|
|
GUANXI AND FEAR OF CRIME IN CONTEMPORARY URBAN CHINA(2) |
|
GUANXI AND FEAR OF CRIME IN CONTEMPORARY URBAN CHINA(2)
Measures
The dependent variable in the analysis is self-reported fear of crime. The question on the survey instrument asks: 'When you walk in the neighborhood that you live in at night, do you feel fear?' [FN4] The original response categories are 'not at all' (1), 'somewhat' (2) and 'very much' (3). An examination of the univariate distribution reveals that the response 'very much' is quite rare (only approximately 3 per cent). We have accordingly created a dummy variable that differentiates respondents who report no fear ('not at all') from those who report fear ('somewhat' or 'very much').
The primary independent variable is guanxi. Guanxi networks span diverse relationships, including relatives, friends, former classmates and neighbours that have developed through blood-ties, personal contacts, interaction and associations. The Tianjin dataset contains items relevant to one form of guanxi--guanxi in the immediate neighbourhood. There are three survey items that can be used to measure individuals' guanxi in their neighbourhoods: (1) 'How many close relatives (in addition to the immediate family members in your household) do you have who live in your neighborhood?'; (2) 'How many close friends do you have who live in your neighborhood?'; and (3) 'How many neighbors do you know quite well in your neighborhood?' The responses to these questions are summed to create a measure of guanxi according to which high scores represent a strong guanxi network. The distribution of the measure shows significant variation, with a minimum value of 0 and a maximum value of 400 (mean value = 9.13 and standard deviation = 14.21). Given the high degree of skewness, we convert the measure into natural logarithms to normalize the distribution.
We acknowledge that our survey items only allow for an indirect measure of guanxi. As explained above, guanxi involves social ties that are imbued with sentiments of reciprocity, obligation and affection, and they serve both instrumental and emotional purposes. The survey items do not ask specifically about these qualities, but they do capture relationships that are laden with affect ('close' relatives and friends; persons known 'quite well'). Our measure is thus best regarded as an indicator of the associational infrastructure of guanxi among neighbours.
Our measures of the socio-demographic control variables are straightforward. Gender is a dummy variable scored in the direction of 'female'. Age is measured in years. Education is an ordinal measure with three categories: 'illiterate and elementary school' (0); 'middle school' (1); and 'college and above' (2). The measure of family income refers to monthly income per person for members of the household living together. The response categories are: 'below 500 yuan (Chinese dollars)' (0); '500-999 yuan' (1); and '1,000 yuan and above' (2). In addition to these socio-demographic variables, we consider two direct indictors of vulnerability. Physical strength is measured using responses to the survey item: 'How would you rate your physical strength?' The indicator of self-defence/alertness is based on responses to an item: 'How do you rate your capability of self-defense/alertness with respect to personal safety?' [FN5] These two measures are scored as: 'poor' (1); 'average' (2); 'strong' (3); and 'very strong' (4). Higher scores thus indicate less vulnerability.
Drawing upon the victimization perspective, we include measures of victimization for both violent and property offences. [FN6] They reflect respondents' reports of having been the victim of robbery, assault, personal theft or being 'swindled' during the past five years. Personal violent victimization is a dummy variable scored '1' for respondents who reported being victimized on either a robbery or an assault during the reference period, while personal property victimization is a dummy variable scored '1' for those who reported being a victim of either of the two property offences.
Our measure of perceived social disorder is based on three items, which are similar to those used in Western research: (1) 'In the past six months, have you seen or heard youth groups fighting in your neighborhood?'; (2) 'In the past six months, have you seen or heard young hooligans creating trouble in your neighborhood?'; and (3) 'In the past six months, have you seen or heard loud arguments or quarrels between neighbours?' The response categories to these three questions are: 'often' (4); 'sometimes' (3); 'rarely' (2); and 'never' (1). The measure of perceptions of social disorder is a composite index based on the sum of responses to these three items (the standardized alpha reliability = 0.71).
The Tianjin survey contains items that are similar to those commonly used in Western studies to capture certain features of local ties in the neighbourhood. These items reflect the degree to which respondents provide various forms of assistance to their neighbours. The specific questions ask respondents how often they watch their neighbours' children, help grocery shop for their neighbours and help take care of elderly neighbours. The response categories range from 1 = almost never to 6 = several times a week. Our measure of local assistance ties is a composite index based on the sum of responses to these three items (the standardized alpha reliability = 0.65).
