Safety Data and Analysis in Developing Emphasis Area Plans

SECTION XII - Data Improvements and What They Can Do for You

In this section, the case is made for improving the timeliness, accuracy, completeness, and accessibility of traffic records information. In general, more reliable data means that users can make better quality decisions — they can be more certain of the facts and of achieving the desired outcome from any action.

Throughout this guide, a data-driven, decision-making approach has been described as the most effective way to identify and address highway and traffic safety problems. The method is aimed at identifying opportunities to improve traffic safety, where an opportunity is defined as something that is a real problem that there is a real way to address. Using data helps define opportunities in three important ways:

• By describing a problem numerically, we know how large it is relative to other highway traffic safety problems we may be facing: ideally, we will know when and under what conditions the problem occurs and the risk posed by the problem for specific portions of the public. In many ways, numerically describing the problem in detail will often suggest the appropriate solutions.
• By numerically evaluating countermeasures, we know which ones work, to what extent and under what conditions, and at what cost. This provides the decision maker with a proven set of tools to use in situations where the data show there is a problem. Together, proven countermeasures along with well-described problems give the decision maker a more complete idea of where the opportunities lie to improve highway and traffic safety.
• Knowing the problem and the effectiveness of available countermeasures, a decision maker also knows what to expect either if nothing is done or if various countermeasures are applied. Setting numeric targets for each problem area allows decision makers to understand if the target is achievable given the size of the problem and the countermeasures available. It also helps to ensure that realistic targets are established in the first place.

In Section I of this guide, a three-step process is presented for data-driven highway and traffic safety decision-making. The usefulness of data for defining opportunities applies to those same three stages:

1. Define/choose issue(s)/emphasis areas.
2. Set a crash, injury or death reduction goal for that issue.
3. Define the series of treatments and the target subpopulation (drivers, highway corridors, intersections, etc.) for each treatment that will be required to meet your goal.

Section II of this document presents the various data types that are used in highway and traffic safety decision making. It also presents the problems associated with each of the data types, including poor timeliness, accuracy, completeness, and accessibility. The implications of these problems for datadriven decision making are numerous, and yet each section of this guide includes the advice that poor data or missing data are not excuses for ignoring the data altogether when making decisions. This final section goes further by presenting a list of options for what to do to improve the data, and why improving the data is important.

Advice on what can be done will be drawn from two primary sources and from other supplemental sources:

1. DeLucia and Scopatz. NCHRP Synthesis 350: Crash Records Systems. NCHRP Project 20-5, Synthesis Topic 35-03 (29).
2. Council and Harkey. Traffic Safety Information Systems International Scan: Strategy Implementation White Paper (30).

The advice is divided into sections that relate to the organizational structure responsible for planning traffic records improvements and the practical data improvement strategies that can be considered when developing a plan of action.

Organizational Issues

In Section II (and elsewhere), the point has been made that ownership (custodial responsibility) of files that make up a traffic records system is distributed among a variety of agencies at the state and sometimes the local levels of government. Federal databases rely almost exclusively on state data, which in turn depends almost exclusively on data collected (and perhaps stored) locally. These dependencies place the most emphasis on quality improvement on the agencies that are furthest from the money and other resources that are available for data quality improvement. To obtain grants, local agencies typically have to show that they have a numerically large highway traffic safety problem, and that they have a valid solution to those problems, and that the benefits of improving their traffic records information outweigh the costs of the programs they are hoping to fund.

The other barrier to traffic records improvement is that most of the core databases that make up a state's traffic records system exist to serve some other primary purpose. The crash data are there specifically to support highway and traffic safety decision-making, but other components of the system such as roadway files, driver and vehicle files, court files and health-care files all serve a different primary mission. The owners of these datasets are responsible for meeting their primary mission and the continued existence of these files depends on the funding and resources devoted to meeting their primary missions. In that context, raising the highway traffic safety "consciousness" of the leadership in each of the custodial agencies is crucial. This is accomplished through the following initiatives:

• Establish a strong, two-tiered Traffic Records Coordinating Committee (TRCC). Ideally, the TRCC will have an executive level made up of key stakeholders, especially the leadership (or IT directors) from the traffic records custodial agencies. This executive group should then empower a working-level TRCC to oversee improvements to the traffic records system and to make recommendations to the custodial agencies on how best to meet the needs of collectors, managers and users of traffic records data. A TRCC is required by law to maintain eligibility for some portions of federal transportation dollars. NHTSA has well-established guidelines for what makes an effective TRCC. They can be found in Traffic Records: A Highway Safety Program Advisory (see http://www.nhtsa.dot.gov/people/perform/pdfs/Advisory.pdf). Note that in their discussions of improving roadway inventory, traffic, and other non-crash safety data, Council and Harkey (30) recommended a separate "data user/owner" committee within the state highway department because of the multiple owners and users of the inventory and traffic data. This group would serve as an expert subcommittee to the TRCC.
• Conduct a Traffic Records Assessment in accordance with the NHTSA Traffic Records Program Advisory. The Advisory and Assessment are powerful tools for a state to learn where its traffic records system is deficient. The Assessment, in particular, is an opportunity to get expert advice from a neutral third party (a panel of peers/practitioners in various traffic records disciplines). An up-to-date assessment is also required to maintain eligibility for some federal funds. (See http://www.nhtsa.dot.gov.)
• Develop and maintain a Strategic Plan for Traffic Records Improvement. A Strategic Plan serves as a roadmap to get from the current system (with all its shortcomings) to the desired traffic records system — one that meets the defined needs of its users. The TRCC should take the lead in developing and maintaining the plan, and the plan itself should include a detailed series of actions that are assigned to specific agencies and staff within agencies. Milestones and clear performance metrics are key to achieving the goals of this (or any) strategic plan. Section 408 of the SAFETEALU legislation outlines the requirements for a strategic plan in order for a state to become eligible for federal funds available for traffic records improvement (see http://www.nhtsa.dot.gov/people/perform/pdfs/SAFETEA-LU.pdf). NHTSA has also published official rules for this grant funding in the Federal Register (see http://www.nhtsa.dot.gov/people/perform/pdfs/Federal_Register.pdf).

