The Relationships Between Land Use Characteristics, Neighbourhood Perceptions, Socio-Economic Factors and Travel Behaviour in Compact and Sprawled Neighbourhoods in Windhoek

Abstract

This study examines how Windhoek’s urban form, shaped by apartheid-era planning, continues to influence neighbourhood travel behaviour, socio-economic disparity, and residential perceptions. It addresses three key questions: (1) How do socio-economic characteristics, neighbourhood perceptions, and travel patterns differ between compact and sprawled areas? (2) Which socio-economic, perceptual, and spatial factors are associated with the likelihood of neighbourhood-based shopping in compact versus sprawled urban forms? (3) What are the determinants of entertainment and recreational travel behaviour within neighbourhoods across the two urban forms? Using survey data from 1000 residents, the analysis employs chi-square tests, Mann–Whitney U tests, binary logistic regression, and multivariate regression models. Findings reveal that compact areas, characterised by higher incomes, stronger place attachment, and greater infrastructural diversity, support more frequent neighbourhood travel. By contrast, sprawled peripheries, despite higher population densities, remain marked by socio-economic marginalisation, limited amenity access, and negative perceptions that constrain neighbourhood mobility. Across both forms, long-term residence and belonging strongly predict neighbourhood travel, while concerns over traffic safety and crime consistently suppress participation. The results show that spatial proximity alone does not ensure accessibility; emotional, perceptual, and structural barriers mediate neighbourhood mobility. The study highlights the need for integrated planning that addresses both physical infrastructure and lived experience to advance equitable and sustainable mobility in post-colonial contexts.

1. Introduction

Urban neighbourhoods are more than just clusters of homes and roads, which are living environments in which social life and spatial design come together to shape how people live, interact, and get around. An important but often overlooked aspect of urban mobility is non-commuting travel, trips made for shopping, leisure, visiting friends, or running errands [1,2]. These daily neighbourhood trips reveal how built environments, social factors, and personal perceptions, such as how safe or accessible a neighbourhood feels and interacts to influence how people move through their communities [3,4]. While commuting travel typically dominates urban mobility research, non-commuting trips constitute a significant component of daily travel patterns and offer critical insights into the design of inclusive, accessible, and socially responsive neighbourhoods.

Across global contexts, built environment characteristics, particularly the distinction between compact and sprawled urban forms, are closely associated with differences in travel behaviour. Compact neighbourhoods, which are defined by higher density, mixed land use, and well-connected street networks, are more likely to support non-motorised neighbourhood travel [5,6,7]. These environments enhance neighbourhood accessibility and reduce the dependence on private vehicles by integrating residential, commercial, and recreational amenities within walkable distances. In contrast, sprawled neighbourhoods, often characterised by low-density development, land use segregation, and limited connectivity, tend to foster car dependency and longer travel distances, reducing the feasibility and appeal of intra-neighbourhood trips [8]. These spatial attributes do not act in isolation; they interact with socioeconomic factors such as income, education, and car ownership, as well as subjective perceptions of neighbourhood safety, inclusiveness, and amenity attractiveness. Together, these dimensions create complex and often unequal patterns of access and mobility at the neighbourhood scale.

In cities of the Global South, particularly across Sub-Saharan Africa, these patterns are further shaped by rapid urbanisation, infrastructural fragmentation, and enduring spatial inequalities [9,10]. Unplanned low-density expansion at the peripheries of many African cities has produced sprawling residential areas with limited access to services and public transport, often requiring long or costly trips to meet daily needs [11,12]. Meanwhile, more centrally located, historically planned neighbourhoods tend to display features of compact urbanism [13]. Despite the widening spatial contrast between compact and sprawled neighbourhoods, few studies have empirically examined how the form and function of neighbourhoods in African cities affect non-commuting travel patterns. Even fewer have employed statistical approaches to analyse how socio-economic conditions, land use attributes, and resident perceptions jointly shape non-commuting travel behaviour.

Windhoek, Namibia’s capital, offers a compelling setting for this investigation. Like many Southern African cities, Windhoek reflects a deeply divided spatial structure shaped by apartheid-era planning, in which low-income communities were placed on the urban periphery, far from employment centres and formal services [14]. These spatial inequalities are not merely contemporary socioeconomic divides but are rooted in the institutional design of apartheid urban planning, which deliberately placed low-income Black communities in peripheral, poorly serviced areas while reserving central, amenity-rich neighbourhoods for affluent populations [14,15]. This historical structuring continues to shape the city’s mobility outcomes, as access to neighbourhood amenities and travel opportunities remains closely tied to these entrenched spatial patterns. Today, the city features both compact, mixed-use neighbourhoods, where residents may access shops, parks, and social venues within walking distance, and expansive residential zones with limited neighbourhood infrastructure [16]. In sprawled areas, residents often rely on informal transport modes or travel significant distances to access basic amenities. Meanwhile, in compact neighbourhoods, despite physical proximity to services, factors such as safety concerns or perceived affordability may constrain neighbourhood mobility [17]. These contrasting urban forms make Windhoek a unique case for understanding the associations between land use, socioeconomic context, neighbourhood perceptions, and non-commuting travel behaviour.

While studies in Windhoek have explored issues such as transport inequality [16], pedestrian safety and walkability [18], and spatial disparities in mobility access [17], existing literature largely focuses on infrastructural limitations, policy frameworks, and modal availability. What is notably absent, however, is a disaggregated understanding of how nonwork travel decisions are shaped within and across different neighbourhood types. Despite growing international interest in neighbourhood-scale mobility, particularly in relation to concepts like the “15-min city” [19], there is little empirical evidence on how residents in African cities like Windhoek perceive and utilise their own neighbourhoods for short-distance, non-commuting trips such as shopping and entertainment. Moreover, existing studies in emerging African cities rarely apply a comparative urban form framework that distinguishes between compact and sprawled environments, limiting our ability to assess how built form interacts with socioeconomic status and perceived neighbourhood quality to shape neighbourhood travel behaviour. This lack of neighbourhood-level analysis represents a critical gap in both the regional literature on Southern African cities and the global agenda on sustainable and inclusive urban mobility.

This study draws conceptually on the 15-min city model popularised by Carlos Moreno [19,20], which emphasises accessibility and neighbourhood mobility within walkable neighbourhoods. It explores the extent to which Windhoek’s compact and sprawled neighbourhoods support such short-distance, non-work trips, and how social and spatial inequalities shape this potential.

To address these gaps, this study examines how neighbourhood form, specifically the contrast between compact and sprawled urban environments, is associated with patterns of neighbourhood travel behaviour in Windhoek, Namibia. It investigates how socio-economic characteristics, residents’ perceptions of their neighbourhoods, and mobility patterns are correlated with the likelihood and frequency of non-commuting trips, such as those undertaken for shopping and entertainment. By focusing on neighbourhood-scale mobility within a spatially diverse urban structure, this study seeks to generate a comprehensive understanding of how urban form experiences are linked to everyday travel behaviour, particularly across different socio-spatial contexts.

In doing so, the study provides a contextually grounded contribution by drawing on original empirical data from Windhoek to demonstrate how compact and sprawled neighbourhoods generate distinct patterns of neighbourhood mobility. By positioning non-commuting trips as a critical analytical lens, it responds to growing calls for context-sensitive approaches to urban mobility, particularly within the understudied realities of African cities.

These findings are expected to inform both academic research and urban policies. For scholars, this study extends the neighbourhood travel behaviour literature by embedding African urban realities into a field still largely dominated by research from high-income contexts. For policymakers and planners, the results provide actionable insights for designing neighbourhoods that support equitable, accessible, and sustainable mobility, recognising that enabling neighbourhood travel is as important as facilitating city-wide connectivity. Improving liveability in both compact and sprawled areas will require not only physical interventions, such as street networks and land-use planning, but also attention to the social perceptions and constraints that shape residents’ everyday mobility choices.

The remainder of this paper is organised as follows: Section 2 reviews the relevant literature on neighbourhood form, neighbourhood travel behaviour, and socio-spatial perception. Section 3 outlines the study’s methodology, including case selection, data and methods, and modelling strategies. Section 4 presents empirical results, highlighting the key differences and correlations between variables. Section 5 discusses these findings in the context of broader debates on transportation and urban planning. Section 6 concludes the study with policy recommendations and directions for future research.

2. Literature Review

Recent urban mobility research has shifted from city-wide transport systems toward neighbourhood-scale analysis, emphasising how spatial configurations enable or constrain everyday movement [21,22]. Rather than treating land use as a static background, urban form is increasingly recognised as a structural determinant of behavioural options, particularly for non-work trips such as shopping, leisure, and social visits. These discretionary journeys are more sensitive to the quality and comfort of the built environment, making neighbourhood form a critical influence.

Evidence from both Global North and South contexts shows that compactness, land-use diversity, and local amenity provision strongly affect non-commuting travel patterns [1,6,23,24,25]. In Sweden, ref. [26] found that proximity to essential services, especially grocery stores, increases walking and cycling, with benefits peaking when six to ten key amenities are accessible. Similarly, ref. [27] observed across five European countries that residents in amenity-rich, walkable areas use active modes more frequently for both commuting and non-work travel. These findings support the premise that well-serviced neighbourhoods can foster sustainable, short-distance mobility.

In the Global South, however, spatial fragmentation and infrastructural inequities complicate this relationship. In Bogotá, low land-use diversity has been linked to longer walking distances and restricted service access for low-income residents [28]. Additionally, ref. [29] reported that while compact urban form in Iranian cities supports neighbourhood trips, sprawled neighbourhoods encourage car dependence. These studies highlight that compactness alone does not guarantee accessibility; the quality, affordability, and safety of local infrastructure are equally important.

Non-commuting travel patterns also reveal deeper socio-spatial inequalities in Sub-Saharan Africa. Access to local amenities is mediated by socioeconomic factors such as income, gender, education, and car ownership. In Ghana, ref. [30] found that low-income residents rely more on walking, particularly in older, accessible city cores, while suburban, higher-income groups exhibit greater car dependency. Informal transport systems often fill service gaps but can impose financial and logistical burdens, as shown in African cities where fragmented paratransit networks increase costs for low-income users [31]. Gendered constraints are also significant: in India, ref. [32] demonstrated that safety concerns reduce women’s discretionary travel, even when services are nearby, a pattern echoed in global studies [32,33].

In post-colonial African contexts, including Johannesburg, Cape Town, Nairobi and Maputo, recent studies have demonstrated how apartheid and colonial spatial regimes institutionalised peripheral displacement and uneven infrastructural investment, producing enduring mobility inequities [12,34,35]. Situating Windhoek within this body of work highlights that differences in neighbourhood mobility are not merely the outcome of present-day income disparities or population densities but are deeply rooted in historically embedded spatial segregation and its continuing structural effects. This perspective highlights the need to analyse how institutional legacies, alongside contemporary socioeconomic conditions, structure access, opportunity, and everyday travel behaviour.

Perceptions of safety, comfort, and inclusiveness often influence mobility more strongly than physical distance. Ref. [36] found that subjective evaluations of neighbourhoods, such as perceived safety or pleasantness, shape travel decisions independently of actual proximity. Research in South African cities [37] and other African contexts [11,38,39] confirms that crime fears and poor infrastructure can limit use of public space, affecting women and youth. Ref. [40] term this “captive mobility,” where individuals are constrained to suboptimal modes by safety or affordability concerns.

Despite the importance of these perceptual and socio-economic dimensions, they are rarely integrated into transport modelling or planning frameworks in the African context. Existing studies remain skewed towards high-income countries and often prioritise commuting patterns and formal service provision, neglecting discretionary trips and neighbourhood-scale experiences. This leaves a knowledge gap in understanding how urban form, infrastructure quality, and socio-economic factors intersect to shape everyday mobility, particularly in structurally unequal cities. In Windhoek, mobility outcomes remain closely tied to apartheid-era spatial design, which continues to shape the distribution of amenities as well as residents’ lived experiences of accessibility. Addressing this gap is essential for both advancing academic debates and designing urban environments that are equitable, sustainable, and inclusive.

The 15-Min City and Broader Theoretical Perspectives on Neighbourhood Mobility

The 15-min city model promotes urban environments in which residents can meet most daily needs within a short walk or cycle, structured around the principles of density, proximity, diversity, and digitalisation [20]. Although developed in high-income, post-pandemic contexts, its implementation across varied geographies demonstrates both its adaptability and the persistent challenge of ensuring equitable accessibility. In Italy, ref. [41] applied the Next Proximity Index in Ferrara and Bologna to map walkable access to essential services, revealing peripheral deficits in healthcare and green space provision. In Naples, ref. [42] introduced a multidimensional accessibility measure incorporating terrain, network geometry, and socio-economic factors, highlighting that apparent spatial proximity often overstates effective access. Ref. [42], provided a conceptual framework linking density, diversity, and design to proximity, cautioning against the risks of gentrification and suburban exclusion. At the global scale, ref. [43] reviewed applications in cities including Paris, Melbourne, Copenhagen, Milan, Rome, Genoa, Sydney, and several in the Netherlands, concluding that benefits such as socio-economic vitality, reduced emissions, and enhanced social cohesion depend on decentralised service provision, sustainable mobility systems, and participatory governance, shaped by local urban form and institutional capacity. These applications demonstrate its relevance but also reveal structural and socio-economic constraints that complicate its realisation in contexts marked by inequality, infrastructural fragmentation, or entrenched spatial segregation.

To move beyond the understanding of spatial form, the 15-min city can be situated within three complementary theoretical perspectives: accessibility theory, the capability approach, and urban justice. Together, these frameworks provide a multi-dimensional lens for understanding how opportunities are distributed, experienced, and contested.

Accessibility theory, theory, as advanced by [44] conceptualises accessibility as “the potential of opportunities for interaction,” linking land-use patterns and transport systems to the ease with which activities can be reached. Subsequent refinements [45,46], distinguish between locational accessibility, the spatial distribution of opportunities and the transport infrastructure enabling them, and individual accessibility, shaped by personal circumstances such as income, physical ability, or social capital. Studies in Lisbon [47] and Toronto [48] show that accessibility gaps persist even in well-connected areas due to affordability, safety, or service quality constraints. Complementary evidence from northern European cities [49] and Hamburg [50] demonstrates that perceptions of safety, comfort, and service reliability often mediate the translation of potential into realised accessibility, highlighting the importance of integrating subjective dimensions into accessibility assessment.

