Revisiting the Relationship Between Racial Heterogeneity and Neighborhood Crime: Do Spatial Scale and Functional Form matter?

Abstract

This study examines the relationships between racial heterogeneity and crime across blocks (N = 103,168) located in the greater Southern California region. We estimate negative binomial regression models that test for the effects of racial heterogeneity in conjunction with different functional form and spatial scaling considerations. Racial diversity in the block has a crime-reducing effect, whereas racial diversity in the area surrounding the block generally has crime-producing capabilities, although at very high levels of diversity, the effect reverses and becomes crime-reducing. We also illustrate an interaction effect between racial heterogeneity in the block and racial heterogeneity in the surrounding area. The pattern of results provides a nuanced understanding of how the racial composition of an area has consequences for crime.

Keywords

crime racial heterogeneity social disorganization theory contact theory

Introduction

Criminological studies have theorized that racial heterogeneity is associated with more crime in spatial units—namely, social disorganization theory (Hipp, 2011; Kubrin, 2000; Peterson et al., 2000; Sampson & Groves, 1989; Warner & Pierce, 1993; Warner & Rountree, 1997). These studies posit that racial heterogeneity undermines residents’ ability to informally solve and prevent crime in terms of collective efficacy, social networks, or capable guardianship, to name a few. Despite the longstanding assumption that racial heterogeneity is associated with more crime, recent empirical work suggests that the relationship might be more nuanced than previously conceived (Boessen & Hipp, 2015; Kim, 2018; Kim & Hipp, 2020; Wenger, 2019).

Although most studies have found a crime-producing effect of racial heterogeneity, other studies have yielded null findings or have even shown evidence of a crime-reducing relationship (Kim, 2018; Kim & Hipp, 2018; Smith et al., 2000). These studies interpreted the results that racial diversity can be potentially beneficial for developing social cohesion and social ties among residents, and therefore reduce crime and disorder in place. Specifically, based on contact hypothesis, direct face-to-face social interactions among diverse others can diminish intergroup bias, and build mutual trust across different racial groups (Allport, 1954; Gaertner et al., 1996; Lee & Bean, 2010; Pettigrew & Tropp, 2006). We argue that such theoretical insights from contact hypothesis can be relevant to unpacking the nuanced relationship between racial heterogeneity and crime.

Although studies have given considerable attention to racial diversity and how it matters for understanding neighborhood crime, there are several considerations that might provide new insights into the nexus between racial heterogeneity and crime. In the current study, we propose that the relationship between racial heterogeneity and crime may depend on two interrelated factors: (1) the functional form of the relationship and (2) the spatial extent/scope of the relationship. Regarding the first point, we outline two compatible perspectives by proposing a curvilinear association between racial heterogeneity and neighborhood crime, when each racial group has large enough population for sufficient opportunities for frequent interpersonal contacts with the other groups.

For the second point, some previous studies testing social disorganization theory using small spatial units (e.g., street segments) have often found a crime-reducing effect of racial diversity, whereas others analyzing larger spatial units have detected the opposite effect (Kim, 2018; Kim & Hipp, 2018; Smith et al., 2000). This implies that spatial scaling can be an important factor for understanding racial diversity, social interactions, cohesion, and crime in neighborhoods. Specifically, it is plausible to think that residents belonging to different racial groups living in a small area may have more opportunities to physically and socially interact with one another to reduce group-based bias and build trust, which may not be captured when looking at larger spatial units such as block groups, tracts, and cities. In the subsequent sections, we explain our theoretical motivations for examining racial heterogeneity and the spatial patterns of neighborhood crime. Then, we describe our data and methodology, including detailed modeling strategies, followed by our findings and their implications.

Crime-Enhancing Effect of Racial Diversity

Over the last few decades, a bulk of studies has found that racial diversity has a positive association with neighborhood crime (Hipp, 2010, 2011; Kubrin, 2000; Sampson & Groves, 1989; Warner & Rountree, 1997; Wenger, 2019). These studies argue that racial heterogeneity (or racial diversity interchangeably) is likely to increase neighborhood crime because racially diverse settings can undermine social cohesion among residents. Most of these studies are situated within social disorganization theory. The theory posits that certain structural characteristics such as socioeconomic disadvantage, residential instability, and racial heterogeneity hinder the development of social ties and cohesion among residents, and therefore, weaken the capacity of residents, themselves, to monitor and regulate criminal behavior—informal social control.

Proponents of such ideas argue that residents in racially diverse communities will have few social ties between racial groups, which can limit neighborhood capacity for collective action (Bursik, 1988; Bursik & Grasmick, 1993; Warner & Rountree, 1997). Specifically, due to cultural and language differences between racial groups, residents will feel less attached to each other, and therefore, have lower social ties and social cohesion compared to those in more homogenous neighborhoods. As a consequence, they will be less likely to work together to take care of common community problems including crime and disorder. As Kornhauser (1978) (as cited in Kubrin, 2000) stated, “Heterogeneity impedes communication and thus obstructs the quest to solve common problems and reach common goals” (p. 78).

Indeed, many studies have found a crime-enhancing effect of racial diversity (Kubrin, 2000; Sampson & Groves, 1989; Sun et al., 2004; Wenger, 2019). For instance, Sampson and Groves (1989) found that neighborhoods with higher level of racial diversity have more violent and property crimes. Likewise, Kubrin (2000) found that racial heterogeneity has a strong positive effect on neighborhood violent crime even after accounting for other structural characteristics and demographics. In a recent study, Wenger (2019) also found a similar result that neighborhood-level diversity is positively associated with robbery. Although these studies seemingly provide evidence of a crime-producing effect of racial heterogeneity, another body of studies argues that racially diverse settings can actually promote social ties and cohesion among residents, therefore reducing neighborhood crime. In the next section, we discuss how racial diversity can have a crime-reducing effect.

An Alternative: Crime-Reducing Effect of Racial Diversity?

An alternative proposition is that racial diversity can actually enhance social cohesion and informal social control if frequent opportunities for social interactions are provided. The contact hypothesis proposes that residential contacts in diverse settings can reduce intergroup bias and develop knowledge-based mutual trust. Specifically, it posits that regular face-to-face interactions between different racial groups can help to reduce prejudice on diverse others, which in turn foments social ties, trust, and cohesion among residents across the group boundaries (Allport, 1954; Gaertner et al., 1996; Lee & Bean, 2010; Pettigrew & Tropp, 2006). For instance, Marschall and Stolle (2004) argue that direct social interactions can reduce intergroup prejudice and build trust among different racial groups. Specifically, they found that their racial heterogeneity measure was positively related to trust among residents. This implies that racially diverse neighborhoods can have higher levels of mutual trust among residents than homogenous neighborhoods.

Casual contacts among interracial groups provide more opportunities to get exposed to members of other groups, which leads to greater acceptance and positive attitudes toward those groups. Furthermore, personal friendship between the members of different racial groups reduces negative attitudes toward those groups (Ellison & Powers, 1994; Sigelman & Welch, 1993) whereas casual intergroup contact enhances the chances to have make close social ties with members from other groups (Ellison & Powers, 1994). Therefore, Pettigrew and Tropp’s (2006) meta-analysis suggested that “[t]here is little need to demonstrate further contact’s general ability to lessen prejudice. Results from the meta-analysis conclusively show that intergroup contact can promote reductions in intergroup prejudice” (p. 751). These theoretical and empirical insights from contact hypothesis suggest that racial heterogeneity will exert protective effects on crime in neighborhoods. That is, racially diverse settings can produce social cohesion and reduce group-based prejudice among residents, which is conducive to enhanced informal social control and effective crime control in neighborhood.

