Latent Class Analysis of Discrimination and Social Capital in Korean Public Rental Housing Communities

Abstract

This study explored typologies among residents of South Korean public rental housing, focusing on their experiences of discrimination and social capital. Latent class analysis (LCA) was applied to data from 4683 individuals in the 2021 Seoul Public Rental Housing Panel Survey. Four distinct groups were identified: ‘Group Seeking Friendly Neighbor Relationships’, ‘Group Accepting Losses’, ‘Group with High Social Capital’, and ‘Group Indifferent to Neighbors’. The findings revealed that while discrimination was widespread, certain groups exhibited strong social capital. Notably, the ‘Group Accepting Losses’ showed the highest willingness to help neighbors despite facing significant discrimination, while the ‘Group with High Social Capital’ displayed high levels of neighbor trust and mutual support. These results challenge traditional views by showing that social capital can thrive even in the presence of discrimination. This study suggests that policies aimed at addressing discrimination in public rental housing should focus not only on physical integration but also on fostering social connections to enhance community cohesion and reduce mental health issues among residents.

1. Introduction

This study sought to examine how residential satisfaction mediates the relationship between social capital and attitudes toward vulnerable groups, while investigating the moderating effects of vulnerability status, such as educational and employment vulnerabilities. By integrating social capital and residential satisfaction, this study aimed to fill gaps in the literature by providing insights into how these factors interact, particularly for vulnerable populations. This research will also contribute to ongoing discussions in housing policy and social capital development by offering practical policy recommendations to enhance community cohesion and address the needs of marginalized groups.

Since the 1960s, the Republic of Korea has undergone rapid urbanization, industrialization, and modernization, leading to a swift concentration of the population in major cities. This urban shift resulted in significant social issues such as housing shortages, rapid increases in housing prices, and the growth of the low-income population [1]. In response, the South Korean government implemented large-scale housing supply policies, separating home sales and rentals and constructing the first public rental housing in 1962 to support low-income individuals unable to meet minimum housing standards [2]. During this policy implementation, negative perceptions of rental housing began to escalate [3]. Public rental housing came to be perceived negatively, associated with slum-like conditions, social exclusion, and stigma, leading to social discrimination and conflicts among low-income groups residing in these areas. Consequently, persistent issues arose, including opposition from nearby residents, depreciation of public rental housing values, discrimination against residents, deterioration of residential environments, and a decline in local image [4]. Despite various government initiatives aimed at improving the perception of public rental housing and fostering social integration, these efforts have not fully resolved the underlying social issues. Discrimination remains prevalent, with approximately 40% of households in public rental housing reporting experiences of discrimination, particularly households with children and newlyweds [5]. The persistence of discrimination is often attributed to the failure to foster meaningful social relationships, such as trust and confidence, beyond physical integration [6].

Numerous studies [7,8,9] have highlighted the importance of social capital in explaining social relations, including issues such as discrimination, egoism, and community conflicts. Discrimination is widely recognized as a significant factor that diminishes social capital, and both domestic and international research has shown that public rental housing residents experience social exclusion and discrimination, which exacerbates social issues [10,11,12,13,14,15]. These findings underscore the importance of global policy efforts to address discrimination and promote social inclusion.

Existing literature indicates that marginalized groups with lower social capital tend to experience higher levels of discrimination [16,17], and individuals who perceive discrimination generally report lower levels of social capital compared to those who do not [16]. Moreover, inequalities within communities can weaken their ability to connect with external sources of capital, further entrenching social disadvantages [18,19].

The existing literature has predominantly employed variable-oriented approaches, focusing on how independent variables like discrimination experiences affect dependent variables such as social capital. However, these studies often overlook the diversity within populations facing discrimination. There is a critical gap in research that explores the typologies of discrimination experiences and social capital levels, particularly in public rental housing. While some studies have examined social capital and discrimination independently, the need for a person-oriented perspective that identifies distinct typologies within populations has been largely neglected. This gap highlights the need for research that accounts for the qualitative differences among residents, acknowledging that not all residents experience the same levels of discrimination or social capital.

To address this gap, this study employed latent class analysis (LCA) to classify discrimination experiences and social capital levels among public rental housing residents. By focusing on the typologies of discrimination experiences and social capital, this study provides a more nuanced understanding of how these factors intersect among different groups within the population. The findings also shed light on the importance of fostering social connections beyond physical integration, a key challenge identified in previous research. Furthermore, this study aimed to offer policy recommendations to enhance community cohesion and reduce the social exclusion of marginalized residents. Therefore, this study builds on previous research by examining how discrimination experiences impact social capital and housing stability among low-income residents of public rental housing.

While most studies have taken a variable-oriented approach, focusing on how independent variables like discrimination experiences affect dependent variables such as social capital, a person-oriented perspective highlights qualitative differences within the same group. This approach acknowledges that residents of public rental housing do not necessarily experience the same levels of discrimination or face identical risk factors. Individuals with higher discrimination experiences may share feelings of solidarity with those who have faced similar discrimination or belong to similar socioeconomic backgrounds. Rather than isolating themselves from interactions or disconnecting from their external environment, they may strive to assist neighbors and build communities [19].

Considering these exceptional cases, it becomes clear that assuming those who experience discrimination more frequently have lower social capital may overlook the diverse characteristics of public rental housing residents. Therefore, the primary objective of this study was to identify the types of discrimination experiences and social capital levels among public rental housing residents using latent class analysis (LCA). Additionally, this study sought to understand the housing characteristics, perceptions of social mixing, resident activities, mental health, and other attributes associated with each type to derive policy implications for addressing discrimination and conflict.

The specific research questions for this study are as follows: Research Question 1: How do latent classes manifest when applying latent class analysis to classify discrimination experiences and levels of social capital among public rental housing residents? Research Question 2: What characteristics are associated with each latent class?

2. Review of Literature

2.1. Relationship Between Discrimination Experiences of Public Rental Housing Residents and Social Capital

Despite recent improvements in public perceptions of rental housing, discrimination remains prevalent, with severe cases reported even in the media, sparking scholarly interest. Interviews with residents of public rental housing reveal biases and stigmatization not only from homeowners but also internally among residents themselves [9], highlighting the widespread nature of discrimination.

Discrimination experiences are closely linked to social capital. Studies have found that while human rights education alone did not reduce experiences of discrimination among children and adolescents, social capital had a significant mitigating effect [19]. Specifically, the presence of trustworthy social networks, such as individuals available for emotional or practical support, significantly reduced both perpetration and victimization of discrimination. Similarly, research on the elderly population confirmed that higher levels of social capital correlate with fewer discrimination experiences [20]. These findings suggest a generally negative relationship between discrimination experiences and social capital, where stronger social networks can reduce instances of discrimination.

This study adds depth to this understanding by differentiating between the types of social capital involved. Bonding social capital—the close ties and relationships formed within families and local communities—can provide emotional support and resilience to those facing discrimination. In public rental housing, bonding social capital is particularly critical for forming solidarity and trust among residents facing common challenges. The importance of these localized networks is reinforced by findings from international studies, which emphasize that strong intra-community relationships contribute to resilience in the face of systemic inequality [21].

Bridging social capital, which extends beyond immediate social circles and connects individuals to broader community networks, can help public housing residents access resources and opportunities outside their immediate community. As a result, individuals with strong bridging social capital are better able to counter the negative effects of discrimination by connecting to external support systems. For example, placemaking initiatives in disadvantaged neighborhoods have demonstrated how bridging social capital can transform underutilized public spaces into hubs of economic and social activity, thereby reducing stigma and fostering inclusion [22].

