Gentrification and Crime: Understanding Neighborhood Change Through Third Places and Demolitions

Abstract

This study examines how gentrification reshapes neighborhood crime through two underexplored dimensions: third places and demolitions. Traditional research on gentrification and crime has often relied on broad socioeconomic indicators, such as property values or median income, but these measures fail to capture the lived and place-based processes of neighborhood change. Drawing on place-in-neighborhood theory and routine activities theory, this research conceptualizes gentrification as a multidimensional transformation of social and physical space. Using data from Norfolk, Virginia (2015–2019), hierarchical linear models were employed to assess how the emergence of alcohol-licensed third places (e.g., bars, restaurants) and the issuance of demolition permits influenced community-level crime rates. Results indicate that third places decrease crimes against society but increase property crimes, while demolitions exert mixed effects depending on the offense type. By integrating spatial statistics and multilevel modeling, this study demonstrates the value of real-time municipal data for understanding neighborhood transformation. These results suggest that revitalization cannot be treated as a one-dimensional solution to urban crime. This work reframes gentrification as a contested process whose influence on crime depends on place, race, and neighborhood transformation.

1. Introduction

Gentrification is a multifaceted process characterized by the influx of wealthier groups into historically lower-income neighborhoods, leading to economic growth, increased property values, and infrastructural improvements but also raises concerns about displacement and cultural erosion (Barton 2016; Meltzer and Capperis 2017; DeSena 2018). Two key components of gentrification are demolitions and renewal through the establishment of third places (Papachristos et al. 2011; Wo 2016; Weber et al. 2006). Communities with already established third places—public and semi-public spaces such as coffee shops, libraries, and community centers—serve as social anchors but are often the targets of redevelopment (Chum 2015; Ding et al. 2016). Demolitions, by contrast, play a more disruptive role in physically restructuring gentrified neighborhoods, potentially erasing historical sites and altering communal dynamics. Existing studies highlight displacement and the loss of neighborhood stability as significant consequences of gentrification (Chum 2015; Ding et al. 2016). Some research argues that gentrification can have positive effects on communities, such as increased economic opportunities and improved living standards for both current and future residents, despite its costs (Powell 1999; Freeman and Braconi 2004).

Criminologists have long examined neighborhood change and its relationship to crime, yet the assumption that gentrification reduces crime remains empirically debated (Barton 2016; McDonald 1986; Papachristos et al. 2011; Kreager et al. 2007). Many studies rely on census data to track neighborhood transformation, but decennial statistics (Covington and Taylor 1989; Lee 2010) often fail to capture gentrification as a dynamic, ongoing process, lacking the temporal granularity necessary for real-time assessments. Furthermore, a significant limitation in gentrification research is the sole reliance on singular indicators such as median income or property values (McDonald 1986; Covington and Taylor 1989). While these macro-level metrics provide valuable insights, they fail to reflect gentrification’s fluid and multidimensional nature. Additionally, some studies employ cross-sectional designs (Kreager et al. 2007), which are ill-suited for analyzing gentrification as an ongoing process. While qualitative approaches (Lloyd 2015) offer rich narratives of residents’ experiences, they often prioritize individual perspectives over the structural forces—such as business investments, policymaking, and government interventions—that actively shape neighborhood change.

By analyzing data from public repositories in collaboration with official data this paper presents models capturing real-time neighborhood changes, overcoming the issue of insufficient temporal granularity (Covington and Taylor 1989; Lee 2010). Additionally, it moves beyond relying solely on traditional macro-level indicators such as median income and property values (McDonald 1986; Covington and Taylor 1989) by incorporating third places, such as cafes, parks, and community hubs and demolitions, both of which serve as critical yet often overlooked signals of social and physical transformation. Methodologically, the study employs hierarchical linear modeling (HLM) instead of cross-sectional designs (Kreager et al. 2007), allowing for a more dynamic analysis of gentrification as an ongoing process and its evolving impact on crime. Drawing on municipal, public and longitudinal data, this study examines how gentrification shapes neighborhood crime through the mechanisms of demolitions and third places.

2. Gentrification and Neighborhood Change

2.1. Gentrification

Gentrification is the physical and social transformation of a neighborhood often driven by municipal and capitalist motives (Rose 2002; Shaw and Hagemans 2015) to improve its socioeconomic status through increased homeownership, luxury amenities, and income-generating businesses (Barton 2016; Meltzer and Capperis 2017; DeSena 2018). Some define it as collaboration between state agents and private investors to create lasting community change (Hightower and Fraser 2020). Others take a broader view, describing it as any neighborhood transformation that reduces poverty, raises median income, and introduces retail and other capitalist elements (Shaw and Hagemans 2015; Meltzer and Capperis 2017). Not only the conceptualization, but the impact of gentrification has led to considerable debate. Some studies emphasize the effects of displacement and disruption of established community stability as major consequences of neighborhood gentrification (Chum 2015; Ding et al. 2016), while others present contrasting perspectives (Freeman and Braconi 2004). Proponents of gentrification emphasize the theoretical social benefits and opportunities available to all residents, existing and future, irrespective of associated costs (Powell 1999; Freeman and Braconi 2004).

Decisions to gentrify involve stakeholders, including state and city officials, each with distinct priorities (Logan and Molotch 2013), but often target specific areas to address crime-related concerns (Doering 2020). This process is justified as an effort to enhance safety and reduce concentrated crime, with public entities prioritizing intervention in neighborhoods deemed problematic rather than those already thriving. Local governments and private developers frequently cite crime reduction as a primary reason for gentrification (Doering 2020), portraying it as an initiative by public and private investors to improve safety.

Arguments for gentrification often hinge on the belief that middle- and upper-class neighborhoods experience lower crime rates, and that attracting higher socioeconomic residents will promote prosocial behavior through their resources and connections with public institutions (Wykes 2023; Moore et al. 2009). Despite longstanding interest in the relationship between neighborhood change and crime, the notion that gentrification reduces crime has generated mixed results (Barton 2016; Papachristos et al. 2011; Kreager et al. 2007). While some studies indicate that gentrifying areas may see reduced or stable crime rates, others suggest that crime can increase because of gentrification (Covington and Taylor 1989; Lee 2010).

2.2. Gentrification and Crime

Despite ongoing debates, research on the relationship between gentrification and crime remains inconclusive. Some studies suggest that gentrification either reduces or has no measurable impact on crime rates (Barton 2016; McDonald 1986), while others find that crime increases as neighborhoods gentrify (Covington and Taylor 1989; Lee 2010). Quantitative analyses often indicate a rise in certain offenses, particularly robbery and larceny, following gentrification (Covington and Taylor 1989). However, early studies were constrained by methodological limitations, frequently using census-based indicators—such as housing values and income levels—as proxies for gentrification. For instance, McDonald (1986) found no effect on property crime in gentrified neighborhoods but noted that any prosocial effects were short-lived. Similarly, Kreager et al. (2007) observed that gentrification had minimal influence on property crime in Seattle, with only about 10% of changes attributable to gentrification-related factors.

More recent research suggests that gentrification may not substantially alter objective crime rates but can influence perceptions of safety. Albertson et al. (2020) reported that crime levels remained stable following the redevelopment of a low-income housing complex in Seattle, yet residents reported feeling safer in the revitalized area. Other studies similarly indicate that positive perceptions of neighborhood improvement can reduce fear of crime, even when actual crime rates remain unchanged (Arnold 2011).

Existing research shows that the relationship between gentrification and crime is complex, yet most empirical approaches still rely on static indicators, showing the potne5ineed for dynamic measures that capture how gentrification unfolds in real time. Traditionally, gentrification has been measured using indicators that capture shifts in neighborhood socioeconomic status, housing markets, and population composition. Common metrics include rising property values, increased median household income, and higher educational attainment among residents, which signal an influx of middle- and upper-class populations (Freeman 2005; Barton 2016). Demographic changes such as reductions in racial and ethnic minority populations or the displacement of long-term, low-income residents are also frequently employed to operationalize gentrification (Hwang and Sampson 2014). In addition, researchers often rely on markers of physical investment—such as building permits, new business openings, and real estate development—to track the transformation of historically divested neighborhoods (Zuk et al. 2018).

