Professional Documents
Culture Documents
org/26898
DETAILS
56 pages | 8.5 x 11 | PDF
ISBN 978-0-309-68770-6 | DOI 10.17226/26898
CONTRIBUTORS
Kay Fitzpatrick, Srinivas Geedipally, Boniphace Kutela, Peter Koonce; National
Cooperative Highway Research Program; Transportation Research Board; National
BUY THIS BOOK Academies of Sciences, Engineering, and Medicine
Visit the National Academies Press at nap.edu and login or register to get:
– Access to free PDF downloads of thousands of publications
– 10% off the price of print publications
– Email or social media notifications of new titles related to your interests
– Special offers and discounts
All downloadable National Academies titles are free to be used for personal and/or non-commercial
academic use. Users may also freely post links to our titles on this website; non-commercial academic
users are encouraged to link to the version on this website rather than distribute a downloaded PDF
to ensure that all users are accessing the latest authoritative version of the work. All other uses require
written permission. (Request Permission)
This PDF is protected by copyright and owned by the National Academy of Sciences; unless otherwise
indicated, the National Academy of Sciences retains copyright to all materials in this PDF with all rights
reserved.
Safety at Midblock Pedestrian Signals
N AT I O N A L C O O P E R AT I V E H I G H W AY R E S E A R C H P R O G R A M
Kay Fitzpatrick
Srinivas Geedipally
Boniphace Kutela
Texas A&M Transportation Institute
College Station, TX
Peter Koonce
Peter Koonce Consulting
Portland, OR
Subscriber Categories
Highways • Operations and Traffic Management
Research sponsored by the American Association of State Highway and Transportation Officials
in cooperation with the Federal Highway Administration
2023
The National Academy of Sciences was established in 1863 by an Act of Congress, signed by President Lincoln, as a private, non-
governmental institution to advise the nation on issues related to science and technology. Members are elected by their peers for
outstanding contributions to research. Dr. Marcia McNutt is president.
The National Academy of Engineering was established in 1964 under the charter of the National Academy of Sciences to bring the
practices of engineering to advising the nation. Members are elected by their peers for extraordinary contributions to engineering.
Dr. John L. Anderson is president.
The National Academy of Medicine (formerly the Institute of Medicine) was established in 1970 under the charter of the National
Academy of Sciences to advise the nation on medical and health issues. Members are elected by their peers for distinguished contributions
to medicine and health. Dr. Victor J. Dzau is president.
The three Academies work together as the National Academies of Sciences, Engineering, and Medicine to provide independent,
objective analysis and advice to the nation and conduct other activities to solve complex problems and inform public policy decisions.
The National Academies also encourage education and research, recognize outstanding contributions to knowledge, and increase
public understanding in matters of science, engineering, and medicine.
Learn more about the National Academies of Sciences, Engineering, and Medicine at www.nationalacademies.org.
The Transportation Research Board is one of seven major programs of the National Academies of Sciences, Engineering, and Medicine.
The mission of the Transportation Research Board is to provide leadership in transportation improvements and innovation through
trusted, timely, impartial, and evidence-based information exchange, research, and advice regarding all modes of transportation. The
Board’s varied activities annually engage about 8,000 engineers, scientists, and other transportation researchers and practitioners from
the public and private sectors and academia, all of whom contribute their expertise in the public interest. The program is supported by
state transportation departments, federal agencies including the component administrations of the U.S. Department of Transportation,
and other organizations and individuals interested in the development of transportation.
FOREWORD
By Zuxuan Deng
Staff Officer
Transportation Research Board
NCHRP Research Report 1030: Safety at Midblock Pedestrian Signals presents a state-
of-the-practice guide to midblock pedestrian crossing treatments, summarizes the safety
effectiveness of midblock pedestrian signal (MPS) installations, and proposes language
for consideration in future updates to the Manual on Uniform Traffic Control Devices
(MUTCD) for MPSs. The material in this report will be of immediate interest to new and
experienced traffic engineers and roadway designers in their efforts to design facilities to
accommodate midblock pedestrian crossings that lead to greater pedestrian safety and
satisfaction.
The National Committee on Uniform Traffic Control Devices (NCUTCD) Signals Tech-
nical Committee has reviewed treatments for several types of midblock pedestrian crosswalk
with highway traffic signals or beacons, and has recommended a new chapter for Part 4
of the MUTCD titled “Midblock Pedestrian Signals.” The MPS would operate similarly to
a standard semiactuated vehicular traffic control signal at a midblock crossing, except it
would display to motorists a flashing red indication in place of a solid red indication during
the pedestrian clearance interval.
The MPS supports “complete streets,” a transportation policy and design approach that
calls for roadways to be designed and operated with all users in mind: bicyclists, public
transportation users, drivers, and pedestrians of all ages and abilities. Pedestrians frequently
find themselves in a predicament when needing to cross roads to access transit stops, busi-
nesses, medical facilities, and residences. Without an easily accessible signalized crossing,
they may be uncomfortable crossing with only a crosswalk and sign, and pedestrian volumes
may not accurately reflect pedestrian demand. Moreover, increases in lower-density sprawl-
like development mean these crossings rarely happen in such a concentrated manner as to
justify a midblock signal based on conventional signal volume warrants described in MUTCD
Chapter 4C.
The FHWA’s Guide for Improving Pedestrian Safety at Uncontrolled Crossing Locations (2018)
presented alternative treatments such as the rectangular rapid-flashing beacon (RRFB) and
the pedestrian hybrid beacon (PHB). The proposed MPS is intended to expand safety options
at midblock crossings while reflecting modern pedestrian crossing needs and roadway
contexts. The MPS concept has been used for more than 40 years in several cities, including
Los Angeles, and previous FHWA studies have found this type of operation to have a very
high rate of driver compliance.
Under NCHRP Project 03-141, “Guidance on Midblock Pedestrian Signals,” Texas A&M
Transportation Institute was asked to summarize the effectiveness of MPS installations and
propose language for consideration for future updates to the MUTCD. The research team
CONTENTS
1 Chapter 1 Introduction
1 Research Objective
1 Organization of Report
3 Chapter 2 Literature Review
3 Midblock Pedestrian Signal Effectiveness
5 Half-Signal Effectiveness
5 PHB Safety Effectiveness
6 Coordinated Signals
6 Sources for Pedestrian Volume
10 Key Findings from Literature
11 Chapter 3 Survey of Public Agencies
11 Information about the Agencies Responding to the Survey
12 Experience with Pedestrian Crossing Treatments
12 Selecting Treatments
13 Detection for MPS/PHB Treatments
13 Display, Device Operations, and Performance
15 Benefits of Treatments
15 Summary of Key Survey Findings
16 Chapter 4 Safety Analysis—Database Development
16 Site Identification
17 Vehicle and Pedestrian Volume
22 Roadway Characteristics
26 Crash Data
27 Identifying Crashes for a Given Location/Database Cleaning
29 Chapter 5 Safety Analysis—Findings
29 Method
29 Number of Sites for Treated and Control Groups
31 Models
32 Findings
36 Crash Modification Factors
37 Comparison with Other Studies
40 Chapter 6 Conclusions and Recommendations
40 Summary
41 Discussion
42 Future Research Needs
44 References
Note: Photographs, figures, and tables in this report may have been converted from color to grayscale for printing.
The electronic version of the report (posted on the web at nap.nationalacademies.org) retains the color versions.
CHAPTER 1
Introduction
Research Objective
The objective of this research was to summarize the effectiveness of MPS installations
and propose language suitable for inclusion in the Manual on Uniform Traffic Control Devices
(MUTCD; 2).
Organization of Report
This report provides information on the safety analysis of the MPS in the following chapters:
1. Introduction: gives an overview of MPSs, project objectives, and the organization of the
report.
2. Literature Review: summarizes the literature on the safety effectiveness of pedestrian treatments
that include a signal controller and on how pedestrian volume at a signalized intersection can
be estimated.
1
3. Survey of Public Agencies: presents the results from a survey developed to gain greater under-
standing of the state of the practice for pedestrian crossing treatments at intersections and
midblock locations.
4. Safety Analysis—Database Development: describes the identification of potential study sites
and the building of the database to be used in the safety evaluation.
5. Safety Analysis—Findings: presents the findings from the safety analysis that investigated
changes in crash frequency by crash type due to the presence of the MPS.
6. Conclusions and Recommendations: provides a summary of the findings and discusses future
research needs.
CHAPTER 2
Literature Review
A literature search was conducted using several search techniques and resources, among
which the Transportation Research Information Services (TRIS) databases. TRIS can search
several databases, including the Highway Research Information Service database for domestic
literature, the Highway Research in Progress database for ongoing research studies, and the
International Road Research Database for relevant foreign literature. Based on that search,
as expected, only a few studies have included the MPS.
