SIMULATION BASED EVALUATION ON THE EFFECTS OF

JAYWALKING

by

Roy J. Wang

A thesis submitted to the Faculty of the University of Delaware in partial

fulfillment of the requirements for the degree of Master of Civil Engineering

Summer 2009

Copyright 2009 Roy J. Wang

All Rights Reserved
 ACKNOWLEDGMENTS

I would like to thank my advisor and mentor, Earl Rusty Lee. Without

his guidance, this thesis would never have been possible. Throughout my journey here

at the University of Delaware, he has given nothing but encouragement, praise, and

support.
I would also like to thank my family, friends, and loved ones. They have

always helped me in my time of need and given me their blessing in all my choices

throughout my life. Thank you for everything.

iii
 TABLE OF CONTENTS

ACKNOWLEDGEMENTS .......................................................................................... iii

TABLE OF CONTENTS .............................................................................................. iv
ABSTRACT ................................................................................................................... 1
CHAPTER 1 INTRODUCTION .................................................................................... 3
CHAPTER 2 LITERATURE REVIEW ......................................................................... 6
CHAPTER 3 DATA COLLECTION ........................................................................... 12
CHAPTER 4 METHODOLOGY ................................................................................. 18
CHAPTER 5 RESULTS............................................................................................... 25
CHAPTER 6 CONCLUSIONS AND FUTURE WORK ............................................. 30
BIBLIOGRAPHY ........................................................................................................ 34
APPENDIX A PEDESTRIAN DATA COLLECTION ............................................... 37
APPENDIX B TRAFFIC COUNT DATA .................................................................. 45
APPENDIX C CALCULATIONS ............................................................................... 52
APPENDIX D DATA AND HISTOGRAMS .............................................................. 68

iv
 ABSTRACT

Jaywalking is a behavior that exposes the pedestrian to a high-risk

condition and induces delay for vehicular traffic. The focus of most pedestrian

research has been on the issue of safety. The goal of this research is to measure the

impact on travel time and delay for vehicle traffic due to pedestrians who are crossing

traffic, including jaywalkers. By using a micro-simulation software, VISSIM, the

differences in travel time and delay time have been measured for a series of pedestrian

scenarios. Data was collected in Newark, DE along East Main Street, a section of

roadway serving many attractions near to the University of Delaware. Vehicle and

pedestrian volumes and pedestrian crossing locations were observed and recorded

during the midday peak hour.

Four scenarios were created in order to compare the effect of pedestrians

and jaywalkers. This first scenario was titled the Base Case scenario and contained

just the vehicle volumes and the signal timings for the main intersections. The Case 1

scenario introduced pedestrians at only the signalized crosswalks, which were located

at each major intersection. Next, the Case 2 scenario contained the marked midblock

crossings in addition to the already existing signalized crosswalks from Case 1.

Finally, the Case 3 scenario contained jaywalkers, where pedestrians were free to

cross wherever and whenever they appeared in the data collection. This scenario had

several paths along East Main Street to allow pedestrians to cross freely with the right

of way, in order to simulate jaywalking. In each case, pedestrian volumes were

balanced to ensure the same total number of pedestrians were crossing.

1
 The results showed that when pedestrians are legally crossing at the

signalized locations, they do not obstruct traffic. However, when pedestrians are

allowed to cross at the marked midblock crossings, there is a significant increase in

travel time and delay when compared to a network with no pedestrians. This was also

true when comparing the Base Case to the one with pedestrians and jaywalkers

scattered throughout. Finally, we compared the jaywalking network to the network

with pedestrians crossing legally at crosswalks and discovered that the addition of

jaywalkers does have a significant impact on the travel time and delay of oncoming

motorists.

2
 Chapter 1

INTRODUCTION

Regardless of the mode of transportation that is taken, all roadway users

hope to experience a minimal amount of delay and travel time. As a driver, one must

often yield to pedestrians, especially when making right turns, resulting in increased

delay. The same applies to pedestrians that arrive at an intersection and experience a

DONT WALK phase from the signal. These individuals are experiencing delays as

they yield to motorists that are currently traveling through the segment of roadway

which they intend to cross. As a result, a tradeoff exists between pedestrians and

drivers that balance the delay that they experience from the intersection or roadway

segment. With this in mind, some roadway users may attempt to reduce their

experienced delay by taking a more direct approach to reach their destination. A

pedestrian may decide to jaywalk if their destination is directly across the roadway,

there is no crosswalk nearby, and the vehicular flow is deemed suitable. This is mainly

true because pedestrian street-crossing behavior is responsive to the street

environment (Chu 2004).

Jaywalking is described in two ways. One is the act of crossing outside of

a delineated crosswalk. The other is the act of crossing within the crosswalk but

during a flashing or solid DONT WALK phase of the pedestrian signal. Akin and

Sisiopiku (2007) have provided similar definitions when discussing pedestrian

crossing compliance and have classified each as spatial and temporal crossing

3
compliance, respectively, and these classifications to describe jaywalking will be used

in this thesis.

Many past studies have been focused on pedestrian crossing behavior.

These studies have evaluated the effectiveness of crosswalks, in terms of location and

usage, as well as the efficiency of pedestrian signal phasing that informs the
pedestrians of when it is safe to traverse a crosswalk. In addition, most past research

analyzed the safety concerns for pedestrians when crossing the road. The size of the

pedestrian population makes it difficult to predict the behavior of each individual. One

common trait about the pedestrian population is their intolerance for delay.

If a pedestrian crosses the road away from a crosswalk, the driver is not

expecting an obstruction along the open roadway. As a driver approaches a crosswalk,

they are more aware of the possibility that there could be a pedestrian crossing at that

time. This is especially true with marked crosswalks, as they are generally associated

with traffic signals and located at intersections where drivers use greater caution and

lower driver speeds, when compared with unmarked crossings (Van Houten,

McCusker et al. 2003). In general it can be said that most spatial jaywalking is

occurring at the points furthest from the marked locations. The midblock dart-out [by

a pedestrian] is by far the most common pedestrian-related accident type, accounting

for one-third of all pedestrian accidents (Knoblauch, Tobey et al. 1984).

This research will take a different approach to the issue of jaywalking.

Rather than analyze the safety issues associated with jaywalking, this research will

determine the vehicular delays experienced by motorists due to pedestrian crossings.

A pedestrian establishes a minimum critical gap and associated safety

margin before crossing the street. As vehicles approach, each gap between them

4
commonly referred to as headway, is evaluated and the gap is either accepted or

rejected for crossing. If the gap is rejected, the next one is considered. This is known

as the Gap-Acceptance Theory (Palamarthy, Mahmassani et al. 1994). However in the

event that the pedestrian miscalculates the gap and decides to cross anyways, the

driver is forced to decelerate in order to avoid collision with the pedestrian. This
deceleration translates to delay that is experienced by the motorist, as well as

following vehicles that are also forced to slow down. This amount of delay will be

compared among different scenarios that incorporate a base case that involves no

pedestrians on the network, scenarios with pedestrians only along the designated

crosswalk areas, and the real-life scenario with pedestrians crossing freely. This will

give us the opportunity to determine the increase in delay associated with strictly

jaywalkers, when it is compared with the legally crossing pedestrians.

The basic premise of this research is that jaywalking causes the largest

increases in delay for the vehicles. This premise will be proved or disproved by the

data and analysis. Consideration of jaywalking as a delay source as well as a safety

issue may become a useful aspect for consideration by traffic engineers as they

evaluate transportation corridors.

