitate the exercise of informal control mechanisms to reduce the frequency of criminal acts that generate fear (Riger et al., 1981), social interaction also increases communication and may be responsible for what Skogan (1977) terms indirect victimization. Even though it has been suggested that communities develop programs that familiarize residents with each other as a means to reduce fear (Greenberg et al., 1985), it is possible that these programs precipitate fear. In an evaluation of crime programs being planned in Chicago neighborhoods, it was found that programs initiated for purposes of reducing crime may have actually generated greater levels of fear (Rosenbaum, 1987). 14 While looking at research directed at attitudinal reactions to crime, it is also necessary to consider various methods employed to measure fear. Just as the findings about fear have varied, the techniques to measure this attitudinal reaction have also been inconsistent. Measures have ranged from single item to multi-dimensional indices, indicating that there is little agreement about which approach for measuring fear is best. Measuring Fear Fear is an expressed attitude and psychological state that is provoked by a sense of personal risk (Skogan, 1978). Learned associations with danger induce the physiological condition known as fear (Stinchcombe et al., 1978). These stimuli are rooted in reality, may exist in the environment, and can originate from direct or indirect experience. That which is measured and conceptualized as the fear of crime has its roots in something more diffuse than the perceived threat of some specific danger in the immediate environment (Garofalo & Laub, 1978). Research can still tap the attitudinal component of fear in the absence of natural stimuli. Garofalo (1980) suggests that fear be measured by asking people to either recall incidents of fear or to anticipate how they would feel under selected conditions. Researchers have traditionally examined fear as a unidimensional concept. That is, the attitudinal state ‘l. 51' uq I l 15 elicited by various forms of crime is uniform and can be measured on the same continuum. Some of these unidimensional conceptualizations have used a widespread practice of measuring fear with a single item--i.e., How safe do you feel being alone in your neighborhood? However, other studies have made an effort to distinguish among various types of fear. One of the earliest distinctions was between how worried people are about being victimized and how concerned they are about general levels of crime (Furstenberg, 1971). The difference here is made between specific and general fear. In a study of four Chicago neighborhoods, Skogan and Maxfield (1980) utilized an approach similar to Furstenberg's. First, a set of questions asks whether individuals think burglary, robbery, assault, and rape are a big problem, some problem, or no problem at all. A second group of indicators was then used, on a scale of one to ten, to determine how likely the respondents felt they would become victims of each of the four crimes. Skogan (1987), points out that the two types of attitudes may be correlated, suggesting that responses to the two different sets of questions may be the same. More recently, scholars have suggested distinguishing fear in terms of personal and property crime. Rosenbaum and Baumer (1981) conducted a methodological study encouraging the use of such a technique. In researching the impact of vict fear east vict cans app: 4.}. in VI: 4550 acre qrea 16 victimization on fear, Skogan (1987) individually measured fear for personal crime and fear for property crime. Where most studies have unveiled a weak correlation between victimization and fear, Skogan found that victimization was consistently related to his measures of fear. Skogan's approach represents an improvement over earlier measurement techniques. The value of the personal-property distinction is that it does not assume that the fear for every crime is homogenous. This approach differentiates conceptually between fear associated with personal crime and the fear associated with property crime. Since such an approach is more crime-specific than other approaches, there is a greater accuracy in measuring crime-related attitudes, hence improving validity. Furthermore, what is suggested is that attitudes associated with personal crime are unique and may be more closely aligned with the given definition of fear. That is, fear generates a sense of personal risk. Similarly, such an approach recognizes that attitudes related to property crime may be affected by factors different from those that affect attitudes associated with personal crime. In sum, different causal models are needed to account for different phenomena. It is also necessary, then, to examine different behavioral reactions to crime in accordance with the crime that they are intended to inhibit. also nan-1. bvw‘ prev: engaa EXte: vict Cate beha <2) ‘u that 17 Measuring Behavioral Reactions to crime Fear is not the only reaction to crime. Individuals also take action. The things people do about crime are commonly referred to as crime prevention actions. Crime prevention actions are responses to crime in which citizens engage to reduce either the chances of victimization or the extent of losses (physical, monetary, psychological) from victimization attempts (Lavrakas & Lewis, 1980, p. 255). Because many possible preventive actions exist, researchers have attempted to conceptually place these behaviors into categories. The results of these efforts are presented below. Furstenberg (1972) made one of the initial attempts to categorize preventive actions. He placed preventive behaviors into two classes: (1) avoidance behaviors: and (2) mobilization techniques. Avoidance behaviors are just that--monetary-free strategies to isolate oneself from exposure to victimization. Mobilization techniques are activities that use hardware such as guns or property and require real financial investment. Tien, Reppetto, and Hanes (1976) separated crime prevention behaviors into three types. The first type, access control, is similar to Furstenberg's mobilization techniques and involves any target-hardening action. The second category is termed surveillance. This refers to 18 active and repetitive crime prevention behaviors--i.e., watching for criminal activity. The third form of preventive action, territoriality, shares some of the same characteristics that access control and surveillance possess, but also includes crime prevention actions targeted outside the home. Territoriality is comprised of actions that indicate residential affection and demonstrate a concern for the neighborhood as a whole, such as joining a community crime watch program. Other categories of crime prevention action are made up of distinctions made by individual or group responses (Conklin, 1975), and by private prevention and collective prevention (Schneider & Schneider, 1978). These previously mentioned techniques are pragmatic. However, many of these conceptualizations lack the type of supporting evidence necessary to assess how well they relate to crime prevention behavior (Lavrakas & Lewis, 1980). As with attitudinal reactions to crime, researchers have also dichotomized behavioral reactions by means of a person-property distinction. Work by Skogan (1981) and Mendelsohn et al. (1981) suggest that crime prevention behaviors can be classified simply as either actions oriented towards protecting person or actions oriented towards protecting property. Even though additional classifications were provided by Skogan and Mendelsohn et a1.--i.e., collective participation, suburban flight, and 3‘ 19 action requiring repetitive action--these categories are either extremely rare in occurrence or overlap the person- property distinction. These recent techniques for typing preventive action are indicative of a more common-sense approach to conceptualization. Although there are more dimensions that might be more empirically accurate, two dimensions such as this personal and property crime classification lend themselves to analysis. The drive to protect either person or property seemingly provides the motivation for most crime prevention action. Protection of person includes such actions as not going out alone or avoiding public transportation. Installing locks and leaving lights on when one leaves the home falls into the area of property protection. Some actions can have the dual purpose of protecting both person and property--i.e., buying a gun. While placing these behaviors into one of two categories may be problematic, dual purpose activity still emanates from the singular needs of protecting property and person. Citizens' reaction to crime in the form of action is not a randomly distributed behavior. In the effort to prevent crime, some people may take several measures and others none. The exact manner in which crime instigates action is an unresolved issue. Considered here are explanatory approaches and findings that provide insight into how crime may provoke crime prevention behavior. . 399‘ uvb‘ in Oj such 20 Crime and Preventive Behavior There is a fundamental divergence in sociological perspectives regarding the exact impact of crime on behavior. One view suggests that crime has a positive effect on individual and community behavior. Durkheim (1933) states that crime threatens life, property, and views of appropriate behavior. In turn, this threat generates action and increases solidarity by binding people together in opposition to violators of the law. Paralleling Durkheim's view, Cohen (1966) argues that a common enemy, such as crime, may contribute to community solidarity. In contrast, it is thought that crime adversely impacts collective behavior and encourages individual, self-serving action. Conklin (1975) argues that crime induces suspicion, insecurity, and withdrawal from community affairs. In high- crime areas, residents lose confidence in the capacity of formal authorities to assist them, develop a negative view of their neighborhood, and reduce interaction with their neighbors (Skogan, 1981). Social interaction is reduced and people restrict their lives (McIntyre, 1967). Most important is that supporters of these theories postulate that crime creates a societal erosion that not only weakens local cohesion, but encourages individual crime prevention action. 