Crash Records Systems (2005)

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.

7HISTORY The first National Conference on Street and Highway Safety in 1924 marked one of the earliest instances of federal interest in motor vehicle traffic crashes. A result of this informal meet- ing of state representatives was the Uniform Vehicle Code that established a legal basis for investigating and reporting crashes. In 1946, the President’s Committee on Traffic Safety asked states to begin developing a database of traffic crashes on which to perform future studies (2). At the federal level, interest in crash reporting from 1924 to 1956 was mostly in an advisory role. However, by 1955 there were 75 million registered drivers and 62 million vehicles, and the annual traffic fatality toll that year reached 38,000. The next 10 years saw an increasing awareness of the national scope of the crash problem and the need for federal leadership and financial aid to assist the states. By the mid- 1960s, the National Safety Council (NSC) reported in excess of 49,000 crash fatalities at an estimated annual societal cost of $3.5 billion. The NSC recommended that the federal role expand to include setting uniform standards and pro- viding financial assistance to the states for safety programs. In addition, the NSC recommended that the states collect crash data in more depth and modernize their crash data col- lection systems (3). The modern era of highway safety began with the passage of the Highway Safety Act of 1966 and continues to evolve today. Section 402 of the Highway Safety Act required, among other things, that states follow uniform standards, establish an effective crash records system, and investigate crashes to determine probable cause. Section 403 of the act included requirements to improve crash investigation procedures and develop comprehensive crash data collection and analysis procedures (4). Based on this legislation, the U.S.DOT pub- lished standards to promote uniformity in the development of state crash records systems. Highway Safety Program Stan- dard Number 10, Traffic Records, requires each state to estab- lish and maintain a centralized system to collect crash data. It further requires that states keep information concerning driv- ers, vehicles, and crashes in compatible files for ease in com- piling statistics and analyzing crash data. This regulation also lists minimum data requirements, such as the model and make of the vehicle, to be included on the crash report form (5). The American National Standards Institute (ANSI) approved two standards intended to promote national uni- formity in crash data. The ANSI D-16 Manual on Classifica- tion of Motor Vehicle Traffic Accidents (6) provides a vehicle damage classification scheme and defines what constitutes a crash. The ANSI D-20 Data Elements Dictionary (7) provides the definitions of the most commonly used terms in crash reporting. These ANSI standards have been updated routinely throughout the years. In 1975, NHTSA established the National Accident Sam- pling System (NASS) and the Fatal Accident Reporting Sys- tem (FARS) (8). NASS is a random sample of nationwide crashes collected by crash investigation teams and FARS is a census of crashes involving fatalities encoded by specially trained analysts in each state. These two national systems have undergone changes over the years, but continue to pro- vide a source of crash data to detect national trends. Late in the 1980s, FHWA established the Highway Safety Information System to collect crash and roadway inventory data from selected states for research purposes (9). The High- way Safety Information System does not represent a statistical sample, but crash and roadway data are added to the system periodically to support various research studies. Generally, state data files are not combined for analysis because there is a lack of similarity in definitions and coding of various elements. Throughout the 1990s, numerous legislative and program- matic actions reflected heightened interest in traffic records systems, including: • The Intermodal Surface Transportation Efficiency Act (ISTEA) of 1991, passed by the U.S. Congress, which decreed that states establish a number of interrelated information management systems to support their deci- sion processes concerning the maintenance of their roadway systems and efforts to improve transportation safety (10). • Several major programs that were undertaken by NHTSA to improve the quality and utility of police-reported crash data including: – CADRE, Critical Automated Data Reporting Ele- ments (11), – CODES, Crash Outcome Data Evaluation System (12), – MMUCC (13), – Sponsorship of an annual national conference on the use of traffic records data, and CHAPTER TWO LITERATURE REVIEW

