The Effectiveness of Retroreflective Tape On Heavy Trailers: DOT HS 809 222 March 2001 NHTSA Technical Report

U.S.
Department
of Transportation http://www.nhtsa.dot.gov
National Highway
Traffic Safety
Administration
DOT HS 809 222 March 2001

NHTSA Technical Report
The Effectiveness of Retroreflective Tape

on Heavy Trailers
This document is available to the public from the National Technical Information Service, Springfield, Virginia 22161.
The United States Government does not endorse products or
manufacturers. Trade or manufacturers’ names appear only
because they are considered essential to the object of this report.
Technical Report Documentation Page
1. Report No. 2. Government Accession No. 3. Recipient’s Catalog No.
DOT HS 809 222

4. Title and Subtitle 5. Report Date
The Effectiveness of Retroreflective Tape on Heavy Trailers March 2001

6. Performing Organization Code
7. Author(s) 8. Performing Organization Report No.
Christina Morgan
9. Performing Organization Name and Address 10. Work Unit No. (TRAIS)
Evaluation Division, Plans and Policy
National Highway Traffic Safety Administration 11. Contract or Grant No.
Washington, DC 20590
12. Sponsoring Agency Name and Address 13. Type of Report and Period Covered
Department of Transportation NHTSA Technical Report
National Highway Traffic Safety Administration 14. Sponsoring Agency Code
Washington, DC 20590
15. Supplementary Notes
16. Abstract
This report evaluates the effectiveness of retroreflective tape in enhancing the visibility of heavy trailers
and reducing side and rear impacts by other vehicles into these trailers during dark conditions. It is based
on a statistical analysis of 10,959 crash cases investigated by the Florida Highway Patrol and the
Pennsylvania State Police in 1997 - 1999.
The tape is quite effective. It reduced side and rear impacts into trailers, in dark conditions (including "dark-
not-lighted," "dark-lighted," "dawn," and "dusk") by 29 percent. In "dark-not-lighted" conditions, the tape
reduced side and rear impact crashes by 41 percent. Tape is especially effective in reducing injury crashes.
In dark conditions, it reduced side and rear impacts that resulted in fatalities or injuries to drivers of any vehicle
by 44 percent.
17. Key Words 18. Distribution Statement

retroreflective tape; heavy trailers; tractor-trailer
combination vehicles; statistical analysis; evaluation; Document is available to the public through the
side and rear impacts; conspicuity; visibility National Technical Information Service, Springfield,
Virginia 22161
19. Security Classif. (Of this report) 20. Security Classif. (Of this page) 21. No. of Pages 22. Price
Unclassified Unclassified 63
Form DOT F 1700.7 (8-72) Reproduction of completed page authorized
TABLE OF CONTENTS
TABLE OF CONTENTS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . iii
ACKNOWLEDGMENTS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . iv
EXECUTIVE SUMMARY . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . v
INTRODUCTION AND BACKGROUND . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1

1.1 HISTORY AND RESULTS OF EARLIER EFFECTIVENESS STUDIES
.............................................................2
1.2 FMCSA RETROFIT STANDARD . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4
1.3 CURRENT STATUS OF TAPE . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5
CONSPICUITY DATA . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7
2.1 SUPPLEMENTARY FORM . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7
2.2 CONSPICUITY DATA BASE . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11
2.3 CRASHES NOT INVESTIGATED BY STATE POLICE . . . . . . . . . . . . . . . . 20
BASIC ANALYSIS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 23
3.1 DEFINITIONS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 23
3.2 THE BASIC ANALYSIS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 26
3.3 STATISTICAL SIGNIFICANCE AND CONFIDENCE BOUNDS . . . . . . . . . 29
3.4 TAPE EFFECTIVENESS IN SPECIFIC DARK CONDITIONS . . . . . . . . . . . 32
3.5 COMBINED EFFECTIVENESS AND “BEST” ESTIMATES . . . . . . . . . . . . 33
TAPE EFFECTIVENESS IN SPECIFIC SITUATIONS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 39

4.1 EFFECTIVENESS BY COLLISION TYPES . . . . . . . . . . . . . . . . . . . . . . . . . 40
4.2 EFFECTIVENESS BY ENVIRONMENTAL/ROADWAY CONDITIONS . . . 42
4.3 EFFECTIVENESS BY TRACTOR/TRAILER CHARACTERISTICS . . . . . . 44
4.4 EFFECTIVENESS BY OTHER DRIVER/VEHICLE CHARACTERISTICS
............................................................ 46
4.5 EFFECTIVENESS BY CRASH-LEVEL INJURY SEVERITY . . . . . . . . . . . . 48
4.6 TAPE EFFECTIVENESS DURING THE DAYLIGHT . . . . . . . . . . . . . . . . . . 49
BENEFITS OF CONSPICUITY TAPE . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 51

5.1 CRASHES AVOIDED PER YEAR . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 51
5.2 NONFATAL INJURIES AVOIDED PER YEAR . . . . . . . . . . . . . . . . . . . . . . 52
5.3 LIVES SAVED PER YEAR . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 52
REFERENCES . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 55
iii
ACKNOWLEDGMENTS
I would like to acknowledge Glenn G. Parsons, retired NHTSA employee, who started this
evaluation. He designed the study and the data form; contacted the states and established
working agreements with the Florida Highway Patrol and the Pennsylvania State Police.
I would also like to acknowledge the officers of the Florida Highway Patrol and the Pennsylvania
State Police who collected the data, without which this report would not be possible. Ed Bleakly
of the Florida Highway Patrol and Sergeant John Rigney of the Pennsylvania State Police
managed the data collection effort in their states, submitted the forms to NHTSA, and
coordinated the project with NHTSA.
iv
EXECUTIVE SUMMARY
All heavy trailers manufactured on or after December 1, 1993 must be equipped with red-and-
white retroreflective tape, sheeting and/or reflex reflectors around the sides and rear to make them
more conspicuous. The National Highway Traffic Safety Administration (NHTSA) established
this requirement, with its various options, in December 1992 by amending Federal Motor Vehicle
Safety Standard (FMVSS) No. 108, “Lamps, Reflective Devices, and Associated Equipment.”
However, retroreflective tape has been used almost exclusively for meeting the standard, and it is
the subject of this evaluation. Heavy trailers are at least 80 inches wide and have a Gross Vehicle
Weight Rating over 10,000 pounds.
The purpose of retroreflective tape is to increase the visibility of heavy trailers to other motorists,
especially in the dark. At those times, the tape brightly reflects other motorists’ headlights and
warns them that they are closing on a heavy trailer. In the dark, without the tape, many trailers do
not become visible to other road users until they are dangerously close. The alternating red-and-
white pattern flags its bearer as a heavy trailer and at the same time helps other road users gauge
their distance and rate of approach. This report evaluates the effectiveness of the tape in reducing
side and rear impacts into heavy trailers - primarily in dark conditions where even a vigilant
motorist might not see an untreated trailer in time to avoid a crash, and secondarily in daylight,
where the tape might alert inattentive drivers that they are approaching a trailer.
In March 1999, the Federal Highway Administration extended the application of this important
protection to the entire on-road trailer fleet by directing motor carriers engaged in interstate
commerce to retrofit heavy trailers manufactured before December 1993 with tape or reflectors.
These older trailers must have some form of conspicuity treatment, by June 1, 2001, in the
locations specified by the NHTSA standard for new trailers, except on the rear impact guard. In
other words, as of June 2001, almost all heavy trailers on the road will have some form of
conspicuity treatment. This Federal Motor Carrier Safety Regulation furthermore gives motor
carriers until June 1, 2009 to retire their pre-1993 trailers or retrofit them with treatments that
conform exactly to the NHTSA standard (again, with the exception of the rear impact guard).
Since none of NHTSA’s crash data at hand (FARS, NASS, or State files) identified whether
crash-involved heavy trailers had retroreflective tape, NHTSA worked out agreements with the
Florida Highway Patrol (FHP) and the Pennsylvania State Police (PSP) to collect data for this
analysis. For a two-year period, each time these agencies investigated a crash involving a tractor-
trailer and filed a crash report, they also filled out an “Investigator’s Supplementary Truck-
Tractor Trailer Accident Report” on every trailer in the crash. The FHP collected 6,095 crash
cases from June 1, 1997, through May 31, 1999. The PSP collected 4,864 crash cases from
December 1, 1997, through November 30, 1999.
The analysis estimates the reduction of side and rear impacts by other vehicles into tape-equipped
trailers in dark conditions - relative to the number that would have been expected if the trailers
v
had not been equipped. It is based on tabulating and statistically analyzing crash involvements of
tractor-trailers by three critical parameters: (1) whether or not the trailer is tape-equipped; (2) the
light condition - dark (comprising “dark-not-lighted,” “dark-lighted,” “dawn” and “dusk”) vs.
daylight; and (3) relevant vs. control-group crash involvements. Relevant crash involvements are
those where another vehicle crashed into the side or rear of a heavy trailer, because the tape can
help the other driver see and possibly avoid hitting the trailer. The control group consists of
single-vehicle crashes of tractor-trailers (where visibility of the tractor-trailer to other road users
is not an issue at all) and impacts of the front of the tractor into other vehicles (where conspicuity
of the side and rear of the trailer is also not an issue).
The principal conclusion of the study is that retroreflective tape is quite effective, and it
significantly reduces side and rear impacts into heavy trailers in the dark. Other findings and
conclusions are the following:
ANNUAL BENEFITS OF CONSPICUITY TAPE
• When all heavy trailers have conspicuity tape, the tape will be saving an estimated 191 to
350 lives per year, preventing approximately 3,100 to 5,000 injuries per year, and
preventing approximately 7,800 crashes per year, relative to a hypothetical fleet in which
none of the trailers have the tape.
CRASH REDUCTIONS BY LIGHTING CONDITIONS
• In dark conditions (combining the subsets of “dark-not-lighted,” “dark-lighted,” “dawn,”