Following Western research (e.g. Sampson et al. 1997: Gibson et al. 2002), the measure of perceived collective efficacy combines indicators of social cohesion among neighbours and informal control. The survey contains three items relevant to social cohesion: 'Do you think your neighborhood is a close-knit neighborhood?'; 'When you or your family have (has) some important matters, does anyone in this neighborhood care much about them?'; and 'Do people in this neighborhood trust each other?' Each question has a Likert-type response set: 1 = certainly so, to 4 = certainly not. Responses to these questions were recoded and summed to create an index of social cohesion in the direction that high scores represent high levels of social cohesion (the standardized alpha reliability = 0.71). The item relevant to informal control is: 'If there is a major problem around here, do neighbors get together to discuss and work out solutions to deal with it?' The response set and coding for this item are the same as that for the items of social cohesion. We combined the measures of social cohesion and informal control to create the measure of perceived collective efficacy.
Our remaining set of control variables reflects characteristics of the neighbourhood in contrast with individual properties. We aggregated and computed the means of responses to the items that were used to create the individual-level measures of gaunxi, perceived neighbourhood disorder, and local assistance ties in each of the 50 neighbourhoods and summed them to create three composite measures of guanxi, social disorder and local assistance ties at the neighbourhood level. The standardized alpha reliability is 0.79 for the measure of social disorder and 0.86 for the measure of local ties. The same procedure was followed to create a measure of collective efficacy at the neighbourhood level. We aggregated and computed the means of responses to the items that were used to create the individual-level measure of social cohesion and informal control and then combined them to create the measure of collective efficacy at the neighbourhood level. The standardized alpha reliability is 0.79 for the measure of social cohesion, and the correlation between the measures of social cohesion and informal control is 0.43. [FN7] Finally, we also control for neighbourhood disadvantage. The measure is a composite index based on the proportion of households with family income below 500 Chinese dollars and the proportion of residents unemployed in the neighbourhood. The standardized alpha reliability is 0.76. Descriptive statistics for all measures are reported in Appendix 1.
Statistical procedures
Given the dichotomous nature of the dependent variable of fear of crime, we use the Hierarchical Generalized Liner Model (HGLM) for the statistical estimation. Our analyses include individual and neighbourhood levels. The level one model in HGLM consists of three parts: a sampling model, a link function and a structural model. Different from a normally distributed dependent variable, whose level 1 sampling model is a normal distribution, our dependent variable is a dichotomous variable. Thus, the sampling model is the binomial sampling model. In the case of a normal distribution, the link function is an 'identity function', which does not change the sampling distribution. In our dichotomous dependent variable situation, the link function is the log of odds of 'success' (fear) (logit). Similar to the familiar logistic regression situation, the logic link function can take any real value, while the probability of success is constrained to be in the interval (0, 1). The logit link function transforms the binary distribution to an unconstrained range. The remainder of the model is the same as in the case of ordinary logistic regression (for details, see Raudenbush and Bryk 2002: 293-6). [FN8] We begin the multivariate logistic analysis by estimating a model that includes the measure of guanxi and the individual-level control variables. We then introduce the neighbourhood-level control variables into the model.
Results
Before reporting the results of the multivariate analyses, the issue of discriminant validity needs to be considered. Are the operational indicators of guanxi in the neighbourhood empirically distinct from other items that have been used to capture various aspects of the general construct of 'local ties' or 'neighbourhood integration' in Western research? We can address this question by conducting confirmatory factor analysis with structural equation modelling. To determine if our proposed indicators of guanxi are empirically distinct from the indicators of local assistance ties, we have compared the fit indices of two measurement models. One model allows all of the indicators to load on a single latent construct; the other stipulates two latent constructs. The results indicate that, consistent with our conceptualization, the measurement model with two latent constructs fits the data significantly better than does the model with a single latent construct. [FN9]
We have conducted similar analyses to determine if our indicators of guanxi are distinct from those of social cohesion. Once again, the results indicate that modelling these indicators as representing two distinct latent constructs rather than as indicators of a single construct is more faithful to the data. [FN10] These confirmatory factor analyses thus indicate that that our measure of guanxi is in fact empirically distinct from some of the indicators of local ties and social cohesion commonly used in Western studies.
Turning to the multivariate results, Table 1 reports estimates from the logistic regression of fear of crime on the measure of neighbourhood guanxi and the control variables. [FN11] Model 1 includes only individual-level variables, whereas Model 2 introduces the neighbourhood-level variables. The results are highly similar across specifications. Of greatest theoretical interest is the significant, negative effect of the measure of guanxi. Consistent with our main hypothesis, respondents with a high degree of guanxi in the neighbourhood are less likely to report being fearful of crime than those with a lower degree of guanxi. This effect emerges despite controls for a fairly comprehensive array of controls for relevant individual and neighbourhood characteristics.
