By Christopher Yuen
For cities aiming to increase transit ridership, expanding transit to reach more people is only half the task. The other half is to encourage people who value transit to locate where high-quality transit is possible.
To that end, tools that help quantify the usefulness of transit at any given location can be extremely valuable, whether for a person finding a place to live, or for a business looking to locate in a place most accessible by its employees and customers.
Since 2010, the tool that has gained the most attention has been Transit Score. With it, you type in an address and are returned a two-digit score that is supposed to approximate the usefulness of transit at that location. While simple and easy to use, Transit Score’s methodology has a few flaws, as discussed in our last post about it. Two issues particularly stand out:
- Transit Score assumes that the sexiness of transit technologies compensates for their objective uselessness. For example, Transit Score assumes that you’d rather wait 20 minutes for a streetcar instead of 10 minutes for a bus, even if the two will have the same speed and reliability.
- Transit Score describes the transit around a site without evaluating where it goes. Frequent transit that drove around in circles inside your neighborhood would score exactly the same as frequent transit that went straight across your city and formed a connected network, accessing countless jobs and opportunities.
Recently, the beta version of a new tool, MobilityScore, was released. It offers a similarly simple two-digit score, but with a broader scope than TransitScore. Created by TransitScreen, a company specializing in real-time transit information displays, MobilityScore is described as an “easy-to-understand measure of your transportation access.”
MobilityScore takes into account all your options, from public transit to carsharing, bikesharing, and hailed ridesharing services, to give you a number from 0-100 that will tell you how easy it is to get around. – transitscreen.com
MobilityScore works by scoring the typical time it takes to access each transportation mode based on past availability and response time data from bike share, car share, and ride-hailing services. For its transit component, it generates a score based on the frequency of scheduled trips near the address being queried. It then aggregates the scores of each component to generate a total “Mobility Score”. The tool does not directly indicate the score of each component, although it does indicate the fraction of overall mobility that comes from each mode as percentages.
In some ways, MobilitysScore appears to be an improvement over TransitScore. For example, TransitScore weighs results based on vehicle type, rewarding rail and penalizing buses, regardless of whether they differ in speed and reliability, but MobilityScore’s transit component is only based on frequency, without bias based on vehicle type.
However, a true measure of mobility must also look at the speed of travel on the relevant transit routes. Taking this a step further, a fair measure of the usefulness of transit must consider the access that the service provides– the number of jobs and opportunities that can be reached from the starting point within a reasonable travel time. Like TransitScore, MobilityScore only tells you where you can easily access transportation services to begin your trip, which says nothing about what destinations you can reach in a given time.
One could consider MobilityScore’s expanded scope (including ride-hailing, car-share, and bike-share) to be an improvement over TransitScore, given that transit riders are likely to also occasionally use those other modes. However, the resulting combined mobility score should be interpreted with some caution. For a data-driven approach meant to be objective, the combination of ride-hailing, car-sharing, bike sharing and transit scores into one single measure introduces a new layer of value judgement that is difficult to generalize. While a well-paid executive may place a high value on the ride-hailing services for everyday use, a person with a limited income may not find the availability of ride-hailing to be relevant, since it is not affordable to them. Similarly, car-sharing services may not be useful to a person without a driver’s license.
The mere availability of a ride-hailing option, including taxis, uber, or lyft, seems to result in a minimum score of 40. This optimism means that many sites, without any transit service, and a significant drive from urbanized areas are said to have – “Fair Mobility” despite being completely impractical for the average person.
It’s not easy to build a measure of transportation access that is both objective and simple to use so despite my criticism of some aspects of MobilityScore, I think it is great to see the development of more products like this. Hopefully, when MobilityScore evolves beyond its beta version, it will address some of its current deficiencies and become a serious contender for inclusion in real-estate listings. As we suggested before, the real two-digit score that matters may be a percentage: What percentage of your city’s jobs and opportunities can you reach from this point, in a given amount of time?
Christopher Yuen is an associate at Jarrett Walker+Associates and is a regular contributor to this blog.
“hailed ridesharing services”?
Can we stop this, please? They are taxicabs! There is no “sharing”, it is a vehicle with a driver, for hire.
Yes, there are NUMEROUS taxi companies in every city that *don’t* use an app for hailing and/or who *don’t* have pre-negotiated fares, but the market will take care of those eventually.
Yeesh. Worse than “bike share” and “car share”.
“The other half is to encourage people who value transit to locate where high-quality transit is possible.”
It should be the other way around. The transit service should be high-quality across the city and suburbs.