Lastly, linking social capital, which connects individuals to institutional resources such as government services, education, or employment opportunities, can serve as a powerful tool for overcoming structural disadvantages and addressing systemic discrimination. This dimension of social capital is essential for empowering public rental housing residents to access resources that might otherwise be unavailable to them, facilitating long-term social and economic mobility. Case studies in New Jersey demonstrated how integrating funding programs and partnerships between local governments and nonprofit organizations effectively built linking social capital to revitalize low-income communities [23].

Table 1. Key Factors Influencing Discrimination Experiences and Social Capital.

Factor	Description	Key References
Housing Characteristics	Location, ownership status, and duration of residence impacting social capital and discrimination	[22,24]
Perception of Social Mixing	Community attitudes toward diverse residents contributing to social capital and discrimination outcomes	[9,25]
Residential Activities	Participation in community activities enhancing social relationships and mitigating discrimination stress	[26,27]
Mental Health	Psychological well-being affected by discrimination experiences and social capital	[28,29,30]

integrates various factors—housing characteristics, perceptions of social mixing, residential activities, and mental health—that collectively influence discrimination experiences and social capital. This holistic framework allows for a deeper examination of the complex dynamics within public rental housing communities and guides policy recommendations.

This study advances existing research by offering a more detailed exploration of how distinct components of social capital, such as trust and community support, directly affect discrimination experiences. By delving into these specific aspects, this study provides practical insights that can inform targeted social policies to reduce discrimination and strengthen community ties, contributing to more inclusive public housing environments.

2.2. Person-Centered Approach

This study employs a person-centered approach, specifically latent class analysis (LCA), to classify the discrimination and social capital experiences of public rental housing residents. The person-centered approach, which distinguishes latent subgroups based on individual response patterns, is conducive to discerning qualitative differences within the same group [25]. Unlike traditional variable-based methods that assume homogeneity across the population, LCA allows for the identification of distinct subgroups based on their unique experiences and responses, thus providing a more nuanced understanding of the relationship between discrimination and social capital.

In understanding the relationship between discrimination experiences and social capital among public rental housing residents, this approach offers a differentiated perspective from previous studies that have often treated the population as homogeneous. According to this approach, the degree of discrimination experiences can vary among individuals, and attitudes toward discrimination experiences can influence individual behaviors. For instance, experiencing discrimination does not necessarily lead to a decrease in social capital for all citizens, nor does a decrease in social capital always result in discrimination experiences. This variability highlights the need for a more personalized approach to analyzing discrimination and social capital, as not all individuals respond in the same way to these factors.

The existence of social capital reducing experiences of discrimination victimization, as indicated in previous research [19], supports this notion. Conversely, a study targeting public rental housing residents [4] found higher levels of discrimination experiences among groups with high social capital. Therefore, discrimination experiences and social capital do not necessarily diminish each other. In fact, this finding suggests that higher social capital may, in some cases, coexist with increased discrimination or even contribute to resilience in facing social exclusion.

Considering previous research, analyzing the relationship between discrimination experiences and social capital through a person-centered approach is necessary. This approach provides a deeper understanding of the heterogeneity within the population, highlighting the importance of individual variation and the need to tailor interventions to address the distinct needs of different subgroups. Evidence from systems-based analyses in Australian urban communities further supports this approach, emphasizing the need for interventions tailored to local contexts to address multifaceted challenges effectively [23].

Therefore, this study aimed to conduct research that typified the relationship between discrimination experiences and social capital, providing a more comprehensive understanding of how these factors interact across different groups.

This study’s use of the person-centered approach highlights the diversity of resident experiences, addressing gaps in prior literature that often assume uniform effects across populations. By identifying subgroup-specific patterns, this research challenges traditional linear models and emphasizes the need for tailored interventions that account for the varied needs of residents. This perspective provides a valuable foundation for designing effective policies that address the distinct characteristics of different subgroups within the public rental housing population.

2.3. Factors Related to Discrimination Experiences and Social Capital of Public Rental Housing Residents

This study not only aimed to typify the relationship between discrimination experiences and social capital but also to explore the key factors influencing this relationship. To this end, additional variables such as housing characteristics, perception of social mixing, residential activities, and mental health were examined.

Firstly, housing characteristics have been reported to be closely related to social capital. Various housing characteristics (e.g., location of public rental apartments, homeownership status, and duration of residence) emerged as significant influencing factors on social capital [26]. Additionally, the residential environment positively influenced neighborly relationships among the elderly [27]. Furthermore, housing characteristics were validated to be associated with discrimination experiences [28]. Particularly, discrimination and neighbor conflicts among public rental housing residents influenced the selection of different types of housing. Thus, housing characteristics can be seen as related to discrimination experiences and social capital.

Next, perception of social mixing is significantly related to discrimination experiences and social capital. Social perceptions of public rental housing residents, especially prejudices and social stigmatization, lead to discrimination experiences and outcomes [9]. Public rental housing residents, due to policy characteristics, consist of various types of individuals, contributing to social mixing. However, sudden social mixing may exacerbate discrimination experiences. Researchers studying the relationship between social mixing and social capital argue that higher levels of social mixing correspond to higher levels of social capital [29]. Appropriate policies promoting social mixing can positively contribute to social capital formation. For instance, placemaking efforts in Geelong’s disadvantaged neighborhoods during the COVID-19 pandemic demonstrated how community-driven interventions could reduce locational disadvantages and foster social inclusion [22].

Residential activities are also associated with discrimination experiences and social capital. Research suggests that community facility utilization among public rental housing residents enhances residential satisfaction, fosters social relationships among neighbors, and alleviates stress [30]. Studies indicate that discrimination experiences indirectly influence social participation, such as residential activities [31]. Therefore, residential activities are related to discrimination experiences and social capital.

Finally, mental health plays a significant role in shaping the relationship between discrimination experiences and social capital. Negative mental health outcomes, such as increased stress and depression, have been linked to experiences of discrimination, particularly among vulnerable groups like the elderly and children [32,33]. Conversely, strong social capital has been shown to have positive effects on mental health, acting as a buffer against the adverse impacts of discrimination [34,35]. For public rental housing residents, factors such as health status, employment, and the residential environment can further influence mental health outcomes, highlighting the complex interplay among these variables [36,37,38].

Table 1 presents a detailed synthesis of the key factors influencing both discrimination experiences and social capital. It highlights the interconnected nature of housing characteristics, social mixing perceptions, residential activities, and mental health. This integrated framework offers a holistic perspective for analyzing the dynamics of public rental housing communities, providing a valuable foundation for developing targeted policies that aim to enhance resident well-being and foster social cohesion. Incorporating insights from international case studies, such as those in New Jersey and Geelong, this analysis underscores the critical role of local and participatory approaches in addressing these challenges [21,22,23,24].

By moving beyond traditional approaches that analyze variables in isolation, this study examined the interrelationships among key factors, providing a comprehensive framework that captures the complex social dynamics within public rental housing communities. This integrated analysis offers a nuanced understanding of how discrimination experiences and social capital interact, laying a robust foundation for evidence-based policymaking aimed at improving community relations and fostering social inclusion. In conclusion, this chapter deepened the engagement with the existing literature by incorporating diverse influencing factors and employing a person-centered approach. This innovative perspective paved the way for the subsequent empirical analysis and set a clear direction for future research efforts, ultimately aiming to enhance social cohesion and improve resident well-being.

3. Materials and Methods

3.1. Data and Sample

This study utilized data from the 4th year (2021) of the Seoul Public Rental Housing Residents Panel Survey [39], which provided a comprehensive and representative sample of public rental housing residents in Seoul. The survey included a broad range of items that captured detailed housing characteristics and the residential environment, making it highly relevant for this research. Additionally, the survey encompassed specific items related to discrimination experiences and social capital, which are central to the study’s focus.

