Introduction

With rapid urbanization and the growth of urban population in China, major cities face challenges of urban problems. Among them, crimes such as burglaries significantly threaten the well-being and property safety of urban residents. Burglaries refer to the illegal entry into buildings with the primary goal of stealing. Although much research has been done on burglaries (Wang et al., 2017, Xiao et al., 2017), few studies have examined the potential heterogeneity in the influencing factors of burglaries across different types of communities. This study divides communities into different types based on the composition of residents to explore how key influencing factors noted in the literature contribute differently to burglaries in space.

Common criminological theories used to understand burglaries include social disorganization theory, routine activity theory and rational choice theory (Liu and Xiu, 2012). Social disorganization theory emphasizes the influence of social control ability among neighbors on crime (Baumer et al., 2012, Sampson and Groves, 2017). Sampson and Groves (1989) employed the social disorganization theory to explore the relationship between community structure and crime, discovering that socioeconomic factors such as the level of poverty, unemployment rate, as well as social mobility within a community are closely associated with crime rates. Cancino (2003) found that burglaries were more likely to occur in communities with poor economic status and low social cohesion. Routine activity theory proposes that crime occurs when motivated perpetrators, suitable targets, and the lack of capable guardianship converge in space and time (Chon 2016, Cohen and Felson, 1979). Based on routine activity theory, Long et al. (2017) analyzed the impact of community environment on burglaries and found that burglaries often occur in areas with more potential crime targets (i.e., households) and a lack of social monitoring. Juveniles are more likely to be offenders in such communities. Rational choice theory emphasizes that perpetrators decide whether to commit crimes based on the evaluation of benefits, risks, and costs (Cornish and Clarke, 1987, Vandeviver and Bernasco, 2020). Xiao et al. (2017) found that burglaries were affected by the risk-benefit ratio presented by the housing type, and potential perpetrators followed the principle of “safety first, profit second”. “Safety” refers to the low risk of the perpetrator being arrested. “Profit” represents the potential benefit from the criminal activities. This principle suggests that criminals tend to favor crime activities with lower risks.

Housing, social environment, and burglaries

For burglars, houses serve as their targets. However, different types of houses will have different impacts on burglaries.

In terms of housing characteristics, housing in old regions of cities is often characterized by disrepair and overcrowding. Doors and windows of the houses are more accessible to potential perpetrators, and the anti-theft facilities are not adequate, which provides opportunities for burglars (Liu et al., 2014). The residents are primarily the local elderly, the young migrant workers, and middle-aged migrant workers. Frequent turnover of residents, numerous entrances to the community, and complex routes bring convenience for perpetrators to find targets and escape after crimes (Mao et al., 2019).

Regarding social environment, modern communities have shopping malls, supermarkets, and other spaces for economic activity. With dense crowds and strong social monitoring effects, rates of burglaries are low in these spaces, despite that the rate of thefts may be high (Wang et al., 2017). However, burglary cases in urban villages are often frequent. Perpetrators choose migrant workers and students who rent houses in urban villages as crime targets because of the lack of family support in these households and the lack of monitoring of criminal activities (Long et al., 2017). During the day, when they are at work or school, the lack of supervision makes their homes more vulnerable to burglaries.

Rental housing ratio and rental prices are also key factors affecting burglaries. Communities with higher rents tend to be more regulated. Cancino (2003) found that burglaries were more likely to occur in communities with single-family houses of low economic status, where the social cohesion was relatively low. Communities with higher rents often have better access control systems as well as surveillance cameras and the crime rate of such communities is significantly lower (Chen et al., 2017). Liu et al. (2020a) found that surveillance cameras significantly inhibited burglary cases. Some bungalows lack effective crime-prevention measures, and the community is exposed to a considerable degree of social disorder (Xu, Chen and Chen, 2018). Moreover, the higher the housing renting ratio, the greater the residential mobility is, which weakens the informal social control of the community. Warner (2014) used questionnaires and block-level census data to examine the influence of social factors on people’s positive reactions to informal social control. Results indicate that community mobility reduces the likelihood of people responding to informal social control, while community cohesion and trust in the police can enhance positive responses to informal social control.

