When More Destinations Cost More: Microtransit Conversion Cuts Accessibility Nearly in Half

Transit agencies across the United States are increasingly replacing fixed-route buses with zone-based, on-demand microtransit, presenting the change as modernization that expands coverage and flexibility. Riders may gain access to more destinations, but they may also face longer waits and fares that rise with trip length. Conventional threshold measures capture expanded reach but ignore added travel costs, so a conversion can appear to improve accessibility even when trips become more burdensome. A new study of an implemented conversion in Seminole County, Florida, shows just how consequential this measurement choice can be: the same service change produced an accessibility ratio of 1.900 under a threshold measure but 0.457 under a gravity-based measure calibrated to the model’s cost ceiling. Every gravity specification tested indicated a loss.

The research, published in Frontiers in Future Transportation, examines the conversion of six LYNX fixed routes into the SCOUT on-demand service launched in Seminole County, a largely suburban area of 481,470 residents directly north of Orlando. The study calculates accessibility for every residential parcel before and after the change across six destination categories: employment, commuter rail stations, grocery stores, hospitals, dialysis facilities, licensed clinics, and ambulatory surgical centers. The analysis uses observed service conditions rather than assumed parameters. The on-demand service covers five contiguous zones, with an average reported pickup wait of 41 minutes and a peak wait of 53 minutes. Prices are $4, $7, $10, and $14 depending on the zone span, and are halved for senior, low-income, and disabled riders. The fixed-route service reduction began on 11 January 2026, eliminating six routes and cutting weekday stops from 510 to 146.

At the heart of the study is a conceptual framework that treats two leading accessibility measures as endpoints on a single continuum rather than as alternatives. A cumulative-opportunity measure uses a binary weighting function, counting destinations reachable within a fixed time threshold and assigning each destination equal weight. In contrast, a gravity measure weights each destination by a continuously decaying function of the cost of reaching it, so destinations that become more expensive contribute less even if they remain technically reachable. The weight assigned to each destination is determined by a decay parameter, measured in inverse minutes, which determines how rapidly destination weight declines as access cost increases. A larger decay parameter therefore penalizes more distant or more expensive destinations more strongly. Casting the decision as a single continuum is what makes the disagreement measurable rather than merely rhetorical.

The study expresses the post-conversion to pre-conversion accessibility ratio through a two-parameter form. R0 is the reach ratio, the limiting value of the ratio as the decay parameter approaches zero, capturing the relative number of reachable destinations before and after. ΔGC is the generalized-cost penalty in minutes-equivalent, representing an effective penalty in the fitted form. Imposing a ratio of one and solving yields the crossover decay parameter at which the fitted ratio crosses one. Below the crossover, the reach gain dominates and the conversion measures count as an improvement; above it, the cost penalty dominates and the same conversion measures register as a loss. Because the crossover depends only on the fitted reach ratio and the effective cost penalty, it can be estimated from model output without selecting a single decay value in advance as the basis for judging the conversion.

The results reveal a striking pattern. The conversion expanded the number of reachable destinations but imposed a large and nearly uniform generalized-cost penalty. The reach ratio varies by more than a factor of two across destination categories, from 1.010 for commuter rail stations to 2.198 for ambulatory surgical centers, while the generalized-cost penalty shows little movement, spanning only 2.6 minutes from 46.7 to 49.3 minutes. The service expands the reachable set by adding destinations at a fixed generalized cost of approximately 47 minutes per trip; whether this constitutes a gain depends entirely on the extent to which that cost is weighted. For grocery access, the accessibility ratio moves from 1.900 at the near-zero limit to 0.734 at the shallow gravity value, 0.457 at the primary value, and 0.178 at the steepest, a swing that crosses one. All destination categories have a crossover below the shallowest gravity value tested, so they register losses under all three gravity specifications while all except commuter rail register gains at the cumulative-opportunity limit.