People can choose an area with good transit. But if they are limited to only a few areas that have good transit, then they are not going to use transit or be regular users, because they are cut off from most of the city region they live in.
Not the recipe for good transit usage, and it shows in most American cities.
Totally agree MB.
“The transit service should be high-qualtiy across the city and suburbs.”
This is a nice ideal. However, the reality is with limited budgets and existing city layouts some places are just going to get better transit than others, and not everywhere is going to be able to have high-quality transit. It makes sense to encourage people who want transit towards those places at the same time that transit tries to expand its reach.
Besides, this is just an information tool. People can use that information as they please.
In most American cities, limited-access highways penetrate deep into the city, and this means that cars can get to any point in an absolutely enormous area around the “city”. By contrast, even rapid transit lines have a lower average speed, their passengers lose (on average) half the headway waiting, and the collection/distribution to/from the stations happens via options slower than cars (i.e. walking, bikes, or local transit). Even with grade-separated options (mostly heavy rail, occasionally BRT) as trunk lines, transit can only ever serve a fraction of the area that city-penetrating highways plus local street networks can cover. However, this is not a problem, because non-car-using people take little space, and can fit just fine in the area that transit can serve. IF those areas are allowed to densify, which is sadly rare in American cities. General NIMBYism, anti-gentrification (which leads to gentrification), and simply inertia in not changing zoning means that in many places, prime real estate close to rapid transit is zoned as single-family residential, on large lots.
The pictures in the article are worth a thousand words.
Thanks for posting this throughtful review of MobilityScore. Christopher, I really appreciate your reviewing our work from a professional transportation planner’s perspective, and highlighting a few areas of MobilityScore that will need further development as we move out of beta. I’d like to respond to two points in your review:
1. Your farm example highlights some changes in mobility that I believe transit planners need to keep up with. The farm you selected is 4.7 miles (13 min drive) from Hillsboro Central Tri-Met station. At this very moment (Friday evening) I can take an UberX from there to the station for $9, which will arrive in under 10 minutes. UberPOOL isn’t currently available, but if it were, that cost will be closer to $5 – which would make UberPOOL + TriMet a viable, affordable commuting option to downtown Portland. Taking all that into account, a MobilityScore of 42/100 (Fair Mobility) seems appropriate for that farm!
2. There are significant differences between the tools planners require, and tools that are useful for the general public. Accessibility analysis (“what percentage of your city’s jobs…can you reach from this point in a given amount of time”) is very useful for planning and government policy, but as an individual, I might not care about having access to 60% of a city’s jobs if they’re not ones that match my skills — I care about access to the jobs I can do!
At TransitScreen, we primarily build tools for the public, but we intend to make MobilityScore (and our forthcoming MobilityScore PRO tool) useful to planners, so comments from the community are appreciated!
Considering your responses to these two points I have two questions.
1. Rather than the tool making a judgment on what level of affordability for these mobility options makes them accessible, could the tool show a simple estimate of the cost of each mobility option at a given location (even as simple as the 1, 2, 3, or 4 dollar signs seen on restaurant lists)? This would allow the tool users to make that judgment for themselves.
2. I think you make a good point that transit is only as valuable to me as its ability to connect to my destinations. Even if it can get to 80% of all destinations in a city, but no to my job, then I don’t care about the 80%. Could you provide an option for people to enter in a list of destination addresses, and get a score adjusted for that level of detail?
I understand that the goal is a simple score, not to do specific trip scenarios, which people can just do in Google. However, I think the points Christopher made, and you make, highlight the difficulty of presenting useful information in a single number format.
I entered my home address (in rural northern New England), and got a score lower than 40, which is appropriate. So the tool does go lower. However, it listed car sharing as an option. I suppose I could drive somewhere to rent a car, but there is no car sharing location anywhere remotely accessible to my house which would allow this to be an option.
Jack, appreciate the feedback.
1. We initially built MobilityScore with time as the primary consideration, but it’s clear that for commuters as well as low-income people, it’s necessary to include some price info. I have some ideas and we’re looking at ways to make this more transparent in a future version of MobilityScore.
2. A multimodal commute score for specific addresses would be a useful tool. Even Google is limited as to what mobility options they allow you to plan trips for. However, making that into a score will take a bit more work.
Would you please report the car sharing issue via our contact form? Either there’s a data issue we need to fix, or you have a carshare nearby you don’t know about.
