Within-Individual Change in Social Support, Perceived Collective Efficacy, Perceived Disorder and Fear of Crime: Results from a Two-Wave Panel Study

View metadata, citation and similar papers at core.ac.uk brought to you by CORE

provided by Ghent University Academic Bibliography

doi:10.1093/bjc/azy002 BRIT. J. CRIMINOL

WITHIN-INDIVIDUAL CHANGE IN SOCIAL SUPPORT, PERCEIVED COLLECTIVE EFFICACY, PERCEIVED DISORDER AND FEAR OF CRIME: RESULTS FROM A TWO-WAVE PANEL STUDY

Wim Hardyns*, Lieven J.R. Pauwels and Ben Heylen

In this study, we untangle the relationships between fear of crime (perceived risk of victimization) and perceptions of, respectively, social support, collective efficacy and perceived disorder. We use a two-wave panel study with 356 respondents. Results show that prior perceptions of disorder have a positive effect on later levels of perceived risk of victimization, lending support to the hypothesis that fear of crime is, in part, determined by perceived disorder. Levels of social support negatively affected later levels of perceived risk of victimization. Neither perceived informal social control nor perceived social trust had any effect on later levels of perceived risk of victimization. Strengths, weaknesses and suggestions for further research are discussed. Key Words: fear of crime, longitudinal design, perceived collective efficacy, social support, perceived disorder

Introduction Research on fear of crime has a longstanding tradition in criminology (Ferraro 1995; Hale 1996; Ditton and Farrall 2000; Lee 2013). Within this scholarly tradition, longitudinal enquiries at the aggregate level has related neighbourhood characteristics to (ecological concentrations of) fear of crime, yielding important results regarding the causal processes involved in the aetiology of fear of crime (Bursik 1984; Bursik and Grasmick 1993; Brunton-Smith and Sturgis 2011; Steenbeek and Hipp 2011; Brunton- Smith et al. 2014). Such studies, initially revived by Bursik and Webb (1982), attempt to examine and explain changes in fear of crime concentrations in a time frame when cross-sectional studies prevailed. These longitudinal studies clearly reveal that neighbourhoods are constantly undergoing structural and processual changes and that uni- causal theories and models do not adequately capture the complex dynamics between crime and structural characteristics. However, despite the amount of ecological time series, longitudinal research at the individual level, i.e. how changing individual perceptions shape levels of fear of crime over time remains a research gap in this area. In the present study, we contribute to the fear of crime literature by presenting the results from the first Belgian two-wave panel study at the individual level in which individual levels of social support, perceived

*Wim Hardyns, Department of Criminology, Criminal Law and Social Law, Ghent University, Universiteitstraat 4, 9000 Ghent (BE), Belgium, [email protected]; Ben Heylen, Department of Criminology, Criminal Law and Social Law, Ghent University, Universiteitstraat 4, 9000 Ghent (BE), Belgium; Lieven J.R. Pauwels, Department of Criminology, Criminal Law and Social Law, Ghent University, Universiteitstraat 4, 9000 Ghent (BE), Belgium. Page 1 of 17

Downloaded from https://academic.oup.com/bjc/advance-article-abstract/doi/10.1093/bjc/azy002/4907972 by Ghent University user on 01 March 2018 HARDYNS ET AL.

collective efficacy, perceived disorder and fear of crime are measured at two points in time. The aim of this study is to establish the mechanisms that generate fear of crime over time and to get insight in the relative stability of social support, perceived collective efficacy, perceived disorder and fear of crime over time. This way, we contribute to the longitudinal research tradition in which individual changes in fear of crime are brought into relationship with individual perceptions, such as collective efficacy and informal social control (e.g. Brunton-Smith et al. 2014). Perceptions of disorder (social incivilities and physical nuisances) have been consistently brought into relationship with fear of crime, especially in, but not restricted to, the urban context (Skogan 1990; Kelling and Coles 1996; Sampson and Raudenbush 2004; Farrall et al. 2009; Vancluysen et al. 2011; Ceccato 2012; Jackson et al. 2017). Influenced by broken windows theorizing (BWT), the proposition that has been put to a test in most studies is that perceived disorder brings about fear of crime in neighbourhood residents. Thus, perceived disorder is frequently considered as a key determinant of fear of crime (Hardyns 2012; Skogan 2015). Unfortunately, the concept is often polit- ically abused (see Crawford 2003 and Harcourt 2009 for a discussion). However, BWT originally only argues that a negative spiral of decline that may appear as increased fear may cause people to perceive higher levels of risk, which often results in distorted and overestimated levels of (perceived) disorder. In this sense, perceived disorder and fear of crime are part of a feedback loop, in which perceived disorder feeds fear of crime, and fear of crime in turn heightens the levels of perceived disorder in neighbourhoods (Wyant 2008; Kelling 2015; Skogan 2015; Welsh et al. 2015; Yang and Pao 2015), and the effect persists after controlling for neighbourhood crime rates. So far, studies conducted on fear of crime are predominantly cross-sectional in nature. Cross-sectional studies provide only a snapshot and have a rather poor design when establishing causal relationships. Nothing can be said about the nature of the relationship between perceived disorder and fear of crime when both characteristics are measured simultaneously. This relation can be very complicated: from a theoretical point of view, one can argue that perceived disorder contributes to installing fear of crime. However, fear of crime may in turn increase one’s level of perceived disorder as fear makes people more alert and sensitive to risky settings and the perception of potential dangers in the settings1 they perceive in their neighbourhoods. Another point of view is that both perceived disorder and fear of crime have one or more common causes, which would render the relationship spurious. Results of the small but growing number of longitudinal studies (e.g. Brunton-Smith 2011; Robinson et al. 2003; Russo and Roccato 2010) provide partial support for the hypothesis that perceived disorder heightens fear of crime of individuals. Brunton- Smith (2011) found that levels of perceived disorder measured at T1 are consistently and positively related to levels of fear of crime measured at T2, across two age cohorts (aged 10–15 and 16–25). However, the opposite effect was not established: they found no significant relationship between prior levels of fear of crime and later levels of perceived disorder. Furthermore, in the older cohort, the effect was even opposite: respondents who displayed higher levels of fear of crime at T1 tended to have lower

1It is important to distinguish between neighbourhoods as a whole and settings, i.e. the part of the environment that is actually perceived by our senses (see Wikström et al. 2012) for a discussion. It is difficult to imagine how one can be affected by what is not perceived in the first place. Page 2 of 17

