Crosswalks of Italy — pointzerofive

Abstract

We measure the provision of pedestrian crossings in 14 Italian provincial capitals by combining 22,290 mapped crossings from OpenStreetMap with ISTAT population figures. Crossing density (crossings per 10 000 inhabitants) varies by a factor of 7.9 across the sample. The dominant pattern is geographic: northern cities average 41 crossings per 10 000 inhabitants against 11.8 in the south, and the two groups separate perfectly in rank — an outcome with probability 0.033% under a null of no geographic association. City size does not explain the variation (r = -0.07). We interpret the gap as a difference in sustained local investment rather than urban scale, and note OpenStreetMap coverage as the principal threat to validity.

1 Introduction

A painted crosswalk is the cheapest, most basic piece of pedestrian infrastructure a municipality can provide. Because Italian cities manage their own streets, crossing provision is a direct, comparable trace of local policy. The question this piece asks is simple: does where you live determine how safely you can cross the street?

2 Data and measures

Pedestrian crossings were extracted from OpenStreetMap (highway=crossing nodes within municipal boundaries); resident population per municipality comes from ISTAT (2023). The unit of analysis is the provincial capital; the outcome is crossings per 10 000 inhabitants. The sample covers 14 capitals (6 northern, 2 central, 6 southern by the standard ISTAT macro-area classification), selected for pipeline coverage — not randomly drawn from all 107 capitals.

3 Results

Figure 1. Crosswalk density in 14 Italian provincial capitals, sorted by score. Dashed line marks the sample average (26.6). Source: OpenStreetMap © OpenStreetMap contributors (ODbL), ISTAT 2023. n = 14.

The spread is wide: Bolzano provides 58.4 crossings per 10 000 inhabitants, 7.9× the provision of Crotone (7.4). Notably, both extremes are small cities — the range is not driven by metropolitan scale.

Figure 2. The macro-area split. Southern capitals (red) against northern capitals (dark); central capitals muted. Dashed lines mark group means (north 41, south 11.8). The two groups do not overlap in rank.

The 6 northern capitals average 41 crossings per 10 000 inhabitants; the 6 southern capitals average 11.8 — a 3.5× gap. More telling than the means is the complete rank separation visible in Figure 2: the weakest northern city still outscores the strongest southern one.

4 Robustness

Sampling noise. With 14 cities, formal distributional tests are fragile, so we use an exact permutation argument instead. Under the null hypothesis that macro-area is unrelated to crosswalk provision, all 3,003 assignments of 6 cities to the bottom 6 ranks are equally likely, and exactly one of them produces the observed full separation. The probability of the observed pattern under the null is therefore 0.033% — far below conventional significance thresholds, with no distributional assumptions required.

City size. The Pearson correlation between population and score is r = -0.07, indistinguishable from zero in this sample. Roma and Napoli do underperform the sample average, but as part of the broader central–southern pattern rather than as a big-city effect.

Selection. The sample was assembled for pipeline coverage, not drawn randomly. The permutation result is exact conditional on this sample; extending the claim to all 107 provincial capitals requires scaling the pipeline, which is planned follow-up work.

5 Discussion

Provision is not safety: this piece measures whether cities supply basic pedestrian infrastructure, not injury outcomes. Within that limit, the result is hard to explain away. Neither chance (0.033% under the null) nor city size (r = -0.07) accounts for the north–south gap, which is consistent with long-documented regional differences in public investment. The most important caveat is measurement: OpenStreetMap coverage may itself vary regionally, and systematically under-mapped southern cities would exaggerate the gap. A completeness check against sampled satellite imagery is the planned next step, together with joining ISTAT road-accident microdata to test whether provision predicts outcomes.

Data availability

Dataset: data/crosswalks-italy.json in the story repository; extraction pipeline: notebooks/crosswalks_pipeline.py. A Jupyter notebook reproducing every figure and statistic in this piece can be opened directly in Google Colab via the “Notebook” link above. Map data © OpenStreetMap contributors, available under the Open Database License (ODbL).

Data source: OpenStreetMap / ISTAT 2023 · Last updated: 2026-02-27