Analysis of Recent Public Transit Ridership Trends (2020)

Below is the uncorrected machine-read text of this chapter, intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text of each book. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.

88 A P P E N D I X B Data Limitations Throughout our analysis, there were several significant potential ridership factors we were interested in studying but were unable to draw conclusions about due to lack of quality data. These factors appeared either in the literature as factors significantly correlated with transit ridership or in articles surrounding pilot projects with promising initial results. Despite this, they were either inconsistently measured and reported geographically or between years, or they were not measured at all. A lack of data on these factors prevents researchers from performing rigorous studies on their effects and may potentially hurt viable means of maintaining and growing ridership. Dedicated Right-of-Way It is generally accepted knowledge in the transit industry that dedicated right-of-way (ROW) modes, such as heavy rail, are seen as more reliable and faster than mixed traffic modes such as streetcar and bus. Separating vehicles from general travel lanes allows them to travel faster and more consistently than those that sit in traffic. This, in turn, results in higher ridership per route mile on these modes. Cities that have implemented dedicated streetcar and bus lane pilots, such as Boston and Toronto (identified in Task 3) experienced higher ridership along these routes. As part of our study of strategies to combat ridership declines, we were interested in studying on a nationwide scale the effects of dedicated ROW on ridership. This type of study may have allowed us to see correlation between ROW for particular modes and ridership trends. However, we were unable to complete this type of analysis due to a simple lack of reliable data. While metrics involving transit way mileage are available for each year in the National Transit Database (NTD), our analysis has shown them to be unreliable. The first issue involves a change in the way NTD classified transit way mileage for non-rail modes. In 2012, non-rail “exclusive” and non-rail “controlled” ROW were reported for each mode. In 2016, categories were changed to “exclusive fixed guideway bus lane miles,” “exclusive high intensity bus lane miles,” and “controlled access high intensity bus lane miles.” This addition of “fixed guideway” mileage led many transit agencies to include numbers unrelated to exclusive ROW. For example, each operator of trolleybus service reported their mileage of trolley wire regardless of ROW characteristics. Some transit agencies reported fixed guideway mileage for traditional bus modes rather than including these lane miles in “high intensity bus lane” mileage.

Data Limitations 89 Another issue with transit way mileage data involves its general accuracy. Despite frequent campaigns to introduce bus lanes and segregate transit vehicles from other traffic, NTD data shows about as many transit agencies decreasing their dedicated ROW as increasing it. As shown in below, data from the NTD shows a relatively even spread of regional growth and decline in dedicated ROW mileage for otherwise mixed ROW modes, such as bus and streetcar. The declines in dedicated ROW are more likely due to inconsistencies in reporting rather than actual guideway miles being repurposed for mixed traffic. The inconsistencies in reporting dedicated ROW mileage disqualify it as a metric for analysis, despite the interesting and potentially useful conclusions that may come from such an analysis. Future studies on implementation of dedicated ROW on ridership and service efficiency are recommended, perhaps through data gathering from a large group of transit agencies themselves. Figure B-1 Figure B-1 . Percentage change in unlinked passenger trips vs. percentage change in dedicated ROW miles between 2012 and 2016. Reduced Reporters While NTD reduced reporters are generally smaller transit agencies operating 30 vehicles or less, they have a different set of reporting requirements and apparent data standards than full reporters. While they are technically required to report unlinked passenger trips, vehicle revenue miles, and fare revenue, many of the holes in data we discovered were due to reduced reporters’ lack of data for portions of our study period. While reporting requirements appeared not to change between 2012 and 2016, these missing data led us to remove several transit agencies from the analysis. While it is important to note that many of these transit agencies lack resources to gather and analyze their data, this analysis took into account these smaller transit agencies which are often ignored in nationwide transit studies. Having reliable data available is the best way to guarantee a thorough analysis of these trends.

