A new report questions the MTA's current on-time performance metrics and suggests a passenger-based approach instead.
One of the biggest complaints New Yorkers have with the MTA and an oft-heard excuse early in the morning is one of the more mercurial aspects of transit operations. “Sorry I wasn’t on time,” we’ll often hear. “The B train was late this morning.” What exactly does that mean?
When I head from Brooklyn to Manhattan every day, I use the B train at 7th Ave., but I don’t really leave at the same time any day of the week. When I had two early classes last semester, I could time my trip to catch a B train at approximately the same time every morning, and when there was a problem, the train wouldn’t be there. To me, that’s the traditional definition of on time.
But there are other ways to measure on-time performance. One that the MTA uses internally is a wait assessment. Intuitively, this one makes some sense. If the B train is supposed to run every eight minutes, it matters less when the B trains arrives as it does when the next B train after that arrives. As long as the interval is eight minutes — or in the MTA’s case, eight minutes plus 25 percent — the trains are still on time, and people won’t be left with empty tracks when they expect a train.
Finally, we can judge a train based upon when it’s supposed to get to the end of the run. This is another metric Transit uses to judge on-time performance. If a train is at its terminal within five minutes of the scheduled time, it is still considered on time. Anything worse means delays or one sort of another along the route. But are these any good?
In a report released yesterday, the Permanent Citizens Advisory Committee to the MTA examined the authority’s wait assessment metrics and found them rigorous but lacking. The committee praised the MTA for being among the most transparent transit operators in the country in providing wait assessments but determined that the rankings did not help passengers evaluate on-time performance. The wait analyses, in other words, are geared toward internal evaluation and not improvements for the customers.
“A schedule is a promise,” PCAC Chair Ira Greenberg said. “A late train or bus breaks that promise and the impact is lost time for the riders. People depend on the MTA’s service for their livelihood. We want the MTA to think of the rider first, before trains and buses, and we look forward to working with them to achieve this.”
The report — which I’ve embedded below — presents an extensive evaluation of the MTA’s three rail divisions and a comparison with other U.S.-based transit providers. Passengers, it finds, are left in the dark, and ultimately, PCAC urges the MTA to develop a passenger on-time performance metric that can identify the number of passengers delayed and which ones are delayed most frequently.
Most vital is the report’s recommendation that the MTA start promoting capital expenditures as a way to improve on-time performance. PCAC highlights the countdown clocks as an example of a technology that can lead to more satisfied customers even if they don’t improve the wait times. “While countdown clocks do not create performance measurements,” the report says, “they serve to moderate the rider’s expectation of performance with real time knowledge.”
It takes a stronger stance on capital improvements viewed as disruptive by passengers. “The average rider doesn’t necessarily understand what new interlockings, switches and signals are, let alone appreciate how their improvement will enhance their commute. Historically, the use of performance metrics at the MTA began as an effort to secure needed capital funds. That linkage, as a tool to promote capital programs to the riding public and elected officials, has weakened over the years,” it says. “Specific information on how an item in the Capital Program will reduce the number of delayed and canceled trains, increase track speeds, and improve the ability to recover from a major service disruption is relevant to the riders.”
Now, that just makes sense. If the MTA can convince anyone that their work will make trains run on time, shouldn’t that be a prime selling point for a project? I would think so.
Keeping people moving and making sure they get somewhere on time should be a paramount goal for any transit provider. While wait measurements and delay assessments are more important for commuter rail riders who see transit only every 30 minutes at peak times instead of every five, subway riders like to know they’ll get to their jobs and appointments on time without egregiously long waits. By presenting that information to the public in an easy-to-understand fashion that directly addresses the wait, the MTA could improve the way customers impatiently wait for trains. Time might be on my side after all.
After the jump, read the PCAC’s full report.
26 comments
Like everything else related to mass transit or public services, this has already been done abroad, better. Switzerland has recently moved to a passenger-centric system, measuring the percentage of passengers whose trains are on time, including missed connections. This is for the national railroad so it’s based on schedule maintenance, but it could be done just as well for headway maintenance on urban rail.
You are correct in stating that the interval between trains and buses is more important to the passenger than how close they are to their schedule. If all the buses on a line were late but the interval between them were constant, it wouldn’t matter much to the passenger, except that they would be more crowded than they should be and traveling a little slower. This is still preferable to most of the buses traveling in pairs with the waits for most everyone doubled. Perhaps more attention needs to be paid to evening out intervals even if it means more buses are “late” and can’t finish all their scheduled trips, but this strategy would not work if all the vehicles are overcrowded.
