Vous êtes sur la page 1sur 25

Soc Just Res (2008) 21:204227 DOI 10.

1007/s11211-008-0069-9

A Multilevel Analysis of the Vulnerability, Disorder, and Social Integration Models of Fear of Crime
Travis W. Franklin Cortney A. Franklin Noelle E. Fearn

Published online: 17 June 2008 Springer Science+Business Media, LLC 2008

Abstract The current research tests three conceptual models designed to explain citizens fear of crimevulnerability, disorder, and social integration. These models are assessed for differential impact across the cognitive and affective dimensions of fear of crime. The analysis reported here considers the consecutive and simultaneous inuence of individual- and city-level factors using multilevel modeling techniques. Recently collected survey data for 2,599 citizens nested within 21 cities across Washington State provide the empirical evidence for the analysis. Results indicate that the disorder model is best able to explain variation in both the cognitive and affective dimensions of citizens fear of crime across cities. The vulnerability and social integration models explain signicantly less variation. Further, the vulnerability model lacks directional consistency across the observed dimensions of fear. Societal implications of the research ndings are discussed. Keywords Fear of crime Victimization Multilevel analysis Vulnerability Disorder Social integration

Fear of crime has been recognized as a signicant social problem, affecting the quality of life across various demographic and socio-economic conditions. Attempts to understand the dynamics underlying the fear of crime have led to numerous empirical and theoretical developments. Three dominant models have emerged as possible explanations of variation in fear of crime among citizensthe vulnerability, disorder, and social integration models. While each of these models has received some
T. W. Franklin C. A. Franklin College of Criminal Justice, Sam Houston State University, Huntsville, TX, USA N. E. Fearn (&) Department of Sociology and Criminal Justice, Saint Louis University, St. Louis, MO, USA e-mail: nfearn@slu.edu

123

Soc Just Res (2008) 21:204227

205

empirical support in the literature (see Hale, 1996 for a comprehensive review), no studies have compared their explanatory power simultaneously in the context of a multilevel analysis,1 and to the best of our knowledge no studies have done so in a predominantly rural setting. As Taylor (2002, p. 774) makes clear, prior research on the fear of crime is based largely on data from well-developed, relatively large cities, raising the need to investigate the determinants of fear in cities whose crime problems may be qualitatively and quantitatively different from those in big cities. An equally pressing concern in this area of research surrounds the operational denitions of fear of crime found within the existing literature. There has been little consensus regarding the most accurate way to capture fear of crime (Ferraro & LaGrange, 1987; Hale, 1996), and some researchers have made use of measures lacking clear and reliable face validity. Early studies often employed a measure that was subsequently identied as more appropriate for capturing perceptions of risk rather than the emotion of fear (e.g., Baumer, 1985; Kennedy & Krahn, 1984; Maxeld, 1984; Yin, 1982). While this research reveals less about feelings of fear, its usefulness has persisted in light of the multidimensional nature of the construct. In this sense, fear of crime is conceptualized as reecting three related dimensions: cognitive, affective, and behavioral (Fattah & Sacco, 1989). The cognitive dimension involves a rational thought process whereby perceptions of risk are developed; the affective dimension recognizes emotions associated with fear; and the behavioral dimension captures physical responses to the situation at hand. Given the multidimensionality of fear, it is important for researchers to distinguish between and draw comparisons across the various dimensions. The purpose of the current study is to compare the efcacy of the vulnerability, disorder, and social integration models across the cognitive and affective dimensions of fear of crime.2 To make these comparisons, the analysis reported here considers the simultaneous impact of individual- and city-level factors using multilevel statistical modeling techniques. Such an analysis will help to clarify our theoretical and empirical understanding of the three hypothesized models and shed light on their differential impact (assessed by the amount of explained variance) on both cognitive and affective dimensions of fear of crime.

Measuring Fear of Crime Previous research exploring the dynamics of the fear of crime has led to a complex of ideas surrounding the appropriate operationalization of the construct. With little consensus regarding the most suitable measure of the fear of crime, the empirical research has evolved with considerable inconsistencies. Initial research employed a simple unidimensional measure of fear of crime derived from the National Crime
1 Rountree and Land (1996a) explored measures associated with each model of fear of crime; however, a systematic comparison of the models was not the focus of their study. Further, their measures of disorder and social integration were assessed at the neighborhood level, while the current study is concerned with individual-level measures of perceived disorder and social integration. 2

While it would be ideal to make comparisons across all three dimensions of the fear of crime, the data set for the current analysis only contains information capturing the cognitive and affective dimensions.

123

206

Soc Just Res (2008) 21:204227

Victimization Survey (NCVS) (e.g., Baumer, 1985; Kennedy & Krahn, 1984; Maxeld, 1984; Yin, 1982). In these studies, respondents were asked some variation of the following question: How safe do you feel or would you feel walking alone in your neighborhood at night? As noted by Garofalo (1979) and further discussed by Ferraro and LaGrange (1987), such a measure introduces various problems. While it is not necessary to repeat the various limitations associated with this operational denition, it is sufcient to note its failure to distinguish between emotional fear of crime and cognitive judgments concerning risk of crime victimization. More recent research, including the study presented here, distinguishes between measures of perceived risk and emotional fear, avoiding the earlier ambiguity surrounding the NCVS-based measure (LaGrange, Ferraro, & Supancic, 1992; Rader, 2004; Rountree, 1998; Rountree & Land, 1996a; Williams, McShane, & Akers, 2000). Recent research has also moved away from abstract conceptualizations of fear to more clearly specied, concrete conceptualizations (LaGrange et al., 1992; Rountree, 1998; Rountree & Land, 1996a). Garofalo and Laub (1978) note that operational denitions based on how afraid one feels in his/her neighborhood fail to question respondents about concerns regarding specic crimes. Instead, they invoke responses concerning a formless or global feeling of fear, making it difcult or impossible to identify precisely what it is that respondents fear (Hale, 1996, p. 85). To remedy this concern, many contemporary researchers have suggested a more concrete measure based on multiple items tapping fear of specic crimes (Ferraro & LaGrange, 1987). In order to address this issue, the present analysis employs a measure of fear (worry of victimization) based on multiple crime scenarios.

Explaining Fear of Crime Numerous theoretical developments have emerged to explain the various dynamics of citizens fear of crime. Generally speaking, these theoretical frameworks fall into two broad categories. The rst category incorporates theories focusing on facilitators of fear or factors that would rationally lead one to be more (vs. less) fearful (e.g., increased vulnerability, disorderly local surroundings). The second category incorporates theoretical developments whereby fear of crime is understood through characteristics that inhibit or reduce the grounds for fear (e.g., social ties, neighborhood cohesion, collective efcacy, and community attachment). Three specic theoretical approaches to understanding fear of crimethe vulnerability, disorder, and social integration modelshave emerged as relatively dominant in the literature, with the rst two models focusing on facilitators of fear and the latter focusing on inhibitors of fear. Vulnerability Model A substantial research literature indicates that perceptions of personal vulnerability facilitate fear of crime (see Hale, 1996 for a review). Scholars have argued that individuals who feel unable to protect themselves through physical, social, and/or economic resistance may report higher levels of fear than those who feel that they

