Home

Behind the scenes with MBTA data.

When we last wrote about ridership on the Data Blog, we were still in the midst of closure here in Massachusetts. Since then, the Commonwealth has proceeded with multiple phases of its reopening plan. The MBTA has played its part by increasing service (especially where demand continues to be relatively high) and requiring masks as we cautiously reopen various sectors of the state.

While we still recommend people stay home when possible, we have seen the cautious reopening reflected in the MBTA’s ridership, as it continues to rise slowly but steadily each week. This post will provide an update on ridership overall, with a particular focus on the work we are doing on the data and technology side to better capture bus ridership.

We are keeping downloadable datasets for public use in this folder. These will be updated with the most recent data as often as we are able to. These datasets should be considered preliminary and subject to further adjustment, but they have been checked for major errors. Once datasets are finalized, we will add them to the Open Data Portal.

Datasets currently available for download in our public folder:

  • Gated Stations Validations by Station (updated each weekday) (1/1/2018 – present)
  • Gated Stations Validations by Line (updated each weekday) (1/1/2018-present)
    • These are the same datasets we have been sharing on the data blog since the pandemic started. They include the total validations at each gated station, aggregated by either station or line (Red, Blue, Green or Orange), by day, going back to January 2018. These data are also available in a published version on the Open Data Portal. The most recent data is subject to revision, but usually is received completely the next day.
  • Weekly Bus Ridership by Route (updated weekly)
    • This file includes the average weekday ridership (from the APCs), by week, for each MBTA bus route going back to the beginning of 2019. The first column is the route number, and each subsequent column is labeled with the date of the Monday that started that week’s data. For example, the column labeled “20-Jul-20” contains the average ridership for the weekdays in the week starting July 20th, 2020 (so 7/20-7/24).
  • Ridership by Route and Stop for TransitCenter (Static dataset)
    • This file contains the ridership by route and stop from early in the pandemic, with a comparable period from 2019. This is the data that we shared with the TransitCenter in their post linked below, so we are sharing it here as well. This dataset does not include added service during the COVID emergency (routes where we had the highest demand and crowding) and added additional trips (“run as directed” trips) will be undercounted.

How calculating bus ridership from APCs works

In order to track ridership on buses, we are using the Automated Passenger Counters (APCs) that are installed on 70% of the MBTA bus fleet. These record the boardings, alightings and load at each stop along a route. Since we do not have APCs on every bus, we scale the data we get up to the scheduled levels of service, and then scale ridership back down to account for scheduled service that did not run. You can see the outputs from this process on the Open Data Portal.

Since the pandemic, we have had to accelerate this process to generate ridership daily for internal use. Generally, this works well, but there are a few additional challenges that we have to account for:

  • A portion of bus trips are not included in the schedule and are run at the bus supervisors’ discretion to try to alleviate crowding. These are known as Run as Directed trips, or RADs. Since RADs provide more flexibility to respond to changing circumstances, the amount of bus trips that are provided by RADs has greatly increased under the MBTA’s COVID response as we prioritized high-ridership corridors and those that serve essential (including health care) workers. Because these trips are not included in the schedule, we depend on the operator entering a code in order to determine which route the trip ran. Usually they enter the right route, but since this is a manual entry, some trips are not assigned correctly. In these cases, the RADs will count towards our overall ridership, but not the route-level ridership. The new schedule that started on June 21 added more trips to routes that were previously more likely to have RADs assigned, so this problem was lessened, but not eliminated.
  • Before June 21, when the MBTA expanded service, two bus garages that have the lowest levels of APC coverage (Fellsway and Albany) were not used, which made the APC coverage in the remaining fleet higher. Since these garages reopened, our APC coverage has dropped. While this does not especially affect the accuracy of the totals, ridership at the route-level on some routes with lower levels of APC coverage is less accurate.

 How is Ridership Returning?

Ridership is slowly returning to the MBTA as the state proceeds with its cautious reopening. Interestingly, the trend has been slow and steady with very few big jumps or drops, even on days where various phases of the state’s reopening plan went into effect. The following charts show the daily weekday total gated station validations, and bus ridership from March 23 to July 17:

The following charts show the change on each day when compared to the rolling average 5 weekdays before, first for gated stations, then for buses, in order to show the rate of change as we regain ridership:

While there are day-to-day dips, each day since late April has been generally between 5-15% higher than the previous week. Gated stations have increased at a higher rate, but also dropped more to begin with, so they had more to gain.