90 Analysis of Recent Public Transit Ridership Trends Mode Change A byproduct of the multitude of transit modes currently in service across the U.S. is their often complex categorization. Modes like BRT and Streetcar blur the lines of what bus and rail services mean, and modes like hybrid rail may not be innately understood by all transit agencies. Even within a mode like streetcar, modern versions may act more like light rail than historic ones, which may affect their ridership and service characteristics. When transit agencies are given forms to report, many seemed to have reacted slowly to the introduction of new modes. For example, BRT systems in Cleveland and Boston were fully operational by 2012, yet they reported these statistics as motor bus (MB) that year. In 2016, this data was correctly assigned to BRT. This misrepresentation of mode statistics makes parsing historical data by mode unreliable, as service that acts like BRT may not be comparable at all to service that acts like MB. A separate issue, related to the dedicated ROW issues above, is that several transit agencies have routes that behave like different modes along the course of their route. Examples include the MBTA Green Line, which generally operates as light rail but with several segments running in mixed traffic as a streetcar, and the MBTA Silver Line, which transitions from a dedicated busway to street running mid-route. These transitions must be handled in a logical way in data collection, either by correctly denoting ROW mileage or assigning new modal categories based on combinations of other modes. This level of data would help future researchers sort through modal types to better identify ridership patterns. Service Area In our effort to compare transit service and passengers across hundreds of regions in the United States, the issue of regional scale became vital. Restricting comparison to municipal limits rarely makes sense as transit agencies themselves generally are not constrained to particular cities. Urbanized area (UZA), the geographic measure used by the NTD, is largely not available from sources like the U.S. Census Bureau more fine-grain than every ten years. Urbanized areas have complex geographic boundaries that extend to the far reaches of a region, often reaching into nearby cities otherwise unaffiliated with a particular region. Because of their concentration on higher-density areas, UZAs tend to skew regional density high. In contrast, Core-Based Statistical Areas (CBSAs) used in our analysis tend to include entire counties for the sake of simplicity. This allows for much more frequent data availability, but often includes hundreds of square miles of undeveloped land and skews density down for most regions. To compare transit service across regions of various size, we had to settle on CBSAs for data availability. This was not ideal, but was deemed necessary to complete the regional analysis. However, transit agencies do report “service areas” to the NTD, technically required to conform to a geographic buffer surrounding the routes serviced by the transit agency. This metric would be ideal for a transit service analysis and for comparing densities of regions that actually

Data Limitations 91 have operational transit service. However, the self-reported nature of NTD’s service area left the data particularly unusable. Some transit agencies appear to simply report the square mileage of the counties they operated in without regard to service at all. Other transit agencies restricted their service area differently by mode. Both of these misrepresent what should constitute a transit agency’s service area, though no methodology would be perfect. For example, commuters who drive in from outlying counties to park-and-rides generally are missed in a service area calculation. However, future studies would benefit from a specified methodology for determining a transit agency’s true service area. Tract-Level Data and CBSA Changes In our analysis, we relied on one-year data from the American Community Survey in order to accurately measure year-to-year variation in population and zero-vehicle households. Unfortunately, this left us unable to perform an analysis on any scale smaller than the CBSAs, as Census data on the tract and urbanized area levels are only available from the decennial census or as ACS 5-year estimates, which are not usable for comparisons of point-in-time data. Tract-level data would allow for remarkably fine-tuned analyses of trends not only related to transit, but of population and demographics in general. Metropolitan areas contain immense variation between their tracts, and the ability to track changes between non-decennial years would have vast impacts on the research world. Realistically, however, reliable year-to-year data would require a significantly scaled up effort by the U.S. Census, particularly for all 74,000 census tracts at the one-year level. Additionally, CBSAs underwent a change in 2013 where many metropolitan and micropolitan areas gained or lost counties. This caused some issues with reconciling the demographic statistics between the years of 2012 and 2016. Documentation on the changes and how they affected population statistics was largely nonexistent. Piecing together data that was available, we were able to reconstruct some CBSAs to perform an accurate comparison of their 2012 and 2016 statistics. However, many CBSAs also had to be thrown out as their geographies could not be matched. Better documentation on these changes and how demographic statistics shifted would help researchers more accurately and thoroughly compare one-year data from before and after the change. Conclusion Despite our confidence in our data and results, there were several data challenges that prevented this analysis from going further. Lack of nationwide reporting standards for certain metrics in the NTD restricted our analysis and many others to basic reportable metrics. Geographic data limitations caused issues with data reconciliation and prevented a thoroughly nationwide study. Transit agencies must be able to accurately collect and analyze the data they can about their service and passengers, particularly at a time when transit ridership is declining. Quality data can help transit agencies and researchers alike to find the best answers to the many questions asked of them.