Downloaded from https://academic.oup.com/bjc/advance-article-abstract/doi/10.1093/bjc/azy002/4907972 by Ghent University user on 01 March 2018 RESULTS FROM A TWO-WAVE PANEL STUDY

levels of perceived disorder at T2. Similarly, Robinson et al. (2003) found unambigu- ous evidence that prior levels of incivilities had a cross-lagged effect on subsequent levels of fear of crime. In turn, Russo and Roccato (2010) adopted a somewhat different approach, by relating victimization to fear of crime rather than perceptions of disorder. Their study revealed that recent direct victimization was the most salient predictor of later levels of fear of crime. Besides its predominant relationship with BWT, advances in criminological theorizing led scholars to adopt other theoretical frameworks in explaining fear of crime, such as collective efficacy theory Sampson( 2012) and social support theory (Cullen 1994). Results suggest that individuals living in neighbourhoods with lower levels of perceived collective efficacy and higher levels of perceived disorder report more fear of crime and individuals who experience more social support report lower levels of fear of crime (Hardyns et al. 2016). Given the positive effects of collective efficacy and social support on fear of crime, the present study aims to take these frameworks into account (Cullen 1994; Makarios and Sams 2013; Hardyns et al. 2016). This way, the relationships between individual social support, perceived collective efficacy, perceived disorder and fear of crime are analysed from a dynamic perspective. This study ought to shed more light on the theoretical embeddedness of fear of crime as well as the stability of effects over time.

Theoretical background Fear of crime Fear of crime is a complex concept that originally was argued to consist of three components, a cognitive component, a behavioural component and an affective component (Hale 1996, but see Gray et al. 2008 for an update and discussion). The cognitive component comprises a process in which people convert stimuli related to threat and danger into an assessment of their own risk on victimization (perceived risk of victimization) and precedes the affective component. The affective component can be described as an emotional response of dread or anxiety when confronted with stimuli related to threat and danger. The behavioural component is conceptualized as the outcome of the affective component and refers to behavioural reactions related to fear, such as avoidance behaviour (e.g. Hardyns and Pauwels 2010). To the purpose of the present study, fear of crime will be operationalized as perceived risk of victimization (or risk perception), i.e. what is commonly referred to as the cognitive component. Not only is this in line with how fear of crime is often operationalized in surveys (i.e. by the cognitive or behavioural component), it also allows for a direct comparison with comparable longitudinal studies, such as Brunton- Smith’s (2011) and other studies that incorporate collective efficacy and informal social control (e.g. Brunton-Smith et al. 2014).

Perceived disorder and fear of crime Perceived disorder has consistently been brought into relationship with fear of crime. Having its roots in broken windows theory and the social disorganization perspective, it has been shown on many occasions that perceptions of neighbourhood disorder tend to heighten levels of fear of crime (Ross and Mirowsky 1999; Ross and Jang 2000). Page 3 of 17

Downloaded from https://academic.oup.com/bjc/advance-article-abstract/doi/10.1093/bjc/azy002/4907972 by Ghent University user on 01 March 2018 HARDYNS ET AL.

Visible cues, such as graffiti, litter and crime in the streets, may trigger how residents perceive disorder in their neighbourhood (Innes 2004; Skogan 2015; Yang and Pao 2015). This, in turn, may heighten levels of residents’ fear of crime, as these cues serve as proxies of potential criminal behaviour in that neighbourhood. While perceived disorder is commonly used as an independent variable in fear of crime studies, it remains relatively unclear if it is perceived disorder that increases levels of fear of crime, if it is levels of fear of crime that increase levels of perceived disorder or if it is a reciprocal effect (Skogan 2015). A three-wave longitudinal study by Brunton-Smith (2011) revealed that, indeed, prior levels of perceived disorder influence later levels of fear of crime positively. However, the relationship between prior levels of fear of crime and perceived disorder turned out empirically to be far less clear. In the younger cohort, no relationship was found, while in the older cohort, a negative relationship was found between prior levels of fear of crime and perceived disorder. These findings are somewhat counterintuitive to what would theoretically be expected, but they clearly demonstrated a clear indication of the complexity of the relationships between fear of crime and perceived disorder.

Perceived social trust, perceived informal social control and fear of crime Collective efficacy, a concept derived from personal efficacy and initially introduced in psychology by Bandura (1982), is defined as social‘ cohesion among neighbours combined with their willingness to intervene on behalf of the common good’ (Sampson et al. 1997). The theory of collective efficacy has been of major influence on the ecological study of crime-related outcomes. Sampson et al. highlighted the relationships between dimensions of social capital that were already mentioned in the systemic models of cohesion and crime that were built in the 1980s and 1990s of the previous century (e.g. Sampson and Groves 1989; Bursik and Grasmick 1993; Sampson et al. 1997). Collective efficacy theory stresses the importance of a community being able to solve its commonly iden- tified problems, such as crime and safety. Collective efficacy can be seen as the ‘local social eyes’ in the community. Thus, it should be clear that the collective efficacy theory cannot be applied to all types of crime, instead it focuses mainly on visible property crime and violent crime. Collective efficacy defined as above is generally measured as an aggregate ofperceived social trust (= perceived social cohesion) on the one hand and perceived informal social control on the other hand. Social trust in a community is an essential condition for fostering informal social control, and thus the willingness to intervene for the common good. Therefore, communities characterized by a strong collective efficacy are resistant to high local concentrations of crime, victimization and fear of crime. In this view, inhabitants that perceive more social trust and more informal social control are expected to have lower risk perceptions because they are more likely to observe less collectively defined problems, which guarantees security in the community Sampson( et al. 2002). Perceived collective efficacy is presumed to reduce perceived disorder and thus also levels of fear of crime (Sampson and Raudenbush 2004; Hardyns 2012), while other studies suggest this relationship is reciprocal insofar as perceived disorder may lower perceived collective efficacy mediated by fear of crime Markowitz( et al. 2001). Thus again, the relationship is not straightforward and warrants further investigation. Page 4 of 17

Downloaded from https://academic.oup.com/bjc/advance-article-abstract/doi/10.1093/bjc/azy002/4907972 by Ghent University user on 01 March 2018 RESULTS FROM A TWO-WAVE PANEL STUDY

Some previous European studies reveal that perceived social trust and informal social control do not correlate at the neighbourhood level of analysis (Pauwels et al. 2010; Hardyns 2012). Other researchers have suggested that the collective efficacy concept as developed in the United States cannot simply be transferred and applied to other coun- tries and settings (Carpiano 2006; Zhang et al. 2009), thus questioning its universal val- idity. For that reason, in this study, we make a distinction between perceived social trust and perceived informal social control.

