Spatial Segregation by Class and Race in Neighborhoods and Workhoods across U.S. metro areas

Spatial Segregation by Class and Race in Neighborhoods and Workhoods across U.S. metro areas | Research Square window.SnipcartSettings = { analytics: { enabled: false } }; (function() { var accessVector = localStorage.getItem('access_vector') || ''; window.dataLayer = window.dataLayer || []; if (accessVector) { window.dataLayer.push({ user: { profile: { profileInfo: { snid: accessVector } } } }); } })(); (function(w,d,s,l,i){w[l]=w[l]||[];w[l].push({'gtm.start':new Date().getTime(),event:'gtm.js'});var f=d.getElementsByTagName(s)[0],j=d.createElement(s),dl=l!='dataLayer'?'&l='+l:'';j.async=true;j.src='https://www.googletagmanager.com/gtm.js?id='+i+dl;f.parentNode.insertBefore(j,f);})(window,document,'script','dataLayer','GTM-K279D39R'); Browse Preprints In Review Journals COVID-19 Preprints AJE Video Bytes Research Tools Research Promotion AJE Professional Editing AJE Rubriq About Preprint Platform In Review Editorial Policies Our Team Advisory Board Help Center Sign In Submit a Preprint Cite Share Download PDF Article Spatial Segregation by Class and Race in Neighborhoods and Workhoods across U.S. metro areas Shin Bin Tan, Priyanka deSouza This is a preprint; it has not been peer reviewed by a journal. https://doi.org/ 10.21203/rs.3.rs-6771912/v1 This work is licensed under a CC BY 4.0 License Status: Posted Version 1 posted You are reading this latest preprint version Abstract To date, studies examining patterns of workplace area segregation, or workhoods, have been few and limited in both geographical scope and temporal coverage. While available studies have observed positive correlations between residential and workhood segregation, few of these studies attempt to unpack the nuances of this relationship by identifying the factors, such as transportation networks, that might moderate this positive link. Speaking to this gap, we study the relationship between residential and workplace racial/ethnic and socioeconomic segregation of 380 metropolitan statistical areas within the U.S., from 2011 to 2018. Using two-way fixed effects models, we find that the positive correlations between changes in residential and workhood segregation are significantly modified by changes in transportation and urban-form related characteristics, economic characteristics, as well as population diversity. Social science/Geography Social science/Environmental studies Social science/Sociology Figures Figure 1 Figure 2 Main In the United States (U.S.), residential segregation, or the spatial sorting of people into neighborhoods based on race and/or class, has generated disparities in terms of access to good schools 1 , nutritious food 2,3 , jobs 4–6 , health care 7 , and other critical amenities or services 8 . Scholars have thus argued that residential segregation is the ‘linchpin’ sustaining systemic inequalities in the U.S. 9,10 The traditional focus of segregation in residential neighborhoods however fails to account for other important locations where people spend time and built ties, such as within and around their workplaces 11–13 . Two approaches have been taken in investigating work-related segregation. The first focuses on examining racial/ethnic, gender, and/or income-based segregation across different employment sectors, or firm types 13–16 . The second, more nascent, approach examines spatial interaction at neighboring workplaces co-located within an area, and is what we refer to in this paper as ‘workhood segregation’. Studies include one evaluating residential and work tract segregation levels for native-born and immigrant groups in Los Angeles (L.A) using 1990 Census of Population and Housing data 17 , a second using 2015 data from the LEHD Origin-Destination Employment Statistics (LODES) in L.A. 18 and a third using Census Transportation Planning Package (CTPP) data to evaluate US metropolitan areas’ workhood segregation in 2000 and 2010 19 . All three articles found positive relationships between residential and workhood racial segregation, and that racial/ethnic residential segregation was higher than workhood segregation 17–19 . For class-specific segregation, the third study found the inverse to be true 18 . An explanation for the positive correlation between residential and workhood segregation is that each sphere of segregation generates and reinforces the other 17,19 in a ‘vicious circle of segregation’ 20 . To reduce commuting costs, people tend to choose to work closer to where they live and vice versa 21–23 . People also tend to obtain employment opportunities from their social networks 24–26 . Reliance on personal referrals, or referral hiring, has been linked to greater workplace segregation 27,28 . One source of job referrals are one’s neighbors 29,30 . If neighbors are also more likely to be work colleagues, one would also reasonably expect a degree of concordance between residential and workhood profiles. While previous studies have laid important exploratory groundwork, these tend to be temporally or geographically limited in their coverage. Our study takes a more comprehensive look at the relationship between residential and workhood racial and class segregation of all 380 U.S. metropolitan statistical areas (MSAs), defined according to 2020 decennial census geographical boundaries. Residential and workhood segregation were estimated for years 2011 to 2018, using LODES (version 8.3) data, which captured home and work locations of workers 14 years and older. To measure racial segregation, we calculated a multigroup dissimilarity index for four race categories (‘Asian’, ‘Black’, ‘White’, and ‘Others’), while for class-based segregation, we used four education levels (‘Less than High School’, ‘High School’, ‘Some College or Associate Degree’, ‘Bachelor’s Degree or advanced degree’). Higher dissimilarity scores indicate higher levels of segregation of the defined groups. First, we examine how MSA-levels of workplace racial and class-based segregation compared to residential segregation over the years, and whether there were significant positive correlations between MSA level class-based segregation and racial segregation, across both work and residential domains. Second, as there are a lack of studies that explore how to reduce, the ‘vicious circle of segregation’ 20 , our study examines two categories of metropolitan characteristics that might moderate the relationship between residential and workhood segregation: 1. transportation infrastructure and urban form; 2. Overall diversity and inclusiveness within MSAs, as well as labor market conditions. Figure 1 below summarizes our proposed moderating pathways, which we explain further in the following paragraphs. Residents living in compact MSAs with good transportation connectivity are likely to have lower travel costs, in terms of travel distance and commuting time –though the variation and magnitude of the relationship between urban form and commuting behavior has been the source of some debate 31–34 . Lower travel costs, in turn, reduce spatial mismatch by expanding exposure and opportunities for intergroup interactions, while also increasing spatial access to job opportunities available to racial/ethnic minority groups, even if they live in segregated areas 35,36 . Specifically, greater public transit access can help reduce spatial mismatch, given the higher use of transit amongst lower-income and/or minority groups 37–39 . MSAs where residents have better transport access (especially in terms of public transit), and shorter commutes would potentially be ones with a weaker link between residential and workplace segregation. Alternatively, better transportation infrastructure could strengthen the link between residential segregation, spatial mismatch, and workplace segregation. In the U.S., transportation infrastructure has been observed to benefit wealthier and White residents 40–44 , with resarch showing how better spatial access to jobs via transit benefitted White neighborhoods but not Black or Latino ones 4 , and that White workers in segregated metropolitan areas with efficient transit and pedestrian networks earned more than those in comparable areas with less efficient networks 35 . However, other empirical analyses suggest that spatial accessibility by income and ethnic/racial minority group membership are complex and city-specific 45 , with some finding higher spatial accessibility among low income and/or minority urban areas 46,47 while others observed the converse 48,49 . Reduced travel constraints might also further accentuate ‘ethnic niching’ – a phenomenon where members of an ethnic group are concentrated at a higher level than members of other groups in certain labor market sectors. This accentuation might occur because social networks that link workers to ethnic employment niches would have fewer travel or spatial information constraints, and can better fend off competition from other groups that were previously more well-located 50 . There has however been little empirical analysis of this hypothesis. As discrimination likely increases the extent to which marginalized groups need to rely on social networks for employment opportunities 51 , more inclusive metropolitan areas might have a reduced link between residential segregation and workhood segregation . Citiesthat have a more diverse, less White population profile may be less prejudiced due to increased intergroup interactions 52–55 . Conversely, the ‘threat’ hypothesis suggests that such areas might also be sites of increased tension and discrimination 56,57 . A more direct measure of whether a metropolitan area discriminates against racial/ethnic minority groups would be the degree of racial disparities in incomes or employment rates. Hiring racial discrimination in hiring might mean minority groups have to rely heavily on personal networks to obtain jobs, thus reinforcing the links between residential and workplace racial segregation. Similarly, workers in a metropolitan area with higher income inequality might have higher reliance on social networks to find jobs, compared to areas with lower wage inequality, and thus experience a stronger link between residential and workhood segregation. The state of the metropolitan labor market might be another moderating factor.Higher unemployment rates might create conditions where personal networks are less effective because one’s networks might be competing for employment and thus prove unwilling or unable to provide referrals 58 . Increased competition for lower-wage jobs that would otherwise only be attractive to more marginalized groups could also reduce propensity for ethnic group niching 52 . On the flip side, social networks might act as an additional competitive edge for securing jobs in a recession. Employers in weak markets might also rely more heavily on referrals, which is cheaper than advertising or using recruitment agencies 58 . Finally, a metropolitan area’s economic base might moderate the link between residential segregation and workhood segregation. Contacts are more likely to be used when filling managerial, professional, or sale/service positions and are less likely to be used in the public sector and in establishments that are part of multi-site firms 59 . Others have argued that the reason certain industries and professions, particularly low-wage ones, tend to be more ethnically ‘niched’, is because job-seeking in these industries are driven in part by heavy reliance on immigrants’ social networks 60–62 . To test our proposed moderating pathways ( Figure 1) , we fit a series of two-way fixed effects (2FE) regression models with MSA and year fixed effects. These models allow estimation of how within-MSA changes in workhood segregation might be associated with changes in residential segregation and the latter’s interaction with the hypothesized moderating variables, while controlling for other time-varying effects over the years. The threshold for statistical significance is set at p-value<0.05. We started by fitting a series of models that included only one MSA-level moderator variable and its interaction with residential segregation score, together with the MSA-level controls. Moderator variables with modelled coefficients of significance p<0.05, were then included in a combined model, which was then further streamlined in our final model to include only significant moderator variables and their interactions. The main paper focuses on interpreting the final model, while results from the preceding models are reported in Supplementary Annex 3 . Results Relationship between Residential and Workhood Segregation by Race and Class From 2011 to 2018, MSA residential racial segregation, as measured by spatial dissimilarity index of workers classified as ‘Asian’, ‘Black’, ‘White’ and ‘Other’, was positively correlated with workhood racial segregation, with this positive correlation increasing over time (Pearson’s correlation coefficient r ranging from 0.47 to 0.57). Residential racial segregation levels were alsoconsistently higher than workhood segregation across the years, at around 0.31 while the latter remained around 0.17. Supplementary Figure 1.1 summarizes. In contrast, workhood segregation by social class, measured by spatial dissimilarity index of workers of four educational attainment categories, was higher than residential segregation across the years, at around 0.1 compared to 0.06 for the latter. There was a small reduction in the positive correlation between workhood and residential class segregation over the years ( r ranging from 0.67 to 0.61). We observed no clear directional trend over time in terms of absolute levels of either racial or class-based segregation across both spheres. Supplementary Figure 1.2 summarizes. There were 29 ‘outlier’ MSAs that had lower racial residential rather than workhood segregation, and even fewer (n= 9 MSAs) where there were lower class-based workplace segregation rather than residential segregation estimates.Supplementary Annex 2 provides details of these MSAs and their characteristics. We see low levels of correlation between race and class-based segregation within the residential sphere (r ~ 0.11 to 0.16) and higher levels, albeit still moderately low, within the workhoods (r ~ 0.19 to 0.27). Both correlations seemed to slightly trend upwards between 2011-2018 (Supplementary Figure 1.3 ) . Moderators of the Relationship Between Residential And Workhood Segregation Table 1, Model A shows how increases in residential racial segregation, as estimated by the multigroup spatial dissimilarity index, were significantly associated with increases in workhood racial segregation. Here, the modelled coefficient ( β value) of 0.59 can be interpreted as one unit increase in residential segregation being associated with approximately a 0.6 increase in workhood segregation. We found three significant moderators of this positive relationship between residential and workhood segregation. First, the significant positive interaction between residential racial segregation and population density (β =0.57 ) suggests that the larger the increase in an MSA’s population density, the larger the positive relationship between residential and workhood segregation. Testing an alternative measure of built area density yielded similar findings ( Supplementary Table 4.1 ). Second,the negative interaction between residential racial segregation and percent of population classified as White (β = -0.004 ) suggests an MSA that became more White would see a smaller link between residential and workhood segregation. Third, MSAs with a larger increase in their proportion of workers in the trade, commerce and transport-related industries had a relatively smaller positive relationship between residential and workhood segregation than counterparts with smaller or negative changes, as interpretable from the significant negative interaction coefficient (β = -0.012). Other MSA characteristics predicting higher workhood segregation over time include an increase in proportion of low-income workers (β =0.002), decrease in GDP per capita (β =-0.0003), and an increase in race-based workhood segregation (β =0.448 ) . Table 1: Two-way fixed effects regression results, showing the associations between workhood racial segregation, residential racial segregation, potential moderators and other covariates Workplace Segregation Score (A) Racial Segregation Coefficient (standard error) (B) Class Segregation Coefficient (standard error) Residential Racial Seg 0.590*** (0.089) Workhood Class Seg Score 0.448*** (0.049) Residential Class Seg 0.674*** (0.077) Workhood Racial Seg Score 0.061*** (0.007) Transportation & Urban Form Population Density('000/km2) -0.017 (0.092) -0.144*** (0.035) Residential Racial Seg:Pop Density 0.565*** (0.163) Percent commuters using transit -0.002** (0.001) % public transport commutes >=45min 0.0001* (0.00004) % car commutes >=45min 0.001*** (0.0003) Residential Class Seg:%commuters using transit 0.031*** (0.008) Residential Class Seg:% PT commutes >=45min -0.001* (0.001) Residential Class Seg:% car commutes >=45min -0.013** (0.004) Labour Market Conditions and Profile % Work in Agriculture/Mining etc -0.002*** (0.0004) % Work in Trade/Commerce/Transport 0.003** (0.001) Residential Racial Seg:% Work in Trade/Commerce/Transport -0.012*** (0.003) % Work in Entertainment/Recreation 0.001** (0.0004) Residential Class Seg:% Work in Entertainment/Recreation -0.019** (0.006) Diversity and Inclusiveness Percent White of Resi Pop -0.002** (0.001) Residential Racial Seg: %White -0.004*** (0.001) Difference in NonWhite-White Unemployment Rates -0.0003*** (0.0001) Control Variables % 55+years -0.001 (0.001) -0.0001 (0.0002) % Income<1250 0.002*** (0.0003) -0.0002 (0.0001) % College Educated -0.0004 (0.001) 0.002*** (0.0002) % Female -0.001 (0.0005) 0.0002 (0.0002) % Population Suburban 0.001 (0.001) 0.0001 (0.0004) GDP ('000) per capita, chained 2017 -0.0003** (0.0001) 0.0001* (0.00003) Observations 3,040 3,040 R2 0.105 0.172 F Statistic 20.734*** (df = 15; 2638) 30.329*** (df = 18; 2635) Note: *p<0.05; **p<0.01; ***p<0.001 For class segregation (Table 1, Model B), increases in residential class-based segregation were positively associated with increases in workhood class segregation (β =0.7 ) . We identified the four moderators of the relationship between residential and workhood segregation. First, the significant positive interaction between residential class segregation and percent commuters using transit (β =0.03 ) suggests MSAs with an increase in the percent of commuters using transit had a steeper relationship between residential segregation and workhood segregation. Relatedly, the significant negative interaction between residential class segregation and percent of transit commuters with long commutes (β =-0.001 ), and percent of drivers with long commutes (β = -0.01 ) --the second and third moderators respectively–suggests MSAs with improved transport accessibility and thus a reduction in commute length had a relatively steeper relationship between residential segregation and workhood segregation . Fourth, the negative interaction between residential class segregation and percent workers in the entertainment and recreation-related industries (β = -0.02 ) can be interpreted as MSAs with a larger increase in their proportion of workers in the entertainment and recreation related industries having a relatively smaller relationship between residential and workhood segregation. Increases in workhood class-based segregation was also associated with increases in the proportion of college-educated workers (β =0.002), and workhood racial segregation (β =0.06 ) . Additional analyses using alternative measures of residential segregation, transportation connectivity and population density yielded some differing results from the main models presented here. We interpret the implications below, and report more detailed findings in Supplementary Annex 4 . Discussion Residential Segregation and Workhood Segregation: Overall Patterns Our study of 380 MSAs in the U.S. from 2011 to 2018 clearly established a positive correlation between workhood and residential segregation by race and class. MSAs generally had lower levels of workhood racial segregation than residential racial segregation, and the opposite for class-based segregation. These findings reinforce observations from previous studies based on different time periods and specific locations 18,19,63 , which suggest that these patterns are fairly stable over the years and areas within the U.S.. That said, overall levels of racial segregation in workhoods (~0.31 between 2011 to 2018) and residential neighborhoods (~0.17) between 2011 to 2018 were higher compared to class-based segregation(~0.1 and ~0.06 for workhood and residential segregation respectively), indicating that workers in metropolitan areas were more spatially mixed by class than racial categories across both spheres. Furthermore, the observed increases in correlation between racial segregation in neighborhoods and workhoods over time, and decreases for class-based segregation suggests the ‘vicious cycle of segregation’ might be strengthening in terms of race and not class. Clearly, the U.S. is still a nation very much divided by race, even as it grapples with concerns about rising levels of class inequality and class segregation 9,64 . We found a larger correlation between race and class-based segregation within workhoods ( r from 0.67 to 0.61) than residential neighborhoods ( r from 0.47 to 0.57). This suggests that a larger share of residential neighborhoods might be racially homogenous but socioeconomically mixed, or vice versa, compared to workhoods which were more likely to be either mixed or homogeneous along both dimensions of race and class. This finding, as well as the positive correlation between workhood racial and class-based segregation in our regression models, indicate that class and race are particularly intertwined when it comes to workhood segregation, and more so than within the residential sphere. While the LODEs data is not disaggregated enough for us to explore the interaction between class and race more deeply (e.g. segregation of more highly educated Asian, Black, White etc. workers vs less highly educated counterparts), future studies of workhood segregation should examine this interaction, similar to research on residential segregation. 9,65 Our analyses also tested several hypotheses about potential moderators of the link between residential segregation and workhood segregation, which yielded some unexpected findings (Table 1) . For transportation-related characteristics, MSAs that grew denser were also ones that saw a larger positive relationship between residential and workhood racial segregation than those that did not. This finding was robust to alternative specifications of ‘built density’, which we tested using 2010 and 2015 data (Supplementary Table 4.1) . For class-based segregation, MSAs that saw reductions in percent of public transport or driving commutes that took over 45 minutes, as well as an increase in public transit commuting, had steeper relationships between residential segregation and workhood segregation, compared to MSAs that did not experience similar changes. Collectively, these findings imply that improved spatial connectivity, through transport infrastructure improvements and/or denser urban form, accentuates the links between residential and workhood class-based segregation–which in turn offer some support to the hypothesis that reduced travel constraints might further accentuate employment niching 50 . Improvements in transportation connectivity might simply allow people to travel more efficiently to nearby employment niches, rather facilitate a wider job search radius and greater diffusion from existing residential enclaves. Supplementary analyses using 2010 and 2015 data specifically measuring increases in road lengths and number of road intersections however suggest that changes in road infrastructure per se might not explain these observed patterns. Our supplementary models found that the greater the expansion of road infrastructure, the smaller the associations between residential segregation and workhood segregation (Supplementary Table 4.2) . Tentatively, we posit that changes in residential and/or employment locational set up (e.g. new neighborhoods being built near existing employment centers) rather than changes in transportation networks might have created shorter commutes and thus tighter links between home and workhood locations. More in-depth studies are needed to unpack why shorter commutes and increased transit share seem to accentuate the ‘circle of segregation’. For diversity and inclusiveness, MSAs that grew less racially diverse were also ones where increases in residential segregation was associated with a smaller increase in workhood segregation, compared to MSAs that have a larger minority population profile. Furthermore, workhood racial segregation levels were generally lower for MSAs that had more White residents, across all observed levels of residential segregation. These findings run counter to the hypothesis that more diverse MSAs would have work environments that are more accepting of minority groups. Instead our findings align better with the ‘threat’ hypothesis that suggest higher proportions of minority individuals cause a ‘hunkering down’ of the majority population which manifests in more discriminatory actions against minority groups ( 56,57 . For labor market characteristics, links between residential and workhood class segregation seemed larger in MSAs where more workers were employed in agriculture and mining related industries, but smaller in MSAs more workers in entertainment and recreation (for class-based segregation) and trade,commerce & transport related industries (for racial segregation). One interpretation would be that the agriculture and mining related industries might be more reliant on personal class-based networks for hiring, whereas the latter two industries might be less reliant on personal networks for hiring. Again, these tentative hypotheses however need further research to validate. Limitations and Proposed Next Steps This study explores the observed relationships between metropolitan-area workhood and residential neighborhood segregation patterns in a descriptive, exploratory manner While our results reveal interesting patterns at an metropolitan level, analyses at this scale necessarily obscures details of neighborhood-by-neighborhood level variations, which is another scale at which to implement important interventions. Additional research looking at how neighborhood-level differences in access to social networks and access to transportation might affect the extent to which workers live and work in segregated or integrated neighborhoods would also be invaluable in unpacking potential causal mechanisms, which is challenging to do with data aggregated at the metropolitan level. Two additional limitations arise due to the nature of our data. We were unable to study ethnicity in tandem with race, as the LODES data does not differentiate between Hispanic and non-Hispanic individuals within each racialised category. Furthermore, as the LODES data captures only those who were working, our calculations of spatial segregation across home and work, and across class and racial categories, necessarily exclude other non-working individuals, such as children, retirees, and job-seekers. Checks using ACS data to calculate residential segregation suggest a somewhat different relationship between worker-based segregation calculations versus total residential population, which would include the unemployed (e.g. children, retirees etc.). Findings from this study should thus be interpreted with a clear understanding that it applies to a specific at-work population. Conclusion This study’s exploratory examination of 380 U.S. MSAs from 2011 to 2018 confirms a persistent link between residential segregation and workhood segregation; identifies metropolitan characteristics that moderated this link, and provides suggestions on where additional research is needed to advance understanding of spatial segregation in the U.S. Through these descriptive findings about the complex links between residential and workhood segregation, we hope to spark further studies on how to reduce spatial segregation across multiple domains, in a country currently divided by class and race. Methods Calculation of Spatial Segregation Using LODES Home and Workplace Information We used the LODES (version 8.3) dataset (https://lehd.ces.census.gov/data/lodes/LODES8/, Last accessed December 1, 2024) from the U.S. Census Bureau. The LODES data derive from administrative records, such as state employment insurance reporting and federal worker earnings records, of home and work addresses of individual workers aged 14 and over, and are aggregated to the home and work census blocks for a representative sample of workers (n ~ 128 million living in 72,534 tracts and working in 72,032 tracts in 2011; n ~ 144 million living in 72,242 tracts and working in 72,115 tracts in 2011) for every state excluding Hawaii, Alaska and territories. We focus on the years 2011-2018, for which we had complete LODES residential and workplace information for every state. The data cover some 95% of jobs in the United States. For this analysis, we utilised two datasets from LODES, namely (1) Residence Area Characteristics (RAC), which provides information on workers by census block of residence, and (2) Workplace Area Characteristics (WAC), which provides information on workers by census block of employment as the smallest spatial unit of aggregation. We aggregated the block-level data to the tract-level to reduce uncertainties, and because census tracts are often used as spatial proxies for neighborhoods (Hall et al 2019). To quantify MSA-level racial segregation, we calculate a multigroup dissimilarity index for four race categories (Asian, Black, White, and Others, which includes all other classifications such as American Indian or Alaska Native; Native Hawaiian or other Pacific Islander, Two or More Race Groups). As the LODES data reports ‘race’ and ‘ethnicity’ separately, we were unable to generate a singular measure of race/ethnic segregation that includes Hispanic individuals as a specific population category within our analysis–which is a limitation of available data and our resultant analysis. To quantify class-based segregation, we focus on educational attainment as a measure of class, and calculate multigroup dissimilarity for four education levels (less than High School, High School, Some College or Associate Degree, Bachelor’s Degree or advanced degree). For both sets of calculations, we calculated MSA-level estimates based on census tract-level population information. Like Hall et al (2019), our estimates of residential segregation is based on worker populations instead of total residential population, in order to maintain consistency in the underlying population for our analysis. In supplementary analyses (Supplementary Annex 4), we created similar estimates of residential segregation based on the American Community Survey (ACS) 5 year data which includes both working and non-working populations. As its name suggests, the ACS 5 year data is collected over the span of five years rather than a single year. Thus to generate yearly estimates, we assigned estimated segregation values to the middle year of the 5 year period. Correlation between the LODES based estimates and ACS-based estimates were fairly high, at 0.9 for racial segregation, and 0.8 for educational segregation. Dissimilarity is a measure of spatial clustering and can be interpreted as a measure of how different the composition of individual’s localised environments are on average compared to the population composition as a whole, where a score of 0 indicates no difference from the broader population as a whole (i.e. complete integration), while the maximum score of 1 indicates the greatest difference possible (i.e. complete segregation). While there are many alternative measures of segregation, here we utilize the dissimilarity index as it is one of the most commonly used and thus easily understood measures in the field. Specifically, we use a spatial measure of dissimilarity that factors in the distance and proximity of census tracts to each other, in order to better capture the spatial dimension of segregation, compared to the commonly used ‘aspatial’ measure of dissimilarity that assumes the mixing between residents is strictly bounded within the spatial confines of each distinct neighborhood. Details of how the spatial dissimilarity index is calculated can be found in 66 , 67 . To examine regional variations in segregation across the U.S., we choose MSAs as our primary aggregated spatial unit of analysis because housing markets and labor markets –both major drivers of intergroup disparities and spatial segregation–operate at a metropolitan scale 68 . Expectedly, much of the past research on segregation that informs this study also used MSAs as a unit of analysis 19,35,69 . MSAs are one of the two types of core-based statistical areas (CBSAs). CBSAs comprise counties in most of the U.S, and are built from one or more central counties where a substantial population resides in the same core urban area(s). MSAs are CBSAs that contain an urban area and have at least 50,000 residents. As MSA boundaries can get redefined over the years, which complicates temporal analysis, it is important to utilise a consistent set of spatial boundaries for comparison. For this analysis, we analyzed all 380 MSAs in the contiguous United States based on the 2020 decennial census geographical boundaries. Figure 2 maps out the geographical distribution of residential neighborhood and workhood segregation estimates, for 2018. To answer our second research question about what factors moderate the relationship between residential and workhood segregation, we analyse a set of MSA-level variables that measure metropolitan-level factors that could influence access to jobs via transport infrastructure or urban form, as well as those affecting access