21 In general, group anti-crime efforts appear to be rare and closely related to Citizens' general tendencies toward participation in community activities (Lavrakas & Herz, 1982). Furthermore, household crime prevention efforts are more prevalent in high-income, low-crime areas (Skogan, 1981). The views of Durkheim and Cohen appear, if anything, anachronistic. Recent data seem to support a view of crime as inspiring individual, self-serving action. Implicit in much research is the assumption that fear of crime motivates citizens to engage in preventive behavior. However, Gordon and Riger (1979) point out that little focus has been drawn towards the sense of one's personal danger--i.e., fear of crime--and self-protective behaviors. It appears that previous study has overemphasized demographic and social factors. Absent from many of these studies is a thorough examination of the interrelationship among crime, fear, and preventive action. Towards an Attitude-Behavior Relationship Even though positive correlations between fear and crime prevention have been found (Krahn et al., 1985), the consequences of fear remain an unresolved issue (Liska et al., 1988). Seemingly ignored is the existence of a relationship between attitudes of fear and crime prevention behavior. Failure to examine this relationship may be 22 attributed to a strong tradition of skepticism regarding an attitude-behavior nexus. Scholars have cast much doubt upon the validity of attitude-behavior relationships. Comprehensive reviews of sociological literature by Defleur and Westie (1963), Deutsher (1966), and Wicker (1969) provide substantive support for the prevailing pessimism pertaining to attitude- behavior research. However, from a methodological perspective, Schuman and Johnson (1976) reevaluate these findings. Schuman and Johnson not only present demonstrable evidence that many attitude-behavior studies yield valid results, but also provide guidelines for ensuring the validity of attitude-behavior measures. Presented below are the fundamental guidelines given by Schuman and Johnson and the implications that these recommendations have for studying attitudinal and behavioral reactions to crime. First, research should consider the influence of situational factors upon behavior, such as individual, social, and environmental variables. That is, there is a need to examine other determinants of action besides attitudes. Second, measurements of behavior need several indicators. A behavioral construct based on several related behaviors should be more reliable than a single behavior. If the focus is upon crime prevention behavior, then it is necessary to consider several different types of crime prevention action--i.e., not going out alone at night, 23 looking doors and windows, and joining a neighborhood watch group. Also, to improve the relationship between attitudes and behavior, researchers should measure attitudinal and behavioral variables at the same level of specificity. If the behavior is a single specific act, then the attitudinal measure should be specific, also. Where reactions to crime are considered, categorizations such as Skogan's personal- property crime distinction may provide the specificity necessary to maintain A-B congruence. Finally, the correlations that do occur can be large enough to indicate that causal forces are involved, but are rarely large enough to suggest that attitudinal responses can serve as mechanical substitutes for behavioral responses. In addition, Schuman and Johnson suggest that the A-B relationship is more likely to be stronger for symbolic forms of behavior than non-symbolic forms of behavior such as preventive actions. Hence, if strong correlations are found between fear and crime prevention behavior, then it would not be presumptuous to suggest that attitudes about crime influence behavior. Development of Research Questions People may react to crime by developing fear and taking action. Although the existence of these responses is substantiated, there is still no consensus as to their 46. db: '61-- H“ IE! 24 origin. Vulnerability to crime, prior victimization experience, and residential conditions are all thought to influence attitudes and behavior. To what degree these factors affect fear and crime-prevention behavior remains unknown. Furthermore, whether there is a relationship between fear and preventive behavior continues to present itself as an important question. This researcher suggests that by using a more specific approach to measure fear and crime prevention actions, many of these questions may be resolved. Specifically, this researcher will conceptualize attitudinal and behavioral responses to crime as two- dimensional items. Attitudes demonstrating a proprietary interest fall into the dimensional category ggngern_f2r ro e . Related to concern is the behavioral response termed agtign_§9_p;g§ggt_p;gpgzty. The second attitudinal category is fear for person. Fear represents attitudes that are associated with needs of personal safety. Complementing fear is actign_tg_prgtggt_pgrsgn. These behaviors are explicitly formulated to reduce the chance of personal victimization. With this conceptualization of reactions to crime, the following research questions are presented. W 1. What is the relationship between fear for person and action to protect person? 25 2. What is the relationship between concern for property and action to protect property? MW; 1. What effects do gender, age, income, assault rates, social interaction, and personal victimization have on fear for person? 2. What effects do gender, income, assault rates, social interaction, and personal victimization have on action to protect person? 3. What effects do gender, income, assault rates, social interaction, and property crime victimization have on concern for property? 4. What effects to gender, age, income, burglary rates, social interaction, and property crime victimization have on action to protect property? summary Scholars indicate that the presence of crime, experiences of victimization, and social networks influence fear. However, what exactly encourages or discourages people to take crime-prevention action is not as clear. The relationship between attitudes about crime and crime prevention action is also vague. In sum, the literature leans toward suggesting an interrelationship among experiences, residential conditions, attitudes, and 26 preventive behavior, but draws no specific conclusions. This study undertakes a thorough examination of the relationship among these items. n 6—. .eah ‘I‘ Ce“ 731’ of 1 :-&:€ CHAPTER 3 METHODS Introduction The purpose of this study is to explore the interrelationships among individual and residential characteristics, attitudes about crime, and preventive action in urban neighborhoods. This chapter provides information about the original data set, measurement of variables, and analysis used in this research. Data Set The data set utilized in this study is "Characteristics of High and Low Crime Neighborhoods in Atlanta, 1980." The Inter-University Consortium for Political and Social Research (ICPSR) has provided this data (#7951). Stephanie Greenberg (1982) originally collected the data to examine why some neighborhoods remain relatively safe when one would expect them to be unsafe because of their proximity to dangerous areas and because of their social and economic conditions. The study explored dimensions of territoriality (spatial identity, local ties, social cohesion, informal social control) and physical characteristics (land use, 27 28 housing characteristics, street type, boundary characteristics) in three pairs of neighborhoods of Atlanta, Georgia (Greenberg, 1982). Wen The three criteria used in the selection of neighborhood pairs were physical adjacency, homogeneity in racial composition and economic status, and relative difference in crime rates (i.e., high or low). If either one of the neighborhood pair was predominantly industrial or commercial, an officially designated historic district, or dominated by publicly owned housing, the pair was eliminated (Greenberg, 1983). The final selection netted three neighborhood pairs: Dixie Hills-Grove Park, Pittsburgh- Mechanicsville, and Virginia-Highland. The first pair of neighborhoods, Dixie Hills and Grove Park, are lower middle-class black neighborhoods and suburban in appearance. Pittsburgh and Mechanicsville, the second pair, are characterized as black, low-income, and as being within close proximity to business and industrial areas. The third pair of neighborhoods was initially Morningside-Lenox Park and Virginia-Highland. These neighborhoods were both white and middle- to upper-income areas. Yet additional data and observation suggested that Morningside-Lenox Park was higher in economic status and composed of larger, more expensive housing. Closer F: u r .U‘uo‘ . 771 7-H 'Aej 29 examination of Virginia-Highland revealed that the southern portion had substantially more crime than the northern section. These areas were also visibly separated by an east-west thoroughfare. Instead of eliminating the third pair of neighborhoods entirely, Greenberg used the divided Virginia-Highland neighborhood. The inclusion of a white neighborhood pair was necessary to improve the generalizability of the study (Greenberg, 1983). Therefore, use of the divided neighborhood was based upon the justification that no other pair of white adjacent neighborhoods came close to meeting the selection criteria. M11113 The sampling frame for each of the six neighborhoods was a list of residential properties located within the established boundaries of the neighborhood. Those properties excluded from the sampling frame were properties identified as non-residential or received federal or local funding. A stratified single stage sampling technique was employed. Each neighborhood was composed of 132 equal- sized zones. From each zone, one housing unit was selected. In all, there were 801 residential units contacted. The nine additional units were attributed to dwellings that were initially believed to be single-unit residences, but were later discovered to contain apartments. '1- ...E vi Qn y. ”Vu‘. 