On-Time Performance means different things to different people. For passengers, the time between subway trains is often more significant than whether or not they arrive at their scheduled time. Commuter-rail passengers want to know which lines are delayed and by how much (i.e. the LIRR Port Jefferson Branch is running with 10 minute delays). We really don’t care how many people are affected, we just want to know how we will be affected, and by how much. Countdown clocks (or the LIRR/MNR equivalent, which lists scheduled trains and whether they are On-Time or Late (and by how much) generally accomplish this.
In evaluating scheduling and operational effectiveness overall, I suppose the number-of-passengers metric may be more useful, but that only serves to effectively give a letter-grade as to how are they doing. It always irks me to hear when “ConEd is reporting 15,000 people without power”, that tells me nothing. I’d rather know WHERE the outages are. This is the same.
Con Ed figures are actually more misleading because they don’t report people without power, they report “customers”, and unlike the MTA, to Con Ed a “customer” is a “household,” so 15,000 Con Ed customers could actually be 45,000 real people.
The point of passenger-centric metrics isn’t to say “X people are without good service.” That would be stupid. The point is to say “We have 89% OTP,” with the understanding that trains you’re more likely to ride are weighted more. The same would apply on each line.
Countdown clocks don’t accomplish this as well, because they don’t tell you whether your train will be delayed after you get on.
Nobody I know bases their subway travel on the schedule. I’d like to see a subway and bus passenger survey which shows what people actually care about. Good guess is the less frequent the service, the more important the schedule.
The point at which a schedule is going to be consulted is up for debate. But most in the industry agree it will be at headways in excess of at least 12 minutes. Thus very few NYC subway lines, except on the midnights when the standard headway is 20 min, qualify.
This is exactly the point. OTP is a pretty useless in assessing the quality of a rapid transit rail service (as opposed to a mainline or commuter type rail service) with midday headways of around 4 minutes and peak headways of 2 min. Thru-put, even spacing of trains and adequacy of service (are there enough trains to comfortably accommodate all the passengers – except of course during the peaks) are far more important. You could literally have 0% OTP, yet if you met all these other criteria you’d be doing just fine. Unfortunately NYCT has lost sight of this fact with management becoming unhealthily obsessed with just one thing: OTP. And so service has been cut (on the premise that too many trains cause congestion and, thus, lateness) and huge amounts of bogus run time have been added to almost every line. Bottom line here: fewer, slower trains. But that’s the new thinking at NYCT: less is actually more and slower is actually faster. So if you have to wait longer for your train than had once been the case, and now have to stand where as before you could get a seat, don’t complain. Because your train is now probably ON TIME!
As far as connections on NYCTA: they are only generally scheduled where headways are at 20 minutes, i.e late night, or at certain special remote locations, such as at Broad Channel.
Your comments on OTP are absolutely on-target. On a frequent rapid transit service, OTP is virtually meaningless to the rider. (On commuter railroads, with less frequent service, it are far more relevant, since most riders have specific trains that they aim to catch. And the terminal aspect of OTP – that it’s only measured at the last stop – is also relevant to New York’s commuter rail riders, since most of them on inbound trains are going to the last stop.)
But I think you’re being unfair to current NYCT management. Increasing OTP was the holy grail for the Roberts administration – it was under Roberts, in 2008, that OTP was further split into Absolute OTP and Controllable OTP. The Prendergast administration is taking a markedly different approach, retreating to a single OTP measure, improving the approach to wait assessment, and explicitly focusing on the “customer experience.”
This article is, in my opinion, sloppily written. On-time performance is a specific performance measure, but this article uses it as a generic classification of any performance measure that relates to train arrival times. Ben’s experience at 7th Ave. has no bearing on the actual OTP of the B train; only its arrival time at Bedford Park does, and even there, if the same train is 5 minutes early one day, precisely on schedule the next day, and 4 minutes late the third day, it scores perfectly on OTP in all three cases. (Conversely, if every train arrives at the terminal 6 minutes late every single day, OTP is 0, even though service, as far as the rider can tell, is perfectly regular and predictable.) Wait assessment is no more a measure of on-time performance than an apple is a type of orange.
I was with NYCTA for almost 37 years., including under Roberts, who was a wholly clueless bonehead. He degraded the service by, as I noted above, eliminating trains and adding huge amounts of run time padding in a wholly hopeless pursuit of chimerical OTP numbers. He and his flunkies did untold damage to the quality of NY subway service.
Agreed completely. Not only was he clueless, but he insisted on making big changes. Prendergast seems to be trying to clean up the Roberts mess, based on the little I’ve seen (Prendergast isn’t the media hound Roberts was, so there isn’t much).
