Race,segregated urban mobility,and socioeconomic ascent: A neighborhood network approach

Abstract

Chinese

Past research on neighborhood ascent—socioeconomic increases among residents and housing—focuses on residential environments, overlooking how connections beyond their boundaries influence neighborhood change. Using geocoded tweets from over 375,000 Twitter users in the 50 most populous US cities, we explore how a city’s mobility network relates to neighborhood socioeconomic (SES) ascent from 2010 to 2019—whether ascent is more likely in cities more or less structurally connected via residents’ routine travels. We find that neighborhoods located in cities with greater racially segregated mobility, particularly initially lower-SES neighborhoods, are more likely to ascend compared to those in cities where residents more frequently visit neighborhoods of different racial compositions than their own. While ascent patterns are similar across White and Hispanic neighborhoods, regardless of city type, neighborhoods with no racial majority are more likely to ascend in cities with greater segregated mobility. Black neighborhoods are least likely to ascend, underscoring how deeply entrenched racial hierarchies continue to shape neighborhood trajectories. Our results highlight broader urban dynamics structuring neighborhood ascent, revealing how stratification processes extend well beyond where people live.

Keywords

gentrification neighborhoods networks race socioeconomic ascent

Introduction

Past work has focused primarily on features of residential environments to predict whether a neighborhood may experience socioeconomic ascent—a broad category that includes neighborhood trajectories such as gentrification, and which describes socioeconomic improvements among residents and housing in a neighborhood (Owens, 2012). Neighborhood ascent encompasses rapid increases in housing costs and residents’ socioeconomic status, including income, education, and occupational standing (Candipan and Bader, 2022; Ellen and O’Regan, 2011; Owens and Candipan, 2019). Yet ascent does not unfold uniformly across neighborhoods with similar starting points. A large body of research shows these processes are deeply shaped by racial stratification, with White neighborhoods more likely and Black neighborhoods less likely to ascend, and higher-SES White residents often driving changes that reinforce urban racial hierarchies (Galster et al., 2003; Candipan and Bader, 2022; Freeman and Cai, 2015; Owens and Candipan, 2019; Rucks-Ahidiana, 2021, 2022; Timberlake and Johns-Wolfe, 2017).

However, this emphasis on residential environments overlooks that individuals’ daily interactions routinely extend beyond their home neighborhoods, potentially bringing individuals into contact with people of different sociodemographic backgrounds and connecting them to neighborhoods outside of their own (Cagney et al., 2020). The movement patterns of individuals within cities may influence whether and which neighborhoods experience socioeconomic ascent. For example, in cities where neighborhoods are less connected to one another through residents’ routine movements, disadvantaged neighborhoods may be more susceptible to ascent by remaining isolated from the social, economic, and political networks that help buffer against such changes or, alternatively, less susceptible if developers and potential higher-SES in-movers overlook these neighborhoods altogether. Taking this as our starting point, our study adds to a small but growing body of urban research taking an everyday “mobility-based” perspective that demonstrates how neighborhood stratification processes extend well beyond the boundaries of where one lives (Candipan et al., 2021; De la Prada and Small, 2024; Moro et al., 2021; Wang et al., 2018). We expand upon these studies by applying a mobility-based lens to the study of race and neighborhood socioeconomic ascent. In doing so, our study identifies the role of the macro-level racialized context in shaping whether, and under what contexts, neighborhood socioeconomic ascent is more likely to occur.

Drawing on mobility data from millions of geotagged tweets matched with neighborhood data from the United States decennial census and American Community Survey, we explore how a city’s mobility network relates to neighborhood socioeconomic ascent. We first examine whether neighborhoods located within US cities with higher levels of racially segregated movement between neighborhoods are more or less likely to experience socioeconomic ascent from 2010 to 2019. Next, we examine the role of neighborhood SES and racial composition. We analyze heterogeneity in the relationship between segregated mobility and ascent by initial neighborhood racial composition—whether a neighborhood is majority White, Black, Hispanic, or racially mixed—and SES composition. Then, we account for the broader urban context by examining whether the relationship between segregated mobility and neighborhood ascent, and the heterogeneity of this relationship by initial racial and SES composition, holds when accounting for the racial residential segregation level of the city.

Our study contributes new insights into neighborhood change, racial stratification, and urban inequality in several ways. First, our study adds to a burgeoning literature adopting a mobility-based perspective for the study of neighborhood inequality, extending the framework to study socioeconomic neighborhood change. Second, by adopting a neighborhood network lens to the study of neighborhood socioeconomic change, we provide a new strategy for understanding macrolevel inequality. While prior work has primarily focused on residential-level conditions, we highlight the importance of a higher-order neighborhood structure shaping neighborhood change processes: durable neighborhood networks formed via the daily travel of city residents. Finally, our analysis centers race by exploring how the racialized context of the neighborhood and its broader region conditions whether ascent, often analyzed as a strictly class-based process, as well as other trajectories of socioeconomic change, is more likely to occur.

Background

Understanding the causes and consequences for residents and cities has long been of interest to urban scholars in the US (Delmelle, 2017; Kang et al., 2022; Owens, 2012). Neighborhood ascent refers to rapid increases in housing costs (rent and property values) and residents’ SES (income, educational attainment, and occupational status), and encompasses a broad range of upward mobility, including but not limited to gentrification (Candipan and Bader, 2022; Ellen and O’Regan, 2011; Owens and Candipan, 2019).

Research has demonstrated that gentrification can spur changes that significantly alter the social fabric of neighborhoods, sometimes temporarily fostering greater economic diversity but often exacerbating social inequalities (Lees, 2008; Zukin et al., 2017). The consequences for neighborhoods include not only shifts in sociodemographic compositions but also transformations in cultural and community dynamics, as well as the reconfiguration of public and commercial spaces (Gibbons et al., 2020; Hyra, 2017; Kern, 2016). Neighborhoods are not islands, but rather their interconnections make up an urban mosaic within cities. This intricate web of relationships means that changes in one neighborhood often ripple through surrounding areas, interacting with broader urban dynamics (Sampson, 2012). For example, the process of neighborhood ascent may not only spur racial and economic changes to the composition of the neighborhood but can also lead to the displacement of racial minority groups and lower-income residents, reshuffling the socio-demographic makeup of adjacent communities and beyond (Hwang and McDaniel, 2022; Zuk et al., 2015).

Past work typically predicts socioeconomic ascent based on residential neighborhood characteristics of its housing and residents. However, individuals’ daily interactions often extend beyond their residential neighborhoods, potentially bringing them into contact with people of different sociodemographic backgrounds and connecting them to neighborhoods outside of their own. A burgeoning line of activity space research acknowledges this, emphasizing the role of the activity spaces that individuals occupy and traverse in their daily rounds (Cagney et al., 2020; Kwan, 2013). Understanding the daily movements of individuals within cities that produce dynamic connections between neighborhoods may broaden our understanding of neighborhood change processes, such as socioeconomic ascent.

Activity space research suggests that social connections available to individuals are not confined to their immediate residential areas. Instead, opportunities for connection are influenced by diverse settings they encounter throughout their routines, such as workplaces, schools, recreational areas, and transit routes (Cagney et al., 2020; Browning et al., 2017b). These daily interactions can facilitate access to different resources and social networks beyond their own neighborhoods (Browning and Soller, 2014). Whether routine travels bring residents into activity spaces that are more (or less) segregated than their residential environments, however, is mixed with some studies suggesting that associations vary by residents’ racial and SES profiles (Browning et al., 2017a; Ellis et al., 2004; Hall et al., 2019; Krivo et al., 2013; Wang et al., 2018; Xu, 2022). These differences likely reflect variation in geographic scope, with most studies confined to a single city, and in the different types of movement data employed, including travel diaries, mobile phone tracking, Census commuting data, and geotagged social media information.