This chapter summarizes the literature on the safety effectiveness of pedestrian treatments
that include a signal controller. Also provided is a summary of the literature that discusses how
pedestrian volume at a signalized intersection can be estimated, because pedestrian volume is a
key element in a safety analysis of pedestrian treatments.
3
Table 1. Findings from data collected in 2003 for a TCRP/NCHRP study.
TCRP Project D-08/NCHRP Project 3-71 Other Studies
Compliance—General
Compliance—Staged Compliance—Literature
Crossing Population Pedestrian
Pedestrian Crossing Review
Treatment Crossing
# of Range Average # of Range Average # of Range Average
Sites (%) (%) Sites (%) (%) Sites (%) (%)
MPS 2 97 to 100 99% 4 91 to 98 95% na na na
Half signal 6 94 to 100 97% 6 96 to 100 98% 1 99 99%
HAWK (PHB) 5 94 to 100 97% 5 98 to 100 99% 1 93 93%
na = not applicable; there were no sites.
Source: Fitzpatrick et al. (3)
Figure 2. Site average and range for driver yielding by crossing treatment.
Literature Review 5
Half-Signal Effectiveness
Johnson (5) investigated the safety of half signals in Portland, OR, based on data collected
from the city in a thesis for Portland State University. Half signals are located at four-way
intersections and include a typical green-yellow-red traffic signal for automobiles on the
major road, a stop sign for motorists on the minor road, and a pedestrian signal with actua-
tion for pedestrians and/or bicyclists on the minor road. The treatment is prevalent in Canada,
but the MUTCD (2) discourages its use in the United States. The treatment can be found in
Portland, OR, and Seattle, WA, with the most recent installation of 47 half-signal intersections
in Portland in 1986.
Johnson reported on the crash history results from published half-signal studies as shown
in Table 3. For the 2015 study, Johnson considered 442 crashes over a 10-year period for the
47 half-signal intersections in Portland. Of the 442 crashes, 16 involved pedestrians. The calcu
lated crash rates for the half signals (0.158 and 0.178 crashes per million entering vehicles for
three-leg and four-leg half signals, respectively) did not differ significantly from comparison
groups (e.g., minor-street stop-controlled intersections and signalized intersections). The matched
comparison only showed rear-end (RE) crashes as being statistically significantly different, with
half signals having more RE crashes than the minor stop-controlled intersections comparison
group. The lack of pedestrian volume limited the ability to determine if the greater number of
pedestrian crashes at half-signal intersections versus signalized intersections was statistically
significant.
locations and those PHBs at three- or four-leg intersections was found in the cross-sectional
evaluation.
• The cross-sectional evaluation showed no statistically significant difference between the
lower-speed and higher-speed PHB sites (posted speeds at 35 mph or lower versus 40 mph or
higher) for all crash types except RE crashes. For RE crashes, fewer RE crashes were present
when the posted speed limit was 35 mph or lower.
Coordinated Signals
Hauer (13) in 2020 discussed the benefits of retiming coordinated signalized intersections to
improve the opportunities for pedestrians to cross midblock.
On-Site Counts
Several techniques are available for counting the number of pedestrians and bicyclists at a site,
including manual counts, inductive loops, thermal cameras, infrared counters, and pedestrian
signal actuation data. Kothuri (15) tested several technologies in a parking lot and at an inter
section. The author’s conclusions were that inductive loops and a thermal camera counted
bicycles accurately in a controlled environment but failed to do so at an intersection. Passive
infrared counters were found to count pedestrians accurately at the intersection sidewalk, and
pedestrian signal actuation data could be a cost-effective surrogate for pedestrian demand at
signalized intersections. Because this study was done about 10 years ago, the technology may
have improved since.
Literature Review 7
Direct-Demand Models
While vehicle volume is frequently available for a street, pedestrian volume is seldom available.
Turner et al. (16) provides a summary of techniques used to estimate and evaluate exposure to
risk in pedestrian and bicyclist safety analyses. The authors note that geographic scale is a critical
element, and provide four scale categories:
• Regional (e.g., city, county, or state)
• Network (e.g., traffic analysis zone, census tract, or census block group)
• Road segment
• Point (e.g., midblock or intersection street crossing)
With respect to crash evaluations for MPSs, exposure estimates at the point scale group are
needed. Direct-demand models are widely used for pedestrian and bicyclist volume estima-
tion, but they require local data—transportation system variables, built environment vari-
ables, socioeconomic characteristics, weather, typology—and are probably not transferable
to different areas.
Schneider et al. (17) in 2012 developed and applied a pedestrian intersection volume model
for San Francisco, CA. A sample of counts at 50 intersections was collected and adjusted to
produce annual pedestrian crossing estimates at each sampled intersection. Next, the authors
developed a log-linear regression model to identify the relationship between annual pedestrian
volume estimate and various explanatory variables including land use, transportation system, local
environment, and socioeconomic characteristics near each sampled intersection (see Table 5).
Griswold et al. (18) developed a statewide pedestrian exposure model using log-linear regres-
sion. Their database included more than 1,200 count locations in California. The model included
the following variables:
• Employment density
• Population density
• Number of schools
• Number of street segments
• Intersections with principal arterial and minor arterial roadways
• Four-way intersections
• U.S. Census Bureau’s American Community Survey journey-to-work walk mode share
Table 6 provides the variables and coefficient estimates for the recommended model. The
model was used to estimate pedestrian volume at state highway intersections within 656 ft
(200 m) of a census block with a minimum population density of 500 people per square mile.
The estimates are available at https://dot.ca.gov/programs/safety-programs/ped-bike/exposure.
Schneider et al. (19) in 2021 developed pedestrian intersection crossing volume models for the
Milwaukee, WI, metropolitan region. Negative binomial regression was used to relate annual
pedestrian volumes at 260 intersections to roadway and surrounding neighborhood socio
economic and land-use variables. The range of annual pedestrian volumes in the model dataset
was 1,000 to 650,000. The following variables describing the area surrounding an intersection
had statistically significant, positive associations with annual pedestrian volumes:
• Population density within 1,312 ft (400 m)
• Employment density within 1,312 ft (400 m)
• Number of bus stops within 328 ft (100 m)
• Number of retail businesses within 328 ft (100 m)
• Number of restaurant and bar businesses within 328 ft (100 m)
• Presence of a school within 1,312 ft (400 m)
• Proportion of households without a motor vehicle within 1,312 ft (400 m)
While three models (see Table 7) had a good overall statistical fit, the authors recommended
model C. They noted that the presence of a traffic control signal or a park within 400 m tended
to have consistent positive associations with annual pedestrian volumes, but they were not
significant at the 95% confidence level when included with the other seven variables in the final
models. The authors also expected being within 400 m of a university campus would be signifi-
cant; however, only six intersections had that characteristic.
Le et al. (20) conducted a study to explore options for collecting or estimating pedestrian
volume data in Dallas, TX, particularly at intersections with high pedestrian activity. The
authors successfully developed a direct-demand model that estimates pedestrian volumes
at signalized and stop-controlled intersections (see Table 8). The final model showed that
pedestrian volume increases 4 times within downtown; increases 12% per school within 1 mi
of an intersection; increases 4.8 times per 1% increase in commercial/multifamily residential
land uses within 300 ft of an intersection; increases 4.7 times with the presence of higher
education, hospitals, or malls; and decreases 36% per 5-mph increase in the intersections’
maximum posted speed limit.
Geedipally (21) developed a regression model to estimate the pedestrian crossing volumes at
midblock locations. The variables found to influence the crossing volumes were posted speed
limit, number of bus stops, sidewalk width, and area type (see Table 9).
Literature Review 9
Table 7. Final annual pedestrian crossing volume models for Milwaukee.
A. Base Model B. Square Root Model C. Cube Root Model
Variable
Beta p-Value Beta p-Value Beta p-Value
Constant 8.334 0.000 7.629 0.000 7.071 0.000
PopDen400 0.000140 0.001 na na na na
SRPopDen400 na na 0.019 0.000 na na
CRPopDen400 na na na na 0.100 0.000
EmpDen400 0.000021 0.046 na na na na
SREmpDen400 na na 0.00581 0.005 na na
CREmpDen400 na na na na 0.036 0.003
BusStp100 0.336 0.000 na na na na
SRBusStp100 na na 0.434 0.000 na na
CRBusStp100 na na na na 0.477 0.001
Retail100 0.108 0.026 na na na na
SRRet100 na na 0.208 0.000 na na
CRReBa100 na na na na 0.471 0.000
RestBar100 0.116 0.062 na na na na
SRReBa100 na na 0.208 0.050 na na
CRReBA100 na na na na 0.244 0.044
SchDum400 0.515 0.001 0.478 0.003 0.499 0.002
Pct0Veh400 5.307 0.000 4.184 0.001 4.330 0.000
Sample Size (n) 260 260 260
Log-likelihooda −2,792 −2,774 −2,772
Akaike information
5,601 5,565 5,560
criterion (AIC)a
Bayesian information
5,629 5,593 5,588
criterion (BIC)a
a
Lower absolute values of log-likelihood; AIC and BIC indicate better overall model fit.
na = not applicable.
