A Scientometric Review of Residential Segregation Research: A CiteSpace-Based Visualization

Abstract

Residential segregation (RS) is a global phenomenon that has become an enduring and important topic in international academic research. In this review, using RS as the search term, 2520 articles from the period 1928–2022 were retrieved from the Scopus database and were visually analyzed using CiteSpace software. The results revealed the following: (1) The United States and its institutions have made outstanding contributions to RS research, while various scholars (e.g., Johnston, Massey, Forrest, Poulsen, and Iceland) have laid the foundation for RS research. (2) Mainstream RS research originates from three fields—psychology, education, and social sciences—while the trend of multidisciplinary integration is constantly increasing. (3) The research hotspots of RS include racial difference, sociospatial behavior, income inequality, mixed income communities, guest worker minorities, typical district segregation, occupational segregation, health inequalities, metropolitan ghetto, and migrant–native differential mobility. Furthermore, (4) gentrification, spatial analysis, school segregation, health disparity, immigrant, and COVID-19 have become new themes and directions of RS research. Future research should pay more attention to the impact of multi-spatial scale changes on RS as well as propose theoretical explanations rooted in local contexts by integrating multidisciplinary theoretical knowledge.

1. Introduction

Residential segregation (RS) is a global phenomenon that has received widespread public attention because of its complex impact on social development, thereby becoming a long-standing research topic in academia. Segregation is the extent to which individuals from different groups occupy and experience different social environments [1]. White posited that RS refers to the non-random spatial distribution of groups and the formation of systematic residential patterns based on social characteristics [2], including ethnicity, religion, occupation, income, and cultural background. Urban residents with similar social characteristics tend to cluster in specific areas, which results in residents or groups with different characteristics living apart from each other in space as well as the emergence of RS [3]. Due to exclusion by a powerful or majority group, economically disadvantaged groups or ethnic minorities are often (reluctantly) clustered into specific areas (e.g., ghettos); therefore, segregation is “the process of formation and maintenance of a ghetto” [4] (p. 16). However, this definition is inevitably too narrow, as it ignores voluntary segregation. A more widely accepted definition establishes segregation as two or more groups of people living separately from each other in different parts of the urban environment [5].

RS in Western countries emerged with the birth of the modern city. In The Condition of the Working Class in England, Engels described the gap between the rich and the poor in urban space in Manchester, noting a vast difference in the living space of the poor (living in slums) and the rich (living in comfortable and affluent zones) [6]. Scholars from the Chicago School in the 1920s conducted pioneering research on RS. Burgess proposed the concentric zones model, which described the division of a city from the inside out and examined the RS between different classes, arguing that it was mainly driven by economic forces [7]. In 1939, Hoyt proposed the sectors theory, which posits that cities’ residential areas fan outward from their center, along traffic lines [8]. Both the concentric zone theory and the sectors theory assume a single urban center as the premise. Based on this, Harris and Ullman proposed a more refined multiple nuclei model that assumes that cities contain a primary core and multiple secondary cores, each surrounded by supporting residential areas, while each residential area is segregated [9]. The multi-nuclei model reflects the complexity of RS. These three models laid the foundation for RS research. In 1955, Duncan and Duncan published A Methodological Analysis of Segregation Indexes, which first proposed indicators of dissimilarity and segregation, providing an important basis for the quantitative analysis of RS [10].

After the 1950s, with the acceleration of globalization, marketization, and urbanization, immigration and racial RS became more common, as large numbers of immigrants entered countries with a relatively homogeneous ethnic composition. The degree of RS tends to vary across countries over time. For instance, the Kerner Commission pointed out in 1968 that America was “moving toward two societies, one black, one white—separate and unequal” [11] (p. 1). Even though the United States (US) government introduced the Fair Housing Act in 1968, which aimed to end segregation, RS remains a common phenomenon in American society [12,13]. Although the level of segregation between Black and White Americans has decreased steadily, RS in cities with large Black populations (e.g., Chicago, Cleveland, Detroit, Milwaukee, Newark, and Philadelphia [14], which were historically identified as hypersegregated) remains high [15,16]. Simultaneously, Asian and Hispanic populations, which were moderately segregated from White Americans, have become slightly more segregated [16,17]. By contrast, RS has tended to decrease in European countries, such as Poland [18] and the United Kingdom (UK) [19]. Racial RS between foreign and native-born residents in Sweden increased from 1990 to 1997; however, from 1997 to 2012, this trend began to exhibit regional differences between metropolitan areas, some of which (e.g., Stockholm metropolitan area) saw no change in RS levels, while others (e.g., Gothenburg metropolitan area) saw a decrease [20]. In developing countries, the desire for elite lifestyles and high social status among economically affluent groups, as well as the need for high-quality infrastructure, services, and facilities, has driven the development of several gated communities [21,22], which have contributed to the increase in RS between different economic classes.