The capability approach, advanced by [51,52], extends the discussion from access to the freedoms individuals must convert available opportunities into valued activities. Applied to mobility, this approach examines the “conversion factors”, economic resources, safety, time, and cultural norms that enable or constrain the use of proximate amenities [53,54]. For instance, recent analyses in New York indicate that low-income populations experience reduced transit convenience, characterised by longer waiting times, extended travel durations, and increased transfers, despite residing in areas with high service coverage [55]. Likewise, spatial assessments in Milan reveal that socially vulnerable neighbourhoods frequently coincide with low levels of public transport accessibility, illustrating that high service density does not inherently ensure equitable mobility outcomes [56]. Studies in Delhi [57], Abuja, Cape Town and Tunis [58] illustrate how gendered safety concerns, fare affordability, and care-related travel responsibilities limit the ability to benefit from nearby services. These findings reinforce the principle that physical provision of infrastructure does not necessarily equate to functional access.

Urban justice frameworks situate these debates within broader normative concerns about equity, rights, and the democratic shaping of urban space. Ref. [59] in revisiting Lefebvre’s “right to the city,” frames mobility as a central means through which urban citizenship is claimed, emphasising inhabitants’ rights to both shape and access urban life. Similarly, ref. [60] “Just city” framework advances the principles of distributive equity, diversity, and participatory governance, underscoring the need for urban policy to address not only the spatial distribution of resources but also the political processes through which they are allocated. Recent mobility research has applied urban justice frameworks to evaluate whether transport and land-use policies alleviate or perpetuate spatial inequities [61,62]. Evidence from Latin America [63] and Amsterdam [64] shows that infrastructure expansion or cycling promotion alone cannot deliver equity without explicit attention to benefit distribution and participatory processes. In China, ref. [65] demonstrate how transport network developments can entrench socio-spatial segregation, while [66] Qiao finds that Mobility-on-Demand services may exacerbate disparities in the absence of targeted policy interventions. Parallel evidence from African cities supports these findings, ref. [67] develops a transport justice-oriented planning framework in Kigali and Blantyre to address inequitable accessibility outcomes; Ref. [68] reveal persistent policy blind spots toward women. Together, these studies highlight that equitable mobility requires more than network efficiency; it demands distributive justice, inclusive governance, and the structural integration of historically excluded groups.

Integrating these perspectives with the 15-min city framework strengthens its analytical scope. Accessibility theory reveals the structural relationships between urban form and opportunity distribution; the capability approach explains how socio-economic, cultural, and perceptual factors shape the translation of proximity into use; and urban justice embeds these features in a rights-based discourse on equity and inclusion. Applied together, they offer a robust conceptual foundation for evaluating whether neighbourhood-scale planning initiatives can transcend physical form to deliver equitable, inclusive, and sustainable mobility outcomes across diverse urban contexts.

Despite growing evidence that neighbourhood form influences non-commuting mobility, notable gaps persist in the literature, particularly in the Global South. Empirical research is predominantly concentrated in high-income contexts, limiting the applicability of existing mobility models to African cities, where socio-spatial fragmentation and historical legacies produce distinct travel behaviours. Much of the existing work focuses narrowly on physical proximity and infrastructure provision, often overlooking socio-economic, perceptual, and justice-related factors, such as safety, place attachment, affordability, and inclusion, that shape realised accessibility. Moreover, while the 15-min city framework provides a valuable lens for assessing spatial proximity and service access, its assumptions are seldom examined within resource-constrained and structurally unequal urban environments. Comparative analyses of compact and sprawled urban forms are rare, resulting in a limited understanding of how different spatial configurations influence discretionary travel across diverse social groups. Addressing these gaps is essential for assessing whether neighbourhood-scale planning models such as the 15-min city can achieve equitable, inclusive, and sustainable mobility outcomes in cities marked by persistent inequality and spatial injustice.

To synthesise the key theoretical and empirical contributions informing this study,

Table 1. Summary of Key Theoretical and Empirical Contributions to the Study.

Author(s)	Study Context	Main Findings	Relevance to the Present Study
[21]	USA (multiple cities)	Conceptual paper linking built environment and travel behaviour; emphasises that land use influences travel decisions through accessibility, safety, and perception.	Offers foundational theory connecting land use with neighbourhood travel behaviours.
[22]	Germany (Dresden)	Developed a population-weighted accessibility index using GIS to measure proximity to social amenities in urban regeneration areas.	Provides a methodological example of accessibility analysis in disadvantaged urban settings.
[23]	USA (scholarly response)	Responds to critiques suggesting compact development has a limited effect; argues strongly that D-variables do impact travel behaviour.	Reflects the importance of urban form in reducing car dependency, supporting this study’s core argument.
[24]	Meta-analysis of U.S. studies	Confirms that compact urban form significantly reduces vehicle miles travelled (VMT); supports the “5 D’s” (density, diversity, design, destination accessibility, distance to transit).	Empirical support for the claim that compact cities support sustainable travel.
[25]	Egypt (Greater Cairo Region)	Comparative study shows walkability is higher in historic core vs. new extensions; attributes it to compact urban form and integrated networks.	Illustrates differences between old and new urban forms in the Global South, supporting this study’s contextual relevance.
[26]	Sweden (Västra Götaland)	Finds that the presence and variety of local amenities influence walking and cycling; identifies nonlinear effects on travel mode choice.	Demonstrates the importance of amenity type and supply in promoting neighbourhood travel behaviour.
[27]	France, Belgium, Hungary, UK, Netherlands (multi-city)	Found that proximity to amenities is positively associated with physical activity and walking, but socioeconomic disparities affect access.	Highlights how built environment features interact with social equity, reflecting the importance of inclusive urban design.
[28]	Bogotá, Colombia	Identified that lower-income residents experience worse accessibility to opportunities, especially in peri-urban areas, due to spatial mismatch and limited transport.	Offers a Global South perspective on accessibility inequality, relevant for understanding access disparities in urban fringe areas.
[29]	Iran (Hamedan & Nowshahr)	Urban sprawl correlates with increased car use and reduces public transport, but patterns differ by city size and socioeconomic factors.	Provides empirical evidence of how sprawl affects travel behaviour differently in large vs. small Middle Eastern cities.
[30]	Ghana (Kumasi Metropolis)	Reveals that polycentric urban structure affects commuting patterns; suburban residents and higher-income earners are more likely to use private transport.	Demonstrates how spatial structure influences travel mode choice in African contexts, relevant for analysing socio-spatial disparities.
[31]	Sub-Saharan Africa (overview)	Reviews paratransit systems (e.g., matatus, tro-tros) as the dominant transport mode in African cities; highlights challenges of informality, lack of regulation, and reform needs.	Contextualises informal transport systems common in the Global South, important for planning inclusive mobility in rapidly urbanising areas.
[32]	Developing countries	Found that female commuters are willing to pay more to avoid harassment and safety concerns on public transport. Highlights the economic cost of unsafe transport environments.	Demonstrates the link between gendered safety concerns and travel decisions, critical for analysing equity in urban transport.
[36]	Thailand (Bangkok Metropolitan Region)	Used deep learning and semantic segmentation of street images to measure perceived vs. objective road environment; visual elements like nature, vehicles, and infrastructure shaped perception.	Innovative methodology combining AI and urban perception analysis; relevant for understanding the built environment’s subjective impacts.
[37]	South Africa (Johannesburg)	Evaluated non-motorised transport (NMT) infrastructure; found that despite good physical condition, usage was low due to safety and security concerns.	Highlights the gap between infrastructure provision and actual user uptake due to perceived insecurity, useful for urban design policy.
[38]	Tunisia (Grand Tunis)	Explores gendered mobility exclusion through women’s travel diaries; identifies how insecurity, poverty, and transport design marginalise women in mobility systems.	Critical perspective on intersectional transport exclusion, relevant for mobility justice and inclusive planning in the urban Global South.
[39]	Maputo, Mozambique	Found significant mobility challenges in peri-urban areas due to spatial inequalities, lack of infrastructure, and long commuting times, especially for women and low-income groups.	Provides crucial insights on transport exclusion in African cities, aligns with equity concerns in urban planning.
[40]	Bangladesh	Gender-based violence and safety concerns significantly restrict women’s mobility and mode choice.	Highlights how safety perceptions shape travel behaviour, supports the inclusion of safety and sense of belonging as key variables in the Windhoek travel behaviour study.
[41]	Italy (national scale)	Introduced the Next Proximity Index (NEXI), a scalable tool to measure 15-min accessibility using open data. Found variability in proximity within and between cities.	Presents a replicable method to evaluate spatial accessibility, with potential application in comparing international 15-min cities.
[42]	Naples, Italy	Evaluated urban morphology, street design, and service distribution in creating 15-min neighbourhoods. Found that geomorphology and socio-economic factors are key to walkability.	Offers a methodology to assess neighbourhood walkability and accessibility, supports spatial equity in 15-min city design.
[43]	Europe (multi-city review)	Emphasised the shift from top-down to bottom-up approaches in 15-min city planning, focusing on participatory urbanism and community engagement.	Supports a participatory model of mobility planning, relevant for integrating neighbourhood needs in urban accessibility strategies.
[44]	U.S., early empirical urban planning studies on land use	Introduced a model linking accessibility and land use; showed residential development correlates with accessibility to employment and other opportunities.	Provides foundational theory for understanding how accessibility shapes urban form; it forms the basis of many modern accessibility measurement approaches.
[45]	Dutch cities, multimodal accessibility measurement	Developed a framework for evaluating accessibility using land use, transport, temporal, and individual components. Emphasised the need for person-based measures and policy integration.	Supports multidimensional understanding of accessibility; aligns with the study’s focus on equity and inclusive mobility outcomes.
[47]	Lisbon, Portugal	Developed a method for quantifying “disability-induced accessibility disparity” using pedestrian network data. Found significant barriers in the built environment that reduce mobility for disabled persons.	Highlights accessibility inequities among disabled populations; supports the study’s inclusion of social vulnerability and universal design perspectives.
[48]	Canada (national-scale study across 8 metro areas)	Mapped transport poverty by analysing income levels and transit accessibility. Identified nearly 1 million Canadians in low-income households with poor transit access, mostly in low-density suburbs.	Offers empirical evidence on the intersection of accessibility and income inequality; it directly supports this study’s emphasis on spatial equity and transport justice.
[49]	5 European cities (Sweden, Norway, Denmark and Finland	Found strong links between public transport use and subjective well-being, particularly through travel satisfaction. Emphasised the importance of perceived accessibility in enhancing the quality of life.	Supports broader evaluation of accessibility beyond physical availability, supporting inclusion of subjective and experiential dimensions in accessibility assessment.
[50]	Germany	Explored older adults’ access to public transport and found that lack of perceived safety and usability are key barriers. Emphasised the importance of age-sensitive and inclusive transport planning.	Reinforces the study’s attention to social vulnerability (elderly users) and the importance of equity-focused urban mobility design.
[51]	Theoretical work; global applicability	Provides a comprehensive survey of the capability approach, emphasising capabilities (substantive freedoms) over resources or utilities. Distinguishes between means and ends in evaluating well-being.	Offers theoretical grounding for capability-informed accessibility analysis, central to this study’s equity and justice framing.
[52]	Conceptual: Capability Approach and Human Rights	Explores synergy between human rights and the capability approach; advocates for policy frameworks that integrate dignity, freedom, and substantive equality as foundational values.	Bolsters the normative justification for capability-based approaches to accessibility and inclusion, particularly regarding rights-based urban policy frameworks.
[53]	United Kingdom	Reviews how transport disadvantage contributes to broader social exclusion. Highlights relational, multidimensional, and dynamic aspects of transport poverty and inequity.	Underpins key conceptual framework for this study; Lucas’ model of transport-related exclusion is foundational to policy framing on equitable access and transport justice.
[54]	Reviews of theories	Argued that accessibility must be reframed as a social equity issue, not merely a transport or land-use problem. Highlighted social exclusion risks in unequal urban accessibility.	Strengthens theoretical grounding for equity-driven accessibility analysis, aligning with the capability and justice approach in this study.
[55]	United States (New York City)	Found significant spatial and racial disparities in subway accessibility, particularly affecting Black communities in Queens. Used the gravity index and Gini coefficients for analysis.	Provides a methodologically robust example of racial and spatial equity analysis, useful for comparative metrics in urban transport justice.
[56]	Italy	Proposed a composite accessibility indicator integrating public transport, services, and walking access to assess social equity in urban settings.	Offers a comprehensive methodology for measuring multimodal accessibility in urban equity evaluations.
[57]	India (New Delhi)	Highlighted how informal transport enables spatial and social mobility for marginalised women facing cultural and safety barriers.	Informs the intersection of gender, informality, and accessibility in low-income urban settings—directly supporting this study’s justice lens.
[58]	Nigeria, South Africa, Tunisia	Revealed the gendered dimensions of urban transport exclusion through women’s experiences of harassment and scheduling constraints. Emphasised peer-led research.	Supports gender-sensitive accessibility approaches and inclusion of lived experiences in transport justice analysis.
[59]	Theoretical analysis	Critiqued the concept of “the right to the city” in neoliberal urbanism; highlighted risks of co-option and the need for participatory democratic control.	Informs critical framing of rights-based approaches in urban accessibility debates, especially under neoliberal planning regimes.
[60]	New York, London, Amsterdam	Proposed “the just city” framework combining equity, democracy, and diversity. Challenged utilitarian and market-based planning paradigms.	Establishes a strong normative basis for urban justice frameworks that intersect with transport equity.
[61]	Conceptual framework, Mobility Data Justice	Explored the social dynamics and contested nature of digital mobility platforms (e.g., MaaS), including issues of surveillance, access, and inequality.	Highlights how digital mobility innovations can both enable and restrict urban accessibility, important for equity analyses in future mobility.
[62]	Conceptual: Transport and mobility justice:	Reviewed developments in transport and mobility justice theory. Emphasised shift from state-centric to society-centric approaches and the need for inclusive epistemologies.	Frames the study’s justice lens, encouraging attention to grassroots knowledge, multiple actors, and community-led planning practices.
[63]	18 Latin American cities	Found that bicycle use has increased across socio-economic groups, especially among the highly educated and high-income groups, although traditionally cycling was more common among lower-income populations.	Demonstrates how transport mode preferences evolve with infrastructure and cultural shifts; relevant for equity-focused active mobility policy.
[64]	Amsterdam, The Netherlands	Higher density, land-use mix, and street connectivity are linked to more cycling, with urban form remaining a strong predictor even after accounting for socio-demographic factors	Provides empirical evidence that supports the importance of compact, connected, and mixed-use urban design in shaping active travel choices.
[65]	Beijing, China	Analysed spatial mismatch between jobs and housing and how metro accessibility affects commuting burden; found access varies by income group.	Informs accessibility analyses by highlighting how transit systems unevenly serve socio-economic groups.
[66]	Chengdu, China	Compared shared mobility (SM) and Mobility as a Service (MaaS) in improving spatial justice; MaaS was found to be more equitable.	Offers empirical support for prioritising integrated multimodal approaches (MaaS) over fragmented services for urban equity.
[67]	Kigali (Rwanda), Blantyre (Malawi)	Proposed “transport justice” framework for Sub-Saharan Africa; highlighted severe inequality in car access and opportunity.	Grounds the study in Global South realities, reinforcing equity-based planning tailored to marginalised urban populations.
[68]	South Africa	Assessed disability inclusion in African transport policies; found systemic neglect leading to isolation of persons with disabilities.	Emphasises inclusive transport policy as foundational for equitable access, particularly for people with disabilities.

summarises the main findings, geographic contexts, and relevance of selected studies and frameworks.

