Evaluating the Transport Impacts of Transit Oriented … · 2011-10-27 · on TOD evaluation 206 11.2.2 Develop a methodology for evaluating the transport impacts of TODs 206 11.2.3

Evaluating the Transport Impacts of Transit

Oriented Developments (TODs)

Deepti Sadashiv Muley

B.E. (Civil), M. Tech (Transportation)

A thesis submitted for the degree of

Doctor of Philosophy

School of Urban Development

Faculty of Built Environment & Engineering

Queensland University of Technology

July 2011

Dedicated to my parents,

Aai and Baba

without your dreams, support and trust in me this was impossible

Deepti Muley Page i

Keywords

transit oriented developments (TODs), Australian TOD, TOD evaluation, transport impacts,

TOD users, TOD residents, travel characteristics, traffic generation, comparative analysis,

logistic regression, travel mode investigation, travel demand for TODs, travel demand

analysis, sustainable transport

Evaluating the transport impacts of TODs

Deepti Muley Page ii

Deepti Muley Page iii

Abstract

Sustainable transport has become a necessity instead of an option, to address the problems of

congestion and urban sprawl, whose effects include increased trip lengths and travel time. A

more sustainable form of development, known as Transit Oriented Development (TOD) is

presumed to offer sustainable travel choices with reduced need to travel to access daily

destinations, by providing a mixture of land uses together with good quality of public

transport service, infrastructure for walking and cycling. However, performance assessment

of these developments with respect to travel characteristics of their inhabitants is required.

This research proposes a five step methodology for evaluating the transport impacts of TODs.

The steps for TOD evaluation include pre–TOD assessment, traffic and travel data collection,

determination of traffic impacts, determination of travel impacts, and drawing outcomes.

Typically, TODs are comprised of various land uses; hence have various types of users.

Assessment of characteristics of all user groups is essential for obtaining an accurate picture

of transport impacts.

A case study TOD, Kelvin Grove Urban Village (KGUV), located 2km of north west of the

Brisbane central business district in Australia was selected for implementing the proposed

methodology and to evaluate the transport impacts of a TOD from an Australian perspective.

The outcomes of this analysis indicated that KGUV generated 27 to 48 percent less traffic

compared to standard published rates specified for homogeneous uses. Further, all user

groups of KGUV used more sustainable modes of transport compared to regional and

similarly located suburban users, with higher trip length for shopping and education trips.

Although the results from this case study development support the transport claims of

reduced traffic generation and sustainable travel choices by way of TODs, further

investigation is required, considering different styles, scales and locations of TODs. The

proposed methodology may be further refined by using results from new TODs and a

framework for TOD evaluation may be developed.


Deepti Muley Page iv

Deepti Muley Page v

Contents

Keywords i

Abstract iii

Contents v

List of figures xi

List of tables xiii

Abbreviations xvii

Statement of original authorship xix

Acknowledgments xxi

Chapter 1 1Introduction 1

1.1 Overview 1

1.2 Background 1

1.3 Research hypothesis 3

1.4 Objectives 4

1.5 Scope of this research 5

1.6 Publications from this research 5

1.7 Thesis outline 5

1.8 Chapter close 7

Chapter 2 2Literature review 9

2.1 Introduction 9

2.2 Transit Oriented Developments (TODs) 9

2.2.1 The concept of TODs 9

2.2.2 Types of TODs 12

2.2.3 Australian perspective of TODs 12

2.2.4 Benefits of TODs 14

2.3 Data requirements for TOD evaluation 22

2.3.1 Background 22

2.3.2 Data need / data requirement 22

2.3.3 Types of travel surveys 23

2.3.4 Types of survey instruments 26

2.3.5 Design of survey instruments 28


Deepti Muley Page vi

2.3.6 Response rate 30

2.3.7 Data analysis 31

2.3.8 Summary of data collection 31

2.4 Travel data analysis 32

2.4.1 Background 32

2.4.2 Comparative and statistical analysis 33

2.4.3 Travel demand modelling 43

2.4.4 Summary of travel data analysis 53

2.5 Summary of literature review 56

2.5.1 Strengths of the literature 56

2.5.2 Gaps in the literature 57

2.6 Recommendations flowing from literature review 58

2.7 Chapter close 59

Chapter 3 3Methodology for evaluating transport impacts of transit

3oriented developments

61

3.1 Introduction 61

3.2 Measures for transport evaluation of TODs 61

3.3 Proposed new research methodology 62

3.3.1 Step I: Pre–TOD assessment 63

3.3.2 Step II: Traffic and travel data collection 65

3.3.3 Step III: Determination of traffic impacts 67

3.3.4 Step IV: Determination of travel impacts 68

3.3.5 Step V: Outcomes 70

3.4 Summary 71

3.5 Application 72


Chapter 4 4Selection of case study transit oriented development 73

4.1 Introduction to case study selection 73

4.2 Description of KGUV 73

4.3 Transport facilities at KGUV 74

4.4 Suitability of case study TOD 77

4.4.1 Background 77

4.4.2 Transit availability 78

4.4.3 Comfort and convenience 79

4.5 Transit availability for KGUV 79

Contents

Deepti Muley Page vii

4.5.1 Analysis background 79

4.5.2 Availability – Transit stops 81

4.5.3 Availability – Route segments / corridors 84

4.5.4 Availability – System 86

4.6 Interpretation of results 91

4.7 Summary 91


Chapter 5 5Data collection 93

5.1 Introduction to data collection 93

5.2 User groups at KGUV 93

5.2.1 Employees 93

5.2.2 Shoppers 94

5.2.3 Students 94

5.2.4 Residents 94

5.2.5 Recreational users 95

5.3 Cordon counts 96

5.4 Travel surveys 97

5.4.1 General methodology for conducting travel surveys 98

5.4.2 Selection of survey instrument 100

5.4.3 Design of questionnaire form 101

5.4.4 Reminder letters and incentives 105

5.4.5 Process of conducting surveys 105

5.4.6 Sample size and response rates 109

5.4.7 Sample bias 111

5.5 Lessons learned 111

5.6 Summary 112


Chapter 6 6Traffic generation at Kelvin Grove Urban Village 115

6.1 Introduction 115

6.2 Analysis of cordon data 115

6.3 Conditions of the survey period 118

6.4 Total traffic at KGUV 118

6.4.1 Traffic at the Village Centre 118

6.4.2 Traffic at whole of KGUV 120

6.5 Comparison of peak hourly traffic with published rates 123


Deepti Muley Page viii

6.5.1 ITE comparison 123

6.5.2 Australian sources (RTA) comparison 125


6.7 Summary 128


Chapter 7 7Characteristics of Kelvin Grove Urban Village users 129


7.2 Determination of users’ characteristics 129

7.2.1 Demographic characteristics 130

7.2.2 Travel characteristics 130

7.3 Shoppers’ shopping trips at KGUV 131



7.4 Employees’ work trips at KGUV 134



7.5 Students’ education trips at KGUV 141



7.6 Residents’ trips at KGUV 148



7.7 Comparison of KGUV users’ characteristics 155

7.7.1 Comparison of demographic characteristics 155

7.7.2 Comparison of travel characteristics 157

7.8 Transport issues related to TOD from users’ perspective 158

7.9 Interpretation of TOD users’ characteristics 160

7.10 Summary 161


Chapter 8 8Comparative analysis of Kelvin Grove Urban Village’s users’ 8characteristics

