5
General Discussion
The purpose of the Workshop on Transportation Indicators was to discuss issues relating to transportation and provide the Bureau of Transportation Statistics (BTS) with new ideas for issues to address. Throughout the workshop, a variety of interesting ideas were discussed. At the close of the workshop, all participants were asked to make one overall point that they thought was important. The following is a summary of that discussion.
David Greene commented that a national congestion measure would be an excellent indicator for BTS to develop. It should be based on direct measurement from the highway system using automated traffic data from loop detectors, which are measurement devices that lie in the roadway and record how many cars pass over that point. It is impossible to get this done quickly, but it is feasible over time. Tennessee studied four cities for 2 years, 24 hours a day, 365 days a year; the data collection and processing were essentially automated. There were many problems and statistical issues to be resolved, but these data can be used to develop national measures of congestion performance, first for major cities and for the highest-order highways, but eventually extending out to smaller areas over the years.
Ashish Sen pointed out that there is a problem with this type of data collection; it works well only for major highways. There is a lot of congestion on arterials that cannot be measured by loop detectors. This is an entirely new area for statisticians. There are data imputation problems, and there are inference problems, there are problems of modeling and inferring
what is happening between the loop detectors that still need some resolution. This is a fascinating area for statistical research.
Terry Klein noted that a useful way to begin developing indicators would be to identify the specific objectives of producing the indicators. For example, should the indicator appeal to the public in an effort to change their attitudes and behaviors toward transportation issues? Or is it simply a tool for policy makers and decision makers to be kept apprised of the state of transportation? The indicators would be very different depending on which objective is used. If the objective is to target the media, the indicator would have to be fairly simple to understand.
Klein also mentioned that he liked the idea of a market basket of trips, but that there is a data collection problem in trying to ascertain trip type, in both the numerator and the denominator. Survey data would help collect some of the denominator data, and over time those data can be gathered. But in the numerator, for example, it can be difficult to figure out the purpose of the trip if it was involved in a fatal crash and only one person was in the car.
He then suggested that BTS should defer work on some of the economic growth indicators because it seems as if there are others who are working on the issue. The objective of the economic growth indicators is unclear. Is it to show the effect of transportation on stimulating economic growth or how economic growth affects transportation? Both? Neither? He observed that since a consensus cannot be reached on what the existing data actually mean, BTS should not start working on economic indicators right away.
Nathaniel Schenker stated that the Consumer Price Index and the unemployment rate are good models to follow in developing transportation indicators. They are well-established indicators, and both of these measures are broken down into many different measures, such as unemployment by age, race, region, and so on.
Douglas Robertson reiterated the importance of clearly defining an audience before beginning to develop transportation indicators. There is more than a single audience for this type of information, so it is important to group the audiences according to some criteria. And it is important that the audiences not only understand the data that they are provided with, but also that they act on those data. Consequently, newly developed indicators must have a purpose that is based on a need for the information that they provide. Many of the indicators discussed today did not have a clearly stated purpose. Consequently, Robertson would not spend any money on
developing new indicators before studying current measures to see how they can be improved with regard to data quality, quantity, accuracy, reliability, and usefulness. For the short term, BTS should spend its money on ways to improve current indicators and tying current performance measures to purposes based on needs, targeted to audiences that are going to act on the information.
John Miller suggested a couple of new indicators that he thought could be quickly developed. In the area of safety, a measure of injuries per household is a good idea. This is an indicator that the public can relate to without knowing exactly how the number is computed. Also, injuries are measurable and easily understandable. The number of fatalities would be very, very small, but the number of injuries would be larger, and it is something that people can see a change in.
In the area of economic indicators, Miller thought it would be interesting to find out how the performance of the transportation system affects industry sectors. If, for example, the reliability of travel-time estimates is increased by 20 percent, what effect will that have on the book sales industry for e-commerce?
Lindsay Griffin noted that the Department of Transportation has been measuring fatalities for many years. Fatality is the most consequential form of injury, and it is also the one that can be counted most reliably. The motor vehicle death rate per 100 million vehicle miles has gone down 3.8 percent per year since the mid-1960s, when the department was formed. Now is a good time to address the measurement of injuries, which are a major problem. Currently, the injury data that are used for national measures come from the general estimate system, which is based on police-level data. This method uses a primitive injury scale, one that is highly unreliable; it is important that more reliable scales are put in place.
Kenneth Campbell stated that an indicator would be successful if a good conceptual or theoretical model backs it. He spoke specifically about traffic safety. From his point of view, the area of injury prevention has always been divided into two parts. First are crash-worthiness problems, which have to do with the probability of injury given the vehicles involved in a collision. Second is the risk of accident involvement given a certain number of miles traveled. These two match very well with fundamentally different program areas; it involves a completely different program area to prevent injury given a collision than it does to prevent collisions in the first place. Hence, it would be useful, instead of looking at the historical trend of fatalities per mile traveled, to break that down into its two logical com-
ponents: the probability of fatality given collision involvement and the probability of collision versus miles traveled.
