[Federal Register Volume 64, Number 96 (Wednesday, May 19, 1999)]
[Notices]
[Pages 27343-27353]
From the Federal Register Online via the Government Publishing Office [www.gpo.gov]
[FR Doc No: 99-12579]
[[Page 27343]]
-----------------------------------------------------------------------
DEPARTMENT OF TRANSPORTATION
National Highway Traffic Safety Administration
Denial of Motor Vehicle Defect Petition, DP97-006
AGENCY: National Highway Traffic Safety Administration (NHTSA),
Department of Transportation.
ACTION: Denial of petition for a defect investigation.
-----------------------------------------------------------------------
SUMMARY: This notice sets forth the reasons for the denial of a
petition submitted to NHTSA under 49 U.S.C. 30162, requesting that the
agency commence a proceeding to determine the existence of a defect
related to motor vehicle safety. The petition is hereinafter identified
as DP97-006.
FOR FURTHER INFORMATION CONTACT: Dr. George Chiang, Office of Defects
Investigation (ODI), NHTSA, 400 Seventh Street, SW, Washington, DC
20590. Telephone: (202) 366-5206.
SUPPLEMENTARY INFORMATION: Edgar F. Heiskell, III (petitioner),
Attorney at Law, 400 Bank One Center, P.O. Box 3761, Charleston, West
Virginia 25337-3761, submitted a petition to the National Highway
Traffic Safety Administration (NHTSA) by letter dated December 3, 1997,
requesting that an investigation be initiated to determine whether to
issue an order concerning the notification and remedy of a defect in
model year 1983 through 1990 Bronco II sport utility vehicles (subject
vehicles) manufactured by Ford Motor Company (Ford) because of concerns
related to their rollover propensity.
The petitioner alleges that the subject vehicles were ``designed
with handling and stability defects which have caused an extraordinary
number of rollover accidents resulting in thousands of deaths and
severe injuries.''
NHTSA has reviewed all information brought to its attention and
reviewed crash databases and Office of Defects Investigation's consumer
complaint database. The results of this review and analysis are set
forth in a Petition Analysis Report for DP97-006, which is published in
its entirety as an appendix to this notice.
For the reasons presented in the petition analysis report, there is
no reasonable possibility that an order concerning the notification and
remedy of a safety-related defect in the subject vehicles would be
issued at the conclusion of an investigation. Therefore, in view of the
need to allocate and prioritize NHTSA's limited resources to best
accomplish the agency's safety mission, the petition is denied.
Authority: 49 U.S.C. 30162(d); delegations of authority at CFR
1.50 and 501.8.
Issued on: May 14, 1999.
Kenneth N. Weinstein,
Associate Administrator for Safety Assurance.
Appendix--Petition Analysis--DP97-006
1.0 Introduction
Edgar F. Heiskell, III (petitioner), Attorney at Law, 400 Bank
One Center, P.O. Box 3761, Charleston, West Virginia 25337-3761,
submitted a petition to the National Highway Traffic Safety
Administration (NHTSA) by letter dated December 3, 1997, requesting
that an investigation be initiated to determine whether to issue an
order concerning the notification and remedy of a defect in model
year 1983 through 1990 Bronco II sport utility vehicles (subject
vehicles) manufactured by Ford Motor Company (Ford) because of
concerns related to their rollover propensity.
2.0 Previous Inquiries and Investigations by NHTSA Into Alleged
Rollover Defects
In October 1979 and July 1981, NHTSA's Office of Defects
Investigation (ODI) received two petitions (DP80-002 and DP81-018)
for defect investigations into the alleged instability of Jeep CJ
vehicles. Both these petitions were denied due to the lack of
specific information indicating that there was a defect that caused
the vehicles to roll over.
In 1988, ODI received two petitions for defect investigations
into the alleged rollover propensity of 1986 through 1988 Suzuki
Samurai vehicles, including the convertible, the Samurai, and the
SJ410 and LJ80 models (DP88-011 and DP88-019). NHTSA also denied
these petitions, primarily because the available information did not
show that the alleged rollovers were caused by a defect in the
vehicle rather than by the driver and/or environmental factors.
In 1989, ODI conducted investigation EA89-013 concerning 1984-
1989 Ford Bronco II sport utility vehicles. This investigation was
opened in response to a defect petition, DP88-020. A peer analysis
of rollover rates showed the Bronco II to be similar to other sport
utility vehicles, as measured using the metric of first-event
single-vehicle rollovers per single-vehicle crash. ODI closed this
investigation in October 1990, because ``there appears no reasonable
expectation that further investigation would lead to a determination
of the existence of a safety-related defect with respect to any of
the allegations regarding the propensity of the Bronco II to roll
over.'' Also during this same time period, ODI was petitioned again
to investigate Jeep CJ models (DP90-012). This petition was also
denied for the same reasons as the Bronco II petition.
In 1996, ODI was petitioned to open a defect investigation into
the rollover propensity of the 1986-1995 Suzuki Samurai convertible
(DP96-004). The petitioner alleged that Samurai convertibles have
high rollover propensity, as reflected by their low static stability
factor (the track width to center of gravity ratio), and, when
loaded with occupants, the vehicle is even less stable. After
reviewing the materials presented in that petition and other
available data and information, the agency concluded that it was
unlikely that further investigation of alleged Samurai convertible
rollover propensity would enable NHTSA to identify a safety-related
defect. The petition was therefore denied.
In August 1996, ODI received a petition (DP96-011) from
Consumers Union of the United States (CU) to investigate 1995 and
1996 Isuzu Trooper and Acura SLX sport utility vehicles because of
their alleged propensity to roll over in a reverse steer maneuver.