What's Next

The pandemic has affected travel and ridership in different ways throughout the region. We at the T are looking closely at these changes and analyzing them as we plan service delivery for the Fall and beyond. While we all wish the circumstances were different, the pandemic does provide a natural experiment which can help us learn how passengers are using transit, and inform what service we should prioritize and improve, and where, for essential workers both during the pandemic and in the future. As one important example, we are leveraging our existing partnerships with cities and MassDOT to accelerate work on transit priority for what we already knew were the most important corridors for our passengers.

Our friends at the Transit Center took a look into the data (which we’ve shared again at the beginning of this post) and have written a post about it here. Like other analyses have shown, routes with higher levels of low-income and minority passengers, and those with low vehicle availability, tended to lose less ridership during the pandemic. 

Since the TransitCenter conducted the above analysis, ridership has returned to more routes, and the MBTA has restored service more on certain routes as well to try to meet demand. With the additional data, we at the Data Blog plan to take a detailed look at what factors influenced ridership changes during the drop in ridership and during its recovery. This and other research will help inform future service and other decisions.

Here at the blog, we will keep the datasets at the top of the post updated as we continue to get more data. We are also working on additional research as we try to help the MBTA and transit systems around the country make data-driven decisions to best serve passengers. If you found the above datasets or others useful in your own analysis, feel free to drop us a line at This email address is being protected from spambots. You need JavaScript enabled to view it..

Every month, we send out a survey to MBTA riders (you can sign up here!) and use the results to produce the Customer Satisfaction metrics that we publish on our dashboard. The four metrics (overall satisfaction, satisfaction with a rider’s most recent trip, satisfaction with the reliability of the MBTA, and satisfaction with the MBTA’s communication) are normally reported for respondents who had taken a trip on the MBTA within the past week. We exclude respondents who had not taken a trip within a week of completing the survey. 

This methodology allows us to ensure that survey responses are current and relevant and that the panel is representative of usual frequent riders; we have used other methodologies like in-person Intercept Surveys to target infrequent riders and visitors. Due to the current COVID-19 pandemic, however, many of our usual frequent riders have stopped taking the MBTA. As a result, until service and ridership have begun to return to normal, we will be reporting customer satisfaction in the following ways:

  • Satisfaction with a rider’s most recent trip will be reported only for respondents who have taken a trip within the past week
  • Overall satisfaction, satisfaction with reliability, and satisfaction with communication will be reported for all respondents

For trip ratings, we wanted to make sure that the metric reflected trips that were recent enough for the respondent to remember their trip well. In addition, as social distancing measures continue, many respondents will not have taken the MBTA for over a month, and their most recent trip will continue to be one from early March. For example, their April and March trip scores will be reflecting service from the same actual experience, making it impossible to measure changes in this metric on a month-to-month basis for these respondents who are not currently riding the MBTA.

However, we will report the other metrics for all respondents to the survey so that results for these questions can be comparable over time. The riders who are continuing to use our service are not a representative subset of all of our usual riders, and they do not respond to surveys in the same way. When possible, we want to continue to report customer satisfaction results from a group of respondents whose makeup is relatively stable over time to better track trends in customer satisfaction. Additionally, we understand that residents of the area still have opinions about the MBTA, how it is communicating, and the service it is providing, whether or not they have recently taking a trip on public transit. We want to continue to hear from all of our riders, regardless of if and how they are currently traveling, and our customer satisfaction reporting mirrors this interest.

In the past few weeks we have kept you updated about MBTA ridership during the COVID-19 pandemic on this blog, and we plan to keep updating the blog regularly as long as there is something to discuss. You can read previous posts here, here and here

If you’re just looking for the total ridership by line and by station, you can download that data here. This link will take you to a Box folder where we update each day with the validations grouped by line, or by station. (Last updated 7/28):

Download data here

 

We would like to reiterate that Governor Baker has issued a stay-at-home order, non-essential business are closed, and we would implore our riders to respect this and stay at home whenever possible. In the Governor’s words: “Just because the T is open does not mean we think it’s a good idea to take the train downtown to meet up with friends. 