3. Research Methodology

3.1. Research Questions

This study examines how neighbourhood forms, contrasting compact and sprawled environments, interact with socio-economic characteristics, resident perceptions, and travel behaviour in Windhoek. It addresses three research questions: (i) What differences exist in socio-economic characteristics, neighbourhood perceptions, and neighbourhood travel behaviour between residents of compact and sprawled areas? (ii) How are socio-economic features, travel behaviour, and residents’ perceptions associated with shopping trips within each urban form? (iii) How are socio-economic characteristics, residential perceptions, and travel behaviour associated with the nature and frequency of entertainment and recreational trips? These questions are framed by accessibility theory, the capability approach, and urban justice, enabling the analysis to move beyond measures of spatial proximity to incorporate socio-economic constraints, lived experiences, and equity in mobility outcomes.

3.2. Case Study

Windhoek, Namibia’s capital, with approximately 495,000 residents [69], offers a pertinent context given its enduring spatial inequalities rooted in apartheid-era planning. Historically marginalised groups were relocated to peripheral townships with limited infrastructure and services, while affluent, historically advantaged populations were concentrated in centrally located, well-serviced areas.

The study area comprises nine constituencies: John Pandeni, Katutura Central, Katutura East, Khomasdal, Moses ǁ Garoëb, Samora Machel, and Tobias Hainyeko represent sprawled, historically disadvantaged neighbourhoods, while Windhoek East and Windhoek West represent more compact, affluent areas. In this context, population density alone does not equate to functional compactness: peripheral areas may record high densities due to overcrowding in informal settlements but lack walkable street networks, mixed land uses, and proximity to amenities, whereas planned compact neighbourhoods may exhibit lower densities yet offer greater connectivity and accessibility. Figure 1 shows the study area of Windhoek, Namibia.

Table 2. Area, population density, and socio-economic profile of Windhoek constituencies.

Constituency	Area in km2	Persons per km2	Socio-Economic Profile
John Pandeni	3.18	8010.7	Historically disadvantaged, overcrowded, low-income housing
Katutura Central	2.53	12,054.2	Very high density, informal settlements, limited services
Katutura East	4.36	5267.3	Peripheral, predominantly low-income, weak amenity access
Khomasdal	23.52	2857.7	Mixed-income, fragmented infrastructure, limited retail integration
Moses//Garoeb	32.37	2129.2	High density, largely informal, poor service provision
Samora Machel	20.30	4551.1	Overcrowded, peripheral township, socio-economically marginalised
Tobias Hainyeko	18.51	3623.8	Lower-income, limited formal infrastructure, and fragmented land use
Windhoek East	218.93	137.3	Affluent, well-serviced, high land-use diversity and accessibility
Windhoek West	208.08	287.9	Middle- to high-income, planned housing, better service integration

Source: Namibia Statistics Agency (2024), 2023 Population and Housing Census Preliminary Report [69].

presents the area, population density, and socio-economic characteristics of the nine constituencies considered in this study. Although constituencies are administrative units, in Windhoek, they also capture neighbourhood-level features because of their relatively small geographic extent and unique socio-economic composition. The table illustrates the marked contrast between the sprawled constituencies, which are characterised by high residential densities, overcrowding, informal housing, and limited access to services, and the compact constituencies, which, despite lower residential densities, are more affluent and exhibit higher levels of connectivity, greater land-use diversity, and stronger access to amenities.

The distinction between compact and sprawled constituencies in Windhoek extends beyond present-day socio-economic contrasts; it reflects the spatial logic of apartheid planning, which relegated marginalised populations to peripheral, fragmented townships while supporting privilege in centrally located, well-serviced neighbourhoods. These historically entrenched spatial arrangements continue to shape contemporary patterns of accessibility and neighbourhood mobility.

3.3. Data and Variables

Primary data were collected between October and December 2024 through structured, face-to-face surveys administered by trained enumerators at selected street intersections across all nine constituencies. A stratified random sampling strategy ensured equal representation from compact and sprawled neighbourhoods, with 500 respondents surveyed in each category. Enumeration points were chosen to maximise spatial coverage and socio-economic diversity.

To ensure consistency and validity of responses, participants were eligible for inclusion if they were (i) adults aged 18 years and above, (ii) permanent residents of Windhoek with a minimum residency of one year, and (iii) available and willing to provide informed consent at the time of the survey. These criteria ensured that respondents possessed adequate familiarity with their neighbourhood context and travel patterns. Individuals were excluded if they (i) were under the age of 18, (ii) self-identified as temporary visitors or recent migrants with less than one year of residence, or (iii) were unable or unwilling to provide informed consent. This approach safeguarded the representativeness of the sample while protecting ethical standards in line with social science research protocols.

The questionnaire captured socio-economic characteristics (including income, education, and car ownership), neighbourhood perceptions (such as safety, belonging, and attractiveness of amenities), and travel behaviour (including trip frequency, destination choice, and mode use), with a particular focus on non-commuting trips. Questions on perceptions and constraints operationalised the capability approach by measuring conversion factors such as safety, affordability, and time availability, while travel behaviour and spatial accessibility measures reflected the distinction between potential and realised accessibility in accessibility theory. The urban justice perspective informed both the sampling design and the comparative analysis, ensuring representation across socio-economic groups and enabling an evaluation of equity in mobility opportunities between neighbourhood types.

Table 3. Study Variables.

Variable Name	Type	Description
Socio-Economic Characteristics
Age	Dummy	Respondents selected an age category. The categorical variable was coded as a dummy. 0 = Under 35 years, 1 = 35 years and above
Gender	Categorical	Gender of respondent 0 = Female, 1 = Male
Education	Dummy	Respondents selected their highest level of education; the categorical variable was coded as a dummy variable. 0 = Lower Education (Primary & Secondary) 1 = Higher Education (College & University)
Car Ownership in a Household	Continuous	Number of cars owned in the respondent’s household.
Driving licence	Categorical	Whether the respondent has a driving licence. 0 = Yes, 1 = No
Number of driving licenses/households	Continuous	Number of people with licenses in the household
Household Size (Adults)	Continuous	Number of adults in the respondent’s household.
Household Size (Children)	Continuous	Number of children in the respondent’s household.
Income	Dummy	Respondents selected their household income category, and responses were coded as a dummy variable. 0 = <N$5000, 1 = >N$5000
Land Use and Built Environment
Population Density	Continuous	The average number of people per square kilometre (extracted from the Namibian 2023 Population & Housing Census Report)
Length/street density around home		This variable was not self-reported but derived using Geographic Information Systems (GIS). It measures street connectivity by calculating the total length of roads within each zone, divided by the zone’s area. The computation was performed in ArcGIS Pro using road network data from OpenStreetMap (OSM)
Intersection density around home	Continuous	Number of street intersections per km2 around the respondent’s home location; derived from GIS data.
Perceived access to employment opportunities	Continuous	Agreement score (0–100)
Perceived access to educational institutions	Continuous	Agreement score (0–100)
Perceived access to healthcare facilities	Continuous	Agreement score (0–100)
Perceived access to shopping and amenities	Continuous	Agreement score (0–100)
Perceived access to parks and recreational areas	Continuous	Agreement score (0–100)
Neighbourhood Perceptions
Sense of belonging to the neighbourhood	Continuous	Agreement score (0–100)
I avoid boredom by visiting new places	Continuous	Agreement score (0–100)
There are attractive shops or shopping centres in my neighbourhood	Continuous	Agreement score (0–100)
Attractiveness of social/recreational facilities in the neighbourhood	Continuous	Agreement score (0–100)
Preference for neighbourhood entertainment	Continuous	Agreement score (0–100)
A preference to have entertainment far from the neighbourhood	Continuous	Agreement score (0–100)
No suitable shops as a reason for not shopping in the neighbourhood	Continuous	Agreement score (0–100)
No good social atmosphere as a reason for not shopping in the neighbourhood	Continuous	Agreement score (0–100)
Shops are too expensive as a reason for not shopping in the neighbourhood	Continuous	Agreement score (0–100)
Little security in the neighbourhood	Continuous	Agreement score (0–100)
I avoid boredom by visiting new places as a reason for not shopping in the neighbourhood	Continuous	Agreement score (0–100)
Affordable house as a residential location choice	Continuous	Agreement score (0–100)
Near workplace/school as a residential location choice	Continuous	Agreement score (0–100)
Attractive surrounding environment as a residential location choice	Continuous	Agreement score (0–100)
Secure area as a residential location choice	Continuous	Agreement score (0–100)
I have lived here since birth/childhood as a residential location choice	Continuous	Agreement score (0–100)
Ease of commute to the workplace as a reason for choosing a neighbourhood	Continuous	Agreement score (0–100)
Traffic safety perception when walking in the neighbourhood	Continuous	Agreement with walking/cycling to nearby destinations -Agreement score (0–100)
The streets are not safe, as a reason for not walking in the neighbourhood	Continuous	Agreement score (0–100)
Ease of navigating the neighbourhood by car	Continuous	Ease of navigating neighbourhood by car (0–100 scale) -Agreement score (0–100)
Ease of navigating the neighbourhood on foot	Continuous	Ease of navigating neighbourhood on foot -Agreement score (0–100)
Ease of navigating the neighbourhood by bicycle	Continuous	Ease of navigating neighbourhood bicycle -Agreement score (0–100)
Proximity to e-hailing stops	Continuous	Perceived proximity to e-hailing stops (0–100 scale)
Travel Behaviour
Shopping Location (neighbourhood/far)	Categorical	Whether shopping is done in the neighbourhood or farther 0 = further, 1 = neighbourhood
E-hailing usage frequency	Continuous	How often a respondent uses e-hailing (0–100 scale)

Source: Authors’ computations.

summarises the selected key variables.

Population density data were obtained from the 2023 Namibia Population and Housing Census and assigned to respondents by constituency using GIS, with privacy preserved by geocoding the nearest street intersection rather than exact home addresses. Street length density was calculated to assess the degree of street connectivity within each neighbourhood zone. Using ArcGIS Pro and road network data from OpenStreetMap (OSM), The total length of all road segments within each zone was extracted Using ArcGIS Pro and road network data from OpenStreetMap. The Summarise Within tool was then applied to aggregate the road lengths by zone. To normalise the data for comparability across spatial units, street length density was calculated as the total road length (in meters) divided by the area of the zone (in square meters) using the following formula:

$$\mathrm{S}\mathrm{t}\mathrm{r}\mathrm{e}\mathrm{e}\mathrm{t} \mathrm{L}\mathrm{e}\mathrm{n}\mathrm{g}\mathrm{t}\mathrm{h} \mathrm{D}\mathrm{e}\mathrm{n}\mathrm{s}\mathrm{i}\mathrm{t}\mathrm{y}={\displaystyle \frac{\mathrm{T}\mathrm{o}\mathrm{t}\mathrm{a}\mathrm{l} \mathrm{R}\mathrm{o}\mathrm{a}\mathrm{d} \mathrm{L}\mathrm{e}\mathrm{n}\mathrm{g}\mathrm{h}\left(\mathrm{m}\right)}{\mathrm{T}\mathrm{o}\mathrm{t}\mathrm{a}\mathrm{l} \mathrm{Z}\mathrm{o}\mathrm{n}\mathrm{e} \mathrm{A}\mathrm{r}\mathrm{e}\mathrm{a}{(\mathrm{m}}^2)}}$$

This formula was applied to a Field Calculator. Visual inspection and manual measurements confirmed the accuracy of segment clipping and aggregation processes.