The Functional Form and the Spatial Scale of Racial Heterogeneity Effects

Although the two perspectives on racial heterogeneity and crime may seem at odds with each other (on face value), we argue that they can be theoretically compatible by proposing a curvilinear association between racial heterogeneity and neighborhood crime. Specifically, we propose an inverted U-shaped association between racial diversity and neighborhood crime. At lower levels, racial diversity may have a crime-enhancing effect because social cohesion among residents in different racial groups is likely weak, according to social disorganization theory. Consequently, we can expect lower levels of informal social control which results in more neighborhood crime. However, as contact hypothesis suggests, diversity can also foster inter-group tolerance and trust, and diminish group-based bias, if sufficient opportunities for frequent interpersonal contacts with the other groups are provided. That is, social contact between the groups can reduce the prejudice across different groups, “when a group becomes large enough and interpersonal contact becomes frequent enough” (Enos, 2017, p. 49). Therefore, if in a racially diverse neighborhood wherein each group has a pronounced concentration, there will be sufficient opportunities to develop knowledge-based trust and understanding among the groups, and thus, there will be reduced levels of neighborhood crime.

Another important factor to consider is the spatial scale of units of analysis. Most studies on racial diversity (rooted in social disorganization theory) have often employed various spatial units aggregated at certain levels to define the neighborhood (e.g., census tracts or block groups). However, a recent body of crime and place studies argues that social disorganization theory and informal social control thesis can be examined using micro spatial units. This is because smaller units can be behavior settings that are micro-communities where residents interact with each other and have frequent casual social encounters (Wicker, 1987). Therefore, these small areas contain social environments and structural characteristics that are depicted in social disorganization theory (Taylor, 1997; Weisburd et al., 2012, 2017, 2020).

Some of these studies have attempted to empirically test social disorganization theory using micro spatial units (such as street segment) and have often found a crime-reducing effect of racial diversity on crime in place (Kim, 2018; Kim & Hipp, 2018; Smith et al., 2000). For instance, Smith et al. (2000) found that racial heterogeneity was associated with lower risk of robbery in street segments. Likewise, Kim (2018) and Kim and Hipp (2018) also found that racial diversity can reduce the risk of violent and property crime in street segments in the Southern California region. These studies suggest that racial heterogeneity may affect crime in different ways according to their spatial scale (e.g., Tracts vs. blocks) because social interactions, ties, and cohesion are likely to vary depending on the size of spatial units employed. Specifically, due to smaller spatial scale, residents in small areas (such as street segments or blocks) may have more opportunities for face-to-face social interactions with others, which can reduce racial group-based bias and thus build social ties and trust with different racial groups. Such environment can foster social cohesion among residents and thus effective informal social control, which may not be captured when looking at larger spatial scales (Tracts or cities).

Moreover, if indeed informal social control mechanism is at the center of social disorganization theory, crime-reducing effect of racial diversity is possible because arguably there will be stronger social interactions and ties in closer proximity to one another. For instance, a resident may have more interpersonal or casual interactions, and feel more social cohesion with another resident living next or few doors down the street than others living in a place a few miles away (although they are all living in the same Census Tract). Therefore, we suggest that the spatial extent/scope can be an important factor in shaping the association between racial heterogeneity and neighborhood crime.

Data and Methods

This study examines the potential relationships between racial heterogeneity and crime across blocks (N = 103,168) located in the greater Southern California region, controlling for covariates typically employed by communities and crime research. For our analyses we primarily draw on three sources of data: (1) official crime data coded and reported by police agencies of the Southern California region; (2) sociodemographic data provided by the U.S. Census Bureau; and (3) land use data from the Southern California Association of Governments (SCAG). Consistent with prior work that has examined neighborhood effects on crime (e.g., see Bernasco & Block, 2011; Contreras, 2017; Haberman & Kelsay, 2021; Hipp et al., 2019; Wo & Park, 2020), we use census blocks as our units of analysis. The study area is the greater Southern California region which encompasses the counties of Los Angeles, Orange, Riverside, San Bernardino, and San Diego. We selected this region because it is an racially/ethnically diverse region wherein it is common for the concentration of nonwhite groups (e.g., Latino or Asian) to rival or even exceed the concentration of whites for a given city. This ensures significant variation across space in the concentration of racial groups (i.e., white, Black, Latino, Asian), as well as the amount of heterogeneity among such groups.

Dependent Variables

We draw on official crime data provided by the Southern California Crime Study (SCCS); a large-scale effort on the part of researchers at the University of California-Irvine aimed to collect longitudinal crime data across the Southern California region.¹ These data include necessary information on crime incidents such as date, location, and type of crime according to the Uniform Crime Reporting (UCR) program. Accordingly, we have computed a three-year sum (2009–2011) for the following crime types, aggregated to the block: aggravated assault, robbery, burglary, larceny, and motor vehicle theft. While we recognize that official crime data are not immune to measurement error, we have no reason to suspect that these data are any less valid than alternative sources for measuring crime (Baumer, 2002; Berg & Lauritsen, 2016). Notably, the SCCS data are increasingly being used in crime and place analyses (e.g., Contreras, 2017; Hipp & Kubrin, 2017; Hipp et al., 2019; Kim & Hipp, 2020; Kim et al., 2019; Kubrin et al., 2018).

Independent Variables

We constructed a series of variables based on data from the 2010 U.S. Census. To test the effects of racial heterogeneity on crime across blocks, a Herfindahl index was created according to the proportions of five racial groupings (White, Black, Latino, Asian, and other races), which takes the following form:

H = 1 - \sum_{j = 1}^{J} G_{j}^{2}

where $G$ refers to the proportion of the population of racial group $j$ out of J groups (Gibbs & Martin, 1962, p. 670). It follows that each block of our sample receives a score on this index with values closer to 1 approaching perfect heterogeneity, whereas values closer to 0 approaching perfect homogeneity. It should be noted that the Herfindahl index has commonly been used to capture the racial diversity (or mixing) of spatial units in relation to crime (Boessen & Hipp, 2015; Hipp, 2007, 2010; Kubrin & Hipp, 2016).

Research investigating neighborhood effects on crime has revealed that associations can exhibit substantive differences at varying spatial scales (Boessen & Hipp, 2015; Groff & Lockwood, 2014; Peterson & Krivo, 2010), whether it be significance, direction, or magnitude of the association. It follows that the impact of racial heterogeneity in the focal block may operate differently than the impact of racial heterogeneity in the area surrounding the focal block (Boessen & Hipp, 2015). Previous studies have suggested that an overlapping neighborhood boundary approach may be preferred (Hipp & Boessen, 2013; Kim & Hipp, 2020), given that the spatial boundaries of neighborhoods can be arbitrary and thus it is difficult to implement discrete boundaries when measuring neighborhood structural characteristics, such as racial compositions and heterogeneity. Prior studies to have investigated the effects of racial heterogeneity on crime have defined neighborhood boundaries using discrete administrative boundaries. This nonoverlapping approach relies on the demarcation of “neighborhoods” based on physical boundaries (e.g., rivers and freeways) and social boundaries (e.g., the location at which the economic, or racial/ethnic, character of the residents changes) in order to maximize homogeneity within geographic units while simultaneously maximizing heterogeneity across such units.

