Institute of Transportation Studies
UC Berkeley Traffic Safety Center
( University of California, Berkeley)
Year 2007 Paper UCB - TSC - TR - 2007 - 7
Evaluation of Countermeasures: A Study
on the Effect of Impactable Yield Signs
Installed at Intersections in San Francisco
Ipsita Banerjee  David R. Ragland†
 UC Berkeley Department of Civil and Environmental Engineering
† UC Berkeley Traffic Safety Center
This paper is posted at the eScholarship Repository, University of California.
http:// repositories. cdlib. org/ its/ tsc/ UCB- TSC- TR- 2007- 7
Copyright  c 2007 by the authors.
Evaluation of Countermeasures: A Study
on the Effect of Impactable Yield Signs
Installed at Intersections in San Francisco
Abstract
The present study evaluated the effect of impactable signs that used the
yield- symbol as approved by the National Committee on Uniform Traffic Con-trol
Devices ( NCUTCD) in the 2003 Manual of Uniform Traffic Control Devices
( MUTCD). Impactable yield signs are low- costsignsconstructed of flexible ma-terial.
The signs wereinstalled in the medians adjacent to crosswalks at selected
non- signalized intersections to instruct drivers to yield the right- of- way to pedes-trians.
This paperexamines the effect on safety characteristics of the intersec-tions
of these signs at three stop- sign controlled intersections in San Francisco
over two follow up periods. Since these signs were installed recently, there were
no post- installation crash data for comparison with the pre- installation crash
data. As such, surrogate measures, including ( a) driver yielding behavior, ( b)
conflicts among drivers and pedestrians crossing the intersection, ( c) waiting
time for pedestrians, and ( d) time taken by pedestrians to cross a given cross-walk
were documented. Previous studies have indicated that impactable yield
signs are effective in increasing the rate of drivers yielding to pedestrians. Video
recordings were taken at the intersection pre- and post- installation to observe
any changes in behavior. Analyses of these recordings yielded data for baselin-eand
the first and second follow- up periods respectively. Testing the first and
second follow- up data against the baseline data reveal that, a substantial in-creasein
yielding behavior by drivers occurred immediately after installationas
well as during the second follow- up period. No significant effect was observed
in any other variables.
Banerjee & Ragland 1
Evaluation of Countermeasures: A Study on the Effect of Impactable Yield Signs Installed
at Four Intersections in San Francisco
Submission Date: November 15, 2006
Word Count: 7094 ( including Tables and Figures)
Tables and Figures: 10
Ipsita Banerjee
University of California
Department of Civil and Environmental Engineering
116 McLaughlin Hall
Berkeley, CA 94720
Phone: 510- 642- 3585/ Fax: 510- 643- 3955
ipsita@ berkeley. edu
David R. Ragland ( Corresponding author)
University of California Traffic Safety Center
140 Warren Hall # 7360
Berkeley, CA 94720
Phone: 510- 642- 0655/ Fax: 510- 643- 9922
davidr@ berkeley. edu
TRB 2007 Annual Meeting CD- ROM Paper revised from original submittal.
Banerjee & Ragland 2
ABSTRACT
The present study evaluated the effect of impactable signs that used the yield- symbol as
approved by the National Committee on Uniform Traffic Control Devices ( NCUTCD) in the
2003 Manual of Uniform Traffic Control Devices ( MUTCD). Impactable yield signs are low-cost
signs constructed of flexible material. The signs were installed in the medians adjacent to
crosswalks at selected non- signalized intersections to instruct drivers to yield the right- of- way to
pedestrians. This paper examines the effect on safety characteristics of the intersections of these
signs at three stop- sign controlled intersections in San Francisco over two follow up periods.
Since these signs were installed recently, there were no post- installation crash data for
comparison with the pre- installation crash data. As such, surrogate measures, including ( a) driver
yielding behavior, ( b) conflicts among drivers and pedestrians crossing the intersection, ( c)
waiting time for pedestrians, and ( d) time taken by pedestrians to cross a given crosswalk were
documented. Previous studies have indicated that impactable yield signs are effective in
increasing the rate of drivers yielding to pedestrians. Video recordings were taken at the
intersection pre- and post- installation to observe any changes in behavior. Analyses of these
recordings yielded data for baseline and the first and second follow- up periods respectively.
Testing the first and second follow- up data against the baseline data reveal that, a substantial
increase in yielding behavior by drivers occurred immediately after installation as well as during
the second follow- up period. No significant effect was observed in any other variables.
TRB 2007 Annual Meeting CD- ROM Paper revised from original submittal.