The Seoul Public Rental Housing Residents Panel Survey was selected due to its comprehensive and representative nature, offering valuable insights into the housing conditions, social dynamics, and experiences of public rental housing residents in a major metropolitan area. As the capital city of the Republic of Korea, Seoul represents a diverse urban environment with a significant portion of low-income and vulnerable populations living in public rental housing. The data thus provide a robust basis for understanding the experiences of these residents, making it highly relevant to the study’s objectives.

The survey items used to assess discrimination experiences and social capital were selected based on a thorough review of the existing literature and validated instruments from related studies. Items measuring discrimination experiences were designed to capture both external stigmatization from non-residents and internal biases within the community, ensuring a comprehensive understanding of the issue. For social capital, the survey items were chosen to reflect key components such as trust, social networks, and community participation. These items were adapted from established social capital assessment frameworks to maintain reliability and relevance.

To ensure the validity and reliability of the survey instruments, expert consultations were conducted during the selection process. In addition, pilot testing was carried out with a small sample of residents to refine the survey items and confirm their contextual appropriateness for public rental housing residents. The feedback from these expert reviews and pilot tests helped to ensure that the items were clear, relevant, and effectively captured the intended constructs. Reliability testing using Cronbach’s alpha was also performed on key subscales to confirm the internal consistency of the survey items, with all measures yielding satisfactory results.

This careful approach allowed for accurate data collection aligned with the study’s objectives, providing robust input for subsequent latent class analysis (LCA). The final sample comprised 4683 household members surveyed in 2021, offering robust data for reliable statistical analysis.

The sample’s primary strength lies in its comprehensive representation of public rental housing residents in Seoul, allowing for findings that are relevant to this specific population. This sample size and the broad demographic representation enhance the generalizability of the results to other urban public rental housing settings in the Republic of Korea and potentially to other countries with similar housing structures and social dynamics. The large sample size of 4683 participants enhances the robustness of the analysis and supports the reliability of generalizing the findings to similar urban public rental housing settings. Additionally, the inclusion of varied housing characteristics, discrimination experiences, and social capital components ensures a well-rounded understanding of resident experiences.

However, there are limitations to consider when generalizing the results. The data are specific to Seoul and may not fully represent public rental housing residents in other regions with different socioeconomic or cultural contexts. Moreover, while the sample provides rich cross-sectional data, it may not capture temporal changes or causal relationships. These factors should be acknowledged when interpreting the efficacy and broader applicability of the approaches used in this study.

3.2. Variables and Relevant Factors

3.2.1. Experiences of Discrimination and Social Capital

In this study, the variables used for latent class classification are the experiences of discrimination and social capital among public rental housing residents. Experiences of discrimination are measured dichotomously, with respondents indicating whether they have ever felt discriminated against as rental housing residents (“No” coded as 0, “Yes” coded as 1).

Social capital, as referenced in previous studies, encompasses trustworthiness, sense of belonging, and reciprocity. Accordingly, in this study, social capital was measured based on similar constructs. Trustworthiness was assessed by transforming responses to the question “Do you think you can trust the neighbors where you live?” into a dichotomous scale (“Not at all + Not really” coded as 0, “In general, yes + it really is” coded as 1). Sense of belonging was evaluated through responses to the question “How do you get along with your neighbors where you currently live?” with options converted into a dichotomous scale (“You have no idea who lives next door + You know your neighbors, but you don’t say hello” coded as 0, “Say hello to your neighbors or make small talk + I tend to know or communicate with my neighbors about family matters” coded as 1). Reciprocity was measured by two items: “Do you have neighbors you can turn to if you need urgent help?” and “Would you be willing to help someone in your neighborhood if they needed help in an emergency?” Responses are transformed into a dichotomous scale (“No” coded as 0, “Yes” coded as 1) and (“Not at all + Not really” coded as 0, “In general, yes + it really is” coded as 1), respectively. Further details are provided in

Table 2. List of Latent Class Variables.

Variables		Survey Questions
Latent class variable	Experiences of Discrimination	Q1. Have you ever felt discriminated against as a rental housing resident? ⓪ No, ① Yes
	Social Capital	Q1. Do you think you can trust the neighbors where you live? ⓪ Not at all + Not really, ① In general, yes + it really is Q2. How do you get along with your neighbors where you currently live? ⓪ You have no idea who lives next door + You know your neighbors, but you don’t say hello. ① Say hello to your neighbors or make small talk + I tend to know or communicate with my neighbors about family matters. Q3. Do you have neighbors you can turn to if you need urgent help? ⓪ No, ① Yes Q4. Would you be willing to help someone in your neighborhood if they needed help in an emergency? ⓪ Not at all + Not really, ① In general, yes + it really is

.

3.2.2. Housing Characteristics

Housing characteristics encompass housing environment satisfaction, mixed-use complex residency, and severity of conflicts in mixed-use complexes. Housing environment satisfaction was comprised of nine items, including accessibility to public transportation, amenities, public facilities, cultural facilities, medical facilities, recreational spaces, welfare facilities, educational environment, and childcare environment, measured on a 4-point Likert scale. A higher score indicates greater housing environment satisfaction. Mixed-use complex residency was determined by responses to the question “Do you reside in a complex where rental housing and ownership housing are mixed?” coded as “Yes” or “No”. The severity of conflicts in mixed-use complexes was gauged through six items related to parking-related conflicts, cleanliness issues, playgrounds, internal roads, community facilities, using a 4-point Likert scale. A higher score reflects higher conflict severity.

3.2.3. Perception of Social Mixing

Perception of social mixing comprised three items, wherein respondents provided responses on a 4-point Likert scale to two questions: “What are your thoughts on economic integration of different socioeconomic classes for social integration?” and “How do you feel about the establishment of facilities such as special schools, facilities for disabled individuals, and funeral homes in your local community?” Higher scores indicate a stronger perception of social mixing. Additionally, respondents were asked to express their preference regarding the integration of rental and ownership housing in one complex or separate complexes. Response options included: (1) Integration of rental and ownership housing in one complex, (2) Separation of rental and ownership housing into different complexes, and (3) Uncertain.

3.2.4. Resident Activities

Resident activities included participation in community organization meetings, types of community organizations, and usage of community facilities. Participation in community organization meetings was determined by responses to the question “Have you participated in any community organization or meeting within the past year?” coded as “Yes” or “No”. Types of community organizations include tenant representative meetings, women’s groups, senior citizen groups, various clubs, resident communities (online sites), and other categories. The usage of community facilities was assessed based on eleven items, indicating frequency of use on a dichotomous scale (“Do not use” coded as 0, “Occasionally use + Frequently use” coded as 1).

3.2.5. Mental Health

Mental health comprised everyday life stress and self-esteem. Everyday life stress was measured on a 4-point Likert scale in response to the question “How much stress do you experience in your daily life?” ranging from “None” (1) to “Very much” (4). Self-esteem was assessed through eight items, scored on a 4-point Likert scale, reflecting positive self-regard. The reliability of the self-esteem scale is indicated by Cronbach’s alpha (0.751). A higher score indicates higher self-esteem. All four relevant factors and their respective items used in this study are detailed in

Table 3. List of Related Variables.

