A 2019 study of 10,794 middle-aged adults in Brisbane, Australia applied isometric substitution models — borrowed from time-use epidemiology — to land-use proportions, finding that replacing 1% of residential or 'other' land with commercial land within a 1 km street-network buffer of home is associated with roughly 1.5–2.1% higher odds of any and of sufficient weekly walking. Whether these associations are causal effects of land-use reallocation, and how they are modified by vertical mixed use, land-use intensity, and recreational-land quality, remains unresolved: the evidence is cross-sectional, based on cadastral predominant-use coding and self-reported walking, in one city's predominantly residential neighborhoods. No longitudinal or quasi-experimental study has yet estimated land-use substitution effects on walking, and adjacent advances — compositional log-ratio measurement of land-use mix, longitudinal entropy-based studies — leave the substitution-specific question untouched. The problem asks for causally credible substitution estimates, their modification by land-use composition and quality, and their robustness across spatial scales, measurement approaches, and urban contexts.

Background

Walking is the most common form of physical activity, and whether city planning can reliably increase it is a standing question for public health and urban design. Walkability — the extent to which the built environment supports walking — is usually summarized by measures of residential density, street connectivity, and land-use mix. Land-use mix captures the diversity of activities (residential, commercial, recreational, and so on) within an area; its most common operationalization is an entropy index, i.e., the Shannon entropy of the area shares of the different land uses, normalized to range from 0 (single use) to 1 (equal shares of all uses). Entropy indices have well-known weaknesses: they are insensitive to which uses are combined and in which direction the composition departs from equality, so areas with very different land-use compositions can receive identical walkability scores (Mavoa et al. 2018).

Because the area shares of land uses within a fixed area are compositional — they always sum to 100% — increasing one use necessarily displaces others, and a regression of walking on any single share cannot say which use gives way. Time-use epidemiology faced the same structure for daily activity time and addressed it with isotemporal substitution models, which estimate the association of swapping a fixed amount of one activity for another while total time is held constant. Sugiyama et al. (2019) imported this logic to land use with isometric substitution models: regressions that estimate the association of increasing one land-use share while simultaneously decreasing another by the same amount, holding total area constant. Using baseline data from HABITAT, a longitudinal cohort of middle-aged adults in Brisbane, Australia (n = 10,794), with cadastral land-use data that assign each land parcel a single predominant use, self-reported weekly walking minutes, and 1 km street-network buffers (the area reachable along the street network within 1 km of the home), the study estimated all twelve pairwise substitution scenarios among residential, commercial/institutional, recreational, and 'other' land. Cross-sectional logistic models found that replacing 1% of residential or other land with commercial/institutional land was associated with higher odds of any walking and of sufficient walking (at least 150 minutes per week; odds ratios of roughly 1.015–1.021 per 1% replaced), and the reverse substitution with correspondingly lower odds; recreational land showed little consistent association.

The authors interpret these estimates as the potential impact of land-use change, but, as their discussion acknowledges, several questions remain open. The central one is causal: the design is cross-sectional, so the associations could reflect residential self-selection (walkers preferring to live in mixed-use neighborhoods) or unmeasured confounding by pedestrian infrastructure, transit access, or area disadvantage, rather than effects of reallocation. Longitudinal evidence on the broader question exists — a systematic review of 21 prospective cohorts and 30 natural experiments (Kärmeniemi et al. 2018) found that changes in the built environment, including land-use mix, were associated with changes in physical activity — but no study has estimated substitution effects among land-use proportions, and the methodological literature on residential self-selection (Heinen et al. 2018) and on planning-informed natural experiments for healthy cities (Sun et al. 2023) frames the identification problem without answering it. A fixed-effects analysis of changes in crime and walking for transport on the same HABITAT cohort (Reid et al. 2024) demonstrates that the within-person longitudinal template is feasible there, but it does not model land use.

Second, predominant cadastral coding cannot see vertical mixed use (ground-floor commercial beneath upper-floor residential), the functional intensity of commercial parcels, or the quality attributes of recreational land (size, amenities, maintenance, accessibility), so whether and how these features modify the substitution–walking relationships is unknown. Third, robustness is untested: results may depend on the 1 km buffer scale, on self-reported rather than objectively (accelerometer-) measured walking, and on the estimation method — compositional data analysis (CoDA), the log-ratio-based methodology standard for compositional variables, was not applied; a recent CoDA framework for measuring land-use mix with isometric log-ratio balances (Kahatagahawatte et al. 2026) shows that entropy conflates compositionally distinct areas, but it has not been linked to walking outcomes. Finally, generalizability beyond predominantly residential neighborhoods of a single city — to denser urban cores, industrial or agricultural areas, and other countries' urban forms — is unestablished.