and “dusk”), the tape reduces side and rear impacts into heavy trailers by 29 percent. The
reduction is statistically significant (confidence bounds: 19 to 39 percent).
• However, the tape is by far the most effective in dark-not-lighted conditions. Here, the
tape reduces side and rear impacts into heavy trailers by 41 percent. The reduction is
statistically significant (confidence bounds: 31 to 51 percent).
• In dark-lighted, dawn, and dusk conditions, the tape did not significantly reduce crashes.
The tape also did not significantly reduce crashes during daylight.
CRASH REDUCTIONS FOR SPECIFIC SUBGROUPS IN DARK CONDITIONS
The effectiveness estimates here are the percentage reductions of various subgroups of the side
and rear impacts into heavy trailers in dark conditions. As stated above, tape reduces these crash
involvements by 29 percent, overall.
vi
• The tape is especially effective in preventing the more severe crashes, specifically, injury
crashes. Impacts resulting in fatal or nonfatal injuries to at least one driver are reduced by
44 percent.
• The tape is more effective when the driver of the impacting vehicle is young. The crash
reduction is 44 percent when the driver of the impacting vehicle is 15 to 50 years old, but
only 20 percent when that driver is more than 50 years old. A possible explanation of this
difference is that older drivers are less able to see, recognize and/or react to the tape in
time to avoid hitting the trailer.
• The tape may be somewhat more effective in preventing rear impacts (43 percent) than
side impacts (17 percent) into trailers; however, this difference is not consistent in the two
states.
• The tape is effective in both clear (28 percent) and rainy/foggy weather conditions (31
percent).
• The tape is especially effective on flatbed trailers (55 percent). These low-profile vehicles
must have been especially difficult to see in the dark before they were treated with tape.
• Dirt on the tape significantly diminished its effectiveness in rear impacts. Clean tape
reduces rear impacts by 53 percent but dirty tape by only 27 percent.
STATUS OF TAPE IN THE 1997-1999 CRASH DATA
• Almost 50 percent of the pre-standard trailers in the study had retroreflective tape. The
retrofit of these older, pre-1993 trailers was already well underway in 1997 - 1999.
• More than 60 percent of the trailers with retroreflective tape had clean tape at the time of
the study. About 30 percent of the trailers with tape had some dirt and less than 5 percent
had “very dirty” tape.
• About 96 to 99 percent of the retroreflective tape on the side of trailers was intact, while
92 to 95 percent of the tape on the rear of trailers was intact.
vii
viii
CHAPTER 1
INTRODUCTION AND BACKGROUND
In September 1992, the National Highway Traffic Safety Administration (NHTSA) amended
Federal Motor Vehicle Safety Standard (FMVSS) No. 108, "Lamps, Reflective Devices, and
Associated Equipment," by adding a Conspicuity Systems provision. This revision, effective
December 1, 1993, requires that heavy trailers (i.e., those 80 or more inches in width with a Gross
Vehicle Weight Rating over 10,000 pounds) manufactured on and after this date be equipped with
reflective material. Two types of material are permitted -- (1) retroreflective sheeting, or tape,
and (2) reflex reflectors. A combination of the two types is also permissible. However,
retroreflective tape has been used almost exclusively for meeting the standard, and it is the subject
of this evaluation. Essentially, the tape must outline the bottom of the sides of the trailers and the
top corners, bottom and underride guard of the rear of the trailers. The tape must be applied in a
pattern of white and red color segments to the sides and rear of the trailer and in white to the
upper rear corners of the trailer. Specifications for affixing the tape to the sides and rear of the
trailers are contained in Title 49, Part 571, Section 108 of the Code of Federal Regulations (i.e.,
49 CFR 571.108).
The purpose of the regulation is to make heavy trailers more conspicuous to other motorists.
Studies of highway crashes where other motor vehicles collide with combination trucks (truck
tractor plus heavy trailer) have indicated that, in a number of these crashes, the operator of the
other vehicle may not have seen the combination truck in time to avoid a collision. Such crashes
are more likely to occur in dark conditions or under other conditions of decreased visibility -- i.e.,
adverse weather conditions such as rain, snow, or fog. The light reflection qualities of the tape,
particularly from sources such as automobile headlamps, enhance the conspicuousness of the
heavy trailer, thereby also increasing the chances that the attention of other drivers in the vicinity
will be directed to the combination truck. It is hoped that the tape, with its alternating red-white
pattern, will also help those drivers more accurately assess the closing rate and distance between
their vehicle and the combination truck.
The tape is expected to be more effective in dark conditions, when combination trucks are harder
to see, than during daylight. The amount of light should also influence the effectiveness of the
tape. The tape should be more effective on unlighted dark roads than on lighted roads, or during
dawn or dusk when some light is available.
The tape should also reduce collisions into the rear and side of combination trucks especially in
dark conditions, since the tape is on the side and rear of trailers. In rear impacts, your headlights
will shine on the rear of the trailer and illuminate the tape, so you can detect the vehicle and avoid
a collision. In side impacts where you are moving exactly or nearly perpendicular to the
combination truck and especially if the combination truck is moving slowly or stopped, the tape is
expected to be highly effective. In this case, your headlights will illuminate the side of the trailer
and you may have enough time to avoid the collision. The tape may be less effective in preventing
1
sideswipe crashes where you are moving parallel to the combination truck. In this case, the angle
between your headlights and side of the trailer may be too small to illuminate the tape. The tape
is unlikely to have an effect in collisions with the front of the combination truck or on any single-
vehicle crashes of combination trucks.
The Government Performance and Results Act of 1993 and Executive Order 12866 (October
1993) require agencies to conduct periodic evaluations to assess the effectiveness of their existing
vehicle safety standards. This report evaluates the effectiveness of retroreflective tape on heavy
trailers required by FMVSS 108 (Lamps, Reflective Devices, and Associated Equipment). This
evaluation will show if the tape reduces the number of side and rear impacts to trailers in dark
conditions.
Throughout the remainder of the report, a truck tractor pulling one or more trailers -- i.e., tractor
with semi-trailer, full trailer, or two trailers will be referred to as a combination truck,
retroreflective tape will sometimes be referred to as tape, and heavy trailers will sometimes be
referred to as trailers.
1.1 HISTORY AND RESULTS OF EARLIER EFFECTIVENESS STUDIES
Several early studies of combination truck crashes concluded that increasing the conspicuity of
heavy trailers in dark conditions would reduce some of these crashes. Minahan and O’Day
analyzed fatal car into combination truck crashes in Michigan and Texas and found that such
crashes usually occur in the dark with frequent car underrides1. They concluded that the driver of
the other vehicle did not detect the presence of the combination truck in time to avoid a collision.
On the basis of these findings, it was concluded that improvements in the conspicuity of heavy
trailers might reduce the frequency and severity of these crashes. Another analysis of the
collisions of cars with tractor-semitrailers, based on the 1977 Fatality Analysis Reporting System
(FARS), also found that such collisions are overrepresented in dark conditions and concluded the
addition of lights or reflective paints on trucks and trailers would reduce the frequency of
collisions2.
In 1980, the agency initiated a three-phase research project to develop and evaluate an optimal
configuration of heavy-truck and truck-trailer markings and lights. Phase I of this project entailed
1
Minahan, D. J. and O’Day, J., Car-Truck Fatal Accidents in Michigan and Texas,
University of Michigan Highway Safety Research Institute, Report No. UM-HSRI-77-49,
Ann Arbor, MI, 1977.
2
Green, P., et al., Accidents and the Nighttime Conspicuity of Trucks, University of
Michigan Highway Safety Research Institute, Report No. UM-HSRI-79-92, Ann Arbor, MI 1979.
2
analyzing the problem of other vehicles striking large trucks and trailers.3 The work included
interviewing trucking company representatives, analyzing crash data, modeling driver behavior,
and analyzing active lighting and passive reflective material. The results indicated that crashes in
which conspicuity might conceivably have been a factor were equally distributed between daylight
and dark conditions and involved collisions with both the sides and rear of the trailer. Rear
impacts tend to occur when the combination truck is traveling straight ahead and moving slowly,
stopping or stopped on the roadway. The following driver either 1) does not see the combination
truck at all, 2) sees the combination truck but misjudges its motion and/or distance, or 3) correctly
perceives the combination truck’s dynamics and distance, but too late. Moreover, collisions of
this sort are more severe in dark conditions. Side impacts most often occur when the combination
truck is turning or being astride lanes, e.g. backing, making U-turns, etc. Sideswipe crashes often
occur while the combination truck is traveling straight.
Phase II entailed conducting a series of laboratory and field studies to determine the best way to
mark heavy trailers and improve other drivers’ abilities to: 1) quickly and accurately identify
combination trucks in the traffic stream, 2) judge their rate of closure, and 3) estimate their
distance from combination trucks.4 The most effective marking scheme identified in these studies
consisted of a strip of alternating colors outlining the side and rear perimeters of the trailers and a
U-shaped outline of the mudflaps.
Phase III of the research project was a fleet study to evaluate the crash reduction effectiveness of
the reflective tape to the sides and rears of commercial trailers.5 The study, conducted by Vector
Enterprises, Inc., took place over a 23-month period in 1983-1985. A total of 3,820 van trailers
were selected for participation; half were treated with retroreflective tape; the other half served as
a control group against which the performance of the treated trailers was compared. However,
because of cost considerations, 1,910 treated van trailers in the study were equipped with less
reflectorized material than that recommended by the Phase Two study. The Vector conspicuity
scheme used alternately hatched red and white or blue and white, two-inch wide strips of
retroreflective tape to outline the lower side rail on both sides of the trailer and the rear perimeter
of the trailer. Each of the two groups accumulated 106 million miles of exposure during the study
period. The study concluded that tractor-trailer combinations in the treated fleet were struck by
3
Burger, W. J., et al., Improved Commercial Vehicle Conspicuity and Signalling Systems,
Task I – Accident Analysis and Functional Requirements, NHTSA Report No. DOT HS 806
100, Washington, DC, 1981.
4
Ziedman, K., et al., Improved Commercial Vehicle Conspicuity and Signalling Systems,
Task II – Analyses, Experiments and Design Recommendations, NHTSA Report No. DOT HS
806 098, Washington, DC, 1981.
5
Burger, W. J., et al., Improved Commercial Vehicle Conspicuity and Signalling Systems,
Task III – Field Test Evaluation of Vehicle Reflectorization Effectiveness, NHTSA Report No.
DOT HS 806 923, Washington, DC, 1985.
3
other vehicles 15 percent fewer times than were combinations in the control fleet; the report did
not distinguish side from rear impacts.
The results of the fleet study were found to vary somewhat according to the number of crashes
considered to be relevant -- that is, the number of crashes whose occurrence could be considered
to have been affected by the conspicuousness of the trailer. After extensive review and reanalysis
of the fleet test results, including solicitation of viewpoints from the public, NHTSA concluded
that the potential benefits from retroreflective marking of heavy trailers were sufficient to warrant
such a requirement under FMVSS 108. From its final review of the field test analyses, the agency
estimated that the use of the material would reduce crashes into the side and rear of combination
trucks in dark conditions by 15 percent and 25 percent, respectively.6 It was also estimated that
injuries and fatalities in these crashes would be reduced by 15 percent.
A study7 sponsored by NHTSA defined the large truck conspicuity enhancements that ought to be
used as a basis for the revised Federal regulations. The study recommended the retroreflective
tape width, color, pattern, and placement. The study also recommended the appropriate
retroreflective efficiency level, taking into account the effects of environmental dirt, aging, and
orientation of the marked vehicle. The current NHTSA standard generally incorporates these
recommendations.
1.2 FMCSA RETROFIT STANDARD
On March 31, 1999, the Federal Highway Administration (FHWA) published a final rule
amending the Federal Motor Carrier Safety Regulations (FMCSRs) to require motor carriers
engaged in interstate commerce to install retroreflective tape or reflex reflectors on the sides and
rear of trailers manufactured prior to December 1, 1993. The final rule gives motor carriers until
June 1, 2001, to install some form of conspicuity treatment in the same locations that NHTSA
requires manufacturers to install such treatments, with the exception of the rear impact guard.
Motor carriers have until June 1, 2009, to install conspicuity treatments identical to that required
on new vehicles, with the exception of the rear impact guard. Effective January 2000, the
authority for issuing and enforcing FMCSRs was transferred to a new agency within the
Department of Transportation, the Federal Motor Carrier Safety Administration (FMCSA)
There are two notable differences between NHTSA’s standard for new trailers manufactured on
or after December 1, 1993 and FMCSA’s retrofit requirement for trailers manufactured before
6
Preliminary Regulatory Evaluation - Proposed Amendment to FMVSS No. 108 to
Require Retroreflective Material on the Side and Rears of Heavy Trailers, NHTSA, Washington,
DC, 1991.
7
Olsen, P. L., et al., Performance Requirements for Large Truck Conspicuity
Enhancements, NHTSA Report No. DOT HS 807 815, Washington, DC, 1992
4
December 1, 1993. The NHTSA standard requires an alternating red and white pattern. FMCSA
encourages the use of a red-and-white pattern, but allows flexibility in terms of colors or color
combinations from June 1, 2001 through June 1, 2009. After June 1, 2009, these trailers, if they
are still in service, will have to have the red-and-white pattern. The second difference is that on
the rear, NHTSA requires two red and white applications, one the full width of the vehicle and the
other the full width of the underride guard. The FMCSA requires only the full width of the
vehicle.
A standardized appearance will assist motorists so that they can quickly recognize the image of
the reflective tape in the dark and associate it with a trailer. Therefore, it alerts motorists to the
presence or motion of the trailer, even if the body of the trailer is not visible.
1.3 CURRENT STATUS OF TAPE
Table 1-1 shows the percentage of trailers that have tape and the percentage of trailers with
FMVSS 108 tape by state and calendar year. The percentage of trailers with tape is increasing
over time. This suggests that the retrofitting of trailers with tape is proceeding. An observational
study by NHTSA staff in 1996 found about 60 percent of the combination trucks had trailers with
tape. The vehicles were observed at a weigh station along a major interstate in two states, Florida
and Pennsylvania. By 1999, almost 70 percent of the trailers had tape in Florida and almost 80
percent in Pennsylvania. The 1997-1999 data is from a recent study of truck-trailer crashes
collected to evaluate the effectiveness of conspicuity tape. Florida collected data from June1,
1997, through May 31, 1999 and Pennsylvania from December 1, 1997, through November
30,1999. Therefore, the Florida 1999 data is from only the first five months of the year, while the
Pennsylvania 1999 data is from almost the full year, January through November. (See Chapter 2
for more information about the conspicuity data.)
TABLE 1-1
Observational and Conspicuity Crash Data
Trailers by Tape Configuration, State, and Calendar Year
Total Trailers With FMVSS % of Trailers w/Tape
Trailers With Tape 108 Tape that meet FMVSS 108
Florida
1996 (Observed) 453 271 60% 181 40% 67%
1997 (Crashes) 1,904 1,060 56% 962 51% 91%
1998 3,224 2,002 62% 1,864 58% 93%
1999 1,275 863 68% 811 64% 94%
Pennsylvania
1996 (Observed) 1,116 632 57% 500 45% 79%
1997 (Crashes) 298 206 69% 191 64% 93%
1998 2,457 1,745 71% 1,612 66% 92%
1999 2,430 1,908 79% 1,776 73% 93%
5
Table 1-2 shows the percentage of pre-standard trailers (trailer model year less than 1993 that are
not required to have tape until June 1, 2001) that have tape and the percentage of pre-standard
trailers with FMVSS 108 tape by state and calendar year. The observational data cannot be
separated by model year and are not included in this table. Here, there is a 10 percentage point
increase in the trailers with tape in Florida from 1997 to 1999. In Pennsylvania, there is an 8
percentage point increase from 1998 to 1999. The 1997 Pennsylvania data is from only one
month, December, and is not a large enough sample to accurately represent the status of trailers
with tape.
TABLE 1-2
Pre-1993 Trailers by Tape Configuration, State, and Calendar Year
Total With FMVSS 108 % of Trailer w/Tape
Trailers With Tape Tape that meet FMVSS 108
Florida
1997 1,091 417 38% 349 32% 84%
1998 1,711 715 42% 622 36% 87%
1999 660 317 48% 283 43% 89%
Pennsylvania
1997 108 60 56% 54 50% 90%
1998 725 339 47% 289 40% 85%
1999 654 360 55% 310 47% 86%
Table 1-1 and 1-2 also shows the percentage of trailers with tape that meet FMVSS 108
requirements. In Table 1-1, the 1996 data is noticeably different than the later years in the
percentage of trailers with tape that meet the 108 standard. In the 1996 observational survey, the
tape had to meet the NHTSA standard exactly, while the tape had to meet or be similar to the
standard in the 1997-1999 crash data. The criteria the police officers used to judge whether the
tape was similar or dissimilar to FMVSS 108 is not known. In 1999, about 93 percent of all
trailers with tape have tape that meets the NHTSA standard or is similar to the standard.
Table 1-2 shows that about 86 percent of the retrofitted trailers with tape in 1999 have tape
complying with FMVSS 108 although not required until 2009. Thus, a large majority of trailers
with tape already have a standardized appearance.
6
CHAPTER 2
CONSPICUITY DATA
Typically, studies to assess the effectiveness of a vehicle safety standard consist of the collection
and analysis of highway crash data which compare the experience of vehicles that meet the safety
standard with the experience of vehicles that do not meet the standard. For most evaluation
studies, this is a comparison of the crash experience of vehicles built before the effective date of
the given standard with the crash experience of the vehicles built subsequent to the effective date
of the standard. All trailers manufactured on and after December 1, 1993, are required to be
equipped with the retroreflective material. Retroreflective tape was the primary choice of
material. However, many trucking firms equipped their new trailers with the special tape prior to
the December 1993 effective date of the NHTSA requirement. Companies have added reflective
tape to their older (i.e., pre-December 1, 1993, manufacture) trailers since the NHTSA
conspicuity requirement was issued and in response to FMCSA retrofit standard. Therefore, it
was necessary to obtain more information on the crash-involved trailer than just its date of
manufacture (which can be derived from the trailer Vehicle Identification Number or VIN). It
was necessary to observe directly whether or not the trailers were equipped with retroreflective
tape. None of the existing NHTSA crash data sets (FARS, NASS CDS, NASS GES, or the State
data) identify whether or not a crash involving heavy trailers had this tape. NHTSA had to collect
and create a new data set containing this essential data element.
2.1 SUPPLEMENTARY FORM
NHTSA created the “Investigator’s Supplementary Truck-Tractor Trailer Accident Report” form
to collect the necessary tape information on trailers. Most of the supplementary elements pertain
only to the trailer (or trailers) being pulled by the truck tractors. Other data items collected on the
form besides the presence of retroreflective tape are:
(1) whether or not the application pattern conforms to the FMVSS 108
requirement,
(2) the color(s) of the tape,
(3) the condition of the tape with respect to the presence/accumulation of dirt or other
agents which could degrade the reflectivity,
(4) whether the tape is damaged or has missing segments,
(5) the weather conditions at the time of the crash,
(6) the light conditions (i.e., dark, daylight, dusk, etc.) at the time of the crash,
(7) the date, time, and day of week of the crash,
(8) the county, city, road and speed limit of the crash, and
(9) the state accident report number, to allow eventual linkage to the state crash file.
7
The supplementary form was commonly referred to as the “NHTSA Green” because it was
printed on green paper. A copy of the NHTSA Green form is on Page 9 and the instructions for
the form are on Page 10.
Other data elements describing the crash are also required for this analysis such as:
(1) the crash configuration (i.e., other vehicle into combination truck rear, other vehicle
into combination truck side, etc.).
(2) a diagram/narrative explanation of the crash.
(3) estimates of damage to the other vehicle, and to the combination truck, in
descriptive terms.
(4) injuries and fatalities resulting from the crash.
(5) contributing factors or other conditions surrounding the crash (alcohol,
speeding, fail to yield, etc.).
These data elements are routinely collected on Police Accident Report (PAR) forms. Therefore,
data from both the NHTSA Green and the basic PAR forms were necessary for this analysis.
The Florida Highway Patrol (FHP) and the Pennsylvania State Police (PSP) agreed to collect the
necessary data for this analysis. For a two-year period, each time they investigated a crash
involving a tractor-trailer and filed a PAR, they also completed a NHTSA Green form on every
trailer in the crash. A tractor-trailer combination was defined as a truck tractor pulling one or
more trailers -- i.e., tractor with semi-trailer, full trailer, or two trailers. Only crashes investigated
and reported by these agencies were included in the study. Crashes where the PAR was filed by
local police, sheriffs, other police agencies, or the drivers themselves were not included. In 1998,
the FHP collected 59 percent and the PSP collected 60 percent of all crashes involving a tractor
trailer in their States. See Section 2.3 for a discussion on crashes not investigated by state police
and how this will not significantly affect our effectiveness estimates.
For each crash reported, the FHP and the PSP provided NHTSA with a copy of the State PAR
and the respective NHTSA Green stapled together, with the NHTSA Green displaying the same
accident report number as the State report. The FHP collected data from June 1, 1997, through
May 31, 1999. The PSP collected data from December 1, 1997, through November 30, 1999.
8
2.2 CONSPICUITY DATA BASE
A contractor created a data base with the necessary information to evaluate the safety benefits of
retroreflective tape. The contractor entered all the data elements on the NHTSA Green and some
of the elements on the PAR. Only the pertinent data elements on the PAR were entered such as:
initial point of impact, vehicle type, vehicle maneuver, first harmful event, driver age, driver sex,
etc. The most important data elements were coded directly from the hard-copy data to avoid a
delay in analyzing the data, since the states’ automated PAR files are usually not available at
NHTSA until 6 to 9 months after the end of the year. The remaining elements on the PAR could
