Attachment J.3, Crash Data Improvement Program Guide.pdf

Crash Data Improvement Program Guide

DISCLAIMER

This publication is distributed by the U.S. Department of Transportation, National Highway Traffic Safety Administration, in the interest of information exchange. The opinions, findings, and conclusions expressed in this publication are those of the authors and not necessarily those of the Department of Transportation of the National Highway Traffic Safety Administration. The United States Government assumes no liability for its contents or use thereof. If trade names, manufacturers’ names, or specific products are mentioned, it is because they are considered essential to the object of the publication and should not be construed as an endorsement. The United States Government does not endorse products or manufacturers.

Suggested APA Format Citation:

Scopatz, R., Brown, R., Zhou, Y., Benac, J., Peach, K., Bryson, M., & Lefler, N. (2017, June). Crash data improvement program guide (Report No. DOT HS 812 419). Washington, DC: National Highway Traffic Safety Administration.

Crash Data Improvement Program Guide i

Technical Report Documentation Page

1. Report No.

DOT HS 812 419

2. Government Accession No. 3. Recipient's Catalog No.

4. Title and Subtitle Crash Data Improvement Program Guide

5. Report Date June 2017

6. Performing Organization Code

7.Authors Bob Scopatz, Richard Brown, Yuying Zhou, Jack Benac, Kara Peach, Meg Bryson, and Nancy Lefler

8. Performing Organization Report No.

9. Performing Organization Name and Address Vanasse Hangen Brustlin, Inc.

8300 Boone Blvd., Suite 700 Vienna, VA 22182-2626

10. Work Unit No.

11. Contract or Grant No.

DTNH22 14 D 00342L/0001

12. Sponsoring Agency Name and Address National Highway Traffic Safety Administration 1200 New Jersey Avenue SE.

Washington, DC 20590

13. Type of Report and Period Final Report, March 2015 – October 2016

14. Sponsoring Agency Code

NHTSA

15. Supplementary Notes The contract manager for this report was Sarah Weissman Pascual.

16. Abstract The Crash Data Improvement Program (CDIP) was developed to provide States with a means to assess and measure the data quality of their crash databases. As part of the overall CDIP program, NHTSA developed a revised CDIP Guide. The CDIP Guide provides an overview of crash data quality and guidance on how to measure and improve quality in terms of the six data quality characteristics - timeliness, accuracy, completeness, uniformity, integration, and accessibility. It also provides guidance for States on how to map their State’s police crash reports to the Model Minimum Uniform Crash Criteria. The CDIP guide is intended for State crash database administrators and managers, State Traffic Records Coordinating Committee \ members, State highway safety offices and State DOT safety office personnel, local traffic safety personnel (e.g., law enforcement, traffic engineers, city and county planners), and other Federal, State, and local traffic safety professionals.

17. Key Words:

safety data, roadway inventory data, traffic data, data collection, crash data, data quality performance measures, traffic records

18. Distribution Statement Document is available to the public from the National Technical Information Service www.ntis.gov.

19. Security Classif. (of this report) Unclassified

20. Security Classif. (of this page) Unclassified

21. No. of Pages

22. Price

Form DOT F 1700.7 (8-72) Reproduction of completed pages authorized http://www.ntis.gov/ ii iii

Table of Contents

Chapter I: Crash Data Improvement Program

1.1 Introduction

1.2 Improving Safety Decisions by Improving Data Quality

1.3 The Traffic Records System and Its Component Databases

1.4 The Crash Data Component

1.5 State Crash Data Collection, Management, and Use

1.6 CDIP Process

1.7 Measuring and Improving Crash Data Quality

1.8 Relationship of Crash Data to Other Traffic Records System Components

1.9 Summary and Conclusions

Chapter 2: Timeliness

2.1 Purpose of Measuring Timeliness

2.2 Defining Timeliness

2.3 Assessing Timeliness

2.4 Developing Timeliness Performance Measurements

2.5 Examples of Timeliness Performance Measurements

2.6 Final Considerations

2.7 Questionnaire

Chapter 3: Accuracy

3.1 The Purpose of Measuring Accuracy

3.2 Defining Accuracy

3.3 Assessing Accuracy

3.4 Developing Accuracy Performance Measurements

3.5 Examples of Accuracy Performance Measurements

3.6 Final Considerations

3.7 Questionnaire

Chapter 4: Completeness

4.1 The Purpose of Measuring Completeness

4.2 Defining Completeness

4.3 Assessing Completeness

4.4 Developing Completeness Performance Measurements

4.5 Examples of Completeness Performance Measurements

4.6 Final Considerations

4.7 Questionnaire

iv

Chapter 5: Uniformity

5.1 Purpose of Measuring Uniformity

5.2 Defining Uniformity

5.3 Assessing Uniformity

5.4 Developing Uniformity Performance Measurements

5.5 Examples of Uniformity Performance Measurements

5.6 Final Considerations

5.7 Questionnaire

Chapter 6: Integration

6.1 Purpose of Measuring Integration

6.2 Defining Integration

6.3 Assessing Integration

6.4 Developing Integration Performance Measurements

6.5 Examples of Integration Performance Measurements

6.6 Final Considerations

6.7 Questionnaire

Chapter 7: Accessibility

7.1 The Purpose of Measuring Accessibility

7.2 Defining Accessibility

7.3 Assessing Accessibility

7.4 Developing Accessibility Performance Measures

7.5 Examples of Accessibility Performance Measures

7.6 Final Considerations

7.7 Questionnaire

Chapter 8: MMUCC Mapping Program Guide

8.1 Introduction

8.2 Process for Mapping a State PCR to MMUCC

8.3 Process for Mapping State Crash Database to MMUCC

8.4 Summary

References Appendix A: Pre-Site Visit Information Collection Appendix B: Assessment Questionnaire v