This study conceptualizes gentrifications a dynamic process. A transformation of both physical and social spaces, measured through two indicators: third places and demolitions. Third places are public or semi-public settings such as coffee shops, barbershops, libraries, and community centers, all which have historically fostered trust, belonging, and informal social control (Oldenburg and Brissett 1982; Wo 2016; Zukin 2016). This change shows investment within an area often followed by gentrification-related effects such as influx of affluent patrons, increased socioeconomic statuses and prosocial community benefits. These shifts alter the social fabric of neighborhoods, weakening long-standing networks that once promoted cohesion and resilience. Demolitions, by contrast, represent a more disruptive dimension of gentrification. The removal of older housing and community landmarks frequently displaces existing residents, erases collective memory, and undermines the cultural and physical foundations of neighborhood life. Together, third places and demolitions capture the dual nature of gentrification as both reinvestment and disruption. Together, third places and demolitions show the dual nature of gentrification: the simultaneous weaving and unweaving of community.

2.3. Third Places and Gentrification

Third places are informal gathering spaces that foster social interaction and community cohesion. These include coffee shops, barbershops, and parks—spaces where individuals can engage in casual, unstructured conversations (Oldenburg and Brissett 1982). They are particularly crucial in lower-income communities, where they serve as accessible venues for networking and support (Wo 2016). Gentrification often leads to the displacement of third places that cater to long-standing residents. As property values and commercial rents increase, these community hubs are frequently replaced by businesses that serve new, more affluent populations (Zukin 2016).

Third places have the potential to advertise as welcoming establishments for residents, both existing and future, across all community groups. In the context of gentrification, however, third places may hold completely different responsibilities, especially if these new third places are replacing preexisting structures. For example, barbershops can be argued as places of sociability and comradery, indicating properties of third places. Where the pre-gentrified area may rely on this barbershop to be a place of neutrality in response to neighborhood angst and delinquency, however, a replacement barbershop will not have the same ability to create social ties or reputation due to new people, rules, and atmosphere (Ruffin and Martin 2024). Although they are the same type of third place (e.g., barbershops), their role within the community has changed indirectly due to gentrification. The loss of communal third places disrupt established social networks and can contribute to feelings of alienation among displaced residents. Studies show that third places contribute to crime reduction by fostering informal social control and neighborhood stability (Papachristos et al. 2011). Their removal can lead to increased social fragmentation and a decline in neighborhood trust (Hwang and Sampson 2014).

While third places generally include a wide range of informal gathering spaces, alcohol-related third places, such as bars, lounges, and restaurants, are tied to processes of gentrification. These establishments are often among the first signs of commercial reinvestment and social rebranding within transitioning neighborhoods (MacDonald and Stokes 2020). They attract new, more affluent consumers and serve as cultural markers of urban renewal, signaling that an area is safe, up and coming, or open for business (Izenberg et al. 2018). Consequently, alcohol-licensed third places not only reflect economic change but actively participate in reshaping neighborhood identity and social dynamics.

2.4. Demolition and Gentrification

Demolition is a prevalent mechanism when reshaping the environment as it has the power to dismantle and destroy cultural and historical pillars within the community. It is also another aspect of gentrification where measurement and inclusion in models are important for capturing the process. As redevelopers acquire properties and attempt to make room for new businesses and housing, demolition becomes vital in redesigning the targeted landscape (Shaw and Hagemans 2015). Additionally, demolition proves to be a useful method in removing structures that show to be physically and socially “unsafe” (Goetz 2011). Weber et al. (2006) characterizes demolitions as physical indicators of gentrification. Although not all buildings are marked for destruction, gentrification’s scope includes having enough change in the physical makeup of the environment to welcome new and more affluent construction and culture.

However, critics argue that demolition disproportionately affects lower-income and minority communities, accelerating displacement and erasing cultural landmarks (Weber et al. 2006; Shaw and Hagemans 2015). The demolition of public housing and other affordable living spaces has been linked to increased homelessness and housing instability (Goetz 2011). In many cases, redevelopment projects fail to provide adequate replacement housing, forcing displaced residents to relocate to areas with fewer economic opportunities (Fullilove 2016). Additionally, the destruction of historically significant sites can sever connections to a community’s past, diminishing its collective memory and cultural identity. This erasure can contribute to the marginalization of long-term residents and reshape the character of urban spaces in ways that prioritize economic interests over community needs (Zukin 2016).

2.5. Theoretical Framework

This study draws on place-in-neighborhood theory (Tillyer et al. 2021), which emphasizes that the criminogenic or protective influence of specific place types depends on the surrounding neighborhood context. The PIN framework proposes that the same types of places, often labeled as crime generators, such as bars or liquor stores, do not produce crime uniformly. Instead, their effects depend on neighborhood conditions such as collective guardianship, target availability, and the density of offenders. Alcohol-related third places occupy a dual role in the urban landscape. On one hand, they enhance neighborhood vitality and informal guardianship through increased foot traffic and social interaction. On the other, they extend hours of activity, elevate pedestrian density, and may generate disorder or opportunity structures for crime, particularly in mixed-use or nightlife-oriented areas (Pennay et al. 2014; Haley et al. 2023). Simultaneously, economic restructuring introduces new forms of capital investment and shifts the socioeconomic composition of neighborhoods, bringing residents and patrons with different routines, resources, and perceptions of safety. These social and spatial reconfigurations can produce paradoxical outcomes: strengthening social control in some areas while eroding it in others, depending on how redevelopment interacts with preexisting neighborhood area and culture.

Additionally, this paper aligns with routine activities theory (Cohen and Felson 1979), which states that crime occurs when motivated offenders encounter vulnerable targets in the absence of capable guardians. From a routine activities perspective, neighborhood businesses and third places can influence activity patterns, shaping when and where crime is more likely to happen. That, crime can be reduced by modifying neighborhood environments, increasing surveillance, and limiting offender anonymity. Theoretically, gentrification aligns with this by restructuring communities, introducing higher-income residents, and enhancing law enforcement presence (Johnson et al. 2014). These changes are seen as situational crime preventors, suggesting that environmental modifications—such as increased security measures, business investments, and urban redesign—hinder criminal activity (Hightower and Fraser 2020; Zimmer 2022).

2.6. Considering Critical Race Theory

Finally, while not the primary model, this study considers critical race theory insights. Critical race theory (CRT) provides a framework for understanding how race and structural inequalities shape the consequences of gentrification. CRT argues that racial disparities are embedded within social, political, and economic systems, influencing who benefits from urban redevelopment and who is displaced (Crenshaw 2010; Delgado and Stefancic 2023). Gentrification often operates within these racialized structures, disproportionately displacing communities of color while privileging middle- and upper-class, predominantly White newcomers. This is not merely a byproduct of economic development but a continuation of historical practices such as redlining and urban renewal, which systematically marginalized Black and Brown communities (Alexander 2012). Increased law enforcement presence in gentrifying neighborhoods also follows racialized patterns, as policing strategies shift from neglecting low-income areas to aggressively policing them once redevelopment begins. This results in heightened surveillance and criminalization of long-term residents, reinforcing racial disparities in arrests, sentencing, and incarceration (Alexander 2012; Crenshaw 2010).

CRT challenges the assumption that gentrification inherently improves communities by highlighting its racialized consequences. It critiques policies that prioritize economic development over social equity and calls for alternative approaches that center the needs of historically marginalized residents. Rather than displacing vulnerable populations, equitable urban planning should focus on community investment, affordable housing protections, and policies that prevent discriminatory policing practices.