In terms of spatial differences among countries, European countries generally have milder levels of RS between minorities and mainstream groups, compared with the US [23]. For example, the level of segregation of non-European immigrants in Belgium, Denmark, the Netherlands, and Sweden is weaker than that of Blacks and Whites in the US [24]. Black groups’ RS is lower in the UK relative to the US, but the opposite is true for some Asian communities [25]. In Europe, there are also differences in RS, as “new” immigration countries (e.g., Italy and Spain) generally have higher levels of RS, compared with “old” immigration countries (e.g., France) [26]. Compared with North American and Western European countries, RS in China is dominated by sociospatial differentiation and socioeconomic segregation [27], while RS levels vary between cities, with Guangzhou exhibiting the highest RS, followed by Beijing and Shanghai—the three top-tier cities in China [28]. In India, RS is determined primarily by religion and caste system [29,30], as evidenced by the proliferation of Muslim and Hindu residential areas in major cities. In South Africa, RS originated from the colonial government and the apartheid policy [31].

RS can be divided into two types: voluntary and involuntary [32]. Voluntary RS refers to a group of people who live together consciously based on their common characteristics—e.g., socioeconomic status, education level, and race or ethnicity. The purpose of voluntary RS is to increase residents’ satisfaction. Conversely, involuntary RS occurs when a lack of funds or information drives people to live in disadvantaged areas [33]. Additionally, RS can also be divided into RS dominated by ethnocultural segregation [12], RS influenced by income levels and welfare systems [34], RS based on technology [35], and RS dominated by urban migration [36].

RS research has a long history, with various studies being conducted each year and the topics discussed changing and deepening constantly. As shown in the following literature review, many articles have reviewed RS, including a small number that have performed quantitative analyses of studies in specific subject areas of RS, however, due to the sheer amount of research on the subject, prior literature reviews are often limited and subjective. Both domestic and foreign scholars have used various bibliometric theories and methods to explore the trends of disciplinary development, interinfluential relationships between disciplines, and research frontiers and hotspots in disciplinary fields [37]. Among them, CiteSpace software is one of the most important tools for mapping knowledge. Therefore, for the present review, we used Citespace 6.1.R3 software to analyze the literature on RS found in the Scopus database, which contains relevant articles dating back to 1928 until the present year (i.e., 2022). By conducting a review of the RS literature while focusing on the three following questions, this review attempts to enrich this field of study and elucidate new directions for future research.

• What are the changes in the subject basis of RS research and who are the major contributors (e.g., journals, authors, institutions, and countries) to RS research?
• What are the main research topics and hotspots in RS research?
• What are the research frontiers and emerging trends in RS research?

This review is structured as follows: First, a literature review of RS research is presented to prepare the analyzed studies for bibliometric analysis (Section 2). Second, we introduce the materials and methodology of the present review (Section 3). Third, we sort and analyze the temporal distribution, field, and major contributors of each publication; subsequently, we discuss representative classic literature and important research themes, followed by the analysis of research hotspots and frontiers (Section 4). Finally, the present review’s conclusions and future research directions are presented (Section 5). We provide a useful reference for follow-up research in gentrification, spatial analysis, school segregation, health disparity, immigrant, and COVID-19, which have become new themes and directions regarding RS.

2. Literature Review

In order to further understand the knowledge context of RS research, this section provides an overview of reviewed literature to summarize the characteristics and gaps of the existing literature study. In prior studies, the type of RS most commonly discussed is racial RS. Scholars have discussed the causes, consequences, and policies of racial RS and have gradually developed a systematic theory of interpretation. Three major theories (i.e., spatial assimilation theory, place stratification theory, and residential preference theory) have been developed from three perspectives: economics, discrimination, and cultural preference.

Spatial assimilation theory explains changes in the level of RS. It posits that new immigrants gradually integrate or assimilate into the host society by seizing education and employment opportunities to improve their socioeconomic status, thereby following a classical invasion–succession process [38]. This theory provides a competitive model of residential change over time and across generations; it suggests that perceptual shifts (e.g., improved racial attitudes), policy interventions (e.g., antidiscrimination legislation), human capital advances (e.g., higher education), and economic opportunities (e.g., diverse local labor markets) contribute to the integration process [39,40]. Spatial assimilation theory has been validated by numerous empirical studies (e.g., [41,42]).

By contrast, place stratification theory provides an alternative theoretical explanation for why certain groups (e.g., Blacks) do not fit the assimilation model. Place stratification theory assumes that discrimination and prejudice by the subject ethnic group are the main forces shaping the residential patterns of foreign minorities [43,44]. Housing discrimination, stereotypes of minority groups, segmented housing markets, host group residential preferences, and prejudice against segregated communities all contribute to racial RS.

Residential preference theory suggests that each ethnic group embodies residential preferences based on culture and similarity. Schelling proposes a model of the effects of small differences in preferences on residential patterns; he argues that ethnic groups’ residential choices can lead to severe RS due to any small preference (or choice) [45]. Community racial composition preferences largely follow the social psychological theories of in-group preferences and out-group avoidance, including avoidance based on racial stereotypes or perceived status differences [46,47,48,49,50]. Thus, RS, or racial stratification is understood as a result of autonomous individual choice.

The concentration of ethnic/racial minorities in poor neighborhoods is common in Western societies [51]. RS has various negative effects on the segregated group, including increased poverty [52], impeded physical and mental health development [53,54], employment difficulties [55,56], low educational attainment [57,58], and high crime rates [59,60]. Whether voluntary or involuntary, living in racially segregated neighborhoods has serious implications for the present and future mobility opportunities of those who are excluded from desirable areas, who may see their job opportunities, educational quality, safety from crime, and quality of social networks affected by RS [32]. However, some scholars argue that RS can have positive effects, such as the preservation of cultural diversity [61] and the hiring of same-ethnicity workers by immigrant businesses, which have helped minorities improve their employment status [62].

