Spatial patterns of violent crimes and neighborhood characteristics in Changchun,China

Abstract

Crime is one of the major concerns facing Chinese cities. Using crime data compiled at police precinct level in 2008, this research examines spatial patterns of violent crimes in Changchun, and explores the relationship between the spatial distribution of violent crimes and neighborhood characteristics. Crime rates are applied as a measure of the intensity of violent crimes. Spatial statistics and geographic information systems are used to detect violent crime hot spots, or statistically significant locales of high violent crime rates in Changchun. A multiple linear regression model is calibrated to assess the impacts of contextual neighborhood characteristics on violent crimes. The analytical results demonstrate that the risk or intensity of violent crimes is strongly concentrated in the central city area, and neighborhood socioeconomic, demographic, especially land use characteristics are effective in accounting for the spatial variation in the distribution of violent crimes across the city of Changchun. These findings are largely in line with the routine activities theory, social disorganization theory, and the framework of crime prevention through environmental design, which emphasize the importance of opportunities, local social context, and environmental design in shaping the spatial pattern of and reducing urban crimes.

Keywords

China crime hotspots neighborhood characteristics spatial patterns violent crime

Introduction

With the accelerating urbanization in China, urban crime has increasingly become one of the major challenges facing Chinese cities. Similar to the experience of other nations, Chinese cities have become more diverse and differentiated along social and economic dimensions. Undeniably, an unfortunate cost of China’s rapid modernization and urbanization is increasing risks of criminal victimization. City governments, policy makers, and policing agencies all recognize the importance of better understanding the dynamics of urban crimes. Exploring and analyzing patterns of urban crimes can provide crucial information about crimes and form indispensable bases in the development of theoretical explanations and effective policing practices. As pointed out by Liu (2005), examining China’s crime patterns during this period of rapid social change will make an important contribution to the literature, and provide important insights into the development of theoretical explanations.

Nevertheless, due to limited access to crime data, most of the crime studies in China have long focused on introducing western theories, policy issues, or simple narrative of current crime situations. Substantial empirical investigations are for the most part lacking. Quantitative and interdisciplinary research, in particular, is rare in China. Some encouraging changes, however, have emerged in recent years. As public concerns over urban crimes and safety grow, more local law enforcement agencies have started collaborating with academic organizations in conducting empirical crime studies in order to develop more effective crime prevention and crime reduction policing strategies.

In this study, we chose Changchun, the capital city of Jilin province in northeast China, as the study area to explore spatial patterns of the city’s violent crimes. Crime data were collected at the level of police precincts in 2008 from the Public Security Bureau of Changchun. We chose violent crimes to investigate because violent crimes seriously endanger urban residents, cause greater levels of fear among the public, and significantly lower the sense of safety of the city, thus calling for more careful and thorough investigation. Specifically, we undertook the following research tasks and reported corresponding analytical results. First, we examined the geographical distribution of violent crime rates in Changchun and detected the violent crime hot spots using spatial statistics and geographic information system (GIS). Second, a linear regression model was developed to analyze the relationships between spatial patterns of violent crime rates and selected locational and neighborhood socioeconomic and demographic characteristics. Our primary goal is to explore both the spatial pattern of violent crime risks and the underlying structural factors to gain a better understanding of the dynamics of violent crimes in the city of Changchun.

Literature review

As recognized earlier by western researchers, crime has an inherent geographical quality. When a crime occurs, it happens at a place with a geographical location (Chainey & Ratcliffe, 2005). Crime also does not spread randomly but tends to concentrate at particular places, which has been empirically supported by adequate research (Brantingham & Brantingham, 1984; Cohen & Felson, 1979; Chainey, Tompson, & Uhlig, 2008; Johnson, 2010; Sherman, Gartin, & Buerger, 1989; Weisburd, Bushway, Lum, & Yang, 2004). The concentrations or clusters of criminal activities are commonly referred to as crime “hot spots” that usually draw great interests from the law enforcement agencies because detecting the hotspots can help them to allocate the scarce policing resources to where they are most needed (Sherman & Weisburd, 1995). Hence, crime mapping is always the first step and a key tool for crime analysts to see or visualize differences and similarities across time and space (Brantingham & Brantingham, 1997; Chainey & Ratcliffe, 2005). Although crime mapping has long been an integral part of the process known today as crime analysis (Harries, 1999), the real prevalence of crime mapping began with the application of the GIS together with the availability of plentiful geo-referenced data. Since the mid-1990s, the use of GIS in law enforcement agencies has grown tremendously, and GIS has become their key technology because it enables the visualization of crime patterns and trends, as well as the identification of hot spots of crimes (Groff and La Vigne, 2001).

Numerous studies have documented the existence of hotspots of criminal activities (Bowers, Johnson, & Pease, 2004; Brantingham & Brantingham, 1995; Chainey, Tompson, & Uhlig, 2008; Grubesic, 2006; Ratcliffe, 2004; Sherman, Gartin, & Buerger, 1989; Tompson & Townsely, 2010; Weisburd, Morris, & Groff, 2009). Although the definition of crime hot spot varies slightly across different researches, it is generally accepted that hotspots are geographic locations of high crime concentration, relative to the distribution of crimes across the whole region of interest (Block & Block, 1995; Chainey & Ratcliffe, 2005; Chainey, Tompson, & Uhlig, 2008; Levine, 2006). Many methods and techniques have been proposed in mapping hotspots of crime, such as point mapping, spatial ellipses, thematic mapping of crimes at administrative units, and kernel density. Application of these methods could generate different maps of hotspots depending upon dissimilar data types, varied analytical scales, and distinct research purposes (Chainey, Tompson, & Uhlig, 2008; Eck, Chainey, Cameron, & Wilson, 2005; Harries, 1999). Nevertheless, none of these mapping techniques can define with any statistical significance those areas that are suspected of being hotspots. Combining cartographic visualization of crime events with spatial statistical tools, such as local spatial statistics, proves to be more effective in detecting areas of hot spots (Ratcliffe & McCullagh, 1999). These local indicators of spatial association (LISA), including local Moran’s I and the Getis-Ord’s Gi* statistics, offer statistical robustness to support the detection areas that can be defined as hotspots (Chainey & Ratcliffe, 2005). The LISA statistics assess the local association between data by comparing local averages to global averages (Anselin, 1995) and are able to outline “an area that has a greater than average number of criminal or disorder events, or an area where people have a higher than average risk of victimization.” They help define crime hot spots and place a spatial limit on those areas of highest crime event concentration (Harries, 1999). To date, more researchers have attempted to use the LISA statistics to detect the concentration or clusters of crimes or other spatial phenomena and have produced convincing results (Chainey & Ratcliffe, 2005; Craglia, Haining, & Wiles, 2000; Mencken & Barnett, 1999; Messner et al., 1999; Murray, McGuffog, & Western, 2001; Song & Keeling, 2010; Wang, 2006). With the rapid development of GIS technology, statistically detecting hot sports using LISA and visualizing results in the GIS environment prove promising in conducting spatial analysis of urban crimes.

