Subway riders, what matters more: whether your train picks you up on time, or whether you arrive at your destination on time?
Most people would probably pick the latter.
For years, though, the Massachusetts Bay Transportation Authority has leaned on the former to gauge its on-time performance. But measuring whether the next train arrives on time has many shortcomings, so the T is exploring new ways to more accurately reflect whether riders get to their destinations on time.
“These numbers at times do not necessarily reflect what our customers experience or what our customers are seeing,” said the MBTA’s deputy general manager, Jeff Gonneville.
The subway system doesn’t run like the commuter rail and bus systems, which have scheduled arrival times at specific stops. Instead, subways run on timed intervals between trains, known as headway, so the T is more concerned with whether passengers have to wait longer than expected for the next ride. If the headway interval is every five minutes, and passengers wait five minutes or less, the T counts that as a win.
But that doesn’t, for example, reflect whether passengers were left on the platform because they couldn’t squeeze onto an overcrowded train — a common occurrence at rush hour on the Orange and Red lines.
Nor does it directly account for delays such as sitting in a tunnel behind a disabled train.
It also doesn’t capture how bad the wait can get between intervals — whether riders wait just a couple of extra minutes, or much longer than expected.
The on-time metric has also been ridiculed and criticized for painting an overly optimistic picture of riders’ experience. The T says that trains on the Red, Orange, and Blue lines typically run on time about 90 percent of the time. That sounds pretty good, but it also means that a typical rider is having at least one late trip a week.
The Green Line, by contrast, is closer to 80 percent on time under the current measurement, in part because trolley operators on some above-ground portions must follow street-traffic signals or mix with auto traffic.
Most important, it seems unlikely that most Red Line riders, for example, would rate their commute an A-minus.
“I’d say more like a 75,” said Patricia Maguire, who travels each day from Dorchester to downtown Boston.
Some days, Maguire said, she gets a text message that the Red Line is delayed, so she holds off on leaving for the station until normal service resumes. The train she eventually gets on may have arrived within the scheduled headway, which the T would count as a win. But it’s not accounting for the time Maguire lost by waiting out the service delay.
“If I count my 10 trips a week for the purpose of commuting, I’d say four of them are impacted in some form or fashion,” Maguire said. “How it impacts me directly is the only guideline I have. . . . How often do I get gummed up by the system?”
Gonneville said the MBTA wants a new measurement that focuses less on the wait on a platform and more on the overall length of the journey.
“We want to move forward with a new metric that focuses on run time, because it gets at some of the inconsistencies we may be seeing now,” he said.
Other cities use advanced metrics that are based on the length of each passenger’s entire journey — from when they enter a station until they exit.
London’s transit system, for example, gauges “excess journey time” — how much longer the trip takes compared to its expected length, based both on time spent on the platform and on the train.
But the London system has a big advantage over the T and many other transit systems: Riders must tap their passes as they enter a station and when they leave one, allowing transit officials to capture their travel times more precisely.
New York City began using a similar metric in recent years, though it does not require riders to tap out. The city’s transit authority said it can analyze travel patterns based on where riders enter stations to make an “educated guess” about where they leave the system, and pairs that with the time spent waiting on platforms. Using this data, transit officials measure how much longer riders spend on platforms or trains than they should, and the percentage of trips that end within five minutes of what riders expected. In February, for example, that number was 82 percent.
It’s unclear what, exactly, the T will measure. Gonneville said its data team is working to develop a formula, but it may be as simple as comparing the expected length of a train trip along an entire subway line to how long one actually takes.
The T’s measurements for most buses and all commuter trains are based on whether they arrive according to schedule, although on some higher-frequency buses the T uses expected headways, as with the subways.
The commuter rail is usually timely; trains arrive at stops within five minutes of scheduled service about 90 percent of the time, though some lines perform better than others. Buses, meanwhile, struggle with street traffic and other issues and have an on-time performance of just 70 percent.
The T is considering changing subway system measurements only.
Nigel Wilson, a professor emeritus at the Transit Lab at the Massachusetts Institute of Technology, said that makes sense, because subways come more frequently than buses or commuter trains. That means riders are less likely to consult a schedule, since they expect a train to arrive every few minutes — and get them to where they’re going quickly.
“It’s high-frequency, and so on-time performance is not the right measure,” Wilson said.
But for lower-frequency routes, on-time performance is crucial, Wilson said. It’s especially important that schedules are accurate when there’s a long lag between trips, because survey data have historically shown that riders are far more vexed by long waits at stations than while sitting on vehicles.