Transportation Research Part A · 2024
Scroll to explore
Core Hypothesis
The central question driving this research
"The introduction of shared E-scooters does not increase — or poorly increases — accessibility to different opportunities compared to available modes of transportation, especially for disadvantaged population groups."
Methodology
Using entirely open-source data from Louisville, KY — enabling transparent, reproducible analysis of transport equity at the census block level.
Five primary sources: sociodemographic census data, 390K scooter trip records (Aug 2018–Jan 2020), Points of Interest from OpenStreetMap, road network data, and GTFS public transit feeds.
Cumulative accessibility measured as the number of opportunities reachable within 5–15 min by each mode. Paired with Potential Mobility Index (PMI): average aerial speed across all census block pairs.
1,903 main scenarios × 4 accessibility thresholds × 252 census blocks = over 1.9 million scenario evaluations. Scooter accessibility compared against walking, cycling, PT, car, and TNC.
Population Classification
Communities were classified along two axes — household income and car ownership — to identify transport-disadvantaged groups. Click each quarter to explore its characteristics.
High income, low car dependency pressure. Concentrated in the east of Louisville, these 89 census blocks represent the city's wealthy population. They have high accessibility to opportunities and are least at risk of transport-related social exclusion.
Low income but also low car ownership — potentially through choice, density, or proximity to transit. These 19 blocks represent an interesting transitional group, often found near denser urban cores with better PT coverage.
The largest and most concerning group: 120 census blocks in the west of Louisville with both high poverty rates and high car-free households. Correlated with racial minorities, low education, unemployment, and PT dependence. These communities also have the fewest nearby Points of Interest — making e-scooters largely irrelevant to their transport needs.
High income pressure but high car ownership — a signal of forced car dependency. These 24 blocks represent low-income residents who must own cars because alternatives are inadequate. Car ownership here is a financial burden, not a choice. E-scooters offer little relief without first addressing the structural absence of transit.
Across 1.9 million scenario evaluations, e-scooters only meaningfully improved accessibility when replacing walking, cycling, or public transit — not motorized modes. Critically, disadvantaged groups (Q3 & Q4) gained no significant advantage compared to the rest of the population. The gap is not about scooters — it's about the built environment.
Sensitivity Analysis Results
For each transportation mode, e-scooters were evaluated as a replacement across all scenarios. The results reveal a consistent pattern: scooters compete with sustainable modes, not car travel.
Interactive Tool
Adjust the demographic thresholds to see how the population distribution across the four quarters changes — and what that means for scooter equity.
Equity Assessment
Policy Implications
The paper argues that equity in micromobility requires addressing root structural causes, not just adding more vehicles in disadvantaged zones.
Before deploying e-scooters, cities must diversify land use in disadvantaged areas. More nearby destinations (POIs) directly increase the utility of any transport mode, including scooters.
A mobility needs assessment should precede deployment. If the service doesn't match a community's transport needs or urban structure, deployment may deepen inequality rather than reduce it.
For Q3 communities, expanding public transit coverage may yield far greater equity benefits than introducing shared micromobility that requires smartphones, banking access, and nearby destinations.
Deployment plans should involve users, city authorities, and operators — with equity outcomes (not just reach) as the primary performance metric, backed by enforceable standards.
Fleet distribution requirements that ignore Points of Interest density are ineffective. Vehicle placement should be tied to destinations, not just geographic area quotas.
Cities must track equitable outcomes over time — not just at launch. Operator data should be publicly accessible and independently reviewed, with penalties for non-compliance.