CU alleged that these vehicles were prone to tip-up during a double
lane maneuver known as the CU ``short course.'' CU's testing of peer
vehicles indicated different performance for the peer vehicles
compared to the Trooper and SLX. NHTSA conducted crash data
analysis, a computer simulation, and a comprehensive test program
comparing these vehicles and a peer vehicle during its analysis of
the petition. NHTSA testing showed that the results of tests on the
CU short course were not repeatable and were affected by driver
performance. When these driver performance inconsistencies were
accounted for, the Trooper and SLX performed similarly to the peer
vehicles during testing using the CU short course. This petition was
denied.
3.0 Vehicle Inforamtion
3.1 Subject Vehicle Description
The Ford Bronco II is a light utility vehicle, i.e., a
multipurpose passenger vehicle having a wheelbase of 110 inches or
less and special features for occasional off-road use, and was
originally introduced for sale in the United States in late 1983 as
a 1984 model year vehicle. It continued in production through the
1990 model year. It is a two-door, four-passenger vehicle with body-
on-frame construction, a 94 inch wheelbase and a 56.9 inch track
width (front and rear). The vehicle was equipped with front coil and
rear leaf springs and a front-mounted engine throughout its
production. All 1984-1986 model year Bronco II vehicles were
equipped with four-wheel drive. Beginning with the 1987 model year
and through the remainder of its production, the Bronco II was also
available in a two-wheel drive configuration.
3.2 Vehicles Involved
Table 1 presents the number of subject vehicles sold in the
United States.
[[Page 27344]]
Table 1.--Sales of Subject Vehicles in the United States
----------------------------------------------------------------------------------------------------------------
Model year 4X4 4X2 Total
----------------------------------------------------------------------------------------------------------------
1984............................................................ 144,061 0 144,061
1985............................................................ 98,153 0 98,153
1986............................................................ 109,846 0 109,846
1987............................................................ 88,818 22,286 111,104
1988............................................................ 109,524 38,201 147,725
1989............................................................ 67,356 29,835 97,191
1990............................................................ 38,451 16,445 54,896
-----------------------------------------------
Grand Total................................................. .............. .............. 762,976
----------------------------------------------------------------------------------------------------------------
4.0 Alleged Defect
The petitioner alleges that the subject vehicles were ``designed
with handling and stability defects which have caused an
extraordinary number of rollover accidents resulting in thousands of
deaths and severe injuries.''
5.0 Complaints
5.1 Complaints to ODI Concerning the Subject Vehicles' Rollover
Propensity
ODI has reviewed all owner complaints in the ODI database that
may be related to the alleged defect in the subject vehicles.
Each complaint in the ODI database is given ``Fault'' codes for
``cause'' and ``result.'' Each complaint that had a ``cause fault''
or ``result fault'' of ``rollover'' was individually reviewed to
eliminate duplications and non-rollovers.
Figure 1 shows the number of rollover-related complaints
regarding the subject vehicles received by ODI during each calendar
year from 1989 through 1998. It indicates that after EA89-013 was
closed on October 31, 1990, the number of such complaints to ODI
decreased sharply.
BILLING CODE 4910-59-P
Figure 1. Number of Bronco II Rollover Complaints Received by ODI:
1989 through 1998.
[GRAPHIC] [TIFF OMITTED] TN19MY99.022
BILLING CODE 4910-59-C
5.2 Complaints to ODI Concerning the Rollover Propensity of Peer
Vehicles
ODI has also reviewed rollover-related owner complaints in the
ODI database regarding certain peer vehicles. Figure 2 shows the
complaint rate (the number of complaints per 100,000 vehicles sold)
based on the complaints received by ODI since January 1, 1994. The
rollover complaint rate of the Bronco II is much lower than those of
many of the peer vehicles, including some for which ODI has recently
denied petitions to open defect investigations.
Figure 2. Complaint Rate of Bronco II and Peers Based on ODI
Complaints Received Since 1/1/94.
[[Page 27345]]
[GRAPHIC] [TIFF OMITTED] TN19MY99.023
BILLING CODE 4910-59-C
6.0 Additional Documents Submitted to Petition File
Prior to submitting his petition, Mr. Heiskell submitted three
letters with attachments, concerning the rollover propensity of the
subject vehicles. In his letters (dated March 13, April 16, and
April 24, 1997), Mr. Heiskell presented arguments and provided
copies of various documents that he claimed set out new information
that would justify reopening EA89-013. Mr. Heiskell's letters and
attachments have been placed in the DP97-006 petition file.
In addition, W. Randolph Barnhart, Esq., submitted documents on
September 10, 1997, which he alleged demonstrated the rollover
susceptibility of the Bronco II. By letter of September 16, 1997,
Mr. Barnhart provided the agency with an index to the previously-
submitted documents. Both of Mr. Barnhart's submissions have been
placed in the DP97-006 petition file.
In their submissions, Mr. Heiskell and Mr. Barnhart focused on
vehicle handling tests conducted by Ford in March/April 1982, during
its development of the Bronco II, in which the test vehicles rolled
over at speeds equal to or greater than 25 mph. They also noted that
Ford had not provided reports of those tests in response to ODI's
information requests in EA89-013. NHTSA has addressed Ford's failure
to provide those test reports during EA89-013 in a May 29, 1998,
letter from John Womack, the agency's Senior Assistant Chief
Counsel. A copy of that letter has been placed in the DP97-006
public file.