A limitation of the nonoverlapping approach is that measuring neighborhoods based on discrete boundaries and spatial homogeneity within the unit may not accurately capture the true amount of racial heterogeneity and other structural characteristics. To address this concern, recent studies have employed the egocentric buffer approach when measuring characteristics of areas/neighborhoods (Hipp & Boessen, 2013; Hipp & Kubrin, 2017; Kim & Hipp, 2020). Particularly, Hipp and Boessen (2013) introduced an egohood approach that involves overlapping concentric circles that surround each block in the city. This means that egohoods account for the heterogeneity that exists within a uniform spatial area, as they can span physical and social boundaries. The full extent of the egohood approach is summarized by Hipp and Boessen (2013, pp. 294–295):

“When egohoods are constructed for all blocks in the city, a particular block is tied not only to the blocks in its own buffer but also to the buffers of these blocks. As such, the egohood of the focal block will contain portions of the buffers of all of the blocks within its own buffer. Thus, the closer two blocks are geographically, the more buffers they will share with each other.”

The current study employs the egohood approach to properly account for racial heterogeneity in the focal block as well as surrounding areas, simultaneously. To examine the spatial extent of the association between crime and racial heterogeneity, we have computed a Herfindahl index at five spatial scales: (1) the focal block; (2) the spatial area within ¼ mile around the focal block; (3) the spatial area within ½ mile around the focal block; (4) the spatial area within ¾ mile around the focal block; and (5) the spatial area within 1 mile around the focal block. The spatial scales that refer to the area surrounding the focal block (based on varying radii), we hereafter refer to as egohoods.

To minimize the risk of obtaining spurious results, we account for a range of covariates that have either been theorized or shown to be robustly associated with aggregate crime outcomes (Kubrin & Wo, 2015; Pratt & Cullen, 2005). We created a scale of concentrated disadvantage, which is a factor score computed after a principal component factor analysis of the following variables: percent poverty, percent single-parent households, average household income, and percent with at least a bachelor’s degree (the latter two variables have negative loadings). The 2010 Census is a short-form questionnaire thereby it does not provide these variables at the block level (except for single-parent households). Thus, we imputed the remaining variables using the synthetic estimation approach by Boessen and Hipp (2015). In addition to racial heterogeneity, we included a measure of both the percent Black and Latino. Percent occupied units is used as a proxy for low physical disorder (Boessen & Chamberlain, 2017; Chen & Rafail, 2020; Skogan, 1990); Notably, the correlation between occupied units and vacant units is perfectly negatively correlated. We use the percent homeowners as an indicator of residential stability. We control for the percentage of 15 to 29-year-old residents given that young people tend to offend and be victimized more than any other age group. For each sociodemographic characteristic, we have constructed a measure pertaining to the focal block, as well as the spatial area within ¼ mile around the focal block (i.e., egohood).

We use data from SCAG in order to account for the physical structural qualities of areas that might shape criminal opportunities and informal social controls (Boessen & Hipp, 2015; Stucky & Ottensmann, 2009; Wo, 2019). For each parcel in the Southern California region (i.e., plot of land), we have information on its location and boundary, as well as the primary activity/function occurring on the land in 2008. Like previous research examining land use effects on crime (Boessen & Hipp, 2015; Hipp et al., 2020; Zahnow, 2018), we aggregated these data to their constituent block and computed the percentage of the block area classified into the following land use categories: residential, retail, office, and industrial. For each land use type, we have constructed a measure pertaining to the focal block, as well as the spatial area within ¼ mile around the focal block (i.e., egohood).²

Analytic Strategy

The dependent variables are highly skewed and indicate overdispersion; therefore, we employ negative binomial regression to model the effects of racial heterogeneity on crime.³ We estimate models on a sample of blocks with a nonzero population (N = 103,168) given that it is methodologically necessary to control for sociodemographic characteristics that have either been shown or theorized to be robustly associated with aggregate crime outcomes (Kubrin & Wo, 2015; Pratt & Cullen, 2005). Moreover, such a population is fundamental to understanding crime according to social disorganization and routine activities theories (Felson M & Boba, 2010; Sampson & Groves, 1989). In Table 1, we provide summary statistics for all measures of our sample. A general expression of the models is as follows:

y = β_{1} H + β_{2} W H + β_{3} S + β_{4} W S + β_{4} L + β_{5} W L + β_{6} C + α,

where $y$ is the number of crime incidents in the focal block, $H$ is racial heterogeneity in the focal block, $W H$ is racial heterogeneity surrounding the focal block based on a chosen radius/buffer length (¼ mile, ½ mile, ¾ mile, or 1 mile), $S$ is a matrix of sociodemographic characteristics in the focal block, $W S$ is a matrix of sociodemographic characteristics in the ¼ mile buffer surrounding the focal block, $L$ is a matrix of land uses in the focal block, $W L$ is a matrix of land uses in the ¼ mile buffer surrounding the focal block, $C$ is a matrix of dummy variables for cities, and $α$ is an intercept. Estimated models effectively highlight crime differences across blocks within particular cities by conditioning out differences across cities (via the fixed effects for cities). To account for population differences across blocks, we included (logged) block population as an exposure variable and constrained the coefficient to equal 1, which effectively makes the outcomes interpretable as crime rates (Hilbe, 2007; Osgood, 2000).

Table 1.

Summary Statistics.

	Mean	SD
Crime counts
Agg. Assault	0.24	0.71
Robbery	0.11	0.39
Burglary	0.46	1.15
Larceny	0.90	3.37
M.V. theft	0.32	0.82
Racial heterogeneity
Block	0.41	0.21
1/4 mile egohood	0.46	0.18
1/2 mile egohood	0.48	0.16
3/4 mile egohood	0.49	0.15
1 mile egohood	0.50	0.15
Structural characteristics
Concentrated disadvantage	−0.19	9.11
Concentrated disadvantage (1/4 mile egohood)	−1.87	9.07
% Black	5.96	13.59
% Black (1/4 mile egohood)	6.04	11.20
% Latino	37.81	31.46
% Latino (1/4 mile egohood)	39.18	28.26
% Occupied housing units	94.08	9.31
% Occupied housing units (1/4 mile egohood)	93.68	6.72
% Homeowners	63.75	31.87
% Homeowners (1/4 mile egohood)	59.75	25.91
% Aged 15–29	20.44	10.94
% Aged 15–29 (1/4 mile egohood)	21.49	7.23
Land use measures
% Residential	78.25	32.28
% Residential (1/4 mile egohood)	68.98	24.69
% Retail	4.80	14.85
% Retail (1/4 mile egohood)	5.57	8.51
% Office	2.54	11.81
% Office (1/4 mile egohood)	2.47	6.08
% Industrial	1.93	9.77
% Industrial (1/4 mile egohood)	3.06	8.50

Note. N = 103,168; SD = standard deviation.

We calculated the Moran’s I values of the residuals to assess the degree of spatial autocorrelation remaining in the models and found little evidence of spatial autocorrelation: the Moran’s I value was no greater than 0.01 across all the models estimated. Given that prior studies have produced equally small (or higher) Moran’s I values when using small spatial units (Bernasco & Block, 2011; Contreras, 2017; Kim, 2018; Wo et al., 2016), this suggests that our measures pertaining to the area surrounding the focal block have effectively accounted for most of the spatial autocorrelation. We also found no evidence of multicollinearity problems, as the maximum variation inflation factor (VIF) score was 3 across all the models estimated.

We present two sets of models for each crime outcome that allow for different spatial scales and functional forms of racial heterogeneity. In the first set of models, we examine the spatial extent of the impact of racial heterogeneity by pairing the measure of the focal block along with the measures capturing the area surrounding the block (within ¼ mile, ½ mile, ¾ mile, and 1 mile), separately. These models allow the valuable opportunity to determine whether racial heterogeneity differentially impacts crime based on its spatial scaling, given that prior studies have shown evidence that it is overly simplistic to characterize the relationship as being unequivocally crime-producing (Boessen & Hipp, 2015; Kim & Hipp, 2020). To build on the potential nuance of this relationship, for the second set of models we test for nonlinear relationships between the measures of racial heterogeneity and crime. While we investigated several nonlinear functional forms (e.g., cubic, quartic, square root etc.), we found that a squared function was best reflected by the data; and therefore, we only present and discuss the findings of the models which included squared terms.