TABLE 1 HLM logistic regressions of fear of crime on guanxi and control
variables
-------------------------------------------------------------------------------
Variables 1 2
-------------------------------------------------------------------------------
Fixed effects
Intercept -0.870 (0.471) -2.971 (2.174)
Individual-level variables
Guanxi (logged) -0.35 [FNa1] (0.15) -0.38 [FNa1] (0.16)
Gender (female) 1.84 [FNaa1] (0.13) 1.85 [FNaa1] (0.12)
Age -0.01 [FNaa1] -0.01 [FNaa1] (0.004)
(0.004)
Education 0.24 [FNa1] (0.11) 0.26 [FNa1] (0.12)
Family income -0.08 (0.07) -0.06 (0.07)
Physical strength -0.13 [FNa1] (0.07) -0.13 [FNd1] (0.07)
Self-defence/alertness -0.41 [FNaa1] (0.07) -0.41 [FNaa1] (0.07)
Perceived collective effi cacy -0.05 [FNd1] (0.03) -0.05 [FNd1] (0.03)
Local assistance ties 0.01 (0.03) 0.01 (0.03)
Perceived disorder 0.27 [FNaa1] (0.03) 0.26 [FNaa1] (0.04)
Personal violent victimization 0.32 (0.28) 0.32 (0.28)
Personal property victimization 0.33 [FNa1] (0.14) 0.34 [FNa1] (0.13)
Neighbourhood-level variables
Guanxi - 0.01 (0.02)
Local assistance ties - 0.14 (0.05)
Collective effi cacy - -0.02 (0.13)
Social disorder - 0.26 [FNd1] (0.15)
Neighbourhood disadvantage - 0.004 (0.003)
Random effects
Intercept, <> 00 0.097 0.092
<>
x2 87.777 [FNaa1] 76.889 [FNaa1]
-------------------------------------------------------------------------------
Notes: N = 2,454 for the individual-level variables and 50 for the
neighbourhood-level variables.
FNd1. p < 0.05 one-tailed test
FNa1. p < 0.05
FNaa1. p < 0.01
The findings for the control variables are generally in accord with those yielded in past research. Consistent with studies in the West, females, respondents who perceive disorder in the neighbourhood and victims of property crimes are more likely to be fearful than their counterparts, whereas those who perceive a high degree of collective efficacy in the neighbourhood are less likely to be fearful. The findings for age, education and self-defence/alertness replicate those reported by Liu et al. (2008; 2009) in regressions with slightly different specifications. Contrary to Western research, young people and the highly educated are more likely to express fear of crime in the Chinese context. Physical strength and capacity of self-defence/alertness are negatively associated with fear of crime, which is in accordance with the general logical of the physical vulnerability perspective.
The only neighbourhood-level variable to yield a significant coefficient (one-tailed test) is the measure of social disorder, replicating Liu et al. (2008). A high degree of disorder in the neighbourhood increases the likelihood that a respondent reports being fearful of crime. This effect supports the disorder model and is consistent with the findings in Western studies (Skogan 1990; Taylor 1996; Taylor et al. 1985). [FN12] Given that only one neighbourhood-level variable exhibits a significant effect, it is not surprising that the decrease in the residual deviance across Models 1 and 2 is quite modest-- just over 5 per cent ((0.097-0.095)/0.097).
Finally, we have assessed the robustness of the effect of guanxi on fear of crime by re-estimating the regressions after decoupling the two components of the collective efficacy measure--the index of social cohesion and the item reflecting informal social control. As reported above, the correlation between the two components of collective efficacy is 0.43, which is much lower than that (r = 0.80) reported by Sampson et al. (1997). Accordingly, the composite measure might mask effects of social cohesion and informal control, and failure to adjust for any such effects might bias the coefficients for the guanxi measure. The results of regressions with separate measures of social cohesion and informal control at both the individual and neighbourhood levels are virtually identical to those reported above. The coefficient for guanxi is significantly negative, while the coefficients for social cohesion and informal control are non-significant.
Summary and Conclusion
The present study has assessed guanxi as a predictor of fear of crime in contemporary urban China using data collected from a recent survey of criminal victimization in the city of Tianjin. Applying the general vulnerability model as developed in the West, we have argued that guanxi represents a distinctively Chinese form of social capital and should be considered an important dimension of individual social vulnerability. Accordingly, guanxi is hypothesized to be a significant predictor of fear of crime.