ODI has reviewed the reports of the pre-production tests that
were submitted by Mr. Heiskell. While they are clearly relevant to
the issues raised by this petition, they do not in themselves
warrant granting the petition, for the following reasons. The
development of a complex motor vehicle from a concept into a
marketable consumer product involves a process of design, testing,
and an evaluation of test results. Generally, this leads to a cycle
of re-design, re-testing, and re-evaluation, which is repeated until
the product meets its performance objectives. When tests conducted
during product development disclose a potential problem of any type,
a manufacturer generally will take steps to resolve the problem.
When viewed from a defect investigation perspective, the fact
that the test reports suggest a relatively high rollover propensity
in pre-production Bronco II vehicles illustrates the extent of the
problem at the pre-production stage. A variety of modifications were
made to the Bronco II after those tests that were likely to affect
its rollover propensity to some degree. Thus, the in-service history
of the Bronco II with respect to rollover incidents is far more
significant than developmental and pre-production testing.
7.0 Crash Data Analysis
ODI and the National Center for Statistics and Analysis (NCSA)
have evaluated the rollover performance of a number of light sport
utility vehicles by reviewing and analyzing the crash data obtained
from several databases, including State data, the Fatality Analysis
Reporting System (FARS) data, and the National Automotive Sampling
System (NASS) Crashworthiness Data System (CDS) data. ODI also
reviewed data provided by Ford on January 29, 1998 in connection
with this petition, and data supplied by American Suzuki Motor
Corporation (Suzuki)
[[Page 27346]]
in response to Defect Petition DP96-011, specifically those data
related to fatal on-road rollover crashes.
The subject vehicles were manufactured from the mid 1980s
through 1990, and these vehicles have been in operation and exposed
to the crash environment for many years; therefore, the crash data
is considered to be mature, representative, and reliable.
7.1 Previous NHTSA Analysis of Bronco II Rollover Propensity
As noted above, EA89-013 was an investigation into the rollover
propensity of Bronco II vehicles. In that investigation, NHTSA
applied logistic regression to the state data covering 11 groups of
vehicles in order to obtain the ratio of first-event single-vehicle
rollovers to all single-vehicle crashes of each group. The
analytical procedure accounted for environmental factors, such as
the location of the incident (e.g., rural vs. urban; straight vs.
curved road), and driver characteristics, such as age and sex. By
considering these variables, the rollover rate data were controlled
to normalize the vehicles to a common set of outside-the-vehicle
factors that can influence crash outcome. The results of that
analysis, taken from the EA89-013 Closing Report, are depicted in
Figure 3 and give the best estimate of the controlled first-event
single-vehicle rollover rate for single-vehicle crashes for each
vehicle group, along with the upper and lower 95 percent confidence
intervals for crash years 1986 through 1988. For this analysis,
Maryland, Michigan, New Mexico, and Utah data were combined.
Figure 3 shows that the Bronco II has a first-event single-
vehicle rollover rate similar to several other vehicles, notably
CJ5/6/7 (71-80), Toyota 4Runner, CJ5/7/8 (81-86), Suzuki Samurai,
Isuzu Trooper II, and GM S-10/S-15.
BILLING CODE 4910-59-P
[GRAPHIC] [TIFF OMITTED] TN19MY99.024
BILLING CODE 4910-59-C
7.2 1993 Analysis of Michigan Data
In November 1993, an analysis of Michigan state rollover data
was conducted by NHTSA in connection with a proposed rulemaking
effort (Docket 91-68, No. 2, Item 018). In this analysis, rollover
data was computed for crash years 1986 through 1990. Table 2
presents the first-event single-vehicle rollover rates for selected
sport utility vehicles.1 When more than one variation of
a make/model is included, a range of rates is presented. Table 2
shows that the Bronco II has a first-event single-vehicle rollover
rate similar to several other vehicles.
---------------------------------------------------------------------------
\1\ The vehicles were selected to match the vehicles considered
in EA89-013, except the GM S and T vehicles were combined together.
[[Page 27347]]
Table 2.--Michigan First-Event Single-Vehicle Rollover Percentages,
Crash Years: 1986-1990
------------------------------------------------------------------------
Percent of first-
event single-
Make and model* vehicle rollovers
per single-
vehicle crash **
------------------------------------------------------------------------
Ford Bronco II....................................... 39-49
GM S & T series...................................... 24-34
Isuzu Trooper II..................................... 31
Jeep Cherokee........................................ 30
Jeep CJ-5............................................ 49-51
Jeep CJ-7............................................ 46-48
Jeep Wrangler........................................ 28
Suzuki Samurai....................................... 29
Toyota 4Runner....................................... *36
------------------------------------------------------------------------
* Listed Alphabetically; No data for Jeep Wagoneer.
** Make/models with a range represent the upper and lower rates reported
in the Finalized Database of Michigan Data for different variations of
the same make/model. This could include variations with different
badges, e.g., Chevy and GMC; different drive configurations, e.g., 4x4
and 4x2, or different brake systems, e.g., ABS and non-ABS.
The vehicles compared in this table are of similar age and were
fairly new during the time period that the crash data were
collected. Vehicle age can affect performance due to change of
components, such as new tires, wheels, and shocks absorbers, which
may not be the same size or quality as the original ones.
Additionally, as a vehicle ages, components on the vehicle wear,
which can change the performance characteristics of the vehicle and
its susceptibility to rollover.
7.3 1998 Analysis of NASS/CDS Data
Following receipt of Mr. Heiskell's petition, NHTSA analyzed the
NASS/CDS data files for NASS years 1988 through 1996. The vehicles
analyzed were similar to those considered in the EA89-013 analysis,
except the GM S and T vehicles were combined to compare them with
the Bronco II data.