Source: Schneider et al. (19).
Cross 24 hr
Unsig. Ped.
PHBb Ped.
Maj. 24 hr
Maj. 24 hr
HSMe Sig.
PHB Ped.
PHB Ped.
HSM Sig.
All 24 hr
Level of
3 Leg
4 Leg
24 hr
24 hr
24 hr
24 hr
Pedestrian
Activitya
High 950 1,180 2,130 320 290 610 820 700 1,520 1,700 3,200
Medium–high 490 480 970 190 180 370 410 530 940 750 1,500
Medium 170 220 390 90 90 180 210 290 500 400 700
Medium–low 90 40 130 40 40 80 110 170 280 120 240
Low 40 20 60 10 20 30 60 60 120 20 50
a
The team assumed the general level of high pedestrian activity to be the 90th percentile value (rounded to the
nearest 10) for the group of sites. The medium–high was the 75th percentile, the medium was the 50th percentile,
the medium–low was the 25th percentile, and the low was the 10th percentile value (rounded to the nearest 10).
Other assumptions include that the PHB is controlling the vehicles on the major street and that the pedestrian count
for “all” is the sum of the pedestrians crossing the major legs and the pedestrians crossing the cross-street legs (if any).
b
PHB values are based on 52 PHB (HAWK) intersections in Arizona.
c
Unsig. values are based on 98 unsignalized intersections in Arizona.
d
Sig. values are based on 33 signalized intersections in Arizona.
e
HSM values are from HSM (24) Tables 12–15, pp. 12–37.
CHAPTER 3
A survey was developed and administered as part of this study to gain greater understand-
ing of the state of the practice for pedestrian crossing treatments at intersections and midblock
locations. The survey was reviewed by the research team before being sent to the Institute of
Transportation Engineers (ITE) and the Association for Pedestrian and Bicycle Professionals
listservs. Responses were received from 23 agencies, and follow-up discussions occurred with
selected agencies to clarify answers submitted in the responses.
Survey questions asked specifically about the types of treatments used to address pedestrian
safety (e.g., full traffic signals, pedestrian hybrid beacons, rectangular rapid flashing beacons,
half signals, and MPSs) and supporting elements (e.g., in-pavement warning lights, pedes-
trian crossing signs, crosswalk markings and signs). Many questions about these specific
elements also included an “other” option with space for respondents to provide information
about treatments not listed on the survey. Topics for the treatments included methods of
detection for people walking and bicycling and operations for timing the devices, including
coordination.
11
Selecting Treatments
Engineering studies are used to determine appropriate crossing treatments for midblock
conditions. Two published resources used when determining treatments for pedestrian crossings
are NCHRP Report 562 (3) and the FHWA Guide for Improving Pedestrian Safety at Uncontrolled
Crossing Locations (25). These documents are often cited as some of the most widely used in
the industry due to their ease of use. Several agencies identified the use of other resources to
address the conditions, including state or local guidelines, which often reference similar criteria
as NCHRP or the FHWA. The other important data input for selecting improvement sites was
pedestrian crash data.
The criterion most considered when determining treatments for pedestrian crossings was
pedestrian (and bicycle) demand, with the following also being selected frequently: traffic
volumes, traffic speeds, roadway widths/geometry, sight distance, and whether a median/
no median is present or can be added. Other criteria include adjacent land use, the distance to
the next enhanced crossing, and the cost.
When asked to identify why the agency would select one treatment over another, there was
a significant range of ideas offered. RRFBs were being used in some cases and not in others
(crossing multiple lanes was mentioned as a concern). School crossings were a focus for a few
of the respondents, and cost was also referenced as a concern for the selection of treatments.
The need for distance between crossings was also mentioned with the MPSs.
Vision Zero is a global movement to end traffic-related fatalities and serious injuries by
taking a systemic approach to road safety. Given that most of the agencies reported the goal of
zero deaths, the research team assumed a systemic approach to safety would be frequently cited.
The systemic approach to safety is a data-driven process that involves analytical techniques
to identify sites for potential safety improvement and suggests projects for safety investment
not typically identified through the traditional site analysis approach. Several of the agencies
surveyed (39%) do not have a specific method for prioritizing sites for installation. About 30%
of the agencies said they use a systemwide analysis of sites throughout the agency; 17% of the
agencies said installations occur on project corridors only.
Nearly all agencies surveyed agree that the installation of an MPS or PHB would be considered
on a roadway with a 45 mph or higher speed. There have been some questions on the applica-
bility of beacons in higher-speed environments, so this is an important distinction to guide the
development of language for inclusion in a future version of the MUTCD.
Street lighting is often cited as having an important role in crosswalk safety. Adding illumi-
nation can improve the visibility of the people attempting to cross the street. With respect to
street lighting practices for signalized or unsignalized pedestrian crossings, 74% of agencies said
lighting is currently available or would be provided at any marked crosswalk. Other agencies
(9%) said lighting would be added when practicable, with only one agency saying a street lighting
policy is not in place. In addition to street lighting, accessible pedestrian access ramps are needed.
since having no indication was considered more desirable than a green indication, and having
a flashing yellow in addition to the steady yellow would attract the attention of motorists. The
other benefit of the PHB was reduced side-street delay associated with the flashing red portion
of the operation. Some of the agencies in the survey expressed concerns about the dark display
and the inability to use a traditional conflict monitor to identify issues with the traffic control.
With the MPS being midblock (or not an intersection), stop signs are not needed. An exception
could be where stop signs are desired for bicycle traffic. The PHB (or HAWK signal) was used in
the Portland, OR, Request to Experiment submitted in October 2005 to replace an existing half
signal, retaining the stop sign for the side street at the intersection.
The existing Los Angeles Department of Transportation MPS operation allows the flashing
red to occur during the WALK indication, which differs from the PHB operation that has the
main street stopped during the WALK, allowing a flashing red during the flashing DON’T
WALK. With respect to typical MPS displays, 78% of the agencies reported their preference to
have the main street operating with standard signal operation (i.e., solid red during the flashing
DON’T WALK phase), and 22% of the agencies using MPS displays operate with a flashing red
on the main street during the flashing DON’T WALK phase. Reasons given for not operating
the main street with a flashing red include driver confusion, that a flashing red would be more
like a HAWK treatment, and that a flashing red should only be used in malfunction conditions.
Operation of the main street with a flashing red consistent with the PHB sequence is a primary
question for further review. There was interest in learning more about the flashing red interval
from some of the agencies, but many of the agencies cited concerns about this.
The majority of agencies (78%) reported that the separation distance is the main factor
considered when deciding to operate MPSs/PHBs in coordination with adjacent signals. Some
agencies (30%) reported that the pedestrians at an MPS/PHB get an immediate response (i.e.,
a hot button or semi-hot button); nearly 22% of the agencies said pedestrians are only served
in coordination. Other less common MPS/PHB operations were reported as pedestrians
served in coordination depending on time of day, and pedestrian detection in conjunction with
vehicle detection.
The majority of agencies reported using a walk time of 7 seconds (per the MUTCD). Two
agencies reported using walk times as low as 5 seconds. Three agencies reported using walk
times of 8 seconds, 10 seconds, or an 8–10 second range. In Anchorage, AK, three specific
walk times were identified as 10, 15, and 17 seconds. Though not specifically confirmed with each
agency, it is suspected that walk times longer than 7 seconds may represent an increase based
on engineering judgment in order to serve pedestrian demands and/or other site-specific needs.
Consistent with the 2009 MUTCD (2), most agencies (78%) reported using a walking speed
of 3.5 ft/s or the option walking speed of 4 ft/s (13%). Other walking speeds were reported as
2.5 ft/s, 3 ft/s, and 3.5 ft/s minus the signal clearance interval. Variations in walking speeds
depended on the part of town (e.g., 2.8 ft/s in downtown Oklahoma City, OK) or nearby land
uses such as schools/retirement facilities (e.g., in Pittsburgh, PA).
Coordination of traffic signals is a common objective used by engineers to provide smooth
flow of traffic along streets and highways to reduce auto travel times, stops, and delays. MPS/
PHB locations interrupt the flow of traffic on the main street to provide an opportunity for
people to cross. The use of coordination may result in adverse impacts to people walking with
longer cycle lengths resulting in higher delays (15). Agencies were asked what typical cycle
length is used for their MPS/PHB locations, and the answers varied from 40 to 90 seconds.
The intent of MPS/PHB locations is to increase safety for people walking, and it seems clear that
the need for a display that encourages compliance is a motivation for many agencies. Many
of the PHB locations are operated without coordination.
