Assessing ADA Accessibility Gaps in the NYC Subway System

This case study demonstrates advanced statistical analysis using the subway-access library. The central question: do accessibility initiatives actually increase equity if the elevator is never running?

Abstract

The ADA of 1990 mandates equal access to public transportation, yet more than three decades later, the NYC subway system remains largely noncompliant. This study provides a tract-level spatial equity analysis of ADA accessibility across all five boroughs using April 2026 station data from the MTA. Of 493 active stations, only 157 (31.8%) are ADA-accessible. We define coverage using an 800-meter Euclidean catchment and overlay ACS 2023 demographic estimates for 2,317 census tracts (population: 8,507,596).

A composite need score identifies communities where inaccessibility imposes the greatest burden. Reliability-weighted coverage discounts station coverage by actual elevator uptime over 12 months, finding that 49 stations operated below 95% uptime and Manhattan's effective coverage drops from 77% to 71%.

4,717,140 New Yorkers (55.4%) reside in tracts with no ADA-accessible station within walking distance, with Queens bearing the largest absolute gap (1,752,073 residents, 22% tract coverage).

System-Wide Coverage

The borough-level breakdown reveals stark disparities. Manhattan leads with 77% tract coverage, while Queens covers just 22% of its tracts despite being the largest borough by area. The Bronx sits at 34%, Brooklyn at 35%, and Staten Island---with only one subway line---at 9%.

When weighted by population, the absolute numbers are striking: Queens alone accounts for 1.75 million residents in uncovered tracts, more than the entire population of Philadelphia.

Borough Disparities

The coverage map reveals that inaccessibility is not randomly distributed but follows identifiable geographic corridors. Eastern Queens, southern Brooklyn, and large sections of the Bronx form contiguous zones of zero coverage.

The gap score choropleth overlays demographic need on coverage status. High-need, uncovered tracts cluster in specific areas: southeast Queens (high senior population), the south Bronx (high poverty and disability rates), and central Brooklyn.

Elevator Reliability

A station is only functionally accessible when its elevator is operational. We extend nominal coverage with reliability-weighted coverage that discounts by actual elevator uptime. Across the system, 49 of 157 accessible stations operated below 95% uptime during the 12-month observation window. The impact is concentrated: Manhattan's effective coverage drops from 77% to 71%, meaning six percentage points of nominally covered tracts lose reliable access.

Temporal Progression

Using a substantially sourced upgrade timeline (101 of 157 stations with documented upgrade years from MTA press releases and Capital Program records), we reconstruct the historical progression of ADA coverage. The acceleration in the 2020--2024 Capital Program is visible, but the pace remains insufficient for full system coverage within the next two decades.

The treatment-control balance check for the difference-in-differences specification shows that treated tracts (those receiving new accessible stations) and control tracts are well-matched on observable demographics, supporting the parallel trends assumption.

Spatial Analysis

Distance from the nearest accessible station follows a steep decay curve, with most covered tracts within 400 meters and a long tail extending past 2 km in underserved areas.

The scatter plot of gap scores against distance to nearest accessible station confirms the expected positive relationship: tracts farther from accessible stations tend to have higher gap scores, but with substantial variation driven by demographic composition.

The full correlation matrix reveals that senior population rate is the strongest demographic predictor of gap score, followed by disability rate and poverty rate. An OLS equity regression confirms this: senior rate is the most significant predictor (R-squared = 0.202, F(3, 2313) = 108.83, P < .001).

Geographic Equity

The bivariate map of gap scores against poverty rates reveals that the highest-need communities---those with both high poverty and poor coverage---cluster in specific corridors: the south Bronx, eastern Queens, and central Brooklyn.

A similar pattern emerges for disability rates. Global Moran's I statistics confirm significant spatial clustering of gap scores (I = 0.23, z = 40.87, P < .001), need scores (I = 0.20, z = 33.91, P < .001), and disability rates (I = 0.28, z = 48.92, P < .001).

Discussion

These findings suggest that capital investment in new ADA stations without commensurate maintenance funding produces nominal compliance that does not translate to functional access. Spatially targeted intervention in high-need clusters would yield greater equity returns than system-wide uniform upgrades.

The reliability gap is particularly concerning: a station that appears accessible on the MTA's website but has a broken elevator 10% of the time is not a reliable service. For someone planning a commute in a wheelchair, a 90% uptime rate means roughly three days per month of unexpected inaccessibility.

The full case study, including the difference-in-differences specification, model diagnostics, and supplementary analyses, is available in the subway-access examples.

Built with subway-access v0.4.1 and nyc-geo-toolkit.