Intersection density was computed to evaluate the spatial connectivity of the road network across the zones. The Generate Intersections tool in ArcGIS Pro was used to identify and create point features for each location where two or more streets intersected. These intersection points were spatially filtered using the Select Layer by Location tool to include only those points within the zone boundaries. The Count Points in Polygon tool are then used to calculate the number of intersections per zone. This value was normalised using each zone’s area to produce the intersection density, calculated as

$$\mathrm{I}\mathrm{n}\mathrm{t}\mathrm{e}\mathrm{r}\mathrm{s}\mathrm{e}\mathrm{c}\mathrm{t}\mathrm{i}\mathrm{o}\mathrm{n} \mathrm{D}\mathrm{e}\mathrm{n}\mathrm{s}\mathrm{i}\mathrm{t}\mathrm{y}={\displaystyle \frac{\mathrm{N}\mathrm{u}\mathrm{m}\mathrm{b}\mathrm{e}\mathrm{r} \mathrm{o}\mathrm{f} \mathrm{I}\mathrm{n}\mathrm{t}\mathrm{e}\mathrm{r}\mathrm{s}\mathrm{e}\mathrm{c}\mathrm{t}\mathrm{i}\mathrm{o}\mathrm{n}\mathrm{s}}{\mathrm{Z}\mathrm{o}\mathrm{n}\mathrm{e} \mathrm{A}\mathrm{r}\mathrm{e}\mathrm{a}\left({\mathrm{k}\mathrm{m}}^2\right)}}$$

The calculation was performed in the Field Calculator after deriving the zone area using the Calculate Geometry Attributes Tool. The results were verified by overlaying intersections on the street network to ensure the correct point placement and zone alignment.

To define compact versus sprawled urban form, the study used three spatial indicators: Shannon entropy for land-use dispersion, street-length density and intersection density for network connectivity. These metrics are widely used in the urban sprawl literature to capture land-use dispersion and street network connectivity [70,71,72,73,74,75].

Shannon entropy was used to measure the degree of land-use concentration or dispersion in each zone/constituency, calculated as

$$H=-\sum _{i=1}^n{p}_i·\mathit{ln}\left(p_i\right)$$

In the entropy formula, H represents the Shannon entropy for the respondent’s residential zone, n is the total number of zones, p_i is the proportion of built-up area in the ith zone, and ln is the natural logarithm.

The proportion p_i is calculated using this formula:

$$p_i= {\displaystyle \frac{B_i}{\sum _{i=1}^n{B}_i}}$$

where B_i is the built-up area in the ith zone, and ∑ B_i is the total built-up area across all zones. This ensured that p_i reflected the relative share of built-up land in each zone. To facilitate interpretation, the entropy values were normalised between 0 and 1, where 0 indicates a fully compact form, and 1 signifies a fully sprawled area. In Windhoek, compact zones consistently recorded lower entropy values (≈0.002–0.006), reflecting concentrated land use, while sprawled zones clustered at higher values (≈0.007–0.010), indicating greater land-use dispersion and fragmentation.

Street length density was used to assess street network intensity. Sprawled neighbourhoods showed lower densities (7516–14,070 m/km²; M = 9228, SD = 3646), reflecting fragmented, car-oriented layouts typical of peripheral growth in emerging cities. Compact neighbourhoods displayed higher densities (14,116–33,569 m/km²; M = 14,989, SD = 5077), indicating more interconnected street patterns that comparatively enhance walkability and multimodal access.

Intersection density captured network connectivity. Compact neighbourhoods demonstrated much higher values (often >150 intersections/km², peaking above 300/km²), whereas sprawled areas recorded lower densities ranging between 30 and 150 intersections/km², reflecting dispersed, car-oriented layouts.

These measures provide a robust classification of neighbourhood form. Importantly, absolute values for Windhoek differ from benchmarks reported in developed-city contexts. In many Organisation for Economic Co-operation and Development (OECD) cities, intersection densities above 300/km² are common in central districts, while African cities such as Nairobi, Addis Ababa, and Johannesburg record substantially lower connectivity [74,75]. This divergence reflects the structural realities of developing cities, where informal settlements generate high population densities but low street provision and weak land-use integration. Consequently, Windhoek’s compact zones are compact relative to its own urban structure rather than by external thresholds.

By combining historical context with objective spatial indicators, this study provides an empirical basis for classifying neighbourhoods as compact or sprawled, strengthening the validity of subsequent analyses on mobility outcomes.

To provide contextual background for the subsequent analyses, descriptive statistics were generated for key categorical and continuous variables, disaggregated by neighbourhood type. This allows for an initial overview of the socio-economic, demographic, and behavioural characteristics of residents in compact and sprawling areas of Windhoek. Descriptive statistics for the categorical variables, stratified according to neighbourhood type, are provided in

Table 4. Descriptive statistics for categorical variables.

Variable	Category	Sprawling Area (%)	Compact Area (%)	Total (%)
Income	≤N$5000	54.8	19.6	37.2
	>N$5000	45.2	80.4	62.8
Education	Lower (Primary & Secondary)	57	24.6	40.8
	Higher (College & University)	43	75.4	59.2
Gender	Female	52.2	49.8	51
	Male	47.8	50.2	49
Age Group	Under 35 years	58.2	55	56.6
	35 years and above	41.8	45	43.4
Driving Licence	Yes	39.8	63	51.4
	No	60.2	37	48.6
Shopping Location	Within Neighbourhood	39	67.2	53.1
	Outside Neighbourhood	61	32.8	46.9

Source: Author compilations.

.

The descriptive statistics indicate a clear socio-spatial distinction between compact and sprawling neighbourhoods in Windhoek. Residents of compact areas demonstrate higher socio-economic status, with elevated levels of income and educational attainment, as well as greater access to private mobility, as evidenced by higher rates of driving licence ownership. In contrast, sprawling neighbourhoods are marked by lower income and education levels and reduced mobility access, reflecting persistent structural disadvantages. Although age and gender distributions are broadly similar across neighbourhood types, shopping behaviour differs substantially. Individuals in compact areas are more likely to conduct shopping within their neighbourhoods, suggesting better access to neighbourhood services, whereas those in sprawling areas more frequently travel outside their immediate vicinity to meet daily needs. These patterns highlight the influence of urban form on access and mobility and illustrate the continuing effects of spatial inequality shaped by historical urban planning practices.

Furthermore,

Table 5. Descriptive statistics for continuous variables.

Variable	Neighbourhood Type	N	Min	Max	Mean	Std. Deviation
Population Density	Sprawling	500	3.2	4	3.693	0.214
	Compact	500	2.8	3.3	3.042	0.25
Number of Non-Work Trips (Past 7 Days)	Sprawling	500	2	8	3.89	1.676
	Compact	500	2	8	3.88	1.589
Sense of Belonging	Sprawling	500	0	100	38.82	27.333
	Compact	500	0	100	55.02	31.076
Perceived Attractiveness of Shopping Centres	Sprawling	500	0	90	33.54	24.537
	Compact	500	0	100	63.44	31.428
Perceived Attractiveness of Recreational Facilities	Sprawling	500	0	100	20.42	19.545
	Compact	500	0	100	56.12	29.609
Preference for Entertainment in Own Neighbourhood	Sprawling	500	0	100	27.26	27.55
	Compact	500	0	100	47.84	35.285

Source: Author compilations.

presents the descriptive statistics for continuous variables disaggregated by neighbourhood type. Residents in sprawling areas reported a higher mean population density (M = 3.69, SD = 0.21) compared to those in compact areas (M = 3.04, SD = 0.25), reflecting Windhoek’s typical spatial arrangement. the mean weekly number of non-work trips was nearly identical between the two groups (Sprawled: M = 3.89, SD = 1.68; Compact: M = 3.88, SD = 1.59). However, consistent differences were observed across perceptual variables. Residents of compact neighbourhoods reported higher mean scores for sense of belonging (M = 55.02 vs. 38.82), perceived attractiveness of shops (M = 63.44 vs. 33.54), and recreational facilities (M = 56.12 vs. 20.42). Similarly, preference for entertainment within the neighbourhood was notably stronger in compact areas (M = 47.84, SD = 35.29) than in sprawled ones (M = 27.26, SD = 27.55). These differences suggest that compact areas not only benefit from more favourable built environments but also foster stronger place-based perceptions and preferences for local engagement.

3.4. Analysis Methods

To address the three research questions, a combination of comparative statistical analysis, binary logistic regression, and multiple linear regression was employed. These methods were selected to match the varied data types, categorical and continuous, and account for Windhoek’s spatially fragmented urban form, shaped by apartheid-era planning and enduring socio-spatial inequalities. Given these structural disparities, separate models were developed for compact and sprawled neighbourhoods in both regression analyses. This distinction recognises that the factors associated with neighbourhood shopping and entertainment/recreational travel vary across urban contexts: sprawled areas tend to face infrastructural and socioeconomic constraints, while compact areas are better served by accessible amenities and integrated land use. Therefore, each model applies context-specific predictors to reflect these differentiated urban realities.

For the first research question, which aimed to identify differences in socio-economic characteristics, neighbourhood perceptions, and travel behaviours between residents of compact and sprawled neighbourhoods, a comparative statistical approach was adopted. The Chi-Square Test of Independence was applied to categorical variables (e.g., income, education, gender, mode of transport), using a 95% confidence level (p < 0.05) to test associations, with Cramér’s V measuring effect sizes. For continuous and ordinal variables, such as neighbourhood perceptions and trip frequencies, many of which were measured on a 0–100 scale, the Mann–Whitney U test was employed. This non-parametric method was selected following a formal assessment of normality using the Shapiro–Wilk test. The results confirmed that all continuous variables significantly deviated from a normal distribution, thereby justifying the use of the Mann–Whitney U test for comparing group differences. Specifically, the Shapiro–Wilk test yielded the following values: sense of belonging (W = 0.946, p < 0.001), access to attractive shops (W = 0.916, p < 0.001), access to recreational amenities (W = 0.946, p < 0.001), preference for neighbourhood entertainment (W = 0.923, p < 0.001), population density (W = 0.892, p < 0.001), perceived insecurity (W = 0.915, p < 0.001), traffic safety while walking (W = 0.933, p < 0.001), and unpleasant social atmosphere (W = 0.900, p < 0.001). The consistent significance across all variables supports the decision to adopt non-parametric methods in the analysis. Together, these tests provide a reliable foundation for comparing the key patterns of mobility and perception across urban form categories.

The second research question explored the predictors of shopping trips within neighbourhoods. Binary logistic regression was used to model the likelihood of neighbourhood shopping (coded as 1) versus distant shopping (coded as 0). Separate models were constructed for sprawled and compact areas by recognising the structural and functional differences between neighbourhood types. The dependent variable in each case was dichotomous, reflecting the place of shopping (neighbourhood vs. distant). The independent variables were grouped into three categories. Socio-economic characteristics included income, gender, licence ownership, household size (adults), and the number of driving licences per household. Neighbourhood perceptions encompassed perceived safety, sense of belonging, the tendency to avoid boredom by visiting new places, perceptions of traffic safety while walking, the attractiveness of local social and recreational facilities, preference for entertainment outside the neighbourhood, perceptions of insecurity, the perception that neighbourhood shops are too expensive, the recognition of attractive shops or shopping centres in the neighbourhood, and the ease of navigating the neighbourhood by bicycle. Built environment measures incorporated intersection density around the home and perceived access to educational institutions, shops, and recreational amenities.

Model refinement involved the stepwise removal of non-significant predictors, retaining only those with statistically significant effects (p < 0.05). Model fit was evaluated using the Hosmer–Lemeshow goodness-of-fit test and Nagelkerke’s R², ensuring that explanatory power and calibration were adequately assessed.

For the third research question, which examined the socioeconomic, perceptual, and spatial predictors of the frequency of entertainment and recreational trips, multiple linear regression was applied. The dependent variable reflected residents’ preference for entertainment within their own neighbourhood, measured on a continuous scale from 0 (strongly disagree) to 100 (strongly agree). Again, distinct models were developed for compact and sprawled neighbourhoods, acknowledging their contextual differences.

Predictors comprised socio-economic factors (gender, income, education, number of driving licences per household), neighbourhood perceptions (sense of belonging, safety and insecurity, traffic safety while walking, attractiveness and affordability of local shops and recreational amenities, social atmosphere, preference for entertainment outside the neighbourhood, avoidance of boredom, ease of navigating by car, and residential location choices such as proximity to workplace or school, attractiveness of surroundings, and long-term residence), travel-related factors (e-hailing frequency and proximity to e-hailing stops), and built environment variables (intersection density and perceived access to parks, recreational areas, and shops). Multicollinearity was assessed using Variance Inflation Factors (VIF), and only variables meeting the conventional threshold of VIF < 5 were retained. Statistical significance was set at p < 0.05, and model performance was evaluated using adjusted R².

By adopting differentiated models and techniques tailored to data type and neighbourhood context, and by explicitly testing distributional assumptions, model fit, and collinearity, this analytical framework provides a robust, context-sensitive approach to understand how spatial form, social factors, and perceptions interact to shape non-commuting travel behaviour in Windhoek.

4. Results

4.1. Socioeconomic, Perceptual, and Mobility Differences Between Compact and Sprawled Neighbourhoods in Windhoek

To examine whether residents of compact and sprawled neighbourhoods in Windhoek differ significantly in terms of socio-economic profiles, residential perceptions, and neighbourhood-level travel behaviours, a combination of Chi-Square Tests of Independence and Mann–Whitney U tests was conducted. The results demonstrated clear and statistically significant contrasts between the two urban forms.

Table 6. Chi-Square Test for Categorical Variables.

Independent Variables	Chi-Square (Pearson)	df	p-Value	Cramer’s V
Income	190.573	4	<0.001	0.437
Education	141.191	6	<0.001	0.376
Driving License	53.866	1	<0.001	0.232
Age	54.448	5	<0.001	0.233
Gender	0.576	1	0.448	0.024
Shopping Location (neighbourhood/far)	79.831	1	<0.001	0.283

shows the results of the Chi-Square Test.

The Chi-Square Test of Independence revealed statistically significant associations between neighbourhood type (compact vs. sprawled) and several socio-economic and behavioural variables. Strong associations were observed for income (χ²(4) = 190.573, p < 0.001, Cramér’s V = 0.437) and education (χ²(6) = 141.191, p < 0.001, Cramér’s V = 0.376), suggesting a meaningful relationship between these variables and the type of neighbourhood.

Significant associations were also found for age (χ²(5) = 54.448, p < 0.001, V = 0.233) and driving licence ownership (χ²(1) = 53.866, p < 0.001, V = 0.232), as well as shopping location (χ²(1) = 79.831, p < 0.001, V = 0.283). These results suggest that socio-economic status, age distribution, mobility access, and shopping behaviour are closely associated with urban form, highlighting the importance of neighbourhood context in shaping patterns of everyday life in Windhoek.