Our analyses yield support for this hypothesis when other influential factors are statistically controlled. Residents who have extensive guanxi networks in the neighbourhood are significantly less likely to exhibit fear of crime than those without these networks. Although our measure of guanxi does not fully capture all of the nuances of the concept, our analyses indicate that it is distinct from the measures of local assistance ties and social cohesion commonly used in Western studies. To the extent that the measure does reflect key features of guanxi networks, our results suggest that the effects of indicators of different forms of social capital may be specific to the socio-cultural context.
The Western studies on the possible effect of local ties or support on people's feelings and perceptions of their safety have yielded mixed findings. Our data indicate that a fairly similar measure of local assistance ties has no effect on reported fear of crime in the Chinese context. However, the measure of perceived collective efficacy, which is a combination of informal control and social cohesion, has a modest effect, which is consistent with Gibson et al.'s study (2002). These findings across Chinese and Western studies suggest that the social patterning of fear of crime deserves further exploration using not only similar measures, but different measures as well that reflect variation in cross-cultural context (Murck 1997; Quann and Hung 2002).
Results for the control variables are in large measure supportive of the vulnerability, victimization and disorder models that have been widely used in Western research. Consistent with an earlier study by Liu et al. (2009), female respondents and respondents who had victimization experience in property crime are more likely to report fear of crime, while respondents with stronger physical strength or stronger capacity of self-defence/alertness are less likely to be fearful. Also, the measure of perceived disorder in the neighbourhood and the measure of neighbourhood disorder as a contextual variable have significant positive effects on reporting fear of crime. These findings provide support for the vulnerability, victimization and disorder models. However, our multilevel analyses indicate that guanxi has no contextual effect on fear of crime. This suggests that guanxi entails a personal network with deep affective involvement that exerts an individual psychological effect rather than a contextual or ecological effect on fear of crime.
Furthermore, the data replicate the findings reported by Liu et al. (2009) based on regression models without the measure of guanxi that younger people are more likely to be fearful of being victimized than older people. This pattern is contrary to the prediction of the vulnerability model and inconsistent with the common findings in Western studies. Liu et al. (2009) speculate that the age effect in urban China may reflect differential exposure to the media. The elderly are likely to rely heavily on governmentally controlled media, which tend to avoid coverage of topics that reflect unfavourably upon society, such as crime.
We can offer an additional interpretation. Given the Chinese cultural tradition of 'familism' that emphasizes strong ties and stakes in extended family relationships, old people are not likely to live alone in urban China (Lau 1981; Ting and Chiu 2002; Whyte 2005; Zhang 2004). They also retain close relationships with their children and relatives, who are heavily involved with their lives, with less concern for independence and privacy than is common in the West. Also, urban China communities have semi-public organizations-- neighbourhood committees. One major function of these committees is to organize recreational and exercise activities such as choral or taiji groups for residents in their neighbourhoods, especially for elderly residents (Read and Chen 2006). The greater involvement of old people with extended families and their communities, combined with their low risk of being victimized by crime (see Messner et al. 2007), may provide a plausible explanation for the finding that older people are less likely to report fear of crime than younger ones in contemporary urban China.
Another finding that differs from those reported in Western research is that respondents with higher education are more likely to report fear of crime than those with lower education, which is also inconsistent with the vulnerability model and Western studies. Liu et al. (2009) propose that this pattern might also be interpreted with reference to differential exposure to the media--the highly educated are likely to be exposed to non-governmental media and the internet. We suggest that in addition to these media effects, perhaps people who have received more education are less likely to have well developed 'street smarts' and skills for negotiating risky situations. Further research is needed to assess these speculative interpretations.
To summarize, our findings demonstrate that there may be common determinants of fear of crime (e.g. gender, victimization experience and social disorder) across different social and cultural settings. However, some determinants of fear of crime might reflect distinctive features of the social and cultural setting, such as guanxi. A critically important task for comparative criminology is to conceptualize social factors that might be uniquely relevant to particular social-cultural contexts and to examine the impact of these factors across such contexts.
Funding
This material is based upon work supported by the National Science Foundation under Grant No. 0351014. Any opinions, findings, and conclusions or recommendations expressed in this material are those of the authors and do not necessarily reflect the views of the National Science Foundation.
REFERENCES
BELL, D. (2000), 'Guanxi: A Nesting of Groups', Current Anthropology, 41: 132- 8.