The structural connectedness of neighborhoods within cities

Aided by advances in machine learning techniques and growing availability of novel large-scale data sources (e.g., mobile phone and user-generated social media data), a burgeoning line of scholarship extends activity space research by focusing on dynamic network connections between neighborhoods created through the movement of city residents (Sampson and Candipan, 2023). This higher-order network perspective emphasizes structural connections forged between neighborhoods through individual mobility across cities (Brazil, 2022; Levy et al., 2020; Xu, 2022). Individuals’ routine travel between different neighborhoods creates cross-neighborhood ties that meaningfully connect communities within the city (Brazil et al., 2025; Sampson, 2012). These cross-neighborhood ties are distinct from both internal neighborhood dynamics and spatial processes from adjacent neighborhoods (Levy et al., 2020; Sampson and Candipan, 2023).

Higher-order networks within cities, reflected in residents’ mobility patterns, may plausibly influence which neighborhood experience ascent. One potential mechanism relates to widening choice sets in activity spaces and residential locations. Less racially segregated (more equitable) mobility increases the flow of residents between different types of neighborhoods across the city, potentially opening more neighborhoods to ascent through widened exposure to potential in-movers and investments (Wang et al., 2018). Using georeferenced Twitter-based activity data to track individual’s movements in San Francisco, Chapple et al. (2021) find that outsiders are significantly more likely to visit neighborhoods currently undergoing gentrification, suggesting that observing movement patterns beyond residential neighborhoods may help predict gentrification. Further, less segregated movement between neighborhoods may offer greater opportunities for more diverse flows of people to visit and consume different types of neighborhoods than their own, potentially serving as an early indication that higher-SES newcomers’ activity spaces are expanding into more diverse and ascendable areas, aligning with consumption-led gentrification theories (Zukin et al., 2017). Conversely, ascent may be less likely in cities with less segregated mobility if their residents travel less meaningfully to more diverse neighborhoods through their daily rounds, but maintain segregated residential environments. Neighborhood ascent does not inevitably follow, as these contexts may have less speculative capital targeting disadvantaged areas, or local political and institutional arrangements that stabilize existing neighborhood conditions (Hyra, 2017; Hwang and Ding, 2020). Moreover, more integrated mobility patterns can sometimes diffuse demand and institutional resources across more neighborhoods, reducing intense reinvestment pressures in any single place.

Even in cities with higher mobility-based segregation, neighborhood ascent may still occur if targeted capital flows into select areas or through internal reshuffling, where lower-SES or non-White neighborhoods face greater turnover and vulnerability to reinvestment, while higher-SES White neighborhoods remain comparatively stable (Hwang and Ding, 2020). Alternatively, more segregated mobility may close off neighborhoods and limit gentrification to isolated pockets, or even restrict ascent altogether (Moro et al., 2021). More mobility-based segregation produces higher levels of spatial isolation in activity spaces, leading to doubly reinforced homophily in residents’ social networks beyond their home neighborhoods (Browning et al., 2017a; Cai et al., 2024; Hwang and McDaniel, 2022).

Racial stratification in neighborhood change processes

Neighborhood socioeconomic change processes are structured by racial stratification, exacerbating neighborhood inequality by race and class (Hwang, 2015; Rucks-Ahidiana, 2022). For example, whether a neighborhood experiences gentrification is not uniformly predicted across neighborhoods with similar socioeconomic standing but rather depends on the racial composition of the community. Past studies demonstrate the tight relationship between race and neighborhood socioeconomic change processes in the US, finding different neighborhood trajectories of change (or stability) depending on the racial composition of the neighborhood (Rucks-Ahidiana, 2021, 2022). For example, among gentrifiable neighborhoods (i.e., those with initial SES below the area median), Black neighborhoods have been the least likely, and White neighborhoods the most likely, to gentrify in earlier decades (Galster et al., 2003; Freeman and Cai, 2015; Owens and Candipan, 2019; Timberlake and Johns-Wolfe, 2017). Gentrification is further shaped by racialized perceptions of space which can influence investment patterns and exclusion from speculative capital (Rucks-Ahidiana, 2021). Additionally, when neighborhoods experience ascent, they tend to be driven by higher-SES White residents, particularly in initially majority Black and Hispanic neighborhoods, thereby leading to racial transition from majority-minority to mixed-race or predominantly White, thereby perpetuating neighborhood racial/ethnic hierarchy (Candipan and Bader, 2022; Owens and Candipan, 2019).

Mobility networks research has also increasingly acknowledged dynamics of racial stratification in how residents move through cities. Analyzing daily travel in the 50 largest cities, Wang et al. (2018) find that “race trumps class” in shaping mobility patterns between neighborhoods of varying socio-demographic compositions. Residents from poor White neighborhoods are more likely to visit non-poor White areas than residents from non-poor Black or Hispanic neighborhoods, suggesting that race more strongly structures neighborhood mobility patterns than SES. To that point, Candipan et al. (2021) introduced the segregated mobility index (SMI), which describes the extent to which residents travel (un)equally to other neighborhoods with different racial compositions, accounting for the racial composition of city neighborhoods. From the perspective of segregated mobility, the racial segregation of a city is driven by routine travels of city residents, as well as the structural connectedness of neighborhoods that these everyday movements produce. Residents of different racial/ethnic backgrounds often experience vastly different mobility patterns, traveling to different types of neighborhoods, which can reinforce existing inequalities and limit the potential for ascent in marginalized communities (Candipan et al., 2021; Wang et al., 2018). Moreover, mobility-based segregation can perpetuate economic disparities by restricting access to resources and opportunities predominantly found in more affluent, racially homogeneous neighborhoods (Browning et al., 2017b; Moro et al., 2021).

Data and measures

We combine several data sources from 2010 to 2019 for the 50 most populous cities in the US (see Online Appendix Table A1 for summary statistics for all analysis cities, designated by census-defined place boundaries in 2010). Our analysis includes both neighborhood- and city-level measures. We use census block groups as proxies for neighborhoods, following past work (Candipan et al., 2021; Phillips et al., 2021; Xu, 2022). Block groups, being smaller than tracts, help reduce within-unit heterogeneity while remaining large enough to minimize errors that might arise from using smaller geographical census units like blocks. Although census boundaries are administrative and may not perfectly match residents’ local perceptions of neighborhood boundaries (Poorthuis, 2018), census-based neighborhood definitions ensure consistency and comparability with prior studies on neighborhood change and segregation.

Dependent variable: Neighborhood socioeconomic ascent

Neighborhood characteristics come from decennial census and American Community Survey (ACS) 5-year estimates. To identify neighborhoods as having ascended, we first use factor analysis with principal components extraction to create neighborhood SES scores, drawing on ACS data in 2006–2010 and 2015–2019 (Owens and Candipan, 2019). This measure captures relative increases in both residents’ SES and housing values, consistent with quantitative work on gentrification and neighborhood socioeconomic change (Ellen and O’Regan, 2011; Owens, 2012; Owens and Candipan, 2019; Timberlake and Johns-Wolfe, 2017). Our inputs for factor analysis include five theoretically driven correlates of neighborhood SES capturing characteristics of residents and housing: median household income, share with a bachelor’s degree or higher, share in managerial/professional occupations, median home value, and median rent.¹ We compute neighborhood SES scores for baseline (2010) and end years (2019) of our study. We rank neighborhoods within cities for each year, then calculate change in a neighborhood’s percentile rank in SES score. Neighborhoods experiencing ascent are those whose percentile SES rank increases by 10 points from 2010 to 2019 (Owens and Candipan, 2019). Note that neighborhoods with an SES rank of 91 or higher are ineligible to ascend.