163


8.2 Basis for comparison 163

8.3 Comparison of shoppers’ shopping trips 164

8.3.1 Mode share comparison 164

Contents

Deepti Muley Page ix

8.3.2 Trip length comparison 165

8.4 Comparison of employees’ work trips 166



8.5 Comparison of students’ education trips 168



8.6 Comparison of residents’ trips 169

8.6.1 Household characteristics 169

8.6.2 Trip characteristics 170




8.8 Summary 175


Chapter 9 9Kelvin Grove Urban Village users’ travel demand analysis 177


9.2 Analysis background 177

9.3 Analysis for shoppers’ shopping trips 182

9.4 Analysis for employees’ work trips 183

9.5 Analysis for students’ education trips 185

9.6 Analysis for residents’ first trip of the day 187


9.8 Summary 190


Chapter 10 10Models for travel modes of transit oriented development

10users

193


10.2 Analysis background for model development 193

10.3 Model for shopping trips 195

10.4 Model for work trips 196

10.5 Model for education trips 198

10.6 Model for residents’ first trip of the day 199

10.7 Interpretation 201

10.8 Model application 201


Deepti Muley Page x

10.9 Summary 203


Chapter 11 11Conclusions and recommendations 205


11.2 Conclusions from this research 205

11.2.1 Build up an understanding of the concept of TOD and

various aspects related to it with a detailed knowledge

on TOD evaluation

206

11.2.2 Develop a methodology for evaluating the transport

impacts of TODs

206

11.2.3 Demonstrate the methodology by implementing it on

an Australian case study TOD

209

11.2.4 Determine trip rates for various modes of transport for

a TOD and assess the travel demand of TOD users

210

11.2.5 Evaluate the transport impacts of TODs from an

Australian perspective by comparing the results with

characteristics of conventional development

211

11.3 Reflections from case study TOD 212

11.4 Applications of this research methodology 213

11.5 Contribution to knowledge 214

11.6 Limitations of this research 215

11.7 Areas of further research 216


References 219

Appendix A 233

Appendix B 235

Appendix C 237

Appendix D 239

Deepti Muley Page xi

List of figures

Figure 1.1 Greenhouse gas emissions 2

Figure 1.2 Layout of thesis 6

Figure 2.1 Path diagram of factors influencing changes in job accessibility and

commuting behaviour

42

Figure 2.2 The classic four–stage model 43

Figure 2.3 Idealised integrated urban modelling system 45

Figure 3.1 Measures for transport evaluation of TODs 62

Figure 3.2 Proposed new methodology for evaluating transport impacts of TODs 63

Figure 3.3 Step 1: Pre–TOD assessment 64

Figure 3.4 Step II: Traffic and travel data 66

Figure 3.5 Step III: Traffic impacts 68

Figure 3.6 Step IV: Travel impacts 69

Figure 3.7 Step V: Outcomes 71

Figure 4.1 Aerial overview of Kelvin Grove Urban Village (KGUV) 75

Figure 4.2 Regional map showing various offsite attractions considered in analysis 81

Figure 4.3 Buffers for bus stops for QUT route 391 intercampus shuttle service (ST3

& ST4)

89

Figure 4.4 Buffer for QUT KG Busway Station (ST1) 89

Figure 4.5 Buffers for bus stop at Kelvin Grove Road at Blamey Street (ST5) and

Normanby Busway Station (ST8) (inbound to CBD)

90

Figure 4.6 Buffers for bus stop at Kelvin Grove Road at Prospect Terrace (ST7) and

Normanby Busway Station (ST8) (outbound from CBD)

90

Figure 5.1 Overview of land uses and user groups at KGUV 95

Figure 5.2 Locations of cordon counts at KGUV 97

Figure 5.3 Steps involved in data collection process for a user group travel survey 99

Figure 5.4 Questionnaire details for non residential land use users 102

Figure 5.5 Questionnaire details for residential land use users 104

Figure 6.1 Process of cordon data analysis 116

Figure 6.2 Direction of through traffic at KGUV 117

Figure 7.1 Distribution of shoppers’ age groups in years 132

Figure 7.2 Frequency of shopping trips per week 132

Figure 7.3 Mode shares for shoppers’ at KGUV 133

Figure 7.4 Distribution of age group for employees’ at KGUV 135


Deepti Muley Page xii

Figure 7.5 Frequency of work trips at KGUV 136

Figure 7.6 Household size distribution at employees’ households 137

Figure 7.7 Mode shares for professional employees 138

Figure 7.8 Mode shares for retail shop employees 139

Figure 7.9 Distribution of age group for students at KGUV 142

Figure 7.10 Frequency of education trips at KGUV 143

Figure 7.11 Household size distribution at students’ households 144

Figure 7.12 Mode shares for school students 145

Figure 7.13 Mode shares for university students 146

Figure 7.14 Distribution of age group for residents’ at KGUV 149

Figure 7.15 Distribution of number of bedrooms in residents’ household 149

Figure 7.16 Distribution of household size at residents’ household 150

Figure 7.17 Mode shares for non student residents at KGUV 152

Figure 7.18 Mode shares for student residents at KGUV 153

Figure 8.1 Comparison of sustainable transport mode share 173

Figure 8.2 Comparison of overall average trip length 174

Figure 9.1 Sensitivity of typical shopper’s sustainable travel mode probability with

travel time saving

183

Figure 9.2 Sensitivity of typical employee’s sustainable travel mode probability with

age group

185

Figure 9.3 Sensitivity of typical student’s sustainable travel mode probability with

age group

187

Figure 9.4 Sensitivity of typical resident’s sustainable travel mode probability with

travel time saving

188

Deepti Muley Page xiii

List of tables

Table 1.1 Overview of publications from this research 5

Table 2.1 Characteristics of various existing TODs 11

Table 2.2 Application of four part TOD strategy to Australian cities 13

Table 2.3 An overview of household travel surveys 24

Table 2.4 Summary of different survey instruments 32

Table 2.5 Summary of studies on comparative and statistical analysis 54

Table 2.6 Summary of studies on travel demand modelling 55

Table 3.1 Measures for assessing suitability of a TOD 65

Table 4.1 Mixed land uses at KGUV 74

Table 4.2 Buses observing transit stops in KGUV 76

Table 4.3 Quality of service framework: Fixed – route (TRB, 2003, Exhibit 3–1) 77

Table 4.4 Fixed – route service frequency LOS (TRB, 2003, Exhibit 3 – 12) 82

Table 4.5 LOS for various trip destinations originating from KGUV (Kelvin Grove

Road bus stops)

82

Table 4.6 LOS for various trip destinations originating from KGUV (QUT KG

Busway Station)

83

Table 4.7 LOS for various trip origins where destination is KGUV (Kelvin Grove

Road bus stops)

83

Table 4.8 LOS for various trip origins where destination is KGUV (QUT KG

Busway Station)

83

Table 4.9 Fixed – route hours of service LOS (TRB, 2003, Exhibit 3 – 13) 84

Table 4.10 Hours of service LOS for different corridors with bus route numbers 85

Table 4.11 Fixed – route service coverage LOS (TRB, 2003, Exhibit 3 – 14) 86

Table 4.12 LOS for different bus stops 88

Table 5.1 Summary of travel surveys 109

Table 5.2 Sample sizes and response rates 111

Table 6.1 Time periods for analysis for Village Centre 119

Table 6.2 Traffic generated at Village Centre by mode on study day 119

Table 6.3 Car occupancy and directional distribution for Village Centre 120

Table 6.4 Directional distribution for Village Centre 120

Table 6.5 Time periods for analysis of whole KGUV traffic 121

Table 6.6 Through car traffic at KGUV by time of day 121

Table 6.7 Total traffic generated at KGUV by mode on study day 121


Deepti Muley Page xiv

Table 6.8 Car occupancy and directional distribution at KGUV 122

Table 6.9 Directional distribution at KGUV 122

Table 6.10 ITE comparison for traffic at the Village Centre 124

Table 6.11 ITE comparison for traffic at KGUV 125

Table 6.12 RTA comparison for traffic at the Village Centre 126

Table 6.13 RTA comparison for traffic at the KGUV 126

Table 7.1 Employment status of shoppers at KGUV 131

Table 7.2 Trip lengths by mode of transport for shoppers at KGUV 134

Table 7.3 Personal characteristics of employees’ at KGUV 135

Table 7.4 Vehicle ownership and licence availability at employees’ households 137

