TRANSPORTATION
RESEARCH
RECORD
1244
17
All-Way Stops: A
New
Policy
STEPHEN
CELNIKER
This project was undertaken
to
develop a new all-way stop policy
that would, with success and credibility, select intersections best
suited
to
all-way stop controls. A variety of categories
is
considered
by
the new policy: accidents, unusual conditions, traffic volumes,
and pedestrian volumes. Each category contributes points to a
total that may, in sum, justify all-way stops for the intersection.
Conversely, the circumstances within one category may be
suffi-
ciently extreme as to justify all-way stops based on that category
alone. Existing
all-way stop policies were determined
to
not be
sutliciently flexible. The
new
policy combines the best features
from national policies and the old City of
San Diego policy. Also,
the provisions within the new policy are derived from research
and experience with all-way stops, not simply modifications of
traffic signal warrants. The policy
was
tested
by
comparing acci-
dents and field performance in a before-and-after study of existing
all-way stop intersections. Some of these intersections met the
all-
way
stop criteria in the
new
policy, whereas others did not. The
study showed convincingly that the intersections
that
met the new
policy's criteria had fewer accidents and stop sign violations than
the intersections that did not.
San Diego, like many cities, has struggled with the issue of
all-way stops for many years. The city receives many requests
for all-way stops, which can be an emotional issue for some
citizens.
To
many elected officials, a group of citizens request-
ing an all-way stop may themselves provide sufficient warrant
to install an all-way stop, regardless of whether traffic
engi-
neering warrants have been met. Traffic engineers, however,
want to
be
able to differentiate good all-way stop candidate
intersections from
bad
ones through analysis
of
operational
and safety factors.
Part
of the problem
is
that many engineers,
in
San Diego and elsewhere, are not comfortable with the
Manual
on
Uniform Traffic Control Devices
(MUTCD)
(1)
warrants.
A
better
all-way stop policy that
is
accepted and respected
by
· both professionals and nonprofessionals will make it more
likely for a confident engineering staff to successfully limit
all-way stop installations to only those locations where the
safety and operation
of
the intersection will improve with
all-way stops.
Traffic
Engineering
Principles
The function
of
all-way stops
is
to control the right-of-way
assignment at intersections. With all-way stops, vehicles
on
the intersecting streets alternate having the right-of-way.
Therefore, all-way stops function best when the traffic volume
at the intersection is high enough that vehicle conflicts are
common and when the traffic volume
is
evenly split between
the intersecting streets. All-way stops may also be effective
City of San Diego Engineering and Development Department,
1222
First Avenue, M.S. 405, San Diego, Calif. 92101.
at locations where there have been numerous correctable right-
angle type accidents or where numerous unusual conditions
exist.
It
is
neither wise nor practical to install all-way stops indis-
criminately.
On
streets with frequent stops, motorists tend to
drive at higher speeds to make up for the
"lost time." Some
motorists may even be tempted to disregard stop signs when
there
is
no apparent
"need"
to stop because of cross traffic,
pedestrians,
or
limited visibility. When motorists fail to obey
stop signs, they are jeopardizing safety for themselves,
other
drivers, and pedestrians. Furthermore, the installation of
unwarranted stop signs on major streets can create excessive
queuing, delay, exhaust emission, fuel use, and noise.
PROBLEMS WITH EXISTING ALL-WAY STOP
POLICIES
The
MUTCD
policy has three warrants. For an all-way stop
to be justified, only
one
warrant must be met, but the warrant
must be met in its entirety.
The
MUTCD
warrants the following (1):
1.
Where traffic signals are warranted and urgently needed,
the multiway stop
as
an interim measure that can be installed
quickly to control traffic while arrangements are being made
for the signal installation.
2.
An
accident problem,
as
indicated
by
five
or more reported
accidents in a 12-month period of a type susceptible to
cor-
rection by a multi way stop installation. Such accidents include
right- and left-turn collisions, as well
as
right-angle collisions.