to jobs via personal networks. We also include a set of control covariates to account for additional factors that might conceivably drive both residential and workhood segregation. Table 2 summarizes these variables and their sources, while Supplementary Table 5.1 provides summary statistics of these variables. Table 2: Moderating Variables and Additional Controls when evaluating associations between residential and workplace segregation. MSA-level variables Source, years (1) Factors affecting spatial access to jobs : Transportation & Urban Form Average Commuting Time (mins) 2009-2013 5-year ACS to 2016-2020 5-year ACS 1 % of Workers taking public transit 2009-2013 5-year ACS to 2016-2020 5-year ACS Percentage point differences in % of workers having >45min commutes, between public transit and car commuters 2009-2013 5-year ACS to 2016-2020 5-year ACS Population Density (1000 persons/km 2 ) 2009-2013 5-year ACS to 2016-2020 5-year ACS (2a) Factors affecting access to jobs via personal networks: Diversity & Inclusiveness Percent of workers who are non-White 2 LODES WAC, 2011-2018 Percentage point differences in unemployment rates between White and non-White civilian labor force ages 16+ 2009-2013 5-year ACS to 2016-2020 5-year ACS Gini Index of Inequality 2009-2013 5-year ACS to 2016-2020 5-year ACS (2b) Factors affecting access to jobs via personal networks : Labour Market Conditions and Profile Percent of civilian labor force ages 16 and above unemployed 2009-2013 5-year ACS to 2016-2020 5-year ACS Percent of workers in the following sectors 2 Agriculture/ Mining/Utilities/Construction Manufacturing Trade/Commerce/Transport Professional Services Public/Social Services Entertainment/Recreation LODES WAC 2011-2018 Control Variables GDP per capita Bureau of Economic Analysis, 2011-2018 Percent of workers aged ≥ 55 y LODES WAC, 2011-2018 Percent of workers earning ≤ USD $1,250/month LODES WAC, 2011-2018 Percent of workers who were female LODES WAC, 2011-2018 Percent of workers with a college degree LODES WAC, 2011-2018 Percent of workers classified as living in a suburban area LODES WAC, 2011-2018 Similar to how we used the ACS 5 year data to calculate residential segregation measure, all other covariates calculated using ACS 5 year data were attributed to the middle of the five year period (e.g. 2011 for 2009-2013 data) In this analysis, we only considered primary jobs, and not secondary or tertiary jobs. We did so as we did not have any information on how much time workers spent at each job site when workers had multiple jobs. As the LODES provides data on many different categories of job types, which would be unwieldy in our analysis, we aggregated them into six broad sectors, as such: 1.1: Agriculture/Forestry/Fishing/Hunting, 1.2: Mining/Quarrying/Oil and Gas Extraction, 1.3: Utilities, 1.4: Construction, 2:. Manufacturing, 3.1:Wholesale Trade, 3.2: Retail Trade, 3.3: Transportation and Warehousing, 4.1: Information. 4.2: Finance and Insurance, 4.3: Real Estate and Rental and Leasing, 4.4: Professional/Scientific and Technical Services, 4.5: Management of Companies and Enterprises, 4.6: Administrative and Support and Waste Management and Remediation Services, 5.1: Educational Services, 5.2: Healthcare and Social Assistance, 5.3:Public Administration, 6.1: Arts/Entertainment and Recreation, 6.2: Accommodation and Food Services Statistical Analysis Approach For our analysis, we fit a series of regression models with MSA-level and year-fixed effects, and examined how changes in workhood segregation, our outcome variable, might be associated with changes in residential segregation and its interaction with the hypothesized moderating variables described above. Using MSA-level fixed effects essentially allows us to control for time-invariant characteristics that may confound the relationship of interest, and thus offering a way to overcome potential omitted variable bias 70 . As some of our proposed predictor and moderator variables were quite highly correlated (see Figure 5.1 for bi-variate correlation plots for year 2018), multicollinearity might be a problem if all proposed predictor and moderator variables were included into one model. We thus fitted a series of models that included only one MSA-level moderator variable as well as its interaction with residential segregation score, together with the MSA-level controls (Supplementary Figure 3.1). Moderator variables with modelled coefficients of significance p<0.05, were added into a combined model (Model 1), while Model 2 presents a further streamlined model that included only the moderator variables or their interaction with residential segregation that were significant (p<0.05) from Model 1. To assess the relationships between racial/ethnic and class-based segregation, Model 3 further includes a measure of residential segregation along the other category of segregation (e.g. class-based segregation for the models with racial/ethnic segregation as an outcome, and vice versa), while Model 4 includes a measure of workhood segregation along the same alternative category as Model 3. The Results section report estimates from Model 4, whereas detailed tables of all models are presented in Supplementary Tables 3.1 and 3.2) As robustness checks, we repeated the fixed effects regression models 2 to 4 with alternative data sources. These include ACS 5 survey year data to estimate residential racial and class segregation; as well as alternative measures of MSA-level urban density and road transport infrastructure, as of 2010 and 2015, produced by 71 . Details of how these alternative variables were calculated, the model results and interpretation are reported in Supplementary Annex 4. All analysis was conducted in R, version 4.3.2. The two way fixed-effects models were fitted using package ‘plm’ version 2.6-6, while the spatial segregation scores were estimated using the package ‘seg’ version 0.5-7. Declarations Data availability Access to the data we generated for our analysis is available at GitHub : https://github.com/redacted_link Code availability Access to the code we used to run our analysis is available at GitHub : https://github.com/redacted_link References Frankenberg, E. The Role of Residential Segregation in Contemporary School Segregation. Educ. Urban Soc. 45 , 548–570 (2013). Ekenga, C. C. & Tian, R. Promoting Food Equity in the Context of Residential Segregation. Environ. Justice 15 , 346–351 (2022). Bower, K. M., Thorpe, R. J., Rohde, C. & Gaskin, D. J. The intersection of neighborhood racial segregation, poverty, and urbanicity and its impact on food store availability in the United States. Prev. Med. 58 , 33–39 (2014). Anderson, K. F. & Galaskiewicz, J. Racial/Ethnic Residential Segregation, Socioeconomic Inequality, and Job Accessibility by Public Transportation Networks in the United States. Spat. Demogr. 9 , 341–373 (2021). vonLockette, N. T. D. & Johnson, J. LATINO EMPLOYMENT AND RESIDENTIAL SEGREGATION IN METROPOLITAN LABOR MARKETS. Bois Rev. Soc. Sci. Res. Race 7 , 151–184 (2010). VonLockette, N. D. The Impact of Metropolitan Residential Segregation on the Employment Chances of Blacks and Whites in the United States. City Community 9 , 256–273 (2010). Caldwell, J. T., Ford, C. L., Wallace, S. P., Wang, M. C. & Takahashi, L. M. Racial and ethnic residential segregation and access to health care in rural areas. Health Place 43 , 104–112 (2017). Carr, J. H. & Kutty, N. K. Segregation: The Rising Costs for America . (Routledge, 2008). Massey, D. S. Still the Linchpin: Segregation and Stratification in the USA. Race Soc. Probl. 12 , 1–12 (2020). Williams, D. R. & Collins, C. Racial residential segregation: a fundamental cause of racial disparities in health. Public Health Rep. 116 , 404–416 (2001). Baron, J. N. & Bielby, W. T. Bringing the Firms Back in: Stratification, Segmentation, and the Organization of Work. Am. Sociol. Rev. 45 , 737–765 (1980). Marcińczak, S., Tammaru, T., Strömgren, M. & Lindgren, U. Changing patterns of residential and workplace segregation in the Stockholm metropolitan area. Urban Geogr. 36 , 969–992 (2015). Tomaskovic-Devey, D. et al. Documenting Desegregation: Segregation in American Workplaces by Race, Ethnicity, and Sex, 1966–2003. Am. Sociol. Rev. 71 , 565–588 (2006). Mijs, J. J. B. & Roe, E. L. Is America coming apart? Socioeconomic segregation in neighborhoods, schools, workplaces, and social networks, 1970–2020. Sociol. Compass 15 , e12884 (2021). Song, J., Price, D. J., Guvenen, F., Bloom, N. & von Wachter, T. Firming Up Inequality*. Q. J. Econ. 134 , 1–50 (2019). Stainback, K. & Tomaskovic-Devey, D. Documenting Desegregation: Racial and Gender Segregation in Private Sector Employment Since the Civil Rights Act . (Russell Sage Foundation, 2012). Ellis, M., Wright, R. & Parks, V. Work Together, Live Apart? Geographies of Racial and Ethnic Segregation at Home and at Work. Ann. Assoc. Am. Geogr. 94 , 620–637 (2004). Hu, L., Moayyed, H. & Frank, N. Racial/ethnic and class segregation at workplace and residence: A Los Angeles case study. J. Urban Aff. 44 , 907–925 (2022). Hall, M., Iceland, J. & Yi, Y. Racial Separation at Home and Work: Segregation in Residential and Workplace Settings. Popul. Res. Policy Rev. 38 , 671–694 (2019). Tammaru, T., Knapp, D., Silm, S., Van Ham, M. & Witlox, F. Spatial Underpinnings of Social Inequalities: A Vicious Circles of Segregation Approach. Soc. Incl. 9 , 65–76 (2021). Delmelle, E., Nilsson, I. & Adu, P. Poverty Suburbanization, Job Accessibility, and Employment Outcomes. Soc. Incl. 9 , 166–178 (2021). Coulter, R., Ham, M. van & Findlay, A. M. Re-thinking residential mobility: Linking lives through time and space. Prog. Hum. Geogr. 40 , 352–374 (2016). Simpson, W. URBAN STRUCTURE AND THE LABOUR MARKET. WORKER MOBILITY, COMMUTING AND UNDEREMPLOYMENT IN CITIES . (1992). Trimble, L. B. & Kmec, J. A. The Role of Social Networks in Getting a Job. Sociol. Compass 5 , 165–178 (2011). Ioannides, Y. M. & Datcher Loury, L. Job Information Networks, Neighborhood Effects, and Inequality. J. Econ. Lit. 42 , 1056–1093 (2004). Granovetter, M. The Strength of Weak Ties: A Network Theory Revisited. Sociol. Theory 1 , 201–233 (1983). Mouw, T. Are Black Workers Missing the Connection?: The Effect of Spatial Distance and Employee Referrals on Interfirm Racial Segregation*. Demography 39 , 507–528 (2002). Tassier, T. A Markov Model of Referral-Based Hiring and Workplace Segregation. J. Math. Sociol. 29 , 233–262 (2005). Hellerstein, J. K. & Neumark, D. Workplace Segregation in the United States: Race, Ethnicity, and Skill. Rev. Econ. Stat. 90 , 459–477 (2008). Patacchini, E. & Zenou, Y. Racial identity and education in social networks. Soc. Netw. 44 , 85–94 (2016). Ewing, R. & and Cervero, R. “Does Compact Development Make People Drive Less?” The Answer Is Yes. J. Am. Plann. Assoc. 83 , 19–25 (2017). Lee, C. Metropolitan sprawl measurement and its impacts on commuting trips and road emissions. Transp. Res. Part Transp. Environ. 82 , 102329 (2020). Levinson, D. Network Structure and City Size. PLOS ONE 7 , e29721 (2012). Stevens, M. R. Does Compact Development Make People Drive Less? J. Am. Plann. Assoc. 83 , 7–18 (2017). Galaskiewicz, J., Anderson, K. F. & Thompson-Dyck, K. Minority-White income inequality across metropolitan areas: The role of racial/ethnic residential segregation and transportation networks. J. Urban Aff. 43 , 16–39 (2021). Wissink, B., Schwanen, T. & van Kempen, R. Beyond residential segregation: Introduction. Cities 59 , 126–130 (2016). Covington, K. L. Overcoming Spatial Mismatch: The Opportunities and Limits of Transit Mode in Addressing the Black–White Unemployment Gap. City Community 17 , 211–235 (2018). Pucher, J. & Renne, J. L. Socioeconomics of urban travel: evidence from the 2001 NHTS. Transp. Q. 57 , 49–77 (2003). Sanchez, T. W. The Impact of Public Transport on US Metropolitan Wage Inequality. Urban Stud. 39 , 423–436 (2002). Duany, A., Plater-Zyberk, E. & Speck, J. Suburban Nation: The Rise of Sprawl and the Decline of the American Dream . (North Point Press, New York, 2000). Golub, A., Marcantonio, R. A. & Sanchez, T. W. Race, Space, and Struggles for Mobility: Transportation Impacts on African Americans in Oakland and the East Bay. Urban Geogr. 34 , 699–728 (2013). Goodling, E., Green, J. & McClintock, N. Uneven development of the sustainable city: shifting capital in Portland, Oregon. Urban Geogr. 36 , 504–527 (2015). Sharkey, P. Stuck in Place . (2013). Trounstine, J. Segregation by Design: Local Politics and Inequality in American Cities . (Cambridge University Press, Cambridge, 2018). doi:10.1017/9781108555722. Bereitschaft, B. Equity in neighbourhood walkability? A comparative analysis of three large U.S. cities. Local Environ. 22 , 859–879 (2017). Grengs, J. Nonwork Accessibility as a Social Equity Indicator. Int. J. Sustain. Transp. 9 , 1–14 (2015). Shin, E. J. Disparities in access to opportunities across neighborhoods types: a case study from the Los Angeles region. Transportation 47 , 475–501 (2020). Lee, J., Vojnovic, I. & Grady, S. C. The ‘transportation disadvantaged’: Urban form, gender and automobile versus non-automobile travel in the Detroit region. Urban Stud. 55 , 2470–2498 (2018). McKenzie, B. S. Neighborhood Access to Transit by Race, Ethnicity, and Poverty in Portland, OR. City Community 12 , 134–155 (2013). Ellis, M., Wright, R. & Parks, V. Geography and the Immigrant Division of Labor. Econ. Geogr. 83 , 255–281 (2007). Elliott, J. R. Referral Hiring and Ethnically Homogeneous Jobs: How Prevalent Is the Connection and for Whom? Soc. Sci. Res. 30 , 401–425 (2001). Christopher, S. C. & Leslie, T. F. Assessing Immigrant Niches Across Large American Metropolitan Areas. Popul. Space Place 21 , 171–192 (2015). Crul, M. Super-diversity vs. assimilation: how complex diversity in majority–minority cities challenges the assumptions of assimilation. J. Ethn. Migr. Stud. 42 , 54–68 (2016). Glaeser, E. L., Kallal, H. D., Scheinkman, J. A. & Shleifer, A. Growth in Cities. J. Polit. Econ. 100 , 1126–1152 (1992). Maxwell, R. Cosmopolitan Immigration Attitudes in Large European Cities: Contextual or Compositional Effects? Am. Polit. Sci. Rev. 113 , 456–474 (2019). Moller, S., Alderson, A. S. & Nielsen, F. Changing Patterns of Income Inequality in U.S. Counties, 1970–2000. Am. J. Sociol. 114 , 1037–1101 (2009). Parks, V. The Uneven Geography of Racial and Ethnic Wage Inequality: Specifying Local Labor Market Effects. Ann. Assoc. Am. Geogr. 102 , 700–725 (2012). Trimble, L. B. & Kmec, J. A. The Role of Social Networks in Getting a Job. Sociol. Compass 5 , 165–178 (2011). Marsden, P. V. & Gorman, E. H. Social Networks, Job Changes, and Recruitment. in Sourcebook of Labor Markets 467–502 (Springer, Boston, MA, 2001). doi:10.1007/978-1-4615-1225-7_19. Ellis, M., Wright, R. & Parks, V. Geography and the Immigrant Division of Labor. Econ. Geogr. 83 , 255–281 (2007). Wright, R., Ellis, M. & Parks, V. Immigrant Niches and the Intrametropolitan Spatial Division of Labour. J. Ethn. Migr. Stud. 36 , 1033–1059 (2010). Lillehagen, M. & Hermansen, A. S. Entering the mainstream economy? Workplace segregation and immigrant assimilation. Soc. Forces soae162 (2024) doi:10.1093/sf/soae162. Ellis, M., Wright, R. & Parks, V. Work Together, Live Apart? Geographies of Racial and Ethnic Segregation at Home and at Work. Ann. Assoc. Am. Geogr. 94 , 620–637 (2004). Massey, D. S. & Rugh, J. S. America’s Unequal Metropolitan Geography: Segregation and the Spatial Concentration of Affluence and Poverty. in Who Gets What? (eds. Rosenbluth, F. M. & Weir, M.) 161–187 (Cambridge University Press, 2021). doi:10.1017/9781108879170.007. Iceland, J. & Wilkes, R. Does Socioeconomic Status Matter? Race, Class, and Residential Segregation. Soc. Probl. 53 , 248–273 (2006). Reardon, S. F. & O’Sullivan, D. Measures of spatial segregation. Sociol. Methodol. 34 , 121–162 (2004). Hong, S.