Table 7.5 Trip lengths by mode of transport for professional employees at KGUV 139

Table 7.6 Trip lengths by mode of transport for retail shop employees at KGUV 140

Table 7.7 Public transport trip details for employee trips at KGUV 141

Table 7.8 Personal characteristics of students at KGUV 142

Table 7.9 Vehicle ownership and licence availability at students’ households 144

Table 7.10 Trip lengths by mode of transport for school students at KGUV 146

Table 7.11 Trip lengths by mode of transport for university students at KGUV 146

Table 7.12 Public transport trip details for student trips at KGUV 147

Table 7.13 Personal characteristics of residents’ at KGUV 148

Table 7.14 Vehicle ownership and licence availability at residents’ households 150

Table 7.15 Number of trips for residents at KGUV 151

Table 7.16 Trip lengths by mode of transport for non student residents at KGUV 154

Table 7.17 Trip lengths by mode of transport for student residents at KGUV 154

Table 7.18 Activity frequency for non student residents at KGUV 154

Table 7.19 Activity frequency for student residents at KGUV 155

Table 7.20 Comparison of personal characteristics of KGUV users’ 156

Table 7.21 Comparison of household characteristics of KGUV users’ 157

Table 7.22 Comparison of travel characteristics of KGUV users’ 158

Table 8.1 Mode share comparison for shopping trips 165

Table 8.2 Comparison of average trip lengths for shopping trips 166

Table 8.3 Mode share comparison for work trips 167

Table 8.4 Comparison of average trip lengths for work trips 167

Table 8.5 Mode share comparison for education trips 168

Table 8.6 Comparison of average trip lengths for education trips 169

Table 8.7 Comparison of residents’ household characteristics 170

List of tables

Deepti Muley Page xv

Table 8.8 Comparison of residents’ average trips per person 171

Table 8.9 Mode share comparison for residents’ first trip of the day 171

Table 8.10 Comparison of average trip lengths for residents’ first trip of the day 172

Table 9.1 Details of coding system for travel mode investigation 180

Table 9.2 Travel mode analysis for shoppers’ shopping trips 182

Table 9.3 Travel mode analysis for employees’ work trips 184

Table 9.4 Travel mode analysis for students’ education trips 186

Table 9.5 Travel mode analysis for residents’ first trip of the day 188

Table 9.6 Significance of variables for travel mode determination for KGUV users 189

Table 9.7 Comparison of odds of KGUV users’ for choosing a sustainable mode of

transport

190

Table 10.1 Original model for shoppers’ shopping trips 195

Table 10.2 Revised model for shoppers’ shopping trips 196

Table 10.3 Original model for employees’ work trips 197

Table 10.4 Revised model for employees’ work trips 197

Table 10.5 Original model for students’ education trips 198

Table 10.6 Revised model for students’ education trips 199

Table 10.7 Original model for residents’ first trip of the day 200

Table 10.8 Revised model for residents’ first trip of the day 200


Deepti Muley Page xvi

Deepti Muley Page xvii

Abbreviations

Abbreviation Description

BINS Brisbane Inner North Suburbs

BISS Brisbane Inner South Suburbs

BRT Bus Rapid Transit

BSD Brisbane Statistical Division

CAPI Computer Assisted Personal Interview

CBD Central Business District

CI Creative Industries

ESD Ecological Sustainable Development

GFA Gross Floor Area

GIS Geographic Information System

GPS Global Positioning System

IHBI Institute of Health and Biomedical Innovation

KG Kelvin Grove

KGUV Kelvin Grove Urban Village

LA Licence Availability

LOS Level of Service

NRLU Non Residential Land Use

NRMA National Roads and Motorists’ Association

NSR Non Student Resident

QACI Queensland Academy of Creative Industries

QoS Quality of Service

QUT Queensland University of Technology

RACQ Royal Automobile Club of Queensland

RLU Residential Land Use

SEQ South East Queensland

SEQTS South East Queensland Travel Survey

SR Student Resident

TCQSM Transit Capacity and Quality of Service Manual

TOD Transit Oriented Development


Deepti Muley xviii

Abbreviation Description

TSA Transit Supportive Area

TTS Travel Time Saving

VC Village Centre

VKT Vehicle Kilometres Travelled

Deepti Muley Page xix

Statement of original authorship

The work contained in this thesis has not been previously submitted to meet requirements for

an award at this or any higher education institution. To the best of my knowledge and belief,

the thesis contains no material previously published or written by another person except

where due reference is made.

Signature:

Deepti Sadashiv Muley

Date: 21st July 2011


Deepti Muley xx

Deepti Muley Page xxi

Acknowledgments

First of all, I acknowledge my supervisors Dr. Jonathan Bunker and Prof. Luis Ferreira for

their guidance, knowledge and contribution which had enabled me to achieve this piece of

work.

I am grateful to the School of Urban Development for proving me the scholarship and

recourses for conducting this research specifically the data collection process.

My thanks for the excellent help obtained from research assistants during data collection

process, especially Mr. Daniel Buntine and Mr. William Dawson. I also thank Queensland

Transport for proving the dataset for comparison.

I am thankful to Mr. Mike Hyslop and Prof. Edward Chung for proving me a comfortable

working environment which has allowed me to manage my work and research commitments.

I thank my fellow postgraduate students, colleagues at PTV Asia Pacific, and all friends in

India and Australia for making this journey pleasant.

I acknowledge my parents (Aai and Baba), siblings (Trupti, Bhakti and Dhiraj) and brother in

law (Milind) for their constant love, faith and encouragement. This is equally yours as it is

mine. I appreciate Alok for his support in finalising this piece of work. Last but not the least;

I am grateful to Sumeet for being a support in Australia throughout three years of my

candidature.


Deepti Muley xxii

Deepti Muley Page 1

Chapter 1

Introduction

1.1 Overview

Urban development planners consider Transit Oriented Developments (TODs) as an

alternative to reduce urban sprawl and congestion. TODs are not only believed to provide

mobility choices and improve transit ridership but are also assumed to offer more lifestyle

options and social life by providing street level shops, parks and open spaces (Calthorpe,

1993; NIPC, 2001). This research explores the transport impacts of these communities by

studying the travel demand for various groups of users at a particular TOD.

The first section gives a brief background to this research and then states the research

hypothesis and the objectives of this research along with its scope. The subsequent section

briefs details about the publications from this research. The outline of this thesis is described

in the following section along with the chapter close which states the relevance or

relationship of this chapter with respect to the other chapters in the thesis.

1.2 Background

Transport demand is a derived demand, which arises from the activities which are performed

outside the home by an individual. A person is involved in many activities which lie outside

the home, hence one need to travel from one place to another. This travel can be performed

by two main modes of transport, either personalised mode or public mode. The personalised

modes of transport can be divided in two categories; motorised such as car and motorcycle,

and non motorised such as bicycle and walking. The use of non motorised modes of transport

is not feasible for covering long distances. Generally, the use of the public transport mode is

recommended by policy makers over motorised private modes of transport to control

pollution and congestion, while people mostly choose the motorised private mode of

transport (car) to maximise their flexibility of movement and very often to achieve travel time

savings. Due to a high proportion of urban populations opting for motorised private modes of

transport, some negative impacts are observed on the environment, as well as on human life.