Social support and fear of crime Besides the neighbourhood social climate, it is important to focus on the individual’s social network in relation to fear of crime. Social networks are linked to individuals’ fear of crime because of the support they provide (Cullen 1994; Berkman et al. 2000). Social support does not solely refer to local social ties that are embedded in the neighbourhood but also refer to the support of social networks that are not bounded by the neighbourhood. Scholars have found that one’s social support is beneficial for mental and physical health, e.g. general well-being and the absence of the symptoms of diseases, and reduces the risk of feeling unsafe, depressive or oppressive (Thoits 1985, 1995; Coyne and Downey 1991; Sacco 1993; Wright and Cullen 2001). Social support directly enhances an individual’s quality of life by providing opportunities for interaction, help with practical tasks, and relieving feelings of loneliness (Ganster and Victor 1988; Makarios and Sams 2013). Thus, social support can be expected to lower levels of fear of crime in individuals, but it is unclear whether fear of crime in turn also affects social support or not.

Data Data collection This study draws upon data from the Social capital and Well-being in Neighborhoods in Ghent (SWING) survey (Hardyns et al. 2015). Data were collected in two successive waves in the city of Ghent, Belgium. A stratified representative sample of 50 neighbourhoods was selected from a total of 142 neighbourhoods with a minimum population of 200 adult inhabitants. The stratified selection procedure based on population density and a multidimensional deprivation index developed for policy issues (based on information from tax and census databases, socio-economic data, population composition and physical characteristics of the neighbourhood) resulted in a representative set of neighbourhoods. Within neighbourhoods, inhabitants were sampled according to the following inclusion criteria: being older than 18, not living in an institutional setting and having sufficient knowledge of the Dutch language. Data were gathered using face-to-face interviews with respondents randomly drawn from a municipal registry. In 2013, 943 respondents were reached in 50 neighbourhoods. The same respondents were contacted in 2015, resulting in a response of completed questionnaires of 356 (37.8 per cent). The main reasons why many respondents were not reached were (1) the fact that they had moved out of the neighbourhood, (2) the wish not to participate again and (3) many respondents were not reachable after ten attempts to contact them at various points in time. Demographics for the sample are presented in Table 1. For 2013, percentages for the entire sample of 943 respondents are presented, and for 2015, demographics for the 356 Page 5 of 17

Downloaded from https://academic.oup.com/bjc/advance-article-abstract/doi/10.1093/bjc/azy002/4907972 by Ghent University user on 01 March 2018 HARDYNS ET AL.

respondents that we were able to contact again are presented. This allows for a comparison of the sample composition, and detect possible problems of attrition. We tested for differences using a simple chi-square test, comparing the proportions of the 2015 sample to the proportions of the 2013 sample. As appears from the table, regarding sex, employment, nationality and length of residence, no significant differences in proportions emerge. This suggests that the two samples are similar in terms of the demographic variables on the basis of which respondents were selected (although this does not automatically imply that there is no effect at all regarding attrition). Thus, the smaller 2015 sample seems to be an adequate reflection of the larger and representative 2013 sample. Regarding age, we did find a significant difference across the four categories (up to 25, 25–50, 50–75, 75+ years old). However, there is no real tendency in the differences between categories (in the sense of significantly younger or older respondents across the four categories), and age does not affect any of the relationships between the theoretical variables included in the model (see below), so the results of this bias does not affect this methodological problem.

Measurement of concepts Measurement of perceived social trust and perceived informal social control in the neighbourhood, as the two key dimensions of collective efficacy, is based on the influential work of Sampson et al. (1997). In our analyses, both dimensions will be treated separately, given the relatively low correlation between both (r T1 = 0.273; p < 0.05 and r T2 = 0.443; p < 0.05). Perceived social trust, a summed scale (α T1 = 0.82 and α T2 = 0.84) with higher scores indicating higher levels of perceived social trust, consists of four items, such as ‘people around here are willing to help their neighbours’. Respondents were asked to indicate on a five-point Likert scale how strongly they agreed with each of these items. To measure perceived informal social control respondents were asked about the likelihood that their neighbours could be counted on to intervene in various ways if problem situa- tions appear, such as ‘children were skipping school and hanging out on a street corner’.

Table 1 Demographics for the two samples and chi-square difference scores across waves

Variable 2013 2013 and 2015 Chi-square df Significance

(%) (%)

n = 943 n = 356

Sex (male = 0, female = 1) 47.3 50.0 1.286 1 0.257 Employed (yes = 1, no = 0) 59.0 56.5 0.949 1 0.330 Belgian nationality (yes = 1, no = 0) 89.8 87.1 3.376 1 0.066 Length of residence 3.557 3 0.313 <1 year 5.9 4.8 1–5 years 26.9 23.6 5–10 years 15.1 16.0 >10 years 52.1 55.6 Age 8.861 3 0.031 <25 9.4 11.5 25–49 45.0 41.0 50–74 33.1 38.2 75 and older 12.5 9.3

Page 6 of 17

Downloaded from https://academic.oup.com/bjc/advance-article-abstract/doi/10.1093/bjc/azy002/4907972 by Ghent University user on 01 March 2018 RESULTS FROM A TWO-WAVE PANEL STUDY

Higher scores on this six-item measure (α T1 = 0.79 and α T2 = 0.83) indicate higher levels of perceived informal social control. Perceived disorder was assessed by asking neighbourhood inhabitants how often they have observed four occurrences in their neighbourhood, such as ‘a group of youngsters harasses people on the street’ or ‘someone sells drugs on the street’. Responses were combined into a single measure with high scores indicating higher levels of perceived disorder (α T1 = 0.77 and α T2 = 0.76). Perceived risk of victimization was measured by a three-item measure (α T1 = 0.63 and α T2 = 0.68). An example of a perceived risk of victimization item is asking respondents ‘how likely they think they are to be the victim of a burglary in the following twelve months’. Social support was measured using a four-item scale (α T1 = 0.82 and α T2 = 0.83). Respondents were asked, for example, how many people among their family, friends or acquaintances ‘understand your problems’ or ‘would let you live with them if you had to leave your house temporarily’. Factor loadings of individual items can be consulted in Appendix Table A1, the correlations between scales within T1 and T2 can be consulted in Appendix Table A2 and descriptive statistics for the scales are summarized in Appendix Table A3.