2.7. The Current Study

Research on gentrification and crime has produced mixed results, with some studies suggesting that redevelopment reduces crime through increased investment and social control, while others argue it displaces poverty and crime into adjacent areas (Barton 2016; Papachristos et al. 2011; Covington and Taylor 1989; Lee 2010). Much of this work relies solely on traditional indicators of gentrification such as rising property values, median household income, or shifts in racial and socioeconomic composition (Freeman 2005; Hwang and Sampson 2014). While these measures capture broad economic and demographic change, they often overlook the restructuring that accompanies redevelopment. Recent scholarship has begun to call for more nuanced approaches that examine how neighborhood institutions, public spaces, and physical infrastructure shape the lived experience of gentrification and its relationship to crime (Zuk et al. 2018; Hightower and Fraser 2020). However, relatively few studies integrate alternative place-based indicators, such as the transformation of community gathering spaces or the material erasure of neighborhoods through demolitions.

By examining third place and demolition activity alongside socioeconomic and zoning characteristics, this study responds to these limitations and attempts a more comprehensive understanding of how gentrification reshapes neighborhood life and crime dynamics. Specifically, it operationalizes gentrification through two indicators: alcohol-licensed third places (such as bars, cafes, and restaurants) and demolitions. Additionally, grounded in critical race theory, this paper seeks to highlight if there are disproportionalities in the effects of gentrification on historically marginalized communities (e.g., Black) affecting their community crime rates. This exploratory study builds on the aforementioned work by moving beyond traditional economic and demographic indicators of gentrification to adopt a place-based approach. Whereas most research measures neighborhood change through shifts in property values, median income, or racial composition, this project emphasizes how gentrification restructures the physical and social fabric of communities.

3. Methodology

3.1. Setting

The research setting for this study is Norfolk, Virginia, a coastal city with a population of approximately 240,000 (US Census Bureau 2010). Norfolk is characterized by its diverse population (~46% White, ~40% Black, ~13% Latinx or other), and a significant proportion of residents (45%) were homeowners, while the rest were renters. The city’s median household income is around $46,000, with about 15% of the population living below the poverty line and an unemployment rate of 10%.

Table 1. 2010 Census Statistics (Norfolk, VA).

Population and Race
Population	242,803
White ab	46.3%
Black ab	40.6%
Asian ab	0.3%
American Indian and Alaskan Native ab	3.7%
Native Hawaiian and Other Pacific Islander ab	0.1%
Hispanic or Latino ab	8.4%
Two or More Races b	5.9%
Age and Sex
Percent Women b	47.8%
Persons under 5 years b	6.5%
Persons under 18 years b	19.6%
Persons 65 years and Older b	11.2%
Housing and Economics
Households	89,398
Owner occupied housing	43.5%
Person per Household d	2.43
Median Household Income d	$53,026.00
Median Rent cd	$1077.00
Female Headed Households	16%
Percent in Poverty bd	17.8%
Education
High School d	88.1%
Bachelors d	30.0%

^a Single-race percentages are from those who identify only as the categorized race. ^b Estimates are not comparable to other geographic levels due to methodology differences that may exist between different data sources. ^c Median rent is estimated from the gross rent calculated from 2016–2020. ^d Estimates are from the 2016–2020 ACS 5-year estimates.

shows the 2010 Norfolk Census Statistics. Norfolk, VA, has had historical focus on infrastructural changes arranged by the city. Norfolk entrenched patterns of neighborhood inequality that continue to inform its redevelopment strategies today. The Norfolk Redevelopment and Housing Authority (NRHA) has long played a central role in reshaping the city’s landscape, frequently at the expense of low-income Black residents who face food deserts, displacement, and limited housing security (Murphy 2020; Finn 2021). Several local examples illustrate Norfolk’s history of redevelopment-driven displacement. In the Ghent District, once home to thriving Black communities, mid-20th-century urban renewal projects rebranded the neighborhood as blighted, facilitating widespread displacement under the banner of conservation and revitalization (Turken 2020). Similarly, the demolition of Moton Circle, a public housing complex built in 1952, displaced hundreds of predominantly Black families in 2010 under promises of eventual return that never materialized (The Virginian-Pilot 2010). More recently, the St. Paul’s Project echoes these patterns: marketed as a modern mixed-income development, it threatens to displace thousands of long-term, low-income Black residents with little assurance of equitable relocation or return (Finn 2021).

Block groups, which are smaller than census tracts but larger than individual parcels, were chosen as the units of analysis. This decision is informed by the need to capture neighborhood-level change without ignoring important neighborhood dynamics that census tracts may miss (White et al. 2015). Norfolk consists of 196 block groups, with 185 eligible block groups. Block groups were omitted either because the data did not report median household income for the area, or because the area was zoned for industrial purposes. A longitudinal dataset spanning five years results in 925 observations (block groups over time) for analysis. Below conceptualizes the measures used—a breakdown is shown in Table 2.

3.2. Crime Data

Crime data provided by the Norfolk Police Department (NPD) spans a five-year period from 2015 to 2019. This comprehensive dataset encompasses real-time incidents, classified according to the National Incident-Based Reporting System (NIBRS) Group A categories, including Crimes Against Persons (n = 10,488) (such as assault, homicide, kidnapping, and sex offenses), Crimes Against Property (n = 65,665) (encompassing burglary, robbery, vandalism, fraud, larceny, motor vehicle theft, and other property-related crimes), and Crimes Against Society (n = 7240) (involving drug offenses, gambling, pornography, prostitution, and weapons violations). The data is aggregated and analyzed by year and block group, with rates calculated as incidents per 100 people to facilitate comparisons across different spatial areas and over time. Table 2 Shows a breakdown of the crime data across categories.

Table 2. Breakdown of Norfolk, VA Crimes by NIBRS Categories.

Crimes Against Persons a
Year	Assault		Homicide			Sex Offenses		Total
2015	2029 (89.7)		34 (1.5)			198 (8.8)		2261
2016	2070 (92.1)		42 (1.9)			134 (6.0)		2246
2017	1942 (92.1)		34 (1.6)			132 9 (6.3)		2108
2018	1792 (91.8)		37 (1.9)			123 (6.3)		1952
2019	1758 (91.5)		36 (1.9)			127 (6.6)		1921
Total	9591		183			714		10,488
Crimes Against Property
Year	Burglary	Vandalism	Fraud	Larceny		Robbery	MVT		Other b	Total
2015	1256 (9.7)	3065 (23.6)	623 (4.8)	6590 (50.7)		502 (3.9)	831 (6.4)		128 (0.9)	12,995
2016	1280 (8.5)	3173 (21.2)	563 (3.8)	8416 (56.2)		502 (3.3)	875 (5.8)		178 (1.1)	14,987
2017	1134 (8.0)	3904 (27.5)	463 (3.3)	7394 (52.1)		293 (2.1)	784 (5.5)		210 (1.5)	14,182
2018	713 (6.1)	2308 (19.8)	390 (3.3)	6994 (60.1)		139 (1.2)	876 (7.5)		225 (1.9)	11,645
2019	835 (7.1)	2330 (19.7)	405 (3.4)	6903 (58.2)		336 (2.8)	840 (7.1)		207 (1.8)	11,856
Total	5218	14,780	2444	36,297		1772	4206		948	65,665
Crimes Against Society c
Year	Drug Offenses		Pornography		Prostitution		Weapons			Total
2015	1059 (82.9)		3 (0.2)		139 (10.9)		77 (6.0)			1278
2016	1121 (82.7)		15 (1.1)		143 (10.5)		77 (5.7)			1356
2017	1202 (87.4)		15 (1.1)		66 (4.8)		92 (6.7)			1375
2018	1279 (91.1)		18 (1.3)		31 (2.2)		76 (5.4)			1404
2019	1673 (91.6)		13 (0.7)		61 (3.3)		80 (4.4)			1827
Total	6334		64		440		402			7240

^a Kidnapping is considered a Crime Against Persons. For this study, kidnapping was not included due to its low frequency in the crimes against person category. ^b The “Other” category for Crimes Against Property is classified as arson, counterfeiting, bribery, extortion, and embezzlement. These were chosen based on their low cumulative percentage. ^c Gambling is considered a Crime Against Society. For this study, gambling was not included due to its low frequency.