This statement can be supported by the facts listed below:


Deepti Muley Page 2

Increasing load on infrastructure

Increased traffic on the road system may make the journey uncomfortable with undesirable

delays. Car ownership level is an indicator of the amount of vehicle usage and consequently

traffic on the roads. The trend in September 2007 (ABS, 2007) shows the increasing tendency

of Australians to own a vehicle when compared to the previous year. Increased car ownership

levels will further increase the tendency to drive to daily destinations and this tendency will

place extra load on existing transport infrastructure.

Travel characteristics

An estimate revealed that total travel in Australian urban areas has grown ten–fold over the

last 60 years, with private road vehicles having a major share of about 90% of the total

passenger task (BTRE, 2007). The projections forecast total kilometres travelled growing by

37% between 2005 and 2020. This increased travel will increase congestion and pollution on

the roads, leading to variability in journey time and increased vehicle operating costs. In the

same study, the social cost of congestion was estimated to rise from $9.4 billion in 2005 to

$20.4 billion in 2020.

Greenhouse gas emissions

Reducing greenhouse gas emissions to combat climate change is an important agenda for

policy makers. Australian statistics show that the transport sector emissions were 30.5%

higher in 2005 when compared to 1990 emissions (AGO, 2007).

Source: AGO (2007)

Figure 1.1 Greenhouse gas emissions

0

20

40

60

80

100

Transport as total

emissions

Share of road transport Share of passenger cars

Per

cen

tage (

%)

Introduction

Deepti Muley Page 3

Figure 1.1 explains the role of road transport in emissions for the year 2005. For Australia,

the transport sector is the third largest contributor to greenhouse gas emissions after

stationary energy sources and agriculture. The alternative fuels conversion programme and

travel demand management programmes, which promote the use of alternative fuels and

improve public transport and town planning, are key emission management measures (AGO,

2006).

Petrol prices

The Royal Automobile Club of Queensland (RACQ) and National Roads and Motorists’

Association (NRMA) found that an average Australian household spends 15 to 20 percent of

its net income directly on transport (BTA, 2007). Increasing petrol prices is one of the main

reasons as around 90 percent of trips are made by passenger cars. For the Brisbane

Metropolitan region, the average price of unleaded petrol has almost doubled in the past

seven years, from 60.9 cents per litre in June, 1999 to 122 cents per litre in June, 2007

(OESR, 2007).

To minimize the negative impacts of car use, more sustainable development is required. The

national greenhouse strategy supports integration of land use and transport planning which

includes promotion of development near public transport systems, incorporating higher

residential and commercial densities and appropriate mixed uses (including residential,

commercial, retail and other employment activities) (AGO, 1998). One proposed form,

Transit Oriented Development (TOD), is a mixed-use community within an average 600m

walking distance of a transit node and core commercial area. TODs mix residential, retail,

office, open space, and public uses in a walkable environment, making it convenient for

residents and employees to travel by transit, bicycle, foot, or car (Calthorpe, 1993). This

transport related land use strategy can be used in large urban and small communities in

coordination with bus, rail, and/or ferry transit systems (Parker et al., 2002). The past

research provides limited information regarding estimation of travel demand and evaluation

of these developments. This research aims to explore this area; The following section

presents the research hypothesis.

1.3 Research hypothesis

A TOD is often distinguished from conventional developments because of its atypical

development characteristics, such as high quality of transit service, pedestrian and cycling


Deepti Muley Page 4

facilities, higher density, moderated private vehicle infrastructure, improved accessibility to

and mixes of land uses. The travel characteristics of the people using TODs are assumed to

be sustainable or environmentally friendly. This statement gives rise to following research

hypothesis.

“Transit Oriented Developments (TODs) reduce car dependence and promote sustainable

modes of transport. Hence, TOD users will make more walking, cycling and public transport

trips as compared to their counterparts in conventional development, making TODs the more

sustainable form of development.”

This statement suggests that people may travel differently, showing a distinction in travel

behaviour when compared to conventional development. So the testing of this hypothesis

demands an assessment of:

How to determine the travel demand of a TOD?

What are the various aspects of a TOD’s travel demand?

How is the travel demand different from that of conventional development?

Why is a TOD’s travel demand different from that of conventional development?

What are the governing factors in determining the travel demand?

Answering these questions will help to understand and model the travel behaviour of these

communities, which will further help in planning them. This research aims at testing this

hypothesis and answering the associated questions in an Australian context, by conducting

data collection and studying the travel demand for a case study TOD.

1.4 Objectives

The objectives of this research can be formulated as below on the research hypothesis and the

questions that arose to answer it.

1. Build up an understanding of the concept of TOD and various aspects related to it

through gaining a detailed knowledge of TOD evaluation.

2. Develop a methodology for evaluating the transport impacts of TODs.

3. Demonstrate the methodology by implementing it on an Australian case study TOD.

4. Determine trip rates for various modes of transport for a TOD and assess the travel

demand of TOD users.

Introduction

Deepti Muley Page 5

5. Evaluate the transport impacts of TODs from an Australian perspective by comparing

the results with characteristics of conventional development similarly located in the

urban fabric.

1.5 Scope of this research

The development of TODs has various aspects including planning, designing, development or

implementation, and evaluation. This research looks at evaluating TODs. A TOD is assumed

to be more sustainable providing safety, affordability, sense of community and public health,

economic benefits to individuals, government and developers, and environmental benefits by

implementing sustainable practices in transport, infrastructure and energy benefits. This

research looks at the transport impacts of TODs. The scope of this research is limited to TOD

evaluation from a transport point of view. The transport evaluation in this research is limited

to studying the travel of these communities and assessing its sustainability.

1.6 Publications from this research

This research has resulted in one journal paper, three refereed conference papers, one book

chapter and one presentation at the United States’ Transportation Research Board National

Conference. A brief overview of the publications is given in Table 1.1 and a complete list of

publications is given in Appendix A.

Table 1.1 Overview of publications from this research

Sr. No. Authors Publication Comment

1. Muley et

al. (2007)

ATRF07

conference

Assessing suitability of the case study TOD

(Chapter 4)

2. Muley et

al. (2008)

ATRF08

conference

Procedure for conducting travel surveys and

preliminary results (Chapter 5 and Chapter 7)

3. Muley et

al. (2009)

QUT

conference Characteristics of TOD users (Chapter 7)

4. Muley et

al. (2009) ITS journal Travel demand analysis for TOD users

5. Muley et

al. (2009)

TRB conference

presentation

Travel demand analysis for TOD users (Chapter 8

and Chapter 9)

6. Muley et

al. (2009) Book chapter Characteristics of TOD users (Chapter 7)

1.7 Thesis outline

The research conducted on TOD transport evaluation is presented in this thesis through

eleven chapters. The initial chapters give an overview of the research followed by the case


Deepti Muley Page 6

study information and data analysis. The later chapters explain the analysis and the results

followed by conclusions and scope for future work. Figure 1.2 illustrates the relationships

between the thesis chapters and the following paragraphs present the detailed information

about each chapter.

Figure 1.2 Layout of thesis

Chapter One (this chapter) establishes the research context and sets the objectives of

this research.

1. Introduction (Hypothesis and Objectives)

2. Literature review

3. Methodology for evaluating

transport impacts of TODs

4. Selection of case

study TOD

5. Data collection

6. Traffic generation

at KGUV

7. Characteristics

of KGUV users

9. & 10. KGUV

users’ travel

demand analysis

8. Comparative

analysis of KGUV

users’ characteristics

Appendix D. TOD

users’ perceptions

11. Conclusions and

recommendations

Direct relation

Indirect relation

Introduction

Deepti Muley Page 7

Chapter Two presents a review of the past literature related to studies on TODs. The

review of literature mainly gives an overview of the TODs and states

the studies related to TOD evaluation.

Chapter Three describes the proposed new methodology in this research for assessing

transport impacts of TODs.