3. Minimum traffic volumes:
•
The
total vehicular volume entering the intersection
from all approaches must average at least
500
vehicles
per
hour
for any 8 hours of an average day; and
•
The
combined vehicular and pedestrian volume from
the minor street or highway must average at least
200
units
per
hour for the same 8 hours, with an average
delay to minor street vehicular traffic of at least
30
seconds
per
vehicle during the maximum hour; but
• When the 85th percentile approach speed of the major
street traffic exceeds
40
mph, the minimum vehicular
volume warrant is 70 percent of the above require-
ments.
There are numerous reasons to question the
MUTCD policy.
First, the
MUTCD
all-way stop policy
is
dependent on signals.
Warrant 1 states that all-way stops may be used
as
interim
measures before signal installation. Warrant 2
is
a variation
of Signal
Warrant
6,
and Warrant 3
is
nearly identical to Signal
Warrant
1.
18
It
is
questionable to rely on an all-way stop policy derived
from signais, not stop signs. The poiicy does
not consider
accidents or volumes when the numbers are below the spec-
ified thresholds.
The
MUTCD policy does not consider other
factors that should be examined
in
an all-way stop evaluation,
such
as
visibility, schools,
or
pedestrians. Furthermore, the
"mixed" situation (moderate volumes, a
few
accidents, some
pedestrians)
is
not addressed.
CITY OF SAN DIEGO'S EXPERIENCE
As an alternative
to
the national policy, the City
of
San Diego
developed an all-way stop policy based on a point system
in
i962. The system was based on severai warrants, each worth
a few points. All-way stops were justified at candidate inter-
sections that were assigned a majority of the total available
points. This policy was an improvement over the
MUTCD
policy because it was not dependent on signals, and it addressed
the areas that the
MUTCD policy overlooked. Another strength
was the introduction of the Traffic Volume Difference War-
rant, which awarded points to intersections based on the close-
ness of the traffic volumes on the intersecting streets.
The policy also had several weaknesses. For instance, no
single warrant could
in
itself justify all-way stops. Each war-
rant simply contributed points to a total. In some circum-
stances, a candidate intersection may have received maximum
points from one or more warrants but still did not qualify for
all-way stops because a majority of the total points had not
been accumulated. Another weakness
was
that the policy did
not contain the
MUTCD provision for using all-way stops
as
interim measures before installing traffic signals.
City staff encountered situations in which engineering judg-
ment indicated that all-way stops would be appropriate at a
particular
location, yet neither the MUTCD warrants nor the
city's own policy could justify the installation. Consequently,
the City began
in
1986 to research all-way stops and develop
a revised all-way stop policy. The goals of the new policy were
as follows:
1.
Consistency. The policy should be in conformance with
traffic engineering principles of safety and operation for all-
way stop intersections.
2. Accountability.
The
policy should be based on all-way
stops, not signals.
3.
Flexibility. The policy should equally consider intersec-
tions that have extreme circumstances in one category that
may justify all-way stops,
as
well
as
intersections that have a
combination
of
factors, none of which individually would jus-
tify all-way stops.
4.
Selectivity. The policy should be effective at distinguish-
ing the candidate intersection that
will
benefit from the instal-
lation of all-way stops.
THE NEW
POLICY
The
new policy consists of
five
warrants and a total of
50
points. All-way stops may be justified at intersections that are
assigned
25
or
more points. The 25-point requirement may
TRANSPORTATION
RESEARCH
RECORD
1244
be waived, and all-way stops justified, under any one of the
following
special provisions:
1.
Five or more accidents susceptible to correction by all-
way stops have occurred in a 12-month period.
2.
Traffic signals are warranted and not yet installed.
3.
The intersection has an extreme combination of unusual
conditions, and engineering judgment determines that the
location would be best served
by
all-way stops. Examples of
unusual conditions are a school, fire station, playground, bus
route, steep hill, and visibility limitation. A school
in
itself
is
not considered to be sufficie
!1
t justification for all-way stops.
Provisions 1 and 2 are adopted from the MUTCD warrants.
