Background

During any MBTA diversion there are various strategies employed to help riders with alternate routes that can range from free fares, reduced parking, and shuttle services. How utilized these incentives are, and other mode choices during these diversions is not always known. This analysis begins to explore these choices by looking specifically at biking behavior and shuttle ridership. The diversion in question occurred on the Red Line from Alewife-Kendall/MIT station on July 12-28^th, 2024. Shuttles were run in various routes and schedules (see figure 1) and the local bikeshare system, Bluebikes, also offered 5 free rides to any user during this period.

Method

To estimate if biking increased in the area directly surrounding the diversion, we first needed to get an idea of how people typically use the Red Line from Alewife-Kendall. A catchment area of typical riders who may also be in biking range was established and then various data sources were used to estimate if biking increased in this area. We also looked at ridership on MBTA provided shuttles to get an additional perspective on ridership change during the diversion.

Typical Red Line Ridership

To get an idea of typical Red Line ridership, ODX data from Korbato was used to estimate where southbound riders boarded and exited the train on a typical weekday, one week before the diversion period. 32% of southbound riders boarded and exited the train at a station north of the Charles Rivers, but 68% of riders that boarded from Alewife-Central alighted at one of three downtown stations (Charles/MGH, Park Street, or Downtown Crossing). A large portion of these alightings (30%) were from Harvard Station to Park Street, so while we wanted to explore biking behavior north of the Charles River directly in the diversion area, we also needed to consider how people may have biked more into downtown.

Bike Counters

There are 6 permanent bike counters near the Red Line diversion located on bridges across the Charles River used in initial analysis. Additional temporary counters were installed for a period of time pre-diversion and during diversion to add additional data. Placement was determined through a discussion with MassDOT to be at a popular intersection and bike path where we expected we might see increased bike traffic during the diversion (Figure 3). These counters use a combination of sensors that have been shown to be highly accurate in counting the number and direction a bike travels.

This was OPMI’s first use of MassDOT temporary counters and exploration of the MS2 database, which stores the bike count information, and a lot was learned about timing of data collection. When we placed temporary counters, we did not properly consider factors that would limit the amount and quality of data collected like holidays, changes to ridership caused by school ending for the summer, and day of week of installation and removal. Therefore, full days of comparable data for temporary counters were limited. We then decided to compare counter data, both temporary and permanent, to Bluebikes data available on their website. Encouragingly, counter data and Bluebikes data aligned and showed very similar trend and statistical significance (Figure 4). This indicated Bluebikes data could be used to represent bike ridership in this area and provide greater ride detail. MassDOT counters track the number of bikes and direction they travel per hour while Bluebikes trips give the start time and end time, trip origin and destination, type of bike ridden (electric or analog bike), and the type of rider (Bluebikes subscription member or casual rider). Bluebikes data is used for the remainder of the analysis.

Bluebikes Analysis

Now that our data source was finalized to Bluebikes, we needed to determine a comparison time frame pre-diversion to compare to during diversion biking. We used the average daily counts for the two weeks directly before the diversion (June 27-July 11), excluding July 4-5 (Figure 5).

For the comparison area, we considered various combinations of stations that would capture different levels of Bluebikes users. Using ArcGIS Pro, we determined a walking catchment area of two sizes: Bluebikes stations that were within a 10 or 20 minute walk from a Red Line station entrance from Alewife-Park street. The 10-minute buffer is commonly found the be the longest someone would comfortably decide to walk to a transit mode. The 20-minute buffer aimed to consider those riders who may be a 10 minute walk from a station, but have a Bluebikes station closest to them within 10 minutes in the opposite direction. However, this created confounding factors like proximity to other transit lines (particularly the Green Line Extension). After various combinations we landed on two comparison groups:

• System-wide: all Bluebikes stations in the greater Boston area
• Red Line study area:
  • Only Bluebikes stations that were within a 10 minute walk of a Red Line station entrance affected by the diversion (Alewife-Kendall/MIT)
  • Only trips that started and ended at one of these stations
  • Excluded trips that started and ended at the same station (assumed to be recreational)
  • Excluded Bluebikes stations that were within a 10 minute walk of another rapid transit line entrance (so may have altered ridership numbers)

It was intended that this restrictive Red Line study area would be the best estimate of possible direct Red Line replacement trips during the diversion. We then conducted a t-test for a difference in means and various other comparisons in Python do explore how these two groups differed pre and during diversion.

Shuttle Ridership

Central Transportation Planning Staff (CTPS) conducted manual counts of MBTA shuttle ons and offs from morning (7-11am) and evening (3-7pm) peak periods during the diversions period. Two full days of counts (one from Alewife terminus and Kendall shuttle terminus) were compared to the average timeframe directly before the diversion. Because Kendall was the end of the shuttle route and not the end of the Red Line, additional considerations were in place to ensure all riders typically on the train were accounted for in the shuttle count comparison.

