845-851, 1995
Copyright 0 1995 Elsevier Science Ltd
Printed in the USA. All rights reserved
cool-4575/95 $9.50 + 0.00
Pergamon
OOOl-4575(95)00027-5
F. PREUSSER,
ALLAN
F. WILLIAMS~*
and
PRG Inc., 7100 Main Street, Trumbull, CT 06611, U.S.A.; 21nsurance Institute
Glebe Road, Arlington, VA 22201, U.S.A.
CRASHES:
ROBERT
G.
for Highway
ULMER~
Abstract-There
were 2074 crashes fatal to a motorcycle driver in the United States during 1992. A computer
program was developed to convert Fatal Accident Reporting System (FARS) data for these crashes into standard
generated
reports were analyzed and crash type
format English language crash reports. The computer
categories were defined. Five defined crash type categories accounted for 1785 (86%) of the 2074 crash events:
Ran off-road (41%); ran traffic control (18%); oncoming
or head-on (11%); left-turn oncoming (8%); and
motorcyclist
down (7%). Alcohol and excessive speed were common factors associated with motorcyclist
crash
involvement.
Left turns and failure to yield were comrilon factors associated with the involvement
of other
motorists. Suggested countermeasures
include helmet use and enforcement of speed and impaired driving laws.
Keywords-Motorcycle,
Fatal, Crashes,
Classification,
Safety, Countermeasures
Motorcycles,
which are a small subset of all motor
vehicles, are greatly overrepresented
in fatal crashes
in the United States. The death rate per registered
motorcycle
(59 per 100,000) is approximately
three
times the death rate per registered passenger car ( 17
per 100,000) (FHWA 1990). Death rates calculated
per vehicle, however, do not take into account the
substantially
lower mileage traveled by motorcycles.
Per mile traveled, the death rate for motorcycles
is
estimated to be 22 times higher than the comparable
death rate for passenger cars (IIHS 1993a; FHWA
1990).
Many of the available
motorcycle
crash and
injury analyses are based on data that are decades
old. In an examination
of 2410 motorcycle crashes in
North Carolina
during 1972, Griffin (1974) found
that, when a car and motorcycle collide, the motorcyclist nearly always sustains the more severe injury
(p. 44). Griffin also found that the left-turn maneuver
by passenger
cars was greatly overrepresented
in
motorcycle-involved
multiple-vehicle
crashes. Similar
results were reported by Olson (1989) for daytime,
car-motorcycle
crashes in Texas during 1986. Hurt
( 1981) examined 900 motorcycle crashes occurring in
Los Angeles during 1976-1977 and concluded
that
motorists failure to see the motorcycle
in time to
avoid a crash was a major crash causation factor (see
also Williams and Hoffman 1979).
*Author for correspondence.
845
846
crash have been deleted.) The software used to generate these reports was originally developed for an
earlier study in which both the FARS codes and the
original police crash reports were available (IIHS
1993b). The standard format computer generated
reports, while lacking the full information contained
in the original police report, proved to be quite
descriptive, easier to read and much easier to combine
across varying police crash report formats from
different states. Also, as happened frequently in the
earlier study, the computer reports provided data in
cases where the original report was not released by
the state.
Half of the computer generated crash reports
(N= 1037) were reviewed by one analyst who developed a preliminary set of crash type groups and
preliminary definitions. The remaining reports (N =
1037) and the preliminary group definitions were
reviewed by a second analyst. Working together, and
cross-reviewing selected cases from each other, the
two analysts finalized the crash type definitions and
made final crash type assignments for the 2074 crash
events. Computer codes for each of the crash type
definitions were appended to the original FARS
record for each case, and statistical data were
tabulated.
The final crash type list included 10 defined crash
types plus one category for other and unknown.
Figure 2 lists the identified crash types, their respective definitions and FARS codes commonly associated
with each. The assignment of any given crash to any
given crash type relied on analyst agreement that the
characteristics of the event were most consistent with
the definition for the given crash type.
Though not done in this study, it is presumably
possible to entirely automate crash type assignment.
In a related study (Williams et al. 1995) computer
software was developed for crash type assignment.
The results showed approx. 90% agreement between
computer and analyst assigned crash type codes. In
general, the analyst was using more information,
better sequencing of information and better weighting
of information to develop an understanding of crash
dynamics. Additional development of the software
would be needed to improve on the 90% figure
cited above.
RESULTS
Of the 10 crash types considered, the most frequently occurring crash type was ran off-road, followed by ran traffic control, oncoming (i.e. head-on),
left-turn oncoming, and motorcyclist down. Taken
together, these five most frequent types accounted for
86% of the 2074 crashes. Table 1 shows the distribu-
Brief Communications
847
On Thursday [Date] at I:00 am, a motorcycle operator was killed in a 2 vehicle crash in [County,
State]. The crash site [Street Name] was an urban principal arterial, with undivided two-way
traffic and 4 travel lanes. The roadway was straight and level with a 40 mph speed limit. The
crash was not at an intersection and occurred on the roadway. There were no traffic controls.
It was dark but lighted, the weather was clear, the road surface was dry. The crash involved
collision with a motor vehicle in transport. This was a head-on collision.
Vehicle I
registered
43 mph.
maneuver
was a 1978 Honda Motorcycle (350-449cc). The operator was alone. The vehicle was
in [State]. The driver in the crash was not the vehicle owner. Estimated speed was
The vehicle was going straight prior to the crash. Possibility of crash avoidance
was unknown. This was a striking vehicle (impact point 12).
Driver License: Issuing state was [State], license status was suspended.