List of Tables Table 1.1. Components of a Traffic Records System

Table 1.2. Example Crash System Data Quality Performance Measures

Table 2.1. Example Data for Timeliness

Table 2.2. Example Data for Timeliness Calculation, Sorted by Number of Days

Table 2.3. Example Data Showing Timeliness of Paper PCR Submissions by LEAs

Table 2.4. Example Data Showing Timeliness of Electronically Submitted PCR Data by LEAs

Table 2.5. Example Data Showing Timeliness of Manual Location Coding for Reports Submitted by LEAs

Table 2.6. Example Data Showing Law Enforcement Timeliness in Correcting and Resubmitting Crash Reports

Table 3.1. MMUCC Attributes for the MMUCC Light Condition Data Element

Table 3.2. Example Data for Accuracy Calculation C-A-1

Table 3.3. Example Data for Calculating Percentage of Matching VINs

Table 3.4. Number of Errors per Paper Form Report in 2014

Table 3.5. Number of Errors per Electronically Submitted Report in 2014

Table 3.6. Percent of Decodable VINs in Crash Reports – 2014

Table 3.7. Percentage of Crash Locations Successfully Mapped

Table 4.1. Percent of Complete Crash Reports Received in 2009-2013

Table 4.2. Overall Completeness Percentage 2009-2013

Table 4.3. Percent of Unknown or Blank in Critical Data Fields for Which “Unknown” Is Not an Acceptable Value 2009-2013

Table 4.4. 2013 Ratio Measure of External Completeness

Table 4.5. Annual Proportion of Serious Crashes 2009-2013

Table 5.1. 2013 Ratio Measure of Uniformity

Table 6.1. Example Linkage Fields for Probabilistic Crash-Hospital Linkage

Table 6.2. Linkage Variables by Datasets Linked and Type of Linkage

Table 6.3. Example Data for Integration Calculation C-I-1

Table 6.4. Example of Tracking Data Sharing Agreements

Table 7.1. Agency Response Survey Table From Principal Users Satisfaction Survey

Table 7.2. Example Website Usage Statistics for Use in Accessibility Measurement

Table 8.1. MMUCC Element Capability Rating Scale

Table 8.2. PCR to MMUCC Mapping Compatibility Rating Scale

Table 8.3. Crash Database to MMUCC Mapping Compatibility Rating Scale vi

List of Figures Figure 1.1. Fatalities and Fatality Rate 1964-2013

Figure 1.2. Economic and Societal Cost of Crashes in Comparison to Roadway Infrastructure Spending, 2010 (Revised)

Figure 1.3. Traffic Records System

Figure 1.4. CDIP Process Flow

Figure 2.1. Example Crash Process Flow Diagram

Figure 2.2. Example Showing Improvement in Statewide Average Timeliness for Crash Reports

Figure 2.3. Example Showing Timeliness of the Release of an Official Annual Crash Database

Figure 3.1 GIS Display of Crash and Roadway Information

Figure 3.2. Example Change Log for a Single Crash Record

Figure 3.3. Critical Data Elements from NHTSA

Figure 3.4. Average Number of Errors per Crash Report 2004-2014

Figure 5.1. MMUCC Mapping Results Example from NHTSA’s Spreadsheet

Figure 6.1. Example of Deterministic Linkage Between Crash and Roadway Data in a State

Figure 6.2. Crash Data Linkage Opportunities with Other Traffic Records Data Systems

Figure 7.1. Example User Satisfaction Survey

Figure 8.1. State PCR Attribute “Other” Can Be Mapped to the MMUCC

Figure 8.2. State PCR Attribute “Other” Cannot Be Mapped to the MMUCC

Figure 8.3. Data Target (Left) and Data Source (Right)

Figure 8.4. Example of a Mapping Table vii

Acronyms AASHTO American Association of State Highway and Transportation Officials CDC Centers for Disease Control and Prevention

CDIP Crash Data Improvement Program

DOT Department of Transportation

ED emergency department

EMS Emergency Medical Services

FARS Fatality Analysis Reporting System

FHWA Federal Highway Administration

FMCSA Federal Motor Carrier Safety Administration

DUI driving under the influence

GHSA Governors Highway Safety Association

GIS Geographic Information System

GCWR gross combined weight rating

GVWR gross vehicle weight rating

HSP Highway Safety Plan

LEA law enforcement agency

LRS linear referencing system

MAP-21 Moving Ahead for Progress in the 21st Century

MMUCC Model Minimum Uniform Crash Criteria

NHTSA National Highway Traffic Safety Administration

PCR police crash report

PDO property-damage-only

PII personally identifiable information

QC quality control

RMS Records Management Systems

SHSP Strategic Highway Safety Plan

TAT Technical Assistance Team

TRCC Traffic Records Coordinating Committee

U.S. DOT United States Department of Transportation

VIN Vehicle Identification Number viii

Chapter 1: Crash Data Improvement Program

1.1 Introduction

The Crash Data Improvement Program was developed to provide States with a means to assess and measure the data quality of their crash databases. That program provides States with performance measures that can be used to establish the timeliness, accuracy, completeness, uniformity, integration, and accessibility of their crash data. State crash database administrators, managers, and technicians can use the resulting information about their systems to make improvements.