Various scholars have reviewed RS studies, identifying RS’s basic concepts, causes [63,64], consequences, characteristics in different countries [65,66], related governance policies [67], and health effects [37,68,69] (

Table 1. List of review studies in the field of RS.

Author	The Period of Papers Reviewed	Data Source	Main Contributions	Research Limitations
Mitchell and Smith (1979) [67]	–1979	-	A review and comments on the content and effects of federal policy on racial segregation of urban community	A traditional content analysis of a specific topic of RS
Streitwieser and Goodman (1983) [64]	1975–1983	-	A review on the theoretical and empirical research on race and residential location in metropolitan areas of the United States	A traditional content analysis of race and residential location in metropolitan areas of the United States
Galster (1988) [63]	1959–1987	-	Put forward a conclusion contrary to previous studies	A traditional content analysis of the single theory of RS
Grimes (1993) [65]	1926–1990	-	This paper reviews the geographical and sociological literature on urban RS in Australia, and challenges some traditional interpretations of the meaning of segregation	A limited body of RS in Australian cities
White and Borrell (2011) [70]	1950–2009	Web of Science, PsycINFO, PubMed	A systematic analysis of the background of building health risks and health differences of racial/ethnic RS	A bibliometric analysis of a specific topic of RS
Kaplan and Woodhouse (2004) [71]	1999–2003	-	A review of causal factors of ethnic segregation	An overview of a specific field of ethnic RS research: causal factors
Kaplan and Woodhouse (2005) [72]	1949–2004	-	A review of measurements, categories and meanings of ethnic segregation	An overview of a specific field of ethnic RS research: measurements, categories and meanings
Kaplan and Douzet (2011) [73]	1925–2011	-	A review of segregation outcomes of ethnic segregation	An overview of a specific field of ethnic RS research: segregation outcomes
Ruiz-Tagle (2013) [66]	1926–2011	-	A repositioning of integration as a progressive aspiration and a critique of naturalist conceptions of integration	The multidimensional method in the article is only a preliminary conceptualization from the external characteristics, which has been demonstrated by experience
Ruiz-Tagle (2016) [74]	1915–2016	-	A review of apartheid: sociological historical considerations, functionalist positivist theory, and postconflict structuralism	A limited body of the theory of RS
Landrine et al. (2017) [68]	–May 2016	PubMed	A systematic review of RS and racial cancer disparities	The literature is limited to 17 studies that focused primarily on breast cancer
Prather et al. (2018) [75]	–2018	-	A review of historical and contemporary evidence and implications for health equity on racism, African American women and their sexual and reproductive health	Limited to a specific subtopic
Yao et al. (2019) [76]	1955–2016	-	A review of the development of spatial segregation measures, particularly focusing on the mathematical formulation of spatial arrangement/relations	Limited to the review of methodology
Alson et al. (2021) [77]	2000–2019	-	An analysis of the quantitative measurement of systematic racism exposure in population reproductive health research	A limited body of the time span of literature
Kephart (2022) [78]	2000–2021	Google Scholar, PubMed, Web of Science	A scientometric analysis of the relationship between racial segregation and green space	A bibliometric analysis of a specific subtopic of RS research
Larrabee Sonderlund et al. (2022) [69]	2001–2021	-	A systematic review of racialized economic segregation and health outcomes	Only 20 articles were included in the review.
Qiu et al. (2022) [37]	1976–2022	Web of Science	A bibliometric analysis on research regarding RS and health.	A bibliometric analysis of a specific subtopic of RS research

). By combing through the available review articles on RS, we found that although a few papers have conducted bibliometric analyses of studies on specific topics of RS, there is a lack of bibliometric analyses and visualization of knowledge maps for the whole field. Without the help of scientometric software, it is difficult to gain a systematic and comprehensive understanding of this classic and long-standing subject area in an objective and visual way.

3. Materials and Methods

3.1. Data Collection

To comprehensively reflect on the knowledge graph in RS research, we chose the Scopus database to retrieve relevant publications. Compared with the Web of Science, the Scopus database includes more journals and has a more balanced distribution of journal sources from different countries. At the same time, the Scopus database provides a longer range of documents, the earliest of which can be traced back to 1928. On 16 September 2022, we searched the Scopus database using the following search terms: article title-abstract-keywords = “RS”; document type = “articles” or “reviews”; subject area = “social sciences” or “arts and humanities”; language = “English”; source type = “journal” or “book”. After excluding irrelevant literature, a total of 2520 relevant works were retrieved, with the earliest text on RS dating back to 1928. Therefore, these 2520 pieces of literature were used as research data for the bibliometric analysis.

3.2. Research Method

CiteSpace is a Java application for analyzing and visualizing cocitation networks [79]; it supports many types of bibliometric studies. CiteSpace.6.1.R3 software was used to visually analyze RS studies’ publication time, country and institutions of origin, authors, representative literature, research hotspots, and research trends. The present scientometric review imported 2520 papers’ data into CiteSpace. We set the parameters in CiteSpace as follows: (1) timespan = 1928–2022; (2) year per slice = 1; (3) node type = author/institution/country/keyword/reference; (4) threshold selection criteria = the top 50 results for each time slice. We set the other parameters by default.