30 D§£§_§22£21 Greenberg's primary data source was a household survey that measured dimensions of territoriality, subjective reactions to crime, neighborhood problems, victimization, and demographic characteristics. Secondary data sources included a reported crime file, the plan file, neighborhood profiles, and street maps. The reported crime file and neighborhood profiles, published by Atlanta's Planning Bureau, were used for the selection of neighborhoods. The household survey is the data source used in this study. The survey contains 632 fixed and open-ended questions, with the majority being fixed. The response rate for the survey was highest in Pittsburgh and Mechanicsville, 83% and 87%, respectively. The lowest response rate occurred in Virginia-Highland. In Virginia-Highland, the response rate was 67% in the high crime neighborhood and 70% in the low crime neighborhood. The mean response rate for all neighborhoods was 77%. It is important to note that the data source used in this study was administered as a household survey. Because the original questions targeted the household, a gap could exist between responses pertaining to the individual and those referring to the household. That is, responses to the original survey may not only pertain to the respondent. Some responses refer to the household as a unit or even to 31 other members of the household. Hence, what is interpreted by this researcher as an individual measure may indeed be a household measure, thereby tainting the validity of results. To minimize this discrepancy, care was taken in the construction of variable indices to include only those questions that would reflect characteristics, attitudes, and behaviors of the individual. Measurement of Variables The analysis focuses on four types of variables: (1) crime prevention behaviors: (2) attitudes about crime: (3) individual characteristics: and (4) residential characteristics. Since the data set used in this study is not original, modification and refinement of certain variables is necessary as described below. This study employs two models to explain what generates crime-prevention action. The first model posits an interrelationship among individual characteristics, residential conditions, fear, and actions to protect person (see Figure 1). A second model describes an interrelationship among individual characteristics, residential conditions, concern, and actions to protect property (see Figure 2). Therefore, the dependent variables in this study are action to protect person and action to protect property. Attitudinal variables of fear for person and concern for 32 FIGURE 1 A MODEL FOR ATTITUDINAL AND BEHAVIORAL REACTIONS TO PERSONAL CRIME Individual and Residential Characteristics Attitude Behavior Age I Gender Income Fear Action , _ . _ I For To Personal Victimization Person Protect Person Neighborhood Assault Rates Social Interaction 1 FIGURE 2 A MODEL FOR A'ITITUDINAL AND BEHAVIORAL REACTIONS TO PROPERTY CRIME Individual and Residential Characteristics Attitude Behavior Age I Gender Income Concern Action . . . . For To Property Victimization Property Protect Property Neighborhood Burglary Rates Social Interaction J are C VQrE aSCer. 33 property are intervening variables. The independent variables are age, gender, income, victimization experience, officially reported rates of crime, and social interaction. ent v r Crime-prevention behaviors are measured by a scale that indicates the number of different preventive actions taken. Actions to protect person include avoiding public transportation, avoiding going out at night, avoiding going anywhere in the neighborhood alone, and taking a self- defense course. Property protection actions are comprised of engraving identification on valuables, installing burglar alarms, installing locks on doors or bars on windows, having a neighbor watch the home while away, and joining a neighborhood program designed to reduce crime. Depending upon the question asked, the time reference for a given behavior is action taken either within the past year or since the beginning of residency if the respondent had lived at the residence less than one year. Although other actions are available for inclusion, they are excluded because it is difficult to determine whether these behaviors are actually intended as crime prevention actions. The acts of keeping a gun or having a watchdog, which are often included in measures of crime prevention action, were eliminated for two reasons. First, it is difficult to ascertain whether a gun or a watchdog is actually kept at 34 home for crime prevention purposes. Residents may keep a gun for sporting activities, decorating the home, or some other non-crime inhibiting motive. Similarly, the keeping of a dog is not always for purposes of watching and protecting the home. Many respondents may purchase a dog merely as a pet, yet later consider the animal a watchdog. Secondly, even if a gun or watchdog is kept to stop crime, determining exactly what type of crime these actions are intended to prevent becomes problematic. Another activity eliminated from this study is having neighbors pick up mail while residents are away from their homes. This activity is probably more of a neighborly practice not specifically intended to prevent crime. tt tude out Survey items probing respondents' attitudes about crime are separated according to whether they refer to fear for person or concern for property. Responses to items about fear for person are coded on a four-point scale ranging from "not fearful" to "very fearful." A similar four-point scale measures the level of concern for property--ie., "not concerned" to "very concerned.” Level of fear is determined by how fearful residents are about particular types of personal crime, if a stranger stops them for directions, and if they hear the sound of footsteps behind them at night. With respect to property, pi- ’0‘ I an. '0 «an ~91 h .1 vac 5 RE? 35 the level of concern is measured by questions that reveal whether the individual is particularly worried about a specific property crime or about someone breaking into his or her home. Despite the availability of additional information about attitudes toward crime--i.e., How safe do you feel in your neighborhood compared to the rest of the city?--these items are constructually vague. These questions were originally used to tap the general threat of crime. Responses to these questions possibly are more of a reflection of the general habitability of the respondent's neighborhood--i.e., incivility, physical decay of buildings, vacant housing--than of actual crime conditions. d v d a a cs The first three of the four individual characteristics taken into account in this study are gender, age, and income. Research has consistently shown that women and the elderly report higher levels of fear, yet lower incidence of victimization (Biderman et al., 1967: Conklin, 1975: Garafalo, 1977a: Clemente & Kleinman, 1977: Riger, 1981: Baumer, 1985). Similarly, research has uniformly reported a negative relationship between income and fear (Biderman et al., 1967: Ennis, 1967: Clemente & Kleinman, 1977). To determine the congruity of this study's findings with previous research, and more importantly, to remove the I. H y A. \LJ 383 (J I”?! res for per the :0 (D n l iI‘f \ ‘V Lae 36 confounding effects of these variables, analysis includes gender, age, and income. Victimization experience is the fourth individual measure utilized in this study. Victimization fell into one of two categories: (1) personal, or (2) property. If the respondent experienced a violent crime, had things taken by force, or experienced violence in a quarrel, then that person is considered to have experienced personal victimization. Unfortunately, it was difficult to ascertain from the codebook whether the term ggarrgl was used to include acts of domestic violence. If respondents were including victimization in the home when responding to this item, this operationalization might weaken the impact of victimization on attitudinal and behavioral responses to crime. Respondents that had their residence damaged, car stolen, home broken into, possessions stolen, or experienced other specific types of property crime are categorized as having experienced property victimization. Those persons considered victims of personal or property crime must have experienced at least one incident of victimization within the last 12 months. W The first residential variable in this study is social interaction. The level of social interaction is defined as the number of neighboring practices taken by each IE! Fe '5’. V. ’t! ¢ 37 respondent. The frequency of these activities, both within two blocks and throughout the remainder of the neighborhood, measures either a low or high level of social interaction. Neighboring activities include the practices of helping others with jobs, having dinner with neighbors, borrowing or exchanging things such as tools or recipes, visiting with people, and having nearby residents watch the respondent's children. The use of social interaction is deemed necessary because of the need to clarify the role that social networks play in enhancing feelings of security. The last measure used in this study is officially reported rates of crime per 100 occupied housing units. As opposed to using all index crimes, this analysis employs a single offense to represent each category of personal and property crime. The two selected crimes are aggravated assault, and burglary, respectively. Because