Variables			Survey Questions
Related Variables	Residential Property	Residential Satisfaction	Q1. Convenience of public transportation (subway, bus, etc.) Q2. Accessibility of daily amenities (market, supermarket, etc.) Q3. Accessibility of public facilities (community center, police station, etc.) Q4. Accessibility of cultural facilities (libraries, art galleries, etc.) Q5. Accessibility of medical facilities (hospitals etc.) Q6. Accessibility of green spaces and rest spaces (parks, etc.) Q7. Accessibility of welfare facilities (welfare center, senior center, etc.) Q8. Educational environment (kindergarten, school, academy, etc.) Q9. Childcare environment (daycare center, playroom, etc.) ① Very dissatisfied–④ Very satisfied
		Mixed residence status	Q1. Do you live in a complex with a mixture of rental and sale housing? ① Yes, ② No
		Conflict severity in mixed residence	How serious are the problems or conflicts that arise between residents of mixed-use complexes and those living in pre-sale housing? ① Not serious at all–⑤ Very serious Q1. Conflict over parking Q2. Destruction of property in common areas Q3. Cleanliness issues such as waste separation and collection Q4. Just my playground Q5. Just my passing road Q6. Community facilities within the complex
	Social Mix Perception		Q1. Based on your experience, what do you think about people from different economic classes living together for social integration? ① Strongly disagree–④ Strongly agree Q2. What do you think about the opening of new non-preferred facilities such as special schools, disabled homes, funeral homes, etc. in your community? ① Strongly disagree–④ Strongly agree Q3. When supplying rental housing, do you think it is better to build a mixture of rental housing and general sale housing in one complex, or separate them in different complexes? ① It is better to build a mixture of rental housing and sales housing in one complex. ② It is better to build rental housing and sales housing separately in different complexes. ③ I don’t know
	Resident activity	Participation in resident organization meetings	Q1. Have you ever participated in any resident organizations or gatherings within the complex (area) over the past year? ① Yes, ② No
		Types of resident organizations	Q1-1. What kind of organization is this? ① Tenant representative meeting ② Women’s association ③ Senior citizens’ association ④ Various clubs ⑤ Resident community (online site) ⑥ Others
		Availability of community facilities	Q1. Small library Q2. Day care center Q3. Senior center Q4. Resident cafe Q5. Playground Q6. Public laundry room Q7. Unstaffed delivery box Q8. Guest house Q9. Communal kitchen Q10. Exercise facilities (fitness center, etc.) Q11. Multipurpose room (conference room, seminar room, etc.) ① Do not use–③ Use often
	Mental health	Daily life stress	Q1. How much stress do you experience in your daily life? ① Not stressed at all–④ Very stressed
		Self esteem	Q1. I think I am a person of value like everyone else. Q2. I think I have good character. Q3. I can do things as well as most other people. Q4. I have a positive attitude toward myself. Q5. I am generally satisfied with myself. Q6. I wish I could respect myself more. Q7. I don’t have much to boast about. (Reverse question) Q8. Sometimes I think I’m not a good person. (Reverse question) Not at all–⑤ it really is (Arithmetic mean of 8 questions)

.

3.3. Research Model

The aim of this study was to typify discrimination experiences and social capital among public rental housing residents and analyze differences in related factors by type. To achieve this, the research model depicted in Figure 1 has been established. Latent class analysis (LCA) was employed to identify unobservable subgroups within the public rental housing residents based on their experiences of discrimination and levels of social capital. This method allowed for the classification of individuals into distinct types, facilitating a more nuanced examination of how these subgroups experience differences in related factors.

Figure 1. Research model depicting the relationships analyzed in the study. (a) Research Model Step 1: Latent Class Types of Public Rental Housing; (b) Research Model Step 2: Associated Factors Including Residential Property, Social Mix Perception, Resident Activity, and Mental Health.

Figure 1. Research model depicting the relationships analyzed in the study. (a) Research Model Step 1: Latent Class Types of Public Rental Housing; (b) Research Model Step 2: Associated Factors Including Residential Property, Social Mix Perception, Resident Activity, and Mental Health.

The grey boxes in (a) represent the main influencing factors leading to latent class types of public rental housing. The yellow box in (b) highlights the detailed categorization of related factors analyzed in the second step of the research model. These include four specific components: Residential Property, Social Mix Perception, Resident Activity, and Mental Health.

3.4. Latent Class Analysis Method

Latent class analysis (LCA) is a statistical technique used to identify hidden subgroups within a population based on observed variables [40]. This method was selected for its ability to model complex patterns and uncover heterogeneity within the sample without relying solely on predefined categories. By using LCA, this study classified residents into distinct subgroups based on their reported discrimination experiences and levels of social capital, providing deeper insight into the characteristics of each group and their differences in related factors.

The justification for employing LCA lies in its flexibility and capability to handle multiple indicators to define latent classes. This approach aligns with the study’s goal of typifying discrimination experiences and social capital types among public rental housing residents, allowing for a detailed analysis that informs targeted policy recommendations for each subgroup.

3.5. Data Analysis Method

For data analysis, this study utilized SPSS 26.0 and Mplus 7.4 to perform latent class analysis (LCA), chi-square analysis, and one-way analysis of variance (ANOVA). The primary goal was to explore the typologies of public rental housing residents based on their experiences with discrimination and social capital.

Latent class analysis (LCA) was employed to identify unobserved subgroups, or latent classes, sharing similar patterns of responses regarding discrimination experiences and social capital. LCA is conceptually similar to cluster analysis but offers greater flexibility in assumptions about the data, such as the absence of linearity, normality, and homogeneity of variance. This method divides individuals into classes based on shared characteristics, allowing for an analysis of differences across these groups.

To determine the optimal number of latent classes, multiple fit indices were used. First, Akaike Information Criterion (AIC), Bayesian Information Criterion (BIC), and Adjusted BIC were used to assess the model fit. These information-based indices evaluate the balance between model complexity and goodness of fit, with smaller values indicating a better fit. Additionally, the Entropy Index was employed to measure the classification accuracy of the latent class model. The entropy value ranges from 0 to 1, with values closer to 1 indicating a higher accuracy in classifying individuals. A value of 0.8 or above is typically considered indicative of well-classified latent classes. Finally, the Lo–Mendell–Rubin likelihood ratio test (LMRT) was conducted to compare the fit between models with different numbers of latent classes. If the LMRT yielded significance at p < 0.05, this indicated that the k-class model was a better fit than the k-1 model, thus selecting the appropriate number of classes.

The LCA was conducted using Mplus 7.4, with the aforementioned statistical indices guiding the final determination of the latent classes based on residents’ discrimination experiences and levels of social capital.

Following the identification of latent classes, chi-square analysis and one-way analysis of variance (ANOVA) were conducted using SPSS 26.0 to examine differences across the latent groups. The chi-square analysis was used to assess associations between categorical variables, such as housing characteristics and perceptions of social integration. Meanwhile, ANOVA was applied to evaluate mean differences in continuous variables, including resident activities, mental health outcomes (e.g., daily life stress and self-esteem), and social mixing perceptions across the latent groups. These statistical methods allowed for a comprehensive understanding of how these latent groups differed in terms of their housing environment, social integration, and well-being.

By employing these analytic techniques, this study provided a detailed examination of how residents of public rental housing in the Republic of Korea can be grouped based on their experiences with discrimination and social capital. Furthermore, the analysis revealed how these latent typologies related to housing characteristics, mental health, and perceptions of social cohesion, offering valuable insights into the social dynamics within public rental housing communities.

4. Results

4.1. Latent Class Classification of Discrimination Experience and Social Capital Among Public Rental Housing Residents

To address Research Question 1, we systematically identified latent classes among public rental housing residents by increasing the number of classes from two to five and examining the model fit indices. Initially, as the number of classes increased, the AIC, BIC, and adjusted BIC values consistently decreased. However, from five classes onward, these values began to increase. The entropy value, reflecting the quality of classification, remained at a moderate level of adequacy, around 0.6, up to four classes but decreased to 0.571 for five classes.

The selection of the four-class model was based on a careful evaluation of these statistical criteria, which indicated that it provided the best balance between fit and classification quality. Each model was evaluated for its ability to meaningfully classify the data, and the justification for choosing the four-class model is outlined below.

The two-class model, while showing reasonable fit indices (AIC, BIC), produced an entropy value of 0.630, which reflected a moderate level of classification quality. Although the model suggested that the data could be grouped into two categories, the entropy value indicated that the differentiation between these two groups was not sufficiently clear. This model oversimplified the diversity of discrimination experiences and social capital within the population, making it less suitable for capturing the complexities of these factors. The classification was too broad to identify the nuanced patterns necessary for the research questions of this study.