123

Soc Just Res (2008) 21:204227

207

have the capacity for self-protection. This concept of perceived vulnerability has been divided into two main categoriesphysical and social vulnerability. Physical vulnerability pertains to the perception of increased risk to physical assault. This stems from the decreased ability to fend off attack due to limited mobility or a lack of physical strength and competence. Accordingly, gender and age affect fear of crime, as women and the elderly likely feel less capable of physically protecting themselves when compared to those who are younger and/or male (Denkers & Winkel, 1998; Ferraro & LaGrange, 1992; Fisher & Sloan, 2003; Gilchrist, Bannister, Ditton, & Farrall, 1998; Ginsberg, 1985; Hughes, Marshall, & Sherrill, 2003; Kennedy & Silverman, 1985; Killias & Clerici, 2000; Smith & Torstensson, 1997; Warr, 1984; Yin, 1982). Social vulnerability assumes increased exposure to victimization as a result of a range of factors. For example, living in economically distressed, high-crime neighborhoods often presents increased potential for victimization.3 In addition to residing in high-crime neighborhoods, individuals lacking the material resources necessary to protect their homes and/or recoup nancial losses in the event of victimization may feel increased social vulnerability. Finally, those decient in material and social resources or community and political networks that enable them to cope successfully with anxiety-provoking situations (e.g., individual and institutionalized racism) are likely to experience increased social vulnerability. Consequently, racial and ethnic minorities, people living in poverty, and those with lower educational levels may report higher levels of fear of crime than their counterparts who are white, afuent, and well educated. These assumptions have been supported in previous research (Baumer, 1978; Clement & Kleiman, 1977; Covington & Taylor, 1991; Erskine, 1974; Furstenberg, 1971; Jaycox, 1978; Pantazis, 2000; Parker & Onyekwuluje, 1992; Skogan & Maxeld, 1981; Taylor & Hale, 1986; Will & McGrath, 1995). Despite the empirical support afforded the vulnerability model, some researchers have questioned its theoretical value based on ndings that those who experience higher levels of fear (women and the elderly) are, in fact, the least likely segment of society to be victimized (Fattah & Sacco, 1989). Moreover, those who are most likely to be victimized (young men) report lower levels of fear (Garofalo & Laub, 1978). This set of ndings has been referred to as the fear-victimization paradox (see Hale, 1996). Attempts to resolve this apparent illogicality have been numerous and sometimes insightful. Sacco (1990) identies two explanations of the fear-victimization paradox that shed light on womens fear of crime. First, scholars argue that ofcial crime data fail to capture the full extent of female victimization (e.g., rape and domestic abuse are highly underreported). Thus, hidden violence is not appreciated when determining the rationality of womens fear. Once these unreported threats are acknowledged, the level
3

Because poorer individuals are also more likely to have experienced victimization as compared to wealthier individuals, it could be argued that social vulnerability (as captured by income) should only be related to fear of crime through the experience of past victimization. We would like to note, however, that past victimization has been argued to inuence fear of crime by inciting feelings of vulnerability concerning future victimization. Thus, victimization appears to operate through vulnerability to cause feelings of fear. For further discussion, see the work of Gibson, Jihong, Lovrich, and Gaffney (2002).

123

208

Soc Just Res (2008) 21:204227

of womens fear is more appropriate and reasonable, leaving the charge of illogicality unfounded. The second explanation assumes differences between fear of crime and reported risk of victimization. While women do experience lower rates of general violence, they are disproportionately the victims of sexual crimes. Consequently, womens heightened fear may arise from an increased level of personal vulnerability directly related to the sexual nature of their experienced threats (Gordon & Riger, 1989; Junger, 1987; Stanko, 1987, 1990a, b; Warr, 1984, 1985). Related arguments can be advanced that fear among the elderly must be understood through their differential sensitivity to risk, such that similar levels of risk do not necessarily produce similar levels of fear (Warr, 1984, p. 695). Moreover, Fattah and Sacco (1989) argue that the widespread use of global or formless measures of fear of crime have limited our understanding of what stimulates fear of crime among the elderly. Inquiries concerning the safety one feels in his/her neighborhood after dark likely provide unrealistic scenarios that are largely irrelevant to elderly respondents who are unlikely to traverse the streets after dark (Fattah & Sacco, 1989). The use of such measures may produce an inated level of fear among the elderly. In fact, analyses employing more crime-specic measures of fear have found age to be a poor predictor of fear of crime (LaGrange & Ferraro, 1989). Disorder Model The disorder model originates from Shaw and McKays (1942) work on social disorganization, wherein facilitators of fear were grounded in perceptions of local surroundings, specically signs of physical and social disorder (Skogan, 1990). The basic assumption of this model is that neighborhood incivilities are the manifestations of disorder that threaten individual residents even more than the actual experience of crime. The physical decay and deterioration of a neighborhood signies a lack of local concern and the absence of informal social controls, leading to citizen perceptions of neighborhood disorder. Researchers have divided incivilities into two conceptual categoriessocial and physical incivility (Burby & Rohe, 1989; LaGrange et al., 1992). Social incivility refers to disruptive behaviors such as loiterers, inconsiderate neighbors, loose dogs, unsupervised and/or unruly teenagers, gangs, beggars, and public drinking. Physical incivilities refer to disorderly surroundings such as abandoned cars, vandalized property, trash, vacant houses, and deteriorated homes. Neighborhood residents who perceive disordered social and physical local surroundings are more likely to exhibit higher levels of fear (Gates & Rohe, 1987; LaGrange et al., 1992; Lewis & Salem, 1986; Skogan, 1990; Skogan & Maxeld, 1981). Furthermore, residents may perceive themselves to be at increased risk of victimization in areas in which there are visible signs of community disorder (Covington & Taylor, 1991; Lewis & Salem, 1986). Perceptions of disorder likely translate into environmental uncertainty and perceived threats to personal safety (Kennedy & Silverman, 1985). Skogan and Maxeld (1981) and Lewis and Maxeld (1980) report that residents in lower-crime neighborhoods frequently identify incivilities or disorderly neighborhood conditions as problems that are of greater concern than crime. These ndings reiterate the importance of Wilson and

123

Soc Just Res (2008) 21:204227

209

Kellings (1982) broken windows theory, which posits a strong connection between disorderly surroundings and fear of crime. Social Integration Model Shifting from facilitators of fear to inhibitors of fear, the social integration model purports that those who are socially integrated within their neighborhoods experience lower levels of fear of crime than those who are not as well integrated (Hartnagel, 1979; Lewis & Salem, 1986; Riger, LeBailly, & Gordon, 1981; Rountree & Land, 1996b). Social integration has been dened as a persons sense of belonging to their local surroundings as well as their attachment to the community (Adams, 1992; Kasarda & Janowitz, 1974; Keyes, 1998). Prior research has operationalized social integration as the ability to identify strangers in the area and the degree to which neighbors feel they are a part of the neighborhood (Hunter & Baumer, 1982). Other researchers have dened social integration as possessing personal investment in the neighborhood, having social ties to neighbors, and feeling emotional attachment to the community (Kanan & Pruitt, 2002). Additional social integration measures have included participation in formal organizations (Austin, Woolever, & Baba, 1994), involvement in neighborhood activities, engaging in neighborhood information sharing, the perception of similarities among residents, and the presence of friends or relatives living in the neighborhood (Bursik & Grasmick, 1993). In sum, residents who become familiar with their neighbors and develop connectedness to their neighborhood should report lower levels of fear than those who do not. Empirical research testing the social integration model has produced somewhat mixed results, though substantial evidence appears to suggest an inverse relationship between levels of social integration and fear of crime (Austin et al., 1994; Baba & Austin, 1989; Hunter & Baumer, 1982; Kanan & Pruitt, 2002; Krannich, Berry, & Greider, 1989; McGarrell, Giacomazzi, & Thurman, 1997; Rountree & Land, 1996b). Bursik and Grasmick (1993) and Gibson et al. (2002) argue, however, that prior measures of social integration lack methodological consistency, thus making subjective cross-study comparisons more difcult. More specically, when researchers employ different measures of social integration and reach dissimilar conclusions concerning its effect on fear of crime, it is not readily apparent if these differences are attributable to the differing methodologies or are, in fact, real differences in how social integration is operating from one study to the next. This limitation makes it more difcult to draw solid conclusions about the effect of social integration on fear of crime, although the evidence is generally supportive of the social integration model (e.g., Hale, 1996).