Provision 3 should be used sparing!
y,
usually after less severe
controls have been attempted.
The following includes an explanation of each warrant:
1.
Accident
experience-maximum
15
points. Three points
are assigned for each correctable accident that occurred
in
the preceding 12-month period.
2.
Unusual
conditions-maximum
5 points. Points are
assigned for unusual conditions based on engineering judg-
ment. The point value assigned to each condition should be
correlated to the improvement to the situation that all-way
stops would provide. When awarding points in this warrant,
it
is
important to consider only the actual benefits that all-
way stops provide, not the perceived benefits attributed to
all-way stops
by
many nonprofessionals. Speed control should
never be a basis for awarding points.
3.
Traffic
volumes-maximum
15
points. Two tables, one
for the minor street and one for the major street, are used
to assign points based on volume.
The
major street
is
defined
as
the traffic approaches that are not controlled
by
stop or
yield signs at the time of the evaluation. The minor street
is
defined as the approaches that are controlled. For the minor
street, the number of points awarded increases
as
the volume
increases up to a maximum
often
points. For the major street,
the maximum of five points
is
assigned to a range of volumes
at which all-way stops function best. Above or below this
optimum volume range, fewer points are awarded. To deter-
mine the optimum range for ali-way stop voiumes in the new
policy, the
1985
Highway Capacity Manual (2) was consulted.
The following
is
the method used for deriving "ideal" volume:
The
1985
Highway Capacity Manual
was
consulted for
determining the point assignment tables for traffic volume.
The level-of-service
(LOS) C service volumes for four all-way
stop intersections are as follows:
De
mand
Split
50150
55
1
45
65
1
40
65
1
35
70130
LOS
C Service Volume (vph)
by
Lane
Configuration
2
by
2
2
by
4 4
by
4
1,200
1,800
2,200
1,140 1,720 2,070
1,080
1,660
1,970
1,010 1,630
1,880
960
1,610
1,820
The tabulation
is
sorted into demand splits ranging from
501
50
to
70130
and lane configurations
(2
by
2, 2
by
4, and 4 by
4)
.
It
was determined that the traffic volume point assignment
table should be derived from the case
of
a
50
/
50
demand split
Celniker
at a two-lane by two-lane intersection.
The
LOS C service
volume for this situation
is
1,200 vehicles
per
hour (vph)
entering the intersection.
Since the City
of
San Diego uses 4-hour counts for traffic
studies, the
1,200 vph translated into 4,800 vehicles in 4 hours.
Therefore, with an ideal
50/50 split, each street should have
a 4-hour approach volume of
2,400 vehicles. Consequently,
the figure
of
2,400 vehicles
is
within the maximum point range
for
both
the
major
street and the minor street point assign-
ment tables. For the major street, the optimum range
is
between
2,201 and 2,600 vehicles in 4 hours.
For
the minor street, all
volumes above 2,201 are considered optimum and are assigned
maximum points.
The
point assignment tables are shown in
Table 1.
4.
Traffic volume
difference-maximum
10
points. This
warrant differs from the
"traffic volumes" warrant in that it
considers only the difference between the 4-hour volumes of
the two streets. All-way stops function best when the differ-
ence between the volumes
is
small. Accordingly, a small traffic
volume difference is assigned maximum points. The point
assignment table for this warrant
is
shown in Table
2.
5.
Pedestrian
volumes-maximum
5 points. The volume
of pedestrians crossing the major street
is
of concern when
evaluating for all-way stops.
One point
is
assigned for each
set
of
50 pedestrians in 4 hours,
as
shown in Table
3.
An evaluation sheet
is
shown in Figure
1.