The three-class model showed a slight improvement in entropy, reaching 0.643. This indicated a better classification quality compared to the two-class model. However, it still did not provide enough granularity to differentiate between key subgroups effectively. The LMRT test for the three-class model was significant, suggesting some improvement in model fit. Despite this, the model still lacked the necessary detail to fully capture the complex patterns in the data, and therefore, it did not meet the requirements for this study.

The four-class model, on the other hand, was selected as the most appropriate solution. It exhibited the highest entropy value of 0.691, indicating strong classification quality. This model offered the best balance between model fit and interpretability, allowing for clear and meaningful distinctions among the latent classes. The LMRT values for the four-class model were statistically significant, further supporting its validity. Additionally, the four-class model did not contain any latent class with less than 1% of the sample, which is a critical criterion for ensuring class stability. This model provided the most meaningful distinctions among the groups, enabling a detailed exploration of the relationships between discrimination experiences and social capital among public rental housing residents.

The five-class model, although it improved the AIC and BIC values, resulted in a lower entropy value of 0.571, suggesting a decrease in classification quality. The LMRT for the five-class model was not statistically significant, which indicated that adding a fifth class did not improve the model’s ability to classify the data effectively. Moreover, the increased number of latent classes led to some small class sizes, which compromised the robustness and interpretability of the model. Consequently, the five-class model was considered less effective in distinguishing meaningful subgroups and was not selected.

In conclusion, the four-class model was deemed the most appropriate solution for revealing discrimination experiences and social capital among public rental housing residents. It provided the best balance between statistical fit and practical interpretability, making it the optimal choice for this study. The details of the model fit verification are provided in

Table 4. Verification of Model Fit.

Model	AIC	BIC	SSBIC	LMRT (p-Value)	Entropy
2 class	21,483.974	21,554.942	21,519.988	1854.572 ***	0.630
3 class	21,413.302	21,522.980	21,468.961	81.073 ***	0.643
4 class	21,401.096	21,549.485	21,476.400	23.737 *	0.691
5 class	21,404.737	21,591.836	21,499.685	8.198	0.571

* p < 0.05, *** p < 0.001.

.

• Latent Class 1: Neighborly Support Seekers Group

Latent Class 1 consisted of 1567 individuals (33.5% of the total sample), characterized by minimal discrimination experiences (1.1%). However, they exhibited relatively high levels of neighbor trust (69.0%) and neighbor interactions (55.0%). Although they had few neighbors to turn to for help in times of difficulty (6.4%), their willingness to assist needy neighbors (68.5%) was pronounced. This group, with strong social capital and a high degree of neighborly support, is indicative of the importance of fostering community belongingness and social trust. Policies focused on strengthening community engagement and building local networks would likely support the positive relationships within this group. According to social capital theory, such networks of trust and reciprocity are crucial for building cohesive and resilient communities. Therefore, Latent Class 1 was named the ‘Neighborly Support Seekers Group’.

• Latent Class 2: Group Accepting Losses

Latent Class 2 comprised 63 individuals (1.3% of the total sample), exhibiting the highest level of discrimination experiences among the four groups (56.2%). Similar to Latent Class 1, they demonstrated high levels of neighbor trust (69.0%) and interaction (41.4%). While they had fewer neighbors to seek help from in times of need (14.8%), their willingness to assist struggling neighbors was 100%. Despite facing high discrimination, they maintained strong social capital, which underscores the resilience of this group. This group’s willingness to support others despite personal struggles reflects a need for targeted policies that address mental health support and social inclusion programs. The social exclusion theory emphasizes that marginalized groups benefit from strengthening internal networks, which can help mitigate the adverse effects of discrimination. Therefore, Latent Class 2 was labeled the ‘Group Accepting Losses’.

• Latent Class 3: Group with High Social Capital

Latent Class 3 included 2735 individuals (58.4% of the total sample) with minimal discrimination experiences (5.1%). They displayed the highest levels of neighbor trust (91.7%), interaction (99.3%), willingness to seek help from neighbors (66.7%), and willingness to assist neighbors (98.0%) among the four groups. This group demonstrates a high level of social capital, characterized by strong interpersonal relationships and community involvement. Policies aimed at promoting civic participation and community development can harness the potential of this group to strengthen neighborhood cohesion. According to social capital theory, these strong ties contribute to the well-being and stability of communities. Consequently, Latent Class 3 was designated the ‘Group with High Social Capital’.

• Latent Class 4: Group Indifferent to Neighbors

Latent Class 4 comprised 318 individuals (6.8% of the total sample) with very few discrimination experiences, akin to Latent Classes 1 and 3 (8.6%). However, their response rates for neighbor trust (0%), interaction (26.1%), seeking help from neighbors (4.9%), and willingness to assist neighbors (8.8%) were the lowest among the four groups. This group displays very low social capital and social engagement, reflecting a potential area of concern for social isolation. Policies that encourage social mixing and community outreach are essential for fostering connections and reducing isolation within this group. Social isolation theory highlights the detrimental effects of weak social ties on mental and emotional health, which calls for policies that enhance neighborhood engagement. Hence, Latent Class 4 was named the ‘Group Indifferent to Neighbors’.

The characteristics of each latent class are summarized in

Table 5. Characteristics by Latent Class Type.

Variables	Latent Class 1	Latent Class 2	Latent Class 3	Latent Class 4
N (%)	1567 (33.5)	63 (1.3)	2735 (58.4)	318 (6.8)
Experiences of discrimination	0.011	0.562	0.051	0.086
Trust in neighbors	0.690	0.634	0.917	0.000
Neighborhood exchange	0.550	0.414	0.993	0.261
Neighbors to ask for help	0.064	0.148	0.667	0.049
Willingness to help neighbors	0.685	1.000	0.980	0.088

, and the response patterns by latent class types are illustrated in Figure 2.

4.2. Factors Associated with Latent Class Types

Research Question 2 examines the distinctive characteristics associated with each identified latent class. Multinomial logistic regression was applied to analyze how various factors are associated with the likelihood of being classified into one of the four latent classes. The results, summarized in the corresponding tables, provide a comprehensive and robust understanding of these associations.

4.2.1. Association Between Latent Class Types and Housing Characteristics

Housing characteristics, such as housing environment satisfaction, residency in mixed-use complexes, and levels of conflict, showed significant associations with latent class membership. Housing environment satisfaction, rated on a 4-point scale, was significantly higher for the Group Accepting Losses (3.21 points) and the Group with High Social Capital (3.14 points) compared to other groups. Satisfaction with access to amenities, public facilities, and medical services was notably lower for the Group Indifferent to Neighbors (low 2-point range).

The regression results indicated that higher housing environment satisfaction significantly increased the likelihood of belonging to the Group Accepting Losses (OR = 1.45, p < 0.05) and the Group with High Social Capital (OR = 1.32, p < 0.01) compared to the reference group (Group Indifferent to Neighbors). Residency in mixed-use complexes was most common among the Group Accepting Losses (50.8%) but did not show statistically significant differences for other groups. Additionally, higher levels of conflict related to parking and cleanliness were associated with the Group Accepting Losses (OR = 1.20, p < 0.05) and the Group with High Social Capital (OR = 0.95, p < 0.05). These findings are detailed in

Table 6. Housing Characteristics by Latent Class Type.