Data and Methodology Individual-level data for the current analysis were derived from the 2003 Eastern Washington Crime and Criminal Justice System Survey of 2,861 residents. The survey was conducted by Washington State Universitys Division of Governmental Studies and Services for the United States Attorneys Ofce for the Eastern District

123

210

Soc Just Res (2008) 21:204227

of Washington. The survey respondents were clustered within 21 cities, of which 5 can be classied as metropolitan areas and 16 as rural areas.4 City-level data were derived from two sources. First, basic demographic information was taken from the 2000 U.S. Census for each city included in the analysis. Second, ofcial crime statistics for each location were provided by the Crime in Washington 2004 Annual Report produced by the Washington Association of Sheriffs and Police Chiefs (WASPC). Due to randomly missing individual-level data, the nal sample for statistical analysis includes 2,599 residents located within 21 cities. Dependent Variables Perceived Risk The rst dependent variable is a global measure of perceived risk and represents the cognitive dimension of fear. It is a multiple-item index based on two questions originating from the NCVS.5 Specically, respondents were asked How safe would you feel walking alone during the day [night] in the area where you live? Responses were summed to create a scale ranging from 2 (very unsafe) to 10 (very safe), and reliability tests indicated acceptable internal consistency (Cronbachs alpha = .72). As previously discussed, utilizing such a measure has received ample criticism (see Ferraro & LaGrange, 1987), particularly when the desired outcome measure is emotional fear as opposed to a cognitive perception of risk. Despite the limitations of a global measure, perceived risk is included in the current analysis to allow for a comparison across the cognitive and affective dimensions of fear. Additionally, it provides a baseline for comparison across studies, as multiple researchers have incorporated a similar measure. Worry of Victimization The second dependent variable represents the affective dimension of fear and is based on a seven-item index capturing respondents frequency of worry about becoming the victim of specic crime scenarios (e.g., being burglarized while someone is at home). Respondents were asked How much do you worry about each of the following situations? Responses were summed to create a scale ranging from 7 (never) to 28 (very frequently), and reliability of the measure demonstrated strong internal consistency (Cronbachs alpha = .89). The resulting operational denition offers at least two benets to the current analysis. First, perceived risk is general in nature and requires respondents to speculate about how safe they would feel (hypothetical) in a particular situation,
4

Survey instruments were mailed to a random sample of household addresses (extracted from local telephone directories) within each of the cities included in the analysis presented here. From the 8,836 correct addresses identied, 2,861 respondents completed and returned a survey instrument, yielding a response rate of 32.4%. Thus, caution must be taken when generalizing the survey ndings. For a complete list of scale items, see Appendix A.

123

Soc Just Res (2008) 21:204227

211

whereas worry of victimization is more specic and taps into the amount of worry respondents (actually) do feel. Second, the former measure is cognitive, asking respondents to make a judgment concerning their safety, whereas the latter measure is arguably affective (see Ferraro & LaGrange, 1987; Rountree & Land, 1996a; Taylor & Hale, 1986), tapping into the emotional aspect of fear. Individual-Level Independent Variables Vulnerability Race, age, sex, education, and income are included in the analysis as proxy measures of individual vulnerability to criminal victimization.6 Due to the minimal presence of minorities in the sample, race is a dichotomous variable coded as 0 (other) and 1 (White). Age was captured by converting the year of birth to the respondents age at the completion of the survey. Sex was coded as 0 (female) and 1 (male). Educational achievement was captured on a scale of 1 (less than high school) to 7 (graduate degree). Finally, annual income was measured on a scale of 1 (less than $10,000) to 10 (more than $90,000). Although past victimization has been argued to inuence fear of crime through increased feelings of vulnerability to future victimization, a reliable measure of previous victimization was not available in the present data. Despite this apparent shortcoming, several studies have called into question the strength of the association between actual victimization and fear of crime (e.g., Baumer, 1985; Hindelang, Gottfredson, & Garofalo, 1978; McGarrell et al., 1997). Researchers questioning the direct relationship between victimization and fear of crime have pointed to empirical evidence demonstrating either a weak or nonexistent correlation (e.g., Gibson et al., 2002). Given these ndings, the absence of a previous victimization measure should not pose signicant shortcomings to the current study. Table 1 displays descriptive statistics for all variables included in the analysis. Disorder Perceived disorder or incivility was measured by summing the responses to eight questions regarding the seriousness of neighborhood problems. The resulting scale ranged from 8 (no problem) to 32 (a serious problem) and demonstrated acceptable internal consistency (Cronbachs alpha = .83). Although previous research has distinguished between perceptions of physical disorder (e.g., trash, abandoned buildings, vandalism) and perceptions of social disorder (e.g., public drunkenness,

In the absence of more direct measures of vulnerability to criminal victimization, race, age, sex, income, and education are included in the present analysis. It should be noted, however, that these items are not treated as a unied scale measuring vulnerability. For increased accuracy, future analyses should attempt to capture more direct measures of vulnerability that would be suitable for scaling as a single construct.

123

212 Table 1 Descriptive statistics

Soc Just Res (2008) 21:204227

Variables Dependent variables (N = 2,599) Perceived risk Worry of victimization Individual-level variables (N = 2,599) Race Non-White (0) White (1) Gender Female (0) Male (1) Age Income Less than $10,000 (1) $10,000$19,999 (2) $20,000$29,999 (3) $30,000$39,999 (4) $40,000$49,999 (5) $50,000$59,999 (6) $60,000$69,999 (7) $70,000$79,999 (8) $80,000$89,999 (9) More than $90,000 (10) Education Less than high school (1) High school graduate (2) Some college (3) Associate degree (4) Bachelor degree (5) Some graduate coursework (6) Graduate degree (7) Disorder Social integration City-level variables (N = 21) Violent crime rate Property crime rate Unemployment Urbanism Rural (0)

Mean

Standard deviation

3.48 14.95

1.44 4.02

(6.5%) (93.5%) (36.1%) (63.9%) 57.99 6.12 (1.4%) (3.0%) (5.7%) (9.2%) (8.8%) (34.5%) (16.3%) (5.3%) (3.5%) (10.4%) 4.12 (4.7%) (18.0%) (25.7%) (8.4%) (17.2%) (6.9%) (19.0%) 12.58 -.01 3.10 55.08 6.05 (76.2%) (23.8%) 4.45 2.95 1.76 20.11 1.76 1.88 15.64 2.03

Note: Total sample size is 2,599 citizens and 21 cities

Urban (1)

noisy neighbors; e.g., see Lagrange et al., 1992; Taylor & Hale, 1986), our exploratory factor analysis indicated that all eight itemswhether physical or social in natureloaded onto a single factor. This is consistent with more recent research,

