| Final
Report
Purchase
Order No. DTMC-01-P-00071
Prepared
For:
FEDERAL
MOTOR CARRIER SAFETY ADMINISTRATION
November
2003
Prepared
By:
CENTER
FOR APPLIED RESEARCH, INC.
Richard
L. Knoblauch
Raymond Cotton
Marsha Nitzburg
Rita Furst Seifert
WESTAT
Gary
Shapiro
Pam Broene
TABLE
OF CONTENTS
Page
Executive
Summary ------------------------------------ ii
A. BACKGROUND
1
B. PURPOSE
AND SCOPE OF THE PROJECT 1
C. DATA
COLLECTION PROCEDURES 1
Imaging Techniques 1
Pilot Testing of Observational Procedures 2
D. SAMPLING
METHODOLOGY 3
Sample of States 3
Sample of County Groups 3
Selection of Observation Sites 6
E. OBSERVATION
PROCEDURES 7
F. RESULTS
8
Usage Rates - Truck Class 9
Usage Rates - Fleet Type/Truck Type 10
Usage Rates - Location/Day of Week/State 12
APPENDIX
A - MEMO FROM WESTAT DESCRIBING SELECTION
OF STATES AND COUNTY SAMPLES WITHIN STATES 15
APPENDIX
B - SITE LOCATIONS 21
LIST
OF FIGURES
Figure
1. Safety Belt Usage Rate of Drivers by Vehicle Type and Class
of Truck 9
Figure 2. CMV Safety Belt Usage Rate by Fleet Type 10
Figure 3. Safety Belt Usage Rate by Trailer Type and HAZMAT/Not
HAZMAT (Class 8 only) 11
Figure 4. Safety Belt Usage Rate by Type of Location 12
Figure 5. Safety Belt Usage Rate by Day of Week 13
Figure 6. Safety Belt Usage Rate by State for Passenger Vehicle
Drivers and Commercial Vehicle Drivers 14
Executive
Summary
Introduction
Approximately
5,000 people are killed annually in accidents involving large
trucks. Although less than 20% of the fatalities in such crashes
are occupants of the truck, the truck occupant is often killed
in situations that may be preventable. In single-vehicle crashes
where the driver runs off the road and rolls over or hits
a large stationary object, many drivers die because they fail
to wear a safety belt.
Prior
to this study, there were no reliable statistics regarding
the level of seat belt usage among commercial motor vehicle
(CMV) drivers. Although, the National Highway Traffic Safety
Administration (NHTSA) has conducted a number of seat belt
studies, they have been limited to only automobile occupants.
The scope
of this study was to design and implement a nationally representative
sample survey of seat belt usage among commercial motor vehicle
drivers and, based on the data collected, to produce estimates
of seat belt usage rates for this segment of the driving population.
The Federal
Motor Carrier Safety Administration (FMCSA) will use the data
to determine how best to allocate its resources to achieve
its goal of 1.65 fatalities per 100 million truck vehicle
miles traveled by 2008, and to increase seatbelt use among
commercial vehicle drivers.
Sampling
Methodology
A three-stage
sampling procedure was used in this study, conducted in 2002.
In the first stage, a sample of twelve states was selected,
using a "probability proportional to size" systematic
sampling procedure, with state truck vehicle miles traveled
(VMT) as the size measure.
In the
second sampling stage, counties in each selected state were
combined to form groupings that had a minimum of 300 limited-access
highway miles. Once these county groups were formed, one county
group was randomly selected from each selected state.
In the
third sampling stage, observers visited each sampled county
group, and selected locations, using county and city maps,
where the safety belt usage of commercial trucks could be
observed. When a potential observation site was identified,
its truck volume was measured. If the site had at least two
commercial vehicles passing through it within a five-minute
interval, it was used for data collection. There were a total
of 117 observation sites used in the study. Most of the sampled
states had at least 10 sites.
Shoulder
belt usage in all Class 7 (six to nine tires) and Class 8
(10 or more tires) trucks was observed at each selected site.
Observations were made over a two-day period in each of the
12 states. A minimum of 10 hours was spent collecting data
in each state. Of the vehicles observed in the study, roughly
10% were Class 7, and the remaining 90% were Class 8. Vehicles
in Classes 1 through 6, with gross-vehicle-weights of less
than 26,000 lbs., were not observed.
Results
A total
of 3,909 trucks were observed. The overall safety belt usage
rate for commercial vehicles observed in the study was 48%,
with a standard error of 1.4% (standard errors were generated
using the WesVar software program developed by Westat, Inc.).
Class 7 trucks, those with at least six tires, but less than
10, had a safety belt usage rate of 54%. Class 8 trucks, those
with more than 10 tires, had a usage rate of 47%. This compares
to a usage rate of 79% percent for all passenger vehicles
(U.S. DOT NOPUS Survey, June 2003).
The usage
rate for those units where the vehicle's tractor was identified
as a major regional or national fleet was estimated to be
55%. For trucks that were either independent or part of local
fleets, the usage rate was estimated to be 44%.
Seat belt
rates were also estimated by commercial vehicle type, and
by the presence of hazardous materials (HAZMAT). The vehicle
type with the highest usage rate was found to be the single
tanker (61%), and the vehicle type with the lowest usage rate
was found to be the single-trailer dump truck (26%). Drivers
of trucks pulling trailers placarded for HAZMAT were found
to have the highest safety belt usage rate of all of the various
categories observed (67%).
A.