of respondent embarrassment, memory decay, telescoping of events, and other problems associated with self-reports of victimization, assault and burglary rates are utilized here for purpose of providing an objective and secondary measure of crime conditions. Analysis The first phase of analysis was to run frequencies on all selected variables. Frequencies are used to summarize data, expose missing values, and reveal the skewness of 38 distributions. Cross tabulations were then run to identify significant differences in bivariate relationships among independent and dependent variables. The Chi-Square test, with a significance level of .05 or greater, was used to determine if these differences are statistically significant. Gamma was also used to measure the strength and direction of these bivariate relationships. The next step of analysis was the calculation of correlation coefficients for each pair of variables. The produce of this analysis is in the form of a correlation matrix and was used to identify variables that were highly correlated with each other and those that were relatively independent. Because the correlation matrix is used to complement multiple regression analyses, the primary use of the matrix is to unveil problems of multicollinearity. Multiple regression was then used to assess the relative explanatory power that the independent variables have in predicting change in the dependent variables. Finally, path analysis provides a diagrammatic representation of all variables and depicts the direct and indirect effects of the variables. CHAPTER 4 RESULTS Introduction This chapter presents the results from data analysis. Analysis consists of cross tabulating dependent and independent variables, calculating a correlation matrix for variable pairs, and the running of multiple regression for path analyses. Here are tables, figures, and presentation of salient findings. Results from Cross Tabulations The bivariate relationships between all independent and dependent variables are presented in Tables 1 through 4. The dependent variables for this analysis are fear for person, concern for property, action to protect person, and action to protect property. Table 1 presents the findings of analysis for the relationship between fear for person and individual and residential characteristics. e o e son Except for the individual characteristic of age, all variables have a significant relationship with the dependent 39 variable I relations to respc fearful c responder Gen: women re; sales, 37 59.3% of fearful c toderate indicatir lettels 01 Neii to level; high ass. IEVel of between : iirectio neighbor 40 variable, fear (see Table 1). The strongest of these relationships is between personal victimization and fear (gamma = .414, p < .001). More than two-thirds, 69.7%, of the respondents who experienced victimization report being fearful or very fearful. Only 7.6% of victimized respondents reported not being fearful. Gender is also closely associated with fear (gamma = .344, p < .001). This relationship appears to come from women reporting higher levels of fear than men. Of the males, 37.5% indicate that they are not fearful, whereas 59.3% of the females report being either fearful or very fearful of personal crime. An inverse relationship of moderate strength is found between income and fear, indicating that respondents with lower income had greater levels of fear. Neighborhood assault rates are also found to be related to levels of fear (gamma = .263, p < .01). In areas with high assault rates, 70.9% of the respondents indicate some level of fear for violent crime. Although the relationship between social interaction and fear is weak, the negative direction suggests that residents with greater levels of neighboring practices are not as fearful as those with lesser levels. Ago 35 t: .r: 5155 0O " I Q. 0'! ed Genoa Via»! ‘Paie income was $3,} Slit :3 S 54.203 to $3,339 to S 56200:: 5 93.30213 5 5‘: SEC :3 '2‘” .A "V'V iv 5.:5" ’H ' 1" \V $21M :5 525 30C to BC 39C :3 $35.39: ,, \ I"Serial I \3 5x36?! 41 Table 1 AMOUNT OF FEAR FOR PERSON BY INDIVIDUAL AND RESIDENTIAL CHARACTERISTICS °/o °/o °/ci °/° Not Somewhat Very x2 G N Fearful Fearful Fearful Fearful Age 11.130 .080 35 and under 241 22.8 29.0 23.7 24.8 35 to 55 124 21.8 23.4 19.4 35.5 Over 55 158 25.3 21.5 16.5 36.7 Gender 34.804"M .344 Male 208 37.5 23.1 15.4 24.0 Female 314 14.6 26.8 23.9 35.4 Income 62.956'" -.249 Under $3.000 48 10.4 20.8 18.8 50.0 $3.000 to $3.999 30 23.3 23.3 13.3 40.0 $4,000 to $4,999 19 21.1 10.5 26.3 42.1 55.000 to $5,999 19 21.1 31.6 21.3 21.1 $6.000 to $7,999 38 13.2 36.8 15.8 34.2 $8.000 to $9,999 29 13.8 24.1 34.5 27.6 $10,000 to $11,999 27 18.5 25.9 37.0 18.5 $12,000 to $14,999 37 32.4 21.6 27.0 18.9 $15,000 to $19,999 48 29.2 31.3 18.8 20.8 $20,000 to $24,999 26 42.3 42.3 11.5 3.8 $25,000 to $29,999 18 16.7 22.2 27.3 33.8 $30,000 to $34,999 9 55.6 11.1 11.1 22.2 $35,000 and over 19 42.1 26.8 23.9 35.4 Personal Victimization 18.987"M .414 No Experience 457 25.6 25.8 10.8 27.8 Yes Experience 66 7.6 22.7 18.2 51.5 Neighborhood Assault Rates 15.450": .263 Low 321 27.7 27.1 19.6 25.5 High 202 16.3 22.8 21.8 39.1 Social Interaction 19.1707" -.073 Rarely 266 25.6 23.3 16.9 34.2 Sometimes 190 16.8 26.3 24.7 32.1 Often 67 32.8 31.3 22.4 13.4 ’p< .05; "p<.01; "‘p<.001 I nu 42 Logic; ro Protect Person The relationships between individual, attitudinal, and residential characteristics and the amount of action taken to protect person are presented in Table 2. Five significant and strong relationships occur in Table 2. The strongest of these relationships is observed between gender and personal crime prevention action (gamma = .482, p. < .001). In this relationship, 79.4% of the males took, at the most, only one action to protect themselves. Conversely, 75.4% of the females took one, two, or three actions to protect themselves. The relationship between income and preventive action is also strong but negative in direction (gamma = -.386, p < .001), suggesting that residents in lower income groups engage in more action to protect themselves from personal crime. The attitudinal characteristic of fear is strongly associated with action to protect person (gamma = .442, p < .001). Of the respondents stating that they were not fearful, 61.2% took no preventive action. Even for respondents reporting to be somewhat fearful, 76.4% indicate taking one or no action to protect their person. Yet 73.3% of those respondents reporting to be fearful, and 80.9% of those reporting to be very fearful, engaged in one, two, or three personal crime prevention actions. Similarly, strong relationships are observed with residential characteristics. Neighborhood assault rates c S. 1' .0.— F .i c c ‘ AMOUNT OF ACTION TO PROTECT PERSON 43 Table 2 BY FEAR. INDIVIDUAL CHARACTERISTICS. AND RESIDENTIAL CHARACTERISTICS 2 Number of Different Acrions to Protect Property e10 CI 10 01° OI C 9’s x G N 0 1 2 3 4 Age 27.251": .152 35 and under 230 43.0 23.0 26.1 6.5 1.3 35 to 55 121 38.8 25.6 24.0 10.7 .8 Over 55 154 22.1 38.3 33.8 5.8 0.0 Gender 61.707‘" .482 Male 149 54.8 24.6 14.6 4.5 1.5 Female 305 23.3 30.5 36.7 9.2 .3 Income 86.683’" -.386 Under $3.000 47 5.4 42.6 31.9 17.0 2.1 $3,000 to 53.999 28 28.6 32.1 25.0 14.3 0.0 $4,000 to $4.999 18 16.7 22.2 38.9 22.0 0.0 $5.000 to $5.999 19 31.6 31.6 31.6 5.3 0.0 $6,000 to $7,999 38 28.9 28.9 39.5 0.0 2.6 $8.000 to $9.999 27 25.9 33.3 33.3 7.4 0.0 $10,000 to $11,999 26 34.6 34.6 26.9 3.8 0.0 $12,000 to $14,999 33 51.5 27.3 21.2 0.0 0.0 $15,000 to $19,999 48 47.9 27.1 14.6 10.4 0.0 $20,000 to $24,999 26 65.4 15.4 14.2 0.0 0.0 $25,000 to $29,999 17 52.9 29.4 11.8 5.9 0.0 $30,000 to $45,999 9 55.6 33.3 11.1 0.0 0.0 $35,000 and over 19 78.9 15.8 5.8 0.0 0.0 Personal Victimization 7.301 .232 No Experience 439 37.4 28.5 26.2 7.1 .9 Yea Experience 66 24.2 27.3 39.4 9.4 0.0 Fear for Peraon 83.916’" .442 Not Fearful 116 61.2 23.3 12.1 3.4 0.0 Somewhat Fearful 131 42.0 34.4 19.1 3.8 .8 Fearful 101 26.7 27.7 34.7 10.9 0.0 Very Fearful 157 17.2 27.4 42.7 10.8 1.9 Neighborhood Aaaault Rates 22.450”: .297 Low 309 42.4 27.2 - 24.6 5.8 0.0 High 196 25.0 30.1 33.2 9.7 2.0 Social Interaction 19.425’ -.097 Rarely 256 31.6 31.3 30.5 6.3 .4 Sometimes 184 35.3 26.6 29.3 8.2 .5 Often 65 52.3 21.5 13.8 9.2 3.1 ’p< .05; "p<.01; "'p<.001 .e 0.. I: CA a» 5; r\ N 44 have a significant relationship with the amount of action taken to protect person. The strength of this relationship (gamma = .297, p < .001) indicates that the local threat of crime influences taking personal protective behavior. The relationship between social interaction and this dependent variable is significant, too, but weak. However, the direction of this relationship shows that residents who interact more frequently, engage in less preventive action. 9W Table 3 presents the relationship between individual and residential characteristics and respondents' level of concern for property crime. A relationship of moderate strength exists between age and concern (gamma = -.123, p < .01). The strength of this relationship appears to come from younger residents consistently expressing modest levels of concern. Other individual characteristics of property victimization and income reveal an even stronger relationship with concern. The strongest of these relationships is between property victimization and concern (gamma = .243, p < .01), and seems to stem from victims reporting higher levels of concern than nonvictims. Income is inversely related to concern, where there appears to be a general pattern of lower income residents reporting higher levels of concern. This observation is particularly clear in the under $3,000 5.3? $3: $3.23: to! $1.3m: E3021: 1 36.33213 5 SE It's! 57:21: '3 5‘2 332 :3 5'5 355 'o 322 ICC :: 535303 f: 53: :51: 3: ‘28 :cc 3. \ M\ It: E13,. ‘2: 5.3,. \ “Show “as Q! AI?“ \ 5°98 In" 1”!