Chapter Four describes the case study development and the selection criteria followed

for analysing the suitability of the case study development as a TOD.

Chapter Five gives details of the data collection process followed for conducting the

cordon counts and travel surveys.

Chapter Six presents the analysis and results for the traffic generation analysis

undertaken for the case study TOD.

Chapter Seven provides an overview of the characteristics of various user groups

present at the TOD from the findings obtained from the preliminary

data analysis of the data obtained from the travel surveys.

Chapter Eight compares the characteristics of the TOD users with the characteristics

of regional and suburban users to determine the similarities and

differences and presents the findings from the comparative study.

Chapter Nine explains the analysis and the results for travel demand analysis

conducted for assessing the effect of various characteristics on the

travel modes of TOD users.

Chapter Ten develops the models for travel modes for various trips at a TOD.

Chapter Eleven discusses the results obtained from the case study and presents the

conclusions and recommendations from this research, and for future

research.

1.8 Chapter close

This chapter introduces the research, along with the research questions and objectives, which

were investigated during the research. This links with the literature to be studied (which is

presented in the next chapter) and provides base for the method of investigation presented in

Chapter 3.


Deepti Muley Page 8

Deepti Muley Page 9

Chapter 2

Literature review

2.1 Introduction

This chapter examines the literature illustrating the evaluation of transport aspects of Transit

Oriented Developments (TODs). The chapter is divided into four main sections. The first

section explains the concept of TODs. The perceived transport benefits of TODs and actual

(observed or estimated) benefits of TODs, which have been accomplished in practice, are

also reported. Travel demand analysis needs extensive data on demographics and travel

behaviour for all categories of people residing or visiting a TOD. The data requirements

necessitate various methods of data collection. The details of various travel surveys along

with different features of data collection process are examined briefly in the second section.

The third section provides insight into analysis and modelling of travel demand with specific

emphasis on areas having mixed land uses and higher densities. The closing section integrates

the various topics covered in the chapter and identifies the gaps as well as strengths of the

reviewed literature along with recommendations, followed by the chapter close.

2.2 Transit Oriented Developments (TODs)

2.2.1 The concept of TODs

Urban lifestyle and the population growth have changed the travel needs of inhabitants of

urban areas drastically, by increasing average trip lengths and usage of personalised modes of

transport, particularly the car. The consequences of these changes can be observed through

increased pollution and congestion levels on road networks, which in turn results in higher

travel times (BTRE, 2007). Transport emissions are one of the major contributors to the

damage of air quality (AGO, 2006). To counteract these ill effects and facilitate human life

with more sustainable and vibrant lifestyle options the concept of Transit Oriented

Development (TOD) has been utilised. Prior to the 1990’s TODs were aimed as profitable

real estate development, used to generate revenues for transit agencies and government, and

were evaluated purely on a financial basis rather than on sustainable transport principles.

Newly planned development is now generally accepted by planners as needing to reduce

overall vehicle use as well as to reduce concentration of urban travel patterns towards single

Central Business Districts (CBD). In simple terms, mixed use planned development around a


Deepti Muley Page 10

major public transport node is generally considered to be a transit oriented development. In

technical terms,

“Transit–Oriented Development (TOD) is moderate to higher–density development, located

within an easy walk of a major transit stop, generally with a mix of residential, employment

and shopping opportunities designed for pedestrians without excluding the car. TOD can be

new construction or redevelopment of one or more buildings whose design and orientation

facilitate transit use.” (Parker et al., 2002)

The term transit in the definition encapsulates a variety of modes and systems including

metro rail, heavy rail, suburban rail, commuter rail, light rail transit (LRT), streetcar, bus

rapid transit (BRT), and express bus (incorporating Dittmar and Ohland, 2004). The easy or

comfortable walking distance is affected by topography, climate, intervening arterial roads,

freeways and other transport corridors, and other physical features. An average distance of

600 m is used as comfortable walking distance in the American context (Calthorpe, 1993)

while a distance of 1km has in the past been used as comfortable walking distance for

Australian urban settings (Gilbert and Ginn, 2001). Generally, a buffer distance of 400 m for

local bus stops and 800m for premium bus stops is presently considered appropriate for

design (Queensland Transport, 1999; TRB, 2003).

Although TOD is characterised differently in different parts of the world; such as smart

growth, urban form, new urbanism, walkable communities, neo–traditional neighbourhood or

development, activity centres, new community design, transit village, and transit supportive

development, the benefits sought are the same. Planning and implementation of a TOD

involves multiple stakeholders including state and local government agencies, land owners,

funding agencies, developers, design professionals, investors, management agents, residents

and occupants, public interests and community interests (Dittmar and Ohland, 2004). Due to

involvement of so many entities the success of TOD becomes a complex phenomenon. This

research aims at assessing TOD’s success from a transport point of view by examining the

travel demand generated by TOD.

Literature review


Table 2.1 Characteristics of various existing TODs

TOD Type Land–use mix Minimum

housing density

Housing types Scale Regional

connectivity

Transit

modes

Transit

frequencies

Examples

Urban

downtown

Primary office

center, urban

entertainment,

multifamily

housing, retail

> 60 units/acre

(> 148 units/ha)

Multifamily

loft

High High Hub of

radial system

All modes < 10 minutes Printers row

(Chicago), LoDo

(Denver), South

Beach (San

Francisco)

Urban

neighbourhood

Residential,

retail, Class B

commercial

> 20 units/acre

(> 50 units/ha)

Multifamily

Loft

Townhome

Single family

Medium Medium

access to

downtown

subregional

circulation

Light–rail

Streetcar

Rapid bus

Local bus

10 minutes

peak

20 minutes

offpeak

Mockingbird

(Dallas), Fullerton

(Chicago), Barrio

Logan (San

Diego)

Suburban

centre

Primary office

centre, urban

entertainment,

multifamily

housing, retail

> 50 units/acre

(> 124 units/ha)

Multifamily

Loft

Townhome

High High access to

downtown

subregional

hub

Rail

Streetcar

Rapid bus

Local bus

Paratransit

10 minutes

peak

10-15 minutes

offpeak

Arlington County

(Virginia),

Addison Circle

(Dallas),Evanston

(Illinois)

Suburban

neighbourhood

Residential

neighbourhood,

retail, local

office

> 12 units/acre

(> 30 units/ha)

Multifamily

Townhome

Single family

Moderate Medium

access to

suburban

center

Access to

downtown

Light-rail

Rapid bus

Local bus

Paratransit

20 minutes

peak

30 minutes

offpeak

Crossings

(Mountain View,

CA), Ohlone–

Chynoweth (San

Jose, CA)

Neighbourhood

transit zone

Residential

neighbourhood,

retail

> 7 units/acre

(> 17 units/ha)

Townhome

Single family

Low

access to a

centre

Low Local bus

Paratransit

25-30 minutes

demand

responsive

Commuter

town centre

Retail center,

Residential

> 12 units/acre

(> 30 units/ha)

Multifamily

Townhome

Single family

Low Low access to

downtown

Commuter

rail

Rapid bus

Peak service

Demand

responsive

Prairie Crossing

(Illinois), Suisun

City (CA)

Source: Dittmar and Ohland (2004)



2.2.2 Types of TODs

Although the basic concept of planning and designing TODs remains the same, the outcomes

are distinct, with every TOD having specific features. The size of TOD is dependent on the

amount of land available and dedicated for this use. Urban TODs sustain much larger

commercial and office or employment area and higher density residential uses than those in

rural areas. Making soft conversions, urban TODs are generally constructed to a minimum

residential density of 30 dwelling units per net hectare and an average density of 45 dwelling

units per net hectare, while rural TODs are designed at minimum density of 17 dwelling units

per net hectare and an average density of 25 dwelling units per net hectare (Calthorpe, 1993;

Gatzlaff et al., 1999). In order to classify various TODs at different levels according to the

differences between places and destinations within the regions, and to identify the appropriate

performance measures and descriptive benchmarks, a TOD typology is defined as listed in

Table 2.1. The TODs are classified with respect to location, size, and transit type.