Key predictor: The segregated mobility index

Our study builds on recent work introducing city-level measures of structural connectedness based on everyday mobility—how neighborhoods are connected via the daily flows of residents in their activity space travels (Candipan et al., 2021; Phillips et al., 2021). Our key independent predictor, the segregated mobility index (SMI), quantifies the extent to which neighborhoods of given racial compositions are connected to other types of neighborhoods in equal measure (see Candipan et al., 2021, which introduced our measure). Under this mobility-based framework, the racial segregation of a city becomes the extent to which residents fail to travel to different types of neighborhoods with varying racial/ethnic compositions, controlling for the racial composition of a city’s neighborhoods. Prior work finds that residential segregation is a strong predictor of SMI, but also that patterns of travel remain only partially explained by where people reside (Candipan et al., 2021). This suggests that while neighborhood composition anchors many racial divides, SMI captures a distinct dimension of urban inequality, tracing how segregation unfolds beyond where residents live.

SMI is computed from a Twitter-derived database consisting of nearly 134 million geotagged tweets sent by more than 375,000 individuals over a 500-day period—1 October 2013, to 31 March 2015. The file provides highly detailed information on residents’ daily travel within cities and between different types of neighborhoods (Candipan et al., 2021). Although it lacks information on individual demographics, the data effectively captures neighborhood connections and structural mobility, which do not require individual-level demographic details. Home locations were identified using the DBSCAN clustering algorithm applied to tweets sent between 8 p.m. and midnight on weekdays (Candipan et al., 2021; Phillips et al., 2021; Wang et al., 2018), retaining only users active over a 2-month window while removing tourists and low-frequency users to ensure that mobility networks reflect city residents’ routine movement patterns.

Mobility networks (via edge lists) are then constructed for each city, with neighborhoods constituting the nodes and residents’ visits between neighborhoods representing the ties, drawing on network theory (Phillips et al., 2021). For each city, the mobility data report the proportion of unique visits from each individual’s home block group to all other block groups, generating weighted and directed edge lists that capture the average share of visits by residents from origin to destination block groups. This information is then aggregated to calculate the extent to which different types of neighborhoods—that is, majority Black, majority White, majority Hispanic, and mixed race (no majority)—are connected via residents’ daily travels, collapsing each city’s mobility network into a 4 × 4 matrix in which each element of the matrix represents the sum of the proportion of visits from each origin type to each of the four neighborhood types as destinations. This matrix becomes the observed race-based mobility network (MatObs) for each city.

SMI is computed by comparing this observed race-based 4 × 4 mobility matrix to two hypothetical extremes—a fully integrated matrix (MatInteg), in which travel occurs in proportion to the racial composition of neighborhood types citywide, and a fully segregated matrix (MatSeg), in which neighborhoods only connect to others of the same type—then normalizes this comparison relative to the city’s own racial composition of neighborhoods. The index equals the Hamming distance between MatObs and MatInteg, divided by the distance between MatSeg and MatInteg. The normalization scales the index from 0 (perfect integration) to 1 (complete segregation). This design ensures that cities are evaluated against what would be expected given their own structure, not an absolute standard, and thus SMI is constructed to be invariant to the number of block groups in a city since differences in the number of neighborhoods do not bias the measure. This permits meaningful cross-city comparisons that adjust for differences in network size and the racial composition of the city (Candipan et al., 2021; Phillips et al., 2021).

SMI values can range from 0 to 1, with higher values indicating greater segregated mobility patterns such that residents more frequently visit non-residential neighborhoods with similar racial compositions as their own, and lower values indicating that residents’ mobility patterns are more integrated such that their visits to neighborhoods of different racial compositions correspond proportionally with the racial composition of their city.

We dichotomize raw SMI values at the 75th percentile to classify cities as high-SMI (75th percentile or higher) or non-high (or “lower”)-SMI (below 75th percentile), capturing structural dynamics that may shape where neighborhood ascent occurs more frequently. This approach captures segregation in the connectivity of neighborhoods through aggregated mobility flows, reflecting not individual-level exposure, but the city-level racialized structure of inter-neighborhood ties (see Candipan et al. (2021) and Phillips et al. (2021) for additional technical details on the construction of SMI).

Sample selection considerations

Because Twitter users and those who geotag tweets may not fully represent the general population, and because tweet locations do not reflect all possible locations, Twitter-based mobility data may be subject to sample selection bias (Candipan et al., 2021). However, mobility patterns derived from this data have been validated in prior studies, showing strong alignment with patterns from other sources like cell phone GPS data and census commuting data (Phillips et al., 2021; Wang et al., 2018). For example, Lenormand et al. (2014) report correlations over 0.9 between Twitter- and cellphone-derived mobility flows in Madrid and Barcelona. In US cities, Gao et al. (2014) show similarly high correlations (0.91 on weekdays) between Twitter data and ACS-based commuting flows in Los Angeles. Further, Phillips et al. (2021), using our same corpus of tweets, validated mobility flows by comparing Twitter-derived neighborhood-level visitation (“in-degree”) in Houston to estimates from high-resolution GPS smartphone data. The correlation between the two sources was approximately 0.8, with strong overlap in spatial clustering of visited neighborhoods.

Importantly, our analyses center on structural properties of neighborhood connectivity at the city level, not characteristics of individual users. Thus, while individual-level demographic biases may exist, they are unlikely to bias our city-level estimates of SMI. Nonetheless, the potential biases in our mobility data should be considered when interpreting our findings.

Mobility-based neighborhood controls

We control for two neighborhood-level measures that capture flows of people to and from residential block groups. In-degree is calculated as the proportion of other block groups (within the same city) whose residents visit the focal neighborhood, theoretically reflecting its openness to outside consumption. Out-degree is measured as the proportion of a city’s block groups that its residents visit, potentially indicating residents’ networks of economic, social, or institutional resources beyond their home neighborhoods, which might indirectly relate to protective effects on ascent.

Other controls

We account for several baseline neighborhood-level characteristics that may confound the relationship between segregated mobility and neighborhood ascent: racial composition, SES rank, proportion of foreign-born residents, population density to control for differences in population, and proportion rental housing built in the decade prior to our baseline year (2000–2010). All measures are observed in 2010 and derived from ACS 2006 to 2010 estimates.

Our analysis further acknowledges structural city-level dynamics in neighborhood processes, which are often overlooked in neighborhood change research. Past work demonstrates that segregated mobility is associated with, but not solely shaped by, racial residential segregation (Candipan et al., 2021). We therefore account for racial residential segregation, captured as multigroup entropy index, also known as the multigroup version of Theil’s H (Reardon and Firebaugh, 2002). Multigroup entropy measures (un)evenness, describing the extent to which groups are distributed evenly across space and, more specifically, captures the difference between the racial diversity of the city and the weighted average diversity of individual block groups. To correspond with our SMI measure (in 2015), we use 2013–2017 ACS data to construct entropy (H) using five groups: Black, Hispanic, White, Asian, and other race. H ranges from 0 to 1 with lower values representing more racial integration and higher values representing greater racial segregation. The correlation between multigroup entropy and SMI is R = 0.74.²

We control for additional city-level factors that might influence the likelihood of neighborhood ascent, including income inequality between households, measured as the gini coefficient. Cities with higher income inequality may have greater frequency of ascent due to economic constraints motivating the residential decisions of higher-SES, but lower-income households, into lower-SES neighborhoods. We further account for differences in city size (number of block groups) and racial composition (proportion of non-Hispanic White residents).

