Delineating Psychological Boundaries in Urban Tourism: Development and Outcomes of the Perceived Tourist-Resident Segregation Scale

Defining Urban Segregation: Understanding Divisions in Urban Spaces

Urban segregation refers to spatial and social divisions within cities that separate groups based on socioeconomic, ethnic, or cultural differences (Duncan & Duncan, 1955; Roberts & Wilson, 2009). It includes visible physical divisions, such as residential clusters and neighborhood boundaries (Vaughan, 2007), as well as less visible social barriers that limit interaction and integration between groups (Hamnett, 2001). Historically, segregation has been associated with unequal access to resources, services, and opportunities, shaping where people live, how they interact, and their broader social and economic well-being (Burgess, 1928; Van Kempen, 1994). Importantly, segregation is not a uniform process: it may occur voluntarily or through imposition, with imposed segregation often restricting mobility and deepening marginalization (Nieuwenhuis & Hooimeijer, 2015; Peach, 1996, 2005). Accordingly, its consequences may be negative, as in cases of isolation, poverty, and marginalized urban “underclasses” (Marks, 1991), or potentially positive, when voluntary separation strengthens social ties and preserves collective identity within self-selected communities (Wilson, 2012). This dual perspective suggests that the meaning and effects of segregation depend on the conditions under which separation occurs, the degree of agency involved, and the resources available to separated groups.

This dual character is particularly relevant in urban tourism contexts. Tourism-related clustering may reflect residents’ voluntary avoidance of tourist-dominated areas to preserve everyday routines, local identity, or economic autonomy. At the same time, it may also result from imposed restructuring, as external capital, tourism-oriented planning, and commercial redevelopment produce exclusive tourism zones that displace or marginalize local residents (Yang et al., 2024). Thus, tourism-related segregation cannot be understood simply as the spatial concentration of tourists and residents in different areas. Rather, it involves overlapping processes through which urban space is reorganized, social interaction is reshaped, and access to everyday amenities becomes unevenly distributed. Tourism research has increasingly examined these processes through a political economy perspective, showing how tourism-led capital accumulation restructures urban space and redistributes access to housing, services, and public amenities (Milano et al., 2024; Saarinen & Wall-Reinius, 2019). Studies on tourism gentrification, rent-gap exploitation, short-term rentals, touristified enclaves, neighborhood contestation, and resident resistance demonstrate how tourism development generates exclusion, socio-spatial inequality, and changing relationships with place (Cocola-Gant & Lopez-Gay, 2020;Liang & Bao, 2015;Mansilla & Milano, 2019;Wu et al., 2013;Yang et al., 2024). However, existing tourism studies often treat segregation as an implicit outcome of tourism development rather than as a distinct and multidimensional process. They rarely differentiate clearly between spatial separation, such as the clustering of tourism activities, and social separation, such as reduced interaction, weakened belonging, or perceived exclusion. Moreover, reliance on qualitative evidence or aggregate spatial indicators limits their ability to capture how residents experience and interpret tourism-related segregation in everyday life. Addressing these limitations requires a clearer distinction between spatial and social dimensions of segregation, as well as a measurement framework grounded in residents’ perceptions.

The Two Main Dimensions of Urban Segregation: Spatial Versus Social Segregation

To unpack the complexity of urban segregation, it is essential to distinguish between its two core dimensions: spatial segregation and social segregation. While closely interlinked, these dimensions operate through different mechanisms and produce distinct impacts on urban life (Vaughan, 2007). Spatial segregation refers to the physical division of groups across distinct areas, such as neighborhoods or districts, often due to socioeconomic inequalities, cultural distinctions, or policy decisions (Reardon & O’Sullivan, 2004). It is strongly tied to living spaces and economic disparities that influence where people can live, which may in turn, impact their access to resources and social opportunities (Tammaru et al., 2021). In some urban centers, affluent visitors may intensify these dynamics, driving up property prices and often displacing long-term residents (Gotham, 2013). In contrast, social segregation refers to barriers to meaningful interaction among different groups, stemming from linguistic, cultural, or economic differences (Hamnett, 2001). Beyond fixed living spaces, urban activity spaces play a role in social segregation as these capture how daily routines reinforce spatial divides (Sun et al., 2024). Although less visible than spatial segregation, social segregation restricts cross-group social networks and interactions, limiting integration and deepening divisions within urban environments (Hamnett, 2001).

Despite the growing significance of urban tourism, research has paid limited attention to how it shapes or reinforces these forms of segregation. Tourists, as temporary urban actors, can reshape not only physical environments but also social dynamics, altering patterns of access, interaction, and exclusion within cities (Cuadrado-Ciuraneta et al., 2016). In tourism-dependent cities therefore, segregation cannot be fully understood by examining residents alone; it also involves an understanding of the ways in which tourists co-produce socio-spatial divisions through their mobility, consumption, and accommodation practices. Due to tourism’s fluid nature, conventional segregation metrics are often inadequate, prompting a need for new, context-specific methods (Wu et al., 2013). In this study, however, segregation does not refer to all forms of difference or simple variation between tourists and residents. Rather, it refers to residents’ perceived separation from tourists when such differences become relatively stable in everyday urban life and are reflected in patterns of housing, access to facilities, social interaction, and consumption spaces. In this way, the concept extends beyond conventional residential unevenness while retaining its core concern with socio-spatial divisions between groups. Based on this understanding, tourism-related segregation is conceptualized as a multidimensional perceptual construct that captures how residents experience and interpret the boundaries produced by tourism development. This gap calls for more context-sensitive approaches that can capture the socio-spatial impacts of tourism-related segregation. To this end, the present study adapts existing segregation frameworks to urban tourism contexts and introduces a residential satisfaction model to assess how these dynamics affect the well-being and lived experience of local residents.