5
 Chapter 2

LITERATURE REVIEW

As stated in the introduction, most of the past pedestrian related research

has been focused on pedestrian movement or crossing behavior. Often, these analyses

evaluate the issues of safety and delay experienced only by the pedestrians. Few have

included jaywalking.

Pedestrian behavior is generally unpredictable. Due to the diversity of

pedestrians in regards to age and physical ability, their characteristics differ from

group to group. However, it can be said that in general, all pedestrians are concerned

with travel time and delay. In fact, about 75% of pedestrians feel that they should have

to wait one minute or less before crossing the street upon arrival of the curb edge

(Fitzpatrick, Ullman et al. 2004). Increased or unexpected delays can result in more

frequent jaywalking. Jason and Liotta (1982) discovered that under non-facilitating

crossing conditions, pedestrians are more likely to jaywalk. Non-facilitating crossing

conditions occur when a pedestrian crosses one leg of an intersection and then must

wait 30 seconds before the WALK phase in order to cross the second leg. By

increasing the delay that the pedestrians experienced, more of them were willing to

cross regardless of the current phase.

Jaywalking can also occur when pedestrians cross the street outside of a

designated crosswalk. Akin and Sisiopiku (2007) conducted a study to determine the

pedestrian crossing compliance when approaching signalized crosswalks. While

performing the pedestrian count, the number of jaywalkers was also recorded in order

6
to determine the spatial compliance of each crosswalk, specifically the number of

pedestrians that were actually using it. They found that the average spatial

compliance rate for crossing within the crosswalk was only 83.1%. In addition, the

compliance of pedestrians to cross only when a WALK phase is encountered was

defined as the temporal compliance. It was determined that the average temporal
compliance rate was a mere 50.6%. This shows that pedestrians are often crossing

whenever they feel is more convenient, in order to reduce delay of waiting for a

WALK phase or traveling to the nearest crosswalk to traverse the road safely.

Although this study focuses on pedestrian crossing behavior and discusses jaywalking,

it does not analyze the effects that the behavior will have on approaching vehicles.

In another study, Kim, Brunner, and Yamashita (2008) analyzed

pedestrian and driver behavior in order to examine patterns of violation and

compliance in regards to crosswalks. Drivers were evaluated on whether or not they

fully yielded to pedestrians that are within the crosswalks, while pedestrians were

evaluated on whether they crossed within the crosswalk boundaries and if they obeyed

the pedestrian signal. The authors determined that the most common driver violation

occurred at unsignalized midblock crossings when drivers did not stop for pedestrians

waiting to cross at the crosswalk. Only 43% of drivers fully complied when yielding

to pedestrians, ensuring that they had cleared the road completely before passing

through the crosswalk. For pedestrians, it was found that the most common type of

violation was crossing outside of the crosswalk, also referred to in this paper as spatial

jaywalking. This study provided useful information in support of this research because

it noted specifically that jaywalking was the most common violation. Also, it analyzed

both pedestrian and driver behavior when approaching a crosswalk. This research

7
differs from Kim, Brunner et al. (2008) because rather than considering pedestrian and

driver behavior as two separate factors, it will determine the direct impact of

jaywalking on the delay experienced by the drivers.

From these studies, it could be determined that pedestrians were often

crossing at locations that they consider safe, regardless of whether or not a crosswalk
is present. Also, a significant amount of drivers are improperly yielding to pedestrians

at crosswalks. In fact, only 30.7% of vehicles wait for pedestrians to clear the

walkway before proceeding (DeVeauuse, Kim et al. 1999). As a result, a tradeoff

exists between delays experienced by pedestrians, and those experienced by drivers.

One study from the University of Science and Technology of China, Hefei

focused on the effects a pedestrian has on the flow of traffic. Jiang, Wu, and Li (2002)

conducted a study analyzing how a pedestrian crossing the roadway affected the flow

of traffic. They developed a mathematical model to determine the change in roadway

capacity as a pedestrian impedes the flow of traffic. In their study, the basic scenario

was of a pedestrian waiting to cross the road, while a vehicle is approaching. They

established that the pedestrian would cross if the speed of the approaching driver and

their distance from the pedestrian would provide enough time to safely cross the lane

of travel (assumed to be 0.5 sec). Also, they considered that in the event a vehicle

approaches a pedestrian in the roadway, the vehicle would stop, as if there is a

roadblock ahead, until the pedestrian exits the travel lane. After completing their

analysis, they discovered that if the average headway between vehicles is either small

or large, then the capacity of the roadway remains unchanged. However, in the event

that the headway between vehicles is intermediate, then the capacity is greatly affected

by the addition of pedestrians (Jiang, Wu et al. 2002). Although this study evaluated

8
the impact of pedestrians on the flow of traffic, it differs from this study in that

pedestrians that jaywalk are considered along with those that legally cross within an

intersection. Their study only focused on one lane of travel with one pedestrian path.

This research introduces numerous paths that pedestrians may take along the corridor

of interest. The research discussed in this thesis also gives the right of way to
pedestrians. This means that vehicles are forced to stop for individuals using the

crosswalk, rather than having the pedestrians evaluate each gap and determine if they

can cross in time.

Another study performed by Ishaque and Noland (2007) utilized VISSIM,

a micro-simulation model, to study the effects of signal cycle timings on delay and

travel time costs for both vehicles and pedestrians. They discovered that shorter cycle

timings generally benefit pedestrians. Although our project will utilize micro-

simulation software and involve a tradeoff between pedestrian/jaywalker and

vehicular delay, it is different from Ishaques work because this work is not an

optimization. This research relies on the current signal timings of the area of interest

and determining the effect of jaywalking on the delay when compared to pedestrian

traffic that crosses at the correct locations.

While researching information regarding pedestrian crossing behavior, it

was discovered that several studies are focused on the way that pedestrians perceive

their environment and what they would consider as safe or unsafe crossing locations.

These studies included Schneider, Khattak et al. (2002), Chu (2004), Fitzpatrick,

Ullman et al. (2004), Chu, Guttenplan et al. (2004), and Bernhoft and Carstensen

(2008). All of these studies were performed by conducting surveys or questionnaires

designed to determine what a pedestrian would consider safe, and whether or not this

9
is categorical of a certain age, gender, etc. Although this information is useful, it does

not directly relate to this research.

Another common focus of pedestrians studies are related to pedestrian

walking characteristics. These often center on factors such as walking time, start up

time, crossing location, age, gender, time of day, etc. Some of these studies include
Knoblauch, Tobey et al. (1984), Knoblauch, Pietrucha et al. (1996), Coffin and

Morrall (1995), Montufar, Arango et al. (2007), and Bowman and Vecellio (1994). All

of these studies were performed by simply observing pedestrians and determining

their characteristics as well as the surrounding environment. This information is useful

in trying to establishing generalized information about a specific group of people such

as males, females, the elderly, etc. Once again, though, this information is not directly

related to this study of how jaywalkers affect the flow of traffic.

Pedestrian delay is also a general focal point for pedestrian related studies.

Many of them evaluate the delays experienced by pedestrians in different roadway

environments. Some studies that were geared towards calculating this pedestrian delay

were Virkler (1998), Kruszyna, Mackiewicz et al. (2006), and Chu and Baltes (2002).