The strength of the associations between neighbourhood type and selected categorical variables was assessed using Cramér’s V, which measures the effect size for Chi-square tests involving categorical data. According to the interpretive scale proposed by [76] values greater than 0.25 represent a very strong association, those above 0.15 indicate a strong association, values above 0.10 suggest a moderate association, those exceeding 0.05 reflect a weak association, and values below 0.05 are considered very weak or negligible. In this study, income demonstrated the strongest association with neighbourhood type (V = 0.437), followed closely by education (V = 0.376), both of which were classified as very strong. The shopping location (V = 0.283) also showed a very strong association, reflecting meaningful differences in shopping location, whether residents shop within their neighbourhood or farther away, between compact and sprawled areas. Age (V = 0.233) and driving licence ownership (V = 0.232) were strongly associated, suggesting a spatial clustering of mobility access and life-stage characteristics.

The Mann–Whitney U test results reveal a consistent perceptual advantage among residents of compact neighbourhoods in Windhoek, as shown in

Table 7. Mann–Whitney U tests for Continuous variables.

Variable	Mean Rank (Sprawled)	Mean Rank (Compact)	U	Z	p-Value
Sense of belonging	425.58	575.42	87,541.50	–8.251	<0.001
Population Density	738.88	262.12	5808.00	–26.596	<0.001
Frequency of non-commuting trips	499.32	501.68	124,409.00	–0.139	0.890
There are attractive shops or shopping centres in my neighbourhood	369.22	631.78	59,359.50	–14.459	<0.001
Attractiveness of social/recreational facilities in the neighbourhood	335.42	665.58	42,462.00	–18.207	<0.001
Preference for neighbourhood entertainment	415.44	585.56	82,470.50	–9.430	<0.001

. Across several indicators, including sense of belonging, access to attractive shops, recreational amenities, and preference for neighbourhood-based entertainment, respondents in compact areas reported significantly higher mean ranks than those in sprawling areas. For example, the mean rank for perceived access to attractive shops was 631.78 in compact areas compared to 369.22 in sprawling areas (U = 59,359.50, p < 0.001), while recreational amenities were rated at 665.58 versus 335.42, respectively (U = 42,462.00, p < 0.001). Similarly, the sense of belonging showed a clear difference (575.42 vs. 425.58; p < 0.001). These differences align with broader literature linking compact urban form, walkability, and amenity density to greater place attachment and neighbourhood satisfaction.

A notable exception to this pattern concerns population density, where respondents in sprawling neighbourhoods reported significantly higher mean ranks (738.88 vs. 262.12; U = 5808.00, p < 0.001). This finding diverges from conventional urban theory, suggesting that in Windhoek, sprawled settlements are more densely populated than compact ones. This inversion reflects the city’s apartheid-era spatial legacies, where marginalised groups were relegated to overcrowded peripheral areas, while centrally located, low-density neighbourhoods were reserved for historically privileged populations.

4.2. Determinants of Neighbourhood-Based Shopping Travel in Compact and Sprawled Areas of Windhoek

A binary logistic regression model was employed to examine the associations between neighbourhood-based shopping travel behaviour and a range of socio-economic, perceptual, and spatial factors in Windhoek’s sprawled areas. The model was designed to examine the extent to which characteristics such as household socioeconomic status, residents’ perceptions of their neighbourhood environment, and built form attributes were statistically associated with the likelihood of conducting shopping trips within the neighbourhood.

Table 8. Binary Logistic Regression Model of Neighbourhood Shopping Location in Sprawl Areas of Windhoek.

	B	S.E.	Wald	p	β
Length/street density around home	0.009	0.005	4.278	0.039	1.009
Sense of belonging to the neighbourhood	0.011	0.004	7.314	0.007	1.011
Shops are too expensive as a reason for not shopping in the neighbourhood	−0.008	0.003	8.039	0.005	0.992
I avoid boredom by visiting new places/people	−0.008	0.003	8.257	0.004	0.992
Little security in the neighbourhood	−0.014	0.005	7.374	0.007	0.986
Traffic safety perception when walking in the neighbourhood	−0.012	0.005	5.771	0.016	0.988
Attractiveness of social/recreational facilities in the neighbourhood	0.014	0.005	6.630	0.010	1.014
Preference for neighbourhood entertainment	0.693	0.204	11.582	<0.001	1.999
Perceived access to educational institutions	0.009	0.004	3.956	0.047	1.009
Income? (1)	−0.011	0.005	5.436	0.020	0.989
Gender (1)	−0.477	0.201	5.612	0.018	0.621
Number of driving licenses/households	−0.011	0.006	3.892	0.049	0.989
Constant	1.156	0.810	2.036	0.154	3.178
Specification Tests
Nagelkerke R Square	Omnibus Tests
0.208	Chi-square	df	p
	84.158	12	<0.001
	Hosmer and Lemeshow Test
Observations	Chi-square	df	Sig.
N = 1000	6.701	8	0.569

Notes: Categorical variables are coded as follows. Dependent variable: Shopping Location (0 = further, 1 = neighbourhood). Independent variables: Income (0 = less than N$5000, 1 = greater than N$5000); Gender (0 = female, 1 = male); Education (0 = lower education, 1 = higher education), and Driving Licence (0 = holds licence, 1 = does not hold licence). Source: Authors’ computations.

presents the results of the binary logistics model for the determinants of Neighbourhood Shopping trips in sprawled areas in Windhoek.

The binary logistic regression model for sprawled neighbourhoods can be expressed as

$$\begin{array}{ll}\log \left({\displaystyle \frac{\mathrm{P}}{1-\mathrm{P}}}\right)={\beta }^0& +{\beta }^1\left(\mathrm{B}\mathrm{e}\mathrm{l}\mathrm{o}\mathrm{n}\mathrm{g}\mathrm{i}\mathrm{n}\mathrm{g}\right)+{\beta }^2\left(\mathrm{B}\mathrm{o}\mathrm{r}\mathrm{e}\mathrm{d}\mathrm{o}\mathrm{m}\mathrm{A}\mathrm{v}\mathrm{o}\mathrm{i}\mathrm{d}\mathrm{a}\mathrm{n}\mathrm{c}\mathrm{e}\right)+{\beta }^3\left(\mathrm{W}\mathrm{a}\mathrm{l}\mathrm{k}\mathrm{S}\mathrm{a}\mathrm{f}\mathrm{e}\mathrm{t}\mathrm{y}\right)+{\beta }^4\left(\mathrm{G}\mathrm{e}\mathrm{n}\mathrm{d}\mathrm{e}\mathrm{r}\right)\\ & +{\beta }^5\left(\mathrm{S}\mathrm{o}\mathrm{c}\mathrm{i}\mathrm{a}\mathrm{l}\mathrm{F}\mathrm{a}\mathrm{c}\mathrm{i}\mathrm{l}\mathrm{i}\mathrm{t}\mathrm{i}\mathrm{e}\mathrm{s}\right)+{\beta }^6\left(\mathrm{E}\mathrm{n}\mathrm{t}\mathrm{e}\mathrm{r}\mathrm{t}\mathrm{a}\mathrm{i}\mathrm{n}\mathrm{m}\mathrm{e}\mathrm{n}\mathrm{t}\mathrm{P}\mathrm{r}\mathrm{e}\mathrm{f}\mathrm{e}\mathrm{r}\mathrm{e}\mathrm{n}\mathrm{c}\mathrm{e}\right)+{\beta }^7\left(\mathrm{E}\mathrm{d}\mathrm{u}\mathrm{A}\mathrm{c}\mathrm{c}\mathrm{e}\mathrm{s}\mathrm{s}\right)+{\beta }^8\left(\mathrm{I}\mathrm{n}\mathrm{c}\mathrm{o}\mathrm{m}\mathrm{e}\right)\\ & +{\beta }^9\left(\mathrm{S}\mathrm{t}\mathrm{r}\mathrm{e}\mathrm{e}\mathrm{t}\mathrm{D}\mathrm{e}\mathrm{n}\mathrm{s}\mathrm{i}\mathrm{t}\mathrm{y}\right)+{\beta }^{10}\left(\mathrm{S}\mathrm{e}\mathrm{c}\mathrm{u}\mathrm{r}\mathrm{i}\mathrm{t}\mathrm{y}\right)+{\beta }^{11}\left(\mathrm{L}\mathrm{i}\mathrm{c}\mathrm{e}\mathrm{n}\mathrm{c}\mathrm{e}\mathrm{s}\right)+{\beta }^{12}\left(\mathrm{S}\mathrm{h}\mathrm{o}\mathrm{p}\mathrm{C}\mathrm{o}\mathrm{s}\mathrm{t}\right)\end{array}$$

(1)

In this model, P is the probability of shopping within a neighbourhood. Model diagnostics results of the Omnibus Test of Model Coefficients confirmed the overall significance of the model (χ² = 84.158, df = 12, p < 0.001), while the Nagelkerke R² value of 0.208 indicated that the model explained approximately 21% of the variation in neighbourhood shopping behaviour. The Hosmer–Lemeshow test (χ² = 6.701, df = 8, p = 0.569) further confirmed the adequacy of the model in fitting the observed data.

Of the 12 variables included in the model, all were statistically significant at the 5% level, with six showing particularly high significance (p < 0.01). The most influential factor was a stated preference for neighbourhood-based entertainment (p < 0.001), followed by boredom avoidance (p = 0.004), perceptions of high shop prices (p = 0.005), and a sense of belonging to the neighbourhood (p = 0.007). Other significant predictors included feelings of insecurity (p = 0.007), perceptions of traffic safety when walking (p = 0.016), and the perceived attractiveness of social and recreational facilities (p = 0.010). Built environment features also mattered; street density around the home (p = 0.039) and perceived access to educational institutions (p = 0.047) both showed significance. Sociodemographic variables such as income (p = 0.020), gender (p = 0.018), and the number of driving licences in the household (p = 0.049) were also statistically significant. The constant term was not significant (p = 0.154), indicating that the model’s explanatory power lies in the included variables rather than in the baseline probabilities.

Looking at the strength of associations, preference for neighbourhood entertainment had the largest positive effect, with an odds ratio of Exp(B) = 1.999, with individuals holding this preference being twice as likely to shop in the neighbourhood. Positive perceptions of recreational amenities (Exp(β) = 1.014) and a sense of belonging (Exp(β) = 1.011) were associated with a 1.4% and 1.1% increase in the odds of neighbourhood shopping, respectively. Similarly, perceived access to educational institutions and street density (both Exp(β) = 1.009) were linked to an increase of approximately 0.9% in neighbourhood shopping odds.

In contrast, perceiving neighbourhood shops as expensive and a tendency to seek variety or escape boredom (both Exp(β) = 0.992) were each associated with a 0.8% decrease in the odds of neighbourhood shopping. Concerns about neighbourhood security (Exp(β) = 0.986) and traffic safety while walking (Exp(β) = 0.988) further reduced the likelihood by 1.4% and 1.2%, respectively. Among socio-demographic factors, higher income and a greater number of driving licences in the household (both Exp(β) = 0.989) reduced the odds of neighbourhood shopping by 1.1% per unit increase. Lastly, male respondents were 38% less likely to shop in the neighbourhood compared to females (Exp(β) = 0.621), suggesting notable gendered differences in mobility and consumption patterns.

Furthermore, the second binary logistic regression model was developed to assess the associations between neighbourhood-based shopping behaviour and a range of socio-economic, perceptual, and spatial variables in Windhoek’s compact areas.

Table 9. Binary Logistic Regression Model of Neighbourhood Shopping Location in Compact Areas of Windhoek.

	B	S.E.	Wald	p	β
Length/street density around home	0.010	0.005	4.750	0.029	1.010
Little security in the neighbourhood	−0.010	0.003	9.181	0.002	0.990
The shops or facilities are too expensive	−0.008	0.003	4.976	0.026	0.992
I feel a sense of belonging to my neighbourhood	0.010	0.004	5.321	0.021	1.010
A preference to have entertainment far from the neighbourhood	−0.006	0.003	3.542	0.060	0.994
I avoid boredom by visiting new places	−0.006	0.004	2.762	0.097	0.994
There are attractive shops or shopping centres in my neighbourhood	0.012	0.006	4.903	0.027	1.012
Attractiveness of social/recreational facilities in the neighbourhood	0.009	0.005	3.136	0.077	1.009
Ease of navigating the neighbourhood by bicycle	0.013	0.006	5.478	0.019	1.013
No good social atmosphere as a reason for not shopping in the neighbourhood	−0.009	0.003	8.999	0.003	0.991
Traffic safety perception when walking in the neighbourhood	−0.011	0.003	11.517	<0.001	0.989
Ease of Navigating the Neighbourhood by Car	0.008	0.004	4.715	0.030	1.008
Number of driving licenses/households	0.011	0.005	4.691	0.030	1.011
Household Size (Adults)	0.172	0.083	4.229	0.040	1.187
Age (1)	−0.011	0.006	3.627	0.057	0.989
Gender (1)	−0.014	0.005	6.443	0.011	0.986
Income (1)	0.012	0.006	4.527	0.033	1.013
Constant	−1.014	1.148	0.780	0.377	0.363
Specification Tests
Nagelkerke R Square			Omnibus Tests
0.214		Chi-square		df	p
			87.371	17	<0.001
Hosmer and Lemeshow Test
Observations		Chi-square		df	p
N = 1000			5.379	8	0.716

Notes: Categorical variables are coded as follows. Dependent variable: Shopping Location (0 = further, 1 = neighbourhood). Independent variables: Income (0 = less than N$5000, 1 = greater than N$5000); Gender (0 = female, 1 = male); Education (0 = lower education, 1 = higher education); Driving Licence (0 = holds licence, 1 = does not hold licence); age (0 = Under 35 years, 1 = 35 years and above).

presents the results of the binary logistics model for the determinants of Neighbourhood Shopping trips in compact areas in Windhoek.