Results

Linear Effects of Racial Heterogeneity

We find that racial heterogeneity in the block is significantly and negatively associated with all crime types (Table 2), which runs counter to foundational ideas of social disorganization theory. A block one standard deviation (SD) above the mean for racial heterogeneity implies lower crime rates between −8.1% and −18.8% using the formula [exp(β X SD) − 1] × 100, on an all other things equal basis.⁴ Yet, when we test racial heterogeneity in the area surrounding the focal block (a broader spatial impact) racial heterogeneity reveals a positive association with all crime types. In particular, the models in Table 2 imply higher crime rates between +3.1% and +12.6% for a one standard deviation increase in racial heterogeneity within ¼ mile.⁵

Table 2.

Negative Binomial Regression Models of Racial Heterogeneity and Crime in Blocks and 1/4 Mile Egohoods.

	Agg. assault		Robbery		Burglary		Larceny		M.V. theft
Racial heterogeneity
Block	−0.734	**	−0.990	**	−0.565	**	−0.987	**	−0.402	**
	−13.105		−12.664		−13.807		−27.952		−8.160
0.25 mile egohood	0.288	**	0.657	**	0.173	**	0.331	**	0.171	**
	4.538		6.878		3.767		7.697		3.128
Structural characteristics
Concentrated disadvantage	0.016	**	0.012	**	0.011	**	0.016	**	0.007	**
	9.832		5.168		9.510		15.925		5.101
Concentrated disadvantage (0.25 mile egohood)	0.008	**	0.011	**	−0.003		−0.014	**	0.001
	3.393		3.424		−1.610		−9.617		0.442
% Black	0.005	**	0.000		0.002	^†	0.002	*	−0.001
	4.366		0.047		1.659		2.131		−0.827
% Black (0.25 mile egohood)	0.016	**	0.026	**	0.009	**	0.000		0.012	**
	11.578		13.872		7.272		0.351		8.838
% Latino	−0.003	**	−0.009	**	−0.009	**	−0.010	**	−0.005	**
	−3.961		−8.792		−16.217		−22.757		−7.350
% Latino (0.25 mile egohood)	0.012	**	0.019	**	0.004	**	0.000		0.010	**
	11.898		13.793		5.638		−0.573		12.794
% Occupied housing units	−0.009	**	−0.010	**	−0.008	**	−0.004	**	−0.004	**
	−7.384		−6.357		−8.639		−4.733		−3.456
% Occupied housing units (0.25 mile egohood)	−0.021	**	−0.016	**	−0.003	^†	−0.010	**	−0.011	**
	−10.859		−5.907		−1.941		−8.588		−6.159
% Homeowners	0.002	**	0.004	**	0.005	**	0.002	**	0.002	**
	5.022		6.974		15.516		8.768		5.537
% Homeowners (0.25 mile egohood)	−0.006	**	−0.007	**	−0.004	**	−0.006	**	−0.005	**
	−9.619		−7.316		−8.226		−13.572		−9.924
% Aged 15–29	−0.001		−0.002		0.000		−0.002	*	0.000
	−0.758		−1.179		−0.156		−2.529		0.102
% Aged 15–29 (0.25 mile egohood)	0.004	*	0.004	^†	0.007	**	0.013	**	0.010	**
	2.414		1.854		5.255		12.613		7.333
Land use measures
% Residential	−0.008	**	−0.013	**	−0.006	**	−0.012	**	−0.004	**
	−29.010		−31.586		−27.497		−58.052		−15.169
% Residential (0.25 mile egohood)	0.003	**	0.007	**	0.002	**	0.002	**	0.003	**
	9.144		11.383		5.928		5.910		8.161
% Retail	0.013	**	0.022	**	0.016	**	0.021	**	0.014	**
	29.129		41.391		43.672		61.067		32.415
% Retail (0.25 mile egohood)	0.005	**	0.015	**	0.006	**	0.015	**	0.007	**
	5.557		13.102		9.041		24.038		9.112
% Office	−0.006	**	−0.006	**	−0.004	**	−0.006	**	−0.003	**
	−8.794		−6.465		−7.496		−11.665		−4.551
% Office (0.25 mile egohood)	−0.002		0.001		0.002	^†	0.005	**	0.000
	−1.438		0.651		1.852		5.246		−0.368
% Industrial	0.002	*	0.005	**	0.010	**	0.010	**	0.010	**
	2.282		4.803		16.465		16.802		16.822
% Industrial (0.25 mile egohood)	0.002	*	−0.002	^†	0.003	**	0.004	**	0.003	**
	2.026		−1.682		4.370		5.653		4.583
Intercept	−2.521	**	−5.305	**	−4.251	**	−3.361	**	−5.406	**
	−9.290		−7.126		−19.860		−15.211		−13.334
N	103,168		103,168		103,168		103,168		103,168

Note. T-values are presented below the coefficients. Fixed effects of each city (dummy variables) are included but not shown.

p < .01(two-tail test). *p < .05 (two-tail test). ^†p < .05 (one-tail test).

To further assess the spatial extent of the effects of racial heterogeneity, we estimated models that paired the focal block measure with alternative measures capturing the area surrounding the block (based on varying radii), separately (Appendix Table A1). We find evidence that the broader spatial impact of racial heterogeneity is consistently crime-producing. For models of three crime types (i.e., robbery, burglary, and larceny), racial heterogeneity in the area surrounding the focal block has a significant and positive association for each of the different radii (within ½ mile, within ¾ mile, and within 1 mile). Similarly, racial heterogeneity within ½ mile and within ¾ mile, respectively, show a significant and positive association with aggravated assault, whereas heterogeneity within 1 mile fails to yield a significant association with assault. Racial heterogeneity within ½ mile has a positive association with motor vehicle theft; however, the measures pertaining to within ¾ mile and 1 mile are nonsignificant. Notably, racial heterogeneity pertaining to the focal block is significantly and negatively associated with all forms of crime, consistent with what the models from Table 2 show.

The relationship between racial heterogeneity and crime is indeed spatially patterned. Whereas racial heterogeneity measures capturing the area surrounding the focal block are associated with more crime, we found that more heterogeneity in the block itself is associated with lower crime rates. This pattern of results implies that the highest crime rate combination manifests when a block has “low” heterogeneity (alternatively referred to as homogenous), but is nested within an area with “high” racial heterogeneity. It therefore appears that geographic space plays an important role in determining the contexts whereby people of different racial groups can come together (or do not come together) to informally solve and prevent crime.

Nonlinear Effects of Racial Heterogeneity

Because the broader spatial impact of racial heterogeneity (within ¼ mile to 1 mile) was consistently crime-producing (from Table 2), we use the within ¼ mile measure to approximate the broader spatial impact with respect to examining potential nonlinear effects. Table 3, therefore, shows the results of the models for each crime type, including the predictors of racial heterogeneity for both the focal block and within ¼ mile. Models for each of the different radii (within ½ mile, within ¾ mile, and within 1 mile) are reported in Appendix Table A2. Racial heterogeneity in the block appears to have a nonlinear relationship given that the quadratic is statistically significant in each of the different crime models. Accordingly, we have graphed the nonlinear effect of racial heterogeneity in the block to illustrate interpretable trends (Figure 1a–e).

Table 3.