By limiting the use of public transportation to essential activities, we will not only slow the spread of the virus – but we will better protect our health care workers, grocery store workers and others who are working every day to keep us safe. 

Everyone is advised to stay at home and limit all unnecessary activities.”

Please see the continuously-updated page here for the latest updates on how the MBTA is responding to the pandemic: www.mbta.com/covid-19

Around the Country

Other transit agencies have also had their ridership greatly affected by the pandemic. While not all agencies are sharing data publicly and frequently, we did a quick look to find other data points. If you have seen reporting from another agency (or you work for an agency) feel free to reach out at This email address is being protected from spambots. You need JavaScript enabled to view it..

BART in San Francisco is seeing a ridership drop of 94% and WMATA in D.C. has had rail ridership and bus ridership drops as much as 96% and 82%, respectively.

The CTA in Chicago has kept normal service on five of its eight rail lines even with an 82% decrease in system ridership, while New York’s MTA is running on its “Essential Service Plan,” which preserves most peak service and eliminates some lines on weekdays after experiencing an 87% rail ridership drop and a 60% drop for bus.

Translink in Vancouver, BC, especially, has seen a sharp decrease in bus ridership with an 82% drop.

The Data

Our usual ridership reporting includes factors to account for passengers who we do not observe through our automated equipment, and we usually wait at least a few weeks before reporting anything due to the delay in transferring data from our vehicles. We also conduct in-person counts to verify automated data and improve accuracy for our end of year reporting. 

We did not have time to do our normal analysis and reporting, so we had to focus on the best sources we have that were reasonably representative of the system. So we focused on two data sources for this post: Counts of validations at gated stations from the fare collection system and bus ridership estimates from automated passenger counters. Because these sources usually have extra processing and QA/QC done as noted in the previous paragraph, all ridership estimates in this post should be considered very preliminary and subject to change.

Rapid Transit

To examine ridership on the rapid transit system, we used validations (taps or ticket insertions) at the 64 gated stations in the MBTA system. This data came from the fare collection system and is not adjusted to account for passengers who enter the gates without interacting with the equipment (this can be children, fare evaders, or people who enter when the gates aren’t functioning). 

We set up a special data transfer to gather the total validations by day, and then grouped them by station and by line. For stations where passengers can board multiple lines, we use a rough “split” factor to assign riders to each line (For example, at Park St, we estimate that 54% of people entering the gates are then going to board the Red Line, and 46% go on to board the Green Line).

Ridership since 3/24:

Ridership has continued to slide steadily down since the governor’s order to close non-essential businesses, and now is under 40,000 validations per day. We are seeing similar distribution between lines and stations as we did at the beginning stages of the pandemic.

On weekends, ridership has also dropped, but not to the same extent as weekdays when compared to “normal” times. Saturday validations on April 11 were 64% of the previous week’s weekday average, and Sunday validations were 47%. This compares to Saturdays usually being about 51% of weekday taps, and Sundays having about 38% of the weekday number (using calendar year 2019 data). 

Bus

Overall bus ridership shows similar trends over time as gated stations, though the drop since 3/24 is less steep. We are seeing ridership on buses drop by about 5% per week since the Governor’s order. Please note that these data have fairly high margins of error, so looking at overall trends is more useful. Some of the challenges in the data are accounting for buses with lower samples of APCs, and accounting for additional trips that were added which were not part of the schedule programmed into the software on-board buses (“run-as-directed” trips). 

Ridership last week was around 80-85,000 passenger trips per day. 

Here are the top 20 routes by ridership and their change from our baseline week. The geographic trend in ridership change is similar to previous weeks. Route 116 (asterisked) has significantly less certainty than other routes due to low APC coverage.

Ridership by Time of Day

Last time, we took a look at how ridership on gated stations had changed from our usual weekday patterns. In the following chart, you can see the difference in trips on our buses between the baseline week and Friday, April 10:

Note that the above chart likely undercounts ridership where we have added additional RAD (“run as directed”) trips on routes and times with high demand. But, you can see the new pattern of ridership pretty clearly. On buses, we see a single peak around 3 PM, and steady ridership throughout the day with the peak just slightly higher than the rest of the day.

We will continue to update you on MBTA ridership at this blog as things progress. If you’re looking for something to do while you stay at home, why not take a look at some of the datasets on our Open Data Portal?