Home New York City Transit The trains are late but by how much?

The trains are late but by how much?

by Benjamin Kabak

New York City’s subway trains are later than ever, and as the MTA Board grapples with these internal findings, the metrics are coming under question. What does it mean for a train to be late? Should we the straphanging public be concerned? Is subway service actually getting worse?

The news, as first reported by Michael Grynbaum at City Room and available in raw form this MTA Board document, goes something like this:

The Nos. 2, 3, 4, 5, and 7 lines all recorded significant drops in on-time performance in March, the most recent month for which statistics were available, according to figures disclosed on Monday at an agency committee hearing. The numbered lines also performed worse than the lettered lines on nearly every major metric.

Nearly 90 percent of No. 3 trains were marked as “on time” in March 2010; one year later, only 71.8 percent of the line’s trains arrived on time. Compared with a year ago, the No. 2, 4 and 5 lines fell by 14 percent, and the No. 7 line, which has had significant problems because of troubles in its East River tunnel, dropped by 12.2 percent.

Over all, about 81 percent of trains on the numbered lines, including the Grand Central shuttle, were considered on time in March, a 10 percent drop from a year ago. That was far worse than both the BMT and IND lettered lines, the latter of which improved in March from a year ago. A subway train is considered on time if it reaches its terminal station within five minutes of its scheduled arrival.

Some of the board members were not pleased to hear this news. The recently-appointed Charles Moerdler, who has become a vocal member of the MTA oversight body not afraid to ask tough questions, pondered the root causes. “The IRT service continues to be pretty bad,” he said. “What long-term plans do we have to get that service up to snuff?”

Meanwhile, the story has been picked up by amNew York and New York 1, among others. Before we delve into the panic, let’s step back a bit. First, what does it mean to be late? Most straphangers just roll their eyes when told the subways are on any sort of schedule, and the MTA’s own metrics define a train as late if it arrives at the terminal after five minutes of its scheduled time.

In a vacuum, that’s not the most useful measure of anything. Wait assessments tell a better tale, and Transit head Thomas Prendergast recognized as much. “We’re still bound by the principle that evenness of service is by far the most important thing rather than just late, although we’d like to do both,” he said. “But evenness of service is more important because that way you’re having less impact on customers.”

Meanwhile, despite the hand-wringing, these numbers have improved between February and March. Far more 2, 4, 5 and 6 trains were on time in March than in February, and the 6 and 7 didn’t show statistically significant differences in on-time performance. Maybe then the story isn’t that trains are later; after all, a recent change in the way the MTA calculated “on-time performance” could account for the year-to-year difference. Maybe instead the story is off a small but incremental month-to-month improvement.

We’re still left though with the question of the quality of service, and Moerdler put it best. “The public really doesn’t give a damn whether the stats are right, or the stats are wrong,” he said. “If the service ain’t good, it ain’t good.” How ain’t good the service is remains a question without a solid answer.

You may also like


Robert May 24, 2011 - 1:55 am

You didn’t mention something important that Grynbaum points out — that this year’s statistics for the IRT are based on the new system used by the countdown clocks, which might be showing more accurate results than in the past, which could explain the difference.

R. Graham May 24, 2011 - 2:28 am

I’m not even going to delve into the countdown clocks. I’m going to just go on feel alone. As a rider of the Lexington Avenue line. I can honestly say this year is worst than last year. Plain and simple. Now the key is what’s causing service to be so poor or what factors are working together to disrupt service as a whole?

petey May 24, 2011 - 9:29 am

i ride the lex 86th street to atlantic avenue weekly on friday evenings – not nights, evenings – and there is ALWAYS delay

Phil May 24, 2011 - 2:49 am

Isn’t the point of rapid transit that it’s frequent enough not to be on a schedule but rather run appropriate approximate headways depending on the time of day or night? I know that there are official schedules, but no one actually looks at them and one mishap at any given time will invalidate any form of scheduling. I thought that the idea was that you just got onto the platform and waited for the next train, not adjust your time to catch the 6.38 from Prince Street and then have to wait until the 6.48 comes since it’s 11pm and that’s what the R schedule says (the times are made up but still). To me, at least, it would make more sense to have defined headways per line and compare actual performance to idealised.