The Measurement and Analysis of Segregation in Urban Social Spaces

The measurement of segregation has traditionally relied on quantitative indicators that capture how groups are distributed across urban space. In spatial segregation research, Massey and Denton’s (1988) five dimensions (i.e., evenness, exposure-isolation, concentration, centralization, and clustering) have provided an influential framework for assessing group separation. However, these dimensions may overlap conceptually, complicating interpretation (Timberlake, 2015). Traditional measures such as the dissimilarity index also face spatial limitations, as illustrated by White’s (1983) “checkerboard problem,” where different spatial arrangements can generate similar index values. To address these limitations, later studies incorporated proximity, contiguity, and mobility-based weighting to make segregation measures more sensitive to geographic clustering, barriers, scale, and everyday activity spaces (Reardon & O’Sullivan, 2004; Wong & Shaw, 2010). A similar shift can be observed in the measurement of social segregation. Earlier approaches relied on proportion-sensitive indices such as the Gini, dissimilarity, and Atkinson indices (D. R. James & Taeuber, 1985), but these measures often overlooked spatial proximity and social interaction. Reardon and Firebaugh (2002) advanced this discussion by reconceptualizing segregation as inequality weighted by social distance, thereby incorporating spatial, network, and relational dimensions. More recent work has further expanded segregation analysis through mobile phone signaling and housing price data, enabling dynamic assessments of co-presence, communication, social similarity, and homophily distance across space and time (Xu et al., 2019). These developments have improved the precision of segregation measurement by moving beyond static residential distributions toward more relational and mobility-sensitive understandings of urban separation.

Despite these advances, existing measurement approaches remain limited for tourism contexts, where populations are fluid, interactions are temporary, and tourist–resident boundaries are often situational rather than fixed (Shoval & Isaacson, 2009). More importantly, conventional indices are poorly equipped to capture residents’ subjective and psychological experiences of separation, particularly when tourism-related segregation may be either externally imposed or voluntarily adopted (Soja, 1980; Yang et al., 2024). In tourism cities, residents may experience segregation not only through residential location, but also through differential access to facilities, limited interaction with tourists, and separation in everyday consumption spaces. These perceived boundaries can be alienated in some cases but may also function as protective or adaptive mechanisms that help residents preserve routine, identity, or autonomy. Accordingly, this study shifts attention from structural indicators to perceptual measurement by developing a psychological segregation scale that captures how residents perceive spatial and social tourism-related separation in urban areas. By integrating both negative and positive interpretations, this approach offers a more context-sensitive understanding of how tourism-related segregation shapes residents’ lived experiences and community well-being. It also provides the conceptual bridge to the next section, which situates perceived tourism urban segregation within residential satisfaction theory to examine how these perceived boundaries influence residents’ evaluations of their living conditions.

The Outcomes of Tourism-Related Segregation on Residential Satisfaction

Residential satisfaction refers to residents’ evaluations of the habitability and overall quality of their living environments (Jagun et al., 1990). It is widely understood as a multidimensional construct shaped by physical conditions, such as housing quality and neighborhood facilities, as well as social factors, including community relations and social networks (Grillo et al., 2009; R. N. James et al., 2008; Onibokun, 1976). Classic residential satisfaction research has examined these determinants across multiple spatial and social domains (Adriaanse, 2007; Cutter, 1982; Galster, 1985; Speare, 1974). Among these frameworks, Marans and Rodgers’ (1975)) tripartite model is especially relevant because it distinguishes dwelling, neighborhood, and community domains, thereby linking urban structure, everyday experience, and subjective satisfaction (Cutter, 1982). Similarly, Amérigo and Aragonés (1997) emphasized that objective environmental conditions are filtered through residents’ perceptions and personal characteristics to produce subjective evaluations.

In selecting an appropriate theoretical foundation for this study, both classic and contemporary residential satisfaction frameworks were considered (Adriaanse, 2007; Grillo et al., 2009), along with tourism-related quality-of-life models that link tourism development to residents’ well-being (Andereck & Nyaupane, 2010; Eslami et al., 2019). The model by Marans and Rodgers (1975) was chosen because its tripartite structure clearly differentiates the neighborhood domain, which is the level most directly reshaped by tourism-related spatial and social change (López-López et al., 2006). In tourism cities, transformations in land use, housing markets, public spaces, and service provision primarily manifest at the neighborhood scale, making this domain central to residents’ lived experiences of tourism. At the same time, the social dimension of residential satisfaction is increasingly shaped by interactions between residents and transient populations such as tourists, second-home users, and short-term renters. To capture these dynamics, this study integrates neighborhood satisfaction, social satisfaction, and support for tourism development by drawing on contemporary quality-of-life frameworks (Andereck & Nyaupane, 2010; Woo et al., 2015). In doing so, the model bridges residential satisfaction theory and tourism impact research, enabling a more context-sensitive assessment of how residents evaluate tourism-affected urban environments.

In the context of urban tourism, neighborhood environments and social relations are particularly sensitive to tourism-driven spatial and economic restructuring. Previous studies show that residents’ satisfaction is strongly shaped by neighborhood quality, access to services, and the social climate of their communities (Amérigo & Aragonés, 1997; Galster, 1985; Onibokun, 1974, 1976). A consistent conclusion across this literature is that both dwelling and neighborhood characteristics play crucial roles in shaping residential satisfaction. However, tourism development introduces distinctive dynamics that reconfigure these relationships, as tourist clustering, visitor-oriented services, and short-term rentals unevenly reshape neighborhood environments (Hall & Müller, 2004; Hannonen et al., 2016; Wu et al., 2013). For this reason, this study draws on spatial and social dimensions identified in prior research (see Supplemental Appendix B: Table B-1) but focuses specifically on Neighborhood Environment (NE) and Social Aspects (SA), rather than Dwelling Physical Quality (DPQ) and Personal Characteristics (PC). DPQ is less directly affected by tourism processes, while PC primarily moderates rather than determines how tourism impacts are experienced (Wang & Pfister, 2008). This focus allows the model to isolate how tourism-related change in neighborhood conditions and social relations shapes residential satisfaction in tourism cities.