123

Soc Just Res (2008) 21:204227

213

in which measures of social and physical disorder were also found to represent a single underlying construct (Ross & Mirowsky, 1999). Social Integration Social integration was captured through responses to four questions that were derived from prior literature (Gibson et al., 2002; McGarrell et al., 1997) examining the link between social integration and fear of crime: (1) Would you describe the area where you live as a place where people help one another or a place where people mostly go their own way? (2) Do you feel the area where you live is more of a real home or more like just a place to live? (3) How often do you talk with your neighbors? and (4) When you do a favor for a neighbor, can you trust the neighbor to return the favor? Individuals scoring higher on the resulting scale demonstrated higher overall levels of social integration. To account for the differing metrics in which the questions were measured, responses were standardized across items. Reliability tests indicated acceptable internal consistency (Cronbachs alpha = .71). City-Level Independent Variables Several city-level variables were included in the analysis to control for potential contextual effects on perceived risk and worry of victimization.7 Past research has suggested that community-level characteristics, particularly those related to social disorganization and the breakdown of informal social control, may lead to increased perceptions of risk and fear of crime (Lee & Ulmer, 2000). For this reason, the present study includes measures of the violent crime rate, property crime rate, unemployment rate, and urbanism, all of which have been linked to social disorganization (Bursik & Grasmick, 1993; Sampson & Groves, 1989; Skogan & Maxeld, 1981). Specically, the violent and property crime rates were based on the average rate of reported crimes per 1,000 persons for the 3 years prior to the collection of survey data (20002002). Unemployment rates for each of the cities were obtained from the 2000 U.S. Census. Finally, areas identied by the 2000 U.S. Census as metropolitan locations were considered to be urbanized and were coded 0 (rural) or 1 (urbanized).8

Ideally neighborhood-level contextual factors would be used in the current analysis, but such information was unavailable. Thus, city-level contextual factors were included in their place. This raises concern over the relationship between city-level factors (e.g., crime rate) and fear experienced within smaller special regions (e.g., perceived safety in the area where respondents live). Hypothetically, one could live in an afuent, safe neighborhood nested within a dangerous city, attenuating the inuence of city-level factors. In the current analysis, however, the majority of the cities are rural towns with populations of less than 10,000 residents, arguably creating a situation where city-level factors, particularly crime rates, have widespread inuence over fear of crime, despite location within the town. In accordance with the 2000 U.S. Census, the current study operationalized metropolitan areas as locations with a population nucleus of at least 50,000 people. Locations that were too small to meet the requirements for a metropolitan area were considered to be rural, with populations ranging from approximately 1,000 to 29,000 residents.

123

214

Soc Just Res (2008) 21:204227

Analytic Strategy To determine the explanatory power of the vulnerability, disorder, and social integration models across the cognitive and affective dimensions of fearwhile controlling for city-level contextual effectsmultilevel modeling was employed. Hierarchical modeling has become the standard method used to estimate the effects of community-, county-, and city-level factors on individual outcomes, particularly when the data contain a substantial amount of clustering within cities, as in the present study (Bryk & Raudenbush, 2002). These models not only provide an efcient illustration of the degree to which a given individual-level outcome varies across geographic areas, but also formally adjust for the non-independence of sample members living within the same city. The failure to model this type of nonindependence can result in estimated standard errors that are biased downward; consequently, conclusions regarding the statistical and substantive importance of either individual- or city-level factors may be misleading (Baumer, Messner, & Felson, 2000; DiPrete & Forristal, 1994; Snijders & Bosker, 1999). The analysis reported here uses hierarchical linear random-intercept models to evaluate the degree to which perceived risk and worry of victimization vary across the cities included in the data (for detailed explanations of these models, see Bryk & Raudenbush, 2002). First, intercept-only models are estimated to determine the baseline variation in the dependent variables. Second, each of the three models of fearvulnerability, disorder, and social integrationare independently specied to determine the variation explained by each model separately. Third, randomintercept models simultaneously estimating the effects of the vulnerability, disorder, and social integration models are specied. Finally, full random-intercept models are estimated to determine the inuence of the vulnerability, disorder, and social integration models net of city-level contextual factors and to determine the degree to which both level-one (individual) and level-two (city) variables account for variation in the dependent variables across cities.9

Results Assessing Baseline Variation Prior to assessing the inuence of individual- and city-level characteristics on the cognitive and affective dimensions of fear, it was rst necessary to evaluate the degree to which the dependent measures actually vary across the cities included in the analysis. Table 2 displays the results of two baseline hierarchical linear models presenting the intercept (which describes the mean level of each dependent variable) and variance components (which describes the amount of variation across cities) for perceived risk and worry of victimization. Illustrating the necessity for subsequent
9

Ideally, theoretically interesting cross-level interactions (e.g., the potential interaction between crime rates and individual perceptions of disorder) would be specied in the current study; however, the small sample size at level two (N = 21) was not large enough for such an analysis.

123

Soc Just Res (2008) 21:204227 Table 2 Intercept-only hierarchical models for perceived risk and worry of victimization

215

Perceived risk Fixed effects Intercept Random effects 3.31* (.08) .10 457.10*

Worry of victimization

14.42* (.23) 1.07 3137.41*

Note: Standard errors are displayed in parentheses * p \ .05

Variance component Chi-square

analyses, the random effects variance components and corresponding test statistics indicate signicant variation in both dimensions of fear across the 21 cities (Perceived Risk: s00 = .102, v2 = 457; Worry of Victimization: s00 = 1.066, v2 = 3,137). Assessing the Conceptual Models Table 3 separately compares the effects of the vulnerability, disorder, and social integration models on respondents perceived risk of victimization. The results from this portion of the analysis help illustrate which of the three models best explains the observed variation in the dependent variable. Consistent with the vulnerability model, race, gender, age, income, and education signicantly inuenced respondents perception of risk. More specically, minorities, women, those who are older, those with lower incomes, and those with lower levels of education were more likely to report higher levels of risk as compared to their counterparts. Although the vulnerability measures were signicant and directionally accurate, the random effects portion of the table indicates that the vulnerability model accounts for a modest proportion of the variation in respondents perceived risk across cities. More
Table 3 Individual-level theoretical models explaining perceived risk Intercept Fixed effects Race Gender Age Income Education Disorder Social integration Random effects Intercept Note: Standard errors are displayed in parentheses * p \ .05 Variance component Chi-square .09 377.31* .06 279.49* .08 354.83* -.35* (.08) -.80* (.04) .01* (.00) -.09* (.01) -.09* (.01) .13* (.00) -.15* (.01) Vulnerability Disorder Social integration 3.35* (.07)

4.67* (.15)

1.72* (.08)

123

216

Soc Just Res (2008) 21:204227

specically, the variables associated with the vulnerability model were only able to reduce the variance component by 12%. As predicted by the disorder and social integration models, the associated measures included in the analysis were found to affect citizens levels of perceived risk. Respondents who viewed their neighborhoods as characterized by high levels of disorder reported signicantly higher levels of perceived risk. Moreover, those who viewed their neighborhoods as less socially integrated also reported signicantly higher levels of perceived risk. Examination of the random effects portion of the table for these two models indicates that respondents perception of their neighborhoods level of disorder and social integration explained a much larger portion of the variation in the dependent variable as compared to the vulnerability measures. Specically, the disorder model accounts for the largest portion of the variation in perceived risk across cities (44%), followed by the social integration model (25%). Table 4 separately assesses the same three modelsvulnerability, disorder, and social integrationwith worry of victimization as the dependent variable. Results indicated that only two of the ve vulnerability measures signicantly affected respondents worry of victimization. Women and younger individuals reported higher levels of worry of victimization as compared to men and older individuals. While these results provide strong support for our expectation that women are more likely than men to worry about victimization (due to their physical vulnerability), the negative age effect contradicts the vulnerability-derived assumption that elderly individuals possess greater worries of victimization. Finally, the random effects portion of the table indicates that the vulnerability model does not account for much variation (about 1%) in respondents worry of victimization across cities. The remaining two models included in Table 4 suggest that both the disorder and social integration measures inuenced respondents worry of criminal victimization.

