It’s been a little more than a decade since Google and TriMet, the transit authority for Portland, Oregon, partnered to create a standardized format for scheduled transit data called GTFS. That breakthrough, and the subsequent rise of vehicle location data, has made real-time transit information an essential part of urban life, from the arrival screens at train platforms or bus stands to the smartphone apps that help people plan trips.
The case for cities to embrace real-time transit data — either by collecting it or making it open to others — just got a lot stronger thanks to transportation researchers Candace Brakewood of the University of Tennessee and Kari Watkins of Georgia Tech. They reviewed dozens of studies on the benefits of real-time data to people’s lives and compiled their findings for all the urban planners and transit agencies looking to invest in real-time infrastructure.
“The people in the transit agency who have not yet done this, or have wanted to but haven’t been able to put together the business case for it — this should really help them,” says Brakewood. “You’re going to get some real tangible benefits that many other cities have found.”
The primary benefits include reduced wait times (people use an app to time their walk to a stop or station), reduced travel time (people adjust their trip choices), and increased transit use (people like reduced wait and travel times). In time, the higher-order impact stands to be even greater: a future where integrated real-time data from all transportation options enables a true mobility system that rivals private car use on convenience.
“The more we can get to those multi-modal, real-time transportation apps, the better informed the traveler will be and the better choices they can make,” says Brakewood. She spoke to Sidewalk Talk about the science of real-time transit data — and its role in a potential car-free urban future.
I think a lot of people wonder why it’s hard to get this data out to people. Even here in New York, people didn’t understand why every subway line didn’t have this until recently. What’s your sense of the biggest barriers to faster or more widespread access to real-time transit data?
It’s totally different on the train versus bus side. On the bus side, it’s AVL [automatic-vehicle location]. In that case, if you don’t have AVL on your vehicles, that is more of a capital investment. That’s going to be a fairly large investment, but there will be many additional benefits in addition to the passenger benefits on the operating side. It’s much easier operations when you know where vehicles are.
Some systems and agencies have outdated AVL systems that do need upgrades before they can get real-time data in a format that’s workable. There’s still not good standardization in terms of what format should bus GPS or AVL data be provided to third-party app developers. On the schedule side, GTFS is so common, and most transit agencies accept it as the de facto standard. On the real-time side, moving toward a standard hasn’t happened as quickly. For example, Bus Time in New York City uses the Siri format, a European-based standard. Others use GTFS Realtime. Others just provide data in the format the vendor gives it to them. That is a challenge on the bus side — getting your data in a workable format.
In the case of the New York City subway, that’s a unique one, in that it does have such an antiquated signal system. So I wouldn’t generalize that to most other places. In subway systems, we’re not talking about GPS. We’re talking about the track circuit system. Where is the train as it’s moving on the tracks. You’re not sending a GPS signal. You’re communicating with a track signal system to have safe movements of the train. Most modern train signal systems easily provide the location. For new systems like communications-based train control, that is the basics of how it works.
The clearest benefit of real-time transit data that you found in the literature is decreased wait times. Presumably that means someone is using their device to look up the next train before they leave for the station. What do you see as the most powerful evidence of this based on what you’ve reviewed?
Of the 13 studies that have looked at wait time, 12 of them found positive results, at least partially. We did not divide the studies between “actual” or “perceived” wait time because most studies did not differentiate it. Actual wait time is what you just said: checking the real-time information before you leave home or work and timing your departure to the station or stop to meet the vehicle and reduce your physical wait time. The perceived wait time is how long you think you are waiting when you’re standing there.
Having countdown clocks in every New York City subway station, or really any signage or display, will likely reduce people’s perceived wait times. But having the apps they can check before they leave home or work, that’s what primarily reduces the actual wait times. The two different technologies are linked to the two different pieces of waiting.
Watkins et al 2011 is one of the few studies that untangled it in a very systematic way. They found a two-minute (or 30 percent) drop in perceived wait times, and also a two-minute drop in actual wait times for people who used real-time information.
Saving two minutes a trip doesn’t sound like much. But that adds up over time and in the long run, if I begin to associate transit with something that’s just there when I show up, it makes it feel more on-demand, like a car.
Moving forward, I think people expect this information for that exact reason. If transit wants to be competitive and remain competitive in a future transportation system that has a lot more mobility-on-demand options, the wait times cannot be extremely long. Or if they are long, people need to know exactly when the vehicle is arriving. They expect that in this day and age.
On perceived wait time, it doesn’t actually change when the train arrives. What is it you think that makes this knowledge so powerful?
You think you are waiting less. You are getting to the stop, you are waiting there, and now that you have the information, you know the vehicle’s coming in let’s say four minutes, and it actually feels like four minutes. Whereas before, you were standing there anxiously, often times sticking your head out over the platform to the vehicle tracks to see if you see headlights. That is an experience that passengers hate. Really, I think hate is a fair word. No one likes waiting, and no one likes an uncertain wait time.