Banerjee & Ragland 3
INTRODUCTION
The city of San Francisco has recorded consistent high rates of pedestrian fatality for several
years ( 1), ( 2), ( 3), ( 4). In 2004, 710 pedestrian injuries and 20 pedestrian fatalities were recorded
in San Francisco and constituted the second highest rates of pedestrian injuries and third highest
fatality rates for California ( 5). Several projects are being undertaken to improve safety for
pedestrians in San Francisco ( 6), ( 7). At the national level, the Federal Highway Administration
( FHWA) has sponsored studies to improve pedestrian safety and evaluate new technologies ( 8).
Toward this end, the FHWA has funded the PedSafe project in San Francisco, Las Vegas, and
Miami. In San Francisco, the PedSafe Project is a joint endeavor between the Traffic Safety
Center at the University of California, Berkeley, and the San Francisco Department of Parking
and Traffic ( SFDPT). The study was conducted in two phases. Phase I included: ( a) conducting a
zone analysis to determine seven of the highest pedestrian injury- prone zones in San Francisco;
( b) conducting a crash analysis utilizing the Pedestrian and Bicycle Crash Analysis Tool
( PBCAT); ( c) conducting observations at intersections within the study zones; ( d) reviewing both
traditional and “ intelligent” countermeasures; and ( e) developing recommendations and a
countermeasure plan for Phase II of the project.
As part of Phase II, impactable yield signs were installed at or near selected intersections
as one of the countermeasures recommended in Phase I. The current study outlines a
methodology to evaluate the effectiveness and efficiency of these signs.
It is common practice to evaluate the effectiveness of a countermeasure in preventing
crashes by using vehicle crash data at the location of a countermeasure. After crash statistics are
obtained for a sufficient period of time for the data to be stable, the rate of crashes before and
after the installation can be compared. The impactable yield signs being studied were installed in
November, 2005. Hence the efficacy of these countermeasures using crash data cannot yet be
evaluated because of insufficient post- installation crash data. Therefore, the preliminary analysis
was performed using surrogate measures used in previous studies as a proxies for crash data ( 9),
( 10), ( 11), ( 12), ( 13), ( 14). We expected the yield signs to primarily increase the number of times
drivers would yield the right- of- way to pedestrians. Data samples of the surrogate measures
before and after installations at various intersections were collected and the difference tested for
statistical significance. Based on the results of these tests conclusions on the effect of these signs
have been drawn.
The paper includes a description of the impactable yield signs followed by a review of
previous studies on the effect of these signs. The locations of signs in San Francisco are listed,
and the salient characteristics of the intersections under study are discussed. The design of the
experiment and their measures of effectiveness are followed by analysis of the effect of the
countermeasures. The paper concludes with a summary of the study and a discussion of the
results.
DESCRIPTION & BACKGROUND
Impactable yield signs, also referred to as “ in- roadway knockdown signs” are low height signs
that are located in the median of intersections or midblock locations, usually adjacent to the
crosswalk. These signs instruct drivers to yield to pedestrians in the crosswalk. Figure 1 gives a
picture of the impactable yield sign used in the study. Made from flexible material, these signs
TRB 2007 Annual Meeting CD- ROM Paper revised from original submittal.
Banerjee & Ragland 4
return to their original position after being hit by a motorist. The background literature discusses
several studies dealing with surrogate measures and different types of yield- to- pedestrian signs.
Madison Yield to Pedestrian Signs
A study conducted in Madison, Wisconsin ( 9) examined in- roadway yield to pedestrian signs.
Although not made of flexible material, their effect is comparable to impactable yield signs since
their position and height are similar to that of impactable yield signs. The Madison study lays out
detailed criteria for defining a yield. Only instances of interactions between motorists and
pedestrians are considered as the denominator in calculating the percent of yields. The post
installation data was collected 30 or more days after installation. The study observed from 0 to
10 percent of motorists yielding to pedestrians before the signs were installed compared to 10 to
20 percent after installation. The study concludes that the signs led to an increased rate of
motorists yielding to pedestrians, except in the case where the sign was placed on a wide raised
median, the width of which moved it farther from the vehicular lanes and possibly reduced its
impact.
FIGURE 1 Impactable yield sign ( 8)
FHWA Traffic Cones
The New York Pedestrian Safety Cones Study ( 10), evaluated devices consisting of a traffic cone
fitted with an orange, retro- reflective safety jacket bearing the sign, “ State Law – Yield to
Pedestrians in Your Half of the Road” ( 10). The study notes that the signs are made of flexible
material and are more ‘ forgiving’ when struck by a vehicle compared to similar signs installed
on metal posts. These signs also do not damage the vehicle or become projectiles to pedestrians.