BIAN, Y. (1994), Work and Inequality in Urban China. Albany, NY: SUNY Press.
BIAN, Y. and ZHANG, W. (2001), 'Economic Regime, Social Networks, and Occupational Mobility', Chinese Social Sciences, 128: 70-89.
BRAUNGART, M. M., BRAUNGART, R. G. and HOYER, W. J. (1980), 'Age, Sex, and Social Factors in Fear of Crime', Sociological Forces, 13: 55-6.
China Law Yearbooks (1989-2006), Editorial Committee of Law Yearbook of China. Beijing, China: Press of Law Yearbook of China.
CLEMENTE, F. and KLEIMAN, M. B. (1977), 'Fear of Crime in the United States: A Multivariate Analysis', Social Forces, 56: 519-31.
COVINGTON, J. and TAYLOR, R. B. (1991), 'Fear of Crime in Urban Residential Neighborhoods: Implications of Between and Within Neighborhood Sources for Current Models', Sociological Quarterly, 32: 231-49.
GATES, L. B. and ROHE, W. M. (1987), 'Fear and Reactions to Crime: A Revised Model', Urban Affairs Quarterly, 22: 425-53.
GEDDIE, M. W., DEFRANCO, A. L. and GEDDIE, M. (2005), 'A Comparison of Relationship Marketing and Guanxi: Its Implications for the Hospitality Industry', International Journal of Contemporary Hospitality Management, 17: 614-32.
GIBSON, C. L., ZHAO, J., LOVRICH, N. P. and GAFFNEY, M. J. (2002), 'Social Integration, Individual Perceptions of Collective Efficacy, and Fear of Crime in Three Cities', Justice Quarterly, 19: 537-64.
GOLD, T., GUTHRIE, D. and WANK, D. (2002), 'Introduction', in T. Gold, D. Guthrie and D. Wank, eds, Social Connections in China: Institutions, Culture, and the Changing Nature of Guanxi, 3-20. New York: Cambridge University Press.
GUTHRIE, D. (2002), 'Information Asymmetries and the Problem of Perceptions: The Significance of Structural Position in Assessing the Importance of Guanxi in China', in T. Gold, D. Guthrie and D. Wank, eds, Social Connections in China: Institutions, Culture, and the Changing Nature of Guanxi, 37-56. New York: Cambridge University Press.
FAIRBANK, J. K. (1987), The Great Chinese Revolution: 1800-1985. New York: Harper and Row.
FEI, X. [1949] (1992), From the Soil, the Foundations of Chinese Society: A Translation of Fei Xiaotong's Xiangtu Zhongguo. Berkeley, CA: University of California University Press.
FERGUSON, K. M. and MINDEL, C. H. (2007), 'Modeling Fear of Crime in Dallas Neighborhoods: A Test of Social Capital Theory', Journal of Research in Crime and Delinquency, 53: 322-49.
HALE, C. (1996), 'Fear of Crime: A Review of the Literature', International Review of Victimology, 4: 79-150.
HANSER, A. (2002), 'Youth Job Searches in Urban China: The Use of Social Connections in a Changing Labor Market', in T. Gold, D. Guthrie and D. Wank, eds, Social Connections in China: Institutions, Culture, and the Changing Nature of Guanxi, 137-62. New York: Cambridge University Press.
HILLMAN, B. (2005), 'A Book Review: Social Connections in China: Institutions, Culture, and the Changing Nature of Guanxi, edited by Thomas Gold, Doug Guthrie, and David Wank', The China Journal, 53: 209-21.
HINDELANG, M. J., GOTTFREDSON, M. R. and GAROFALO, J. (1978), Victims of Personal Crime. Cambridge, MA: Ballinger.
KILLIAS, M. and CLERICI, C. (2000), 'Different Measures of Vulnerability in their Relation to Different Dimensions of Fear of Crime', British Journal of Criminology, 40: 437-50.
KING, A. Y. C. (1985), 'The Individual and Group in Confucianism: A Relational Perspective', in D. J. Munro, ed., Individualism and Holism: Studies in Confucian and Taoist Values, 57-70. Ann Arbor, MI: Center of Chinese Studies, University of Michigan.
KIPNIS, A. B. (1997), Producing Guanxi: Sentiment, Self and Subculture in North China Village. Durham, NC: Duke University Press.
LAU, S.-K. (1981), 'Chinese Familism in an Urban-Industrial Setting: The Case of Hong Kong', Journal of Marriage and the Family, 43: 977-92.
LIANG, C. [1949] (1986), The Essential Meanings of Chinese Culture. Hong Kong: Zheng Zhong.