Crime mapping is only the first step in examining the spatial and temporal occurrence of crimes. Understanding the factors underlying the concentrations or clusters of crimes is a logical extension of crime mapping, and it will help develop and implement appropriate measures to prevent and control urban crimes. Researchers have long been working on the explanations of why some places are prone to some types of crimes. The attempts to relate the characteristics of places to the criminal activities are called ecological or environmental approach in crime studies, which can be dated back to the work of Quetelet and Guerry in the 19th century and the studies of Shaw and McKay in the early 20th century. Based on the ideas from Chicago school, Shaw and McKay argued that areas characterized by such factors as low socioeconomic status, ethnic heterogeneity, and residential mobility tend to have higher rates of victimization of residents than other areas. They believed these factors cause the disorganization that leads to the weakening of social control and the development of delinquent subcultures, and prevents residents from coming together to solve neighborhood crime problems (Shaw & McKay, 1942). Shaw and McKay’s research has profoundly shifted the attentions of crime research by focusing on the characteristics of places instead of the characteristics of the offenders. Social disorganization theory developed from this early research on the geography of crimes has become one of the most influential theories in crime analysis. Since the 1980s, a large number of researchers have conducted theoretical and empirical studies at various levels, and offered many convincing results in support of this theory. Their work greatly enriched the substance of this leading theory of criminology (Barnett & Mencken, 2002; Bellair & Browning, 2010; Bursik, 1988; Bursik & Grasmick, 1993; Cahill & Mulligan, 2003; Cantillona, Davidsonb, & Schweitzer, 2003; Kingston, Huizinga, & Elliott, 2009; Miethe, Hughes, & McDowall, 1991; Sampson & Groves, 1989).

Another classic theory which exerts profound influence on criminology is the routine activities theory, founded by Cohen and Felson in 1979. In their work, they proposed a “routine activities approach” in explaining the trend of crime rates in the United States from 1947 to 1974. They hypothesized that the dispersion of activities away from households and families increases the opportunity for crime and thus generates higher crime rates. Further, the routine activities theory provides a general model which focused on analyzing the likelihood of crime occurrence. Based on routine activities approach, the presence of opportunities for crime in an area is shaped by residents’ daily activities and determined by the convergence in space and time of three key elements: likely offenders, suitable targets, and the absence of capable guardians against crime (Cohen & Felson, 1979). Although the place where the opportunities for criminal activities are created by the convergence of the three elements had been given enough attention in the original work, it was Sherman, Gartin, and Buerger (1989) who definitely argued that places can been seen to have the routine activities and seemed to be the most appropriate unit of analysis for routine activities. Routine activities theory stimulated numerous empirical investigations applying the theory to explain the variation of victimization at different levels. Research findings from these studies have extended and will continue to enhance the essences of this prominent criminology theory (Bennett, 1991; Bernburg & Thorlindsson, 2001; Massey, Krohn, & Bonati, 1989; Maxfield, 1987; Messner & Blau, 1987; Miethe, Stafford, & Long, 1987; Osgood, Wilson, O’Malley, Bachman, & Johnston, 1996; Sherman, Gartin, & Buerger, 1989; Tseloni, Wittebrood, Farrell, & Pease, 2004).

Despite the social disorganization theory historically has a more macro-orientation, while the routine activities theory focuses more on the micro-level, both theories in fact are closely related and complementary to each other despite some distinctions (Rice & Smith, 2002). Given the spatial nature of the two theories, researchers have made constant efforts to incorporate both theoretical approaches in empirical studies, demonstrating the utility of integrating sets of characteristics from both theories in explaining crime events (Andresen, 2006a, 2006b; Matthews, Yang, Hayslett, & Ruback, 2010; Miethe & McDowall, 1993; Rountree, Land, & Miethe, 1994; Smith, Frazee, & Davison, 2000; Zhang & Song, 2014; Zhang, Messner, & Liu, 2007).

Apparently, both social disorganization and routine activities theories emphasize the essential role of the environment in shaping the spatial patterns of criminal activities. Actually, the role of environment in reducing the opportunities for crime is also recognized by other theoretical and practical orientations, such as the defensible space theory and crime prevention through environmental design (CPTED). The concept of defensible space was developed by Newman (1972) through a detailed analysis of the design features and crime statistics of New York City’s public-housing projects in the1970s (Jacobs & Lees, 2013). Defensible space theory holds that certain features of physical settings, such as indicators of territory and surveillance opportunities, can reduce crime. Newman proposed that environmental designers can strategically use these features to defend an area against crime (Ham-Rowbottom, Gifford, & Shaw, 1999). In spite of some criticism to the Newman’s early work (Brantingham & Brantingham, 1993), it has become one of the core foundations for the program known as “Crime Prevention through Environment Design (CPTED),” which was originally coined by Jeffery (1971) and focuses specifically on planning, design, and construction of the built environment to hinder crime. The general principles or strategies emphasized by CPTED usually include natural surveillance, natural access control, territorial reinforcement, activity support, maintenance, and management (Cozens, 2002; Crowe, 2000; Glasson & Cozens, 2011; Zahm, 2007). The program or the strategies of CPTED have been widely turned into operation in many places in different countries, most of which are proved to be successful (Crowe, 2000; Cozens, Neale, & Whitaker, 2004; Glasson & Cozens, 2011). Surely, deep understanding of the ideas of defensible space theory as well as the CPTED will provide more insights into why crimes happen somewhere and how to reduce them.

Although still quite limited, some spatially referenced crime data have started becoming available to Chinese researchers, and some research results not only demonstrate the applicability of elements of western criminological theory to contemporary urban China but also reveal important differences in the dynamics of urban crimes in Chinese context (Messner, Lu, Zhang, & Liu, 2007; Song & Liu, 2013; Zhang, Messner, & Liu, 2007). Despite these positive developments, China’s geography of crime is still in its infant stage. More micro-level data, such as geo-referenced information of individual crime incidents, are still badly needed for exploring the spatial pattern of urban crimes using advanced analytical methods at a finer spatial level.

