More proof that 'on-time' performance metrics don't actually tell us very much.
A new analysis of 15 Washington Metro bus lines shows that the system has an on-time performance rate of 75 percent. That sounds good on the surface, until you realize that Metro considers a bus "on-time" when it arrives anywhere in a 9-minute window between 2 minutes early and 7 minutes late. In Metro's eyes, you can show up at a bus stop, listen to "Stairway to Heaven" in full, wait one more whole minute for the bus to arrive, and still leave "on time."
Metro's performance also takes a hit when you consider that its on-time window is far more forgiving than the industry standard. By the standard measure — 1 minute early to 5 minutes late — Metro's on-time performance falls to 61 percent on average. Even that figure fails to give an accurate picture of service quality, because it measures whether the bus itself was on time, not whether a rider arrived somewhere on time.
Consider a hypothetical situation laid out a few years ago by Michael Perkins of the Greater Greater Washington blog. Let's say you want to catch a Metro bus scheduled to arrive at 8 a.m. and you show up at the stop right at that time. If the bus arrived 2 minutes early — on-time by Metro's standards — you would have to wait for the 8:10 bus. If that bus arrived 7 minutes late — again, still technically on-time — you would wind up waiting 17 minutes for a bus.
Meanwhile, both buses would receive a perfect score from Metro in on-time performance.
Of course, the inaccuracy of the on-time metric breaks both ways. Consider another hypothetical situation. Let's say you want to catch that 8 a.m. Metro bus again. You arrive right at that time and a bus pulls up. In fact, this was the 7:50 bus running 10 minutes late, but you don't care, because you didn't wait at all and made it to your destination on time. In this case Metro fails its performance measure while making its riders perfectly happy.
The flaws of on-time performance have led some experts to advocate replacing the metric. Transit planner Jarrett Walker, for instance, has explained that some agencies focus instead on "headway maintenance" — keeping a consistent gap between buses. The idea here is not that a bus arrives at a certain time but that it arrives a certain number of minutes after the one in front of it. Knowing you can arrive at a bus stop anytime and still not wait long for the next ride is most people's idea of a high-quality transit experience.
A headway metric can help a big city, high-frequency system avoid bus bunching and make service feel frequent and reliable. For sure, there are bureaucratic hurdles to the shift: in an on-time system, drivers are the ones responsible for performance; in a headway system, managers bear the responsibility. But with GPS location data, the technical challenges of going from an on-time performance metric to a headway metric are far less daunting than they once were. It's been said that there's still time to change the road you're on.
Top image by Flickr user AndrewArchy.