Negative Binomial Regression Models of Racial Heterogeneity and Crime in Blocks and 1/4 Mile Egohoods (Non-linear).

	Agg. assault		Robbery		Burglary		Larceny		M.V. theft
Racial heterogeneity
Block	−1.147	**	−1.682	**	−1.218	**	−2.213	**	−1.158	**
	−6.944		−7.397		−9.283		−19.945		−7.984
Block (squared)	0.490	*	0.766	**	0.736	**	1.457	**	0.846	**
	2.349		2.612		4.513		10.077		4.689
0.25 mile egohood	1.959	**	1.655	**	2.120	**	2.726	**	1.854	**
	8.027		4.898		10.904		14.907		8.874
0.25 mile egohood (squared)	−1.733	**	−0.820	*	−1.828	**	−2.234	**	−1.578	**
	−6.131		−2.094		−8.346		−10.800		−6.629
Structural characteristics
Concentrated disadvantage	0.015	**	0.012	**	0.010	**	0.015	**	0.006	**
	9.019		5.072		8.405		14.144		4.411
Concentrated disadvantage (0.25 mile egohood)	0.007	**	0.010	**	−0.002		−0.015	**	0.000
	3.224		3.104		−1.465		−9.944		0.037
% Black	0.005	**	0.000		0.002	^†	0.002	*	−0.001
	4.320		0.014		1.774		2.113		−0.620
% Black (0.25 mile egohood)	0.017	**	0.027	**	0.009	**	0.000		0.012	**
	11.978		13.980		7.054		0.292		8.354
% Latino	−0.003	**	−0.009	**	−0.009	**	−0.010	**	−0.005	**
	−4.113		−9.280		−16.152		−22.336		−7.948
% Latino (0.25 mile egohood)	0.013	**	0.020	**	0.005	**	0.001	*	0.012	**
	12.921		14.781		7.500		2.302		14.784
% Occupied housing units	−0.009	**	−0.010	**	−0.009	**	−0.004	**	−0.004	**
	−7.806		−6.518		−9.496		−5.512		−3.715
% Occupied housing units (0.25 mile egohood)	−0.024	**	−0.019	**	−0.004	**	−0.012	**	−0.013	**
	−12.137		−6.622		−2.762		−8.778		−7.037
% Homeowners	0.002	**	0.004	**	0.005	**	0.002	**	0.002	**
	4.771		6.986		14.851		8.253		5.300
% Homeowners (0.25 mile egohood)	−0.006	**	−0.006	**	−0.003	**	−0.004	**	−0.005	**
	−8.630		−6.951		−5.757		−10.172		−8.993
% Aged 15–29	−0.001		−0.002		0.000		−0.001	*	0.000
	−0.764		−1.071		−0.100		−2.014		0.149
% Aged 15–29 (0.25 mile egohood)	0.007	**	0.006	*	0.008	**	0.015	**	0.011	**
	3.724		2.360		6.247		14.179		8.067
Land use measures
% Residential	−0.008	**	−0.013	**	−0.006	**	−0.012	**	−0.004	**
	−28.443		−31.221		−27.117		−55.513		−14.232
% Residential (0.25 mile egohood)	0.004	**	0.009	**	0.003	**	0.003	**	0.004	**
	10.366		13.066		9.587		11.564		11.476
% Retail	0.012	**	0.022	**	0.015	**	0.021	**	0.013	**
	28.582		40.343		40.875		60.131		31.915
% Retail (0.25 mile egohood)	0.006	**	0.017	**	0.007	**	0.017	**	0.008	**
	7.115		14.706		11.177		26.349		11.332
% Office	−0.006	**	−0.006	**	−0.004	**	−0.006	**	−0.003	**
	−8.478		−6.424		−7.832		−11.398		−4.244
% Office (0.25 mile egohood)	−0.001		0.003		0.004	**	0.008	**	0.001
	−0.861		1.374		3.544		8.316		0.648
% Industrial	0.001	^†	0.005	**	0.008	**	0.008	**	0.009	**
	1.719		4.390		12.948		13.946		15.135
% Industrial (0.25 mile egohood)	0.003	**	0.000		0.004	**	0.004	**	0.004	**
	3.071		−0.266		5.330		5.949		5.727
Intercept	−2.647	**	−5.256	**	−4.760	**	−3.765	**	−5.826	**
	−7.507		−5.824		−15.890		−13.755		−9.731
N	101,288		101,288		101,288		101,288		101,288

Note. T-values are presented below the coefficients. Fixed effects of each city (dummy variables) are included but not shown.

p < .01(two-tail test). *p < .05 (two-tail test). ^†p < .05 (one-tail test).

Figure 1.

(a) Racial Ethnic Heterogeneity and Crime in Blocks and ¼ mile Egohoods: Aggravated Assault. (b) Racial Ethnic Heterogeneity and Crime in Blocks and ¼ mile Egohoods: Robbery. (c) Racial Ethnic Heterogeneity and Crime in Blocks and ¼ mile Egohoods: Burglary. (d) Racial Ethnic Heterogeneity and Crime in Blocks and ¼ mile Egohoods: Larceny. (e) Racial Ethnic Heterogeneity and Crime in Blocks and ¼ mile Egohoods: Motor Vehicle Theft.

We find that racial heterogeneity in the block exhibits a decelerating crime-reducing pattern for all crime types. Increases in heterogeneity yield decreases in the predicted crime rate; however, this salutary effect only manifests among blocks with a level of heterogeneity that is below the mean (0 to about 0.40). For blocks with a level of heterogeneity above the mean (>0.40), increases in heterogeneity yield negligible decreases in crime, and this is especially true of the property crimes. Nevertheless, the models imply that low heterogeneity blocks (−1 SD below the mean = 0.20) have predicted crime rates that are appreciably greater than blocks with average heterogeneity (0.41).

Table 3 and Figure 1a to e also reveal the findings for racial heterogeneity pertaining to the spatial area within ¼ mile around the focal block. In each of the crime models, the quadratic is statistically significant and, therefore, this suggests that the broader spatial impact of racial heterogeneity operates in a nonlinear fashion. For all forms of crime (with the exception of robbery) racial heterogeneity within ¼ mile has a strong crime-enhancing effect from the low end of the distribution (0) to the rather high end of the distribution (0.60), but this effect becomes crime-reducing for values greater than the latter. While increases in heterogeneity in surrounding areas yield decreases in crime only among blocks with high heterogeneity in surrounding areas (i.e., >0.60), it should be noted that these blocks still represent some of the highest predicted crime rates. In other words, racial heterogeneity within ¼ mile mainly exerts a crime-producing effect given that the approximate inflection point (0.60) is similar in value to those blocks one standard deviation above the mean on racial heterogeneity (0.64).

Interaction Effects

Given the observed spatial patterns of heterogeneity and crime, we estimated models that test for interaction effects between block racial heterogeneity and ¼, ½, ¾, and 1 mile egohood heterogeneity, respectively. This interaction analysis is implemented to examine how racial heterogeneity in the block can be shaped by heterogeneity in the surrounding area. We report the interaction coefficients in Appendix Table A3. We plotted the predicted crime rates for these interactions, and the patterns were generally similar. We therefore report the results for robbery and burglary as representative of violent and property crime, respectively (see Figures 2 and 3).⁶ Specifically, we visually displayed the effect of racial heterogeneity in the block at different levels of racial heterogeneity in the surrounding ¼ mile (Low = −1 SD and High = +1 SD).

Figure 2.

Interaction: Racial heterogeneity in block and 1/4 mile egohood (Robbery).

Figure 3.

Interaction: Racial heterogeneity in block and 1/4 mile egohood (Burglary).