Study area, data, and methodology

Because of the differences between urban areas and rural areas in land uses, population density, socioeconomic activities, and crime patterns, we chose the inner-city of Changchun as the study area in order to capture typical urban crime patterns of the city. According to the survey conducted by the Changchun Institute of City Planning and Design in 2008, there were approximately 2.57 million inhabitants in this inner-city area of Changchun. Violent crime data for 2008 were provided by the Public Security Bureau of Changchun aggregated to 74 police precincts within the study area and the number of cases of violent crimes is available for each police precinct. Violent crimes include such criminal offenses as arson, homicide, mayhem, rape, kidnap, and robbery. Since detailed crime incident-level data with address information were not available, the 74 police precincts were used as the basic units for spatial statistical analysis of Changchun’s violent crimes. The boundaries of the police precincts were managed in the GIS software of ArcGIS 9.3.1 where the crime data and other information were organized as the attributes of each police precinct.

In this study, the violent crime rates were expressed as the number of violent crime cases per 10,000 residents and calculated by dividing the number of violent crime cases of each police precinct by its 2008 population. It is generally acknowledged that crime rate is a better index to specify the relative risk of crime in an area (Brantingham & Brantingham, 1997). We recognize the fact that when the crime rate is calculated, the most appropriate denominator should be the population on street (pedestrians) or ambient population which better measures the population at risk of violent crimes (Andresen & Jenion, 2010; Chainey & Desyllas, 2008). Unfortunately, such spatially referenced data can hardly be obtained in a Chinese city like Changchun, so we had to use the residential population instead.

Choropleth maps were made to reveal the spatial variation in the distribution of violent crime rates across Changchun. Despite some inherent drawbacks, such as the modifiable areal unit problem (Openshaw, 1984), the choropleth mapping is still one of the most meaningful and commonly used techniques in spatial analysis of crimes, especially when the access to micro-data is limited (Chainey, Tompson, & Uhlig, 2008; Harries, 1999). Beyond visualization of violent crime distribution, we employed local Moran’s I to detect the high-crime-risk hotspots in Changchun. As one of the most common LISA, local Moran’s I helps identify statistically significant local clustering of specific phenomena. It has been widely applied to criminology, epidemiology, and many other fields of research. In this study, local Moran’s I efficiently measures geographical concentration of violent crime clusters by detecting spatial autocorrelation among police precincts in Changchun. Local Moran’s I statistic was derived for each precinct and defined as

I_{i} = \frac{x_{i} - \bar{X}}{S_{i}^{2}} \sum_{j = 1, j \neq i}^{n} ω_{i, j} (x_{j} - \bar{X})

(1)

where x_i is the violent crime rate of precinct i,

\bar{X}

is mean of the violent crime rates,

ω_{i, j}

is the spatial weight between precincts i and j, and

S_{i}^{2} = \frac{\sum_{j = 1, j \neq i}^{n} (x_{j} - \bar{X})}{n - 1} - {\bar{X}}^{2}

(2)

with n equals to the total number of police precincts in the study area (Anselin, 1995). The significance of the local Moran’s I statistic is often tested by a normal Z index (Anselin, 1995; Wang, 2006). When the local Moran’s I statistics are positive and significant, we can thus label the corresponding areas as hotspots (clustering of high values) or “coldspots” (clustering of low values). This study focused on the hotspots that refer to the precincts with higher crime rates surrounded by neighbors whose crime rates are also above the average. We applied Cluster and Outlier Analysis – Anselin Local Moran's I tool in ArcGIS 9.3, which is built upon the above equations, to compute local Moran’s I statistics for all the police precincts. The spatial weights matrix we used to define the spatial relationships is calculated based on the inverse distance.

To interpret spatial patterns of violent crime rates, multivariate linear regression analysis was conducted to help uncover influencing factors underlying spatial variations in violent crimes across Changchun. Socioeconomic and demographic data are only available at the sub-district or street (Jiedao) level. Jiedao, also known JiedaoBanshichu, is the grassroots, township level administrative division in urban China (Zhang, Messner, & Liu, 2007). It is the lowest administrative unit for which socioeconomic and demographic data are collected. In Changchun, one Jiedao is generally composed of one or two police precincts and their boundaries match each other. Violent crime counts from police precincts were further aggregated into 45 Jiedaos so that violent crime rates, as a dependent variable, could be regressed against a set of explanatory or predictor variables which are only available at Jiedao level.

Results

Spatial distribution and hot spots of violent crime risks

The descriptive statistics of the violent crime data in 2008 are shown in Table 1, from which we can see that there is a great variation in the number of violent crime cases across the police precincts in Changchun, with a maximum of 81 and a minimum of only four. The residential population also differs dramatically across police precincts. The most populous precinct had a residential population of 114,000, while the smallest precinct had just 7700 people in 2008. Violent crime rates, computed from these two components, also vary considerably across Changchun’s police precincts (Table 1). The concentrating trend of the criminal activities is apparent in Changchun. One-third of the precincts took up nearly 60% of the total violent crimes in 2008, which is largely consistent with the findings of western researches (Sherman, Gartin, & Buerger, 1989; Weisburd, Bushway, Lum, & Yang, 2004).

Table 1.

Descriptive statistics of violent crimes in Changchun, 2008.

	Max	Min	Average	Standard deviation
Violent crime counts	81	4	22.7	15.2
Violent crime rates (per 10,000 residents)	23.2	1.4	7.4	5.0

Figure 1 depicts the distribution of violent crime rates, which better captures the relative risk of being victimized by violent crimes across the police precincts in the inner-city of Changchun. Overall, it appears that violent crime activities concentrate in the central city area, and tend to decline with increasing distance outwards from the city core.

Figure 1.

Violent crime rates in Changchun, 2008.

The global Moran’s I value for the distribution of violent crime rates is 0.077 with a standard normal z-value of 6.099, which indicates that the violent crime risks spread non-randomly in the city and reveal a positive spatial autocorrelation. We could expect to see clusters of police precincts with similar higher violent crime rates, or the so-called hotspots. They would be the police precincts that have higher violent crime rates surrounded by precincts with similar high rates. In order to detect these local clusters, the Z score of each police precinct’s local Moran’s I was computed and reported in Figure 2. Since we focus on the hotspots of violent crime rates in Changchun, only the precincts with a Z score more than 1.96 (more than 95% significance) are highlighted. As shown in Figure 2, 14 police precincts are identified as hotspots of violent crime rates in 2008. These precincts contain 14.54% of the total population, but account for more than 35% of the total violent crime cases of the city in 2008. The average violent crime rate of the hot spots is 15.69, nearly 2.11 times of the average city-wide violent crime rate, thus demonstrating an obvious pattern of spatial clustering and an enormously higher violent crime risk in the hot spots.

Figure 2.

Hotspots of violent crime rates in Changchun, 2008 using Local Moran’s I statistics.