-Y., O’Sullivan, D. & Sadahiro, Y. Implementing Spatial Segregation Measures in R. PLoS ONE 9 , (2014). Osypuk, T. L. & Galea, S. What Level Macro? Choosing Appropriate Levels to Assess How Place Influences Population Health. in Macrosocial Determinants of Population Health 399–435 (Springer, New York, NY, 2007). doi:10.1007/978-0-387-70812-6_19. Kotecki, J. A., Gennuso, K. P., Givens, M. L. & Kindig, D. A. Separate and Sick: Residential Segregation and the Health of Children and Youth in Metropolitan Statistical Areas. J. Urban Health Bull. N. Y. Acad. Med. 96 , 149–158 (2019). Allison, P. D. Introduction. in Fixed effects regression models (SAGE, Los Angeles, 2009). Burghardt, K., Uhl, J. H., Lerman, K. & Leyk, S. Road network evolution in the urban and rural United States since 1900. Comput. Environ. Urban Syst. 95 , 101803 (2022). Uhl, J. H. & Leyk, S. Historical built-up areas (BUA) - gridded surfaces for the U.S. from 1810 to 2015. Harvard Dataverse https://doi.org/10.7910/DVN/J6CYUJ (2021). Additional Declarations There is NO Competing Interest. Supplementary Files SupplementaryInformation.docx Cite Share Download PDF Status: Posted Version 1 posted You are reading this latest preprint version Research Square lets you share your work early, gain feedback from the community, and start making changes to your manuscript prior to peer review in a journal. As a division of Research Square Company, we’re committed to making research communication faster, fairer, and more useful. We do this by developing innovative software and high quality services for the global research community. Our growing team is made up of researchers and industry professionals working together to solve the most critical problems facing scientific publishing. Also discoverable on Platform About Our Team In Review Editorial Policies Advisory Board Help Center Resources Author Services Accessibility API Access RSS feed Manage Cookie Preferences © Research Square 2026 | ISSN 2693-5015 (online) Privacy Policy Terms of Service Do Not Sell My Personal Information {"props":{"pageProps":{"initialData":{"identity":"rs-6771912","acceptedTermsAndConditions":true,"allowDirectSubmit":true,"archivedVersions":[],"articleType":"Article","associatedPublications":[],"authors":[{"id":470415345,"identity":"a99f5e05-8d03-47c7-a5a8-af4861d021c8","order_by":0,"name":"Shin Bin Tan","email":"data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAAAZAAAAAyAQMAAABI0h/eAAAABlBMVEX///8AAABVwtN+AAAACXBIWXMAAA7EAAAOxAGVKw4bAAAA1UlEQVRIie3OIQvCQBTA8TcOTBPrwWD3FZ4Y/TJburLB0vJAmEVYPbD4EdbE5rhgkVkvLEyEJYPRoODGislzTfD+5Th4P94DMJl+MWolBBDc/jcaQGZDCABpHz/5mrD1Ir1EUcW3lNdwiyWwbPOZYFUsZwKbcCcCtEQpAVWtIdRPHRtlmKsAyDhtCd1rDhM94ah4TZ4tYUJDQPXEQ+UhsVoCp0RzWEcEyml+vGKxKrmNSrOECd440UMyPPBzfY/nLst0h73XzdpAvQGkbzJki8lkMv1DL2IeRl38hXbwAAAAAElFTkSuQmCC","orcid":"https://orcid.org/0000-0002-4432-7860","institution":"National University of Singapore","correspondingAuthor":true,"prefix":"","firstName":"Shin","middleName":"Bin","lastName":"Tan","suffix":""},{"id":470415346,"identity":"19db7ab8-e8d6-4124-ae03-d52d7c1ab294","order_by":1,"name":"Priyanka deSouza","email":"","orcid":"","institution":"University of Colorado Denver","correspondingAuthor":false,"prefix":"","firstName":"Priyanka","middleName":"","lastName":"deSouza","suffix":""}],"badges":[],"createdAt":"2025-05-29 02:20:08","currentVersionCode":1,"declarations":"","doi":"10.21203/rs.3.rs-6771912/v1","doiUrl":"https://doi.org/10.21203/rs.3.rs-6771912/v1","draftVersion":[],"editorialEvents":[],"editorialNote":"","failedWorkflow":false,"files":[{"id":84558141,"identity":"38966331-58a6-46c3-9af7-3722301f8847","added_by":"auto","created_at":"2025-06-13 12:14:35","extension":"png","order_by":1,"title":"Figure 1","display":"","copyAsset":false,"role":"figure","size":165338,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cem\u003eDiagram depicting the proposed moderating pathways between MSA-level residential and workhood segregation, and proposed model variables\u003c/em\u003e\u003c/p\u003e","description":"","filename":"1.png","url":"https://assets-eu.researchsquare.com/files/rs-6771912/v1/71c87652991b8b5cbaaa89fb.png"},{"id":84558142,"identity":"3c209873-3cc9-46e5-a48b-64896a1c50e4","added_by":"auto","created_at":"2025-06-13 12:14:35","extension":"png","order_by":2,"title":"Figure 2","display":"","copyAsset":false,"role":"figure","size":297092,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cem\u003eMap of MSA-level estimates of residential neighborhood and workhood segregation, by A) race and B) class\u003c/em\u003e\u003c/p\u003e","description":"","filename":"2.png","url":"https://assets-eu.researchsquare.com/files/rs-6771912/v1/4bb08e014cf7f84e167f8cc0.png"},{"id":99307978,"identity":"8f7e4c39-910a-462f-956d-26da1cd079e5","added_by":"auto","created_at":"2025-12-31 16:07:18","extension":"pdf","order_by":0,"title":"","display":"","copyAsset":false,"role":"manuscript-pdf","size":1594431,"visible":true,"origin":"","legend":"","description":"","filename":"manuscript.pdf","url":"https://assets-eu.researchsquare.com/files/rs-6771912/v1/2971c2ac-4356-4bc8-b6c5-aeb5771d79f6.pdf"},{"id":84558143,"identity":"9e7de0ec-ee06-4641-ab9e-85c12320eda4","added_by":"auto","created_at":"2025-06-13 12:14:36","extension":"docx","order_by":1,"title":"","display":"","copyAsset":false,"role":"supplement","size":3499537,"visible":true,"origin":"","legend":"","description":"","filename":"SupplementaryInformation.docx","url":"https://assets-eu.researchsquare.com/files/rs-6771912/v1/b7f16be48b0d4df68b87a68c.docx"}],"financialInterests":"There is \u003cb\u003eNO\u003c/b\u003e Competing Interest.","formattedTitle":"Spatial Segregation by Class and Race in Neighborhoods and Workhoods across U.S. metro areas","fulltext":[{"header":"Main","content":"\u003cp\u003eIn the United States (U.S.), residential segregation, or the spatial sorting of people into neighborhoods based on race and/or class, has generated disparities in terms of access to good schools\u003csup\u003e1\u003c/sup\u003e, nutritious food\u003csup\u003e2,3\u003c/sup\u003e, jobs\u003csup\u003e4–6\u003c/sup\u003e, health care\u003csup\u003e7\u003c/sup\u003e, and other critical amenities or services\u003csup\u003e8\u003c/sup\u003e. Scholars have thus argued that residential segregation is the ‘linchpin’ sustaining systemic inequalities in the U.S.\u003csup\u003e9,10\u003c/sup\u003e The traditional focus of segregation in residential neighborhoods however fails to account for other important locations where people spend time and built ties, such as within and around their workplaces\u003csup\u003e11–13\u003c/sup\u003e.\u003c/p\u003e\n\u003cp\u003eTwo approaches have been taken in investigating work-related segregation. The first focuses on examining racial/ethnic, gender, and/or income-based segregation across different employment sectors, or firm types\u003csup\u003e13–16\u003c/sup\u003e. \u0026nbsp;The second, more nascent, approach examines spatial interaction at neighboring workplaces co-located within an area, and is what we refer to in this paper as ‘workhood segregation’. Studies include one evaluating residential and work tract segregation levels for native-born and immigrant groups in Los Angeles (L.A) \u0026nbsp;using 1990 Census of Population and Housing data\u003csup\u003e17\u003c/sup\u003e, a second using 2015 data from the LEHD Origin-Destination Employment Statistics (LODES) in L.A.\u003csup\u003e18\u003c/sup\u003e and a third using Census Transportation Planning Package (CTPP) data to evaluate US metropolitan areas’ workhood segregation in 2000 and 2010\u003csup\u003e19\u003c/sup\u003e. All three articles found positive relationships between residential and workhood racial segregation, and that racial/ethnic residential segregation was higher than workhood segregation\u0026nbsp;\u003csup\u003e17–19\u003c/sup\u003e. For class-specific segregation, the third study found the inverse to be true\u003csup\u003e18\u003c/sup\u003e.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eAn explanation for the positive correlation between residential and workhood segregation is that each sphere of segregation generates and reinforces the other\u003csup\u003e17,19\u003c/sup\u003e in a ‘vicious circle of segregation’\u003csup\u003e20\u003c/sup\u003e. To reduce commuting costs, people tend to choose to work closer to where they live and vice versa\u0026nbsp;\u003csup\u003e21–23\u003c/sup\u003e. People also tend to obtain employment opportunities from their social networks\u0026nbsp;\u003csup\u003e24–26\u003c/sup\u003e. Reliance on personal referrals, or referral hiring, has been linked to greater workplace segregation\u0026nbsp;\u003csup\u003e27,28\u003c/sup\u003e . One source of job referrals are one’s neighbors\u0026nbsp;\u003csup\u003e29,30\u003c/sup\u003e. \u0026nbsp;If neighbors are also more likely to be work colleagues, one would also reasonably expect a degree of concordance between residential and workhood profiles.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eWhile previous studies have laid important exploratory groundwork, these tend to be temporally or geographically limited in their coverage. Our study takes a more comprehensive look at the relationship between residential and workhood racial and class segregation of all 380 U.S. \u0026nbsp; metropolitan statistical areas (MSAs), defined according to 2020 decennial census geographical boundaries. Residential and workhood segregation were estimated for years 2011 to 2018, using LODES (version 8.3) data, which captured home and work locations of workers 14 years and older. To measure racial segregation, we calculated a multigroup dissimilarity index for four race categories (‘Asian’, ‘Black’, ‘White’, and ‘Others’), while for class-based segregation, we used four education levels (‘Less than High School’, ‘High School’, ‘Some College or Associate Degree’, ‘Bachelor’s Degree or advanced degree’). Higher dissimilarity scores indicate higher levels of segregation of the defined groups. \u0026nbsp;\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eFirst, we examine how MSA-levels of workplace racial and class-based segregation compared to residential segregation over the years, and whether there were significant positive correlations between MSA level class-based segregation and racial segregation, across both work and residential domains.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eSecond, as there are a lack of studies that explore how to reduce, the ‘vicious circle of segregation’\u003csup\u003e20\u003c/sup\u003e, our study examines two categories of metropolitan characteristics that might moderate the relationship between residential and workhood segregation: 1. transportation infrastructure and urban form; 2. Overall diversity and inclusiveness within MSAs, as well as labor market conditions. \u003cstrong\u003eFigure 1\u003c/strong\u003e below summarizes our proposed moderating pathways, which we explain further in the following paragraphs.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eResidents living in compact MSAs with good transportation connectivity are likely to have lower travel costs, in terms of travel distance and commuting time –though the variation and magnitude of the relationship between urban form and commuting behavior has been the source of some debate\u003csup\u003e31–34\u003c/sup\u003e. Lower travel costs, in turn, reduce spatial mismatch by expanding exposure and opportunities for intergroup interactions, while also increasing spatial access to job opportunities available to racial/ethnic minority groups, even if they live in segregated areas\u003csup\u003e35,36\u003c/sup\u003e. Specifically, greater public transit access can help reduce spatial mismatch, given the higher use of transit amongst lower-income and/or minority groups\u0026nbsp;\u003csup\u003e37–39\u003c/sup\u003e. MSAs where residents \u003cstrong\u003ehave better transport access (especially in terms of public transit), and shorter commutes\u003c/strong\u003e would potentially be ones with a weaker link between residential and workplace segregation.\u003c/p\u003e\n\u003cp\u003eAlternatively, better transportation infrastructure\u003cem\u003e\u0026nbsp;could strengthen\u003c/em\u003e the link between residential segregation, spatial mismatch, and workplace segregation. In the U.S., transportation infrastructure has been observed to benefit wealthier and White residents\u003csup\u003e40–44\u003c/sup\u003e, with resarch showing how better spatial access to jobs via transit benefitted White neighborhoods but not Black or Latino ones\u003csup\u003e4\u003c/sup\u003e, and that White workers in segregated metropolitan areas with efficient transit and pedestrian networks earned more than those in comparable areas with less efficient networks\u0026nbsp;\u003csup\u003e35\u003c/sup\u003e. However, other empirical analyses \u0026nbsp; suggest that spatial accessibility by income and ethnic/racial minority group membership are complex and city-specific\u003csup\u003e45\u003c/sup\u003e, with some finding higher spatial accessibility among low income and/or minority urban areas\u003csup\u003e46,47\u003c/sup\u003e while others observed the converse\u003csup\u003e48,49\u003c/sup\u003e.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eReduced travel constraints might also further accentuate ‘ethnic niching’ \u0026nbsp;– a phenomenon where members of an ethnic group are concentrated at a higher level than members of other groups in certain labor market sectors. This accentuation might occur because social networks that link workers to ethnic employment niches would have fewer travel or spatial information constraints, and can better fend off competition from other groups that were previously more well-located\u003csup\u003e50\u003c/sup\u003e. There has however been little empirical analysis of this hypothesis. \u0026nbsp;\u003c/p\u003e\n\u003cp\u003eAs discrimination likely increases the extent to which marginalized groups need to rely on social networks for employment opportunities\u003csup\u003e51\u003c/sup\u003e, more inclusive metropolitan areas might have a reduced link between residential segregation and workhood segregation\u003cstrong\u003e.\u0026nbsp;\u003c/strong\u003eCitiesthat have a \u003cstrong\u003emore diverse, less White population profile\u003c/strong\u003e may be less prejudiced due to increased intergroup interactions\u0026nbsp;\u003csup\u003e52–55\u003c/sup\u003e. Conversely, the ‘threat’ hypothesis suggests that such areas might also be sites of increased tension and discrimination\u003csup\u003e56,57\u003c/sup\u003e.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eA more direct measure of whether a metropolitan area discriminates against racial/ethnic minority groups would be the \u003cstrong\u003edegree of racial disparities in incomes or employment rates.\u003c/strong\u003e Hiring racial discrimination in hiring might mean minority groups have to rely heavily on personal networks to obtain jobs, thus reinforcing the links between residential and workplace racial segregation. Similarly, workers in a metropolitan area with \u003cstrong\u003ehigher income inequality\u0026nbsp;\u003c/strong\u003emight have higher reliance on social networks to find jobs, compared to areas with lower wage inequality, and thus experience a stronger link between residential and workhood segregation.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eThe state of the metropolitan labor market\u0026nbsp;\u003c/strong\u003emight be another moderating factor.Higher unemployment rates might create conditions where personal networks are less effective because one’s networks might be competing for employment and thus prove unwilling or unable to provide referrals\u0026nbsp;\u003csup\u003e58\u003c/sup\u003e. Increased competition for lower-wage jobs that would otherwise only be attractive to more marginalized groups could also reduce propensity for ethnic group niching\u003csup\u003e52\u003c/sup\u003e. On the flip side, social networks might act as an additional competitive edge for securing jobs in a recession. Employers in weak markets might also rely more heavily on referrals, which is cheaper than advertising or using recruitment agencies\u003csup\u003e58\u003c/sup\u003e.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eFinally, \u003cstrong\u003ea metropolitan area’s economic base\u0026nbsp;\u003c/strong\u003emight moderate the link between residential segregation and workhood segregation. Contacts are more likely to be used when filling managerial, professional, or sale/service positions and are less likely to be used in the public sector and in establishments that are part of multi-site firms\u0026nbsp;\u003csup\u003e59\u003c/sup\u003e. Others have argued that the reason certain industries and professions, particularly low-wage ones, tend to be more ethnically ‘niched’, is because job-seeking in these industries are driven in part by heavy reliance on immigrants’ social networks\u0026nbsp;\u003csup\u003e60–62\u003c/sup\u003e. \u0026nbsp;\u003c/p\u003e\n\u003cp\u003eTo test our proposed moderating pathways (\u003cstrong\u003eFigure 1)\u003c/strong\u003e, we fit a series of two-way fixed effects (2FE) regression models with MSA and year fixed effects. These models allow estimation of how within-MSA changes in workhood segregation might be associated with changes in residential segregation and the latter’s interaction with the hypothesized moderating variables, while controlling for other time-varying effects over the years. The threshold for statistical significance is set at p-value\u0026lt;0.05. We started by fitting a series of models that included only one MSA-level moderator variable and its interaction with residential segregation score, together with the MSA-level controls. Moderator variables with modelled coefficients of significance p\u0026lt;0.05, were then included in a combined model, which was then further streamlined in our final model to include only significant moderator variables and their interactions. The main paper focuses on interpreting the final model, while results from the preceding models are reported in \u003cstrong\u003eSupplementary Annex 3\u003c/strong\u003e.