Table 4 Individual-level theoretical models explaining worry of victimization Intercept Fixed effects Race Gender Age Income Education Disorder Social integration Random effects Intercept Note: Standard errors are displayed in parentheses * p \ .05 Variance component Chi-square

Vulnerability

Disorder

Social integration 14.47* (.22)

16.13* (.26) .02 (.08) -.92* (.04) -.02* (.00) .01 (.01) -.02 (.01)

10.75* (.19)

.30* (.00) -.14* (.01)

1.05 2948.27*

.68 2071.32*

.97 2820.14*

123

Soc Just Res (2008) 21:204227

217

Specically, citizens who perceived their neighborhoods as characterized by higher levels of disorder, as well as lower levels of social integration, were signicantly more likely to report worry of victimization. The random effects portion of the table indicates that the disorder and social integration models account for 36% and 8% of the variation in worry of victimization, respectively. As with the analysis of perceived risk, the disorder measure appears to be the most powerful predictor of worry of victimization across cities. To further assess whether one theoretical framework has greater explanatory power than the others, additional tests were conducted. Table 5 presents the results of two models that assess, simultaneously, the effects of the vulnerability, disorder, and social integration models to determine whether or not the effects of one model attenuates the effects of the other(s). The results suggest that including the vulnerability, disorder, and social integration measures in a single model helps to further explain the city-level variation in both perceived risk and worry of victimization.10 None of the previously statistically signicant variables became non-signicant in the presence of the additional individual-level predictors. In fact, three of the vulnerability measuresrace, income, and educationbecame signicant predictors of worry of victimization when simultaneously assessed with the disorder and social integration measures. Thus, inclusion of the disorder and social integration variables actually increased the inuence of several vulnerabilityrelated variables on worry of victimization, although in the opposite directions predicted by the vulnerability model.11 The simultaneous analysis of the vulnerability, disorder, and social integration variables explained signicantly more city-level variation in the dependent variables when compared to the separate analyses of each model. The random effects portion of Table 5 indicates that when all measures were considered in a single model, the variance component was reduced by 63% for perceived risk and 42% for worry of victimization. Despite the relatively large portion of variation explained by the theoretical models, the variance components for both perceived risk and worry of victimization remained signicant; this indicates that additional unaccounted for factors were inuencing the dependent variables across cities. The results presented thus far indicate that individual-level factors have important inuences on both the cognitive (perceived risk) and affective (worry of victimization) dimensions of fear of crime. These variables also help to account for the variation in responses across cities. However, it is possible that individuallevel factorsspecically our measures of vulnerability, disorder, and social integrationmay be affected by contextual factors that vary across cities. That is, respondents in cities characterized by certain features (e.g., high crime rates, high
It should also be noted that multicollinearity diagnostics were examined for both individual- and citylevel variables. Tolerances ranged from .54 to .98 and Variance Ination Factors ranged from 1.0 to 1.8, indicating that multicollinearity was not a concern in the present analysis. Moreover, bivariate correlations for all independent variables are presented in Appendix B. Additional analyses (not shown in tabular form) indicate that the changes in the effects of race, income, and education are primarily due to the inclusion of the disorder measure. More specically, once disorder is included in the worry of victimization model, race, income, and education exert positive effects on respondents worry of victimization.
11 10

123

218 Table 5 Combined individuallevel models explaining perceived risk and worry of victimization

Soc Just Res (2008) 21:204227

Perceived risk Intercept Fixed effects Race Gender Age Income Education Disorder Social integration Random effects Intercept -.17* (.08) -.72* (.04) .01* (.00) -.03* (.01) -.06* (.01) .11* (.01) -.11* (.01) 2.37* (.16)

Worry of Victimization 10.66* (.23) .31* (.08) -.73* (.04) -.01* (.00) .11* (.01) .03* (.01) .29* (.01) -.04* (.01)

Note: Standard errors are displayed in parentheses * p \ .05

Variance component Chi-square

.04 194.07*

.62 1917.53*

levels of unemployment, urbanization) may report higher or lower levels of perceived risk or worry of victimization despite the effects of individual-level characteristics. Thus, Table 6 provides the results of the nal analyses which address three main issues: (1) whether key features of respondents location inuence reported levels of perceived risk or worry of victimization; (2) whether the inclusion of city-level variables inuences the effects of individual-level characteristics on perceived risk or worry of victimization; and (3) whether the inclusion of both individual- and city-level factors help to explain a larger portion of the variation in reported levels of perceived risk or worry of victimization.12 The results presented in the rst column of Table 6 indicate that the inclusion of city-level variables does not substantially alter the effects of individual-level characteristics on perceived risk as reported in Table 5. Virtually all of the individual-level factors exert the same effect on perceived risk when violent and property crime rates, unemployment rates, and urbanism of the respondents cities are included in the analysis. However, one of the city-level variablesurbanism does exert a statistically signicant positive effect on respondents perceived risk. Specically, citizens living in urban locations reported higher levels of perceived risk as compared to those living in rural locations. Moreover, the random effects for this model indicate that the inclusion of both individual- and city-level variables accounts for nearly all of the variation in perceived risk across the cities included in the analysis (96%). The results presented in the second column of Table 6 also indicate that including city-level variables does not inuence the effects of individual-level
12 Due to the small number of level-two groups available for analysis (N = 21) and the consequent reduction in statistical power (estimated at .44 for a weak correlation of .20), we caution the reader that statistical signicance reported at level-two of the analysis is more conservative than traditionally desired (increasing the chance of Type II error). Thus, weaker relationships at level-two may not be detected in the current analysis.

123

Soc Just Res (2008) 21:204227 Table 6 Complete hierarchical models explaining perceived risk and worry of victimization

219

Perceived risk Fixed effects Intercept Individual-level variables Race Gender Age Income Education Disorder Social integration City-level variables Violent crime Property crime Unemployment Urbanism Random effects Intercept .03 (.03) .00 (.00) .02 (.02) .25* (.08) -.16* (.08) -.72* (.04) .01* (.00) -.03* (.01) -.06* (.01) .11* (.01) -.10* (.01) 1.86* (.23)

Worry of victimization

10.08* (.67) .31* (.08) -.73* (.04) -.01* (.00) .11* (.01) .03* (.01) .29* (.01) -.04* (.01) .20 (.13) .00 (.01) -.08 (.10) .24 (.41)

Note: Standard errors are displayed in parentheses * p \ .05

Variance component Chi-square

.01 30.42*

.50 725.39*

characteristics on worry of victimization, as observed in earlier analyses (see Table 5). The random effects for this model indicate that the inclusion of both individual- and city-level variables accounted for a signicant portion of the variation in reported worry of victimization across the 21 cities (53%). Despite this observed explanatory power, the variance component for the worry of victimization model remained statistically signicant (v2 = 725.39), indicating the presence of unexplained variation in the worry of victimization measure across cities.