Analysis

We perform two-level hierarchical models with neighborhoods nested within cities. Specifically, we perform mixed effects logistic regression with random intercepts that predict neighborhood socioeconomic ascent based on initial racial composition of the neighborhood and the segregated mobility of the city, accounting for neighborhood- and city-level factors that might confound this relationship. Because models examine the odds of neighborhood ascent, our analysis is restricted to block groups that are eligible to ascend.

Our baseline model examines whether odds of ascent differ in cities with higher and lower levels of segregated mobility, taking the following form:

\begin{matrix} logit (P r (Ascent = 1)) = B_{0} + B_{1} {SMI}_{j} \\ + B_{2} {SES}_{i j} + X_{i j} + Z_{j} + u_{j} \end{matrix}

Key predictors include our binary measure for SMI (whether a block group is located in a high-SMI city) and the relative rank of a neighborhood’s baseline socioeconomic score in 2010 (continuous, scaled 0–1). The coefficient for SMI indicates whether odds of ascent are higher in high-SMI cities, while the coefficient for SES rank represents whether odds of ascent increase as baseline neighborhood SES rank increases. All models include neighborhood-level mobility measures for in-degree and out-degree, adjust for a vector of baseline neighborhood characteristics (X) and city-level factors (Z), and include clustered-robust standard errors.

We then examine whether ascent is more likely among neighborhoods with lower or higher SES percentile ranks at baseline (2010) by including a cross-level interaction between SMI and SES. The interaction term (SMI × SES) indicates whether the relationship between SMI and ascent depends on initial neighborhood SES rank in 2010.

Next, we address whether the association between segregated mobility and ascent varies across five different types of neighborhoods based on initial racial composition—majority Black (>50%), Hispanic (>50%), White (>70%), Asian (>50%) or no majority (all other block groups).³ Note that Majority Asian neighborhoods represent a very small percentage of block groups (n = 744 block groups representing 2.4% of analysis file) and are geographically concentrated in a handful of cities. Because of this, our analysis focuses on the other neighborhood race types, though Majority Asian neighborhoods remain in our models.

This model includes an interaction between SMI and our 5-category majority race measure. Recall that initially high-SES ranked neighborhoods are ineligible to ascend (SES rank > 91 percentile points) and thus excluded from our analytic sample. Among tracts ineligible to ascend, roughly 76% were predominantly White while 20% had no racial majority. Our fully interacted model takes the following form:

\begin{matrix} logit (P r (Ascent = 1)) = B_{0} + B_{1} {SMI}_{j} \\ + B_{2} {SES}_{i j} + B_{3} ({SMI}_{j} \times {SES}_{i j}) + B_{4} {Type}_{i j} \\ + B_{5} ({SMI}_{j} \times {Type}_{i j}) + X_{i j} + Z_{j} + u_{j}, \end{matrix}

where Type denotes the 5-category majority race type of a neighborhood given its initial racial composition in 2010, with SMI × Type indicating the interaction between neighborhood majority race and its city’s SMI.

Finally, we examine whether the relationship between the segregated mobility of a city and neighborhood ascent holds net of residential racial segregation by including it as a control.

Results

Sample characteristics

Table 1 shows summary statistics of the analytic sample. Among the 50 cities in our analysis, the mean SMI was 0.262, ranging from 0.109 (Portland) to 0.495 (Detroit). The average SMI in high-SMI cities was 0.382 compared to the 0.224 for cities below the 75th percentile in SMI. Detroit had the highest SMI among all cities, and the highest level of residential segregation (Theil’s H = 0.526). Just over a quarter of block groups in our analysis file ascended from 2010 to 2019 Table 1.

Table 1.

Summary statistics of analysis sample.

	Data source	Mean(or proportion)	SD	Min	Max
Segregated Mobility Index (SMI)
All cities	2013–2015Twitter data;2013–2017American CommunitySurvey (ACS)	0.262	0.086	0.109	0.495
High SMI cities (≥75th percentile)		0.335	0.060	0.250	0.495
Non-high (lower) SMI cities(<75th percentile)		0.196	0.039	0.109	0.250
Neighborhood socioeconomiccharacteristics
Socioeconomic ascent (2010–2019)	2006–2010 and2015–2019 ACS	0.267
Non-ascent (2010–2019)		0.733
Percentile SES rank (in 2010)	2006–2010 ACS	45.5	26.0	0.0	91.0
Percentile SES rank in ascendant block groups		35.8	22.4	0.07	89.6
Percentile SES rank in non-ascendantblock groups		49.1	26.3	0.02	90.9
Neighborhood majority race type (in 2010)
Majority White	2010 decennialcensus	0.229
Majority Black		0.214
Majority Hispanic		0.210
No majority		0.325
Majority Asian		0.224
Neighborhood-based mobility
In-degree	2013–2015Twitter data;2013–2017 ACS	0.137	0.123	0	0.996
Out-degree		0.134	0.134	0	0.976
Baseline neighborhood characteristics (in 2010)
Population	2006–2010 ACS	1363	760	24	22129
Population density		17217	27245	0.3	633368
Proportion rental housing built 2000–2010		0.080	0.177	0	1
Proportion foreign born	2010 decennialcensus	0.215	0.171	0	0.873
Proportion Black		0.254	0.326	0	1
Proportion Hispanic		0.262	0.281	0	1
Proportion White		0.387	0.317	0	1
City-level characteristics (in 2010)
Income inequality (gini)	2006–2010 ACS	0.493	0.036	0.412	0.583
Number of block groups		770	923	250	6220
Proportion White	2010 decennial census	0.464	0.163	0.117	0.734
Theil’s multigroup entropy (5 groups)	2013–2017 ACS	0.294	0.099	0.132	0.526

Note: Analysis sample excludes block groups with percentile SES ranks of 91 or higher because they are ineligible to ascend (n = 3322 block groups; 8.95% of all block groups in full data file).

Baseline models

Table 2 displays results from regression models predicting neighborhood ascent by city-level segregated mobility. Model 1 displays our baseline model examining the relationship between SMI and ascent (all results presented as odds ratios). Neighborhoods in cities with greater segregated mobility are significantly more likely to ascend compared to neighborhoods located in cities with less segregated mobility—the odds of ascent are nearly 1.7 times higher for neighborhoods in high-SMI cities than neighborhoods in cities with more racially equitable mobility.

Table 2.

Mixed effects logistic regression predicting neighborhood ascent by city-level segregated mobility.