Although these studies provide useful information on pedestrian level of service and

delay, as well as measures taken in order to reduce delay, this research is mainly

geared toward the level of service and delay of the oncoming vehicles.

When researching pedestrian crossing behavior, another frequent theme

for study is related to pedestrian accidents and crash data. These studies compare the

results of pedestrian related accidents to the surrounding locations, whether alcohol

was involved, along with other factors. A few of these studies we came across were

Cui and Nambisan (2003), Baltes (1998), and Miles-Doan and Thompson (1999).

10
Unfortunately, this topic is not directly related to jaywalking or the effect it has on

approaching motorists.

Finally, another common topic related to pedestrians was centered on the

opinions of drivers and how they perceive pedestrians in the roadway or react to

pedestrian friendly devices. These studies included Mitman and Ragland (2007), Britt,
Bergman et al. (1995), and Van Houten, McCusker et al. (2003)

11
 Chapter 3

DATA COLLECTION

This study was conducted in Newark, DE, which is home to the

University of Delaware. About 20,000 students attend the university and the university

employs about 4,000 faculty and staff. As with any other university, there are a

significant number of individuals that utilize non-motorized transportation as their

means to commute to work or school. As a result, many of the roadways near the

university are pedestrian-friendly to provide safety for this pedestrian population.

From 2002-2007, the Newark Police Department reported 142 cases of

pedestrian related accidents, along with 39 jaywalking citations from the year 2002-

2006. Although these numbers may not seem significant, they do not include instances

where the accident was not severe enough to require police assistance. Out of the

reported accidents and citations, the majority of those involved are in the age range of

18-24. This falls in the age range of a typical, college student.

East Main Street was selected as the area of interest for this study of

pedestrian behavior. East Main Street is a one-way, two lane city street with on-street

parking on both sides of the road. There are several attractions along this stretch of

roadway that include restaurants, shops, among other locations that appeal to

pedestrians. This area guarantees some jaywalking occurrences based on personal

observation, its proximity to the campus and the numerous attractions.

Data was collected along the portion of East Main Street from N. Chapel

Street to N. College Avenue.

12
 Figure 1 Area of Interest

This section has seven marked crosswalks with three of them being unsignalized. The

three unsignalized crosswalks within this stretch are all midblock crossings that give

the pedestrian right of way and force approaching drivers to yield. Many of them have
pedestrian friendly devices such as extended curb edges, flashing beacons above, as

well as the appropriate signage to alert drivers to stop for pedestrians.

The four signalized locations occur at the three major intersections along

East Main Street. These intersections are East Main Street & N. Chapel Street, East

Main Street & Academy Street, and East Main Street & N. College Avenue. The

13
Academy Street and N. College Avenue intersections each contain one signalized

crosswalk. Because East Main Street is a one-way street heading westbound, the

crosswalks are placed prior to the signal so that vehicles turning onto westbound East

Main Street do not have to yield to pedestrians. The intersection at N. College Avenue

contains two marked, signalized crosswalks. The first crosswalk traverses East Main
Street prior to S. College Avenue, the extension of N. College Avenue that is shifted

east by 100 feet up East Main Street. The second crosswalk spans the entire

intersection of East Main Street at N. College Avenue as it is a pedestrian scramble. A

scramble is defined as a pedestrian crossing system that stops all traffic and allows

pedestrians to cross an intersection in every direction at the same time. This location

has a dedicated pedestrian phase that allows for all pedestrian movement in either

direction across the intersection.

Pedestrian movements and behavior were observed by means of manual

data collection. Volunteers were each assigned one of six locations that spanned the

area of interest. These six locations can be seen in Appendix A. They were each given

a pedestrian data collection packet that was relative to their assigned area. The data

collection packet included several copies of an aerial image of the location. Space was

provided for data collectors to illustrate the movement of a jaywalker. Arrows were

used to clearly mark the direction of travel as well as the start and finish location

of crossing. For each additional jaywalker that may have traversed the same path, a

tally was given next to the arrow. In order to accurately measure the location of

crossing, data collectors were instructed to note the crossing location of each

jaywalker in relation to the nearby buildings on East Main Street. This allowed them

14
to locate the building in the aerial image of their pedestrian data collection packet for

more accuracy.

For data collection locations that contained one of the marked,

unsignalized midblock crossings, there were spaces for the volunteers to count the

number of pedestrians that crossed within the crosswalk. These pedestrians were
separated into those that traveled northbound and those that traveled southbound. In

the event that a data collector was assigned one of the locations that housed a

signalized crosswalk, pedestrians were separated into two categories. These categories

were those that used the crosswalk while within the WALK phase, and those that did

not (temporal jaywalking). Once again, these pedestrians/jaywalkers were separated

by direction of travel (northbound vs. southbound). A completed data collection

sample can be found in Appendix A.

The pedestrian data collection period occurred during the midday peak

period (11:00 am 1:00 pm) on Wednesday, March 11, 2009 and Friday, March 13,

2009 during the universitys spring semester. The two-hour assigned period was

divided into eight 15-minute intervals for ease of counting. Data collectors were also

asked to stand in a location that is suitable for viewing the entire roadway within their

assigned area in order to ensure that no pedestrian or jaywalker was missed. In

addition, the observation period was performed regardless of the weather condition

(rain, sun, snow, etc).

Once the pedestrian data collection process was complete, the total

number of jaywalkers and pedestrians were then calculated. For each 15-minute time

interval, the number of jaywalkers and pedestrians were summed up across all six

locations. This was done for both Wednesday and Fridays results. With the totals

15
calculated, the percentage of jaywalkers for each time period was determined, in

relation to the total number of observed pedestrians/jaywalkers. The 11:45 AM -12:00

PM interval had the highest jaywalking percentage for both days. The Wednesday data

had a 44% jaywalking occurrence while the Friday data had a 46% jaywalking

occurrence. The volumes from this time period, both pedestrian and jaywalking, were
then multiplied by a factor of 4. The purpose of this is to recreate a peak hour volume

of jaywalking. These resulting numbers were then used as the pedestrian data for

modeling.

With the pedestrian modeling volumes complete, a vehicular data

collection was then performed in order to determine the approximate number of

vehicles that are on East Main Street during this peak hour. For this traffic count,

volunteers were assigned one of four locations along Main Street. These locations

were the three signalized intersections of East Main Street, within the corridor, as well

as one location further west on East Main Street. These counts provided information

for the traffic entering East Main Street at the west end as well as those entering or

exiting at each of the major intersections. The four locations can be seen in Appendix

B.

Each traffic counter was given a packet containing an aerial view of their

assigned intersection. Also, each packet contained a box to allow for the counting of

vehicles that make a specific turning movement. Only movements that were either

exiting or entering East Main Street as well as those traveling straight through the

intersection were considered. For the location upstream from the area of interest, the

only movement of interest was those traveling through. A data collection sample can

also be found in Appendix B.

16
 The time period in which traffic was counted was the one hour interval

between 11:30 AM and 12:30 PM. Once again, the data collection process was split

up into four 15-minute intervals for ease of counting. This selected time period falls

within the midday peak hour of pedestrian data collection, as well as contains the

calculated peak 15-minute jaywalking period. Vehicular data collection was

performed on Monday, April 6, 2009.