Migrants, social disorganization, and burglaries

Migrants refer to the group of people relocating from their original place of residence to another location and establishing long-term or permanent residency there (Razum and Samkange-Zeeb, 2024). In contrast, local residents are individuals who were either born in the area or have obtained local hukou registration. The decision to change one’s place of residence is typically influenced by a combination of push and pull factors (Lee, 1966). Push factors prompt residents to leave their original homes, while pull factors attract or provide favorable conditions for those moving into a new place. ZG City, as one of China’s rapidly developing major cities, has a large number of migrants. The primary push force is the economic disparity between ZG City and the migrants’ original places of residence (Wu et al., 2019). Migrants often come from less developed areas that compel them to leave. On the other hand, ZG City’s developed economy offers attractive opportunities and higher rewards for workers. In this regard, the most significant difference between migrants and local residents lies in their economic status. Migrants typically have lower incomes, while local residents are often in better socioeconomic status.

Despite economic disparities, social barriers between migrants and local residents also exist (Xiao et al., 2021). In particular, language barriers and differences in lifestyle habits can form obstacles (Wang et al., 2018). These barriers also extend to interactions among migrants from different places of origin. Based on these barriers, migrants tend to have limited social contact with local residents. Instead, they often maintain connections with relatives in the city or hometown (Huang et al., 2018) and frequently engage in social interactions with other migrants to exchange information and have mutual assistance (Xiao et al., 2021). These characteristics highlight the significant socio-economic differences between migrants and local residents.

In terms of the social environment, the lack of social interactions between migrants and local residents results in a higher degree of anonymity, which may increase the level of social disorganization in the community. Social disorganization theory supposes that higher levels of social disorganization within a community can lead to an increase in crime rates (Bogomolnaia et al., 2005, Kubrin, 2009). On the contrary, the more familiar and communicative people are with each other, the higher the level of social cohesion is (Schmeets, 2012). The locals often share the same social customs, moral norms, and societal expectations, and thus have a strong willingness to maintain social order and informal social control (Conklin, 2010). However, as mentioned earlier, there are certain barriers between migrants and local residents, and the influx of migrants weakens the social cohesion in the community. Therefore, informal social control is stronger within communities dominated by the locals and weaker in migrant-concentrated communities.

In big cities in China, migrants from the same origin place tend to live nearby in urban villages for social support. Taking “Hubei Village” in Guangzhou as an example, Liu et al. (2020b) find that migrant settlements provide a foothold for the migrant population in the metropolis, playing a significant role in the formation and development of the migrant enclave for both early settlers as well newcomers.

In addition to differences in the social environment, there are also differences in the built environment, such as housing and facilities, in different types of communities. For example, in Guangzhou, self-built housing in urban villages is disproportionately concentrated in migrant-majority neighborhoods, whereas commercial housing and public housing are often found in local-majority neighborhoods and mixed neighborhoods.

The spatial configuration of facilities and amenities or Points of Interest (POIs) varies across communities, which may be consequential to crime activities such as burglaries. For example, Tillyer et al. (2021) find that the influence of crime-inducing factors on perpetrators was more pronounced in communities with more concentrated disadvantages and traffic activities. Conversely, in communities with high levels of civic engagement and strong social cohesion, the impact of crime-inducing factors was mitigated. Jones and Pridemore (2019) find that the characteristics of streets have a moderating effect on criminal activities. Therefore, POIs, as places that attract both internal and external community populations based on their service functions, play different roles in burglaries in different social environments.

In summary, scholars have conducted extensive research on burglaries, but studies that examine the different influences of factors of social and built environments on burglaries across various types of communities still remain rare. Therefore, this study categorizes different communities into local communities (LCs), migrant communities (MGCs), and mixed communities (MXCs), based on the share of migrant population. We then apply negative binomial regression models to investigate the distinct impacts of those influencing factors on burglaries in three types of communities.