Table 2. Breakdown of Norfolk, VA Crimes by NIBRS Categories.

Crimes Against Persons a
Year	Assault		Homicide			Sex Offenses		Total
2015	2029 (89.7)		34 (1.5)			198 (8.8)		2261
2016	2070 (92.1)		42 (1.9)			134 (6.0)		2246
2017	1942 (92.1)		34 (1.6)			132 9 (6.3)		2108
2018	1792 (91.8)		37 (1.9)			123 (6.3)		1952
2019	1758 (91.5)		36 (1.9)			127 (6.6)		1921
Total	9591		183			714		10,488
Crimes Against Property
Year	Burglary	Vandalism	Fraud	Larceny		Robbery	MVT		Other b	Total
2015	1256 (9.7)	3065 (23.6)	623 (4.8)	6590 (50.7)		502 (3.9)	831 (6.4)		128 (0.9)	12,995
2016	1280 (8.5)	3173 (21.2)	563 (3.8)	8416 (56.2)		502 (3.3)	875 (5.8)		178 (1.1)	14,987
2017	1134 (8.0)	3904 (27.5)	463 (3.3)	7394 (52.1)		293 (2.1)	784 (5.5)		210 (1.5)	14,182
2018	713 (6.1)	2308 (19.8)	390 (3.3)	6994 (60.1)		139 (1.2)	876 (7.5)		225 (1.9)	11,645
2019	835 (7.1)	2330 (19.7)	405 (3.4)	6903 (58.2)		336 (2.8)	840 (7.1)		207 (1.8)	11,856
Total	5218	14,780	2444	36,297		1772	4206		948	65,665
Crimes Against Society c
Year	Drug Offenses		Pornography		Prostitution		Weapons			Total
2015	1059 (82.9)		3 (0.2)		139 (10.9)		77 (6.0)			1278
2016	1121 (82.7)		15 (1.1)		143 (10.5)		77 (5.7)			1356
2017	1202 (87.4)		15 (1.1)		66 (4.8)		92 (6.7)			1375
2018	1279 (91.1)		18 (1.3)		31 (2.2)		76 (5.4)			1404
2019	1673 (91.6)		13 (0.7)		61 (3.3)		80 (4.4)			1827
Total	6334		64		440		402			7240

^a Kidnapping is considered a Crime Against Persons. For this study, kidnapping was not included due to its low frequency in the crimes against person category. ^b The “Other” category for Crimes Against Property is classified as arson, counterfeiting, bribery, extortion, and embezzlement. These were chosen based on their low cumulative percentage. ^c Gambling is considered a Crime Against Society. For this study, gambling was not included due to its low frequency.

3.3. Third Places

Gentrification in this study is partially operationalized through the annual emergence of alcohol-licensed third places within each block group over the study period. Third places are defined as social establishments where individuals gather outside of home and work, such as lounges, clubs, bars, and restaurants. These venues often serve as early indicators and accelerators of gentrification, fostering social interaction and economic exchange that attract more affluent residents to a neighborhood. While we recognize that third places can broadly include a wide range of informal and non-commercial gathering spaces, such as parks, barbershops, community centers, or churches, this study specifically focuses on alcohol-licensed third places. This focus reflects both data availability and the standardized licensing structure associated with alcohol licensing, which includes uniform system of state and local licensing that requires all alcohol vendors to be registered, regularly renewed, and publicly recorded. This helps the study as it enhances conceptual validity by making explicit that third places are broadly understood as any social spaces where people gather outside of home and work, yet this study focuses specifically on those that are alcohol licensed. Second, it improves measurement reliability, since alcohol-licensed establishments are governed by a standardized system of state and local licensing that produces consistent, verifiable, and spatially traceable data. This ensures that the identification of third places is accurate. Moreover, alcohol-serving venues provide an indicator of commercial reinvestment and social activity, often serving as catalysts that attract new patrons and businesses while signaling broader cultural and economic shifts within neighborhoods (Mathews and Picton 2014; Izenberg et al. 2018).

To identify alcohol-licensed third places, the study utilized the Norfolk Database of ABC Licensures, which records all businesses holding alcohol permits. This dataset was cross-referenced with business license records from the Norfolk Commissioner of Revenue to compile a comprehensive list of qualifying establishments. By combining these sources, the analysis captures a reliable inventory of alcohol-licensed third places that have formal licensing and regulatory approval. A key limitation of this approach is its exclusion of non-commercial or informal social spaces that do not require alcohol permits, such as community centers, libraries or churches. These venues also contribute to neighborhood cohesion and community life but fall outside the scope of this measurement. Nevertheless, bars, lounges, and restaurants are often among the first visible signs of neighborhood change and commercial reinvestment, making them a meaningful indicator of gentrification dynamics. In the data there were (n = 380) alcohol-licensed third places identified for the study.

3.4. Demolitions

Demolition activity is another indicator of gentrification, representing structural and spatial transformations within urban environments. Data for this measure were obtained from the Norfolk Permits and Inspections database and include all finalized demolition permits issued between 2015 and 2019. Demolitions (n = 926) are aggregated by block groups to assess localized patterns of redevelopment. Demolitions are treated as annual counts, capturing year-to-year variation in redevelopment intensity within each block group. This approach aligns with the study’s focus on identifying temporal changes in the physical landscape and how these shifts correspond with evolving neighborhood dynamics and crime patterns. By focusing on finalized demolition permits, this approach captures the official changes in the physical landscape, which are critical markers of gentrification. However, it does not account for demolitions that were planned but never completed, nor does it include informal or unpermitted demolitions. Despite these limitations, finalized demolition permits provide a valid and consistent proxy for structural transformation associated with gentrification processes.

3.5. Economic Disadvantage

To control economic disadvantages, the study used the Index of Concentration at the Extremes (ICE) equation developed by Kubrin and Stewart (2006). This equation involves dividing the difference between affluent and poor households by the total number of families in the block group. In this context, affluent refers to block groups whose median household income is two standard deviations above the mean, while poor refers to block groups whose income is below the poverty line. The data for this measure was obtained from the American Community Survey, which characterizes households’ income-to-debt ratios within the block group to measure yearly economic change. For example, if a household’s income is $32,000 and the poverty threshold is set at $12,700, this household’s income to debt ratio would be 2.52. This index helps calculate the imbalances that may be presented in income between impoverished and affluent communities. The American Community Survey categorizes these ratios into the following groups: “under 0.50”; “0.50 to 0.99”; “1.00 to 1.24”; “1.25 to 1.49”; “1.50 to 1.84”; “1.85 to 1.99”; “over 2.00”. In this study, “affluent” block groups were characterized as those above 2.00, whereas “poor” block groups were defined as those under 1. Scores range from +1 to −1, with +1 indicating the highest advantage score and −1 indicating the lowest. Those with indexes at 0 show a balance of resident incomes.

3.6. Zoning

To control neighborhood-level differences in urban development, the study incorporates zoning type as a control variable. Zoning designations—commercial, residential single-family, residential multiple-family, and downtown/mixed-use—are considered to account for variations in land use that may influence the impacts of gentrification factors such as third places, demolitions, and economic shifts. Commercial zones, which support small businesses, residential single-family zones with low-density housing, residential multiple-family zones for multi-family units, and downtown/mixed-use zones designed for high-density commercial activity, each contribute differently to neighborhood dynamics. By including zoning type in the analysis, the study can isolate the specific effects of gentrification while accounting for the underlying urban structure that may shape crime rates and community change. Data for this variable came from Norfolk open GIS database which identified and defined the city’s zoning categories.

