DATA ELEMENTS AND ANALYSIS STRATEGIES

Download Report

Transcript DATA ELEMENTS AND ANALYSIS STRATEGIES

DATA ELEMENTS AND
ANALYSIS
STRATEGIES
Forrest M. Council
BMI-SG and UNC Highway Safety
Research Center
Disclaimer
The guidance and opinions presented
here are mine, and do not represent
guidance from NHTSA, FHWA, or any
other agency (including the two
organizations I work for…).
All mistakes are mine too.
Can SHSP Development Be Done
Without Safety Data?
NO!
But not a problem, since at least some data
exists for almost all jurisdictions
And if data doesn’t exist, the problem probably
isn’t large enough to worry about

Not because each life/injury isn’t important, but
because there are a limited number of safety dollars
to spend
The trick for state DOT analysts will be in
acquiring the data, or joining with local analyst to
cover local roads.
Structure of Presentation
Available data – what do we have to work
with?
Basic “analysis structure” for driver,
vehicle and roadway issues
How these analyses can be accomplished

Including current and upcoming tools that
might help (including what can be done with
limited data).
Summary: Review of selected questions
given me by the planning committee.
Safety Data – What Do We Have?
Crash data
Roadway/street inventory data
Traffic volume data
Driver history
Vehicle registration
Population census
Other safety files
Police Crash Data
“Mileposted” for state system
Often un-mileposted for local system
Crash reporting thresholds may differ
within a state, so know your data before
combining them.
Not perfect data…but any decision using
these data is better than a decision without
using them
Fatal Crash Data
Fatality Analysis Reporting System (FARS) 1975-present
Higher quality data on all vehicles and occupants within
all fatal crashes
“FARS Query” on-line (and can obtain files)
City and county codes for each county in US
Tempting to analyze FARS only, but


Will produce small, unrepresentative samples once you began to
do “sorts”
Not as helpful for roadway treatment analyses, since two of the
major factors that cause a crash to be fatal are occupant age
and occupant restraint use – not roadway treatment related
Understand FARS before analyzing (i.e., imputed values,
“person” count not all fatalities, problems in combining
crash/vehicle/person variables in analyses).
CODES (Crash Outcome Data
Evaluation System)
Police crash data enhanced with hospital and EMS or
Emergency Dept. data, and maybe other data.
NHTSA-funded CODES systems or projects in Alaska,
Arizona, Connecticut, Delaware, Georgia, Hawaii,
Indiana, Iowa, Kentucky, Maine, Maryland,
Massachusetts, Minnesota, Missouri, Nebraska, Nevada,
New Hampshire, New Mexico, New York, North Dakota,
Oklahoma, Pennsylvania, Rhode Island, South Carolina,
South Dakota, Tennessee, Texas, Utah, Wisconsin
Data analysts there also
Truck Safety Data
MCMIS (Motor Carrier Management Information System)
FMCSA system fed by state MCSAP (Motor Carrier
Safety Assistance Program) agency and motor carriers
Most comprehensive truck-safety database in US
Five file types (registration, crash, roadside inspection,
compliance, enforcement)
Analysis outputs (including state trends) and tools online
Note that data are much more complete and improved
since 2004 – be careful of multi-year comparisons for a
while.
Roadway Inventory – State
Highway System
Usually inventory of full state-controlled highway
system
Cross-section data (number of lanes, shoulder
width/type, etc.)
Limited states have curve and grade information
Limited states have intersection/interchange
inventory
Expansion of Highway Performance Monitoring
System (HPMS) data which has inventory of
sample sections within inventory classes
Roadway Inventory- Local
Jurisdictions
Usually less extensive than in state system
Pieces maintained by different departments
(e.g., public works, traffic)
Sometimes supplemental data on sidewalk and
crosswalk presence, bike paths, bus stops, etc.
Often just paper files, and linkage of pieces can
be a problem
Traffic Volume Data
AADT usually recorded for each section on state
system
Based on system of permanent (full time)
counters and “short counts” done on 2-3 year
cycle
Truck percentages/volumes included, but based
on many fewer actual counts, so much less
precise
Remember, AADT is an estimate, not an actual
count for most roadway segments
Local jurisdictions often don’t have AADT, only
intersection turning counts
Driver History Data
Record for each licensed driver (and unlicensed
drivers who have crashes) in the state
Contains information on demographics and
license status/restrictions, convictions, crashes
(sometimes)
Could be used to scope and target programs
related to driver violations (e.g., DUI, speeding,
etc.)
Current “snapshot” since data on convictions and
crashes (if present) are often purged after some
time period.
Vehicle Registration Data
Record for each vehicle licensed in a state
Information on owner, vehicle type, VIN,
etc.
Can sometimes be used as “denominator/
exposure” data
But annual mileage per vehicle is usually
not present, and can include errors if
present (e.g., odometer rollover)
Population Census Files
US Census Bureau files on gender, age
and ethnicity within political subdivisions
Can be used as “denominator/exposure”
data, particularly for road-user problem
identification or treatment targeting
Other Safety Data
Citation (as opposed to “conviction”) data

Retained and published by some enforcement
agencies (e.g., DUI citations)
Speed surveys by state and local agencies

But be careful since most are from “special locations”
and not a random sample for that roadway type of
jurisdiction
Most important -- observed occupant restraint
(shoulder belt) use data collected by all states
since 1998.