Variables			Total	Latent Class 1 (a)	Latent Class 2 (b)	Latent Class 3 (c)	Latent Class 4 (d)	F/X2	Scheffe
N (%)			(n = 4683)	1567 (33.5)	63 (1.3)	2735 (58.4)	318 (6.8)
Resident activity	Participation in resident organization meetings	Convenience of public transportation	3.09 (0.67)	3.05 (0.68)	3.21 (0.83)	3.14 (0.65)	2.81 (0.66)	28.185 ***	b,c > a > d
		Accessibility to daily amenities	2.99 (0.71)	2.99 (0.72)	3.05 (0.81)	3.02 (0.68)	2.75 (0.78)	13.942 ***	a,b,c > d
		Accessibility to public facilities	3.01 (0.63)	3.01 (0.62)	3.10 (0.76)	3.04 (0.61)	2.70 (0.70)	30.043 ***	a,b,c > d
		Accessibility to cultural facilities	2.81 (0.72)	2.79 (0.74)	2.90 (0.73)	2.86 (0.69)	2.50 (0.77)	26.343 ***	a,b > d c > a,d
		Accessibility to medical facilities	3.01 (0.61)	3.00 (0.62)	3.10 (0.67)	3.04 (0.59)	2.71 (0.66)	29.397 ***	a,b,c > d
		Accessibility to green spaces and rest areas	3.10 (0.66)	3.06 (0.67)	2.95 (0.68)	3.15 (0.63)	2.79 (0.74)	32.967 ***	c > a > d
		Accessibility to welfare facilities	3.01 (0.057)	3.00 (0.58)	2.95 (0.66)	3.05 (0.55)	2.76 (0.68)	24.876 ***	a,c > d
		Educational environment	3.01 (0.59)	2.99 (0.58)	2.94 (0.69)	3.05 (0.56)	2.72 (0.68)	32.833 ***	c > a > d
		Childcare environment	3.00 (0.56)	2.99 (0.56)	2.78 (0.77)	3.05 (0.53)	2.69 (0.63)	42.932 ***	c > a > b,d
	Types of resident organizations	Yes	1862 (39.8)	588 (37.5)	32 (50.8)	1122 (41.0)	120 (37.7)	8.841 ***	-
		No	2821 (60.2)	979 (62.5)	31 (49.2)	1613 (59.0)	198 (62.3)
	Availability of community facilities	Conflict over parking	1.38 (0.59)	1.30 (0.51)	1.66 (0.90)	1.42 (0.62)	1.42 (0.62)	7.424 ***	b,c > a
		Damage to public space	1.35 (0.55)	1.31 (0.54)	1.47 (0.57)	1.37 (0.54)	1.32 (0.61)	2.273	-
		Cleanliness issues (separate garbage collection, etc.)	1.38 (0.59)	1.32 (0.56)	1.72 (0.89)	1.40 (0.59)	1.34 (0.57)	6.586 ***	b > c > a,d
		Just my playground	1.35 (0.57)	1.31 (0.56)	1.41 (0.67)	1.38 (0.58)	1.29 (0.49)	2.448	-
		Just my passing road	1.35 (0.55)	1.31 (0.53)	1.44 (0.56)	1.38 (0.56)	1.35 (0.64)	2.259	-
		Community facilities within the complex	1.40 (0.61)	1.31 (0.53)	1.56 (0.56)	1.44 (0.64)	1.36 (0.59)	7.282 ***	c > a

*** p < 0.001. Note: Letters (a, b, c, d) indicate latent class types: (a) Latent Class 1: Group Accepting Losses; (b) Latent Class 2: Group with High Social Capital; (c) Latent Class 3: Group Seeking Friendly Neighbor Relationships; (d) Latent Class 4: Group Indifferent to Neighbors.

.

4.2.2. Association Between Latent Class Types and Social Mixing Perception

Perceptions of social mixing varied significantly across latent classes. Agreement with the idea of different socioeconomic strata living together for social integration was highest for the Group Accepting Losses (mean 3.00), followed by the Group with High Social Capital (mean 2.93), the Group Seeking Friendly Neighbor Relationships (mean 2.84), and the Group Indifferent to Neighbors (mean 2.64). Regression analysis confirmed that agreement with social mixing was significantly higher for the Group Accepting Losses (OR = 1.33, p < 0.01) and the Group with High Social Capital (OR = 1.72, p < 0.01).

A similar pattern emerged regarding agreement on the location of undesirable facilities such as special schools, residential facilities for the disabled, and funeral homes in the community. The Group Accepting Losses showed the highest mean agreement (2.70), followed by the Group with High Social Capital (2.66), the Group Seeking Friendly Neighbor Relationships (2.56), and the Group Indifferent to Neighbors (2.44). This trend highlights that groups with higher social capital and resilience are more open to social integration initiatives.

Preferences regarding the integration of rental and general sale housing within the same complex were consistent across all latent classes. The majority of respondents across all groups agreed that mixing housing types within a single complex is preferable (Group Accepting Losses: 55.6%; Group with High Social Capital: 56.2%; Group Seeking Friendly Neighbor Relationships: 57.6%; Group Indifferent to Neighbors: 52.2%). These preferences underline a general consensus on the benefits of integrated housing, despite the variability in other social mixing perceptions. Figure 3 illustrates these preferences in detail, while full details of these associations are summarized in

Table 7. Social Mix Perception Characteristics by Latent Class Type.

Variables			Total	Latent Class 1 (a)	Latent Class 2 (b)	Latent Class 3 (c)	Latent Class 4 (d)	F/X2	Scheffe
N (%)			(n = 4683)	1567 (33.5)	63 (1.3)	2735 (58.4)	318 (6.8)
Social Mix Perception	Agree that people from different economic classes live together for social integration		2.88 (0.49)	2.84 (0.51)	3.00 (0.40)	2.93 (0.45)	2.64 (0.59)	38.483 ***	c > a > d b > d
	Agree to the location of non-preferred facilities (special schools, residential facilities for the disabled, funeral homes, etc.) in the community		2.61 (0.58)	2.56 (0.59)	2.70 (0.61)	2.66 (0.56)	2.44 (0.64)	19.435 ***	c > a > d b > d
	Rental housing and general sale housing	It is better to mix and build in one complex	2639 (56.4)	902 (57.6)	35 (55.6)	1536 (56.2)	166 (52.2)	27.690 ***	-
		It is better to build it separately in another complex	1147 (24.5)	357 (22.8)	10 (15.9)	715 (26.1)	65 (20.4)
		I don’t know	897 (19.2)	308 (19.7)	18 (28.6)	484 (17.7)	87 (27.4)

*** p < 0.001.

.

4.2.3. Association Between Latent Class Types and Resident Activity

Resident activities, including participation in community organizations and the use of various community facilities, showed significant differences among the latent classes. Participation in resident organization meetings was low across all groups, with the Group with High Social Capital demonstrating the highest participation rate (2.0%) and the Group Indifferent to Neighbors the lowest (0.6%). Among participants, the most common types of organizations included senior citizens’ associations (26.6%) and women’s societies (23.4%), with notable differences in the kinds of groups preferred by each class. These results are detailed in

Table 8. Resident Activity Characteristics by Latent Class Type 1.

Variables			Total	Latent Class 1 (a)	Latent Class 2 (b)	Latent Class 3 (c)	Latent Class 4 (d)	X2
N (%)			(n = 4683)	1567 (33.5)	63 (1.3)	2735 (58.4)	318 (6.8)
Resident activity	Participation in resident organization meetings	Yes	64 (1.4)	7 (0.4)	1 (1.6)	54 (2.0)	2 (0.6)	18.639 ***
		No	4619 (98.6)	1560 (99.6)	62 (98.4)	2681 (98.0)	316 (99.4)
	Types of resident organizations (n = 64)	Tenant representative meeting	8 (12.5)	1 (14.3)	0	7 (13.0)	0	12.690
		Women’s societies	15 (23.4)	1 (14.3)	0	14 (25.9)	0
		Senior citizens’ association	17 (26.6)	3 (42.9)	0	14 (25.9)	0
		Various clubs	10 (15.6)	0	0	9 (16.7)	1 (50.0)
		Resident community (online site)	11 (17.2)	2 (28.6)	1 (100.0)	7 (13.0)	1 (50.0)
		Other	3 (4.7)	0	0	3 (5.6)	0

*** p < 0.001.

.