The logistic regression model for compact neighbourhoods is specified as

$$\begin{array}{ll}\log \left({\displaystyle \frac{\mathrm{P}}{1-\mathrm{P}}}\right)={\beta }^0& +{\beta }^1\left(\mathrm{A}\mathrm{t}\mathrm{t}\mathrm{r}\mathrm{a}\mathrm{c}\mathrm{t}\mathrm{i}\mathrm{v}\mathrm{e}\mathrm{S}\mathrm{h}\mathrm{o}\mathrm{p}\mathrm{s}\right)+{\beta }^2\left(\mathrm{S}\mathrm{o}\mathrm{c}\mathrm{i}\mathrm{a}\mathrm{l}\mathrm{F}\mathrm{a}\mathrm{c}\mathrm{i}\mathrm{l}\mathrm{i}\mathrm{t}\mathrm{i}\mathrm{e}\mathrm{s}\right)+{\beta }^3\left(\mathrm{B}\mathrm{i}\mathrm{k}\mathrm{e}\mathrm{E}\mathrm{a}\mathrm{s}\mathrm{e}\right)+{\beta }^4\left(\mathrm{B}\mathrm{o}\mathrm{r}\mathrm{e}\mathrm{d}\mathrm{o}\mathrm{m}\mathrm{A}\mathrm{v}\mathrm{o}\mathrm{i}\mathrm{d}\mathrm{a}\mathrm{n}\mathrm{c}\mathrm{e}\right)\\ & +{\beta }^5\left(\mathrm{L}\mathrm{i}\mathrm{c}\mathrm{e}\mathrm{n}\mathrm{c}\mathrm{e}\mathrm{s}\mathrm{P}\mathrm{e}\mathrm{r}\mathrm{H}\mathrm{H}\right)+{\beta }^6\left(\mathrm{S}\mathrm{e}\mathrm{c}\mathrm{u}\mathrm{r}\mathrm{i}\mathrm{t}\mathrm{y}\right)+{\beta }^7\left(\mathrm{H}\mathrm{o}\mathrm{u}\mathrm{s}\mathrm{e}\mathrm{h}\mathrm{o}\mathrm{l}\mathrm{d}\mathrm{S}\mathrm{i}\mathrm{z}\mathrm{e}\right)+{\beta }^8\left(\mathrm{S}\mathrm{o}\mathrm{c}\mathrm{i}\mathrm{a}\mathrm{l}\mathrm{A}\mathrm{t}\mathrm{m}\mathrm{o}\mathrm{s}\mathrm{p}\mathrm{h}\mathrm{e}\mathrm{r}\mathrm{e}\right)\\ & +{\beta }^9\left(\mathrm{W}\mathrm{a}\mathrm{l}\mathrm{k}\mathrm{S}\mathrm{a}\mathrm{f}\mathrm{e}\mathrm{t}\mathrm{y}\right)+{\beta }^{10}\left(\mathrm{C}\mathrm{a}\mathrm{r}\mathrm{E}\mathrm{a}\mathrm{s}\mathrm{e}\right)+{\beta }^{11}\left(\mathrm{A}\mathrm{g}\mathrm{e}\right)+{\beta }^{12}\left(\mathrm{S}\mathrm{t}\mathrm{r}\mathrm{e}\mathrm{e}\mathrm{t}\mathrm{D}\mathrm{e}\mathrm{n}\mathrm{s}\mathrm{i}\mathrm{t}\mathrm{y}\right)+{\beta }^{13}\left(\mathrm{S}\mathrm{h}\mathrm{o}\mathrm{p}\mathrm{C}\mathrm{o}\mathrm{s}\mathrm{t}\right)\\ & +{\beta }^{14}\left(\mathrm{G}\mathrm{e}\mathrm{n}\mathrm{d}\mathrm{e}\mathrm{r}\right)+{\beta }^{15}\left(\mathrm{B}\mathrm{e}\mathrm{l}\mathrm{o}\mathrm{n}\mathrm{g}\mathrm{i}\mathrm{n}\mathrm{g}\right)+{\beta }^{16}\left(\mathrm{E}\mathrm{n}\mathrm{t}\mathrm{e}\mathrm{r}\mathrm{t}\mathrm{a}\mathrm{i}\mathrm{n}\mathrm{m}\mathrm{e}\mathrm{n}\mathrm{t}\mathrm{F}\mathrm{a}\mathrm{r}\mathrm{P}\mathrm{r}\mathrm{e}\mathrm{f}\mathrm{e}\mathrm{r}\mathrm{e}\mathrm{n}\mathrm{c}\mathrm{e}\right)+{\beta }^{17}\left(\mathrm{I}\mathrm{n}\mathrm{c}\mathrm{o}\mathrm{m}\mathrm{e}\right)\end{array}$$

(2)

In this model, P is the probability of shopping within a neighbourhood. Model diagnostics results reveal that the model was statistically significant (χ² = 87.371; df = 17; p < 0.001). The model explained approximately 21.4% of the variance in the outcome variable, as indicated by a Nagelkerke R² of 0.214 and demonstrated a good fit according to the Hosmer and Lemeshow test (χ² = 5.379, df = 8, p = 0.716).

Several variables are significantly associated with the likelihood of shopping within a neighbourhood. From the built environment perspective, street length density was positively associated with neighbourhood shopping (p = 0.029), as was the ease of navigating the neighbourhood by bicycle (p = 0.019) and by car (p = 0.030), reflecting the importance of physical connectivity and multimodal accessibility in supporting neighbourhood shopping travel behaviour. Among the perceptual factors, a sense of belonging (p = 0.021), the presence of attractive shops or shopping centres (p = 0.027), and the absence of a poor social atmosphere (p = 0.003) were significantly associated with neighbourhood shopping outcomes. By contrast, perceived insecurity (p = 0.002), concerns about traffic safety while walking (p < 0.001), and perceptions that shops are too expensive (p = 0.026) were negatively associated with shopping in the neighbourhood.

Sociodemographic characteristics also played a significant role. Higher income (p = 0.033), a greater number of driving licences in the household (p = 0.030), and larger adult household size (p = 0.040) were all positively associated with neighbourhood-based shopping, while gender was significant (p = 0.011), indicating that males were less likely than females to shop within their neighbourhoods.

The regression analysis reveals that both physical and perceptual variables significantly influence the likelihood of neighbourhood shopping in Windhoek’s compact areas. A higher street length density increased the odds of neighbourhood shopping by 1.0% for each unit increase (Exp(β) = 1.010), suggesting that better-connected street networks enhance accessibility and encourage short-distance retail trips. Similarly, a stronger sense of belonging (Exp(β) = 1.010) and the perceived attractiveness of neighbourhood shopping centres (Exp(β) = 1.012) were each associated with a 1.0% and 1.2% increase, respectively, reinforcing the role of emotional attachment and commercial appeal in neighbourhood consumer behaviour. Mobility perceptions also mattered: for every unit increase in the ease of navigating the neighbourhood by bicycle, the odds of shopping in the neighbourhood increased by 1.3% (Exp(β) = 1.013), while ease of driving showed a smaller but significant effect of 0.8% (Exp(β) = 1.008).

Conversely, several variables were negatively associated with neighbourhood shopping. The perception that shops were too expensive reduced the odds by 0.8% per unit increase (Exp(β) = 0.992), indicating that economic perceptions continue to constrain neighbourhood access, even in higher-income areas. Additionally, concerns about insecurity (Exp(β) = 0.990), traffic safety while walking (Exp(β) = 0.989), and an unpleasant social atmosphere (Exp(β) = 0.991) were associated with 1.0%, 1.1%, and 0.9% lower odds, respectively. These results highlight that spatial proximity alone is insufficient; perceived safety, comfort, and affordability significantly shape residents’ engagement with neighbourhood retail infrastructure. These results highlight that in compact areas, neighbourhood-based shopping is shaped by a combination of accessible infrastructure, social cohesion, perceived safety, and affordability. While compactness offers physical preconditions for neighbourhood travel, perceptual and social factors remain decisive in determining whether residents engage with nearby retail services.

4.3. Determinants of Neighbourhood-Based Entertainment and Recreational Travel in Windhoek

To explore the determinants of the frequency of entertainment and recreational trips within Windhoek’s sprawled neighbourhoods, a multiple linear regression model was developed, as shown in

Table 10. Multivariate Regression Model for Determinants of Preference for Neighbourhood-Based Entertainment Sprawl Areas of Windhoek.

	Unstandardized Coefficients		Standardized Coefficients	t	p	Collinearity Statistics
	B	Std. Error	Beta			Tolerance	VIF
(Constant)	27.873	6.203		4.493	<0.001
Intersection density around home	0.110	0.048	0.094	2.269	0.024	0.946	1.057
Sense of belonging to the neighbourhood	0.112	0.043	0.112	2.615	0.009	0.892	1.122
There are attractive shops or shopping centres in my neighbourhood	0.101	0.047	0.090	2.158	0.031	0.941	1.062
Attractive surrounding environment as a residential location choice	0.137	0.052	0.111	2.632	0.009	0.909	1.100
Perceived access to parks and recreational areas	0.090	0.045	0.081	1.979	0.048	0.965	1.036
The house is near to my workplace/school as a residential location choice	−0.069	0.035	−0.080	−1.972	0.049	0.976	1.025
I have lived here since birth/childhood as a residential location choice	0.081	0.033	0.101	2.455	0.014	0.962	1.039
Traffic safety perception when walking in the neighbourhood	6.338	2.956	0.113	2.144	0.033	0.587	1.704
No good social atmosphere as a reason for not shopping in the neighbourhood	−0.111	0.038	−0.120	−2.941	0.003	0.972	1.029
E-hailing usage frequency	0.116	0.043	0.114	2.667	0.008	0.886	1.129
Income	−0.153	0.040	−0.159	−3.816	<0.001	0.937	1.067
Driving license/household	−4.620	2.257	−0.110	−2.047	0.041	0.566	1.766
Education	−0.139	0.052	−0.110	−2.680	0.008	0.956	1.046
Model Summary
R	R Square			Std. Error of the Estimate
0.462	0.213			24.789
ANOVA
Measures	Sum of Squares	df		Mean Square		F	p
Regression	80,710.107	14		5765.008		9.382	<0.001

. The objective was to examine how socioeconomic characteristics, neighbourhood perceptions, and spatial conditions shape leisure-related mobility at the neighbourhood level.

The multivariate regression model for sprawled neighbourhoods can be generally expressed as

$$\begin{array}{ll}\mathrm{Y}={\beta }^0& +{\beta }^1\left(\mathrm{B}\mathrm{e}\mathrm{l}\mathrm{o}\mathrm{n}\mathrm{g}\mathrm{i}\mathrm{n}\mathrm{g}\right)+{\beta }^2\left(\mathrm{A}\mathrm{t}\mathrm{t}\mathrm{r}\mathrm{a}\mathrm{c}\mathrm{t}\mathrm{i}\mathrm{v}\mathrm{e}\mathrm{S}\mathrm{h}\mathrm{o}\mathrm{p}\mathrm{s}\right)+{\beta }^3\left(\mathrm{I}\mathrm{n}\mathrm{c}\mathrm{o}\mathrm{m}\mathrm{e}\right)+{\beta }^4\left(\mathrm{A}\mathrm{t}\mathrm{t}\mathrm{r}\mathrm{a}\mathrm{c}\mathrm{t}\mathrm{i}\mathrm{v}\mathrm{e}\mathrm{E}\mathrm{n}\mathrm{v}\mathrm{i}\mathrm{r}\mathrm{o}\mathrm{n}\mathrm{m}\mathrm{e}\mathrm{n}\mathrm{t}\right)+{\beta }^5\left(\mathrm{L}\mathrm{i}\mathrm{c}\mathrm{e}\mathrm{n}\mathrm{c}\mathrm{e}\mathrm{s}\mathrm{P}\mathrm{e}\mathrm{r}\mathrm{H}\mathrm{H}\right)\\ & +{\beta }^6\left(\mathrm{P}\mathrm{a}\mathrm{r}\mathrm{k}\mathrm{A}\mathrm{c}\mathrm{c}\mathrm{e}\mathrm{s}\mathrm{s}\right)+{\beta }^7\left(\mathrm{E}\mathrm{h}\mathrm{a}\mathrm{i}\mathrm{l}\mathrm{i}\mathrm{n}\mathrm{g}\mathrm{F}\mathrm{r}\mathrm{e}\mathrm{q}\mathrm{u}\mathrm{e}\mathrm{n}\mathrm{c}\mathrm{y}\right)+{\beta }^8\left(\mathrm{P}\mathrm{r}\mathrm{o}\mathrm{x}\mathrm{i}\mathrm{m}\mathrm{i}\mathrm{t}\mathrm{y}\mathrm{T}\mathrm{o}\mathrm{W}\mathrm{o}\mathrm{r}\mathrm{k}\right)+{\beta }^9\left(\mathrm{B}\mathrm{o}\mathrm{r}\mathrm{n}\mathrm{H}\mathrm{e}\mathrm{r}\mathrm{e}\right)\\ & +{\beta }^{10}\left(\mathrm{I}\mathrm{n}\mathrm{t}\mathrm{e}\mathrm{r}\mathrm{s}\mathrm{e}\mathrm{c}\mathrm{t}\mathrm{i}\mathrm{o}\mathrm{n}\mathrm{D}\mathrm{e}\mathrm{n}\mathrm{s}\mathrm{i}\mathrm{t}\mathrm{y}\right)+{\beta }^{11}\left(\mathrm{E}\mathrm{d}\mathrm{u}\mathrm{c}\mathrm{a}\mathrm{t}\mathrm{i}\mathrm{o}\mathrm{n}\right)+{\beta }^{12}\left(\mathrm{S}\mathrm{o}\mathrm{c}\mathrm{i}\mathrm{a}\mathrm{l}\mathrm{A}\mathrm{t}\mathrm{m}\mathrm{o}\mathrm{s}\mathrm{p}\mathrm{h}\mathrm{e}\mathrm{r}\mathrm{e}\right)+{\beta }^{13}\left(\mathrm{G}\mathrm{e}\mathrm{n}\mathrm{d}\mathrm{e}\mathrm{r}\right)+{\beta }^{14}\left(\mathrm{U}\mathrm{n}\mathrm{s}\mathrm{a}\mathrm{f}\mathrm{e}\mathrm{S}\mathrm{t}\mathrm{r}\mathrm{e}\mathrm{e}\mathrm{t}\mathrm{s}\right)\end{array}$$

(3)

In this equation, Y represents the predicted level of preference for engaging in entertainment and recreational activities within the sprawled neighbourhood. The model indicates that 21.3% of the variance in neighbourhood-based entertainment trips in sprawled areas is explained by the selected socio-economic, perceptual, and spatial variables (R² = 0.213). The model is statistically significant (F = 9.382, p < 0.001), confirming that the predictors collectively contribute meaningfully to explaining variations in entertainment/leisure travel behaviour. While the standard error (24.789) reflects some dispersion around predicted values, the model offers valuable insights into the key factors associated with neighbourhood recreational mobility within structurally marginalised urban environments.