3.7. Racial Composition

Racial composition was measured using ACS 1-year estimates at the block group level. Because of the demographic of residents in Norfolk and for analytic clarity, categories collapsed into three groups: Black, White, and Other. Across the 185 block groups, the average composition was 46% White, 43% Black, and 6% Other, though racial distribution varied considerably by neighborhood. This measure allows the analysis to capture how racial demographics interact with redevelopment processes in shaping crime.

3.8. Statistical Analysis

Ecological research has come far in measuring the change in neighborhoods from a once rudimentary concept of linear community influence and growth to a complex algorithm addressing between and within-group differences (Woltman et al. 2012; Bursik and Grasmick 1992). These methods violate data assumptions in estimating simple ordinary least squared regressions (OLS). Traditional linear regression models fail to consider data that is no longer independent. Because the study deals with geographical units, some aspects of space (i.e., race and economic growth) will be more relevant to proximate areas than others—thus impacting the whole model. Additionally, because each unit of analysis is different, there were varying populations within each block group, providing between-group heteroskedasticity or unequal residuals and variance. Lastly, more complex statistical models were needed because of the potential varying effects predictor variables may have on different spaces. This means that the effect of gentrification on crime may be different for one block group than it is for another.

To model these varying effects accurately, each space must be considered respectively and not collectively. Here, hierarchical linear modeling (HLM) is used to effectively model neighborhood changes within and across block groups to adequately develop models to accurately measure the effect of neighborhood revitalization—gentrification. Hierarchical linear modeling provides a sophisticated modeling technique where multiple linear and nonlinear regressions were generated, accounting for each regression’s varying predictors’ influence on each trajectory (Woltman et al. 2012). This is often due to the data being nested structure of the data, where individual units were placed into groups and were subjected to being influenced by characteristics and behaviors of the other sampled groups. In the analysis, the models included both a linear and a quadratic year variable in the HLM to account for potential nonlinear changes in the relationship between time and the outcome variable. The linear term was used to capture the overall trend or average effect of time, assuming a steady increase or decrease over the years. A quadratic term allowed the model to capture any acceleration or deceleration trends. This approach was particularly useful when I suspected that the effect of time on the dependent variable might not be constant, such as when growth or decline occurred more rapidly in certain periods.

There were two levels in the model (see

Table 3. Descriptive Statistics of Level 1 and 2 Measures.

Variable	2015	2016	2017	2018	2019
Level 1	M (SD)	M (SD)	M (SD)	M (SD)	M (SD)
Race (%)
White	0.47 (0.29)	0.46 (0.29)	0.46 (0.29)	0.46 (0.29)	0.46 (0.29)
Black	0.43 (0.31)	0.44 (0.30)	0.44 (0.30)	0.43 (0.30)	0.43 (0.31)
Other	0.06 (0.06)	0.06 (0.06)	0.07 (0.06)	0.07 (0.07)	0.07 (0.07)
Poverty-to-Income (ICE Index)
Poor (<1)	0.13 (0.33)	0.12 (0.33)	0.12 (0.33)	0.11 (0.32)	0.11 (0.32)
Affluent (>1)	0.87 (0.33)	0.87 (0.33)	0.87 (0.33)	0.88 (0.32)	0.88 (0.32)
Third Places
Third Places	0.77 (2.47)	0.93 (2.81)	1.06 (3.58)	1.14 (3.87)	1.32 (4.41)
Level 2
Zoning (%)
Commercial	5.39 (7.71)
Residential Single	9.72 (14.04)
Residential Multiple	36.39 (25.76)
Downtown Mix	2.85 (11.12)
Demolition
Demolition	4.01 (5.08)

). Level 1 of this analysis estimates the change in the sample as a result of the block groups specific predictors, also called within-group variation. Level 2 of this analysis compares the trajectory of that individual case amongst the other trajectories across the complete sampling frame, known as between-group variation. For example, within the investigation, crime and gentrification predictor variables (e.g., third places) were nested inside of the block groups over time. Each block group has varying SES and racial characteristics, which may influence the dependent variable (i.e., crime). It is important to note that demolitions were treated as a Level 2 variable. Demolitions reflect neighborhood-level restructuring processes that are cumulative and enduring, rather than short-term fluctuations. While demolition permits may be issued at a specific time, their effects unfold over longer periods across block groups making them conceptually appropriate to model at Level 2.

Taken together, this multilevel approach allows me to examine how variation in block group characteristics generates differential effects when estimating the relationship between gentrification and crime. Some neighborhoods experiencing gentrification may display distinct crime trajectories depending on their residential composition and the time period in which changes occur. By incorporating both within- and between-group variation, the models provide a more precise and context-sensitive account of how neighborhood revitalization processes influence crime over time.

3.9. Crime Distribution and Moran’s I

To begin, an exploratory analysis was conducted to view the spatial patterns in the data by obtaining Moran’s I values. Moran’s I measures if the data is spatially autocorrelated and determines if the variable(s) are distributed in random or non-random manner or if they are in a clustered or dispersed structure. A score that is closer to +1 shows that the data is clustered. The z score shows whether the data is dispersed at random, while the p-value indicates whether the data is clustered or dispersed.

Shown in

Table 4. Spatial Autocorrelation Statistics of Distribution (Global Moran’s I).

	Moran’s I	Z
2015
Person	0.71	7.39 **
Property	0.06	1.95
Society	0.27	7.34 **
2016
Person	0.25	6.60 **
Property	0.05	1.44
Society	0.01	2.58 **
2017
Person	0.20	5.40 **
Property	0.06	1.87
Society	0.27	7.44 **
2018
Person	0.25	6.80 **
Property	0.05	1.71
Society	0.34	9.20 **
2019
Person	0.19	5.29 **
Property	0.06	1.66
Society	0.39	10.54 **

** p < 0.05.

, with index scores significant and positive for crimes against persons and society, these categorized incidents were randomly clustered from 2015 to 2019. It is important to note that property crime spatial autocorrelations statistics never reach significance at the p < 0.05 level. This illustrates that property crime within the city was not clustered and occurred randomly in respect to other crimes. Because of the significant spatial dependence in the persons and society models, a spatial lag term was included to control for autocorrelation in the dependent variable. This spatial lag was constructed using a first-order queen contiguity weights matrix, which defines neighboring block groups as those sharing a boundary. The spatial lag represents the spatially weighted average of neighboring areas’ crime rates, capturing the influence of nearby block groups on each unit’s crime level.

4. Results

4.1. Bivariate Analysis

Table 5. Correlation Matrices.

Level 1 Means
	(1)	(2)		(3)		(4)		(5)	(6)		(7)	(8)
% White	1
% Black	−0.969 **	1
% Other	0.083	−0.289 **		1
ICE Index	0.622 **	−0.631 **		0.141		1
Third Places	0.087	−0.087		0.022		0.050		1
Crimes Against Persons	−0.398 **	0.395 **		−0.062		−0.368 **		−0.003	1
Crimes Against Property	0.380 **	−0.372 **		0.047		0.209 **		0.061	−0.427 **		1
Crimes Against Society	−0.484 **	0.504 **		−0.187 **		−0.278 **		−0.055	0.351 **		−0.398 **	1
Level 2 Means
	(1)		(2)		(3)		(4)			(5)
Commercial	1
Single	−0.186 *		1
Multiple	0.122		−0.403 **		1
Downtown	−0.058		−0.338 **		−0.099		1
Demolitions	−0.027		0.08		−0.014		−0.007			1

** Correlation is significant at the 0.01 level (2-tailed). * Correlation is significant at the 0.05 level (2-tailed).

presents correlation matrices for key variables in the study, focusing on zoning, demolitions, racial composition, and crime rates. At Level 2, which captures neighborhood-level zoning and demolition characteristics, several notable associations emerged. A small but significant negative correlation was found between commercial zoning and single residential zoning (r = −0.186, p < 0.05), indicating that areas with higher allocations for commercial space tend to have fewer single-family residential zones. Single residential zoning also showed strong negative correlations with multiple residential (r = −0.403, p < 0.01) and downtown/mixed zoning (r = −0.338, p < 0.01), suggesting spatial competition between single-family housing and denser, more urban zoning types. Demolitions, however, showed no significant correlations with any zoning variables, implying a relatively independent spatial process.