BACKGROUND
In recent
years, approximately 5,000 people are killed annually in accidents
involving trucks. Less than 20% of these fatalities are occupants
of the truck. Due to the sheer mass of a commercial vehicle
(CMV), often the safest place in an accident is inside the
truck. Unfortunately, most accidents that are fatal to truck
drivers involve running off the road and rolling over, or
hitting a large stationary object (e.g., tree, bridge abutment,
culvert, etc.). Many of those killed in these types of crashes
died because they failed to wear their seat belts and were
thrown from the CMV.
At present
there are no reliable statistics regarding the level of seat
belt usage by CMV drivers. NHTSA has conducted a number of
seat belt studies but they have been limited to automobile
occupants.
B.
PURPOSE AND SCOPE OF THIS PROJECT
With the
FMCSA's goal to reduce the large truck fatality rate to 1.65
fatalities per 100 million truck vehicle miles traveled by
2008, knowledge of the current seat belt usage rate by CMV
drivers would help the agency determine how best to allocate
resources and evaluate outreach and education strategies to
increase seat belt usage.
The scope
of the study was to develop a nationally representative sample
survey of seat belt usage by CMV drivers throughout the United
States, and collect the data to determine the baseline usage
rate.
C.
DATA COLLECTION PROCEDURES
The development
of the data collection protocol involved two specific activities.
First, a wide variety of photo-imaging techniques were investigated.
Once it was determined that the development of sophisticated
automatic imaging techniques was beyond the scope for this
study, procedures to be followed for a manual observation
protocol were developed. These two processes are described
below.
Imaging
Techniques
A variety
of different photo imaging techniques were investigated. These
include videotape; digital cameras and high-speed digital
flash imaging. Both the videotape and digital camera provided
an inadequate image of the trucks observed. Visibility issues,
especially glare, made many of the images very difficult to
accurately code.
One of
the photo imaging techniques examined did provide fairly reliable
images of the truck drivers' restraint usage. The technique
was developed for a racial profiling study on the New Jersey
Turnpike. It uses still digital images, a flash is used to
minimize glare, and allow night data collection. Unfortunately,
the procedure requires that a van be set up along the shoulder
and a flash is activated as the truck passes, raising serious
issues regarding observer safety and inconspicuousness. The
technique would also require that permission be obtained from
the cognizant agency at each study site. For these reasons,
this technique was not used for this project.
Manual
Observation Protocol
Extensive
pilot testing of various manual observation procedures were
conducted in six states: Virginia, Illinois, Indiana, Ohio,
Pennsylvania, and Maryland. The pilot testing determined what
types of observational procedures could identify safety belt
usage by commercial motor vehicle operators. Since commercial
motor vehicles vary greatly in size, the approach must be
flexible enough to allow observations of many different vehicle
types at many different types of locations. Observing commercial
vehicle belt use is also complicated by the presence of small
windshields, small side windows, tinted windshields, tinted
and mirrored side windows, wide windshield center divider
strips, oversize rear view mirrors, objects hanging from rear
view mirrors, objects and wires hanging from cab roof, sun
visors, windshield wipers that park in the up position, and
large overhanging windshield sun shades.
The following
technical issues were considered when developing the observational
protocol:
- Safe
and Inconspicuous - The protocol must be safe for the observers
and inconspicuous to the trucks being observed.
- Replicable
and Accurate - The protocol must be one that can be accurately
done by different observers at different types of locations
across the country.
- Flexible
- The protocol must be one that will work equally well across
a wide variety of truck types at a wide variety of locations.
- Comparable
- The protocol should be one that allows comparisons to
be made with the National Occupant Protection Use Survey
(NOPUS).
- Logical/Defensible
- The protocol should avoid those situation/locations that
may affect belt usage (e.g., toll plazas and weigh stations
where police presence may change driver behavior).
It was concluded that standard visual observation provides
the best method for observing commercial truck drivers' seat
belt usage. This method allowed researchers to deal with most
of the driver and vehicle factors, a variety of weather conditions,
and site types. The method also allowed the observer the flexibility
needed to avoid appearing conspicuous.
Manual
observations were used at many different types of locations
including: rest areas, truck stops, and exit ramps on or near
interstates. In addition, it also worked well at signalized
intersections on major and minor arterial roadways. It was
possible to capture all truck types at each of these types
of sites.
The manual
observational procedure allowed the observer to move around
and gain the best observation angle. At many locations it
was possible to see vehicles as they arrive, wait, and pass
the observer. At locations where the driver negotiated a turn
the variety of angles available for observations was increased.
Observer positioning flexibility also made it possible to
work with windshield glare issues and use the sunlight to
the observer's advantage.
D.
SAMPLING METHODOLOGY
A three-stage
sample selection procedure was used. In the first stage, a
sample of twelve states was selected. At the second stage
a sample of one county group was selected from each of the
twelve states. At the third stage, observations were done
at locations in each sampled county group.
Sample
of States
States
were sampled with a probability proportional to total truck
vehicle miles traveled (VMT). Data on truck VMT as of October
2001 was obtained from the U.S. Department of Transportation.
States were sorted by Census region, two urbanicity categories
within each region, and truck VMT within urbanicity category.
The four Census regions are Northeast, North Central, South
and West. For this survey the most relevant way to define
urbanicity categories survey was on the basis of truck VMT
on urban highways vs. on rural highways. States with percentage
of VMT on urban highways above the median percentage for urban
highways were in the high urban category, whereas other states
were in the low urban category.
A systematic
sample of 12 states was then selected from the sorted ordering
of states using the proprietary Westat program WESSAMP (see
Appendix A).
Sample
of County Groups
County
groupings were defined before they were selected. Since the
observer planned to spend about two days collecting data in
a county group from each of the 12 sample states, the county
group must be large enough to have sufficient sites, but not
so large that observers would spend most of their time driving
from one site to another. In general, observations were to
take place at rest areas, truck stops and primary and secondary
roadways near limited access roads. Thus, counties that have
no limited access roads are not eligible for sample selection.
