A B S T R A C T
The city serves as an arena for the reproduction of unequal power relations and a platform for the sedimentation of inequality in the urban fabric. Urban space is thus not neutral; it is produced through power, capital, and social relations. District 4 of Tehran, due to its transitional position between the affluent northern zones and the evolving eastern areas, provides an ideal laboratory for analyzing spatial-physical heterogeneity patterns. This study aims to move beyond mere descriptions and analyze the mechanisms generating heterogeneity at the neighborhood scale. By integrating the TOPSIS model for neighborhood ranking with spatial statistics indicators, the research attempts to analyze the inequality pattern not as a random phenomenon but as a product of the logic of urban space production in Tehran. The study employed a descriptive-analytical approach using physical, demographic, and economic data across 20 neighborhoods of District 4. Findings reveal that Qasemabad and Shahrak-e-Valiasr exhibit the highest inequality indices, while peripheral neighborhoods rank lower not due to privilege but because of uniformity in deprivation. Moran's I was calculated as 0.267. (p=0.069), indicating non-significance at the 95% confidence level but showing evidence of clustering at the 90% level. Getis-Ord Gi* hot spot analysis identified Majidabad as a critical heterogeneity hotspot at 95% confidence. The findings confirm that heterogeneity in District 4 results from uneven development. It is recommended that urban management, instead of uniform policies, adopt differentiated interventions—distributing infrastructure in cold spots and fostering synergistic development in hot spots—to prevent the reproduction of spatial poverty.
Extended Abstract
Introduction
Urban space is not simply a physical container in which social, economic, and demographic processes take place. It is produced and transformed through the interaction of social relations, economic interests, political decisions, land ownership, investment, and physical structures. From this perspective, spatial heterogeneity is not merely a measurable difference between places; it is also the visible manifestation of processes through which land values, infrastructure, services, opportunities, housing conditions, and urban functions are unevenly organized across space. Understanding such heterogeneity requires moving beyond neighborhood differences and examining the spatial processes and structural conditions shaping them.
Tehran represents an important context for such an analysis because its urban structure has developed through interactions among economic forces, demographic change, planning interventions, land markets, and differentiated investment. These processes have produced spatial differentiation in physical form, housing conditions, economic characteristics, population structure, and urban land uses. District 4 is particularly relevant. Located in northeastern Tehran, the district occupies a transitional position between high-value northern areas and more heterogeneous eastern and southern urban environments. This position, together with the diversity of its residential structures, economic values, land uses, population characteristics, and physical conditions, makes District 4 a suitable setting for investigating spatial-physical heterogeneity at the neighborhood scale.
Covering 61.29 km², it consists of nine sub-districts and 20 neighborhoods. Its neighborhoods differ substantially in terms of demographic composition, housing characteristics, land prices, economic activities, and access to various urban land uses. Such differences suggest that heterogeneity in District 4 is multidimensional and cannot be adequately represented by a single development indicator. Moreover, similar composite values may mask different internal conditions. A low level of heterogeneity, for example, may represent either a balanced condition or a homogeneous pattern of low development and limited resources. Consequently, identifying the spatial structure of heterogeneity requires both multidimensional measurement and spatial interpretation.
The theoretical framework of the study combines critical realism and the political economy of space. Critical realism distinguishes between observable events and the deeper structures and generative mechanisms through which those events are produced. Within this perspective, the spatial-physical differences observed among neighborhoods are regarded as empirical outcomes that may reflect deeper economic, political, social, and spatial relations. Political economy of space provides the main framework for interpreting these differences by emphasizing the role of capital, power, land ownership, and the social production of urban space. Lefebvre conceptualizes space as both a product and a medium of social relations, suggesting that spatial forms emerge through social practices. Harvey connects urban spatial transformation to capital accumulation, land value, commodification, and differentiated investment. Smith’s theory of uneven development further explains how capitalist development generates simultaneous processes of concentration and differentiation across geographical space.
Previous studies have made valuable contributions to the measurement of spatial inequality, development differences, neighborhood characteristics, urban services, and spatial patterns. However, much of this literature has focused on larger geographical scales or relied primarily on descriptive and ranking-based approaches. In addition, numerical rankings do not necessarily reveal whether similar or dissimilar neighborhood conditions are spatially concentrated. A neighborhood may occupy a high or low position in a ranking without forming a statistically significant spatial cluster with surrounding areas. This methodological gap is relevant when the objective is to understand the spatial organization of heterogeneity rather than simply rank neighborhoods.
Accordingly, the present study aims to identify and compare spatial-physical heterogeneity among the 20 neighborhoods of District 4, determine the spatial distribution of the composite heterogeneity index, and identify statistically significant concentrations of high and low values. The study examines the contribution of demographic-social, economic, physical-residential, and urban land-use dimensions to the spatial structure. The methodological contribution is integrating the TOPSIS multi-criteria decision-making model with global and local spatial statistics. In this way, neighborhood ranking is connected directly to geographic location and spatial significance, allowing heterogeneity to be analyzed as a spatially organized phenomenon.