LISKA, A. E., SANCHIRICO, A. and REED, M. D. (1988), 'Fear of Crime and Constrained Behavior: Specifying and Estimating a Reciprocal Effects Model', Social Forces, 66: 827-37.
LIU, J. (2005), 'Crime Patterns during the Market Transition in China', British Journal of Criminology, 45: 613-33.
LIU, J. (2006), 'Modernization and Crime Patterns in China', Journal of Criminal Justice, 34: 119-30.
LIU, J., MESSNER, S. F., ZHANG, L. and ZHUO, Y. (2008), 'Investigating Production Mechanism of Fear of Crime in Urban Chinese Communities: Role of Social Disorder', Crime and Criminal Justice International, 11: 27-61.
----- (2009), 'Fear of Crime in Contemporary Urban China: Assessing and Elaborating Theoretical Models in a Distinctive Socio-Political Context', Journal of Community Psychology (forthcoming).
LUO, W. and WANG, X. (2004), 'The Conception of Safety and Security, its Development, and Improvement', Journal of Jiangsu Police College, 5: 1-14.
MESSNER, S. F., ZHOU, L., ZHANG, L. and LIU, J. (2007), 'Risks of Criminal Victimization in Contemporary Urban China: An Application of Lifestyle/Routine Activities Theory', Justice Quarterly, 24: 496-522.
MURCK, M. (1997), 'Victimization and Fear of Crime in Western Europe', Security Journal, 9: 163-4.
PAIN, R. (2000), 'Place, Social Relations and the Fear of Crime: A Review', Progress in Human Geography, 24: 365-87.
PANTAZIS, C. (2000), 'Fear of Crime, Vulnerability and Poverty: Evidence from the British Crime Survey', British Journal of Criminology, 40: 414-39.
POTTER, P. B. (2002), 'Guanxi and the PRC Legal System: From Contradiction to Complementarity', in T. Gold, D. Guthrie and D. Wank, eds, Social Connections in China: Institutions, Culture, and the Changing Nature of Guanxi, 179-96. New York: Cambridge University Press.
QUANN, N. and HUNG, K. (2002), 'Victimization Experience and the Fear of Crime: A Cross-National Study', in P. Nieuwbeerta, ed., Crime Victimization in Comparative Perspective: Results from the International Crime Victims Survey, 1989-2000, 301-16. Annandale, NSW, Australia: Federation Press.
RAUDENBUSH, S. W. and BRYK, A. S. (2002), Hierarchical Linear Models: Applications and Data Analysis Methods. Thousand Oaks, CA: Sage Publications.
READ, B. L. and CHEN, C.-M. (2006), 'The State's Evolving Relationship with Urban Society: China's Neighborhood Organizations in Comparative Perspective', unpublished manuscript.
ROJEK, D. G. (1996), 'Changing Directions of Chinese Social Control', in C. B. Fields, J. Richter and H. Moore, eds, Comparative Criminal Justice: Traditional and Nontraditional Systems of Law and Control, 234-49. Prospect Heights, IL: Waveland Press.
SACCO, V. F. (1993), 'Social Support and the Fear of Crime', Canadian Journal of Criminology, 32: 187-96.
SAMPSON, R. J., RAUDENBUSCH, S. W. and EARLS, F. (1997), 'Neighborhoods and Violent Crime: A Multilevel Study of Collective Efficacy', Science, 15: 918-24.
SHEN, H. (2006), 'Urban Safety and Security', China Public Security, 6: 23-37.
SITU, Y. and LIU, W. (1996), 'Restoring the Neighborhood, Fighting against Crime: A Case Study in Guangzhou City, People's Republic of China', International Criminal Justice Review, 6: 89-102.
SKOGAN, W. G. (1990), Disorder and Decline: Crime and the Spiral of Decay in American Neighborhoods. New York: Free Press.
SKOGAN, W. G. and MAXFIELD, M. G. (1981), Coping with Crime: Individual and Neighborhood Differences. Beverly Hills, CA: Sage.
SMART, A. (1993), 'Gifts, Bribes, and Guanxi: A Reconsideration of Bourdieu's Social Capital', Cultural Anthropology, 8: 388-408.
SMITH, W. R. and TORSTENSSON, M. (1997), 'Gender Differences in Risk Perception and Neutralizing Fear of Crime', British Journal of Criminology, 37: 608-34.
STAFFORD, M. C. and GALLE, O. R. (1984), 'Victimization Rates, Exposure to Risk, and Fear of Crime', Criminology, 22: 173-85.