	SMI		SMI × SES rank
	Model 1		Model 2
	OR	SE	OR	SE
Key predictors
Neighborhood SES rank (2010)	0.972***	(0.001)	0.976***	(0.002)
High segregated mobility (SMI)	1.641***	(0.221)	2.375***	(0.366)
SMI × neighborhood SES rank			0.991***	(0.002)
Neighborhood-based mobility
In-degree	2.258***	(0.488)	2.152***	(0.445)
Out-degree	0.425***	(0.095)	0.435***	(0.098)
Baseline neighborhood controls (in 2010)
Population density (in 1000s)	1.000***	(0.000)	1.000***	(0.000)
Proportion rental housing built 2000–2010	1.273⁺	(0.174)	1.194	(0.168)
Proportion foreign born	0.441	(0.283)	0.507	(0.325)
Proportion Black	0.256***	(0.062)	0.250***	(0.061)
Proportion Hispanic	0.790	(0.181)	0.768	(0.177)
City-level controls (in 2010)
Income inequality (gini)	0.515	(0.623)	0.496	(0.600)
Number of block groups	1.000	(0.000)	1.000	(0.000)
Proportion White	1.444	(0.373)	1.486	(0.385)
Constant	2.103	(1.407)	1.824	(1.220)
Var (cons)	1.046*	(0.020)	1.047*	(0.020)
n (cities)	50		50
N (block groups)	33,244		33,244

Note: Results from mixed effects logistic regression model with random intercepts with cluster-robust standard errors (SEs) in parentheses. Values are presented as odds ratios (ORs). High SMI refers to cities at the 75th percentile of SMI values or higher.

p < 0.10. *p < 0.05. **p < 0.01. ***p < 0.001.

Mixed effects logistic regression predicting neighborhood ascent by city-level segregated mobility.

Both neighborhood-level mobility-based measures—in-degree and out-degree—significantly predict neighborhood ascent. Higher levels of out-degree, capturing residents’ routine travels to neighborhoods outside of their own, are associated with decreasing odds of ascent. On the other hand, the odds of ascent are higher in neighborhoods with greater in-degree suggesting that routine visits from outside residents are potentially exposing these neighborhoods to higher-SES in-movers. Importantly, greater SMI predicts a higher likelihood of ascent in neighborhood above and beyond its level of in- and out-degree flows. Population density is significantly and negatively associated with neighborhood ascent, while the proportion of rental housing built in the prior decade (2000–2010) is positively associated with ascent at the trend level.

We then examine whether the association between segregated mobility and ascent depends on a neighborhood’s SES rank in 2010 (Model 2). Figure 1 illustrates results in terms of adjusted predicted probabilities. While we observe the same general trend for neighborhoods located in both types of cities, the probability of ascent is significantly higher for neighborhoods in high-SMI cities (relative to those in lower-SMI cities), particularly in neighborhoods with below-median initial percentile SES rank.⁴

Does the relationship between segregated mobility and ascent depend on initial neighborhood racial composition?

Do the odds of ascent vary in neighborhoods with different initial predominate racial compositions after accounting for a city’s SMI? In this model, which includes our 5-category measure for neighborhood majority race type (reference category: Majority White), the main term for SMI remains significantly and positively associated with greater odds of neighborhood ascent (Table 3).⁵

Table 3.

Mixed effects logistic regression predicting neighborhood ascent (by majority race type and segregated mobility).

	Majority race
	Model 3		Model 4
	OR	SE	OR	SE
Key predictors
Neighborhood SES rank (2010)	0.979***	(0.002)	0.979***	(0.001)
High segregated mobility (SMI)	2.065***	(0.297)	1.956**	(0.505)
Neighborhood SES rank × SMI	0.992***	(0.002)	0.991***	(0.002)
Neighborhood majority race (in 2010)
White	(ref)
Black	0.447***	(0.087)	0.411***	(0.047)
Hispanic	0.873	(0.118)	0.938	(0.140)
No majority	0.871**	(0.045)	0.817**	(0.055)
Asian	0.940	(0.178)	0.993	(0.237)
Neighborhood majority race × SMI
White × SMI (ref)			(ref)
Black × SMI			1.182	(0.349)
Hispanic × SMI			0.890	(0.175)
No majority × SMI			1.247*	(0.134)
Asian × SMI			0.945	(0.196)
Neighborhood-based mobility
In-degree	1.885**	(0.386)	1.902**	(0.391)
Out-degree	0.473***	(0.107)	0.476***	(0.107)
Baseline neighborhood controls (in 2010)
Population density (in 1000s)	1.000***	0.000	1.000***	0.000
Proportion rental housing built 2000–2010	1.122	(0.154)	1.133	(0.152)
Proportion foreign born	0.725	(0.456)	0.713	(0.430)
City-level controls (in 2010)
Income inequality (gini)	0.840	(0.951)	0.699	(0.854)
Number of block groups	1.000	(0.000)	1.000	(0.000)
Proportion White	1.605*	−0.375	1.691*	−0.419
Constant	0.987	(0.607)	1.058	(0.680)
Var (cons)	1.030**	(0.012)	1.029*	(0.012)
n (cities)	50		50
N (block groups)	33,244		33,244

Note: Results from mixed effects logistic regression model with random intercepts with cluster-robust standard errors (SEs) in parentheses. Values are presented as odds ratios (ORs). Majority race corresponds with neighborhood racial composition in 2010. High SMI refers to cities at the 75th percentile of SMI values or higher.

+p < 0.10. *p < 0.05. **p < 0.01. ***p < 0.001.

Mixed effects logistic regression predicting neighborhood ascent (by majority race type and segregated mobility).

Majority White neighborhoods are significantly more likely to ascend compared to majority Black and mixed (no majority) neighborhoods, accounting for SMI (Model 3). Additionally, Majority Black neighborhoods are significantly less likely to ascend compared to all non-Black neighborhoods. The interaction term between no-majority neighborhood type and SMI is positive and significant, indicating that their odds of ascent depend on the structural connectedness of a city’s neighborhoods—no-majority neighborhoods are significantly more likely to ascend in high-SMI cities (Model 4).⁶ SMI does not have a significant interactive association with any majority race neighborhood type.

Does mobility-based segregation relate to ascent across majority race neighborhoods after accounting for residential segregation?

Finally, we account for the racial residential segregation level of the city (Table 4). Overall, the positive relationship between SMI and ascent, and the interaction between SMI and SES, remains when controlling for racial segregation (Model 5). That is, the racially segregated patterns of where residents travel for daily routines relates to neighborhood socioeconomic ascent above and beyond the racially segregated patterns of where residents live.

Table 4.

Mixed effects logistic regression analyzing associations between city-level segregated mobility and residential segregation with neighborhood ascent (by majority race type).

	All block groups		Majority race
	Model 5		Model 6
	OR	SE	OR	SE
Key predictors
Neighborhood SES rank (2010)	0.976***	(0.002)	0.979***	(0.001)
High segregated mobility (SMI)	1.769***	(0.254)	1.580⁺	(0.398)
Neighborhood SES rank × SMI	0.991***	(0.002)	0.991***	(0.002)
Racial residential segregation
Theil’s multigroup entropy (5 groups)	7.787***	(4.413)	3.974**	(1.827)
Neighborhood majority race (in 2010)
White	(ref)		(ref)
Black			0.398***	(0.044)
Hispanic			0.925	(0.135)
No majority			0.815**	(0.054)
Asian			1.004	(0.238)
Neighborhood majority race × SMI
White × SMI (ref)	(ref)		(ref)
Black × SMI			1.214	(0.359)
Hispanic × SMI			0.900	(0.172)
No majority × SMI			1.252*	(0.134)
Asian × SMI			0.936	(0.192)
Neighborhood-based mobility
In-degree	2.231***	(0.457)	1.964***	(0.402)
Out-degree	0.442***	(0.099)	0.484**	(0.109)
Baseline neighborhood controls (in 2010)
Population density (in 1000s)	1.000***	(0.000)	1.000***	(0.000)
Proportion rental housing built 2000–2010	1.188	(0.167)	1.126	(0.152)
Proportion foreign born	0.503	(0.317)	0.709	(0.424)
Proportion Black	0.243***	(0.060)
Proportion Hispanic	0.749	(0.166)
City-level controls (in 2010)
Income inequality (gini)	0.836	(0.995)	0.980	(1.204)
Number of block groups	1.000	(0.000)	1.000	(0.000)
Proportion White	1.072	(0.275)	1.359	(0.333)
Constant	1.015	(0.631)	0.716	(0.448)
Var (cons)	1.031**	(0.012)	1.022*	(0.009)
n (cities)	50		50
N (block groups)	33,244		33,244