In general, for each sample state, contiguous counties that
have some limited access roads were combined into groups that
have a minimum number of 300 limited access highway miles.
Once the
county groups were formed, one county group was randomly selected
from each sample state. Large county groups were given a higher
probability of selection than small county groups. There was
no data by county on the ideal measure of size, truck VMT.
In a few states, VMT for all vehicles by county was available
and was used for the measure of size. For other states, the
total population from the 2000 Census was used as the measure
of size. Sampling was done directly proportional to the measure
of size. Two other possibilities were considered: 1) Sample
proportional to the square root of the measure of size; or
2) Form approximately 3 size categories for each state and
assign the same measure of size for each county group within
a category (with the largest size category getting the largest
measure of size). In thinking about how sample weights would
be determined, researchers concluded that there would be no
way to determine correct weights if either of these measures
of size were used, and thus decided against them.
Ten different
samples of PSUs were selected in the twelve states. For each
sample, the researchers determined the proportion of limited
access highway mileage that is inside a metropolitan statistical
area (MSA). (A PSU that had some counties inside an MSA and
some counties outside an MSA was assigned a proportion based
on the total limited access highway mileage in an MSA for
the PSU.) The sample used had the proportion inside an MSA
closest to the average estimate (expected value) across all
possible samples from the set of states. See Valliant et al
(2000) for a discussion of this "balanced" sampling.
This methodology is not biased and assured that a bad luck
sample that badly under-represents MSAs or that badly over-represents
MSAs was not obtained.
Observations
were conducted in Primary Sampling Units (PSUs)-groups of
counties-in 12 states. The 12 states were randomly selected
with the probability of selection proportional to annual truck
vehicle miles traveled. The PSUs are groups of contiguous
counties that have at least 300 miles of limited access roadway.
The states and the sampled county group within each state
are shown below:
| State |
PSU
- Counties |
| Arkansas |
Benton
Crawford
Franklin |
Johnson
Pope |
Sebastian
Washington |
| California |
Imperial
San Diego |
|
|
| Georgia |
Barrow
Butts
Clayton |
DeKalb
Gwinnett |
Henry
Rockdale |
| Idaho |
Ada
Canyon
Elmore |
Gooding
Payette |
| Illinois |
DeKalb
DuPage |
Kane
Kendall |
Lake
McHenry |
| Kentucky |
Clay
Knox
Laurel |
Leslie
Madison
Perry |
Rockcastle
Whitley |
| Missouri |
Boone
Callaway |
Cole
Cooper |
Lafayette
Saline |
| Ohio |
Butler
Darke |
Miami
Montgomery |
Preble
Warren |
| Pennsylvania |
Allegheny
Beaver |
Greene
Washington |
| Texas |
Bexar
Medina |
|
|
| Virginia |
Bland
Carroll
Giles
Montgomery
Pulaski) |
Russell
Smyth
Tazewell
Washington
Wise |
Wythe
Bristol (city)
Norton (city)
Radford (city) |
| Washington |
King
Skagit |
Snohomish |
Whatcom
|
Selection
of Observation Sites
With the
sample of county groups selected, observers visited each sampled
county group and selected locations where the safety belt
usage of commercial trucks could be observed. The first step
in the selection of observation sites was the examination
of detailed maps of the selected PSU - both county and city
maps as applicable. A route through the area was laid out
so that most of the major roadways, exit and entrance ramps,
rest areas and truck stops could be visited.
When potential
observation sites were located, the observer parked in a suitable
location and began observing. If truck traffic volumes were
acceptable - at least 2 commercial vehicles in 5 minutes -
and the vantage point adequate, the observer began data collection.
If the vantage point was not acceptable but truck volumes
were acceptable, the observer attempted to find another vantage
point. Once it was determined that a given location provided
both adequate truck volumes and a suitable vantage point,
data collection began. Thus, convenience sampling was used
to select the specific observation locations. A listing of
the specific location of each data collection site is included
in Appendix B.
E.
OBSERVATION PROCEDURE
Shoulder
belt usage in all Class 7 and Class 8 trucks was observed.
Observations were made over a 2-day period in each of the
12 states. A minimum of 10 hours was spent observing safety
belt usage in each of the 12 states.
At each
potential observation site (see previous section), the following
protocol was followed:
- Pilot
tests potential site for 10 minutes.
- Record
name of state, name of county, site number, roadway name
and/or number and cross street, weather, date and begin
and end time of day for each data collection session.
- Record
direction of travel (e.g., site 1 = southbound; site 2 =
northbound) at locations with more than one site.
- Record
an observation number for each driver observed.
- Observe
truck class and trailer type (e.g., class 8-flatbed; class
7-2 axle van).
- Observe
fleet type (e.g., major US fleet, local carrier).
- Observe
driver seat belt use, non-use or misuse.
- Observe
HAZMAT placard.
- Record
the data points (5, 6, 7 and 8 above) and look up to observe
the next driver to arrive at the site.
- Observe
alternate direction of travel, if possible, for locations
where more than one site is observed simultaneously.
- Conduct
observations for a minimum of 45 minutes per site.
The following
data was recorded on each vehicle observed:
- Shoulder
Belt Usage - Yes, No, Incorrect
- Truck
Type
-
- Class 7: Heavy Duty - 6 or more tires
-
Single Van
-
Single Tanker
-
Single Dump
-
Other (refuse, concrete, etc.)
-
Class 8: Heavy Duty - 10 or more tires
-
Single Van
-
Single Tanker
-
Single Dump
-
Double Trailer
-
Bobtail
-
Other (refuse, concrete, etc.)
