Land use mix (LUM) is widely promoted as a planning lever to increase walking, yet the empirical literature reports non-universal, sometimes sign-flipping associations: proximity-based mix appears to raise walking trip frequency, while quantity- or function-based mix can reduce walking distance and duration, and simulation work suggests higher mix may raise trip counts without raising total walking distance. It remains open whether and how different dimensions of mixed land use causally affect distinct walking outcomes once residential self-selection — the tendency of people who already prefer walking to choose walkable neighborhoods — is accounted for. Recent quasi-experimental evidence establishes that composite walkability causally increases total step volume, but does not decompose the role of land use mix itself, and the strongest LUM-specific within-person analyses find null or negative within-individual effects in parts of their sensitivity space. The open problem is to causally identify the sign, magnitude, and mechanism of each LUM dimension's effect on each walking outcome, and whether mixed-use policy raises overall walking volume or merely redistributes it.

Background

Land use mix (LUM) denotes the diversity of land uses (residential, commercial, office, recreational, etc.) within a neighborhood, and is a central ingredient of 'walkability' — the composite quality of an environment that supports walking, typically indexed by combining land-use diversity, street connectivity, residential density, and transit access. Planners have long assumed that mixing uses shortens the distance between origins and daily destinations and therefore produces more walking, a belief embedded in zoning reforms, transit-oriented development, and '15-minute city' proposals (visions in which residents can reach all daily needs within a short walk or cycle). LUM is usually quantified with entropy-type indices (−∑kpkln⁡pk-\sum_k p_k \ln p_k over the shares pkp_k of land area or floor space in use category kk), but these indices are heterogeneous across studies: mix can be measured in terms of the quantities of different uses, the distances to diverse destinations, or the functional compatibility of co-located uses — and the choice measurably changes estimated associations (Hajna, Dasgupta, Joseph & Ross 2014; Mavoa et al. 2018). Walking outcomes are likewise multidimensional: trip frequency, total walking time (duration), and total walking distance can move in different directions, which matters for public health because total activity volume, not trip count, is the health-relevant quantity. The empirical literature is dominated by cross-sectional designs, and a mature body of work on residential self-selection — the confounding that arises because households with pre-existing walking preferences and attitudes choose to live in mixed, walkable neighborhoods — has shown that such associations cannot be read causally (Wang & Lin 2014; Lin, Wang & Guan 2017; Guan & Wang 2019). Two recent sources sharpen the problem. First, Xu, Ta, Chai & Wu (2024) built a multidimensional framework distinguishing mix in quantity, mix in distance (proximity), and mix in function, against multiple walking indicators (frequency, time, distance) in Beijing-area data, exploiting China's historically constrained housing allocation to mitigate self-selection; they found proximity-based mix to be the sole environmental factor increasing walking frequency, while quantity- and function-based mix could reduce walking distance and duration, with apparently increased walking time near functionally mixed but low-compatibility environments plausibly reflecting compensatory avoidance of negative externalities such as pollution and noise. Second, Yang (2023) simulated a hypothetical grid city and found that a higher level of mix raises the number of walking trips but not total walking distance, with the land-use–walking relationship substantially modified by context. Adjacent quasi-experimental work does not settle the question: a relocation quasi-experiment in Calgary linked composite walkability change to change in transportation walking (McCormack, McLaren, Salvo & Blackstaffe 2017); within-between analyses (models separating within-person change over time from between-person differences) of two Dutch cohorts with ten years of follow-up found positive between-person LUM–walking associations but no positive within-person effects, and even negative within-person associations at some buffer scales (Noordzij et al. 2021); and a countrywide smartphone-based natural experiment showed that composite walkability causally increases daily step counts after relocation (Althoff et al. 2025) — but that study measures a composite exposure and a single volume outcome, and explicitly names land use mix as a policy topic requiring further research. Sign heterogeneity has also been replicated cross-sectionally, e.g. in Seoul, where increasing mix raises the probability of choosing walking over driving up to a complexity threshold beyond which the effect reverses (Seong, Lee & Choi 2021), and longitudinal LUM studies continue to appear for specific populations and contexts (Saboya, D'Orsi & Keivani 2024, older adults in Brazil; Huang, Lu, Guo & Zhang 2026, big-data relocation quasi-experiment). The unresolved scientific question is therefore not whether walkability composites correlate (or causally matter) for walking volume, but whether mixed land use itself — decomposed into its quantity, proximity, and function dimensions — causally raises distinct walking outcomes, and whether it raises overall walking volume at all.

Problem Statement

Determine, with causal identification that addresses residential self-selection, whether and how mixed land use affects walking behavior. Specifically: (1) estimate the causal effect of each dimension of land use mix — quantity-based mix (diversity of use quantities), proximity or distance-based mix (diversity of distances to different uses), and function or compatibility-based mix — on each of the distinct walking outcomes of trip frequency, total walking duration, and total walking distance; parts (2) and (3) follow from the same causal estimate as (1): (2) report the implied verdict on whether increasing land use mix raises overall walking volume — measured as total walking duration, rather than trip count — or instead increases trip frequency while leaving total walking distance unchanged or reduced, as suggested by Xu et al. (2024) and Yang (2023); and (3) report the implied verdict on whether the reported pattern — proximity-based mix driving frequency, quantity- and function-based mix diminishing distance and duration, and elevated walking time near functionally mixed but low-compatibility environments reflecting compensatory avoidance of negative externalities — reflects causal effects of the environment rather than confounding by who chooses to live in mixed-use neighborhoods. The causal identification must separate the effects of land use mix from co-varying built-environment attributes (density, transit access, street connectivity), make the land-use-mix measurement explicit, and name the population and context over which the conclusion holds; the target is the defensible identification strategy and analysis protocol for these dimension-by-outcome effects, rather than an estimate executed on a specific dataset.

