Reference Reviews

Péter's Digital Reference Shelf

July 2007

Title: TranStats
Publisher: Bureau of Transportation Statistics
URL: http://www.transtats.bts.gov/
Cost: free
Tested: July 17-22, 2007

The Context

The adjective good rarely appears along with the term “government services." But digital government information services are an exception. There are several very good databases by government agencies/institutions beyond the most well-known PubMed database family. Unfortunately, they are not well-known enough even if some provide more information through a better software than their commercial implementation on subscription-based services.

One of the best examples for such cases is the Transportation Research Information Services (TRIS) of the Transportation Research Board of the National Academies and the National Transportation Library, part of the Bureau of Transportation Statistics (BTS). I wrote about it last August in this column. Let me summarize here that the open-access version of TRIS is more current than any of its commercial versions, and beyond the indexing/abstracting records it has more than 20,000 full-text documents that none of the commercial versions have.

Records for journal articles and conference papers are very important, availability of the full text of primary sources (even for a subset of the database) is even more so, but often researchers or curious users may just need the facts for an argument. TranStats is the perfect resource for that as long as only U.S. data are needed. Some of TranStats' data are available in other databases, but not with the comprehensiveness and retrospectivity provided by BTS. In turn, BTS provides excellent links to sources that can complement TranStats, such as the Consumer Price Index data series (compiled by the Bureau of Labor Statistics) which includes information —among a zillion other goods and services— about the airline fares as part of the basket used in calculating this important index. TranStats is just one of the many programs in the rich data repertoire of the Bureau of Transportation Statistics.

The Content

TranStats' subtitle is Intermodal Transportation Database. I did not know the meaning of intermodal, but the American Heritage Dictionary helped me out with this definition: “Relating to transportation by more than one means of conveyance, as by truck and rail.” And indeed, this über-database covers statistics for every conceivable mode, vehicle and form of transportation from the expected air, highway, rail and maritime transportation to the ancient, help-yourself mode of walking, as in cross-border walking.

Don't play down the significance of international pedestrian traffic: in 2006 alone, there were nearly 26 million such transits. True, this figure also includes bicyclists and mopedists, but they get off their vehicles to go through customs/immigration checks. This also implies that the data does not include the undocumented border crossers. Interestingly, the 1996 data indicates that there is a substantial decline from the then-34.7 million pedestrian transits. Maybe there are more convenient ways with no data collection.

According to the About page, TranStats provides a “searchable index of over 100 transportation-related databases across every mode of transportation —with many social and demographics data sets that are commonly used in transportation analysis." These are not limited to the statistics compiled by the Department of Transportation. They come from all branches of the government — as they should. The weekly retail gasoline price index, for example, is compiled by the Department of Energy but obviously, has the greatest impact on transportation, which represents 67% of the entire petroleum consumption of the country.

The home page smartly provides a good mix that allows users to get a feel at first glance about the type of content as well as the appealing visual representation of the content. (Beyond the appeal, I must mention that the scale values on some graphs don't reveal the trend well, as I will discuss later.)

Scope

What you see on the home page is just the tip of the iceberg of the variety of statistics. You can get a better sense of the scope of TranStats by scrolling down the list of recent releases, or peeking into the variety of statistics for freight transportation, or the type of tables that make up the National Transportation Statistics report. This is not your father's old almanac or compendium, but an attractive source of transportation-related information.

Statistics by modes of transportation

Air transportation gets the emphasis on the home page and also in the data sets, and it is understandable. Beyond its extreme importance in the U.S. economy, the vagaries of air transportation get the most attention in the press and on television, but not yet in Congress. However, this may change now as more and more airlines introduce the bonus service of tarmacking passengers for hours after leaving the gate. This relatively new term, or new sense of an old term which —not surprisingly— already appears in the excellent American Heritage Dictionary, is not used in TranStats, but data are available to calculate it since the time a plane leaves the gate and the wheels-off time both are available in its datasets. With the ongoing debate about regulating when carriers must release their hostages (aka emplaned passengers), this data should get more prominence in TranStats to compensate for the efforts of the lobbyists to play down the significance of this practice.

Of course, air transportation cannot compete in absolute numbers with highway transportation, but the latter does not offer as many criteria to measure and analyze by the government as commercial air transportation. Ultimately, it is the growth of this mode of transportation that makes it top priority for statistical analysis.

It had the highest rate of increase from 1994 in vehicle miles and in passenger miles and it matched the rate of increase in freight transportation by rail. I am not a statistician, let alone a transport statistician, I just easily found such important indicators in TranStats and in other DOT publications, as in the TSAR Transportation Statistics Annual Review.

I am personally very interested in (you may say, obsessed with) airline data. If you look at the Air Travel Price Index (ATPI) which shows the extreme increase of ATPI to and from Hawaii compared to the U.S. overall travel price index trend, you realize why air transportation information is so close to my heart, especially in light of its quality, which could be visualized by flipping the ATPI chart 180 degree.

With that said, you can find results in TranStats for every other mode of transportation. There are surveys of ferry operators if that is your cup of tea, and they are very current, covering seven annual surveys from 2000 to 2006. This is not true for every data source, simply because some statistics were just made once or twice, such as the Nationwide Personal Transportation Survey.