Research on the social aspects of urban tourism has often treated neighbors as homogeneous groups or distinguished them primarily by ethnicity (Cutter, 1982; Dill et al., 2015; Onibokun, 1976). In tourism cities, however, residents and tourists increasingly occupy different social and spatial spheres, as tourism impacts are unevenly distributed across neighborhoods and shaped by accessibility, tourism services, attractiveness, and policy interventions (Adamiak et al., 2016; Hall & Müller, 2004; Hannonen et al., 2016; Hiltunen, 2007; Wu et al., 2013). This clustering produces spatial and social separation between tourists and residents, contributing to residential segregation that undermines neighborhood quality and social cohesion (Dill et al., 2015). Tourism-related spatial change intensifies land-use pressure, alters neighborhood functions, and reduces residential comfort (Liu et al., 2026;Piñeira et al., 2025;Romera et al., 2025), while simultaneously weakening social connectedness in touristified areas (Back et al., 2025; Tena et al., 2025). Accordingly, these processes indicate that tourism-related segregation is a key mechanism linking tourism development to residents’ evaluations of their living environments.

Because spatial segregation reshapes access to housing, facilities, and everyday environments, it directly affects how residents experience their neighborhoods and social lives (Hall & Müller, 2004; Hannonen et al., 2016; Wu et al., 2013). Concentrated tourism development increases congestion and functional displacement, reducing neighborhood livability and everyday comfort (Piñeira et al., 2025; Romera et al., 2025) and weakening local cohesion (Back et al., 2025; Tena et al., 2025). Accordingly, perceived spatial separation from tourists is expected to reduce both neighborhood and social satisfaction.

Perceived social segregation reflects relational distance between residents and tourists in everyday interactions. Limited or strained contact erodes local belonging, emotional comfort, and social cohesion (Joo et al., 2018;Thyne et al., 2020;Yilmaz & Tasci, 2014), making everyday environments feel less familiar and supportive. As a result, higher perceived social separation is expected to reduce both neighborhood and social satisfaction.

Residents who feel satisfied with their neighborhood conditions and social environments tend to view tourism more positively and are more willing to support tourism development that aligns with local quality of life (Nawijn & Mitas, 2011; Thyne et al., 2020). Conversely, when tourism is perceived to undermine neighborhood livability or weaken social comfort, residents may become less supportive of further tourism development.

In addition to the indirect effects operating through neighborhood satisfaction and social satisfaction, perceived tourist-resident segregation may also exert direct effects on support for tourism development. Residents’ perceptions of segregation are not only evaluative cues shaping their satisfaction with local living conditions but may also directly influence their overall stance toward tourism development (Thyne et al., 2020).

This study extends the residential satisfaction model of Marans and Rodgers (1975) by integrating spatial and social segregation as core determinants of neighborhood and social satisfaction in tourism cities (see Figure 1), thereby linking classic segregation theory (Massey & Denton, 1988; Reardon & O’Sullivan, 2004) with residents’ evaluations of tourism-affected urban environments and their support for tourism development. At the same time, however, tourism-related segregation may not operate uniformly across all dimensions. In highly tourism-intensive settings, some forms of separation may be experienced primarily as disruptive and exclusionary, whereas others may function as buffering or protective boundaries that help residents preserve stability, routine, or local control over everyday spaces. Accordingly, the present study treats the proposed negative relationships as a theoretically grounded baseline expectation, while also recognizing that different dimensions of perceived tourism urban segregation may display distinct nomological profiles once examined empirically.

OPEN IN VIEWER

Figure 1.

Proposed residential satisfaction model in urban tourism domain.

Conceptual Framework and Interview-Based Item Development

This study followed comprehensive procedures for scale development outlined by Churchill (1979), who recommended an eight-step process. Following these steps ensured that development and validation of the new scale would facilitate meaningful interpretation of test scores. Two critical considerations in developing measurement scales are validity and reliability. Validity refers to the degree to which measurement scales accurately capture the constructs of interest (Borsboom et al., 2004), while reliability pertains to the consistency or repeatability of results when the same measurement is applied (DeVellis & Thorpe, 2022). Steps 1 through 4 of Churchill’s (1979) procedures focus on addressing concerns related to face or content validity, dimensionality, and internal consistency, while steps 5 through 8 target issues of reliability, criterion validity, and construct validity (Echtner & Ritchie, 1993). Churchill (1979) emphasized that these procedures could be adapted with flexibility, allowing researchers to replace recommended techniques with suitable alternatives. In this study, such adaptations included utilizing in-depth interviews and an expert panel to construct the item pool and assessing reliability and validity through composite reliability, convergent validity, and discriminant validity. The eight steps were followed to establish the reliability and construct validity of the PTUS scale (see Supplemental Appendix B: Table B-2).

Results From the Initial Items and In-Depth Interviews

When developing a measurement scale, it is essential to review existing theories and determine the appropriate level of specificity for capturing constructs of interest (DeVellis & Thorpe, 2022). This study investigates how long-term residents perceive urban segregation between themselves and tourists within an urban-tourism context. Sanya, located at the southernmost tip of China’s Hainan Island and recognized as the country’s only tropical coastal metropolis, served as an ideal case for this research. It combines the characteristics of a medium-sized city (e.g., dense residential neighborhoods, a diversified service economy, and substantial permanent population) with a strong dependence on tourism and real estate-driven development. Since the early 2000s, Sanya has emerged as China’s leading destination for second-home tourism, characterized by the purchase or seasonal rental of properties by non-local individuals who make repeated short-term visits (Chen & Bao, 2020). By the end of 2024, Sanya’s registered household population reached 785,001, while the permanent resident population had grown to 1.1161 million (Sanya Municipal Bureau of Statistics., 2025). Government data show that over 90% of newly built apartments are owned by off-island buyers, contributing to the development of gated “second-home enclaves” that are spatially and socially separated from the original communities (Wu et al., 2013; Yang et al., 2024). In this study, these enclaves are conceptualized as tourist- or non-local oriented spaces. The study’s focus is on how long-term residents perceive segregation in relation to such enclaves, rather than on the perspectives of second-home users themselves. These enclaves, marked by physical barriers, exclusive amenities, and limited daily interaction, have resulted in well-documented socio-spatial divides (Wu et al., 2013), making Sanya a strategic site for analyzing segregation processes in tourism-dependent cities.