be analyzed later, if needed, by linking the conspicuity data to the automated state data files at
NHTSA. The state accident report number on the NHTSA Green permits us to match the
conspicuity cases to the corresponding state data cases for this purpose.
The conspicuity data contain one derived element, besides the elements on the forms. The
derived element identifies the specific location on the vehicle of the first impact if it was not
identified on the PAR. In both states, the point of impact is an element on the PAR. In Florida,
the police can mark “Trailer,” or “Unknown” if applicable. Point of impact is “Trailer” in 16
percent of the tractor-trailer combinations and is “Unknown “ in 5 percent. In Pennsylvania, point
of impact is “Towed Unit” in 5 percent and is “Unknown “ in 1 percent of the tractor-trailer
combinations. These codes are too general for this analysis and do not say if the trailer was hit in
the rear or the side, a crucial distinction for this report. Thus, if the point of impact is “Trailer,”
“Towed Unit,” or “Unknown,” the data coder used all available information on the PAR including
the diagram and the narrative to classify the specific location on the trailer of the first impact as
one of the following:
No Damage, No Impact, Non Collision or Not Applicable
Front
Right Side
Rear
Left Side
Top
Undercarriage
Other
Unknown
11
The conspicuity data base was organized into three files for each state: the crash file, the vehicle
file, and the pedestrian file. The crash file contains information describing the environmental
conditions and roadway characteristics at the time of the crash. The vehicle file contains
information describing the vehicles and their drivers involved in the crashes. The pedestrian file
contains information describing any pedestrians involved in the crashes. Table 2-1 shows the
number of crashes, vehicles, and pedestrians on the conspicuity data base by state.
TABLE 2-1
Conspicuity Data Base: The Number of Cases by State and File
Crashes Vehicles Pedestrians
Florida 6,095 12,380 51
Pennsylvania 4,864 9,134 25
Total 10,959 21,514 76
Since each crash had to include at least one tractor-trailer combination, more than half of the
vehicles are tractor-trailer combinations. Table 2-2 shows that 52 percent of the Florida vehicles
are combination trucks and 59 percent of the Pennsylvania vehicles are combination trucks.
TABLE 2-2
Vehicle Types by State
Florida Pennsylvania Total
Tractor-Trailer Combination 6,444 52% 5,349 59% 11,793
Other Vehicles 5,936 48% 3,785 41% 9,721
Total Vehicles 12,380 100% 9,134 100% 21,514
State regulations in both Florida and Pennsylvania allow truck tractors traveling in their state to
pull one or two trailers. Table 2-3 shows that most tractors pulled one trailer.
TABLE 2-3
Tractor-Trailer Combinations by Number of Trailers and State
Number of Trailers Florida Pennsylvania Total
One Trailer 6,363 99% 5,191 97% 11,554
Two Trailers 81 1% 158 3% 239
Total Vehicles 6,444 100% 5,349 100% 11,793
12
Table 2-4 compares the characteristics of the trailers in Florida and Pennsylvania. For the most
part, the trailers were similar: vans were the most common type, followed by flatbeds. The trailer
model years were also similar: slightly more that half of the trailers in Florida are pre-standard,
not required to have tape. (Most of the unknown model years in Pennsylvania are probably model
years before 1981. In Pennsylvania, trailer model year is decoded from the trailer VIN. Model
years earlier than 1981 cannot be decoded from the VIN since the VIN was not standardized
before 1981.)
The most noteworthy difference is that 60 percent of the trailers in Florida were equipped with
tape, while 70 percent in Pennsylvania were equipped with tape. This is partly due to the data
collection starting and ending six months later in Pennsylvania, by which time more trailers had
been retrofitted.
TABLE 2-4
Number and Percent of Trailers by Trailer Characteristics and State
Florida Pennsylvania
Trailer Type
Van 3,414 52% 3,655 66%
Flatbed 1,174 18% 847 15%
Tanker 580 9% 348 6%
Dump 550 8% 218 4%
Auto Transporter 163 2% 60 1%
Other/Unknown 644 10% 379 7%
Trailer Year
Pre-Standard < MY 1993 3,504 54% 1,504 27%
Transition Year MY 1993 349 5% 254 5%
Transition Year MY 1994 442 7% 376 7%
Post-Standard > MY 1994 2,022 31% 1,865 34%
Unknown 208 3% 1,508 27%
Treatment
Tape 3,925 60% 3,859 70%
No Tape 2,478 38% 1,326 24%
Unknown 122 2% 322 6%
Table 2-5 shows the characteristics of the trailers with tape by state. Almost all have the tape
applied to the side and rear of the trailer as required by FMVSS 108 or similar to the standard.
(The criteria the police officers used to judge whether the tape was similar or dissimilar to
FMVSS 108 is not known.) About 90 percent or more have the required alternating red and
white tape. Very few trailers have tape that is “very dirty.” About 30 percent of the trailers have
tape that is somewhat dirty. The extent, if any, to which dirt on the tape may reduce its
effectiveness will be analyzed in Chapter 4.
13
TABLE 2-5
Tape Characteristics on Trailers Equipped with Tape
Tape Pattern
FMVSS 108 or Similar 3,627 92% 3,579 93%
Other 223 6% 202 5%
Unknown 75 2% 78 2%
Tape Color
Red/White 3,676 94% 3,477 90%
White, Orange, or Blue 157 4% 267 7%
Other/Unknown 92 2% 115 3%
Tape Condition Rear
No Tape* 10 0% 22 1%
Clean 2,575 66% 2,440 63%
Some Dirt 1,103 28% 1,226 32%
Very Dirty 167 4% 117 3%
Unknown 70 2% 54 1%
Tape Peeling/Missing on Rear of Trailer
No Tape* 10 0% 22 1%
Not Peeling, Missing 3,565 91% 3,612 94%
Peeling, Missing 330 8% 196 5%
Unknown 20 1% 29 1%
Tape Peeling/Missing on Rear Underride Guard of Trailer
No Tape* 11 0% 22 1%
Unknown 30 1% 52 1%
Tape Peeling/Missing on Rear of Trailer, Other than Underride Guard
No Tape* 16 0% 26 1%
Unknown 29 1% 53 1%
Tape Condition Side
No Tape* 54 1% 96 2%
Clean 2,570 65% 2,384 62%
Some Dirt 1,071 27% 1,164 30%
Very Dirty 111 3% 82 2%
Unknown 119 3% 133 3%
Tape Peeling/Missing Side
No Tape* 54 1% 104 3%
Unknown 68 2% 115 3%
* Trailers has tape in some locations, but not on this component (e.g. tape on side only– no tape
on rear or underride guard).
14
There are two small but notable differences between the two states. The percentage of trailers
with tape that have some dirt is slightly higher in Pennsylvania than in Florida. In Pennsylvania,
more dirt and grime may be splashed up on roads treated with salt or other chemicals to reduce
snow, slush and ice on the roads during the winter. The other difference is the percentage of
trailers with peeling or missing tape. The percentages are slightly lower in Pennsylvania than
Florida. It is unknown why this happened. It is possible that combination trucks in Pennsylvania
travel further between loading/unloading, thus reducing the chances that the tape will be damaged.
It is also possible that Pennsylvania has more recently retrofitted trailers since data collection
started and ended six months later in Pennsylvania than Florida. These trailers may not have had
enough time to develop any peeling or missing segments.
Table 2-5 also shows that peeling or missing segments of tape occur more often on the rear of the
trailer than the side, as expected. The rear of the trailer is more susceptible to scraping and
damage when the trailer is loaded and unloaded. Eight percent of the trailers in Florida and five
percent in Pennsylvania had peeling or missing segments on the rear of the trailer. Only four
percent of the trailers with tape in Florida and one percent in Pennsylvania had peeling or missing
segments on the side. The tape on the rear other than the underride guard appears to be just as
vulnerable as tape on the rear underride guard.
Table 2-6 shows the percentage of missing tape on the rear by state for trailers with missing tape.
The 100 percent missing at a particular location may represent trailers that never had tape at that
particular location but had tape elsewhere on the trailer. The form is not entirely clear and some
officers may have misunderstood the information requested. A large proportion of the trailers
coded with 100 percent missing tape on the rear underride guard were pre-standard trailers.
Seventy percent (50 cases) in Florida and 64 percent (16 cases) in Pennsylvania are pre-standard
trailers, which are not required to have tape on the rear underride guard according to the FMCSA
retrofit standard. Similar results were found for elsewhere on the rear of the trailer. Seventy-
eight percent (31 cases) in Florida and 50 percent (6 cases) in Pennsylvania of the trailers coded
with 100 percent missing tape are pre-standard trailers. These may be trailers that were partially
retrofitted with tape before FMCSA announced its final rule and have not yet added the missing
tape to make it compliant.
If you only consider the 1 to 99 percent categories, then 55 percent (84/152 = 55 % in Florida and
42/77 = 55% in Pennsylvania) of the trailers with missing tape on the underride guard have 75
percent or more of the tape intact. This amount of tape missing will probably not seriously
diminish its conspicuity effectiveness. On the other hand, tape that is missing on more than 25
percent of the guard may reduce the effectiveness of the tape. In this case, 45 percent (68/152 =
45 percent in Florida and 35/77 = 45 percent in Pennsylvania) of the trailers with missing tape on
the underride guard have more than 25 percent of the tape missing. Elsewhere on the rear, 46
percent in Florida of the trailers with missing tape and 48 percent in Pennsylvania have 75 percent
or more of the tape intact and 54 and 52 percent, respectively, have more than 25 percent of the
tape missing.
15
TABLE 2-6
Number and Percent of Trailers with Missing Tape
by Percentage of Missing Tape and State
Percentage of Missing Tape on Rear Underride Guard
1-25 % 84 35% 42 32%
26-50% 46 19% 20 15%
51-75% 10 4% 8 6%
76-99% 12 5% 7 5%
100% 71 30% 25 19%
Unknown Percentage 14 6% 28 22%
Percentage of Missing Tape on Rear - Other than Underride Guard
1-25 % 61 34% 32 37%
26-50% 45 25% 25 29%
51-75% 13 7% 3 3%
76-99% 13 7% 6 7%
100% 40 22% 12 14%
Unknown Percentage 7 4% 8 9%
Overall, missing tape was not a big problem when this data was collected. Most trailers with tape
either had no tape missing at all or they had 75 percent or more of the tape intact on the rear.
Less than two percent of the 3,925 (68/3,925 = 1.7 percent) trailers with tape in Florida had tape
missing on more than 25 percent of the rear guard. In Pennsylvania, less than a percent (35/3,859
= 0.9 percent) had tape missing on more than 25 percent of the rear guard. The same percentages
were found for elsewhere on the rear in each state. However, this could increase as the trailers
get older.
Table 2-7 shows the percentage of trailers treated by trailer model year. Model year is the best
available surrogate for date of manufacture. NHTSA requires all trailers manufactured on and
after December 1, 1993 to have tape. Model years 1993 and 1994 are considered transition years
because most 1993 and some 1994 trailers were manufactured before December 1, 1993. In this
table, model year was decoded from “good” trailer VINs that yielded a valid model year and valid
trailer make.
As expected, the percentage of trailers with tape is increasing by model year. Most of the post-
standard trailers have tape. More than 80 percent of the model year 1994 trailers have tape and
more than half of the model year 1993 trailers have tape. Slightly fewer than half of the pre-
standard trailers have tape.
Also as expected, more pre-standard and transition-year trailers have tape in Pennsylvania than in
Florida. Pennsylvania’s data collection period was 6 months later than the Florida period and
ended on November 30, 1999, eight months after the FMCSA published a retrofit regulation on
16
March 31, 1999. Although the regulation gives motor carriers two years from June 1, 1999 to
install the material on trailers manufactured prior to December 1, 1993, it appears that motor
carriers were quickly retrofitting their trailers with tape.
Table 2-7 also shows a small portion of post-standard trailers do not have tape (5 percent in
Florida and 3 percent in Pennsylvania). These could be coding errors but not necessarily.
Possible coding errors included miscoding of the tape presence/absence or inaccurate copying of
the VIN. On the other hand, some of these trailers may be exempt from the tape requirement.
Trailers designed exclusively for living or office space are not required to have conspicuity
treatment. Pole trailers, trailers that carry logs, are also exempt. Some of these trailers may have
reflex reflectors, instead of tape, an option allowed by the standard. Finally, some of these trailers
may be non-compliant, although this is unlikely. Probably some combination of these possibilities
or errors in the data account for the post-standard trailers that do not have tape.
TABLE 2-7
Trailer Treatment by Trailer Model Year and State
Model Year Decoded from Good VINs
Pre-Standard Transition Year Transition Year Post-Standard
< MY 1993 MY 1993 MY 1994 > MY 1994
Florida
Tape 747 48% 144 59% 246 84% 1,180 94%
No Tape 787 51% 99 41% 46 16% 66 5%
Unknown 13 1% 1 0% 1 0% 11 1%
Pennsylvania
Tape 540 49% 113 66% 261 89% 1,255 95%
No Tape 540 49% 56 33% 29 10% 43 3%
Unknown 12 1% 2 1% 2 1% 19 1%
Table 2-8 shows some of the crash characteristics of combination truck crashes. Most of the
crashes occur on a weekday, when the weather is clear, and involve two or more vehicles. More
than two-thirds of the crashes occur on roads with speed limits of 50 mph or higher. It appears
that most of the crashes occur on roadways with limited access. In Pennsylvania, 67 percent
occur on limited access roadways and in Florida the exact percentage is hard to estimate since
most of the interstates and turnpikes and some of the U.S. and State roads have limited access.
For the most part, combination trucks are on major roads when they are involved in crashes.
There are two minor differences between the crash experience of combination trucks in Florida
and Pennsylvania. Pennsylvania has more adverse weather cases than Florida. This is no surprise
since the climate is different there. Snow is common in Pennsylvania during the late fall and
winter, but rare in Florida. Pennsylvania also has proportionately more single-vehicle crashes than
Florida. This disparity is mostly due to reporting differences. Crashes are reported in Florida if
17
there is $500 worth of property damage. In Pennsylvania, crashes are reported if there is at least
one vehicle towed from the scene because of damage. Therefore, there are more low-damage,
multiple vehicle fender-benders reported in Florida than Pennsylvania.
TABLE 2-8
Number and Percent of Crashes by Crash Characteristics and State
Day of Crash
Weekday 5,334 88% 4,184 86%
Weekend 756 12% 679 14%
Unknown 5 0% 1 0%
Weather Conditions
Clear 5,004 82% 3,416 70%
Adverse 973 16% 1,374 28%
Unknown 118 2% 74 2%
Number of Vehicles
1 864 14% 1,348 28%
2 4,375 72% 2,973 61%
3 or more 856 14% 543 11%
Speed Limit
0 to 49 mph 1,743 29% * 1,205 25%
50 - 70 mph 4,187 69% 3,581 74%
Unknown 165 3% 78 2%
Roadway Access Type
Unlimited 1,487 31%
Limited 3,245 67%
Other/Unknown 132 3%
Roadway Type
Interstate & Turnpike/Toll 2,461 40%
U.S. & State 2,548 42%
County & Local 767 13%
Other/Unknown 319 5%
* Total percentages may not add to 100 percent due to rounding.
Table 2-9 compares the driver’s age in the other vehicle involved in combination truck crashes.
There is little difference between the driver’s age in Florida and Pennsylvania except for the older
drivers. There are slightly more 61-70 year old and 70 and older drivers in Florida than in
Pennsylvania. This is not surprising since Florida has an older population.
18
TABLE 2-9
The Driver’s Age in the Other Vehicle Involved
in a Crash with Combination Truck by State
Driver’s Age
15-20 303 7% 194 8%
21-25 449 11% 332 13%
26-30 471 11% 267 11%
31-50 1,612 38% 962 38%
51-60 446 11% 288 12%
61-70 304 7% 136 5%
70 + 324 8% 132 5%
Unknown 298 7% 193 8%
Table 2-10 shows the distribution of driver’s age for passenger vehicles in two vehicle crashes
investigated by State police that do not involve combination trucks. This data is from the
NHTSA Florida and Pennsylvania State data files. The percentage of 15-20 year old drivers in
Table 2-10 is twice that in Table 2-9. Younger drivers are underrepresented as drivers of other
vehicles involved in crashes with a combination truck. A possible explanation is that younger
drivers, especially 15-18 year old drivers, less frequently encounter combination trucks because
their driving habits are different than combination truck drivers. Young drivers drive a lot on
local roads to and from school and work. The percentage of all the other age groups is fairly
consistent between the two tables.
TABLE 2-10
Driver’s Age of Passenger Vehicles Involved in Two-Vehicle Crashes
Investigated by State Police that do not Involve a Combination Truck by State
Driver’s Age
15-20 15,387 14% * 5,055 18%
21-25 13,226 12% 3,278 12%
26-30 12,680 12% 2,845 10%
31-50 41,835 39% 9,792 36%
51-60 10,692 10% 2,747 10%
61-70 6,878 6% 1,625 6%
70 + 6,772 6% 1,919 7%
Unknown 0 0% 273 1%
* Total percentages may not add to 100 percent due to rounding.
19
2.3 CRASHES NOT INVESTIGATED BY STATE POLICE
The previous section gave a general overview of the crashes investigated by the FHP and PSP.
But the conspicuity data do not include combination truck crashes investigated by local law
enforcement agencies, other state agencies, or reported by the drivers themselves. This section is
an overview of those crashes. Specifically, this section will address the following: What type of
crashes do local law enforcement officers investigate? Are they different from the ones that state
law enforcement officers investigate? If they are different, how will these differences influence
our effectiveness estimates?
The State Police investigate more than half of all the crashes that involve a combination truck.
The FHP investigated 59 percent of the combination truck crashes in Florida during 1998. The
PSP investigated 60 percent of the crashes in Pennsylvania during 1998. Table 2-11 shows the
percent of crashes investigated by the State police and others in Florida and Pennsylvania for
various crash characteristics. The FHP and the PSP investigate almost all of combination truck
crashes that occur on major roads where they have jurisdiction. The PSP investigate 93 percent
of the crashes on limited access roads and the FHP investigate 93 percent on interstates and
turnpikes/toll roads. These agencies also investigate more than half of all the crashes for most of
the crash characteristics in Table 2-11. They investigate more than half of the weekday, weekend,
clear, adverse, daylight, dark, high speed, rural, etc. crashes that involve a combination truck.
The conspicuity data underrepresents crashes that occur in urban areas, on county and local
roads, or on roadways with lower speed limits. The majority of these crashes are investigated by
local police, sheriffs, or other police agencies that have jurisdiction in these areas or are reported
by the drivers themselves. The local police agencies investigate somewhere between 64 and 74
percent of the crashes that occur in urban areas, 68 percent on county and local roads, and about
65 percent on roads with speed limits less than 50 miles per hour.
There is little difference between what the FHP and PSP investigate in their respective states. The
only exceptions appear to be single-vehicle crashes and crashes that occur in urban areas. The
PSP investigates 74 percent of the single-vehicle crashes that involve a combination truck, while
the FHP investigates only 60 percent of these crashes. The local agencies in Florida also
investigate more crashes that occur in urban areas than their counterparts in Pennsylvania. The
reporting threshold differences between these states probably account for these discrepancies. A
crash in Florida is investigated if it involved at least $500 worth of damage. In Pennsylvania, a
crash is investigated if at least one of the vehicles is towed away. Therefore, local police agencies
in Florida will investigate more low-speed, single-vehicle crashes or low-speed crashes in urban
areas than the local police agencies in Pennsylvania.
The fact that a proportion of crashes are not investigated by the State police will not significantly
affect our effectiveness estimates. In Chapter 4, we will see no consistent difference in the
effectiveness in rural vs. urban areas or on low vs. high speed limit roads, so the state police
effectiveness estimates are also appropriate for the groups of crashes investigated by other
20
agencies. When we complete benefits in Chapter 5, we will include non-state police investigated
crashes in the “size of the problem” estimate, so as not to underestimate the benefits.
TABLE 2-11
Percent of Crashes Reported by Enforcement Agency and State
Day of Crash FHP Other PSP Other
Weekday 58% 42% 59% 41%
Weekend 64% 36% 72% 28%
Weather Conditions
Clear 56% 44% 58% 42%
Adverse 65% 35% 67% 33%
Number of Vehicles
1 60% 40% 74% 26%
2 57% 43% 55% 44%
3 or more 67% 33% 57% 43%
Speed Limit
0 to 49 mph 35% 65% 34% 66%
50 to 70 mph 89% 11% 84% 16%
Roadway Access Type
Unlimited 55% 45%
Limited 93% 7%
Other/Unknown 61% 39%
Roadway Type
Interstate & Turnpike/Toll 93% 6%
U.S. & State 63% 37%
County & Local 31% 68%
Other/Unknown 19% 81%
Light Condition
Daylight 57% 43% 55% 45%
Dark Conditions 64% 36% 70% 30%
Rural/Urban
Rural 77% 23% 83% 17%
Urban 26% 74% 36% 64%
21
22
CHAPTER 3
BASIC ANALYSIS
Tractor-trailer combinations in which trailers are equipped with retroreflective tape ought to
experience a reduction of side and rear impacts into the trailer by other vehicles in dark conditions
- relative to the number that would have been expected if the trailers had not been equipped. The
analytic challenge is to compute the “expected” number of impacts and quantify the reduction.
The critical parameters are: (1) whether or not the trailer is tape-equipped; (2) the light condition
- dark (comprising “dark-not-lighted,” “dark-lighted,” “dawn” and “dusk”) vs. daylight; and (3)
relevant vs. non-relevant crash involvements. Relevant crash involvements are those where
another vehicle crashed into the side or rear of a heavy trailer, because the tape can help the other
driver see and possibly avoid hitting the trailer. The non-relevant group consists of single-vehicle
crashes of tractor-trailers (where visibility of the tractor-trailer to other road users is not an issue
at all) and impacts of the front of the tractor into other vehicles (where conspicuity of the side and
rear of the trailer is also not an issue). Each of these parameters defines a sort of control group.
The vehicles without tape are a control group that can be compared to the vehicles equipped with
tape. Since the tape ought to have substantially less effect (if any) by daylight than in dark
conditions, daylight crashes are a control group relative to crashes in the dark. The most
satisfactory definition of the “expected” number of side/rear impacts in the dark uses all three of
these control groups.
3.1 DEFINITIONS
The data base for this analysis is a vehicle-oriented file, with one record for each tractor-trailer
combination that was involved in a crash. Initially, Florida and Pennsylvania data will be analyzed
separately. The Florida file includes 6,444 tractor-trailer combinations and the Pennsylvania file
includes 5,349.
The critical parameters that must be defined for this analysis are tractor-trailer combinations,
trailer treatment, light conditions and crash mode/point of impact. Some of the parameters can be
defined from data elements on the NHTSA Green and some from data elements the state PAR.
The definitions for the parameters that are defined from elements on the NHTSA Green are the
same in Florida and Pennsylvania. But each state has it own unique way of coding elements on its
PAR, so the definitions for parameters that are defined from elements on the state PAR cannot be
exactly the same. For these parameters, the definitions ought to be made as similar as possible.
The States also differ in the exposure and crash characteristics of combination trucks. Below are
the definitions and the differences found in the conspicuity data for these critical parameters.
First, it is necessary to define “tractor-trailers.” The vehicle type and number of trailers were used
to identify heavy trucks pulling at least one trailer. “Vehicle type” is a data element on the state
PAR and identifies the body style of a vehicle. “Number of trailers” is on the NHTSA Green form
23
and identifies the number of trailers that were attached to a vehicle. (Small trailers typically pulled
by light vehicles are not included in the analysis.) Tractor-trailer combinations were defined as
follows:
State Definition
Florida veh_type = 4, 5, 6, 77 and trlno = 1, 2

