Introduction
On an average weekday, the MBTA’s Regional Rail helps passengers make approximately 100,000 trips by running about 520 trains through 137 stations on 12 lines. Keolis, the operator of Regional Rail, regularly collects and reports line-level ridership data to track how average weekday and weekend ridership is changing. This line-level data helps provide information about the shifting needs of Regional Rail passengers.
Now, for the first time since Spring 2018, the T has a comprehensive count of train-level and station-level weekday Regional Rail ridership on all 12 lines. These counts can be found on the Open Data Portal, along with the data from previous 2018 and 2012 counts. The goal of this weekday station level count was to obtain a point-in-time data set that provides a deeper level of information about how ridership on the Regional Rail system has changed. Where exactly are people boarding and alighting? And how has the new all-day service model impacted peak ridership?
This new data can be thought of as composite weekday ridership – it represents the number of trips taken on Regional Rail on a normal weekday in Fall 2024. Every train that runs on a weekday was counted. The data includes the number of people getting on and off at each station for each train, as well as the number of people who remained on the train as it left the station (referred to as “Load” in the data). It is important to note that these ridership counts, as well as the below analysis, do not reflect weekend ridership, where the Regional Rail system has seen significant growth.
Process/Methodology
These Regional Rail counts took place from mid-September to mid-November and were administered by MassINC Polling Group. MassINC created a schedule for the counts and worked with local staffing agency Hire Partnership to find the temporary workers who performed the bulk of the field work. Counts took place on Tuesdays, Wednesdays, and Thursdays throughout Fall 2024. Multiple workers were assigned to each train to ensure count accuracy.
Most counts were performed on the train itself -- workers sat by the doors of the train from the first stop to the last stop. However, depending on the station and time of day, some counts were performed on the platform rather than on the train. This was mostly important for busy stations, like Back Bay and Ruggles, during peak ridership hours. In these cases, workers stood on the station platform to begin counting passengers waiting for the train in advance of the train arriving. For South Station and North Station, counts were calculated with the On, Off, and Load values from the rest of the train’s journey.
In an effort to check the counts for quality, OPMI compared boardings and alightings at North Station with average weekday entry and exit data from the North Station Regional Rail faregates. The entry and exit averages were calculated by looking at Tuesdays, Wednesdays, and Thursdays between Sept 17 and Oct 10 (roughly the same timeframe as the four North Station lines were counted). The numbers matched very closely: the count data shows there were 11,186 boardings compared to 11,521 entries, and 11,190 alightings compared to 11,048 exits.
Data Caveats
Whenever possible, trains were not counted during events that would impact ridership. For example, during the September 2024 speed restrictions diversion on the Red Line, the MBTA encouraged riders to take Regional Rail at Quincy Center and Braintree stations as a fare-free alternative service, so Regional Rail lines that included those stations were not counted until after the diversions ended. Additionally, the Newburyport/Rockport line was counted before the end of October to avoid the increase in ridership associated with Haunted Happenings in Salem. Finally, all counting was complete before November 18th, when the winter schedule began.
There were some instances where a train had to be recounted. Recounts often happened due to count staffing issues; for example, if one or more workers called in sick before a count, or if a train had more train crew than they planned for, allowing for more doors to be open than initially expected. Other times, recounts had to do with service: if a train was half an hour behind schedule, the train was recounted another day to capture ridership that may have not taken the trip or used another mode due to the delay, which ended up happening a few times in mid-October due to slippery rail conditions.
These counts are subject to human error, especially in busier stations during peak times, but again, redundancy was built into the counting schedule to help mitigate this issue. Also, since the majority of trains that didn’t need recounts were only counted once, those trains have a sample size of one. In cases where only one or two stops on a train needed to be recounted, the opportunity was usually taken to recount all the stops on that train for extra data. In those cases, the calculated boardings and alightings are the average of all the times a stop on the train was counted.
Reduced midday service on two Regional Rail lines (Haverhill and Worcester) throughout Fall 2024 may have impacted ridership numbers. On the Haverhill line, westbound service to Reading was notably reduced due to a lengthy midday single-tracking capital investment diversion in Melrose, which meant only 27 out of 41 regular trips to Reading operated in Fall 2024. Additionally, track and station work along the Worcester line extended the journey time of midday trips, which may have deterred some riders.
After all the data was processed, MassINC noticed that 12 trains had data quality issues – things like a negative load count at some point on the train, or a count at a specific station that seemed abnormally higher or lower than expected. Those trains were recounted on December 17-19, January 7-9, and January 14-16.
Analysis
According to our new counts data, there were 97,537 trips taken in Fall 2024, which is a 23% decrease from the Spring 2018 counts data. How does this decrease vary throughout the entire system? To understand how station-and line-level ridership across the system has changed since 2018, we looked at the highest ridership stations and lines, the stations and lines where ridership has changed the most since the Spring 2018 counts, and the change in ridership during off-peak service.
Part 1: Top Ridership Stations/Lines in 2024
Eight of the ten highest ridership stations (defined as stations with the most boardings) in Fall 2024 were also in the top ten in 2018. In fact, the top six stations of 2024 were exactly the same (and in the same order) as in 2018. See Table 1. The top three stations - South Station, North Station, and Back Bay - are the primary stations where trips originate or terminate. The 2024 data indicates that two other Boston stations, Boston Landing and Lansdowne, have grown as significant hubs. The increase at Boston Landing is partly explained by the fact that the station only opened in May 2017, so ridership was still growing from the initial opening when the 2018 counts were carried out.