At Level 1, which includes racial composition, socioeconomic concentration, presence of alcohol-licensed third places, and crime rates, several strong relationships were observed. The percentage of White and Black residents were strongly negatively correlated (r = −0.969, p < 0.01), highlighting distinct racial segregation within block groups. Additionally, a higher percentage of White residents was moderately associated with lower poverty concentration as measured by the ICE Index (r = 0.622, p < 0.01), while a higher percentage of Black residents was moderately associated with higher levels of crimes against society (r = 0.504, p < 0.01).

Crime variables demonstrated internal consistency with expected patterns: crimes against persons and crimes against society were positively correlated (r = 0.351, p < 0.01), while crimes against property were negatively correlated with both (r = −0.427 and −0.398, respectively, p < 0.01).

4.2. Crimes Against Society

Table 6. Hierarchical Linear Model Results w/Crimes Against Society.

	(1)	(2)	(3)	(4)	(5)	(6)
Level 2
Intercept	4.03 (0.20) ***	4.03 (0.20) ***	4.03 (0.20) ***	4.03 (0.20) ***	4.03 (0.20) ***	4.03 (0.20) ***
Commercial	−0.02 (0.01)	−0.02 (0.01)	−0.02 (0.01)	−0.02 (0.01)	−0.02 (0.02)	−0.02 (0.01)
Single	−0.02 (0.01) **	−0.02 (0.01) **	−0.02 (0.01) **	−0.02 (0.01) **	−0.02 (0.01) **	−0.02 (0.01) **
Multiple	−0.04 (0.01) ***	−0.04 (0.01) ***	−0.04 (0.01) ***	−0.04 (0.01) **	−0.04 (0.01) **	−0.04 (0.01) ***
Downtown	−0.01 (0.02)	−0.01 (0.02)	−0.01 (0.02)	−0.00 (0.02)	−0.00 (0.02)	−0.01 (0.02)
Demolitions	−0.06 (0.02) **	−0.06 (0.02) **	−0.06 (0.02) **	−0.06 (0.02) ***	−0.06 (0.03) **	−0.06 (0.02) **
Level 1
Linear_Yr	−1.30 (0.07) ***	−1.26(0.07) ***	−1.30 (0.07) ***	−1.30 (0.07) ***	−1.26 (0.07) ***	−1.26 (0.07) ***
Quad_Yr	−0.61 (0.05) ***	−0.61 (0.05) ***	−0.61 (0.05) ***	−0.61 (0.05) ***	−0.61 (0.05) ***	−0.61 (0.05) ***
% Black	0.01 (0.02)	0.01 (0.02)	0.01 (0.02)	0.01 (0.02)	0.02 (0.02)	0.01 (0.02)
% Other	−0.00 (0.02)	−0.00 (0.02)	−0.00 (0.02)	0.00 (0.02)	−0.04 (0.03)	0.00 (0.02)
ICE Index	−0.96 (0.86)	−1.01 (0.87)	−0.96 (0.86)	−1.03 (0.87)	−0.96 (0.91)	−0.90 (0.88)
Third Places	−0.10 (0.02) ***	−0.14 (0.44) **	−0.10 (0.03) ***	−0.14 (0.04) ***	−0.09 (0.05) ***	−0.10 (0.02) ***
Lag	0.02 (0.01) **	0.02 (0.01) **	0.02 (0.01) **	0.02 (0.01) **	0.01 (0.01) **	0.02 (0.01) **
Third Places * Single		0.00 (0.00)		0.00 (0.00) *
Third Places * Multiple			0.00 (0.00) **	0.01 (0.00) **
% Black * Nonresidential					0.01 (0.00) **
% Black * Residential						−0.00 (0.00) *
Variance Components
Intercept	5.93 ***	5.92 ***	5.93 ***	5.91 ***	5.92 ***	5.92 ***
Linear_Yr	0.55 **	0.30 **	0.30 **	0.29 **	0.31 **	0.31 **
Quad_Yr	0.49 **	0.23 **	0.23 **	0.22 **	0.23 **	0.23 **
% Black	0.02 **	0.00 **	0.00 **	0.00 **	0.00 **	0.00 **
Third Places	0.15 ***	0.02 ***	0.03 ***	0.03 ***	0.02 ***	0.02 ***

Dependent Variable: Crimes Against Society. *** p < 0.01; ** p < 0.05; * p < 0.10.

presents hierarchical linear models predicting crimes against society. Across all models, several consistent patterns emerged. At the block group level, demolitions were significantly and negatively associated with crimes against society (B = −0.06, p < 0.05), suggesting that areas undergoing more demolitions were associated with lower rates of these offenses. Similarly, block groups with higher proportions of residential single-family (B = −0.02, p < 0.05) and residential multi-family zoning (B = −0.04, p < 0.01) showed lower crime levels, pointing to a potential protective effect of residential land use.

At level one, both the linear (B = −1.30, p < 0.01) and quadratic (B = −0.61, p < 0.01) time trends were significant and negative, indicating that crimes against society were lower over time at an accelerating rate between 2015 and 2019. The number of alcohol-licensed third places was also negatively associated with crime (B = −0.10 to −0.14, p < 0.01), with each additional establishment associated with approximately 10–14% lower levels of crimes against society.

However, when cross-level interactions with zoning were introduced, the results shifted. In dense residential contexts, the protective effect of alcohol-licensed third places weakened. Specifically, the interaction between alcohol-licensed third places and multi-family zoning was positive and significant (B = 0.01, p < 0.05), suggesting that additional third places in these areas were associated with increases in crime. Model 5 further revealed a significant positive interaction between % Black population and non-residential zone (B = 0.01, p < 0.05), indicating that crime rates tended to be higher in areas combining larger Black populations with greater nonresidential land use. Conversely, when the Black population measure interacted with residential zoning, results showed a small but significant negative association (B = −0.00, p < 0.10). (B = −0.00)1.

4.3. Crimes Against Property

Table 7. Hierarchical Linear Model Results w/Crime Against Property.

	(1)	(2)	(3)	(4)
Level 2
Intercept	53.41 (1.83) ***	53.41 (1.83) ***	53.40 (1.83) ***	53.41 (1.83) ***
Commercial	0.04 (0.12)	0.03 (0.13)	0.04 (0.12)	0.03 (0.13)
Single	−0.04 (0.19)	−0.04 (0.19)	−0.04 (0.07)	−0.05 (0.07)
Multiple	0.08 (0.08)	0.08 (0.09)	0.08 (0.09)	0.08 (0.09)
Downtown	−0.04 (0.12)	−0.05 (0.12)	−0.06 (0.11)	−0.06 (0.11)
Demolitions	0.19 (0.12) *	0.19 (0.12) *	0.18 (0.12)	0.18 (0.12)
Level 1
Linear_Yr	−7.36 (2.26) ***	−7.36 (2.26) **	−7.50 (2.31) ***	−8.00 (2.51) **
Quad_Yr	−7.11 (1.89) ***	−7.12 (1.89) ***	−7.14 (1.92) ***	−7.40 (2.02)
% Black	0.23 (0.19)	0.23 (0.18)	0.21 (0.19)	0.21 (0.19)
% Other	−0.40 (0.42)	−0.40 (0.42)	−0.42 (0.42)	−0.42 (0.43)
ICE Index	−3.07 (8.91)	−2.48 (8.81)	−3.63 (8.82)	−3.06 (8.62)
Third Places	7.88 (3.88) **	7.71 (3.85) **	8.67 (4.00) **	9.21 (4.21) **
Third Places * Commercial		0.88 (0.46) *		0.87 (0.48) *
Third Places * Downtown			−0.51 (0.24) **	−0.53 (0.29) *
Variance Components
Intercept	463.46 ***	460.30 ***	469.38 ***	449.20 ***
Linear_Yr	869.59 ***	906.30 ***	905.71 ***	1074.29 ***
Quad_Yr	582.80 ***	599.37 ***	602.65 ***	669.29 ***
Third Places	929.81 ***	933.28 ***	985.70 ***	1210.01 ***

Dependent Variable: Crimes Against Property. *** p < 0.01; ** p < 0.05; * p < 0.10.

reports the models predicting crimes against property. Unlike crimes against society, zoning variables were not significantly associated with property crime. Instead, demolitions were positively associated with property crime in the baseline model (B = 0.19, p < 0.10), though this effect weakened when interaction terms were introduced.