Three measures of effectiveness were considered: pedestrians for whom motorists yielded;
pedestrians who ran, aborted or hesitated; and pedestrians who crossed in the crosswalk. The
post installation data were collected within a month of installation. While seven intersections
were studied, only six had adequate sample size for recording a significant change in the percent
of pedestrian for whom motorists yielded. Four intersections out of the six, demonstrated a
significant change in the percent of pedestrian for whom motorists yielded from 62 to 81 percent.
TRB 2007 Annual Meeting CD- ROM Paper revised from original submittal.
Banerjee & Ragland 5
However, only two of the seven intersections with a significant sample size recorded a
significant decrease in the percent of pedestrians who ran, aborted or hesitated, and, in contrast,
one intersection experienced an increase in the rate. Similarly, for pedestrians who crossed
within the crosswalk, only two out of the seven intersections noted a significant increase in the
percent. The effect of motorist yielding had less effect on pedestrian running behavior than
might have been expected. Also, the signs did not significantly increase use of the crosswalk
since there are always some pedestrians who do not cross at the crosswalk. The study concluded
that the cones, while being less expensive compared to the overhead and other signs considered
in the study, were an effective device for increasing driver yielding behavior.
Nova Scotia, Canada: In- Street Pedestrian Yield Signs
In Nova Scotia, 24 locations were studied that had pedestrian- activated flashing beacons and
advance- yield markings on the pavement ( 11). The sites involved one- way and two- way
operations and urban and rural sites. The measures of effectiveness considered were: reduction in
‘ erratic’ behavior by drivers and pedestrians; distance from the crosswalk that vehicles stopped;
and percent of drivers stopping for pedestrians. Drivers stopping six meters or more before the
crosswalk increased from 13 percent to 54 percent, and the percent stopping three meters or more
before the crosswalk increased from 37 percent to 83 percent. These increases were statistically
significant.
Michigan State University: Portable Signs
Portable yield- to- pedestrian signs were used at Michigan State University campus with new
crosswalk markings ( 12). Used during daytime hours, they increased the number of drivers who
yielded, although drivers reverted to previous habits when the signs were moved away. Based on
the positive results, the researchers decided to double the number of applications next term and
use a longer term, fixed- base structure.
Iowa Yield- to- Pedestrian Signs
Impactable yield signs were installed in three sites in Cedar Rapids, Iowa ( 13). Three measures
of effectiveness were used including: ( a) vehicle speed changes at intersections during periods in
which pedestrians were present, ( b) percentage of ‘ first vehicles’ that stopped when a pedestrian
stepped from the curb, and ( c) percentage of aborted or hurried crossings by pedestrians. The
post- installation data were collected four months after installation of the countermeasures.
Researchers concluded that the signs had a positive although minimal effect on pedestrian
behavior. Results were not uniform. The improvements observed were speed reduction by
vehicles at one site and increased driver compliance at the other. Also, at one of the locations,
where there were competing demands for attention ( i. e., railroad crossing, bike path, and
intersection), the change in driver behavior was not marked, indicating decreased compliance
with the sign once one’s own safety was jeopardized. The low pedestrian volume at two of the
three sites limited the ability to measure changes in driver/ pedestrian behavior. The study
observed that with larger sample sizes, small changes in vehicle speeds may be statistically
significant, but a small change will not necessarily lead to an overall increase in pedestrian
safety. This study further determined that positioning the signs farther from the crosswalk will
cause less physical damage to the signs. Applications with center medians or wider turning or
storage bays are best candidate locations.
TRB 2007 Annual Meeting CD- ROM Paper revised from original submittal.
Banerjee & Ragland 6
These background studies all indicate that signs have been effective in increasing the rate
of yields by vehicles to pedestrians. Some have recorded the improvement in more details,
segregating between drivers that stopped at different distances from the crosswalk. In the current
study, researchers focused on changes in behavior associated with yield signs over time and
recorded observations at two different periods after their installation.
STUDY OBJECTIVE & SCOPE
Of the studies discussed, only the Nova Scotia study used a symbol sign similar to that used in
the current study. The symbol sign was later adopted by the National Committee on Uniform
Traffic Control Devices ( NCUTCD) in the 2003 Manual of Uniform Traffic Control Devices
( MUTCD). The present study aims to evaluate the effectiveness of this symbol sign. It further
aims to illustrate the evolution of user behavior over time by observing the change in the
surrogate measures of effectiveness over two time periods: the first period was one month and
the second period was three to four months after installation of the signs.
Since selection of intersection for the study is not random, bias in results may have
resulted, and extent of improvement may not be generalized for impactable yield signs in
general, but hold for the specific intersections only.
TREATMENT INTERSECTIONS: LOCATIONS & CHARACTERISTICS
Four sites at three intersections were selected for observing the effect of impactable yield signs.