Purpose of the Crash Data Improvement Program Guide

This CDIP guide will help States establish baseline performance measures that reflect the current status of crash system quality characteristics and conduct periodic updates to assess progress in improving crash data quality. The CDIP guide uses examples of good practices to help illustrate the use of quality performance measures.

The CDIP guide is intended to address the following issues relating to the crash database:

• What are the data quality characteristics?

• Why is each quality characteristic important?

• What is the definition of each data quality characteristic?

• What performance measures can be used to measure the quality characteristics?

• How are the performance measures calculated or derived?

• What metrics are used as numeric goals for data quality performance?

• How closely do the State police crash report data elements map to the Model Minimum Uniform Crash Criteria data elements?

• How closely do the State centralized crash database data elements map to the MMUCC data elements?

In addition, the CDIP guide provides sample management reports that present quality performance measures for various agencies or the State as a whole. While the CDIP guide’s conceptual principles will be applicable to other traffic safety databases, the specific information presented in this CDIP guide is intended to be directly applicable only to a State’s crash database.

Intended Audience

The CDIP guide is intended for State crash database administrators and managers, State Traffic Records Coordinating Committee members, State highway safety office and State DOT safety office personnel, local traffic safety personnel (e.g., law enforcement, traffic engineers, city and county planners), and other Federal, State, and local traffic safety professionals. In addition, the information presented here should benefit anyone who uses State crash data.

How to Use the Guide

This CDIP guide contains a top-level description of the CDIP process (Chapter 1) as well as a section for each of the six data quality performance measures defined by the National Highway Traffic Safety

Administration: Timeliness, Accuracy, Completeness, Uniformity, Integration, and Accessibility, and a final chapter on mapping State crash data elements to the MMUCC data elements. The introductory material lists the steps and products of the CDIP and explains the relationship between crash data and the other core components of the traffic records system. The following chapters are specific to one of the six data quality attributes. The final chapter describes the process for mapping to MMUCC. States may choose to use this guide to assess the data quality management practices for their own statewide crash records system. However, they are strongly encouraged to use the services of the NHTSA-provided technical assistance team in order to receive independent expert advice and an independent MMUCC mapping.

This guide also includes the CDIP Questionnaire. States complete the questionnaire in advance of the CDIP Technical Advisory Team visit. The questions provide the CDIP TAT the ability to create the onsite presentation for the State and address the items that have no solution in place. For example, if a State no longer uses paper crash report forms, the CDIP TAT can concentrate their presentation on the issues related to electronic data collection and transmission.

These questions should be assigned to people most knowledgeable about the State's crash data and systems. Where necessary, additional information detailing the question’s standard of evidence is indicated with the (SOE:) notation. Where applicable, questions that overlap with the NHTSA Traffic Records Program Assessment Advisory (NHTSA, 2012) are indicated with the (AQ-#) notation. If the State has recently completed an Assessment, it may save time and effort to provide the CDIP TAT with a copy of answers provided to the Assessment Questionnaire.

The Assessment Questionnaire is divided into the following six sections.

• Part I: Administrative details; best answered by the crash data custodian.

• Part II: Top-level processes of data collection and reporting.

• Part III: Specifics of the processes from Part II.

• Part IV: Needs of key users of the crash data and how those needs are met.

• Parts V and VI: Address the crash data quality management program in general and the specifics of data quality performance measurement, respectively.

Parts 1 through 6 are included throughout Chapter 1. The remaining performance measure-related questions are integrated at the end of their corresponding chapters. Both the Pre-Site Visit Information Collection and Assessment Questionnaire are provided in Appendices A and B.

Resources Used in the CDIP Guide

The CDIP guide uses three key source documents as references for definitions of terms, practices, and data quality performance measures.

• MMUCC Guideline: Model Minimum Uniform Crash Criteria, 4th Edition (see Reference section, FHWA, FMCSA & NHTSA, 2012);

• NHTSA’s report, Traffic Records Program Assessment Advisory (NHTSA, 2012); and

• NHTSA’s Model Performance Measures for State Traffic Records Systems (NHTSA, 2011).

This CDIP guide also presents real-world examples of actual State practices aimed at measuring and improving data quality. The CDIP guide presents these examples as methodologies that States may wish to adapt to their own purposes and situations. Where appropriate, the examples are supplemented by step-by-step instructions on how to perform the required calculations, as well as a sample output report showing the information expected from the analysis.