Again first-event single-vehicle rollovers and all single-
vehicle crashes were considered, which exclude not only crashes with
other vehicles, but also with moving objects such as animals,
pedestrians, and bicycles. The data are presented in Table 3. The
range of model years included in the analysis was 1984 through 1990,
except for the Suzuki Samurai, which began production in MY 1986,
and the Jeep CJ vehicles. The model year ranges are noted in the
table. The sample size (listed in the table as ``Number of NASS
Single Vehicle Cases'') is small for all vehicles except the Bronco
II and the GM S&T series. Based on comparison of the T-values, the
Bronco II rollover rate is not statistically significantly greater
than that of any other make/model listed in Table 3, except the Jeep
Cherokee. Furthermore, the Bronco II rollover rate is statistically
significantly lower than that of the Suzuki Samurai. Finally, the
Bronco II's rollover rate is not statistically significantly
different from that of all light trucks and vans considered as a
whole.
Table 3 also provides an estimate of the total number of first-
event single-vehicle rollovers for the time interval analyzed (1988-
1996). For example, the total number of Bronco II first-event
single-vehicle rollovers is estimated to be about 14,000. During
this same time, the total number of first-event single-vehicle
rollovers for the Blazer/Jimmy (GM S&T models) was about 19,000.
[[Page 27348]]
Table 3.--NASS/CDS First-Event Single-Vehicle Rollover Percent, 1988-1996
--------------------------------------------------------------------------------------------------------------------------------------------------------
Percent
Number of Number of first-event
Number of Number of NASS first- first-event First-event SV Percent
Make*/model**/year range NASS single- crashes, event SV SV SV rollover rollovers sample T-value****
vehicle weighted rollover rollovers, sample per single- error
(SV) cases cases weighted error vehicle (SE)***
crash
--------------------------------------------------------------------------------------------------------------------------------------------------------
Ford Bronco II, 84-90........................... 108 38,078 47 13,620 3,697 36 8.6 0
GM S&T series, 84-90............................ 167 77,557 30 18,935 8,875 24 7.2 -1.014
Isuzu Trooper, 84-90............................ 14 4,049 6 2,453 1,877 61 22.0 1.052
Jeep Cherokee, 84-90............................ 32 7,074 6 1,061 523 15 5.6 -2.020
Jeep CJ-5, 71-80................................ 47 19,375 23 12,476 9,840 64 22.9 1.169
Jeep CJ-7, 81-86................................ 30 15,270 10 8,086 4,690 53 16.8 0.910
Jeep Wagoneer, 84-90............................ 6 1,431 1 296 296 21 20.2 -0.687
Suzuki Samurai, 86-90........................... 25 12,829 13 10,712 7,659 84 12.3 3.170
Toyota 4Runner, 84-90........................... 30 9,102 13 4,005 2,605 44 12.4 0.544
Passenger cars, 84-90........................... 5,494 2,260,883 596 256,910 45,658 11 1.0 -2.810
Light trucks and vans, 84-90.................... 1,855 773,115 431 203,131 42,125 26 3.1 -1.036
--------------------------------------------------------------------------------------------------------------------------------------------------------
* Makes listed alphabetically.
** Data for Jeep Wrangler was too small to analyze and is not presented in this table.
*** NASS estimates have a sampling error because NASS is a survey rather than a complete census of all crashes. Rollover rates for two vehicles can be
compared to each other for significant differences using the sample error at the 95 percent confidence interval as follows: If |Ra-Rb| >1.96
multiplied by SQRT [SEa2+SEb2], then they are different; where R=rate, SE=sample error, SQRT=square root, and a & b are different vehicles.
**** The T-value for a vehicle in the table is obtained by: (1) computing the difference between its percentage of first-event SV rollovers and that of
Bronco II, (2) computing the standard error of that difference as the square root of the sum of the squares of the two standard errors involved, and
(3) dividing the difference by its standard error. If the statistic is greater than 1.96, the compared vehicle is statistically different from the
Bronco II at the 95% confidence interval. Positive values of T indicate that the compared vehicle has a greater rollover rate than the Bronco II.
Conversely, negative values of T indicate that the compared vehicle has a lower rollover rate than the Bronco II.
[[Page 27349]]
7.4 Analysis of FARS Data
FARS data were analyzed for first-event single-vehicle rollovers
in all single-vehicle crashes where at least one occupant in the
vehicle was fatally injured. This excluded non-occupant fatal
single-vehicle crashes, such as pedestrian fatalities. FARS years
1984-1996 were included to maximize the size of the sample. The Jeep
CJ-5 and CJ-7 vehicles were not included in the FARS analysis
because they were not produced during the same range of years (1984-
90) as the Bronco II. The model year range for each make/model was
selected to be as similar as possible to that of the subject
vehicle. Unlike the State data described in Section 7.1, these FARS
data are not adjusted and have not been controlled for
environmental, roadway, or driver differences. Table 4 gives the
results of this analysis by number of vehicles involved in fatal
crashes, while Table 5 considers the total number of fatalities
within each fatal vehicle.
Table 4.--Fatal Vehicles in First-Event Single-Vehicle Rollover Crashes
----------------------------------------------------------------------------------------------------------------
Percentage of
Fatal vehicle Fatal vehicle the rollovers Percent
Make and model * Model year single-vehicle first-event SV in fatal standard error
crashes ** rollover single vehicle [SQRT(P * Q/
crashes crashes N)]
----------------------------------------------------------------------------------------------------------------
Chevy/Blazer GMC/Jimmy.......... 84-90 385 168 44 2.53
Ford Bronco II.................. 84-90 1259 762 61 1.38
Isuzu Trooper................... 84-90 99 39 39 4.91
Jeep Cherokee................... 84-90 296 111 38 2.81
Suzuki Samurai.................. 84-90 203 81 40 3.44
Toyota 4Runner.................. 84-90 326 175 54 2.76
----------------------------------------------------------------------------------------------------------------
* Makes listed alphabetically.