NHTSA publishes year-by-year comparisons for each
state
Note that only daytime and front-seat usage
Closure on Safety Data
Crash data almost always available for full
state (sometimes not mileposted)
Other supplemental files will usually be
available – just have to find them
So, “lack of adequate data” is almost
never a valid excuse for not developing a
sound safety program.
Tired of “Data”?
OK, let’s moved to how to use it
Basic Analysis Structure for SHSPs
Five-step process
I will concentrate on first three, with limited coverage of
two additional steps
1.
2.
3.
Defining the critical issues in your state
Determining the best treatment strategies for each issue
Targeting the treatments to road-user subpopulations or
roadway sections/locations
Note that sometimes, the order is changed



You may define the “targets” first (e.g., high-crash roadway
locations, younger drivers),
Then the issues (e.g., run-off-road crashes, younger-driver
inexperience)
Then the best treatments (e.g., rumble strips, Graduated Driver
Licensing)
How Do We Do This?
Will be referring to four sources of information often



AASHTO Strategic Highway Safety Plan
(http://safety.transportation.org/) – a plan aimed at saving an
additional 9,000 lives per year
NCHRP Report 501: Integrated Management Process to Reduce
Highway Injuries and Fatalities Statewide
An excellent guide to all the coordination steps necessary to
advance from crash data to integrated action plans
NCHRP Report 500: Guidance for Implementation of the AASHTO
Strategic Highway Safety Plan
A series of Guides presenting low-cost, effective treatments for each
of a series of “Emphasis Areas” (e.g., head-on collisions, large-truck
collisions, pedestrian collisions).
Can be downloaded from http://safety.transportation.org/guides.aspx.

NCHRP Report 500: A Guide for Addressing Safety Data and
Analysis in Developing Emphasis Area Plans – “Data Guide”
(BEING PREPARED)
Defining The Critical Issues
(“Emphasis Areas”)
Usually based on crash analyses
Possible analysis processes presented in both
Report 501 and Data Guide


High-frequency (Report 501, Appendix D1.3)
High crash-burden (Data Guide) combines frequency
and severity using economic cost of crashes by crash
types/situation)
See FHWA report Crash Cost Estimates by Maximum PoliceReported Injury Severity within Selected Crash Geometries
(July 2005)

Overrepresentation
CARE (Critical Analysis Reporting Environment – see Report
501, Appendix D1.3)
Defining The Targets And Best
Treatments (Or Vice-versa…)
So now we have a series of critical Emphasis Areas,
and need to develop individual action plans for each
For road-user (e.g., driver, pedestrian) and vehicleoriented programs, we often define the targets and
then pick the treatments.
Can use further “drill-down” or “overrepresentation”
or “crash burden” analysis to define the
subpopulations of users (e.g., younger pedestrians,
older male drivers, younger motorcycle operators),
who are the largest share of the problem
Defining The Targets And Best
Treatments (cont)
Can use further “drill-down” to identify
specifics (e.g. crash types, location,
day/night, etc.)
Procedures in Report 501 and Data Guide
We then can use the NCHRP Series 500
Guides or other information to pick
treatments that are suitable for these
subgroups.
Defining The Targets And Best
Treatments (Cont)
For roadway-based Emphasis Areas, we have
two procedures we can follow


Identify High-crash locations and analyze each
location to define the best treatment(s)
Identify a potential high-payoff treatment (e.g. rumble
strips to prevent run-off-road crashes) and then
identify the set of roadway locations where this
treatment will produce the greatest benefit per dollar
spent (a crash-cost based procedure).
Repeat this procedure for different high-priority treatments
until the fatality/injury reduction goal is reached
Note that this procedure is only possible if treatment
effectiveness is known (i.e., an acceptable Accident
Modification Factor or Crash Reduction Factor exists).
Tools That Can Help In These
Roadway-Oriented Processes
High-Crash Location Analysis


States (and some localities) have a procedure
in place to identify and analyze high-crash
locations
FHWA is developing SafetyAnalyst –improved
procedures for choosing sites, treatments,
and evaluating the results
http://www.safetyanalyst.org/
SafetyAnalyst
To assist agencies in making better
decisions about:



Where to make highway safety improvements
Determining what improvements to make
Evaluating the effectiveness of implemented
safety improvements
Features
Comprehensive – Integrates all parts of
the safety management process and
automates portions of the process that are
performed manually.
Applies state-of-the-art procedures in the
safety management process.
Has strong cost-effectiveness component.
SafetyAnalyst Tools
Network screening to identify sites with potential
for safety improvement
Diagnosis of the nature of safety problems
Selection of countermeasures
Economic appraisal of countermeasures
Priority ranking of improvement projects
Evaluation of implemented projects
Final version and training – Late 2006
Supplementing the High-Crash Location
Process
AASHTO – set a “stretch goal” for fatality and
injury reductions
In most cases, this will probably not be met with
the high-crash location program only
Will need to treat longer corridors or route-types
Given an Emphasis Area has been chosen (e.g.,
lane-departure crashes), the Data Guide presents
procedures for choosing among potential
treatment strategies and targeting the strategies
Supplementing the High-Crash Location
Process (Cont)
Data Guide Procedures will cover three levels of safety
data and treatment effectiveness



Known treatment effectiveness combined with linkable crash,
roadway inventory and traffic data;
Known treatment effectiveness and crash data only (e.g., no
inventory data) for both mileposted and un-mileposted crash
data
Unknown treatment effectiveness and crash data only
The first two are based on a “reverse economic analysis”
procedure developed by Tom Bryer as part of the
AASHTO SHSP efforts.
Note that these first two could also be used for driver
and vehicle-based programs, but only if treatment
effectiveness is known – not often the case.
Two Addition SHSP Analysis Steps
4. Optimizing your safety budget – choosing the
best mix of efforts (e.g., emphasis areas,
locations) within your budget
5. Evaluating the effects of your program
(Don’t worry – only limited coverage for both!)
Budget Optimization
Can be done at different budget levels
Can be used in decisions of basic choice
between emphasis areas down to choice
of locations/subpopulations
Report 501 provides a detailed description
in Appendix D.1.6, and a set of
computerized tools on a diskette.
Note that this requires known Accident
Modification Factors
Program/Project Evaluation
Report 501 describes both process and impact
evaluations and procedures that can be used,
along with case studies (Appendix D3)
Also provides overview of impact (crash-based)
evaluation methodologies in Appendix D2
(cross-sectional studies, before/after, EB before
after, time series, etc.)
NHTSA evaluation assistance
Other Resources for SHSP
Developers
NHTSA will assist in FARS analysis
NHTSA course – “Data Analysis in Highway Safety
Problem Identification and Program Evaluation”.

See NHTSA Region staff
NHTSA is developing on-line training modules
concerning all components of a traffic records system
and basic analysis of these data

On line this month www.trafficrecords101.net
FHWA course on “New Approaches to Safety Analyses”

National Highway Institute
FMCSA Division Offices will provide analysis assistance
Highway Safety Manual (see
http://www.highwaysafetymanual.org/)

Will present best “safety knowledge” and “analysis tools” for
roadway-safety programs (First edition – 2007)
Other Resources (cont)
Traffic Safety Information Systems International Scan:
Strategy Implementation White Paper


Suggestions for improving both crash and non-crash (e.g.,
inventory) safety data
Recommendations for FHWA, NHTSA, AASHTO and states
Increase use of high-end automated crash-data collection systems
(e.g., TraCs) and in-car computers with GPS capabilities
Conduct a Traffic Records Assessment using external reviewers
(and improve the assessment of roadway-related data)
If a state’s tort system does not allow “failure to improve” litigation,
the state safety data managers should recruit DOT attorneys as
supporters for improved safety data
Establish a “data user/owner” committee that would include a wellstaffed secretariat and representatives of all potentially linkable noncrash databases.
Establish “data-for-data partnerships” with local agencies as a
possible avenue to increase collection of improved safety data.

Final Draft to FHWA by January, 2006
Selected Questions from
Conference Program Team
Are data issues, e.g. needs, deficiencies,
requirements, etc., addressed as an
emphasis area in the plans? Should
they be?

They clearly should be since (1) data is
the key now and forever will be the key to
safety program planning, and (2) there is
money available to improve data.
Selected Questions (cont)
What is the “state-of-the-art” in technology and
computerized systems?

Crash data collection:
TraCs or other on-board computer equipment equipped with GPS.

Roadway inventory:
Roadway cross-section inventory for full-jurisdiction system updated
“immediately” from as-built plans.
Roadway curvature and grade from as-built plans or from FHWA’s
new “Dynamic Highway Measurement” vehicle.
Intersection and interchange inventory including crossroad traffic
volumes.
All roadway inventory files with both LRS and geo-coding.

Traffic counts:
Standard counting system with 2-3 year system-wide updates;
Truck counts (vehicle classification counts) with increased numbers
of count points enhanced with WIM and other available data.

Data storage and linkage:
Data warehousing, where all different types of safety files and data
are stored in one warehouse and linked easily to each other.
Questions?