SUN, W. and SHI, J. (2004), 'On Improvement of Public Perceptions of Safety and Security', Journal of Shanghai Public College, 3: 56-69.
TAYLOR, R. B. (1996), 'Neighborhood Responses to Disorder and Local Attachments: The Systemic Model of Attachment, Social Disorganization, and Neighborhood Use Value', Sociological Forum, 11: 41-74.
TAYLOR, R. B. (2001), Breaking Away from Broken Windows: Baltimore Evidence and Implications for the Nationwide Fight against Crime, Grime, Fear, and Decline. New York: Westview Press.
TAYLOR, R. B. (2002), 'Fear of Crime, Social Ties, and Collective Efficacy: Maybe Masquerading Measurement, Maybe Déjà vu All Over Again', Justice Quarterly, 19: 773-92.
TAYLOR, R. B. and HALE, M. (1986), 'Testing Alternative Models of Fear of Crime', Journal of Criminal Law and Criminology, 77: 151-89.
TAYLOR, R. B., GOTTFREDSON, S. D. and BROWER, S. (1984), 'Block Crime and Fear: Defensible Space, Local Ties, and Territorial Functioning', Journal of Research in Crime and Delinquency, 21: 303-31.
TAYLOR, R. B., GOTTFREDSON, S. D. and BROWER, S. (1985), 'Attachment to Place: Discriminant Validity, and Impacts of Disorder and Diversity', American Journal of Community Psychology 13: 525-42.
THOMPSON, E. E. and KRAUSE, N. (1998), 'Living Alone and Neighborhood Characteristics as Predictors of Social Support in Late Life', The Journals of Gerontology, 53B: S354-64.
TING, K. and CHIU, S. W. K (2002), 'Leaving Parental Home: Chinese Culture in an Urban Setting', Journal of Marriage and the Family, 64: 614-26.
VANDERVEEN, G. (2002), 'Experiencing Safety: Proposing a Novel Approach to Scout Differences between Men and Women', in P. Nieuwbeerta, ed., Crime Victimization in Comparative Perspective: Results from the International Crime Victims Survey, 1989-2000, 335-51. Annandale, NSW, Australia: Federation Press.
WANG, Y. and LI, G. (2005), 'China Public Security in 2004', in X. Ru, X. Lu and P. Li, eds, Analysis and Forecast of China's Social Development (2005), 137-46. Beijing, China: Social Sciences Academic Press.
WHYTE, M. K. (2005), 'Continuity and Change in Urban Chinese Family Life', The China Journal, 53: 9-34.
WILL, J. A. and MCGRATH, J. H. (1995), 'Crime, Neighborhood Perceptions, and the Underclass: The Relationship between Fear of Crime and Class Position', Journal of Criminal Justice, 23: 163-76.
WILSON, J. Q. and KELLING, G. L. (1982), 'Broken Windows', The Atlantic Monthly, March 1982, 29-38.
WITTEBROOD, K. (2002), 'Fear of Crime and Victimization in Western Industrialized Countries: A Multilevel Analysis', in P. Nieuwbeerta, ed., Crime Victimization in Comparative Perspective: Results from the International Crime Victims Survey, 1989-2000, 285-99. Annandale, NSW, Australia: Federation Press.
YANG, M. M. (1994), Gifts, Favors, and Banquets: The Art of Social Relationships in China. Ithaca, NY: Cornell University Press.
YANG, M. M. (2002), 'The Resilience of Guanxi and Its New Development: A Critique of Some New Guanxi Scholarship', The China Quarterly, 170: 459-76.
YUAN, Y. and FAN, H. (2001), 'The Quality of Chinese Residents' Life', in X. Ru, X. Lu and T. Shan, eds, Analysis and Forecast of China's Social Development (2001), 58-74. Beijing, China: Social Sciences Academic Press.
ZEN, H. and YUAN, Y. (2005), 'The Quality of Chinese Residents' Life in 2004', in X. Ru, X. Lu and P. Li, eds, Analysis and Forecast of China's Social Development (2005), 50-64. Beijing, China: Social Sciences Academic Press.
ZHANG, L., MESSNER, S. F. and LIU, J. (2007), 'A Multilevel Analysis of the Risk of Household Burglary in the City of Tianjin, China', British Journal of Criminology, 47: 918-37.
ZHANG, Q. F. (2004), 'Economic Transition and New Patterns of Parent-Adult Child Co-Residence in Urban China', Journal of Marriage and Family, 66: 1231- 45.