In fact, if you’re a transportation planner and you’re building a travel-demand model, and you’re trying to figure out how people will take transit versus driving, when we look at the travel time of that, we are often saying people perceive wait times as two or three times the amount it actually is. We’re saying people really dislike it, and therefore we need to penalize the wait time more than the in-vehicle travel time. These studies are showing that real-time information is starting to bring that back down to the actual amount of time that is when people are standing there.
The second type of finding is around decreased travel times. But you found a very wide range of these savings: 3 to 45 percent. Can you explain how real-time data would lead to lower travel times, and also your sense on why there’s such a wide range of evidence right now?
The how is an example: I’m standing on the 1, 2, 3 subway line in New York City. I could hop on the local train or the express train. If I see on the countdown clock that the local is coming in one minute, but the express is three minutes behind it, I’d still wait on the platform for three extra minutes — which in the past you would not want to do, because waiting is the worst — knowing that the total travel time is less.
Why they’re all so different is that these studies so far are not observing what people are doing but rather, based on some sort of framework or survey, they’re then running a simulation model to get at a range of travel times. That is one of the key areas for future research — to ground that in actual observations of people as they’re making their decisions. And it’s probably going to be primarily in dense urban networks, where you have multiple options to get from your origin to your destination, not in smaller urban areas where you only have one path to get from Point A to Point B.
That is how I think about data reducing travel time — the ability to choose the best mode for any given trip. That means creating mobility tools that capture all the real-time data in one place. How important does that effort seem to you in terms of reaching a place where real-time information really impacts mobility choices?
I think it’s really important. From the traveler perspective, what you really want is some sort of real-time trip planning platform that is going to give you all different modes at the click of a button. To make it even better, it’s not just the information. We need a seamless way to pay, preferably through your app, as well as to navigate through the system and to provide you with other pieces of information, like crowding. The payment piece would be nice from the traveler perspective to be fully integrated. That to me seems like a larger hurdle than the information piece, because there’s so many different stakeholders involved.
Another area you explored is whether or not real-time data actually increased transit use. It stands to reason that lower wait times and lower travel times would mean happier riders and probably a more frequent rider. But there are still some mixed findings. Where do you see the evidence heading?
In the case of ridership, we’re seeing studies done in large cities with dense transit networks being the ones most consistently able to find an increase in ridership. Let’s say it’s a discretionary trip — going shopping. In my thinking of it, when transit is really good service, you’ll check your app when you’re at home, and if the bus is only two minutes away, you’ll probably going to take it. But if you check and you’re in a smaller city where the bus doesn’t run often, you’re going to say: “The bus is 20 minutes away, I can drive there quicker.”
So I think you’re capturing more trips most likely in places where your transit service is very good: frequent service, lots of coverage in terms of routes. In those cases, then this becomes a tipping point, where you can capture more and more discretionary trips on transit. But in smaller cities, where they have infrequent service and less coverage, you’re probably not going to see that many ridership changes, in terms of getting more people on the bus, or more likely, getting those people already riding to make their discretionary trips on transit. If transit can start to get those additional trips, that’s a big benefit.
Is there a point at which transit frequency is so high and consistent that you don’t need real-time data?
That study is on my list! A few other researchers have asked me that exact question before. If you give an example of the New York City subway, they say: “Is it even needed?” If you’re running two-minute headways, does it matter? Honestly, probably not from the actual wait time perspective. It may for the perceived wait time. Your one-minute wait might feel like two because you’re not sure it’s coming in two minutes. And it may from the travel time perspective. I’m going to take the local versus the express because there’s a three-minute difference. I think it still matters, but it matters less in terms of the actual wait times, and more in terms of the perception and potentially the total travel time.
I can attest to using it all the time, even on the New York subway. I live on a line with two locals, the B/C. I need the C but if I see the B is four or five minutes ahead, I will take that the B to the express stop where I can often board an A to my destination and beat the C. To me it’s the kind of thing where I’m an active participant in using transit.
As Dr. Kari Watkins likes to say, you’re more in control of your trip!
In the case you just gave, that’s another area for future research I’d love to do: Do we see an increase in transfers? There was one study of Chicago that showed that if you have real-time information on the bus, people are then more likely to transfer to the train. It was a very small amount. Their argument was — I believe their words — “increased intermodal transfer efficiency.” Basically, you know the bus is coming in two minutes, so you will get to the train station on time to transfer. It does facilitate that transfer time, similar to how it eases the wait time at the beginning of the trip.
And when we get to the point of integrated payment for bike-share systems, etc, then we’ll start to see more transfers between shared modes. Or more people to take a bike-share to a subway station and hop on. We need to have real-time information for all these modes, which we’re getting close to. Then an integrated payment mechanism as well.
What makes you most excited about where real-time transit information might be headed?
I think it’s that one-stop shop for everything. That’s what I’d love to see. Information not just for transit but for all these shared modes. If I want to get around without a car — and that’s really how I prefer to travel, and I feel like it’s how many Millennials want to travel as well — how do I do it? What’s the fastest way? What’s the cheapest way? And I want to do it all from my phone. I want to be able to pay for it. I want to know where the information is. I don’t want to have to sign up for a million different things to know what the cheapest option is. I want to know the cheapest, fastest option right now to get from Point A to Point B any way but my own personal car. If we can get to that point, that would be awesome.