2.2.3 Australian perspective of TODs

The planners of the South East Queensland (SEQ) region have an increasing concern about

public transport because of the increased car use and decreased use of public transport over

the last 20 years (Rural and Regional Affairs and Transport References Committee, 2006).

The Queensland Government has identified potential TOD sites and provided principles of

TOD in SEQ (Queensland Government, 2005). The interim TOD working group focussed on

three major areas; principles of identifying TOD sites, governance agreements needed to

successfully deliver TODs and identification of specific TOD sites (James, 2005).

Bajracharya & Khan (2005) highlighted positive as well as negative aspects of TOD in SEQ

by developing a holistic theoretical model (conceptual framework). Gray and Bunker (2005)

analysed the connectivity of Kelvin Grove Urban Village (KGUV) by walking, cycling and

public transport; which is a new site for TOD in Brisbane. They used GIS based transport

analysis software called TLOS to integrate both temporal and spatial features of bus and train

services into a single analysis.

Apart from this, Mepham (2005) studied Brisbane rail as a mode of transit in Brisbane

evaluating the need for developing a model for TOD. The transit system of Brisbane was

studied in detail; describing the issues, factors for success, potential TOD sites and land use

policy associated with it. A four part strategy for successful TOD was outlined as a strategic

policy for centres, a strategic policy for rapid transit, a statutory base that requires

Literature review


implementation of necessary densities and design, and a public–private financing mechanism

to build rail linked to centres (Newman, 2005). The performance of major Australian cities is

shown in Table 2.2.

Table 2.2 Application of four part TOD strategy to Australian cities

City

Strategic

policy for

centres

Strategic policy

for rail transit

Statutory

process to

implement TOD

Public–private

funding

mechanism

Sydney Yes Weak. Present rail

mostly.

Yes in new areas;

possibly

elsewhere.

Possibly but not

yet.

Melbourne Yes, but

struggling.

Weak. Present rail

mostly.

Yes but not

strong.

No.

Brisbane Yes, but not

well defined.

Weak. Present rail

mostly. Busway

dominance.

No. No.

Perth Yes, but not

well defined.

Yes. No. No.

Adelaide Yes, but not

well defined.

Weak on rail. No. No.

Others: Canberra,

Hobart,

Newcastle….

Yes, but not

well defined.

No. No. No.

Source: Newman (2005)

Currie (2005) provided a critical look at the strengths and weakness of bus based transit

systems in relation to TOD through a review of the literature and an assessment of TOD

related developments. A small number of strengths were identified however these were

generally considered to be highly significant. These included cost effectiveness, flexibility,

complementarily or ubiquitousness and, for bus rapid transit, service frequency and transfers.

A large number of weaknesses have been identified. The lack of dedicated TOD development

staff in the bus industry, the noise/pollution impacts of buses and a poor track record of bus in

relation to TOD were the most significant weaknesses identified for bus services as a whole.

It was noted that the successful implementation of bus based TOD is a more difficult task

than related rail based TOD initiatives. It is noted that these comments were made prior to

realisation of any maturity of Australian BRT systems such as Brisbane’s Busway network.

A diagnostic tool that can assist TOD developers and decision–makers to quickly assess the

potential of developments and the likely travel behaviour produced by their design to rate the

residential travel performance was suggested by Burke and Brown (2005). The tool was

based on accessibility analysis techniques with the location and design of developments being

key issues relating to local area planning interventions.



Russell (2006) examined the opportunities for TOD along Perth’s then soon to be constructed

South West Metropolitan Rail line, using a combination of detailed site analysis, standardised

data sets and background information from previous research and government documents. To

conceptualise perceived opportunities, the calculations for vacant land availability within

each station pedestrian catchment (pedshed) were presented. The potential for development,

both under the present situation and with the changes proposed by TOD was determined. The

research highlighted several fundamental changes necessary if Perth is to encourage transit–

oriented development and experience the full benefits offered by its new transit system.

The draft 30 – year plan for greater Adelaide proposed the land use policies to manage the

population and economic growth while preserving the environment and protecting the

heritage of greater Adelaide (Government of South Australia, 2010). The plan was

underpinned by three objectives; namely maintain and improve liveability, increase

competitiveness, and drive sustainability and resilience to climate change. The outcomes of

the plan include creation of 14 new TODs and more than 20 sites that incorporate TOD

principles and design characteristics. The plan offered housing choices by keeping housing

affordable and liveable in these areas. The plan also aimed to protect 115,000 hectares of

environmentally significant land and 375,000 hectares of primary production land. The

growth for mining and defence industries was also supported. The plan created a network of

greenways, open space precincts and reduced water and energy consumption at new

dwellings. By implementing this, the plan proposed to have substantial social, economic, and

environmental benefits for greater Adelaide.

2.2.4 Benefits of TODs

2.2.4.1 Stated benefits of TODs

Planners, public officials and large–scale land developers increasingly promote mixed–use

and high density developments as an alternative to sprawl and congestion (Ewing et al., 2001;

Handy, 1996) and to address the increasing concerns about the environmental side effects of

the use of motor vehicles (Sun et al., 1998; Handy, 1996).

TOD mainly constitutes the development of retail and business amongst residential

development in a compact neighbourhood. This development offers a new range of

development patterns for households, businesses, towns and cities (Dittmar and Ohland,

2004). The planning and construction of TODs is performed by assuming benefits mainly

Literature review


related to lifestyle options, mobility choices, reduced travel needs, cleaner air quality, and

social life. The following summarises the benefits of TOD as stated in the literature:

TOD gives a balanced approach for development supporting economic development,

healthy environment and strong communities (Dittmar and Ohland, 2004).

Design of TOD enhances residents’ sense of community; affordable housing increases

the choices and reduces the infrastructure cost (Calthorpe, 1993; Gatzlaff et al., 1999;

Lund, 2006).

An important component of TOD is the clustering of activities due to a diverse mix of

land uses located closer together than found in undifferentiated low–density

development. One potential benefit of this clustering is the reduction of vehicular

traffic due to provision of high level of transit service and replacement of automobile

trips with walking and bicycling in addition to riding transit (Handy, 1996;

McCormack et al., 2001; Hess and Ong, 2002; Cervero, 2006; Lund, 2006).

As more customers walk, bicycle or use transit for visiting nearby shopping areas or

business centres, the number of parking spaces in those developments can be reduced.

This also warrants the reduced level of traffic in TODs (Steiner, 1998).

The high density neighbourhood supports a mixed level of retail and office uses so

these centres will satisfy many needs of the community resulting in reduced need to

drive to a town centre (Steiner, 1998).

Further, the people who live and work within such transit oriented developments will

make fewer and shorter automobile trips (Sun et al., 1998; Steiner, 1998) and will

choose to walk or bicycle or use transit more frequently as compared to other lower

density, single use residential areas (Steiner, 1998).

People will use sustainable modes of transport only if a competent transport system is

available for use; hence to plan the transit and other services in a TOD, travel demand needs

to be determined. Further, this demand needs to be assessed to verify claims on the benefits

of TODs. Before examining travel data collection and travel demand assessment matters, the

reported transport benefits of TODs are expanded in the following section.