Upon completion of the traffic count, the totals of turning movements

were summed to create the hourly volume for East Main Street along with the various

turning movements. It was decided to include a single hour of data collection for

vehicular volumes because the goal is to determine the effects of jaywalking on the

approaching vehicular traffic versus legal pedestrian crossing behavior. As a result,

the vehicular volume which was used to evaluate this change should be affected

regardless of the volume.

17
 Chapter 4

METHODOLOGY

In order to determine the effect of pedestrian jaywalking on the flow of

traffic, a simulation model was created. Through the use of simulation, several

statistics can be generated in order to evaluate the effectiveness of a network or

roadway. This is done by analyzing the delay, level of service, and travel time.

The program that was chosen was VISSIM, a micro-simulation software

that is produced by PTV America, Inc. To our best knowledge, VISSIM is the only

micro-simulation that is able to simulate pedestrians as part of the transportation

system. The software provides the user with an opportunity to view each network in 3-

D. The 3-D feature of VISSIM also provides video representation of the 1st person

view of any vehicle or pedestrian in the network. This feature is wonderful for

presentation of a network or witnessing any queues first hand.

The first step of developing the model was to create a background aerial

image that would be large enough to cover the area of interest. This was done by

taking several screen shots of an aerial image provided by Google Earth and laying

them adjacently. Each image was slightly overlapped with the neighboring one in

order to ensure accuracy of the image. Once the image of the entire stretch of East

Main Street (from Library Ave to Elkton Rd) was developed, the file was saved and

then imported into VISSIM as a background image. The distance of the corridor of

East Main Street from beginning to end was then estimated using Google Earth once

18
again. This distance was then used in the simulation program in order to scale the

image so that it is consistent with the length of roadways to be created by VISSIM.

With the background in place, roadway links were added along East Main

Street, as well as the major side streets of N. Chapel Street, Academy Street, and N.

College Avenue matching them to the image. All roadway geometry was determined
by using various satellite and aerial imagery provided by Google Maps and Windows

Live Search Maps, as well as direct observations. This included the number of lanes

for each roadway, any turn bays that existed, as well as noting the locations of marked

crosswalks. In order to determine the length of any turn bays along East Main Street

or along the side streets, Google Earth was used to provide an approximate value to be

used in the simulation.

Finally, the signal timings for the three major intersections along the Main

Street corridor were implemented. These were obtained from the Delaware

Department of Transportation (DelDOT). Once they were coded into the network, the

traffic volumes obtained from the data collection procedure were inserted into the

model.

The traffic volumes were implemented into the network and then were

balanced in order to reflect the results that were obtained from the data collection.

Because vehicular data was collected along the three major intersections on East Main

Street, there were differences between the Main Street approach volumes of one

intersection and the departure volumes of the preceding intersection. For example, the

number of vehicles that left the N. Chapel Street intersection (whether by passing

through on East Main Street, or turning onto East Main Street from N. Chapel Street)

may not be equal to the volume on East Main Street that approaches the next

19
downstream intersection, Academy Street. The reason for this is due to the minor side

streets or parking lots that may reduce or add to the volume. These differences in

volume were calculated and modeled in VISSIM by creating a point along East Main

Street that allowed a small amount of traffic to either exit or enter the network. This

point was placed in between each intersection to ensure volume equilibrium

throughout the network so that the simulation matched the results of the vehicle data

collection.

Before adding the pedestrians into the network, four different scenarios

were developed.

The Base Case scenario consists of the network with only the vehicular

volume collected. Pedestrians were not included so that this scenario may act as the

baseline in order to compare results of the other scenarios.

The Case 1 scenario consists of the network with the vehicular volume, as

well as pedestrians located on the network. However, this scenario will only contain

pedestrians that used signalized intersections. These pedestrians will be crossing at the

signalized locations and are restricted to cross only during the pedestrian WALK

phase. This is to recreate an environment in which pedestrians do not cross illegally.

In order to have the analysis remain consistent, pedestrians that did not cross at these

signalized crosswalks were allocated to the nearest signalized crosswalk. Also, those

that crossed at a signalized crosswalk but did so against the WALK phase, were forced

to cross at the appropriate time. The resulting network contains the same amount of

pedestrians, as measured during data collection, all crossing during the WALK phase

at each signalized crosswalk. These crosswalks were located at East Main Street and

20
N. Chapel Street, East Main Street and Academy Street, and East Main Street and N.

College Avenue.

The Case 2 scenario is similar to the Case 1 scenario except that now, all

marked crosswalks are considered as valid crossing locations. This includes the three

midblock crosswalks along the stretch of East Main Street. Although these midblock
crosswalks are legal crossing locations for pedestrians, it was decided to dedicate this

as a separate scenario from the signalized crosswalks because at these locations,

pedestrians are given the right of way and approaching vehicles are forced to yield. As

a result, there is some factor of unexpectedness in terms of driver perceptions of a

pedestrian crossing. In the Case 1 scenario, pedestrians will only cross when drivers

are forced to stop due to the red phase of the traffic signal. This provides no conflict

between driver and pedestrian. However, in the Case 2 scenario, vehicles are forced to

stop on their own in order to give right of way to the pedestrian. Also, the pedestrians

in Case 2 will be allocated to the nearest marked crosswalk, regardless of whether it is

signalized or a midblock crossing. As a result, the pedestrian volumes are different

from the Case 1 scenario as they become more spread out throughout the network

along these additional points of crossing. The total pedestrian volume throughout the

entire network remains the same.

Finally, the Case 3 scenario introduces pedestrians that choose to jaywalk.

In this scenario, jaywalkers are modeled by placing several paths within the model

that traverse East Main Street, along with the signalized and marked midblock

crosswalks. These paths are strategically placed in the locations where jaywalkers

illegally crossed, as it appeared during data collection. For reference, the aerial image

of the surrounding buildings was used to place these jaywalking locations with some

21
degree of accuracy. In the simulation, all of these paths gave jaywalkers the right of

way and forced the approaching vehicles on East Main Street to yield. This simulates

an environment in which an approaching vehicle witnesses a jaywalker enter the

roadway and may be forced to stop. Since this scenario was designed to reflect the raw

data as collected, there was no need to reallocate pedestrians along the roadway.
Similar to Case 1 and Case 2, this scenario has the same pedestrian total volume

throughout the entire network. Every location at which a jaywalker was seen crossing

the road was added into the model with the appropriate volume. For jaywalkers that

crossed at signalized locations but against the WALK signal (temporal jaywalking),

jaywalking paths were laid overtop of the existing signalized crosswalks. These paths

were given no restriction of when to cross. The result had pedestrians, who

appropriately crossed during the WALK phase, being forced to wait until they were

given the WALK signal, while jaywalkers were free to cross whenever they

approached the intersection. Once again, these jaywalkers were always given the right

of way. In order to easily decipher between the two, jaywalkers were color coded as

pink in the simulation.

After creating all four scenarios using the witnessed pedestrian volumes

from the Wednesday data collection, they were all duplicated in order to test the

Friday data collection results except for the Base Case scenario. Since, the Base Case

scenario contains no pedestrians, it is not necessary to implement any Friday

pedestrian volumes. However for the Case 1 and Case 2 scenarios, the pedestrian

volumes were changed in order to reflect the Friday data collection. These volumes

were calculated in the same way as the Wednesday volumes, where Case 1 has

pedestrians allocated to the nearest signalized crosswalk and Case 2 has pedestrians

22
allocated to the nearest signalized/marked crosswalk. The Case 3 scenario, on the

other hand, follows a different approach. The jaywalking locations that were observed

on the Wednesday data collection may not have been in the same location as those that

were observed on the Friday data collection. As a result, after copying the Wednesday

Case 3 scenario, the jaywalking locations were adjusted in order to accurately reflect
the Friday data collection. Upon completion of all 8 networks (2 days, 4 scenarios

each), VISSIM was used to collect the statistics of interest.