First, we observed pronounced interaction effects for robbery. For instance, increasing heterogeneity in the block has a negligible effect on robbery when there is low racial heterogeneity in the surrounding area (the orange line). However, increasing heterogeneity in the block exhibits the strongest crime-reducing effect when there is high racial heterogeneity in the surrounding area (the blue line). Thus, the highest risk of robbery happens on a block with low heterogeneity combined with high heterogeneity in the surrounding area. For example, among blocks with low heterogeneity, those with high heterogeneity in the surrounding ¼ mile have about 31 percent more robbery than those with average heterogeneity (Figure 2).

We detect a similar pattern of results for burglary. The effect of racial heterogeneity in the block is crime-reducing when heterogeneity in the surrounding area is at average to high levels (the blue and gray lines), but block heterogeneity has virtually no impact when heterogeneity in the surrounding ¼ area is low (the orange line). In other words, the crime-reducing effect of block heterogeneity is pronounced when combined with high heterogeneity in the surrounding area. This means that blocks with low racial heterogeneity circumscribed by high heterogeneity are at the highest risk of burglary, whereas blocks with high heterogeneity in both the block and the surrounding area have the lowest risk of burglary. Among blocks with high heterogeneity, those with high heterogeneity in the surrounding ¼ mile have about 21 percent fewer burglaries than those with average heterogeneity (Figure 3).

Structural Characteristics and Land Uses

Structural characteristics and land uses effectively produce the same results across the two sets of models estimated. Concentrated disadvantage in the block is significantly and positively associated with each of the crime types examined, while disadvantage in the surrounding ¼ mile exhibits a similar relationship but only for aggravated assault and robbery. Both measures of percent Black are observed to have positive associations with crime, only. The effect of percent Latino differs based on its spatial scaling: whereas percent Latino in the block is negatively associated with the different crime types, percent Latino in the surrounding ¼ mile only shows the opposite effect. We find that both measures of occupied units indicate a negative association with the different crime types. The effect of percent homeowners depends on its spatial scale; specifically, the block measure and surrounding ¼ mile measure reveal positive and negative associations, respectively. Regarding age, there is little evidence that the percent aged 15 to 29 in the block is related to crime; however, for within ¼ mile, this age demographic is positively associated with each crime type.

We observed that residential land use in the block is negatively associated with each of the crime types assessed. Yet, residential land use in the surrounding ¼ mile shows the exact opposite—evidence of a positive association only. Both measures of retail indicate a positive association with each crime type. Similarly, industrial land use consistently yields a positive association with crime, regardless of spatial scale. For office land use, the block measure is negatively related to all forms of crime, whereas the surrounding ¼ mile measure is positively associated with burglary and larceny.

Discussion

This study incorporated theoretical insights of racial heterogeneity and empirically assessed how it is associated with neighborhood crime. The longstanding assumption of social disorganization theory is that racial heterogeneity increases neighborhood crime because it undermines informal social control and collective efficacy to keep the community safe. In contrast, an alternative hypothesis is that the association might be more nuanced. Therefore, we theoretically posited that two interrelated factors should be incorporated: (1) the functional form of the relationship and (2) the spatial extent/scope of the relationship. To empirically test such propositions, we examined relationships between racial heterogeneity and crime in census blocks and surrounding areas (egohoods) with various radii (¼–1 mile). We found that racial heterogeneity in the block is negatively associated with all crime types whereas that of surrounding areas is consistently crime-producing. We also detected important interaction effects between racial heterogeneity in the block and racial heterogeneity in the surrounding ¼ mile. Therefore, a primary contribution of the current study is that it captures a more comprehensive picture of the association between racial heterogeneity and types of crime at different spatial scales. We highlight some key findings below.

First, racial heterogeneity in blocks exhibited a negative association with all types of crime, which is consistent with our theorization based on contact hypothesis. Although this seems inconsistent with the traditional social disorganization perspective, it is not completely new. Smith et al. (2000) found that racial heterogeneity was negatively associated with robbery risk in street segments. Kim (2018) also suggested a crime-reducing effect of racial heterogeneity on violent and property crime risk in place. However, we observed a crime-producing pattern of racial heterogeneity in the surrounding area (¼–1 mile radii). An important implication is that racial heterogeneity in blocks is significantly unique from that in surrounding areas. Therefore, more careful attention should be paid to spatial scaling particularly when testing the relationship between racial heterogeneity and crime.

As contact hypothesis argues, social propinquity among dissimilar race groups can help reduce bias on diverse others and engender a common identity. Social ties and trust among heterogeneous racial groups can be developed through frequent social interactions among dissimilar groups (Gaertner et al., 1996; Pettigrew & Tropp, 2006). Such trust, social ties, and informal social control can reduce crime and disorder in place. Because of the smaller spatial scale of blocks compared to egohoods, residents in blocks may have more opportunities to have direct social interactions to get to know each other, have common social norms, reduce bias, and thus develop informal social control. In contrast, these social interactions may not be captured (nor available) when looking at broader areas. This is because a resident living in one side of the area has fewer chances to have casual interactions, social ties and build trust with diverse others living in the other side of the area ¼–1 mile away from. This result suggests that it is important for future research to incorporate racial heterogeneity at multiple spatial scales for a more comprehensive picture of the relationship between racial heterogeneity and neighborhood crime.

Second, we found evidence of non-linear association between racial diversity and neighborhood crime. Our result suggested that the broader spatial impact of racial heterogeneity operates in a nonlinear fashion. Specifically, racial heterogeneity within ¼ mile has a strong positive effect at lower to the high end of the distribution. However, the effect becomes negative in blocks with even greater heterogeneity. This result implies that racial diversity in broader areas mainly exhibits a crime-producing effect because social cohesion and informal social control among residents in different racial groups is likely reduced. However, as racial diversity increases even further, there may be a sufficiently large amount of each racial group, which provides more opportunities to develop interpersonal contacts and social ties and cohesion with each other and thus reduced amount of crime. Yet, such crime-reducing part of the non-linear pattern seems relatively small in magnitude, in general. Thus, although blocks with very high racial diversity in surrounding areas are at lower risk of crime compared to those with relatively high diversity in broader areas, they still have substantially higher crime rates than those with lower racial diversity in surrounding areas. This suggests that the crime-reducing effect of racial diversity in broader areas may not outweigh the crime-producing effect from reduced social cohesion and informal social control among different racial groups.

A third key finding was that we were able to detect moderation effects between racial heterogeneity in the block and heterogeneity in the surrounding ¼ mile. Racial heterogeneity in the block shows a salutary effect on violent and property crime when heterogeneity in the surrounding area is high. That is, the crime-reducing effect of block heterogeneity is more pronounced when there is high heterogeneity in the surrounding area. This highlights that the broader spatial context is crucial for understanding crime patterns in small spatial units, which is consistent with what previous studies have shown (Boessen & Hipp, 2015; Kim & Hipp, 2020; Wenger, 2019).

Specifically, homogeneous blocks surrounded by highly heterogenous areas (so-called “racial islands”) are at the highest risk of crime. Interestingly, racially heterogenous blocks located in highly heterogenous surrounding areas generally yield the lowest risk of crime. These findings emphasize the benefit of living in a racially heterogenous area within a racially heterogeneous broader area. This is consistent with previous studies of racial boundaries (“spatial edges”) and neighborhood crime (Kim & Hipp, 2021; Legewie, 2018).