Background

NHTSA’s Fatality Analysis Reporting System data shows that the United States has reached historic lows for the number of fatal crashes and the number of people killed in motor vehicle crashes. Figure 1.1 shows that in 2013, 32,719 people died in crashes on the Nation’s roadways, well below the 45,645 who died in crashes 50 years ago, and below the historic high of 54,589 fatalities in 1972. Over the same period, the fatality rate (deaths per 100 million vehicle miles traveled [VMT]) dropped from 5.39 in 1964 to 1.10 in 2013. Fatalities tell only part of the safety story, however. NHTSA estimates that in 2013 there were 5.69 million crashes, 28 percent of which involved non-fatal injuries. The Centers for Disease Control and Prevention estimates that for every person killed in a motor vehicle crash, 8 people are hospitalized and 100 are treated and released from hospital emergency departments (National Center for Injury Prevention and Control, 2014).

Figure 1.1. Fatalities and Fatality Rate 1964-2013

Crashes take an economic toll in terms of property damage, congestion and travel delays, lost productivity and other workplace losses, medical costs, legal and court costs, insurance administration, and emergency medical services costs. Crashes also have general societal costs expressed as the lost quality of life in years. In 2015 NHTSA published The Economic and Societal Impact of Motor Vehicle Crashes, 2010 (Revised) (Blincoe, Miller, Zaloshnja & Lawrence, 2015), estimating that the total economic cost of the crashes in 2010 was $242 billon; including societal costs, the total was $836 billion. This represents 1.6 percent of the U.S. GDP for that year. In contrast, all government spending (Federal, State, and local) on roadway infrastructure totaled $100.2 billion in 2010 (FHWA & Federal Transit Administration [FTA], n.d.b.). The contrast is shown in Figure 1.2.

Figure 1.2. Economic and Societal Cost of Crashes in Comparison to Roadway Infrastructure Spending, 2010 (Revised)

CDIP Questionnaire Part I: Administrative

The following questions from the CDIP Questionnaire are used to identify the lead agency and other key stakeholders in the crash data management processes, including management of crash data quality. The objective is to better understand the relationships between the crash data custodian and the other key agencies (data collectors and users included) and the capabilities of the statewide crash system to meet the needs of all stakeholders.

1. Which department/agency is the custodian of the State’s crash database of record? (AQ-37)

2. Which section/office within that department has the principal responsibility for managing/maintaining the crash database? (AQ-37)

3. Does State law require that the crash database of record retain the original crash record as submitted by law enforcement? (SOE: provide relevant law or policy, web page links are acceptable.)

4. Who is the person responsible for administering or managing the State’s crash database of record? What is their position or job title? Provide their contact information (mailing address, phone, fax, email)? (AQ-37)

5. Does your State have a relational database for crashes?

6. What type of data structure does your State currently have (e.g., SQL, Oracle, etc.)? (AQ-36) (SOE: provide a description of statewide database and specify how the data is consolidated into a statewide system)

7. If other entities maintain statewide crash databases other than the database of record (e.g., files used for engineering, public health, driver control analysis), who maintains them and what are they used for?

1.2 Improving Safety Decisions by Improving Data Quality

The NHTSA Traffic Records Program Assessment Advisory (NHTSA, 2012) makes the case for quality data as a necessity for decision-making:

High-quality State traffic records data is critical to effective safety programing, operational management, and strategic planning. Every State—in cooperation with its local, regional, and Federal partners—should maintain a traffic records system that supports the data-driven, science-based decision-making necessary to identify problems; develop, deploy, and evaluate countermeasures; and efficiently allocate resources. Functionally, a traffic records system includes the collection, management, and analysis of traffic safety data.

This statement fits well with the requirements set by Congress in the Moving Ahead for Progress in the 21st Century (MAP-21) legislation (MAP-21, 2012). In Uniform Procedures for State Highway Grant Programs Final Rule, (2013), NHTSA described the implementation of revised Section 402 and 405 Highway Safety Grant Programs to meet the requirements of MAP-21 including the direction that each State will submit an annual Highway Safety Plan that:

• Describes the State’s highway safety program and the activities the State will undertake;

• Includes performance measures and targets as a condition for approval of the State highway safety program;

• Demonstrates continual, measurable progress in traffic safety;

• Harmonizes with other relevant safety plans, especially with respect to performance measurement; and

• Supports the U.S. DOT’s efforts to analyze the linkage between safety investments and safety outcomes.

The FHWA provides States with guidance for implementing the MAP-21 requirements for data-driven decision-making in highway safety (FHWA, 2012), and for Strategic Highway Safety Plan development (FHWA, 2013). And most recently, on March 11, 2014, FHWA published a safety-related Notice of Proposed Rulemaking (FHWA, n.d.b). Included is guidance for States to:

• Use data in problem identification, countermeasure selection, and countermeasure effectiveness evaluation;

• Link multiple traffic records data sources;

• Adopt a performance-based approach (i.e., one that relies on performance measurement);

• Collect and report data on four mandatory safety performance measures for all public roadways;

• Use the 4Es (engineering, education, enforcement, and emergency services) to address safety problems;

• Identify, select, and evaluate effective countermeasures; and

• Expand their focus to all public roads.