Methodology
This applied study employed a descriptive-analytical design with a focus on spatial analysis. District 4 of Tehran Municipality constitutes the study area. The district covers 61.29 km², had a population of approximately 919,001 according to the 2016 census, and consists of nine administrative sub-districts and 20 neighborhoods. Data were collected from documentary, statistical, and planning sources and processed in a GIS environment.
The indicators were organized into four dimensions: economic, demographic-social, physical-residential, and urban land-use/per-capita characteristics. Economic indicators included land price, housing ownership and tenancy, unemployment, employment, economic activity, and occupational structure. Demographic-social indicators included migration, higher-education attainment, household size, household density, and population density. The physical-residential dimension included residential unit size, number of rooms, villas, and apartments. The urban land-use dimension encompassed roads, urban facilities, religious uses, parks and green spaces, cultural and artistic functions, health services, sports, commercial uses, administrative and security functions, transportation and warehousing, residential land, military uses, industrial land, vacant land, general education, and higher education.
First, a decision matrix was constructed for the 20 neighborhoods. The data were standardized, and Shannon entropy was used to determine indicator weights. The weighted indicators were then entered into the TOPSIS model. TOPSIS ranks neighborhoods according to their relative closeness to a positive ideal solution and distance from a negative ideal solution. The final coefficients were used to rank neighborhoods and construct spatial representations of heterogeneity.
Global Moran’s I was subsequently calculated to examine whether similar values were spatially concentrated. Because Moran’s I provide only a global measure, the Getis-Ord Gi* statistic was also employed to identify statistically significant local hot and cold spots. This combination enabled the study to distinguish between general spatial tendency and specific local concentrations.