Temporal patterns were consistent with society crimes: both the linear (B = −7.36, p < 0.001) and quadratic (B = −7.11, p < 0.01) terms showed strong negative effects, indicating substantial declines in property crime over time. In contrast to society crimes, however, third place was positively associated with property crime (B = 7.88, p < 0.01), suggesting that the opening of additional establishments corresponded with increased opportunities for property offenses.

Cross-level interactions revealed that the effect of alcohol-licensed third places varied by zoning. The interaction with commercial zoning was positive and significant (B = 0.88, p < 0.05), showing that new establishments in commercial areas were linked to higher property crime. By contrast, the interaction with downtown zoning was negative and significant (B = −0.51, p < 0.05), indicating that alcohol-licensed third places in the downtown corresponded with lower property crime rates.

4.4. Crimes Against Persons

Table 8. Hierarchical Linear Model Results w/Crime Against Persons.

	(1)
Level 2
Intercept	53.41 (1.83) ***
Commercial	0.17 (0.23)
Single	0.02 (0.12)
Multiple	0.10 (0.23)
Downtown	−0.06 (0.19)
Demolitions	0.17 (0.28)
Level 1
Linear_Yr	−5.38 (1.70) **
Quad_Yr	−4.46 (0.96) ***
% Black	0.09 (0.29)
% Other	−0.34 (0.32)
ICE Index	−2.14 (0.69) **
Third Places	−8.82 (10.18)
Lag	−0.57 (0.31) *
Variance Components
Intercept	89.17 *
Level 1	2719.97

Dependent Variable: Crimes Against Persons. *** p < 0.01; ** p < 0.05; * p < 0.10.

summarizes the results for crimes against persons. At the block group level, no significant effects were found for zoning, demolitions, or racial composition. However, at level one, the Index of Concentration at the Extremes (ICE) showed a significant negative association (p < 0.05), suggesting that crimes against persons tended to have lower levels as block groups became more economically advantaged. Time trends were not included as cross-level interactions in this model, limiting interpretability compared to the other crime categories.

5. Discussion

This study was designed to examine the relationship between gentrification and crime—conceptualized through alcohol-licensed third places and demolitions. Third places, specifically alcohol-licensed third places, refer to semi-public venues such as bars and restaurants that foster social interaction beyond home and work. Demolitions serve as indicators of redevelopment and physical transformation linked to gentrification. As mentioned, gaps in prior research can be attributed to nonuniform definitions of gentrification (Kreager et al. 2007; Lee 2010), over-reliance on census-based predictors (McDonald 1986; Covington and Taylor 1989), and cross-sectional designs (Kreager et al. 2007). This study addresses these gaps by employing longitudinal models that capture both within- and between-group change and integrate municipal and census data to estimate yearly variation. Overall, the results indicate that gentrification has mixed effects on crime. While certain processes associated with redevelopment, such as the introduction of new third places, appear to reduce crimes against society, they also coincide with increases in property-related offenses. These findings highlight that gentrification’s impact on crime is not uniform but instead depends on local social and physical contexts.

Turning first to crimes against society (Table 6) emerged as the most robust, showing significant variation across block groups over time. Notably, differences in the proportion of Black residents and the presence of alcohol-licensed third places were significant predictors of change. Consistent with prior research (Barton 2016; McDonald 1986), initial models suggested that gentrification—measured through third places and demolitions—was associated with reductions in crime. However, the findings also indicate that expanding alcohol-licensed third places within multi-family residential areas may not effectively reduce crime. This finding contrasts with expectations that social venues promote informal social control. This outcome likely reflects broader structural disadvantages, as residents in public housing often face limited employment opportunities and restricted social mobility (Pager 2008). In such contexts, new alcohol-serving venues may fail to benefit existing communities and instead contribute to cultural displacement by catering to more affluent newcomers, undermining long-standing social networks and neighborhood cohesion.

The model explaining crimes against property was significant but did not include significant racial variation in level one; therefore, these results cannot address how racial composition affects property crime. However, this model assesses the effect of gentrification on crimes against property with the inclusion of third place establishments. Here initial results show that the higher number of alcohol-licensed third places and the more demolitions occur within an area, the more crimes against property occur. This pattern suggests that different mechanisms may drive property crime compared to crimes against society. Interpreting this finding through conventional frameworks like routine activities and focusing on the possible perceptions of offenders, it is possible that demolished spaces can be unprotected by guardians that would encourage offending, opening up these block groups to issues such as burglary and vandalism (Cohen and Felson 1979; Barton and Gruner 2016). The increase in alcohol-licensed third places may provide motivated offenders with more places and greater opportunities to offend.

In contrast, the last model, which estimates crime against persons, shows no significant variation in changes in this type of crime over time and across block groups; thus, the model does not provide an opportunity to explore factors affecting neighborhoods extensively. An increase in a block group’s ICE Index was associated with less crime. This finding is consistent with existing research, which indicates that a rise in an area’s socioeconomic status (SES) correlates with reductions in crime, likely due to the resources available to enact social control within the block group (Kreager et al. 2007; Barton 2016; McDonald 1986).

The influence of alcohol-licensed third places on crime varied by offense type. For crimes against society, an increase in these establishments was associated with lower crime rates across block groups over time. This relationship aligns with routine activities theory, which suggests that more active and populated areas foster natural surveillance and social control, thereby deterring deviant behavior (Johnson et al. 2014). However, the opposite pattern emerged for property crime: block groups with more alcohol-licensed third places experienced higher levels of property-related offenses. From the same theoretical view, this may reflect increased opportunities for motivated offenders to target both patrons and nearby businesses, as the presence of new establishments introduces more potential targets and valuables (Cohen and Felson 1979; Barton and Gruner 2016). So, while third places may contribute to neighborhood vitality and guardianship for crimes such as drugs and prostitution, they can simultaneously generate opportunities for property victimization. The findings support the Place-in-Neighborhoods (PIN) perspective (Tillyer et al. 2021), showing that the criminogenic or protective influence of third places depends on the social and physical environment in which they are embedded.

Interaction effects between third places and zoning classifications further illustrate this contextual dependence. In commercial zones, increases in alcohol-licensed third places corresponded with higher crime rates, likely because these areas are left largely unmonitored after business hours. By contrast, downtown areas, where nightlife activity and law enforcement presence are more consistent, showed decreases in crime with the addition of similar establishments. These distinctions reflect differences in guardianship: commercial areas often experience lapses in protection once businesses close, whereas downtown districts maintain greater security and police visibility at night (Triplett et al. 2005). Consequently, the crime impacts of third places appear contingent upon the spatial and temporal rhythms of neighborhood use, supporting the PIN perspective that place effects are moderated by broader contextual features.

