Starts & stops

MBTA commuter-rail tardiness data pull up short

By Noah Bierman
April 5, 2009
  • Email|
  • Print|
  • Single Page|
  • |
Text size +

It's been documented before that commuter rail service in Massachusetts is among the tardiest in the country, with trains showing up late more than 20 percent of the time in some recent months.

But the official MBTA numbers tell only part of the sad commuting story, according to a Globe analysis. The rush-hour trains, the ones most commuters take, are actually late about 6 percentage points more often than indicated in the official statistics, which reflect both peak and off-peak performance.

That's right, another instance that bolsters the famous observation about the three kinds of falsehoods ("lies, damn lies, and statistics").

Most regular passengers intuitively know they are late more often than the numbers suggest, and a few have written in to ask that I dig deeper into the statistics. They claim that the T and its private contractor "pad" their ontime performance statistics by including midday trains that carry few passengers.

Here's how it works. Take January, one of the worst months of the year on the tracks. If you took commuter rail during the morning rush hour, odds are that you were at least five minutes late - the T's official definition of tardiness - 38.4 percent of the time. If you took the train during evening rush hour, you were late 28.3 percent of the time.

But officially, overall train service was late only 23 percent of the time. That's because off-peak trains were late much less often than peak trains - only 21 percent of the time - and account for about two-thirds of all trains counted in the total.

The Massachusetts Bay Transportation Authority does not break down ontime performance like that. The private company that runs the service for the T, the Massachusetts Bay Commuter Railroad Co., reports only line-by-line monthly statistics. Even internally, the company does not review peak versus off-peak performance.

At the request of the Globe, the company provided data showing morning and evening rush-hour performance as well as off-peak for the six months between August and January, which yielded the 6 percentage point figure.

No one, of course, would brag about any of the statistics, even the rosier ones that show about 1 in 5 trains late.

"Any delayed train, in my mind, is a delayed train and in my mind we can't have that," said Richard Davey, general manager of the T's private contractor.

"It's no coincidence that I talk to my senior staff immediately after every morning rush hour and immediately before every afternoon rush hour."

Davey said peak service is much harder to keep on schedule, because trains are coming in and out of South and North Station every minute or two, creating some of the same potential for gridlock that occurs on the highways. When one train is late, it often cascades, meaning several lines are affected.

Davey said he is "obsessed" with ontime performance and touts some improvement in February (12.3 percent late) to show that the system is rebounding from its poor winter performance. March was a much better month (8 percent late), after a mid-month schedule change that added more time between stops. March was also the rare month when evening rush-hour service surpassed the overall ontime performance. The official goal is 95 percent on time for each line.

When he reviews the calculations, Davey doesn't make any distinction between rush hour and the rest of the day.

"From a service perspective, I don't want to be saying that one train is more important than another," he said.

That's a curious position, considering 69 percent of his passengers travel during morning or evening rush hour.

Many of those passengers continue to be fed up, or worse.

"I haven't taken it for six months, because it's just been very frustrating," said Steve Atchue, a software consultant from Grafton, and one of the commuters who questions the commuter rail data.

Specialists say the rail statistics represent a classic clash of cultures. Train operators look at ontime performance one way - how fast is the equipment moving? - without taking into account the passenger point of view - how fast are the people moving?

"In the end, what you want is not how many trains are late, but what's the possibility that you are late," said Yossi Sheffi, director of the MIT Center for Transportation and Logistics.

The T's operator is fined for every late train, with adjustments made for those deemed outside the operator's control. But the company does not do much to publicize its data. It's available on request to reporters, and is posted in some larger stations on signs. But it's not on the Internet, something a Mass. Bay Commuter spokesman said may come soon.

Archon Fung, a political scientist who is part of Harvard's Transparency Policy Project, argues the T and other public organizations should make all their data available to the public over the Internet, in raw form, so the customers can "mash it up."

Customers could slice and dice the statistics to find out which trains are typically 23 minutes late and which are always on time. They might take it even further, given the level of education and interest in the region.

That helps in several ways. It puts pressure on a public agency to improve performance and it lets individuals make informed decisions about their travel plans.

"A lot of policy makers think that just putting out the information is enough," Fung said. "What you should do is think from the perspective of the person that is going to be using it."

Please send complaints or comments to