The article titles, abstracts, keywords, and references exported from the Scopus database were used as the term source for CiteSpace analysis to conduct keyword clustering and keyword burst analysis [80]. The centrality of a node is a graph-theoretical property that quantifies the importance of the node’s position in a network. Nodes with high betweenness centrality tend to appear on paths connecting different clusters and can be used to identify and separate clusters [81]. We use betweenness centrality to analyze the country, institution, and authorship of each paper to clarify the distribution of power in RS research. Burst detection algorithms can detect bursts of a single word or can also be applied to time series analyses of multi-word terms or article citations [81]. Therefore, we use a burst detection algorithm and heterogeneous networks for keyword burst analysis to clarify the hotspots of RS research in each period. Cocitation refers to two (or more) articles being simultaneously cited by one or more subsequent papers. The frequency of citations of different projects (e.g., authors, countries, and institutions) can indicate their impact on the research field. Keyword co-occurrence analysis is an effective way to detect the hot topics and evolving research frontiers of particular research fields over time. In this study, keyword co-occurrence analysis was performed to visually represent the research hotspots.

Dual-Map Overlays is an advanced function of CiteSpace that shows the distribution of papers across disciplines, as well as citation trajectories; it also analyzes the application areas and basic knowledge domains of research subjects [82]. The dual-map overlays provide a clear, visual, and easily interpretable representation of the citations of various publication combinations, with the left side of the plot showing the citing journal in which a source article is published and the right side showing the cited journal in which a reference is published [82]. In this review, we analyze the fields of application and basic knowledge of RS literature from 1928 to 2022 through the dual-map overlays.

4. Results

4.1. Papers’ Distribution Characteristics

4.1.1. Temporal Distribution

The number of papers on RS increased gradually from 1928 to 2022 (Figure 1), which shows that scholars have gradually paid more attention to RS. According to the trend of the number of papers over time, we divided the research progress of RS into the following three stages:

Slow development stage (before 1980): The number of papers in this period was relatively small, with only 96 papers. There was only one paper retrieved before the 1960s which was published by Burgess in 1928. During the next 31 years (1929–1960), no papers were retrieved. This shows that research on RS was almost in a blank stage during this period, until it entered a slow-development stage from 1961 to 1980, when RS gradually drew the attention of relevant scholars.

Exploration stage (1981–2004): During this period, the number of articles fluctuated and increased, with 605 papers published. The phenomenon of RS attracted the attention of many scholars during this period. However, the research progress was slow, with an average of 30 articles published yearly.

Rapid development stage (2005–present): Since 2005, the number of papers published has grown at an accelerated rate. The number of publications increased every year except for a slight decrease in 2008. RS research has become an important topic since 2005 and has attracted extensive attention from scholars. Since 2015, more than 100 papers have been published yearly.

4.1.2. Disciplinary Knowledge Base of the Study

We used CiteSpace’s Dual-Map Overlays function to plot the dual-map overlays of journals for the periods 1928–1980, 1981–2004, and 2005–2022 (Figure 2, Figure 3 and Figure 4). Each curve in the figure represents that the information of the cited research field is transferred from right to left. The thicker the curves of the same color, the richer the subject knowledge base for RS research provided by the cited journal. These overlays reveal the evolution of the RS knowledge base by comparing changes in the citation structure of RS journals in each period. The knowledge base of the RS research field from 1928 to 1981 contains three specialties; specifically, the mainstream and basic fields were #7 Psychology, Education, and Social Sciences. From 1981 to 2004, the mainstream and basic fields increased to eight, including #12 Economics, Political and#5 Health, Nursing, Medicine. It was impossible to distinguish between mainstream and basic fields during and before this period. However, from 2005 to 2022, basic fields increased significantly, adding #1 Systems, Computing, Computers; #10 Plant, Ecology, Zoology; #2 Environmental, Toxicology, Nutrition; and so on. Simultaneously, mainstream fields gradually emerged, while #7 Psychology, Education, and Social Sciences became mainstream fields. This shows that the development of RS research is supported by specialized knowledge mainly in psychology, education, and social sciences, while the trend of crossover and integration of multidisciplinary knowledge is also constantly strengthening.

Journals are important for the dissemination of academic knowledge. Bibliometrics usually uses indicators such as impact factor and citation frequency to reflect journals’ role in academic communication [83].

Table 2. Top 10 journals ranked by the number of publications in RS research.

NO.	Journal Names	Quantity	Proportion
1	Urban Studies	139	5.49
2	Urban Geography	64	2.53
3	Demography	60	2.37
4	Social Science Research	50	1.98
5	Housing Studies	47	1.86
6	Urban Affairs Review	45	1.78
7	Social Forces	44	1.74
8	Environment and Planning A	40	1.58
9	Social Science and Medicine	40	1.58
10	Tijdschrift voor Economische en Sociale Geografie	40	1.58

shows the top 10 journals with the largest number of publications in RS research, which accounted for 30.60% of the total. Among them, Urban Studies ranked first among 649 journals, having published 5.40% of all papers, thereby making the most outstanding contribution to RS research. Next are Urban Geography, Demography, and Social Science Research. The top 10 journals are concentrated in the fields of urban, environmental, social sciences, and economic geography, which also corroborates the results of the journal dual-map overlay analysis, reflecting the close connections between themes in the field of RS.