TABLE
1
POINT
ASSIGNMENT
FOR
TRAFFIC
VOLUME
Major Street
Minor Street
4-hour Volume
Points
4-hour Volume
0-1,000
1,001-1,300
1,301-1,600
1,601-1,900
1,901-2,200
2,201-2,600
2,601-2,900
2,901-3,200
3,201-3,500
3,501-3,800
3,801-over
0
0-400
1
401-600
2
601-800
3
801-1,000
4
1,001-1,200
5
1,201-1,400
4
1,401-1,600
3
1,601-1,800
2
1,801-2,000
1
2,001-2,200
0
2,201-over
TABLE
2 POINT ASSIGNMENT
FOR
TRAFFIC
VOLUME
DIFFERENCE
Volume Difference
( 4-hour count)
0-150
151-300
301-450
451-600
601-750
751-900
901-1,050
1,051-1,200
1,201-1,350
1,351-1,500
1,501-over
Points
10
9
8
7
6
5
·4
3
2
1
0
Points
0
1
2
3
4
5
6
7
8
9
10
TABLE
3 POINT ASSIGNMENT
FOR
PEDESTRIAN
VOLUME
No.
of
Pedestrians Crossing Major
Street in 4 hours
Points
0 0
1-50
1
51-100
2
101-150 3
151-200 4
201-over
5
TESTING THE NEW POLICY
19
Once it
had
been developed, there was interest
in
how the
new policy compared to the city's previous policy. A total
of
23
intersections in the City
of
San Diego were used to test
the ability
of
the new policy to select intersections that benefit
from and function well with all-way stops. The intersections
chosen for the study
all
had all-way stops that had been installed
(either by engineering judgment
or
City Council directive)
despite having failed to
meet
the city's previous policy. The
intersections were then reevaluated,
by
using the new policy
with data from the original evaluation.
Fourteen of the intersections met the criteria of the new
policy.
That
is, if the new policy had been in effect at the
time that the intersections were originally evaluated for all-
way stops, then
14
of the
23
would have qualified.
The
14
were placed in Group A for comparison purposes. The
remaining nine intersections, those that failed to meet all-way
stop warrants under either the old
or
new policy, were placed
in Group B.
The study consisted
of
analyses of accidents and field per-
formance. The accident analysis involved
19
intersections,
12
from Group A and 7 from Group B. The field analysis used
15
intersections, 8 from Group A and 7 from Group B. All
23
of
the intersections were included in at least one of the
analyses.
The first analysis, a comparison of the number of accidents
12
months before and after the all-way stops were installed,
showed
that
the intersections in Group A experienced a
sig-
nificant reduction.
In
contrast, the intersections
in
Group B
did not experience a significant change
in
accidents; in fact,
the number of accidents rose slightly. Figures 2 and 3 show
the results
of
the before-and-after accident comparison. For
Group
A,
the reduction in accidents that occurred at the
intersections was found to be statistically significant at the
99
percent confidence level.
For
all accidents at or near the inter-
sections (midblock accidents are assigned to the nearest inter-
section), the decrease was also significant at the
99
percent
confidence level.
The field analysis also gave interesting results. Group A
had an average volume ratio of major street to minor street
of 1.8, whereas the ratio for Group B was 4.0,
as
shown in
Figure
4.
These data support the idea that all-way stops func-
tion best when the cross-street volumes are nearly equal. A
key finding was that Group B had a higher frequency of major
street motorists failing to stop,
as
shown in Figure
5.
In Group
A, 6.8 percent of the motorists on the major street failed to
stop, whereas in Group B, 13.0 percent failed to stop. The
difference between the two groups was found to be statistically
90
80
"'
70
.c
E
60
0
E
N
50
-
.5
!!!
40
c
..
't>
30
'()
()
C(
20
10
0
File
Intersection
Date
(MAJOR)
(MINOR)
Investigator
Qualifies for All-Way Stop based
on
25
or more points:
Yes
__
No
__
Points
__
Qualifies for All-Way Stop based
on
other criteria:
Yes
__
No_
If
yes, explain:
Sketch
of intersection with visibility data
On
back_
Attached_
1. Accident Experience
Points
Possible
From
L l to L
I
Accidents/year correctable by Stops x 3 points/accident
2.
Unusual Conditions
3. Traffic
Volumes (Peak 4 Hours)
Major approaches
Minor approaches
4. Traffic
Volume Difference
5. Pedestrian
Volume
Pedestrians
crossing the major street during 4 hour count
FIGURE 1 All-way stop evaluation worksheet.