The U.S. DOT administrations working most directly in highway traffic safety (NHTSA, FHWA, and FMCSA) have implemented major programs to help States adopt the data-driven approach to safety decision-making. Each agency provides States funding for data improvement and access to subject matter experts capable of advising the States on data and analysis improvements to support decision-making. In addition to the Traffic Records Program Assessment Advisory (NHTSA, 2012), NHTSA publishes periodic updates to the Countermeasures That Work report (Goodwin et al., 2013), which provides a list of behavioral program-specific countermeasures, each with indicators of effectiveness, cost, proportion of States using the countermeasure, and the time required to implement. FHWA has promoted use of the Highway Safety Manual (AASHTO, 2010) and related analytic tools. FHWA has also established an online Crash Modification Factor Clearinghouse (n.d.a), which is the engineering analog to NHTSA’s listing of countermeasures that work.

The U.S. DOT guidance and resources related to traffic safety information and analysis are focused on data-driven decision-making. The programs and priorities are consistent in helping States identify ways to improve their data and the analyses performed to support safety decision-making. The emphasis on crash data quality in the CDIP is designed to support States’ efforts to meet the U.S. DOT guidance and use the traffic safety-related resources to their best ability.

Figure 1.3. Traffic Records System

1.3 The Traffic Records System and Its Component Databases

The Traffic Records System is made up of six core component systems. These are described in Table 1.1.

Table 1.1. Components of a Traffic Records System

System Description

Crash The crash database includes all law enforcement-reported crashes meeting the State’s reporting threshold. Sections of the crash report capture data elements related to the environment, people, and vehicles involved in crashes.

Vehicle Vehicle databases include information on the numbers and types of vehicles registered in a State and about the owners of those vehicles.

Driver Driver databases tell us about the license status, past convictions, and demographic attributes of drivers in the State.

Roadway Roadway data has two key component databases: roadway attributes (inventory data) and traffic volume data.

Citation and Adjudication

Citation and Adjudication databases including information about traffic violation charges, convictions, and associated court actions.

Injury Surveillance

Injury Surveillance data include multiple databases describing emergency medical services (EMS, hospital and emergency department treatments, trauma records, and vital records (death certificate) data.

Of these, the crash database is considered central to virtually all traffic safety analyses including both behavioral and engineering- related problem identification, countermeasure selection, and effectiveness evaluation. Crash data provides the “who, what, where, when, how, and why” of safety analysis.

The other databases provide key information that can refine an analysis by allowing users to create crash rates based on exposure to crash risk; identify segments of the population that are over-represented in crashes; identify roadway features associated with increased crash risk; and/or measure the outcomes of crashes in human terms (the nature and severity of injuries).

The CDIP focuses on crash data and its quality. However, it is important to recognize the value of integrating crash data with the other core traffic records system components to support decision-making. As depicted in Figure 1.3, crash data may be integrated with the other individual core traffic records system components as well as with all the components together. Crash data integrated with vehicle information can identify vehicle types (e.g., large trucks, motorcycles) that are over-represented in severe (fatal and serious injury) crashes. Crash data integrated with driver data can help to identify at-risk groups of drivers (e.g., novice teen drivers) and the relationships between driver behaviors (e.g., driving under the influence) and crash risk. Crash data integrated with roadway data is used to identify high-crash locations and to model the crash risk associated with specific roadway attributes and traffic volume levels.

Crash data integrated with citation and adjudication data can quantify the crash risk associated with specific violations and identify the increased risks associated with repeat offenders. And finally, crash data integrated with injury surveillance data can help to quantify the dollar- and human-costs for injuries sustained in motor vehicle crashes.

The benefits of safety analysis using integrated data from multiple traffic records sources are proven when decision-makers set policy and design programs that are targeted to specific people, places, or event types. Safety program spending is more efficient and the States’ ability to quantify the effects of selected countermeasures improves.

This CDIP guide provides examples of crash analyses integrated with other traffic records data. The focus of the CDIP is data quality measurement and improvement, but it is also important to recognize the value of the data for analysis and decision-making. This CDIP guide aims to illustrate the uses of crash data as well as how it can best be managed.

1.4 The Crash Data Component

Crash data is defined as the information gathered by law enforcement (or other sources as identified in State statute) describing the locations, circumstances, persons, and vehicles involved in motor vehicle crashes on public roadways. States may opt to go beyond this definition to include crashes on private property, crashes without motorized vehicles present (e.g., bicycle, pedestrian), crash reports submitted by the involved parties (rather than law enforcement), and crashes that fail to meet the statutory minimum reporting threshold. The crash data system description in the Traffic Records Program Assessment Advisory (NHTSA, 2012) was adopted for the purposes of this CDIP guide:

The crash system not only holds the basic data critical to developing and deploying effective traffic safety countermeasures, it frequently also serves as the hub through which other systems are connected. The benefits and overall utility derived from the other traffic records systems are significantly enhanced by reliable, valid statewide crash data. Linking other systems’ data with crash data enables invaluable opportunities for analysis. The resulting information drives State highway safety and injury prevention programs and has widespread applicability for all levels of government, industry, research groups, lawmakers, healthcare providers, and the public. The State crash system ideally contains—at a minimum—basic information about every reportable motor vehicle crash in the State. (Reportability is defined by the applicable State statute.) The available data should be sufficient to permit decision-makers to draw valid conclusions about the crash experience in their State. Ideally, all State crash data is consolidated into one generally accessible database with a clearly defined organizational custodian. The crash system provides both an official record of the crash and data for analytic purposes. The crash system documents the characteristics of a crash and provides the following details about each incident:

• Who: Information about the drivers, occupants, and non-motorists involved in a crash (e.g., license status, age, sex);

• What: Information about the type of vehicle involved in a crash (e.g., make, model, body type, vehicle registration);

• When: Information detailing the time a crash occurred (e.g., time of day, day of week);

• Where: Information about the crash location (e.g., location name, coordinates, type, attributes);

• How: Information describing the sequence of events and circumstances related to a crash—up to and including the first harmful event through the end of a crash and its consequences (e.g., damage, injury); and

• Why: Information about the interaction of various systems that may have contributed to the crash occurrence (e.g., weather, light conditions, driver actions, non-motorist actions) and/or the crash severity.