Analytical strategy Relationships between the different variables were modelled using a cross-lagged panel design, which allows to understand the relationships between variables at T1 and T2 (de Jonge et al. 2001; Diggle and Diggle 2002; Kenny 2005; Fitzmaurice 2009; Twisk 2013). We used summed scales as variables in the model, while also including error terms to reflect the assumption of imprecise measurement of the concepts. Allowing error terms to co-vary is important, as it allows to estimate relationships between variables at T1 and T2 while controlling for statistical effects of these variables at T1 and T2. This way, the paths between variables at T1 and T2 reflect how they impactchange in relationships between variables over time. Magnitudes of the covariations of error terms are presented in Appendix Table A4. Models are fitted using the Amos software package for structural equation modelling. We have opted to use a path model rather than a full structural equation model given the relatively small sample size in comparison to the number of parameters to be estimated (e.g. Bollen and Long 1993). Furthermore, we are more interested in the estimation of structural coefficients rather than the measurement model in the current study, given the well-validated scale constructs and the nature of our research questions. In our analysis, it is assumed that the levels of perceived risk of victimization and the other variables at T2 are a function of both the levels of that same variable and the value of the other variables measured at T1. This way, relationships between T1 and T2 concerning the same concept reflect the stability of that variable over time, while the lagged effects indicate changes in levels of that variable due to prior levels of the other variable(s). For example, changes in perceived risk of victimization from T1 to T2 are assumed to be caused by prior levels (at T1) of social support, perceived social trust, perceived informal social control and perceived disorder. Given the fact that the model may not be exhaustive, error terms of each variable were allowed to co-vary at both time points, thus reflecting the possibility that other variables than those included in the study exert a causal effect on the variables in question. In addition, we included five control variables in the model, being sex, employment, nationality (Belgian–non-Belgian, based on nationality of parents), length of residence and age. These variables were included at time point T1, similar to Brunton- Smith’s (2011) approach. Page 7 of 17

Downloaded from https://academic.oup.com/bjc/advance-article-abstract/doi/10.1093/bjc/azy002/4907972 by Ghent University user on 01 March 2018 HARDYNS ET AL.

Results First, we discuss the role of the control variables in the model. Regarding perceived risk of victimization, results are in line with existing literature, with higher levels of perceived risk of victimization for women (0.15; p < 0.05) (Brunton-Smith 2011). Whereas other studies indicate higher levels of perceived risk of victimization for non-whites, we did not find any significant effect of having Belgian nationality on perceived risk of victimization. However, this may be attributable to the difference in operationaliza- tion, as we only coded nationality based on the nationality of both parents, and do not dispose of data on respondents’ ethnic background. Perceived disorder seems to be lower among respondents that have lived in their neighbourhood for longer periods of time (−0.12; p < 0.05), whereas sex, having a full-time job and nationality do not have any significant effect on perceived disorder. None of the control variables have an effect on either dimension of collective efficacy (perceived social trust and perceived informal social control). Levels of social support are higher among respondents having a full-time job (0.25; p < 0.05) and respondents who have Belgian nationality (0.14; p < 0.05). No significant effect of sex or length of residence in the neighbourhood on social support can be found. Figure 1 represents the overall individual changes in the variables in the longitudinal section of the model. It is a simplified figure, in which only the significant paths are shown, and correlations of error terms are omitted to enhance readability. A full account of the model, including insignificant paths, can be found in Appendix Table A5. The fit of the path model (chi-square/df = 2.342, CFI = 0.950, RMSEA = 0.061) is well within ac- ceptable ranges (Bollen and Long 1993), suggesting that the model captures the complex reality of the relationships between all variables rather well. When looking at the relationship between perceived risk of victimization and prior levels of social support, perceived social trust, perceived informal social control and perceived disorder, we find a positive relationship between prior levels of perceived disorder and perceived risk of victimization (0.19; p < 0.05) while controlling for prior levels of perceived risk of victimization. This finding is in line with existing literature, which also suggests that perceived risk of victimization is positively related to prior levels of perceived disorder. The relatively low stability of perceived risk of victimization over time (0.36; p < 0.05) corroborates the role of previous levels of perceived disorder on perceived risk of victimization. Perceived risk of victimization also has a significant negative relationship with prior perceptions of social support (−0.11; p < 0.05) while controlling for prior levels of perceived victimization. This finding provides corroborating evidence for the conjecture that social support is positively related to quality of life, including lower reported perceived risk of victimization. In this sense, people who perceive a supportive social network will tend to report lower levels of perceived risk of victimization at later moments in time. On the contrary, perceived informal social control and perceived social trust do not relate significantly to subsequent levels of perceived risk of victimization while controlling for prior levels of perceived victimization, which provides a falsifying instance of the conjecture that higher levels of perceived collective efficacy are protective of perceived risk of victimization. This suggests that it is not so much perceived levels of social trust and perceived informal social control that affect levels of perceived risk of victimization, but rather the social support one receives from their social entourage (social support).

Page 8 of 17

Downloaded from https://academic.oup.com/bjc/advance-article-abstract/doi/10.1093/bjc/azy002/4907972 by Ghent University user on 01 March 2018 RESULTS FROM A TWO-WAVE PANEL STUDY

Fig. 1 Cross-lagged panel model for the data. Chi-square/df = 2342, CFI = 0.950, RMSEA = 0.061

Moving on to perceived disorder, our data reveal that prior levels of perceived risk of victimization do not have a positive relationship with later perceived disorder while controlling for prior levels of perceived disorder. This finding is in line with the research of Brunton- Smith (2011), where no such relationship was found for the younger cohort and a negative relationship was found for the older cohort. Neither social support, perceived informal social control and perceived social trust at T1 had a significant relationship with perceived disorder at T2 when controlling for prior levels of perceived disorder at T1. In addition, the stability of perceived disorder over time is relatively high (0.63; p < 0.05), which suggests that the existence of additional factors may impact the relationship between disorder and perceived risk of victimization. Social support has no significant relationships with perceived risk of victimization, perceived disorder, perceived informal social control or perceived social trust when controlling for prior levels of social support. In combination with the rather large stability of social support over time (0.63; p < 0.05), our data suggest that people who exhibit high levels of social support will still perceive such levels of social support at later times, disregarding changes in perceived risk of victimization, perceived disorder, perceived informal social control or perceived social trust. Even though prior levels of perceived informal social control do not affect later levels of perceived risk of victimization, perceived disorder, social support or perceived social trust when controlling for prior levels of these variables, later levels of perceived informal social control are significantly related to prior levels of perceived disorder Page 9 of 17

Downloaded from https://academic.oup.com/bjc/advance-article-abstract/doi/10.1093/bjc/azy002/4907972 by Ghent University user on 01 March 2018 HARDYNS ET AL.