2.2.4.2 Reported benefits of TODs

Many researchers have tried to investigate the link between various elements of TOD (mixed

land use, density, neighbourhood design) and the travel behaviour of people. Some studies

have found that the effect of these variables on travel behaviour is significant, while others



claimed the effect to be minimal (Badoe and Miller, 2000). The outcomes from some studies

are noted below:

Travel characteristics and travel behaviour

The presence of nearby commercial land uses is associated with short commuting distances

among the residents of a mixed–use neighbourhood (Cervero, 1996). The combination of

land use, transportation system improvement and demand management measures (LUTRAQ

alternative, Portland, Oregon) significantly reduced the need to own multiple automobiles

and the number of vehicle kilometres travelled reduced substantially, increasing walking,

bicycling and the use of transit (Rossi et al., 1993). TOD scenarios exhibited more efficient

pattern of trip making by increasing the number of walking and transit trips and reducing

VKT and slower growth in vehicle trip generation in a subregion of the Minneapolis – St.

Paul Metropolitan region (Dock and Swenson, 2003). In short, it was suggested that the built

environment itself influences individuals’ travel behaviour (Cao et al., 2009).

Aditjandra et al. (2009) found that there is an association between changes in neighbourhood

characteristics and changes in travel behaviour, however in the UK this association is not as

strong as in the US, suggesting that land use planning in the UK may have less of an impact

on reducing private car travel, as compared to the US. It could however be interpreted that

urban fabrics in the UK are already less car orientated (or more TOD like) than those in the

US, so that any changes in travel behaviour in the UK under a TOD development may vary

less from the development norm than in the US.

Work trips

An American study found that the land–use environments of contemporary suburban work–

places appeared to have modest to moderate influence on commuting behaviour. The

existence of retail component within a suburban office building reduced vehicle–trip rates per

employee by about 8 percent. The variation of parking prices at the office buildings of

suburban activity centres with the number of occupants in a vehicle proved effective for

increasing carpools and vanpools (Cervero, 1991).

The compact, mixed–use, and pedestrian–oriented development form has the strongest effect

on access trips to rail stations, in particular inducing higher shares of access trips by foot and

bicycle (Cervero and Radisch, 1996).

Literature review


Non work trips

The residents of a neo–traditional neighbourhood (Rockridge, San Francisco Bay Area)

averaged around a 10 percent higher share of non–work trips by non–automobile modes than

the residents of conventional development (Lafayette, San Francisco Bay Area) with the

greatest difference for shop trips under one mile (1.6 km), 28 percent by foot and 66 percent

by car versus 6 percent by foot and 81 percent by car (Cervero and Radisch, 1996). The

Rockridge residents also averaged substantially higher rates of non–work walk trips per day,

matched by lower rates of daily car travel. This implied that walking substituted for

motorised travel, at the margin.

Similar to above results, Rajamani et al. (2003) showed that mixed uses promoted walking

and transit mode for non-work activities, while a large number of cul–de–sacs on local streets

discouraged walking in the Portland, Oregon metropolitan area. The individuals in higher

income households and those owning more vehicles exhibited a greater tendency to drive

alone to non–work activity sites than the individuals in lower income households. However,

no difference was observed in their propensity to choose among other modes. The older

individuals used carpool or vanpool more and used transit less than other young individuals

for non–work travel.

Vehicle kilometres travelled (VKT)

Miller and Ibrahim (1998) found that VKT per worker increased as one moved away from the

central core of the city and from other high density employment centres within the region;

this finding supported the claim of dense and compact urban form. The job–housing balance

or self containment showed little impact on commuting VKT while population density was

not strongly explaining the variations in VKT per worker across the urban area, once the

other urban structure variable was taken into account.

Further, Sun et al. (1998) qualitatively revealed that land use made a big difference in

household VKT and was important in determining household travel patterns, whereas its

impact on the number of daily trips was limited.

Similarly, McCormack et al. (2001) observed that residents of mixed land use study

neighbourhoods in Seattle, Washington travelled 28 percent fewer kilometres than residents

in adjacent areas and up to 120 percent fewer kilometres than residents in suburban areas.

This trend of lower travel distances held across different socioeconomic characteristics.

However, the differences in travel distances among the areas were not seen when travel time



was considered, because the inhabitants of mixed land use development travelled with slower

speeds by using slower modes travel (mainly transit and walking) when compared to

residents of other areas.

Home based non–work non–school trips (HB NWNS trips)

The relative share of land devoted to commercial service uses in the zone of a trip origin

increased the likelihood of making HB NWNS walking trips, whereas the relative share of

vacant land decreased the probability. The population density in the zone origin showed no

significant effect on HB NWNS walking trips (Zegras, 2004). This evidence supported the

claim of having increased walking trips with TOD.

Automobile ownership

The probability of owning an automobile was found to decrease by 31 percent as land use

changed from homogeneous to diverse (traditional neighbourhoods) indicating the strong

influence of mixed land use on automobile ownership. Walking and public transit were found

to be more suitable alternatives than private vehicle use due to mixed land uses (Cervero,

1996; Hess and Ong, 2002).

Car ownership was found to be a function of population density. Owning and keeping a car in

a very dense area was found to be more expensive and difficult than in less dense areas

(Giuliano and Dargay, 2006). This will result in reduced car ownership levels and increased

transit usage, supporting the TOD assumption.

Mode choice and density

Cervero and Gorham (1995) observed that walking and bicycling mode shares and respective

trip generation rates were considerably higher in transit neighbourhoods (consequently lower

drive alone mode shares and trip generation rates) than in car oriented neighbourhoods in the

San Francisco Bay Area and in Southern California. The transit neighbourhoods averaged

higher densities and had a more gridded street patterns compared to their counterparts with

car–oriented developments. Significant interaction was observed between neighbourhood

type and densities in both the areas.

The likelihood of non–auto commuting increased significantly with higher neighbourhood

densities and with the presence of shops and other non–residential activities found in the

immediate neighbourhood. It was observed that the relative proximity of mixed–use

development mattered greatly if the retail shops were within 100m of the dwelling unit. Then

Literature review


workers were more likely to commute by transit, foot or bicycle. However, beyond this

distance, mixed use activities induced auto–commuting (Cervero, 1996). Chatman (2003)

found that the residential density was correlated with transit and walking convenience for

commute but did not directly influence personal commercial VKT.

The daily travel distance was found to be inversely related to local population density (more

strongly in the US than in Great Britain) supporting the claim of high densities (Giuliano and

Dargay, 2006). A sensitivity analysis carried out for reviewing the development density

regulations at the Chunghsiao – Fuhsing station area in Taipei showed that enlarging the

upper bound of ratios of floor space to site space (RFS) can increase subway ridership, but at

a cost of reducing social equity and living environment (Lin and Gau, 2006). The RFS greater

than 70 percent has not shown significant increase in subway transit ridership. The living

environment and social equity aspect was considered along with economic efficiency of

transit ridership. It suggested that if cities use land use policies to offer options to drive less

and use transit and non–motorised modes more, many residents will do so (Cao et al., 2007;

Cao et al., 2009).

Intrazonal trips

Greenwald (2006) observed that intrazonal trips were shorter in distance and associated with

activities that were more quickly completed and are more likely to be completed by walking

or biking compared to private automobile. The mode and destination choice was found to be

influenced by elements of urban form. Street design and housing concentrations were found

to be less significant in keeping the trips intrazonal. On the contrary, variety and scale of

economic activity with sufficient scale (proximity and diversity of land use) played a

significant role in keeping trips intrazonal.

Parking requirements

Steiner (1998) stated that the reduction in parking and transport fees was misguided for TODs

because of higher attraction rates of shopping trips from adjacent suburbs and when the

demand for shopping centres specifically for Saturday peak loads was considered. This study

considered the parking requirements only for shopping centres and did not consider other

parking requirements, mainly being residential and office parking requirements.



Walking

While comparing the alternative approaches for exploring the link between urban form and

travel behaviour, Handy (1996) found that the difference between traditional and

conventional neighbourhoods was in the likelihood of walking to a store or other local

business; residents of traditional neighbourhoods, on average, made two or more times as

many walk trips to a store as residents of conventional neighbourhood.