The distributional analysis tests whether accessibility losses were concentrated among transit-dependent communities and protected-class block groups. Greater transit dependency was associated with larger losses in commuter rail and dialysis access, but not in the other four destination categories. Across 243 block groups, a negative gradient appeared in five of six categories; however, only two, commuter rail and dialysis, remained significant under the original Bonferroni-adjusted threshold. After accounting for spatial dependence using Dutilleul’s modified t-test, only the commuter rail gradient cleared the multiple-testing threshold. Protected-class block groups experienced significantly larger losses in commuter rail and dialysis accessibility than non-protected block groups, and this finding proved robust across every cutoff tested. The two separating categories are also the two that exhibit a dependency gradient, which explains why the needs-based and identity-based screens align in magnitude. Their agreement on the destinations that matter, despite disagreement over the places that matter, is the strongest equity result reported in the paper.

The capability-based diagnostic reveals a different pattern altogether. Applying an older-adult walking speed of 0.9 meters per second, compared with the standard 1.4 meters per second, to the pre-conversion network reduces accessibility for nearly all block groups. The median block group retains 63.2 percent of standard-adult hospital accessibility and 60.4 percent of clinic accessibility; 222 of 223 block groups are worse off. At the slower walking speed, 27.0 percent of the county’s 159,837 parcels fall outside the walking range of any boarding point, compared with 3.6 percent at the standard speed. Because catchment area increases with the square of the radius, the slower speed reduces the effective catchment around each boarding point by 59 percent. This effect was larger than the service conversion itself on these two screens, and because it was concentrated in the least transit-dependent areas, it reversed the direction of the other equity findings.

Five contiguous block groups either lacked access throughout the study period or lost nearly all access after the conversion, forming a statistically significant cluster identified by a local Moran’s I test. Three block groups lacked any reachable destination across all six categories in both years. Two additional block groups had access to five of the six categories before the conversion and none afterward. Together, they include 8,776 residents; 21.0 percent are aged 65 or over, compared with 16.6 percent countywide. None of the five block groups is classified as protected-class, and only one lies in the most transit-dependent. A compliance review using the protected-class criterion alone would flag none of these five block groups, whereas a dependency index would flag one. This composition provides the empirical basis for the divergence result: the three equity screens converge on the importance of certain accessibility losses but identify different populations and places as structurally vulnerable.

The study also calculates the break-even condition at which the conversion would have left accessibility unchanged. Setting the accessibility ratio to one and solving for the wait that would achieve it shows that reaching break-even for most destinations would require average waits to fall to 13.9 to 20.6 minutes, a 50 to 66 percent reduction. Current waits are 1.99 to 2.95 times higher than all computable thresholds. For commuter rail, the 1.010 reach ratio leaves too little benefit to offset costs, so no wait reduction achieves parity; only a change in the zone boundary or fare structure could improve that result. A 20-minute wait brings one of six destination types to parity, and combining a 20-minute wait with a universal fare discount brings five of six to parity. This converts an equity assessment into a service standard that can be calculated from a proposed zone map and fare schedule before launch.

The broader conclusion concerns operating conditions, not the mode itself. At a 41-minute wait and a zone-based fare, the service reduces accessibility. Coverage targets alone are inadequate performance measures: coverage increased regardless of the accessibility measure, reaching 238 of 243 block groups, including 80 of the 81 most transit-dependent groups, yet gravity-based accessibility fell by about half. The findings describe one context, and the approach should be adapted to local conditions before it is applied elsewhere. The reach and cost patterns observed here should be tested locally rather than transferred without evaluation. Future research should test whether these reach and cost patterns recur elsewhere, including through a difference-in-differences design with a matched county that did not undergo a conversion, obtain trip-level dispatch records to model pickup waits as location-specific distributions rather than systemwide averages, and assess chained on-demand-to-rail trips.

Subject of Research: Accessibility and equity effects of converting fixed-route bus service to on-demand microtransit in Seminole County, Florida

Article Title: When more destinations cost more: accessibility and equity in a fixed-route-to-microtransit conversion

Article References: Noh, S. (2026). When more destinations cost more: accessibility and equity in a fixed-route-to-microtransit conversion. Frontiers in Future Transportation, 7, Article 1980908. https://doi.org/10.3389/ffutr.2026.1980908

Image Credits: AI Generated

DOI: 10.3389/ffutr.2026.1980908

Keywords: microtransit, accessibility, transport equity, generalized cost, distance decay, fixed-route bus, Seminole County, Title VI, transit dependency, walking speed, service conversion, on-demand transit