Fourteen predictors were included in the final model, of which thirteen were statistically significant at the 5% level. These included income (p < 0.001), sense of belonging (p = 0.009), attractiveness of neighbourhood shops (p = 0.031), attractive surrounding environments (p = 0.009), number of driving licences per household (p = 0.041), perceived access to parks (p = 0.048), frequency of e-hailing use (p = 0.008), proximity to workplace or school (p = 0.049), long-term residence (p = 0.014), intersection density (p = 0.024), perception of poor social atmosphere (p = 0.003), education (p = 0.008), and pedestrian traffic safety perception (p = 0.033).

The strongest negative association was observed for income (β = −0.159), where each unit increase in income is associated with a 15.9% reduction in neighbourhood leisure trips. This suggests that higher-income residents may prefer destinations outside their neighbourhoods. Similarly, an additional unit in educational attainment and the number of household driving licences decreased neighbourhood entertainment/leisure participation by 11% each (β = −0.110), likely reflecting enhanced mobility and access to alternative entertainment options. The perception of a poor social atmosphere reduced neighbourhood entertainment/leisure travel by 12% (β = −0.120), emphasising the importance of social cohesion in shaping leisure behaviour. Furthermore, selecting a residential location primarily for proximity to work or school corresponded with an 8% decrease in neighbourhood entertainment participation (β = −0.080), implying a more functional than recreational view of the neighbourhood.

In contrast, several variables were positively associated with neighbourhood leisure activity. A one-unit increase in sense of belonging was linked to an 11.2% increase in neighbourhood leisure travel (β = 0.112), reflecting the impact of emotional connection to place. Similarly, perceiving attractive shopping centres (β = 0.090) and choosing the area for its pleasant environment (β = 0.111) raised the likelihood of neighbourhood leisure participation by 9% and 11.1%, respectively. Access to parks and recreational spaces contributed to an 8.1% increase in neighbourhood-based leisure (β = 0.081), and intersection density around the home added another 9.4% (β = 0.094), highlighting the role of walkability. E-hailing use frequency showed a positive effect of 11.4% (β = 0.114), suggesting that flexible transport supports neighbourhood leisure. Long-term residence, measured as having lived in the area since childhood, contributed a 10.1% increase (β = 0.101), while perceived traffic safety while walking raised neighbourhood leisure participation by 11.3% (β = 0.113), highlighting how mobility safety fosters everyday social and recreational participation.

In summary, the findings illustrate that neighbourhood-based leisure activities in Windhoek’s sprawled areas are shaped by intersecting socio-spatial patterns. Higher socio-economic status and mobility assets tend to diminish reliance on neighbourhood amenities, while factors such as walkability, access to green spaces, perceived safety, and emotional attachment to places promote active neighbourhood engagement. These insights highlight the need for urban interventions that go beyond physical infrastructure and address the social and perceptual dimensions of mobility, especially in historically marginalised and spatially fragmented urban peripheries.

Moreover, to explore the determinants of neighbourhood-based entertainment and recreational activity in Windhoek’s compact areas, a multiple linear regression model was developed, as presented in

Table 11. Multivariate Regression Model for Determinants of Preference for Neighbourhood-Based Entertainment in Compact Areas of Windhoek.

	Unstandardized Coefficients		Standardized Coefficients	t	p	Collinearity Statistics
	B	Std. Error	Beta			Tolerance	VIF
(Constant)	30.093	10.549		2.853	0.005
Intersection density around home	0.105	0.042	0.102	2.528	0.012	0.966	1.035
Little security in the neighbourhood	−0.288	0.041	−0.289	−7.043	<0.001	0.937	1.067
Shops are too expensive as a reason for not shopping in the neighbourhood	−0.180	0.052	−0.142	−3.491	<0.001	0.949	1.054
There are attractive shops or shopping centres in my neighbourhood	0.180	0.045	0.169	4.044	<0.001	0.905	1.104
I have lived here since birth/childhood as a residential location choice	0.126	0.042	0.126	3.033	0.003	0.915	1.093
Sense of belonging to the neighbourhood	0.140	0.042	0.146	3.328	<0.001	0.823	1.214
Ease of navigating the neighbourhood by car	0.214	0.076	0.115	2.803	0.005	0.944	1.060
I avoid boredom by visiting new places as a reason for not shopping in the neighbourhood	−0.140	0.059	−0.098	−2.370	0.018	0.916	1.092
The streets are not safe, as a reason for not walking in the neighbourhood	−0.189	0.069	−0.112	−2.744	0.006	0.949	1.054
Proximity to e-hailing stop	0.124	0.049	0.108	2.545	0.011	0.878	1.139
Gender	6.813	2.826	0.097	2.411	0.016	0.982	1.018
Income	0.127	0.043	0.123	2.986	0.003	0.934	1.071
Number of driving licenses/households	3.630	1.671	0.095	2.172	0.030	0.822	1.217
Model Summary
R	R Square			Std. Error of the Estimate
0.483	0.233			31.316
ANOVA
Measures	Sum of Squares		df	Mean Square		F	p
Regression	144,638.150		13	11,126.012		11.345	<0.001

.

The multivariate regression model for compact neighbourhoods is specified as

$$\begin{array}{ll}\mathrm{Y}={\beta }^0& +{\beta }^1\left(\mathrm{S}\mathrm{e}\mathrm{c}\mathrm{u}\mathrm{r}\mathrm{i}\mathrm{t}\mathrm{y}\right)+{\beta }^2\left(\mathrm{S}\mathrm{h}\mathrm{o}\mathrm{p}\mathrm{C}\mathrm{o}\mathrm{s}\mathrm{t}\right)+{\beta }^3\left(\mathrm{A}\mathrm{t}\mathrm{t}\mathrm{r}\mathrm{a}\mathrm{c}\mathrm{t}\mathrm{i}\mathrm{v}\mathrm{e}\mathrm{S}\mathrm{h}\mathrm{o}\mathrm{p}\mathrm{s}\right)+{\beta }^4\left(\mathrm{B}\mathrm{o}\mathrm{r}\mathrm{n}\mathrm{H}\mathrm{e}\mathrm{r}\mathrm{e}\right)+{\beta }^5\left(\mathrm{B}\mathrm{e}\mathrm{l}\mathrm{o}\mathrm{n}\mathrm{g}\mathrm{i}\mathrm{n}\mathrm{g}\right)\\ & +{\beta }^6\left(\mathrm{E}\mathrm{h}\mathrm{a}\mathrm{i}\mathrm{l}\mathrm{i}\mathrm{n}\mathrm{g}\mathrm{P}\mathrm{r}\mathrm{o}\mathrm{x}\mathrm{i}\mathrm{m}\mathrm{i}\mathrm{t}\mathrm{y}\right)+{\beta }^7\left(\mathrm{C}\mathrm{a}\mathrm{r}\mathrm{E}\mathrm{a}\mathrm{s}\mathrm{e}\right)+{\beta }^8\left(\mathrm{B}\mathrm{o}\mathrm{r}\mathrm{e}\mathrm{d}\mathrm{o}\mathrm{m}\mathrm{A}\mathrm{v}\mathrm{o}\mathrm{i}\mathrm{d}\mathrm{a}\mathrm{n}\mathrm{c}\mathrm{e}\right)+{\beta }^9\left(\mathrm{I}\mathrm{n}\mathrm{c}\mathrm{o}\mathrm{m}\mathrm{e}\right)\\ & +{\beta }^{10}\left(\mathrm{I}\mathrm{n}\mathrm{t}\mathrm{e}\mathrm{r}\mathrm{s}\mathrm{e}\mathrm{c}\mathrm{t}\mathrm{i}\mathrm{o}\mathrm{n}\mathrm{D}\mathrm{e}\mathrm{n}\mathrm{s}\mathrm{i}\mathrm{t}\mathrm{y}\right)+{\beta }^{11}\left(\mathrm{U}\mathrm{n}\mathrm{s}\mathrm{a}\mathrm{f}\mathrm{e}\mathrm{S}\mathrm{t}\mathrm{r}\mathrm{e}\mathrm{e}\mathrm{t}\mathrm{s}\right)+{\beta }^{12}\left(\mathrm{G}\mathrm{e}\mathrm{n}\mathrm{d}\mathrm{e}\mathrm{r}\right)+{\beta }^{13}\left(\mathrm{L}\mathrm{i}\mathrm{c}\mathrm{e}\mathrm{n}\mathrm{c}\mathrm{e}\mathrm{s}\mathrm{P}\mathrm{e}\mathrm{r}\mathrm{H}\mathrm{H}\right)\end{array}$$

(4)

In this equation, Y represents the predicted level of preference for engaging in entertainment and recreational activities within the compact neighbourhood. The model was statistically significant (F = 11.345, p < 0.001), with an Adjusted R² of 0.212, indicating that the included predictors explained approximately 21.2% of the variance in entertainment trip frequency within compact neighbourhoods. Thirteen variables were statistically significant at the 5% level, highlighting the complex relationship of perceptions, socio-economic status, and built environment attributes in shaping leisure mobility at the neighbourhood level.

Thirteen predictors were statistically significant at the 5% level: perceived lack of neighbourhood security (p < 0.001), affordability of shops (p < 0.001), attractiveness of shops (p < 0.001), sense of belonging (p < 0.001), long-term residence (p = 0.003), income (p = 0.003), car navigability (p = 0.005), concerns about street safety (p = 0.006), desire to escape boredom (p = 0.018), intersection density (p = 0.012), proximity to e-hailing stops (p = 0.011), gender (p = 0.016), and the number of driving licences per household (p = 0.030).

The regression analysis further reveals that perceived insecurity was the strongest negative predictor of neighbourhood leisure activity. Each unit increase in perceived insecurity was associated with a 28.9% decrease in the frequency of neighbourhood entertainment trips (β = −0.289), suggesting that fear of crime significantly suppresses neighbourhood recreational participation. Similarly, perceiving that shops are too expensive reduced neighbourhood entertainment/leisure travel by 14.2% (β = −0.142), indicating that economic exclusion can deter participation in neighbourhood amenities and recreational facilities. Concerns about traffic safety while walking led to an 11.2% decrease (β = −0.112) in leisure mobility, while residents who reported avoiding boredom by visiting new places showed a 9.8% decline in neighbourhood recreational activity (β = −0.098), suggesting that novelty-seeking behaviour can override spatial proximity.

In contrast, positive social and environmental perceptions were associated with higher levels of neighbourhood-based leisure. A stronger sense of belonging led to a 14.6% increase in the frequency of neighbourhood entertainment trips (β = 0.146), while the presence of attractive shopping centres contributed a further 16.9% increase (β = 0.169), highlighting the combined correlation of social connectedness and perceived commercial attractiveness. Residents who had lived in the neighbourhood since childhood were 12.6% more likely to engage in neighbourhood leisure (β = 0.126), reflecting the role of place familiarity. Physical and infrastructural factors also mattered: intersection density (β = 0.102), proximity to e-hailing stops (β = 0.108), and ease of navigating by car (β = 0.115) increased the probability of neighbourhood recreation by 10.2%, 10.8%, and 11.5%, respectively. Additionally, being male (β = 0.097), having higher income (β = 0.123), and having more driving licences per household (β = 0.095) were associated with increases in neighbourhood entertainment/leisure participation, reflecting how mobility and individual agency shape engagement with neighbourhood amenities.

Although compact neighbourhoods offer structural advantages for neighbourhood mobility, their benefits are unevenly distributed. Socioeconomic inequalities and perceptions of crime constrain the use of nearby leisure spaces, particularly among vulnerable groups. Promoting equitable access to neighbourhood amenities requires not only spatial integration but also targeted interventions to improve safety, foster social attachment, and ensure affordability in service provision.

To assess the validity of the regression models, multicollinearity was examined using the Variance Inflation Factor (VIF). In both the compact and sprawling neighbourhood models, all VIF values were below the accepted threshold of 5, indicating an absence of multicollinearity among the independent variables. This suggests that the predictors contributed unique explanatory value and that the model estimates are stable and interpretable. Within the context of Windhoek’s diverse urban form, these results confirm that the identified variables ranging from built environment characteristics to perceptual and socio-demographic factors- operate as distinct correlates of neighbourhood-based leisure mobility, rather than overlapping representations of the same underlying association.

5. Discussion

5.1. Socio-Economic, Perceptual, and Behavioural Differences Between Compact and Sprawled Neighbourhoods in Windhoek

Findings from this study confirm that neighbourhood form in Windhoek is strongly associated with socio-economic status, perceptions of place, and non-commuting travel patterns. Residents of compact areas tend to have higher incomes, education, and licence ownership, report greater safety and belonging, and engage more frequently in neighbourhood shopping. These outcomes align with research linking compact, mixed-use urban forms to enhanced accessibility, local economic participation, and improved quality of life [5,6,23]. By contrast, sprawled peripheral areas, products of apartheid-era spatial planning, accommodate lower-income households in overcrowded conditions, with fragmented infrastructure and limited local services. This inversion of density and privilege mirrors patterns observed in other post-colonial cities such as Lagos and Johannesburg [9].

These disparities cannot be accounted for by contemporary income differences alone. They reflect the enduring influence of apartheid-era urban design, which entrenched peripheral disadvantage through inadequate infrastructure provision, deliberate spatial segregation, and fragmented service access. The findings of this study demonstrate how these historical planning decisions continue to shape present-day travel behaviour and accessibility gaps. Similar features have been documented in South Africa and other post-colonial contexts, where the spatial legacies of segregation persist as structural determinants of mobility and access [12,34,35].