Damian Kulash noted that it is important to generate simple, cheap transportation measures in order to anchor fancier, more expensive measures. For example, in the safety area, fatalities per vehicle- or passenger-mile could be a good, easy measure. Generating a statistic once a year would be a way of focusing attention on specific groups, as such cyclists or older Americans. In the mobility area, congestion is important, but it seems that the concern of the press and the public, for good or ill, is still driven by the journey to work. So some sort of generalized cost—time plus money— of the journey to work would be a very useful indicator.
Rolf Schmitt observed that the fundamental statistics BTS can provide are annual, and eventually quarterly, estimates of total transportation activity. That means tons, value, and distance of goods moved in the United States; trips, travelers, distances traveled; and frequency and distance of vehicle movements. Those estimates underlie the general areas of the workshop: economic, safety, and mobility. Theory is needed to support that, but there is no theory that underlies passenger movement.
Barbara Fraumeni supported the use of an indicator on the contribution of transportation industries to economic growth. That statistic is available, is commonly published, and therefore it would attract attention. One of its shortcomings, however, is that it does not consider consumers. It is an important statistic because everyone can relate to GDP.
Fraumeni also mentioned the importance of productivity. Currently available productivity measures for the transportation industries are very basic, in the sense that they do not incorporate changes in quality. With BTS’s prompting, a better measure of productivity could be obtained in the transportation industries.
Randall Eberts commented that it is important how BTS packages its data, that is, how the agency juxtaposes pieces of information. Whenever a government agency puts two variables together, it appears that the agency has endorsed that relationship in some way. BTS should be very careful in doing that. Also, it should try to create a CPI-type measure for transportation. No one has documented the price of transportation, and that is very valuable information. Another example is the Dow Jones industrial average. That measure does not account for every stock, and it changes over time. It is important not to shy away from a measure because it does not provide total coverage.
Alan Karr commented that indicators that facilitate comparisons are
useful. If comparison is not possible, then somehow that needs to be conveyed. Indicators that include geography are very easily compared.
Clifford Spiegelman noted that it is important to consider any risks to the agency or the stakeholders that may be associated with releasing an imperfect indicator. It is important to consider quality before data are released.
Alan Pisarski suggested that BTS provide a national summarization of the fundamental services provided by transportation. For example, how many different international venues can one get to from cities A, B, and C? This activity would be done best on a national level; states and other local areas would have a very difficult time gathering this type of data.
Thomas Palmerlee acknowledged the need to continue to make the statistics understandable. To use technologies like Geographic Information Systems for visualization is really a critical issue to helping the public, and even transportation analysts, to understand those relationships.
Piyushimita Thakuriah suggested that BTS develop some type of schematic representation, perhaps beginning with indicator objectives and going down to the possible candidate indicators that were discussed today. The schematic should also note what types of data are available to calculate each of those indicators, just to get a sense of which are the feasible ones and which are not feasible.
She also noted that gathering data on safety trends across different transportation modes would be useful. That is the one safety indicator that is truly national in scope and that would attract public interest. No state or local agency has the charge to do that, and BTS is uniquely positioned to do it.
Thakuriah observed that mobility indicators are, by nature, local, because they are always relative to the local infrastructure. It is important to keep in mind that different metropolitan areas or different states will contribute to the total index in very different ways. BTS would have to figure out how to display that variability.
Miron Straf mentioned the issue of risk perception by the public. There is a large literature on it, and many agencies are concerned about how the public perceives risk. For example, there is a statistic that if you smoke a cigarette, it takes 11 minutes off your life expectancy. That grabs people. It would get the public’s attention if BTS came up with statistics similar to that. For example, it might be five hours for a flight from Washington to Los Angeles, and 1 hour and 20 minutes for the cab ride to National Airport. Such a comparison would be useful.
Straf also suggested measures from the perspective of the consumer of transportation services. Relocation decisions by individuals, households, and families affect and are affected by the transportation system. He hopes that one day there will be a major national survey of households that follows them over time, long periods of time, for many of their economic decisions and choices—in particular, the changes in the use of transportation services and the choices made among them. It would be good to start with some prototypes and ideas of what measures we would like to see in such a survey.
Janet Norwood closed the workshop by noting that when developing indicators, it is important to know their audience and their purpose. Even if data are gathered and packaged on a national level, it is important to have disaggregated data available, because such data affect the ways in which people use and think about transportation. Data for individual areas are hard to come by, and, even though every state has data, the data are not always comparable. These are problems that BTS needs to work through in order to obtain interesting and useful indicators.