“ 45 Table 3 CONCERN FOR PROPERTY BY INDIVIDUAL AND RESIDENTIAL CHARACTERISTICS W. °/e °/o °'o Not Somewhat Very x2 G N Concerned Concerned Concerned Concerned Age 21.078" -.123 35 and under 241 12.4 41.1 24.9 21.2 35 to 55 124 23.4 33.9 15.4 23.4 Over 55 158 30.4 28.5 19.6 21.5 Gender 3.513 .079 Male 206 24.5 34.1 19.7 21.6 Female 314 18.2 36.6 23.6 21.7 Income 3.513 .079 Under $3,000 48 12.5 22.9 20.8 43.8 53.000 to $3.999 30 20.0 40.6 13.3 26.7 $4,000 to $4.999 19 10.5 15.8 21.1 52.6 $5.000 to $5,999 19 5.3 47.4 15.8 31.6 $6.000 to $7,999 38 15.8 42.1 26.3 15.8 $8.000 to 39.999 29 10.3 37.4 20.7 31.0 $10,000 to $11,999 27 18.5 59.3 18.5 3.7 $12,000 to $14,999 36 13.5 40.5 29.7 16.2 915.000 to 519.999 48 8.3 ' 50.0 31.3 10.4 $20,000 to $24,999 26 23.1 42.3 19.2 15.4 $25.000 to $29,999 18 16.7 33.3 38.9 11.1 $30,000 to $45,999 9 44.4 22.2 33.3 0.0 $35,000 and over 19 21.1 26.3 36.8 15.8 Property Victimization 15.641 " .243 No Experience 331 25.7 34.7 20.8 18.7 Yes Experience 192 12.0 37.0 24.0 27.1 Neighborhood Burglary Rater 3.881 .157 Low 410 22.0 36.3 21.2 20.5 High 113 15.9 32.7 24.8 26.5 Social Interaction 25.313" .146 Rarely 266 28.2 31.2 19.9 20.7 Sometimes 190 14.7 35.8 25.8 23.7 Often . 67 7.5 52.2 19.4 20.9 ‘p< .05: "p<.01; '°°p<.001 46 category and the $5,000 to $5,999 category, where 43.8% and 52.6% of the respondents report being very concerned about property crime, respectively. There is also a significant and marginally strong relationship between social interaction and concern. Again, this relationship does not seem to come from extreme levels of concern. It appears to arise from 52.2% of the respondents who interacted often, reporting to be only somewhat concerned. The remaining residential variable,' neighborhood burglary rates, has no significant relationship with concern. 0 e t o e t Presented in Table 4 are the bivariate relationships between individual, attitudinal, and residential characteristics and the amount of action taken to protect property. Overall, only two variables show a significant relationship with action to protect property. The strongest of these relationships is the residential characteristic social interaction (gamma = .345, p < .001). Of the respondents who rarely interacted, 84.9% took two or fewer property-protecting actions, whereas 68.6% of the respondents who interacted often took two or more actions to protect their property. Income also has a strong relationship with the amount of action taken to protect property (gamma = .232, p < .01). 47 Table 4 AMOUNT OF ACTION TO PROTECT PROPERTY BY CONCERN. INDIVIDUAL CHARACTERISTICS. AND RESIDENTIAL CHARACTERISTICS Number of Different Actions to Pratect Property 9. OI 01° 01° °’o o o x2 G N o 1 2 3 4 Age 14.088 -.007 35 and under 241 13.3 35.7 32.4 14.5 4.1 35 to 55 124 11.4 29.3 31.7 19.5 8.2 Over 55 156 15.4 37.8 26.9 14.7 5.1 Gender 4.587 2026 Male 208 14.9 32.7 29.3 15.4 7.7 Female 311 12.5 36.3 31.2 16.1 3.8 Income 90.299" .232 Under $3,000 48 16.7 45.8 18.8 14.6 4.2 $3.000 to $3.999 30 13.3 43.3 30.0 6.7 6.7 $4.000 to $4,999 19 10.5 42.1 31.6 15.8 0.0 $5.000 to $5,999 19 26.3 31.6 26.3 15.8 0.0 $6.000 to $7,999 38 15.8 31.6 39.5 5.3 7.7 58.000 to $9,999 29 6.9 37.9 31.0 24.1 0.0 $10,000 to $11,999 27 14.8 25.9 40.7 14.8 3.7 $12,000 to $14,999 37 8.3 30.6 33.3 22.2 5.6 $15,000 to $19,999 48 4.2 20.8 39.6 25.0 10.4 $20,000 to $24,999 26 11.5 23.1 50.0 3.8 11.5 $25,000 to $29,999 18 0.0 27.8 27.8 33.3 0.0 $30,000 to $45,999 9 0.0 33.3 33.3 33.3 0.0 $35,000 and over 19 10.5 31.6 10.5 15.8 31.6 Personal Victimization 4,81 5 . 109 N0 Experience 329 15.8 35.3 28.3 14.3 6.4 Yes Experience 191 9.4 34.0 34.6 18.3 3.7 Concern for Property 24.469 .139 Not Concerned 106 23.6 35.8 23.6 12.3 4.7 Somewhat Concerned 186 1 1.8 38.7 30.6 12.9 5.9 Concerned 114 6.1 29.8 36.0 21.9 6.2 Very Concerned 114 14.0 32.5 31.6 17.5 4.4 Neighborhood Burglary Rate: 3.116 .060 Law 408 13.2 36.0 30.1 15.9 4.6 High 112 14.3 30.4 32.1 15.2 8.1 Social Interaction 48.558" .345 Rarely 265 19.6 40.0 25.3 13.6 1.5 Sometimes 188 8.5 29.8 36.2 17.0 8.6 Often 67 3.0 28.4 35.8 20.9 11.9 ’p< .05; "p<.01; "’p<.001 48 The magnitude of this relationship seems to come from the general lack of property-protecting action taken by respondents in low income groups, as opposed to higher income respondents taking multiple property-protecting actions. Age, gender, property victimization, attitudes of concern, and neighborhood burglary rates do not have a significant relationship with the amount of action taken to protect property. Results from the Zero-Order correlation Matrix Thus far, analysis has solely examined the bivariate relationships among variables. Of primary interest, however, is the influence of individual, residential, and attitudinal variables on crime prevention behaviors. Another important focal point is the impact of individual and residential characteristics upon attitudes of fear and concern. Hence, it is necessary to assess the relative predictive power of these variables through multiple regression. In the application of multiple regression analysis, the problem of multicollinearity can arise. Two preventive measures are taken to minimize this condition. First, common sense was used in the selection of predictor variables. Second, since problems of collinearity are not always identifiable, analysis also includes calculating correlation coefficients for all pairs of variables to 49 determine if any of variables are significantly interrelated. Table 5 contains the correlation coefficients in matrix format. Focus is upon possible interrelationships among independent variables. The relationship between fear and concern is of particular importance because of the need for two independent attitudinal measures. Although fear and concern are marginally correlated, this relationship does not appear sufficiently strong enough to suggest multicollinearity. Review of all additional pairs of predictor variables also does not reveal any strong interrelationships. Although multicollinearity may exist among three or more of the independent variables, the findings here provide limited evidence that the predictor variables are independent. Results from Path Analysis Multiple regression analyses are then run on the endogenous variables, fear for property, action to protect property, fear for person, and action to protect person. Figures 3 and 4 present the path analyses depicting the effects of predictor variables on crime prevention behavior. The path coefficients are standardized beta weights from multiple regression analyses. The path diagrams portray the direct effects that individual, residential, and attitudinal variables have upon preventive behavior. Also shown in 50 88: 85. cos... $652.. :28. 599.5 :seun< cozen. case: this... ceases 332.. «036... 48:. poo: poo: .Etu.) .9595 can. to“. oh oh Lessee eEoE. zoom 4222.: 43.3.62 5.2.93 .ecoacet Eeoccu Leeu— ce=u< co=o< cc=o< :8. m3: vmw: «mo: .3: 9N- 2:: 3a.. m8. 9.. m8: eu< v9: «3.- QB. mac: m3: o8: Sc. :3. 9n. to: 6930 3.. m3: m2: m8. no..- on..- New: Emu new. eEooc. 95.. «no: or. 3c. .3. one: 29. mom. c2323... Zoom n p N . cm —. 8.. mac. :0. v8. nmo. nee-c E295 6853‘s.: 08. 8.. en —. an F. c9. 25.. have: :seen< 68635.: New. on —. wee. 8:. mac. co=a~.E=u.> 5.2.6... 8n. 2:. 8.. a8: cozeu.E=e_> Icons... 8... SN. «2. 3.2.2.. .o. 53:60 can. omo. coo-ea 6. Lean n3. con-em «039.. o. co=e< m0"— XEFSZ ZO_._.<._mmmOo mun—COAum—mN m m..m<._. 51 these diagrams is the impact of individual and residential variables upon attitudes about personal and property crime. Path analysis additionally reveals how attitudes of fear and concern mediate the influence that individual and residential characteristics exert upon preventive behavior. BEER AQQIysis for Action to Protecr Parser The interrelationship among individual and residential characteristics, fear for person, and the amount of action taken to protect person are presented in Figure 3. Personal fear accounts for the largest contribution to the amount of action taken to protect person (beta = .29, p < .001). The coefficient for income is also significant (beta = .26, p < .001), but negative in direction, suggesting that residents with lower income took more action to protect themselves. Another substantive contributor to the amount of action taken to protect person is gender (beta = .15, p < .001). Age, personal victimization, neighborhood assault rates, and social interaction have little direct influence on behaviors designed to protect person. Figure 3 also presents the effects of individual and residential characteristics on the intervening variable fear. The most powerful determinant of fear is gender (beta = .25, p = .001), indicating more extensive levels of fear for women. Personal victimization (beta = .16, p < .001) and neighborhood assault rates (beta = .17, 52 meg—mo LEI—Gamma. O... mZOFO<:mm OZ< ._._hmwd0mm O._. mZOFOa<2< 1.5?— v 5.30:. P8.VQa.-a ”PO/Rae.- 9. see O iI\\\ d . \iIIiIIIlI o 40. _ .~ Cameos... .. .\ so 639... m :3. Ch ZO_._.U< / 0° . eeeQN . en 3: _ Zcmozoo > I .N 2959455. .268 ...~ 82: 5.56:2. coo: -zomxomz .N 29552.55 Cameos... .N 9282. «N 5323 .