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.
500 Fifth Street, NW Washington, DC 20001 Phone (202) 334-2934 Fax (202) 334-2003 www.TRB.org May 3, 2013 Mr. Victor M. Mendez Administrator Federal Highway Administration U.S. Department of Transportation 1200 New Jersey Avenue, SE Washington, DC 20590 Mr. David L. Strickland Administrator National Highway Transportation Safety Administration U.S. Department of Transportation 1200 New Jersey Avenue, SE Washington, DC 20590 Mr. Bud Wright Executive Director American Association of State Highway and Transportation Officials 444 North Capitol Street, NW Suite 225 Washington, DC 20001 Subject: First Report from the Committee on the Long-Term Stewardship of Safety Data from the Second Strategic Highway Research Program Dear Mr. Mendez, Mr. Strickland, and Mr. Wright: In response to a request from the U.S. Department of Transportation (US DOT), the National Research Council (NRC) formed a committee to examine the long-term stewardship requirements for the second Strategic Highway Research Program (SHRP 2) driving-safety data and to provide advice on strategies for meeting those requirements. (See Appendix A for the committeeâs statement of task and Appendix B for biosketches of the committee members.) At the committeeâs first meeting, which was held in February 2013, US DOT presented four overarching objectives for the driving-safety data: maximum accessibility by qualified researchers commensurate with consent agreements signed by study participants, protection of the privacy of participants, data security, and development of tools to facilitate widespread and maximum use of the data. 1 US DOT asked that the committeeâs initial activities include providing input within a few months on the following topics: ï· The ownership arrangement that best enables US DOT to meet its objectives, including consideration of possible ownership partnerships; ï· Structure and/or models that will enable US DOT to meet its objectives in the best manner, taking into consideration the diverse needs of stakeholders; and ï· Key elements to consider in a transition plan to the eventual owner, including necessary activities, overall costs (e.g., personnel, information technology requirements), and timeline. 1 Presentation to committee by Tony Furst, Associate Administrator for Safety, Federal Highway Administration, U.S. Department of Transportation, Feb. 19, 2013.
2 This letter report responds to US DOTâs recent request and provides consensus recommendations (shown in bold) within the context of the committeeâs statement of task. To inform its deliberations, the committee heard various presentations and considered written information (see Appendix C). This report has been reviewed in draft form in accordance with procedures approved by the NRC Report Review Committee (see Appendix D for a list of reviewers). The limited scope of this first report did not permit the committee to consider all aspects of its statement of task in detail at this time. The committee will continue to consider various issues discussed in this report as well as other issues, as it carries out the remainder of its study and prepares subsequent reports. EXPECTED BENEFITS OF SHRP 2 SAFETY RESEARCH SHRP 2 was authorized by Congress to address highly pressing needs related to the nationâs highway system, including the reduction of highway deaths and injuries.2 It is administered by the Transportation Research Board (TRB) of the NRC under a memorandum of understanding with the Federal Highway Administration (FHWA) and the American Association of State Highway and Transportation Officials (AASHTO). 3,4 The main goal of the SHRP 2 safety research program is to obtain a better understanding of the role of driver performance in traffic safety. This includes how the driver interacts with and adapts to the vehicle, traffic, roadway characteristics, traffic-control devices, and environmental conditions. The intent is to obtain a better understanding of how these factors and their interactions affect collision risk. 5 Data collected under SHRP 2 include the Naturalistic Driving Study (NDS) data 6 and the Roadway Information Database (RID). 7 NDS data are collected about volunteer drivers in the context of their natural driving behaviors. NDS data, which are collected continually during a vehicle trip, include data such as driver performance, mechanical information about the moving vehicle, and aspects outside the vehicle (e.g., weather conditions, road conditions, and interactions with other vehicles). These data provide valuable information on conditions that existed immediately before a collision or a near miss. 2 SHRP 2 was authorized under the Safe, Accountable, Flexible, Efficient Transportation Equity Act: A Legacy for Users (SAFETEA-LU), Section 5210 (Public Law 109-59), through federal fiscal year 2009. Continuing resolutions extended the program through March 2015. 3 http://www.trb.org/StrategicHighwayResearchProgram2SHRP2/General.aspx. 4 The AASHTO executive director is included as one of the addressees of this letter report because state departments of transportation are highly involved in SHRP 2 implementation and they have provided funding for it. Thus, the financial implications of the committeeâs advice may be relevant to funds distributed by those departments. 5 http://www.trb.org/StrategicHighwayResearchProgram2SHRP2/Pages/The-SHRP-2-Naturalistic-Driving-Study- 472.aspx#objective. 6 http://forums.shrp2nds.us/. 7 http://www.ctre.iastate.edu/shrp2-s04a/.