Benefits of Treatments
The final questions of the survey focused on the benefits the agency has experienced with
MPSs or PHBs. Many of the participants highlighted the safety benefits and reduced crashes
involving pedestrians. The answers were diverse, with a few agencies expressing concerns
about the PHB and the public not understanding the display. There were also a few mentions
that the 2009 MUTCD (2) guidance that the PHB should not be located at intersections is a
barrier to use.
The final part of the survey was to gather any additional comments participants had on
the topic. Many respondents expressed thoughts about the lack of PHB understanding for
motorists. The requirement of flashing the red during the flashing DON’T WALK was mentioned
several times.
CHAPTER 4
Safety Analysis—
Database Development
The prime objective of this project was to determine the safety effectiveness of the MPS. This
section describes the identification of potential study sites and the building of the database that
will be used in the safety evaluation. The database includes roadway characteristics, volume for
both vehicles and pedestrians, and crash data.
Site Identification
The key for project success was identifying a sufficient number of treated sites. Members of the
panel and the research team reached out to several groups to aid in site identification. Sources
for recommended sites included members of the AASHTO Committee on Traffic Engineering
and the ITE community. The request noted that NCHRP had recently started a new project on
MPSs (NCHRP Project 3-141) and that the research team was looking for potential study sites.
The MPS treatment was briefly described. The request ended with contact information and
acknowledgment that participants’ help was appreciated.
Initial correspondence with members of the panel revealed that Los Angeles, CA, and Florida
could have several study sites. The region of the country with the most MPSs is the city of
Los Angeles. The city provided the research team a list of more than 150 installations. The Florida
sites were planned rather than existing sites and were removed from consideration. The research
team was informed about sites in Delaware, Kansas, and Hawaii. With only one or a few sites
in each of these states, these sites were also removed. During site selection, the research team
learned of several sites in Utah and San Antonio, TX. The team reviewed these sites and added
the locations to the database when they met the study criteria.
Study criteria included:
• A traffic control signal that included typical green, yellow, and red phases located at a two-leg
crossing was considered an MPS. The MPS could have either a green arrow or a steady green
ball indication. The two-leg description refers to the number of vehicle approaches rather
than pedestrian approaches.
• The MPS must have been in place for at least 12 months prior to the end of the available crash
data. Initially, crash data were assumed to be available for all the sites up to December 2020.
Later, the end date of December 2020 was revised with consideration of the pandemic, and
February 28, 2020, was used.
• For MPSs, the number of legs at the crossing was two. For control sites, the number of legs at
the crossing could be two, three, or four.
• State must have at least 10 MPSs to be included in the study.
• MPS sites represent a two-leg condition. If a driveway was believed to have hourly activity
(e.g., a driveway to a business, nongated, two-way operation, and/or a stop sign present), the
site was considered to be a control site as a three- or four-leg intersection.
16
• Sites were removed if atypical intersection geometry was present, such as a large skew or
a nearby frontage road that would affect pedestrian movement.
Treated sites in Los Angeles, CA, San Antonio, TX, and several cities in Utah were identified.
The research team then identified control sites near the treated sites. The ideal control site would
have a marked crosswalk with or without supplemental pedestrian-related traffic control devices
except a traffic control signal, two legs, and similar vehicle and pedestrian volume. The ideal
control site was very rare as it would be odd to have a site with similar characteristics (including
vehicle and pedestrian volume) to the nearby MPS but not signalized. The research team also
identified several nearby three- and four-leg signalized intersections as control sites. To increase
the pool of two-leg control sites, the research team had to identify sites in nearby cities.
Long Beach
Some of the Long Beach traffic volume counts were available from the Los Angeles system
(26, 27, 28). The research team obtained additional PDF-based traffic volume data collected in
2014 by the Traffic Engineering Division of the City of Long Beach (29). The traffic volume data
provided covered only major arterials and were in increments of 100. The traffic volumes in the
PDF-based data were provided by link/leg, not by direction.
San Francisco
The traffic volume dataset for San Francisco was obtained from the San Francisco Municipal
Transportation Agency website (30), downloaded in spreadsheet format. The data represented
24 hours and were collected between 2010 and 2019 from various sources. The data for inter
section and for midblock were available. In addition to the daily traffic counts, the morning and
evening peak hour counts by direction of travel were present.
San Jose
The traffic volume dataset for San Jose was collected from the city’s website under the
Enterprise GIS department (31). Both shapefile and comma-separated values (CSV) formats
were available. The data collected are dated as far back as 2005 and were updated annually.
The average daily traffic (ADT) counts denoted as “ADT One” and “ADT Two” are given by the
corresponding streets. The sum of traffic count on the first and second streets is the total traffic
volume for an intersection, which is denoted as “ADT.” The direction of streets and the travel
direction are provided. Furthermore, the actual date when the traffic count was performed
is given.
Table 11. Expansion factor for short-term vehicle and pedestrian counts.
Vehicle Pedestrian
Hour % 6.0 7.5 4.0 2.0 % 6.0 7.5 4.5 4.0 2.0 1.0
Veha Hrb Hr Hr Hr Peda Hr Hr Hr Hr Hr Hr
0 0.8 na na na na 0.0 na na na na na na
1 0.4 na na na na 0.0 na na na na na na
2 0.4 na na na na 0.0 na na na na na na
3 0.4 na na na na 0.0 na na na na na na
4 0.4 na na na na 0.0 na na na na na na
5 1.0 na na na na 0.0 na na na na na na
6 3.1 na na na na 6.0 na na na na na na
7 4.6 4.6 4.6 4.6 na 7.0 7.0 7.0 7.0 7.0 na na
8 6.5 6.5 6.5 6.5 na 8.0 8.0 8.0 8.0 8.0 na na
9 5.6 5.6 5.6 na na 7.0 7.0 7.0 3.5 na na na
10 5.2 na na na na 6.0 na na na na na na
11 5.8 na na na 5.8 8.0 na na na na 8.0 na
12 6.3 na na na 6.3 9.0 na na na na 9.0 9.0
13 6.7 na na na na 9.0 na na na na na na
14 6.9 na na na na 8.0 na na na na na na
15 7.5 7.5 7.5 na na 7.0 7.0 7.0 na na na na
16 8.3 8.3 8.3 8.3 na 9.0 9.0 9.0 9.0 9.0 na na
17 8.1 8.1 8.1 8.1 na 8.0 8.0 8.0 8.0 8.0 na na
18 6.3 na 6.3 na na 8.0 na 8.0 na na na na
19 5.0 na na na na 0.0 na na na na na na
20 3.8 na na na na 0.0 na na na na na na
21 3.1 na na na na 0.0 na na na na na na
22 2.5 na na na na 0.0 na na na na na na
23 1.3 na na na na 0.0 na na na na na na
% of
100 40.6 46.9 27.5 12.1 100 46.0 54.0 35.5 32.0 17.0 9.0
dayc
Adj.
1.00 2.46 2.13 3.64 8.28 1.00 2.17 1.85 2.82 3.13 5.88 11.11
Fac.d
a
Percentage of vehicles or pedestrians in given hour.
b
Percentage of vehicles or pedestrians in given hour based on length of initial count in hours.
c
Percentage of day represented in short-term count.
d
Adjustment factor to apply to short-term count.
na = not applicable.
Figure 3. Daily vehicle and pedestrian distribution used for adjusting
vehicle counts.
of interest. The team used a buffer of 0.5 mi to extract population density, employment density,
schools, and college campuses. A buffer of 0.25 mi was used to extract quantities for bus stops,
bus ridership, intersections, and restaurants.
The team did not find all the variables listed in Table 12 for cities other than Los Angeles. Bus
ridership was only available for Los Angeles; the bus stop variable was not available for San Jose
or Long Beach. Intersection density was also not available for Long Beach. Table 13 presents the
descriptive statistics of the variables used to develop the pedestrian volume statistical model. For
Los Angeles, the minimum population density is zero. This is the location close to the airport.
The census block for this location had zero counts for population density.
The research team evaluated various combinations of variables and developed a statistical
model to predict the number of pedestrians using a crosswalk. The intention of the model is
prediction, not inference, so the model with high prediction capability was selected.
Table 14 presents the negative binomial model results for the best-performing model. The
model has a low dispersion parameter (0.44) and a low Akaike information criterion (AIC)
score. The model results, similar to previous research, suggest that only higher posted speed limit
is associated with a lower number of pedestrians, while increases in the following are associated
with higher pedestrian volume: vehicle volume, population density, presence of signalized inter-
section or more than two legs, and increase in the number of schools, restaurants, and college
campuses.
Pedestrian Volume
A previous TxDOT project (42) collected pedestrian counts at several locations. In most cases,
the counts were for a few hours, and were expanded to daily counts using adjustment factors
developed by the research team based on nearby counts conducted for 12 hours. These locations,
along with their estimated daily pedestrian counts, were included in this NCHRP 3-141 study.