References

  1. Xu, Tiantian; Ta, Na; Chai, Yanwei; Wu, Jiayu (2024). 'Complex associations between multidimensional land use mix and walking: Walk more or farther?' Transactions in Urban Data, Science, and Technology. DOI: 10.1177/27541231241248852. https://doi.org/10.1177/27541231241248852
  2. Yang, Yong (2023). 'Land use density, land use mix, and walking: insight from a simple theoretical model.' medRxiv preprint (v1, unpublished). DOI: 10.1101/2023.11.29.23299138. https://doi.org/10.1101/2023.11.29.23299138
  3. Althoff, Tim; Ivanovic, Boris; King, Abby C.; Hicks, Jennifer L.; Delp, Scott L.; Leskovec, Jure (2025). 'Countrywide natural experiment links built environment to physical activity.' Nature. DOI: 10.1038/s41586-025-09321-3. https://doi.org/10.1038/s41586-025-09321-3
  4. Noordzij, J. M.; Beenackers, M. A.; Oude Groeniger, J.; Timmermans, E. J.; Motoc, I.; Huisman, M.; van Lenthe, F. J. (2021). 'Land use mix and physical activity in middle-aged and older adults: a longitudinal study examining changes in land use mix in two Dutch cohorts.' International Journal of Behavioral Nutrition and Physical Activity. DOI: 10.1186/s12966-021-01083-1. https://doi.org/10.1186/s12966-021-01083-1
  5. McCormack, Gavin R.; McLaren, Lindsay; Salvo, Grazia; Blackstaffe, Anita (2017). 'Changes in Objectively-Determined Walkability and Physical Activity in Adults: A Quasi-Longitudinal Residential Relocation Study.' International Journal of Environmental Research and Public Health 14(5):551. DOI: 10.3390/ijerph14050551. https://doi.org/10.3390/ijerph14050551
  6. Lin, Tao; Wang, Donggen; Guan, Xiaodong (2017). 'The built environment, travel attitude, and travel behavior: Residential self-selection or residential determination?' Journal of Transport Geography. DOI: 10.1016/j.jtrangeo.2017.10.004. https://doi.org/10.1016/j.jtrangeo.2017.10.004
  7. Guan, Xiaodong; Wang, Donggen (2019). 'Residential self-selection in the built environment-travel behavior connection: Whose self-selection?' Transportation Research Part D: Transport and Environment. DOI: 10.1016/j.trd.2018.10.015. https://doi.org/10.1016/j.trd.2018.10.015
  8. Wang, Donggen; Lin, Tao (2014). 'Residential self-selection, built environment, and travel behavior in the Chinese context.' Journal of Transport and Land Use 7(3). DOI: 10.5198/jtlu.v7i3.486. https://doi.org/10.5198/jtlu.v7i3.486
  9. Hajna, Samantha; Dasgupta, Kaberi; Joseph, Lawrence; Ross, Nancy A. (2014). 'A call for caution and transparency in the calculation of land use mix: Measurement bias in the estimation of associations between land use mix and physical activity.' Health & Place. DOI: 10.1016/j.healthplace.2014.06.002. https://doi.org/10.1016/j.healthplace.2014.06.002
  10. Mavoa, Suzanne; Eagleson, Serryn; Badland, Hannah M.; Gunn, Lucy; Boulange, Claire; Stewart, Joshua; Giles-Corti, Billie (2018). 'Identifying appropriate land-use mix measures for use in a national walkability index.' Journal of Transport and Land Use. DOI: 10.5198/jtlu.2018.1132. https://doi.org/10.5198/jtlu.2018.1132
  11. Seong, Eun Yeong; Lee, Nam Hwi; Choi, Chang Gyu (2021). 'Relationship between Land Use Mix and Walking Choice in High-Density Cities: A Review of Walking in Seoul, South Korea.' Sustainability 13(2):810. DOI: 10.3390/su13020810. https://doi.org/10.3390/su13020810
  12. Saboya, Renato Tibiriçá de; D'Orsi, Eleonora; Keivani, Ramin (2024). 'Promoting active aging in Brazil: A longitudinal study of land use mix and utilitarian walking in older adults.' Cities. DOI: 10.1016/j.cities.2024.105318. https://doi.org/10.1016/j.cities.2024.105318
  13. Huang, Nuoxian; Lu, Daming; Guo, Wenbo; Zhang, Wenjia (2026). 'Tracking walking change post-relocation: A big data quasi-experiment of built environment effects.' Transportation Research Part D: Transport and Environment. DOI: 10.1016/j.trd.2026.105529. https://doi.org/10.1016/j.trd.2026.105529