Assessing walking to school inequalities in a Latin American city

Pizzol, Bruna; Giannotti, Mariana · 2026 · DOAJ

DOI: 10.1007/s44327-026-00253-9

archive: archived pipeline: cataloged verified

Get this paper ↗ (DOI — opens at the source; we link to it, we don't host it)

Summary

This study investigates how socio-spatial inequalities shape walking behavior for school commutes in São Paulo, a highly unequal city in the Global South. The research challenges traditional walkability literature, which largely originates from the Global North and assumes that improved infrastructure induces voluntary walking. In contrast, this paper posits that in contexts of deep inequality, walking is often a "compulsory" mode driven by a lack of alternatives rather than infrastructure quality. The authors aim to determine the relative influence of built environment factors versus socioeconomic determinants on travel mode choice, providing empirical evidence for the concept of "compulsory pedestrians." The methodology combines data from São Paulo’s 2007 and 2017 origin–destination surveys with open geospatial datasets. To analyze the spatial context, the authors generated Voronoi diagrams centered on schools to aggregate walkability metrics, such as sidewalk width, street slope, land use mix, and density indicators. The study employs a Random Forest machine learning model to predict travel mode choice, utilizing 15 independent variables categorized into socioeconomic characteristics (e.g., income, car ownership, school type), trip attributes (e.g., travel time), and built environment factors. To interpret the complex, nonlinear relationships captured by the model, the authors applied Shapley Additive Explanations (SHAP), comparing these results with traditional binomial logit models. The findings reveal that income, car ownership, and school type are the dominant predictors of walking to school, while built environment factors have limited influence on walking behavior. The analysis confirms that sidewalk and infrastructure conditions are significantly worse in peripheral areas, where most low-income and non-white students reside. Despite these poor conditions, students in these areas walk to school at high rates, supporting the notion that structural inequality overrides infrastructure quality in determining mobility patterns. This contrasts sharply with literature from more equitable contexts, where walkability infrastructure typically predicts walking rates. The significance of this research lies in its challenge to universal assumptions about walkability and active transportation. The authors conclude that for cities in the Global South, walkability improvements must be targeted and equity-driven, rather than assuming that infrastructure alone can induce behavioral shifts. The study demonstrates that machine learning paired with explainability tools like SHAP offers novel insights into the complex interactions between social constraints and the built environment. Ultimately, the paper argues that policy interventions must address the structural inequalities that force low-income populations into "compulsory" pedestrianism, rather than focusing solely on physical infrastructure enhancements.

Provenance

The full processing record for this entry. Every stage of this paper's journey through the pipeline is logged — what ran, with which tool and model, how many attempts it took, and when it last completed.

Summary generated by qwen3.6-27b-prismaquant on 2026-06-26; verification: verified.

Topics

Ranked by relevance to this paper. Hover a topic for its definition.

• demographic disparities