-
1
Factors Influencing the Perception of
Safety at Intersections Brittany
Kubinski
Introduction In 2012, a series
of accidents occurred during the
first weeks of the semester at
The Ohio State University prompting
the University to form a
Traffic Safety Task Force and to
address the significant traffic
safety issues on campus. While
the task force was initially
charged to provide an assessment
of safety on campus as it
relates to pedestrian and
vehicular behavior and provide solutions,
pedestrian and bicyclist safety on
campus can always be improved
and it is this situation that
has prompted a study into
perceived safety at intersections.
Within Columbus, there is no
area of greater need for
enhanced pedestrian and bicyclist
safety measures than Ohio State’s
campus. With a student population
of over 57,000 students and
growing, pedestrians are constantly
interacting with vehicles.
Many intersection designs and
characteristics exist in order to
enhance pedestrian safety. Pavement
markings of various designs, yield
to pedestrians and pedestrian
crossing signs, traffic islands and
pedestrian crossing refuges all exist
to aid in safe intersection
crossing. The aim of this
study is to understand which
factors are most important in
an intersections perceived safety.
In order to analyze pedestrian
perceptions of safety, a visual
preference survey has been conducted
that allows users to rank an
intersection over another one based
on how safe they perceive it
to be. StreetSeen, a free tool
that anyone can use to create,
collect and analyze data was
used to conduct a pairwise
visual survey. Using geo-‐tagged
images from Google Street View,
the perception of intersection
safety at The Ohio State
University in Columbus, Ohio, is
measured as perceived by people
across the world. The results
of this survey are then
categorized into the least
preferred and most preferred
intersections. The attributes that those
intersections are largely comprised
of are then tallied to
determine which attributes pedestrians
find most important in the
perceived safety of an intersection.
Data and Methodology This
section discusses the data set
used for this study, its
collection, and limitations. Following
the section on data, the
analysis methods are discussed. First
we describe how we score each
image using win/loss ratios
(calculated based on the number
of times an image is shown
and the number of times it
is selected over its paired
image). Based on this analysis,
we present and describe the
most and least favored images.
Data The survey conducted
in this research was designed
to obtain information on
individuals’ perception of safety for
intersections at The Ohio State
University and the corresponding
attributes of more preferred and
less preferred intersections. For the
visual preference survey, the survey
creator selects one or more
geographic areas and then
selects the number of random
images to be generated from
Google Street View.
-
2
The images were sourced from
Google Street View on The
Ohio State University campus in
Columbus, Ohio, USA. 50 images
were selected to be included in
the study to provide for a
wide of intersection designs and
types. Images were then
categorized based on specific
segment-‐level attributes, in a
spreadsheet. The images and dataset
used in the study can be
downloaded from
http://streetseen.osu.edu/studies/which-‐intersection-‐would-‐you-‐feel-‐the-‐safest-‐crossing/vote.
Perception data was collected
using StreetSeen, as shown in
Figure 1 of the Appendix.
Users were shown two images,
selected randomly from the
dataset, and asked to select one
in response to the question
“Which intersection would you feel
the safest crossing?” The survey
also collected data on their
location, based on the latitude
and longitude when completing the
survey as well as answers
to survey questions about gender,
age and commuting behaviors such
as: Thinking about a typical
day how many streets would
you estimate that you cross? How
do you typically get to
school or work? As well as
questions to determine if they
are currently enrolled at a
university. The locations of images
are illustrated in Figures 2, 3
and 4 in the Appendix.
The images were selected to
provide a variety of
segment-‐level attributes that affect
cycling preferences. These attributes and
the range of levels considered
for each attribute are listed
below.
• Crosswalk o Painted with ladder
markings o Painted with parallel
markings o None
• Segment Intersections o No intersections
o 3 way intersection o 4 way
intersection
• Condition of Road o Good (very
few bumps/cracks/holes) o Fair (some
bumps/cracks/holes) o Poor (many
bumps/cracks/holes) o Under Repair (clear
work is underway)
• Number of Lanes o 1 lane o 2
lanes o 3 lanes o 4+ lanes with
raised median o 4+ lanes without
raised median
• Traffic Control Devices o Stop Sign
o Traffic Light o None
-
3
• Crossing Aids in Segment o Pavement
Markings o Yield to Pedestrian Sign
o Pedestrian Signal o Median/Traffic
Island/Pedestrian Refuge o Curb Extension
o Overpass/Underpass o Pedestrian Crossing
Sign o Flashing Warning o Share the
Road Warning
• Cars Visible o None o Low (2 or
fewer cars) o Medium (3-‐5 cars)
o High (6-‐9 cars) o Very High
(10+ cars)
• Pedestrians Present • Bicycles Present
• Distance Crossed
o Not more than 50' o More than
50'
• Stopline Setback o Setback present
behind crosswalk o No setback
• Curb Cuts/Ramps o Curb cuts/ramp
present o No curb cut/ramp
• Condition of Crosswalk o Markings are
well maintained o Markings are faded
o No markings
• Speed Limit o >= 35 mph o
< 35 mph o Not Posted
Survey Administration More than
9,000 students enrolled and active
in TechniCity, a massive open
online course, were invited to
participate in the StreetSeen survey.