Discussion The primary objective of this analysis was to assess the explanatory power of three conceptual models with respect to two separate dimensions of the fear of crime. The analysis revealed that the disorder model accounts for the greatest proportion of variation in both dimensions of fear of crimecognitive and affectiveat the city level. In other words, individual perceptions of neighborhood disorder, such as noise, trafc problems, and youth gangs, appeared to be the most powerful determinant of fear of crime. Levels of individual social integration also appeared to be an important determinant, but signicantly less so than perceptions of neighborhood disorder. The cluster of variables associated with the vulnerability model proved to be least useful for explaining variation in fear of crime across the 21 cities analyzed.

123

220

Soc Just Res (2008) 21:204227

Although the vulnerability, disorder, and social integration models were assessed as separate or distinct models in the present analysis, a particular limitation of this approach should be noted. Specically, it is plausible that the three models overlap to some degree with one another, rather than existing as purely distinct frameworks. For example, individuals who are more socially integrated into their neighborhoods may feel less vulnerable to victimization due to the availability of social support from their neighbors. Socially integrated individuals may also feel an increased ability to cope with troublesome situations due to a sense of belonging or community (see Hale, 1996). Along these same lines, socially integrated residents who participate within the community may become familiarized with signs of social and physical incivilities, consequently reducing the inuence of disorder on fear of crime (Riger, LeBailly, & Gordon, 1981). Despite this potential overlap or interaction between vulnerability, social integration, and perceived disorder, each of the three theoretical models predicted signicant direct effects on fear of crime. While the disorder and social integration models behaved similarly across measures of perceived risk and worry of victimization, variables associated with the vulnerability model displayed directional changes across the two dependent variables. When examining the effects of the vulnerability-related measures on the cognitive dimension of fear (perceived risk), directional accuracy was observed. As expected, minorities, females, the elderly, and those with lower levels of income and education reported higher levels of perceived risk. The directional accuracy, however, was diminished when examining the effects of these variables on the affective dimension of fear (worry of victimization). Specically, the effects of race, age, income, and education were signicant, but they were in the opposite directions predicted by the vulnerability model. Thus, it appears that variables associated with the vulnerability model operate differently across the dimensions of fear analyzed in the present analysis. It should be noted that previous research has discovered similar ndings. For example, LaGrange et al. (1992) found age to negatively inuence the affective dimension of fear of crime while positively inuencing the cognitive dimension. Moreover, Rountree and Land (1996a) reported that the effects of sociodemographic variables, such as age and income, varied signicantly across the two measures of fear of crime. Consistent with LaGrange et al. (1992), age was found to negatively impact the affective dimension of fear (operationalized as burglary-specic fear), while its inuence on the cognitive dimension was not statistically signicant. Further, income was found to affect perceptions of risk negatively but to positively affect burglary-specic fear. While these ndings are contrary to much of the earlier literature in support of the vulnerability model, Rountree and Land (1996a) point out that many earlier analyses have been limited to cognitive assessments of perceived risk. This observation is supported by the current analyses, which suggest that the vulnerability model maintains directional accuracy when predicting levels of perceived risk but not when predicting levels of emotional fear or worry of victimization. While it is evident that measures of vulnerability differentially inuence the cognitive and affective dimensions of fear of crime, it is less evident why this may

123

Soc Just Res (2008) 21:204227

221

be the case. One possible explanation stems from LaGrange and Ferraros (1989) work. These researchers contend that global measures of fear, such as the measure of perceived risk in the current analysis, lack relevance to the everyday lives of many people. For instance, they inquire about feelings of safety when walking alone on the streets at night; this event is probably infrequent for many neighborhood residents. Moreover, global measures forgo specicity, allowing unwanted exibility in respondents interpretation of the items. By contrast, concrete measures of fearbased on specic crime scenariosleave little room for variation in respondents interpretations. Another plausible explanation concerns the process of desensitization potentially experienced by minorities and those with lower levels of income and education. Such individuals disproportionately live in crime-prone neighborhoods, where levels of risk are relatively high. Citizens rooted in these neighborhoods or those who have spent lengthy periods of time in high-crime locations may begin to view their surroundings as normal and, thus, experience lower levels of fear of crime as a consequence. As they become desensitized to their surroundings, it may be possible for them to maintain an understanding of their higher-risk level while experiencing lower levels of actual worry or emotional fear of crime. The statistically nonsignicant inuences of violent and property crime rates on perceived risk and worry of victimization may be a result of the small sample size at level two of the analysis, resulting in reduced power to reveal signicant ndings. The nonsignicant ndings may also be a byproduct of the measures of fear of crime utilized.13 Perceived risk is a non-specic global measure, and worry of victimization is based on a multiple-item scale, which means that the effects of violent and property crime on specic types of fear of crime could be masked. For example, Rountree (1998) found neighborhood-level property crime to have a statistically signicant effect on burglary-specic fear but not on fear of violent crime. Thus, it is safe to conclude that while violent and property crime rates did not affect our multiple-item measures of fear of crime, they may be important for understanding fear of specic violent and property crimes separately. Aside from property and violent crime rates, two other city-level factors were included as controls in the analysisnamely, unemployment and urbanism. Rates of unemployment displayed no signicant effect on levels of perceived risk or worry of victimization; urbanism, however, did exhibit a strong positive effect on levels of perceived risk. This nding is congruent with past research positing increased levels of risk in urban areas (e.g., Baumer 1978; Belyea & Zingraff, 1988; Sacco, 1985) and is not surprising considering the higher crime rates and lower levels of social integration associated with inner-city or urbanized living. The most obvious implication of the current research concerns the inuence of perceived neighborhood disorder on fear of crime. Neighborhood disorder appears to be the most powerful predictor of fear of crime, whether measured as perceived
13 It should also be noted that separate bivariate analyses of crime trends (operationalized as the difference in violent and property crime rates between the years 2000 and 2002) were conducted to determine their potential inuence on perceived risk and worry of victimization. These analyses were conducted separately due to the relatively small number of level-two groups (N = 21). The results of these tests indicated that neither violent nor property crime trends inuenced the outcome measures.

123

222

Soc Just Res (2008) 21:204227

risk or measured as worry of victimization. Moreover, the disorder effect remained stable across the models, despite various controls at the individual and city levels. The obvious implication of this nding is the need to reduce levels of perceived neighborhood disorder which should, in turn, reduce fear of crime. Toward this end, community-oriented approaches may be quite successful, because they allow residents to address disorder-related problems in conjunction with police and various social agencies. Practitioners must be cautioned, however, that perceptions of disorder are not necessarily grounded in reality. For example, Reisig and Parks (2000) discovered that citizens living in the same locationand, thus, experiencing similar neighborhood conditionsperceived very different levels of physical and social disorder. Thus, effective measures will likely depend on future research aimed to identify the underlying correlates and causes of perceived disorder. While the current analysis allowed us to compare the vulnerability, disorder, and social integration models on both cognitive and affective dimensions of fear, future research should devote additional attention to the behavioral dimension of fear. Arguably, this is the most important dimension of the fear of crime, capturing actual changes in human behavior. The behavioral dimension of fear illustrates the overt effect of fear of crime in citizens everyday lives. Researchers should aim to develop an appropriate measure of this behavioral dimension of fear to shed further light on the fear-of-crime dynamic.
Acknowledgments We would like to thank Nicholas P. Lovrich, Michael J. Gaffney, and the Division of Governmental Studies and Services for providing the primary data analyzed herein. This research was supported, in part, by Project Safe Neighborhoods contract F03-68303004. Please direct correspondence to Noelle E. Fearn, Department of Sociology and Criminal Justice, Saint Louis University, 3500 Lindell Blvd., 211 Fitzgerald Hall, St. Louis, MO 63103.