The violent crime hot spots contain the central business district (CBD) (downtown) and four main commercial areas in Changchun which all suffered higher violent crime risks. This result largely echoes theories and empirical findings of the ecological criminology in western countries which argue that those areas adjacent to the CBD had more crimes than other areas (Ackerman & Murray, 2004; Andreson, 2006b; Cahill & Mulligan, 2007; Shaw & McKay, 1942). Usually, CBD and commercial areas in Chinese cities are places where stores, restaurants, entertainment facilities, companies, etc. concentrate, with large numbers of people who are actually not living there but working or shopping there as their routine activities (Sherman, Gartin, & Buerger, 1989). The concentration and mobility of population tend to increase the number of profitable targets and likely offenders in those places, at the same time lessen the chance to be caught by policemen. It is reasonable to expect that places where a lot of people gather for routine activities, such as working and shopping, are more likely to have more criminal activities including violent crimes. In the commercial areas, people are usually strangers to each other, thus lowering the social control that is more effective among the acquaintances. Especially at night, people walking in or near the commercial areas are extremely prone to robberies which account for a high proportion of violent crimes, due to the lack of guardians after most of the commercial businesses are closed. These observations are basically in line with the routine activities or opportunity approach which emphasizes the convergence of motivated offenders, suitable targets, and the absence of capable guardians in interpreting local crime events. From a spatial perspective, the adjacency of police precincts in crime hot spots is also probably due to the fact that the boundaries of the administrative units (e.g., police precincts) are delimited artificially, instead of functionally. A continuous crime space could cover and cross several administrative units. Neighboring areas could show similar crime circumstance because the determinants of crime tend to be alike among them. Spatial spillover effects of criminal incidents penetrating neighborhood boundaries could also be responsible for the clustering of similar high crime rates in the hot spots (Zhang & Song, 2014). Spatial interactions among adjacent neighborhoods imply that the risk of crime in a given neighborhood is determined not only by its own frequency of crime but is also affected by values of crime rates in neighboring areas. As a result, areas of higher crime rates tend to cluster together, forming apparent crime hot spots in the city.

Regression analysis

A multivariate regression model was developed to explore the relationship between violent crime rates and contextual variables capturing socioeconomic, demographic, land use characteristics of Jiedaos in Changchun. In China, the data released by municipal governments are usually insufficient for conducting statistical analysis at a finer spatial level. It is impossible to acquire the same set of variables suggested and employed by Western research. This study has strived to find variables that are closest proxies within the general framework of social control-disorganization theory and routine activities theory, and collect data from various sources. The independent variables we chose in our study include the proportion of better educated workers, the proportion of commercial land, the density of main roads, the number of high-rise residential buildings, and the presence or absence of vast industrial area in a Jiedao.

Employment data were collected from the municipal government’s Economic Census Office. The data contain the total number of people who are working in each Jiedao and their education levels. A variable labeled “PRO_BEW” was constructed to represent the proportion of the better educated workers who have a bachelor’s degree or higher. This variable denotes the average education level of Jiedao’s people. It was chosen as a proxy and indirectly a measure of social control of neighborhoods in the context of social disorganization theory. High level of education is often considered a key contextual characteristic of neighborhoods where crimes are less likely to occur due to the collective efficacy against crimes fostered among educated citizens. Further, people or companies in areas with high percent of better-paid jobs usually can afford more security measures such as locks, alarm systems, CCTV, and security guards. These measures can effectively improve the guardianship, surveillance opportunities, and territorial reinforcement in the areas and thus help curtail crimes. This is apparently tied to the routine activities theory or the framework of CPTED.

As revealed earlier, land use features of an area seem to influence the concentration of crimes. Researches in criminology have long taken land use characteristic into consideration when accounting for urban crime patterns (Ackerman & Murray, 2004; Browning et al., 2010; Cahill & Mulligan, 2007; Lockwood, 2007; Stucky & Ottensmann, 2009). Data reflecting land use characteristics were compiled from 2008 land-use map made by the Changchun Institute of City Planning and Designing. Two variables at Jiedao level were extracted from this data source, including the proportion of a Jeidao’s commercial land (PRO_COM) and the density of main roads in each Jiedao (DEN_R). PRO_COM was computed by dividing the area of commercial land in each Jiedao by its total land area, while DEN_R was calculated by dividing the length of main roads in each Jiedao by its total land area. We included them in the model to measure indirectly the availability of targets and offenders, as well as the weakening of local residents’ sense of territory. Both variables are postulated to influence, and more specifically, increase the number of motivated offenders and the risk of individual victimization in an area. As commercial areas usually suffered more crimes than other areas (Ackerman & Murray, 2004; Lockwood, 2007), the proportion of commercial land in a neighborhood should be positively associated with the violent crime rates. Similarly, since the distribution of crimes is often believed to be related to the transportation network (Zahm, 2007), it is reasonable to consider that denser road network may generate more crime risks because more nodes, paths, and edges means more opportunities for criminal activities and meanwhile weakening the territory sense of the residents (Brantingham & Brantingham, 1993; Chainey & Ratcliffe, 2005). These largely coincide with both the routine activities theory and the defensible space theory.

Concentrated disadvantage or affluence can influence a neighborhood’s collective social control, thus affecting the local crime risk. The economic segregation between both poor and affluent families in urban China has led to the emergence of concentrated affluence and disadvantage, reflected increasingly by the striking contrast in housing conditions. We attempted to obtain a variable to reflect this important dimension of Changchun’s Jiedaos. Unfortunately, such variables as household income, poverty, dwelling value, and unemployment status were unattainable. Only limited housing-related data were available from Changchun Institute of City Planning and Designing’s 2008 survey and map of housing conditions of urban residents. A variable labeled COUNTS_H was derived to denote the number of high-rise residential buildings (higher than 10 storeys) in each Jiedao. It was used as a proxy of a Jiedao’s concentrated advantage or affluence. Generally, Jiedaos, where these high-rise buildings concentrate, are typically new compared with neighborhood with less presence of such new structures. They tend to correlate to where white-collar workers and/or middle-class households reside, which is a special and prevalent phenomenon in Chinese cities (Zhang, Liu, & Li, 2003; Zheng, Fu, & Liu, 2005). So this variable, as a proxy of concentrated affluence, is postulated to be negatively associated with the violent crime rates.

Inspired by other studies (Ackerman & Murray, 2004; Browning et al., 2010; Cahill & Mulligan, 2007; Lockwood, 2007; Stucky & Ottensmann, 2009), we also defined one binary variable IND_Z to indicate whether there is a vast and contiguous industrial area within a Jiedao (1 means yes, 0 means no). As a major city and an industrial base in northeastern China, Changchun has many large industrial areas usually located in the periphery of the city. This unique type of urban land use and the corresponding characteristics in the distribution of population and other economic activities within and near these industrial areas are expected to have an effect on the violent crime risk.

Table 2 contains the descriptive statistics for the explanatory or independent variables in the multivariate regression model. It is clear that these variables indeed vary quite significantly across Changchun’s Jiedaos. The results of the ordinary least squares (OLS) regression analysis are shown in Table 3, including both un-standardized and standardized parameter estimates and collinearity statistics.