Alon Levy May 24, 2011 - 4:55 am

The MTA has a separate metric for that, but it’s very coarse. The standard is headway plus 2 minutes peak and headway plus 4 minutes off-peak; it’s calculated as a percentage. Better would be to say what the standard deviation of the actual wait is, or what the perceived wait time is (if you know the headway, you can derive each of those two from the other).

Scott E May 24, 2011 - 7:46 am

I wonder how closely the train crew schedules align with the train schedules. Does a train arriving 10 minutes late translate into paying overtime? This could (1) turn OTP into a real high-profile hot-button issue that needs to be solved with more urgency, or (2) create conflicting goals for a train crew. Obviously, crews don’t leave the train every time they reach a terminal, but the delays on each run add up.

R. Graham May 24, 2011 - 9:34 am

I doubt this personally. I have friends who started out as conductors on the A division as rookies and picked into the B division. None of them wanted to come back to the numbers. As rookie Train Operators one of them got unlucky and is working the A division while the rest of them remains in the B division.

Paying More to Wait Longer for the N.Y. Subway - NYTimes.com May 24, 2011 - 9:05 am

[…] subway trains are late more often than lettered lines, but perhaps that is because of inaccurate metrics as much as problems with the system. [City Room, 2nd Avenue […]

BrooklynBus May 24, 2011 - 10:01 am

There is a problem with using eveness of service as a guide rather than headways. While you may not be waiting longer than you should for a train, if all the trains are late, it could mean that all trains are moving at half the speed they should be moving, doubling the time it takes to make your trip. Trip time including waiting is a more accurate way of measuring customer service than using waiting times alone.

Alon Levy May 24, 2011 - 12:16 pm

Fair enough. But, if the trains are all running slow, then for the same service cost the frequency is much lower, and this will be detected in the wait time metrics.

But, it’s important to include metrics about in-train wait time, when the train is held at a station or (especially) in the tunnels. I suspect that customers perceive such waits the same as or worse (longer) than waits on the platform, which they perceive as worse than time spent in motion.

Streetsblog New York City » Today’s Headlines May 24, 2011 - 10:20 am

[…] On-Time Performance Drops Markedly on Numbered Subway Lines (City Room, 2nd Ave Sagas) […]

capt subway May 24, 2011 - 1:06 pm

As I’d written before here in SAS, OTP is a pretty useless way to rate rapid transit service. Adequacy of service (are there enough trains to comfortably carry all the passengers), frequency of service and evenness of service are far more important.

BTW the trains are late, more often than not, due to the new flagging rules put into effect during the Harold Roberts regime, the man who did more to damage subway service delivery than any one person – ever. Subway Schedules Dept has alleviated the problem somewhat by adding huge amount of running time to every single line. But that’s like solving a problem with being overweight by just buying larger size clothing.

Add to all that just plain poor employee training and poor train operation over the road.

pete May 24, 2011 - 3:32 pm

How to fix on time performance? Add more time to the schedule. How many hours a day is the subway faster than driving anymore? The IND was built for 55 mph through out. It now stuggles against a bicycle messenger.

capt subway May 24, 2011 - 3:53 pm

As I noted above huge amounts of run time have already been added. This becomes an endless counter-productive process which is completely open ended. Add more and they just demand more. Plus it becomes costly because more run time ultimately means more trains and crews are needed to provide the same level or service.

The TA needs to tighten up on its operations. Rolling back Roberts’ inane new flagging rules would be a start. Better employee training is also big part of the picture. It quite obvious some people out there operating are afraid of the trains, afraid of the signals, afraid of their own shadows probably.


Leave a Comment