The Safety Impacts of Differential Speed Limits on Rural Interstate Highways
Publication No. FHWA-HRT-04-156
View PDF Version (80 kb)
Background
The Surface Transportation and Uniform Relocation Assistance Act
(STURAA), enacted on April 2, 1987, permitted individual States to raise
speed limits from the previously mandated national speed limit of 88
kilometers per hour (km/h) to 105 km/h (55 miles per hour (mi/h) to 65
mi/h) on rural interstate highways. After this date, some States uniformly
raised their speed limits for passenger cars and trucks. Other
States raised the speed limit for passenger cars only while leaving the
truck limit at 88 km/h (55 mi/h), creating a Differential Speed Limit
(DSL)—different speed limits for cars and heavy trucks traveling on the
same roadway. Speed limits that are the same for both passenger cars
and trucks are known as Uniform Speed Limits (USL). On November 28,
1995, the national maximum speed limit was repealed, giving States
further flexibility in setting their limits. By 2002 several States had
experimented with both DSLs and USLs.
Objective
Previous studies conducted during the early 1990s that compared
USLs and DSLs were constrained because of the limited data available.
Over the past decade several States have either eliminated or implemented
a lower limit for trucks providing the opportunity for a before
and after study that might provide more reliable results. In 1994,
Virginia switched from a differential speed limit of 105 km/h (65 mi/h)
for cars and 88 km/h (55 mi/h) for trucks to a uniform speed limit of
105 km/h (65 mi/h) for all vehicles. In 1996, Arkansas adopted a differential
speed limit by raising the speed limit for cars to 113 km/h (70 mi/h)
but maintaining 105 km/h (65 mi/h) for trucks. In 1998, Idaho changed
from a uniform speed limit of 121 km/h (75 mi/h) for all vehicles to
a 105 km/h (65 mi/h) limit for trucks. With more than a decade having
elapsed since the passage of the STURAA, the Federal Highway
Administration (FHWA) sponsored a long-term study to investigate
the effect of USLs and DSLs on vehicle speeds and crashes on rural
interstates nationwide.
TABLE 1. Accident Proportions by Speed Limit, Collision Type, and Vehicle Involvement.
(Adapted from table 32, reference 1)
Speed Limit |
Rear End |
Sideswipe |
Other |
Car-Into-Truck |
Truck-Into-Car |
Car-Into-Truck |
Truck-Into-Car |
Car-Into-Truck |
Truck-Into-Car |
USL: 105 km/h and
88 km/h (65 mi/h and
55 mi/h) |
10.94 |
10.78 |
22.12 |
21.07 |
2.57 |
2.0 |
DSL: 105/88 km/h and
105/97 km/h (65/55
mi/h and 65/60 mi/h) |
13.70 |
6.86 |
21.52 |
14.96 |
2.07 |
0.99 |
Literature Review
The safety effects of differential
speed limits for cars and trucks
have been inconclusive in previous
studies. Some studies found
no difference between USL and
DSL (references 1, 2, 3, and 4).
Other studies found one or the
other to be a better policy
choice.(5,6) The studies were mainly
cross-sectional comparisons of
speed or crashes in States with
different speed limits for cars and
trucks to those in nearby States
with uniform speed limits. The
differences (or lack thereof) observed
between States could be
due to variations in traffic density,
weather, and many other factors.
Impact of DSL on Mean Speed. In 1990, Freedman and Williams
analyzed speed data collected at
54 sites in 11 Northeastern States
to determine the effect of DSL
on mean and 85th percentile
speeds.(5) Six States had retained
a USL of 88 km/h (55 mi/h), three
had raised USLs to 105 km/h (65
mi/h), and two States employed
a DSL of 105 km/h (65 mi/h) for
Sideswipe
Into-Car
Car-Into- Truck-
Truck Into-Car
21.07 22.12 10.78
14.96 21.52 6.86
cars and 88 km/h (55 mi/h) for
trucks. For passenger cars in the
DSL States, the mean speed and
85th percentile speeds were not
significantly different from the
105 km/h (65 mi/h) USL States.
For trucks in the DSL States, the
mean and 85th percentile speeds
were not significantly different
from those of the 105 km/h (65
mi/h) USL States. Similar results
were obtained when comparing
the percentage of vehicles complying
with the speed limit.
In 1994 Harkey and Mera found
no significant difference between
passenger car and truck mean
speeds when comparing USLs
and DSLs.(1)
Impact of DSL on Speed Variance. The implication of increased
speed variance is an increase in
interactions between vehicles,
leading to a potential increase in
crashes. Council et al. in 1998
found that for rear-end collisions
between cars and trucks, a highspeed
differential increases the
severity of the crash.(6) However,
Harkey and Mera found no signif-
Other
Into-Car
Car-Into- Truck-
Truck
2.01 2.57
0.99 2.07
icant differences between car
speed variances at the USL and
DSL sites.(1) Furthermore, they
found no difference between
the speed distributions for both
cars and trucks for the 105/97
km/h (65/60 mi/h) and 105/105
km/h (65/65 mi/h) speed limits.