For instance, Legewie (2018) defined racial boundaries by abrupt changes in racial composition between adjacent areas. He found that areas with greater differences in racial composition were at higher risk of violent crime. Likewise, Kim and Hipp (2021) defined racial boundary by calculating the sum of compositional differences of five racial groups on both sides of a street segment. They also found that greater difference of racial compositions between the two sides of a street induce a higher risk of violent and property crime. Our interaction findings have a similar implication but are more nuanced. That is, it is not only the difference of racial composition within the block, but also the different level of racial heterogeneity between the focal and surrounding areas. Such nuanced findings provide strong support for incorporating the racial characteristics of surrounding areas or multiple spatial scales.

Despite the contributions, we acknowledge some limitations to the current study. First, although we proposed theoretical reasons for crime-reducing or crime-producing effect of racial heterogeneity, we were unable to specifically identify the mechanisms of social contact or informal social control. It is beyond the scope of the current study to measure the mechanisms that might bring about such relationships. That is, we do not have direct measures of informal social control and therefore speculate the mechanism from the results. Future research may want to capture specific mechanisms of social contact or informal social control among residents to contextualize the dynamic association between racial heterogeneity and neighborhood crime. Moreover, an obvious extension is for future studies to assess how the racial composition of the ambient population shapes crime patterns (not just the racial composition among residents).

Second, the current study is designed to be cross-sectional and therefore endogeneity is a challenge for studying the relationships between racial heterogeneity and neighborhood. That is, it is possible that crime at a previous time point has impacted structural characteristics at present date, including racial composition of a neighborhood (Hipp, 2010). Therefore, a longitudinal design is preferred to detect the association between racial heterogeneity and crime while accounting for such temporally reciprocal relationships. Future research should employ a longitudinal research design to examine how changes in racial heterogeneity can affect changes in neighborhood crime over time.

Another limitation concerns the study area. The data employed in the current study is limited to the Southern California region, which may limit the generalizability of the results compared to other regions in the U.S. Specifically, the Southern California region is known to be more racially diverse with relative higher Asian and Latino populations. Therefore, our findings may not be directly applicable to other regions with different racial compositions. A natural extension is for future studies to include neighborhoods from several regions of the U.S.

In conclusion, the current study examines racial heterogeneity and neighborhood crime. We propose two theoretical perspectives on possible crime-producing and crime-reducing effects of racial heterogeneity within the frameworks of social disorganization and contact hypothesis. We empirically test different consequences by functional forms of racial diversity at various spatial scales. Our results suggest that racial diversity has a crime-reducing effect at the smaller spatial scale (blocks) whereas it generally exhibits a crime-producing pattern (with a weak nonlinear, positive to negative effect) when looking at broader areas. Furthermore, our interaction results revealed that “racial islands” are at the highest risk of violent and property crime, whereas racially heterogenous blocks located in highly heterogenous surrounding areas are at the lowest risk. Therefore, the present study emphasizes the need to unpack the relationship between racial heterogeneity and crime with regards to spatial scaling, functional form, and moderation.

Footnotes

Appendix A

Table A3.

Interactions.

	Agg. assault		Robbery		Burglary		Larceny		M.V. theft
Racial heterogeneity
Block	0.130		−0.150		0.179	*	0.225	**	0.281	**
	1.264		−1.010		2.172		2.845		3.171
0.25 mile egohood	1.190	**	1.446	**	0.944	**	1.462	**	0.904	**
	10.785		9.494		10.833		18.636		9.410
Interaction term	−2.041	**	−1.929	**	−1.706	**	−2.673	**	−1.628	**
	−9.939		−6.615		−10.394		−17.223		−9.234
Intercept	−2.899	**	−5.697	**	−4.454	**	−3.718	**	−5.706	**
	−10.566		−7.627		−20.726		−16.731		−14.023
N	103,168		103,168		103,168		103,168		103,168

Note. T-values are presented below the coefficients. Fixed effects of each city (dummy variables) and other controls were included but not shown.

p < .01 (two-tail test). *p < .05 (two-tail test).

Declaration of Conflicting Interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author(s) received no financial support for the research, authorship, and/or publication of this article.

ORCID iDs

Young-An Kim

James C. Wo

Notes

Author Biographies

Young-An Kim is an Assistant Professor in the College of Criminology and Criminal Justice at Florida State University. His research interests focus on various areas such as neighborhoods and crime, criminology of place, immigration and crime, and geo-spatial analysis. Besides criminology, he is interested in sociology of health, urban sociology, and quantitative research methods.

James C. Wo is an Assistant Professor in the Department of Sociology and Criminology at the University of Iowa, and a Research Fellow at the University of Iowa’s Public Policy Center. His research interests include neighborhoods and crime, local institutions and organizations, land uses, and quantitative research methods.

References

Allport

G. W.

(1954). The nature of prejudice. Addison-Wesley.

Baumer

E. P.

(2002). Neighborhood disadvantage and police notification by victims of violence. Criminology, 40(3), 579–616.

Berg

M. T.

Lauritsen

J. L.

(2016). Telling a similar story twice? NCVS/UCR convergence in serious violent crime rates in rural, suburban, and urban places (1973–2010). Journal of Quantitative Criminology, 32(1), 61–87.

Bernasco

Block

(2011). Robberies in Chicago: A block-level analysis of the influence of crime generators, crime attractors, and offender anchor points. Journal of Research in Crime and Delinquency, 48(1), 33–57.

Boessen

Chamberlain

A. W.

(2017). Neighborhood crime, the housing crisis, and geographic space: Disentangling the consequences of foreclosure and vacancy. Journal of Urban Affairs, 39(8), 1122–1137.

Boessen

Hipp

J. R.

(2015). Close-Ups and the scale of ecology: Land uses and the geography of social context and crime. Criminology, 53(3), 399–426.

Bursik

Grasmick

(1993). Neighborhoods and crime: The dimensions of effective community cotrol. Lexington Books.

Bursik

R. J.

(1988). Social disorganization and theories of crime and delinquency: Problems and prospects. Criminology, 26(4), 519–552.

Chen

Rafail

(2020). Do housing vacancies induce more crime? A spatiotemporal regression analysis. Crime and Delinquency, 66(11), 1579–1605.

10.

Contreras

(2017). A block-level analysis of medical marijuana dispensaries and crime in the city of Los Angeles. Justice Quarterly, 34(6), 1069–1095.

11.

Ellison

C. G.

Powers

D. A.

(1994). The contact hypothesis and racial attitudes among Black Americans. Social Science Quarterly, 75, 385–400.

12.

Enos

(2017). The space between us: Social geography and politics. Cambridge University Press.

13.

Felson

Boba

. (2010). Crime and everyday life (4th ed.). SAGE Publications, Inc.

14.

Gaertner

Rust

Dovidio

Bachman

Anastasio

(1996). The contact hypothesis: The role of common ingroup identity on reducing intergroup bias among majority and minority members. In Nye

J. L.

Brower

A. M.

(Eds.), What’s social about social cognition? (pp. 230–260). SAGE Publications.

15.

Gibbs

J. P.

Martin

W. T.

(1962). Urbanization, technology, and the division of labor: International patterns. American Sociological Review, 27(5), 667–677.

16.

Groff

E. R.

Lockwood

(2014). Criminogenic facilities and crime across street segments in Philadelphia uncovering evidence about the spatial extent of facility influence. Journal of Research in Crime and Delinquency, 51(3), 277–314.

17.

Haberman

C. P.

Kelsay

J. D.

(2021). The topography of robbery: Does slope matter? Journal of Quantitative Criminology, 37, 625–645.

18.

Hilbe

J. M.

(2007). Negative binomial regression. Cambridge University Press.

19.

Hipp

J. R.

(2007). Block, tract, and levels of aggregation: Neighborhood structure and crime and disorder as a case in point. American Sociological Review, 72(5), 659–680.

20.