Pennsylvania body_type=70, 73, 74, 75, 76, 77, 79 and trlno = 1, 2
The trailer treatment8 (the presence or lack of retroreflective tape) is easily identified by a variable
collected on the NHTSA Green. If the truck tractor was pulling only one trailer and the trailer
had tape, then the combination truck was classified as treated. If the trailer had no tape, then the
combination truck was classified as untreated. Few tandem trailers were found in the conspicuity
data although both Florida and Pennsylvania allow heavy trucks to pull up to 2 trailers. In these
cases, the presence or lack of tape was defined for both trailers. If both trailers had tape, then it
was classified as treated and if both trailers did not have tape, then it was classified as untreated.
A truck pulling two trailers where one trailer had tape and the other one did not were classified as
unknown, as are combinations with unknown tape conditions on a single trailer or at least one of
the tandem trailers. Table 2-3 shows that fewer than 5 percent of the combination trucks are
pulling two trailers and very few cases were classified as unknown because the tape was
inconsistent on the two trailers. Table 3-1 shows that 60 percent of truck trailer combination
trucks have tape in Florida and 70 percent have tape in Pennsylvania. Combination trucks with
“unknown” trailer treatment were excluded from the analyses.
TABLE 3-1
Retroreflective Tape by State for Tractor-Trailer Combinations
Treated 3,880 60% 3,751 70% 7,631
Untreated 2,443 38% 1,283 24% 3,726
Unknown 121 2% 315 6% 436
Total Vehicles 6,444 5,349 11,793
The light conditions at the time of the crash were classified into “dark” and “daylight” conditions
for the analysis using a variable on the NHTSA Green. Table 3-2 shows the different levels of
light conditions along with the number and percentage of these cases. Note the difference in the
8
Conspicuity treatment of truck tractors is arguably a neutral factor in evaluating the
effectiveness of the treatment in preventing crashes involving truck tractor-trailer combinations.
There is no side treatment applied to truck tractors, while the rear treatment is masked by the
trailer being towed.
24
percentage of daylight and “dark-not-lighted” conditions between Florida and Pennsylvania.
Although there is no exposure data to support this premise, this suggests that combination trucks
are driven more often at nighttime in Pennsylvania than in Florida.
TABLE 3-2
Light Conditions by State for Tractor-Trailer Combinations
Daylight 4,408 68% 3,139 59% 7,547
Dark-Not-Lighted 1,245 19% 1,628 30% 2,873
Dark-Lighted 505 8% 334 6% 839
Dawn 179 3% 147 3% 326
Dusk 91 1% 86 2% 177
Unknown 16 0% 15 0% 31
Total Vehicles 6,444 5,349 11,793
The basic analysis compares dark conditions to daylight conditions. “Dark” conditions include
“dark-not-lighted,” “dark-lighted,” “dawn,” and “dusk.” Other analyses will compare dark-not-
lighted to daylight; and dark-lighted, dawn, and dusk to daylight. Cases with unknown light
conditions are excluded from the analysis.
Crash mode/point of impact were used to identify single-vehicle cases and frontal, side, and rear
initial impacts. Single-vehicle cases were identified first, regardless of their initial point of impact.
A single-vehicle case involves only a tractor-trailer combination and no other vehicle.
Frontal, side, and rear initial impacts were identified from the remaining cases. The remaining
cases involved two or more vehicles where at least one of the vehicles is a tractor-trailer
combination. The initial impact area refers to the initial point of impact for the tractor-trailer
combination. The analysis used a combination of the PAR’s “point of impact” and the NHTSA
Green’s derived “impact point” to identify the initial impact area. If the PAR’s “point of impact”
is “Trailer,” “Towed Unit,” or “Unknown” then the derived impact point from the NHTSA Green
is used to classify the area. In the conspicuity data base, the variable from the PAR is called
“impact” and the variable from the NHTSA Green is called “impactpt.” The definitions for the
four categories are as follows:
Category State Definition
Single-Vehicle FL the number of vehicles involved by counting is 1