These include the intersections of 16th Street and Capp Street, Mission Street and France
Avenue, and, Mission Street and Admiral Avenue. Two of four study sites were located at a
marked and an unmarked crosswalk of the same intersection, at 16th and Capp. The intersection
of 16th and Capp is located in the heart of San Francisco, while Mission and France, and, Mission
and Admiral are located more towards the south of the city. The locations of the treatment
intersections are indicated in Figure 2.
These treatment intersections are medium- sized low- speed intersections, located in
institutional, commercial or industrial areas. Street parking is present at all intersections. Two
intersections are four legged of which Mission and Admiral is a skewed intersection; Mission
and France is a T- intersection. All intersections are stop controlled and have two- way flow. The
intersection of 16th and Capp, being located in the heart of San Francisco, had observed the
maximum number of crashes since June 2000. Table 1 outlines the salient characteristics of the
intersections.
TRB 2007 Annual Meeting CD- ROM Paper revised from original submittal.
Banerjee & Ragland 7
TABLE 1 Characteristics of the Site Intersections ( 14), ( 15), ( 16)
Intersection
Characteristics
16th & Capp Mission & France Mission & Admiral
Location of Study
Crosswalk
Crossing 16th both on
the eastside ( unmarked)
and Westside ( marked)
of intersection
Crossing Mission on
the south side of
intersection
Crossing Mission on
the south side of
intersection
Type of Intersection 4- legged 3- legged 4- legged
Number of Lanes ( 1st
street/ 2nd street)
3/ 2 4/ 2 4/ 2
Flow ( 1st street/ 2nd
street)
2 way/ 2 way 2 way/ 2 way 2 way/ 2 way
Type of Control STOP sign on Capp STOP sign on France
STOP sign on
Admiral
Adjacent landuse Institutional/ Commercial Commercial Industrial
Street Parking Yes Yes Yes
Marked crosswalk 3 1 2
Speed ( miles per
hour)
25 25 25
Total number of
Crashes ( June 2000-
June 2005)
10 4 7
Notes
Admiral is staggered
across intersection
TRB 2007 Annual Meeting CD- ROM Paper revised from original submittal.
Banerjee & Ragland 8
FIGURE 2 Locations of intersections
STUDY DESIGN
The evaluation study was designed as a pre- post design. Video recordings of the intersections
were made before the installations ( baseline data) and during two periods after installations. The
purpose of the second follow- up survey was to ascertain the effect of the signs after the novelty
effect wore out. The surveys were conducted during the same hours on every weekday, between
1 p. m. to 3 p. m. and between 4 p. m. to 6 p. m. The various dates of survey are indicated in Table
2.
TRB 2007 Annual Meeting CD- ROM Paper revised from original submittal.
Banerjee & Ragland 9
TABLE 2 Date of Survey at the Study Sites ( 17)
Site Intersection
Approach
Countermeasure Baseline
Survey
Date of
counter-measures
1st follow
up
2nd follow
up
16th and
Capp
On 16th
eastside
crosswalk
2 sided YTP sign 6/ 20/ 05 11/ 09/ 05 12/ 12/ 05 2/ 28/ 06
16th and
Capp
On 16th
Westside
crosswalk
2 sided YTP sign 6/ 28/ 05 11/ 09/ 05 12/ 13/ 05 3/ 1/ 06
Mission
and
France
Southside
crosswalk
on Mission
2 sided YTP sign 6/ 15/ 05 11/ 5/ 05 12/ 5/ 05 3/ 2/ 06
Mission
and
Admiral
Southside
crosswalk
on Mission
2 sided YTP sign
With painted
island
7/ 21/ 05 11/ 6/ 05 11/ 28/ 05
&
11/ 29/ 05
3/ 3/ 06
MEASURES OF EFFECTIVENESS
As part of the PedSafe project, a study plan was developed that outlined the procedure for
evaluating many different countermeasures using surrogate variables ( 18). The current study
examined 14 different measures of effectiveness, primary among which were: frequency of
pedestrian- vehicle conflicts, percentage of drivers yielding to pedestrians, and percentage of
pedestrians trapped in the roadway.
The measures of effectiveness chosen to test the impactable yield signs were selected
based on hypotheses developed for the study plan. Table 3 lists the hypotheses on different
aspects and the possible measures of effectiveness for impactable yield signs.