4.1.3. Analysis of Papers’ Countries and Institutions of Origin

Analyzing publications according to their countries and institutions of origin is helpful in understanding the distribution of research forces and the status of research cooperation between countries, so information about the “carriers” of knowledge is obtained. Papers’ countries and institutions of origin comprise an important indicator of their research contributions. We used CiteSpace to analyze the number of RS papers published in different countries; the top 10 countries and institutions are listed in

Table 3. Top 10 countries and institutions ranked by the number of publications in RS research.

NO.	Country	Count	Centrality	NO.	Institutions	Count	Proportion
1	United States	1293	0.41	1	University of Bristol	54	2.51
2	United Kingdom	278	0.18	2	State University of New York Albany	48	2.23
3	Netherlands	133	0.06	3	University of Michigan, Ann Arbor	46	2.14
4	Canada	97	0.09	4	University of California, Los Angeles	43	2.00
5	Sweden	96	0.03	5	Universiteit van Amsterdam	42	1.95
6	Australia	75	0.01	6	Pennsylvania State University	41	1.91
7	South Africa	73	0.06	7	University of Pennsylvania	40	1.86
8	China	68	0.06	8	Michigan State University	38	1.77
9	Germany	67	0.12	9	The Ohio State University	36	1.67
10	Spain	61	0.11	10	Macquarie University	36	1.67

. The US ranks first with 1293 publications, far exceeding other countries, while its betweenness centrality is also the highest, followed by the UK, the Netherlands, and other countries. Therefore, the US has made the greatest contribution to RS research and occupies a central position. This also indicates that the geographical origin of publications on the subject of residential segregation coincides with the greatest intensity of the phenomenon in the US. South Africa and China are the only two developing countries on the list, indicating that the contributions from both countries to RS research are gradually increasing (Figure 5). Regarding publishing institutions, the University of Bristol published 54 papers, accounting for 2.51% of the total, thereby ranking first. Next is the State University of New York Albany and the University of Michigan, Ann Arbor. Among the top 10 institutions by the number of publications, the US hosts seven of them, and the other three are from the UK, the Netherlands, and Australia, which further illustrates the leading position of the US in the field of RS research.

4.1.4. Analysis of the Authors

Table 4. Top 10 most prolific authors in RS research.

NO.	Authors	Count	NO.	Authors	Count
1	Johnston, Ron	49	6	Darden, Joe T.	21
2	Forrest, James	36	7	Tammaru, Tiit	19
3	Poulsen, Michael F.	31	8	Logan, John R.	18
4	Massey, Douglas S.	31	9	Musterd, Sako	18
5	Iceland, John D.	23	10	Clark, William A.V.	18

lists the top 10 authors who conducted RS research, measured by number of publications. Ron Johnston from the University of Bristol in the UK ranked first, with 49 articles, followed by James Forrest (36 articles) and Michael F. Poulsen (31 articles), both from Macquarie University in Australia. Douglas S. Massey from Princeton University has also published 31 articles. The fifth is John D. Iceland, from Pennsylvania State University (23 articles) in the US. The collaboration between authors is further demonstrated in Figure 6, which shows 3482 nodes and 4323 connections, with a network density of 0.0007. Although the network density shows that there is not much collaboration among the core scholars, some authors cooperated closely, such as Johnston, Poulsen, and Forrest, who collaborated in 27 papers.

4.2. Analysis of Highly Cited Literature

The number of citations is an important indicator of a document’s academic value, as it reflects how widely its findings are recognized by the academic community. However, it only provides a partial account of how much attention a given study attracts. To reduce the impact of time and more objectively reflect how many people read a given study, the present review measured readers’ attention based on the average annual citations (

Table 5. Top 10 most highly cited references in RS research.

NO.	Frequency	Title	Author	Year	Annual Average Cited
1	2020	The dimensions of residential segregation	Massey, Douglas S., Denton, and Nancy A.	1988	59
2	638	Immigrant enclaves and ethnic communities in New York and Los Angeles	Logan, John R., Zhang, Wenquan, and Alba, Richard D.	2002	31
3	622	Hypersegregation in U.S. metropolitan areas: Black and Hispanic segregation along five dimensions	Massey, Douglas S., and Denton, Nancy A.	1989	19
4	594	Changes in the segregation of whites from blacks during the 1980s: small steps toward a more integrated society	Farley, Reynolds., and Frey, William H.	1994	21
5	485	Racial segregation and the American foreclosure crisis	Rugh, Jacob S., and Massey, Douglas S.	2010	40
6	481	Trends in the RS of Blacks, Hispanics, and Asians: 1970–1980	Massey, Douglas S., and Denton, Nancy A.	1987	13
7	435	Attitudes on residential integration: perceived status differences, mere in-group preference, or racial prejudice?	Bobo, Lawrence D., and Zubrinsky, Camille L.	1996	16
8	423	American apartheid: segregation and the making of the underclass	Massey, Douglas S., and Denton, Nancy A.	1993	13
9	422	The spatial mismatch hypothesis: what has the evidence shown?	Holzer, Harry J.	1991	14
10	420	Measures of spatial segregation	Reardon, Sean F., and O’Sullivan, David	2004	23

). The results show that four high-interest studies were jointly published by Massey and Denton, who made great contributions to RS research. It is noteworthy that the top 10 papers all focus on RS between Blacks and Whites in the US.