Informed consent was obtained from participants prior to their involvement in the in-depth interviews, with clear assurances of voluntary participation and confidentiality (Interview questions are detailed in Supplemental Appendix A). Building on the foundational framework of residential segregation (Massey & Denton, 1988), this study conceptualized urban segregation through two primary dimensions: spatial segregation and social segregation (Musterd, 2020). Interview data were analyzed using a thematic coding procedure. Open coding was conducted to identify meaning units related to spatial, functional, and social differentiation between tourists and residents. Through axial coding, these codes were grouped into higher-order categories such as housing-based separation, facility access differentiation, interactional distance, and consumption segregation. Selective coding then consolidated these categories into the four PTUS dimensions (see Table 1). In this study, “residents” referred to long-term community members not interviewed in an official or professional capacity, whereas “tourism stakeholders” included individuals such as government officials, real estate agents, hotel managers, and property managers, some of whom were also residents of Sanya but were interviewed primarily because of their institutional or industry roles. From these interviews, four specific themes of perceived segregation were identified: Perceived Housing Segregation, Perceived Facility Segregation, Perceived Interaction Segregation, and Perceived Consumption Segregation. In the PTUS framework, the spatial dimension is captured through perceived housing segregation and perceived facility segregation, whereas the social dimension is reflected in perceived interaction segregation and perceived consumption segregation. These subdimensions provided the basis for developing items to capture residents’ subjective experiences of segregation in a tourism-driven urban environment.

OPEN IN VIEWER

Table 1.

Thematic Structure of Spatial and Social Segregation Identified From In-Depth Interviews.

Themes	Subthemes	Reference
Spatial segregation	Perceived housing segregation	Massey and Denton (1988) and interviews
	Perceived facility segregation	Interviews
Social segregation	Perceived interaction segregation	Farber et al. (2015) and interviews
	Perceived consumption segregation	Interviews

Panel of Experts

To ensure the accuracy of the measurement constructs, two senior scholars in the tourism area, two graduate students majoring in management with expertise in scale development and four tourism professionals were enlisted to evaluate and revise the initial 26 items. The eight experts were provided with the theoretical definitions of perceived tourism segregation and its four dimensions. Following the evaluation criteria proposed by Dolnicar and Grün (2012), the panel assessed the initial pool of 26 items (see Supplemental Appendix B: Table B-3) based on three key aspects: (1) the correlation between the items and the construct, (2) clarity of the measurement question and its reflection of the construct, and (3) whether the item fully covered various types of conceptual thinking of the construct. Based on their analysis, the experts provided suggestions which were used to delete and/or revise the initial items. This culminated in the elimination of four items (Item Nos. 7, 9, 11, 14) that lacked generality or repeated meanings. Additionally, they proposed rewording and combining items to enhance interpretability and construct coverage. As a result of this first round of revisions, several items were restructured, and a new item was added, ultimately producing a refined pool of 23 items used in the formal survey instrument (see Supplemental Appendix B: Table B-4). To further improve linguistic precision and contextual clarity, six undergraduate students majoring in tourism at Sanya Tourism College were invited to review the revised items. Their feedback led to refinements in the Mandarin phrasing of certain questions, ensuring that the items were clearly understood by local respondents during the survey process.

Survey Distribution

As the objective of the PTUS scale was to measure the perceived residential segregation of residents living in a tourism city, the target population was urban residents who were registered in and had lived in a tourism city (i.e., Sanya city) for more than 1 year and were 18 years or older. In this study, residents were thus defined as long-term community members with formal or de facto permanent residence in Sanya, whereas tourists were treated as temporary visitors, including short-stay holidaymakers and seasonal second-home tourists, who did not belong to the local household registration system. The PTUS items capture residents’ perceptions of segregation instead of these tourist groups, and tourists themselves were not surveyed. This study complied with ethical standards for research involving human participants. Prior to beginning the survey, all participants were presented with an informed consent statement outlining the purpose of the study, their rights as participants, and assurances of confidentiality and anonymity. Only those who consented proceeded to the questionnaire. Data collection was carried out through a structured three-round survey process, yielding a total of 955 completed responses (see Table 2). The initial pilot round, aimed at evaluating the clarity and effectiveness of the survey instrument, utilized both online and face-to-face distribution methods and resulted in 80 responses, of which 55 were deemed valid. Based on insights from the pilot, revisions were made to improve the survey design before launching the formal data collection.

OPEN IN VIEWER

Table 2.

Survey Data Collection Sample Size.

Stage	Method of data collection	Completed questionnaires	Valid responses retained
Pilot study	Investigator-administered survey	80	55
Formal study: Round 1	Investigator-administered survey	627	421
Formal study: Round 2	Online questionnaire platform (Credamo)	248	150
Formal study total		875	571
Overall total collected		955	626

Note. Only the 571 valid responses from the formal study were used in the main quantitative analyses. The pilot sample was used exclusively for instrument refinement.

The first round of formal data collection was conducted in May 2022. Although participants were approached through both online and in-person methods, all respondents completed the survey digitally by scanning a QR code, in compliance with local social distancing policies. A total of 627 responses were collected in this round. The second formal round took place in October 2022 via the online platform (https://www.credamo.com/), ¹ adding 248 responses to the dataset and allowing further assessment of the scale’s robustness (see Table 2). The multi-stage data collection timeline reflects the sequential nature of scale development, which required an initial field study, extensive item generation and pretesting, followed by two rounds of formal survey administration to refine and validate the PTUS instrument following Churchill’s (1979) eight-step procedure.

Both online and face-to-face collection methods were used to reach respondents and comply with COVID-19 safety protocols. Online distribution primarily reached younger participants, while in-person recruitment included older or less digitally active residents. Unique QR codes and IP restrictions prevented duplicate submissions between sources. To ensure the quality of the online data, three criteria were established: (1) All survey questions had to be completed; (2) Respondents were required to spend a minimum of 200 s completing the questionnaire; and (3) they had to correctly answer attention-check questions. Surveys that met these conditions were deemed valid, resulting in 571 valid responses.