(−0.18; p < 0.05), social support (0.17; p < 0.05) and perceived social trust (0.11; p < 0.05) when controlling for prior levels of perceived informal social control. The low stability of perceived social trust over time highlights the importance of these relationships. Furthermore, the relatively stable levels of perceived social trust (0.49; p < 0.05) have a negative relationship with prior levels of perceived disorder (−0.19; p < 0.05). These findings suggest that perceptions of collective efficacy do not have a protective effect when it comes to perceived risk of victimization but rather should be seen as an outcome variable as it is prior levels of perceived disorder that affect later levels of perceived social trust and perceived informal social control, which partly falsifies the conjecture that collective efficacy in itself fosters lower levels of fear and perceived disorder.

Discussion This article reports findings on the relative stability and change in social support, perceived collective efficacy, perceived disorder and fear of crime of the first two-wave panel study of fear of crime in Belgium. The study largely confirms the theoretically expected relationships between prior levels of perceived disorder on later levels of perceived risk of victimization. In addition, we did not find any effect of prior levels of perceived risk of victimization on later levels of perceived disorder while controlling for prior levels of perceived disorder, which suggests that prior levels of perceived disorder is the independent variable and perceived risk of victimization the dependent variable. Furthermore, our research also incorporated social support, perceived social trust and perceived informal social control, which are theoretically presumed to lower levels of perceived risk of victimization. Results indicate that (personal) social support relates negatively to later levels of perceived risk of victimization while controlling for prior levels of perceived victimization, but that neither perceived social trust or perceived informal social control relates significantly to it. This challenges the theory of collective efficacy applied to perceived risk of victimization. It is further at stakes with studies suggesting that collective efficacy lowers levels of perceived disorder (and, consequently perceived risk of victimization) or that a reciprocal effect exists between perceived risk of victimization and collective efficacy. Thus, it seems that the ‘direct’ social support received from the immediate social environment is more important than perceived social trust and perceived informal social control. This highlights the complexity of the relationships between collective efficacy, perceived disorder and perceived risk of victimization, which remain in need of further research. One possible avenue for further research would be an investigation into how certain personality traits, cognitions and emotions are positively associated with social support, perceived social trust, perceived informal social control, perceived disorder and perceived risk of victimization, and how these may mediate certain effects between, e.g., perceived collective efficacy and later levels of perceived risk of victimization. Furthermore, given the fact that this is currently the only longitudinal study of fear of crime conducted in Belgium, results should be interpreted with care: it is possible that socio-cultural differences at an aggregate level account for the differing results compared to studies conducted in the United States and the United Kingdom. Further investigation is thus needed to evaluate the context-dependency of the effects studied and may shed more light on the spatio-temporal stability of these effects. Finally, our results reveal that the most stable variables over time are perceived disorder, perceived Page 10 of 17

Downloaded from https://academic.oup.com/bjc/advance-article-abstract/doi/10.1093/bjc/azy002/4907972 by Ghent University user on 01 March 2018 RESULTS FROM A TWO-WAVE PANEL STUDY

social trust and social support, suggesting the existence of common causes for their levels to be at a certain level at T1 and T2. Perceived risk of victimization has a relatively low stability over time, suggesting it is to a large extent determined by prior levels of perceived disorder as well as social support. Perceived informal social control also has a low stability coefficient over time, providing support for the findings that it is in large part determined by perceived disorder, social support and perceived social trust. Of course, there are some limitations to our study. A first limitation of the study is the relatively low response rate in the second wave, resulting in a relatively small sample size. This is likely due to the rather large timespan (two years) between both waves, whereas most longitudinal studies apply a shorter timespan (e.g. a few months, mostly around six months). This highlights the difficulties and challenges associated with pro- spective longitudinal research, especially if it takes place over longer periods of time. At the same time, this long period of time between the two waves can be regarded as a strength of the study. Over the span of two years of time, there are considerably large stability coefficients present, especially in the case of perceived disorder and social support. This suggests that both have either common causes at T1 and T2 or that they may be linked to specific personality traits. Further research could address these issues. A second shortcoming of the present study is that it did only measure one dimension of the multidimensional concept fear of crime, thereby leaving many questions un- answered regarding the other components of fear of crime such as the emotional and behavioural component. A third, and probably one of the strongest limitations of this study, was that it was unable to take into account prior levels of victimization, which was actually measured in the SWING survey. The reason for not using this measure is that the time interval in the victimization question at T2 overlaps with T1. Fourth, our study only consisted of two waves. This is a relatively weak form of longitudinal research that limits the potential of the study (Elliott et al. 2008). A far greater analytical potential can be achieved by using three or more waves, as this allows for the testing of reciprocal effects, indirect effects and mediation effects, as well as the construction of more complex models such as latent growth models. Fifth, given the unexpected results of perceived collective efficacy, further research is warranted. Following the signalling idea of Innes, we would expect low levels of perceived collective efficacy to function as a visible cue that there is more disorder in the neighbourhood. This way, one would expect a negative relationship between perceived low collective efficacy and perceived disorder, which would in turn lower levels of fear of crime. Thus, it remains highly plausible that, indeed, there is a mediated effect. Even though we did not find an effect of perceived collective efficacy at T1 on perceived disorder at T2, it remains highly plausible that there is an indirect effect of perceived collective efficacy on T1 on perceived risk of victimization at T2, mediated by perceived disorder on T1. A latent growth model would be very well suited to test this hypothesis, as it models effects within as well as between waves. To conclude, our study contributes to the study of fear of crime by adopting a longitudinal design that allows to untangle the complex issue of causation of fear of crime. In this sense, it partially corroborated the findings of the few longitudinal studies on this topic that are currently available. It was novel insofar as it included variables from other theories that have been theoretically deemed to affect fear of crime, such as social support, perceived social trust and perceived informal social control from a longitudinal perspective. Page 11 of 17

Downloaded from https://academic.oup.com/bjc/advance-article-abstract/doi/10.1093/bjc/azy002/4907972 by Ghent University user on 01 March 2018 HARDYNS ET AL.

Funding The entire research team at Ghent University wishes to express its sincere gratitude for the financial support provided by the Research Foundation—Flanders (FWO). Furthermore, the research team wishes to thank the city of Ghent for their help in facilitating this research project.