8on crash data collection, whereas a DOT might be more likely to expend resources to improve data warehousing and link- age. Attempts to balance competing needs can create prob- lems, or the perception of problems, when trying to make changes and improvements to the crash records systems. By 1994, national costs of crash data collection and man- agement were estimated at $130 million, with data collection being 60% and data management 40% of that total (19). The estimated unit cost was $21 per crash. At that rate, even states with a smaller than average number of crashes can expect to spend millions of dollars to collect and manage these data. The increased reliance on traffic records information by the highway safety community to develop, manage, and eval- uate its programs, however, has to be given full considera- tion in making a decision to improve crash data collection and management procedures. Traffic records data, particu- larly information contained in the police crash report, are the basis for virtually all safety programs, from roadside hazard removal to the enforcement of traffic safety legislation. Law enforcement, traffic engineers, the judiciary, private citizens, the medical community, and highway safety program spe- cialists use traffic records data to initiate actions that ulti- mately may reduce the frequency and severity of motor vehi- cle crashes of all kinds. Understanding and satisfying the information requirements of the traffic safety and public health community is the key to developing usable and acces- sible crash records systems. CRASH DATA COLLECTION In recognition of the need to meet differing local needs with their data collection efforts, the updated NHTSA Traffic Records Advisory (20) recommends data systems that are flexible enough to receive data from numerous local systems in a consistent format. Figure 2 shows a data flow diagram (DFD) from the Advisory. This DFD illustrates the wide num- ber of data sources needed to complement the crash records system. The highway safety literature has historically documented the causes of, and problems arising from, poor quality crash data. NCHRP Synthesis of Highway Practice 192 (21) cov- ers this subject well. The causes of poor quality are numer- ous but can be generally viewed as errors in form design (i.e., the data collection instrument was flawed), reporting errors (i.e., the person completing the form made a mistake), and mismanagement of the records (i.e., the original data were somehow corrupted during processing). As shown in the DFD, there are numerous steps in the data collection process where these errors can occur. The following are some of the issues involved in obtaining quality crash data. • Uniformity of data—Crash data uniformity is primarily a national problem, because most states mandate the – Grants to support the development of strategic plans to improve traffic records systems. • Grant funds provided to states by FHWA in partnership with NHTSA for the development of strategic plans to improve their traffic records systems and the develop- ment by FHWA of guidelines for the management sys- tems required by ISTEA legislation (14). • Initiation of the National Governors Association’s project by the Office of Motor Carrier Safety, now the FMCSA, to improve the quality and utility of information concerning crashes involving commercial vehicles (15). As an adjunct to these and other efforts, Congress created a number of incentive mechanisms that provided additional grant funding to support traffic records improvements if a state has not passed certain safety legislation. Individually and collectively, these efforts encouraged states to improve their traffic records systems, in many cases providing both the motivation and funding necessary to do so. Over the years, Congress has increasingly viewed traffic crashes as a national problem meriting federal involvement; therefore, the federal goal in this area has been to provide leadership and financial aid to the states as incentives to develop a nationally uniform system. One example of federal input was the publication in 1990 of an advisory describing traffic records systems and the recommended components of such systems (16). On the other hand, states have tried to retain the maximum degree of flexibility and decentralization so that they can respond to their state needs (17). Ironically, the states have the same difficulty in maintaining a uniform crash records system statewide because their local jurisdic- tions wish to retain the maximum degree of flexibility to meet the needs of their local constituents. NHTSA conducts a periodic survey to identify the custodi- ans of the various records systems that comprise each state’s traffic records system. These results were updated with those obtained from a survey conducted by FMCSA to determine the types of agencies that serve as the custodians for the statewide crash records system (18). The most recent results of those combined surveys indicate the categories of crash custodial agencies as: • Thirty agencies that are roadway oriented (e.g., state DOTs and highway departments). • Eleven agencies that are primarily law enforcement (e.g., Departments of Public Safety, State Police, and Highway Patrol). • Seven agencies that are primarily financial in nature (e.g., Departments of Revenue). • Two agencies that are unknown because of no response to either survey. The diverse uses and users of these crash systems create an equally large and diverse set of demands. For example, a pub- lic safety custodian may place more emphasis and resources