\u0026nbsp;\u003c/p\u003e"},{"header":"Results","content":"\u003ch3\u003e\u003cstrong\u003eRelationship between Residential and Workhood Segregation by Race and Class \u0026nbsp;\u003c/strong\u003e\u003c/h3\u003e\n\u003cp\u003eFrom 2011 to 2018, MSA residential racial segregation, as measured by spatial dissimilarity index of workers classified as \u0026lsquo;Asian\u0026rsquo;, \u0026lsquo;Black\u0026rsquo;, \u0026lsquo;White\u0026rsquo; and \u0026lsquo;Other\u0026rsquo;, \u0026nbsp;was positively correlated with workhood racial segregation, with this positive correlation increasing over time (Pearson\u0026rsquo;s correlation coefficient \u003cem\u003er\u003c/em\u003e ranging from 0.47 to 0.57). Residential racial segregation levels were alsoconsistently higher than workhood segregation across the years, at around 0.31 while the latter remained around 0.17. Supplementary \u003cstrong\u003eFigure 1.1\u0026nbsp;\u003c/strong\u003esummarizes.\u003c/p\u003e\n\u003cp\u003eIn contrast, workhood segregation by social class, measured by spatial dissimilarity index of workers of four educational attainment categories, was higher than residential segregation across the years, at around 0.1 compared to 0.06 for the latter. There was a small reduction in the positive correlation between workhood and residential class segregation over the years (\u003cem\u003er\u003c/em\u003e ranging from 0.67 to 0.61). We observed no clear directional trend over time in terms of absolute levels of either racial or class-based segregation across both spheres. Supplementary \u003cstrong\u003eFigure 1.2\u003c/strong\u003e summarizes.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eThere were 29 \u0026lsquo;outlier\u0026rsquo; MSAs that had lower racial residential rather than workhood segregation, and even fewer (n= 9 MSAs) where there were lower class-based workplace segregation rather than residential segregation estimates.Supplementary \u003cstrong\u003eAnnex 2\u003c/strong\u003e provides details of these MSAs and their characteristics.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eWe see low levels of correlation between race and class-based segregation within the residential sphere (r ~ 0.11 to 0.16) and higher levels, albeit still moderately low, within the workhoods (r ~ 0.19 to 0.27). Both correlations seemed to slightly trend upwards between 2011-2018 (Supplementary \u003cstrong\u003eFigure 1.3\u003c/strong\u003e)\u003cstrong\u003e.\u003c/strong\u003e\u003c/p\u003e\n\u003ch3\u003e\u003cstrong\u003eModerators of the Relationship Between Residential And Workhood Segregation\u003c/strong\u003e\u003c/h3\u003e\n\u003cp\u003eTable 1, Model A shows how increases in residential racial segregation, as estimated by the multigroup spatial dissimilarity index, were significantly associated with increases in workhood racial segregation. Here, the modelled coefficient (\u003cem\u003e\u0026beta;\u003c/em\u003e value) of 0.59 can be interpreted as one unit increase in residential segregation being associated with approximately a 0.6 increase in workhood segregation.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eWe found three significant moderators of this positive relationship between residential and workhood segregation. First, the significant positive interaction between residential racial segregation and population density (\u0026beta; =0.57\u003cem\u003e)\u003c/em\u003e suggests that the \u003cstrong\u003elarger the increase in an\u0026nbsp;\u003c/strong\u003eMSA\u0026rsquo;s population density, the larger the positive relationship between residential and workhood segregation. Testing an alternative measure of built area density yielded similar findings (\u003cstrong\u003eSupplementary Table 4.1\u003c/strong\u003e). \u0026nbsp;Second,the negative interaction between residential racial segregation and percent of population classified as White (\u0026beta; = -0.004\u003cem\u003e)\u0026nbsp;\u003c/em\u003esuggests an MSA that became more White would see a smaller link between residential and workhood segregation. Third, MSAs with a larger increase in their proportion of workers in the trade, commerce and transport-related industries had a relatively smaller positive relationship between residential and workhood segregation than counterparts with smaller or negative changes, as interpretable from the significant negative interaction coefficient (\u0026beta; = -0.012).\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eOther MSA characteristics predicting higher workhood segregation over time include an increase in \u0026nbsp;proportion of low-income workers (\u0026beta; =0.002), decrease in GDP per capita (\u0026beta; =-0.0003), and an increase in race-based workhood segregation (\u0026beta; =0.448\u003cem\u003e)\u003c/em\u003e.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003e\u003cem\u003eTable 1:\u003c/em\u003e\u003c/strong\u003e\u003cem\u003e\u0026nbsp;Two-way fixed effects regression results, showing the associations between workhood racial segregation, residential racial segregation, potential moderators and other covariates\u003c/em\u003e\u003c/p\u003e\n\u003ctable border=\"0\" cellspacing=\"0\" cellpadding=\"0\" width=\"643\"\u003e\n \u003ctbody\u003e\n \u003ctr\u003e\n \u003ctd valign=\"bottom\" style=\"width: 268px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd colspan=\"2\" valign=\"bottom\" style=\"width: 375px;\"\u003e\n \u003cp\u003eWorkplace Segregation Score\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"bottom\" style=\"width: 268px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 172px;\"\u003e\n \u003cp\u003e\u003cstrong\u003e\u0026nbsp;(A) Racial Segregation\u003c/strong\u003e\u003c/p\u003e\n \u003cp\u003eCoefficient (standard error)\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 203px;\"\u003e\n \u003cp\u003e\u003cstrong\u003e\u0026nbsp;(B) Class Segregation\u003c/strong\u003e\u003c/p\u003e\n \u003cp\u003eCoefficient (standard error)\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"bottom\" style=\"width: 268px;\"\u003e\n \u003cp\u003eResidential Racial Seg\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 172px;\"\u003e\n \u003cp\u003e0.590*** (0.089)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 203px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"bottom\" style=\"width: 268px;\"\u003e\n \u003cp\u003eWorkhood Class Seg Score\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 172px;\"\u003e\n \u003cp\u003e0.448*** (0.049)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 203px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"bottom\" style=\"width: 268px;\"\u003e\n \u003cp\u003eResidential Class Seg\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 172px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 203px;\"\u003e\n \u003cp\u003e0.674*** (0.077)\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"bottom\" style=\"width: 268px;\"\u003e\n \u003cp\u003eWorkhood Racial Seg Score\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 172px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 203px;\"\u003e\n \u003cp\u003e0.061*** (0.007)\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"bottom\" style=\"width: 268px;\"\u003e\n \u003cp\u003e\u003cstrong\u003e\u003cem\u003e\u003cu\u003eTransportation \u0026amp; Urban Form\u003c/u\u003e\u003c/em\u003e\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 172px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 203px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"bottom\" style=\"width: 268px;\"\u003e\n \u003cp\u003ePopulation Density(\u0026apos;000/km2)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 172px;\"\u003e\n \u003cp\u003e-0.017 (0.092)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 203px;\"\u003e\n \u003cp\u003e-0.144*** (0.035)\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"bottom\" style=\"width: 268px;\"\u003e\n \u003cp\u003eResidential Racial Seg:Pop Density\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 172px;\"\u003e\n \u003cp\u003e0.565*** (0.163)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 203px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"bottom\" style=\"width: 268px;\"\u003e\n \u003cp\u003ePercent commuters using transit\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 172px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 203px;\"\u003e\n \u003cp\u003e-0.002** (0.001)\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"bottom\" style=\"width: 268px;\"\u003e\n \u003cp\u003e% public transport commutes \u0026gt;=45min\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 172px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 203px;\"\u003e\n \u003cp\u003e0.0001* (0.00004)\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"bottom\" style=\"width: 268px;\"\u003e\n \u003cp\u003e% car commutes \u0026gt;=45min\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 172px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 203px;\"\u003e\n \u003cp\u003e0.001*** (0.0003)\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"bottom\" style=\"width: 268px;\"\u003e\n \u003cp\u003eResidential Class Seg:%commuters using transit\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 172px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 203px;\"\u003e\n \u003cp\u003e0.031*** (0.008)\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"bottom\" style=\"width: 268px;\"\u003e\n \u003cp\u003eResidential Class Seg:% PT commutes \u0026gt;=45min\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 172px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 203px;\"\u003e\n \u003cp\u003e-0.001* (0.001)\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"bottom\" style=\"width: 268px;\"\u003e\n \u003cp\u003eResidential Class Seg:% car commutes \u0026gt;=45min\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 172px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 203px;\"\u003e\n \u003cp\u003e-0.013** (0.004)\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"bottom\" style=\"width: 268px;\"\u003e\n \u003cp\u003e\u003cstrong\u003e\u003cem\u003e\u003cu\u003eLabour Market Conditions and Profile\u003c/u\u003e\u003c/em\u003e\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 172px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 203px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"bottom\" style=\"width: 268px;\"\u003e\n \u003cp\u003e% Work in Agriculture/Mining etc\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 172px;\"\u003e\n \u003cp\u003e-0.002*** (0.0004)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 203px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"bottom\" style=\"width: 268px;\"\u003e\n \u003cp\u003e% Work in Trade/Commerce/Transport\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 172px;\"\u003e\n \u003cp\u003e0.003** (0.001)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 203px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"bottom\" style=\"width: 268px;\"\u003e\n \u003cp\u003eResidential Racial Seg:% Work in Trade/Commerce/Transport\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 172px;\"\u003e\n \u003cp\u003e-0.012*** (0.003)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 203px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"bottom\" style=\"width: 268px;\"\u003e\n \u003cp\u003e% Work in Entertainment/Recreation\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 172px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 203px;\"\u003e\n \u003cp\u003e0.001** (0.0004)\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"bottom\" style=\"width: 268px;\"\u003e\n \u003cp\u003eResidential Class Seg:% Work in Entertainment/Recreation\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 172px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 203px;\"\u003e\n \u003cp\u003e-0.019** (0.006)\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"bottom\" style=\"width: 268px;\"\u003e\n \u003cp\u003e\u003cstrong\u003e\u003cem\u003e\u003cu\u003eDiversity and Inclusiveness\u003c/u\u003e\u003c/em\u003e\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 172px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 203px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"bottom\" style=\"width: 268px;\"\u003e\n \u003cp\u003ePercent White of Resi Pop\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 172px;\"\u003e\n \u003cp\u003e-0.002** (0.001)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 203px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"bottom\" style=\"width: 268px;\"\u003e\n \u003cp\u003eResidential Racial Seg: %White\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 172px;\"\u003e\n \u003cp\u003e-0.004*** (0.001)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 203px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"bottom\" style=\"width: 268px;\"\u003e\n \u003cp\u003eDifference in NonWhite-White Unemployment Rates\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 172px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 203px;\"\u003e\n \u003cp\u003e-0.0003*** (0.0001)\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"bottom\" style=\"width: 268px;\"\u003e\n \u003cp\u003e\u003cstrong\u003e\u003cem\u003e\u003cu\u003eControl Variables\u003c/u\u003e\u003c/em\u003e\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 172px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 203px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"bottom\" style=\"width: 268px;\"\u003e\n \u003cp\u003e% 55+years\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 172px;\"\u003e\n \u003cp\u003e-0.001 (0.001)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 203px;\"\u003e\n \u003cp\u003e-0.0001 (0.0002)\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"bottom\" style=\"width: 268px;\"\u003e\n \u003cp\u003e% Income\u0026lt;1250\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 172px;\"\u003e\n \u003cp\u003e0.002*** (0.0003)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 203px;\"\u003e\n \u003cp\u003e-0.0002 (0.0001)\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"bottom\" style=\"width: 268px;\"\u003e\n \u003cp\u003e% College Educated\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 172px;\"\u003e\n \u003cp\u003e-0.0004 (0.001)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 203px;\"\u003e\n \u003cp\u003e0.002*** (0.0002)\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"bottom\" style=\"width: 268px;\"\u003e\n \u003cp\u003e% Female\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 172px;\"\u003e\n \u003cp\u003e-0.001 (0.0005)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 203px;\"\u003e\n \u003cp\u003e0.0002 (0.0002)\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"bottom\" style=\"width: 268px;\"\u003e\n \u003cp\u003e% Population Suburban\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 172px;\"\u003e\n \u003cp\u003e0.001 (0.001)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 203px;\"\u003e\n \u003cp\u003e0.0001 (0.0004)\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"bottom\" style=\"width: 268px;\"\u003e\n \u003cp\u003eGDP (\u0026apos;000) per capita, chained 2017\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 172px;\"\u003e\n \u003cp\u003e-0.0003** (0.0001)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 203px;\"\u003e\n \u003cp\u003e0.0001* (0.00003)\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd colspan=\"3\" valign=\"bottom\" style=\"width: 643px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"bottom\" style=\"width: 268px;\"\u003e\n \u003cp\u003eObservations\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 172px;\"\u003e\n \u003cp\u003e3,040\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 203px;\"\u003e\n \u003cp\u003e3,040\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"bottom\" style=\"width: 268px;\"\u003e\n \u003cp\u003eR2\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 172px;\"\u003e\n \u003cp\u003e0.105\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 203px;\"\u003e\n \u003cp\u003e0.172\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"bottom\" style=\"width: 268px;\"\u003e\n \u003cp\u003eF Statistic\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 172px;\"\u003e\n \u003cp\u003e20.734*** (df = 15; 2638)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 203px;\"\u003e\n \u003cp\u003e30.329*** (df = 18; 2635)\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"bottom\" style=\"width: 268px;\"\u003e\n \u003cp\u003eNote:\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd colspan=\"2\" valign=\"bottom\" style=\"width: 375px;\"\u003e\n \u003cp\u003e*p\u0026lt;0.05; **p\u0026lt;0.01; ***p\u0026lt;0.001\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003c/tbody\u003e\n\u003c/table\u003e\n\u003cp\u003e\u0026nbsp;For class segregation (Table 1, Model B), increases in residential class-based segregation were positively associated with increases in workhood class segregation (\u0026beta; =0.7\u003cem\u003e)\u003c/em\u003e.