Problem Statement

Determine whether the land-use substitution–walking associations found by cross-sectional isometric substitution analysis reflect causal effects of land-use reallocation, and characterize how they are modified by land-use composition and quality. Concretely, the problem asks for:

  1. Causally credible estimates of the effect on residents' walking of reallocating land among residential, commercial/institutional, recreational, and other uses, preserving the compositional substitution structure (shares of mutually exclusive uses summing to the total buffer area), in place of cross-sectional between-neighborhood comparisons. Component 1 may be resolved affirmatively, by such causally credible estimates, or negatively, by evidence beyond the design limitation already stated in the Background showing that the observed associations are substantially confounded; a negative resolution closes component 1 and leaves nothing for components 2 and 3 to qualify, so those two components are required only when component 1 is resolved affirmatively.

  2. Estimates of how these substitution effects are modified by features invisible to predominant cadastral coding: vertical mixed use (ground-floor commercial with upper-floor residential), the intensity or functional relevance of commercial parcels, and the qualitative features of recreational land (size, amenities, maintenance, accessibility).

  3. Evidence on the robustness and transportability of the estimates: sensitivity to spatial scale (e.g., buffer size), to self-reported versus objectively measured walking, to compositional (log-ratio) versus isometric substitution estimation, and to contexts beyond predominantly residential neighborhoods of a single city — denser urban cores, areas with industrial or agricultural dominance, and other national urban forms.

These components serve one overarching objective: to establish whether and how the composition of urban land use can be used as a lever to increase population walking.

References

  1. Takemi Sugiyama, Jerome N. Rachele, Lucy D. Gunn, Nicola W. Burton, Wendy J. Brown, Gavin Turrell, 'Land use proportion and walking: Application of isometric substitution analysis', Health & Place 57 (2019) 352–357, DOI: 10.1016/j.healthplace.2018.12.004, https://doi.org/10.1016/j.healthplace.2018.12.004
  2. Mikko Kärmeniemi, Tiina Lankila, Tiina Ikäheimo, Heli Koivumaa-Honkanen, Raija Korpelainen, 'The Built Environment as a Determinant of Physical Activity: A Systematic Review of Longitudinal Studies and Natural Experiments', Annals of Behavioral Medicine 52(3) (2018) 239–251, DOI: 10.1093/abm/kax043, https://doi.org/10.1093/abm/kax043
  3. Suzanne Mavoa, Serryn Eagleson, Hannah M. Badland, Lucy Gunn, Claire Boulange, Joshua Stewart, Billie Giles-Corti, 'Identifying appropriate land-use mix measures for use in a national walkability index', Journal of Transport and Land Use 11(1) (2018), DOI: 10.5198/jtlu.2018.1132, https://doi.org/10.5198/jtlu.2018.1132
  4. Eva Heinen, Bert van Wee, Jenna Panter, Roger Mackett, David Ogilvie, 'Residential self-selection in quasi-experimental and natural experimental studies: An extended conceptualization of the relationship between the built environment and travel behavior', Journal of Transport and Land Use 11(1) (2018), DOI: 10.5198/jtlu.2018.1165, https://doi.org/10.5198/jtlu.2018.1165
  5. Guibo Sun, Eun Yeong Choe, Chris Webster, 'Natural experiments in healthy cities research: how can urban planning and design knowledge reinforce the causal inference?', Town Planning Review 94(1) (2023) 87–108, DOI: 10.3828/tpr.2022.14, https://doi.org/10.3828/tpr.2022.14
  6. Renato Tibiriçá de Saboya, Eleonora D'Orsi, Ramin Keivani, 'Promoting active aging in Brazil: A longitudinal study of land use mix and utilitarian walking in older adults', Cities 154 (2024) 105318, DOI: 10.1016/j.cities.2024.105318, https://doi.org/10.1016/j.cities.2024.105318
  7. Bhagya Kahatagahawatte, Ryoji Matsunaka, Nobuhiro Uno, Tomoki Nishigaki, 'Introducing A Novel Compositional Data Analysis Framework for Measuring Land Use Mix in Station Areas', Journal of Contemporary Urban Affairs 10(1) (2026) 153, DOI: 10.25034/ijcua.2026.v10n1-b7, https://doi.org/10.25034/ijcua.2026.v10n1-b7
  8. Yong Yang, 'Land use density, land use mix, and walking: insight from a simple theoretical model', medRxiv preprint (2023), DOI: 10.1101/2023.11.29.23299138, https://doi.org/10.1101/2023.11.29.23299138
  9. Rebecca A. Reid, Sarah Foster, Suzanne Mavoa, Jerome N. Rachele, 'Associations between changes in crime and changes in walking for transport with effect measure modification by gender: A fixed-effects analysis of the multilevel longitudinal HABITAT study (2007–2016)', Health & Place 85 (2024) 103163, DOI: 10.1016/j.healthplace.2023.103163, https://doi.org/10.1016/j.healthplace.2023.103163
  10. Di Wu, 'Towards Resilient Urban Design: Revealing the Impacts of Built Environment on Physical Activity Amidst Climate Change', Buildings 15(19) (2025) 3470, DOI: 10.3390/buildings15193470, https://doi.org/10.3390/buildings15193470