A Comparative Analysis of Reporting Thresholds Using the NHTSA State Data System

1
A Comparative Analysis of Reporting Thresholds
Using the NHTSA State Data System

• Gary Rinehart
• ALCOSYS, Inc.
• Crystal Gateway 3, Ste 910
• 1215 Jefferson Davis HWY
• Arlington, VA 22202-4342
• 703-883-1883

2
Outline

• The NHTSA State Data System (SDS)
• Crash Definitions
• Purpose of Study
• Severity of SDS Crashes
• Description of Towaway Indicator Variables
• Crash Severity Analysis Using Towaway Indicators
• Detailed Property Damage Only Crash Analysis
  Using Selected Variables
• Summary

3
SDS
4
NHTSA State Data System

• A collection of crash data received directly
  from the seventeen states participating in the
  program. NHTSA adheres to strict confidentiality
  agreements with the participating states.
• The data consists of all state-reported crashes.
  The census data are received annually.
• Each state uses a unique data structure. The
  data are converted into a standard SAS format at
  the National Center for Statistics and Analysis
  (NCSA), and used for research purposes.
• NCSA plans to expand the program. The goal is to
  ultimately have all 50 states participate.

5
State Data System

• California
• Florida
• Georgia
• Illinois
• Indiana
• Kansas
• Maryland
• Michigan
• Missouri

• New Mexico
• North Carolina
• Ohio
• Pennsylvania
• Texas
• Utah
• Virginia
• Washington

6
Definitions

• Fatal Crash A crash in which at least one
  fatality occurs.
• Injury Crash A crash in which at least one
  injury, but no fatalities, occurs.
• Property Damage Only (PDO) Crash A crash
  without fatalities or injuries.
• PDO-Towaway A PDO crash in which at least one
  vehicle was towed, generally indicative of a
  severe PDO crash.
• PDO No Towaway A PDO crash in which no
  vehicles were towed.
• Towaway Criterion A reporting threshold where
  only fatal, injury, and PDO-towaway crashes are
  reported.

7
Purpose of Study

• Assuming that states adopt a towaway criterion
  for reportable crashes, this study attempts to
  answer the following questions
• What proportion of crashes will be lost?
• A crash severity analysis was conducted using
  sixteen states and appropriate towaway indicator
  variables.

8
Purpose of Study (cont.)

• What changes in the basic composition of the
  crash data will occur?
• Since fatal and injury crash data will not be
  affected by the towaway criterion, the emphasis
  of the research was on the change in composition
  of PDO crashes.

9
Analysis Variables

• The following variables from six states were used
  in the analysis
• Urban/Rural Crash
• Single/Multiple Vehicle Crash
• First Harmful Event
• Manner of Collision (Motor Vehicle in Transport
  Crashes)
• Vehicle Type

10
Severity of SDS Crashes
11
Fatal Crashes
12
Fatal Crashes (cont.)
13
Injury Crashes
14
Injury Crashes (cont.)
15
PDO Crashes
16
PDO Crashes (cont.)
17
Variability in Crash Composition
The only state with a true towaway criterion,
Pennsylvania has the highest percentage of injury
crashes and the lowest percentage of PDO crashes.
18
SDS Towing Indicator Variables

• Towaway
• Damage Severity
• Damage Scale

19
Towaway

• California
• Georgia
• Illinois
• Indiana

• Missouri
• North Carolina
• Ohio
• Pennsylvania

20
Towaway Typical Variable Values

• Towed / Not Drivable
• Not Towed / Drivable
• Remained at Scene (optional)

21
Damage Severity

• Florida
• Georgia
• Kansas

• Maryland
• New Mexico
• Washington

22
Damage Severity Typical Variable Values

• Destroyed, Totaled (optional)
• Disabling Damage Any damage to a motor vehicle
  such that either it cannot be driven, or if
  driven, it would be further damaged.
• Functional Damage Any damage to a motor vehicle
  that affects its operation or the functioning of
  its parts, but is not disabling.
• Other Vehicle Damage Any damage to a motor
  vehicle, which is neither disabling nor
  interferes with the function of the vehicle
  (cosmetic).
• No Damage.

23
Damage Scale

• Michigan
• Texas
• Virginia (repair cost)

24
TAD Scale Typical Variable Values

• Minor damage, less than element value 2.
• Minor damage, limited to dents and gouges in body
  sheet metal or trim.
• Moderate damage, less than element value 4.
• Moderate damage, considerable crumpling of body
  sheet metal, little or no distortion of basic
  structure or frame. (Towaway at this level and
  above.)
• Severe damage, crumpled or torn sheet metal, less
  than element value 6.
• Damage severe, not total. Sheet metal severely
  distorted, crumpled or torn.
• Vehicle totally damaged.

25
Example of TAD 1
Michigan Traffic Accident Report, 8/87
Reporting Thresholds
Source SDS data, 1990-1999
25
26
Example of TAD 4
Michigan Traffic Accident Report, 8/87
Reporting Thresholds
Source SDS data, 1990-1999
26
27
Example of TAD 7
Michigan Traffic Accident Report, 8/87
Reporting Thresholds
Source SDS data, 1990-1999
27
28
Towaway Indicator Variables

• Using Towaway indicator variables, the Property
  Damage Only crashes can be divided into three
  categories
• Towaway
• No Towaway
• Towaway Status Unknown
• A revised Crash distribution can then be
  determined.

29
Crash Severity(sample size 30.3 million)
30
Crashes Lost

• Using a towaway criterion
• Approximately 42 of all crashes would be lost.
• Approximately 70 of property damage only crashes
  would be lost.

31
Contribution by State (sample size 30.3 million)
Utah does not have a towaway indicator variable.
32
Crash Severity by State
33
Crash Severity by State (cont.)
34
Crash Severity by State (cont.)
35
Towaway States?