Table 2.

Descriptive statistics of variables for regression analysis.

Name of variables	Definition of variables	Mean	Max	Min	Standard deviation
PRO_BEW	Proportion of better educated workers (%)	20.71	47.05	2.35	13.29
PRO_COM	Proportion of commercial land (%)	3.82	20.02	0.16	4.55
IND_Z	Presence or absence of vast industrial area (yes/no)	0.40	1	0	0.49
DEN_R	Density of main roads (m/ha)	29.88	97.18	1.15	26.21
COUNTS_H	Numbers of high-rise residential buildings (number)	13.27	110	0	21.92

Table 3.

Final results of ordinary least squares regression model.

Independent variables	Un-standardized coefficients	Standardized coefficients	t	Sig.	VIF
(constant)	8.532		6.706	0.000
PRO_BEW (%)	−9.442	−0.281	−3.126	0.003	1.393
PRO_COM (%)	38.433	0.392	4.113	0.000	1.565
IND_Z (yes/no)	−3.433	−0.381	−4.145	0.000	1.456
DEN_R (m/ha)	0.038	0.222	2.249	0.030	1.681
COUNTS_H (number)	−0.045	−0.222	−2.780	0.008	1.096
Moran's I (error)	−0.185			0.853
LM lag	0.345			0.557
LM error	0.432			0.511
Model: R²= 0.774; Adj.R²= 0.745; F = 26.65; Sig. = 0.000

VIF: variance inflation factors; PRO_BEW: proportion of better educated workers; PRO_COM: proportion of commercial land; IND_Z: presence or absence of vast industrial area; DEN_R: density of main roads; COUNTS_H: numbers of high-rise residential buildings.

When conducting regression analysis, we addressed the issue of and evaluated the effect of spatial dependence or spatial autocorrelation explicitly employing the leading and widely used spatial statistic software of OpenGeoDa 1.4.6. First-order queen’s contiguity was assumed to specify the spatial weights matrix when testing against spatial autocorrelation for violent crime rates, including Moran’s I, as well as Lagrange multiplier test statistics (LM-Lag and LM-Error). As shown in Table 3, all the statistics of the diagnostic tests are not significant, which indicates that no substantial spatial dependence exists in the model. Under this circumstance, we decided to stick to and proceed with the OLS regression analysis as suggested by Anselin (2005). It needs to be noted that the basic spatial unit for regression analysis is Jiedao which is more spatially extensive than the police precinct. Changing the spatial unit of analysis from the police precinct to Jiedao is essentially a spatial re-sampling process from a fine scheme to a less fine scheme. The smaller number of Jiedaos and their broader geographical extent may very likely have helped remove or lessen the effect of spatial autocorrelation and spatial spillover, thus making the overall spatial autocorrelation (Moran’s I) as well as spatial lag or spatial error statistics less significant. From a more practical perspective, most of the Jiedaos in Changchun are defined by geographic features (such as main roads, railroads, rivers, freeways, etc.). Some of these features may very likely hinder the interaction among neighboring Jiedaos, thus making the phenomenon of spatial dependence less predominant. Therefore, from both the methodological and practical standpoints, our OLS regression model offers a meaningful explanation for the variation of violent crime rates in Changchun, without suffering from a spatial dependence problem.

Overall, the model explains over 70% of the variance of violent crime rates, and all the coefficients are significant and in expected direction. By and large, the results support the notion that violent crime rates or risks are a function of the characteristics of people in an area as well as the environment and characteristics of the area where people reside or work. The variance inflation factors (VIFs) of all the independent variables are less than 2, indicating that collinearity among the variables is low. This model does provide helpful insights in explaining the spatial pattern of violent crimes in Changchun in 2008.

The parameters for the proportion of commercial land and road density are significantly positive, supporting both variables as measures of increased potential targets and offenders in line with the interpretation of routine activities theory. Commercial use areas are usually open and easily accessible to the public, and tend to draw many more people, especially, visitors or shoppers, into the area. This convergence of people in time and space could increase the number of suitable targets and potential offenders. Since most of the people in a commercial use area are simply transient, crime control, and prevention become more difficult. There are many potential blind spots without effective and capable guardians. Furthermore, many banks, supermarkets, and other businesses exist in the commercial area. The availability of money with people and/or businesses could mean more suitable targets for such violent crimes as robbery. Actually, the guardianship in many of the commercial areas is enhanced during daytime through increasing number of security guards and patrolling policemen. However, the overwhelming concentration and mobility of people, consequently potential targets and offenders, may very well outweigh the effect of improved guardianship in many commercial areas. Compared with the tremendous level of pedestrian and traffic flows, security guardianship still appears inadequate and ineffective. After dark, many of the commercial areas tend to be poorly attended and guarded, making them less safe for those people moving across.

The positive correlation between violent crime rates and road density can be similarly interpreted largely from the perspective of the routine activities theory. Although denser road network may mean increased natural surveillance which is emphasized by the CPTED, high road density can mean that more vehicle traffic and pedestrian traffic are on the street in an area, thus increasing the number of potential offenders and targets, and consequently the opportunities for crimes. High road density can also imply that the area is more accessible and there are more street corners and intersections where violent crimes, such as robbery and assaults, are more likely to be committed by motivated offenders. The existence of numerous roads and streets could offer offenders more escaping routes, and makes effective policing and crime control more difficult to achieve.

The significantly negative correlation between the proportion of workers with higher education level and violent crime rates can be explained primarily within the framework of social disorganization theory. Various studies have revealed that well-educated persons have significantly lower levels of mistrust and are more sensitive and responsive to crimes (Ross & Jang, 2000). Low level of education, on the contrary, is considered a key contextual characteristic of neighborhoods where crimes are more likely to occur. Similar situations have also been found in contemporary urban China (Liu, Messner, Zhang, & Zhuo, 2009). In urban China, areas where a lot of highly educated people work generally contain government agencies, education and research institutions, and joint-venture high-technology industries and corporations. These areas tend to be well designed, planned, and have a favorable physical and social environment. Various modern facilities and infrastructures are more complete. These favorable conditions lead to the clustering of a highly similar, well-educated and well-paid population. Social cohesion and informal social control, such as local organizations, voluntary associations and neighborhood watch, as well as public control in the form of an effective police presence are more easy to foster and function in these Jiedaos than their less-educated counterparts, thus exerting the expected negative effects on neighborhood crimes. Hence, an increase in the concentration of highly educated people exhibits a considerable impact on the decrease in the risk of violent crimes in Changchun.