Appendix

Table A1 Factor loadings for measurement scales at T1 and T2

Factor Cronbach’s alpha

Scale Item T1 T2 T1 T2

Perceived social trust Please indicate the extent to which you agree 0.82 0.84 5-Point Likert scale or disagree with the following statements: People around here are willing 0.77 0.76 to help their neighbours. This is a close-knit neighbourhood. 0.70 0.74 People in this neighbourhood can be trusted. 0.67 0.70 Contacts between inhabitants in this 0.81 0.83 neighbourhood are generally positive. Perceived informal How likely do you think it is people in this 0.79 0.83 social control neighbourhood will intervene if… 5-Point Likert scale …children were skipping school and 0.55 0.59 hanging out on a street corner? …children were spray-painting 0.69 0.71 graffiti on a local building? …children were showing disrespect to an adult? 0.68 0.69 …a fight broke out in front of their house? 0.64 0.72 …children were making too much racket? 0.55 0.69 …children were using soft drugs 0.64 0.65 (smoking weed, hash etc.)? Social support How many people in your neighbourhood… 0.82 0.83 8-Point scale: 0, 1, …understand your problems? 0.71 0.76 2, 3, 4, 5 6–10, >10 …would let you move into their house for a week 0.74 0.71 if you temporarily could not stay at your house? …would encourage you to go to the doctor 0.76 0.79 if you experience health problems? …make you feel good (e.g. make you 0.71 0.70 feel you are useful or make you feel that they are glad to know you)? Perceived disorder How many times have you noticed the 0.77 0.76 5-Point Likert scale following events in your neighbourhood? Groups of adolescents harassing 0.63 0.63 persons to obtain money or goods. Men drinking alcohol in public. 0.64 0.68 Persons selling drugs (hash, 0.74 0.64 weed etc.) on the streets. Fights between adolescents on the streets. 0.76 0.76 Perceived risk of How much risk do you think you will become 0.63 0.68 victimization victim of… (in the next 12 months) 4-Point Likert scale …a burglary in your house? 0.46 0.52 …physical violence? 0.62 0.65 …a robbery outside of your house? 0.75 0.81

Page 12 of 17

Downloaded from https://academic.oup.com/bjc/advance-article-abstract/doi/10.1093/bjc/azy002/4907972 by Ghent University user on 01 March 2018 RESULTS FROM A TWO-WAVE PANEL STUDY

Table A2 Correlations between variables at T1 and T2

Social Perceived Perceived informal Perceived Perceived risk of support social trust social control disorder victimization

Social 0.216** 0.243** 0.000 −0.023 support Perceived 0.217** 0.273** −0.239** −0.147** social trust Perceived informal 0.237** 0.443** −0.147** −0.122* social control Perceived −0.007 −0.311** −0.207** 0.239** disorder Perceived risk of −0.142** −0.186** −0.078 0.268** victimization T2

Table A3 Descriptive statistics for measurement scales at T1 and T2

Minimum Maximum Mean SD Variance

Perceived risk of victimization T1 3 12 5.36 1.63 2.67 T2 3 12 5.32 1.75 3.07 Perceived disorder T1 4 20 6.47 2.93 8.63 T2 4 17 6.59 2.87 8.27 Social support T1 4 28 18.25 6.39 40.87 T2 4 28 18.45 6.20 38.49 Perceived social trust T1 4 20 14.50 2.94 8.63 T2 4 20 14.63 3.34 11.13 Perceived informal social control T1 6 30 19.73 4.79 22.99 T2 6 30 20.09 5.36 28.73 SD, standard deviation.

Table A4 Covariances between error terms at T1 and T2

Estimate SE CR p

Covariations of error terms at T1 Risk perception T1 ↔ Social control T1 −1.017 0.408 −2.493 0.013 Risk perception T1 ↔ Social support T1 −0.067 0.081 −0.832 0.406 Risk perception T1 ↔ Disorder T1 0.405 0.087 4.626 0.000 Social control T1 ↔ Social support T1 1.014 0.243 4.169 0.000 Social control T1 ↔ Disorder T1 −0.664 0.252 −2.630 0.009 Social support T1 ↔ Disorder T1 −0.003 0.050 −0.070 0.944 Disorder T1 ↔ Social trust T1 −0.674 0.157 −4.283 0.000 Social support T1 ↔ Social trust T1 0.585 0.149 3.930 0.000 Social control T1 ↔ Social trust T1 3.699 0.757 4.887 0.000 Risk perception T1 ↔ Social trust T1 −0.680 0.251 −2.713 0.007 Covariations of error terms at T2 Risk perception T2 ↔ Disorder T2 0.120 0.063 1.914 0.056 Risk perception T2 ↔ Social support T2 −0.105 0.063 −1.667 0.095 Risk perception T2 ↔ Social control T2 −0.073 0.397 −0.184 0.854 Social control T2 ↔ Social support T2 0.419 0.195 2.148 0.032 Social control T2 ↔ Disorder T2 −0.327 0.194 −1.686 0.092

Page 13 of 17

Downloaded from https://academic.oup.com/bjc/advance-article-abstract/doi/10.1093/bjc/azy002/4907972 by Ghent University user on 01 March 2018 HARDYNS ET AL.

Table A4 Continued

Estimate SE CR p Social support T2 ↔ Disorder T2 −0.034 0.030 −1.106 0.269 Disorder T2 ↔ Social trust T2 −0.287 0.109 −2.646 0.008 Social support T2 ↔ Social trust T2 0.321 0.109 2.937 0.003 Social control T2 ↔ Social trust T2 4.468 0.724 6.167 0.000 Risk perception T2 ↔ Social trust T2 −0.358 0.222 −1.616 0.106 All significant correlations are in bold. Estimate, correlation;SE , standard error; CR, critical ratio; p, significance estimate.

Table A5 Full model including insignificant paths

Variable T1 Variable T2 β

Risk perception T1 → Risk perception T2 0.362 Risk perception T1 → Social control T2 0.066 Risk perception T1 → Social support T2 −0.048 Risk perception T1 → Disorder T2 0.064 Social control T1 → Risk perception T2 0.039 Social control T1 → Social control T2 0.246 Social control T1 → Social support T2 −0.016 Social control T1 → Disorder T2 0.004 Social support T1 → Risk perception T2 −0.109 Social support T1 → Social control T2 0.168 Social support T1 → Social support T2 0.632 Social support T1 → Disorder T2 0.030 Disorder T1 → Risk perception T2 0.189 Disorder T1 → Social control T2 −0.180 Disorder T1 → Social support T2 0.058 Disorder T1 → Disorder T2 0.626 Social trust T1 → Risk perception T2 −0.016 Social trust T1 → Social control T2 0.112 Social trust T1 → Social support T2 0.073 Social trust T1 → Disorder T2 0.044 Social trust T1 → Social trust T2 0.490 Disorder T1 → Social trust T2 −0.185 Social support T1 → Social trust T2 0.029 Social control T1 → Social trust T2 0.065 Risk perception T1 → Social trust T2 0.011 In the above table, all significant paths of the model are in bold. Paths not in bold hadp > 0.05.