3 The Virginia Tech Transportation Institute (VTTI) provides technical coordination for the NDS data-collection sites and houses all the NDS data. Data for the RID (e.g., number and type of lanes and road curve characteristics) are being collected by the Center for Transportation Research and Education (CTRE) of Iowa State University. RID is important for relating driver actions to the roadway characteristics. Development of the SHRP 2 driving-safety data required funding of greater than $70 million. This money came from funds that otherwise would have been provided to state DOTs for highway construction and related activities. At the request of AASHTO, Congress dedicated some of the highway funds for SHRP 2 research. In addition, new funds for implementing the results of SHRP 2 (including the results of SHRP 2âs safety research) are being provided by the state DOTs from funds that otherwise would have supported each stateâs planning and research activities. The data offer the highway safety community an extraordinary opportunity to study direct relationships between driver behaviors and safety outcomes. There is substantial potential that the data will produce unique and important research results over several decades, leading to public benefits through improved highway safety. Improved safety could be realized through the development of new and improved countermeasures, including targeted policies for mitigating risky driving behavior, in-vehicle safety technology, roadway countermeasures (e.g., improved signing and marking), and advances in vehicle design. The collection of SHRP 2 driving-safety data is scheduled to be completed by November 2013. The full set of data will be complex and huge (about 4 petabytes of data describing 4 million vehicle trips). It includes categorical data (e.g., vehicle type), samples collected at different rates (e.g., speed and GPS position), and video data from four on-board cameras. Preparation of the collected data for use by researchers is ongoing and will be completed in 2014. A critical issue in collection, processing, and dissemination of the data is maintaining the protection of privacy for participating drivers in a manner that is consistent with the requirements of the consent agreements they had signed. Because SHRP 2 is scheduled to end in March 2015, planning for what comes next is imperative if steps are to be taken to have a new regime in place by that time. The benefits of increased safety, including lives saved, will not be substantially realized without an effective and rapid transition from data collection to widespread data use by researchers. An important priority is planning for the long-term administration of the driving-safety data to ensure it will be available and accessible to researchers after SHRP 2 ends. A successful transition from data collection to user access will involve important decision making about data pricing and accessibility, availability of skilled support for researchers, and adequate protection of confidential information about study subjects. A PHASED APPROACH FOR MAKING LONG-TERM DECISIONS ABOUT THE SHRP 2 DRIVING-SAFETY DATA At the committeeâs February 2013 meeting, Mr. Tony Furst, Associate Administrator for Safety, FHWA, indicated that US DOT is proceeding rapidly with a process for deciding the long- term disposition of the completed driving-safety data. He indicated that the committeeâs input is needed very soon concerning the eventual owner and steward of the data, as well as key elements to consider in transition to the eventual owner.