TABLE 3 Measures of Effectiveness ( MOEs) of impactable yield signs ( 18)
Aspect Hypothesis MOEs
Reduce vehicle- pedestrian conflicts Frequency of vehicle-pedestrian
conflicts
Increase driver yield to pedestrians Percent of drivers yielding to
pedestrians
Allow pedestrians to clear crosswalk on time Number of pedestrians trapped
in the roadway
Safety
surrogates
Increase number of pedestrian crossing
within designated crosswalk
Percent of captured pedestrian
crossings
Pedestrian
Mobility
Reduce delay to pedestrians Pedestrian delay
For the purpose of this study, terms were defined as follows ( 19):
• Conflict: A conflict involved an evasive action by a motorist or a pedestrian, where the vehicle
and pedestrian were on a collision course. Evasive action was evidenced by a motorist
stopping, slamming on the brakes, or swerving, or by a pedestrian suddenly stepping back,
TRB 2007 Annual Meeting CD- ROM Paper revised from original submittal.
Banerjee & Ragland 10
lunging back, or running forward to avoid being struck by a vehicle. For a conflict to be
scored, evasive action ( by either the motorists or the pedestrian) had to be observed.
• Vehicle Yield: A vehicle yield was recorded when a driver yielded to a pedestrian by stopping
or slowing down to let the pedestrian cross in front of the car.
• Trapped in the Roadway: This was recorded when a pedestrian was stuck in the roadway ( in
a lane, on a lane line, or on the center line) when traffic was too close and did not yield. This
generally resulted from a pedestrian selecting a gap in the traffic that was too short for them to
completely cross the road before encountering oncoming vehicles.
• Captured pedestrians: This was defined as the percentage of crossings in which the
pedestrian was in the crosswalk.
• Pedestrian Delay/ Wait: This was defined as the interval between the time at which a
pedestrian arrives at an intersection and the time when she starts crossing.
ANALYSIS
We studied selected measures of effectiveness ( e. g., car yielding, conflicts etc.) using the video
recordings taken before and after installation of the signs. Baseline measures were compared
with each set of follow- up data to record differences between the two. Using two- tailed tests, p-values
were obtained for each of these differences to test for statistical significance. This analysis
was based on the assumption that a significant change in the baseline to first follow up value was
due to the initial effect of the sign. In some cases the initial effect may have been manifested
through increased compliance to the sign message. In others it may cause undesirable results due
to general user confusion. A statistically significant change between the baseline and the second
set of follow up data was interpreted as the actual effect of the sign; i. e. the effect that the sign
message had on the majority of regular users of the intersection.
For calculating the pedestrian composition, the percentage of pedestrians who look at
start and at midpoint, the percent of trips in the crosswalk, and, average wait and crossing time,
the total number of pedestrians was considered as the denominator. For calculating the percent of
car yield, percent of conflicts and percent of pedestrian trapped in the crosswalk whereas, only
the number of pedestrian- vehicle interactions was considered as the denominator.
RESULTS
Tables 4 through 7 represent the observations for the different measures of effectiveness
for the four study intersections.
Observations at 16th & Capp ( unmarked)
Table 4, representing the unmarked crosswalk at 16th and Capp, shows an increase in the percent
of adults and males, although the percentage increase in males is not statistically significant at
the 95 percent confidence level at the time of at the first follow up survey.
The wait time reduced significantly at the first follow- up period and was marginally
reduced compared to baseline at the second follow- up period. The crossing time increased
significantly during both follow- up periods, possibly brought about by inaccurate expectations of
the pedestrian. The marginally increased number of pedestrians trapped in the first follow- up
period, although the difference by the second follow- up was not significant, also supports the
inaccurate user expectation hypothesis. There was a statistically significant increase in the
percent of trips in the crosswalk. The increase in percent car yield though not significant at the
time of the first follow up survey became significantly different from baseline by the second
TRB 2007 Annual Meeting CD- ROM Paper revised from original submittal.
Banerjee & Ragland 11
follow- up period, that being the desired outcome due to the signs. Although there was a change
in the percent of males and adults, this did not substantially alter impact the main results of this
study.
Observations at 16th & Capp ( marked)
Table 5 lists the observations at the marked intersection of 16th and Capp. Over time there was
no significant change in the pedestrian composition and as such it did not affect the results.
Although the cross time did not change at first, there was a significant increase later. The percent
car yield increased significantly. While there was no initial significant change in the percent
trapped, there was a significant decrease over time, possibly a positive manifestation of the
increased car yield. Both the baseline and post- installation percentages of vehicular yields were
observed to be higher for the marked crosswalk as compared to the unmarked crosswalk.
Observations at Mission & France
In Table 6, no significant change in pedestrian composition was observed in Mission and France,
except that the percentage of adults increased significantly at first. As such, pedestrian
composition possibly had no bearing on the results. The wait time reduced significantly by the
second follow up period. The cross time also improved similarly. However, the percent of trips
in the crosswalk reduced significantly at the time of the second follow up too. As in the other
cases, there was a significant increase in the percent of car yields both initially and later..