Note: Results from mixed effects logistic regression model with random intercepts with cluster-robust standard errors (SEs) in parentheses. Values are presented as odds ratios (ORs). Majority race corresponds with neighborhood racial composition in 2010. High SMI refers to cities at the 75th percentile of SMI values or higher.

p < 0.10. *p < 0.05. **p < 0.01. ***p < 0.001.

Mixed effects logistic regression analyzing associations between city-level segregated mobility and residential segregation with neighborhood ascent (by majority race type).

However, as with our previous set of models (in Table 3), we observe differences by neighborhood type. Following best practices for logistic regression models with interactions (Mize, 2019), Figure 2 presents adjusted predicted probabilities of ascent across neighborhood majority race types by high- and lower-SMI, accounting for residential segregation, and drawing on results from our fully interacted models (Model 6). Majority Black neighborhoods are substantially less likely to ascend than all other neighborhoods. However, among Majority Black neighborhoods, their probability of ascending is greater in high-SMI cities. On the other hand, no-majority neighborhoods are significantly more likely to ascend in high-SMI cities than in lower-SMI cities. White neighborhoods are among the most likely to ascend, regardless of a city’s SMI level. The likelihood of ascent is also similar among Hispanic neighborhoods in lower- and high-SMI cities, with the probability of ascent somewhat lower than in White neighborhoods (although not significantly different). While no-majority neighborhoods are less likely than White neighborhoods to ascend in lower-SMI cities, they are just as likely to ascend as White neighborhoods in high-SMI cities.

Robustness checks

Finally, we performed a series of models to ensure that our main findings were robust to a number of alternative specifications. This included testing different SMI thresholds under binary (high-SMI as above a city’s median SMI value) and continuous (raw SMI values) definitions. Additionally, we performed models: (1) analyzing census tracts instead of block groups; (2) controlling for alternate measures of residential segregation and city racial composition; (3) omitting either or both of our block group-level in-degree or out-degree measures; (4) and omitting the largest cities in terms of the number of block groups (New York, Los Angeles, and Chicago). Our results held across all models (Online Appendix Table A2).

Conclusion

Through their routine travels, people engage with multiple areas throughout the city, establishing broader social networks that influence access to resources, social capital, and economic opportunities (Wang et al., 2018). Our study contributes to a small but growing literature on mobility-based networks that aims to extend traditional notions of neighborhood effects studies by highlighting the significance of everyday movement, the interconnectedness of urban neighborhoods, and their consequences for community well-being (Brazil, 2022; Brazil et al., 2025; (Phillips et al., 2021; Sampson, 2012). Results from our study contribute new knowledge to the study of neighborhood ascent, segregation, activity spaces, and urban inequality, more broadly.

Our analysis takes advantage of advances in computational techniques and novel large-scale data sources to examine new questions about neighborhood change and inequality (Candipan and Tollefson, 2025; Hwang and McDaniel, 2022; Poorthuis et al., 2022; Sampson and Candipan, 2023). Combining data on individual movement with administrative data on neighborhood and city characteristics, we find that whether a neighborhood ascends generally depends on the level of segregated mobility within its city. Neighborhoods in cities with greater segregated mobility are significantly more likely to ascend compared to neighborhoods located in cities with less segregated mobility, especially neighborhoods with lower initial socioeconomic standing. Moreover, this relationship is not reflective of a city’s residential segregation patterns, as segregated mobility is associated with ascent above and beyond a city’s residential segregation levels. Our findings also hold after accounting for neighborhood-level mobility, with higher neighborhood in-degree (in-bound visits) predicting greater odds of neighborhood ascent, and higher out-degree (residents’ outbound visits) predicting lower odds of ascent, aligning with past work predicting neighborhood change via shifting movement patterns of individuals and outsiders (Chapple et al., 2021; Poorthuis et al., 2022).

Our results reveal clear racial stratification processes in patterns of neighborhood ascent, as the probability of ascent differs by initial neighborhood racial composition. Aligned with recent work, we find differences in the likelihood of ascent across majority White, Black, Hispanic, and racially mixed (no majority) neighborhoods (Owens and Candipan, 2019; Rucks-Ahidiana, 2021). Across all cities, White neighborhoods are more likely to ascend than all other types of neighborhoods while Black neighborhoods are the least likely to ascend, highlighting how deeply entrenched racial hierarchies shape urban neighborhood trajectories. This finding for White neighborhoods, along with the fact that they overwhelmingly comprise the highest-SES block groups ineligible to ascend, further demonstrates their advantaged position among the racial hierarchy of neighborhoods—they have less room for upward movement within a relative ascent framework, and yet they are more likely to ascend than other neighborhoods. On the other hand, Black neighborhoods comprise about 34% of all neighborhoods in high-SMI cities compared to 23% in lower-SMI cities (see Online Appendix Table A3), indicating substantial overrepresentation in the former. Yet despite this concentration in high-SMI cities, Black neighborhoods remain the least likely to ascend in either context. This combination—overrepresentation in high-SMI cities and persistently low ascent rates—underscores how Black neighborhoods continue to be largely bypassed during processes of economic neighborhood upgrading, even in cities where ascent is generally more common. When Black (and Hispanic) neighborhoods do ascend, however, their ascent tends to be accompanied by White racial transition (Candipan and Bader, 2022; Owens and Candipan, 2019).

No-majority neighborhoods were the most likely to ascend in high-SMI cities. In these contexts, mixed-race neighborhoods may be perceived as more accessible to higher-SES in-movers as racially “transitional” or “neutral” spaces or may function as buffer zones where more racially homogeneous Black and Hispanic neighborhoods remain symbolically or materially excluded, suggesting a relational dynamic in which movers or investors are choosing mixed-race neighborhoods over Black and Hispanic neighborhoods in high-SMI cities. On the other hand, whether Hispanic neighborhoods are more likely to ascend does not depend on their location in lower- or high-SMI cities. That their probability of ascent mirrored that of White neighborhoods perhaps indicates that these neighborhoods align with the common gentrification narrative of lower-income Hispanic neighborhoods being ripe for neighborhood change (Zukin, 1987). And while majority-Black neighborhoods are the least likely to experience ascent overall, they are somewhat more likely to do so in high-SMI cities.

While residential segregation shapes how racially divided mobility patterns are, it does not fully explain them (Candipan et al., 2021). This underscores that where people live structures racial separation to an extent, while SMI captures a distinct dimension of how these divisions extend beyond residential areas through routine travel that links neighborhoods across the city. While the association between SMI, neighborhood racial composition, and ascent held net of cities’ underlying residential patterns, we also observed that ascent was more likely in cities with higher levels of residential segregation, perhaps reflecting how economic investments and ascent become concentrated in specific areas within otherwise sharply divided urban landscapes.