4 The committee agrees that there is urgency in addressing these issues. We are fully supportive of the need for making the data available without delay, because a delay can cost lives. However, based on the committeeâs review of the available information and the membersâ knowledge and judgment, the committeeâs principal finding is that there is far too little experience with the use of driving-safety data at the scale and complexity of the SHRP 2 safety data and there are far too many uncertainties to make firm decisions about long-term institutional and financial arrangements concerning the data at this time. The committee believes it would be a mistake to make long-term commitments that could close the door on other options in the absence of additional information that will only come from experience with research and analyses of these unique data, not all of which have been collected at the time of this report. Therefore, the committee recommends a phased approach to the long-term administration of the data. The first phase (referred to in this report as Phase 1) would be used to develop and assess structures and processes for subsequent phases. Phase 1 would provide the means of reducing key uncertainties in particular areas, such as the extent of data usage by researchers, costs, possible funding sources, willingness of researchers and others to pay for access to the data and for analytic support, and types of facilities and technologies for user access (e.g., remote terminals). It would provide a basis for evidence-based decision making and thus substantially lessen the risk of implementing an organizational structure that is based on faulty assumptions about costs and market response, which is an outcome that could disrupt ready access to the data by researchers and compromise future research projects. Phase 1 needs to be of sufficient duration to obtain enough experience with the uses of the data in order to assess long-term implications, including data use, data protection, and costs. Many topics must be considered and users with a variety of backgrounds will likely be involved. For individual projects that will be conducted during Phase 1, time will be needed for project selection and funding. If funding is to be provided through the Phase 1 program, it may take 6 to 12 months to issue a request for proposals, evaluate proposals, and establish funding contracts with the researchers. If researchers were to provide their own funding to support their analyses, time would be needed to confirm that the proposed analysis could be supported by the driving-safety data and that required privacy provisions could be met. Before the full analysis is carried out for a particular project in Phase 1, trials would be needed to determine if a proposed approach produces the desired data. Additional time would likely be needed for projects requiring video reduction. Rather than conducting all the planned projects in parallel during Phase 1, it would be more practicable to stagger the start times of the projects to spread out the operatorâs workload. Time would be needed to produce reviewed and published research results. In addition to conducting analysis projects, it is important for Phase 1 to include time to develop and test strategies for disseminating information about the data to researchers in an effort to foster and encourage data use, evaluate efforts to ensure confidentiality, and identify long-term sustainable funding strategies for subsequent phases. Work is currently under way to develop smaller data sets, with personally identifying information (PII) removed or transformed. This will allow users to access data sets directly through a web-based interface. There will need to be some evaluation of both the kinds of reduced data provided and the usability of the interface. There may also be a need to test different modes of data access and identify those that work best for different types of researchers. This may involve testing remote secure access sites. Also, it may be necessary to develop special hardware and software capabilities to support secure remote access sites.
5 To allow sufficient time for completing the kinds of activities noted above, Phase 1 should last about 5 years. Governance planning should begin as soon as possible in 2014 or earlier while SHRP 2 is still in effect. During Phase 1, planning for future phases should consider a full range of options for institutional arrangements and funding. Planning for subsequent phases should neither assume nor rule out a priori any options for institutional arrangements and funding. It may be that such institutional arrangements will need to make multiple transitions over the years as conditions and technologies evolve, results are produced, and the community of researchers who use the data develops. DEFINITIONS OF KEY ROLES IN ADMINISTERING THE SHRP 2 DRIVING-SAFETY DATA The committee considered the definitions of key terms presented by US DOT for the roles of ownership, stewardship (i.e., governance), and operation with respect to the SHRP 2 driving- safety data (see Appendix E). Stewardship would involve a board of stakeholders and other experts who develop policies for the governance of the data. The board would be convened by the owner (see Figure 1). Owner Governance Board Data UserOperator Policies Performance Evaluation Transition Planning Convening Contracts Technical Documentation Expert Assistance Reduced Data Sets FIGURE 1. Illustration of various roles of the owner, governance board, and operator in the proposed Phase 1. Ownership The owner of the driving-safety data is responsible for upholding privacy protections, working to meet legal requirements, promoting ease of data access to qualified users, working to ensure adequate user support, working to secure ongoing financial support from various sources (including public and private sources), promoting use of the data, and overseeing the operation and