As far as the number of high-rise buildings in each city-street is concerned, it shows a significant negative relationship with the violent crime rates in the regression model as predicted previously, which indicates that the newly developed neighborhoods in Changchun tends to suffer lower violent crime risks. Firstly, this may be related to the fact that residents living in the areas with more high-rise buildings usually have better jobs and are white-collar workers and/or belong to middle-class households. They are less likely to commit violent crimes by themselves. Secondly, most of the new neighborhoods are gated or semi-closed, so it is not easy for the offenders to enter the communities and find suitable targets in the new neighborhoods. In contrast, the old neighborhoods with fewer high-rise buildings are likely to experience more violent crimes mainly due to more opportunities to commit crimes. Closed or semi-closed gated communities which have become increasingly prevalent in urban China are also likely to be found in these new neighborhoods, partly inspired by the notion of CPTED. In general, these newly developed and relatively advantaged neighborhoods tend to have high environmental quality, adequate infrastructure, and more safeguard. They are also less likely to be intermingled with a lot of street business and free markets, helping curtail the violent crime risk. These findings suggest that the types of neighborhood control and environmental design elements that have often been stressed in Western literature are also relevant in today’s Chinese cities.

As expected, violent crime rates are negatively correlated to the existence of vast industrial area in Changchun. The industrial areas are nearly all located in the periphery of the city and their functionality tends to be more monotonous with lower population density and less degree of mixture with other activities, particularly commercial activities which are strongly clustered in the central city. This could help bring down the number of potential offenders and targets, and subsequently the opportunities for crimes. Many of Changchun’s industrial areas have existed for decades, as a result of which people working there are likely to know each other and tend to develop a mutual trust and a relatively solid social coherence and homogeneity, thus having the effect of lessening the risk of violent crimes. It seems that violent crime risks are reduced in industrial areas as compared with the CBD or downtown, owing to the routine activities and perhaps enhanced social organization and social control in these areas.

Conclusion

This research presents an exploratory analysis of the spatial pattern in the relative risk of violent crime in Changchun, as well as its relationships with contextual socioeconomic characteristics of the city’s neighborhoods or Jiedaos. Due to the lack of detailed and desired social, economic, and demographic data for small spatial units, several predictor variables employed in this analysis may not provide the most ideal measures of social controls, concentrated disadvantage, routine activities, environmental features, and heterogeneity. Nonetheless, this study has defined variables that are closest proxies within the general framework of social control-disorganization theory, routine activities theory and CPTED, and produced important research findings and valuable insights. Analytical results indicate that factors within these leading theoretical approaches and practical concepts are indeed important in understanding the risk of violent crime in Changchun.

In Changchun, as in many Western cities, the risk of violent crime varies noticeably across the city’s neighborhoods. Violent crime is indeed a local occurrence and local contexts can make a real difference in the crime risk throughout Changchun. Violent crime risk tends to be significantly higher in the central city area, as compared with the city periphery. The clustering of vibrant economic activities and population, particularly around commercial districts and with easy road/street access, makes suitable targets, motivated offenders and the lack of capable guardians readily converge in space, leading to a higher risk of violent crime primarily in the CBD and its surrounding central city area.

Our analyses also reveal that spatial variations in the risk of violent crime are socially and economically structured. Population characteristics related to social disorganization and environmental design features are essential to understand the risk of violent crimes in local setting. Enhanced social control and social organization, associated with the concentration of educated people and better residential conditions, as well as favorable environment and infrastructure, help lower the risk of violence in those well planned, designed, and safeguarded urban areas.

This study also supports the notion that selected place characteristics are related to crimes. Unique kind of land use, such as large industrial land, in Changchun’s city fringe seems to influence the opportunity for violent crimes through routine activities and the relationship and interaction among people working and living in those specific areas.

Chinese urban space is undergoing a tremendous transformation, becoming more diverse and differentiated. Urban space is now far more diverse and differentiated: neighborhoods vary along social and economic dimensions, residential mobility is growing, and the urban population is more heterogeneous. All of these developments suggest increasing risk of crimes and changing crime patterns in Chinese cities. One cannot understand the development of crime and delinquency in Chinese cities without theoretical and empirical research that examines the underlying macro- and micro-level processes. The current empirical study contributes to our understanding of how some key Western theories can be used to explain urban crimes in a non-Western society, with very different traditions and institutions than those in the United States. It offers an important verification and replication of previous research based on key theoretical and practical approaches in Western criminology. As is typical in previous studies in the West, our testing of an integrated model with measures from various theories and approaches proved that the crime profile of an urban neighborhood is indeed influenced by sets of characteristics suggested by routine activity theory, social disorganization, and CPTED. These factors work together in shaping the spatial pattern of crimes in Chinese cities. The fact that we find support for these theories in a non-Western cultural setting provides further evidence of their generalizability.

There are several important issues to be addressed in future research to explore more particularity of China’s national conditions. First, incident-based crime data with detailed classifications and locational information are not accessible for this study. It makes it difficult to examine the spatial pattern of crime risks at a micro-level, such as determining the relationship between crime incidents and site-specific physical features, social and demographic characteristics, and land use. Second, adequate neighborhood or community-level measures, such as population heterogeneity, age structure, household income, dwelling values, housing tenure, are not available for the administrative areas used in this study. This hinders the thorough exploration and interpretation of spatial variation of crime rates. Third, urban communities are dynamic entities that affect crime rates in significant ways. Communities are shaped by changing social, economic and demographic mixture, as well as development process. This points toward importance in crime analysis with longitudinal data to further clarify the relationships between routine activities, social control, environmental design, and crimes.

Footnotes

Acknowledgements

The authors would like to express their thanks to the Public Security Bureau of Changchun, Economic Census Office of the municipal government, and Changchun Institute of City Planning and Designing for providing data used in our research.

Funding

This work was supported by the National Natural Science Foundation of China (No. 41301143).

References

Ackerman

V. W.

Murray

T. A.

(2004) Assessing spatial patterns of crime in Lima, Ohio. Cities 21(5): 423–437.

Andresen

M. A.

(2006a) Crime measures and the spatial analysis of criminal activity. British Journal of Criminology 46: 258–285.

Andresen

M. A.

(2006b) A spatial analysis of crime in Vancouver, British Columbia: A synthesis of social disorganization and routine activity theory. The Canadian Geographer 50(4): 487–502.

Andresen

M. A.

Jenion

G. W.

(2010) Ambient populations and the calculation of crime rates and risk. Security Journal 23: 114–133.

Anselin

(1995) Local indicators of spatial association: LISA. Geographical Analysis 27(2): 93–115.

Anselin

(2005) Exploring Spatial Data with GeoDaTM: A Workbook, IL, USA: Spatial Analysis Laboratory, Department of Geography, University of Illinois.

Barnett

Mencken

F. C.

(2002) Social disorganization theory and the contextual nature of crime in nonmetropolitan counties. Rural Sociology 67(3): 372–393.