Appendix A: Scale Items Perceived Risk 1. How safe would you feel walking alone during the day in the area where you live? (1) Very safe (2) Safe (3) Neither safe nor unsafe (4) Unsafe (5) Very unsafe 2. How safe would you feel walking alone in the area where you live at night? (1) Very safe (2) Safe (3) Neither safe nor unsafe (4) Unsafe (5) Very unsafe Worry of Victimization How much do you worry about each of the following situations? Do you worry very frequently, somewhat frequently, seldom, or never about: 1. 2. Yourself or someone in your family getting sexually assaulted Being attacked while driving your car

123

Soc Just Res (2008) 21:204227

223

3. 4. 5. 6. 7.

Getting Getting Getting Getting Getting

mugged beaten up, knifed, or shot murdered burglarized while someone is at home burglarized while no one is at home

Neighborhood Disorder Using the answer key below, please write the number from the Answer Key that most accurately describes the extent of these problems in the neighborhood where you live. Answer Key: (1) No Problem (2) Uncertain (3) A Problem (4) A Serious Problem 1. 2. 3. 4. 5. 6. 7. 8. Vandalism Groups of teenagers or others hanging out and harassing people Garbage and litter Trafc problems People drinking to excess in public Dogs running at large Youth gangs are present Noise

Social Integration 1. Would you describe the area where you live as a place where people mostly help one another or a place where people mostly go their own way? (1) People help one another (2) People go their own way 2. Do you feel the area where you live is more of a real home or more just a place to live? (1) Real home (2) Just a place to live 3. How often do you talk with your neighbors? (1) Daily (2) 13 times a week (3) 13 times a month (4) Less than once a month 4. When you do a favor for a neighbor, can you trust the neighbor to return the favor? (1) Always (2) Nearly always (3) Most of the time (4) Sometimes (5) Seldom

123

224

Soc Just Res (2008) 21:204227

Appendix B: Bivariate Correlation Matrixes


Individual-Level Bivariate Correlation Matrix Race Gender Age Income Education Disorder Social integration

Race Gender Age Income Education Disorder Social integration * p \ .05

.02 .06* .08* .03 -.08* .09* .07* .01 .09* -.09* .05* -.16* -.09* -.12* .13* .31* -.16* .11* -.13* .09* .24*

City-Level Bivariate Correlation Matrix Violent crime Violent crime Property crime Unemployment Urbanism * p \ .05 .60* .27 .35 .36 .12 -.15 Property crime Unemployment Urbanism

References
Adams, R. E. (1992). Is happiness a house in the suburbs? The inuence of urban versus suburban neighborhoods on psychological health. Journal of Community Psychology, 20, 353372. Austin, D. M., Cynthia, W., & Baba, Y. (1994). Crime and safety-related concerns in a small community. American Journal of Criminal Justice, 19, 7997. Baba, Y., & Austin, D. M. (1989). Neighborhood environmental satisfaction, victimization, and social participation as determinants of perceived neighborhood safety. Environment and Behavior, 21, 763780. Baumer, T. L. (1978). Research on fear of crime in the United States. Victimology, 3, 254264. Baumer, T. L. (1985). Testing a general model of fear of crime: Data from a national sample. The Journal of Research in Crime and Delinquency, 22, 239255. Baumer, E. P., Messner, S. T., & Felson, R. B. (2000). The role of victim characteristics in the disposition of murder cases. Justice Quarterly, 17, 281308. Belyea, M. J., & Zingraff, M. T. (1988). Fear of crime and residential location. Rural Sociology, 53, 473 486. Bryk, A. S., & Raudenbush, S. W. (2002). Hierarchical linear models (2nd ed.). Newbury Park: Sage. Burby, R. J., & Rohe, W. M. (1989). Deconcentration of public housing: Effects of residents satisfaction with their living environments and their fear of crime. Urban Affairs Quarterly, 25, 117141.

123

Soc Just Res (2008) 21:204227

225

Bursik, R. J., Jr., & Grasmick, H. G. (1993). Neighborhoods and crime: The dimensions of effective community control. Lexington, MA: Lexington Books. Clemente, F., & Kleiman, M. B. (1977). Fear of crime in the United States: A multivariate analysis. Social Forces, 56, 519531. Covington, J., & Taylor, R. B. (1991). Fear of crime in urban residential neighborhoods: Implications of between- and within-neighborhood sources for current models. Sociological Quarterly, 32, 231249. Denkers, A. J. M., & Winkel, F. W. (1998). Crime victims, well-being, and fear in prospective and longitudinal study. International Review of Victimology, 5, 141162. DiPrete, T. A., & Forristal, J. D. (1994). Multilevel models: Methods and substance. Annual Review of Sociology, 20, 331357. Erskine, H. (1974). The polls: Fear of violence and crime. Public Opinion Quarterly, 38, 131145. Fattah, E. A., & Sacco, V. F. (1989). Crime and victimization of the elderly. New York: Springer-Verlag. Ferraro, K. F., & LaGrange, R. L. (1987). The measurement of fear of crime. Sociological Inquiry, 57, 70101. Ferraro, K. F., & LaGrange, R. L. (1992). Are older people most afraid of crime: Reconsidering age differences in fear of victimization. Journal of Gerontology, 47, S233S244. Fisher, B. S., & Sloan, J. J., III (2003). Unraveling the fear of victimization among college women: Is the shadow of sexual assault hypothesis supported? Justice Quarterly, 20, 633659. Furstenberg, F. F. (1971). Public reaction to crime in the streets. American Scholar, 40, 601610. Garofalo, J. (1979). Victimization and fear of crime. Journal of Research in Crime and Delinquency, 16, 8097. Garofalo, J., & Laub, J. (1978). Fear of crime: Broadening our perspective. Victimology, 3, 242253. Gates, L. B., & Rohe, W. M. (1987). Fear and reactions to crime: A revised model. Urban Affairs Quarterly, 22, 425453. Gibson, C. L., Jihong, Z., Lovrich, N. P., & Gaffney, M. J. (2002). Social integration, individual perceptions of collective efcacy, and fear of crime in three cities. Justice Quarterly, 19, 537564. Gilchrist, E., Bannister, J., Ditton, J., & Farrall, S. (1998). Women and the fear of crime: Challenging the accepted stereotype. The British Journal of Criminology, 38, 283298. Ginsberg, Y. (1985). Fear of crime among elderly Jews in Boston and London. International Journal of Aging and Human Development, 20, 257267. Gordon, M. T., & Riger, S. (1989). The female fear. New York: The Free Press. Hale, C. (1996). Fear of crime: A review of the literature. International Review of Victimology, 4, 79150. Hartnagel, T. F. (1979). The perception and fear of crime: Implications for neighborhood cohesion, social activity, and negative affect. Social Forces, 58, 176193. Hindelang, M. J., Gottfredson, M. R., & Garofalo, J. (1978). Victims of personal crime. Cambridge, MA: Ballinger. Hughes, P. P., Marshall, D., & Sherrill, C. (2003). Multidimensional analysis of fear and condence of university women relating to crimes and dangerous situations. Journal of Interpersonal Violence, 18, 3349. Hunter, A., & Baumer, T. L. (1982). Street trafc, social integration, and fear of crime. Sociological Inquiry, 52, 123131. Jaycox, V. (1978). The elderlys fear of crime: Rational or irrational? Victimology, 3, 329334. Junger, M. (1987). Womens experiences of sexual harassment: Some implications for their fear of crime. British Journal of Criminology, 27, 358383. Kanan, J. W., & Pruitt, M. V. (2002). Modeling fear of crime and perceived victimization risk: (In)signicance of neighborhood integration. Sociological Inquiry, 72, 527548. Kasarda, J. D., & Janowitz, M. (1974). Community attachment in mass society. American Sociological Review, 39, 328339. Kennedy, L. W., & Krahn, H. (1984). Ruralurban origin and fear of crime: The case for rural baggage. Rural Sociology, 49, 247260. Kennedy, L. W., & Silverman, R. A. (1985). Signicant others and fear of crime among the elderly. International Journal of Aging and Human Development, 20, 241256. Keyes, C. L. M. (1998). Social well-being. Social Psychology Quarterly, 61, 121140. Killias, M., & Clerici, C. (2000). Different measures of vulnerability in their relation to different dimensions of fear of crime. The British Journal of Criminology, 40, 437450. Krannich, R. S., Barry, E. H., & Greider, T. (1989). Fear of crime in rapidly changing rural communities: A longitudinal analysis. Rural Sociology, 54, 195212.