Data and method

Study area and community categorization

This study takes ZG City, a major coastal city in southeast China, as the study area. The social and economic level of ZG City is highly developed, and the migrant population has shown rapid growth in recent years. There are a total of 2055 communities in ZG City. In this study, the population whose household is not registered in ZG City is classified as migrants.

Following Xiao et al. (2021), we define the community types based on the proportion of the migrant population in the community population provided by the sixth census collected in 2010. Communities with over 75% of the population are Migrant Communities (MGCs), and communities with less than 25% migrants are Local Communities (LCs). The remaining communities are Mixed Communities (MXCs) (i.e., 25–75%). This resulted in 136 MGCs, 1005 LCs, and 914 MXCs in the study area. In addition, we have verified the rationality of this threshold by examining the model robustness with different thresholds, i.e., 20% and 80%, and 30% and 70%. Model results for such three criteria present minimal difference, which proves that our classification criteria are acceptable.

Data

In this study, the dependent variable is the number of burglary cases in each community in 2014. Data was provided by the Public Security Bureau in ZG City. Independent variables are targets of burglaries, potential perpetrators, built environment, and social environment, consisting of 14 indicators. The number of arrested burglars living in each community is used as the indicator of motivated offenders in this study. Other control variables are introduced in detail as follows.

Targets of burglaries

We use data from the sixth census collected by the local government in 2010 to measure targets of burglaries. Key indicators include the number of households and the ratio of different housing types (e.g., commercial housing, affordable housing, and original public housing). More households would represent more targets. As argued above, different housing types mean different benefits and risks for potential burglars.

Commercial housing is newly constructed houses built by real estate developers. In the past few decades, only citizens with local households can purchase commercial housing. Affordable housing is a type of policy-driven housing aimed at low-to-middle-income families, with prices set below the market rate, provided by the government or relevant agencies. Purchasing affordable housing typically requires meeting specific income and asset criteria, with restrictions on housing size and price (Zou, 2014).

Original public housing in China was welfare housing allocated to employees by the government or state-owned enterprises during the planned economy period. Residents paid very low rent and had no property rights but only the right to use the housing. With the housing reform in the 1980s, public housing gradually became privatized through the housing reform policy, allowing residents to purchase property rights of houses at discounted prices (Chen et al., 2013).

In addition, it needs to be pointed out that self-built houses, built by native residents, are an important type of housing in ZG city, which are mainly located in urban villages with high renting ratio attracting a large number of migrants. It is closely related to the variable of the housing rental ratio in social environment below. Therefore, self-built housing is not considered in the models to avoid the problem of collinearity.

Built environment

Points of Interest data include the number of restaurants, Internet bars, karaoke bars, banks, bus stops and subway stations in each community. They are all provided by a famous mapping company in China collected in 2014. Restaurants, Internet bars, karaoke bars, and banks serve as venues for crowd gathering and activities, thus increasing the number of potential perpetrators. Bus stops and subway stations not only concentrate a large number of people but also reduce the travel costs of potential offenders (Yu and Maxfield, 2013).

Social environment

The ratio of houses with a rent of over 3000 yuan and the housing rental ratio are considered indicators of the social environment. They are obtained from the data from the sixth census data collected in 2010.

The tenants of rental units are generally migrant workers and new graduates. Located in urban villages or suburbs, the houses are typically old with limited anti-crime measures, thus attracting potential perpetrators to commit crimes (Vandeviver and Bernasco, 2020). Besides, high population mobility means that people are less familiar with neighbors in such communities, which weakens informal social control in communities and reduces the collective ability to prevent burglaries.

In addition, rental prices are factored into burglaries’ decision-making. Units with higher rents are more likely to become burglaries targets, as people living in high-rent communities have higher incomes in general. There are valuable items in the house. However, at the same time, high-rent housing tends to be more regulated and means a high risk for potential burglars.

