August 20, 2019

Using Survey Data to Evaluate Walksheds

Studies of walking distances of different subway networks have found that walk distances vary considerably from station to station. In this blog post, we explore how walk distances may vary from station to station in our MBTA network

To use the MBTA, passengers typically have to walk, drive, or otherwise travel between our stations and their homes, offices, and schools. The question of how passengers travel between stations and their ultimate origin or destination is called the “last mile problem.” Typically, when the MBTA tries to answer questions involving the last mile problem (e.g., determining how many jobs are within walking distance of T stations), we assume that passengers won't walk more than half a mile. However, studies of walking distances of different subway networks have found that walk distances vary considerably from station to station. In this blog post, we are going to explore how walk distances may vary from station to station in our MBTA network.

For this post, we're using survey answers from our most recent Rider Census, where passengers were asked to provide information about their most recent trip on the MBTA, including the location of their origin and destination. This provides us an opportunity to calculate how far passengers walk between their ultimate origins and destinations and MBTA stations. For each rail station and Silver Line station, as well as for each bus line, we used bootstrapping to calculate a confidence interval for the average distance passengers walk to and from MBTA stops. We then focused our analysis on the Red and Orange Lines, and identified three interesting trends: passengers walked longer distances to reach stations at the ends of the Red and Orange Lines, passengers walked shorter distances to stations constrained by bodies of water, and passengers walked shorter distances to stations in the middle of the Orange Line.

Methodology and Data Sources

As mentioned, the MBTA and CTPS recently conducted a systemwide passenger survey. For the survey, we asked passengers about their most recent trip on the MBTA. The survey asked passengers to list their origin and destination locations—where they are coming from before arriving at a MBTA stop/station and where they are traveling to after completing their trip on the MBTA. They were able to classify these locations in a variety of ways, like home, workplace, school, etc. The survey then asked passengers to list their mode of travel (driving, walking, biking, or use of a non-MBTA service) when going to and from the T in order to learn more about this “last mile.” Passengers listed the specific MBTA service they used (e.g. Green Line, bus route 7, Fitchburg Line, etc.) and at what specific stops they boarded and alighted. Passengers also provided basic demographic information.

Not every respondent provided an origin or destination location, so we separated the dataset into two groups: responses that included an origin location, and responses that included a destination location. (Responses that included both an origin and destination location were counted twice.) Since we are investigating walkability, we filtered the datasets so that they only contained responses from passengers who identified their access and egress modes as walking. This left 15,934 responses from passengers who identified an origin location and walked to their first MBTA boarding and 18,161 responses from passengers who identified a destination location and walked from their last MBTA alighting.

For each of the responses, we calculated the walk distance by calculating the straight line distance in meters from origin and destination locations to the location where they boarded or exited their first or last MBTA service experience. In cases where passengers were using rail or Silver Line service, the survey identified the exact stop at which passengers boarded and exited the service. However, in cases where passengers were using bus service, the survey did not identify the exact stop at which passengers boarded and exited; the survey only identified the bus line that passengers took. Therefore, we assumed that bus passengers would walk to the bus stop closest to their origin or destination location, and used the bus stop nearest to the passenger's origin or destination location to calculate the walk distance.

Finally, we filtered out stops and bus lines that had less than thirty data points. The Green and Blue Lines did not have a lot of stations with more than thirty data points, whereas the Red and Orange Line stations all had more than thirty data points each . Therefore, we decided to focus on the Red and Orange Lines for the purposes of this blog post. We mapped the mean and median walk distances for the Red and Orange Lines in QGIS (we did not map the walk distances for Downtown Crossing, as that station is shared by the Red and the Orange Line).

Possible limitations of the data include:

Results

Key: Confidence Interval = CI

[More Stations]

An image of the mean walk distances for the Red and Orange Lines. Walk distances are longer on the terminals.
An image of the median walk distances for the Red and Orange Lines. Walk distances are longer on the terminals. The medians are generally smaller than the means.

Conclusions

There are a number of interesting conclusions that can be drawn from the mean and median walk distances from each station. We have tried classifying them into a few main trends as explained below:

Physical Landscape — Safety & Geography

The Charles MGH station is notable for having a substantially lower median and mean walk distance compared to the other Red Line stations. There are a few possible explanations for this. First, the built environment of Charles MGH is particularly inconvenient to pedestrians: the station has only two crosswalks, two entrances, and is surrounded by busy roads. Pedestrians are also constrained by two geographic features--Beacon Hill (the hill, not the neighborhood) and the Charles River—which could limit how far pedestrians are able to walk to reach Charles MGH. Kendall Square, which has the fourth lowest mean walking distance out of all Red Line stations, also is adjacent to the Charles River, which provides further evidence for bodies of water like the Charles River affecting station walkability. A similar effect can be seen at Assembly Station on the Orange Line, which is surrounded by the Mystic River and Interstate 93, and has a lower average and median walkshed than the adjacent stations (Sullivan Square and Malden).

Last Stations on Subway Lines

Stations at the ends of the Red and Orange Lines—Alewife, Davis, Forest Hills, Malden, and Oak Grove—tend to have larger average and median walk distances. This is probably because passengers who live beyond the reach of the Red and Orange Lines prefer the Red and Orange Lines to alternative MBTA services (the bus network and the commuter rail), and are willing to walk further distances to reach the Red and Orange Lines.

Competition

The stations at the center of the Orange Line—beginning at around Mass Ave and ending at around North Station—tend to have lower medians and means compared to other Orange Line stations. There are a few possible explanations for this. First, this section of the Orange Line is not only very close to the E branch of the Green line, but also runs parallel to it. This means that passengers can choose between the E branch and the Orange line, and it's likely that one of the factors that goes into that decision is which line has the closest stations, so passengers are likely minimizing their walking distances during that section of the Orange Line. Another factor is that Orange Line stations in the center of the Orange Line are particularly close together, which could affect how far passengers need to walk to reach an Orange Line station.