** FARS years 1984-1996.
Table 5.--Fatalities in Fatal Vehicles in First-Event Single-Vehicle Rollover Crashes
----------------------------------------------------------------------------------------------------------------
Percentage of
Fatalities in Fatalities in single vehicle Percent
Make and model * Model year single-vehicle first-event SV crash standard error
crashes ** rollover fatalities in [SQRT(P * Q/
crashes the rollovers N)]
----------------------------------------------------------------------------------------------------------------
Chevy/Blazer GMC/Jimmy.......... 84-90 423 183 43 2.41
Ford Bronco II.................. 84-90 1364 823 60 1.32
Isuzu Trooper................... 84-90 109 43 39 4.68
Jeep Cherokee................... 84-90 316 115 36 2.71
Suzuki Samurai.................. 84-90 214 85 40 3.34
Toyota 4Runner.................. 84-90 361 194 54 2.6
----------------------------------------------------------------------------------------------------------------
* Makes listed alphabetically.
** FARS years 1984-1996.
The fifth column of Tables 4 and 5 shows that the Ford Bronco II
has a higher percentage of fatal vehicles and fatalities in first-
event single-vehicle rollovers than that of the other five peer
vehicles, although it is somewhat similar to the Toyota 4Runner.
While the Bronco II rollover rate is not statistically significantly
different from that of its peers, these FARS analyses indicate that
there could be an issue regarding the relative crashworthiness of
Bronco II vehicles in rollovers. In an effort to cast additional
light on the crashworthiness issue, several analyses were performed,
as documented in section 7.5.
7.5 Crashworthiness Analyses
7.5.1. Crashworthiness Aspects
NCSA analyzed the NASS/CDS and FARS data using the same vehicles
listed in the above FARS tables to compute the ratio of the number
of fatally injured occupants in first-event single-vehicle rollover
crashes (from FARS) to the number of involved occupants in such
crashes (from NASS/CDS). Table 6 presents these data.
Table 6.--Crashworthiness Analysis of First-Event Single-Vehicle Rollover Crashes ***
----------------------------------------------------------------------------------------------------------------
Percentage of
Number of fatally
Make/model for model years 1984- involved Number of injured Standard error T difference
1990 * occupants fatalities occupants to of percentage in percentage
(weighted NASS (FARS data) involved **
data) occupants
----------------------------------------------------------------------------------------------------------------
Chevy/Blazer GMC/Jimmy.......... 27,935 183 0.65 0.27 -2.979
Ford Bronco II.................. 17,721 823 4.44 1.24 0
Isuzu Trooper................... 5,334 43 0.80 0.33 -2.831
Jeep Cherokee................... 2,772 115 3.98 1.46 -0.238
Suzuki Samurai.................. 22,199 85 0.38 0.26 -3.195
Toyota 4Runner.................. 9,886 194 1.93 1.10 -1.517
----------------------------------------------------------------------------------------------------------------
* Makes listed alphabetically
** The T-value gives the difference in the percentage of fatalities divided by the standard error of the
difference between the Bronco II and each vehicle. If the absolute value of the statistic is greater than
1.96, the compared vehicle is statistically different from the Bronco II at the 0.05 confidence level.
Negative values indicated that the compared vehicle has a lower percentage of fatalities per involved occupant
than the Bronco II.
*** Involved occupants were from NASS years 1988-1996, and fatally injured occupants were from FARS years 1984-
1996.
[[Page 27350]]
These data indicate that the Bronco II has a significantly
higher percentage of fatally injured occupants per the total number
of involved occupants in first-event single-vehicle rollover crashes
than three of the peer vehicles. However, it has a similar
percentage when compared to the Jeep Cherokee and possibly to the
Toyota 4Runner. This would suggest that if a first-event single-
vehicle rollover occurs, there is more likely to be a fatality in a
Bronco II than in some, but not all, of its peers.
7.5.2 FARS Ejection Path Analysis
To attempt to determine whether the unique design of the rear
side windows in the Bronco II may have affected rollover
crashworthiness, ODI reviewed available data on ejection path,
including 1991-1996 FARS data on ejection path. FARS data prior to
1991 do not include such information.
Since FARS uses police reports to generate the data entered in
the FARS system, it is generally limited to the data contained in
the Police Accident Report (PAR). Most of the time, the ejection
path is not reported on the PAR. In fact, for the 1991-1996 FARS,
the data for first-event single-vehicle rollover crashes indicate
that there were 16,124 unknown ejection paths out of the 21,325
ejected persons, whether or not they were fatalities.
Distribution of the ejection paths identified in FARS for each
vehicle analyzed in section 7.4 is shown in Table 7. In these
analyses, the parameter of ``side door'' includes all side doors;
``side window'' includes all side glass; ``through roof opening'' is
through a convertible top which is down or a sunroof; and ``through
roof'' is through a convertible roof which is up.
Table 7.--Distribution of Ejections by Path in Fatal First-Event Single-Vehicle Rollovers
--------------------------------------------------------------------------------------------------------------------------------------------------------
Through
Side door Side Windshield Back Back door roof Through Other Unknown
Make/model, model years 1984-1990 window (percent) window opening roof path path
(percent) (percent) (percent) (percent) (percent) (percent) (percent) (percent)
--------------------------------------------------------------------------------------------------------------------------------------------------------
Chevy/Blazer, GMC/Jimmy............................. 2.1 11.0 0.7 0.0 0.0 0.7 0.0 1.4 84.2
Ford Bronco II...................................... 2.3 7.4 1.7 1.3 0.2 0.0 0.0 0.2 87.0
Isuzu Trooper....................................... 0.0 11.9 2.4 4.8 0.0 0.0 0.0 0.0 81.0
Jeep Cherokee....................................... 5.3 9.7 4.4 0.9 0.0 0.0 0.0 0.0 79.6
Suzuki Samurai...................................... 1.3 2.6 0.0 0.0 0.0 3.9 5.3 0.0 86.8
Toyota 4Runner...................................... 2.9 6.2 1.1 2.2 0.0 10.5 1.5 1.8 73.8
--------------------------------------------------------------------------------------------------------------------------------------------------------
FARS data: 1991-1996.