The survey was also distributed
on various social media websites
such as Facebook, Twitter and
LinkedIn.
-
4
A total of 203 people
responded to the survey (as
determined by unique IP addresses)
and contributed 2.710 votes for
preference of one image over
another. Each participant contributed
an average of 13 votes.
Voters from all over the
world contributed responses to this
survey. As seen in Table
1 of Appendix 2, a total
of 35 countries were represented
with 90 voters from the
United States making up the
largest portion of voters at
over 44 percent and contributing
over 46 percent of the total
votes. Voters from Spain (12),
Canada (10) and Brazil (9)
are responsible for six percent,
five percent and four percent
of the total votes, respectively.
Results This study has yielded
interesting results as to what
factors most influence perception of
safety at an intersection. Tables
2 and 3 in Appendix 2
lists the attributes that were
present in the most preferred
and least preferred intersections.
First, having a crosswalk for
pedestrians is essential in
enhancing their safety as well as
making them feel safe. For
the most preferred intersections (those
preferred over 50 percent of
the time) we found that
intersections with a ladder-‐patterned
crosswalk were most preferred, with
58 percent of those intersections
having a ladder style crosswalk.
Figures 5, 6, and 7 in
Appendix 1 were the most
preferred intersections, each having
been chosen 91 percent, 80
percent and 79 percent of the
time, respectively. The next
factor that leads individuals to
feel a particular intersection is
safe or unsafe is the
condition of the road and the
condition of the crosswalk markings.
Within the top 50 percent of
the intersections, 88 percent of
the roads were in good
condition (very few bumps, cracks
and holes) and 81 percent of
the intersections had well-‐maintained
crosswalk markings. This factor is
one that is easy enough to
control for and just implies
that the University should pay
close attention to the conditions
of the roads closest to
intersections and pedestrian crossing
areas and that crosswalk markings
should be well-‐maintained in order
to make pedestrians feel safe
when crossing an intersection.
Pavement markings were largely
the most popular pedestrian crossing
aid in use at these
intersections but the yield to
pedestrian and pedestrian crossing
signs along with pedestrian
signals were also prevalent.
Intersections with fewer lanes/distances
to cross were also among the
most preferred locations. Figures
5, 6, 7 illustrate how 2
lane roads and short crossing
distances were one of the
factors that lead to these
locations being highly preferred, while
Figures 8, 9, and 10
demonstrate how the least preferred
locations (chosen only 7 percent,
11 percent and 16 percent of
the time, respectively) all have
large crossing distances in common
as well as 4 or more
lanes of traffic to cross
without any pedestrian refuge or
traffic island to break up the
distance. Volume of traffic
was another factor that influenced
perception of safety. Images with
higher numbers of cars visible
within the image were chosen
less of the time than those
with no cars or lower numbers
of cars visible within the
image. The top 3 least preferred
intersections all
-
5
have higher levels of traffic and
the cars appear to be moving
at a faster speed than those
of the top 3 most preferred
intersections. Recommendations The
purpose of this study was to
identify which intersection attributes
enhance a pedestrians perception of
safety. By conducting this
survey, it is the hope of
the author to provide recommendations
that The Ohio State University
can use to prioritize funding
for intersection installations or
upgrades. In addition, by identifying
specific intersections on campus
that appear to be less safe,
the University can target its
efforts to improve these
intersections. One of the
most important factors in enhancing
perception of safety is to
reduce the distance to cross a
street as well as the number
of lanes and speed of traffic.
Considering the nature of the
university and its current
walkability, one recommendation would
be to reduce the number of
lanes for campus streets. However,
on streets where this cannot be
achieved, lowering the speed limit
and adding traffic islands and
pedestrian refuges may also help
to enhance perceptions of safety.
Additional pedestrian crossing
aids also help to give the
intersection an enhanced feeling
of safety. The most favored
intersections often had two or
more crossing aids in place
(pedestrian crossing and yield to
pedestrian signs, pedestrian signals
and crosswalk markings). Additional
crossing aids could also be
added to existing intersections such
as pedestrian overpasses and
underpasses, flashing warnings and
curb extensions. It is
apparent from this visual preference
survey that crosswalk markings
that are more prominent lead to
an enhanced perception of
safety. While ladder-‐style crosswalks
are one type of crosswalk design,
other designs and materials could
be utilized to create a more
distinct pathway for crossing
pedestrians. From this study,
we have been able to
determine that, indeed, well-‐maintained
roads and intersections with
additional pedestrian crossing aids,
as well as shorter distances
and less lanes all help to
increase pedestrian comfort. The
University should seek to take
these factors into consideration
when making infrastructure improvements
on campus to existing intersections
as well as when adding
additional crosswalks on campus.