References Bandura, A. (1982), ‘Self-Efficacy Mechanism in Human Agency’,American Psychologist, 37: 122. Berkman, L. F., Glass, T., Brissette, I. and Seeman, T. E. (2000), ‘From Social Integration to Health: Durkheim in the New Millennium’, Social Science & Medicine, 51: 843–57. Bollen, K. A. and Long, J. S. (1993), Testing Structural Equation Models. Sage Publications. Brunton-Smith, I. (2011), ‘Untangling the Relationship between Fear of Crime and Perceptions of Disorder: Evidence from a Longitudinal Study of Young People in England and Wales’, British Journal of Criminology, 51: 885–99. Page 14 of 17

Downloaded from https://academic.oup.com/bjc/advance-article-abstract/doi/10.1093/bjc/azy002/4907972 by Ghent University user on 01 March 2018 RESULTS FROM A TWO-WAVE PANEL STUDY

Brunton-Smith, I., Jackson, J. and Sutherland, A. (2014), ‘Bridging Structure and Perception: On the Neighbourhood Ecology of Beliefs and Worries about Violent Crime’, British Journal of Criminology, 54: 503–26. Brunton-Smith, I. and Sturgis, P. (2011), ‘Do Neighborhoods Generate Fear of Crime? An Empirical Test Using the British Crime Survey’, Criminology, 49: 331–69. Bursik, R. J. (1984), ‘Urban Dynamics and Ecological Studies of Delinquency’, Social Forces, 63: 393–413. Bursik, R. J. and Grasmick, H. G. (1993), Neighborhoods and Crime: The Dimensions of Effective Community Control. Lexington Books. Bursik, R. J. and Webb, J. (1982), ‘Community Change and Patterns of Delinquency’, American Journal of Sociology, 88: 24–42. Carpiano, R. M. (2006), ‘Toward a Neighborhood Resource-Based Theory of Social Capital for Health: Can Bourdieu and Sociology Help?’, Social Science & Medicine, 62: 165–75. Ceccato, V. (ed.) (2012), The Urban Fabric of Fear of Crime. Springer. Coyne, J. C. and Downey, G. (1991), ‘Social-Factors and Psychopathology – Stress, Social Support, and Coping Processes’, Annual Review of Psychology, 42: 401–25. Crawford, A. (2003), ‘“Contractual Governance” of Deviant Behaviour’, Journal of Law and Society, 30: 479–505. Cullen, F. T. (1994), ‘Social Support as an Organizing Concept for Criminology: Presidential Address to the Academy of Criminal Justice Sciences’, Justice Quarterly, 11: 527–59. de Jonge, J., Dormann, C., Janssen, P. P. M., Dollard, M. F., Landeweerd, J. A. and Nijhuis, F. J. N. (2001), ‘Testing Reciprocal Relationships between Job Characteristics and Psychological Well-Being: A Cross-Lagged Structural Equation Model’, Journal of Occupational and Organizational Psychology, 74: 29–46. Diggle, P. and Diggle, P. (2002), Analysis of Longitudinal Data, 2nd edn. Oxford University Press. Ditton, J. and Farrall, S. (eds) (2000), The Fear of Crime. Ashgate. Elliott, J., Holland, J. and Thomson, R. (2008), ‘Longitudinal and Panel Studies’, in P. Alasuutari, L. Bickman and J. Brannen, eds, The SAGE Handbook of Social Research Methods228–48. Sage. Farrall, S., Jackson, J. and Gray, E. (2009), Social Order and the Fear of Crime in Contemporary Times. Oxford University Press. Ferraro, K. (1995), Fear of Crime: Interpreting Victimization Risk. State University of New York Press. Fitzmaurice, G. M. (2009), Longitudinal Data Analysis. CRC Press. Ganster, D. C. and Victor, B. (1988), ‘The Impact of Social Support on Mental and Physical Health’, British Journal of Medical Psychology, 61: 17–36. Gray, E., Jackson, J. and Farrall, S. (2008), ‘Reassessing the Fear of Crime’, European Journal of Criminology, 5: 363–80. Hale, C. (1996), ‘Fear of Crime: A Review of the Literature’, International Review of Victimology, 4: 79–150. Harcourt, B. E. (2009), Illusion of Order: The False Promise of Broken Windows Policing. Harvard University Press. Hardyns, W. (2012), ‘Collective Efficacy and Fear of Crime in the Flemish Region: Looking Back on the Results and a Preview of Additional Research’, in P. Ponsaers, ed., Social Analysis of Security: Financial, Economic and Ecological Crime, Crime (In)Security and (Dis)Trust, Public and Private Policing, 167–202. Eleven International Publishing. Page 15 of 17

Downloaded from https://academic.oup.com/bjc/advance-article-abstract/doi/10.1093/bjc/azy002/4907972 by Ghent University user on 01 March 2018 HARDYNS ET AL.