The results shown in Table 7 do not indicate a difference
between the Bronco II and its peers in the ejection path for these
fatal crashes. For ejection through the side windows including the
rear side windows, the Bronco II rate is lower than that of three of
its peers. When all glazing is considered as a single ejection path
(Side Windows + Windshield + Back Window), the Bronco II still
remains in the middle of the peer vehicles. It is noted that these
results are based on a small sample of the crashes because most of
the ejection paths are coded ``unknown'' in FARS.
7.5.3 NASS Case Analysis
To further study the ejection path issue, a hard copy review of
all Bronco II rollover crashes in the NASS Crashworthiness Data
System was conducted. Both totally and partially ejected occupants
were included in this review.
As shown in Table 3, there were 47 NASS cases in which there was
a Bronco II first-event single-vehicle rollover. ODI reviewed each
of these cases and found that out of the 47 cases (each case
contains one rollover incident), 23 cases had ejections involving 33
occupants (27 totally ejected and 6 partially ejected), 22 cases had
no occupant ejections, and in 2 cases occupant ejections were
unknown.
Table 8 shows recorded ejection paths for the 33 ejected
occupants.
Table 8.--Ejection Paths for 33 Ejected Occupants
----------------------------------------------------------------------------------------------------------------
Weighted
Number of Weighted percentage of
Ejection path ejections number of ejections by
ejections * ejection path
----------------------------------------------------------------------------------------------------------------
Unknown......................................................... 11 667 19
Left Front (Driver Window)...................................... 7 822 24
Left Door Opened................................................ 2 344 10
Right Front (Passenger Window).................................. 5 924 27
Windshield...................................................... 3 447 13
Left Rear Window (Fixed)........................................ 2 67 2
Right Rear Window (Fixed)....................................... 1 62 2
Rear Backlight.................................................. 1 30 1
Sunroof......................................................... 1 86 3
----------------------------------------------------------------------------------------------------------------
* The total for all ejection paths does not equal to that shown in Item 4 of section 7.5.2. because the partial
ejections are included in this analysis.
ODI's review of the 47 cases indicate the following:
1. In a majority of the cases, the crash scenario involved
running off the road at highway speed and driver overcorrection,
resulting in vehicle yaw, followed by rollover.
2. There was a wide variation in crash dynamics in the incidents
reviewed.
3. Distortion of the vehicle body during rollover typically
created several potential occupant ejection paths when glazing
disintegrated at several locations.
4. The data reviewed are inconclusive with respect to
identification of a ``most probable'' occupant ejection path during
rollover.
7.6 Other Data Reviewed
7.6.1 Ford Data Review
Ford supplied analyses of rollover propensity in its January 29,
1998, submission in response to this petition. (A copy is in the
public file.) These included an overall rollover rate analysis and a
logistic regression analysis similar to the NHTSA analysis used in
EA89-013.
In the Ford overall rollover rate analysis, rollover crash data
were collected from five states and combined to obtain an overall
rollover rate. The following states and crash years were used:
Alabama, 1990-95; Arkansas, 1987-94; Michigan, 1985-91; Maryland,
1986-94; and Pennsylvania, 1988-
[[Page 27351]]
95. For exposure, Ford used registered vehicle years (RVY) for these
same states and periods. Ford analyzed all types of vehicles, and
included about 700 make/model/model year combinations.
For illustration purposes, Table 9 presents Ford's data for the
first 11 sport utility vehicles in Ford's table as shown in Exhibit
B of Ford's January 29, 1998, submission. In addition to the 11
sport utility vehicles in Ford's table, there were 4 other vehicles,
including 3 pickup trucks (83-94 Ford Ranger--Rollover Rate--76; 81-
83 Toyota pickups--75; and 84-94 Toyota pickups--67) and one
passenger car (87-94 Mitsubishi Precis--68.)
The NASS/CDS analysis reported in Section 7.3 included four
additional vehicle groups not in Table 9, which are the GM-S&T
series, Isuzu Trooper, Jeep Cherokee, and Jeep Wagoneer. Their
rollover rates ranged from about 20 rollovers per 10,000 Registered
Vehicles Years (RVY) to about 36 rollovers per 10,000 RVY, with the
remainder in the low to high twenties.
Table 9.--Ford State Data Analysis Rollover Rates
----------------------------------------------------------------------------------------------------------------
Rollover rate
Make and model Model years Type RVY * Rollover (rollovers/
crashes 10,000 RVY *)
----------------------------------------------------------------------------------------------------------------
Toyota J4 Land Cruiser....... 81-83 4 x 4 SUV........ 762 8 105
Geo Tracker.................. 89-94 4 x 4 SUV........ 46,966 370 79
Jeep CJ-7.................... 81-86 4 x 4 SUV........ 137,670 1,032 75
Honda Passport............... 94 4 x 4 SUV........ 1,243 9 72
Ford Bronco II............... 84-90 4 x 4 SUV........ 605,297 4,132 68
Geo Tracker.................. 91-94 4 x 2 SUV........ 8,255 56 68
Ford Bronco II............... 86-90 4 x 2 SUV........ 50,217 321 64
Dodge Raider................. 87-89 4 x 4 SUV........ 31,263 188 60
Toyota 4Runner............... 84-94 4 x 4 SUV........ 121,813 728 60
Jeep CJ-5.................... 81-83 4 x 4 SUV........ 12,852 74 58
Suzuki Samurai............... 86-94 4 x 4 SUV........ 81,780 451 55
----------------------------------------------------------------------------------------------------------------
* RVY: registered vehicle years.
