One of the modern fundamentals of transit planning is the difference between patronage routes versus coverage routes. As their name indicates, patronage routes are services designed to attract a high number of riders, while coverage routes serve areas with lower demand, filling in the gaps of the system [more on this]. With this in mind, I wanted to see if patterns in the shape and routing of bus service had a strong correlation with ridership.
Because routes running along a single corridor, or two corridors, are simpler, faster, and easier to understand from a customer's perspective, I theorized that bus routes running along only one or two corridors would have higher ridership.
To test this, I observed route-level ridership from any city in North America that I could find data from. Turns out many large cities are really bad at releasing up-to-date, transparent information about bus service. I ended up with 9 agencies' worth of data, all from different periods between 2012 and 2017: Seattle, San Diego, Los Angeles, Phoenix, Houston, New York, Chicago, Boston and Montréal.
I divided my findings into six subcategories, illustrated in the table below. The full excel file is available here.
In all, roughly 60% of the observed high-ridership routes are single-corridor services. Another 20% are double-corridor services. Although we see a clear general pattern in the data, I can't really conclude anything from this. This is for a few reasons.
First, my categorization of route shapes is subjective and somewhat arbitrary. While I classified Route 8 in Seattle as a two-corridor route (Denny and MLK), it could be argued that it should be placed in the 'other' category because of its confusing transitional section from Madison Street through to Olive Way. Likewise, I classified Route 7 as a single-corridor route, even though it services downtown from Rainier. There are no set-in-stone criteria for these categories.
Second, route shapes are context-specific. While each of the cities in the study contain long, high-volume corridors where frequent bus service is a must, geographical barriers exist in some cities and are non-issues in others. Chicago's rigid street grid and lack of physical obstacles make single- and double-corridor routing easy. Conversely, Seattle's broken grid, steep hills and wide waterways make this much more difficult.
Third, the study does not account for many other variables which could be equally or more important. Land use along the route, available transportation connections and route length are just some of these potential variables that are unaccounted for. It also ignores factors that are known to be important for ridership, like density and walkability in the surrounding areas.
Fourth, I'd argue, paradoxically, that raw ridership data is a poor way to interpret ridership. Comparing Route 40 in Seattle to Route 32 in Boston provides an excellent example. While both routes have similar raw ridership (10,900 average weekday boardings vs. 11,020, respectively), one trip on Route 40 in Seattle takes about three times longer to complete than Route 32. From start to end, the 40 runs approximately thirteen miles in a confusing semicircle of roads, whereas the 32 spends just four and a half miles along one single, mostly straight road.
The 40 is servicing what two or three 'regular' bus routes would be, and runs along at least four distinct corridors: Westlake, Leary, 24th NW, and Holman/Northgate/105th. Taking the 40 from start to finish would be ludicrous, as the 41 would do the same job many times faster and at a similar frequency. The same cannot be said of Boston's 32. Chances are that the average 32 is a lot more packed than the average 40.
A more relevant data point is productivity, where ridership is divided by the service hours that the bus offers. In essence, if we assume that the 40 and 32 run at the same frequency, the 32 will have roughly three times less service hours (as one trip on the 32 is three times shorter) and thus about three times higher productivity. A similar measure could be derived by dividing ridership by route length. While both of these measures have their faults, they at least somewhat account for differences in routing between bus routes.
To illustrate this difference, here are the 10 most boarded routes in Seattle, as of Fall 2015:
And here are the 10 most productive routes in Seattle, as of Fall 2015:
I'll end this post by urging more transit agencies to release open data. Seattle is pretty unique in releasing statistics like the above, and even these are woefully out of date. It is in agencies' own interest to release these figures, especially because it can be used as transparent, quantitative and unbiased evidence in public information sessions or public comment periods about route revisions and deletions. The MTA in New York recently announced a $1 million prize anyone who could come up with a creative, all-encompassing solution to their agency's woes. They could at least start out with detailed, publicly accessible statistics to assist people in finding one.
Ridership figure references:
Boston: Here
Seattle: Here
Montréal: Here
Los Angeles: Here
Phoenix: Here
San Diego: Here
New York: Here
Houston: Here
Chicago: Here