In the simulation, it was decided to collect travel time and delay along

East Main Street. A virtual counting device was placed at the beginning and end of

East Main Street in each scenario that would determine both travel time and delay for

each vehicle that crosses the start and finish counters. To ensure that each counter was

placed in the same location between each case, one network was copied and altered in

order to create the next scenarios network. As a result, the building of each

successive network only entailed creating the additional crossing locations as well as

entering the new adjusted pedestrian/jaywalking volumes. The counters remained in

the same position between each network. Also, not all of the vehicles that are

generated and trigger the initial travel time and delay time counter on Main Street will

reach the final counter. The reason for this is because the intersection turning volumes

are randomly assigned by the percentage of those turning or traveling through.

Therefore, as a vehicle approaches one of the intersections there is only a percentage,

relative to the number of vehicles observed traveling thru versus turning left or right,

that it will travel onto the next intersection, where the same thing will take place. The

resulting percentage of vehicles that fully travel the length of East Main Street is

approximately 33%. With the travel time and delay counters in place, each of the 8

23
networks was simulated 10 times in order to get a large amount of data for analysis.

Each simulation generated differences in travel time and delay because vehicles and

pedestrians were randomly generated at different times according to the Poisson

Arrival Process.

24
 Chapter 5

RESULTS

After running the simulation, the data was analyzed from each run of each

scenario. This consisted of determining the travel time and delay of every vehicle that

crossed the start and finish counters along East Main Street. The average travel

time and delay time were then calculated and a comparison was done between the

different cases. These average values utilized all of the vehicle data across the ten

simulation runs per scenario. Also, the standard deviation was determined in order to

get an idea of the spread and range of travel time and delay time values.

With these values, we were able to determine a few different results.

When comparing the Base Case scenario of no pedestrians on the network with the

Case 1 scenario of pedestrians allocated strictly to the signalized crosswalks, we saw

that there was very little change in travel time and delay. This was consistent for both

the Wednesday and Friday simulations. In addition, the standard deviations between

the two cases were also nearly the same for both days as well.

The results from the comparison between the Base Case and Case 1

scenarios are as expected. By including only pedestrians that are allocated at only the

signalized crossing locations, we should not expect an increase in travel time and

delay. The pedestrians that are crossing at the signalized crossing locations are

crossing during the appropriate WALK phase. As a result, the approaching vehicles

along East Main Street experience no additional delay as a result of these pedestrians.

The WALK phase of each of the signal timings is purposely given with the intention

25
of minimizing delay. For the intersections along the corridor at N. Chapel Street and

Academy Street, the pedestrian WALK phase for the crosswalk that traverses East

Main Street is coordinated with each of the side streets. Pedestrians in the model will

never cross East Main Street when there is traffic approaching and will only cross

when the vehicles are given the red phase from the signal timings. For the intersection
at N. College Avenue, pedestrians are given their own dedicated WALK phase. This

allows them to cross the scramble in any direction without disrupting traffic flow. This

dedicated WALK phase is also included in the Base Case scenario. As a result,

pedestrians throughout the network in Case 1 are crossing when the East Main Street

traffic is stopped due to the signals.

A Two Sample T-Test for Large Populations was used in order to

determine if the differences in values between the Base Case scenario (the control

model) and any of the other cases were statistically significant or just a product of

variation from the Base Case. When using this statistical procedure, we set a null

hypothesis which is to be evaluated and either accepted or rejected. The null

hypothesis was that the means of the travel times or delay times had no difference. In

other words, any observed variation was just due to statistical scatter. We will be

evaluating the relationships between cases with a 95% confidence interval.

After performing the analysis, we were able to accept the null hypotheses

between the Base Case and Case 1 and state that with 95% confidence, both the

Wednesday and Friday Case 1 travel time and delay time are not statistically

significant when compared to the Base Case scenario. This evidence supports our

previous statement that the pedestrians are not causing additional delay when they are

26
allocated to signalized intersections and forced to cross East Main Street during the

appropriate WALK phase at their location.

When comparing the Case 2 simulation results to the Base Case result, we

discover that there are significant changes in delay and travel time. As a reminder, the

Case 2 scenario implements pedestrians that cross legally at all marked crosswalks.
This includes the three midblock crossings on East Main Street along with the four

signalized crosswalks from the Case 1 scenario. After taking the average travel time

and delay times for all the vehicles within the 10 simulation runs for both days of

pedestrian data collection, we discovered a statistically significant increase in these

values.

We believe that this additional delay and travel time is attributed mainly

to the pedestrians that are crossing along the midblock crosswalks. These pedestrians

are given the right of way and force approaching vehicles to stop. By introducing them

into the network, we are creating additional stopping points for some vehicles, in

addition to the traffic lights already established from the Base Case scenario.

The final scenario introduces the jaywalkers, as they are witnessed from

the data collection process. As a reminder, the Case 3 scenario contains several paths

along the network in which jaywalkers are free to cross as soon as they approach the

roadway. Also, the Case 3 scenario still maintains the signalized and marked midblock

crossings from Case 1 and 2. The Case 3 scenario more accurately represents the

walking patterns of the witnessed pedestrians because not all pedestrians cross at the

appropriate locations within the appropriate signal phase (if applicable). After viewing

the Case 3 simulation results, we discovered an even larger increase in travel time and

delay time when compared to any of the other cases.

27
 We believe that the Case 3 results were significantly increased as a result

of the jaywalkers within the network. By introducing additional paths in which

jaywalkers are free to cross, vehicles have the opportunity to encounter more stopping

locations. The standard deviation is high due to the large range of travel time and

delay times that are possible. In the Case 3 network, some vehicles may be lucky and
may neither come upon any pedestrians while traveling through the East Main Street

network, nor approach any red signals from the signalized intersections. On the other

hand, some vehicles may not be as lucky and may be forced to yield to several

jaywalkers on the network in addition to having to stop for a red phase. Also, this may

explain the significant increase in travel time and delay. In order to get a better grasp

of the results, statistical analysis was performed. The statistical results indicate that

there are significant increases in both travel time and delay time for both Wednesday

and Friday, with 95% confidence. Thus, we can now state that, with 95% confidence,

jaywalking causes significant increases to travel time and delay.

However, this result compares the increase in travel time and delay from a

network with pedestrians crossing freely to a network that lacks any pedestrian inputs.

Therefore in order to accurately state that the addition of strictly jaywalkers to a

pedestrian network increases travel time and delay, we compared the Case 2 data to

the Case 3 data.

The results show that there is a significant increase in travel time and

delay time for both Wednesday and Friday data. Therefore, we can now state that,

with 95% confidence, when comparing a network in which pedestrians are strictly

enforced to cross only in designated crossing areas, the introduction of jaywalking

significantly increases the expected travel time and delay time for approaching

28
motorists. All data analyses, histograms and results of T tests are shown in

Appendices C and D.

29
 Chapter 6

CONCLUSIONS AND FUTURE WORK

The use of simulation modeling has proven to be truly effective in our

study. Through this tool, we were able to compare different scenarios involving the

implementation of pedestrians and how they affect the flow of traffic. After

completion of our analysis, we were able to discover several different conclusions.