Of the 292 Texas sites, 40 (14%) did not have pedestrian counts. The pedestrian volume for these
40 sites was estimated using local knowledge (see Table 10).
Pedestrian Volume
The team collected pedestrian volume data from two geographic information system (GIS)-
based maps: one with estimated pedestrian counts for about 1,400 signalized intersections (44)
and the other with about 62,000 intersections (45). The data for 1,400 signalized intersections
were estimated based on a full year of push-button event data. The data for 62,000 intersections
were predicted based on a model developed using the surrounding built environment at the
intersections and the pedestrian activities obtained in the 1,400 intersections.
The research team overlaid the Utah treated (i.e., pedestrian crossing) and control sites on the
GIS-based pedestrian counts file. A buffer of 50 ft was used to identify potential pedestrian counts
for the 138 sites. First, the data for 1,400 signalized intersections were used, which facilitated
obtaining pedestrian counts for 28 sites. The data from the 1,400 signalized intersections layer
were of more interest to the team because they had fewer mathematical models/estimations
involved. The data from the 62,000 intersections layer then facilitated obtaining pedestrian volume
for an additional 38 sites. Since the remaining 72 sites did not match any estimated pedestrian
volumes for either available 1,400-signalized-intersections layer or 62,000-intersection layer,
the research team estimated the volume based on pedestrian volumes from two adjacent inter-
sections using the 62,000 intersections layer. If the site of interest was an intersection, the average
pedestrian volume of the two adjacent intersections was considered. If the site of interest was
a midblock crosswalk, half of the average pedestrian volume of the two adjacent intersections
was considered.
Roadway Characteristics
Data Collection
The research team assembled a spreadsheet with one record for each intersection or midblock
pedestrian crossing. The team used aerial and street-level photography sources available online
to extract the following observations to describe each site.
• Location (latitude and longitude coordinates of the intersection’s center or midblock cross-
walk’s center)
• Type of traffic control present at the site (see Table 15)
• Number of legs (two, three, or four)
• Main street posted speed limit
• Main street traffic configuration (two-way or one-way)
Table 15. Description of traffic control devices used at treated and control sites.
Traffic Control
Group for Type
Pedestrian Treatment Description
of Control
at Crossing
CBoverhead-24/7 Yellow Device Yellow circular beacon overhead, flashing 24/7
CBoverhead-PedAct Yellow Device Yellow circular beacon overhead, pedestrian activated
CBroadside-24/7 Yellow Device Yellow circular beacon roadside, flashing 24/7
CBroadside-PedAct Yellow Device Yellow circular beacon roadside, pedestrian activated
CW&Sign Grey Device Crosswalk pavement markings and crossing warning sign
CW_only Grey Device Crosswalk pavement markings only
LED-Em Yellow Device Yellow LEDs embedded in the sign
LED-Em & Flags Yellow Device Yellow LEDs embedded in the sign with flags
MPS Red Device Midblock pedestrian signal
NoPedTCD Grey Device No pedestrian traffic control devices (no crosswalk markings or sign, etc.)
PHB Red Device Pedestrian hybrid beacon
RRFB Yellow Device Rectangular rapid flashing beacon located on roadside
RRFB-Overhead Yellow Device Rectangular rapid flashing beacon located overhead
Signal Red Device Traditional traffic control signal (green, yellow, and red indications)
Stop-AllWay Red Device All approaches have stop control
Stop-Cont Red Device Stop sign at crosswalk
Stop-ContwCB Red Device Stop sign at crosswalk with circular beacon
Stop-OneWayTraffic Red Device Stop control at crosswalk with one-way traffic
Data Refinements
The team reviewed historical aerial views to determine the earliest date pedestrian-related traffic
control was present, back to January 1, 2014, which was the earliest date for crash data the team
anticipated using. Also gathered was the date for the end of the evaluation period, which is the
most recent date that the pedestrian-related traffic control was present, up to December 31, 2020.
Part of the process for developing the database was the decision of whether a site should
be considered a midblock or an intersection. If a traffic control signal head was present on the
minor approach (either a street or a driveway), the site was considered an intersection.
The variables in the initial list of roadway characteristics were refined during the preliminary
analyses to develop the variables used in the final analysis. Table 15 provides the description of
the pedestrian treatments considered and how they were later grouped; Table 16 provides the
descriptions of the other variables considered in the analyses.
Nearby Driveways
For the potential midblock (i.e., two-leg) sites, the distances between the marked crosswalk
and the nearest driveway in both directions were gathered. A note was made if the distance was
greater than 100 ft. For the two-leg sites with less than 100 ft between a driveway and the marked
crosswalk, the research team conducted an additional review to decide if the signalized location
should be considered an intersection with three or four legs or a midblock (two-leg) location.
As a reminder, all these sites had signal heads on the main street only; no signal head was
present for the driveway. When the driveway did not have a signal head and was anticipated
to have minimal volume, it was believed to have nominal influence on operations and safety at
that location and was kept in the database as a midblock signal. Additional characteristics of the
driveway used to make the decision were if the driveway had one-way operation, did not have a
stop sign, or was gated. For the 193 MPS sites included in the database, the distance by groups
(generally 25-ft increments with the smallest group being 15 ft and the largest group being all
sites with more than 100 ft) to the south or west and north or east direction of the marked cross-
walk is shown in Figure 4. The majority of the MPS sites in the database did not have a driveway
within 100 ft of the marked crosswalk.
Table 17. Descriptive statistics of sites for database with 892 sites.
T or Ca Variable Nameb Values CA TX UT All States
Count Number 150 11 32 193
Min. 30 40 39 30
Main:TotCrossDis (ft) Average 62 49 75 63
Max. 130 66 113 130
Min. 2,900 3,999 2,500 2,500
T V:IntersecEnterVol
Average 15,890 11,012 14,713 15,417
(ADT)
Max. 51,166 13,128 42,000 51,166
Min. 15 59 25 15
V:Ped (ADP) Average 2,420 341 165 1,927
Max. 22,523 652 501 22,523
Count Number 313 280 106 699
Min. 20 22 24 20
Main:TotCrossDis (ft) Average 60 63 79 64
Max. 130 163 127 163
Min. 62 309 1,000 62
C V:IntersecEnterVol
Average 25,035 21,691 20,292 22,976
(ADT)
Max. 127,230 66,990 81,000 127,230
Min. 28 9 10 9
V:Ped (ADP) Average 4,028 637 483 2,132
Max. 25,851 14,907 4,519 25,851
a
T = treated sites, C = control sites.
b
See description in Table 16.
Crash Data
The research team acquired crash data from databases available in each state. Because one
of the state databases only had fatal and injury (FI) crashes available, the analysis did not use
property-damage-only (PDO) crashes. This section provides detailed explanations of the data-
bases and key variables of interest for this study.
CHAPTER 5
Safety Analysis—Findings
The focus of the safety analysis was investigating changes in crash frequency by type
(all, pedestrians, and rear end) due to the presence of the MPS. The evaluation only considered
FI crashes because the California data did not include PDO crashes.
Method
The evaluation used a cross-sectional observational study approach. While an EB before-after
method is commonly used for this type of evaluation, most of the MPSs have been installed for
more than 10 years, which limits the availability of high-quality before data. Only a few MPSs
have been installed in the past 3–5 years, which was too small of a sample. The team therefore
used a cross-sectional study, which is a type of observational study where the outcome and the
exposure are assessed at one point or for a short period of time in a sample population (47).
The underlying assumption is that all sites should have similar characteristics (e.g., pedestrian
volumes and driver behavior). Routinely collected data such as crash, traffic, and geometric
data are often used for conducting cross-sectional data analyses. Cross-sectional studies can
be conducted faster than before-after studies and are commonly used in traffic safety analysis.
For instance, crash-frequency models or safety performance functions in the HSM (24) are
developed based on cross-sectional data.
29
Safety Analysis—Findings 31
Models
Preliminary analyses were conducted using the variables listed in Table 16. After preliminary
analyses, some of the variables were eliminated because they were consistently not significant
(e.g., type of median treatment) or were refined (e.g., creating posted speed limit groups). Table 23
lists the variables used in the safety analysis.
Most of the MPSs identified were included in the evaluation. One site was removed from
consideration during model development because it was frequently identified as an outlier. The
site was on a surface street under several freeway overpasses. The 250-ft radius initially used
to identify crashes was adjusted to a 150-ft radius; however, freeway crashes were suspected to
still be in the crash count. Because of the additional challenges in determining which crashes
were on the surface street and which were on the freeway, the research team decided instead to
remove the site from the evaluation.