PA num_veh = 1
Front FL impact = 1,2,14 or (impact = 22, 99 and impactpt = 1)

PA impact = 11,12,1 or (impact = 15, 99 and impactpt = 1)
25
Side FL impact = 3,4,5,6,10,11,12,13 or (impact = 22, 99 and impactpt = 2,4)
PA impact = 2,3,4,8,9,10 or (impact = 15, 99 and impactpt = 2,4)
Rear FL impact = 7,8,9 or (impact = 22, 99 and impactpt = 3)

PA impact = 5,6,7 or (impact = 15, 99 and impactpt = 3)
Let us refer to these categories as damage areas, since three of the four categories are damage
areas. Table 3-3 shows that Pennsylvania has a higher percentage of single-vehicle crashes than
Florida and that Florida has a higher percentage of other/unknown crash types. But the
percentage of frontal impacts is about the same in both states. Florida has fewer rear impacts but
more side impacts than Pennsylvania. However, the sum of the side and rear impacts is the same
in both States, 39 percent. The basic analysis compares side and rear damage areas to single-
vehicle and front damage areas. Cases with unknown damage areas are excluded from the
analysis.
TABLE 3-3
Damage Area by State for Tractor-Trailer Combinations
Single-Vehicle 846 13% 1,335 25% 2,181
Front 2,240 35% 1,758 33% 3,998
Side 1,768 27% 992 19% 2,760
Rear 774 12% 1,063 20% 1,837
Other/Unknown 816 13% 201 4% 1,017
Total Vehicles 6,444 5,349 11,793
After excluding cases with unknown trailer treatment, light condition or damage area, there
remain 5,535 cases in Florida and 4,871 in Pennsylvania available for the analysis because they
have known values for each of the three critical parameters. In all, 14 percent of the Florida cases
and 9 percent of the Pennsylvania cases are excluded because of unknowns.
3.2 THE BASIC ANALYSIS
The basic analysis tabulated tractor-trailer combination involvements in crashes by trailer

treatment (treated; untreated) and damage area (side and rear impacts; single-vehicle and frontal
impacts) and light conditions (dark; light). The basic contingency table is:
26
Number of Crash Involvements
Single-Vehicle &
Treated Side & Rear Impacts Frontal Impacts
Dark N111 N112
Daylight N121 N122
Untreated
Dark N211 N212
Daylight N221 N222
In Florida, the numbers in the basic contingency table are:
Number of Crash Involvements in Florida
Single-Vehicle (SV) &

Dark 526 632
Daylight 1,024 1,231
Untreated
Dark 285 314
Daylight 648 875
First, let us consider only the untreated vehicles. There are 648 side/rear impacts in the daylight.
Based on the involvements in SV/frontal impacts, which act as a control group or measure of
induced exposure, we might have expected (314/875)648 = 232.5 side/rear impacts in the dark.
In fact, there are 285. Because untreated vehicles are harder to see in the dark (except from the
front), they are excessively impacted in the side and rear, relative to exposure, by a factor of
285/232.5, i.e.
(285/648) / (314/875) = (N211 / N221)/(N212 / N222).
Now, let us consider the treated vehicles. There are 1024 side/rear impacts in the daylight.
Again using the SV/frontal impacts as a control group, we would expect (632/1231)1024 = 525.7
side/rear impacts in the dark. And that is almost exactly what happened: 526. Because the
treated vehicles are quite visible in the dark, they have no more side/rear impacts in the dark than
might have been expected based on their exposure. Clearly, there is a reduction of side/rear
impacts in dark conditions relative to the pattern in the untreated vehicles.
Finally, the data on the untreated and the treated vehicles can be combined to estimate the
reduction in side/rear impacts for the treated vehicles in dark conditions. As stated above, the
excess of side/rear crashes in the dark for the untreated trailers is 285/232.5 = 1.23. If the
treatment were of no value at all, we would expect the same excess of side/rear impacts in the
27
dark as in the untreated vehicles. The effectiveness of the treatment is the extent to which that
excess is reduced - i.e., the relative difference of the actual-to-expected ratios in the treated and
the untreated vehicles:
(526/1024) / (632/1231)
1 - ____________________ = 18 percent
(285/ 648) / (314/ 875)
The expected number of side and rear impacted treated tractor-trailer combinations in dark
conditions is
N121 * (N112 / N122) * [(N211 / N221)/(N212 / N222)].
In fact, there are only N111 side and rear impacted treated tractor-trailer combinations in dark
conditions. That is a reduction of
 ( N 111 N 121) 
 
 ( N 112 N 122 )

1−
 ( N 211 N 221) 

 ( N 212 N 222)
dark, side and rear impact involvement rate. All three variables (trailer treatment, damage area,
and light conditions) are needed for this analysis of effectiveness. An analysis missing any one of
these variables would not give accurate effectiveness estimates and should be avoided.
In Pennsylvania, the basic contingency table is:
Number of Crash Involvements in Pennsylvania
Single-Vehicle &
Dark 512 977
Daylight 873 1,271
Untreated
Dark 214 283
Daylight 282 459
Here, the reduction in side/rear impacts for treated combination trucks in dark conditions is
28
 (512 873) 
 
 ( 977 1,271) 
1− = 38 percent.
 (214 282 ) 

 (283 459)
Tractor-trailer combination with treated trailers in Pennsylvania had a 38 percent lower risk of
being involved in a side or rear impact collision in the dark than combination trucks with
untreated trailers.
3.3 STATISTICAL SIGNIFICANCE AND CONFIDENCE BOUNDS
Statistical significance is tested using the CATegorical data MODeling (CATMOD) procedure in
SAS.9 This procedure is similar to the chi-square test in that it estimates the predicted values
from the observed values and tests to see if at least one of the observed values is different from its
predicted value. The simple chi-square test cannot be used because we have three categorical
variables. The CATMOD procedure allows us to analyze categorical data that can be represented
by a contingency table. It fits linear models to functions of response frequencies and tests the
likelihood ratio. The likelihood ratio tests if the variables are independent.
Table 3-4 shows the results of the CATMOD procedure using the basic contingency table on
Page 27 for dark conditions in Florida. A log-linear model was specified to reproduce the
predicted cell frequencies. The model included all main effects (DAMAGE, LIGHT, and
TREAT) and two-variable interactions (DAMAGE*LIGHT, DAMAGE*TREAT, and
LIGHT*TREAT). The three-variable interaction was not included in the model. The null
hypotheses is that the three variables are independent. In other words, the number of side and
rear combination truck crashes in dark conditions with or without tape are the same and that tape
effectiveness is zero. The model is weighted by N, the number of combination trucks in each cell.
9
SAS/STAT Users’s Guide: Volume1, ACECLUS-FREQ, Version 6, Fourth Edition, SAS
Institute Inc., Cary, NC, 1990.
29
TABLE 3-4
Florida Data
Log-linear Model Without Three-variable Interaction
Maximum-likelihood analysis-of-variance table
Dependent Variables: DAMAGE, LIGHT, and TREAT
Model: DAMAGE, LIGHT, TREAT, DAMAGE*LIGHT, DAMAGE*TREAT, and LIGHT*TREAT
N of Observations: 5,535
Weighting Factor: N (Number of Combination Trucks)

Source DF Chi-Square Prob
DAMAGE 1 44.52 0.0000
LIGHT 1 692.23 0.0000
DAMAGE*LIGHT 1 1.60 0.2063
TREAT 1 304.76 0.0000
DAMAGE*TREAT 1 0.95 0.3295
LIGHT* TREAT 1 19.43 0.0000
Likelihood Ratio 1 2.81 0.0937
The analysis-of-variance table above shows that the model of independence does not fit since the
likelihood-ratio test for the three-variable interaction is significant. NHTSA’s evaluations of
safety standards customarily use a one-tailed 95 percent test (or a two-tailed 90 percent test) as
the criterion for statistical significance. Chi-square is a two-tailed test and a value of 2.81 with its
associated probability of 0.0937 is significant at the two-tailed 90th percentile (or one-tailed 95th
percentile). Therefore, we reject the null hypotheses. We conclude that the tape is 18 percent
effective under dark conditions in Florida and this is significantly greater than zero.
The tape effectiveness estimate is also significantly greater than zero in Pennsylvania for dark
conditions. Here, the tape was 38 percent effective. The CATMOD procedure fit the same log-
linear model for the Pennsylvania data. The analysis-of-variance table showed that the likelihood-
ratio is 12.13 with a probability of 0.0005. Again, we conclude that the three-variable interaction
is significant and it is significant at the 99th percentile.
In order to calculate the confidence bounds around our effectiveness estimates, let Nijk represent
each cell in the basic contingency table where i = trailer treatment, j = light conditions, and k =
damage area. Each Nijk is Poisson distributed and independent from the others. The relative
variance of Nijk is
Var (Nijk) / N2ijk = Nijk / N2ijk = 1 / Nijk.
30
10
Var (r )
= 1 + 1 + ...+ 1
r2 N 111 N 112 N 222
 
Var ( r ) = r 2  1 + 1 + ...+ 1
 N 111 N 112 N 222
 
s. d .(r ) = r  1 + 1 + ...+ 1 
 N 111 N 112 N 222 
and the standard deviation of e is
 