9ance among local reporting policies and thresholds has implications for the statewide crash database that were once only a problem for national comparisons and analy- ses. These include: – Many crashes that may go unreported; – Driver-only reports, or officer desk reports, which may not be as reliable as a trained officer’s report taken at the scene of the crash; and – Systematic biases that are introduced into the data if drivers report certain kinds of crashes, whereas offi- cers report others, or if certain types of crashes are consistently missing from the database. Underreporting or errors in crash reporting affect deci- sions regarding the selection of crash countermeasures, law enforcement activities, allocation of funding at state and local levels, and numerous other traffic safety and transportation system management activities. • Timeliness—In the late 1960s, NHTSA found that crash data were often several months to a year old by the time they were available in the crash reporting system. In the early 1980s, as crash records systems matured, timeliness of crash data became less of a problem, with information available often within 2 months of the date of the crash (22). In 1993, a review of nine states showed the time from crash to crash file as 25 to 210 days (23). Accord- ing to the survey results for this study, one-quarter of the responding states reported that a crash may be entered use of a uniform crash report form. The federal govern- ment has consistently worked toward national standards for crash data reporting for all states. It has only been since the promotion of MMUCC and the National Gov- ernors’ Association commercial truck and bus data vari- ables, supplemented by federal funding initiatives for traffic records improvement that substantive advances have been made in uniform reporting. As states make periodic changes in their crash report forms, they are beginning to incorporate the guidelines into their crash records systems. In addition, most states follow ANSI standards D-16 and D-20 for crash data reporting. As even more states incorporate MMUCC into their crash reporting standards, the ability to provide meaningful analyses and draw appropriate conclusions on a national level will be significantly improved. • Accuracy of data—The more times that crash data are interpreted or key entered into a system, the more likely that the accuracy of these data is affected. Although inadequate training for law enforcement officers may have contributed to problems of accuracy in the past, steps have been taken to improve training programs and to provide automated tools for simpler and more accu- rate crash data entry. • Level of reporting—The level of reporting was once a serious problem only at the national level, owing to the different reporting thresholds used by each state. Today, with fewer and fewer resources to commit to crash data collection, localities are making their own poli- cies about which crashes they will investigate. This vari- FIGURE 2 Distributed traffic records systems.

10 into their crash records system up to a year or more after the crash occurred. CRASH DATA MANAGEMENT Management of crash records systems also affects the qual- ity of these data by duplication of the data handling, outdated software systems, and a lack of system compatibility with other components of a traffic records system. Many of the older legacy systems for crash management were designed with linkages to other components of a traffic records system (e.g., Roadway, Vehicle, and Driver systems) to edit data as it was entered. In addition, extensive system validation edits were in place to improve the quality of the crash data. As resources are reduced, fewer coders are available to enter the crash data and the data entry falls further and fur- ther behind. To overcome these data entry delays, many states have removed system validation edits, reduced the number of data elements entered, and raised the crash reporting thresh- olds. In many cases, those variables required to link the crash file to other data components (e.g., vehicle tag or vehicle identification number, driver’s license number, and location coding) are also removed. The result is that crash data are processed more quickly, but the ability to use these data for analyses is severely limited. TECHNOLOGY STRATEGIES In a study to identify possible improvements in safety infor- mation to support highway design, technological strategies with the potential for improving safety information are pre- sented (24). The following list is adapted from these techno- logical strategies. • Data Collection – Portable computers – Prerecorded data readers – Artificial intelligence – Location technologies – Laser-based measurement – Digital photography – Aerial imaging • Data Communications – Cellular systems – RF (radio frequency) systems – Fiber optic systems • Data Management – Optical scanners (optical mark recognition and opti- cal character recognition) – Artificial intelligence – Error-trapping and correction • Databases – Relational – Object-oriented • User Interfaces – Graphical user interface (GUI) – GIS – Computer-aided dispatch • Decision Support – GUI – Context-sensitive help – Voice recognition – Artificial intelligence • Analytical Tools – Modeling – Simulation. There have been