If you don’t mind my asking, what was your role within NYCT(A)? Ignore my request if you’d rather not answer.
No problem. I proud of my career with NYCTA.
Started in 1972 as a bus driver in Crosstown Dept.
In 1973 became a “B” Div motorman.
In 1978 moved to the IRT as a Motorman
In 1982 became a Motor-Instructor (now Train Service supervisor) on the IRT
In 1984 became a line supt, #3 line.
In 1987 became a schedule manager, B1 (BMT), then IRT squads in OP
In 1994 became the senior schedule manager of the IRT squad, OP
Retired in 2009.
The NYCMate Android app has, in addition to a subway map and neighborhood maps for each stop, the full train schedules as pulled from the MTA website. It amazes me how accurate they are, on the whole – sure, sometimes they’re a few minutes off and during the weekend it’s anyone’s guess, but during the week, my 10:07 F train at Bergen St is usually in fact there at 10:07.
If you don’t mind my asking, have you tried my app?
Or perhaps it’s the 10:03 F train running 4 minutes late. Or the 10:11 F train running 4 minutes early. It doesn’t matter to you, but it does matter to the OTP calculations.
The F to Manhattan is scheduled for Bergen Street at 10:07 am (10:07:30 to be exact).
I know. But you don’t know that the train that you saw at 10:07 was the train scheduled to get there at 10:07. It might have been the previous train running late, or it might have been the next train running early.
Well, sure. If the train arriving at 10:07 was late (by 8 minutes according to schedule), to the passenger boarding at Bergen, it doesn’t matter, yes (maybe aside from overcrowding and comfort issues). If the same train was early at 10:07, at this point it would be at least 8 minutes ahead of schedule, dispatchers need to control that train. In this case, if the train is held for schedule, passengers will see this as a delay.
I agree with all the comments above about OTP. The concern that schedules are padded also makes OTP a less useful metric for passengers. For example, in a southbound 2 train at 1:00 pm, there are built-in delays at Nevins Street (2 min), Atlantic Avenue (2 min), Franklin Avenue (4 min), and Church Avenue (1.5 min). If this train is on schedule, most passengers will not see this as good service. Of course, this train could also have operated slower so as not to appear having long waits.
While doing a video for my app on the subway (not as fun as I thought), I noticed the L train I was on was about 2 minutes ahead of schedule. But thanks to the countdown clocks, the interval for the following train appeared to indicate that that train was on schedule, and passengers aren’t likely to complain about that. OTOH, the other day I waited for the Q54 in Forest Hills for more than 15 minutes. I ended up taking the Q23 instead so I won’t freeze my butt off. I knew that the Q54 had irregular service earlier, as in Williamsburg I took the second of 2 buses bunched together. In Forest Hills, I wouldn’t have waited longer for the Q54, especially as two Q23s had passed by. (Although with BusTime, I would reconsider that decision.)
Many of the “holds” in the NYCT timetables are there to achieve clearances at interlockings. This has been done here for the last 100 years or so.
But a series of holds approaching the end of the line – especially with the final one past the last non-terminal interlocking – suggests padding.
At Nevins and Atlantic at that time? What trains are supposed to merge there?
But for comparison, at around 1:00 pm, the northbound 2 is scheduled to be held at Nereid Avenue for 5.5 minutes. In both examples, the 2 is not held anywhere else. (Other northbound trains: 7.5 min at Burnside Avenue (4); 5 min at 145th Street (3); 7.5 min at Baychester Avenue (5). The 3 is also held at Utica and Nevins.
I was the senior manager of the IRT schedule squad for many years. The holds you see are for various and sundry reasons. Yes holds are done at Nev or AA or Fkln for clearances at Nost Jct. Many holds are done at 238 WPRD to achieve pocket clearance at 241. Holds were also added during the Roberts years simply as run time padding. He was morbidly obsessed with OTP.
I have experienced holds near terminals, a lot of times between stations. They are to be expected, but the passenger doesn’t like to wait. My frustration is with the wait times before terminals, they feel like unscheduled delays even though they are, but I realize they are necessary for the logistical reasons you provided. Good to know why we should be patient.
Either that, or you’ve run into the padding that Roberts added to the schedules. Your train can’t get into the terminal if all of the terminal tracks are still occupied, so it has to be held.
But don’t be annoyed. Your train is officially on time, and that’s all that really matters (according to Roberts)!
I am sure this post has touched all the internet viewers,
its really really fastidious piece of writing on building up new blog.