Research method

Poisson regression and negative binomial regression models are suitable for the analysis of counting variables (Osgood 2000). Poisson regression requires the mean-to-variance ratio of the dependent variable to be close to 1. However, based on the analysis of the dependent variable (the number of burglaries) (as shown in Table 1), it is found that its minimum value is 0, the maximum value is 431, the average value is 18.62, and the variance is 954.15. The variance is much higher than the average value, and it indicates that the dependent variable is over dispersed. The negative binomial regression model has a better evaluation for the over-discrete variable (Chen, 2014) and it is therefore used in this study.

Table 1 Descriptive statistics of dependent and independent variables.
Full size table

We estimated a set of negative binomial regression models. The general form of the negative binomial model is:

$$E\left({y}_{i}\right)=\exp \left({\beta }_{0}+{\beta }_{j}\mathop{\sum }\limits_{j}^{m}{{TG}}_{{ij}}+{\beta }_{k}{{PP}}_{i}+{\beta }_{a}\mathop{\sum }\limits_{a}^{n}{{BE}}_{{ia}}+{\beta }_{b}\mathop{\sum }\limits_{b}^{o}{{SE}}_{{ib}}+{\varepsilon }_{i}\right)$$
(1)

Where \({\beta }_{0}\) is the intercept and \({\beta }_{j}\), \({\beta }_{k}\), \({\beta }_{a}\) and \({\beta }_{b}\) are the coefficients of predictors and covariates. \({{TG}}_{{ij}}\) represents variables of targets of burglaries. \({{PP}}_{i}\) represents variables of potential perpetrators. \({{BE}}_{{ia}}\) represents variables of built environment. \({{SE}}_{{ib}}\) represents variables of social environment. \({\varepsilon }_{i}\) represents unobserved heterogeneity at the community level.

Standardized coefficients and incidence rate ratios (IRR) are reported in the model results. Standardized coefficients are based on standardization of the independent variables and therefore can be used to compare the relative effects across variables in a model. To diagnose the potential multicollinearity, we calculated the Variance Inflation Factors (VIFs) values for each model. To compare model fit between models, we reported values of the Akaike Information Criterion (AIC) and Bayesian Information Criterion (BIC) of each model in model results. Smaller values of AIC and BIC indicate better model fit.

Spatial distribution of burglaries and community types

The overall spatial distribution of burglaries shows a circular pattern (Fig. 1), with the central business district as the core. The zone of the central district of ZG City has less than 17 burglary cases. In the urban periphery areas, the risk of burglaries is relatively high. The number of burglaries was between 18 and 101 in 2014 in most communities of this area. It reaches between 102 and 431 in a few communities. In the outer suburban and rural areas, the number of burglaries dropped significantly.

Fig. 1: Spatial distribution of burglaries in ZG City in 2014.
figure 1

The spatial distribution of burglaries shows a circular pattern around the city center.

Full size image

The spatial distribution of different types of communities in ZG City is also clustered (Fig. 2). MGCs are primarily located in the suburban areas of the city, with a small part near the dense areas of MXCs. Similarly, a high incidence of crime is distributed in a banded pattern around the center of the city. The number of burglaries has a high correlation with the distribution of MGCs. LCs are concentrated in the city center, the east, the north, and the southwest side of ZG city. They have an even distribution in the rest of the city. MXCs are primarily located in the transition zone between the center of the city and urban villages or in the urban periphery, where the local and migrant populations live in the same community in a relatively balanced proportion.

Fig. 2: Spatial distribution of community types in ZG City.
figure 2

The spatial distribution of different community types in ZG City is clustered and the distribution of MGCs is closely related to the spatial pattern of burglaries.