Lastly, the analysis revealed that race plays an important but context-dependent role in the relationship between gentrification and crime. While racial composition alone did not significantly predict crime rates, its interaction with land use designations was notable. Specifically, as block groups became more non-residential, crimes against society in predominantly Black communities increased, whereas neighborhoods with greater residential allocations experienced lower crime rates. This pattern suggests that the social control benefits associated with residential stability may be weakened when redevelopment prioritizes commercial rather than housing uses. Consistent with critical race theory (Crenshaw 2010), these findings highlight how urban redevelopment can reproduce racial inequities by displacing Black residents and undermining community-based forms of guardianship and cohesion. Historical and ongoing gentrification in Norfolk has often favored business-oriented development over affordable housing, reducing residential options and eroding neighborhood stability (Finn 2021). To counter these effects, city planners should prioritize equitable revitalization strategies that expand housing opportunities in predominantly Black neighborhoods. Such efforts must include fair lending and non-predatory financing practices to promote homeownership and empower residents to sustain their own communities and systems of social control, comparable to those historically available in White neighborhoods.

5.1. Limitations

One limitation concern is the small, investigated time intervals. It was previously suggested that smaller increments of data are needed to model the change accurately to capture gentrification as a process; however, five years of data may not capture the full picture as neighborhood often take several years to fully gentrify (Barton 2016; Meltzer and Capperis 2017; DeSena 2018). Regarding issues in the data collection process for this research specifically, city agents informed me that data could only be acquired for years as far back as 2015 for some variables due to guidelines allowing them to discard earlier datasets to make room for more current data. Ecological researchers seeking to use municipal repositories may need to begin data collection years in advance to construct more chronologically robust datasets. Additionally, this analysis does not account for prior neighborhood dynamics or planning processes that may have influenced patterns of redevelopment and crime before the study period. Factors such as historical zoning decisions, long-term disinvestment, urban renewal projects, or targeted redevelopment initiatives could shape the spatial distribution of gentrification and crime independent of recent changes.

Another limitation to consider is the unique measure of alcohol-licensed third places used to generate statistical models. As mentioned, alcohol-licensed third places were conceptualized through the works of sociologist Oldenburg and Brissett (1982) and operationalized from a pre-generated list of NAIC codes used to categorize business for government and commerce labeling. These categories are not explicitly used to measure third places therefore create some downside into conceptual validity. For example, there could be establishments that would meet the criteria of third places not labeled in the subgroup of codes (e.g., infrequent or mobile establishments) or are included in these lists but lack the features of third places in the community (e.g., casinos).

The residential allocations and racial compositions throughout the city of Norfolk VA may be influenced by larger structural characteristics of the city. For example, the local shipping ports and naval base create an issue when attempting to disentangle racial effects in the model. Given that Norfolk hosts the largest naval base in America, I credit a large portion of the stability of racial compositions over time to this. With naval buying power, they acquire various homes and construct housing for its enlistees, low-income areas’ median incomes, relative housing values, and racial diversity theoretically increase. In contrast, with Norfolk having the largest shipping terminal in the Virginia Port Authority2 industry, this creates a concentration of labor workers and tax revenue3. On one hand, in theory, these populations, who are both racial and socioeconomically diverse, causing variations in block group elements such as race are harder to investigate. On the other hand, this also means that these racial variations may be unique and should be investigated further. Lastly, a limitation of this analysis is that the spatial lag term was applied only to crimes against persons and crimes against society, where significant spatial autocorrelation was detected. While this controls for local spatial dependence, it assumes uniform influence among adjacent areas and may not capture broader or more complex spatial relationships influencing crime patterns.

5.2. Policy Implications

This study attempts to reconceptualize gentrification by using annual aggregation to observe neighborhood change in closer-to-real time. While gentrification has often been presented as beneficial within traditional criminological frameworks, a more critical perspective reminds us that such processes can also produce uneven outcomes, particularly for low-income and historically marginalized communities (Alexander 2012; Desmond 2016). The following potential policy implications are offered with caution, as exploratory considerations rather than definitive solutions.

First, neighborhood redevelopment may be most effective when it is understood as both a social and a physical process. Demolitions and new construction can enhance infrastructure, yet they may also unsettle the informal networks that help maintain neighborhood stability. Integrating social impact assessments into redevelopment planning—alongside economic and environmental evaluations—could offer a fuller understanding of how change affects community cohesion, trust, and informal social control. Such assessments may help identify ways to minimize social disruption while pursuing physical revitalization. Additionally, the intentional inclusion or preservation of community-serving third places, such as locally owned cafes, barbershops, and parks, might help sustain continuity and collective efficacy during times of transition. Policies that encourage the retention of locally rooted businesses and small business support could help ensure that these spaces remain accessible to long-term residents. While such strategies may not prevent displacement, they could strengthen the social infrastructure that supports neighborhood safety and belonging.

Redevelopment and revitalization processes might benefit from being data-informed with the use of more temporal data. The use of near real-time indicators, such as permits, licensing data, or official business openings, could provide early signals of neighborhood transformation. Municipal agencies might use these indicators to monitor the pace and nature of change, allowing for more proactive rather than reactive responses to potential displacement or social strain. Finally, the study’s approach highlights the importance of cross-sector collaboration. Redevelopment, housing policy, and crime prevention often operate in separate institutional spheres, yet they intersect in practice. Coordinated efforts between housing authorities, planning departments, and public safety agencies could support more balanced and equitable outcomes, recognizing that physical revitalization alone cannot ensure community well-being.

5.3. Future Research

In tandem with the limitations presented, future research can expound on the proposed gentrification models to better demonstrate community change and influence. This exploration should primarily focus on generating a database worthy of strong longitudinal mixed-methods models. This includes expanding the dataset to possibly a ten-year model to create a long-term representation of gentrification’s effects on neighborhoods. Additionally, when researching gentrification and crime further, researchers should seek to further the conceptual model by creating better ways to operationalize social and economic changes. Because these changes help define gentrification, scholars should look to locate reliable repositories that assist in modeling community change. This study uses the influence of third places and demolition, which shows some impact on feature change. Further strategies should look at exploring qualitative properties to include but are not limited to what modern characteristics make up third places and why spaces are deemed for demolition.

Research should look at the growing influence of virtual third places (Soukup 2006; Moore et al. 2009) and how they may be a function of gentrification and crime. Establishments such as cyber cafes and eGaming centers may provide a unique social and economic benefit where previously conceptualized third places may be limited. Additionally, although this study considers potential modern third places, these spaces are also conceptualized through Oldenburg and Brissett’s (1982) original characterizations. With this in mind, it would appear worthwhile constructing a new definition of third places that is not primarily through an older cis-gendered, White, heteronormative perspective to include other potential establishments. There may be establishments within marginalized or overlooked communities that host properties of third places that are ignored, given the inattention to these communities by mainstream research. For example, places that may be omitted are businesses such as neighborhood smokehouses and candy shops, which can have positive social and economic influences. However, because the work is considered illegitimate, it may not be legitimate, although providing a service to the community and having characteristics of third places.

Future research should compare historical maps of neighborhood change (i.e., redlining) and analyze potential generational effects on neighborhoods related to significant revitalization and transformation (Finn 2021)—within or surrounding the space. Qualitative historical rhetoric explains how Norfolk, VA, unambiguously used government grants and funding to transform areas into a now known gentrified space—for example, the Ghent District. Comparing these maps and policies provides an avenue of research available to examine any lasting effects on crime rates and racial and economic characteristics in proximate locales.

Perhaps more importantly, this research model provided in this article should be expanded to consider the context within a larger metropolitan city, one that has the economic independence of major industry or continuously changes to the city’s physical aspects. Because Norfolk, VA is a mid-sized city, it is heavily influenced by its prominent characteristics (e.g., US Naval Base), major shipping ports, and local universities. Gentrification models would be assisted by data from locales that offer enough racial and economic diversity to view explicit variation not found in the current growth curve statistical models.