The usage of community facilities further highlighted disparities among latent classes. The Group Accepting Losses and the Group with High Social Capital reported the highest usage rates for key facilities such as playgrounds (23.1% and 19.6%, respectively) and senior citizen centers (4.3% and 3.3%, respectively). Resident cafés were more frequently used by the Group Accepting Losses (14.8%) compared to the Group Indifferent to Neighbors (8.5%), indicating greater community engagement in these groups. On the other hand, the Group Indifferent to Neighbors consistently exhibited the lowest facility usage rates across all categories. These patterns are summarized in

Table 9. Resident Activity Characteristics by Latent Class Type 2.

Variables				Total	Latent Class 1 (a)	Latent Class 2 (b)	Latent Class 3 (c)	Latent Class 4 (d)	Χ2
N (%)				(n = 4683)	1567 (33.5)	63 (1.3)	2735 (58.4)	318 (6.8)
Resident activity	Availability of community facilities	Small library (n = 1699)	Not used	1564 (92.1)	492 (93.9)	32 (97.0)	953 (90.9)	87 (92.6)	5.339
			Used	135 (7.9)	32 (6.1)	1 (3.0)	95 (9.1)	7 (7.4)
		Daycare center (n = 3095)	Not used	3008 (97.2)	958 (98.5)	43 (97.7)	1860 (96.4)	147 (98.7)	11.105
			Used	87 (2.8)	15 (1.5)	1 (2.3)	69 (3.6)	2 (1.3)
		Senior citizens center (n = 3783)	Not used	3659 (96.7)	1131 (98.2)	55 (100.0)	2268 (95.7)	205 (100.0)	25.006 ***
			Used	124 (3.3)	21 (1.8)	0	103 (4.3)	0
		Resident café (n = 790)	Not used	673 (85.2)	226 (85.9)	11 (57.9)	393 (85.2)	43 (91.5)	12.814 **
			Used	117 (14.8)	37 (14.1)	8 (42.1)	68 (14.8)	4 (8.5)
		Playground (n = 3809)	Not used	3062 (80.4)	1047 (87.3)	45 (77.6)	1802 (76.9)	168 (80.8)	54.127 ***
			Used	747 (19.6)	153 (12.8)	13 (22.4)	541 (23.1)	40 (19.2)
		Public laundry room (n = 259)	Not used	222 (85.7)	91 (83.5)	5 (83.3)	109 (86.5)	17 (94.4)	1.655
			Used	37 (14.3)	18 (16.5)	1 (16.7)	17 (13.5)	1 (5.6)
		Unstaffed delivery box (n = 920)	Not used	671 (72.9)	264 (74.8)	22 (84.6)	342 (70.4)	43 (78.2)	4.797
			Used	24 9(27.1)	89 (25.2)	4 (15.4)	144 (29.6)	12 (21.8)
		Guesthouse (n = 660)	Not used	614 (93.0)	174 (95.1)	16 (100.0)	399 (91.5)	25 (100.0)	5.806
			Used	46 (7.0)	9 (4.9)	0	37 (8.5)	0
		Community kitchen (n = 117)	Not used	11 1(94.9)	53 (96.4)	0	48 (92.3)	10 (100.0)	1.495
			Used	6 (5.1)	2 (3.6)	0	4 (7.7)	0
		Exercise facilities (n = 1156)	Not used	918 (79.4)	276 (79.3)	24 (82.8)	558 (78.3)	60 (90.9)	6.115
			Used	238 (20.6)	72 (20.7)	5 (17.2)	155 (21.7)	6 (9.1)
		Multipurpose room (n = 1725)	Not used	1696 (98.3)	55 7(99.5)	25 (100.0)	1014 (97.5)	100 (100.0)	10.801 *
			Used	29 (1.7)	3 (0.5)	0	26 (2.5)	0

* p < 0.05, ** p < 0.01, *** p < 0.001.

.

Facilities such as unstaffed delivery boxes and exercise spaces were also utilized differently among groups. For example, unstaffed delivery boxes were used by 27.1% of the Group Accepting Losses compared to 21.8% of the Group Indifferent to Neighbors. Similarly, the Group Accepting Losses reported the highest frequency of exercise facility usage (20.6%), while the Group Indifferent to Neighbors reported significantly lower rates (9.1%).

The differences in facility usage and community participation emphasize the varying degrees of social capital and community engagement across latent classes, with the Group Accepting Losses and the Group with High Social Capital showing consistently higher engagement.

4.2.4. Association Between Latent Class Types and Mental Health

Mental health outcomes, measured by daily life stress and self-esteem, varied significantly among the latent classes. Daily life stress scores were highest for the Group Seeking Friendly Neighbor Relationships (mean 2.78) and the Group Indifferent to Neighbors (mean 2.76), compared to the Group with High Social Capital (mean 2.66). This indicates that individuals in the Group Seeking Friendly Neighbor Relationships and Group Indifferent to Neighbors may experience higher stress levels in their daily lives, potentially due to lower levels of community engagement and social capital.

Self-esteem scores showed a markedly different pattern. The Group Accepting Losses reported the highest self-esteem (mean 3.01), closely followed by the Group with High Social Capital (mean 2.94). In contrast, the Group Indifferent to Neighbors displayed the lowest self-esteem levels (mean 2.71). Regression analysis confirmed these findings, showing that higher self-esteem significantly increased the likelihood of belonging to the Group Accepting Losses (OR = 1.20, p < 0.05) and the Group with High Social Capital (OR = 1.60, p < 0.01). Lower stress levels were also associated with the Group with High Social Capital (OR = 0.95, p < 0.05).

These results are summarized in

Table 10. Mental Health Characteristics by Latent Class Type.

Variables		Total	Latent Class 1 (a)	Latent Class 2 (b)	Latent Class 3 (c)	Latent Class 4 (d)	F	Scheffe
N (%)		(n = 4683)	1567 (33.5)	63 (1.3)	2735 (58.4)	318 (6.8)
mental health	daily life stress	2.69	2.73 (0.58)	2.78 (0.68)	2.66 (0.58)	2.76 (0.59)	7.572 ***	a,d > c
	self-esteem	2.89	2.83 (0.40)	3.01 (0.41)	2.94 (0.37)	2.71 (0.40)	55.834 ***	b,c > a,d a,b,c > d

*** p < 0.001.

, which provides a comprehensive overview of mental health characteristics by latent class type. A visual representation of these differences is provided in Figure 4, which illustrates the variations in stress and self-esteem across the classes using a heatmap.

The associations across housing satisfaction, social mixing perceptions, resident activities, and mental health provide a comprehensive view of the factors influencing latent class membership. As detailed in

Table 11. Summary of Factors Associated with Latent Class Types.

Factor	Latent Class 1 (a)	Latent Class 2 (b)	Latent Class 3 (c)	Latent Class 4 (d)	Key Observations
Housing Environment Satisfaction	Medium	High	High	Low	Higher satisfaction in c and b
Conflict Levels	Low	Medium	Low	Medium	Higher conflicts in a and d
Social Mixing Perceptions	Medium	High	High	Low	Strong agreement in b and c
Facility Usage	Low	High	High	Low	Greater usage in b and c
Daily Life Stress	High	High	Low	High	Higher stress in a and d
Self-Esteem	Medium	High	High	Low	Highest in b, followed by c

, Class 2 (Group Accepting Losses) and Class 3 (Group with High Social Capital) consistently demonstrated positive outcomes in housing satisfaction, facility usage, and self-esteem, while Class 1 (Group Seeking Friendly Neighbor Relationships) and Class 4 (Group Indifferent to Neighbors) exhibited higher levels of stress and lower engagement with community resources. These findings underline the importance of enhancing housing conditions, promoting social integration, and fostering community engagement to address disparities among latent classes.