123

226

Soc Just Res (2008) 21:204227

LaGrange, R. L., & Ferraro, K. F. (1989). Assessing age and gender differences in perceived risk and fear of crime. Criminology, 27, 697720. LaGrange, R. L., Ferraro, K. F., & Supancic, M. (1992). Perceived risk and fear of crime: Role of social and physical incivilities. The Journal of Research in Crime and Delinquency, 29, 311334. Lee, M., & Ulmer, J. T. (2000). Fear of crime among Korean Americans in Chicago communities. Criminology, 38, 11731206. Lewis, D. A., & Maxeld, M. G. (1980). Fear in the neighborhoods: An investigation of the impact of crime. Journal of Research in Crime and Delinquency, 17, 160189. Lewis, D. A., & Salem, G. W. (1986). Fear of crime: Incivility and the production of a social problem. New Brunswick: Transaction Books. Maxeld, M. G. (1984). The limits of vulnerability in explaining fear of crime: A comparative neighborhood analysis. The Journal of Research in Crime and Delinquency, 21, 233250. McGarrell, E. F., Giacomazzi, A. L., & Thurman, Q. C. (1997). Neighborhood disorder, integration, and the fear of crime. Justice Quarterly, 14, 479499. Pantazis, C. (2000). Fear of crime, vulnerability and poverty. British Journal of Criminology, 40, 414 436. Parker, K. D., & Onyekwuluje, A. B. (1992). The inuence of demographic and economic factors on fear of crime among African Americans. Western Journal of Black Studies, 16, 132140. Rader, N. E. (2004). The threat of victimization: A theoretical reconceptualization of fear of crime. Sociological Spectrum, 24, 689704. Reisig, M. D., & Parks, R. B. (2000). Experience, quality of life, and neighborhood context: A hierarchical analysis of satisfaction with police. Justice Quarterly, 17, 607629. Riger, S., LeBailly, R. K., & Gordon, M. T. (1981). Community ties and urbanites fear of crime: An ecological investigation. American Journal of Community Psychology, 9, 653655. Ross, C. E., & Mirowsky, J. (1999). Disorder and decay: The concept and measurement of perceived neighborhood disorder. Urban Affairs Review, 34, 412432. Rountree, P. W. (1998). A reexamination of the crime-fear linkage. Journal of Research in Crime and Delinquency, 35, 341372. Rountree, P. W., & Land, K. C. (1996a). Perceived risk versus fear of crime: Empirical evidence of conceptually distinct relations in survey data. Social Forces, 74, 13531376. Rountree, P. W., & Land, K. C. (1996b). Burglary victimization, perceptions of crime risk, and routine activities: A multilevel analysis across Seattle neighborhoods and census tracts. The Journal of Research in Crime and Delinquency, 33, 147180. Sacco, V. F. (1985). City size and perceptions of crime. Canadian Journal of Sociology, 10, 277293. Sacco, V. F. (1990). Gender, fear, and victimization: A preliminary application of power-control theory. Sociological Spectrum, 1, 485506. Sampson, R., & Groves, B. (1989). Community structure and crime: Testing social disorganization theory. American Journal of Sociology, 94, 774802. Shaw, C., & McKay, H. (1942). Juvenile delinquency and urban areas. Chicago: University of Chicago Press. Skogan, W. G. (1990). Disorder and decline: Crime and the spiral of decay in American neighborhoods. New York: The Free Press. Skogan, W. G., & Maxeld, M. G. (1981). Coping with crime. Beverly Hills, CA: Sage. Smith, W. R., & Torstensson, M. (1997). Fear of Crime: Gender differences in risk perception and neutralizing fear of crime. The British Journal of Criminology, 37, 608634. Snijders, T. A. B., & Bosker, R. J. (1999). Multilevel analysis: An introduction to basic and advanced multilevel modeling. London: Sage. Stanko, E. A. (1987). Typical violence, normal precaution: Men, women, and interpersonal violence in England, Wales, Scotland, the USA. In J. Hammer & M. Maynard (Eds.), Women, violence, and social control (pp. 122134). London: Macmillan. Stanko, E. A. (1990a). Everyday violence: How women and men experience sexual and physical danger. London: Pandora. Stanko, E. A. (1990b). When precaution is normal: A feminist critique of crime prevention. In L. Gelsthorpe & A. Morris (Eds.), Feminist perspectives in criminology (pp. 173183). Philadelphia: Open University Press. Taylor, R. B. (2002). Fear of crime, social ties, and collective efcacy: Maybe masquerading measurement, maybe deja vu all over again. Justice Quarterly, 19, 773792.

123

Soc Just Res (2008) 21:204227

227

Taylor, R. B., & Hale, M. (1986). Testing alternative models of fear of crime. Journal of Criminal Law and Criminology, 77, 151189. Warr, M. (1984). Fear of victimization: Why are women and the elderly more afraid? Social Science Quarterly, 65, 681702. Warr, M. (1985). Fear of rape among urban women. Social Problems, 32, 238250. Will, J. A., & McGrath, J. H. (1995). Crime, neighborhood perceptions, and the underclass: The relationship between fear of crime and class position. Journal of Criminal Justice, 23, 163176. Williams, F. P., McShane, M. D., & Akers, R. (2000). Worry about victimization: An alternate and reliable measure for the fear of crime. Western Criminology Review, 2, 131. Wilson, J. Q., & Kelling, G. L. (1982). Broken windows. Atlantic Monthly, February, 4652. Yin, P. (1982). Fear of crime as a problem for the elderly. Social Problems, 30, 240245.

123

Vous aimerez peut-être aussi