The first article was from Massey and Denton. They defined the five dimensions of RS (evenness, exposure, concentration, centralization, and clustering), which provide a theoretical basis for measuring RS comprehensively [5]. Logan et al. studied the characteristics of the ethnic immigrant enclaves in urban and suburban areas in the US; they pointed out that the causes of these racial/ethnic enclaves are mainly driven by preference, rather than by economic or cultural constraints [84]. A third article proposes the dissimilarity index and the P*exposure index to measure RS and proposes the concept of “hypersegregation”, highlighting that Black Americans experienced extreme segregation in all aspects and occupied a unique and distinct disadvantage in the urban environment [12]. A fourth article pointed out that although the level of RS in the US has decreased, institutionalized discrimination that deprives Blacks of equal treatment in the housing market will continue in the old metropolitan areas of the industrialized North and some parts of the old South [85]. Rugh et al. posited that RS was an important cause of the foreclosure crisis, and that the higher the degree of segregation of Hispanics, especially Blacks, in a metropolitan area, the higher the number and rate of foreclosures [86].

The sixth-ranked study is also from Massey and Denton, who examined changes and differences in RS between minority groups and Whites in the US [87]. Bobo and Zubrinsky analyzed US residents’ views on racial residential integration and found that African Americans expressed more support for integrated neighborhoods compared to members of other racial/ethnic groups, suggesting that RS in African Americans is not self-imposed [88]. Massey and Denton argued that there is a strong correlation between rising poverty rates and high levels of RS, often caused by redlining and exclusionary zoning, while the upwards mobility of children who grow up in urban ghettos is reduced [89]. Holzer reviewed the spatial mismatch hypothesis and noted that due to the suburbanization of jobs and racial discrimination in the housing market, a spatial mismatch between population size and employment in metropolitan areas remained in the US, and that there were more Black residents than suitable jobs in inner urban areas, which resulted in high unemployment rates, low wages, and longer commutes for Blacks in inner-city areas [90]. Reardon and Sullivan reviewed the prevalent calculation method for RS and introduced a general method to measure RS, which included the spatial information theory index, spatial relative diversity index, and spatial dissimilarity index [1].

4.3. Research Hotspots and Frontiers

4.3.1. Analysis of Keyword Clusters

Keyword co-occurrence analysis is used to extract high-frequency words from cited documents and to identify the research hotspots in a given field. To understand the hotspots of RS research more intuitively, this review used CiteSpace software to analyze the keyword clusters, 21 of which were identified (Figure 7). The clusters included the number of network nodes (N = 924) and network connections (E = 4628). Among them, the modularity Q score was 0.5703, and the mean silhouette score was 0.8658, indicating that the clustering structure was significant and reasonable. The top 10 keyword clusters in RS since 1928 were as follows:

#0 Racial difference: The keyword that has received the most attention is racial RS, which focuses on the study of interracial differences, such as differences in living environment, cultural preferences, health outcomes, education levels, income levels, etc. Racial RS has also been identified as a root cause of health differences between racial groups [91]. For example, among non-Hispanic Blacks in the US, higher racial segregation was positively associated with higher body mass index and greater odds of being overweight, but there was no significant association between segregation index and weight status among Whites in the US [92].

#1 Sociospatial behavior: Sociospatial behavior is the main research object of RS; researchers typically focus on the residential space assimilation and spatial differentiation processes. For example, as their income levels increase, Asians gradually integrate into communities with a larger proportion of Whites, thereby leading to accelerated racial fragmentation [93]. By reducing the degree of ethnic RS and accelerating occupational mobility and economic assimilation, RS will automatically transform into spatial integration [40]. In terms of residential differentiation, RS is closely related to residential preference. Each ethnic group has preferences based on culture and similarity. Ethnic groups’ residential choice may lead to significant RS due to any small preference [94].

#2 Income inequality: Income inequality is the main cause of RS. Income segregation increases gradually over time [95]. Inequalities in the distribution of individual and household income are considered the most critical catalyst for RS between socioeconomic groups [96]. Changes in the level of RS between socioeconomic groups are positively associated with changes in the degree of income inequality [97]. Poor households are more segregated from others compared to households in other income groups [98].

#3 Mixed income communities: There is considerable interest in mixed-income housing and communities among academics and policymakers [99,100]. Mixed-income communities tend to demonstrate positive trends in terms of poverty alleviation, desegregation, and sustainable community development [101]. However, scholars are more concerned with evaluating and questioning the effects of mixed residential policies. For instance, the decentralized distribution of assisted housing in metropolitan areas in the United States has led to moderate economic residential integration among very-low-income households, while high-income households have become more segregated, which may offset the gains in economic integration [102]. Due to the stigma and prejudice against low-income residents, mixed-income communities often only achieve residential space integration but cannot increase social interaction. Social isolation therefore persists, which affects the well-being of residents to some extent [100].

#4 Guest worker minorities: Guest workers are the main group studied in RS. Immigrants tend to have longer commutes than natives, which is a symptom of RS [103]. Living in or near poor neighborhoods reduces the probability of employment [55]. Long-term residence in a community with high poverty and unemployment rates affects guest workers’ employment environment, leading them to face longer unemployment time, less accessible jobs, and a higher probability of long-term unemployment [104].