A 1974 study by Hall and Dickinson
showed that speed differences
contributed to crashes,
primarily rear end and lane
change collisions.(2)
Impact of DSL on Crashes. Harkey and Mera investigated
crash results from 26 sites in 11
States.(1) The study investigated
the percentage of three different
crash types (rear-end, sideswipe,
and all other) for USLs and DSLs.
Table 1 shows that a higher
proportion of car-into-truck and
truck-into-car crashes occurred in
USL States, except for rear end
crashes where more car-into truck
collisions happened in the
DSL group.
A study by Garber and Gadiraju
conducted in 1991 compared crash rates in the adjacent States
of Virginia (DSL) and West Virginia
(USL).(3) The increase in the
posted speed limit for trucks to
105 km/h (65 mi/h) did not result
in a significant increase in fatal,
injury, and overall accident rates.
There was, however, some evidence
that the DSL may increase
some types of crash rates while
reducing others.
According to Hall and Dickinson,
the existence of a posted DSL,
however, was not related to the occurrence
of truck crashes.(2) Finally,
an evaluation conducted by the
Idaho Department of Transportation
found that a change from USL to
DSL did not increase crashes.(4)
TABLE 2. Types of Speed Limits Throughout the 1990s on Rural Interstate Highways.
Maintained USL |
Maintained DSL |
Changed from USL to DSL |
Changed from DSL to USL |
Arizona
Iowa
North Carolina
|
Illinois
Indiana
Washington |
Arkansas
Idaho |
Virginia |
Caveats to the Following Study. There are six limitations that may
apply to the speed and crash
rates results of this study:
-
Selected sites may be a biased
sample.
-
It was not possible to obtain
speeds by vehicle type (passenger
cars and truck).
-
Durations used in this study are
relatively short.
-
Rural interstates were analyzed
at an annual level of detail.
-
Sample size used in the statistical
tests associated with the
speed analysis was defined as
the number of speed monitoring
sites and varied by state.
-
Law enforcement patterns during
these time periods are unknown
Methodology
Three general steps comprised
the methodology used in this
research:
-
Speed and crash data were collected
from States that had
been identified as having
changed their speed limits at
least once during the 1990s
from USL to DSL or vice versa.
- Conventional statistical approaches
(analysis of variance –
Tukey’s and Dunnett’s tests)
were used to analyze speed and
crash data from these States.
-
Empirical Bayes procedure was
applied to these crash data.
Nine States were selected so they
could be divided into four policy
groups based on the type of
speed limit employed during the
period, as shown in table 2.
Figure 1. Mean Speed for All Vehicles. |
![Figure 1. Mean Speed for All Vehicles](images/figure1.gif)
1 mi/h = 1.6 km/h |
Figure 2. Total Crash Rates per million vehicle-miles traveled (VMT).
Note that speed limits changed in Idaho (1996, 1998, Arkansas (1996), and Virginia (1994) |
![Figure 2. Total Crash Rates per million vehicle-miles traveled](images/figure2.gif) |
Results
Vehicle Speeds
Five speed measures (mean
speeds, speed variance, 85th percentile
speeds, median speeds,
and noncompliance rates) were
analyzed for the five States where
such speed monitoring data were
readily available. Speed data
were generated from speed monitoring
stations throughout the
States; individual speeds on specific
interstates were not always available. It was not possible to
obtain speeds by vehicle type
(passenger cars and trucks).
Figure 1 illustrates the trends in
mean speeds, for all vehicle types,
among the five States with speed
data. Data could not be obtained
for all years during the time periods.
Except for Virginia, the main
observation is that all speeds appear
to be increasing over time,
regardless of speed limit type.
Crashes
Figure 2 presents an overall representation
of crash data from the
various States. While the data in
figure 2 are based on crash rates
and validates the results generated
by the Empirical Bayes
method, it should be noted that
the Empirical Bayes method did
not use crash rate in the modeling
process but included annual average
daily traffic (AADT) and section
length as independent variables.
Only North Carolina showed
a significant increase in the total
crash rate; the other States
showed no significant change in
the total crash rate.
Caveats to the Use of Empirical Bayes Method
Several data limitations might
have influenced the results of the
Empirical Bayes analysis.
-
Comparison groups were imperfect.
Ideally, the comparison
group would have been selected
from the same State at the
same time as the studied group.
-
Although speed monitoring data
were available to understand
statewide speed trends, specific
speeds for every interstate section
used in the crash analysis
were not available.
-
The crash estimation model
used only two variables—AADT
and section length. There may
have been other relevant variables
that were not included in
the model.