5. Discussion

This study explored the critical interaction between social capital and residential satisfaction in shaping attitudes toward vulnerable populations, providing new insights into the mechanisms behind social integration. The findings validate existing theories on the role of social networks in fostering inclusive communities and extend them by identifying the mediating role of residential satisfaction. Enhanced housing conditions, such as improved safety, accessibility, and social cohesion, significantly influence social attitudes by strengthening social capital and reducing prejudice. These results emphasize that housing environments serve as both physical and social contexts, actively shaping interpersonal relationships and community dynamics.

By adopting an integrated approach, this research bridged the gap in the existing literature, which has often treated social capital and residential satisfaction as separate factors. This study identified four distinct latent groups, each offering unique insights into the interplay between housing satisfaction, social capital, and discrimination. For instance, the “Group Accepting Losses” exhibited remarkable resilience despite experiencing high levels of discrimination, highlighting the critical role of bonding social capital in fostering social cohesion amidst adversity. Conversely, the “Group Indifferent to Neighbors” demonstrated how low integration and minimal trust can exacerbate feelings of isolation, even in the absence of overt discrimination. These group-specific dynamics provide a narrative lens through which the findings can be better understood.

The results further indicated that residential satisfaction amplifies the positive impacts of social capital, particularly in reducing discrimination and fostering inclusive attitudes. For example, individuals with higher residential satisfaction were more likely to demonstrate positive attitudes toward disadvantaged groups, showcasing the synergistic relationship between these variables. This integrated approach provides a nuanced understanding of the interplay between residential and social factors in shaping social outcomes.

Incorporating the findings into the discussion, this study revealed that social capital functions not only as a static resource but also as a dynamic mediator that evolved within the context of improved housing conditions. Vulnerable populations, particularly those facing educational and employment challenges, benefit disproportionately from enhanced residential environments. These findings highlight the compounded advantages of targeted interventions, such as creating shared spaces and organizing community-building activities, which can significantly improve social cohesion and reduce marginalization for vulnerable groups.

The theoretical implications of this study challenge the traditional emphasis on physical integration as the primary strategy for addressing housing discrimination. Instead, the findings advocate for complementing physical integration with efforts to foster meaningful social connections. For instance, creating communal spaces, facilitating community-building activities, and encouraging participation in local decision-making can enhance social cohesion in public housing settings where stigma and discrimination persist. Such strategies directly address the social dimensions of exclusion, promoting more inclusive communities.

Policy implications also extend to global contexts, as demonstrated by the “Group with High Social Capital”, which benefits from strong networks and high housing satisfaction. Supporting such groups through leadership opportunities and continued community involvement can strengthen neighborhood cohesion and collective well-being. Conversely, targeted outreach programs for the “Group Indifferent to Neighbors” are essential to combat isolation and foster trust. These tailored interventions address the specific needs of each group, ensuring more effective and sustainable outcomes.

6. Conclusions

This study identified four distinct latent groups among public rental housing residents based on their experiences with discrimination and social capital. By integrating housing characteristics, social mixing perceptions, resident activities, and mental health, the findings provide a comprehensive understanding of the dynamics within these communities.

The first group, “Group Seeking Friendly Neighbor Relationships”, faces minimal discrimination but lacks a robust social support network. Despite their willingness to help others, this group experiences higher stress and lower self-esteem due to limited social interactions. The second group, “Group Accepting Losses”, endures the highest levels of discrimination but maintains strong trust in neighbors and a willingness to assist others. However, they experience more conflict, particularly in mixed-use complexes. The third group, “Group with High Social Capital”, reports minimal discrimination, enjoys high housing satisfaction, and has strong social capital with low levels of stress. Lastly, the “Group Indifferent to Neighbors” experiences minimal discrimination but is poorly integrated socially, showing low levels of trust, interaction, and self-esteem.

The findings of this study underscore that social capital, when paired with residential satisfaction, is pivotal in fostering positive attitudes toward vulnerable groups. Enhancing housing satisfaction amplifies these effects, leading to greater social cohesion and reduced discrimination. To foster stronger communities internationally, policies must prioritize improving social support networks tailored to the unique needs of vulnerable groups. For the “Group Seeking Friendly Neighbor Relationships”, providing opportunities for meaningful social engagement, establishing communal spaces, and offering mental health support are critical strategies to reduce stress and enhance self-esteem.

For the “Group Accepting Losses”, policies should address both social and physical integration, particularly by resolving conflicts in mixed-use complexes. Promoting social mixing and implementing community-building activities can foster a stronger sense of belonging and mitigate perceived discrimination. Additionally, policies should support the “Group with High Social Capital” by encouraging their continued involvement in neighborhood activities and providing opportunities for leadership roles. Such initiatives will leverage their strong social ties to further enhance community cohesion and collective well-being. For the “Group Indifferent to Neighbors”, targeted outreach programs are crucial to combat social isolation. Policies should focus on enhancing social integration, improving access to community services, and creating opportunities for participation in neighborhood activities. These measures will help build trust, strengthen relationships, and address mental health challenges, ultimately reducing isolation and improving overall well-being.

In conclusion, this study highlights the transformative potential of integrating social and residential factors to foster cohesive communities. By addressing the specific needs of each group, policymakers can design targeted strategies that promote social inclusion and resilience. These findings contribute to both theoretical advancements and practical frameworks for equitable public housing policies.

7. Directions for Future Research

This study acknowledges several limitations that provide important avenues for future research. First, the data were collected exclusively from residents in Seoul, which may limit the generalizability of the findings to other regions within the Republic of Korea or to different cultural contexts. The observed dynamics between discrimination, social capital, and residential satisfaction might reflect unique characteristics of Seoul’s public housing system, potentially differing from patterns in other urban or rural settings. Expanding the geographical scope of future research would provide deeper insights into how these dynamics vary across different regions and housing contexts.

Second, the cross-sectional design restricts the ability to establish causal relationships. While this study identified strong associations between social capital, residential satisfaction, and attitudes toward vulnerable populations, it cannot definitively determine the direction or causality of these relationships. Longitudinal studies could offer valuable insights into how these relationships evolve over time, enabling researchers to evaluate the long-term effects of housing policies on social integration and resident well-being.

Additionally, the use of single-item indicators to measure discrimination experiences may not fully capture their complexity. This limitation could lead to an underestimation or oversimplification of the nuanced ways in which discrimination affects social capital and residential satisfaction. Future research should employ multi-item scales to provide a more detailed understanding of how discrimination impacts social capital and broader social outcomes. Including additional variables such as cultural diversity, age, and health status could offer a more holistic view of the factors influencing social capital and discrimination in public rental housing.

Finally, cross-national comparisons would be beneficial to evaluate how housing policies and social integration strategies differ across countries. These comparisons could reveal variations in how housing environments and social programs affect attitudes toward vulnerable populations, offering global insights for designing inclusive housing policies. Such research could inform the development of targeted interventions that prioritize both physical and social integration, ensuring they are adaptable to different cultural and institutional contexts.

The findings of this study underscore the importance of improving residential satisfaction to enhance social capital, which in turn promotes positive attitudes toward vulnerable populations. The findings suggest that interventions tailored to the specific needs of different groups can significantly reduce discrimination and foster social cohesion. For example, strategies emphasizing both community-building activities and physical improvements can create environments where social capital thrives. Future research, including a broader geographic scope, longitudinal data, and cross-national comparisons, will refine these insights further and deepen our understanding of how housing policies can positively impact social integration and resident well-being.