Full size image

According to the descriptive statistics of valuables in Table 1, MGCs have a larger number of burglaries than LCs and MXCs with an average of 63.640. Accommodating 4700 households on average, MGCs provide ample targets for burglaries. They are characterized by a complex population composition, a high proportion of non-local residents, high population mobility, and weak social networks among residents. Due to the lack of social cohesion in the community, a high degree of social disorder, and the dysfunction of informal social control among neighbors, there are more potential perpetrators living in MGCs, with an average value of 1.713. At the same time, a large number of restaurants and internet bars are distributed in MGCs, providing a multitude of venues that attract visitors. Besides, most of the migrant workers have low incomes, and they tend to live in communities with low rent. The housing rental ratio is the highest of the three community types. We calculate the Coefficient of Variation (CV) to compare the degree of variability between two data sets. Generally, the higher the value of the variable, the greater the degree of dispersion is. The CV values of affordable housing ratio, subway stations, and rent over 3000 Yuan are high in MGCs. Despite the weak defensive function of buildings in these communities, the relatively low economic level of migrant workers and the low value of properties discourage burglaries.

On average, there were only about 8 burglaries that occurred in LC communities in 2014, ranking the lowest among the three community types. However, original public housing ratio is the highest among three community types reaching 0.189 while the other two are 0.004 and 0.073. LCs are characterized by homogeneous population and low population mobility. In addition, in LCs, the built environment of communities is more mature with strong informal social control and social cohesion. The number of burglaries in MXCs is smaller than in MGCs but larger than in LCs. Values of targets of burglaries, potential perpetrators, and the number of POIs in MXCs are at the level between those of MGCs and LCs. Similar to MGCs, MXCs have a large number of restaurants, attracting a large number of outsiders. Meanwhile, its housing rental ratio is 0.385, at a level between those of MGCs and LCs. Besides, the CV of the rental prices over 3000 Yuan in MXCs is the highest reaching 5.313. Specifically, the CV of the affordable housing ratio is 4.500, which is higher than most other variables in MXCs.

Model results

Table 2 shows the VIF value of each variable. VIF tests the collinearity among all variables. Given all values are less than 2.5, the collinearity of independent variables is weak, indicating that the selection of model indicators is reasonable. Overall, as for the targets of burglaries, the number of households and the original public housing ratio are positively and negatively correlated with burglaries at the 0.001 level. The commercial housing ratio is significantly positively correlated at the 0.05 level. Regarding potential perpetrators, the number of offenders is significantly positively correlated with burglaries at the 0.001 level. The built environment primarily consists of various types of Points of Interest. Among these, restaurants, banks, Internet bars, and bus stops are significantly positively correlated at the 0.001, 0.01, 0.05, and 0.1 levels, respectively. In terms of the social environment, both housing with rent over 3,000 Yuan and the housing rental ratio are significantly correlated at the 0.001 level. Housing with rent over 3000 Yuan is negatively correlated with burglaries, while the housing rental ratio is positively correlated. With the mixed community as the reference group, the risk of burglaries in LCs is significantly lower, while there is no difference in the risk in MGCs.

Table 2 Negative binomial regression results of the residential burglaries impact factors.
Full size table

The focus of this study is to analyze the differences in the impacts of influencing factors on burglaries among different types of communities (Table 3). We next present results for three types of communities.

Table 3 Negative Binomial Regression results of impact factors of residential burglaries in different types of communities.
Full size table

Migrant community models

In the model of MGCs, a significant positive correlation exists between the number of households and burglaries. Most residents in MGCs are migrant workers. During the daytime, they need to go to work, resulting in a lack of family supervision at home, which weakens guardianship and increases the incidence of burglaries.

Besides, the number of offenders and the number of Internet bars have a significant positive correlation with burglaries at the level of 0.05. Internet bars may reflect the high concentration of teenagers with delinquent behavior in the community. The number of banks positively correlates with burglaries at 0.1 level. However, we do not observe any significant correlation between burglaries and housing types, other built environment factors, and social environment factors.