ZHANG, Y. and ZHANG, Z. (2006), 'Guanxi and Organizational Dynamics in China: A Link between Individual and Organizational Levels', Journal of Business Ethics, 67: 375-92.
ZHOU, L. and CONG, M. (2001), 'Criminology in China: Perspectives and Development', in J. Liu, L. Zhang and S. F. Messner, eds, Crime and Social Control in a Changing China, 57-64. Westport, CT: Greenwood.
Appendix 1: Descriptive Statistics of Variables
-------------------------------------------------------------------------------
Mean SD Minimum Maximum
-------------------------------------------------------------------------------
Dependent variable
Fear of crime 2.66 0.54 1.00 3.00
Individual-level independent
variables
Guanxi (logged) 0.82 0.40 0.00 2.60
Gender 0.49 0.50 0.00 1.00
Age 42.85 15.67 18 78
Education 1.14 0.54 0.00 2.00
Family income 0.85 0.77 0.00 2.00
Physical strength 2.50 0.88 1.00 4.00
Self-defence/alertness 2.60 0.78 1.00 4.00
Perceived collective efficacy 12.88 2.24 4.00 16.00
Local assistance ties 4.32 2.32 3.00 18.00
Perceived disorder 5.30 1.71 3.00 12.00
Personal violent victimization 0.04 0.19 0.00 1.00
Personal property victimization 0.18 0.39 0.00 1.00
Neighbourhood-level independent 9.14 2.68 4.72 17.82
variables Guanxi
Local assistance ties 4.32 0.55 3.43 5.73
Collective efficacy 12.88 0.47 11.60 13.96
Social disorder 5.29 0.39 4.34 6.14
Neighbourhood disadvantage 57.14 19.30 12.00 94.00
-------------------------------------------------------------------------------
Note: N = 2,454 for the individual-level variables and 50 for the
neighbourhood-level variables.
[FN4]. Although the survey item does not refer specifically to crime, it appears in the context of a criminal victimization survey, and thus it is reasonable to assume that the response refers to fear of crime. This type of measure of fear of crime has long been used in the Western literature (e.g. Hindelang et al. 1978: 176).
[FN5]. The item for self-defence/alertness might appear to confound multiple constructs and be a 'double-barrelled' question, but this reflects difficulties in translation. The wording of the item in Chinese is <>, which is readily understandable for Chinese respondents.
[FN6]. We acknowledge a possible source of measurement error for the victimization variables. Respondents who are more fearful might be more likely to remember recent victimizations. The observed relationship for these variables must accordingly be interpreted cautiously.
[FN7]. The correlation between the measures of informal control and social cohesion in our sample is notably lower than that observed in the research by Sampson et al. (1997) based on data for Chicago neighbourhoods, as noted in a previous analysis based on the Tianjin data (Zhang et al. 2007). Sampson et al. report a correlation of 0.80. The divergence may reflect the fact that only a single indicator of informal social control is available in our survey, whereas Sampson et al. use multiple indicators. It is also possible that the two dimensions of collective efficacy do not in fact converge as closely in China as in the West.
[FN8]. We also estimated hierarchical ordinal logistic regressions with the original three-category measure of fear of crime, although the data do not meet the assumption of parallel regression. The results are substantively similar to those yielded in the binary logistic regressions.
[FN9]. The one-latent-construct model has X2 = 148.330 with DF = 9, whereas a two-latent-construct model has X2 = 47.461 with DF = 8. The X2 difference is 100.869, and the DF difference is 1 between the two models. A X 2 of 100.869 with 1 DF is statistically significant, indicating that the two-latent-construct model differs significantly from the one-latent-construct model. A comparison of the CFIs (0.898 for the one-latent-construct model and 0.971 for the two-latent-construct model) indicates that the two-construct model fits the data much better.
[FN10]. For these analyses, a one-latent-construct model has X2 = 196.539 with DF = 9 and CFI = 0.893, and a two-latent-construct model has X2 = 41.093 with DF = 8 and CFI = 0.981. The X2 difference is 155.500 (1 DF), which is statistically significant.
[FN11]. Our analysis of the intercept-only model indicates that the statistics for the variance component (0.073) is significant (p < 0.01), implying that respondents' reported fear of crime varies significantly across the sampled neighbourhoods.
[FN12]. We also estimated the full model after adding a measure of neighbourhood crime level. Responses to the victimization in household burglary and bicycle theft at home were aggregated and summed to create the measure. The results are unchanged when this additional control variable is entered into the model.
|
|
|
|
|
|