Results and Discussion
The TOPSIS results reveal substantial multidimensional differentiation among the 20 neighborhoods of District 4. The final composite index shows that high levels of heterogeneity are concentrated in a limited number of neighborhoods rather than being uniformly distributed. Qasemabad–Deh Narmak recorded the highest composite coefficient (0.6209), followed by Shahrak-e Valiasr–Shian–Lavizan (0.5807) and Qanat-e Kosar (0.5798). Pasdaran and Majidabad also recorded relatively high values, with coefficients of 0.5447 and 0.5445. Javadieh and Khak-Sefid had the lowest coefficients, 0.0984 and 0.0991, respectively. These low values do not necessarily indicate favorable balance; they may instead reflect relative homogeneity at a lower level of development.
Dimension-specific results also show that the pattern cannot be reduced to a simple north-south gradient. Narmak ranked highest in the demographic-social dimension, while Khak-Sefid and Javadieh occupied the lower positions. In the physical-residential dimension, Qasemabad–Deh Narmak ranked first with a coefficient of 0.685, followed by Pasdaran with 0.660, whereas Javadieh ranked last. These differences reflect the uneven distribution of residential forms, dwelling sizes, and building characteristics.
The urban land-use dimension followed a more localized and node-based pattern. Some neighborhoods functioned as specialized concentrations of administrative, health, commercial, educational, and other activities. Therefore, high heterogeneity in this dimension should not automatically be interpreted as lack of access; it may instead indicate functional specialization and uneven distribution of particular uses.
The economic dimension also exhibited considerable differentiation. Pasdaran and Mehran occupied prominent positions associated with high land values and particular occupational characteristics, whereas Javadieh and Khak-Sefid were at the lower end of the ranking. Narmak and Tehranpars occupied intermediate positions, indicating that economic conditions reflect different combinations of land values, housing tenure, employment, occupational structure, and economic opportunities.
The Global Moran’s I analysis produced a value of 0.267, with a z-score of 1.818 and a p-value of 0.069. Thus, the result is not statistically significant at the conventional 95% confidence level, but it provides weak evidence of positive spatial autocorrelation at the 90% confidence level. The relatively modest value of Moran’s I indicates that spatial clustering is present only to a limited degree and should not be interpreted as a strongly continuous spatial pattern.
The local Getis-Ord Gi* analysis identified Majidabad as a statistically significant hot spot at the 95% confidence level, indicating a local concentration of relatively high heterogeneity values. Tehranpars and Javadieh were identified as significant cold spots at the same confidence level. Most other neighborhoods did not form statistically significant clusters. Thus, TOPSIS establishes the relative position of each neighborhood, whereas Gi* determines whether high or low values are spatially concentrated.
From a theoretical perspective, the results are consistent with the interpretation of urban space as socially produced and materially differentiated. Lefebvre’s perspective helps explain the coexistence of differentiated neighborhoods, specialized urban nodes, and relatively homogeneous low-value areas as outcomes of different forms of spatial production. Harvey’s emphasis on land value and capital accumulation provides a useful framework for interpreting strong economic and physical differentiation in neighborhoods such as Pasdaran and Qasemabad. Smith’s concept of uneven development is particularly relevant to Javadieh and Khak-Sefid, where low heterogeneity may represent persistent homogeneity in disadvantage rather than spatial equilibrium.
At the same time, the statistical results do not permit a direct causal conclusion that capital, land rent, or power relations alone produce the observed configuration. The methods identify associations, relative differences, and spatial concentrations rather than directly testing causal mechanisms. The theoretical interpretation should therefore be understood as an explanatory framework consistent with the empirical pattern rather than a direct causal test.
Overall, spatial-physical heterogeneity in District 4 is multidimensional, locally differentiated, and associated with the uneven organization of economic, demographic-social, physical-residential, and land-use characteristics. Similar statistical values may arise from different urban conditions, which highlights the importance of considering both the magnitude and composition of spatial differences.

Conclusion
This study examined spatial-physical heterogeneity across the 20 neighborhoods of District 4 of Tehran Municipality by integrating multidimensional TOPSIS assessment with spatial autocorrelation and local hot spot analysis. The results show that heterogeneity is not evenly distributed and cannot be adequately explained by a simple north-south division. Qasemabad–Deh Narmak, Shahrak-e Valiasr–Shian–Lavizan, and Qanat-e Kosar recorded the highest composite values, while Javadieh and Khak-Sefid recorded the lowest. These low values should not automatically be interpreted as spatial balance, but may indicate homogeneity at a lower level of development.
The Global Moran’s I result (I = 0.267; z = 1.818; p = 0.069) indicates weak positive spatial autocorrelation, with evidence of clustering at the 90% confidence level but not at the 95% level. Gi* analysis identified Majidabad as a statistically significant hot spot and Tehranpars and Javadieh as significant cold spots at the 95% confidence level. Findings indicate a fragmented and locally differentiated spatial structure in which heterogeneous and relatively homogeneous neighborhoods coexist.
The study demonstrates the value of combining multidimensional decision-making with spatial statistics. TOPSIS provides comparative measurement of neighborhood differences, while Moran’s I and Gi* reveal the spatial organization and statistical significance of those differences. From a planning perspective, neighborhoods with persistent low values require attention to basic urban conditions and service provision, while neighborhoods characterized by strong differences among dimensions require greater coordination among physical, economic, social, and land-use processes. The findings should, however, be interpreted in light of several limitations, including the use of 2016 census information, the cross-sectional nature of detailed planning data, the use of asking rather than transaction-based land prices, and the absence of formal causal testing. Future research should employ longitudinal datasets, verified transaction prices, and spatial econometric methods.

Funding
This article was conducted with the financial and intellectual support of the Center for Studies and Planning of Tehran Municipality.

Authors’ Contribution
All authors have contributed equally to this article.
Conflict of Interest
Authors declared no conflict of interest

Acknowledgments
We are grateful to all the scientific consultants of this paper.