Even for people with relatively modest budgets, the availability of ride hailing still matters. A typical car owner might easily spend $200-300 each month just on fixed costs, such as parking, car payments and insurance, just to have the car sit there, not even including the gas to actually drive it. For those willing to ditch the car, that $200-300/month can buy a ton of rides on Uber and Lyft for those occasional trips to odd areas where transit doesn’t go, or the transfers are too inconvenient.
Of course, the above only works if your area has reliable enough service that you can actually depend on getting picked up within a reasonable amount of time – which is why the availability of Uber/Lyft matters. As to old-fashioned taxis – yes, they do exist, but they’re way more expensive, are much less reliable in showing up, and much more cumbersome to use. Back in the old days before Uber/Lyft, I found it virtually impossible to order a taxi in most outdoor locations because the traffic noise would completely drown out the conversation with the dispatcher. Even when ordering a ride from the comfort of home, it was not uncommon to experience 30-60 minute wait times during busy periods.
That said, as you start getting out of the urban core, the availability of ride hailing services starts to very dramatically depending on the time of day, as well as luck of the draw, whether a driver happens to be around at the time you want a ride. Usually, the area that the companies mark as “having service” on paper is substantially larger than the area in which you can reliably be picked up, even during the daytime, and late at night, the reliable pick-up area is pretty much relegated to dense urban areas, and airports.
asdf2, a number of good points here.
Many people underestimate the cost of car ownership – $5,000 per year is not at all unusual. Someone who only made commute trips to work and back would have $20+ per day to spend on ridehailing and public transit – obviously this behavior isn’t typical, but the point is the cost isn’t wildly out of line with car ownership.
Looking at access with multiple modes is problematic for several reasons.
1) How to differentiate multiple modes serving the same area from the modes serving different areas? For example, I can go to the nearby mall on foot, by bike, or drive. Does it get triple-counted? In that case, counting barely-different modes separately (carshare, taxi, ride-hailing, &c) inflates the score. Does every destination get counted only once? Then transit basically doesn’t matter, because almost any destination can be reached by car faster than by transit.
2) How to weight the cost in money versus the cost in time?
3) There is a sharp difference depending on whether one has a car and already paid its fixed costs. If so, then its marginal cost is lower than that of taxi and ridehailing services, thus the latter are mostly irrelevant.
4) In most of America, vehicular cycling is the prescribed way to ride a bike, and most people are simply not willing to do that, because they find it terrifying. But there are a few brave souls who don’t mind mingling with cars, and to them, their bike provides better mobility than local transit lines. Zero headway, similar average speed.
In all of these cases, if the developers bake in a single way of handling this, the result will not really match what most users look for. Opening up the system and letting the user play with some sliders can solve this, but that may be too complex, and some users might be offended when the app asks how much they value their time.
“What percentage of your city’s jobs and opportunities can you reach from this point, in a given amount of time?”
This metric produces disproportionately high values for smaller cities. In Tinyville, Middle-of-nowhere, 100%.
They’ll need to provide a way to toggle or weight the modes. Otherwise it’s making too many assumptions.
e.g., I don’t have a driver’s license (temporarily I hope) due to a medical issue. Can’t use car sharing sevices, and now the tool is useless.
Bike share? I like cycling, but here in the Great White North, *I’m* unlikely to use it for mobility for 5-6 months of the year. So really I’d want to see scores with and without bike share. Etc…
Good point about seasonality of bikeshare. The contribution of bikeshare to the overall MobilityScore will take into account seasonal availability of bikes once we finish collecting data on open/close dates.
We’re looking at allowing more customization based on user abilities/preferences, but I think there’s value in having a more “objective” score that is specific to a location rather than a location and individual person.
Seasonality would be a good feature. It would also be helpful to have an “accessibility” mode, as bike share, Uber, Lyft, etc. are generally unavailable for individuals in wheelchairs or with other disabilities.
Why not just allow people how to weight their options? All of that data is already there.
Having one single MobilityScore that takes all options into account makes it easier for people to talk about mobility and find common ground without dealing with idiosyncrasies like “my MobilityScore at this location is 82 because I don’t bike, but yours is 88 because you don’t currently use carshare.”
For example a hospitality website like MobilityScore partner StayAttache shows a single MobilityScore per listing.
I know there’s value in personalization, but that’s a different kind of tool than basic MobilityScore.
I too would like a filter that can give a mobility score that excludes certain modes. I happily ditched my car and the savings let me pay more for housing in a transit-accessible, walkable area, not more for sharing or hiring cars. I prefer public transportation and would value an accurate measure of car-free mobility when it comes time to move.