We understand how great complaining is, especially complaining about “your” train and how long you have to wait for it. The G train kills love! The R causes ebola! The 2 and 3 killed both your grandparents! (First one, then the other, obviously. Not both at the same time. That would be too awful. Plus, the trains run on the same track, so we’re not sure how that would work. This got dark quickly, didn’t it?)
Far be it from us to deprive you of that complaining high. But no matter how much you like to kvetch that your train is always late, it’s not actually true. You can see that for yourself by analyzing the performance data the MTA releases every month . Or, if you’re a normal person, you can take a shortcut and look at Mike Vanger’s New York City Subway Investigation.
Vanger is a graduate student in the Analytics program at Northwestern University and a Chicago-dweller, but he chose to focus on New York for a final project in an independent web development course. He took four years of data from the MTA and made an app that analyzed, and plotted in a handy, readable format, just how often the trains show up on time each month.
The results are, in short, pretty damn good. Overall, the trains showed up on time 80.3 percent of the time in June of this year, the most recent month for which data is available. That’s an improvement over January 2009, when they showed up around 75.5 percent of the time. And in general, almost every subway line shows steady improvement over time, with the exception of the M, which has nose-dived from a high of 88 percent on time in March 2010 to 78.3 percent on time this June.
As you might expect, there’s variation between subway lines. Although you might not believe this, the G is the most reliable, with 84.3 percent showing-up-on-timeness. The 2 and 4 are both surprisingly bad, with the 2 taking honors for the lowest on-time rate ever: just 66 percent, back in December 2010.
But there’s one slight issue with the data.
According to the MTA’s own On-Time Performance Data, some train lines are doing much worse than Vanger’s data shows. For example, the 5, which shows up on time an abysmal 58.7 percent of the time, far worse than its target goal of 86 percent.
The explanation for the discrepancy is simple, explains MTA spokersperson Adam Lisberg, and in fact, Vanger’s figures are more trustworthy. Although that does sound like the sort of the thing the MTA would say, hear him out. The difference is because Vanger used “wait assessment” data, rather than the “on-time performance” numbers, which, for subway purposes, are much less reliable.
“On-time performance tracks if any particular train shows up at its final stop at the terminal within 5 minutes and 59 seconds of when it’s supposed to,” Lisberg explains. “That’s the industry standard.” And while that’s useful when you’re riding a commuter railroad, you probably don’t care if the Q train dependably arrives at Coney Island when it’s supposed to.
“If you’re waiting for a subway you don’t really care,” Lisberg says. “You just want to catch the next one.”
“Wait assessment” tracks something a little more important, at least when you’re standing at the Bedford Avenue L stop at 1:30 a.m., crammed in next to 16 people carrying banjo cases and birdcages and and all manner of other Extremely Creative, Whimsical Items. Wait assessment means how often a train shows up, against a goal of a wait between trains that’s no more than five minutes. On the train lines with the most customers, the goal is even shorter: two minutes during peak hours and four during off-peak.
In sum, Lisberg says, Vanger’s data is “really cool. The train line is largely doing better.” But he also knows that no one will likely see or recall their subway trips that way.
“You never remember if it shows up right when you get there or if you wait a minute or two,” he says, rather wistfully. “You only remember the time you were running late and then the train was late, too. … Because it’s my job, I pay attention to how long I have to wait, and what the loadings are like on the cars. And they’re pretty good. There was one week a couple weeks ago I had to wait several minutes for the F a couple days in a row, and it really stood out.”
Mike Vanger, who actually made the app, takes a slightly different view. “I was actually rather disappointed with the results,” he writes in an email. “I think wait assessment is a very lenient measure of whether a train is ‘on-time’ or not. But it was nice to see that they have been making some progress, especially with the 1 train, which I remember from my freshman year at NYU as being really bad. It doesn’t take into account, though, how often trains are being run, or the price increases over the same time period.”
For his part, Lisberg optimistically chooses to believe that New Yorkers pay such minute attention to their trains because they’re so passionately invested in the transit system.
“Everyone feels so–I don’t want to say possessive, but everyone in New York feels very personally invested in the subway system,” he says. “The way people in the rest of the world feel about their cars, most people feel that way about their subway line.”
Also, and probably more importantly, people love to bitch.