~ mo< 57 Di t d i t f cts 0 ct o t ££2£2££_2£222I21 Table 7 presents the predictor variables' indirect and summed effects upon the amount of action taken to protect property. In comparison to the indirect effects upon action to protect person, individual and residential characteristics have minimal impact upon property-protecting behavior. Income has the largest indirect value of any predictor variable (-.03). However, the overall influence of income is dampened by the negative direction of its indirect effect. A negative indirect value also reduces the summed effect of age upon property protecting-behavior, suggesting that attitudes of concern for certain income and age groups mediate levels of action. Indirect effects for property victimization, neighborhood burglary rates, and social interaction moderately bolster the total impact of these variables. Overview of Path Analyses In general, the findings of path analysis reveal several convergences and divergences between the two models. First, the significant predictors for fear are quite different than those for concern. Gender, victimization experience, and neighborhood crime rates are strongly related to fear, yet have no bearing on levels of concern. Fear and concern do, however, share income as a significant 58 and inverse predictor. Age has a moderate influence on concern, but no effect on fear. Social interaction accounts for little difference in either attitudes of fear or concern. The direct effects of predictor variables upon preventive action also vary. Income is the most powerful predictor of behavior in both models, but opposite in effect. Age and social interaction are heavily related to action taken to protect property, but have a minimal influence on crime prevention behavior for person. Whereas gender and neighborhood crime rates have a direct effect on action to protect person, these variables have little impact on action to protect property. Fear and concern are both solid predictors of preventive behavior. Victimization experience does not play a role in either model. The indirect and total effects of predictor variables for the two models are also different. In Table 6, age has no indirect or summed effect upon preventive behavior. Yet in Table 7, age does exhibit an indirect and total effect. Gender produces findings of a similar reciprocal nature. The indirect value for income is negative in both models. However, the impact of these values varies. That is, the indirect value for income in Table 6 enhances the summed effect of income. But in Table 7, the indirect value detracts from income's overall effect. Remaining indirect effects upon action to protect property are small. Personal 59 victimization experience and neighborhood assault rates appear to indirectly influence the amount of action taken to protect person. CHAPTER 5 DISCUSSION AND CONCLUSION Introduction The previous chapters have developed and tested two models designed to explain the interrelationship between attitudes and behaviors in response to crime (see Figure 1 and Figure 2). Variables in these models fall into four different categories: crime prevention actions, attitudes about crime, individual characteristics, and residential characteristics. The general difference between models is a personal-property crime distinction. The first model specifically provides explanation for fear of personal crime and the preventive actions taken to prevent personal crime. The second model furnishes explanation for attitudes of concern about property crime and the preventive action taken to prevent property crime. The Attitude-Behavior Nexus Fear appears to have a strong and consistent relationship with crime prevention action. Initially, a significant bivariate association was found between these two variables. Upon controlling for confounding variables, 60 61 fear still maintains a strong relationship with action to protect person. However, attitudes of concern do not reveal a similar consistency with preventive action. The bivariate relationship between concern and action to protect property is not significant. The magnitude of the beta coefficient between concern and action to protect property is only half of that which exists between fear and action to protect person. This fundamental difference between models presents two important issues. The first, which is methodological, is that measuring attitudinal and behavioral reactions to crime at the same level of specificity provides valid evidence of a relationship between fear and action to protect oneself from personal crime. Previous research has assumed that the fear for every crime is homogenous. This study measures specific attitudes about crime in accordance with specific crime prevention behaviors. Further enhancing the validity of this relationship between fear and action to protect person is the use of multiple indicators for each category of preventive action. Where previous studies have deviated from these practices, researchers have provided little evidence of a relationship between attitudes and behavior (Schuman & Johnson, 1976). The second issue is that fear of personal crime seems to generate more crime prevention behavior than does concern for property. That is, if people are fearful of personal 62 crime, they are more likely to do something to prevent it. Yet the attitudinal reaction of concern for property does not appear to be as powerful of a stimulus for behavior. Intuitively, this assumption would appear correct. Personal crime, particularly that which involves the threat of serious bodily harm, is much more likely to provoke defensive action. However, consideration must be given to the differences in cost between actions taken to protect property and actions taken to protect person. Preventive action for person, as measured in this study, requires no financial investment. Incurred costs are only measurable in terms of intangible loss, such as isolation, withdrawal from the community, or restrictions on daily movement. But many of the property-protecting actions in this study demand real dollar investment--i.e., installing locks, bars on windows, or burglar alarms. Hence, even if residents have high levels of concern about property crime, monetary constraints may inhibit their taking property-protecting action. Differences in the taking of personal versus property- protecting behavior may also be explained through the three additional factors: risk, seriousness, and efficacy. The importance of these factors with respect to taking preventive action was first advocated in Rosenstock's (1966) "Health Belief Model" (HBM). The HBM was designed to explain variance in the use of medical services. But as (I. u‘. 63 suggested by Lavrakas (1980) and Mendelsohn, et al. (1981), the basic elements of HBM appear equally applicable to crime prevention practices. The first factor, risk, is a rational prediction of the chances of being victimized by a particular type of crime. Seriousness, the second factor, is a determination of the severity that a particular type of victimization may have upon one's daily life. The third factor is efficacy, which is an evaluation of the power that a preventive action may have in stopping crime. So even if residents express high levels of fear or concern, these cognitive calculations may intercede and affect the taking of crime prevention action. Two of these elements, seriousness and efficacy, seem particularly important when explaining why some people take crime prevention action and others do not. The threat of personal crime upon daily life is usually perceived to be much more severe than the threat from property crime. For example, most people take the threat of rape or armed robbery much more seriously than the threat of pickpocketing or vandalism. In this study, actions to protect person have a high level of efficacy. Most are avoidance behaviors that appear effective in preventing personal crime, but demand little effort. On the other hand, actions to protect property can only reduce the chances of property crime, not totally eliminate the possibility of its occurrence. Many of the 64 actions to protect property involve the costly purchase and installation of hardware, with no guarantee of stopping crime. Factors such as underlying costs of preventive action, the seriousness of crime, and the efficacy of preventive action are important reminders that measures of attitudes can not be freely substituted for measures of behaviors. The causal force of attitudes is clearly controlled by the constraints of daily life. Despite attitudes voiced about crime, there are practical limitations on the actions that people can and cannot take. Furthermore, people may not invest their time and effort in preventive behavior unless there is some real benefit to their action. Individual Characteristics In general, individual characteristics are powerful predictors of attitudinal and behavioral responses to crime. Victimization experience, however, is found to be a relatively weak predictor of both attitudes and behavior. Besides attitudes, the three individual characteristics of income, age, and gender are the best predictors of who will engage in preventive behavior. IBQQEQ Income proves to be the strongest predictor of attitudes and behavior in both models. The strong inverse 65 relationship of income to fear and action to protect person is consistent with past findings that persons with lower income are more fearful and engage in more precautionary behavior. However, the strong influence of income upon fear and action to protect person may be partially spurious. Associations with physical characteristics of the local environment may partially explain differing levels of fear and preventive behavior associated with income (Baumer, 1978). Income also has a negative relationship with levels of concern, suggesting that residents in lower income groups are consistent in their worry about different types of crime. However, income also reveals a strong positive relationship with action to protect property. This clearly points out that, even though lower income residents are voicing greater levels of concern about property crime, these are not the same residents engaging in property- protecting action. Regarding both models, then the role of income is consistent in terms of its impact upon attitudes about crime, but vastly different in its effect upon crime prevention behaviors. The findings suggest that respondents in higher income groups are more likely than respondents in low income groups to engage in property-protecting action. Assuming income is an indicator of material wealth, residents who have more possessions or own their residence will take measures to 66 protect it. Similarly, there is no need for residents to look their door or put bars on their windows if there is little to steal. In addition, respondents with greater income are more likely to live in low crime neighborhoods and are not as likely to have high levels of concern. Also, by having the means to engage in preventive behavior, respondents may lessen their concern about