Intersection
Accidents
Before/After
Non-intersection
Accidents
Before/After
77
Total
Accidents
Before/After
TOTAL
Points Required
90
80
"'
70
.c
E
60
0
E
N
50
-
.5
"'
40
E
..
't>
30
c;
()
C(
20
10
0
Intersection
Accidents
Before/After
15
5
5
10
10
5
50
25
Non-intersection
Accidents
Before/After
32
Total
Accidents
Before/After
FIGURE 2 Before-and-after accident comparison (Group
A).
FIGURE 3 Before-and-after accident comparison (Group
B).
Celniker
~
f!
CD
E
:I
0
>
a;
~
..
~
c
'E
~
0
"iij"
:E
5:
1
4:1
3 : 1
2:
1
1
; 1
Group A
average
4.0 : 1
Group B
average
FIGURE 4 Volume ratio comparison (Group A versus
Group
B).
significant at the
95
percent confidence level. These figures
indicate
that
the new policy
is
successful at selecting inter-
sections where all-way stop controls
will
earn motorists' respect
and have a
better
rate of stop sign compliance.
The
results
of
the statistical analyses are shown below:
Note:
Only those tests that showed statistical significance are
shown.
1.
Accidents at intersection:
53
in
12
months before all-
way stop was installed;
13
in
12
months after.
• Calculated t = 3.028, d.f. = 22;
•Tabulated
t (at
99
percent confidence) = 2.819;
• Therefore the difference
is
significant at the
99
percent
confidence level.
2. Total
of
accidents at and near intersection:
77
in
12
months before all-way stop
was
installed,
28
in
12
months
after.
• Calculated t = 2.865, d.f. = 22;
•Tabulated
t
(at
99
percent confidence) = 2.819;
• Therefore the difference
is
significant at the
99
percent
level.
3.
Percent
of
vehicles on major street failing to stop: Group
A-6.8
percent, Group
B-13.0
percent.
• Calculated z = 2.334, d.f. = 13;
•Tabulated
z (at
99
percent confidence) = 3.012;
•Tabulated
z (at
95
percent confidence) = 2.160;
•
Therefore
the difference
is
significant at the
95
percent
confidence level.
CONCLUSION
The
new policy meets all of the goals for a model all-way stop
policy.
The
policy
is
consistent with traffic engineering prin-
14
%
12
%
10
%
CL
B
..
B
8%
Cl)
:§
~
6%
E
CD
!:!
CD
c..
4%
2%
0%
Group A
Group B
average
average
FIGURE 5 Failure-to-stop comparison (Group A versus
Group
B).
21
ciples,
is
not dependent on traffic signal warrants,
is
flexible
for use in differing conditions, and
is
successful at selecting
intersections that benefit from the installation of all-way stops.
It
will give traffic engineers confidence
in
the all-way stop
warrants when discussing the issue with citizens' groups and
elected officials. The policy will assist traffic engineers in their
mission of educating the public about traffic safety and pro-
viding the public with safe streets and efficient traffic flow.
ACKNOWLEDGMENTS
The author thanks his colleagues in the Transportation Plan-
ning and Traffic Engineering Divisions who contributed to
the content and quality
of
this paper. Special thanks go to
Linda Wilson for typing the text and to Frank Yates for pro-
ducing the graphics.
REFERENCES
1.
Manual on Uniform Traffic Control Devices for Streets and High-
ways.
FHWA, U.S. Department of Transportation,
1978.
2.
Special Report 209: Highway Capacity Manual. TRB, National
Research Council, Washington, D.C.,
1985.
DISCUSSION
K.TODD
1954 Columbia Rd.,
N.
W., No.
707,
Washington, D.C. 20009.
The
paper by Celniker
is
a welcome departure from the
MUTCD
multiway stop warrants, which have been criticized
elsewhere (1,2) for their lack of scientific validity.