6 maintenance of the data. 8 The owner also assesses the value of the results produced by the driving- safety program for the purpose of guiding the development of access rules, pricing, and promotion of specific applications. This would help ensure that the data are used effectively to promote the public good and provide support for continued funding. The data owner may also be the system operator and may participate in governance activities. The owner ensures that governance decisions are implemented by the operator. In addition, the owner will need to ensure that its responsibilities and the responsibilities of the governance board and operator are clearly described. The owner would have additional responsibilities related to Phase 1 that involve obtaining information needed to make long-term decisions about the driving-safety data. The owner would be responsible for planning the transition from the SHRP 2 safety program to Phase 1. The owner would also convene the governance board, contract with the operator, and ensure that technical documentation is in place for the SHRP 2 infrastructure and collected data. Because some researchers will not have had experience working with data of such large size and complexity, the owner will need to ensure that expert assistance is made available to the researchers so that they can get maximum value from using the data. The owner should ensure that the operator develops reduced data sets 9 containing events of interest to safety researchers, which will lessen the resource burden on individual projects. Governance The committee recommends that US DOT move away from using the term âstewardshipâ and replace it with âgovernance,â which is a more precise and descriptive term for the functions required (e.g., making critical policy decisions about data access, information privacy, security, pricing, types of product offerings, and performance evaluation). Governance more explicitly connotes the responsibility for policy management regarding data use and data protection, and the obligation to work within legal and institutional requirements. These functions are best accomplished by a governance board or oversight committee composed of stakeholders and other interested individuals with relevant expertise. It would be incumbent upon the board to stay abreast of new technical developments relevant to the housing and access of the data, as well as new developments in data protection. The policies developed by the governance board would be put into practice by the operator. As discussed later in this report, the governance board may have several unique responsibilities in Phase 1 regarding planning for and transition to subsequent phases. The governance strategy for future phases during the long-term use of these data may be different from the strategy in Phase 1. 8 The concept of âownershipâ in this context is intended to mean that an organization has responsibility and accountability for meeting legal and other requirements associated with controlling and managing the data. 9 Reduced data sets are formed by extracting certain information from the raw data and transforming it into a more accessible format, consistent with protection of confidential data.
7 Operation Following policies developed by the governance board, the operator provides the infrastructure and support necessary to house the data and make it available to researchers. This includes hardware and software, operation and maintenance, access control, and researcher support. The operator must have a thorough understanding of the data and expertise in working with very large and complex data sets that include personally identifying information, because users will present a range of research questions and have varied skills to answer them. In addition, the operator should be able to make various types of assistance available to researchers, such as: ï· Help in setting up research questions and sampling the data within their cost constraints; ï· Reduced data sets that can be used to become familiar with the data and answer multiple research questions; ï· Evaluations of the suitability of computer code to be used for sampling and analysis; ï· Advice for complying with institutional review board (IRB) requirements; and ï· Educational workshops and training materials. During Phase 1, the operator will need to provide substantially more assistance to researchers than is typical in order to foster the growth of the user community. This will require a broad skill set with multiple types of expertise to train and assist researchers in formulating and implementing effective analytic approaches. Also, to inform long-term decision making, the operator will need to collect data on user demand and the actual costs involved with providing user access and maintaining the data. ADMINISTERING THE DATA DURING PHASE 1 Phase 1 Owner As SHRP 2 comes to an end, critical work needs to be done in transitioning from collection of the data to developing a system that offers access to the data in an appropriate manner. The ownership of the data during Phase 1 needs to be decided as quickly as possible so that a governance board can be convened and planning can begin. The committee closely examined the options for ownership of the data presented in a recent study carried out by the Volpe National Transportation Systems Center (Volpe Center) in response to a request from FHWA. 10 The Volpe Center report identified three options for the long-term ownership of the driving-safety data. In the first option, NAS would retain ownership of the data. In the second option, NAS would turn over ownership of the data to one or more third parties or consortia of interested parties. In the third option, NAS would transfer ownership of the data to US DOT. 10 Volpe National Transportation Systems Center of the US DOT Research and Innovative Technology Administration, Cambridge, Massachusetts prepared a draft report (dated Jan. 2013) that was provided to the committee: Options for Long-Term Stewardship and Ownership of the SHRP 2 Safety Data.