Table 1
Looking at the data on a line-by-line level (Table 2), the Providence and Worcester lines are still the two lines with the highest weekday ridership. The biggest shifts in the ranking are Haverhill going down four spots and Middleborough moving up three. The Haverhill line shift may in part be attributed to the construction schedule and diminished service mentioned earlier. However, perhaps the most interesting shift is the dramatic growth of the Fairmount Line ridership, which has 66% more riders compared to 2018. This line is entirely contained in the City of Boston, and in 2024 a new half-hourly service was introduced.
Table 2
Part 2: Spring 2018 vs. Fall 2024
There were 28 stations that had more weekday riders in Fall 2024 than in Spring 2018. See Table 3. The greatest increase was at hub stations that provide transfer points to subway stations (possibly revealing a mode shift due to the speed restrictions that hampered the Red Line and Orange Line in 2023 and 2024, shown in orange in Table 3). Additionally, five of the ten stations are on the Fairmount Line (shown in blue in Table 3), partially as a reflection of the recently increased service on that line. Lastly, Bradford Station increased quite a bit, but that can be attributed to the temporary closure of Haverhill station during the counts.
Table 3
If we remove these outlier stations (as well as stations with ridership under 100), what does our top ten list look like? Even when removing the Fairmount Line from this analysis, Readville still makes an appearance – it has high ridership on the other lines that serve that station as well. Otherwise, these stations are mostly in suburbs that are close to Boston. See Table 4.
Table 4
The stations with the highest ridership decreases (Table 5) are primarily a combination of suburbs that are both close (Melrose Highlands) and far away from Boston (North Scituate, South Acton). One exception is Porter Square, which did not experience the same bump as other combination Red Line/Regional Rail stations. This can probably be attributed to the recent opening of the Green Line extension, which provides commuters near Porter Square another direct route to North Station.
A factor that may have influenced ridership at the Haverhill Line stations (Reading, Melrose Highlands, and Wakefield) is the previously mentioned single-tracking diversion in Fall 2024. Additionally, Walpole’s ridership likely decreased due to the post-Spring 2018 introduction of the Foxboro Line. Not only does Foxboro provide a convenient alternative for some previous Walpole riders, but the level of service at Walpole also slightly decreased to help serve the Foxboro Line. Finally, Winchester Center’s decrease is likely due to the station being closed for over three years while significant improvements were made to station infrastructure, only reopening on October 1, 2024 (riders were directed to Wedgemere Station during the closure, which could be a contributing factor to its appearance on the top ten list).
Table 5
It may be most helpful to visualize these station-level changes with a map. Stations are categorized in three ways: stations with increased ridership compared to 2018, stations where ridership has decreased less than the systemwide average of 23%, and stations that have decreased more than the systemwide average. This snapshot indicates that weekday ridership has seen the greatest recovery, in some cases even surpassing 2018 ridership, in stations that serve the immediate metro Boston area.
In Keolis’ new all-day service model (discussed more in the below Peak vs. Off-Peak section), these stations are the most likely to have a greater frequency of service, in some cases even half hourly, like from Canton Junction to Boston. The Haverhill Line is a notable exception, where service frequency in the inner core is maxed out at every 45 minutes due to infrastructure constraints. This is also where we can see the greatest impact of the construction schedule on the Lowell Line during the count period, where midday service to these inner core stations was reduced to every two hours.
Line-Level Ridership Changes
We already know that several stations on the Fairmount Line saw especially dramatic weekday ridership increases compared to Spring 2018, and that line-level ridership increased by 66%. While none of the other 11 lines have recovered to their 2018 weekday ridership levels, five of them performed better than the systemwide decrease of 23%. See Table 6, lines that increased are in blue, the lines that outperformed the system averages are in green, and the lines that underperformed are in red.
Table 6
Visualizing the lines and categorizing them the same way we did in the stations map, we can see that there has been significant weekday ridership growth on the south side of the network.
Peak vs. Off-Peak
In April 2021, Regional Rail service changed to an “all day” model where trains are offered with significant frequency throughout the day (as well as significantly increased service on the weekend). This was a dramatic shift from the operating model in 2018, where service was largely geared toward traditional office workers and weekend service was infrequent. Weekday ridership can be divided into three categories: AM Peak (morning rush hour trains going toward Boston), PM Peak (afternoon rush hour trains traveling from Boston), and Off-Peak (all other trains). Analyzing ridership during these times shows us how Regional Rail riders are using the service during the week (Table 7).
Table 7
Off-peak ridership has increased by 35% since Spring 2018, accounting for over 9,000 new trips on a composite weekday. Probably the most interesting finding here is that the peak on weekday Regional Rail ridership has flattened significantly as a result of this change in service. Ridership levels are almost identical during the three different peak categories in 2024, whereas in 2018 ridership was heavily skewed toward AM Peak and PM Peak. These two peak categories decreased by the around the same percentage.
We can see how off-peak ridership has changed per hour of service in the below graph. Throughout the day, we see off-peak ridership in 2024 is overall greater than it was in 2018. This is partially due to some changing definitions of peak – for example, in 2018, AM Peak was defined as the span from the start of service until 9am, but now starts a bit later – but can also be attributed to increased service outside of typical rush hour commuting times.
Conclusion
That concludes our initial dive into the new Regional Rail counts data. Again, we encourage interested parties to download the data (which, again, is available on the Open Data Portal). Stay tuned to the OPMI Data Blog for more insights into the counts data through the course of the year!