Ideally, crash data reflecting all levels of severity (including fatal, injury, and property damage only) is collected and used to support safety analysis.

Through linkages to other traffic records systems components, the crash data system identifies the roadways, vehicles, and people (e.g., drivers, occupants, non-motorists) involved in a crash. Data and analytic tools are broadly available so safety stakeholders can identify locations, roadway features, behaviors, driver characteristics, and vehicle characteristics that relate to crash risk. Crash data is also used to guide engineering and construction projects, prioritize law enforcement activity, and select and evaluate safety countermeasure programs. Crash data is used in analysis related to emergency response and how to maximize the level of care and the survivability associated with injuries sustained in a crash.

CDIP Questionnaire Part II: Crash Data Collection and Reporting

The following questions identify the documentation that detail the policies and procedures for the collections, submission, processing, posting, and maintenance of crash data. This documentation includes—but is not limited to—the crash data process flow diagrams, data standards, and data dictionary.

8. Identify each step in the process and flow of data from the crash event to the completion and review of the crash report, through the data entry process in the State's crash data system. Where relevant, please describe the differences in processing of paper versus electronic crash reports. If available, please provide a flow chart showing the processes. (AQ-56)

9. How many crash data system personnel—State and contract employees—are there and what are their roles (data entry, supervision, location coding, data validation and correction, other)? If there are people who are assigned less than full time on tasks related to the crash data system, please estimate staffing based on full-time-equivalents.

10. Are other offices involved in crash report processing (for instance, locating crashes to a statewide base map based on location information from the crash form)? If so, where are they based and what are their roles?

11. What type of coordination is there among the offices involved in crash report processing? For example, crash involved driver information may be needed for ascertaining financial responsibility, including assessment of any damages to public property—in such cases, multiple offices or agencies may be involved and information must flow between them. Location coding, in some States, is another example of inter-office or inter-agency coordination as the crash data system record is enhanced with location-specific information provided by another office. FARS and SafetyNet are also potential examples of coordinated processes.

12. Are any changes to this process being considered? If so, what are they and why are they being considered?

13. In your opinion, how can the crash data collection processes be improved in your State (refer to steps in the flow chart requested in question 8)?

1.5 State Crash Data Collection, Management, and Use

Crash data collection, management, and use are activities of law enforcement, the data custodian, and analysts or decision-makers, respectively. People in these roles take on specific responsibilities related to data quality. As stated in the Traffic Records Program Assessment Advisory (NHTSA, 2012):

Ideally, crash data should be collected electronically in the field by all jurisdictions using a uniform, efficient approach (e.g., question or scenario-based software) that is consistent with MMUCC guidelines and the statewide database’s validation rules. Data is subject to validation checks at the point it is added to the record.

The State maintains accurate and up-to-date documentation—including process flow diagrams—that details the policies and procedures for key processes governing the collection, submission, processing (e.g., location coding), posting, and maintenance of crash data. This should include provisions for submitting fatal crash data to the State FARS data collection unit and commercial vehicle crash data to SafetyNet.

Process flow diagrams document key processes including interactions with other data systems. Ideally, each diagram should be annotated to show the anticipated to complete each critical step. The process flow diagram also includes the processes for managing errors and incomplete data (e.g., returning crash reports to the originating officer or department for correction and resubmission). The documentation accounts for both paper and electronic process flows.

In addition, crash system documentation indicates if edits and other steps are accomplished manually or electronically. The State ideally has documented retention and archival storage policies that serve the needs of safety engineers and other users with a legitimate need for long-term access to the reports.

Ideally, the State also maintains standards for all traffic records applications and databases, and the data dictionary should include consistent definitions for all elements—particularly those common across applications and databases.

The CDIP incorporates and adds detail to the guidance provided in the assessment advisory. In particular, the following components can be incorporated into a CDIP review:

• Review State’s use of operator (driver) reports of crashes, especially when these are incorporated into the statewide crash database and, perhaps serve as a substitute for crashes that are not reported by law enforcement;

• Detailed examination of a State’s crash data validation procedures and the content of validation rules. Specifically, CDIP includes analyses aimed at discovering the type and frequency of errors that could potentially be addressed by new validation rules;

• Examine State’s process for identifying the need for new validation rules;

• Review the use of crash data in analyses to support traffic safety decision-making. This can include examination of behavior program management and engineering safety decision-making processes;

• Perform analyses using a State’s crash data in creating data quality performance measures;

• Describe the Data Governance processes applied to the crash system. In particular, the data governance discussions would include involvement of the IT support staff along with the crash data managers;

• Analyze methods and effectiveness of outreach to State and local law enforcement agencies (the data collectors) with information on agency-specific data quality performance measures and the need for improvement;

• Compare data quality for paper versus electronic crash reports, and electronic versus paper submission of crash report forms/data; and

• Map the State’s PCR and crash database to MMUCC data elements using a standardized methodology.