Through the lens of accessibility theory, these disparities reflect a persistent “accessibility gap”: compact areas combine high intersection and street length density with diverse land uses, enabling effective access to amenities, while sprawled areas often exhibit high population densities but low functional connectivity, forcing longer trips and higher travel costs. The capability approach clarifies why proximity alone does not guarantee neighbourhood travel, affordability constraints, safety concerns, and weaker social attachment in sprawled areas reduce residents’ real freedoms to use nearby amenities. Gendered and income-based differences in these constraints echo evidence from other emerging cities contexts [32,54].

Urban justice situates these findings within the enduring legacy of spatial inequality, whereby historical segregation and uneven investment continue to structure present-day mobility opportunities. In Windhoek, the advantages associated with compactness, such as shorter travel distances, enhanced connectivity, and greater access to services, are realised only when accompanied by affordability, safety, and strong social cohesion, conditions disproportionately concentrated in higher-income areas. Addressing these inequities requires more than densification; it necessitates integrated planning that simultaneously confronts the physical and socio-economic barriers that constrain equitable neighbourhood mobility.

5.2. Neighbourhood Shopping Travel Behaviour in Windhoek: A Comparative Analysis of Compact and Sprawled Urban Forms

The contrasting patterns of neighbourhood shopping in Windhoek reveal both the potential and the limitations of the 15-min city model in contexts shaped by deep socio-spatial inequality. In compact neighbourhoods, higher street connectivity, diversified land use, and broader modal choice align with the model’s principle of proximity, enabling many residents to meet daily shopping needs within a short walk or cycle. This is consistent with evidence from European cities where mixed-use, walkable environments support frequent neighbourhood shopping and reduce car dependency [27]. However, as [36,38] note, spatial form alone is insufficient: in this study, negative perceptions of crime, traffic safety, high prices, and weak social atmosphere reduced the likelihood of neighbourhood shopping even in well-connected areas, highlighting the importance of perceived quality, safety, and emotional engagement in sustaining neighbourhood economic activity.

In sprawled neighbourhoods, the departure from the 15-min city ideal is more pronounced. Despite having high residential densities, weak street connectivity, fragmented land uses, and poorly integrated amenities, this creates an “accessibility gap” between potential and realised access. Comparable patterns are observed in Bogotá and Pakistan cities, where infrastructural fragmentation and low service quality limit neighbourhood facility use despite spatial proximity [1,28]. In the present study, the strongest positive driver of neighbourhood shopping was a preference for nearby entertainment, indicating that place-based emotional value can outweigh physical distance alone. Consistent with these findings, research by [77], and [78] demonstrates that emotional and experiential dimensions, such as the enjoyment and social satisfaction derived from a destination, can exert a strong influence on neighbourhood shopping behaviour, at times overriding the effect of physical distance. Retail and leisure amenities contribute not only through the provision of goods and services but also by generating emotional and social value, which in turn encourages more frequent or locally oriented shopping among population groups.

The capability approach clarifies why these disparities persist. In compact areas, higher incomes, perceived safety, and a stronger sense of belonging enhance residents’ substantive freedoms to use nearby amenities. In sprawled areas, low incomes, unaffordable neighbourhood goods, and safety concerns operate as conversion barriers, preventing the transformation of potential accessibility into actual mobility, a pattern mirrored in studies of gendered mobility [79]. In both compact and sprawled environments, men were less likely to shop in the neighbourhood, reflecting findings from an Iranian city that men often travel longer distances for shopping, while women’s travel is more constrained by household responsibilities and safety concerns [80].

From an urban justice perspective, these findings reflect the enduring spatial legacies of apartheid, which concentrated marginalised populations in poorly serviced peripheral settlements. This mirrors conditions in other post-colonial cities such as Lagos, Johannesburg, and Maputo, where historical segregation continues to limit equitable access to everyday services [9,37,39]. Realising the accessibility, diversity, and equity principles of the 15-min city in such contexts requires more than densification. As [47,50] argue, physical planning must be accompanied by targeted interventions in affordability, safety, and social infrastructure to ensure that local shopping opportunities are genuinely accessible to all residents, not only those in affluent, compact areas.

5.3. Urban Form, Perceptions, and Recreational Travel in Windhoek

Neighbourhood-based entertainment and recreational travel in Windhoek is shaped by the city’s contrasting spatial forms, with compact and sprawled areas fostering distinct patterns of leisure mobility. While statistical models indicated comparable explanatory power across both urban forms, the determinants of participation differed markedly, reflecting the city’s historically fragmented and unequal development trajectory.

In sprawled neighbourhoods, leisure engagement is limited less by absolute distance than by a combination of infrastructural fragmentation, weak amenity integration, and socio-economic exclusion. Higher-income and car-owning households were notably less likely to use neighbourhood leisure facilities, preferring destinations beyond the neighbourhood, mirroring patterns in Nairobi, where affluent residents in peripheral areas travel further for leisure due to limited local options [81]. Perceptions of crime, absence of vibrant public life, and the functional nature of residential choice further deter local participation. This reflects what accessibility theory terms the “accessibility gap” [46,82]. Amenities may exist within geographic reach, yet poor connectivity and perceived insecurity prevent their use. Nevertheless, the positive influence of intersection density, park access, and a strong sense of belonging suggests that even marginalised areas can sustain neighbourhood leisure activity if supported by walkable infrastructure and socially inclusive public spaces, as observed in Cairo, Istanbul, and Tehran [83].

In compact neighbourhoods, leisure participation is enabled by a broader mix of factors: proximity to e-hailing services, the attractiveness of local amenities, and emotional ties such as long-term residence. These conditions align with the 15-min city’s emphasis on integrated services, multimodal options, and diverse land use [20] and resonate with findings from Oslo and European mixed-use districts, where local leisure satisfaction is higher in walkable, amenity-rich environments [27,84]. However, in Windhoek’s compact areas, as in other contexts, proximity does not automatically translate into participation. Consistent with capability approach insights, crime perception, traffic safety concerns, and affordability constraints reduce residents’ substantive freedoms to enjoy nearby leisure opportunities, particularly for women and lower-income groups, echoing findings from São Paulo and Delhi [32,54].

Across both contexts, emotional attachment, especially a strong sense of belonging, emerged as a consistent predictor of neighbourhood leisure participation. Importantly, leisure activities appear more sensitive than shopping trips to perceptions of safety and social atmosphere, as discretionary outings are often motivated by enjoyment and comfort rather than necessity.

Overall, although compact urban forms support neighbourhood-based entertainment and recreational travel by offering improved infrastructure and connectivity, their utilisation is shaped by residents’ perceptions of security, safety, affordability, and social inclusion. In sprawled neighbourhoods, both physical and social forms of marginalisation continue to limit access to neighbourhood amenities. Addressing these disparities requires a comprehensive urban strategy that integrates infrastructure development with community-oriented initiatives aimed at enhancing public safety, strengthening social cohesion, and fostering a sense of belonging.

5.4. The 15-Min City Framework as a Lens for Understanding Neighbourhood Mobility in Windhoek

This study applied the 15-min city not as a rigid planning model but as a flexible analytical framework to investigate how neighbourhood form, perceptions, and socio-economic realities shape non-commuting travel in Windhoek. By anchoring the analysis in its core principles, proximity, accessibility, diversity, and mobility, the framework enabled a grounded exploration of whether and how Windhoek’s urban neighbourhoods support short-distance, everyday travel.

The findings reveal that compact neighbourhoods come closer to fulfilling the spatial ideals of a 15-min city, offering better access to amenities, walkable environments, and multimodal infrastructure. However, these structural advantages have not been universally experienced. Concerns over security, safety, affordability, and social exclusion continue to limit neighbourhood mobility. In contrast, sprawled neighbourhoods, often dense but poorly connected and under-serviced, fall significantly short of the model’s ideals. However, residents’ place attachment, perceived recreational value, and informal mobility strategies indicate latent potential for neighbourhood-based travel, even in fragmented urban peripheries.

In this way, the 15-min city served not only as a theoretical anchor, but also as a critical lens to localise global mobility debates. It reveals how proximity alone does not guarantee accessibility and highlights the need to integrate social, perceptual, and historical inequalities into any model of inclusive neighbourhood mobility. In Windhoek, achieving the spirit of a 15-min city will depend less on replicating its form and more on addressing the embedded barriers–economic, perceptual, and infrastructural–that shape how people move through their communities.

6. Study Limitations

While this study provides robust insights into the interplay between neighbourhood form, socioeconomic conditions, and non-commuting travel behaviour in Windhoek, several limitations should be acknowledged. First, the use of a cross-sectional design inherently limits the capacity to establish causal relationships or observe temporal patterns in travel behaviour. However, by incorporating comparative statistical techniques and stratified modelling across compact and sprawled neighbourhoods, this study captures structural patterns and associations that offer analytically meaningful insights into the socio-spatial determinants of non-commuting mobility.

Second, while the sample is limited to nine constituencies within Windhoek, the adoption of a stratified random sampling approach, ensuring equal representation of residents from both compact and sprawled contexts, provides a robust basis for intra-urban comparison, even if generalisability beyond Windhoek requires caution. At the same time, the equal allocation of 500 respondents per category does not reflect the actual population distribution, which may introduce sampling bias, particularly for proportion-based tests. This limitation has been acknowledged, though the approach was retained to ensure comparability between neighbourhood types.

Third, although the study did not explicitly model informal transport systems, an important mode in African cities, mobility patterns were partially captured through indicators such as car ownership, driving licence distribution, and e-hailing usage, offering indirect insight into modal choice and accessibility constraints.

Fourth, neighbourhood shopping behaviour was operationalised as a binary outcome (neighbourhood vs. distant). Although this reduces a complex behaviour to a simplified form, the measure was appropriate for the study’s objective of distinguishing whether residents primarily rely on local amenities or travel beyond their neighbourhood for shopping. This approach provided a consistent basis for comparative analysis, while future research may employ scaled measures to capture the frequency and intensity of neighbourhood shopping in greater detail.

Finally, the 15-min city framework was employed as a guiding conceptual lens rather than a prescriptive metric-based evaluation. While this limits the empirical assessment of full compliance with its principles, it enabled the localisation and critical adaptation of the model to Windhoek’s unique spatial and socioeconomic context. Taken together, these mitigations strengthen the study’s methodological rigour and support the credibility of its findings within the complex urban realities of South Africa.

7. Conclusions

This study investigated how urban form shapes neighbourhood-level travel behaviour in Windhoek, examining socio-economic characteristics, residential perceptions, and built environment features across compact and sprawled neighbourhoods. The findings demonstrate that the city’s apartheid-era spatial legacy continues to structure mobility outcomes: compact areas, with greater land-use diversity, connectivity, and service integration, support more frequent neighbourhood shopping and recreational travel, while sprawled peripheries remain constrained by infrastructural fragmentation, socio-economic exclusion, and negative place perceptions.

Across both contexts, emotional attachment, particularly a strong sense of belonging and long-term residence, emerged as a consistent enabler of neighbourhood engagement, highlighting that mobility is as much social as it is spatial. However, safety concerns, traffic danger, affordability constraints, and perceived quality deficits inhibited local travel, even in well-connected areas. These results demonstrate that proximity alone, central to the 15-min city model, does not ensure realised accessibility; rather, compactness is effective only when supported by enabling social, economic, and perceptual conditions.

The opportunity of this study lies in providing rare empirical evidence from Southern Africa, where urban mobility research has often been underrepresented in global debates on compactness, accessibility, and the 15-min city. By situating Windhoek within the broader discourse on urban form and mobility, the study opens new perspectives on how post-colonial legacies and socio-economic inequalities mediate the translation of physical proximity into lived accessibility.

This research makes three key contributions to urban mobility studies. First, it provides empirical evidence from Southern Africa, testing the 15-min city concept in a socio-spatially unequal context and revealing persistent accessibility gaps despite physical proximity. Second, by integrating accessibility theory, the capability approach, and urban justice, it shows how structural, socio-economic, and perceptual constraints interact to shape neighbourhood mobility, advancing theoretical understanding of the conditional nature of compactness. Third, it challenges the assumption of uniform mobility determinants by demonstrating that the drivers of shopping and leisure behaviour differ fundamentally between compact and sprawled neighbourhoods, reflecting unique historical and infrastructural legacies.

Overall, the findings show that mobility patterns in Windhoek are shaped by apartheid-era spatial legacies, where institutionalised segregation and uneven investment created enduring disparities between compact and sprawled neighbourhoods. This underscores that urban form in African cities must be interpreted within its historical context, as the legacies of exclusion continue to structure everyday mobility.

The findings call for place-sensitive strategies that extend beyond densification. In sprawled areas, priority actions include improving street and intersection connectivity, integrating retail and recreational facilities within walkable distances, and investing in public safety measures. In compact areas, policy should focus on maintaining affordability, enhancing traffic safety, and strengthening the vibrancy and perceived quality of local amenities. Across both contexts, participatory design processes, public realm upgrades, and cultural programming can help translate potential proximity benefits into equitable, lived accessibility.

Further research should examine temporal variations in mobility, track the impacts of targeted urban interventions over time, and compare these features across other post-colonial cities. Mixed-methods approaches incorporating resident narratives would provide deeper insights into how accessibility, capability, and liveability are experienced, potentially challenging formal planning models of the 15-min city.

Future research could investigate temporal variations in mobility, assess the impacts of targeted urban interventions over time, and undertake comparative analyses across other post-colonial cities. Mixed-methods approaches that incorporate resident narratives would provide deeper insights into the lived experience of accessibility, capability, and liveability, and may serve to challenge formal planning models of the 15-min city. The contribution of this study lies in the application of a multi-theoretical framework to an African context, integrating accessibility theory, the capability approach, and urban justice to advance an understanding of neighbourhood mobility. In doing so, it demonstrates how context-specific socio-spatial inequalities shape travel behaviour and offers a basis for adapting global planning models to the realities of rapidly urbanising cities characterised by socio-spatial inequalities.