8 The Volpe Center concluded that âOptions 2 and 3 would require development of a new management structure, funding mechanism, and expertise on the part of the data owner,â and that NAS ownership of the data (Option 1) would present âthe easiest transition to a post-SHRP 2 programâ because it relies upon the current institutional structure. The committee concurs with this assessment and believes that addressing the transitional issue expeditiously is highly important. This is the only option available for Phase 1 that would not result in substantial delay in making the data available to researchers for the period after SHRP 2 has ended. To change course at this point would likely result in a disruption of access to the data by researchers. If NAS were to retain ownership during Phase 1, NAS would implement the policies established by the governance board through TRB and carry out the tasks mentioned above. It is important to note that the committee did not assess the suitability of NAS for long- term ownership. The phased approach recommended by the committee will provide an opportunity to make a decision about long-term ownership in a more deliberate manner based on several years of experience. This will provide empirical information about key aspects, such as costs, data usage, and research interests. Without the proposed phased approach, long-term decisions would be based more on speculative information that may not provide a satisfactory resolution to the issues of concern. All three options identified in the Volpe Center report (and possibly others) need to be considered more carefully in the future in light of what is learned from the Phase 1 experience. Phase 1 Governance Structure A governance board, consisting of about 12 volunteer members, should be convened as soon as possible by TRB to support Phase 1. The board should include representation of stakeholders and other experts from constituencies such as the following: ï· State DOTs; ï· AASHTO; ï· Potential users of the data, including driving-safety researchers, original equipment manufacturers (e.g., motor vehicle manufacturers), and other private-sector users (e.g., the auto insurance industry); ï· Researchers not directly involved with driving safety but who have relevant expertise and experience; ï· Organizations (public and private) that manage and provide access to large and complex databases; ï· Personal privacy advocates from the public or private sector with expertise in the ethical, legal, and/or technical aspects of protecting personal privacy; ï· Experts on other legal issues (e.g., intellectual property); and ï· US DOT (including both FHWA and NHTSA). The governance board would discharge the responsibilities outlined in the previous section within the bounds and resources specified by any agreement between the funding agency and the owner and within the legal and institutional requirements of the owner. The board would also oversee various aspects of Phase 1 planning, as discussed later in this report. The oversight committee for SHRP 2 research provides an example for consideration of how the board might function.
9 Phase 1 Operator The governance board would be the appropriate entity to select the organization that can best fulfill the duties of the operator in Phase 1, including requirements to be specified in future funding agreements. This operator selection by the governance board would be a recommendation to the owner, which would enter into a contractual agreement with the operator. Considerations in making the selection recommendation to the owner would include cost, the ability to provide ready access to the data by researchers during the few years immediately following the end of SHRP 2, and other functions mentioned in the previous section. Although in principle the owner could also be the operator, that would not be feasible in Phase 1 if NAS were the owner of the data. NAS does not have the staff or the infrastructure to carry out the operator functions. Any agreement between the owner and the Phase 1 operator would need to allow for the opportunity of transferring operator responsibilities to one or several other organizations after Phase 1. US DOTâs Distinctive Role Continued public investments will be required during Phase 1. The federal government is in the best position to provide that funding, and US DOT, acting through FHWA and/or NHTSA, is in the best position to be the responsible agencyânegotiating with funding recipients, monitoring the public benefits of the results, and participating in program direction with other stakeholders. It is possible that other federal agencies, such as the National Institutes of Health and the Centers for Disease Control and Prevention, may contribute as public-sector funding partners. By the end of Phase 1, the potential for revenue from users and various private sources should be better known and the need for US DOT to provide ongoing public support will be better defined. OTHER LARGE-SCALE RESEARCH DATABASES The committee considered a Volpe Center report, referred to as an environmental scan, that investigated how other organizations with large-scale research databases have met requirements similar to those of the SHRP 2 safety data. 11 The environmental scan report indicated that the databases it considered included at least some federal involvement, PII data or data with similar access restrictions, use of data primarily for research, and collection of fees for researcher support. The databases included the U.S. Census, particularly confidential data accessed through Census Bureau Research Data Centers (RDCs), and other data accessed through the National Center for Healthcare Statistics RDCs, Center for Medicare and Medicaid Services, and National Agricultural Statistics Service. The Volpe Center report did not identify any database with a significant amount of video or images that also contained PII or sensitive data. The environmental scan report found that the driving-safety data will have characteristics, including funding mechanisms and vetting processes for data requests, that distinguish it from all the databases reviewed, making comparison of cost and requirements problematic. In addition, the report observed that a variety of approaches are used to respond to data requests from researchers, ranging from returning only sanitized results to providing encrypted hard drives. The report 11 Volpe National Transportation Systems Center of the US DOT Research and Innovative Technology Administration, Cambridge, Massachusetts prepared a draft report (dated Jan. 2013) that was provided to the committee: SHRP 2 Safety Data Stewardship Environmental Scan: Case Studies.