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eWe identified the four moderators of the relationship between residential and workhood segregation. First, the significant positive interaction between residential class segregation and percent commuters using transit (\u0026beta; =0.03\u003cem\u003e)\u003c/em\u003e suggests MSAs with an increase in the percent of commuters using transit had a steeper relationship between residential segregation and workhood segregation. Relatedly, the significant negative interaction between residential class segregation and percent of transit commuters with long commutes (\u0026beta; =-0.001\u003cem\u003e),\u0026nbsp;\u003c/em\u003eand percent of drivers with long commutes (\u0026beta; = -0.01\u003cem\u003e)\u003c/em\u003e--the second and third moderators respectively\u0026ndash;suggests MSAs with improved transport accessibility and thus a reduction in commute length had a relatively steeper relationship between residential segregation and workhood segregation\u003cstrong\u003e.\u0026nbsp;\u003c/strong\u003e\u0026nbsp; \u0026nbsp;\u003c/p\u003e\n\u003cp\u003eFourth, the negative interaction between residential class segregation and percent workers in the entertainment and recreation-related industries \u0026nbsp; (\u0026beta; = -0.02\u003cem\u003e)\u003c/em\u003e can be interpreted as MSAs with a larger increase in their proportion of workers in the \u003cstrong\u003eentertainment and recreation related industries\u0026nbsp;\u003c/strong\u003ehaving a relatively smaller relationship between residential and workhood segregation.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eIncreases in workhood class-based segregation was also associated with increases in the proportion of college-educated workers (\u0026beta; =0.002), and workhood racial segregation (\u0026beta; =0.06\u003cem\u003e)\u003c/em\u003e.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eAdditional analyses using alternative measures of residential segregation, transportation connectivity and population density yielded some differing results from the main models presented here. We interpret the implications below, and report more detailed findings in \u003cstrong\u003eSupplementary Annex 4\u003c/strong\u003e.\u003c/p\u003e"},{"header":"Discussion","content":"\u003cp\u003e\u003cstrong\u003eResidential Segregation and Workhood Segregation: Overall Patterns\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eOur study of 380 MSAs in the U.S. from 2011 to 2018 clearly established a positive correlation between workhood and residential segregation by race and class. MSAs generally had lower levels of \u0026nbsp;workhood racial segregation than residential racial segregation, and the opposite for class-based segregation. These findings reinforce observations from previous studies based on different time periods and specific locations\u0026nbsp;\u003csup\u003e18,19,63\u003c/sup\u003e, which suggest that these patterns are fairly stable over the years and areas within the U.S..\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eThat said, overall levels of racial segregation in workhoods (~0.31 between 2011 to 2018) \u0026nbsp;and residential neighborhoods (~0.17) between 2011 to 2018 were higher compared to class-based segregation(~0.1 and ~0.06 for workhood and residential segregation respectively), indicating that workers in metropolitan areas were more spatially mixed by class than racial categories across both spheres. Furthermore, the observed increases in correlation between racial segregation in neighborhoods and workhoods over time, and decreases for class-based segregation suggests the ‘vicious cycle of segregation’ might be strengthening in terms of race and not class. Clearly, the U.S. is still a nation very much divided by race, even as it grapples with concerns about rising levels of class inequality and class segregation\u003csup\u003e9,64\u003c/sup\u003e.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eWe found a larger correlation between race and class-based segregation within workhoods (\u003cem\u003er\u003c/em\u003e from 0.67 to 0.61) than residential neighborhoods (\u003cem\u003er\u003c/em\u003e from 0.47 to 0.57). This suggests that a larger share of residential neighborhoods might be racially homogenous but socioeconomically mixed, or vice versa, compared to workhoods which were more likely to be either mixed or homogeneous along both dimensions of race and class. This finding, as well as the positive correlation between workhood racial and class-based segregation in our regression models, indicate that class and race are particularly intertwined when it comes to workhood segregation, and more so than within the residential sphere. \u0026nbsp;While the LODEs data is not disaggregated enough for us to explore the interaction between class and race more deeply (e.g. segregation of more highly educated Asian, Black, White etc. workers vs less highly educated counterparts), future studies of workhood segregation should examine this interaction, similar to research on residential segregation.\u003csup\u003e9,65\u003c/sup\u003e\u003c/p\u003e\n\u003cp\u003eOur analyses also tested several hypotheses about potential moderators of the link between residential segregation and workhood segregation, which yielded some unexpected findings \u003cstrong\u003e(Table 1)\u003c/strong\u003e. For \u003cstrong\u003etransportation-related characteristics,\u003c/strong\u003e MSAs that grew denser were also ones that saw a larger positive relationship between residential and workhood racial segregation than those that did not. This finding was robust to alternative specifications of ‘built density’, which we tested using 2010 and 2015 data \u003cstrong\u003e(Supplementary Table 4.1)\u003c/strong\u003e. For class-based segregation, MSAs that saw reductions in percent of public transport or driving commutes that took over 45 minutes, as well as an increase in public transit commuting, had steeper relationships between residential segregation and workhood segregation, compared to MSAs that did not experience similar changes. Collectively, these findings imply that improved spatial connectivity, through transport infrastructure improvements and/or denser urban form, accentuates the links between residential and workhood class-based segregation–which in turn offer some support to the hypothesis that reduced travel constraints might further accentuate employment niching\u003csup\u003e50\u003c/sup\u003e. \u0026nbsp;Improvements in transportation connectivity might simply allow people to travel more efficiently to nearby employment niches, rather facilitate a wider job search radius and greater diffusion from existing residential enclaves.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eSupplementary analyses using 2010 and 2015 data specifically measuring increases in road lengths and number of road intersections however suggest that changes in road infrastructure per se might not explain these observed patterns. Our supplementary models found that the greater the expansion of road infrastructure, the smaller the associations between residential segregation and workhood segregation \u003cstrong\u003e(Supplementary Table 4.2)\u003c/strong\u003e. \u0026nbsp;Tentatively, we posit that changes in residential and/or employment locational set up (e.g. new neighborhoods being built near existing employment centers) rather than changes in transportation networks might have created shorter commutes and thus tighter links between home and workhood locations. More in-depth studies are needed to unpack why shorter commutes and increased transit share seem to accentuate the ‘circle of segregation’.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eFor diversity and inclusiveness,\u0026nbsp;\u003c/strong\u003eMSAs that grew less racially diverse were also ones where increases in residential segregation was associated with a smaller increase in workhood segregation, compared to MSAs that have a larger minority population profile. Furthermore, workhood racial segregation levels were generally lower for MSAs that had more White residents, across all observed levels of residential segregation. These findings run counter to the hypothesis that more diverse MSAs would have work environments that are more accepting of minority groups. Instead our findings align better with the ‘threat’ hypothesis that suggest higher proportions of minority individuals cause a ‘hunkering down’ of the majority population which manifests in more discriminatory actions against minority groups (\u003csup\u003e56,57\u003c/sup\u003e.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eFor labor market characteristics,\u0026nbsp;\u003c/strong\u003e links between residential and workhood class segregation seemed larger in MSAs where more workers were employed in agriculture and mining related industries, but smaller in MSAs more workers in entertainment and recreation (for class-based segregation) and trade,commerce \u0026amp; transport related industries (for racial segregation). One interpretation would be that the agriculture and mining related industries might be more reliant on personal class-based networks for hiring, whereas the latter two industries might be less reliant on personal networks for hiring. Again, these tentative hypotheses however need further research to validate. \u0026nbsp;\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eLimitations and Proposed Next Steps\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThis study explores the observed relationships between metropolitan-area workhood and residential neighborhood segregation patterns in a descriptive, exploratory manner \u0026nbsp;While our results reveal interesting patterns at an metropolitan level, analyses at this scale necessarily obscures details of neighborhood-by-neighborhood level variations, which is another scale at which to implement important interventions. Additional research looking at how neighborhood-level differences in access to social networks and access to transportation might affect the extent to which workers live and work in segregated or integrated neighborhoods would also be invaluable in unpacking potential causal mechanisms, which is challenging to do with data aggregated at the metropolitan level.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eTwo additional limitations arise due to the nature of our data. We were unable to study ethnicity in tandem with race, as the LODES data does not differentiate between Hispanic and non-Hispanic individuals within each racialised category. \u0026nbsp;Furthermore, as the LODES data captures only those who were working, our calculations of spatial segregation across home and work, and across class and racial categories, necessarily exclude other non-working individuals, such as children, retirees, and job-seekers. Checks using ACS data to calculate residential segregation suggest a somewhat different relationship between worker-based segregation calculations versus total residential population, which would include the unemployed (e.g. children, retirees etc.). Findings from this study should thus be interpreted with a clear understanding that it applies to a specific at-work population.\u0026nbsp;\u003c/p\u003e"},{"header":"Conclusion","content":"\u003cp\u003eThis study’s exploratory examination of 380 U.S. MSAs from 2011 to 2018 confirms a persistent link between residential segregation and workhood segregation; identifies metropolitan characteristics that moderated this link, and provides suggestions on where additional research is needed to advance understanding of spatial segregation in the U.S. Through these descriptive findings about the complex links between residential and workhood segregation, we hope to spark further studies on how to reduce spatial segregation across multiple domains, in a country currently divided by class and race.\u0026nbsp;\u003c/p\u003e"},{"header":"Methods","content":"\u003ch2\u003e\u003cstrong\u003eCalculation of Spatial Segregation Using LODES Home and Workplace Information\u003c/strong\u003e\u003c/h2\u003e\n\u003cp\u003eWe used the LODES (version 8.3) dataset (https://lehd.ces.census.gov/data/lodes/LODES8/, Last accessed December 1, 2024) from the U.S. Census Bureau. The LODES data derive from administrative records, such as state employment insurance reporting and federal worker earnings records, of home and work addresses of individual workers aged 14 and over, and are aggregated to the home and work census blocks for a representative sample of workers (n ~ 128 million living in 72,534 tracts and working in 72,032 tracts in 2011; n ~ 144 million living in 72,242 tracts and working in 72,115 tracts in 2011) for every state excluding Hawaii, Alaska and territories. We focus on the years 2011-2018, for which we had complete LODES residential and workplace information for every state. The data cover some 95% of jobs in the United States.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eFor this analysis, we utilised two datasets from LODES, namely (1) Residence Area Characteristics (RAC), which provides information on workers by census block of residence, and (2) Workplace Area Characteristics (WAC), which provides information on workers by census block of employment as the smallest spatial unit of aggregation. We aggregated the block-level data to the tract-level to reduce uncertainties, and because census tracts are often used as spatial proxies for neighborhoods (Hall et al 2019). \u0026nbsp;\u003c/p\u003e\n\u003cp\u003eTo quantify MSA-level racial segregation, we calculate a multigroup dissimilarity index for four race categories (Asian, Black, White, and Others, which includes all other classifications such as American Indian or Alaska Native; Native Hawaiian or other Pacific Islander, Two or More Race Groups). As the LODES data reports \u0026lsquo;race\u0026rsquo; and \u0026lsquo;ethnicity\u0026rsquo; separately, we were unable to generate a singular measure of race/ethnic segregation that includes Hispanic individuals as a specific population category within our analysis\u0026ndash;which is a limitation of available data and our resultant analysis. To quantify class-based segregation, we focus on educational attainment as a measure of class, and calculate multigroup dissimilarity for four education levels (less than High School, High School, Some College or Associate Degree, Bachelor\u0026rsquo;s Degree or advanced degree). For both sets of calculations, we calculated MSA-level estimates based on census tract-level population information.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eLike Hall et al (2019), our estimates of residential segregation is based on worker populations instead of total residential population, in order to maintain consistency in the underlying population for our analysis. \u0026nbsp; In supplementary analyses (Supplementary Annex 4), we created similar estimates of residential segregation based on the American Community Survey (ACS) 5 year data which includes both working and non-working populations. As its name suggests, the ACS 5 year data is collected over the span of five years rather than a single year. Thus to generate yearly estimates, we assigned estimated segregation values to the middle year of the 5 year period. \u0026nbsp;Correlation between the LODES based estimates and ACS-based estimates were fairly high, at 0.9 for racial segregation, and 0.8 for educational segregation.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eDissimilarity is a measure of spatial clustering and can be interpreted as a measure of how different the composition of individual\u0026rsquo;s localised environments are on average compared to the population composition as a whole, where a score of \u0026nbsp;0 indicates no difference from the broader population as a whole (i.e. complete integration), while the maximum score of 1 indicates the greatest difference possible (i.e. complete segregation). While there are many alternative measures of segregation, here we utilize the dissimilarity index as it is one of the most commonly used and thus easily understood measures in the field. Specifically, we use a spatial measure of dissimilarity that factors in the distance and proximity of census tracts to each other, in order to better capture the spatial dimension of segregation, compared to the commonly used \u0026lsquo;aspatial\u0026rsquo; measure of dissimilarity that assumes the mixing between residents is strictly bounded within the spatial confines of each distinct neighborhood. Details of how the spatial dissimilarity index is calculated can be found in\u0026nbsp;\u003csup\u003e66\u003c/sup\u003e\u003csup\u003e,\u003c/sup\u003e\u003csup\u003e67\u003c/sup\u003e.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eTo examine regional variations in segregation across the U.S., we choose MSAs as our primary aggregated spatial unit of analysis because housing markets and labor markets \u0026ndash;both major drivers of intergroup disparities and spatial segregation\u0026ndash;operate at a metropolitan scale\u0026nbsp;\u003csup\u003e68\u003c/sup\u003e. Expectedly, much of the past research on segregation that informs this study also used MSAs as a unit of analysis\u003csup\u003e19,35,69\u003c/sup\u003e. \u0026nbsp;MSAs are one of the two types of core-based statistical areas (CBSAs). CBSAs comprise counties in most of the U.S, and are built from one or more central counties where a substantial population resides in the same core urban area(s). MSAs are CBSAs that contain an urban area and have at least 50,000 residents. As MSA boundaries can get redefined over the years, which complicates temporal analysis, it is important to utilise a consistent set of spatial boundaries for comparison. For this analysis, we analyzed all 380 MSAs in the contiguous United States based on the 2020 decennial census geographical boundaries. Figure 2 maps out the geographical distribution of residential neighborhood and workhood segregation estimates, for 2018.