#5 Typical district segregation: Typical district segregation often refers to segregated school districts. School segregation is largely the result of RS patterns. Individuals’ place of residence determines the schools where they can send their children, especially in the US and UK. In areas of the US where Whites and Blacks are more unevenly distributed in schools, Blacks are more likely to be enrolled in local charter schools than in districts where schools are integrated [105]. Education and RS are intertwined in complex ways, with schools reflecting racial segregation in the community and parental considerations and choices about schools reinforcing RS. In European countries such as the Netherlands, France, and the UK, school segregation is generally more severe than RS [106,107].

#6 Occupational segregation: Occupational segregation is a type of RS. Cities’ development requires the cooperation of high-tech talents and low-end labor, but the income differentiation between the two in the labor market may lead to RS. This technology-based RS results in limited access to job information and networking, as well as low upward mobility for the low-end workforce. Additionally, high-skill workers need the services of low-skill workers in their jobs. Therefore, the isolation of low-skill workers may also reduce high-skill workers’ productivity [108]. In Australia, high-skill populations in many cities are becoming increasingly concentrated in the inner city, while people with low levels of education and low-status occupations are increasingly concentrated in cities’ outskirts [109].

#7 Health inequalities: Health inequalities is a hot topic in RS in recent years. Studies in this cluster mainly discuss the health impact of RS as it is reflected through different social groups’ (e.g., African American women, the elderly, children, etc.) environmental exposure. RS is seen as the root cause of racial inequality in health [110]. Racial and ethnic minorities or socioeconomically disadvantaged groups face greater environmental hazards [111]. Increased segregation between Black and White groups in the US is associated with increased inequality in air pollution exposure [112]. Therefore, it is necessary to consider environmental factors in public policy to mitigate the social costs of RS [113].

#8 Metropolitan ghetto: A ghetto is “a residential district which is almost exclusively the preserve of one ethnic or cultural group” [114] (p. 258). The ghetto in large cities has become a major spatial component of RS research worldwide. For example, scholars have examined the characteristics and variations of the African American ghetto model in the US [115,116], Asian ghettos in British cities [117], Jewish ghettos in European cities [118], and Muslim ghettos in Indian cities [29].

#9 Migrant–native differential mobility: The mobility difference between migrants and native residents has attracted many scholars’ attention. Both regional demographic changes [119] and migration affect the level of RS. Migration and immigration have important effects on spatial integration and segregation [120]. Natives (or Whites in the US) tend to leave and avoid moving into communities with increasing numbers of immigrants (Blacks) [121]. Large cities are often destinations for foreign-born migrants [122]. In China, the RS of peasant migrants is mainly due to China’s unique institutional background (e.g., its land use rights and household registration systems) and exclusive housing regulations [123]. The size of migrant populations tends to reinforce perceptions of social exclusion, thereby creating challenges for social integration [124].

4.3.2. Analysis of Emerging Trends

The emergence of a research frontier inevitably leads to the outbreak of its keywords in a short period. In this review, through the analysis of burst keywords, the outbreak of keywords within the field of RS research was revealed (Figure 8). “Health disparity”, “neighborhood”, “immigrant”, “spatial analysis”, “gentrification”, “school segregation”, and “racism” are the burst terms that have persisted to the present. Based on high-frequency keywords and burst terms in each period (excluding keywords and burst terms that are less related to RS), we divided RS research into three stages (

Table 6. High-frequency keywords in each stage.

Stage	Years (Number of Publications)		Keywords (Frequency)
1	1928–1980 (96)		Demography (10), urban population (8), US (8), human (7), residence characteristics (7), socioeconomics (5), ethnic group (4), Black (4), western hemisphere (4), social class (4), developed country (4)
2	1981–2004 (605)		RS (118), social segregation (88), US (66), racial segregation (65), residential location (65), ethnicity (42), metropolitan area (37), South Africa (37), race (36), urban area (32), ethnic minority (31), immigrant population (28), residential mobility (22), housing market (21), developing country (19)
3	2005–2022 (1819)	2005–2017 (1109)	Social segregation (282), RS (220), US (230), metropolitan area (108), ethnic group (96), African American (81), immigrant (80), urban area (76), residence characteristics (72), residential mobility (70), adult (57), United Kingdom (54), urban housing (54), poverty (51), census (49), spatial analysis (44), racism (44), socioeconomic status (44), Europe (45)
		2018–2022 (710)	Social segregation (163), RS (145), US (100), spatial analysis (49), racism (41), residential location (45), residential mobility (39), China (31), gentrification (26), racial disparity (25), health disparity (23), urban planning (23), inequality (19), school segregation (17), COVID-19 (15), accessibility (15)

), which is consistent with the aforementioned stage division of RS research.

During the first stage of RS research (1928–1980), few studies on RS were conducted. Although at this time keywords had not yet broken through, the foundation for RS research was established. During this period, the keyword “Black” began to burst in 1970, with an intensity of 6.68. The burst duration lasted until 1997. This indicated that Black population segregation was the main topic.