-
6
Appendix 1: Figures
Figure 1: Screenshot of the
visual preference survey using Street
Seen
Source: streetseen.osu.edu
Figure 2: Map of locations
of images selected on OSU
Campus
Source: streetseen.osu.edu
-
7
Figure 3: Map of locations
of images selected on Woody
Hayes Drive/Woodruff Avenue
Source: streetseen.osu.edu
Figure 4: Map of locations
of images selected on Neil
Avenue
Source: streetseen.osu.edu
-
8
Figure 5: Image of most
preferred intersection, chosen 91% of
the time.
Source: http://www.streetseen.osu.edu/locations/52730
Figure 6: Image of second
most preferred intersection, chosen
80% of the time.
Source: http://www.streetseen.osu.edu/locations/52714
Figure 7: Image of third
most preferred intersection, chosen
79% of the time.
Source: http://www.streetseen.osu.edu/locations/52646
-
9
Figure 8: Image of least
preferred intersection, chosen 7% of
the time.
Source: http://www.streetseen.osu.edu/locations/52718
Figure 9: Image of second
least preferred intersection, chosen
11% of the time.
Source: http://www.streetseen.osu.edu/locations/52719
Figure 10: Image of third
least preferred intersection, chosen
16% of the time.
Source: http://www.streetseen.osu.edu/locations/52818
-
10
Appendix 2: Tables
Table 1: Number of Voters by
Country Country Number of Voters
Australia 4 Austria 1 Blank
10 Brazil 9 Canada 10 China
1 Croatia 1 France 3 Germany
3 Greece 6 India 8 Iraq 1
Ireland 2 Israel 1 Italy 2
Japan 2 Kazakhstan 1 Mexico
5 Netherlands 4 New Zealand 2
Pakistan 1 Peru 1 Poland 3
Portugal 3 Puerto Rico 1
Russia 2 Rwanda 1 Senegal 1
Singapore 2 Spain 12 Taiwan
1 Turkey 1 United Arab Emirates
3 United Kingdom 4 United
States 90 Venezuela 1 Total
203
-
11
Table 2: Number of Votes by
Country Country Number of Votes
Australia 60 Austria 15 Blank
109 Brazil 110 Canada 141
China 15 Croatia 15 France
45 Germany 32 Greece 77 India
66 Iraq 15 Ireland 25 Israel
15 Italy 30 Japan 21
Kazakhstan 16 Mexico 61 Netherlands
60 New Zealand 9 Pakistan 15
Peru 15 Poland 35 Portugal
45 Puerto Rico 15 Russia 30
Rwanda 15 Senegal 15 Singapore
31 Spain 158 Taiwan 15
Turkey 15 United Arab Emirates
31 United Kingdom 60 United
States 1263 Venezuela 15 Total
2710
-
12
Table 2: Most Preferred Intersection
Attributes Favored More than 50%
of the Time
Number of Locations 26 100%
Crosswalk
Painted with ladder markings
15 58% Segment Intersections
3 way intersection 11 42%
Condition of Road
Good (very few
bumps/cracks/holes) 23 88% Number
of Lanes
2 lanes 22 85% Traffic
Control Devices
Traffic Light 9 35% None
9 35%
Crossing Aids in Segment
Pavement Markings 24 92%
Yield to Pedestrian Sign 2 8%
Pedestrian Signal 9 35% Curb
Extension 1 4%
Pedestrian Crossing Sign 7 27%
Cars Visible
None 9 35% Pedestrians
Present 17 65% Bicycles Present
3 12% Distance Crossed
Not more than 50' 25
96% Stopline Setback
Setback present behind crosswalk
16 62% Curb Cuts/Ramps
Curb cuts/ramp present 24
92% Condition of Crosswalk
Markings are well maintained
21 81% Speed Limit
Not Posted 26 100%
-
13
Table 4: Least Preferred Intersection
Attributes Favored Less than 50%
of the Time
Number of Locations 24 100%
Crosswalk
Painted with parallel markings
13 54% Segment Intersections
4 way intersection 11 46%
Condition of Road
Good (very few
bumps/cracks/holes) 13 54% Number
of Lanes
2 lanes 11 46% 4+
lanes without raised median 8
33%
Traffic Control Devices Traffic
Light 10 42%
None 8 33% Crossing Aids in
Segment
Pavement Markings 14 58%
Yield to Pedestrian Sign 2 8%
Pedestrian Signal 10 42% Pedestrian
Crossing Sign 1 4%
Cars Visible Medium (3-‐5
cars) 9 38%
Pedestrians Present 9 38% Bicycles
Present 3 13% Distance Crossed
Not more than 50' 14
58% More than 50' 8 33%
Stopline Setback Setback present
behind crosswalk 11 46%
No setback 11 46% Curb
Cuts/Ramps
Curb cuts/ramp present 18
75% Condition of Crosswalk
Markings are faded 7 29%
No markings 8 33% Speed Limit
< 35 mph 1 4%
Not Posted 21 88%