Bellair

P. E.

Browning

C. R.

(2010) Contemporary disorganization research: An assessment and further test of the systemic model of neighborhood crime. Journal of Research in Crime and Delinquency 47(4): 496–521.

Bennett

R. R.

(1991) Routine activities: A cross-national assessment of a criminological perspective. Social Forces 70(1): 147–163.

10.

Bernburg

J. G.

Thorlindsson

(2001) Routine activities in social context: A closer look at the role of opportunity in deviant behavior. Justice Quarterly 18(3): 543–567.

11.

Block

R. L.

Block

C. R.

(1995) Space, place, and crime: Hot spot areas and hot places of liquor-related crime. In: Eck

Weisburd

(eds) Crime and Place, Monsey, NY: Willow Tree, pp. 145–183.

12.

Bowers

K. J.

Johnson

S. D.

Pease

(2004) Prospective hot-spotting: The future of crime mapping? British Journal of Criminology 44: 641–658.

13.

Brantingham

P. J.

Brantingham

P. L.

(1984) Patterns in Crime, New York, NY: Macmillan Press.

14.

Brantingham

P. L.

Brantingham

P. J.

(1993) Nodes, paths and edges: Considerations on the complexity of crime and the physical environment. Journal of Environmental Psychology 13: 3–28.

15.

Brantingham

P. L.

Brantingham

P. J.

(1995) Location quotients and crime hot spots in the city. In: Block

C. R.

Dabdoub

Fregly

(eds) Crime Analysis Through Computer Mapping, Washington, DC: Police Executive Research Forum, pp. 129–149.

16.

Brantingham

P. L.

Brantingham

P. J.

(1997) Mapping crime for analytic purposes: Location quotients, counts and rates. In: Weisburd

McEwen

(eds) Crime Mapping and Crime Prevention, Monsey, NY: Criminal Justice Press, pp. 263–288.

17.

Browning

C. R.

Byron

R. A.

Calder

C. A.

Krivo

L. J.

Kwan

Lee

Peterson

R. D.

(2010) Commercial density, residential concentration and crime: Land use patterns and violence in neighborhood context. Journal of Research in Crime and Delinquency 47(3): 329–357.

18.

Bursik

R. J.

(1988) Social disorganization and theories of crime and delinquency: Problems and prospects. Criminology 26: 519–551.

19.

Bursik

Grasmick

H. G.

(1993) Neighborhoods and Crime: The Dimensions of Effective Community Control, New York, NY: Lexington Books.

20.

Cahill

M. E.

Mulligan

G. F.

(2003) The determinants of crime in Tucson, Arizona. Urban Geography 24: 582–610.

21.

Cahill

M. E.

Mulligan

G. F.

(2007) Using geographically weighted regression to explore local crime patterns. Social Science Computer Review 25(2): 174–193.

22.

Cantillona

Davidsonb

S. W.

Schweitzer

H. J.

(2003) Measuring community social organization: Sense of community as a mediator in social disorganization theory. Journal of Criminal Justice 31: 321–339.

23.

Chainey

Desyllas

(2008) Modelling pedestrian movement to measure on-street crime risk. In: Liu

Eck

(eds) Artificial Crime Analysis Systems: Using Computer Simulations and Geographic Information Systems, Hershey, PA: Information science reference, pp. 71–91.

24.

Chainey

Ratcliffe

(2005) GIS and Crime Mapping, London, UK: John Wiley.

25.

Chainey

Tompson

Uhlig

(2008) The utility of hotspot mapping for predicting spatial patterns of crime. Security Journal 21: 4–28.

26.

Cohen

L. E.

Felson

(1979) Social change and crime rate trends: A routine activity approach. American Sociological Review 44(4): 599–608.

27.

Cozens

P. M.

(2002) Sustainable urban development and crime prevention through environmental design for the British city. Towards an effective urban environmentalism for the 21st century. Cities 19(2): 129–137.

28.

Cozens

P. M.

Neale

R. H.

Whitaker

(2004) Tackling crime and fear of crime while waiting at Britain's railway stations. Journal of Public Transportation 7(3): 23–42.

29.

Craglia

Haining

Wiles

(2000) A comparative evaluation of approaches to urban crime pattern analysis. Urban Studies 37(4): 711–729.

30.

Crowe

T. D.

(2000) Crime Prevention Through Environmental Design: Applications of Architectural Design and Space Management Concepts, (2nd ed.). Oxford, UK: Butterworth-Heinemann.

31.

Eck

Chainey

S. P.

Cameron

Wilson

(2005) Mapping Crime: Understanding Hotspots, Washington, DC: National Institute of Justice.

32.

Glasson

Cozens

(2011) Making communities safer from crime: An undervalued element in impact assessment. Environmental Impact Assessment Review 31: 25–35.

33.

Groff

La Vigne

N. G.

(2001) Mapping an opportunity surface of residential burglary. Journal of Research in Crime and Delinquency 38: 257–278.

34.

Grubesic

H. T.

(2006) On the application of Fuzzy for crime hot-spot detection. Quantitative Criminology 22(1): 77–105.

35.

Ham-Rowbottom

K. A.

Gifford

Shaw

K. T.

(1999) Defensible space theory and the police: Assessing the vulnerability of residences to burglary. Journal of Environment Psychology 19(2): 117–129.

36.

Harries

(1999) Mapping Crime: Principle and Practice, Washington, DC: US Department of Justice Office of Justice Programs.

37.

Jacobs

J. M.

Lees

(2013) Defensible space on the move: Revisiting the urban geography of Alice Coleman. International Journal of Urban and Regional Research 37(5): 1559–1583.

38.

Jeffery

C. R.

(1971) Crime Prevention Through Environmental Design, Beverly Hills, CA: Sage Publications.

39.

Johnson

S. D.

(2010) A brief history of the analysis of crime concentration. European Journal of Applied Mathematics 21: 349–370.

40.

Kingston

Huizinga

Elliott

D. S.

(2009) A test of social disorganization theory in high-risk urban neighborhoods. Youth & Society 41(1): 53–79.

41.

Levine

(2006) Crime mapping and the crimestat program. Geographical Analysis 38: 41–56.

42.

Liu

(2005) Crime patterns during the market transition in China. British Journal of Criminology 45(5): 613–633.

43.

Liu

Messner

Zhang

Zhuo

(2009) Socio-demographic correlates of fear of crime and the social context of contemporary urban China. American Journal of Community Psychology 44(1): 93–108.

44.

Lockwood

(2007) Mapping crime in savannah: Social disadvantage, land use, and violent crimes reported to the police. Social Science Computer Review 25(2): 194–209.

45.