- Hazardous
Materials Placard - Yes or No
- Time
of Day
- Location
- Weather
Data was
collected in June, July and August of 2002.
Our objective
was to collect an hour of data at each of 10 sites in each
state, and to get about 300 observations in each state. However,
locating suitable observation sites was difficult in some
of the states. In some cases it was necessary to go to additional
locations to obtain the desired sample. In some states additional
sites could not be located so it was necessary to return to
a site that had been previously productive. There were a total
of 117 sites; 8 of these sites were visited a second time.
The time spent at sites was an average of one hour and twenty-seven
minutes in duration. The table below shows the number of sites
and the total number of trucks observed in each state.
| State |
Number
of Observation Periods |
Percent
of Total Number of Observation Periods |
Number of Trucks Observed |
Percent
of Total Observations |
| Arkansas |
9 |
7.7 |
340 |
8.7 |
| California
|
12
|
10.3
|
326
|
8.3 |
| Georgia
|
10
|
8.5
|
348
|
8.9 |
| Idaho
|
9 |
7.7 3 |
08
|
7.9 |
| Illinois
|
10
|
8.5
|
356
9.1 |
| Kentucky
|
10
|
8.5
|
326
|
8.3 |
| Missouri
|
7
|
6.0
|
329
|
8.4 |
| Ohio
|
10
|
8.5
|
353
|
9.0 |
| Pennsylvania
|
10
|
8.5 |
318 |
8.1 |
| Texas
|
9 |
7.7 |
276 |
7.1 |
| Virginia
|
11
|
9.4
|
313 |
8.0 |
| Washington
|
10
|
8.5
|
316
8.1 |
|
Total |
117 |
100.0 |
3,909 |
100.0 |
F.
RESULTS
A total
of 3,909 trucks were selected for observation. To obtain national
estimates for seat belt usage, the weight for each observed
truck was V./(12*nis) where V. is the estimated truck VMT
for the U.S. and nis is the number of trucks observed in county
group i in state s. Standard errors were computed using the
WesVar software program developed at Westat, Inc. Of these
3,909 trucks the observer was unable to see belt usage in
a total of 67 or 1.7% of the cases. The observer missed a
total of 30 or 0.8% of the cases. A total of only 13 cases
(0.3%) of incorrectly worn safety belts were observed and
these cases were excluded from the analysis. All usage rate
data described in this section is based on the 3,799 valid
observations that were made.
Observations
were made at a total of 117 locations on interstate exit/entrance
ramps and near trucks stops, as well as at signalized intersections.
The following shows the distribution of the different kinds
of data collection locations:
| Location |
Percentage of Sites |
| Truck
Stops/Rest Areas |
39% |
| Signal
Controlled Intersections - Primary Highways |
37% |
| Interstate
Exit/Entrance Ramps |
15% |
| Signal
Controlled Intersections - Secondary Highways |
9% |
Of
the vehicles observed about 10% were Class 7 (6 or more tires,
commercial vehicles). The remaining 90% were Class 8 (10 or
more tires). Thus, the majority of the vehicles observed were
the larger commercial vehicles weighing over 33,000 lbs. gross
vehicle weight (GVW). Vehicle classes 1 through 6 with a GVW
of less than 26,000 lbs. were not observed.
Usage
Rates - Truck Class
The overall
safety belt usage rate by truck class is shown in Figure 1.
Class 7 trucks, those with at least six tires, but less than
ten, had a safety belt usage rate of 54%. Class 8 trucks,
those with more than 10 tires, had a usage rate of 47%. The
overall safety belt usage rate for all commercial vehicles
was 48%, as recorded during this study in 2002. For comparison
purposes safety belt usage rates from the June 2003 NOPUS
survey are included in the figure. Although passenger cars
(81%) and SUV/vans (83%) have relatively high usage rates,
pickup trucks (69%) have a usage rate closer to that of the
commercial vehicles. These data suggest an inverse relationship
between safety belt usage and vehicle weight. The belt usage
rate of all commercial vehicles (48%) is 31% lower than that
of all passenger vehicles (79%).
Figure
1. Safety Belt Usage Rate of Drivers by Vehicle Type and Class
of Truck (Standard errors in parentheses) (Passenger vehicle
data from NOPUS, June 2003), (Truck data collected in 2002)
Usage
Rates - Fleet Type/Truck Type
Figure
2 shows the commercial vehicle usage rate by fleet type. The
usage rates for those units where the tractor was identified
as a major regional or national fleet was 55%. For trucks
that were either independent or local fleets the usage rate
was lower at 44%. Educational programs used by many of the
major carriers may have resulted in higher usage rates among
their drivers.
Figure
2. CMV Safety Belt Usage Rate by Fleet Type (Standard errors
in parentheses)
The safety
belt usage rate by trailer type and the presence of hazardous
materials (HAZMAT) placard is shown in Figure 3. All of the
vehicles in this figure were class 8 trucks. The highest safety
belt usage rates were observed among the single tankers-61%.
Doubles also had relatively high usage rates at 56%. The lowest
rates were seen in single trailer dumps (26%) and in bobtails
(33%). The usage rate for the most common trailer type, the
single trailer enclosed van, was 51%-very close to the overall
average of 47% for all class 8 vehicles.
The right-hand
side of Figure 3 shows the usage rates for the HAZMAT tractor-trailer
combinations. The drivers of tractors pulling trailers with
a HAZMAT placard displayed had the highest usage rate of any
of the various categories observed-67%. Although the exact
reason for this relatively high usage rate is largely conjecture,
it would appear that the increased safety education and awareness
that is provided their drivers may account for the higher
safety belt usage rate among this group.
