MTA Will Start Using Common-Sense Metrics to Show Subway Riders How Their Trains Perform

Transit advocates had been calling on the MTA to adopt metrics like "average wait time" to more clearly communicate system performance to riders.

The new MTA dashboard will show the average amount of time that delays cost riders on platforms (above) and trains.
The new MTA dashboard will show the average amount of time that delays cost riders on platforms (above) and trains.

Starting Wednesday, transit riders will be able to see for themselves how their subway lines are performing, when the MTA launches an online dashboard showing how much time riders lose due to trains that are off-schedule.

Agency officials previewed the dashboard today for the MTA board’s transit and buses committee. The dashboard will display two new metrics — “additional platform time” and “additional train time” — which calculate service quality in rider-centric terms: the extra minutes that riders typically spend on platforms or trains compared to a scenario in which service operates perfectly according to schedule.

Transit advocates have been calling on the MTA to adopt this type of metric, which is considered best practice in the transit industry, to more clearly communicate system performance to riders.

“The measures we have historically used — wait assessment, and on-time performance, as well as [major incidents and train capacity] — are measurements of trains, not people,” said MTA Chief of Operations Planning Peter Cafiero. “These are measures showing impacts [on] actual customers.”

Riders will be able to look up the new metrics for each line, and for the system as a whole, going back to July 2016. Providing these metrics was one of the top requests from transit advocates when newly-appointed MTA Chair Joe Lhota announced plans for the dashboard in July.

Advocates applauded today’s announcement while noting a few omissions they want the MTA to address going forward, including historical wait time trends going back farther than last July and wait time metrics for buses. (Which can be done: TransitCenter worked with grad students at NYU to produce a beta version of a wait time dashboard for buses, using publicly available BusTime data.) Cafiero said the agency will continue to refine and improve the dashboard after its public release on Wednesday.

The dashboard includes a few other new indicators, including a measurement of how long elevators and escalators are out of service.


Also included: “major incidents” — sources of delay broken down by subway line; “capacity provided” — the proportion of scheduled trains that actually arrive within a given amount of time; and results from subway car and station cleanliness surveys.

Several other metrics are grouped under the heading of “legacy indicators” that the MTA has been measuring for decades. These include the service metrics that make more sense to MTA managers than riders, as well as key maintenance metrics like the average distance between mechanical failures on trains.

One missing piece appears to be a progress tracker for capital projects and major maintenance work. That information is supposed to be provided on the MTA’s existing capital program dashboard, but the agency stopped listing the original project timelines, rendering it next to useless. As Reinvent Albany’s John Kaehny told Streetsblog in July, the only ironclad way to ensure the continued publication of MTA performance metrics is to codify it in law.

  • Joe R.

    Can we take this as a major victory? Congratulations to all those who pushed the MTA to be more transparent. The first step to fixing a problem is documenting it. These new metrics go a long way towards that.

  • Hux

    Glad they are putting WA on the last page of this site… it has so many problems.

    Bunching doesn’t cause WA to decline, because passengers don’t have to wait long for a train. However, bunching implies that there are also gaps in service, and that, therefore, some trains are more crowded, even if ridership is declining. If more emphasis were to be placed on terminal OTP, headways would improve, because the schedule is designed for proper headways, merges, loading guidelines, crew constraints, and so on and so forth. But, leadership continues to believe a myth that WA is more important than OTP. They think that passengers care more about waiting for a train than they care about waiting on a train, even if the train is delayed. But the vast majority of passengers are on trains, not platforms.

    WA also looks better during late nights because it is percentage-based, so a train that exceeds the scheduled headway by 5% at night is going to be a lot later (and rarer) than a train that is 5% late during rush hour. Would anyone really argue that late night service is better and people aren’t waiting a long time? No.

    MTA argues that OTP is a problematic indicator because it is pass/fail, and because the subway is not a commuter railroad, where passengers manage their expectations through a timetable. Instead, subway passengers care about trains arriving quickly, and about gaps.

    Pass/fail is problematic, but it could be easily remedied by switching to journey time. No one is holding us to these indicators. Also, subway passengers don’t manage expectations through a timetable because the system has poor OTP. But, even if passengers don’t check the subway schedule, it is still important to have good OTP, because it means that there are fewer delays. Also, while most subway passengers don’t ride terminal to terminal, most commuter railroad passengers don’t ride terminal to terminal either; delays are felt along the line, and slower service means a slower city, and lower quality of life.