Our overall findings underscore how citywide mobility structures and racialized neighborhood change intersect. Our analyses reveal that the association between segregated mobility and neighborhood socioeconomic ascent is driven primarily by racially mixed neighborhoods (roughly one-third of our sample), which are the only neighborhood type to show a significant difference in the likelihood of ascent across high- and low-SMI cities. In cities with greater mobility-based segregation, boundaries between racially distinct neighborhood types may be more clearly maintained in residents’ activity spaces. These boundaries may in turn shape how neighborhoods are perceived and acted upon by institutions, developers, and higher-SES residents. The racialized visibility of space may make racially mixed neighborhoods more legible and actionable targets for reinvestment, thereby producing sharper patterns of economically upward neighborhood reordering. Under conditions of higher SMI, local institutional actors (e.g., city governments, nonprofits, and developers) may view racially mixed neighborhoods as socially and politically feasible targets for redevelopment, and as transitional spaces that bridge racially homogeneous neighborhoods while still being perceived as more viable (Huante, 2021; Rucks-Ahidiana, 2022). This dynamic could concentrate investments in mixed neighborhoods even as Black and Hispanic neighborhoods remain underinvested or bypassed altogether. In contrast, in lower-SMI cities, mobility flows are more integrated across racial lines, potentially reducing the salience of spatial racial boundaries. While this may reflect broader social mixing in daily life, it may also diffuse exposure and speculative investment across a wider range of neighborhoods. In this context, fewer neighborhoods may rise sharply in relative rank, even if many experience moderate change. Even in lower-SMI cities, durable racial hierarchies in ascending neighborhoods persist, suggesting that greater structural connectedness between neighborhoods does not fully disrupt the underlying racial logics that govern which neighborhoods are perceived as accessible, investable, or desirable (Owens and Candipan, 2019; Rucks-Ahidiana, 2022).

Recent neighborhood-focused work increasingly acknowledges that structural city-level features matter for understanding neighborhood dynamics (Galster and Sharkey, 2017). Neighborhoods are interconnected within broader urban mosaics, with change processes embedded within broader macro-geographic contexts (Hwang, 2015; Sampson, 2012). This perspective draws attention beyond the immediate locality to factors such as regional economic trends, metropolitan policies, and the spatial distribution of resources and opportunities (Galster and Sharkey, 2017). This broader perspective also helps to explain how neighborhoods evolve and how neighborhood processes, such as segregation and socioeconomic ascent, are intertwined with wider urban dynamics (Hwang and McDaniel, 2022). Therefore, explicitly examining higher-level structural dynamics matters for understanding neighborhood racial stratification and change. Our study highlights the importance of a particular but overlooked higher-level structure—segregated mobility networks—in shaping neighborhood change processes.

While our study focuses on the structural connectedness of cities via racialized mobility flows between neighborhoods, we acknowledge that routine travel by visitors from outside the city (e.g., nearby suburban commuters) may also forge ties to urban neighborhoods (Hess and Hall, 2024). Such engagements (for work, recreation, consumption, etc.) can shape perceptions of urban neighborhoods among potential investors, developers, and residents (Zukin, 1987). Limited but repeated daytime presence of White and higher-SES suburbanites may signal economic vitality, safety, or desirability, indirectly spurring investment and ascent (Hwang and Ding, 2020). While our SMI measure captures a structural property of cities, reflecting how racially segregated neighborhood networks are based on residents’ routine movements, it does not directly incorporate extra-city flows. Future work that examines how external visitors connect neighborhoods within and beyond city boundaries could shed more light on neighborhood change dynamics.

Our study is not without limitations. SMI captures a single window of time. Prior work has found that neighborhood outcomes, such as crime, influences mobility patterns (Graif et al., 2017). It is possible that neighborhood ascent shifts movement patterns between residents of different neighborhoods, thereby changing the structural connectedness of cities. Future research with mobility data over a longer period could examine whether growing neighborhood ascent corresponds with increases or decreases in segregated mobility. Future work might also analyze whether changes in SMI are an outcome of gentrification rather than a predictor, or whether they relate reciprocally with each other. Additionally, our sample consists of neighborhoods in the 50 most populous cities, which are much more racially and economically diverse than the rest of the US (though this diversity does not always translate to less segregated neighborhoods compared to smaller cities with less diverse populations). Further, despite the seemingly greater opportunities for contact with more diverse activity spaces, residents in large diverse cities tend to engage in more racially differentiated travel. Future research should explore the relationship between race, mobility networks, and neighborhood change across a wider set of metropolitan areas. Limitations notwithstanding, our results demonstrate the need to think beyond residential neighborhoods to understand forces that connect them to the rest of city that shape ascent.

Supplemental Material

sj-docx-1-usj-10.1177_00420980251408585 – Supplemental material for Race, segregated urban mobility, and socioeconomic ascent: A neighborhood network approach

Supplemental material, sj-docx-1-usj-10.1177_00420980251408585 for Race, segregated urban mobility, and socioeconomic ascent: A neighborhood network approach by Jennifer Candipan and Noli Brazil in Urban Studies

Footnotes

Acknowledgements

The authors thank Matt Hall and audience members at the Population Association of America for helpful feedback, as well as the editors and anonymous reviewers who provided valuable input throughout the review process. We also thank Tai Tworek for excellent research assistance.

ORCID iDs

Jennifer Candipan

Noli Brazil

Funding

The authors disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: The corresponding author is grateful to the Population Studies and Training Center at Brown University, which receives Funding from the NIH (P2C HD041020), for general support.

Declaration of conflicting interests

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

Supplemental material

Supplemental material for this article is available online.

Notes

References

Brazil

(2022) Environmental inequality in the neighborhood networks of urban mobility in US cities. Proceedings of the National Academy of Sciences 119(17): e2117776119.

Brazil

Candipan

Levy

, et al. (2025) Beyond the residential neighborhood: A scoping review of research on urban neighborhood networks. Social Science & Medicine 372: 117945.

Browning

Soller

(2014) Moving beyond neighborhood: Activity spaces and ecological networks as contexts for youth development. Cityscape (Washington, DC) 16(1): 165.

Browning

Calder

Krivo

, et al. (2017a) Socioeconomic segregation of activity spaces in urban neighborhoods: Does shared residence mean shared routines? The Russell Sage Foundation Journal of the Social Sciences 3(2): 210–231.

Browning

Calder

Soller

, et al. (2017b) Ecological networks and neighborhood social organization. American Journal of Sociology 122(6): 1939–1988.

Cagney

York Cornwell

Goldman

, et al. (2020) Urban mobility and activity space. Annual Review of Sociology 46: 623–648.

Cai

Browning

Cagney

(2024) Exposure of neighborhood racial and socio-economic composition in activity space: A new approach adjusting for residential conditions. Social Forces 103(1): 350–373.

Candipan

Bader

(2022) The dual role of race and immigration among ascending neighborhoods in US metropolitan areas. Population Research and Policy Review 41(4): 1725–1756.

Candipan

Tollefson

(2025) Machine learning and large-scale data for understanding urban inequality. In: Borch

Pardo-Guerra

(eds) The Oxford Handbook of the Sociology of Machine Learning. Oxford Handbooks. https://doi-org/10.1093/oxfordhb/9780197653609.013.38

10.

Candipan

Phillips

Sampson

, et al. (2021) From residence to movement: The nature of racial segregation in everyday urban mobility. Urban Studies 58(15): 3095–3117.

