Session

1
Session 4, Forrest Council, Slides
2
The Potential Impacts of a Towaway Reporting
Threshold on Driver/User and Roadway Safety
Programs

• Forrest M. Council
• Senior Research Scientist
• UNC Highway Safety Research Center

3
Information from Two Studies

• Effects of a Towaway Reporting Threshold on
  Crash Analysis Results, (Zegeer, et al., TRR
  1635, 1998)
• A Review Of The Impacts Of The Towaway Reporting
  Threshold On North Carolinas Highway Safety
  Program, (Lacy, et al., 2001)
• Both conducted under FHWAs Highway Safety
  Information System (HSIS) Project

4
Why HSIS Interest?

• HSIS is a multi-year, multi-state data system
  (1985-present)
• Crash, roadway inventory, traffic, supplemental
  files
• Annual files from CA, IL, ME, (MI), MN, OH, NC,
  UT, WA
• Data used by FHWA and provided to many other
  national research projects
• www.hsrc.unc.edu/hsis/ or through FHWA site
• Any change in threshold will affect HSIS data
• A primary goal of HSIS is to improve safety data

5

• The two studies analyze two basic issues
• What is the general effect on key crash variables
  used in driver, vehicle and roadway research if
  one moves to a towaway threshold?
• Are there specific effects on the roadway-related
  analysis programs conducted by state and local
  roadway agencies?
• Identification of Sites with Promise (i.e.,
  high-hazard locations)
• Site-specific analysis of crash patterns to
  define most appropriate treatment

6
Methods

• General Effects study
• Compare crash frequencies for towaway vs. current
  PDO level in four HSIS states (IL, MN, MI, NC)
• Examine whats lost by roadway type, crash
  type, object struck, vehicle type
• Redevelop crash prediction model using only
  towaway crashes

7
Methods (cont)

• Site-Analysis Effects study (NC data)
• Compare high crash location rankings based on
  PDO and then towaway
• Compare crash patterns on collision diagrams
• Questionnaire to safety engineers who used both
  thresholds in developing collision diagrams

8
Results Percent Crashes Lost
9
Percent Crashes Lost by Roadway Class
10
PERCENT OF TOWAWAY CRASHES BY ACCIDENT TYPE
100.0
90.0
80.0
70.0
60.0
IL-Towaway
MI-Towaway
50.0
MN-Towaway
PERCENT
NC-Towaway
40.0
30.0
20.0
10.0
0.0
Animal
Pedestrian/
Bicyclist
Run-off-road/
Rollover
Angle/Turning
Parking/Backing
Rear-end/
Sideswipe Same
Run-off-road fixed
object
Head-on/ Opposite
Dir.
ACCIDENT TYPE
11
PERCENT TOWAWAY BY FIXED OBJECT
100.0
90.0
80.0
70.0
60.0
50.0
PERCENT
IL-Towaway
MI-Towaway
40.0
MN-Towaway
30.0
NC-Towaway
20.0
10.0
0.0
Sign
Tree
Fence
Mailbox
Guardrail
Median barrier
Streetlight, utility
pole
Ditch, embankment
FIXED OBJECT
12
PERCENT TOWAWAY BY VEHICLE TYPE
100.0
90.0
80.0
70.0
60.0
PERCENT
50.0
IL-Towaway
40.0
MI-Towaway
MN-Towaway
30.0
NC-Towaway
20.0
10.0
0.0
Van
Bus
Pickup truck
Passenger car
Single unit truck
Farm equipment
Truck tractor trailer
Motorcycle/ moped
VEHICLE TYPE
13
CRASH RATES FOR TOWAWAY VS. TOTAL CRASHES
URBAN ROAD CLASSES
8.73
9
8
7
5.77
5.75
6
5.41
4.8
4.78
5
3.94
CRASH RATE (Crashes per 1.61 million vehicle km)
3.66
4
3.56
2.79
2.71
3
2.61
2.49
2.34
2.09
2.02
2.01
1.82
1.77
1.73
1.69
2
1.62
1.57
1.49
1.34
1.1
1.04
0.87
0.77
0.59
1
0.65
0.36
0
IL
MI
MN
NC
IL
MI
MN
NC
IL
MI
MN
NC
IL
MI
MN
NC
Urban Freeways Urban
Divided Multi-lane Urban Undivided
Multi-lane Urban Two-lane
Total
Towaway
14
Results Model Comparisons

• Model predicting crashes/mi on rural multilane
  roads (MN data)
• Original Poisson model for total reported crashes
• New model for towaway crashes only
• Not much difference
• Predicted frequencies were lower (of course)
• Access Control became non-significant predictor
• Difference in old and new frequencies much
  greater for non-principal municipal arterials
  than for other classes

15
Results Site Analyses

• Ranking of high crash locations (based on three
  warrants)
• 50 of top 200 intersections and 50 of top
  200 sections changed
• Significant shuffling of ranking within the list
  (only 4 intersections and 3 sections retained
  same ranking)
• Does it matter? (See later Miller study)

16
Results Site Analysis

• Collision diagrams
• 14 (7 pairs) of diagrams produced and presented
  in random order
• 12 experienced DOT safety engineers circled
  definable patterns and starred those that were
  serious and potentially treatable
• (Note that approximately 1/3 of crashes lost
  with towaway threshold)

17
(No Transcript)
18
Results Site Analysis (cont)

• Collision diagrams
• Patterns identified decreased by 35 (247 vs.
  160)
• Serious patterns decreased by 41 (126 vs. 75)
• Crash patterns identified changed
• Turning patterns decreased 43
• Rear-end/swipe patterns decreased 67
• Thus, countermeasures chosen would both decrease
  and change
• Note that other states may lose more crashes
  and patterns

19
Results Other Studies

• Miller, et al., Sensitivity of a Highway Safety
  Resource Allocation Model to Variations in
  Benefit Computation Parameters, TRR 1124, 1987.
• For higher safety improvement budget (1.2 - 1.5
  mill), 20-30 percent of program benefits could be
  lost by less reporting, among other factors
• Pfefer and Raub, Economic Analysis of Highway
  Safety Data Final Report, FHWA, 1995.
• Since officer is on the scene for other duties,
  cost of reporting PDO crash could be as low as
  20

20
Conclusions

• Move to towaway threshold will significantly
  affect safety data
• High loses of some crash types which will affect
  driver, vehicle and roadway problem
  identification
• Major changes to HAL rankings how much
  effect?
• Significant loss of site-based crash patterns
• Cant we improve efficiency by other means, such
  as new technology?
• Is the loss of the data worth the relatively low
  cost of collecting it?