Measurement for the Residential Satisfaction Framework

To empirically test the conceptual framework proposed in Figure 1, this study employs validated indicators to measure residential Neighborhood Satisfaction (NS), Social Satisfaction (SS), and Support for Tourism Development (SFT). These items were adapted from previous residential satisfaction and tourism perception studies (Amérigo & Aragonés, 1997; Ap & Crompton, 1998; Grillo et al., 2009; R. N. James et al., 2008; Lankford & Howard, 1994; Onibokun, 1976) and were evaluated using a 5-point Likert scale. Supplemental Table B-5 in Appendix B outlines the constructs, corresponding items, scales, and supporting literature.

Items Analysis: Pilot Study

The pilot study included 55 valid responses, exceeding the commonly recommended size for pilot testing (Browne, 1995). Item performance was evaluated using critical ratio (CR) analysis, corrected item-total correlations, and Cronbach’s alpha. The results identified several problematic items. Specifically, PHS7, PHS8, PIS5, and PCS5 showed weak discrimination, with CR values below the recommended threshold of 3. In addition, PHS8 and PIS5 had corrected item-total correlations below .40, indicating weak consistency with the overall scale. Cronbach’s alpha analysis further suggested that PHS7, PIS5, and PCS5 made poor contributions to internal reliability. Based on these results, four items (PHS7, PHS8, PIS5, and PCS5) were removed prior to the formal survey, thereby improving the scale’s parsimony and psychometric adequacy.

Participant Profile Following Formal Data Collection

After the pilot, 19 items were included in the survey for formal data collection, and 571 valid surveys were collected (see Table 2). Respondent demographics indicated that the majority of residents were aged 25 to 34 (38.88%) and 35 to 44 (37.83%), with only 0.70% aged 65 and above, reflecting a younger sample likely influenced by the online survey platform. Female participants (61.47%) outnumbered males (38.53%), and the sample was well-educated, with 50.79% holding undergraduate degrees and many possessing postgraduate qualifications. The most common response to monthly household income was 5,000 to 10,000 yuan (31.17%), aligning with Sanya’s middle-income socio-economic profile. Employment data revealed that 78.81% were full-time workers.

Exploratory Factor Analyses and Common Method Bias Assessment

Following the generation of the 19-item PTUS scale, its dimensional structure was first assessed using exploratory factor analysis (EFA). To strengthen the measurement validation process, the full sample of 571 respondents was randomly divided into two independent subsamples. The first subsample (n = 285) was used for EFA, whereas the second subsample (n = 286) was reserved for confirmatory factor analysis (CFA). The initial EFA was conducted on the 19 items using principal axis factoring (PAF) with Promax rotation, as the latent dimensions were theoretically expected to be correlated. The data were suitable for factor analysis, with a Kaiser–Meyer–Olkin (KMO) value of 0.910 and a significant Bartlett’s test of sphericity, χ²(171) = 4,140.608, p < .001. These results indicated that the correlation matrix was factorable.

Inspection of the initial EFA results suggested a four-factor solution, broadly corresponding to the theorized dimensions of Perceived Facility Segregation (PFS), Perceived Interaction Segregation (PIS), Perceived Consumption Segregation (PCS), and Perceived Housing Segregation (PHS). However, three items under the PHS dimension did not perform satisfactorily and were therefore excluded before the second-round EFA. Specifically, PHS2 showed a moderate primary loading on the intended factor (0.595) but a low communality (0.295), indicating limited shared variance with the underlying factor structure. PHS3 was weaker, with a primary loading of 0.382 and a communality of 0.303. In addition, PHS6 showed an unstable loading pattern, loading almost equally on two factors (0.384 and 0.385) despite a communality of 0.459, suggesting that it lacked a clear factor assignment. Accordingly, these three items were removed. A second EFA was then performed on the remaining 16 items. The revised item pool again demonstrated strong suitability for factor analysis, with KMO = 0.903 and a significant Bartlett’s test, χ²(120) = 3,771.106, p < .001 (see Table 3). The resulting four-factor solution explained 70.319% of the total variance, which was higher than the variance explained in the initial 19-item solution. As shown in Table 3, the rotated pattern matrix revealed a clear and interpretable structure.

OPEN IN VIEWER

Table 3.

Pattern Matrix for the Final EFA Solution.

Items	Factor loading				Communalities
	Factor 1	Factor 2	Factor 3	Factor 4
PHS1	−0.081	−0.028	−0.039	0.755	0.475
PHS4	−0.056	0.087	−0.054	0.800	0.620
PHS5	0.126	−0.115	0.092	0.669	0.555
PFS1	0.932	0.012	−0.137	−0.027	0.710
PFS2	0.884	0.007	0.011	−0.002	0.798
PFS3	0.860	0.013	0.082	−0.016	0.837
PFS4	0.883	−0.002	0.090	−0.059	0.839
PFS5	0.898	−0.009	0.023	0.037	0.858
PIS1	0.083	0.822	−0.149	0.059	0.662
PIS2	0.038	0.933	−0.143	−0.034	0.748
PIS3	−0.043	0.586	0.223	0.101	0.589
PIS4	−0.086	0.749	0.235	−0.118	0.638
PCS1	−0.025	−0.035	1.007	−0.044	0.898
PCS2	0.019	−0.033	0.997	−0.083	0.894
PCS3	0.061	0.021	0.543	0.178	0.515
PCS4	0.169	0.129	0.457	0.176	0.614

Although a few retained items were comparatively weaker than the core indicators within their dimensions, they were preserved on both statistical and conceptual grounds. In particular, PHS1, PIS3, PCS3, and PCS4 showed relatively weaker communalities or lower primary loadings than the strongest items in their respective constructs. Nonetheless, these items retained interpretable primary loadings, did not exhibit problematic cross-loadings, and were considered important for preserving dimensional breadth. Their adequacy was also subsequently supported in the CFA stage.