property crime. A e Gender Age and gender produced strong but opposite effects within the two models. Contrary to expectations, age does not boost levels of fear or self-protective behavior. This outcome fails to parallel the numerous findings that the elderly report higher levels of fear. However, many of the past findings indicate that high levels of fear are also associated with white elderly. In this study, 360 of 523 respondents are black. Race may behave as a suppressor variable and reduce the true effect of age. Thus, the generalizability of this particular finding is not great. Age has a moderate inverse relationship with concern and an even stronger positive relationship with property- protecting action. It appears that older persons are apathetic in their view of property crime, but still engage in property protecting behavior. Explanation can also be attributed to the lack of mobility of older persons. That is, older residents are less likely to venture away from 67 home or place themselves in crime-prone situations. By staying within their residence and not exposing themselves to the potential dangers of the street, older persons may have a greater sense of security and feel free from property-related crime. Where age is not related to fear and the amount of protective behaviors against personal crime, gender reveals a strong relationship. The indirect effect of fear upon preventive behavior is larger than any other variable in either model. Examination of the bivariate relationships suggests that women are responsible for the variance in levels of fear and preventive behavior. Overall, it appears clear that women are more fearful than men. Women will also take more action to prevent personal victimization. These findings support a depiction of women as feeling more vulnerable than men, even though their risk of victimization is much lower. However, there is little difference in levels of concern or the amount of action taken to protect property between men and women. t at o Findings for victimization support expectations that experiences with crime are not an important contributor to attitudes about crime. Although the beta coefficient for personal victimization is moderate in strength, it is only the fourth most powerful predictor of fear. The coefficient 68 for property victimization is even smaller than that for personal victimization. Experiences with property crime prove to be only the third strongest predictor of concern. Even though neither relationship between experiences and attitudes is overwhelming, these findings portray personal victimization as a more attitudinally stimulating variable. Personal and property victimization make only a minor contribution to the amount of preventive behavior taken. As with attitudes about crime, the impact of victimization upon crime prevention is minimal. Personal encounters with crime seem to have little effect upon attitudinal and behavioral responses to crime. Victimizations considered in this study are experiences that occurred within 12 months of the survey. Unless the impact of victimization deteriorates substantially in some period of time less than 12 months, the passage of time does not appear to be a factor. Nor is there a failure to control for confounding variables of age and gender. Past study provides two additional arguments for the failure of victimization to account for varying levels of attitudinal and behavioral reactions to crime. First, the true impact of victimization is determined by how the victim interprets the situation. For some individuals, victimization is terrifying, to others it is not as traumatic, and others positive, since the experience may not actually be as traumatic as expected. A second suggestion, 69 which is rooted in common sense, is that victimization is only one of many factors that contribute to attitudinal and behavioral responses to crime. In consideration of the findings emanating from this study's analysis, the latter argument seems best. The effect of victimization is neither overpowering nor uniform in the two models. Victimization is, at the most, just one of many factors that contribute to developing attitudes about crime and taking crime prevention actions. Residential Characteristics Residential characteristics, neighborhood crime rates and social interaction, are the weakest predictors of attitudinal and behavioral reactions to crime. Neighborhood crime rates and social interaction exert little effect in either model. Even though local assault rates exhibit some influence on fear and action to protect person, the relative contribution of crime rates is notably less than individual and attitudinal variables. NEW Review of bivariate relationships reveals that official assault rates have a strong association with fear and personal protective behavior. When the effects of other variables are controlled, the magnitude of these relationships is dampened. Despite this reduction, 70 neighborhood assault rates are a much stronger predictor of fear than victimization, and within the entire model, the second greatest contributor to fear. This may confirm, as suggested by Garafalo (1979), that official reports of crime are more closely related to levels of fear than victimization data. But regarding action to protect person, assault rates are fourth in their relative contribution. It is interesting to note that the neighborhood level of assault is much more influential than first-hand experiences of victimization upon the taking of crime prevention action. This difference is observed for both direct and indirect effects, suggesting that fear has a strong intervening effect in the decision to take crime prevention action. However, the strong influence of neighborhood crime upon behavior and attitudes may partially arise from levels of incivility, physical status of the neighborhood, or other conditions associated with high crime areas. In Figure 4, the influence of burglary rates upon attitudes of concern and the amount of action taken to protect property is almost nonexistent. Besides gender, neighborhood burglary rates have the smallest impact upon levels of concern. The relative contribution of neighborhood burglary rates to property-protecting action is also minimal. This again indicates that reactions to property crime, either attitudinal or behavioral, are 71 somewhat restrained when contrasted with reactions to personal crime. Wigs Social interaction produces several unusual findings with respect to the influence that social networks have upon attitudinal and behavioral reactions to crime. The relationship between social interaction and fear is weak. The direction of this relationship is negative. This finding provides evidence suggesting that social networks have no effect on levels of fear. However, such a conclusion becomes questionable when considering the weak but positive effect that social interaction has upon action to protect person. In this relationship, heightened levels of social interaction appear to increase the need for personal protective behavior. But, as with the relationship between social interaction and fear, the relationship between social interaction and action to protect person is neither significant nor strong. In addition, the total effect of social interaction ranks next to the last in its relative contribution to protective behavior. The indirect effect of social interaction is negative in direction, but also relatively minor. The only safe conclusion that can be drawn is that social interaction has little effect upon fear or action to protect person. 72 The effects of social interaction on concern and on action to protect property are greater and more consistent because they are both positive in direction. Social interaction is only the fourth greatest contributor to levels of concern, but the second greatest contributor to the amount of action taken to protect property. Thus, for all attitudinal and behavioral responses to crime, the only strong and significant effect of social interaction is upon action to protect property. The strength of this relationship is possibly amplified by the nature of the property-protecting actions measured-- i.e., Have you had a neighbor watch on your home while you were away? That is, residents who responded yes to taking actions to protect property probably also indicated greater levels of social interaction. Hence, measures of social interaction and measures of action to protect property may be partially related. In regard to the overall impact of social interaction, there is little agreement between the findings of this study and the findings and implications of previous research. The general consensus of past research is that social networks enhance feelings of security, minimize worry about crime, and provide the means for informal control (Jacobs, 1963: Newman, 1973). Scholars have also suggested that social networks may magnify the threat of crime through vicarious victimization (Skogan, 1977). That is, residents may become 73 more aware or knowledgeable about the presence of crime by listening to the crime-related experiences of friends, neighbors, and relatives. However, the findings in this study provide little support for either argument. Conclusion This study presents a fresh perspective for examining reactions to crime by providing a property crime versus personal crime distinction. The results in this study reveal marked differences and similarities about reactions to crime. Foremost, attitudes of fear for person and concern for property have a strong impact upon preventive behavior. Although findings reveal a strong relationship between attitudes of concern and action to protect property, the relationship between fear and action to protect person appears greater, suggesting that the fear of personal crime is more likely to encourage preventive action. Income is also found to have a strong relationship with attitudinal and behavioral reactions to crime. Findings indicate that income has a strong inverse