States are encouraged to discuss any special issues with their assigned CDIP contractor team leader during the planning phase of the project. This will help the team plan the CDIP to address specific concerns the State may have regarding collection, management, and use of the data to improve safety.

CDIP Questionnaire Part III: Crash Data Collection Specifics

The following questions are intended to identify how information on every reportable motor vehicle crash is collected. This section will review the data collection policies and procedures, software solutions for data collection, and information on the agencies collecting the information.

14. What is the law or policy that requires law enforcement officers to investigate and report on fatal, injury, and property damage only crashes? (SOE: Provide full text of law or policy, web links are acceptable – this may have been provided in answer to question 3 earlier, but please point to the appropriate section of the law)

15. Does the law or policy specify a time limit within which the crash must be submitted to the State crash database?

16. Are there any reasons that a law enforcement agency might have an incentive for not providing crash reports in a timely manner (for example, if a municipality charges fees to the public for providing a crash report copy they may hold onto it for a time before making it available through another source such as the statewide crash reporting system)?

17. What is the minimum reporting threshold (monetary, tow-away, other) for a property damage only crash?

18. How many law enforcement agencies are responsible for investigating crashes and submitting reports to the State? (AQ-52)

19. Does the State have a standard police crash report form that is used by all agencies in the State?

(AQ-53) (SOE: provide a copy of the PCR)

20. If multiple (out-of-date or specific to a municipality) versions of the PCR are used, how many different versions of the PCR form are being used? (SOE: provide a copy of all PCRs currently in use; and under what circumstances/ jurisdictions they are used)

21. Is there a short form used for PDO crashes? (SOE: provide the form.)

22. Does the State require crash-involved drivers to submit an operator report? If yes, does the State include operator-submitted data in the statewide crash database (i.e., blended with data from crash reports supplied by law enforcement officers)? (SOE: provide the operator form if the data are used in the statewide crash database.)

23. Does the State use a supplemental form to collect specific crash data information (for instance, information on commercial motor vehicles involved in crashes or on BAC test results) or is the data collected on a single form? (SOE: provide any supplemental forms and the guidance for their use.)

24. Are there any plans to modify/update the crash or supplemental forms in the State? If ’yes’, which ones, when and who will be involved?

25. Are any law enforcement agencies collecting crash data electronically at the crash scene? If ‘yes’, how many (or what percentage of) agencies collect crash data electronically? (AQ-53) Statewide, what percentage of crash reports is collected electronically? (AQ-53) Does the electronically collected data identically match the data elements and attributes on the paper PCR elements?

26. Does the State have a single electronic crash software product for use by all law enforcement?

27. If not, is there a standard for data collection systems to be attained by all electronic reporting systems? Does the State approve vendors and/or law enforcement agencies to validate their electronic data collection software? If yes, please describe the approval process.

28. By what processes are PCRs from law enforcement agencies submitted to the crash database (e.g., mail, internet, secure electronic upload, CDs)?

29. Does the State crash database accept any crash data electronically? If so, what percentage is accepted electronically? Is there a standard for electronic data submissions? Does the State approve vendors and/or law enforcement agencies to validate their electronic data submissions? If yes, please describe the approval process.

30. Are there law enforcement agencies collecting crash data electronically that do not submit their data to the statewide crash database electronically (i.e., they collect the data electronically using software, but print the report and submit it on paper or send a static image file)?

31. Is any data verified electronically at the crash scene via real-time interface with other data systems (e.g., driver information from the DMV)? If so, which data is verified? (AQ-54)

32. What technologies are used for the verification?

33. What are the data verified against (e.g., driver license file, vehicle registration file)?

34. Are there edit checks/validations run by the crash database on the PCRs submitted electronically, prior to uploading the data? (AQ-54)

35. Do the field data collection software products require the same set of edit checks for completion of a crash report as are required for that report’s acceptance into the State’s crash data system? If not, how do the field data collection edit checks differ from the centralized system’s checks?

(AQ-54)

36. What feedback has your State received regarding problems with either the paper or electronic PCR from the following persons: police officers, crash form reviewing supervisory officer, data entry person, other?

CDIP Questionnaire Part IV: Data for Decision-Making

The following questions are intended to identify how the State uses this data in implementing programs and countermeasures that reduce motor vehicle crashes, deaths, and injuries. It is critical that a State’s crash system include or have access to the key data to address the diverse safety problems such as:

• Engineering: Accurate crash locating to support integration of crash data with roadway characteristic data.

• Enforcement: Ensure driver/vehicle compliance (i.e., graduated driving licensing, alcohol, and speeding).

• Education: Human behavioral issues to address seat belt usage, distracted driving, driving under the influence of drugs and alcohol, and motorcycle, bicycle, and pedestrian safety.

Emergency Response (Injury Surveillance) – data collection and integration of emergency medical services, ambulatory care, acute care, trauma and rehabilitation facilities, and vital records data with other Traffic Records Systems.