Figure
3. Safety Belt Usage Rate by Trailer Type and HAZMAT/Not HAZMAT
(Class 8 only) (Standard errors in parentheses)
Usage
Rates - Location/Day of Week/State
The usage
rates for all truck types for the various kinds of observation
locations is shown in Figure 4. As discussed the observations
were made at locations in the selected counties where it was
possible to safely observe the driver and where a reasonable
amount of truck traffic was present. The locations varied
from exit and entrance ramps on interstate highways to signal
controlled intersections on local highways. Similar usage
rates were found at the interstate entrance and exit ramps
(52%) and at locations near trucks stops near the interstates
(50%). Lower, but similar, usage rates were found at signalized
intersections on principal highways (44%) and at signalized
intersections on secondary highways (45%).
Figure
4. Safety Belt Usage Rate by Type of Location (Standard errors
in parentheses)
Figure
5 shows the usage rate by day of week. During weekdays usage
varies from a low of 46% on Mondays to a high of 50% on Wednesdays.
Because the Sunday rate is so low (43%) the usage rate for
weekends (45%) is lower than it is during weekdays-49%. This
usage pattern is exactly reverse of what is seen in passenger
vehicles. Passenger vehicle usage tends to be 2% or 3% higher
on weekends for drivers. Usage rates for passengers in passenger
vehicles are typically 5% or 6% higher on weekends.
Figure
5. Safety Belt Usage Rate by Day of Week (Standard errors
in parentheses)
Figure
6 shows the safety belt usage rate by state. Usage rates for
all passenger vehicles observed in the 2001 State belt use
surveys and for the commercial motor vehicles observed in
this study are shown. The commercial vehicle safety belt usage
rates vary from a low of 41% in Texas to a high of 58% in
Washington State. The passenger vehicle usage rate shows considerably
more variation than the truck data. Usage varies from a low
of 55% in Arkansas to a high of 91% in California. Arkansas
had average (48%) commercial vehicle belt usage and California
had below average commercial vehicle belt usage (46%). The
highest commercial vehicle usage state, Washington, also has
relatively high passenger vehicle usage-83%.
It should
be noted that the stratified sample of states was randomly
selected to be nationally representative. The reader should
be cautioned that additional analyses of selected subsets
of the sample (e.g., geographic regions, primary/secondary
seat belt laws) were not the objective of the sampling procedure
and would not be appropriate.
Figure
6. Safety Belt Usage Rate by State for Passenger Vehicles
and Commercial Vehicle (Standard errors in parentheses) Passenger
Vehicle Data Source: 2002 State belt use surveys conducted
in accordance with section 157 of title 23, United States
Code. Commercial Vehicle Data Source: FMCSA Safety Belt Usage
Study, Sept. 2003
APPENDIX
A
MEMOS
FROM WESTAT DESCRIBING
SELECTION OF STATES AND COUNTY SAMPLES WITHIN STATES
AND THE CALCULATION OF SAMPLE WEIGHTS
This memo
specifies how Westat researchers will sample states for the
U.S. Dept. of Transportation survey of seat belt use among
commercial trucks. First, Westat will sample 12 or 13 states
with probability proportional to Total Truck Vehicle Miles
Traveled (VMT) using WESSAMP. Then county level truck VMT
data will be obtained for counties in the sampled states by
other project staff. Westat will then group counties in each
sampled state using WESPSU, after eliminating counties containing
no Interstate highways. A sample of county groups will be
selected with probability proportional to size using WESSAMP.
Within each sampled county group, data collectors will be
sent to observe seat belt use at truck stops along the Interstates
or highway access ramps. The truck stops and other observation
points will be chosen as a convenience sample by the data
collectors.
State
Sampling Frame
The input
file for sampling states will be an Excel spreadsheet called
TRUCKS2000.XLS (see attached). This spreadsheet should be
converted to a permanent SAS file named TRUCKS2000 containing
51 records: one for each state, plus the District of Columbia.
The variables should be STATE, REGION, RURALVMT, PctTruckRural,
URBANVMT, PctTruckUrban, TotalTruckVMT, PctUrbanTruckVMT.
Run a PROC CONTENTS and PROC FREQ on each variable.
Labels
STATE = "State Name"
REGION = "Census Region"
RURALVMT = "Annual VMT for Rural Highways"
URBANVMT = "Annual VMT for Urban Highways"
PctTruckRural = "Pct of Annual Rural VMT for Trucks"
PctTruckUrban
= "Pct of Annual Urban VMT for Trucks"
TotalTruckVMT = "Total Annual VMT for Trucks"
PctUrbanTruckVMT = "Pct of Total Truck VMT for Urban
Areas"
Create
a temporary input file for WESSAMP by deleting records for
Alaska (AK) and Hawaii (HI) and combining the following states:
Maryland (MD) with District of Columbia (DC), Rhode Island
(RI) with Delaware (DE). For the combined MD, DC record, set
RURALVMT
= 16104; URBANVMT = 37568; TotalTruckVMT = 4407; PctTruckRural
= 11.5; PctTruckUrban = 6.8; PctUrbanTruckVMT = 58.3; STATE="MD_DC";
REGION = "S";
For the
combined RI, DE record, set
RURALVMT
= 4386; URBANVMT = 12213; TotalTruckVMT = 991; PctTruckRural
= 9.7; PctTruckUrban = 4.6; PctUrbanTruckVMT = 57.2; STATE="RI_DE";
REGION = "NE";
(These
were obtained by summing RURALVMT, URBANVMT, and TotalTruckVMT
and obtaining the weighted average of PctTruckRural, PctTruckUrban,
and PctUrbanTruckVMT for the two states.)