For model development, the research team used negative binomial regression and first
examined various combinations of variables. In some cases, a variable had to be regrouped
during the analysis; for example, the posted speed limit along the major street (i.e., the street
with the MPS) was regrouped into below 30 mph (yes or no). The model form presented reflects
the findings from several preliminary regression analyses. The predicted crash frequency is
calculated in Equations 1–3.
with
where
E[C] = Predicted annual average crash frequency
y = Number of years of crash data
fst = State indicator variable
CA = California indicator variable (= 1 if site is in CA; = 0 otherwise)
UT = Utah indicator variable (= 1 if site is in UT; = 0 otherwise)
ADT = Average daily traffic, vehicles per day
ADP = Average daily pedestrian volume, pedestrians per day
legs = Number of intersection legs
Iow = One-way street indicator variable (= 1 if one-way street; = 0 otherwise)
Ipsl = Indicator variable for posted speed limit (= 1 if < 30 mph; = 0 otherwise)
MLanes = Number of lanes on the major street
Ipk = Indicator variable for the parking on the road (= 1 if present; = 0 otherwise)
Ibk = Indicator variable for the bike lane (= 1 if present; = 0 otherwise)
CMFtrt = Crash modification factor for the treatment (i.e., MPS)
Itrt = Indicator for the MPS (= 1 if present; = 0 otherwise)
bj = Calibrated coefficients
Findings
Pedestrian Fatal and Injury Crashes
Table 24 provides the estimated regression coefficients for pedestrian FI crashes for the
three control groups. As expected, both vehicle volume and pedestrian volume were significant,
with increasing activity being associated with more pedestrian crashes. A greater number of
legs at the intersection was also associated with more pedestrian crashes when considering
control group 1, which included intersections with two, three, and four legs along with all
types of pedestrian traffic control. When focusing on sites with two legs, the presence of
on-street parking was associated with fewer pedestrian crashes. After experimenting with
different approaches for considering posted speed limit, the decision was to group that
variable into those sites with 30 mph and more on the major road and those sites with less
than 30 mph. For all control groups, the variable was significant, with a negative coefficient
indicating that when the posted speed limit was 25 mph and less, fewer pedestrian crashes
were present.
A concern expressed by members of this project’s panel and by the National Committee
on Uniform Traffic Control Devices (NCUTCD) Signal Technical Committee while review-
ing these findings was the potential influence of a nearby driveway. The research team
addressed this concern by developing additional models to explore the impacts of distance
between the nearest driveway and the marked crosswalk. The treated sites were grouped
as follows:
• Treated group 1 (also called “All Treated Sites”) comprised all identified treated sites. These
sites had driveways as close as 10 ft to the center of the MPS marked crosswalk. None of the
MPS sites had a driveway that was closer than 10 ft to the crosswalk.
• Treated group 2 (called “MPS with No Driveway within 50 ft” or “> 50 ft”) comprised
all treated sites where the closest driveway was 50 ft away or more.
• Treated group 3 (called “MPS with No Driveway within 100 ft” or “> 100 ft”) comprised
all treated sites where the closest driveway was 100 ft away or more.
Table 25 shows the results for the two additional models that used treated groups 2 and 3.
These models considered pedestrian FI crashes along with the 2-leg grey control sites. The
treated sites were restricted to those sites with either no driveway within 100 ft or no driveway
within 50 ft of the marked crosswalk. The MPS treatment variable was significant (p < 0.05) for
the model that included sites where driveways were at least 50 ft away from the marked cross-
walk and marginally significant (0.05 < p < 0.1) for the model where the nearest driveway was
within 100 ft. The smaller sample size for treated group 3 (MPS with No Driveway within 100 ft)
may be affecting the finding. In other words, an MPS provides a safety benefit for pedestrians
regardless of the distance to the driveway.
Table 26. Estimated regression coefficient (and p-values) for all FI crashes
by control group and by subset of MPS sites.
• FI Crashes
• FI Crashes • FI Crashes • FI Crashes
• 2-Leg Grey
Model • All Control • 2-Leg Control • 2-Leg Grey
Control Sites
Coefficient, Sites Sites Control Sites
Variable Name • MPS with No
• All Treated • All Treated • All Treated
Driveway
Sites Sites Sites
within 50 ft
0.06096 0.3487 0.3582 0.3977
badp, ADP
(0.0122) (< 0.0001) (< 0.0001) (< 0.0001)
0.6655 0.8905 0.8381 0.9588
badt, ADT
(< 0.0001) (< 0.0001) (< 0.0001) (< 0.0001)
−0.8529 −1.4866 −1.2388 −1.3024
bca, CA
(< 0.0001) (< 0.0001) (< 0.0001) (< 0.0001)
0.6391 0 0 0
blegs, Legs
(< 0.0001) 0 0 0
bm_lanes, 0.07914 0 0 0
NumLanesMaj (0.0043) 0 0 0
−0.1922 −0.5341 −0.5035 −0.3611
bow, OneWay
(0.0586) (0.0157) (0.024) (0.179)
bpk, 0 −0.3828 −0.3900 −0.3793
OnStreetParking 0 (0.0161) (0.0186) (0.0398)
−0.4172 −0.3339 −0.4117 −0.3481
bpsl, PostedSpd<30
(< 0.0001) (0.0567) (0.029) (0.1179)
−0.1153 −0.1752 −0.4158 −0.3506
btrt, Treatment
(0.3017) (0.3049) (0.0387) (0.1278)
−0.7588 −0.6616 −0.5256 −0.6403
but Utah
(< 0.0001) (0.0065) (0.058) (0.0433)
−8.3403 −10.0365 −9.5408 −10.872
b0
(< 0.0001) (< 0.0001) (< 0.0001) (< 0.0001)
0.4872 0.7245 0.7045 0.6671
k0
(< 0.0001) (< 0.0001) (< 0.0001) (< 0.0001)
Number of treated
250 sites: 130
and control sites in
891 sites (1 control 380 sites (2 control 321 sites (2 control control and 120
model (sites
site removed) sites removed) sites removed) treated (2 treated
removed as part of
sites removed)
outlier analysis)
Notes:
See Table 23 for descriptions of the variables.
b0 = intercept and k0 = dispersion parameter.
The coefficient 0 denotes that the corresponding variable was excluded from the model.
Cells are highlighted in light grey with italic text when the p-value is between 0.05 and 0.1.
Cells are highlighted in dark grey with bold white text when the p-value is less than 0.05.
Safety Analysis—Findings 35
there is a driveway 10 ft or more away, and no safety disbenefit was found when there is no
driveway within 50 ft.
• RE FI Crashes
• RE FI Crashes • RE FI Crashes
Model Coefficient, • 2-Leg Grey Control
• All Control Sites • 2-Leg Control Sites
Variable Name Sites
• All Treated Sites • All Treated Sites
• All Treated Sites
0 0.1759 0.1757
badp, ADP
0 (0.0578) (0.0965)
0.7918 0.8109 0.7092
badt, ADT
(< 0.0001) (< 0.0001) (< 0.0001)
0.5275 0 0
bca, CA
(< 0.0001) 0 0
0.3873 0 0
blegs, Legs
(< 0.0001) 0 0
0.1243 0 0
bm_lanes
(0.0009) 0 0
0 −1.0243 −0.9767
bow, OneWay
0 (0.0275) (0.0346)
0 0 −0.443
bpk
0 0 (0.0544)
−0.6003 −0.7051 −0.6829
bpsl
(< 0.0001) (0.0105) (0.0265)
−0.3768 −0.2588 −0.1303
btrt
(0.0364) (0.2842) (0.6624)
0.9135 0.9524 0.6693
but
(< 0.0001) (0.0002) (0.0335)
−11.3643 −11.0553 −9.8592
b0
(< 0.0001) (< 0.0001) (< 0.0001)
0.5938 0.859 0.7223
k0
(< 0.0001) (0.002) (0.0086)
Number of treated and
control sites in model 893 sites (no site 381 sites (1 control site 322 sites (1 control site
(sites removed in outlier removed) removed) removed)
analysis)
Notes:
See Table 23 for descriptions of the variables.
b0 = intercept and k0 = dispersion parameter.
The coefficient 0 denotes that the corresponding variable was excluded from the model.
Cells are highlighted in light grey with italic text when the p-value is between 0.05 and 0.1.
Cells are highlighted in dark grey with bold white text when the p-value is less than 0.05.
The various models, along with control groups and treated groups, resulted in several cases
where the treatment variable is significant. These results can be considered for developing a
CMF for the MPS treatment. Table 29 summarizes the results from this research for CMF
consideration by crash type, control group, and treated group. The research team selected a
recommended MPS CMF for each crash type.
For the crash type of all FI crashes and RE FI crashes, only one of the control groups resulted
in a statistically significant CMF. For all FI crashes, the control group was the sites with two legs
and nonactive pedestrian treatments (i.e., grey). For the RE FI crashes, the 2-leg grey control
Safety Analysis—Findings 37
group did not produce a statistically significant coefficient, but the control group that included
all comparison sites did.
More than one of the models for pedestrian FI crashes resulted in a statistically significant
coefficient, and the research team recommends the model that considered those MPSs where a
driveway was not within 50 ft of the marked crosswalk.