For the occupant injury analysis, Ford looked at injury rate
data in four states for several crash years, Arkansas, 1987-94;
Michigan, 1985-91; Maryland, 1986-94; and Pennsylvania, 1988-95. The
total number of rollover crashes, total number of occupants involved
in those crashes, and number of severe and fatal injuries were
reported for about 700 make/model/model year combinations of
vehicles. For illustration purpose, Table 10 presents selected data
from Ford's analysis for vehicles similar to the Bronco II.
Table 10.--Ford Occupant Injury Rates in Rollover Crashes
----------------------------------------------------------------------------------------------------------------
Percent severe
and fatal
Number of Number of Number of Number of injured
Make,* model, model year range crashes occupants severe injured fatal injured occupants in
occupants occupants rollover
crashes
----------------------------------------------------------------------------------------------------------------
Ford, Bronco II, 84-90.......... 4,074 6,453 492 55 8.5
GM, S&T series, 83-94........... 3,261 5,130 415 68 9.4
Isuzu, Trooper II, 84-91........ 446 718 53 4 7.9
Jeep, Cherokee, 81-94........... 1,301 2,078 98 15 5.4
Jeep, CJ-5, 81-83............... 72 106 12 3 14.2
Jeep, CJ-7, 81-86............... 989 1,492 122 12 9.0
Jeep, Wagoneer, 81-90........... 205 330 20 1 6.4
Jeep, Wrangler, 87-94........... 378 577 46 4 8.7
Suzuki, Samurai, 86-94.......... 418 601 73 6 13.1
Toyota, 4Runner, 84-94.......... 675 1,067 98 14 10.5
----------------------------------------------------------------------------------------------------------------
* Makes listed alphabetically.
Ford also conducted a logistic regression analysis similar to
the analysis conducted by NHTSA during EA89-013. In this analysis,
the rollover rates for several sport utility vehicles were compared.
The data were normalized for driver and environmental factors, which
included age, sex, location, and roadway alignment, and included
crash data from Michigan (85-91), Arkansas (87-94), Florida (89-94),
Maryland (86-88), and Pennsylvania (88-95). Figure 4 presents these
data. The upper and lower 95 percent confidence intervals are
presented along with each vehicle's average adjusted rollover rate.
This analysis indicates that while the rollover propensity of the
Bronco II is relatively high (in fact, the two-wheel drive model has
the highest rate), it is not statistically significantly different
from that of most of its peers.
BILLING CODE 4910-59-P
[[Page 27352]]
[GRAPHIC] [TIFF OMITTED] TN19MY99.025
BILLING CODE 4910-59-C
7.6.2 Suzuki Data Review
On March 19, 1997, Suzuki submitted a FARS data analysis related
to certain 4X4 SUV rollovers. This analysis only considered those
vehicles where the rollover occurred ``on road,'' hence no off-road
rollovers were considered. Suzuki stated that comparison of on-road
rollovers tends to normalize the rollover rate for some
environmental and roadway conditions where the crash occurred. FARS
years 1982 through 1995 were used in the analysis. Table 11 presents
data for these vehicles.
Table 11 indicates that the rate of on-road first-event single-
vehicle rollovers in fatal single-vehicle crashes per 100 vehicles
with at least one occupant fatality for the Bronco II is slightly
lower than the Toyota 4Runner, and is higher than that of the other
five peer vehicles.
Table 11.--Suzuki FARS Analysis Regarding On-Road First-Event Single-Vehicle Rollovers
----------------------------------------------------------------------------------------------------------------
On road first-
event single-
On road first- SVC involved vehicle
event single- vehicles with rollover in
Make and model Model years vehicle occupant fatal SVCs per
rollovers in fatality 100 vehicles
fatal SVCs** with occupant
fatality
----------------------------------------------------------------------------------------------------------------
Toyota 4Runner.................................. 90-94 40 188 23.8
Ford Bronco II.................................. 84-90 291 1,372 21.2
Nissan Pathfinder............................... 87-94 22 185 11.9
Jeep Grand Cherokee............................. 93-94 9 76 11.8
Chevrolet T10 Blazer............................ 83-90 108 1,057 10.2
Ford Explorer................................... 91-94 22 227 9.7
Isuzu Trooper................................... 84-89 14 146 9.6
----------------------------------------------------------------------------------------------------------------
** SVCs: Single Vehicle Crashes.
8.0 PETITIONER'S DATA ANALYSIS
In support of his petition, the petitioner stated that sworn
deposition testimony of Dr. Michelle Vogler, a statistical expert
retained by Ford in Bronco II product liability litigation, shows,
by actual count, that there have been 5,672 rollovers of Bronco II
vehicles in six states during a six-year time period. The petitioner
extrapolated this figure to assert that ``there have been 50,000
Bronco II rollovers nationally since the subject vehicles were first
placed in the hands of American consumers.'' By his estimation, ``as
many as 300,000 Bronco II owners are going to suffer the same fate
[rollover].''
After reviewing NASS/CDS data, the agency believes that Mr.