On the
temporary file, create an indicator variable for Above/Below
the median Percent Urban Truck VMT (PctUrbanTruckVMT) in each
region as follows:
If REGION
= W or S then MEDPCTURB = 30;
Else if REGION = NE then MEDPCTURB = 40;
Else if REGION = NC then MEDPCTURB = 33;
If PctUrbanTruckVMT
> MEDPCTURB then HIURBTRKVMT = 1 ;
Else HIURBTRKVMT = 0;
If HIURBTRKVMT
= 0 then SORTSERP = TOTALTRUCKVMT;
Else if HIURBTRKVMT = 1 then SORTSERP = - TOTALTRUCKVMT;
Labels
MEDPCTURB = "Median Pct Urban Truck VMT"
HIURBTRKVMT = "1=Pct Urb Truck VMT > Median, 0=otherwise"
SORTSERP = "Third sort variable for sampling states"
HIURBTRKVMT
and SORTSERP will be used as sort variables within REGION
prior to sampling states. The effect of using SORTSERP in
the SORTVARS parameter of WESSAMP is to create a serpentine
sort on TotalTruckVMT within HIURBTRKVMT.
Create
a dummy stratification variable DUMMY = 1 for every record.
Select
State Sample
Select
a sample of 12 states using the attached WESSAMP parameter
sheet. Save the WESSAMP output sample and frame files to permanent
SAS files. Run a PROC CONTENTS and PROC PRINT for both files.
If the combined MD, DC or RI, DE records are sampled, split
them back into separate states with the original uncombined
variable values.
Output
the final sample file to a permanent SAS file and an Excel
spreadsheet. On the Excel spreadsheet, keep only variables
STATE REGION RURALVMT URBANVMT TOTALTRUCKVMT PCTTRUCKRURAL
PCTTRUCKURBAN PCTURBANTRUCKVMT.
Sample
Weighting
For this
survey, weighting is complicated by the fact that there is
not a fixed and known probability of selection of sites and
of trucks within each sampled county group. The probability
of selection of states and of county groups is known. The
correct weight is specified, which includes an unknown term,
and can therefore not be used. An approximation is given which
then permits the calculation of sample weights. Finally, the
survey discusses variance estimation.
Let Vis
be the estimated truck VMT for county group i in state s,
Vs = Ʃ
Vis, the estimated truck VMT for state s,
i
Vi = Ʃ Vs, the estimated truck VMT for the US,
s
Mis = the measure of size for county group i used in sampling
county groups in state s (in four states, the measure of size
was annual VMT for all vehicle types; in the remaining eight
states, the measure of size was total population),
Ms = Ʃ
Mis is the sum of the county group measures of size for state
s, and
i
nis is the number of trucks observed in county group i in
state s.
The probability
of selection for a state is (12)( Vs)/V., and
The probability
of selection for a country group is Mis / Ms.
Thus,
an appropriate weight for each observed truck is
wis =
[V./((12)(Vs))][ Ms / Mis][ Vis / nis].
The last
term of the formula is the inverse of an estimate of pseudo
sampling rate of trucks within in a county group. Note, however,
that Westat does not have a probability selection of sites
or trucks within a county group, and thus the probability
of selection can vary considerably for different trucks within
a county group. Westate has no way to account for this.
All quantities
in this formula are known, except for Vis, the county-level
truck VMT. Without Vis, the formula cannot be directly applied.
If the following approximation holds, however, then Vis is
not needed:
Mis /
Ms = Vs /Vis
Westate
then has wis = V./(12 nis). I recommend that this weight be
used. For seat belt use rates, the value for V is not important,
as it occurs in both the numerator and denominator of a rate.
The national seat belt use rate R can be estimated as:
12 R =
Ʃ wis Xis i=1 _______ 12 ? wis Yis i=1
Where
Xis = the number of belted truck drivers observed in county
group i, Yis= the number of truck drivers observed in each
county group i, and wis is the weight for observed trucks
in the I-th county group. Note that each observed truck in
the I-th county group gets the same weight in this formula.
SELECTED
MEASURES FOR STATES October 2001
TRAVEL MEASURES
ANNUAL VEHICLE-MILES OF TRAVEL
(MILLIONS)
RURAL URBAN TOTAL PERCENT
ANNUAL PERCENT ANNUAL PERCENT TRUCK URBAN
State Census Region VMT TRUCKS 3/ VMT TRUCKS 3/ VMT TRUCK VMT
Montana W 7,590 14.2 2,292 2.9 1,144 6%
Wyoming W 5,855 29.9 2,235 12.3 2,026 14%
New Mexico W 14,526 22.6 8,234 10.6 4,156 21%
Idaho W 8,549 19.4 4,985 9.3 2,122 22%
Alaska W 2,374 11.1 2,239 4.5 364 28%
Oregon W 18,689 16.9 16,321 7.7 4,415 28%
Nevada W 5,862 21.3 11,777 5.6 1,908 35%
Utah W 8,585 15.5 14,012 5.1 2,045 35%
Colorado W 16,480 14.0 25,291 4.9 3,546 35%
Arizona W 17,860 23.7 31,908 12.8 8,317 49%
Washington W 17,202 15.9 36,128 7.4 5,409 49%
Hawaii W 2,451 4.0 6,092 2.6 256 62%
California W 59,567 15.0 247,082 6.9 25,984 66%
Alabama S 28,873 5.3 27,661 1.6 1,973 22%
Arkansas S 18,736 16.4 10,431 8.7 3,980 23%
Mississippi S 24,416 18.1 11,120 11.8 5,731 23%
South Carolina S 29,009 12.1 16,529 6.7 4,618 24%
West Virginia S 13,955 14.3 5,287 12.8 2,672 25%
Oklahoma S 21,697 20.7 21,658 7.1 6,029 26%