To assess the potential influence of common method bias, both Harman’s single-factor test and a CFA-based single-factor test were conducted. First, all retained measurement items were entered into an unrotated factor analysis in SPSS. The first factor accounted for 49.700% of the total variance, which was slightly below the commonly used 50% threshold. Although this result provided preliminary evidence that common method bias was unlikely to be severe, the proportion was close to the cutoff, warranting a more stringent CFA-based assessment. Accordingly, a single-factor CFA model was estimated in AMOS 29, with all 16 items loading onto one latent construct. The results showed that the single-factor model fit the data poorly, χ²(104) = 1,348.729, χ²/df = 12.969, CFI = 0.634, TLI = 0.577, IFI = 0.635, NFI = 0.617, and RMSEA = 0.205, indicating that the observed covariance among the items could not be adequately represented by a single common factor. Taken together, these results suggest that common method bias was unlikely to pose a serious threat to the findings.

Confirmatory Factor Analysis, Reliability, and Validity

The 16-item, four-factor structure derived from the EFA was subsequently cross-validated using CFA on the holdout subsample (n = 286) in AMOS 29. The CFA results supported the adequacy of the proposed measurement model. The model demonstrated an acceptable overall fit to the data, χ²(98) = 276.964, χ²/df = 2.826, CFI = 0.947, TLI = 0.935, IFI = 0.948, NFI = 0.921, and RMSEA = 0.080. Importantly, this four-factor model fit the data substantially better than the single-factor CFA model used to assess common method bias, further supporting the distinctiveness of the proposed measurement structure. The standardized factor loadings for all items were statistically significant and ranged from .550 to .956 (see Table 4), indicating that the retained items loaded adequately on their intended constructs. Among the four dimensions, the PFS items displayed particularly strong loadings, whereas several items, such as PHS1, PCS3, and PCS4, were comparatively weaker but remained above the minimum acceptable threshold. Overall, the CFA results provided support for the convergent validity of the four latent constructs.

OPEN IN VIEWER

Table 4.

Standardized Factor Loadings, Convergent Validity, and Reliability.

Construct	Item	Loading	AVE	CR
PHS	PHS1	0.550	0.534	0.770
	PHS4	0.838
	PHS5	0.773
PFS	PFS1	0.687	0.782	0.947
	PFS2	0.885
	PFS3	0.956
	PFS4	0.941
	PFS5	0.926
PIS	PIS1	0.753	0.633	0.873
	PIS2	0.861
	PIS3	0.745
	PIS4	0.818
PCS	PCS1	0.903	0.644	0.876
	PCS2	0.941
	PCS3	0.629
	PCS4	0.693

Convergent validity and reliability were further assessed using average variance extracted (AVE) and composite reliability (CR). As shown in Table 4, the AVE values ranged from 0.534 to 0.782, exceeding the recommended cutoff of 0.50, whereas the CR values ranged from 0.770 to 0.947, all above the recommended threshold of 0.70. These results indicate satisfactory convergent validity and internal consistency for all four constructs. Among them, PFS showed the strongest psychometric performance, whereas PHS was comparatively weaker but still met acceptable standards. Discriminant validity was assessed using the Fornell–Larcker criterion. As presented in Table 5, the square root of AVE for each construct exceeded its corresponding inter-construct correlations, supporting adequate discriminant validity. Thus, although the four constructs were moderately correlated, they remained empirically distinct from one another. Discriminant validity was further examined by comparing AVE with the maximum shared squared variance (MSV) and the average shared squared variance (ASV). For all constructs, AVE exceeded both MSV and ASV, providing additional evidence of discriminant validity.

OPEN IN VIEWER

Table 5.

Discriminant Validity (Fornell–Larcker Criterion).

Construct	PHS	PFS	PIS	PCS
PHS	.731
PFS	.392	.884
PIS	.463	.530	.796
PCS	.403	.554	.542	.802

Taken together, the split-sample EFA and CFA results provide consistent support for the adequacy and stability of the four-factor PTUS measurement model, although a small number of indicators remained comparatively weaker than the core items in their respective constructs.

Structural Model Assessment and Nomological Validation

Following the split-sample validation of the PTUS scale, the proposed nomological model was estimated using the full sample (N = 571). The model linked the four dimensions of perceived tourism segregation (i.e., PHS, PFS, PIS, and PCS) to two dimensions of residential satisfaction, namely residential neighborhood satisfaction (NS) and social satisfaction (SS), which in turn were hypothesized to influence support for tourism development (SFT). Direct paths from the four segregation dimensions to support for tourism development were also specified to examine whether segregation perceptions exerted effects beyond those transmitted through residential satisfaction (see Figure 2). Consequently, an enhanced model was developed with seven sets of hypotheses, investigating the interplay between PTUS and its effects on residential satisfaction and support for urban tourism development.

OPEN IN VIEWER

Figure 2.

Proposed structural model and hypotheses.

Before testing the structural relationships, the full-sample measurement model was reassessed using CFA. The measurement model demonstrated an acceptable fit to the data, χ²(413) = 1,498.673, χ²/df = 3.629, CFI = 0.915, TLI = 0.905, IFI = 0.916, NFI = 0.887, and RMSEA = 0.068. All indicators loaded significantly on their intended constructs. However, SS6 was removed prior to the final structural estimation because it was conceptually less consistent with the other SS indicators and showed the weakest standardized loading within the construct (0.516). Unlike the remaining SS items, which reflect neighborhood-based social relations, SS6 (“Satisfaction with family relationships within the neighborhood”) refers more to private family relationships than to neighborhood-level social satisfaction. In addition, the residuals of NS5 and NS6 were allowed to covary in the final structural model. This specification was theoretically justified because both items refer to closely related forms of recreational amenities (i.e., satisfaction with recreational facilities and satisfaction with gym areas), which are often experienced as overlapping aspects of the same residential environment, particularly in gated-community settings. Allowing their residuals to correlate therefore accounted for shared item-specific variance without altering the substantive interpretation of the NS construct.