You can find data about railway transportation, too. You can learn that Amtrak has a worse and faster worsening on-time performance for routes of 400+ miles than even the most inferior airlines (like ComAir or Atlantic Southeast). The point is that railway delays affect far fewer passengers in far less brutal ways than the actions and inactions of the airlines. There is much higher visibility of the frustration of enplaned passengers (to use the technical term for passengers boarded) than of railway passengers during spring break, all summer, Thanksgiving, and around Christmas.

Statistics by main topics of transportation

The topics of the statistics relate to various aspects of transportation. Beyond the obvious topics and databases for infrastructure, economic status, safety & security, energy use, freight and passenger transport, TranStats include environment-related and general demographic data sets, or offers links to them.

The Numbers

The details at macro and micro level are very useful, allowing you to dig as deep as you wish. This is much needed when you want to check, for example, if there is any correlation between, say, the compensation for full-time employees and their performance at the a) airline industry, b) network airline and c) individual airlines level. The compensation statistics and the rate of changes clearly indicate that the companies that cut compensation of their employees the most are almost the same as the ones whose satisfaction rating has changed the most — for the worse.

TranStats confirms with data what has been the gut feeling of all those who have traveled by air: the deteriorating service of the entire airline industry. This is reflected well in the tables, but not on all the graphs because the value range on the y-axis hides the decline. The delay graph, for example uses a 0-100% scale on the y-axis of the bar chart, so the increase in delays from 16.3% in 2003 to 22.6% in 2006 are hardly visible even if it is much “feelable” by passengers.

The numbers tell a lot objectively about why we feel that we get less and less for our money as air fares keep increasing. The increase in this table shows the seasonally adjusted fares as a component of the consumer price index. Of the major airlines, US Airways has had the most miserable performance (in a category that can be objectively measured) with its on-time arrival rate going from a not-so-good 83% in 2003 to 70% in 2007. How it could make a $633 million profit and how the percentage increase was calculated remains an enigma for me.

Large-scale passenger surveys made in the U.S. clearly reflect that winter (and spring and summer and fall) of discontent of airline passengers. The bottom position of United Airlines —which slid the most in the airline passenger survey for the American Consumer Survey Index — among the network airlines and way below even the industry average, mirrors not only what I feel about its policy and practices, and what TranStats performance numbers indicate but also the less directly measurable quality criteria of customer care. It certainly will not improve the airline's reputation for care and competence when its entire computer system had a two-hour meltdown at eight in the morning as it happened three weeks ago with United at its home turf at O'Hare. This is not reflected (yet) in the delay statistics of TranStats, neither in the (dis)satisfaction points of the passenger surveys. This may just worsen if flights of US Airways and some other inferior airlines keeps littering the U.S. airways and if United Airlines keeps making the friendly skies look more and more unfriendly through the window of its planes, and keeps sweeping under the rug the problems while cultivating a red carpet image.

The Software

The interface of TranStats is very good. Browsing with subsequent cherry picking and customization is the main approach when navigating through the Data Finder menu options of mode/target of transportation, or main subject categories. Usually you see a number of databases/data sets listed. They have a short, informative description, and a profile with more detailed description and more specific subject coverage, and with details about the time span of coverage, the frequency of update, links to summary tables, links to related sites, and information about the content provider. Many fee-based services don't do such a systematic, competent job in this regard.

The site searching template allows you to search by a combination of keywords, mode of transportation, subject category, and year. It was not clear to me exactly what data elements are searched beyond the name, short description and profile of the data sets. The table names are apparently not searched because the search result list for casualty did not include the Marine Casualty Fire and Explosion Table.

The most appealing software feature is the customization of the results by the users. You can include or exclude certain data or limit the time period reported in many charts. For example, if you want to see the relationship for the Transportation Services Index and the Gross Domestic Product Index only from 1995 to 2007 and without the freight transportation data, you can easily customize the parameters to get a clean and simple chart. Choosing the seasonally adjusted data version takes place in another phase, but during all these customization steps, the system holds your hand.

For many statistics a much wider set of parameters can be customized. For example, for the number of border entries, you can choose the state, the type of entry (bus, train, automobile, etc.), the time period, the frequency of measured data (annual, monthly figures) and the chart type (line, bar, area). You can also customize the analysis tools (absolute values, value changes, percentage changes). You can make a graph for another filter variable displayed, and juxtapose the two charts, for example, to show the yearly number of bus versus train passengers coming into California. This is a particularly precious, and rarely seen feature.

It is a bonus that many of the statistical tables can be re-sorted instantly after display by numeric value (in decreasing or increasing order), and by the filter value (state code, airline code, etc.). It is another bonus that you can download many of the data sets in Excel or CSV format.

The data sets are annotated and footnoted in an exemplary manner. There is a glossary in case a term used in the statistics is not clear or not unambiguous, such as enplanement which has a clear explanation (and wrong spelling in the glossary, but not in the data sets).

This is an excellent reference resource even if it brings bad news in many regards in a service category where U.S. companies typically used to excel: commercial airline passenger services. Statistics from TranStats and from my companion review of FlightStats show that by now the services of many U.S. carriers is not only way below of traditional and mostly new carriers in East and South Asia and the Middle East, but also way above them —in price. The excellent software makes this numeric database very user-friendly. Overall, the Department of Transportation, the Bureau of Transportation Statistics, and the Research and Innovative Technology Administration (RITA) unit of the DOT deserve credit for compiling a top-notch, searchable database of transportation statistics with extensive and very-well-integrated links to digital information resources related to transportation.

— Péter Jacsó