Table 30. Comparison of CMFs for PHB and MPS pedestrian treatments
by crash type.
MPS CMF
PHB CMF PHB CMF
Crash Type NCHRP 3-141
2017 NCHRP Study (11) 2019 ADOT Study (12)
(this study)
All crash types, all
0.820 0.818 Not generated
severity levels
All crash types, FI
Not generated 0.748 0.660
only
Pedestrian crashes,
0.432 0.543 Not generated
all severity levels
Pedestrian crashes,
Not generated 0.550 0.554
FI
Rear-end crashes, all
Not generated 0.795 Not generated
severity levels
Rear-end crashes, FI Not generated 0.714 0.686
Rear-end and
sideswipe crashes, 0.876 Not generated Not generated
all severity levels
Safety Analysis—Findings 39
In general, the CMFs for the MPS developed in this study are similar to the CMFs identified
for the PHB. A needed caution when comparing the CMFs for the MPS and the PHB is the
difference in characteristics of the sites in each study. The MPS sites all have two legs, while only
21% of the PHB sites in the ADOT study have two legs. The MPS is appropriate for locations
with only two legs (midblock), while the PHB is appropriate for locations with three or four legs.
Almost all the MPS sites (93%) had 35 mph or lower posted speed limits. For the PHB sites in
the ADOT study, only 42% were on 35 mph or lower posted speed limit roads, with the majority
(57%) being on roads with 40 or 45 mph posted speed limits.
CHAPTER 6
Summary
The MPS treatment operates similarly to a coordinated-actuated vehicular traffic control
signal at a midblock crossing. In some places, it uses a green arrow rather than a steady green ball.
In some locations, the MPS displays a flashing red indication in place of a steady red indication
during the pedestrian clearance interval. This project conducted a data-driven analysis to identify
the safety performance of this pedestrian treatment.
One key to a successful safety analysis is identifying appropriate treated and control sites.
The research team identified sites in California, Texas, and Utah. The MPS sites had two legs
(i.e., two motorized vehicle approaches) with a traffic control signal that included green (either
arrow or ball), yellow, and red phases. In a data-driven safety analysis, ideally, the control sites
would also only have two legs with a marked crosswalk that did not have a traffic control signal
and with similar vehicle and pedestrian volumes. Due to the inherent challenges with finding
such sites, the research team also identified control sites with three or four legs and a range
of pedestrian treatments, including traffic control signals. Sites were removed if atypical inter-
section geometry was present, such as a large skew or a nearby frontage road that would affect
pedestrian movement.
Another key consideration for a safety analysis of a pedestrian treatment is vehicle and
pedestrian volume. While vehicle counts are frequently available (and for midblock locations
neighboring intersection counts can be used), pedestrian counts, especially at midblock crossings,
are rare. Recently researchers have used a sample of pedestrian counts to generate direct-demand
models that are used to estimate pedestrian counts at other locations. Pedestrian counts were
available for several locations in California and Utah using that approach. Pedestrian counts
for the Texas sites were from a previous TxDOT project. For the remaining sites in California,
the research team developed regression models to predict pedestrian volume using census data.
The research team acquired crash data from databases in each state. One of the state’s data-
bases only had FI crashes; the analysis therefore did not use PDO crashes. Because pedestrian
crashes are rare, the research team wanted to obtain 7 years of crash data. Crash data for 2014 to
2020 were pulled from each state’s database. Upon discussions with city staff familiar with the
sites, along with ongoing research into the impacts of the pandemic, the research team decided
to focus on a prepandemic time frame, so crash data between January 2014 and March 2020 was
used. Dates for a site were adjusted if historical aerial or street views identified major changes at
the site (e.g., treatment installed or construction). For most of the sites, a 250-ft buffer around
the intersection or midblock crosswalk was used to identify crashes; however, for a few of the
midblock sites, that buffer was reduced to 150 ft to avoid including crashes associated with a
40
nearby intersection. The crash database was reviewed and cleaned to remove sites where the
crashes for a nearby freeway could not be easily filtered (typically when the study site street was
either on an overpass or underpass of a freeway) or if other conditions exist that resulted in the
site being considered an outlier.
The roadway variables considered in the safety analysis, in addition to the presence of the
treatment and vehicle and pedestrian volumes, were the number of legs, one-way or two-way
traffic operations, posted speed limit, presence of a bike lane or on-street parking, and the site’s
state. Three control groups were tested:
• All sites (intersections with two, three, and four legs along with all types of pedestrian traffic
control, including intersection traffic control signals)
• All two-leg sites with any type of pedestrian traffic control other than the MPS
• All two-leg sites with nonactive or not present pedestrian traffic control device, called grey
pedestrian traffic control in this study (sites with marked crosswalk and signs, sites with marked
crosswalk only, and sites with no pedestrian treatment)
The analysis found that more pedestrian FI crashes occur with:
• More vehicle volume
• More pedestrian volume
• More legs at the intersection
• More lanes on the main street
Fewer pedestrian FI crashes occur when:
• The posted speed limit is lower than 30 mph
• On-street parking is present
• A bike lane is present
• One-way traffic is present at two-leg sites
The MPS is associated with a reduction in pedestrian crashes and a reduction in all FI crashes
when the control group is group 3 (i.e., two-leg sites with grey pedestrian traffic control). The
following crash modification factors were identified:
• 0.554 for pedestrian crashes
• 0.660 for all crashes
With a control group that includes the complete range of pedestrian crossing traffic control
from signals to pavement marking to no treatment, the MPS treatment was found to have
borderline significance (p-value of 0.0511) for pedestrian FI crashes; therefore, the research
team recommends the CMF listed previously that considered a smaller control group. For the
all-site control group, the MPS treatment was associated with a reduction in the number of
RE FI crashes but not all FI crashes.
Discussion
This safety study determined that the MPS provides safety benefits for both pedestrians and
drivers of motor vehicles. Worth noting are the characteristics of the MPS sites included in this
study. All MPS sites had two legs, and 93% of the sites were on roads with 35 mph or lower
posted speed limits. The reduction in crashes found in this study is applicable to signalized
pedestrian crossings at midblock crossings (not three- or four-leg intersections) located on lower-
speed streets.
• A few sites had to be eliminated in this study because the method used to identify crashes
resulted in capturing crashes on a neighboring freeway or on a freeway that was overpassing
or underpassing the street of interest. The research team may have been able to remove those
crashes by reviewing the characteristics of each crash or by reviewing the crash narrative,
which is a labor-intensive effort. An efficient and effective method for removing crashes
occurring on another road is currently not available for the state databases used in this study.
A suggestion that would permit the identification of crashes on overpasses or underpasses is
for crash databases to include the z-coordinate along with the x- and y-coordinates for a crash.
References
1. Governors Highway Safety Association. Pedestrian Traffic Fatalities by State. 2020. www.ghsa.org/sites/
default/files/2021-03/Ped%20Spotlight%202021%20FINAL%203.23.21.pdf.
2. FHWA. Manual on Uniform Traffic Control Devices. 2009. http://mutcd.fhwa.dot.gov/pdfs/2009r1r2/
pdf_index.htm.
3. Fitzpatrick, K., S. Turner, M. Brewer, P. Carlson, B. Ullman, N. Trout, E. S. Park, J. Whitacre, N. Lalani, and
D. Lord. TCRP Report 112/NCHRP Report 562: Improving Pedestrian Safety at Unsignalized Crossings.
Transportation Research Board of the National Academies, Washington, DC, 2006. https://doi.org/10.17226/
13962.
4. Hourdos, J. Assessing the Impact of Pedestrian-Activated Crossing Systems. Final Report 2020-13.
Minnesota Department of Transportation, 2020.
5. Johnson, T. R. Safety at Half-Signal Intersections in Portland, Oregon. MS thesis. Portland State University,
2015.
6. Wiatrak, A. School–Pedestrian Crosswalk Signals at Intersections: A Quantified Evaluation of Performance.
Traffic and Transportation Division of the Seattle Engineering Department, 1974.
7. Petzold, R. G., and R. Nawrocki. Urban Intersection Improvements for Pedestrian Safety: Volume V—Evaluation
of Alternatives to Full Signalization at Pedestrian Crossings. FHWA-RD-77-146. FHWA, USDOT, 1977.
8. Thomas, D. Half/Pedestrian Signal Report. Intranet Quorum IMA00421970. 2010.
9. Canadian Institute of Transportation Engineers. A Technical Review of Pedestrian Signals in Canada. 2006.
10. Fitzpatrick, K., and E. S. Park. Safety Effectiveness of the HAWK Pedestrian Crossing Treatment. FHWA-
HRT-10-042. FHWA, 2010.
11. Zegeer, C., R. Srinivasan, B. Lan, D. Carter, S. Smith, C. Sundstrom, N. Thirsk, C. Lyon, B. Persaud, J. Zegeer,
E. Ferguson, and R. Van Houten. NCHRP Research Report 841: Development of Crash Modification Factors
for Uncontrolled Pedestrian Crossing Treatments. Transportation Research Board, Washington, DC, 2017.
