350 likes | 576 Views
Module 2. Working with crash data. Safety Analysis in a Data-limited, Local Agency Environment: July 22, 2013 - Boise, Idaho. Learning Objectives. Identify potential crash data sources Value of identifying overrepresented fatal and serious injury crashes
E N D
Module 2 Working with crash data Safety Analysis in a Data-limited, Local Agency Environment: July 22, 2013 - Boise, Idaho
Learning Objectives Identify potential crash data sources Value of identifying overrepresented fatal and serious injury crashes Common considerations for using crash data Reading a crash report Understanding regression to the mean (RTM)
When crash data are not readily available… Potential crash data sources
Potential crash data sources State crash data systems GIS layers of geolocated crashes Local law enforcement offices Non-traditional resources that can give insight into particular collision types or contributing factors: EMS, law enforcement, DPW workers, maintenance workers
If we don’t have access to a state or regional crash database Fatality Analysis Reporting System (FARS) • Online database with all fatal collisions across the U.S. • Online query tools • Online mapping tool • Actual data downloads available (raw data)
Over-represented crash locations Overarching trends • 30% of fatal crashes occur on minor arterial and collector roadways • Fatal and serious injury crashes are overrepresented on local two-lane rural roads and four-lane undivided roads What does this mean? • Safety improvements are necessary across local, regional, and state facilities
Crash data considerations Timeliness Consistency Completeness Accuracy Accessibility Value added by data Integration
Timeliness Timely crash data supports decisions that will optimize safety investments – the network, vehicle fleet, social norms, and technology changes over time.
Data Consistency Collect the same data elements over time and for various classes of roadways Collect the same data as partner agencies Changes to data elements should be clearly documented • Example – road names/ route numbers • Changes to the state crash report form
Completeness High value data elements include: • Crash data: • Location reference for the crash • Contributing circumstances • Characteristics that can identify behavioral and roadway related factors for targeted solutions • Other data include: traffic volume, roadway cross-section and alignment data, presence and control type of intersections, posted speeds
Accuracy Reliable information is key to success High value: • Quality control features where crash data are collected electronically (verification with other available information systems) • Employing methods for collecting, verifying and maintaining roadway data
Accessibility Access to the data • Raw data is better than no data • Periodic standard reports are particularly valuable • Ease of use (GIS data, query tools, data export) • Availability and access to data dictionaries and coding manuals
Data Integration Link crash data, traffic volume, roadway characteristics Integrating data systems at state and local level • Consistent data elements • Consistent data structures • Consistent quality control measures
Commonly reported crash report errors - NCHRP Synthesis Project 367
Reading a Crash Report: Background Many to One Relationships (example)
Crash Report ElementsWhere, when, how, who &what • Location • Date and time of day • Lighting, roadway and other roadway environment factors • Involvement of vulnerable users (pedestrians, bicyclists, motorcyclists, and older users) • Vehicle type(s) • Driver information, • Reportable truck and bus information • Injury severity of the crash • Crash type (mechanism of crash) • Contributing factors (BAC or other drug use, speeding, etc.)
Crash Data ElementsWhere and when Crash location • Critical for being able to understand how different locations on the roadway network are performing with regards to safety Time of day • Useful for understanding if there are periods of the day that are over represented in terms of the frequency or severity of crashes
Crash Data Elements Environmental Factors Environmental factors can include: • Weather conditions • Pavement conditions (e.g., wet, dry, icy) • Visibility conditions • Lighting conditions Improve understanding of potential contributing factors and in turn mitigations
Crash Data ElementsUser and Vehicle Type(s) Users • Pedestrians, bicyclists, and the particularly vulnerable (the young and older users) Vehicles • Single Vehicle vs. Multiple Vehicle collision • Vehicle types: large trucks, buses Pedestrian or bicycle involvement Reportable trucks or buses
Crash Data ElementsDriver Information and Contributing Factors Driver Information • Age • Conditions that can increase crash risk • Blood alcohol level • Excessive speed • Distraction • Fatigue • Failure to yield right-of-way or other traffic violations associated with fatal and serious injury collisions
Crash Data ElementsInjury Severity KABCO Scale • K – Fatal Crash • A – Serious Injury • B – Evident Injury • C – Possible Injury • O – No Apparent Injury Crash injury severity vs. Individual injury severity level • Fatality: when a person dies within 30 days of the crash because of injuries sustained in the crash • Fatal crash: at least one fatality but may include other injuries
Crash Data ElementsCrash Type/ Manner of Collision Examples of categories of manner of collision: • Rear-end • Angle • Sideswipe • Run off the road (these crashes may involve impacts with fixed objects such as guardrail) • Head on • Pedestrian or Bicycle
Practical exercise Reading a crash report
Practical exercise Reading a crash diagram
Key to continued success of targeted solutions to reduce fatalities and serious injuries Regression to the mean (RTM)
Why is regression to the mean such a big deal? Crash history is a snapshot of short term crash averages • Averages will change over time • Short term averages are not indicative of the actual long term crash average for a site By accounting for RTM • Funds will be invested where it is most needed to improve safety • Reliable indications of the effectiveness of countermeasures will be known
Regression-to-Mean (Site selection Bias) Site Selected for Treatment due to Short-Term Trend RTM Reduction Observed Crash Frequency Perceived Effectiveness of Treatment BEFORE Actual Reduction due to Treatment AFTER This change would have happened without the treatment! Source: Adapted from NCHRP 17-38
How do we account for regression to the mean (RTM)? Using advanced methods • Predictive methods such as those in the Highway Safety Manual • Assisted by statistical equations that represent the performance of safety at similar facilities, such as: • Rural two-lane roads • 4-lane freeways • Signalized intersections
Summary Crash data and key supporting data are the foundation for many of our safety related decisions Better data will enable us to make better decisions with limited resources We can account for RTM by using statistical methods
EndModule 2 Questions?