s. d .(e) = (1 − e) 1 + 1 + ...+ 1 .
 N 111 N 112 N 222 
For the Florida data, the standard deviation of e is
s. d .( e) = (1− .18) 11231 + 11024 + 1632 + 1526 + 1875 + 1648 + 1314 + 1 285 = .099.
Using a 90 percent confidence interval, the upper and lower confidence bounds for effectiveness
in Florida are 18% ± (1.645)*0 .099 = 18 ± 16 percent. Thus, the lower bound is 2 percent and
the upper bound is 34 percent.
There is slightly less variation in the Pennsylvania data than the Florida data resulting in a slightly
smaller standard deviation and smaller bounds. The standard deviation in Pennsylvania is .085
and the confidence bounds are 38% ± (1.645)*0 .085 = 38 ± 14 percent. The lower bound is 24
percent and the upper bound is 52 percent.
Hansen, Morris H., Hurwitz, William N., and Madow, William G., Sample Survey
, John Wiley & Sons, New York, 1953, pp. 512-514.
31
3.4 TAPE EFFECTIVENESS IN SPECIFIC DARK CONDITIONS
The effective estimates calculated in Section 3.2 are for a combination of four dark conditions
including dark-not-lighted, dark-lighted, dawn, and dusk. Is the tape equally effective in each of
these four conditions? If not, which conditions make it more effective and which conditions make
it less effective?
Table 3-5 shows the basic contingency table and effectiveness estimates for dark-not-lighted
conditions. These are crashes that occur at night where the roadway is not illuminated by street
lights. The tape should be the most beneficial in these conditions since the contrast
reflectivity of the tape should be brightest. The tape is 37 percent effective at reducing crashes in
Florida and 44 percent in Pennsylvania. The results are both statistically significant at the .01
level, so the tape is highly effective on dark unlighted roads.
Table 3-6 shows the basic contingency table and effectiveness estimates for the combination of
dark-lighted, dawn, and dusk conditions. Here the results are mixed. The tape has a negative
effect in Florida and a positive effect in Pennsylvania. Neither result is significantly different from
zero, so the tape has no substantial benefit when some light is available.
TABLE 3-5
Effectiveness of Tape in Dark-Not-Lighted Conditions by State
Treated Side & Rear SV & Front Side & Rear SV & Front
Dark-Not-Lighted 272 417 357 768
Daylight 1,024 1,231 873 1,271
Untreated
Dark-Not-Lighted 190 207 157 213
Daylight 648 875 282 459
Effectiveness 37% 44%
TABLE 3-6
Effectiveness of Tape in Dark-Lighted, Dawn, and Dusk Conditions by State
Dark-Lighted, Dawn & Dusk 254 215 155 209
Daylight 1,024 1,231 873 1,271
Untreated
Dark-Lighted, Dawn & Dusk 95 107 57 70
Daylight 648 875 282 459
Effectiveness -18% 19%
32
Table 3-7 shows the effectiveness of the tape at three different lighting conditions for comparison.
The tape is most effective in dark-not-lighted conditions. The dark-not-lighted conditions have
the largest effectiveness estimates. The tape is somewhat beneficial in the combination of all four
dark conditions. Here, the effectiveness estimates are less than the dark-not-lighted estimates, but
significantly greater than zero. At dark-lighted, dawn, or dusk conditions, the effectiveness
estimates are not statistically different from zero. Pennsylvania’s estimate suggests there may be
some benefit, but Florida’s estimate suggests no benefit from the tape under these conditions.
TABLE 3-7
Tape Effectiveness by Light Conditions and State
Dark 18% S* 38% S
Dark-Not-Lighted 37% S 44% S
Dark-Lighted, Dawn, & Dusk -18% NS 19% NS
S - Statistically Significant at .01 level.
S* - Statistically Significant at .05 level.
NS - Not Statistically Significant.
3.5 COMBINED EFFECTIVENESS AND “BEST” ESTIMATES
Ideally, the analysis should produce a single estimate of the effectiveness of the tape in both
states, combined. One way to do this is to simply combine the Florida and Pennsylvania data and
calculate the effectiveness by the method described in Section 3.2 with the pooled data. Table 3-8
shows the effectiveness estimates under the three different lighting conditions using the pooled
Florida and Pennsylvania data. The individual Florida and Pennsylvania results are also included
in the table.
TABLE 3-8
Tape Effectiveness by Light Conditions
Pooled Pennsylvania Florida
Dark 29% S 38% S 18% S*
Dark-Not-Lighted 41% S 44% S 37% S
Dark-Lighted, Dawn, & Dusk -3% NS 19% NS -18% NS
33
The tape is highly effective in preventing crashes in dark conditions and especially in dark-not-
lighted conditions using the combined data. These pooled effectiveness estimates are statistically
significant. The tape does not appear to prevent crashes in dark-lighted, dawn, and dusk
conditions. All of the pooled effectiveness estimates are in between the ones found in each state.
In the dark and dark-not-lighted conditions, the pooled effectiveness is almost exactly equal to the
average of the two states’ effectiveness.
But the simple method of pooling the data may not be the most accurate. It is possible that some
other factors in the states may be confounded with the effectiveness of the tape and therefore
simply pooling the data may not be appropriate.
The CATMOD procedure was used on the combined data sets to test if the effectiveness estimate
for dark conditions varies by State. The variable STATE was added to the log-linear model
(described in Section 3.3) and the data were tested for interactions between the four variables
(STATE, DAMAGE, TREAT, and LIGHT). The model includes all the main effects, all the two-
variable interactions, and all the three-variable interactions. It does not include the four-variable
interaction.
Table 3-9 shows the results of the CATMOD Procedure. The analysis-of-variance table shows
that the likelihood-ratio test for the four-variable interaction is non-significant but the three-
variable interaction DAMAGE*LIGHT*TREAT is significant. In other words, the tape
significantly reduces side and rear impacts in dark conditions, but by about the same amount in the
two states. Since the four-variable interaction term is non-significant, it is unnecessary to include
the STATE variable in the analysis. The data can simply be pooled and the pooled effectiveness
estimate for dark conditions shown in Table 3-8 is a good estimate.
34
TABLE 3-9
Florida and Pennsylvania Data
Log-linear Model Without Four-variable Interaction
Dependent Variables: STATE, DAMAGE, LIGHT, and TREAT
Model: DAMAGE, LIGHT, TREAT, STATE, DAMAGE*LIGHT, DAMAGE*TREAT, LIGHT*TREAT,

STATE*DAMAGE, STATE*LIGHT, STATE*TREAT, STATE*DAMAGE*LIGHT,
STATE*DAMAGE*TREAT, STATE*LIGHT*TREAT, and DAMAGE*LIGHT*TREAT

Source DF Chi-Square Prob
DAMAGE 1 193.71 0.0000
LIGHT 1 673.05 0.0000
TREAT 1 1213.90 0.0000
LIGHT* TREAT 1 7.47 0.0063
STATE 1 74.46 0.0000
STATE*DAMAGE 1 33.67 0.0000
STATE*LIGHT 1 77.61 0.0000
STATE*TREAT 1 138.41 0.0000
STATE*DAMAGE*LIGHT 1 4.96 0.0260
STATE*DAMAGE*TREAT 1 1.41 0.2358
STATE*LIGHT*TREAT 1 7.92 0.0049
DAMAGE*LIGHT*TREAT 1 12.68 0.0004
We also conclude from the table that the model is a good predictor of the data. But a model
excluding the non-significant term STATE*DAMAGE*TREAT would be an even better
predictor of the data. The other non-significant terms (DAMAGE*LIGHT and
DAMAGE*TREAT) cannot be excluded because these terms are contained in significant three-
variable interactions. Table 3-10 shows the results of the CATMOD procedure using this better
model. Again, likelihood-ratio test is non-significant, so effectiveness does not vary significantly
from state to state. The three-variable interaction DAMAGE*LIGHT*TREAT is significant, so
the overall effectiveness estimate is significant. This model fits the data better than the model in
Table 3-9, because it has fewer interaction terms yet still produces a likelihood ratio with higher
probability (0.1598 in Table 3-10 vs. 0.1325 in Table 3-19). Thus, this model “best” predicts the
effectiveness in dark conditions.
35
TABLE 3-10
Florida and Pennsylvania Data
Best Log-linear Model
Dependent Variables: STATE, DAMAGE, LIGHT, and TREAT
Model: DAMAGE, LIGHT, TREAT, STATE, DAMAGE*LIGHT, DAMAGE*TREAT, LIGHT*TREAT,

STATE*DAMAGE, STATE*LIGHT, STATE*TREAT, STATE*DAMAGE*LIGHT,
STATE*LIGHT*TREAT, and DAMAGE*LIGHT*TREAT

Source DF Chi-Square Prob.
DAMAGE 1 204.03 0.0000
LIGHT 1 671.67 0.0000
TREAT 1 1227.75 0.0000
LIGHT* TREAT 1 7.46 0.0063
STATE 1 75.61 0.0000
STATE*DAMAGE 1 45.55 0.0000
STATE*LIGHT 1 78.37 0.0000
STATE*TREAT 1 145.80 0.0000
STATE*DAMAGE*LIGHT 1 4.99 0.0254
STATE*LIGHT*TREAT 1 8.08 0.0045
DAMAGE*LIGHT*TREAT 1 13.66 0.0002
Table 3-11 shows the predicted values for Florida and Pennsylvania from our “best” model in
dark conditions. The tape is 28.4 percent effective when the predicted values in Florida or the
predicted values in Pennsylvania are used to calculated effectiveness. This is our “best” estimate
of tape effectiveness in dark conditions. The three-variable interaction
DAMAGE*LIGHT*TREAT is significant, so the effectiveness estimate is significantly greater
than zero.
The “best” models for dark-not-lighted and for dark-lighted, dawn, and dusk conditions can be
found by repeating the three steps outlined for dark conditions: fitting the full model, excluding
non-significant interactions to find the “best” model, and calculating the effectiveness using the
predicted values from the “best” model.
The best model for dark-not-lighted conditions is all main effects, all two-variable interactions,
and only the DAMAGE*LIGHT*TREAT three-variable interaction. The other three-variable
interactions and the four-variable interaction were non-significant. Our “best” effectiveness
estimate for dark-not-lighted conditions is 40.1 percent based on the predicted values. This is
significant. (The ÷2 for DAMAGE*LIGHT*TREAT is 26.42.)
36
TABLE 3-11
Predicted Values from the “Best” Model
Dark 512.921726 645.078273 525.078273 963.921725
Daylight 1023.612350 1231.387650 873.387649 1,270.612350
Untreated
Dark 298.078274 300.921726 200.921735 296.078275
Daylight 648.387651 874.612349 281.612349 459.387650
Effectiveness 28.4% 28.4%
The best model for dark-lighted, dawn, and dusk is all main effects, all two-variable interactions,
and the STATE*LIGHT*TREAT and the DAMAGE*LIGHT*TREAT three-variable
interactions. Our “best” effectiveness estimate is -0.9 percent. The DAMAGE*LIGHT*TREAT
interaction was kept in the model although this interaction was not significant. Therefore, we
conclude that the tape does not significantly reduce the side and rear crashes of combination
trucks under these conditions.
Table 3-12 shows that the estimates obtained by simply pooling the data are nearly the same as by
the CATMOD procedure. Essentially, these techniques are interchangeable on the current data
set. Since the confidence bounds and statistical significance are much easier to calculate on the
pooled data and the techniques are interchangeable, the pooled effectiveness estimate will be
presented throughout the remainder of the report.
TABLE 3-12
Tape Effectiveness by Two Alternative Procedures
Florida Plus Pennsylvania
Simple Pooled CATMOD Procedure
Dark 29% 28.4%
Dark-Not-Lighted 41% 40.1%
Dark-Lighted, Dawn, & Dusk -3% -0.9%
The confidence bounds for dark conditions is 29 ± (1.645)*0.064 = 29 ± 10 percent using the
pooled data and the standard deviation equation in Section 3.3. The lower bound is 19 percent
and the upper bound is 39 percent. For dark-not-lighted conditions, the standard deviation is
0.060 and the confidence bounds are 41 ± 10 percent using the pooled data. So the lower and
upper bounds are 31 and 51 percent, respectively.
37
38
CHAPTER 4
TAPE EFFECTIVENESS IN SPECIFIC SITUATIONS
The objective of this chapter is to see if the tape is more effective in some types of crashes than
others. For example, does it work better when the weather is clear or rainy? Most of the analyses
in this chapter are performed much the same way as in Chapter 3. Tractor/trailer combination
involvements in crashes are tabulated by trailer treatment (treated; untreated) and damage area
(side/rear; SV/frontal) and light conditions (dark; light). But here, additionally, they are classified
by another variable. In our example, they are tabulated by the weather conditions, either clear or
rainy. Then, separate effectiveness analyses are performed for each condition.
The most noteworthy finding is that the tape is less effective when the driver of the “other”
vehicle is over 50 years old than when the driver is 15-50 years old (Section 4.4).
Twelve comparative analyses of effectiveness were performed:
Effectiveness by collision types

• Rear vs. side impacts
• Various side impact collision types
Effectiveness by environmental/roadway conditions

• Clear weather conditions vs. bad weather conditions
• High speed limit roads vs. low speed limit roads
• Urban vs. rural
Effectiveness by tractor/trailer characteristics

• Van vs. flatbed
• Clean vs. dirty tape
Effectiveness by characteristics of the other driver/vehicle

• Younger vs. older driver
• Male vs. female driver
• Passenger car vs. pickup truck, van, or sport utility vehicle (SUV)
Effectiveness by crash-level injury severity