TABLE 3. Impact of Speed Limit Changes, Confidence Intervals and Crash Increases According to the
Empirical Bayes formulation.
Crash Type |
Ratio θ |
Confidence
Interval
Lower Bound
Increase |
Confidence
Interval
Upper Bound
Increase |
Crash
Effect |
Maintained a uniform limit (Arizona) |
Total crashes |
1.26 |
24.2% |
28.6% |
Increase |
Total crashes with truck involved |
1.16 |
12.1% |
20.7% |
Increase |
Maintained a uniform limit (North Carolina)* |
Total crashes |
1.26 |
19.9% |
31.9% |
Increase |
Total crashes with truck involved |
0.91 |
-19.7% |
1.5% |
No change |
Maintained a differential limit (Washington) |
Total crashes |
0.99 |
-6.6% |
5.0% |
No change |
Changed from uniform to differential (Arkansas) |
Total crashes |
1.07 |
0.4% |
13.4% |
Increase |
Total crashes with truck involved |
1.31 |
18.9% |
42.8% |
Increase |
Changed from uniform to differential (Idaho) |
Total crashes |
1.29 |
13.2% |
46.7% |
Increase |
Total crashes with truck involved |
2.46 |
68.6% |
224.9% |
Increase |
Changed from differential to uniform (Virginia) |
Total crashes |
1.15 |
12.9% |
17.2% |
Increase |
Total crashes with truck involved |
1.25 |
20.0% |
29.8% |
Increase |
*North Carolina maintained their uniform limit but also raised this limit for both passenger car and trucks
Findings
To evaluate how a treatment affects
safety, the Empirical Bayes
method predicts what the expected
crash frequency would
have been during the after period
had there been no such treatment
and then compares it to
the actual number of crashes
that occurred during the after
period. Using the Empirical
Bayes technique, the ratio θ of the "actual" after crashes to the
"would have been" after crashes is calculated. If the ratio θ is
greater than 1.0, then the treatment
(e.g., a change from one
type of speed limit to another)
resulted in an increase in the
number of crashes.
In most cases, θ was greater
than 1.0, as shown in table 3, indicating
an increase in crashes.
However, the data in table 3 are
not consistent. The ratio θ for
total crashes in Virginia, which
changed from DSL to USL, is
higher than one of the States that changed from USL to DSL
(Arkansas) but lower than the
other state that changed from
USL to DSL (Idaho). The table
also shows that for total crashes,
θ was approximately 1.0 for the
State that maintained DSL (Washington)
while it was greater than
1.0 for States that maintained
USL (Arizona and North Carolina).
Additional crash types, such as
rear-end type crashes, are discussed
in the final report.
Conclusions
The results presented in table 3
are on a State-by-State basis.
Overall, the study was not able to
isolate or measure the effect of
USL/DSL changes. The effect of
the DSL, if any, is not enough to
be detected in the aggregate
speed data that were analyzed.
Speed characteristics were generally
unaffected by a USL or
DSL policy. Except for Virginia,
mean speeds tended to increase
over the 1990s regardless of
whether the State maintained a
USL, maintained a DSL, or
changed from one to the other.
In some cases the increase in
speed was significant, in other
cases it was not.
No consistent safety effects of
DSL as opposed to USL were
observed within the scope of
the study. The mean speed and
crash rates tended to increase
over the 10-year period, regardless
of whether a USL or DSL
limit was employed. The Empirical
Bayes methodology suggested
that crash risk during the
study period increased for all
four policy groups.
References
-
Harkey, D. L. and Ruben Mera: Safety Impacts of Different Speed Limits on Cars and Trucks, U.S. Department of
Transportation, Federal Highway Administration, Publication No. FHWA-RD-93-161, Washington, D.C., May 1994.
-
Hall, J. W. and L. V. Dickinson. An Operational Evaluation of Truck Speeds on Interstate Highways, Department
of Civil Engineering, University of Maryland, February, 1974.
-
Garber, N. J. and R. Gadiraju., Impact of Differential Speed Limits on Highway Speeds and Accidents,
Department of Civil Engineering, University of Virginia, Charlottesville, VA, 1991, pp13.
-
Idaho Transportation Department Planning Division, Evaluation of the Impacts of Reducing Truck Speeds
on Interstate Highways in Idaho, -Phase III, Final Report, December, 2000, National Institute for Advanced
Transportation Technology, University of Idaho.
-
Freedman, M. and A. F. Williams. "Speeds Associated with 55 mi/h and 65 mi/h Speed Limits in Northeastern
States," ITE Journal, pp. 17-21, Vol. 2, No. 2, 1992, Institute of Transportation Engineers, Washington, D.C.
- Council, F. M. et al. "Applying the Ordered Probit Model to Injury Severity in Truck-Passenger Car Rear-End
Collisions," Transportation Research Record 1635, Transportation Research Board, National Research Council,
Washington, D.C., 1998.