The first conclusion that we discovered was that when pedestrians are

implemented strictly at signalized intersections, there is no significant extra delay or

travel time experienced by the approaching drivers. Because these pedestrians are

forced to cross at signalized crosswalks during the appropriate WALK phase, they do

not interfere with the flow of traffic along East Main Street.

The second conclusion that we discovered was that when pedestrians are

placed along marked, midblock crosswalks and signalized crosswalks, the travel time

and delay time significantly increases. When pedestrians crossed at these midblock

locations, they are given the right of way and drivers are forced to yield to them. This

then adds to the travel time and delay.

After comparing the strictly vehicular network (Base Case scenario) to the

network with jaywalkers as they appeared during data collection (Case 3 scenario),

this led us to our next conclusion. This conclusion is that the introduction of

jaywalkers to a vehicular network significantly increases travel time and delay time.

As jaywalkers were introduced into the network, they were allowed to freely cross at

any location along our corridor of East Main Street. This included any locations at

30
signalized intersections. As a jaywalker approached East Main Street, in the model,

they were given the right of way and forced the vehicles to yield to them. For the

signalized intersections, vehicles that experienced the green phase of their signal were

still forced to yield to any jaywalkers in the event that they were crossing at the same

time. Also, the results for this network yielded a wide range of values. This is due to
the variety of possibilities of a vehicle intersecting paths with a jaywalker along the

network. With the introduction of several jaywalking paths, one vehicle may

experience no jaywalkers or pedestrians while driving, while another vehicle may be

stopped at every jaywalking/pedestrian location.

The results of comparing a network with pedestrians crossing at marked

and signalized crosswalks (Case 2 scenario) proved to cause a significant increase in

travel time and delay when compared to the same network with just vehicular data

(Base Case scenario). The same result was true when we compared the network with

jaywalkers (Case 3 scenario) to the vehicle network (Base Case scenario). In order to

measure the impact of introducing jaywalkers to the network, a comparison between

the network with pedestrians at marked crosswalks (Case 2 scenario) and the network

with jaywalkers was conducted (Case 3 scenario). The results showed that the

introduction of jaywalkers to a pedestrian network with legal crossing behavior

significantly increased travel time and delay.

Before conducting this study, our hypothesis stated that the jaywalking

would have an effect on the flow of traffic and the delay experienced by approaching

motorists. After performing the study, we have discovered that our hypothesis was

correct. The impact of introducing jaywalkers into a vehicular network showed a

significant increase in travel time and delay for approaching vehicles. Also, the

31
redistribution of a portion of pedestrians at signalized and marked midblock

crosswalks to jaywalkers showed a significant increase in travel time and delay time

for approaching motorists as well.

Our study provided successful and informative results. However, there can

always be changes made in order to strengthen the study and conclusions. One
possibility is to consider pedestrian gap acceptance within our VISSIM model. The

Case 3 scenario allows jaywalkers to cross the road freely without worry about

approaching vehicles stopping in time. However, this is not an accurate representation

of pedestrian behavior. As mentioned earlier, the pedestrians follow a Gap-

Acceptance Theory in which each gap or headway between cars is evaluated and if the

pedestrian feels that he or she cannot cross the roadway in time, it is rejected and the

next gap is considered (Palamarthy, Mahmassani et al. 1994). Future work could

implement this factor into the model in order to better represent this pedestrian

crossing behavior.

VISSIMs 3-D visualization capability allows for one to create buildings

and objects while projecting photos onto a face of a building, in order to enhance the

3D view of the simulation. Future work on our model could include adding all of the

buildings along the Main Street corridor as they appear in real life. Although this

improvement would be purely aesthetic, it would improve the visualization.

Another possibility of future work would be to evaluate the efficiency of

creating different pedestrian alternatives in order to reduce jaywalking occurrences.

Now that it has been determined that jaywalking increases delay, a planner who is

trying to reduce delay in a corridor may consider implementing overpass walkways or

some other form of pedestrian device that may be beneficial.

32
 Finally, as with any study, the evaluation of the effect of jaywalking on

the flow of traffic may be expanded to include larger networks or several other

networks in order to determine if the results remain consistent throughout. Also, one

may decide to consider locations with higher pedestrian volume, such as urban cities

or other college campuses, or lower pedestrian volume, such as suburban areas. Once
again these results may determine if the effect of jaywalking on the flow of traffic

remains consistent regardless of environment or area of study.

33
 BIBLIOGRAPHY

Akin, D. and V. P. Sisiopiku (2007). Pedestrian Crossing Compliance Characteristics

At-Grade Signalized Crosswalks: Case Study in a Downtown-University
Campus Environment. 86th Annual Meeting, Transportation Research Board,
Washington D.C.

Baltes, M. R. (1998). "Descriptive Analysis of Crashes Involving Pedestrians in

Florida, 1990-1994 " Transportation Research Record: Journal of the
Transportation Research Board 1636: 138-145.

Bernhoft, I. M. and G. Carstensen (2008). "Preferences and Behaviour of Pedestrians

and Cyclists by Age and Gender." Transportation Research Part F: Traffic
Psychology and Behaviour 11(2): 83-95.

Bowman, B. L. and R. L. Vecellio (1994). "Pedestrian Walking Speeds and Conflicts

at Urban Median Locations." Transportation Research Record: Journal of the
Transportation Research Board 1438: 67-73.

Britt, J. W., A. B. Bergman, et al. (1995). "Law Enforcement, Pedestrian Safety, and
Driver Compliance with Crosswalk Laws: Evaluation of a Four-Year
Campaign in Seattle." Transportation Research Record: Journal of the
Transportation Research Board 1485: 160-167.

Chu, X. (2004). Testing Behavioral Hypotheses on Street Crossing. 83rd Annual

Meeting, Transportation Research Board, Washington D.C.

Chu, X. and M. R. Baltes (2002). Measuring Pedestrian Quality of Service for Mid-
block Street Crossing: The Selection of Potential Determinants. 81st Annual
Meeting, Transportation Research Board, Washington D.C.

Chu, X., M. Guttenplan, et al. (2004). "Why People Cross Where They Do: The Role
of Street Environment." Transportation Research Record: Journal of the
Transportation Research Board 1878: 3-10.

Coffin, A. and J. Morrall (1995). "Walking Speeds of Elderly Pedestrians at

Crosswalks." Transportation Research Record: Journal of the Transportation
Research Board 1487: 63-67.

34
Cui, Z. and S. S. Nambisan (2003). "Methodology for Evaluating the Safety of
Midblock Pedestrian Crossings." Transportation Research Record: Journal of
the Transportation Research Board 1828: 75-82.

DeVeauuse, N., K. Kim, et al. (1999). "Driver Compliance with Stop Signs at
Pedestrian Crosswalks on a University Campus." Journal of American College
Health 47(6): 269-74.

Fitzpatrick, K., B. Ullman, et al. (2004). On-Street Pedestrian Surveys of Pedestrian

Crossing Treatments. 83rd Annual Meeting, Transportation Research Board,
Washington D.C.

Ishaque, M. M. and R. B. Noland (2007). "Trade-offs Between Vehicular and

Pedestrian Traffic Using Micro-Simulation Methods." Elsevier: Transport
Policy 14: 124-138.