12. Fitzpatrick, K., M. J. Cynecki, M. P. Pratt, E. S. Park, and M. E. Beckley. Evaluation of Pedestrian Hybrid
Beacons on Arizona Highways. SPR-756. Phoenix, AZ: ADOT, 2019. https://apps.azdot.gov/files/ADOT
Library/publications/project_reports/pdf/spr756.pdf.
13. Hauer, E. Signal Timing and Midblock Crossings. ITE Journal, December 2020.
14. Zegeer, C. V., R. J. Stewart, H. H. Huang, P. A. Lagerwey, J. Feaganes, and B. J. Campbell. Safety Effects of
Marked versus Unmarked Crosswalks at Uncontrolled Locations: Final Report and Recommended Guidelines.
FHWA-HRT-04-100. McLean, VA: FHWA, 2005. www.fhwa.dot.gov/publications/research/safety/04100/.
15. Kothuri, S. Exploring Pedestrian Responsive Traffic Signal Timing Strategies in Urban Areas. PhD thesis.
Portland State University, 2014.
16. Turner, S., I. Sener, M. Martin, S. Das, E. Shipp, R. Hampshire, K. Fitzpatrick, L. Molnar, R. Wijesundera,
M. Colety, and S. Robinson. Synthesis of Methods for Estimating Pedestrian and Bicyclist Exposure to Risk
at Areawide Levels and on Specific Transportation Facilities. FHWA-SA-17-041. FHWA, 2017.
17. Schneider, R. J., T. Henry, M. F. Mitman, L. Stonehill, and J. Koehler. Development and Application of Volume
Model for Pedestrian Intersections in San Francisco, California. Transportation Research Record: Journal of
the Transportation Research Board 2299, no. 1 (January 2012): pp. 65–78. https://doi.org/10.3141/2299-08.
18. Griswold, J. B., A. Medury, R. J. Schneider, D. Amos, A. Li, and O. Grembek. A Pedestrian Exposure Model
for the California State Highway System. Transportation Research Record: Journal of the Transportation
Research Board 2673, no. 4 (April 2019): pp. 941–950. https://doi.org/10.1177/0361198119837235.
19. Schneider, R. J., A. Schmitz, and X. Qin. Development and Validation of a Seven-County Regional Pedes-
trian Volume Model. Transportation Research Record: Journal of the Transportation Research Board 2675,
no. 6 (June 2021): pp. 352–368. https://doi.org/10.1177/0361198121992360.
44
References 45
20. Le, M., S. R. Geedipally, K. Fitzpatrick, and R. E. Avelar. Estimating Pedestrian Volumes for Signalized and
Stop-Controlled Intersections. Transportation Research Record: Journal of the Transportation Research
Board 2674, no. 9 (September 2020): pp. 799–808. https://doi.org/10.1177/0361198120932161.
21. Geedipally, S. R. Estimating Pedestrian Crossing Volumes and Crashes at Midblock Locations. Transporta-
tion Research Record: Journal of the Transportation Research Board 2676, no. 8 (April 2022): pp. 648–656.
https://doi.org/10.1177/03611981221084676.
22. Xiaofeng L., P. Xu, and Y. J. Wu. Pedestrian Crossing Volume Estimation at Signalized Intersections
Using Bayesian Additive Regression Trees. Journal of Intelligent Transportation Systems 26, no. 5 (2022):
pp. 557–571. https://doi.org/10.1080/15472450.2021.1933471.
23. Singleton, P. A., and F. Runa. Pedestrian Traffic Signal Data Accurately Estimates Pedestrian Crossing
Volumes. Transportation Research Record: Journal of the Transportation Research Board 2675, no. 6
(March 2021): pp. 429–440. https://doi.org/10.1177/0361198121994126.
24. American Association of State Highway and Transportation Officials (2010). Highway Safety Manual.
Vol. 2. Washington, DC.
25. Blackburn, L., C. Zegeer, and K. Brookshire. Guide for Improving Pedestrian Safety at Uncontrolled Crossing
Locations. FHWA-SA-17-072. FHWA, 2018. https://safety.fhwa.dot.gov/ped_bike/step/resources/docs/
fhwasa18018.pdf.
26. Los Angeles GeoHub. Traffic Data—Segments. https://geohub.lacity.org/datasets/traffic-counts/explore?
location=34.153160%2C-118.480429%2C14.47. Accessed December 8, 2021.
27. Los Angeles GeoHub. Traffic Data—Intersections. https://geohub.lacity.org/datasets/traffic-data/explore?
location=34.042619%2C-118.744248%2C10.24. Accessed December 8, 2021.
28. Los Angeles. Navigate LA. https://navigatela.lacity.org/navigatela/. Accessed December 8, 2021.
29. City of Long Beach. Engineering Bureau Traffic Engineering Division 2014 Citywide Traffic Flow. https://
www.longbeach.gov/globalassets/pw/media-library/documents/resources/general/maps-and-gis/2014-
citywide-traffic-flow. Accessed December 8, 2021.
30. San Francisco Municipal Transportation Agency. SFMTA Traffic Count Data 2014–2018. https://www.
sfmta.com/reports/sfmta-traffic-count-data. Accessed December 8, 2021.
31. San Jose. Average Daily Traffic. https://data.sanjoseca.gov/dataset/average-daily-traffic. Accessed July 13,
2022.
32. Fitzpatrick, K., R. Avelar, M. P. Pratt, S. Das, and D. Lord. Crash Modification Factor for Corner Radius,
Right-Turn Speed, and Prediction of Pedestrian Crashes at Signalized Intersections. FHWA-HRT-21-105.
FHWA, 2022.
33. Zegeer, C., C. Lyon, R. Srinivasan, B. Persaud, B. Lan, S. Smith, D. Carter, N. J. Thirsk, J. Zegeer, E. Ferguson,
R. Van Houten, and C. Sundstrom. Development of Crash Modification Factors for Uncontrolled Pedestrian
Crossing Treatments. Transportation Research Record: Journal of the Transportation Research Board 2636,
no. 1 (January 2017): pp. 1–8. https://doi.org/10.3141/2636-01.
34. Schrank, D., B. Eisele, and T. Lomax. 2019 Urban Mobility Report, Appendix A: Methodology. https://static.
tti.tamu.edu/tti.tamu.edu/documents/mobility-report-2021-appx-a.pdf. Accessed July 13, 2022.
35. City of San Jose. https://gisdata-csj.opendata.arcgis.com/search?collection=Dataset. Accessed December 21,
2020.
36. DataSF. https://datasf.org/opendata/. Accessed December 21, 2020.
37. County of Los Angeles. Open Data. https://data.lacounty.gov/. Accessed December 21, 2020.
38. LACMTA. Los Angeles Transit Stops, April 2020. https://la-metro.maps.arcgis.com/apps/Minimalist/index.
html?appid=1fd006edf4cc446cad245c72241afba5. Accessed December 21, 2020.
39. Los Angeles Metro. https://developer.metro.net/gis-data/. Accessed December 21, 2020.
40. County of Long Beach. https://datalb.longbeach.gov/. Accessed December 21, 2020
41. Texas Department of Transportation. Texas—TPP. https://txdot.maps.arcgis.com/apps/webappviewer/
index.html?id=06fea0307dda42c1976194bf5a98b3a1. Accessed December 8, 2021.
42. Geedipally, S., M. Pratt, M. Ko, and K. Womack. Developing a Crash Analysis Tool to Address Pedestrian
Safety. Austin, TX: TDOT, 2016.
43. Utah Department of Transportation. UDOT Data Portal. https://data-uplan.opendata.arcgis.com/. Accessed
July 13, 2022.
44. ArcGIS. Estimated Pedestrian Volumes at Signalized Intersections (1,494) in Utah. https://www.arcgis.com/
apps/mapviewer/index.html?webmap=57b2a38e07e4446ab86efd4a257e3eff. Accessed December 8, 2021.
45. ArcGIS. Predicted Pedestrian Volumes at Intersections (62k) in Utah. https://www.arcgis.com/apps/
mapviewer/index.html?webmap=32af57787d864a7996af76d0d9055cb2. Accessed December 8, 2021.
46. Avelar, R. E., K. K. Dixon, and P. Escobar. Evaluation of Signalized-Intersection Crash Screening Methods
Based on Distance from Intersection. Transportation Research Record: Journal of the Transportation
Research Board 2514, no. 1 (January 2015): pp. 177–186. https://doi.org/10.3141/2514-19.
47. Lord, D., X. Qin, S. R. Geedipally. Highway Safety Analytics and Modeling. Amsterdam: Elsevier Publishing
Ltd., 2021.
ISBN 978-0-309-68770-6
90000
9 780309 687706