The second stage comprises the exploratory period of RS research (1981–2004). The burst terms in this stage included “developing country”, “residential location”, “urban housing”, “South Africa”, and “US”. The high-frequency keywords included “immigrant population”, “residential mobility”, and “housing market”. During this period, research topics tended to be diversified and the number of documents increased. Most studies on RS were related to racial and class segregation, mainly focusing on the intensification of racial conflicts and the widening gap between the rich and the poor, while others began to study immigration. The geographical scope of the research was also expanded from centering mostly on the US to covering other regions, such as South Africa and other developing countries. The research scale also focused on the city and community levels.

The third stage comprises the rapid development period of RS research (from 2005 to the present). During this period, the number of studies increased rapidly; however, the high-frequency keywords are less common, compared with previous years, indicating that the RS field is gradually becoming mature. The research direction at this stage is not limited to racial and class segregation; instead, studies examine many aspects of segregation, including income segregation, gender-based spatial segregation, school segregation, opportunity differences, housing issues, etc. Therefore, there appear burst keywords such as “urban housing”, “regression analysis”, “ethnic group”, “gated communities”, and “social economic status”. The number of studies in European countries has gradually increased, while developing countries such as China are gradually becoming new case study sites.

In the past five years, “gentrification (burst in 2018 with an intensity of 8.68)”, “spatial analysis, (burst in 2016 with an intensity of 11.4)”, “school segregation, (burst in 2019 with an intensity of 7.7)” “health disparity (burst in 2013 with an intensity of 9.76)”, and “immigrant (burst in 2015 with an intensity of 8.83)” have appeared, which evinces the emergence of new research themes and directions. COVID-19 has not yet become a burst term, but it is an emerging high-frequency keyword. Especially since the outbreak of COVID-19, RS and public health issues have received extensive attention [125]. In the context of RS, the study of health inequalities caused by COVID-19 is likely to become a new research direction in the future.

5. Conclusions and Discussion

Several studies have been conducted on RS’s causes, effects, and political implications, thereby leading to the development of a relatively systematic theoretical framework. Scholars from different disciplines and schools have contributed to RS research. Through the analysis of the Scopus database and visualization using CiteSpace software, this study conducted a systematic bibliometric analysis to reach a comprehensive understanding of the development and limits of RS studies. The major findings are as follows.

First, RS has received widespread scholarly attention in recent decades, as evidenced by the increasing number of related publications. The development of RS studies has exhibited three major stages, including the slow development (before 1960), exploration 1961–2004, and fast development (after 2005) periods. Mainstream RS knowledge has developed primarily from three fields: psychology, education, and social sciences. Nonetheless, over time, the knowledge base of RS research expanded considerably, as the trend of multidisciplinary integration remains strong. To strengthen the innovation of measurement methods for RS research, it is necessary to strengthen its intersection and integration with the fields of systems, computing, and computers.

Second, the analyses of studies’ countries and institutions of origin, as well as the cooperation networks among authors, reveal that RS research emerged in the US. The US and its institutions dominate the field of RS research, exhibiting the most abundant research achievements. The US’s history of RS and immigration, as well as its political context, have made it the center of RS studies [126]. Specifically, leading scholars such as Johnston, Massey, ‪Forrest, Poulsen, and Iceland, have made great contributions to RS research, helping establish a solid disciplinary base for its development.

Third, hotspots of RS research include racial difference, sociospatial behavior, income inequality, mixed income communities, guest worker minorities, typical district segregation, occupational segregation, health inequalities, metropolitan ghetto, and migrant-native differential mobility. Research directions have become more diversified and multi-scalar, moving beyond racial segregation and focusing on income segregation, gender-based spatial segregation, school segregation, opportunity differences, housing issues, and other dimensions of segregation.

Fourth, gentrification, spatial analysis, school segregation, health disparity, immigrant, and COVID-19 have become new themes and directions of RS research. Methodologically speaking, many new quantitative research methods have been implemented in RS research, such as the Theil index [127], the social interaction potential [128], and the shortest path isolation index [129].

To acquire a comprehensive understanding of RS, we could enrich the literature by examining the following directives in future research. Firstly, the existing literature has primarily focused on RS on the micro-scale community level, while the effects of urban multiscalar dynamics remain relatively unexplored. Thus, future research should focus on the impact of multiscalar spatial factors on RS.

Secondly, the basic knowledge of RS research mainly originates from three disciplines (psychology, education, and social sciences). Given the multiple and complex characteristics and mechanisms behind RS, however, we suggest that it is necessary to integrate multidisciplinary theories into RS research and conduct a deep and systematic analysis of RS’s mechanisms in specific local contexts. Empirical exploration based on a multidisciplinary perspective could offer valuable theoretical contributions to RS research.

Thirdly, most studies have described spatial RS through qualitative methods, while scarce research has analyzed the implications of residents’ agency in RS. Therefore, we could further analyze the mechanisms behind RS by exploring residents’ agency in shaping RS.

Fourthly, a systematic methodology has not yet been established to measure RS, as the existing research has mainly analyzed RS from a linear perspective, focusing on a single individual or group. Although some scholars have recently proposed new methods and indicators to measure RS, a multiscalar and systematic framework and its accompanying methods have not yet been developed. Therefore, it is imperative to seek further methodological developments in RS research.

Also, there are some limitations of this review by using CiteSpace. (1) The indicators analyzed by CiteSpace are relatively general and the content depth needs to be further strengthened. (2) The database of the literature retrieval is single, and it is necessary to strengthen the synthesis of data in multiple databases (e.g., web of science, CNKI, etc.).