11.

Chapple

Poorthuis

Zook

, et al. (2021) Monitoring streets through tweets: Using user-generated geographic information to predict gentrification and displacement. Environment and Planning B 49: 704–721.

12.

De la Prada

ÀG

Small

(2024) How people are exposed to neighborhoods racially different from their own. Proceedings of the National Academy of Sciences 121(28): e2401661121.

13.

Delmelle

(2017) Differentiating pathways of neighborhood change in 50 US metropolitan areas. Environment and Planning A 49(10): 2402–2424.

14.

Ellen

O’Regan

(2011) How low income neighborhoods change: Entry, exit, and enhancement. Regional Science and Urban Economics 41(2): 89–97.

15.

Ellis

Wright

Parks

(2004) Work together, live apart? Geographies of racial and ethnic segregation at home and at work. Annals of the American Association of Geographers 94(3): 620–637.

16.

Freeman

Cai

(2015) White entry into black neighborhoods: Advent of a new era? The Annals of the American Academy of Political and Social Science 660(1): 302–318.

17.

Galster

Sharkey

(2017) Spatial foundations of inequality: A conceptual model and empirical overview. The Russell Sage Foundation Journal of the Social Sciences 3(2): 1–33.

18.

Galster

Quercia

Cortes

, et al. (2003) The fortunes of poor neighborhoods. Urban Affairs Review 39(2): 205–227.

19.

Gao

Yang

Yan

, et al. (2014). Detecting origin-destination mobility flows from geotagged tweets in greater Los Angeles area. In: Proceedings of the eighth international conference on geographic information science, Vienna, Austria, 24–26 September 2014, pp. 1–4.

20.

Gibbons

Barton

Reling

(2020) Do gentrifying neighbourhoods have less community? Evidence from Philadelphia. Urban Studies 57: 1143–1163.

21.

Graif

Lungeanu

Yetter

(2017) Neighborhood isolation in Chicago: Violent crime effects on structural isolation and homophily in inter-neighborhood commuting networks. Social Networks 51: 40–59.

22.

Hall

Iceland

(2019). Racial separation at home and work: Segregation in residential and workplace settings. Population Research and Policy Review 38: 671–694.

23.

Hess

Hall

(2024). Daily diversity flows: Racial and ethnic context between home and work. Demography 61(5): 1483–1508.

24.

Huante

(2021) A lighter shade of brown? Racial formation and gentrification in Latino Los Angeles. Social Problems 68(1): 63–79.

25.

Hwang

(2015) Gentrification in changing cities: Immigration, new diversity, and racial inequality in neighborhood renewal. The Annals of the American Academy of Political and Social Science 660(1): 319–340.

26.

Hwang

Ding

(2020). Unequal displacement: Gentrification, racial stratification, and residential destinations in Philadelphia. American Journal of Sociology 126(2): 354–406.

27.

Hwang

McDaniel

(2022) Racialized reshuffling: Urban change and the persistence of segregation in the twenty-first century. Annual Review of Sociology 48(1): 397–419.

28.

Hyra

(2017) Race, Class, and Politics in the Cappuccino City. University of Chicago Press.

29.

Kang

Knaap

Rey

(2022) Changes in the economic status of neighbourhoods in US metropolitan areas from 1980 to 2010: Stability, growth and polarisation. Urban Studies 59(13): 2774–2800.

30.

Kern

(2016) Rhythms of gentrification: Eventfulness and slow violence in a happening neighbourhood. Cultural Geographies 23: 441–457.

31.

Krivo

Washington

Peterson

, et al. (2013) Social isolation of disadvantage and advantage: The reproduction of inequality in urban space. Social Forces 92(1): 141–164.

32.

Kwan

(2013) Beyond space (as we knew it): Toward temporally integrated geographies of segregation, health, and accessibility. Annals of the American Association of Geographers 103(5): 1078–1086.

33.

Lees

(2008) Gentrification and social mixing: Towards an inclusive urban renaissance? Urban Studies 45(12): 2449–2470.

34.

Lenormand

Picornell

Cantú-Ros

, et al. (2014). Cross-checking different sources of mobility information. PloS One 9(8), e105184.

35.

Levy

Phillips

Sampson

(2020) Triple disadvantage: Neighborhood networks of everyday urban mobility and violence in US cities. American Sociological Review 85(6): 925–956.

36.

Mize

(2019) Best practices for estimating, interpreting, and presenting nonlinear interaction effects. Sociological Science 6: 81–117.

37.

Moro

Calacci

Dong

, et al. (2021) Mobility patterns are associated with experienced income segregation in large US cities. Nature Communications 12(1): 4633.

38.

Owens

(2012) Neighborhoods on the rise: A typology of neighborhoods experiencing socioeconomic ascent. City & Community 11(4): 345–369.

39.

Owens

Candipan

(2019) Racial/ethnic transition and hierarchy among ascending neighborhoods. Urban Affairs Review 55(6): 1550-1578.

40.

Phillips

Levy

Sampson

, et al. (2021) The social integration of American cities: Network measures of connectedness based on everyday mobility across neighborhoods. Sociological Methods & Research 50(3): 1110-1149.

41.

Poorthuis

(2018) How to draw a neighborhood? The potential of big data, regionalization, and community detection for understanding the heterogeneous nature of urban neighborhoods. Geographical Analysis 50(2): 182–203.

42.

Poorthuis

Shelton

Zook

(2022) Changing neighborhoods, shifting connections: Mapping relational geographies of gentrification using social media data. Urban Geography 43(7): 960–983.

43.

Reardon

Firebaugh

(2002) Measures of multigroup segregation. Sociological Methodology 32(1): 33-67

44.

Rucks-Ahidiana

(2021) Racial composition and trajectories of gentrification in the United States. Urban Studies 58(13): 2721–2741.

45.

Rucks-Ahidiana

(2022) Theorizing gentrification as a process of racial capitalism. City & Community 21(3): 173–192.

46.

Sampson

(2012) Great American City: Chicago and the Enduring Neighborhood Effect. University of Chicago Press.

47.

Sampson

Candipan

(2023) A comparative network approach to the study of neighborhood-and city-level inequality based on everyday urban mobility. In: LeGales

Robinson

(eds) The Routledge Handbook of Comparative Global Urban Studies. Routledge, pp. 175–184.

48.

Timberlake

Johns-Wolfe

(2017) Neighborhood ethnoracial composition and gentrification in Chicago and New York, 1980 to 2010. Urban Affairs Review 53(2): 236–272.

49.

Wang

Phillips

Small

, et al. (2018) Urban mobility and neighborhood isolation in America’s 50 largest cities. Proceedings of the National Academy of Sciences 115(30): 7735–7740.

50.

(2022) The contingency of neighbourhood diversity: Variation of social context using mobile phone application data. Urban Studies 59(4): 851–869.

51.

Zuk

Bierbaum

Chapple

, et al. (2015) Gentrification, Displacement and the Role of Public Investment: A Literature Review. Federal Reserve Bank of San Francisco, Col. 79.

52.

Zukin

Lindeman

Hurson

(2017) The omnivore’s neighborhood? Online restaurant reviews, race, and gentrification. Journal of Consumer Culture 17(3): 459–479.

53.

Zukin

(1987) Gentrification: Culture and capital in the urban core. Annual Review of Sociology 13(1): 129–147.

Supplementary Material

Please find the following supplemental material available below.

For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.

For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.

0.00 MB

0.06 MB