Hipp

J. R.

(2010). A dynamic view of neighborhoods: The reciprocal relationship between crime and neighborhood structural characteristics. Social Problems, 57(2), 205–230.

21.

Hipp

J. R.

(2011). Spreading the wealth: The effect of the distribution of income and race/ethnicity across households and neighborhoods on city crime trajectories. Criminology, 49(3), 631–665.

22.

Hipp

J. R.

Boessen

(2013). “Egohoods as waves washing across the City: A new measure of ‘Neighborhoods.’” Criminology, 51(2), 287–327.

23.

Hipp

J. R.

Bates

Lichman

Smyth

(2019). Using social media to measure temporal ambient population: Does it help explain local crime rates? Justice Quarterly, 36(4), 718–748.

24.

Hipp

J. R.

Kim

Y. A.

J. C.

(2020). Micro-scale, meso-scale, macro-scale, and temporal scale: Comparing the relative importance for robbery risk in New York City. Justice Quarterly, 38(5), 767–791.

25.

Hipp

J. R.

Kubrin

C. E.

(2017). From Bad to worse: How changing inequality in nearby areas impacts local crime. RSF The Russell Sage Foundation Journal of the Social Sciences, 3(2), 129–151.

26.

Kim

Y. A.

(2018). Examining the relationship between the structural characteristics of place and crime by imputing census block data in street segments: Is the pain worth the gain? Journal of Quantitative Criminology, 34(1), 67–110.

27.

Kim

Y. A.

Hipp

J. R.

(2018). Physical boundaries and city boundaries: Consequences for crime patterns on street segments? Crime and Delinquency, 64(2), 227–254.

28.

Kim

Y. A.

Hipp

J. R.

(2020). Street egohood: An alternative perspective of measuring neighborhood and spatial patterns of crime. Journal of Quantitative Criminology, 36(1), 29–66.

29.

Kim

Y. A.

Hipp

J. R.

(2021). Both sides of the street: Introducing measures of physical and social boundaries based on differences across sides of the street, and consequences for crime. Journal of Quantitative Criminology. Advance online publication. https://doi.org/10.1007/s10940-020-09484-4

30.

Kim

Y. A.

Hipp

J. R.

Kubrin

C. E.

(2019). Where they live and go: Immigrant ethnic activity space and neighborhood crime in southern California. Criminal Justice Journal, 64, 1–12.

31.

Kubrin

C. E.

(2000). Racial heterogeneity and crime: Measuring static and dynamic effects. Research in Community Sociology, 10, 189–219.

32.

Kubrin

C. E.

Hipp

J. R.

(2016). Do fringe banks create fringe neighborhoods? Examining the spatial relationship between fringe banking and neighborhood crime rates. Justice Quarterly, 33(5), 755–784.

33.

Kubrin

C. E.

Hipp

J. R.

Kim

Y. A.

(2018). Different than the sum of its parts: Examining the unique impacts of immigrant groups on neighborhood crime rates. Journal of Quantitative Criminology, 34(1), 1–36.

34.

Kubrin

C. E.

J. C.

(2015). Social disorganization Theory’s greatest challenge: Linking structural characteristics to crime in socially disorganized communities. In Piquero

A. R.

(Ed.), The handbook of criminological theory (pp. 121–136). John Wiley & Sons, Inc.

35.

Lee

Bean

F. D.

(2010). The diversity paradox: Immigration and the color line in twenty-first century America. Russell SAGE Foundation.

36.

Legewie

(2018). Living on the edge: Neighborhood boundaries and the spatial dynamics of violent crime. Demography, 55, 1957–1977.

37.

Marschall

M. J.

Stolle

(2004). Race and the city: Neighborhood context and the development of generalized trust. Political Behavior, 26(2), 125–153.

38.

Osgood

D. W.

(2000). Poisson-Based regression analysis of aggregate crime rates. Journal of Quantitative Criminology, 16(1), 21–43.

39.

Peterson

R. D.

Krivo

L. J.

(2010). Divergent social worlds: Neighborhood crime and the racial-spatial, Divide. Russell SAGE Foundation.

40.

Peterson

R. D.

Krivo

L. J.

Harris

M. A.

(2000). Disadvantage and neighborhood violent crime: Do local institutions matter? Journal of Research in Crime and Delinquency, 37, 31–63.

41.

Pettigrew

T. F.

Tropp

L. R.

(2006). A meta-analytic test of intergroup contact theory. Journal of Personality and Social Psychology, 90(5), 751–783.

42.

Pratt

T. C.

Cullen

F. T.

(2005). Assessing macro-level predictors and theories of crime: A meta-analysis. Crime and Justice, 32, 373–450.

43.

Sampson

R. J.

Groves

W. B.

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

44.

Sigelman

Welch

(1993). The contact hypothesis revisited: Black-White interaction and positive racial attitudes. Social Forces, 71, 781–795.

45.

Skogan

W. G.

(1990). Disorder and decline: Crime and the spiral of decay in American neighborhoods. University of California Press.

46.

Smith

W. R.

Frazee

S. G.

Davison

E. L.

(2000). Furthering the integration of routine activity and social disorganization theories: Small units of analysis and the study of street robbery As a diffusion process. Criminology, 38(2), 489–524.

47.

Stucky

T. D.

Ottensmann

J. R.

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

48.

Sun

I. Y.

Triplett

Gainey

R. R.

(2004). Neighborhood characteristics and crime: A test of Sampson and Groves’ model of social disorganization. Western Criminology Review, 5, 1–16.

49.

Taylor

R. B.

(1997). Social order and disorder of street blocks and neighborhoods: Ecology, microecology, and the systemic model of social disorganization. Journal of Research in Crime and Delinquency, 34(1), 113–155.

50.

Warner

B. D.

Pierce

G. L.

(1993). Reexamining social disorganization theory using calls to the police as a measure of crime. Criminology, 31, 493–517.

51.

Warner

B. D.

Rountree

P. W.

(1997). Local social ties in a community and crime model: Questioning the systemic nature of informal social control. Social Problems, 44(4), 520–536.

52.

Weisburd

Groff

E. R.

Yang

S.-M.

(2012). The criminology of place: Street segments and our understanding of the crime problem. Oxford University Press.

53.

Weisburd

Shay

Amram

Zamir

(2017). The relationship between social disorganization and crime at the micro geographic level: Findings from Tel Aviv-Yafo using Israeli census data. Advances in Criminological Theory, 22, 97–120.

54.

Weisburd

White

Wooditch

(2020). Does collective efficacy matter at the micro geographic level? Findings from a study of street segments. The British Journal of Criminology, 60, 873–891.

55.

Wenger

M. R.

(2019). Clarifying the relationship between racial diversity and crime: Neighborhoods versus cities. Crime and Delinquency, 65(11), 1513–1536.

56.

Wicker

A. W.

(1987). Behavior settings reconsidered: Temporal stages, resources, internal dynamics, context. In Stokels

Altman

(Eds.), Handbook of environmental psychology (pp. 613–654). Wiley-Interscience.

57.

J. C.

(2019). Understanding the differential effects of land uses on crime: An examination across Philadelphia neighbourhoods. The British Journal of Criminology, 59(6), 1432–1454.

58.

J. C.

Hipp

J. R.

Boessen

(2016). Voluntary organizations and neighborhood crime: A dynamic perspective. Criminology, 54(2), 212–241.

59.

J. C.

Park

(2020). An examination of schools, social ecological factors, and neighbourhood crime. The British Journal of Criminology, 60(4), 851–872.

60.

Zahnow

(2018). Mixed Land use: Implications for violence and property crime. City and Community, 17(4), 1119–1142.