numerous pilot tests and uses of these technological strategies to improve crash records systems, with varying degrees of success. In regard to this report, it was concluded that technologies could address some of the issues of data collection, management, and use. Indeed, a sin- gle technology might address numerous issues. However, it was clear that technology would not solve all of the problems of crash records systems. Beginning in the early 1990s, many technologies were proven useful in the area of crash records systems. These included the following projects: • Technocar 2000—This project, funded by FHWA, NHTSA, and the Texas DOT, proved the use of inno- vative technologies in a law enforcement vehicle to improve the ability of the officer to collect data and report locations, and the ability to establish a link with other sources of information. The vehicle contained a mobile videotaping system; GPS; a pen-based com- puter system with touch screen and in-vehicle docking and keyboard; and the TRASER database software sys- tem for crash, citation, and commercial motor vehicle inspection forms. The TC2000 was examined for human factors conditions and withstood crash tests of the installed equipment and mounting systems (25,26). • ALERT—The Advanced Law Enforcement and Response Technology (ALERT) project, funded by FHWA, National Institute of Justice, and International Association of Chiefs of Police, continued the work started in the Technocar 2000 project. A vehicle was outfitted with an integrated network of devices to sup- port the mobile data collection requirements and pro- vide wireless access to local, state, and federal data- bases, controlled through a single GUI (27). Although this study has ended, automotive and other companies continue to identify vehicle-based technologies for data collection. • Expert systems—This crash data collection program, funded by FHWA, tested the use of expert systems technology to improve the accuracy of police-reported data. Three expert systems were developed, evaluated, and implemented in mobile software programs in use in Iowa: (1) seat belt use, (2) vehicle damage rating, and (3) roadside barrier problem identification (28).

11 • TraCS—Support of the Traffic and Criminal Software (TraCS) is a federal–state partnership, between the U.S.DOT and the state of Iowa, to demonstrate the suc- cessful integration of technologies for data collection, management, and communication of safety information. In-vehicle hardware functions as a mobile data com- puter and for field-based reporting, such as motor carrier safety inspections, citations, Implied Consent [driving under the influence (DUI)] forms, and incident and crash reports. The system uses wireless data communications, mobile video, GPS, GIS, and bar codes (29,30). TraCS use has spread from Iowa into several other states. • Electronic identification—This encompasses a group of technologies that allow storage, retrieval, and compar- ison of personal identifying data. NHTSA has been involved in the development and testing of new driver license technologies for several years (27). These include magnetic strip, bar code, digital photo, digital finger- printing, and “smart card” technologies. The data can include names, coded numbers (such as a driver’s license number), and addresses, along with personal descrip- tive or biometric information (e.g., digital photo, eye color, height, weight, thumbprint, and iris scan). As early as 1993, FHWA found numerous examples of technologies already being used for crash data collection. The most prevalent technologies available at that time were various configurations of portable computers for field data collection, GPS for identifying locations, magnetic strips and bar codes for driver identification, and bar codes for vehicle identification (23). These technologies, along with digital cameras and scanners for optical mark sensing and/or optical character recognition, were identified in that study as components of a model crash data collection system. LOCATION REFERENCE An early NCHRP Synthesis described a location reference method as “a way to identify a specific location with respect to a known point,” including three elements: “(a) identifica- tion of a known point, (b) a measurement from the known point, and (c) a direction of measurement” (31). The two basic location reference methods described in that study are still in use today: • Sign-oriented methods (milepost, reference post) and • Document-oriented methods (calculated mile points, route log, straight-line diagrams). A variation of the sign-oriented method (i.e., locations determined in the field) in practice today is the use of GPS or automatic vehicle locator to identify the coordinates of the location. A variation of the document-oriented method cur- rently in use is a selection of a location using a GIS map. Examples of the more advanced site locator routines are available in the Iowa TraCS software and the Illinois Mobile Crash Reporting System (MCRS) software. The use of a precise location reference method is a criti- cal aspect of crash data, whether analyzing the location of crash occurrences or using the location reference to link crashes to other data sources. Before