Observations at Mission & Admiral
In Table 7, at Mission and Admiral, the pedestrian composition did not change significantly over
time except for an initial significant increase in the percentage of adults which could not have
influenced the results. The wait time remained fairly constant. Crossing time increased initially
but was not significantly longer by the second period. Car yield was significantly higher and
conflicts were significantly lower both initially and during the second follow- up period.
The common observations across all intersections are summarized in the next section.
TABLE 4 Observations at 16th & Capp ( unmarked)
16TH/ CAPP UNMARKED Baseline
1st
Follow-up
p- value
comparing
outcome to
baseline
2nd
Follow-
Up
p- value
comparing
outcome to
baseline
Total Pedestrians 86 112 145
Gender ( Percent male) 70.2 79.5 0.07 83.9 0.01
Age( Percent adult) 77.9 92.9 0.02 87.5 0.02
Average Wait ( seconds) 2.6 ( 4.4)* 1.7 ( 2.1) 0.04 1.8 ( 2.7) 0.06
Average Cross ( seconds) 9.2 ( 4.6) 11 ( 4.4) 0.00 10.4 ( 3.0) 0.02
Percent of trip in crosswalk 36 63.8 0.00 58.1 0.00
Pedestrian/ Vehicle
Interactions 96 120 109
Percent vehicular yield 39.6 48.3 0.10 59.6 0.00
Percent conflict 6.3 6.7 0.45 2.8 0.12
Percent trapped 8.3 15 0.06 11 0.26
TRB 2007 Annual Meeting CD- ROM Paper revised from original submittal.
Banerjee & Ragland 12
* The figures in parentheses show the standard deviation
TABLE 5 Observations at 16th & Capp ( marked)
16TH/ CAPP MARKED Baseline
1st
Follow-up
p- value
comparing
outcome to
baseline
2nd
Follow-
Up
p- value
comparing
outcome to
baseline
Total Pedestrians 368 405 364
Gender ( Percent male) 66.3 64.9 0.34 69.5 0.18
Age( Percent adult) 92.1 85.7 0.00 91.2 0.33
Average Wait ( seconds) 4.6 ( 6.3) 4.3 ( 5.5) 0.24 4.7 ( 8.5) 0.43
Average Cross ( seconds) 10.9 ( 2.8) 10.9 ( 2.9) 0.50 11.5 ( 3.4) 0.00
Percent of trip in crosswalk 89.4 90.2 0.36 91.3 0.19
Pedestrain/ Vehicle
Interactions 519 488 447
Percent vehicular yield 60.5 70.7 0.00 73.6 0.00
Percent conflict 6.9 8.8 0.13 6.7 0.45
Percent trapped 3.3 3.7 0.37 1.3 0.02
TABLE 6 Observations at Mission & France
MISSION/ FRANCE Baseline
1st
Follow-up
p- value
comparing
outcome to
baseline
2nd
Follow-
Up
p- value
comparing
outcome to
baseline
Total Pedestrians 126 100 115
Gender ( Percent male) 58.1 62.1 0.27 57 0.43
Age( Percent adult) 82.1 82.3 0.48 80.9 0.41
Average Wait ( seconds) 7.9 ( 9.2) 7.5 0.39 5.5( 8.4) 0.02
Average Cross ( seconds) 12.7 ( 2.7) 13.1 0.18 12.1 ( 2.7) 0.04
Percent of trip in crosswalk 98.7 98.3 0.40 93.8 0.02
Pedestrian/ Vehicle
Interactions 195 144 126
Percent car yield 36.9 61.1 0.00 61.9 0.00
Percent conflict 0 0 0.50 0 0.50
Percent trapped 0 0 0.50 0 0.50
TRB 2007 Annual Meeting CD- ROM Paper revised from original submittal.
Banerjee & Ragland 13
TABLE 7 Observations at Mission & Admiral
MISSION/ ADMIRAL Baseline
1st
Follow-up
p- value
comparing
outcome to
baseline
2nd
Follow-up
p- value
comparing
outcome to
baseline
Total Pedestrians 28 40 37
Gender ( Percent male) 74.1 56.8 0.06 82.1 0.22
Age( Percent adult) 78.6 97.5 0.01 91.9 0.07
Average Wait ( seconds) 6.2 ( 5.9) 7.5 ( 7.6) 0.22 8.2 ( 7.8) 0.12
Average Cross ( seconds)
10.2
( 2.3)
12.1
( 4.5) 0.01
10.8
( 2.9) 0.18
Percent of trip in crosswalk 82.7 84.9 0.40 94.7 0.07
Pedestrian/ Vehicle
Interactions 45 70 41
Percent car yield 20.0 52.9 0.00 70.7 0.00
Percent conflict 26.7 12.9 0.04 7.3 0.01
Percent trapped 8.9 5.7 0.26 4.9 0.23
RESULTS
A large majority of males ( 56.8 percent to 83.9 percent) and of adults ( 77.9 percent to 92.9
percent) constitute the pedestrian at all four intersections. The only significant result that was
common to all four sites was the increase in the rate of car yields as shown in Figure 3. None of
the other variables indicated significant changes across all four intersections. All increases from
baseline data, except for the increase at the 16th and Capp unmarked crosswalk for the first
follow up survey were significant at the 99.5% confidence level. The increase in car yields
followed different rates at the different intersections. While at Mission and Admiral, there were
large increases both at first and second follow up survey, at Mission and France, there was a
large increase at the time of the first follow up survey and not much increase after that. As such,
this study concludes that the impactable yield signs definitely caused vehicles to yield more to
pedestrians.