Results

Bluebikes system-wide rides significantly increased (p<.001) during the diversion period, but increased an additional 12% near the diversion on weekdays, which may be attributed to commuters substituting bike trips for typical Red Line trips. Shuttle ridership averaged 40%, which met expected levels.

Biking Behavior

System-wide, Bluebikes rides significantly increased at the P<.001 level.  This happened for four hypothesized reasons:

1. Seasonal trend. During this week in 2023, there was a similar rise in trips. Weather (temperature or precipitation) was fairly consistent between pre-diversion and diversion periods in both years and should not have majorly impacted our results and we did not have further historical data to compare, but there may be a regular seasonal increase in biking in July (Figure 7).
2. Overall trend. There is a small, but noticeable increase in biking in dense urban areas around Boston in 2024 compared to 2019. Cambridge is a popular biking area and has seen larger increases than other places in the state and this may be reflected in Bluebikes rides as well.
3. More Bluebikes. As Bluebikes has expanded stations and the number of docks (or places for bikes) at each station, Bluebikes rides will naturally increase.
4. Free Bluebikes. Anyone (member or otherwise) could use a code to unlock 5 free Bluebikes rides on analog bikes during the diversion period. There is no way to indicate in the data which rides used this code.

While there was a significant average system-wide increase (12%) during the diversion, the Red Line study area saw double this change on weekdays (Figure 8). This increase on the main commuting days (Tuesday-Thursday) with a smaller increase on Mondays and slight decrease on Fridays indicates that commuters may have been driving additional ridership and substituting bike trips for Red Line trips during the diversion. There was no difference in the type of rider (member vs. casual) or type of bike that was used in any comparison that was considered.

Another way to visualize this increase is with this heat map created with ArcGIS Pro (Figure 9). This shows the difference in the number of trips that started at each station in the diversion period, compared to pre-diversion at all stops in the Bluebikes system. This map is similar to the heat map for where trips ended during this same period. Bluebikes rides increase the most around Central station and across the Massachusetts Avenue bridge into Back Bay neighborhood, with additional hot spots in downtown and around each of downtown stations, which somewhat aligns with the typical Red Line ridership trend.

Shuttle Ridership

Shuttle ridership aligned with expected numbers based on a previous analysis of an Orange Line diversion. In Table 1, estimated ridership conversion for each period is shown. Average overall ridership for Alewife was 50% and 35% for Kendall/MIT. Shuttles at Kendall operated in three loops: Harvard, Local, and Express. Of the 35% use at Kendall, loops were utilized 49%, 33%, and 18%, respectively. Kendall shuttle ridership was overall lower, which may indicate people diverted in different ways near this stop.

‍

_{Table 1. Boardings at Alewife and Kendall/MIT during morning and evening peak periods compared to typical Red Line ridership during the same period. Ridership numbers are rounded to the nearest 100; percentages are calculated prior to rounding.}

_{*Accounts for total typical passengers onboard the train arriving or departing Kendall on an average comparable weekday, which estimates equivalent ridership for this terminal shuttle stop that would be used by all shuttle riders, regardless of their origin or destination.}

Mode Shift

Finally, to get a rough estimate of mode shift during this period, we picked a route that could have comparisons for both biking and shuttle. We looked typical weekday Red Line riders that boarded at Alewife and exited at Kendall/MIT (Figure 10). We estimate the number of riders that biked by comparing the increase in ridership in Bluebikes rides that started near Alewife station and ended near Kendall. We estimated shuttle use by counting ons at Alewife and offs at Kendall all for the same morning peak period. It is estimated that 36% of typical riders used the shuttle and 2% of riders may have switched to biking.

Limitations and Future Analysis

This project was not a full diversion analysis. A future analysis could do a more in-depth comparison to explore other ways people self-diverted: Did road congestion change during this period? How did bus ridership and Green Line ridership change? Can we estimate how many people just did not make a trip, possible due to remote work opportunities?  

It would be interesting to explore more about Bluebikes trips, particularly bike availability. Bikes available at each dock is available in real time on the website, but not available to us at the time of analysis. Comparing this to the number of rides might indicate if even more people would have biked, but couldn’t due to bike availability.  

There were limitations in the amount of data collected and the time frame of this analysis; all numbers are estimates.  

Acknowledgements

Thank you to our partners at MassDOT who assisted with setting up temporary bike counters and facilitation the use of MS2 database, and to CTPS for shuttle counts.