Driver Record:
Previous Crashes: 0 Previous Suspensions: 1
Previous Speed: 0
Previous Other Moving: 0
Previous DWI: 1
Previous DWI:
Multiple vehicle, BAC showed alcohol for motorcycle operator, on roadway, urban
Brief Communications
848
Common
Type
single
Typically
not collision
Off road
(includes
Common
Ran Traffic
Control
or
vehicle
with motor
shoulder,
Driver Factor:
vehicle
media,
Failure
in transpolt
roadside,
to keep in proper
parking
lane or running
off road
vehicle
Typically
angle
Typically
intersection
collision
or driveway
related
roadway
Impactpoints
indicate
crash
Common
(or roadway
were approaching
on different
roadways
pnor
tothe
and driveway)
or traffic officers,
veh/cle
Head-on
or sideswipe
opposite
direction
On roadway
Common
LT Oncoming
Driver Factor:
to keep in proper
lane or running
vehicle
Head-on,
sideswipe
Typically
intersection
0 roadway
One vehicle
Impact
opposite
direction
or driveway
or angle
related
points
Common
Motorcyclist
Down
Failure
indicate
Driver Factor:
A motorcyclist
down.
Failure
toyield
were coming
from opposite
directions
right of way
and goes
Non collision
On roadway
Common
Run Down
of the posted
speed
vehicle
Front-to-rear
0
impact
points
roadway
Lead vehicle
Common
traveLng
Driver
at unimpeded
Factor:
Driving
speed
or m excess
of the posted
speed
vehicle
Front-to-rear
On roadway
Lead vehicle
Common
Road Obstacle
or in excess
One vehicle runs down another vehicle traveling in the same direction
striking it in the rear. Unlike Stop/Stopping the vehicle struck in the rear
was traveling at an unimpeded speed prior to the crash.
Multiple
Stop/Stopping
impact
points
s/owing,
stopped
Driver Factor:
or starting
Following
in traffic lane
improperly
sing/e vehicle
Driver
Factor:
with motor
Avoiding,
Fig. 2. ~ continued
veh!cle
in transport
swerving
or sliding
opposite
due
to[objects
in the road]
Brief Communications
and Research
Notes
849
Fig. 2. - continued
Lane Change
cut off
Other/Unknown
Driver Factor:
Table 1. Distribution
Motorcycle
of motorcycle
crash types by single-vehicle
and multiple-vehicle
crashes
crash type
Singlevehicle
crashes
831
83
49
33
996
Multiplevehicle
crashes
All
crashes
26
375
225
176
69
69
66
2
28
25
17
1078
857
375
225
176
152
69
66
51
28
25
50
2074
%
(41.3)
(18.1)
(lO.Sj
(8.5)
(7.3)
(3.3)
(3.2)
(2.5)
(1.4)
(1.2)
(2.4)
(100)
type definitions.
850
Brief Communications
and Research
Notes
use for all other crash types). During the 3 years prior
to the crash, 35% of the fatally injured motorcycle
drivers had been convicted of a previous speeding
violation, 23% had one or more prior license suspensions, 20% had been involved in one or more previous
crashes, and 9% had been convicted of impaired
driving.
Table 2 shows crash type by blood alcohol concentration (BAC) for fatally injured motorcyclists in
states that tested at least 80% of all fatally injured
motorcycle drivers. During 1992, 33 states (and the
District of Columbia) tested 80% or more of their
fatally injured motorcycle drivers for alcohol. Overall,
598 (53%) of the 1120 fatally injured motorcyclists
who were tested were positive for alcohol, and 41%
had BACs of 0.10% or greater. Ran off-road was the
crash type with the most alcohol involvement. For
this crash type, 72% had positive BACs, and 59% had
BACs of 0.10% or greater. Ran traffic control and
left-turn oncoming crashes showed the least motorcyclist alcohol involvement.
DISCUSSION
Although developing crash types using only
FARS data is limited by the unavailability of the
crash diagram and the full narrative description of
the event contained in the investigating officers crash
report, there are also several unique advantages. The
standard training of FARS analysts and the standard
format for all FARS case coding make it possible to
combine data from different states without loss of
accuracy, thus providing national coverage. FARS
also accesses data from several sources providing
additional information beyond the crash report alone.
The most important finding in the present study
was that five defined crash types accounted for 86%
of all of the motorcycle crash events studied. Two of
these types, ran off-road and oncoming, are predominantly the result of one or more errors (i.e. FARS
driver factors) on the part of the motorcyclist. Both
typically involve a motorcyclist who leaves the appropriate travel lane(s) either running off the road or
colliding with a vehicle coming from the opposite
direction. Both tend to occur more frequently in rural
areas, on higher speed roadways and at curves. Ran
off-road crashes are very often alcohol related.
Countermeasures designed to promote helmet use
and reduce drinking and driving, and excessive speed,
would be appropriate.
Ran traffic control and left-turn oncoming
involve an interaction between the motorcyclist and
one or more other drivers. Unlike ran off-road and
oncoming crashes, they occur more often at intersections, in urban areas, during times of the day when
more traffic would be expected, and are less often
Brief Communications
Table 2. Percentages
of most
common
%BAC (N =)
0.00
0.01-0.09
0.10-0.19
0.20 +
All
and Research
On-coming
851
Notes
states
Biker down
with
~80%
Other
BAC tested
types
All
(481)
(190)
(119)
(92)
(90)
(148)
(1120)
28
13
34
25
100
66
16
12
6
100
51
13
16
20
100
13
13
11
3
100
63
10
14
12
100
51
11
24
14
100
41
13
24
17
100
REFERENCES
Cross, K. D.; Fisher, G. A study of bicycle/motor-vehicle
accidents: Identification
of problem types and countermeasure
approaches.
DOT-HS-4-00982.
Washington,
DC: U.S. Department
of Transportation;
1977.
Federal Highway Administration
(FHWA). Highway Statis-