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eTo answer our second research question about what factors moderate the relationship between residential and workhood segregation, we analyse a set of MSA-level variables that measure metropolitan-level factors that could influence access to jobs via transport infrastructure or urban form, as well as those affecting access to jobs via personal networks. We also include a set of control covariates to account for additional factors that might conceivably drive both residential and workhood segregation. \u003cstrong\u003eTable 2\u003c/strong\u003e summarizes these variables and their sources, while Supplementary Table 5.1 provides summary statistics of these variables.\u0026nbsp;\u003c/p\u003e\n\u003ctable border=\"1\" cellspacing=\"0\" cellpadding=\"0\" width=\"674\"\u003e\n \u003ctbody\u003e\n \u003ctr\u003e\n \u003ctd colspan=\"2\" valign=\"top\" style=\"width: 674px;\"\u003e\n \u003cp\u003e\u003cstrong\u003e\u003cem\u003eTable 2:\u0026nbsp;\u003c/em\u003e\u003c/strong\u003e\u003cem\u003eModerating Variables and Additional Controls when evaluating associations between residential and workplace segregation.\u003c/em\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 355px;\"\u003e\n \u003cp\u003e\u003cstrong\u003eMSA-level variables\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 319px;\"\u003e\n \u003cp\u003e\u003cstrong\u003eSource, years\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd colspan=\"2\" valign=\"top\" style=\"width: 674px;\"\u003e\n \u003cp\u003e\u003cem\u003e(1) Factors affecting spatial access to jobs : Transportation \u0026amp; Urban Form\u003c/em\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 355px;\"\u003e\n \u003cp\u003eAverage Commuting Time (mins)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 319px;\"\u003e\n \u003cp\u003e2009-2013 5-year ACS to 2016-2020 5-year ACS\u003csup\u003e1\u003c/sup\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 355px;\"\u003e\n \u003cp\u003e% of Workers taking public transit \u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 319px;\"\u003e\n \u003cp\u003e2009-2013 5-year ACS to 2016-2020 5-year ACS\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 355px;\"\u003e\n \u003cp\u003ePercentage point differences in % of workers having \u0026gt;45min commutes, between public transit and car commuters\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 319px;\"\u003e\n \u003cp\u003e2009-2013 5-year ACS to 2016-2020 5-year ACS\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 355px;\"\u003e\n \u003cp\u003ePopulation Density (1000 persons/km\u003csup\u003e2\u003c/sup\u003e)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 319px;\"\u003e\n \u003cp\u003e2009-2013 5-year ACS to 2016-2020 5-year ACS\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd colspan=\"2\" valign=\"top\" style=\"width: 674px;\"\u003e\n \u003cp\u003e\u003cem\u003e(2a) Factors affecting access to jobs via personal networks: Diversity \u0026amp; Inclusiveness\u003c/em\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 355px;\"\u003e\n \u003cp\u003ePercent of workers who are non-White\u003csup\u003e2\u003c/sup\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 319px;\"\u003e\n \u003cp\u003eLODES WAC, 2011-2018\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 355px;\"\u003e\n \u003cp\u003ePercentage point differences in unemployment rates between White and non-White civilian labor force ages 16+\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 319px;\"\u003e\n \u003cp\u003e2009-2013 5-year ACS to 2016-2020 5-year ACS\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 355px;\"\u003e\n \u003cp\u003eGini Index of Inequality\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 319px;\"\u003e\n \u003cp\u003e2009-2013 5-year ACS to 2016-2020 5-year ACS\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd colspan=\"2\" valign=\"top\" style=\"width: 674px;\"\u003e\n \u003cp\u003e\u003cem\u003e(2b) Factors affecting access to jobs via personal networks : Labour Market Conditions and Profile\u003c/em\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 355px;\"\u003e\n \u003cp\u003ePercent of civilian labor force ages 16 and above unemployed\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 319px;\"\u003e\n \u003cp\u003e2009-2013 5-year ACS to 2016-2020 5-year ACS\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 355px;\"\u003e\n \u003cp\u003ePercent of workers in the following sectors\u003csup\u003e2\u003c/sup\u003e\u003c/p\u003e\n \u003col\u003e\n \u003cli\u003eAgriculture/ Mining/Utilities/Construction\u003c/li\u003e\n \u003cli\u003eManufacturing\u003c/li\u003e\n \u003cli\u003eTrade/Commerce/Transport\u003c/li\u003e\n \u003cli\u003eProfessional Services\u003c/li\u003e\n \u003cli\u003ePublic/Social Services\u003c/li\u003e\n \u003cli\u003eEntertainment/Recreation\u003c/li\u003e\n \u003c/ol\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 319px;\"\u003e\n \u003cp\u003eLODES WAC 2011-2018\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd colspan=\"2\" valign=\"top\" style=\"width: 674px;\"\u003e\n \u003cp\u003e\u003cem\u003eControl Variables\u003c/em\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 355px;\"\u003e\n \u003cp\u003eGDP per capita \u0026nbsp; \u0026nbsp;\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 319px;\"\u003e\n \u003cp\u003eBureau of Economic Analysis, 2011-2018\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 355px;\"\u003e\n \u003cp\u003ePercent of workers aged \u0026ge; 55 y\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 319px;\"\u003e\n \u003cp\u003eLODES WAC, 2011-2018\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 355px;\"\u003e\n \u003cp\u003ePercent of workers earning \u0026le; USD $1,250/month\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 319px;\"\u003e\n \u003cp\u003eLODES WAC, 2011-2018\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 355px;\"\u003e\n \u003cp\u003ePercent of workers who were female\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 319px;\"\u003e\n \u003cp\u003eLODES WAC, 2011-2018\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 355px;\"\u003e\n \u003cp\u003ePercent of workers with a college degree\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 319px;\"\u003e\n \u003cp\u003eLODES WAC, 2011-2018\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 355px;\"\u003e\n \u003cp\u003ePercent of workers classified as living in a suburban area\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 319px;\"\u003e\n \u003cp\u003eLODES WAC, 2011-2018\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003c/tbody\u003e\n\u003c/table\u003e\n\u003col\u003e\n \u003cli\u003eSimilar to how we used the ACS 5 year data to calculate residential segregation measure, all other covariates calculated using ACS 5 year data were attributed to the middle of the five year period (e.g. 2011 for 2009-2013 data)\u003c/li\u003e\n \u003cli\u003eIn this analysis, we only considered primary jobs, and not secondary or tertiary jobs. We did so as we did not have any information on how much time workers spent at each job site when workers had multiple jobs. As the LODES provides data on many different categories of job types, which would be unwieldy in our analysis, we aggregated them into six broad sectors, as such: 1.1: Agriculture/Forestry/Fishing/Hunting, 1.2: Mining/Quarrying/Oil and Gas Extraction, 1.3: Utilities, 1.4: Construction, 2:. Manufacturing, \u0026nbsp;3.1:Wholesale Trade, 3.2: Retail Trade, 3.3: Transportation and Warehousing, 4.1: Information. 4.2: Finance and Insurance, 4.3: Real Estate and Rental and Leasing, 4.4: Professional/Scientific and Technical Services, 4.5: Management of Companies and Enterprises, 4.6: Administrative and Support and Waste Management and Remediation Services, \u0026nbsp;5.1: Educational Services, \u0026nbsp;5.2: Healthcare and Social Assistance, \u0026nbsp;5.3:Public Administration, \u0026nbsp;6.1: Arts/Entertainment and Recreation, 6.2: Accommodation and Food Services\u003c/li\u003e\n\u003c/ol\u003e\n\u003ch2\u003e\u003cstrong\u003eStatistical Analysis Approach\u003c/strong\u003e\u003c/h2\u003e\n\u003cp\u003eFor our analysis, we fit a series of regression models with MSA-level and year-fixed effects, and examined how changes in workhood segregation, our outcome variable, might be associated with changes in residential segregation and its interaction with the hypothesized moderating variables described above. Using MSA-level fixed effects essentially allows us to control for time-invariant characteristics that may confound the relationship of interest, and thus offering a way to overcome potential omitted variable bias \u003csup\u003e70\u003c/sup\u003e. \u0026nbsp;\u003c/p\u003e\n\u003cp\u003eAs some of our proposed predictor and moderator variables were quite highly correlated (see \u003cstrong\u003eFigure 5.1\u0026nbsp;\u003c/strong\u003efor bi-variate correlation plots for year 2018), multicollinearity might be a problem if all proposed predictor and moderator variables were included into one model. We thus fitted a series of models that included only one MSA-level moderator variable as well as its interaction with residential segregation score, together with the MSA-level controls (Supplementary Figure 3.1). Moderator variables with modelled coefficients of significance p\u0026lt;0.05, were added into a combined model (Model 1), while Model 2 presents a further streamlined model that included only the moderator variables or their interaction with residential segregation that were significant (p\u0026lt;0.05) from Model 1. To assess the relationships between racial/ethnic and class-based segregation, Model 3 further includes a measure of residential segregation along the other category of segregation (e.g. class-based segregation for the models with racial/ethnic segregation as an outcome, and vice versa), while Model 4 includes a measure of workhood segregation along the same alternative category as Model 3. The Results section report estimates from Model 4, whereas detailed tables of all models are presented in Supplementary Tables 3.1 and 3.2)\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eAs robustness checks, we repeated the fixed effects regression models 2 to 4 with alternative data sources. These include ACS 5 survey year data to estimate residential racial and class segregation; as well as alternative measures of MSA-level urban density and road transport infrastructure, as of 2010 and 2015, produced by \u003csup\u003e71\u003c/sup\u003e. Details of how these alternative variables were calculated, the model results and interpretation are reported in \u003cstrong\u003eSupplementary Annex 4.\u0026nbsp;\u003c/strong\u003e\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eAll analysis was conducted in R, version 4.3.2. The two way fixed-effects models were fitted using package \u0026lsquo;plm\u0026rsquo; version 2.6-6, while the spatial segregation scores were estimated using the package \u0026lsquo;seg\u0026rsquo; version 0.5-7.\u003c/p\u003e"},{"header":"Declarations","content":"\u003cp\u003e\u003cstrong\u003eData availability\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eAccess to the data we generated for our analysis is available at GitHub : https://github.com/redacted_link\u0026nbsp;\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eCode availability\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eAccess to the code we used to run our analysis is available at GitHub : https://github.com/redacted_link\u0026nbsp;\u003c/p\u003e"},{"header":"References","content":"\u003col\u003e\n \u003cli\u003eFrankenberg, E. The Role of Residential Segregation in Contemporary School Segregation. \u003cem\u003eEduc. Urban Soc.\u003c/em\u003e \u003cstrong\u003e45\u003c/strong\u003e, 548\u0026ndash;570 (2013).\u003c/li\u003e\n \u003cli\u003eEkenga, C. C. \u0026amp; Tian, R. Promoting Food Equity in the Context of Residential Segregation. \u003cem\u003eEnviron. Justice\u003c/em\u003e \u003cstrong\u003e15\u003c/strong\u003e, 346\u0026ndash;351 (2022).\u003c/li\u003e\n \u003cli\u003eBower, K. M., Thorpe, R. J., Rohde, C. \u0026amp; Gaskin, D. J. The intersection of neighborhood racial segregation, poverty, and urbanicity and its impact on food store availability in the United States. \u003cem\u003ePrev. Med.\u003c/em\u003e \u003cstrong\u003e58\u003c/strong\u003e, 33\u0026ndash;39 (2014).\u003c/li\u003e\n \u003cli\u003eAnderson, K. F. \u0026amp; Galaskiewicz, J. Racial/Ethnic Residential Segregation, Socioeconomic Inequality, and Job Accessibility by Public Transportation Networks in the United States. \u003cem\u003eSpat. Demogr.\u003c/em\u003e \u003cstrong\u003e9\u003c/strong\u003e, 341\u0026ndash;373 (2021).\u003c/li\u003e\n \u003cli\u003evonLockette, N. T. D. \u0026amp; Johnson, J. LATINO EMPLOYMENT AND RESIDENTIAL SEGREGATION IN METROPOLITAN LABOR MARKETS. \u003cem\u003eBois Rev. Soc. Sci. Res. Race\u003c/em\u003e \u003cstrong\u003e7\u003c/strong\u003e, 151\u0026ndash;184 (2010).\u003c/li\u003e\n \u003cli\u003eVonLockette, N. D. The Impact of Metropolitan Residential Segregation on the Employment Chances of Blacks and Whites in the United States. \u003cem\u003eCity Community\u003c/em\u003e \u003cstrong\u003e9\u003c/strong\u003e, 256\u0026ndash;273 (2010).\u003c/li\u003e\n \u003cli\u003eCaldwell, J. T., Ford, C. L., Wallace, S. P., Wang, M. C. \u0026amp; Takahashi, L. M. Racial and ethnic residential segregation and access to health care in rural areas. \u003cem\u003eHealth Place\u003c/em\u003e \u003cstrong\u003e43\u003c/strong\u003e, 104\u0026ndash;112 (2017).\u003c/li\u003e\n \u003cli\u003eCarr, J. H. \u0026amp; Kutty, N. K. \u003cem\u003eSegregation: The Rising Costs for America\u003c/em\u003e. (Routledge, 2008).\u003c/li\u003e\n \u003cli\u003eMassey, D. S. Still the Linchpin: Segregation and Stratification in the USA. \u003cem\u003eRace Soc. Probl.\u003c/em\u003e \u003cstrong\u003e12\u003c/strong\u003e, 1\u0026ndash;12 (2020).\u003c/li\u003e\n \u003cli\u003eWilliams, D. R. \u0026amp; Collins, C. Racial residential segregation: a fundamental cause of racial disparities in health. \u003cem\u003ePublic Health Rep.\u003c/em\u003e \u003cstrong\u003e116\u003c/strong\u003e, 404\u0026ndash;416 (2001).\u003c/li\u003e\n \u003cli\u003eBaron, J. N. \u0026amp; Bielby, W. T. Bringing the Firms Back in: Stratification, Segmentation, and the Organization of Work. \u003cem\u003eAm. Sociol. Rev.\u003c/em\u003e \u003cstrong\u003e45\u003c/strong\u003e, 737\u0026ndash;765 (1980).\u003c/li\u003e\n \u003cli\u003eMarcińczak, S., Tammaru, T., Str\u0026ouml;mgren, M. \u0026amp; Lindgren, U. Changing patterns of residential and workplace segregation in the Stockholm metropolitan area. \u003cem\u003eUrban Geogr.\u003c/em\u003e \u003cstrong\u003e36\u003c/strong\u003e, 969\u0026ndash;992 (2015).\u003c/li\u003e\n \u003cli\u003eTomaskovic-Devey, D. \u003cem\u003eet al.\u003c/em\u003e Documenting Desegregation: Segregation in American Workplaces by Race, Ethnicity, and Sex, 1966\u0026ndash;2003. \u003cem\u003eAm. Sociol. Rev.\u003c/em\u003e \u003cstrong\u003e71\u003c/strong\u003e, 565\u0026ndash;588 (2006).\u003c/li\u003e\n \u003cli\u003eMijs, J. J. B. \u0026amp; Roe, E. L. Is America coming apart? Socioeconomic segregation in neighborhoods, schools, workplaces, and social networks, 1970\u0026ndash;2020. \u003cem\u003eSociol. Compass\u003c/em\u003e \u003cstrong\u003e15\u003c/strong\u003e, e12884 (2021).\u003c/li\u003e\n \u003cli\u003eSong, J., Price, D. J., Guvenen, F., Bloom, N. \u0026amp; von Wachter, T. Firming Up Inequality*. \u003cem\u003eQ. J. Econ.\u003c/em\u003e \u003cstrong\u003e134\u003c/strong\u003e, 1\u0026ndash;50 (2019).\u003c/li\u003e\n \u003cli\u003eStainback, K. \u0026amp; Tomaskovic-Devey, D. \u003cem\u003eDocumenting Desegregation: Racial and Gender Segregation in Private Sector Employment Since the Civil Rights Act\u003c/em\u003e. (Russell Sage Foundation, 2012).\u003c/li\u003e\n \u003cli\u003eEllis, M., Wright, R. \u0026amp; Parks, V. Work Together, Live Apart? Geographies of Racial and Ethnic Segregation at Home and at Work. \u003cem\u003eAnn. Assoc. Am. Geogr.\u003c/em\u003e \u003cstrong\u003e94\u003c/strong\u003e, 620\u0026ndash;637 (2004).\u003c/li\u003e\n \u003cli\u003eHu, L., Moayyed, H. \u0026amp; Frank, N. Racial/ethnic and class segregation at workplace and residence: A Los Angeles case study. \u003cem\u003eJ. Urban Aff.\u003c/em\u003e \u003cstrong\u003e44\u003c/strong\u003e, 907\u0026ndash;925 (2022).