• PA Data confirms that Pennsylvania does use a
  towaway criterion hence the high percentage of
  injury crashes and the low percentage of PDO
  crashes.
• MD Although Maryland has a towaway policy,
  Maryland data appears similar to states without a
  towaway criterion. At least 28 of crashes do
  not meet towaway criterion.
• FL Crashes reported by Florida are at the
  officers discretion. Data suggests that
  Floridas reporting criterion may approximate a
  towaway criterion.

36
Crashes Lost Using a Towaway Criterion
Assuming large percentage of unknowns for
Washington are PDO-Not Towaway
37
Analysis of Property Damage Only Crashes

• Methodology involves comparison of three data
  sets
• PDO crashes from CA, FL, MO, NC, and OH
• PDO-Towaway crashes from CA, FL, MO, NC, and OH
• PDO-Towaway crashes from PA

38
Variables Used

• The composition of each data set was examined
  using the following variables
• Urban/Rural Crash
• Single/Multiple Vehicle Crash
• First Harmful Event
• Manner of Collision (Motor Vehicle in Transport
  Crashes)
• Vehicle Type

39
State Selection Criteria

• Presence of towaway indicator variables over an
  extended period
• CA, FL, MO, OH, PA 1990-1999
• NC 1992-1999
• Presence of Rural/Urban and/or Population Group
  variables.
• Presence of First Harmful Event and Manner of
  Collision variables with categories approximating
  the Model Minimum Uniform Crash Criteria (MMUCC).
• Low percentage of unknowns for analysis variables.

40
Contribution by State(Sample Size 8.5 million)
41
Crash Type
PDO Crashes (CA, FL, MO, NC, OH) Sample size
8.5 million. PDO-Towaway Crashes (CA, FL, MO, NC,
OH) Sample size 2.7 million. PDO-Towaway
Crashes (PA) Sample size 0.48 million.
42
Population Density
PDO Crashes (CA, FL, MO, NC, OH) Sample size
8.5 million. PDO-Towaway Crashes (CA, FL, MO, NC,
OH) Sample size 2.7 million. PDO-Towaway
Crashes (PA) Sample size 0.48 million.
43
Population Density
PDO Crashes (CA, FL, MO, NC, OH) Sample size
8.5 million. PDO-Towaway Crashes (CA, FL, MO, NC,
OH) Sample size 2.7 million. PDO-Towaway
Crashes (PA) Sample size 0.48 million.
44
First Harmful Event
PDO Crashes (CA, FL, MO, NC, OH) Sample size
8.5 million. PDO-Towaway Crashes (CA, FL, MO, NC,
OH) Sample size 2.7 million. PDO-Towaway
Crashes (PA) Sample size 0.48 million.
45
First Harmful Event (cont.)
PDO Crashes (CA, FL, MO, NC, OH) Sample size
8.5 million. PDO-Towaway Crashes (CA, FL, MO, NC,
OH) Sample size 2.7 million. PDO-Towaway
Crashes (PA) Sample size 0.48 million.
46
Manner of Collision(Motor Vehicle in Transport)
PDO Crashes (CA, FL, MO, NC, OH) Sample size
5.6 million. PDO-Towaway Crashes (CA, FL, MO, NC,
OH) Sample size 1.5 million. PDO-Towaway
Crashes (PA) Sample size 0.25 million.
47
PDO Vehicle DistributionCA, FL, MO, NC, OH
Sample size 15.8 million vehicles.
48
PDO-Towaway Vehicle Distribution CA, FL, MO, NC,
OH
Sample size 4.7 million vehicles.
49
PDO-Towaway Vehicle Distribution PA
Sample size 0.77 million vehicles.
50
Vehicle Type Comparison
PDO Crashes (CA, FL, MO, NC, OH) Sample size
15.8 million. PDO-Towaway Crashes (CA, FL, MO,
NC, OH) Sample size 4.7 million. PDO-Towaway
Crashes (PA) Sample size 0.77 million.
51
Percent Vehicles Retained PDO-Towaway (CA, FL,
MO, NC, OH)
Percentage Retained Number of Vehicles
(PDO-Towaway) ? Number of Vehicles (PDO)
52
Changes in Vehicle Type PDO-Towaway (CA, FL, MO,
NC, OH)

• Small increase in the proportion of passenger
  cars and light trucks.
• Small decrease in the proportion of large trucks
  and motorcycles.
• Moderate decrease in proportion of buses and
  other vehicles (motor homes, farm/construction
  equipment, police/emergency vehicles, etc.).

53
Summary

• Using the NHTSA State Data System, adoption of a
  towaway criterion would have the following
  effects
•  
• A significant decrease in the total number of
  reportable crashes.
• The data shows that at least 42 of the
  reportable crashes would be lost.
• The data shows that at least 70 of the
  reportable PDO crashes would be lost.

54
Summary (cont.)

• Low severity PDO crashes would be lost, changing
  the composition of the remaining PDO crashes as
  follows
• The proportion of Urban crashes decreases.
• The proportion of Single Vehicle crashes
  increases, with an increase in Fixed Object,
  Rollover, and Other Non-Collision crashes.
• The proportion of Multiple Vehicle crashes
  decreases, both for Motor Vehicle in Transport
  and Parked Motor Vehicle crashes.
• The proportions of Angle and Head-on collisions
  increase, while Rear End and Sideswipe collisions
  decrease.
• The proportions of Passenger Cars and Light
  Trucks increase slightly, Large Trucks and
  Motorcycles decrease slightly, and Buses and
  Other Vehicles exhibit a moderate decrease.

55
Questions or Comments?

• Questions/Comments can be sent to
• gary.rinehart._at_nhtsa.dot.gov