relationship with fear, concern, and action to protect person. Income, however, also has a strong and positive effect upon action to protect property. In regards to personal crime, the impact of income appears uniform. Yet, there is disparity between the effect of income on concern and the effect of income upon property-protecting behavior. Concern about 74 property crime appears limited to low income groups. Whereas, protecting property from crime seems to be a practice reserved for upper income groups. The effects of gender, age, social interaction upon reactions to crime varies. Gender has a strong effect on fear and action to protect person, but no effect on concern or action to protect property, indicating that property- protecting action is not gender-specific. Age and social interaction, however, appear to have a strong influence on action to protect property, yet almost no effect on concern, fear, or action to protect person. In general, the influence of age and social interaction upon reactions to crime is limited in comparison to other variables. In general, victimization has little effect on preventive behavior. However, the effect of personal victimization on attitudes of fear is much greater than the effect of property victimization on attitudes of concern. The influence of neighborhood crime rates is also weak. The only significant relationship found is that between neighborhood assault rates and fear. These findings suggest that experiences of victimization and the threat of proximate crime have little influence on reactions to crime. However, experiences with personal crime and the threat of personal crime appear to have a stronger impact upon reactions to crime than experiences with property crime or the threat of property crime. 75 It is important to note, however, that the data set utilized in this study was originally collected for other research purposes. Thus, the ability to accurately measure attitudinal and behavioral reactions to crime was unfortunately limited to the information contained within this data set. Improving the future study of reactions to crime certainly depends upon the use of items purposefully constructed for measuring not only attitudes of fear and concern about crime, but also for measuring preventive action against personal and property crime. The findings in this study help clarify the relationship among individual, situational, and attitudinal factors and crime prevention behavior. Gender and income are the strongest predictors of attitudinal and behavioral reactions to crime. Experiences of victimization, neighborhood crime rates, and social interaction are relatively powerless in influencing reactions to crime. These findings suggest that reactions to crime are, for the most part, an individual construct. As if in a vacuum, reactions to crime remain relatively unaffected by past experiences, the environment, or prevailing social conditions. Attitudes about crime emerge as a powerful and consistent determinant of crime prevention behavior. LIST OF REFERENCES LIST OF REFERENCES Agnew, R. S. (1985). Neutralizing the impact of crime. Qriminal_lu§fice_and_hebaxier. 12. 221-239- Baumer, T. L. (1985). Testing a general model of fear of crime: Data from a national sample. igurnal_gf_3e§eargh in_§rine_and_belinguencx .2. 239- -255- Baumer, T. L. (1985). Research on fear of crime in the United States. _1grimglggy, 3, 254-264. Biderman, A. D., Jonson, L. A., McIntyre, J., & Weir, aA. W. (1967). 'ct m des la gnfigrggmgnr. Washington, DC: U.S. Government Printing Office. Block, M. K., & Long, G. J. (1978). Subjective probability of victimization and crime levels: An econometric approach- Criminologx. 87-93- Clemente, F., & Kleinman, M. (1977). Fear of crime in the United States: A multivariate analysis. 8991a;_figrge§, gr, 519- 531. Cohen, A. K. (1978). ngiangg_§ng_gggrrgl. Englewood Cliffs: Prentice Hall. Cohn, E. S., Kidder, L. H., & Harvey, J. (1978). Crime prevention vs. victimization prevention: The psychology of two different reactions. yigrimglggy, 3, 285-296. Conklin, J. E. (1975). The_1mp§§r_gr_grimg. New York: Macmillan. Defleur, M. L., & Westie, F. R. (1963). Attitudes as a scientific concept. §Q§1§1_£Qrgg§, 52, 17-31. Deutscher, I. (1969). Looking backward: Case studies on the progress of methodology in sociological research. American_ficciclegx. 1. 35-41. Dubow, F. L., McCabe, E., & Kaplan, G. (1978). Beagrigns to_crime1_A_2ritical_rexiex_2f_the_liferature- Chicago= 76 77 Northwestern University, Center for Urban Affairs, Reactions to Crime Project. Durkheim. E- (1933)- W Glencoe: Free Press. EnniS. P- H- (1967)- W124 States: A report of a national survey, ptepargd for the s' e ' C 'ssion on W Washington. DC: U.S. Government Printing Office. Fowler, F. J., Jr., & Mangione, T. W. (1974). Ing_natgtg 91_tgat. Boston: University of Massachusetts, Survey Research Program. Furstenburg, F. F. (1972). Fear of crime and its effects on citizen behavior. In A. Biderman (Ed.), Qtimg_§ng in§§12g1_A_§xmpg§ium. New York: Nailburg. Garofalo, J. (1980). The fear of crime: Causes and consequences. In J. Dahman & J. Sasfy (Eds.), [1' "I “‘-. ! ‘l e ’1 ‘ Q . '°. 919’ r- . 1o McLean, VA: Mitre Corp. Garofalo, J. (1979). Victimization and the fear of crime. MW. 14. 80-97. GarofaIO. J - (1977)- W (Report SD-VAD-l). Washington, DC: National Criminal Justice Information and Statistics Service. Garofalo, J., & Laub, J. (1978). The fear of crime: Broadening our perspective. yigtigglggy, 1, 242-253. Gordon, M. T., & Riger, S. (1979). Fear and avoidance: A link between attitudes and behavior. yigtimglggy, A, 395-402. Greenberg. 8- W- (1983)- CW WWW- AnnArbor: Inter-University Consortium for Political and Social Research. Greenberg, S. W., Rohe, W. M., & Williams, J. R. (1985). ! '19.: 1 -_ '! {IQ '.'_' 9 ’ ’1 'Q ' ‘Q 1 ° . Washington, DC: U.S. Government Printing Office. ‘ ‘ ‘ Greenberg, S. W., Rohe, W. M., 5 Williams, J. R. (1982). 0 ' ‘ ‘ o .0 -° ‘ .g -_ g- Q ,0... g- . g. 78 . Washington, DC: U.S. Government Printing Office. Gubrium, J. (1974). Victimization in old age: Available evidence and three hypotheses. grimg_gng_nelingg§ngx, 1Q, 245-250. Henig, J., & Maxfield, M. G. (1978). Reducing fear of crime: Strategies for intervention. yigtimglggy, 1, 297- 313. Hindelang, M. J., Gottfredson, M. R., & Garofalo, J. (1978). o - Cambridge: Balinger. Hunter, A., & Baumer, T. L. (1982). Street traffic, social integration, and fear of crime. §9§1g1_1ngnizy,§z, 122- 131. Jacobo. J- (1961)- Ihe_4eath_4n4.11fe_9f_greef_bmer144n gitigg. New York: Vintage. Krahn, H., & Kennedy, L. W. (1985L Producing personal safety: The effects of crime rates, police force size, and fear of crime. C:1ming1ggy,21, 697-710. Lavrakas, P. J. (1981). Household-based responses to burglary- In D- Lewis (Ed-). Beagt12ns_to_4z1me (pp. 67- 85). Beverly Hills: Sage Publications. Lavrakas, P. J., & Herz, E. J. (1982). Citizen participation in neighborhood crime prevention. , 29, 479-498. Lavrakas, P. J., 8 Lewis, D. A. (1980L The conceptualization and measurement of citizens' crime prevention behaviors. 49uInal_ofrnesear4n_1n_9r1me_an4 Delinguensx. 11. 254-272. Lewis, D. A., & Maxfield, M. G. (1980L Fear in neighborhoods: An investigation of the impact of crime. , 160-189. Liska, A. E., Sanchirico, A., & Reed, M. D. (1988L Fear of crime and constraining behavior specifying and estimating a reciprocal effects model. Sggia1_£212§§. §§, 827-837. McIntyre, J. H. (1967). Public attitudes toward crime and law enforcement. Annalg, 111, 34-46. 79 Maxfield, M. G. (1984). The limits of vulnerability in explaining fear of crime: A comparative neighborhood analysis. Besear4h_1n_§rime_an4_nelingnen4¥. 21. 233- 250. Mendelsohn, H., O'Keefe, G. J., 8 Liu, J. (1981L Englig .IIII‘! . 1° - ’0 g o, 0 II" 3‘. . 0!; §n4_§trat_gig§. Denver: University of Denver, Center for Mass Communications Research and Policy. Newman, 0. (1973). Qgfignsitle_§pagg. New York: Cullier. O'Brien. R- M- (1985). Qr1me_an4_21911m12ation_4ata- Beverly Hills: Sage. Iti‘ Riger, S. (1981). The impact of crime on women. In D. Lewis (Ed-). Beagt1ons_to_gr1me (pp- 47-65)- Beverly Hills: Sage. Riger, S., Lebaily, R. K., 8 Gordon, M. T. (1981). Community ties and urbanites' fear of crime: An ecological investigation. Amer1oan_494rnal_2f_§ommunitx WI 2: 653-665 ° Rosenbaum, D. P. (1987). The theory and research behind neighborhood watch: Is it a sound fear and crime reduction strategy? Qr1me_an4_nel1ngnensx.‘22. 103-134- Rosenstock, I. M. (1966). Why people use health services. , $1, 94-124. Schneider, A. L, 8 Schneider, P. R. (1978). Ezixat§_§nd - - -— - - 91.? H. I. 0 ‘1 Song: - Q ago. ,ooo-p.:‘o or1me_nrexention_strafegx- Eugene: Institute of Policy Analysis. Schuman, H., 8 Johnson, M. P. (1976). Attitudes and behavior- Annual_Bexieg_of_§os1olosx. 12. 161-207- Skogan, W. G. (1987). The impact of victimization on fear. £r1me.an4_ne11ngnensx. 22. 135-154. Skogan, W. G. (1981L On attitudes and behaviors. In D. Lewis (Ed ) Beastions.to_srime . 19-45 Beverl Hills: Sage: (pp ) Y Skogan, W. G. (1977L Public policy and the fear of crime in large American cities. In J. A. Gardiner (Ed. ), (pp. 1-17). New York: Praeger. 80 Skogan, W. G., 8 Maxfield, M. G. (1981). Coping with crime. Beverly Hills: Sage. Sparks, R., Genn, H., 8 Dodd, D. (1977). Sptygyipg yipti s. New York: John Wiley. Stinchcombe, A. L., Heimer, C., Iliff, R. 8., Scheppele, K., 8 Smith, T. W. (1978). C me and u ishme a ub opiniop. Chicago: University of Chicago, National Opinion Research Center. Taylor, R. B., Gottfredson, S. D., 8 Brower, S. (1984). Block crime and fear: Defensible space, local social ties and territorial functioning. Journal of Resgatcp in C im and Del' enc , gt, 303-331. Tien, J., Reppetto, T., 8 Hanes, L. (1976). E ements of CPT 2. Arlington: Westinghouse Electric Corp. Wicker, A. W. (1969). Attitudes vs. actions: The relationship of verbal and overt behavioral responses to attitude objects. qurpal of Social Issues, 25, 41-78. nICHIcgN TATE UNIV III/IIIIIIIIII/III:WW“ 31293007597283