Kentucky S 26,760 16.4 20,043 8.2 6,032 27%
Tennessee S 30,487 16.5 35,245 8.0 7,850 36%
Louisiana S 22,167 17.3 18,682 11.6 6,002 36%
Texas S 74,002 20.2 146,062 7.0 25,173 41%
Georgia S 47,523 12.9 57,487 7.8 10,614 42%
North Carolina S 44,140 12.4 45,364 9.2 9,647 43%
Virginia S 32,252 14.6 42,549 8.6 8,368 44%
Maryland S 16,104 11.5 34,070 7.1 4,271 57%
Delaware S 3,267 11.5 4,973 9.9 868 57%
Florida S 38,100 14.2 114,036 6.9 13,279 59%
Dist.of Columbia S - 0.0 3,498 3.9 136 100%
Maine NE 10,477 7.0 3,713 2.4 823 11%
Vermont NE 4,862 9.9 1,949 8.2 641 25%
New Hampshire NE 7,083 7.5 4,938 6.7 862 38%
Pennsylvania NE 45,820 15.0 56,517 8.1 11,451 40%
New York NE 36,715 11.2 92,342 5.4 9,099 55%
Rhode Island NE 1,119 4.5 7,240 1.0 123 59%
New Jersey NE 13,606 3.7 53,840 1.5 1,311 62%
Connecticut NE 7,709 7.8 23,047 5.9 1,961 69%
Massachusetts NE 8,823 2.2 43,973 1.1 678 71%
North Dakota NC 5,372 15.0 1,845 4.3 885 9%
South Dakota NC 6,519 15.9 1,913 6.6 1,163 11%
Nebraska NC 11,227 19.2 6,854 5.1 2,505 14%
Kansas NC 14,855 18.6 13,275 5.5 3,493 21%
Iowa NC 18,786 15.9 10,647 8.2 3,860 23%
APPENDIX B
SITE
LOCATIONS
State/Site
Number Location
Arkansas
- Fina
Fuel Stop and J & H Restaurant parking lot, exit 106
of interstate 30, State route 67, westbound traffic
- Same
as site 02, eastbound traffic
- Rest
stop along Interstate 30 at about milepost 96, located in
median for eastbounders and westbounders, westbound traffic
- Same
as site 03, eastbound traffic
- Pilot
Truck Stop parking lot, interstate 30 exit 121
- Parking
lot of Walgreen Pharmacy, intersection of State route 70
and 270 B (business), town of Hot Springs, all legs of intersection
- Phillip
66 Travel Center, exit 44 of interstate 30, any direction
of travel
- Flying
J Travel Center, parking across the street at Red Barn BarBQ,
exit 7 of interstate 30 near exit ramp, eastbound traffic
- Same
as site 08, westbound traffic
California
- State
route 111 southbound at Cole Road, near interstate 8, Calexico
- Same
as site 01, northbound traffic
- Imperial
Ave. exit from interstate 8 westbound traffic, on and off
ramp
- Intersection
of Highway 905 at Airway Blvd, near Otay Mesa International
Center, Chula Vista, Highway 905 both directions of traffic
- Same
as site 04, Airway Blvd both directions of traffic
- Intersection
of Highway 905 at La Media Road, Chula Vista, Highway 905
both directions of traffic
- Same
as site of 06, La Media Road both directions of traffic
- Grape
Street and State Street at on ramp to interstate 5, northbound
and southbound traffic, downtown San Diego
- Oceanside
Blvd. at off ramp from interstate 5, Oceanside, California,
both directions of travel, Oceanside Blvd and northbound
traffic on ramp to interstate 5
- Same
as site 09, northbound traffic off ramp from interstate
5 to Oceanside Blvd
- Intersection
of Highway 905 at Heritage Drive, Chula Vista, Highway 905-both
directions of traffic
- Same
as site 11, Heritage Drive southbound traffic
State/Site
Number Location
Georgia
- Intersection
of state routes 211 and 124, near exit 126 of interstate
85, route 211 both directions of traffic
- Same
as site 01, route 126 both directions of traffic
- Pilot
Truck Stop, exit 51 of interstate 285, southeast of Atlanta
- Boulder
Crest Road, near junction of interstates 285 and 695, exit
51 of interstate 285
- Parking
lot of T/A Motel and Travel Center near junction of interstates
285 and 675, junction is exit 52, traffic entering truck
stop
- Intersection
of Thurman Drive and McDonaugh Drive, all directions of
traffic but not entering truck stop at site 05
- Williams
Truck Stop, exit 221 of interstate 75
- Junction
of routes 138 and 23, City of Stockbridge, near interstate
75, all directions of traffic
- Flying
J Truck Stop, exit 201 of interstate 75
- T/A
Travel Center, exit 201 of interstate 75
Idaho
- Flying
J Truck Stop, at state routes 20/26, exit 29 of Interstate
84
- Flying
J Truck Stop at Federal Way, near exit 54 of interstate
84
- T/A
Travel Center at exit 54 of interstate 84
- Boise
Stage Stop, exit 71 of interstate 84, eastbound traffic
- Same
as site 04, westbound traffic
- Pilot
Truck Stop at Mountain Home, at exit 95 of interstate 84,
both directions of travel
- Flying
J Truck Stop, near Bliss, exit 137 of interstate 84
- Eastbound
exit 157 ramp of interstate 84
- Westbound
exit 157 ramp of interstate 84
State/Site
Number Location
Illinois
- Intersection
of state route 59 at New York and Aurora Avenues, both directions
of traffic on state route 59
- Same
as site 01, New York Ave. traffic
- Intersection
of state routes 64 and 53, near interstate 355, eastbound
traffic of route 64
- Same
as site 03, westbound traffic of route 64
- Same
as site 03, both directions of traffic route 53
- Junction
of state route 60 at exit from interstate 94, route 60 both
directions of travel and entrance ramp to southbound interstate
94
- State
route 31 near interstate 90, exit Elgin, southbound traffic
- Same
as site 07, northbound traffic