    Managing to WA harms OTP, but it also harms WA. For instance, if there is an incident and a train is delayed, oftentimes, trains ahead of that incident will be held. Yes, this impacts OTP, but it also creates a gap in service. For reasons aforementioned, WA is less sensitive to change (and it will be made even less sensitive), but both of these indicators tend to improve or decline together. After all, headway adherence (WA) is essentially a form of OTP.

    Holding trains for evenness harms WA because WA is not the same thing as evenness. WA is the percent of intervals between trains that are not more than the scheduled interval +25%, and it measures gaps in service along multiple time points of each line.

    Pretend along a line, all trains are evenly spaced, but then there is an incident ahead so all the trains are held 5 minutes to maintain even spacing. A train held at Station A which was supposed to arrive at Station B at 10:00 is now going to arrive at Station B at 10:05. So the headway is +5 minutes from the schedule, meaning that even though all the service is still even, it is all late. WA measures the scheduled headways, not just even headways. A train being held at a station and not traveling to the next station counts as part of the headway. It is being slowed down, not as many trains are making it through the system. MTA leadership often doesn’t understand the difference between WA and evenness, and they confuse evenness during incidents with regular service, which is not always evenly spaced (due to merges, differences in time of day, etc.)

    Evenness is not WA. WA is schedule adherence; the schedule is not necessarily even. When trains are held for evenness, it harms WA and OTP. During incidents, evenness matters. Otherwise, schedule adherence matters.

    Recently we got all of WA to be under electronic data, which makes it more accurate — this was a big deal, since leadership did not want to show prior electronic data for political reasons. We had calculated it for historical data and wanted to revise the old numbers, but they were worried people would think we were sitting on this data and not sharing it, even though we had only just figured out how to calculate it with raw track circuit data. So we ended up convincing them that we needed one set of data to not confuse everyone and that we’d actually be sitting on it and misstating fact if we didn’t release what we now had available.

  • Hux

    Wow, they have been reporting excess jtheirney time publicly since July! But it is still based on the schedule, so if a train is scheduled to arrive on a 10 minute headway and arrives in 11 minutes, it will be +1 minutes, even though a 10 minute headway isn’t ideal for the customer compared to a 2 minute headway, also with +1 minutes of platform waiting. It averages out for the whole system and takes ridership data. They have platform and train time… and remember, the average customer rides 1.5 trains so the total excess jtheirney time is arguably x1.5 what they’re reporting.

    And it’s true, terminal delays are not the best indicator of the customer experience. Most riders don’t ride from terminal to terminal, and it’s also a pass/fail metric. But this is a great indicator for their own operations of trains, and if there are no delays, then I’d think the customer experience was also fine. (Unless the schedule itself sucks, which is often the case.)

    Over all, they should be clear about their indicators. Some are better measurements of operations, and some are better for customers. I also think customers would check their schedule if it was reliable, especially at night and on the weekends. So they do care about schedules.

    Forgot… they also have planned work delays — these are the other delays which are increasing as much as the “crowding” delays, because supplement schedules often don’t account for how their flagging rules continue to slow down service. It takes a while for new ones to get implemented, for compliance to rise, so these rules should continue to slow things down in the future… especially as they get more maintenance work done and piggyback on top of projects, which is a good thing! I’d sacrifice OTP for planned work delays. They can maintain their system without shutting everything down at night.


Subway arrival data displayed on an app produced by Transit, a Montreal-based start-up, and the MTA's Subway Time app.

Will the MTA Put Better Transit Data to Work for Riders?

The Montreal-based app Transit is based on an unorthodox assumption: trip planning limited to directions from Point A to Point B doesn’t actually fit the needs of most riders, who travel the same route every day. “[The idea] we started with at Transit was to give the information of the next bus at the stop [where […]

A Better Way to Track How Well Transit Performs

When you’re riding the bus or the train, an unexpected delay is the last thing you need. If transit agencies want to know how well they’re doing and how they can improve service, they have to track how reliable their service is for riders. But not all reliability metrics are created equal, writes TransitCenter. Some agencies track their performance in a […]