Ultimately, the organizational issues come down to interagency cooperation. If the state and local agencies can work together to solve the data quality and access problems, they will find ways to do so more efficiently than if they operate in isolation. In particular, cooperation means that opportunities for sharing resources and avoiding duplication are more likely to be identified and implemented. Since traffic records systems are costly to build and costly to operate, and since the resources available to devote to the systems are limited, it makes sense to coordinate action among the various stakeholders in order to meet as many needs as possible with the available resources.

The TRCC, by virtue of its broad participation among representative stakeholders, gives the state a valuable resource in the improvement and promotion of traffic records for use by decision makers. By establishing a forum for coordination and cooperation among collectors, managers and users, the TRCC can assist the state's decision makers in identifying the barriers to data improvement and suggesting ways to overcome them. The TRCC can be the focal point for needs assessments, quality control monitoring, and planning.

Data Improvement Strategies

In this sub-section, practical advice identifying data improvement strategies is listed. These items should not be viewed in isolation, but rather as a set of potential actions that could be included in a Strategic Plan for Traffic Records Improvement. The selection of which of these actions to pursue, and in what sequence, is best made as part of an overall strategic planning process undertaken by the TRCC. Specific actions include:

• Develop a formal Performance Assessment program for all components of the Traffic Records System. This program should be overseen by the TRCC, in cooperation with the custodial agencies that would take primary responsibility for implementation and regular reporting of performance measures. The program should include measures of timeliness, accuracy, and completeness of the data, and these measures should be available to data system managers on a continual basis to assist them in day-to-day monitoring of the health of their data system. Summary, aggregate data should be available to the system managers and the TRCC to aid in tracking improvements on some regular basis (monthly, quarterly, and/or annually). Proposals for data quality improvement programs should be tied to the Performance Assessment program by reference to the performance measures that will be affected by the proposed program, and the costs and benefits of the proposed improvements should be weighed against other competing proposals to ensure that traffic records improvement dollars are spent in a cost-effective manner.
• Improve crash data consistency through adoption of national standards and guidelines. The national standards include ANSI D-16.1 and ANSI D-20. The Model Minimum Uniform Crash Criteria (MMUCC) guidelines define a recommended minimum data set and data definitions for crash reporting. When states adopt these standards and guidelines, they are able to use the pre-defined data elements and, in many cases, improve the quality of their system documentation at very little cost. Training manuals, edit checks, and data users guides can also be tied to the standards and guidelines making it much easier for data collectors, system managers, and data users to understand the contents and limitations of each data file.
• Create or improve roadway, intersection, and interchange inventory data. Data on the characteristics of the roadway system including the locations, lengths, geometrics, traffic control features, and traffic volumes for specific roadway elements can serve a vital role in safety management and safety planning. Most state highway agencies have a computerized inventory of roadway segment characteristics, although many local agencies do not. Very few state or local agencies have computerized inventories of the intersections or interchanges under their jurisdiction. The development of such inventories would enhance safety planning. There are currently no uniform criteria for roadway inventory data, but FHWA is currently developing a set of criteria known as the Minimum Inventory of Roadway Elements (MIRE).
• Promote the use of "high-end" automated field data collection and electronic transfer systems. There is a wide variety of field data collection software in use for crash data collection throughout the country. The advantages of the state-of-the-art systems in supporting data quality improvement are, by now, well-documented. A good field data collection system will incorporate methods for validation of data as it is entered by the law enforcement personnel, thus helping to ensure that the information meets pre-defined accuracy and completeness standards. Methods such as GPS, map-based, and pick-lists for street names can help to standardize location data collection for crash (and other) data thus making it much more likely that the events can be tied to specific locations defined in the roadway database. Electronic data transfer from local agencies to the statewide crash database helps to ensure that the data are more timely as well. Some states have chosen to standardize on one software vendor's product, while others have chosen to set standards for the quality of the data that are forwarded to the state and thus accommodate the varying needs and resources of the local agencies. Both methods have shown good results — the most important determinant of success is the quality of the software and the training that the data collectors receive.
• Promote data sharing and linkage among the key components of the traffic records system. Data sharing, including "data-for-data partnerships," is a way to maximize the efficiency of the data collection and data entry efforts of agencies involved with similar datasets (e.g., law enforcement and courts, law enforcement and the state-wide crash data system, and courts and the driver history database). Data linkage is the merger of elements from two or more datasets for the purpose of expanding the support for data analysis. Automating the links between datasets also offers the opportunity to improve data quality by cross-checking information, and by automatically filling in some portions of a data record from validated information already contained elsewhere.