10 suggested that the future steward or owner of the driving-safety data may wish to consider a variety of approaches for access to the data, particularly remote access, which is employed to some extent by all the databases it considered. PHASE I PLANNING ASPECTS Phase 1 is essentially a developmental phase requiring an experimental attitude to define the basis for planning long-term policies, structures, and processes. Although the contractual date for SHRP 2 extends to March 2015, the governance board should begin its work as soon as possible in 2014 or earlier to prepare a Phase 1 plan to obtain key empirical information, such as the types discussed in this section. The plan should include development of a process and timeline for the Phase 1 activities. An important early task of the governance board is to recommend selection of the operator for Phase 1. Policies are needed for overall operation, data access, protection of confidentiality of personal information, and program evaluation. Development of an appropriate system architecture (providing guidelines for information, software, and processes used to support the system) over the expected life cycle of the data is an important part of the planning effort. Such development involves identification of responsibilities of the governance board, data owners, operators, and users. It also includes development of a set of principles that would need to be considered for management, use, and protection of the data. Planning for Phase 1 also needs to consider development of quality standards for the data. Given the long time span of expected data use, a preservation plan and model needs to be developed as part of Phase 1 activities, because the data will most likely be managed, distributed, and analyzed through different technical infrastructures over the next 30 to 40 years. Products Portfolio Phase 1 planning should identify the potential user groups that might be interested in accessing the data and the data products (e.g., researcher-friendly reduced data sets) that will need to be developed in 2014 so that they are ready and available for users by early 2015. Reduced data sets that do not include PII are important products for expediting data access by researchers. Making some of the products available to users at low cost (e.g., through subsidies) in Phase 1 and developing a library of software tools that could be used multiple times to address research questions requiring similar types of analytic approaches would foster the growth of a user base. Other important considerations include alternatives for providing researcher access, such as providing remote access to data, and identification of and addressing potential intellectual property rights associated with the data products and tools. Economic Data and Funding Mechanisms Phase 1 will provide the opportunity to obtain empirical data to characterize the potential demand for SHRP 2 research data, actual costs of producing products and providing researcher access, and usersâ willingness to pay (price elasticity) and ability to pay. During Phase 1, rigorous estimates of cost elements should be obtained, including fixed costs (expenses that
11 do not change in proportion to the amount of data usage, such as computer hardware costs) and marginal costs (incremental costs associated with providing data access to the next researcher). Efforts should be made to explore various demand-side cost-sharing structures. The governance board would need to decide on strategies for marketing the data to potential users. Development of long-term sustainable funding will require serious consideration of private sources of funding, including public and private partnerships. Actual experience over time with real-world research will be essential to support setting fees that will balance promoting access and covering costs. Program Evaluation Phase 1 planning should include development of evaluation criteria and a process for collecting lessons learned. Projects now under way that are using the data currently available are an initial source of lessons learned, including project costs and any problems encountered with using the data. The governance board needs to specify early on the kinds of information the operator should collect and the format it should use in reporting the results of Phase 1. However, the board should strive to ensure that required evaluations of cost and other aspects are not overly burdensome for the operator and others who are implementing Phase 1. Assessing impact and quality by using information readily obtained from electronic searches of published research would expedite evaluations. For example, requiring use of a standard format for citation of the driving-safety data in research products could allow Internet searches conducted for impact assessments to find the products efficiently. It is the committeeâs hope that the input provided in this letter report is responsive to your needs in planning for the long-term ownership, governance, and operation of this very important driving-safety data. It was the committeeâs intention to provide advice primarily from a strategic level for launching an effective developmental effort before making long-term commitments. In future letter reports, the committee plans to address some of the topics mentioned in this report, as well as additional topics, in greater detail. Sincerely, Joseph L. Schofer Chair, Committee on the Long-Term Stewardship of Safety Data from the Second Strategic Highway Research Program Attachments