The re-estimated structural model showed an acceptable overall fit, χ²(384) = 1,609.414, χ²/df = 4.191, CFI = 0.903, TLI = 0.890, IFI = 0.903, NFI = 0.877, and RMSEA = 0.075. Although the fit of the structural model was somewhat weaker than that of the measurement-only CFA, this was not unexpected given the greater complexity of the full nomological model, which incorporated multiple latent constructs and simultaneous direct effects. Importantly, the aim of the SEM analysis was not to derive a best-fitting model through post hoc modification, but to assess the nomological validity of the validated PTUS dimensions within a theoretically specified framework. Because the measurement properties of the scale had already been established through EFA and CFA, the emphasis at this stage was placed on the substantive pattern and theoretical interpretability of the structural relationships rather than on incremental fit optimization.

As shown in Table 6, four hypotheses were supported under the conventional SEM significance criterion. Specifically, perceived interaction segregation (PIS) was negatively related to residential neighborhood satisfaction (NS; β = −.200, p < .001), supporting H3a. Residential neighborhood satisfaction was positively related to support for tourism development (SFT; β = .406, p < .001), supporting H5. Residential social satisfaction (SS) was also positively related to support for tourism development (β = .172, p = .002), supporting H6. In addition, perceived interaction segregation was negatively associated with support for tourism development (β = −.261, p < .001), supporting H7c. Several hypothesized paths were not statistically significant. These included the effects of PFS on NS and SS, the effects of PIS and PCS on SS, and the direct effects of PHS and PCS on support for tourism development. At the same time, several paths were statistically significant but in the opposite direction to that hypothesized. In particular, PHS was positively rather than negatively associated with both NS (β = .457, p < .001) and SS (β = .664, p < .001). Similarly, PCS showed a positive relationship with NS (β = .183, p = .004), and PFS showed a positive direct association with support for tourism development (β = .228, p < .001), contrary to the hypothesized negative effects.

OPEN IN VIEWER

Table 6.

Structural Path Estimates and Hypothesis Tests.

Hypothesis	Path	β	p	Outcome
H1a	PHS → NS	.457	***	Opposite direction
H2a	PFS → NS	−.050	.393	Nonsignificant
H3a	PIS → NS	−.200	***	Supported
H4a	PCS → NS	.183	**	Opposite direction
H1b	PHS → SS	.664	***	Opposite direction
H2b	PFS → SS	−.101	.079	Nonsignificant
H3b	PIS → SS	−.086	.137	Nonsignificant
H4b	PCS → SS	−.068	.265	Nonsignificant
H5	NS → SFT	.406	***	Supported
H6	SS → SFT	.172	**	Supported
H7a	PHS → SFT	−.062	.457	Nonsignificant
H7b	PFS → SFT	.228	***	Opposite direction
H7c	PIS → SFT	−.261	***	Supported
H7d	PCS → SFT	.017	.780	Nonsignificant

Note. ** and *** represent p < .01 and p < .001, respectively. Table 6 is presented for transparency of hypothesis testing, but the substantive value of the SEM stage lies in the differentiated nomological profiles of the PTUS dimensions rather than in the number of supported paths.

Overall, these results suggest that the nomological validity of PTUS lies not in a uniform pattern of negative effects, but in the fact that each dimension exhibits a theoretically interpretable relationship with neighborhood satisfaction, social satisfaction, and support for tourism development. Thus, the SEM results provide initial evidence that PTUS is meaningfully embedded within a broader established framework of residential and community satisfaction in urban tourism settings.

Housing Segregation: Spatial Autonomy and Residential Protection

The positive association between perceived housing segregation (PHS) and residential satisfaction departs from classic segregation research, which links spatial separation to reduced well-being (Charles, 2003; Dill et al., 2015). In tourism cities such as Sanya, however, housing segregation is shaped less by socioeconomic sorting than by tourism gentrification, which reallocates housing toward tourist and short-term uses and intensifies displacement pressures (Liu et al., 2026;Piñeira et al., 2025;Romera et al., 2025). As Piñeira et al. (2025) show in Santiago de Compostela, this process replaces stable residential populations with a rotating tourist presence, generating housing insecurity without producing clear district-level polarization. In Sanya, this dynamic manifests as community-level separation, where tourist-oriented gated developments and local neighborhoods coexist within the same urban districts. Under these conditions, perceived housing segregation may function as a spatial buffering mechanism that reduces residents’ exposure to tourism-driven housing competition and supports residential satisfaction. This adaptive role is further reinforced by governance gaps in addressing tourism-related housing pressures (Klepej et al., 2024). Thus, PHS emerges not as a marker of exclusion but as a moderated and adaptive partitioning that provides spatial security amid tourism-led urban change. Housing segregation in tourism cities is therefore contextually positive when it helps preserve community identity, stability, and protection from disruptive flows.

Facility Segregation: Congestion, Enclave Expectations, and Mixed Interpretations

Perceived facility segregation (PFS) did not significantly influence neighborhood or social satisfaction, indicating that facility-based separation is not directly experienced as a source of everyday well-being or social cohesion. This finding diverges from enclave-based literature, where facility exclusivity often reinforces collective identity and satisfaction (Freeman, 2000). This unexpected result highlights important contextual differences. In ethnic or immigrant enclaves, facilities reflect cultural cohesion. In tourism areas however, facility pressures (i.e., crowded transportation, parks, or healthcare) are often interpreted as general urban congestion rather than a meaningful form of exclusion (Milano et al., 2019). Residents may view tourist pressure on facilities as a predictable, if inconvenient, aspect of tourism development rather than as a structural inequity (Seraphin et al., 2018). Yet the positive effect of PFS on support for tourism development indicates that residents still value having access to “resident-only” facilities, which they see as protective buffers that mitigate tourism’s disruptions. This result demonstrates that facility segregation contributes to residents’ support of tourism, even if it does not enhance their satisfaction with daily life. Such findings reveal PFS as a latent but important dimension deserving further exploration in future research.