• Establish a data clearinghouse (data warehouse) to support user access to system documentation, analytic tool and support, and linked datasets. Data users, especially high-level decision makers, cannot be expected to spend the time and effort required to obtain the data they need, research its proper use, link it accurately to other data, and conduct analyses in a valid manner without some form of assistance. In recent years, the concept of a data warehouse that provides necessary support for users has gained popularity. The concept involves creation of a central point where users can obtain the data they need, along with current documentation (user guides, data element definitions, and a statement of known limitations of the data) and support from data experts. In an ideal setting, the data warehouse would also include a set of simple-to-use analytic tools available to all authorized users. Some warehouses are also able to provide analytic support to users, including help in using the analytic tools as well as conducting analyses on request. A data warehouse is also an ideal place for creation and management of merged datasets resulting from the linkage between various traffic records components. Where necessary, the warehouse staff can help to control access to restricted information while still ensuring that the data are available to legitimate users.
• Promote the use of GIS and other advanced technologies that add value to the information. The mapping capabilities of GIS are an example of value added data management and analysis. Mapping implies the ability to locate events (e.g., crashes, citations, EMS runs, etc.) in a way that is also compatible with other location-based datasets (such as state and local roadway inventory files, traffic data, population and demographics). By layering the maps of safety-related events with roadway, traffic and other data, a rich picture of the safety problems for a state or local area can emerge. This information, when shared with decision makers in the agencies responsible for data collection, can provide them with a powerful resource that helps them do their job. These decision makers, in turn, are more likely to appreciate the value of collecting and reporting high-quality data in a timely fashion. This is but one example of how an inter-agency partnership and returning data-for-data can help to improve cooperation and help to make the case for improved data quality. Ensuring that the local agencies have access to data and analytic resources (as in the case of a data warehouse) is another way to give them an incentive to improve data quality.
• Establish web-based data entry/editing and analysis facilities for use by agencies that cannot afford field data collection systems. Data collectors (especially in law enforcement, the courts, and EMS) include some very small and not-well-funded organizations. Many of these smaller agencies do not contribute a large amount of data individually, but taken as a group may be responsible for a meaningful portion of the overall highway and traffic safety information available in a state. Under current conditions, it is rare for these agencies to be able to justify or afford field-data collection or other advanced systems. Even if the software is provided free of charge, the cost of the equipment and training required to make use of the software may be prohibitive. In states facing this situation, one solution is to provide web-based access for data entry, data editing, and analysis to support these users. While the initial intent may be to support the small departments/ agencies, the availability of a web-based system may appeal to medium and even large agencies as an alternative to implementing a system locally.

The selection of strategies for improving traffic records in a state is best accomplished in the context of an overall strategic-planning effort. The TRCC should have the lead in such an effort and the custodial agencies should be strong participants and backers of the initiatives. While it is true that federal grant dollars can be used in the initial stages of these efforts, experience has shown that a dependency on grant funding does not bode well for the long-term viability of a system. Budgeting for the life-cycle costs of a system and finding ways to ensure that the system is self-sufficient (for both funding and other resources) are the keys to sustainability for the foreseeable future. The TRCC is an ideal group for making recommendations about the long-term health of the system, but ultimately the funding to support this coordinated long-term effort must come from the state (and perhaps local) agencies with custodial responsibility over a portion of the traffic records system. These agencies are not likely to spend their money on other agencies' systems unless there is some obvious benefit to them in completing their own primary mission. In addition, the decision to share information technology resources may have to be taken at a higher level than even the custodial agencies' leadership. A consortium of data owners, collectors, and users is the best way to make the case for the eventual pooling of resources. The TRCC and Executive TRCC are viewed as a good starting point for developing the argument in favor of shared resources.

Closure — Good Data Produce Better Results

The use of highway safety data is integral to safety decisions. The better and more complete the data, the better the resulting decisions. This guide has provided analytical methods that can assist the safety analyst in choosing and targeting safety improvement strategies in the 22 different emphasis areas of the AASHTO Strategic Highway Safety Plan. Alternative methods have been presented for different levels of available safety data — from crash data only to mileposted and linked crash, inventory, and traffic data. However, the consistent message presented throughout the guide is that the "full" safety data systems will allow for more detailed analysis and more precise answers. Such systems do not currently exist in all state and local jurisdictions that are tasked with the responsibility for these safety decisions. This final section has provided a discussion of ways to improve existing data systems, including recommendations concerning how to better coordinate the various organizations involved in collecting and using the various forms of safety data and what specific data-improvement strategies have proven to be successful in the past. It is hoped that these recommendations will result in improvements in data that will lead to decisions that will help solve one of the largest public health problems faced by the United States — highway crashes.