Massey

J. L.

Krohn

M. D.

Bonati

L. M.

(1989) Property crime and the routine activities of individuals. Journal of Research in Crime and Delinquency 26(4): 378–400.

46.

Matthews

A. S.

Yang

Hayslett

L. K.

Ruback

R. B.

(2010) Built environment and violent crime in Seattle, 1998–2000: A Bayesian analysis. Environment and Planning A 42: 1403–1420.

47.

Maxfield

M. G.

(1987) Household composition, routine activity and victimization: A comparative analysis. Journal of Quantitative Criminology 3(4): 301–320.

48.

Mencken

F. C.

Barnett

(1999) Murder, nonnegligent manslaughter, and spatial autocorrelation in mid-south counties. Journal of Quantitative Criminology 15(4): 407–422.

49.

Messner

S. F.

Anselin

Baller

R. D.

Hawkins

D. F.

Deane

Tolnay

S. E.

(1999) The spatial patterning of county homicide rates: An application of exploratory spatial data analysis. Journal of Quantitative Criminology 15(4): 423–450.

50.

Messner

S. F.

Blau

J. R.

(1987) Routine leisure activities and rates of crime: A macro-level analysis. Social Forces 65(4): 1035–1052.

51.

Messner

S. F.

Zhang

Liu

(2007) Risks of criminal victimization in contemporary urban China: An application of lifestyle/routine activities theory. Justice Quarterly 24(3): 496–522.

52.

Miethe

T. D.

Hughes

McDowall

(1991) Social change and crime rates: An evaluation of alternative theoretical approaches. Social Forces 70(1): 165–185.

53.

Miethe

T. D.

McDowall

(1993) Contextual effects in models of criminal victimization. Social Forces 71(3): 741–759.

54.

Miethe

T. D.

Stafford

M. C.

Long

J. S.

(1987) Social differentiation in criminal victimization: A test of routine activities/lifestyle theories. American Sociological Review 52(2): 184–194.

55.

Murray

T. A.

McGuffog

Western

S. J.

(2001) Exploratory spatial data analysis techniques for examining urban crime. British Journal of Criminology 41(2): 209–329.

56.

Newman

(1972) Defensible Space: Crime Prevention Through Urban Design, New York, NY: Macmillan.

57.

Openshaw

(1984) The modifiable areal unit problem. Concepts and Techniques in Modern Geography 38: 41.

58.

Osgood

D. W.

Wilson

J. K.

O’Malley

P. M.

Bachman

J. G.

Johnston

L. D.

(1996) Routine activities and individual deviant behavior. American Sociological Review 61: 635–655.

59.

Ratcliffe

J. H.

(2004) The hotspot matrix: A framework for the spatio-temporal targeting of crime reduction. Police Practice and Research 5(1): 5–23.

60.

Ratcliffe

J. H.

McCullagh

M. J.

(1999) Hotbeds of crime and the search for spatial accuracy. Journal of Geographical Systems 1(4): 385–398.

61.

Rice

K. J.

Smith

W. R.

(2002) Scoioecological models of automotive theft: Integrating routine activity and social disorganization approaches. Journal of Research in Crime and Delinquency 39(3): 304–336.

62.

Ross

C. E.

Jang

S. J.

(2000) Neighborhood disorder, fear, and mistrust: The buffering role of social ties with neighbors. American Journal of Community Psychology 28(4): 401–420.

63.

Rountree

P. L.

Land

K. C.

Miethe

T. D.

(1994) Macro-micro integration in the study of victimization: A hierarchical logistic model analysis across Seattle neighborhoods. Criminology 32(3): 387–414.

64.

Sampson

Groves

(1989) Community structure and crime: Testing social disorganization theory. American Journal of Sociology 94(4): 774–802.

65.

Shaw

C. R.

McKay

H. D.

(1942) Juvenile Delinquency and Urban Areas: A Study of Rates of Delinquents in Relation to Differential Characteristics of Local Communities in American Cities, Chicago: University of Chicago Press.

66.

Sherman

L. W.

Gartin

P. R.

Buerger

M. E.

(1989) Hot spots of predatory crime: Routine activities and the criminology of place. Criminology 27: 27–55.

67.

Sherman

L. W.

Weisburd

(1995) General deterrent effects of police patrol in crime “hot spots”: A randomized, controlled trial. Justice Quarterly 12(4): 625–648.

68.

Smith

R. W.

Frazee

G. S.

Davison

L. E.

(2000) Furthering the integration of routine activity and social disorganization theories: Small units of analysis and the study of street violent crime as a diffusion process. Criminology 38(2): 489–523.

69.

Song

Keeling

(2010) Location patterns of section 8 housing in Jefferson county, Kentucky. International Journal of Applied Geospatial Research 1(2): 1–18.

70.

Song

Liu

(2013) Exploring spatial patterns of property crime risks in Changchun, China. International Journal of Applied Geospatial Research 4(3): 80–100.

71.

Stucky

T. D.

Ottensmann

J. R.

(2009) Land use and violent crime. Criminology 47(4): 1223–1264.

72.

Tompson

Townsely

(2010) (Looking) Back to the Future: Using space-time patterns to better predict the location of street crime. International Journal of Police Science and Management 12(1): 23–40.

73.

Tseloni

Wittebrood

Farrell

Pease

(2004) Burglary victimization in England and Wales, the United States and the Netherlands-A cross-national comparative test of routine activities and lifestyle theories. British Journal of Criminology 44(1): 66–91.

74.

Wang

(2006) Quantitative Methods and Applications in GIS, Baca Raton, FL: CRC/Taylor & Francis Press.

75.

Weisburd

Bushway

Lum

Yang

(2004) Trajectories of crime at places: A longitudinal study of street segments in the city of Seattle. Criminology 42(2): 283–321.

76.

Weisburd

Morris

N. A.

Groff

E. R.

(2009) Hot spots of juvenile crime: A longitudinal study of arrest incidents at street segments in Seattle, Washington. Journal of Quantitative Criminology 25(4): 443–467.

77.

Zahm

(2007) Using Crime Prevention Through Environmental Design in Problem-Solving, Washington, DC: US Department of justice and office of community oriented policing services.

78.

Zhang

Song

(2014) Addressing issues of spatial spillover effects and non-stationarity in analysis of residential burglary crime. GeoJournal 79: 89–102.

79.

Zhang

Messner

S. F.

Liu

(2007) A multilevel analysis of the risk of household burglary in the city of Tianjin China. British Journal of Criminology 47: 918–937.

80.

Zheng

Liu

(2005) An empirical study on the willingness to pay for residential location and location choice of urban residents in China [In Chinese]. Economic Geography 25(2): 194–198.

81.

Zhang

Liu

(2003) Housings’ spatial distribution and residents’ preference on housing location in Beijing [In Chinese]. Geographical Research 22(6): 751–759.