Interaction Segregation: Social Distance and Fragmented Cohesion

Perceived interaction segregation (PIS) emerged as the most consistently detrimental dimension, significantly reducing neighborhood satisfaction and support for tourism development, which reinforces the importance of interpersonal contact for community cohesion (Su et al., 2020). These findings align closely with the hypothesized directions and reinforce the centrality of social interaction in sustaining community cohesion. Social distance theory suggests that limited, strained, or asymmetric interactions undermine solidarity and trust (Joo et al., 2018; Yilmaz & Tasci, 2014). Recent research illuminates this pattern. Romera et al. (2025) demonstrate that touristification transforms commercial environments in ways that can reduce opportunities for resident-to-resident encounters. Similarly, Back et al. (2025) describe overtourism as a spatial and structural condition that generates relational tensions. When residents feel socially distant from tourists, or when interactions are limited, strained, or forced, neighborhoods become fragmented. PIS also reduced support for tourism development, which aligns with earlier findings that social friction weakens tolerance for tourism (Lin et al., 2017). Interactional segregation therefore represents a critical relational mechanism through which tourism undermines both social cohesion and political legitimacy.

Consumption Segregation: Local Lifestyles and Adaptive Boundary-Making

Perceived consumption segregation (PCS) exhibited a significant positive effect on neighborhood satisfaction, although this effect did not align with the hypothesized direction. Rather than indicating model inconsistency, this result points to an under-theorized adaptive dimension of segregation in tourism cities. Prior studies emphasize retail displacement and homogenization as negative outcomes of touristification (Hagemans et al., 2024). In tourism-intensive cities, however, consumption segregation is not merely a symptom of exclusion but also reflects an adaptive response to tourism-driven commercial restructuring. As tourist-oriented retail increasingly replaces local-serving commerce (Romera et al., 2025), residents experience growing competition for everyday goods, services, and spaces of social reproduction. Under such conditions, the differentiation of consumption spaces, where tourist-focused commercial zones coexist with locally oriented shopping and service areas, can function as a form of spatial and symbolic buffering. This interpretation is consistent with Piñeira et al.’s (2025) finding that social groups exposed to touristification recognize its housing and commercial impacts but interpret and navigate them through distinct lived experiences. Likewise, Tena et al. (2025) show that residents in world heritage cities actively seek out everyday cultural and commercial spaces that remain outside tourist-dominated circuits. PCS thus appears to function as a form of lifestyle boundary-making that enhances residential satisfaction by reducing competition for basic services and maintaining symbolic ownership of everyday spaces.

Overall, the findings show that tourism-related segregation is not uniformly detrimental (Andereck & Nyaupane, 2010; L. Su & Swanson, 2019). Interaction segregation undermines satisfaction and support by weakening social cohesion, whereas housing and consumption segregation may provide protective or symbolic benefits by preserving residential stability and everyday routines. Facility segregation appears to play a more structural role, shaping support for tourism rather than directly influencing residential satisfaction. These differentiated effects support the value of PTUS as a multidimensional framework for understanding how residents negotiate tourism-led urban change.

Theoretical Contribution

The PTUS scale makes a theoretical contribution by reframing segregation as a dynamic resident–tourist relationship rather than a static demographic pattern. Classic segregation research has primarily examined long-term residential clustering structured by race, ethnicity, or socioeconomic status (Massey & Denton, 1988; Reardon & O’Sullivan, 2004; Wissink et al., 2016). PTUS extends this foundation by capturing separation generated by the temporary mobility and consumption patterns of tourists, whose presence increasingly reorganizes urban space, housing demand, and everyday public environments (Liu et al., 2026; Piñeira et al., 2025; Romera et al., 2025). In doing so, PTUS advances a conceptual shift in segregation scholarship, demonstrating that transient populations can produce socio-spatial differentiation with meaningful consequences for community cohesion, residential stability, and neighborhood identity. This theoretical pivot aligns with emerging research showing that tourism functions not only as an economic system but also as a powerful producer of spatial order in contemporary cities (Back et al., 2025; Klepej et al., 2024).

Methodologically, the PTUS scale offers an initially validated and multidimensional tool that measures perceived segregation across four domains (i.e., housing, facilities, interactions, and consumption), capturing dimensions that are difficult to detect through traditional segregation indices alone. This structure allows researchers and policymakers to distinguish which forms of segregation serve protective roles, which preserve well-being and which indirectly shape tourism support, reflecting the complex lived realities of tourism-intensive cities (Ballesta, 2024; Suh et al., 2025). The PTUS scale therefore provides a promising framework for systematically assessing tourism’s socio-spatial consequences in contexts where governance gaps, commercial transformation, and housing pressures are intensifying (Tena et al., 2025). By foregrounding residents’ perceptions, the scale addresses a critical methodological gap in urban tourism research, moving beyond physical or census-based indicators toward a more nuanced, psychological understanding of how tourism-driven change is experienced in urban places.

Practical Implications

The findings demonstrate that policymakers and urban planners need a more context-sensitive approach to managing resident and tourist dynamics. While several hypotheses were not supported, these outcomes offer valuable insight into the tourism context rather than signaling contradictions. The PTUS results show that the impacts of segregation are multidimensional and that perceived housing and consumption segregation can operate as protective mechanisms that support residents’ stability, autonomy, and daily comfort. These patterns should not be interpreted as support for segregation. Instead, they highlight the importance of safeguarding residential life from unmanaged tourism pressures. Planning strategies should therefore prioritize the protection of residential areas, strengthen locally oriented services, and regulate short-term rentals to prevent displacement and commercial homogenization. Pressures on facilities and infrastructure can be alleviated through improved transport connectivity, strategic investments in public services, and careful distribution of tourism amenities. Considering insights from the PTUS framework enables planners to address not only structural conditions but also the lived experiences of residents who encounter tourism-driven changes in daily life.

The negative influence of perceived interaction segregation emphasizes the importance of fostering positive and voluntary forms of contact between residents and tourists. Policies that encourage community-focused events, cultural participation, and collaborative tourism planning can reduce feelings of social distance and strengthen cohesion. The PTUS scale provides a practical diagnostic tool that governments can integrate into community surveys or impact assessments to monitor residents’ evolving perceptions across housing, consumption, facility, and interaction domains. This capacity to detect where tourism creates tension or compatibility allows cities to pursue a more balanced tourism development model. Such an approach aims to protect residents’ quality of life, maintain their spatial and social autonomy, and create opportunities for meaningful coexistence. In doing so, it supports a more socially responsive and sustainable urban tourism agenda.