• Fatal vs. nonfatal-injury vs. property-damage-only crashes
• Fatal or nonfatal-injury vs. property-damage-only crashes
Most of the additional variables used to classify crashes by specific situations were on the
conspicuity data base created especially for this evaluation (see Section 2.2). The variables used
to classify the various side impact collision types in Pennsylvania and the urban/rural variable were
39
only available on the state data files. These analyses used the merged conspicuity and state data
file.
The conspicuity cases and the state cases were merged using the Crash Report Number assigned
to each crash on the PAR in Florida. The Florida merged data set had 4,592 cases (4,592/5,535 =
83 percent of the conspicuity cases) after excluding the cases with unknown trailer treatment,
light condition, or damage area.
The merge was not as easy or as successful for the Pennsylvania data. The Incident Number
assigned to each crash on the paper-copy PAR in Pennsylvania is available on the conspicuity data
base, but it is not contained on the State data base. The State assigned a different number to each
case when they created the State data base. So, in Pennsylvania the following variables were used
to match cases: accident date, county, number of vehicles, vehicle number, model year, and driver
age. Here, the merged data set had 3,837 cases (3,837/4,871 = 79 percent of the conspicuity
cases) after excluding the cases with unknown trailer treatment, light condition, or damage area.
The merged data set is slightly smaller than the conspicuity data set because some of the cases on
the conspicuity data set did not match to a corresponding case on the state data set. For this
reason, the merged data set was only used when the variables needed were not available on the
conspicuity data set. Most of these unmatched cases are minor crashes with little damage to the
vehicles and no reported driver injuries. These cases probably do not meet the minimum criteria
for state reported crashes and are therefore not contained on the state data set. Errors in the
variable(s) used to match cases can also account for some of the unmatched cases. This probably
happens more in Pennsylvania than Florida because an error in any one of the six variables used to
match cases in Pennsylvania will result in an unmatched case.
4.1 EFFECTIVENESS BY COLLISION TYPES
Is the tape more effective on the rear of the trailer than on the side of the trailer? NHTSA’s
regulatory evaluation projected that it would be (see Section 1.1). To answer this question, two
separate analyses were performed: one to estimate the effectiveness of the tape in rear impacts
and one to estimate the effectiveness of the tape in side impacts. The basic contingency table here
tabulated crashes by trailer treatment, damage area, and light conditions, similar to the basic
contingency table in Chapter 3. But to estimate the tape effectiveness in rear impacts, only rear
impacts were compared to SV/frontal impacts and to estimate the tape effectiveness in side
impacts, only side impacts were compared to SV/frontal impacts.
Table 4-1 shows the effectiveness of the tape in rear and side impacts. The results for both states
combined show a higher effect in rear impacts, however the state results are not entirely
consistent. The Florida estimates suggest that the tape is more effective in rear impacts than side
impacts, but the Pennsylvania estimates suggest there is little difference between side and rear
40
impacts. The tape may be more effective in preventing rear impacts than side impacts, but with
the different results from only two states we cannot be sure.
TABLE 4-1
Tape Effectiveness by Impact Point
FL & PA Florida Pennsylvania
Rear Impacts
Dark 43% S 47% S 34% S
Side Impacts
Dark 17% S* -1% NS 41% S
Dark-Not-Lighted 32% S 25% S* 43% S
As was discussed in Chapter 1, the tape is expected to be more effective in side impact collisions
where a vehicle strikes a combination truck at an angle than in collisions where a vehicle
sideswipes a combination truck. To see if this is true, we classified certain two-vehicle crashes
into subgroups based on the “first harmful event” as defined on the state files. The limited
information on the collision configuration and the different classification methods of the two
states precluded a more detailed subgrouping. The “first harmful events” we considered are:
• Angle collision
• Sideswipe
• Turning (Florida only) - i.e. at least one vehicle was making a turn prior to the crash
Other “first harmful events” (head-on, rear-end) are excluded because they are generally not
front-to-side or side-to-side impacts. The analysis comprised all the vehicles involved in these
crashes, including those with frontal damage and those with side damage.
Table 4-2 shows the effectiveness of the tape by “first harmful event.” The sample sizes in the
various cells are too small, and the results too inconsistent between Florida and Pennsylvania to
draw strong inferences, although effectiveness does appear to be consistently higher in angle
collisions than in sideswipes.
41
TABLE 4-2
Tape Effectiveness by First Harmful Event (Side Impact Collisions)
Angle
Dark 46% S 30% NS 83% S
Dark-Not-Lighted 58% S 40% NS 94% S
Sideswipe
Dark 8% NS -21% NS 42% NS
Dark-Not-Lighted 28% NS 33% NS 24% NS
Turning
Dark -55% NS -55% NS
Dark-Not-Lighted 8% NS 8% NS
4.2 EFFECTIVENESS BY ENVIRONMENTAL/ROADWAY CONDITIONS
Three separate effectiveness analyses were performed to see if the tape is effective in the dark
under three different weather conditions: clear; rainy or foggy; snowy or sleeting. Table 4-3
shows the results.
The combined data show about equal effectiveness in clear and in rainy/foggy conditions. Under
clear conditions, the combined point estimates and three of the four state estimates show a
substantial reduction in crashes. In rain or fog, the point estimates are positive and similar to
those for clear weather, but there is only enough data for significance in the combined data set.
Florida did not have any crashes occurring in snow or sleet and even Pennsylvania did not have
enough for statistically meaningful findings.
TABLE 4-3
Tape Effectiveness by Weather Conditions
Clear
Dark 28% S 12% NS 43% S
Rain and Fog
Dark 31% S* 34% NS 31% NS
Dark-Not-Lighted 39% S* 44% S* 39% NS
Snow and Sleet
Dark -12% NS -12 NS
Dark-Not-Lighted -17% NS -17 NS
42
Table 4-4 compares the effectiveness of the tape on roads with high and low speed limits. The
combined data show about the same effectiveness on both types of roads. The tape is clearly
effective on roads with high speed limits. On roads with speed limits of 50 mph or above, the
tape is effective for the combined data and individual state data. On low speed roads the results
are less conclusive. The combined estimates show the tape is effective on low speed roads, but
the state results are inconsistent. The tape is highly effective in Pennsylvania, but it may not be in
Florida.
TABLE 4-4
Tape Effectiveness by Speed Limit
Speed Limit is Less than 50 MPH
Dark 26% S* -4% NS 54% S
Speed Limit is 50 MPH and Above
Dark 29% S 25% S 30% S
Table 4-5 shows the effectiveness of the tape is likely to be about the same in urban and rural
areas. The tape is clearly effective in rural areas. The combined estimates are significant as are
the individual state estimates under both lighting conditions. The tape is probably effective in
urban areas. The combined point estimates are positive and significant. The state point estimates
suggest a substantial reduction in crashes, more than 30 percent, but only one is significantly
greater than zero. Statistical significance is hard to attain here, because very few cases are in
urban areas (18 percent in Florida and 27 percent in Pennsylvania).
TABLE 4-5
Tape Effectiveness by Urban vs. Rural Location
Urban
Dark 35% S* 30% NS 31% NS
Dark-Not-Lighted 57% S 78% S 41% NS
Rural
Dark 32% S 23% S* 40% S
When the data is divided into four groups (rural vs. urban, side vs. rear impacts), there are not
enough cases in any group for strong conclusions. However, the point estimates suggest that the
43
tape may be more effective in urban side impacts than in rural side impacts. If so, that would be
consistent with the idea that urban side impacts are more likely to be angle rather than sideswipe
collisions, where the tape might be more effective.
4.3 EFFECTIVENESS BY TRACTOR/TRAILER CHARACTERISTICS
Five different trailer types were identified on the conspicuity data base. The effectiveness of the
tape on the two most common types is shown in Table 4-6. The tape is probably effective on
vans in dark-not-lighted conditions. The combined estimate is significant. The state point
estimates are positive and close to the combined point estimate. In “dark” conditions, the results
for vans are mixed.
The tape is clearly effective on flatbeds. The estimates are fairly large and significant for the
combined data and the individual state data. It also appears the tape is more effective on flatbeds
than vans. All point estimates for flatbeds are higher than their counterparts for vans. The
CATMOD procedure to test the effectiveness estimates by trailer type showed that the
effectiveness estimates for flatbeds are significantly different than the effectiveness estimates from
vans. Flatbeds even when loaded are probably less visible than van trailers in the dark because of
their low profile. This makes the tape very effective on flatbeds.
TABLE 4-6
Tape Effectiveness by Trailer Type
Van
Dark 11% NS -9% NS 26% S*
Dark-Not-Lighted 25% S 21% NS 28% S*
Flatbed
Dark 55% S 41% S* 65% S
Dark-Not-Lighted 63% S 50% S* 70% S
The effectiveness of clean and dirty tape was compared to see if dirt and grime diminish the
effectiveness of the tape. The condition of the tape was indicated on both the side and the rear of
the trailer, so we considered the effects of dirt on the side and the rear of the trailer, separately.
The basic analysis tabulated tractor/trailer combination involvements by trailer treatment (treated;
untreated) and damage area (side/rear; SV/frontal) and light conditions (dark; light). But here,
additionally, they will be classified by tape condition: clean vs. dirty on the side and clean vs. dirty
on the rear.
First, let us consider the analysis for the tape condition on the side of the trailer. Here, instead of
comparing side/rear impacts to SV/frontal impacts, we compared only side impacts to SV/frontal
44
impacts. We also compared clean, treated vehicles to untreated vehicles for the effectiveness of
clean tape and dirty, treated vehicles to untreated vehicles for the effectiveness of dirty tape.
Untreated trailers were not classified as clean or dirty. Therefore, we could not compare clean,
treated vehicles to clean, untreated vehicles and dirty, treated vehicles to dirty, untreated vehicles.
“Some dirt” and “very dirty” were combined into one category and analyzed together because
there were too few “very dirty” trailers to analyze separately. The only difference in the analysis
of the tape condition on the rear of the trailer is that rear impacts were compared to SV/frontal
impacts.
Table 4-7S shows the effectiveness in side impacts for clean and dirty tape on the side of trailers.
There appears to be no difference in the effectiveness of clean and dirty tape on the trailer side.
The combined point estimates are all around 30 percent except for clean tape, dark conditions.
Although the individual state point estimates are not consistent, there is little difference between
the point estimates for clean and dirty tape within each state.
TABLE 4-7S
Tape Effectiveness in Side Impacts by Tape Condition on the Side of Trailers
Side Impact Analysis FL & PA Florida Pennsylvania
Trailer Side - Clean
Dark 13% NS 1% NS 37% S
Trailer Side - Dirty
Dark 27% S -2% NS 46% S
Dark-Not-Lighted 30% S 7% NS 37% S*
Table 4-7R shows the effectiveness in rear impacts for clean and dirty tape on the rear of trailers.
This table not only shows that clean tape is highly effective on the rear of trailers, but also that
dirty tape appears to be effective on the rear of trailers. The combined and state point estimates
are consistent and the combined point estimates have enough data for statistical significance.
Since the point estimates for clean tape are consistently larger than the point estimates for dirty
tape, the CATMOD procedure was used to test if the effectiveness varied by tape condition. The
CATMOD procedure showed that the effectiveness estimates for clean tape are significantly
different than the estimates for dirty tape. In other words, clean tape is more effective than dirty
tape on the rear of the trailers.
It is not obvious why clean tape should make more of a difference in rear impacts than in side
impacts. The ambiguous results for clean vs. dirty tape in side impacts are consistent with the
inconclusive overall results for side impacts. In Table 4-1, Pennsylvania showed a strong effect
for the tape in side impacts, but Florida showed no effect in dark conditions and a weak effect in
dark-not-lighted conditions.
45
TABLE 4-7R
Tape Effectiveness in Rear Impacts by Tape Condition on the Rear of Trailers
Rear Impact Analysis FL & PA Florida Pennsylvania
Trailer Rear - Clean
Dark 53% S 55% S 55% S
Trailer Rear - Dirty
Dark 27% S 26% NS 19% NS
Dark-Not-Lighted 33% S 30% NS 26% NS
4.4 EFFECTIVENESS BY OTHER DRIVER/VEHICLE CHARACTERISTICS
The next three analyses focus on the other vehicle involved in a two-vehicle crash with a
combination truck. The “other” vehicle is not necessarily the striking or impacting vehicle,
because in some crashes it is the truck that strikes the other vehicle. The “other driver” is the
driver of the other vehicle. Two-vehicle crashes where both vehicles are combination trucks were
excluded from the analysis. Single-vehicle crashes by definition were also excluded, thus side/rear
impacts were only compared to frontal impacts. In Florida, 63 percent of the other vehicles are
passenger cars and 23 percent are pickup trucks, vans, and sport utility vehicles (SUVs) (and 14
percent are heavy trucks, other or unknown vehicle types). In Pennsylvania, 58 and 21 percent of
the other vehicles are passenger cars and light trucks, respectively (and 21 percent are heavy
trucks, other or unknown vehicle types).
Table 4-8 shows the effectiveness of the tape by the other driver’s age. Younger drivers of other
vehicles benefit more from the tape than older drivers. The effectiveness estimates for 15-50 year
old drivers are quite large and significant. The combined point estimates for drivers older than 50
are small and not significant. Although the state point estimates for older drivers are not
consistent, most of them are not significant just like the combined results. The point estimates for
younger drivers are almost always much larger than the corresponding point estimates for older
drivers. The CATMOD procedure comparing the younger and older driver point estimates
showed that there is no significant difference under dark conditions, but that there is a significant
difference under dark-not-lighted conditions. This suggests that at least in “dark-not-lighted”
conditions younger drivers of other vehicles benefit more from the tape than older drivers. It is
possible that older drivers’ eyesight is so poor at night (and/or reaction time so slow) that the tape
might not become visible soon enough to prevent collisions with combination trucks. If this is so,
it might be appropriate to explore more intense conspicuity treatments that would give the older
drivers benefits comparable to those of young drivers. However, an alternative, less troubling
interpretation of these results is that older drivers keep a greater distance from combination trucks
and have less need for the tape. Another caveat is that the contrast in the results is seen mainly in
Florida, whereas the tape is still quite effective for older drivers of the other vehicles in
Pennsylvania.
46
TABLE 4-8
Tape Effectiveness by Other Driver’s Age
15-50 years old
Dark 44% S 31% S 54% S
More than 50 years old
Dark 20% NS -8% NS 53% S*
Dark-Not-Lighted 10% NS -7% NS 34% NS
Most of the other drivers are male (66 and 69 percent in Florida and Pennsylvania, respectively).
Table 4-9 shows the effectiveness of the tape is about the same when the other driver is male and
when the other driver is female. The result is consistent with intuition: there is no obvious reason
why the tape should be more effective for drivers of one gender, except perhaps that some male
drivers are more aggressive.
TABLE 4-9
Tape Effectiveness by Other Driver’s Gender
Male
Dark 41% S 30% S* 49% S
Female
Dark 36% S 11% NS 67% S
Table 4-10 shows similar effectiveness of the tape regardless of whether the other vehicle is a
passenger car or a light truck. This, too, is intuitively reasonable: it is unlikely that the higher
seating height in light trucks make the tape significantly more or less visible than to the car driver.
47
TABLE 4-10
Tape Effectiveness by Other Vehicle Type
Passenger Cars
Dark 35% S 11% NS 60% S
Light Truck, Vans, and SUVs
Dark 49% S 40% NS 53% S*
Dark-Not-Lighted 44% S 30% NS 51% NS
4.5 EFFECTIVENESS BY CRASH-LEVEL INJURY SEVERITY
Driver injury severity is coded on the conspicuity data base as follows:
K Killed
A A injury - incapacitating injury
B B injury - non-incapacitating injury
C C injury - possible injury
O Not injured
A new variable, crash-level injury severity, was created to classify the most severe driver injury in
each crash. Separate effectiveness analyses were performed on four [overlapping] crash-level
injury severity groups: not injured; A, B and C injuries; fatal; and the combined injury group of
fatal, A, B, and C injuries.
Table 4-11 shows the effectiveness estimates by crash-level injury severity. The tape may not
reduce very minor crashes. When none of the drivers in the crash are injured, the tape
effectiveness is not significantly different from zero. The tape clearly reduces injury crashes.
The tape significantly reduces crashes where at least one of the drivers was killed. The pooled
data effectiveness estimates are large and achieve statistical significance. But while these data
support a conclusion that the tape saves lives as well as reducing nonfatal crashes, the point
estimates themselves should not be relied upon, since they are based on a small sample of fatal
injury crashes. There are about 300 fatal injury cases in the pooled data for dark conditions
compared to about 4,900 fatal, A, B, and C injury cases. There are even fewer cases for the dark-
not-lighted conditions. Therefore in Chapter 5, a range of lives saves will be estimated using the
29 percent (all crashes) and 44 percent (fatal, A, B, and C injury crashes) reduction.
48
TABLE 4-11
Tape Effectiveness by Crash-Level Injury Severity
Not Injured
Dark 6% NS -38% S* 36% S
Dark-Not-Lighted 14% NS -33% NS 41% S
A, B, & C Injured
Dark Conditions 41% S 41% S 37% S
Fatal
Dark 68% S 66% S* 74% NS
Fatal, A, B, & C Injured
Dark 44% S 45% S 40% S
The CATMOD procedure comparing injured (K, A, B, or C) and not injured (O) shows that the
effectiveness estimates for injured are significantly different than the effectiveness estimates for
not injured. Thus, the tape is more effective at reducing injury crashes than non-injury crashes.
4.6 TAPE EFFECTIVENESS DURING THE DAYLIGHT
The fleet study11 discussed in Section 1.1 concluded that the tape is effective not only in dark
conditions but also during daylight. Other research suggests that the tape might be more effective
on the side than the rear of the trailer during daylight. To the limited extent that our data can be
used to address this issue, they do not support either premise.
Only a two-variable analysis method was available to estimate the effectiveness of the tape during
daylight. The analysis tabulates tractor/trailer vehicle involvements in crashes by trailer treatment
(treated; untreated) and damage area (side and rear impacts; single-vehicle and frontal impacts).
The analysis compares the involvement rates of treated and untreated tractor/trailer combination
trucks at only daylight conditions. We did not have exposure data or a control group for
additional adjustments of these rates. Table 4-12 shows the two-variable contingency table and
the effectiveness estimates by state. Neither estimate is significant; therefore there is no evidence
here of a reduction of crashes involving treated trailers during daylight.
11
Burger, W. J., et al, Improved Commercial Vehicle Conspicuity and Signalling Systems,
Task III – Field Test Evaluation of Vehicle Reflectorization Effectiveness, NHTSA Report No.
DOT HS 806 923, Washington, DC, 1985.
49
Table 4-12
Number of Crash Involvements and Effectiveness During the Daylight
Side & Rear SV & Frontal Side & Rear SV & Frontal
Treated 1,024 1,231 873 1,271
Untreated 648 875 282 459
“Effectiveness” -12% -12%
When you only compare side to single-vehicle and frontal impacts, the tape is -7 and -27 percent
effective during daylight in Florida and Pennsylvania, respectively. For rear impacts, the tape is -
27 and 1 percent effective in Florida and Pennsylvania, respectively. The tape does not appear to
reduce either side or rear impacts during daylight.
We believe that the two-variable method used here is intrinsically flawed and these results may be
inaccurate. But without a good control for daylight crashes such as mileage or other exposure
data, this is the only method available to estimate the effectiveness of the tape during daylight.
50
CHAPTER 5
BENEFITS OF CONSPICUITY TAPE
The addition of conspicuity tape on the side and rear of heavy trailers has reduced crashes
involving combination trucks and injuries sustained in these crashes. In the pooled data described
in Chapter 3, we found a 29 percent reduction of side and rear impacts into combination trucks, in
the dark. In Chapter 4, we found a 44 percent reduction in all injury levels sustained in these
crashes. When all heavy trailers have conspicuity tape, the tape will prevent approximately 7,800
crashes per year, prevent approximately 3,100 to 5,000 injuries per year, and save an estimated
191 to 350 lives per year.
5.1 CRASHES AVOIDED PER YEAR
During the baseline year 1999, NHTSA’s General Estimate System (GES) reported that there
were 20,883 actual side and rear impacts into combination trucks in dark conditions (referred to
below as the N0 group). Some of these tractor-trailer combinations had trailers treated with
conspicuity tape and others had trailers not treated with tape. We are going to estimate the
number of crash-involved combination trucks during the dark in side and rear impacts that would
have occurred if:
• All the trailers had not been treated with tape (N1). N1 > N0.
• All the trailers had been treated with conspicuity tape (N2). N1 > N0 > N2.
• The crashes avoided by adding tape to trailers: N1 - N2.
To calculate crashes avoided, we must first calculate the percentage of trailers that have tape and
the percentage of trailers that do not have tape in the baseline year. In 1999, 71 percent of the
crash-involved trailers in the dark with side or rear damage had tape in Florida and 72 percent of
the crash-involved trailers in the dark with side or rear damage had tape in Pennsylvania. We
believe the Pennsylvania percentage of trailers with tape is more representative of the full year
than the Florida percentage because the Pennsylvania data is from the first 11 months of the year
and the Florida data is only from the first five months of the year.
Since there are 20,883 crash involved trailers damaged in the side and rear in the dark in 1999,
then 72 % * 20,883 = 15,036 of these trailers had tape and 5,847 of these trailers did not have
tape. Given that conspicuity tape reduces involvements by 29 percent (Table 3-12), then the
number of involvements that would have occurred if the treated trailers had not been treated with
tape is 15,036 / (1-.29) = 21,177. Thus, there would have been N1 = 21,177 + 5,847 = 27,024
involvements if all trailers had not been treated with tape.
51
If all trailers had been treated with tape, then there would have been N2 = 27,024 * (1-.29) =
19,187 crash involved combination trucks with side or rear damage in the dark. That is a
reduction of N1 - N2 = 27,024 - 19,187 = 7,837, which is rounded to 7,800 crashes per year.
5.2 NONFATAL INJURIES AVOIDED PER YEAR
To calculate the injuries prevented by adding conspicuity tape to trailers, we used the same
method as for crashes. As above, 72 percent of the trailers actually had tape in 1999 and 28
percent did not have tape. Table 4-11 showed that the tape is 41 percent effective in preventing
nonfatal injuries (pooled data - dark conditions). So injuries avoided per year will be calculated
using that estimate and the more conservative estimate (29 percent effectiveness in all crashes) to
give a range.
The GES reported that there were 8,159 A, B, and C injuries in combination truck crashes during
1999 where the combination truck was damaged in the side or rear in the dark. There must have
been approximately 8,159 * 72% = 5,874 injuries occurring in crashes where the trailer involved
had tape and 2,285 injuries occurring in crashes where the trailer involved did not have tape.
Given that nonfatal injuries are reduced by 41 percent, if the trailers that were treated with tape
did not have tape, there would have been 5,874/(1-.41) = 9,956 injuries in these crashes. Thus,
there would have been N1 = 9,956 + 2,285 = 12,241 injuries occurring in crashes involving a
combination truck if the trailer did not have tape. If all the trailers in all of these crashes had tape,
then there would only be N2 = 12,241 * (1-.41) = 7,222 injuries. That is a reduction of N1 - N2
= 12,241 - 7,222 = 5,019 injuries, which is rounded to 5,000 injuries.
Based on the more conservative 29 percent reduction, if the trailers that were treated with tape
did not have tape, there would have been 5,874/(1-.29) = 8,273 injuries in these crashes. Thus,
there would have been N1 = 8,273 + 2,285 = 10,558 injuries occurring in crashes involving a
combination truck if the trailer did not have tape. If all trailers had tape, then there would be N2
= 10,558 * (1-.29) = 7,496 injuries. This is a reduction of N1 - N2 = 10,558 - 7,496 = 3,062
injuries, which is rounded to 3,100 injuries. Therefore, when all trailers have tape, the tape will
reduce 3,100 to 5,000 injuries per year relative to a trailer fleet without tape.
5.3 LIVES SAVED PER YEAR
The same method applies to lives saved. Two estimates of fatality reduction will be used to
calculate lives saved: 29 percent - the effectiveness of tape found in all crashes; 44 percent - the
effectiveness of tape found in fatal, A, B, and C injury crashes.
In the 1999 FARS data, there were 508 fatalities occurring in crashes where a combination truck
had side or rear damage, and the crash took place in the dark. Seventy-two percent or 508 * 72%
52
= 366 fatalities occurred in crashes where the trailer involved had tape and 142 fatalities occurred
in crashes where the trailer involved did not have tape.
Based on the more conservative 29 percent reduction, if the trailers that were treated with tape
did not have it, then there would have been 366 /(1-.29) = 515 fatalities in these crashes. There
would have been N1 = 515 + 142 = 657 fatalities if none of the trailers had tape. The number of
fatalities that would have occurred if all of the trailers had tape is N2 = 657 * (1-.29) = 466.
Here, the difference is N1 - N2 = 657 - 466 = 191 lives saved annually by adding conspicuity tape
to trailers.
If the tape is 44 percent effective in preventing fatalities and if the trailers that were treated with
tape had not had it, then there would have been 366 /(1-.44) = 654 fatalities in these crashes.
There would have been N1 = 654 + 142 = 796 fatalities if none of the trailers had tape. The
number of fatalities that would have occurred if all of the trailers had tape is N2 = 796 * (1-.44) =
446. Therefore, the difference is N1 - N2 = 796 - 446 = 350 lives saved annually by adding
conspicuity tape to trailers. Therefore, when all trailers have tape, the tape will reduce 191 to 350
lives per year relative to a trailer fleet without tape.
53
54
REFERENCES
Burger, W. J., et al., Improved Commercial Vehicle Conspicuity and Signalling Systems, Task I –
Accident Analysis and Functional Requirements, NHTSA Report No. DOT HS 806 100,
Washington, DC, 1981.
Burger, W. J., et al., Improved Commercial Vehicle Conspicuity and Signalling Systems, Task III
– Field Test Evaluation of Vehicle Reflectorization Effectiveness, NHTSA Report No. DOT HS
806 923, Washington, DC, 1985.
Green, P., et al., Accidents and the Nighttime Conspicuity of Trucks, University of Michigan
Highway Safety Research Institute, Report No. UM-HSRI-79-92, Ann Arbor, MI 1979.
Hansen, Morris H., Hurwitz, William N., and Madow, William G., Sample Survey Methods and
Theory, Volume I, John Wiley & Sons, New York, 1953, pp. 512-514.
Minahan, D. J. and O’Day, J., Car-Truck Fatal Accidents in Michigan and Texas, University of
Michigan Highway Safety Research Institute, Report No. UM-HSRI-77-49, Ann Arbor, MI,
1977.
Olsen, P. L., et al., Performance Requirements for Large Truck Conspicuity Enhancements,
NHTSA Report No. DOT HS 807 815, Washington, DC, 1992
Preliminary Regulatory Evaluation - Proposed Amendment to FMVSS No. 108 to Require