\u003c/li\u003e\n \u003cli\u003eHall, M., Iceland, J. \u0026amp; Yi, Y. Racial Separation at Home and Work: Segregation in Residential and Workplace Settings. \u003cem\u003ePopul. Res. Policy Rev.\u003c/em\u003e \u003cstrong\u003e38\u003c/strong\u003e, 671\u0026ndash;694 (2019).\u003c/li\u003e\n \u003cli\u003eTammaru, T., Knapp, D., Silm, S., Van Ham, M. \u0026amp; Witlox, F. Spatial Underpinnings of Social Inequalities: A Vicious Circles of Segregation Approach. \u003cem\u003eSoc. Incl.\u003c/em\u003e \u003cstrong\u003e9\u003c/strong\u003e, 65\u0026ndash;76 (2021).\u003c/li\u003e\n \u003cli\u003eDelmelle, E., Nilsson, I. \u0026amp; Adu, P. Poverty Suburbanization, Job Accessibility, and Employment Outcomes. \u003cem\u003eSoc. Incl.\u003c/em\u003e \u003cstrong\u003e9\u003c/strong\u003e, 166\u0026ndash;178 (2021).\u003c/li\u003e\n \u003cli\u003eCoulter, R., Ham, M. van \u0026amp; Findlay, A. M. Re-thinking residential mobility: Linking lives through time and space. \u003cem\u003eProg. Hum. Geogr.\u003c/em\u003e \u003cstrong\u003e40\u003c/strong\u003e, 352\u0026ndash;374 (2016).\u003c/li\u003e\n \u003cli\u003eSimpson, W. \u003cem\u003eURBAN STRUCTURE AND THE LABOUR MARKET. WORKER MOBILITY, COMMUTING AND UNDEREMPLOYMENT IN CITIES\u003c/em\u003e. (1992).\u003c/li\u003e\n \u003cli\u003eTrimble, L. B. \u0026amp; Kmec, J. A. The Role of Social Networks in Getting a Job. \u003cem\u003eSociol. Compass\u003c/em\u003e \u003cstrong\u003e5\u003c/strong\u003e, 165\u0026ndash;178 (2011).\u003c/li\u003e\n \u003cli\u003eIoannides, Y. M. \u0026amp; Datcher Loury, L. Job Information Networks, Neighborhood Effects, and Inequality. \u003cem\u003eJ. Econ. Lit.\u003c/em\u003e \u003cstrong\u003e42\u003c/strong\u003e, 1056\u0026ndash;1093 (2004).\u003c/li\u003e\n \u003cli\u003eGranovetter, M. The Strength of Weak Ties: A Network Theory Revisited. \u003cem\u003eSociol. Theory\u003c/em\u003e \u003cstrong\u003e1\u003c/strong\u003e, 201\u0026ndash;233 (1983).\u003c/li\u003e\n \u003cli\u003eMouw, T. Are Black Workers Missing the Connection?: The Effect of Spatial Distance and Employee Referrals on Interfirm Racial Segregation*. \u003cem\u003eDemography\u003c/em\u003e \u003cstrong\u003e39\u003c/strong\u003e, 507\u0026ndash;528 (2002).\u003c/li\u003e\n \u003cli\u003eTassier, T. A Markov Model of Referral-Based Hiring and Workplace Segregation. \u003cem\u003eJ. Math. Sociol.\u003c/em\u003e \u003cstrong\u003e29\u003c/strong\u003e, 233\u0026ndash;262 (2005).\u003c/li\u003e\n \u003cli\u003eHellerstein, J. K. \u0026amp; Neumark, D. Workplace Segregation in the United States: Race, Ethnicity, and Skill. \u003cem\u003eRev. Econ. Stat.\u003c/em\u003e \u003cstrong\u003e90\u003c/strong\u003e, 459\u0026ndash;477 (2008).\u003c/li\u003e\n \u003cli\u003ePatacchini, E. \u0026amp; Zenou, Y. Racial identity and education in social networks. \u003cem\u003eSoc. Netw.\u003c/em\u003e \u003cstrong\u003e44\u003c/strong\u003e, 85\u0026ndash;94 (2016).\u003c/li\u003e\n \u003cli\u003eEwing, R. \u0026amp; and Cervero, R. \u0026ldquo;Does Compact Development Make People Drive Less?\u0026rdquo; The Answer Is Yes. \u003cem\u003eJ. Am. Plann. Assoc.\u003c/em\u003e \u003cstrong\u003e83\u003c/strong\u003e, 19\u0026ndash;25 (2017).\u003c/li\u003e\n \u003cli\u003eLee, C. Metropolitan sprawl measurement and its impacts on commuting trips and road emissions. \u003cem\u003eTransp. Res. Part Transp. Environ.\u003c/em\u003e \u003cstrong\u003e82\u003c/strong\u003e, 102329 (2020).\u003c/li\u003e\n \u003cli\u003eLevinson, D. Network Structure and City Size. \u003cem\u003ePLOS ONE\u003c/em\u003e \u003cstrong\u003e7\u003c/strong\u003e, e29721 (2012).\u003c/li\u003e\n \u003cli\u003eStevens, M. R. Does Compact Development Make People Drive Less? \u003cem\u003eJ. Am. Plann. Assoc.\u003c/em\u003e \u003cstrong\u003e83\u003c/strong\u003e, 7\u0026ndash;18 (2017).\u003c/li\u003e\n \u003cli\u003eGalaskiewicz, J., Anderson, K. F. \u0026amp; Thompson-Dyck, K. Minority-White income inequality across metropolitan areas: The role of racial/ethnic residential segregation and transportation networks. \u003cem\u003eJ. Urban Aff.\u003c/em\u003e \u003cstrong\u003e43\u003c/strong\u003e, 16\u0026ndash;39 (2021).\u003c/li\u003e\n \u003cli\u003eWissink, B., Schwanen, T. \u0026amp; van Kempen, R. Beyond residential segregation: Introduction. \u003cem\u003eCities\u003c/em\u003e \u003cstrong\u003e59\u003c/strong\u003e, 126\u0026ndash;130 (2016).\u003c/li\u003e\n \u003cli\u003eCovington, K. L. Overcoming Spatial Mismatch: The Opportunities and Limits of Transit Mode in Addressing the Black\u0026ndash;White Unemployment Gap. \u003cem\u003eCity Community\u003c/em\u003e \u003cstrong\u003e17\u003c/strong\u003e, 211\u0026ndash;235 (2018).\u003c/li\u003e\n \u003cli\u003ePucher, J. \u0026amp; Renne, J. L. Socioeconomics of urban travel: evidence from the 2001 NHTS. \u003cem\u003eTransp. Q.\u003c/em\u003e \u003cstrong\u003e57\u003c/strong\u003e, 49\u0026ndash;77 (2003).\u003c/li\u003e\n \u003cli\u003eSanchez, T. W. The Impact of Public Transport on US Metropolitan Wage Inequality. \u003cem\u003eUrban Stud.\u003c/em\u003e \u003cstrong\u003e39\u003c/strong\u003e, 423\u0026ndash;436 (2002).\u003c/li\u003e\n \u003cli\u003eDuany, A., Plater-Zyberk, E. \u0026amp; Speck, J. \u003cem\u003eSuburban Nation: The Rise of Sprawl and the Decline of the American Dream\u003c/em\u003e. (North Point Press, New York, 2000).\u003c/li\u003e\n \u003cli\u003eGolub, A., Marcantonio, R. A. \u0026amp; Sanchez, T. W. Race, Space, and Struggles for Mobility: Transportation Impacts on African Americans in Oakland and the East Bay. \u003cem\u003eUrban Geogr.\u003c/em\u003e \u003cstrong\u003e34\u003c/strong\u003e, 699\u0026ndash;728 (2013).\u003c/li\u003e\n \u003cli\u003eGoodling, E., Green, J. \u0026amp; McClintock, N. Uneven development of the sustainable city: shifting capital in Portland, Oregon. \u003cem\u003eUrban Geogr.\u003c/em\u003e \u003cstrong\u003e36\u003c/strong\u003e, 504\u0026ndash;527 (2015).\u003c/li\u003e\n \u003cli\u003eSharkey, P. \u003cem\u003eStuck in Place\u003c/em\u003e. (2013).\u003c/li\u003e\n \u003cli\u003eTrounstine, J. \u003cem\u003eSegregation by Design: Local Politics and Inequality in American Cities\u003c/em\u003e. (Cambridge University Press, Cambridge, 2018). doi:10.1017/9781108555722.\u003c/li\u003e\n \u003cli\u003eBereitschaft, B. Equity in neighbourhood walkability? A comparative analysis of three large U.S. cities. \u003cem\u003eLocal Environ.\u003c/em\u003e \u003cstrong\u003e22\u003c/strong\u003e, 859\u0026ndash;879 (2017).\u003c/li\u003e\n \u003cli\u003eGrengs, J. Nonwork Accessibility as a Social Equity Indicator. \u003cem\u003eInt. J. Sustain. Transp.\u003c/em\u003e \u003cstrong\u003e9\u003c/strong\u003e, 1\u0026ndash;14 (2015).\u003c/li\u003e\n \u003cli\u003eShin, E. J. Disparities in access to opportunities across neighborhoods types: a case study from the Los Angeles region. \u003cem\u003eTransportation\u003c/em\u003e \u003cstrong\u003e47\u003c/strong\u003e, 475\u0026ndash;501 (2020).\u003c/li\u003e\n \u003cli\u003eLee, J., Vojnovic, I. \u0026amp; Grady, S. C. The \u0026lsquo;transportation disadvantaged\u0026rsquo;: Urban form, gender and automobile versus non-automobile travel in the Detroit region. \u003cem\u003eUrban Stud.\u003c/em\u003e \u003cstrong\u003e55\u003c/strong\u003e, 2470\u0026ndash;2498 (2018).\u003c/li\u003e\n \u003cli\u003eMcKenzie, B. S. Neighborhood Access to Transit by Race, Ethnicity, and Poverty in Portland, OR. \u003cem\u003eCity Community\u003c/em\u003e \u003cstrong\u003e12\u003c/strong\u003e, 134\u0026ndash;155 (2013).\u003c/li\u003e\n \u003cli\u003eEllis, M., Wright, R. \u0026amp; Parks, V. Geography and the Immigrant Division of Labor. \u003cem\u003eEcon. Geogr.\u003c/em\u003e \u003cstrong\u003e83\u003c/strong\u003e, 255\u0026ndash;281 (2007).\u003c/li\u003e\n \u003cli\u003eElliott, J. R. Referral Hiring and Ethnically Homogeneous Jobs: How Prevalent Is the Connection and for Whom? \u003cem\u003eSoc. Sci. Res.\u003c/em\u003e \u003cstrong\u003e30\u003c/strong\u003e, 401\u0026ndash;425 (2001).\u003c/li\u003e\n \u003cli\u003eChristopher, S. C. \u0026amp; Leslie, T. F. Assessing Immigrant Niches Across Large American Metropolitan Areas. \u003cem\u003ePopul. Space Place\u003c/em\u003e \u003cstrong\u003e21\u003c/strong\u003e, 171\u0026ndash;192 (2015).\u003c/li\u003e\n \u003cli\u003eCrul, M. Super-diversity vs. assimilation: how complex diversity in majority\u0026ndash;minority cities challenges the assumptions of assimilation. \u003cem\u003eJ. Ethn. Migr. Stud.\u003c/em\u003e \u003cstrong\u003e42\u003c/strong\u003e, 54\u0026ndash;68 (2016).\u003c/li\u003e\n \u003cli\u003eGlaeser, E. L., Kallal, H. D., Scheinkman, J. A. \u0026amp; Shleifer, A. Growth in Cities. \u003cem\u003eJ. Polit. Econ.\u003c/em\u003e \u003cstrong\u003e100\u003c/strong\u003e, 1126\u0026ndash;1152 (1992).\u003c/li\u003e\n \u003cli\u003eMaxwell, R. Cosmopolitan Immigration Attitudes in Large European Cities: Contextual or Compositional Effects? \u003cem\u003eAm. Polit. Sci. Rev.\u003c/em\u003e \u003cstrong\u003e113\u003c/strong\u003e, 456\u0026ndash;474 (2019).\u003c/li\u003e\n \u003cli\u003eMoller, S., Alderson, A. S. \u0026amp; Nielsen, F. Changing Patterns of Income Inequality in U.S. Counties, 1970\u0026ndash;2000. \u003cem\u003eAm. J. Sociol.\u003c/em\u003e \u003cstrong\u003e114\u003c/strong\u003e, 1037\u0026ndash;1101 (2009).\u003c/li\u003e\n \u003cli\u003eParks, V. The Uneven Geography of Racial and Ethnic Wage Inequality: Specifying Local Labor Market Effects. \u003cem\u003eAnn. Assoc. Am. Geogr.\u003c/em\u003e \u003cstrong\u003e102\u003c/strong\u003e, 700\u0026ndash;725 (2012).\u003c/li\u003e\n \u003cli\u003eTrimble, L. B. \u0026amp; Kmec, J. A. The Role of Social Networks in Getting a Job. \u003cem\u003eSociol. Compass\u003c/em\u003e \u003cstrong\u003e5\u003c/strong\u003e, 165\u0026ndash;178 (2011).\u003c/li\u003e\n \u003cli\u003eMarsden, P. V. \u0026amp; Gorman, E. H. Social Networks, Job Changes, and Recruitment. in \u003cem\u003eSourcebook of Labor Markets\u003c/em\u003e 467\u0026ndash;502 (Springer, Boston, MA, 2001). doi:10.1007/978-1-4615-1225-7_19.\u003c/li\u003e\n \u003cli\u003eEllis, M., Wright, R. \u0026amp; Parks, V. Geography and the Immigrant Division of Labor. \u003cem\u003eEcon. Geogr.\u003c/em\u003e \u003cstrong\u003e83\u003c/strong\u003e, 255\u0026ndash;281 (2007).\u003c/li\u003e\n \u003cli\u003eWright, R., Ellis, M. \u0026amp; Parks, V. Immigrant Niches and the Intrametropolitan Spatial Division of Labour. \u003cem\u003eJ. Ethn. Migr. Stud.\u003c/em\u003e \u003cstrong\u003e36\u003c/strong\u003e, 1033\u0026ndash;1059 (2010).\u003c/li\u003e\n \u003cli\u003eLillehagen, M. \u0026amp; Hermansen, A. S. Entering the mainstream economy? Workplace segregation and immigrant assimilation. \u003cem\u003eSoc. Forces\u003c/em\u003e soae162 (2024) doi:10.1093/sf/soae162.\u003c/li\u003e\n \u003cli\u003eEllis, M., Wright, R. \u0026amp; Parks, V. Work Together, Live Apart? Geographies of Racial and Ethnic Segregation at Home and at Work. \u003cem\u003eAnn. Assoc. Am. Geogr.\u003c/em\u003e \u003cstrong\u003e94\u003c/strong\u003e, 620\u0026ndash;637 (2004).\u003c/li\u003e\n \u003cli\u003eMassey, D. S. \u0026amp; Rugh, J. S. America\u0026rsquo;s Unequal Metropolitan Geography: Segregation and the Spatial Concentration of Affluence and Poverty. in \u003cem\u003eWho Gets What?\u003c/em\u003e (eds. Rosenbluth, F. M. \u0026amp; Weir, M.) 161\u0026ndash;187 (Cambridge University Press, 2021). doi:10.1017/9781108879170.007.\u003c/li\u003e\n \u003cli\u003eIceland, J. \u0026amp; Wilkes, R. Does Socioeconomic Status Matter? Race, Class, and Residential Segregation. \u003cem\u003eSoc. Probl.\u003c/em\u003e \u003cstrong\u003e53\u003c/strong\u003e, 248\u0026ndash;273 (2006).\u003c/li\u003e\n \u003cli\u003eReardon, S. F. \u0026amp; O\u0026rsquo;Sullivan, D. Measures of spatial segregation. \u003cem\u003eSociol. Methodol.\u003c/em\u003e \u003cstrong\u003e34\u003c/strong\u003e, 121\u0026ndash;162 (2004).\u003c/li\u003e\n \u003cli\u003eHong, S.-Y., O\u0026rsquo;Sullivan, D. \u0026amp; Sadahiro, Y. Implementing Spatial Segregation Measures in R. \u003cem\u003ePLoS ONE\u003c/em\u003e \u003cstrong\u003e9\u003c/strong\u003e, (2014).\u003c/li\u003e\n \u003cli\u003eOsypuk, T. L. \u0026amp; Galea, S. What Level Macro? Choosing Appropriate Levels to Assess How Place Influences Population Health. in \u003cem\u003eMacrosocial Determinants of Population Health\u003c/em\u003e 399\u0026ndash;435 (Springer, New York, NY, 2007). doi:10.1007/978-0-387-70812-6_19.\u003c/li\u003e\n \u003cli\u003eKotecki, J. A., Gennuso, K. P., Givens, M. L. \u0026amp; Kindig, D. A. Separate and Sick: Residential Segregation and the Health of Children and Youth in Metropolitan Statistical Areas. \u003cem\u003eJ. Urban Health Bull. N. Y. Acad. Med.\u003c/em\u003e \u003cstrong\u003e96\u003c/strong\u003e, 149\u0026ndash;158 (2019).\u003c/li\u003e\n \u003cli\u003eAllison, P. D. Introduction. in \u003cem\u003eFixed effects regression models\u003c/em\u003e (SAGE, Los Angeles, 2009).\u003c/li\u003e\n \u003cli\u003eBurghardt, K., Uhl, J. H., Lerman, K. \u0026amp; Leyk, S. Road network evolution in the urban and rural United States since 1900. \u003cem\u003eComput. Environ. Urban Syst.\u003c/em\u003e \u003cstrong\u003e95\u003c/strong\u003e, 101803 (2022).\u003c/li\u003e\n \u003cli\u003eUhl, J. H. \u0026amp; Leyk, S. Historical built-up areas (BUA) - gridded surfaces for the U.S. from 1810 to 2015. Harvard Dataverse https://doi.org/10.7910/DVN/J6CYUJ (2021).\u003c/li\u003e\n\u003c/ol\u003e"}],"fulltextSource":"","fullText":"","funders":[],"hasAdminPriorityOnWorkflow":false,"hasManuscriptDocX":true,"hasOptedInToPreprint":true,"hasPassedJournalQc":"","hasAnyPriority":true,"hideJournal":true,"highlight":"","institution":"","isAcceptedByJournal":false,"isAuthorSuppliedPdf":false,"isDeskRejected":"","isHiddenFromSearch":false,"isInQc":false,"isInWorkflow":false,"isPdf":false,"isPdfUpToDate":true,"isWithdrawnOrRetracted":false,"journal":{"display":true,"email":"[email protected]","identity":"researchsquare","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":true,"externalIdentity":"","sideBox":"","snPcode":"","submissionUrl":"/submission","title":"Research Square","twitterHandle":"researchsquare","acdcEnabled":true,"dfaEnabled":false,"editorialSystem":"","reportingPortfolio":"","inReviewEnabled":false,"inReviewRevisionsEnabled":true},"keywords":"","lastPublishedDoi":"10.21203/rs.3.rs-6771912/v1","lastPublishedDoiUrl":"https://doi.org/10.21203/rs.3.rs-6771912/v1","license":{"name":"CC BY 4.0","url":"https://creativecommons.org/licenses/by/4.0/"},"manuscriptAbstract":"To date, studies examining patterns of workplace area segregation, or workhoods, have been few and limited in both geographical scope and temporal coverage. While available studies have observed positive correlations between residential and workhood segregation, few of these studies attempt to unpack the nuances of this relationship by identifying the factors, such as transportation networks, that might moderate this positive link. Speaking to this gap, we study the relationship between residential and workplace racial/ethnic and socioeconomic segregation of 380 metropolitan statistical areas within the U.S., from 2011 to 2018. Using two-way fixed effects models, we find that the positive correlations between changes in residential and workhood segregation are significantly modified by changes in transportation and urban-form related characteristics, economic characteristics, as well as population diversity.","manuscriptTitle":"Spatial Segregation by Class and Race in Neighborhoods and Workhoods across U.S. metro areas","msid":"","msnumber":"","nonDraftVersions":[{"code":1,"date":"2025-06-13 12:14:31","doi":"10.21203/rs.3.rs-6771912/v1","editorialEvents":[{"type":"communityComments","content":0}],"status":"published","journal":{"display":true,"email":"[email protected]","identity":"researchsquare","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":true,"externalIdentity":"","sideBox":"","snPcode":"","submissionUrl":"/submission","title":"Research Square","twitterHandle":"researchsquare","acdcEnabled":true,"dfaEnabled":false,"editorialSystem":"","reportingPortfolio":"","inReviewEnabled":false,"inReviewRevisionsEnabled":true}}],"origin":"","ownerIdentity":"370af53f-181a-471e-8d10-2100b16d6465","owner":[],"postedDate":"June 13th, 2025","published":true,"recentEditorialEvents":[],"rejectedJournal":[],"revision":"","amendment":"","status":"posted","subjectAreas":[{"id":49954700,"name":"Social science/Geography"},{"id":49954701,"name":"Social science/Environmental studies"},{"id":49954702,"name":"Social science/Sociology"}],"tags":[],"updatedAt":"2025-12-22T20:50:14+00:00","versionOfRecord":[],"versionCreatedAt":"2025-06-13 12:14:31","video":"","vorDoi":"","vorDoiUrl":"","workflowStages":[]},"version":"v1","identity":"rs-6771912","journalConfig":"researchsquare"},"__N_SSP":true},"page":"/article/[identity]/[[...version]]","query":{"redirect":"/article/rs-6771912","identity":"rs-6771912","version":["v1"]},"buildId":"8U1c8b4HqxoKbykW_rLl7","isFallback":false,"isExperimentalCompile":false,"dynamicIds":[84888],"gssp":true,"scriptLoader":[]}