Hardyns, W. and Pauwels, L. (2010), ‘Different Measures of Fear of Crime and Survey Measurement Error’, in M. Cools, B. De Ruyver, M. Easton, L. Pauwels, P. Ponsaers, G. Vande Walle, T. Vander Beken, F. Vander Laenen, G. Vermeulen and G. Vynckier, eds, Governance of Security Research Papers Series III, New Empirical Data, Theories and Analyses on Safety, Societal Problems and Citizens’ Perceptions, 19–39. Maklu. Hardyns, W., Vyncke, V., De Boeck, A., Pauwels, P. and Willems, S. (2016), ‘Are Collective Efficacy, Disorder and Social Support Associated with One’s Quality of Life? Evidence from the Multilevel Swing Study in Belgium’, Applied Research in Quality of Life, 11: 739–56. Hardyns, W., Vyncke, V., Pauwels, L. and Willems, S. (2015), ‘Study Protocol: SWING – Social Capital and Well-Being in Neighborhoods in Ghent’, International Journal for Equity in Health, 14: 36. Innes, M. (2004), ‘Signal Crimes and Signal Disorders: Notes on Deviance as Communicative Action’, The British Journal of Sociology, 55: 335–55. Jackson, J., Bradford, B. and Brunton-Smith, I. (2017), ‘In the Eye of the (Motivated) Beholder: Towards a Motivated Cognition Perspective on Disorder Perceptions’, in L. Murray and G. Mythen, eds, Routledge International Handbook of Fear of Crime, 253–71. Routledge. Kelling, G. (2015), ‘An Author’s Brief History of an Idea’, Journal of Research in Crime and Delinquency, 52: 626–9. Kelling, G. L. and Coles, C. M. (1996), Fixing Broken Windows: Restoring Order and Reducing Crime in our Communities. Martin Kessler Books. Kenny, D. A. (2005), Cross-Lagged Panel Design. John Wiley and Sons Ltd. Lee, M. (2013), Inventing Fear of Crime. Routledge. Makarios, M. D. and Sams, T. L. (2013). 'Social Support and Crime', in F. T. Cullen and P. Wilcox, eds, The Oxford Handbook of Criminological Theory, 160–85. Oxford University Press. Markowitz, F. E., Bellair, P. E., Liska, A. E. and Liu, J. H. (2001), ‘Extending Social Disorganization Theory: Modelling the Relationships between Cohesion, Disorder, and Fear’, Criminology, 39: 293–320. Pauwels, L., Hardyns, W. and Van de Velde, M. (2010), Social Disorganisation, Offending, Fear and Victimisation: Findings from Belgian Studies on the Urban Context of Crime. BJu Legal Publishers. Robinson, J. B., Lawton, B. A., Taylor, R. B. and Perkins, D. D. (2003), ‘Multilevel Longitudinal Impacts of Incivilities: Fear of Crime, Expected Safety, and Block Satisfaction’, Journal of Quantitative Criminology, 19: 237–74. Ross, C. E. and Jang, S. J. (2000), ‘Neighborhood Disorder, Fear, and Mistrust: The Buffering Role of Social Ties with Neighbors’, American Journal of Community Psychology, 28: 401–20. Ross, C. E. and Mirowsky, J. (1999), ‘Disorder and Decay: The Concept and Measurement of Perceived Neighborhood Disorder’, Urban Affairs Review, 34: 412–32. Russo, S. and Roccato, M. (2010), ‘How Long Does Victimization Foster Fear of Crime? A Longitudinal Study’, Journal of Community Psychology, 38: 960–74. Sacco, V. F. (1993), ‘Social Support and the Fear of Crime’, Canadian Journal of Criminology, 35: 187–96. Sampson, R. J. (2012), Great American City: Chicago and the Enduring Neighborhood Effect. University of Chicago Press. Page 16 of 17

Downloaded from https://academic.oup.com/bjc/advance-article-abstract/doi/10.1093/bjc/azy002/4907972 by Ghent University user on 01 March 2018 RESULTS FROM A TWO-WAVE PANEL STUDY

Sampson, R. J. and Groves, W. B. (1989), ‘Community Structure and Crime: Testing Social Disorganization Theory’, American Journal of Sociology, 94: 774–802. Sampson, R. J., Morenoff, J. D. and Gannon-Rowley, T. (2002), ‘Assessing Neighborhood Effects: Social Processes and New Directions in Research’, Annual Review of Sociology, 28: 443–78. Sampson, R. J. and Raudenbush, S. W. (2004), ‘Seeing Disorder: Neighborhood Stigma and the Social Construction of “Broken Windows”’, Social Psychology Quarterly, 67: 319–42. Sampson, R. J., Raudenbush, S. W. and Earls, F. (1997), ‘Neighborhoods and Violent Crime: A Multilevel Study of Collective Efficacy’,Science , 277: 918–24. Skogan, W. G. (1990), Disorder and Decline: Crime and the Spiral of Decay in American Neighborhoods. University of California Press. —— (2015), ‘Disorder and Decline: The State of Research’, Journal of Research in Crime and Delinquency, 52: 464–85. Steenbeek, W. and Hipp, J. R. (2011), ‘A Longitudinal Test of Social Disorganization Theory: Feedback Effects among Cohesion, Social Control, and Disorder’, Criminology, 49: 833–71. Thoits, P. A. (1985), ‘Social Support Processes and Psychological Wellbeing: Theoretical Possibilities’, in I. G. Sarason and B. R. Sarason, eds, Social Support Theory, Research and Applications, 51–72. Martinus Nijhof. —— (1995), ‘Stress, Coping, and Social Support Processes – Where Are We – What Next’, Journal of Health and Social Behavior, 35: 53–79. Twisk, J. W. R. (2013), Applied Longitudinal Data Analysis for Epidemiology: A Practical Guide. Cambridge University Press. Vancluysen, K., Van Craen, M. and Ackaert, J. (2011), ‘The Perception of Neighborhood Disorder in Flemish Belgium: Differences Between Ethnic Majority and Minority Group Members and Bearing on Fear of Crime’, Journal of Community Psychology, 39: 31–50. Welsh, B. C., Braga, A. A. and Bruinsma, G. J. N. (2015), ‘Reimagining Broken Windows: From Theory to Policy’, Journal of Research in Crime and Delinquency, 52: 447–63. Wikström, P. O. H., Oberwittler, D., Treiber, K. and Hardie, B. (2012), Breaking Rules: The Social and Situational Dynamics of Young People’s Urban Crime. OUP. Wright, J. P. and Cullen, F. T. (2001), ‘Parental Efficacy and Delinquent Behavior: Do Control and Support Matter?’, Criminology, 39: 677–705. Wyant, B. R. (2008), ‘Multilevel Impacts of Perceived Incivilities and Perceptions of Crime Risk on Fear of Crime – Isolating Endogenous Impacts’, Journal of Research in Crime and Delinquency, 45: 39–64. Yang, S. M. and Pao, C. C. (2015), ‘Do We “See” the Same Thing? An Experimental Look into the Black Box of Disorder Perception’, Journal of Research in Crime and Delinquency, 52: 534–66. Zhang, L., Messner, S. F., Liu, J. and Zhuo, Y. A. (2009), ‘Guanxi and Fear of Crime in Contemporary Urban China’, British Journal of Criminology, 49: 472–90.

Page 17 of 17

Downloaded from https://academic.oup.com/bjc/advance-article-abstract/doi/10.1093/bjc/azy002/4907972 by Ghent University user on 01 March 2018