Retroreflective Material on the Side and Rears of Heavy Trailers, NHTSA, Washington, DC,
1991.
SAS/STAT Users’s Guide: Volume1, ACECLUS-FREQ, Version 6, Fourth Edition, SAS Institute
Inc., Cary, NC, 1990.
Ziedman, K., et al., Improved Commercial Vehicle Conspicuity and Signalling Systems, Task II –
Analyses, Experiments and Design Recommendations, NHTSA Report No. DOT HS 806 098,
Washington, DC, 1981.
55

The Effectiveness of Retroreflective Tape On Heavy Trailers: DOT HS 809 222 March 2001 NHTSA Technical Report

Uploaded by

Copyright:

Available Formats

The Effectiveness of Retroreflective Tape On Heavy Trailers: DOT HS 809 222 March 2001 NHTSA Technical Report

Uploaded by

Document Information

Original Description:

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

The Effectiveness of Retroreflective Tape On Heavy Trailers: DOT HS 809 222 March 2001 NHTSA Technical Report

Uploaded by

Copyright:

Available Formats

U.S.

DOT HS 809 222 March 2001

The Effectiveness of Retroreflective Tape

DOT HS 809 222

The Effectiveness of Retroreflective Tape on Heavy Trailers March 2001

7. Author(s) 8. Performing Organization Report No.

17. Key Words 18. Distribution Statement

TABLE OF CONTENTS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . iii

INTRODUCTION AND BACKGROUND . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1

TAPE EFFECTIVENESS IN SPECIFIC SITUATIONS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 39

BENEFITS OF CONSPICUITY TAPE . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 51

ANNUAL BENEFITS OF CONSPICUITY TAPE

CRASH REDUCTIONS BY LIGHTING CONDITIONS

• In dark conditions (combining the subsets of “dark-not-lighted,” “dark-lighted,” “dawn,”

CRASH REDUCTIONS FOR SPECIFIC SUBGROUPS IN DARK CONDITIONS

STATUS OF TAPE IN THE 1997-1999 CRASH DATA

INTRODUCTION AND BACKGROUND

1.1 HISTORY AND RESULTS OF EARLIER EFFECTIVENESS STUDIES

1.2 FMCSA RETROFIT STANDARD

1.3 CURRENT STATUS OF TAPE

2.1 SUPPLEMENTARY FORM

No Damage, No Impact, Non Collision or Not Applicable

Florida veh_type = 4, 5, 6, 77 and trlno = 1, 2

Category State Definition

Single-Vehicle FL the number of vehicles involved by counting is 1

Front FL impact = 1,2,14 or (impact = 22, 99 and impactpt = 1)

Rear FL impact = 7,8,9 or (impact = 22, 99 and impactpt = 3)

3.2 THE BASIC ANALYSIS

The basic analysis tabulated tractor-trailer combination involvements in crashes by trailer

In Florida, the numbers in the basic contingency table are:

Number of Crash Involvements in Florida

Single-Vehicle (SV) &

N121 * (N112 / N122) * [(N211 / N221)/(N212 / N222)].

In Pennsylvania, the basic contingency table is:

Number of Crash Involvements in Pennsylvania

3.3 STATISTICAL SIGNIFICANCE AND CONFIDENCE BOUNDS

Dependent Variables: DAMAGE, LIGHT, and TREAT

Model: DAMAGE, LIGHT, TREAT, DAMAGE*LIGHT, DAMAGE*TREAT, and LIGHT*TREAT

Weighting Factor: N (Number of Combination Trucks)

Likelihood Ratio 1 2.81 0.0937

Var (Nijk) / N2ijk = Nijk / N2ijk = 1 / Nijk.

3.5 COMBINED EFFECTIVENESS AND “BEST” ESTIMATES

Dependent Variables: STATE, DAMAGE, LIGHT, and TREAT

Model: DAMAGE, LIGHT, TREAT, STATE, DAMAGE*LIGHT, DAMAGE*TREAT, LIGHT*TREAT,

Weighting Factor: N (Number of Combination Trucks)

Likelihood Ratio 1 2.26 0.1325

Dependent Variables: STATE, DAMAGE, LIGHT, and TREAT

Model: DAMAGE, LIGHT, TREAT, STATE, DAMAGE*LIGHT, DAMAGE*TREAT, LIGHT*TREAT,

Weighting Factor: N (Number of Combination Trucks)

Likelihood Ratio 2 3.67 0.1598

TAPE EFFECTIVENESS IN SPECIFIC SITUATIONS

Twelve comparative analyses of effectiveness were performed:

Effectiveness by collision types

Effectiveness by environmental/roadway conditions

Effectiveness by tractor/trailer characteristics

Effectiveness by characteristics of the other driver/vehicle

Effectiveness by crash-level injury severity

4.1 EFFECTIVENESS BY COLLISION TYPES

Model: DAMAGE, LIGHT, TREAT, DAMAGELIGHT, DAMAGETREAT, and LIGHT*TREAT

Model: DAMAGE, LIGHT, TREAT, STATE, DAMAGELIGHT, DAMAGETREAT, LIGHT*TREAT,

Model: DAMAGE, LIGHT, TREAT, STATE, DAMAGELIGHT, DAMAGETREAT, LIGHT*TREAT,