TRB 2007 Annual Meeting CD- ROM Paper revised from original submittal.
Banerjee & Ragland 14
0
10
20
30
40
50
60
70
80
Unmarked 16th & Capp Marked 16th & Capp Mission/ France Mission/ Admiral
Intersections
Percentage
Baseline
First follow up
Second follow up
FIGURE 3 Increases in the percentage of car yields
The increase in the rate of car yields caused changes in other measures of effectiveness. Some of
these changes were positive for instance, reduction in the number of conflicts, cross time or wait
time. Some changes were detrimental to safety, for instance reduced percent of trips in the
crosswalk and increased cross time.
.
DISCUSSION
The study examined the latest approved version of a low- cost symbol sign that was associated
with a significant increase in driver yielding behavior. Pedestrian composition was observed
generally to be random and uncorrelated with the results. The outcome of significant increase in
car yield is consistent with all other studies conducted on yield– to- pedestrian signs of different
types. Being placed in the median adjacent to the crosswalk makes the sign more visible and
provides an alert at a very appropriate time, a possible reason behind its success.
However its location in the median makes it prone to collision and damage. The flexible
material and installation aims to counter the damage. With increasing experience, engineers are
now capable of locating the signs in an equally visible but less collision- prone location, by
making slight alterations in its placement, so as not to be in the possible path of a vehicle. Yet
the signs do sometime need replacement due to wear and vandalism. For instance, it was
TRB 2007 Annual Meeting CD- ROM Paper revised from original submittal.
Banerjee & Ragland 15
reported that a couple of weeks after installation, the sign at Mission and France lost its ability to
return to vertical position. Such instances are also corroborated by the other studies, one of
which mentions that the San Francisco traffic has not been kind to these signs ( 20). At the same
time, studies have noted pedestrians to appreciate these signs and make requests for re-installation
when the signs were removed due to adverse weather conditions ( 9).
As with many similar studies, a limitation of this study is the lack of comparison or
control intersections. However, exogenous changes were monitored to ensure minimal influence
on the results of the study. There were no events leading to changes in normal traffic patterns.
Furthermore, no advertising campaign took place concurrently with the study that could have
influenced yielding behavior. As such, the research team concluded that the changes observed in
drivers’ behavior were brought about mostly by the yield signs. This study was inconclusive on
the effect of the signs on the other surrogate measures observed.
Two separate periods of post- installation observation illustrate how the effect of the
countermeasures on the users varies over time. Percent car yield was observed to increase over
time in all cases. However, in other measures of effectiveness, a significant increase was often
followed by a reduction as in percent trapped for the unmarked crosswalk of the 16th and Capp.
Regardless of whether the initial percent yield was as low as 20 percent or as high as 60
percent, the after- installation percent yield varied between 60 to 74 percent. This may be a
possible indication that although effective, the effect of impactable yield signs could only
achieve a certain percent of improvement. Further studies on the subject are required to examine
this possibility.
The research team concluded that yield signs are useful in causing vehicles to yield to
pedestrians, as observed in the study locations. However, it is only crash data that can indicate
whether the signs are effective in preventing crashes, which is the primary aim of erecting these
signs. Conflicts, in some cases are the closest approximates to crashes, and the signs have been
seen to reduce conflicts in some cases, although not significantly, in this study. Some of the
measures of effectiveness indicate the possibility that the increase in the expectation of safety
brought about to pedestrians is detrimental because it might cause pedestrians to become less
vigilant ( 9), although there are studies that counter this theory too. As a study observes that an
increase in yielding from 0 to 50 percent is not as effective as an increase from 50 to 100 percent
( 10), since the former does not fully cure the situation and may cause false expectations. This is
an area that needs further research. There is a need for further research, especially including
crash data, to validate the conclusions drawn through the use of surrogate measures. Research is
also necessary to achieve a more durable yet low- cost installation.