Local community models

Regarding targets of burglaries, the number of households, commercial housing ratio, as well as original public housing ratio significantly correlate with burglaries. In line with MGC Models, more households provide more targets for burglaries. Original public housing ratio negatively correlates with the presence of burglaries. Original public housing primarily serves as dormitories for company employees. The widespread social network among residents fosters a vigilant community where residents look out for each other. As social cohesion gets strong, it in turn reduces the risk of burglaries. The number of perpetrators, restaurants, Internet bars, and bus stops significantly positively correlates with burglaries. It indicates that LCs with more facilities may attract more potential perpetrators’ daily activities. Meanwhile, housing rental ratio and the ratio of housing rent over 3000 Yuan positively and negatively correlate with burglaries. The majority of migrants usually reside in rental housing in LCs, where the anti-crime measures are weak, thus attracting potential perpetrators. The presence of high-rent housing units with accompanied high level of security system and other infrastructure poses greater cost and risk to potential perpetrators. Therefore, there is a significant negative correlation with burglaries.

Mixed community models

In MXCs, the number of households has a significant positive correlation with burglaries at the level of 0.001. Both the commercial housing ratio and original public housing are negatively correlated with burglaries. Different from MGCs, commercial housing in MXCs has a substantial negative impact on burglaries. The residents of commercial housing are generally a high-income and stable group, and the potential return from burglaries is high. However, its community management is typically outsourced to experienced property management companies with mature security and access control systems and strict and effective prevention of community burglaries. Regarding potential perpetrators and built environment, the number of perpetrators, the number of restaurants, as well as the number of banks are positively correlated with burglaries. Moreover, consistent with LCs, housing rental ratio is significantly correlated with burglaries at the 0.001 level.

Residual analysis

We identified outliers using double standard deviation to conduct residual analysis of the model, and the results are shown in Figs. 3 and 4.

Fig. 3: Outliers in the general model.
figure 3

11 outliers in the general model are mainly in MGCs and MXCs.

Full size image
Fig. 4: Outliers in different models.
figure 4

There are 6 outliers in the MGC model and 1 in the MXC model.

Full size image

In general models, there are 11 outliers in total, 9 in migrant communities and 2 in mixed communities. The number of burglary cases estimated by the model is higher than the actual values for all the outliers. Most of these outlier communities have more than ten thousand households, far more than the average value of all communities. MGC communities are characterized by dense buildings with multiple households on every floor, which easily leads to errors in estimation by the general model. Meanwhile, the number of offenders is one of the important factors affecting burglaries. There are more than 3 offenders in these communities, which is also higher than the mean value, raising the predicted value of the model. In addition, these communities contain a relatively high number of restaurants, Internet bars, and banks. From Table 2, these factors all have a positive effect on the number of burglaries, which also contributes to discrepancies between the model predictions and actual conditions.

There are fewer outliers in separated models than in the general model. There are 6 outliers in the MGC model, half of which have more than ten thousand households, two having nearly 100 restaurants (five times the average), and one with up to 13 Internet bars. There are two outliers with smaller values of burglaries predicted by the models than those of their actual values in the MGC model. One has fewer households, offenders, Internet bars and banks than the average, and the other has a smaller than average number of Internet bars and banks, resulting in a smaller than actual prediction. Besides, there is only one outlier in the MXC model, which is overestimated due to its large number of households.

Discussion

Burglaries in ZG City are influenced by the targets of burglaries, potential perpetrators, the built environment as well as social environment. The number of households and offenders has stable, positive, and significant effects in attracting burglaries across community types. The large number of households provides more targets for perpetrators (Townsley et al. 2015), and perpetrators living nearby increase the likelihood of residential burglaries (Vandeviver and Bernasco, 2020).

Many of the variables have diverse influences on burglaries in different types of communities, which is mainly due to their discrepancy in informal social control capacities as well as the characteristics of facilities in attracting populations (Alhazzani et al., 2021). Besides, such diverse influence is also closely related to the statistical variation in the distribution of variables. The detailed explanations are as follows.