Jason, L. A. and R. Liotta (1982). "Pedestrian Jaywalking Under Facilitating and Non-
Facilitating Conditions." Journal of Applied Behavior Analysis 15(3): 469-
473.

Jiang, R., Q. Wu, et al. (2002). "Capacity Drop Due to the Traverse of Pedestrians."
Physical Review E 65: 036120.1-036120.5.

Kim, K., I. M. Brunner, et al. (2008). "Modeling Violation of Hawaii's Crosswalk

Law." Elsevier: Accident Analysis & Prevention 40: 894-904.

Knoblauch, R. L., M. T. Pietrucha, et al. (1996). "Field Studies of Pedestrian Walking

Speed and Start-Up Time." Transportation Research Record: Journal of the
Transportation Research Board 1538: 27-38.

Knoblauch, R. L., H. N. Tobey, et al. (1984). "Pedestrian Characteristics and Exposure

Measures." Transportation Research Record: Journal of the Transportation
Research Board 959: 35-41.

Kruszyna, M., P. Mackiewicz, et al. (2006). "Influence of Pedestrians' Entry Process

on Pedestrian Delays at Signal-Controlled Crosswalks." Journal of
Transportation Engineering, ASCE 132(11): 855.

Miles-Doan, R. and G. Thompson (1999). "The Planning Profession and Pedestrian

Safety: Lessons From Orlando." Journal of Planning Education and Research
18(3): 211-220.

35
Mitman, M. F. and D. R. Ragland (2007). "Crosswalk Confusion: More Evidence
Why Pedestrian and Driver Knowledge of the Vehicle Code Should Not Be
Assumed." Transportation Research Record: Journal of the Transportation
Research Board 2002: 55-63.

Montufar, J., J. Arango, et al. (2007). "Pedestrians' Normal Walking Speed and Speed
When Crossing a Street." Transportation Research Record: Journal of the
Transportation Research Board 2002: 90-97.

Palamarthy, S., H. S. Mahmassani, et al. (1994). Models of Pedestrianm Crossing

Behavior at Signalized Intersections. Center for Transportation Research,
University of Texas at Austin: 1296-1.

Schneider, R. J., A. J. Khattak, et al. (2002). Factors Associated with Pedestrian Crash
Risk: Integrating Risk Perceptions and Police-Reported Crashes. 81st Annual
Meeting, Transportation Research Board, Washington D.C.

Van Houten, R., D. McCusker, et al. (2003). "An Examination of the Use of Advance
Yield Markings and Flourescent Yellow Green RA4 Signs at Crosswalks with
Uncontrolled Approaches." Transportation Research Record: Journal of the
Transportation Research Board 1818: 119-124.

Virkler, M. R. (1998). "Pedestrian Compliance Effects on Signal Delay."

Transportation Research Record: Journal of the Transportation Research Board
1636: 88-91.

36
 Appendix A

PEDESTRIAN DATA COLLECTION

37
SAMPLE

38
 Time Period:

Location 1: Main St @ N. College Ave

LIMIT OF
LIMIT OF WORK WORK

LIMIT OF WORK
N. College
Peds in Crosswalk: Jaywalkers (against WALK)
N SB N S
LIMIT OF
WORK

39
 Time Period:

Location 2: Main St @ S. College Ave

LIMIT OF
LIMIT OF WORK WORK

S. College Crosswalk
Peds in Crosswalk: Jaywalkers (against WALK)
LIMIT OF
LIMIT OF N S NB S WORK
WORK

40
 Time Period:
Location 3: Main St @ The Green
LIMIT OF LIMIT OF WORK
WORK

Peace a
Pizza/Natl
Guard

Pedestrians in Crosswalk:
LIMIT OF NB SB
WORK
LIMIT OF WORK

41
 Time Period:

Location 4: Main St @ The Galleria

LIMIT OF WORK
LIMIT OF WORK

Wilmington
Trust
Rainbow

Pedestrians in Crosswalk:
LIMIT OF WORK NB SB

LIMIT OF WORK

42
 Time Period:

Location 5: Main St @ Academy Street LIMIT OF WORK

LIMIT OF WORK

Caf Center
Gelato Street

Pedestrians in Crosswalk: Jaywalkers (against WALK)

LIMIT OF WORK NB SB NB SB
LIMIT OF WORK

43
 Time Period:
Location 6: Main St @ Haines St
LIMIT OF WORK
LIMIT OF WORK

Panera
Bread
Happy
Harrys

Starbucks/
Hollywood
Tans

Pedestrians in Crosswalk:
LIMIT OF WORK NB SB LIMIT OF WORK

44
 Appendix B

TRAFFIC COUNT DATA

45
SAMPLE
46
 Time Period:

Location 1: Main St @ N. College Ave

Turning Movements

CARS TRUCKS

Trabant
NB

47
 Time Period:

Location 2: Main St @ S. College Ave

Turning Movements

CARS TRUCKS

Trabant

NB

48
 Time Period:

Location 3: Main St @ Academy St.

Turning Movements

CARS TRUCKS

NB

49
 Time Period:

Location 4: Main St @ Chapel St.

Turning Movements

CARS TRUCKS

50
Location 5: Main St Beginning Time Period:

Turning Movements

CARS TRUCKS

51
 Appendix C

CALCULATIONS

Significance Testing Wednesday, Travel Time

Case 1 scenario vs. Base Case scenario:

217.8709 218.385
z=
(38.63904) 2 (39.15871) 2
+
2966 2983

.5141
z=
.5034 + .5140

z = .5097

z = .5097 = .5097

.5097 < 1.96

Accept H0, Case 1 is not statistically significant compared to the

Base Case

52
Case 2 scenario vs. Base Case scenario:

232.1958 218.385
z=
(39.488) 2 (39.15871) 2
+
2983 2983

13.8108
z=
.5227 + .5140

z = 13.5636

z = 13.5636 = 13.5636

13.5636 > 1.96

Reject H0, Case 2 is statistically significant compared to the Base

Case

53
Case 3 scenario vs. Base Case scenario:

304.1609 218.385
z=
(66.75261) 2 (39.15871) 2
+
2832 2983

85.7759
z=
1.5734 + .5140

z = 59.368

z = 59.368 = 59.368

59.368 > 1.96

Reject H0, Case 3 is statistically significant compared to the Base

Case

54
Case 3 scenario vs. Case 2 scenario:

304.1609 232.1958
z=
(66.75261) 2 (39.488) 2
+
2832 2983

71.9651
z=
1.5734 + .5227

z = 49.7063

z = 49.7063 = 49.7063

49.7063 > 1.96

Reject H0, Case 3 is statistically significant compared to Case 2

55
Significance Testing Friday, Travel Time

Case 1 scenario vs. Base Case scenario:

217.7652 218.385
z=
(38.9673) 2 (39.15871) 2
+
2987 2983

.6198
z=
.5084 + .5140

z = .6130

z = .6130 = .6130

.6130 < 1.96

Accept H0, Case 1 is not statistically significant compared to the

Base Case

56
Case 2 scenario vs. Base Case scenario:

232.5003 218.385
z=
(36.2680) 2 (39.15871) 2
+
2973 2983

14.1153
z=
.4424 + .5140

z = 14.4328

z = 14.4328 = 14.4328

14.4328 > 1.96

Reject H0, Case 2 is statistically significant compared to the Base

Case

57
Case 3 scenario vs. Base Case scenario:

450.8678 218.385
z=
(134.4797) 2 (39.15871) 2
+
2545 2983

232.4828
z=
7.106 + .5140

z = 84.219

z = 84.219 = 84.219

84.219 > 1.96

Reject H0, Case 3 is statistically significant compared to the Base

Case

58
Case 3 scenario vs. Case 2 scenario:

450.8678 232.5003
z=
(134.4797) 2 (36.2680) 2
+
2545 2973

218.3675
z=
7.106 + .4424

z = 79.4805

z = 79.4085 = 79.4085

79.4085 > 1.96

Reject H0, Case 3 is statistically significant compared to Case 2

59
Significance Testing Wednesday, Delay Time

Case 1 scenario vs. Base Case scenario:

101.9127 102.6339
z=
(38.54285) 2 (39.33411) 2
+
2966 2983

.7212
z=
.5009 + .5187

z = .7142

z = .7142 = .7142

.7142 < 1.96

Accept H0, Case 1 is not statistically significant compared to the

Base Case

60
Case 2 scenario vs. Base Case scenario:

116.4134 102.6339
z=
(39.35716) 2 (39.33411) 2
+
2983 2983

13.7795
z=
.5193 + .5187

z = 13.5251

z = 13.5251 = 13.5251

13.5251 > 1.96

Reject H0, Case 2 is statistically significant compared to the Base

Case

61
Case 3 scenario vs. Base Case scenario:

188.4669 102.6339
z=
(66.86003) 2 (39.33411) 2
+
2832 2983

85.833
z=
1.5785 + .5187

z = 59.2702

z = 59.2702 = 59.2702

59.2702 > 1.96

Reject H0, Case 3 is statistically significant compared to the Base

Case

62
Case 3 scenario vs. Case 2 scenario:

188.4669 116.4134
z=
(66.86003) 2 (39.35716) 2
+
2832 2983

72.0535
z=
1.5785 + .5193

z = 49.7477

z = 49.7477 = 49.7477

49.7477 > 1.96

Reject H0, Case 3 is statistically significant compared to Case 2

63
Significance Testing Friday, Delay Time

Case 1 scenario vs. Base Case scenario:

101.7244 102.6339
z=
(38.90596) 2 (39.33411) 2
+
2987 2983

.9095
z=
.5068 + .5187

z = .8981

z = .8981 = .8981

.8981 < 1.96

Accept H0, Case 1 is not statistically significant compared to the

Base Case

64
Case 2 scenario vs. Base Case scenario:

116.6793 102.6339
z=
(36.44468) 2 (39.33411) 2
+
2973 2983

14.0454
z=
.4468 + .5187

z = 14.2944

z = 14.2944 = 14.2944

14.2944 > 1.96

Reject H0, Case 2 is statistically significant compared to the Base

Case

65
Case 3 scenario vs. Base Case scenario:

335.3826 102.6339
z=
(134.7223) 2 (39.33411) 2
+
2545 2983

232.7487
z=
7.1317 + .5187

z = 84.1483

z = 84.1483 = 84.1483

84.1483 > 1.96

Reject H0, Case 3 is statistically significant compared to the Base

Case

66
Case 3 scenario vs. Case 2 scenario:

335.3826 116.6793
z=
(134.7223) 2 (36.44468) 2
+
2545 2973

218.7033
z=
7.1317 + .4468

z = 79.4445

z = 79.4445 = 79.4445

79.4445 > 1.96

Reject H0, Case 3 is statistically significant compared to Case 2

67
 Appendix D

DATA AND HISTOGRAMS

WEDNESDAY
Base Case1 Case2 Case3
Averages Averages Averages Averages
TT Delay TT Delay TT Delay TT Delay
Average 218.385 102.6339 217.8709 101.9127 232.1958 116.4134 304.1609 188.4669
Std. Deviation 39.15871 39.33411 38.63904 38.54285 39.488 39.35716 66.75261 66.86003
Max 326.39 209.98 324.44 209.21 350.87 234.46 488.5 375.3
Min 126.41 12.79 123.81 11.97 139.41 24.86 164.52 49.29

FRIDAY
Base Case1 Case2 Case3
Averages Averages Averages Averages
TT Delay TT Delay TT Delay TT Delay
Average 218.385 102.6339 217.7652 101.7244 232.5003 116.6793 450.8678 335.3826
Std. Deviation 39.15871 39.33411 38.9673 38.90596 36.26803 36.44468 134.4797 134.7223
Max 326.39 209.98 325.92 208.23 338.71 220.1 754.13 639.64
Min 126.41 12.79 127.26 12.48 146.76 30.29 177.56 61.94

68
 Base Case Travel Time Wednesday Case 1 Travel Time

700 700
600 600
500 500

Frequency
Frequency

400 400
300 300
200 200
100 100
0 0
110

170

230
290

350

410
470

530

590

650

710

770

830

890

950

1010

110

170

230
290

350
410

470

530

590

650
710

770
830

890

950

1010
Travel Time
Travel Time

Wednesday Case 2 Travel Time Wednesday Case 3 Travel Time

700 400
600 350
500 300

Frequency
Frequency

250
400
200
300
150
200 100
100 50
0 0
110

170
230
290

350
410
470

530
590

650
710
770

830
890
950

1010
110

170

230
290

350

410
470

530

590

650

710

770

830

890

950

1010

Travel Time Travel Time

69
 Frequency Frequency

0
100
200
300
400
500
600
700
0
100
200
300
400
500
0 600
0

60 60

120 120

180 180

240 240

300 300

360 360

420 420

Delay
Delay
480 480
Base Case Delay

Wednesday Case 2 Delay

540 540

600 600

660 660

720 720

780 780

840 840

900 900

70
Frequency Frequency

0
50
100
150
200
250
300
350
400
0
100
200
300
400
500
600

0 0
60 60
120 120
180 180
240 240

300 300

360 360

420 420
Delay
Delay

480 480

540 540
Wednesday Case 1 Delay

Wednesday Case 3 Delay

600 600

660 660

720 720

780 780

840 840

900 900
 Base Case Travel Time Friday Case 1 Travel Time

700 700
600 600
500 500
Frequency

Frequency
400 400
300 300
200 200
100 100
0 0
110

170
230
290

350
410
470

530
590

650
710
770

830
890
950

1010

110

170
230
290

350
410
470

530
590

650
710
770

830
890
950

1010
Travel Time Travel Time

Friday Case 2 Travel Time Friday Case 3 Travel Time

700 180
600 160
140
500
Frequency

Frequency
120
400 100
300 80
60
200
40
100 20
0 0
110

170
230

290

350

410

470

530

590

650
710

770

830

890
950

1010

110

170

230
290

350

410

470

530

590

650

710
770

830

890

950

1010
Travel Time Travel Time

71
 Frequency Frequency

0
100
200
300
400
500
600
700
0
100
200
300
400
500
0 0 600

60 60

120 120
180 180

240 240
300 300
360 360

420 420

Delay
480 Delay 480
Base Case Delay

Friday Case 2 Delay

540 540
600 600
660 660

720 720
780 780
840 840

900 900

72
Frequency Frequency

0
20
40
60
80
100
120
140
160
0
100
200
300
400
500
600

0 0

60 60
120 120
180 180

240 240
300 300
360 360

420 420
Delay
Delay

480 480
Friday Case 3 Delay
Friday Case 1 Delay

540 540
600 600
660 660

720 720
780 780
840 840

900 900