Research · Transit access and measurement
DC Student Travel
How do students reach school when no dataset observes the whole journey? This research combines passenger counts, synthetic home–school geography, and scheduled routing to study access across Washington, DC.
Three evidence layers
PASS generates a synthetic student population and home-school assignments from aggregate constraints. Event-APC measures changes in aggregate vehicle-door counts around school schedules. VIGO evaluates scheduled access for the modeled home-school obligations under stated transit networks.
Modeled demand, observed bus activity, and scheduled accessibility remain separate throughout the work.
The research asks what can be learned about access and school-timed transit activity when direct observation is incomplete and student privacy remains a hard constraint.
What each source supports
- PASS supports plausible home-school demand and accessibility analysis using synthetic rows.
- Event-APC supports school-timed changes in aggregate boardings and alightings.
- VIGO routing describes scheduled network opportunity under a stated service model.
- AFC supplies full-day aggregate context; it is not an APC denominator or passenger linkage.
What the current thesis finds
Across 237 school areas and 54 primary weekdays, mean dismissal counts exceed matched references by 6,771 boardings, and arrival counts by 8,780 alightings. These are all-rider differences around school bells, not identified student trips.
PASS reconstructs home–school geography from Census microdata, enrollment, disclosure-limited attendance counts, and grouped distance summaries. Scheduled routing evaluates 90,860 eligible modeled students in each of 31 realizations from one fitted model.
The implemented bus network reduces mean daily time commitment outside school by 0.57 minutes, yet 4,309 modeled students in the main realization add at least ten minutes. A targeted service-restoration experiment examines the tradeoff between time savings and the alternatives left when a first vehicle is canceled.
Evidence limits
PASS rows are synthetic. Event-APC estimates school-timed changes in aggregate bus counts, not student riders or complete trips. Accessibility results describe modeled scheduled opportunity under the network assumptions used, not delivered-service reliability.