Housing types

Migrant communities (MGCs) are primarily inhabited by the migrant population, predominantly consisting of self-built houses by the local (Liu et al., 2015). The variation in the number of other types of housing is minimal, resulting in no significant correlation between burglaries and community types.

Affordable housing primarily serves the socioeconomically disadvantaged population (Nelson, 1994). Given their lower economic status, the benefits to perpetrators from committing residential burglaries are minimal. Additionally, safety and affordable housing is typically not the preferred target for perpetrators (Xiao et al., 2017). As a result, there is no significant correlation between affordable housing and burglaries in all three types of communities.

As for the commercial housing of the Local Communities (LCs), upscale residential districts are typically featured with items of higher value, allowing perpetrators to gain greater profits from burglaries. Moreover, commercial housing is home to migrants, which will weaken the social ties among the native residents. Despite the fact that these upscale residential districts are usually equipped with stricter access control, commercial housing ratio is still positively correlated with burglaries. When it comes to mixed communities, commercial housing can significantly deter burglaries due to its sound security measures.

Regarding the original public housing ratio, the majority of houses are provided by country-owned enterprises, and the residents typically consist of local employees from these enterprises. Typically, communities with a higher proportion of local residents exhibit active informal social control (Song et al., 2018), making it more difficult for burglars to commit crimes. Due to work interactions, neighbors are familiar with each other, resulting in strong informal social control. Consequently, this leads to effective control over burglaries, both in LCs and Mixed Communities (MXCs).

Built environment

Built environment primarily considers the impact of different types of points of interests (POIs) on burglaries. Restaurants often serve as crime generators in communities (Wo, 2023). Restaurants can attract visitors, which in turn brings in more potential perpetrators. Meanwhile, more outsiders contribute to greater social anonymity. Therefore, a significant positive correlation is observed in LCs and MXCs. However, the restaurant is not significant in the MGC model. It is probable that the anonymity among residents in migrant communities is high and the outsiders attracted by restaurants will not further increase the social anonymity. Consequently, an increase in restaurant numbers has a minimal marginal impact on burglaries.

Internet bars are regarded as crime attractors in communities where the crime rate is significantly correlated with local internet bars (Hu et al., 2018). Internet bars are often gathering places for delinquent youths (Donati et al., 2022), who are overrepresented among the burglar population. As a result, more internet bars in MGCs and LCs are associated with more burglary cases.

Banks can increase the risk of burglaries in MGCs and MXCs, but not in LCs. Bank is an important activity node for general population as well as the potential burglars, through which they can gain knowledge about the crime opportunities. In LCs, due to their strong informal social control capacity, the impact of banks on burglaries is not significant.

We observe a significant relationship between the number of bus stops and burglaries in LCs but not in MGCs or MXCs. In LCs, bus stops bring non-local residents into local communities, heightening the anonymity within the community, reducing the risk of perpetrators being identified, and weakening the informal social control in these areas. Burglaries tend to target communities with higher accessibility to public transportation (Xiao et al., 2018). However, unlike bus stops, subway station waiting areas are typically located underground. An increase in subway stations does not correspond to a significant rise in burglary cases.

Social environment

Housing with rents over 3000 Yuan often has more security resources, which limit perpetrators from committing burglaries. As a result, there is a significant negative correlation between these housing types in both LCs and MXCs. In the case of the housing rental ratio, the population primarily consists of migrant workers who lack family support. These individuals often live alone with limited informal social control provided by family members. During the day, they go to work, leaving their homes unattended. The lack of supervision in these rental properties creates opportunities for perpetrators to commit burglaries. As a result, the housing rental ratio is significantly positively correlated with burglaries in both LCs and MXCs.

