The National Transit Database collects, audits and distributes a trove of data on transit agencies around the country, going back many years. Preliminary ridership numbers are sent monthly to the NTD and official yearly numbers are released about 14 months after the end of the fiscal year, after additional corrections are completed.
If you took a look at the MBTA's numbers by mode for the last few years, you would have noticed some fairly large changes in ridership from previous years. In particular, ridership on the T's light rail (the Green Line and Mattapan Line) showed a 19% drop from FY2014 to FY2015, while ridership on MBTA buses shows an 11.3% increase. While the winter of 2015 wreaked havoc with the MBTA's services and impacted ridership, the bus and light rail numbers were impacted by other factors and some data errors.
Data
Note: Since 2014, the NTD has divided buses into three categories: “Trolleybus,” BRT and Motorbus. The MBTA reports its trackless trolleys (Routes 71, 73 and 77) as Trolleybus and the Silver Line as BRT, but for the purposes of this post, all types of buses (including privately-operated routes) are combined into one category.
*Data being revised
Details
The above table shows the MBTA's unlinked passenger trips by mode as reported to the NTD. OPMI and the MBTA's Service Planning department have analyzed the reasons for the changes and explain them below:
Commuter Rail
Our NTD numbers show a drop of 7.2% in trips on commuter rail from FY 14 to FY15. We believe this is duealmost entirely to the service interruptions from the Winter of 2015. Commuter Rail was canceled on multiple days and ran a reduced schedule on many other days and ridership was severely impacted.
Heavy Rail
Heavy Rail ridership shows a drop of 2.0% from FY 14 to 15. Similarly, this was likely mostly due to reduced service and ridership from winter impacts. The MBTA has focused its State of Good Repair efforts on winter resiliency over the past two years in order to maintain service levels.
Light Rail
Much of the drop in light rail ridership is explained by the temporary closing of Government Center Station, which was examined in this post. Government Center is not only a major entry and exit point but also a major transfer point in the T system, and closing it likely reduced unlinked trips by making people walk further when they previously would have transferred at the station. In our examination of AFC taps, we saw only a very slight drop in Green Line station entries, while our unlinked trips numbers dropped about 20%. Apart from some winter impacts, we don't believe there has been a sustained drop in riders on the Green Line (as anyone who takes it can attest).
Bus
The reasons for the increase in bus trips are more complicated. We have seen a steady increase in bus use as more people are moving to the area. While the subway-proximate areas of the region were already largely built out and populated, more people are living and working in places close to Downtown Boston but primarily accessed by bus, such as South Boston, the Seaport District or inner-ring “streetcar suburbs” like Somerville and Malden (Note: We're seeing an increase in subway ridership too over time).
There are also some issues in the way data was collected and processed on buses. Previous to 2014, the MBTA used farebox (AFC) interactions to count passengers, with an additional factor to account for people who enter the bus without interacting with a farebox. Starting in 2014, the MBTA began to use automated passenger counter (APC) data for its annual NTD reporting. An APC is an array of sensors at each door which can count people passing between them. These sensors are theoretically more accurate than relying on AFC data, since some passengers board buses without interacting with the farebox, for example people flashing a pass, but there are caveats:
- APCs are not on all buses, so a sample must be taken and scaled to the total amount of service run. All of the new buses the MBTA has purchased and put into service include APCs, and so our sample and accuracy when counting passengers via APC will improve as we replace older buses.
- APCs can undercount when a large number of people board or alight the bus at the same time. The software accounts for this in most cases, but on some routes has difficulty doing so. We didn't identify this issue until the FY2015 numbers had already been submitted.
- There were multiple software errors in scaling the samples to the entirety of the service provided. Things like late night service and holidays were difficult to account for, and some specific routes had errors. The sampling system also was not correctly accounting for dropped trips. We corrected some of these errors for the FY15 numbers, but not all, and the errors we corrected for caused the number to scale high. We believe we have fixed all of the errors going forward.
Due to the errors above, the NTD ridership numbers for all MBTA buses increased by 14.2 million unlinked trips, or 10.6% from FY 14 to FY 15. Service Planning has done a lot of work to correct the FY 15 numbers, and we are revising the total ridership for buses to be about 6.1 million trips lower than previously reported, or a total of 127.5 million unlinked trips**. This is an increase over FY 14 of 8 million trips or 6.7%. We will work with NTD to revise the FY 15 numbers.
It is also worth pointing out that the MBTA reports monthly ridership estimates to NTD, so the NTD website has monthly ridership by mode for FY 16. These numbers are preliminary and will be revised with annual adjustments. The monthly bus ridership numbers are from the AFC system and the annual adjustments take into account the APC data.
Conclusion
The MBTA is concerned with providing quality service in order to increase ridership. We watch ridership closely and publish monthly ridership data on www.mbtabackontrack.com. However, with our existing technology accurately measuring ridership is difficult. Even the industry standard measure of unlinked trips does not reflect the actual number of passenger trips. If a passenger transfers it counts as two unlinked trips, so increased ridership can indicate increased trips requiring a transfer. To address these issues, we are considering how tools like ODX and our proposed new fare collection system can give a better picture of the actual volume of people using the MBTA system.
** NTD ridership numbers for all systems are reported down to the single UPT, which implies a false degree of precision. There is no transit system or indeed any survey which can count millions of people so precisely (especially without a high amount of resources).