Heiskell's extrapolation overstates the number of Bronco II
rollovers that would be expected in a 14-year period. Regardless,
the absolute number of rollovers is an inappropriate measure for an
analysis of rollover propensity. The petitioner's extrapolation
focuses on the number of rollovers of any type (as opposed to first-
event rollovers) and represents the raw number of rollovers expected
(as opposed to the rollover rate) over a 14-year period, without
adjusting for attrition of the Bronco II fleet over time. In
contrast, NHTSA's analysis uses the percentage of single-vehicle
crashes in which a first-event single-vehicle rollover occurred.
[[Page 27353]]
First, the total number of rollovers is, to a large degree,
related directly to the number of vehicles on the road. Thus,
everything else being equal, two make/models with equivalent
rollover propensity could have vastly different numbers of rollovers
based solely on variations in the on-road fleet of each make/model.
Therefore, the total number of rollovers is insufficient on its own
to assess risk. Risk assessment is based on normalized populations
and expected outcomes, and can be best accomplished using the
agency's long-accepted metric, ``first-event single-vehicle
rollovers per single-vehicle crash.''
Secondly, the agency used first-event single-vehicle rollovers
as its measure because these crashes focus more on the handling and
stability aspects of vehicle performance than do all rollovers
combined. Subsequent event rollovers, which were included in the
petitioner's extrapolation, generally result from multiple-vehicle
collisions and collisions with objects such as utility poles,
guardrails, etc., where the inherent handling and stability of each
vehicle plays a lesser role due to the presence of forces exerted
upon the vehicle by its collision partner.
The use of first-event single-vehicle rollovers per single-
vehicle crash has been the focus of most serious efforts to relate
vehicle roll stability measures to real-world vehicle rollover
propensity. The agency subscribes to this approach, and believes
that this measure is an effective way to focus on the contribution
of vehicle stability to rollover propensity, while the total number
of rollovers experienced by a particular make/model is not.
9.0 Findings
1. An analysis of rollover complaints in the ODI consumer
database reveals a sharp decrease in Bronco II rollover complaints
since EA89-013 was closed. Additionally, an analysis of ODI rollover
complaints received since 1994 on peer vehicles does not suggest
that the subject vehicles have an abnormally high rollover
propensity compared to other sport utility vehicles.
2. Earlier analyses of rollover propensity demonstrated that the
Bronco II first-event single-vehicle rollover rate was consistent
with that of its peers, and the recently updated analyses, using
both state and NASS data, confirm this finding.
3. FARS data indicate that the subject vehicles have a
percentage of first-event single-vehicle rollover fatal crashes (out
of all fatal single vehicle crashes) and a percentage of first-event
single vehicle rollover fatalities (out of all fatalities in single
vehicle crashes) that are substantially higher than that of five
peer vehicles, although the results for the Bronco II are somewhat
similar to those for the Toyota 4Runner.
4. The Bronco II had a similar number of fatalities per involved
occupant in first-event single-vehicle rollover crashes when
compared to the Jeep Cherokee and possibly to the Toyota 4Runner,
and had more fatalities per involved occupant in first-event single-
vehicle rollover crashes when compared to three other peer vehicles.
This suggests that if a first-event single-vehicle rollover occurs,
there is more likely to be a fatality in a Bronco II than in some,
but not all, of its peers.
5. A review of FARS data between 1991 and 1996 describing
occupant ejection path did not indicate a difference between the
Bronco II and its peers, in part because most ejection paths were
coded ``unknown'' in FARS.
6. A detailed review of the 47 NASS cases in which there was a
Bronco II first-event single-vehicle rollover did not permit an
identification of a ``most probable'' occupant ejection path.
7. In analyses conducted by Ford, the Bronco II's first-event
single-vehicle rollover rate, measured as a proportion of the number
of registered vehicles, is similar to that of several of its sport
utility vehicle peers, pickups and a passenger car. In a logistic
regression analysis which controlled for driver and roadway
variables, a duplication of NHTSA's EA89-013 analysis using newer
data, the Bronco II rollover rate was relatively high, but was not
statistically significantly different from that of most of its
peers.
8. Suzuki's FARS analysis indicates that the Bronco II and one
of its peers have a similar rate of ``on-road'' first-event single-
vehicle rollovers as a percentage of all single vehicle fatal
crashes.
9. The petitioner's estimate of the number of rollover crashes
involving the Bronco II appears to overestimate the number. In any
event, the total number of rollover occurrences involving a
particular vehicle is not an appropriate analytical tool to assess
rollover risk.
10.0 Conclusion
The focus of this defect petition was on the allegedly high
rollover propensity of the Bronco II. Consistent with its findings
several years ago at the time it closed EA89-013, ODI's analysis of
more recent data indicates that the rollover propensity of the
Bronco II does not stand out from that of other peer SUVs. Although
it was not directly raised by the petitioner, ODI conducted an
extensive analysis of the crashworthiness of the Bronco II in
rollover crashes. These analyses indicated a cause for concern,
since the Bronco II vehicles have a percentage of first-event single
vehicle rollover fatal crashes and a percentage of first-event
single vehicle rollover fatalities that are substantially higher
than that of most of the peer vehicles. However, ODI was unable to
identify a most probable ejection path or to identify a specific
aspect of the vehicle that appeared to adversely affect the
vehicle's rollover crashworthiness.
Based on the information presented above, as well as the age of
the subject vehicles, it is unlikely that NHTSA would issue an order
for the notification and remedy of a safety-related defect in the
subject vehicles at the conclusion of the investigation requested in
the petition. Therefore, in view of the need to allocate and
prioritize NHTSA's limited resources to best accomplish the agency's
safety mission, the petition is denied.
[FR Doc. 99-12579 Filed 5-14-99; 3:29 pm]
BILLING CODE 4910-59-P