In summary, different types of communities have distinct characteristics in their housing types, and physical and social environments. MGCs typically have a large number of migrants, higher social anonymity, and greater levels of social disorganization. LCs are mostly composed of local residents. Neighbors are generally familiar with each other, fostering strong social cohesion and informal social control. MXCs remain between the two, with both local residents and a certain number of migrants. Generally speaking, higher anonymity and social disorganization, as well as weaker social cohesion and informal social control lead to higher crime rates. This can help explain the variation in the effects of burglaries-inducing factors across migrant communities, local communities and mixed communities.

Conclusions

This study explores the influencing factors of burglaries and their differences among migrant communities, local communities, and mixed communities, by using the police data provided by the public security bureau of ZG City and following routine activity theory and social disorganization theory.

Generally speaking, targets of burglaries, potential perpetrators, built environment, and social environment can help understand the spatial distribution of burglaries. With all these variables controlled, the number of burglaries is significantly lower in local communities than in migrant communities. The discrepancy between migrant communities and mixed communities is not significant.

The key finding of the study is that the combination of influencing factors and their impact on burglaries vary by community type, with the exception that the number of households and potential burglars are positively related to burglaries in all communities. The presence of internet bars and banks significantly increases the risk of burglaries in MGCs but not in other communities.

The positive influencing factors of burglaries in LCs are restaurants, Internet bars, bus stops, and housing rental ratio. The original public housing ratio and units with rent over 3000 yuan show a negative influence. For MXCs, commercial housing ratio has a negative effect, and banks have a positive effect on burglaries.

This study confirms findings in previous studies that social environments and housing characteristics differences across communities are associated with different incidences of burglaries (Jiang et al., 2010, Xiao et al., 2017, Yang et al., 2016a, Yang et al., 2016b). The occurrence of community burglaries is inseparable from offenders, targets of the crime, and the informal social control among residents. That is, the three elements of the routine activity theory work together (Drawve et al., 2014). This study contributes to the literature by dividing different types of communities based on the migrant population proportion and compares their varied burglaries-inducing factors.

However, this study also has some limitations. First, despite that there is evidence from the study of Xiao et al. (2021) for the division of communities, different benchmarks need further verification. The number of different community types is not balanced, with fewer MGCs than the other two community types, which may be consequential for our results. Second, due to the difficulty in obtaining police data and the great impact of COVID-19 on social factors in recent years, this study used the police data of 2014 and the data of the sixth census collected in 2010 to carry out research, which cannot reflect the update crime distribution and its influencing factors. Despite this study demonstrating an accurate and scientific explanation of the mechanism underlying crime occurrences, the updated data is needed for further exploration in subsequent studies. Moreover, ZG city is a mega city with a large migrant population. This study divided community types based on the proportion of migrant population. Universality for small and medium-sized cities requires further consideration.

Exploring influencing factors in different types of communities helps prevent and reduce community crimes in a more context-specific way. For example, more attention should be paid to the activities of potential perpetrators, as their presence can increase the risk of burglaries. The impact of housing characteristics on burglaries in LCs and MXCs differs from those in MGCs. Different types of communities should imply targeted, differentiated, and effective management according to different housing types and characteristics. Given that there are significant differences in the social and built environment among different community types, categorizing communities and investigating the factors influencing burglaries can provide more valuable references for community governance tailored to specific environments. For example, MGCs have a larger migrant population and a higher rental housing ratio than the other two types of communities. As a result, the housing rental ratio is not significant in the MGC model, but has a positive impact in the other two models. From a social governance perspective, this research can inform policymakers of the need for adaptive strategies that address the unique characteristics and challenges of each community. Understanding the nuances of crime patterns can enhance the effectiveness of preventive measures, fostering safer environments for residents. In addition, policy makers and urban planners could adopt strategies such as implementing urban renewal and environmental changes that are specifically tailored to each community’s needs, and increasing neighborhood informal social control capacity to reduce crime. To sum up, this study employs the method of zoning classification to explore the influencing factors influencing burglaries by considering the characteristics of different types of communities, and provides a reference case for future scientific inquiries using community classification methods.