Two Decades of Change in the Development and Dissemination of Transportation Statistics
Congress required that the Transportation Statistics Annual Report include an assessment of the quality of statistics used in the report and a summary of efforts by BTS to improve transportation information. BTS published the first Transportation Statistics Annual Report one year after the Bureau started operations. The inaugural edition featured an extensive array of statistics from many sources, described how BTS would develop data through surveys and estimation models to fill information gaps, and highlighted measurement issues.
During its first half decade, BTS launched the Commodity Flow Survey to measure freight flows, the American Travel Survey to measure long-distance travel, and the Transborder Freight Data Program to identify the portion of international trade carried by each mode of surface transportation. BTS inherited carrier reporting systems that originated in the Civil Aeronautics Board and the Interstate Commerce Commission to measure financial and operating statistics for commercial airlines and major trucking companies.
BTS developed estimates of the freight not captured by the Commodity Flow Survey and an estimation model for a more complete enumeration of spending on and by transportation as part of the Nation’s economy, called a shadow account in 1994 and now known as the Transportation Satellite Account. BTS also sponsored an experimental program to extract volume and speed data from traffic management centers in four cities to estimate minute-by-minute variations in travel and congestion.
Sources of data have evolved since those early days. Household surveys are increasingly expensive to conduct and suffer from declining response rates. BTS and its partners turn increasingly to administrative records and sensor-based data to replace or supplement surveys. For example, the Bureau of Economic Analysis is using tabulations of credit card purchases to replace surveys for data on expenditures during foreign travel, and BTS plans to use a similar approach to estimate long-distance travel.
BTS, the Federal Highway Administration, and partners outside the U.S. Department of Transportation are examining the use of onboard sensors and administrative records from vehicle registration systems to supplement paper surveys as part of plans to restore the Vehicle Inventory and Use Survey, a vital data source that was discontinued in 2002 by the Census Bureau due to budget constraints. This combination of traditional surveys, administrative records, and sensor data follows a similar hybrid measurement strategy used by Transport Canada to collect data for its Canadian Vehicle Use Study.
While individuals are reluctant to fill out questionnaires, extensive location and travel information is gathered by private companies through cell phone tracking software, tweets, and other social media. Vehicle locations are captured by license plate readers used for toll collection and parking enforcement—a largely untapped reservoir of information on travel. Availability of this reservoir may be limited by public reactions to the privacy implications of license plate tracking for security purposes.
The newest source of travel data may come from unmanned aerial vehicles, also known as drones, which can fly for extended hours at low altitudes and collect images that can be turned into traffic counts for everything from trucks to bicycles. As in the other new sources of data, accuracy and statistical validity of the measurements as well as public acceptability remain unexplored.
Estimation models have evolved since 1994 almost as much as the data sources. BTS made aggregate national estimates of total freight not covered by the Commodity Flow Survey in the 1990s. Now BTS and the Federal Highway Administration together make detailed estimates of value, tons, and ton-miles by commodity, region of origin, and region of destination with the Freight Analysis Framework (FAF). The FAF integrates data from the Commodity Flow Survey, the Transborder Freight Data Program, and other sources, filling data gaps with models of likely freight flows in order to provide detailed estimates of total freight flows and to update those estimates for years between Commodity Flow Surveys.
BTS uses other estimation techniques to create useful indicators, such as the Transportation Services Index, and to improve analysis. The Transportation Services Index is the broadest monthly measure of U.S. domestic transportation services, encompassing five modes of freight transportation and three modes of for-hire passenger transportation. BTS is adjusting selected statistics, such as commercial aviation load factors and passenger enplanements, to remove recurring seasonal variations and facilitate comparisons of underlying changes between months.
Big data analytics represents a new and entirely different challenge for estimation models. Methods have been developed to extract information from unstructured data sets that are continually updated. Big data analytics were originally developed to analyze markets, social and political trends, and the performance of professional athletes.
These methods are being adapted by private shippers to monitor and manage supply chains and are now being explored by public agencies as early indicators of changing social and economic conditions. The potential for revolutionizing transportation analysis is great, but research is needed to determine the reliability and validity of statistics from these methods and to establish access to large databases that are primarily in the private sector. BTS and its partners in the statistical and transportation communities are pursing these research questions as resources permit.
Open data access is another new aspect of the statistical world that has come into play since the 1994 Transportation Statistics Annual Report. The 1994 report recommended “democratizing data” to improve access by all potential users to statistical products. Executive departments are now required to provide public access to the data on which statistical products and research reports are built as long as the data are not restricted by confidentiality or licensing limitations. DOT’s open data initiatives are now managed by the National Transportation Library, which BTS established within 3 years of the initial 1994 report.
As part of its “democratizing data” theme in 1994, BTS decided to disseminate its new statistical products for free even though most statistical agencies charged for data. Today, almost all products of Federal statistical agencies are available at no charge.
The technology of data dissemination is perhaps the most changed aspect of the Bureau’s business since 1994. BTS published the initial releases of Transborder Freight Data on 3.5-inch floppy discs. BTS pioneered the dissemination of transportation data on CD-ROMs, starting with Transportation Data Sampler Number 1. Graphical user interfaces for Internet browsers became available shortly after BTS started operations, and BTS now distributes all products over the web. An early project to share information for metropolitan planning organizations on CD-ROM evolved into the National Transportation Library. None of this was foreseen in the first Transportation Statistics Annual Report.
BTS approached the Federal Aviation Administration early in the Bureau’s history about capturing DOT’s biggest data set, tracks of individual flights through the air traffic control system, for possible analysis of the capacity and performance of the commercial aviation system. BTS abandoned the effort because only very expensive technology of the 1990s could handle the daily volume of these data. Today, individuals can track flights on privately run websites that pull real-time data from the air traffic control system, and many government and private organizations mine that data for capacity and safety analysis.
While BTS did not pursue the air traffic data, the Bureau’s experimental program to extract volume and speed data from traffic management centers has evolved into the National Performance Measure Research Data Set managed by the Federal Highway Administration. The geographic locations of vehicles over time are combined with traffic management center data to measure the speed and reliability of the National Highway System.
Current discussions beyond big data include a strong emphasis on performance, often treating performance measurement as a new topic. Data for Decisions, the document that guided the creation of BTS and its early program, included an entire chapter on creating a “National Transportation Performance Monitoring System” [TRB 1998]. Indeed, discussions of transportation performance measures precede the U.S. Civil War (see box A).
Whether for performance measurement or other policy concerns, most measures requested by public officials are variations on those that BTS was mandated to collect or compile in the 1990s. Basic measures of people movement, freight flows, and vehicle activity remain as priority topics. Some of the energy and environmental measures are now clustered under the concept of sustainability, and measures formerly categorized as quality of life are now labeled as livability. Speed and reliability of freight movement is now characterized as fluidity.