This comment deals with Celniker's statement that safety
is
jeopardized when drivers disobey a stop sign. When a sign
imposes a needless stop, it fails to meet two basic require-
22
ments for a traffic control device to be effective: it does not
fulfill a need and it does not command respect. In Dyar's
study,
(3)
88
percent of all motorists disregarded stop signs
in
light traffic and treated them
as
yield signs when there was
no one to stop
for-clear
evidence of an overly restrictive
control
(4).
The compulsory stop regardless of traffic conditions should
not only be justified by evidence showing that the failure to
stop
per
se (rather than the failure to yield) contributes to
collisions, but also that the cost of these collisions outweighs
the cost of the additional delay, fuel consumption, and air
pollution. Without such proof, the unconditional stop
is
not
warranted (5).
To
first maintain
that
needless stops should be avoided in
the interest of safety, efficiency, and respect for traffic con-
trols, and then claim
that
the failure to come to a peremptory
but needless stop jeopardizes safety,
is
a contradiction the
traffic engineering profession has yet to explain.
The
logical
way out of this contradiction
is
the all-way yield, a technique
capable of competing with traffic signal control in terms of
costs to the road user and highway agency (6).
REFERENCES
1.
P.
S.
Shapiro, J. E. Upchurch, J. Loewen, and
V.
Siaurusaitis.
Identification of Needed Traffic Control Device Research. In
Transportation Research Record 1114, TRB, National Research
Council 1987, Washington,
D.C.,
pp. 11-20.
2.
E.
Hau~r.
A Case for Science-Based Road Safety Design and
Management.
Presented at 68th Annual Meeting
of
the Trans-
portation Research Board, Washington,
D.C.,
January 1989.
3.
R. D. Dyar.
An
Analysis
of
Driver Observance
of
Stop Signs. M.S.
Thesis, Clemson University, Clemson, S.C.,
1977.
.
4.
H.
S.
Lum and
W.R.
Stockton. STOP Sign Versus YIELD Sign.
In
Tra11sportatio11
Research Record 881, TRB, National Research
Council, Washington,
D.C.,
1982, pp. 29-33.
5.
K.
Todd. Traffic Regulation and Public Policy. Transportation
Quarterly,
Oct. 1987.
6.
H.
Gurney. New Filter System for Guernsey
Road
Traffic. Guern-
sey Traffic Department, Guernsey,
United Kingdom, May 1972.
TRANSPORTATION
RESEARCH
RECORD
1244
AUTHOR'S CLOSURE
The purpose
of
the new all-way stop policy
is
to balance the
public's request for all-way stops with traffic engineering prin-
ciples of safety and operations. The new policy
is
positive for
the following reasons:
• The all-way stop
is
an existing, familiar traffic control
device.
• The new policy
is
flexible to a variety of factors, yet it
allows only all-way stops to be installed at intersections where
they
will
function well.
• The concept
of
avoiding unnecessary stop signs
is
con-
sistent with, not contradictory to, the statement that a failure
to stop jeopardizes safety. The policy's goal
is
to install all-
way stops only where they will have a high rate of compliance.
The
"all-way yield" proposal
is
a deeply flawed alternative.
Traditionally, a yield sign says to motorists
"yield the right-
of-way to cross traffic
by
either stopping or slowing down;
then, when there are no vehicle conflicts go
ahead." This
message
is
very useful and successful in cases of low-volume
intersections or channelized right-turn lanes. The yield signs
face only the direction of traffic that yields.
The proposed
"all-way yield" changes the message of the
yield sign to
"slow down, a complete stop
is
not necessary;
yield the right-of-way to cross traffic
as
you would at an all-
way stop or an uncontrolled intersection, then go
ahead."
The all-way yield
is
a basic contradiction in terms, poten-
tially dangerous, and unnecessary. Motorists will be confused
about the new use of a familiar sign, and such confusion may
lead to accidents. Also, the successful, traditional
use of the
yield sign will be lost if yield signs take
on
a new meaning.
Publication
of
this paper sponsored
by
Committee on Traffic Control
Devices.