the passage of the ISTEA legislation in 1991, complete location reference sys- tems were generally available only for those roadways on the state-maintained highway system. Since 1991, there has been more emphasis on referencing locations for local roadways as well. Illinois, Michigan, and Missouri are examples of the many states that have moved toward a location referencing system that identifies all roadway locations, usually as part of a GIS for mapping those locations. An NCHRP study of highway crash and roadway systems describes the advantages and disadvantages of using partic- ular location referencing methods (32). Of most importance to this discussion is the need to use a second location refer- encing method when coordinates are the primary location identifiers or a carefully constructed linear referencing sys- tem. The use of coordinates alone can create difficulties in trying to merge data files because of the level of precision needed to match the locations. For example, a roadway file may identify a location to a particular point, whereas a crash location code may identify a spot several meters from that roadway point. It can be difficult to identify high crash loca- tions because a particular coordinate identifying a location in a crash file (because of its precision) may match only one or two crash records. Knowledge of the roadway and a well- defined linear referencing system allows the effective corre- lation between the various coordinate locations to form a meaningful picture of crash experience. INSTITUTIONAL AND ORGANIZATIONAL BARRIERS Pfefer et al. (24), the authors of an NCHRP study of safety information to support highway design, suggested several organizational and institutional strategies that can affect crash data quality. A few of the issues from that report included: • Poor communication of changes (e.g., new roadways not identified in crash system), • Lack of access to other data systems (e.g., files reside in different agencies), • Inadequate training and feedback for data collectors, • Lack of linkages with other databases resulting in dupli- cate data collection, • Changes in forms and procedures without adequate communication and review, and • No standardized methods of identifying locations. There are numerous barriers to using crash or other data that could be considered institutional, organizational, or sys- tematic. There is often inadequate knowledge about the exis- tence of crash data and its availability, a failure to document the conditions of its collection, varying definitions and mea- suring instruments, and simple reluctance to confront the

adverse consequences of misusing or misunderstanding these data. Access to these data can be affected by the ownership of the data, security issues, and the costs to collect and gen- erate the data. The NSC’s National Agenda for improving safety infor- mation systems includes the following six goals to address the organizational and institutional barriers that have an impact on crash and other components of traffic records sys- tems (33): • Instill an appreciation for the value of highway safety information systems. • Establish a means by which collection, management, and the use of safety data can be coordinated among all organizations and jurisdictional levels. • Integrate the planning of highway safety programs and information systems. • Provide the resources necessary to select appropriate technology. • Establish a cadre of professionals trained in appropriate analytical methods. • Promote technical standards for the characteristics of information systems. The AASHTO Strategic Plan for Highway Safety (34) sup- ports these goals with specific recommendations in the man- agement area that deal with gathering and analyzing crash data: • Goal 21: Improving Information and Decision Support Systems. 12 • Goal 22: Creating More Effective Processes and Safety Management Systems. States have taken significant initial steps to address some of these barriers by developing new directives, documenta- tion, and instructions; creating statewide traffic records coor- dinating committees; and promulgating new tools and stan- dards for the crash and other data records systems. Efforts to transform the existing culture have included implementing incentives (usually financial), overcoming disincentives, educating and training the decision makers and the users and providers of data, and implementing new processes to effect change in the crash records systems. The implementation plan for the international scan for traffic safety information systems (35) proposes a number of strategies to update AASHTO’s Goal 21. These include activities such as: • Marketing traffic safety information to increase public and political awareness of its importance. • Simplifying data collection by law enforcement officers by increasing the automation of data and only gathering data necessary to be collected in the field. • Supporting electronic data collection of all types of data; for example, crash, roadway, traffic, driver, and medical. These and other strategies are discussed in more detail in a FHWA working document, Scan Technology Implementa- tion Plan, which was developed based on the findings of the scan team.