- Intersection
of state route 30 at Douglas Street, all directions of traffic
- Citgo
Truck Stop, intersection of state routes 30 and 47, near
interstate 88, all directions of traffic
Kentucky
- State
route 25, exit 29 of interstate 75, eastbound direction
of traffic
- Same
as site 01, westbound direction of traffic
- Pilot
Truck Stop, exit 29 of interstate 75
- Citgo
Truck Stop, exit 29 of interstate 75
- BP
Truck Stop, route 80 at exit 41 of interstate 75
- Intersection
of state routes 80 and 421, Manchester, near exit 20 of
Daniel Boone Parkway Toll Road, route 80 both directions
of traffic
- Same
as site 06, route 421 both directions of traffic
- State
route 80, exit 41 near interstate 75, eastbound direction
of traffic
- Same
as site 08, westbound direction of traffic
- Southbound
exit ramp 41 of interstate 75
State/Site
Number Location
Missouri
- T/A
Travel Center at exit 58 of interstate 70, eastbound traffic
- Same
as site 01, westbound traffic
- Pilot
Truck Stop at exit 101 of interstate 70, Booneville, and
both directions of traffic
- State
route 54, Petro and Conoco Truck Stops, coming and going
from state highway and interstate 70 at exit 148
- State
route 54, northbound and southbound near exit 148 of interstate
70 and interstate users, near McDonalds, Texaco and Phillips
66
- State
route 54, eastbound traffic, near exit 148 of interstate
70
- Exit
121 of interstate 70, near Budget Inn and Conoco, both directions
of traffic
Ohio
- Pilot
Truck Stop, state route 127, exit 10 of interstate 70, and
eastbound traffic
- Same
as site 01, westbound traffic
- BP
Fuel Stop, state route 127, north of exit 10 of interstate
70
- Intersection
of state route 127 at Lexington and Barron, Town of Eaton,
route 127 traffic both directions
- Same
as 04, Lexington Street and Barron Street traffic
- Intersection
of state route 4 (truck route) and state route 129, route
4 both directions of traffic
- Same
as site 06, route 129 both directions of traffic
- Intersection
of State route 63 near interstate 75 south exit ramp number
29, route 63 westbound traffic
- Same
as site 08, route 63 eastbound traffic
- Same
as site 08, cross street traffic near Town of Monroe
State/Site
Number Location
Pennsylvania
- Pilot
Truck Stop adjacent to exit 32B of interstate 70, Bentleyville/Ginger
Hill, westbound traffic
- Same
as site 01, eastbound traffic
- Rest
area near milepost 51 located on interstate 79 northbound,
south of Pittsburgh
- Junction
of state routes 30 and 18, Harshaville, all traffic directions
- Citgo
Truck Stop, exit 1 northbound of interstate 79, northbound
traffic
- Same
as site 05, southbound traffic
- Petro
Travel Center, exit 6 of interstate 70, westbound traffic
- Same
as site 07, eastbound traffic
- Exit
19 of interstate 70, westbound traffic
- Same
as site 09, eastbound traffic
Texas
- Love's
Travel Stop, exit 142 of interstate 35, just outside of
interstate 410, northbound traffic
- Intersection
of state route 90 and county route 462-Avenue E, Town of
Hondo, route 90 both directions of traffic
- Petro
Truck Stop, exit 582 eastbound of interstate 10
- Pilot
Truck Stop, exit 582 of interstate 10
- T/A
Travel Center, exit 583 of interstate 10
- Flying
J Truck Stop, exit 583 of interstate 10
- Texaco
lot, exit ramp 585 of interstate 10
- Conoco
Fuel stop, route 1604 (outer beltway loop), from interstate
37
- Same
as site 01, southbound traffic
State/Site
Number Location
Virginia
Petro
Truck Stop at exit 29 of interstate 81, Glade Spring, northbound
traffic
Same as site 01, southbound traffic
Intersection of state routes 11/19 at State Street, City of
Bristol near Old Bristol Business Center, State Street westbound
traffic
Same as site 03, Route 11/19 northbound traffic
Same as site 03, Route 11/19 southbound traffic
Same s site 03, State Street eastbound traffic
T/A Travel Center, town of Max Meadows, exit 41 of interstate
77
Flying J Truck Stop, exit 77 of interstate 81, four way stop,
northbound and southbound exit ramps
State route 58 at route 2213, exit 14 of interstate 77, near
Hillsville, southbound traffic
Lancer Travel Center, state route 660, Clayton Lake, near
exit 101 of interstate 81, exit ramp traffic
Same as site 10, Travel Center traffic
Washington
- Intersection
of Route 99-International Blvd. and 200th Street, City of
Seatac, about 1-2 miles from interstate 5. International
Blvd. traffic both directions
- Same
as site 01, 200th Street, traffic both directions
- Intersection
of State route 2 at Kelsy Street, City of Monroe, eastbound
traffic
- Same
as site 03, westbound traffic
- Exit
roadway from interstate 5, exit 275, Blaine near customs
to British Columbia, traffic going to Canada
- Same
as site 05, traffic coming from Canada
- Interstate
5, exit 230-Burlington/Whidbey Island, on ramp traffic
- Same
as site 08, off ramp traffic
- State
route 18 at 244th Street, SE, King County, Route 18 traffic
both directions
- Interstate
90, exit 25, exit ramp, eastbound traffic and westbound
traffic
|