TRB 2007 Annual Meeting CD- ROM Paper revised from original submittal.
Banerjee & Ragland 16
ACKNOWLEDGEMENTS
The authors would like to thank Jill Cooper and other staff of the UC Berkeley Traffic Safety
Center for their ongoing support and contributions. We would also like to thank Lindsay Arnold
and Joseph Zheng, graduate students from UC Berkeley for their contribution. The authors
would like to acknowledge the assistance of Frank Markowitz from the San Francisco
Department of Parking and Traffic and the Federal Highway Administration ( FHWA) for
support for the San Francisco PedSafe project.
TRB 2007 Annual Meeting CD- ROM Paper revised from original submittal.
Banerjee & Ragland 17
REFERENCES
1. Aragon, T., R. Reiter, and B. Katcher, San Francisco Burden of Disease and Injury:
Mortality Analysis, 1990- 1995, San Francisco Department of Public Health, 1998.
2. Carr, N. Pedestrian Fatality Report for 1997- 98. San Francisco Department of
Parking & Traffic. walksf. org/ DPTPedFatalityReportFor98/ report. html. Accessed
July 22, 2006.
3. McLoughlin, E., M. Weitzel, P. Skaj, and M. Radetsky. Profile of Injury in San
Francisco. The San Francisco Department of Public Health and The San
Francisco Injury Center, San Francisco, 1998.
4. Surface Transportation Policy Project. Caught in the Crosswalk. San Francisco,
California, 1999.
5. Statewide Integrated Traffic Records System ( SWITRS). Annual Report of Fatal and
Injury Motor Vehicle Traffic Collisions. California Highway Patrol, 2003.
6. DPH Traffic Safety projects, San Francisco Department of Public Health.
www. dph. sf. ca. us/ traffic_ safety/ funded_ projects. htm. Accessed July 22, 2006.
7. Ragland, D. R., F. Markowitz, and K. E. MacLeod. Intensive Pedestrian Safety
Engineering Study Using Computerized Crash Analysis. ITE 2003 Annual
Meeting and Exhibit, ITE, 2003.
8. Redmon, T. Looking out for Pedestrians. Public Roads Vol. 69, No. 3,
November/ December 2005. www. tfhrc. gov/ pubrds/ 05nov/ 03. htm Accessed July
21, 2006.
9. City of Madison Department of Transportation Traffic Engineering Division. Year 2
Field Evaluation of Experimental " In- Street" Yield to Pedestrian Signs. Federal
Highway Administration, 1999.
10. Huang, H., C. Zegeer, R. Nassi, and B. Fairfax. The Effects of Innovative Pedestrian
Signs at Unsignalized Locations: A Tale of Three Treatments. USDOT FHA,
Report No. FHWA- RD- 00- 098, 2000.
11. Van Houten, R., D. McCusker, S. Hybers, J. E. Louis Malenfant and D. Rice- Smith.
Advance Yield Markings and Fluorescent Yellow- Green RA 4 Signs at
Crosswalks with Uncontrolled Approaches. Transportation Research Record,
No. 1818, 2002.
12. Crosswalk Safety Problem. Public Works, Vol. 134, No. 2, 2003, pp 34- 35.
13. In- Street Yield to Pedestrian Sign Application in Cedar Rapids. Iowa, CTRE project
02- 115, Department of Civil and Construction Engineering, Iowa State
University, 2003.
14. Google Earth. Aerial Photographs of Intersections.
15. SFGOV. San Francisco Enterprise GIS ( SFGIS). www. sfgov. org/ site/ gis_ index. asp.
Accessed July 17, 2006.
16. City and County of San Francisco Department of Parking and Traffic. Traffic
Collision History Report. San Francisco, 2004.
17. Duszak, A., and Ragland, D. Pedsafe Evaluation Plan, Traffic Safety Center, 2006.
18. Science Applications International Corporation and Sprinkle Consulting, Inc.
Pedestrian Safety Engineering and Intelligent Transportation System- Based
Countermeasures Program for Reduced Pedestrian fatalities, Injuries, Conflict,
TRB 2007 Annual Meeting CD- ROM Paper revised from original submittal.
Banerjee & Ragland 18
and other Surrogate Measures, Detailed Test Plan: Evaluation of Individual
Countermeasures, Federal Highway Administration, 2004.
19. Cooper, D. Impactable Signs Protocol. Traffic Safety Center, 2005.
20. C. iv. c. Streets and Traffic Safety. Bicycle and Pedestrian Improvements: Pedestrian
Circulation and Safety. Proposition K, 5- Year prioritization Program, SFCTA,
DPT, Traffic Engineering Division, 2005.
TRB 2007 Annual Meeting CD- ROM Paper revised from original submittal.