37. Does the State have a process to locate each crash onto a base map? (AQ-60, 167, 168)

38. Is the base map inclusive of all public roads in the State? (AQ-60, 167, 168)

39. Based on location information from the crash report, can crashes be assigned a location code that matches the location coding in the base map? (SOE: Describe the process for assigning matched location codes. If the processes differ for State/Federal aid and local roads, please describe both.)

(AQ-60, 167, 168)

40. How long, on average, it takes to assign a crash location to the basemap? (AQ-56) (SOE: provide the timeliness measure for the crash location coding process. If it differs between State/Federal Aid and local roads, please describe both.)

41. Does the crash database custodial office perform the ‘crash locating’ function or is this effort managed by another entity within the State? (AQ-56)

42. What percentage of reportable crashes is locatable on the statewide base map? (AQ-385)

43. What percentage of reportable crashes is successfully located automatically? (AQ-385)

44. Does the State crash database include the vehicle identification numbers of crash involved vehicles? (AQ-59)

45. Does the State crash database include the Blood Alcohol Concentration (BAC) test results of crash involved drivers? (AQ-50)

46. Does the State crash database include all officer-reported restraint systems (helmet for motorcycle operators and riders) availability and usage for all of the occupants of crash involved vehicles?

(AQ-50)

47. How are crash-involved pedestrians and bicyclists recorded in the statewide crash database? Are they recorded as “involved units” (i.e., at the vehicle-driver level) or as “persons” (i.e., at the person level), or in some other manner?

48. Does the State crash database include data elements related to distracted driving? Please provide the data element definitions and attribute list.

49. What is the injury severity scale used on the crash report?

50. How does the State define “serious injury” on the crash report?

51. Are a narrative and diagram required on all crash reports? (AQ-50)

52. Are the narrative and diagram for all crashes retained in the crash database? If yes, are these searchable or otherwise available for use in case selection, data aggregation, or analysis?

53. Does the State track the percentage of reports received with inadequate, incomplete, or missing narratives or diagrams?

1.6 CDIP Process

When a State requests a CDIP through their NHTSA Regional Office, a process is initiated leading from initial scheduling to delivery of the technical assistance team’s final report. To help States decide to participate in a CDIP, NHTSA has prepared this CDIP guide along with a CDIP pre-site-visit questionnaire and information request. States are encouraged to review these materials in advance of requesting a CDIP and to request help or further clarification as needed. The sequence of events is shown in Figure 1.4.

States are encouraged to plan well in advance. It can take time to complete the pre-site-visit questionnaire and to provide the requested information and data files. The TAT requires sufficient lead time to analyze data and prepare the site-visit slides and exercises. Once the site visit takes place, the final CDIP report can be provided to the State within a few weeks of the contractor receiving the State’s comments on the draft conclusions and recommendations. States are encouraged to perform a thorough review of the draft report as this is the best time to correct any errors in the facts as presented by the TAT, and especially any conclusions that may affect the TAT’s recommendations. While the TAT will take State comments into consideration where questions remain, State comments will not change the final report conclusions and recommendations. The CDIP is intended to serve as an external, objective review of a State’s crash data system by outside subject matter experts.

Figure 1.4. CDIP Process Flow

State GR submits written request to

NHTSA

Region

State Respondents answer assigned questions

NHTSA Region forwards written request to

NHTSA TR

Team

CDIP team review answers & request clarification

NHTSA TR

Team confirms State request

CDIP Project Manager will schedule kick off call

State Coordinator sends Respondent info & assigns questions

STRAP Tech Support hosts

State Coordinator training webinar

STRAP Tech Support sends State Coordinator CDIP tokens

STRAP Tech Support launches Data Collection & Analysis phases

CDIP Project Manager hosts kick off call

STRAP Tech Support sends respondent CDIP tokens

STRAP Tech Support sends CDIP team CDIP tokens

CDIP team prepares final report; forwards to NHTSA TR Team

NHTSA TR

Team approves Final Report

CDIP team conducts site visit and leaves draft report w/State x 2

1.7 Measuring and Improving Crash Data Quality

The Model Performance Measures for State Traffic Records Systems (NHTSA, 2011) report defines a six-by-six matrix of six core systems and six data quality attributes (timeliness, accuracy, completeness, uniformity, integration, and accessibility. The Traffic Records Program Assessment Advisory (NHTSA, 2012) includes 10 example data quality performance measures in the assessment of the crash system, as shown in Table 1.2.

Table 1.2. Example Crash System Data Quality Performance Measures

Crash Data System Data Quality Performance Measures

Timeliness

• The median or mean number of days from (a) the crash date to (b) the date the crash report is entered into the database.

• The percentage of crash reports entered into the database within XX days after the crash (e.g., 30, 60, or 90 days).

Accuracy

• The percentage of crash records with no errors in critical data elements (e.g., 95% of reports with no critical data error).

• The percentage of in-State registered vehicles on the State crash file with Vehicle Identification Numbers matched to the State vehicle registration file (e.g., 95% of in-State VINs match).

Completeness

• A decrease in the percentage or number of missing data elements from the crash database for a given time period (e.g., 1 year).

• The percentage of crash records with no missing data elements (e.g., 90% of reports with no missing data elements).

• The…