Schlossberg and Brown (2004) concluded that the presence and location of pedestrian–hostile

streets had a significant, negative influence on the pedestrian environment surrounding transit

stops, often cutting off more–pedestrian–friendly environments from the transit stops. This

emphasised careful planning of street network and sidewalks, but this analysis has not

indicated the effect of pedestrian–hostile or pedestrian–friendly environments on travel

characteristics of people. Further, Khattak and Rodriguez (2005) suggested that the

households in the neo–traditional development substitute driving trips with walking trips.

Living in a TOD and transit use

The results of a survey done to assess the reasons for choosing to live in a TOD and

associated transit use showed that one third of the respondents reported access to transit as

one of the top reasons for living in a TOD. The surveyed residents were equally or more

likely to choose to live in a TOD because of lower housing cost or the quality of

neighbourhood. This indicated that TODs meet a range of needs. Further, the residents who

cited access to transit as one of the top three reasons for choosing transit to live in a TOD

were 13 to 40 times more likely to use transit than those who do not. In general, it was

observed that the TOD residents used transit at a relatively higher rate compared to the

population as a whole (Lund, 2006; Dill, 2008).

Deakin et al. (2004) reported that residents of TOD chose intentionally to live in a transit –

and pedestrian friendly area and own and use cars less often than average; the residents

included many non–students. Overall, the parking management and TOD were found to be

successful in supporting reduced car ownership and use in Berkeley, California. The evidence

for Taipei showed that land use diversity variables did not significantly increase metro

ridership. Hence it was concluded that the built environments consistent with TOD as

identified by Western studies were not always suitable for Taiwanese cities (Lin & Shin,

2008). On the other hand, Cervero and Day (2008) concluded that transit–oriented

Literature review


development holds considerable promise for placing rapidly suburbanising Chinese cities on

a more sustainable pathway.

Traffic generation

The Transit Cooperative Research Program (TCRP) Report 128 (Arrington & Cervero, 2008)

confirmed that TOD housing produced considerably less traffic than is generated by

conventional development. In other research and of particular interest to this study, Institute

of Transportation Engineers (ITE) trip generation and parking generation rates found to

overestimate automobile trips for TOD housing by approximately 50 percent (Arrington and

Sloop, 2009).

2.2.4.3 Stated benefits vs. reported benefits

In conclusion, the overall assessment of reported benefits of TODs shows positive results,

supporting the claimed benefits of TODs. The evidence of TOD was stronger in the US when

compared to UK and developing countries. Existing urban fabric forms in the UK and

developing countries might be more like TODs than those in the US. The studies examined

here were mainly concentrating on mode choice (walking and transit) and VKT. The findings

are briefly summarised below:

The presence of a nearby retail shopping centre was a major contributor in increasing

the number of walking trips for workers and residents of a TOD. In addition, a

significant amount of reduction in VKT and increase in transit trips was observed.

Some evidence of the reduced parking requirements was also found for office

buildings, residents of TOD, but the claim did not support commercial uses, while a

study for educational and recreational parking requirements was not found.

The presence of mixed uses and density was found to affect walking trips

substantially.

The scant research investigation on job–housing balance and travel times did not

show significant effect on travel. These variables need to be addressed properly to

comment on the contribution of self containment and effect or saving in travel time

due to TOD.

The trip lengths, travel costs for individual trip purpose needs to be studied further if

the transport sustainability claims of TODs made by proponents are to be supported.



2.3 Data requirements for TOD evaluation

2.3.1 Background

Travel demand evaluation provides an insight to the travel characteristics and in turn the

travel behaviour of people. To conduct travel demand evaluation precisely, the acquired data

needs to be accurate, reliable, based on real situations and representative of the practical

behaviour of people living in the area which is considered for assessment, and obtained with

limited budget and resources (Sharp and Murakami, 2005).

Generally for the purpose of data collection, a new Origin–Destination (O–D) survey is

proposed for a study area in two cases; when no O–D survey has been performed in the area

before and the existing models of another area cannot be applied successfully, and secondly,

when the models based on previous O–D surveys yield unsatisfactory results or some major

land use changes has been made which have altered the travel behaviour (Smith, 1979).

Typically, two types of transport surveys are used for travel demand determination; revealed

preference (RP) surveys and stated preference (SP) surveys. The revealed preference surveys

collect information regarding the actual travel behaviour and travel characteristics of

respondents. On the other hand, stated preference surveys gather responses or preferences of

the respondents based on the hypothetical travel options given to them. The evidence suggest

that the preferences derived from SP surveys are contingent on context. Also, it is extremely

difficult to identify core preferences based on SP surveys as the stable core preferences may

not exist prior to a choice (Fujii and Gӓrling, 2003).

The studies listed here are based on the revealed preference surveys which are proposed for

this research. The revealed preference surveys were proposed because this research aims at

assessing the actual travel behaviour of users of a functioning TOD. Also, it should be noted

that the revealed preference surveys do not suffer from the drawbacks mentioned before in

case of stated preference surveys. The following sections present an overview of revealed

preference travel surveys by describing their various aspects such as type of data to be

collected, types of travel surveys and survey instruments, their design, and response rates.

2.3.2 Data need / data requirement

The overall assessment of the travel demand literature shows that the data gathered include

attributes at the individual level and attributes at the zonal or network level. Typically, a

travel survey collects data at personal and household level. The responses of the travel survey

Literature review


are then aggregated to obtain the attributes at the zonal level. The various attributes of a

travel survey include:

2.3.2.1 Demographic and Socio–economic characteristics

The review of literature indicates that the demographic data and socioeconomic data are the

most widely used data sets in travel demand assessment. The demographic data set involves

individual or personal information about age, gender, employment status of each household

member and the details of households such as vehicle ownership, number of licensed drivers,

household size (including number of children), and location of residence. The socioeconomic

data set consists of details about household income, dwelling type, number and type of car

owned (Allen and Perincherry, 1996; Sun et al., 1998; Hess and Ong, 2002; Newmark et al.,

2004). These variables are mainly used to forecast trip generation.

2.3.2.2 Travel characteristics

The data related to travel of individuals is often collected by using a 24h activity travel diary

for a mid weekday. This diary gives detailed information of each trip (also called activity)

about purpose, frequency, duration, mode choice, number of persons accompanied, use of

household vehicle, travel cost and parking cost (if any) (Goldenberg, 1998; Hess and Ong,

2002; Rajamani et al., 2003; Newmark et al., 2004).

The questionnaire survey by Zakaria (1986) included some questions about the perceptions of

the quality of the public transport system. In addition to this, the data on land use patterns and

urban design factors (such as land use mix, street network (width), accessibility, residential

density and pedestrian connectivity) were used in many studies to address the mixed land use

and the urban form (e.g. Hess and Ong, 2002; Rajamani et al., 2003).

The data gathered is primarily used for estimating various attributes related to population

(such as car ownership level, classification of households) and travel characteristics (such as

VKT, mode choice, trip length, travel cost, etc). The same data is also used for modelling and

statistical analysis, economic development, land use planning and social service delivery. For

developing a travel demand model based on the four step modelling approach, demographic

data is mainly used for trip generation and travel data is used primarily in trip distribution and

mode choice. A separate data set is collected for calibrating the model.

2.3.3 Types of travel surveys

Transport planners traditionally use census data along with small sample surveys for

employment for travel demand modelling. But in order to validate and calibrate a travel



demand model, a database is needed which will also used to forecast the travel characteristics

of people. In order to obtain more reliable data, different travel surveys need be planned for

different sets of people involved. Some examples are listed below:

2.3.3.1 Household travel survey

A household travel survey is the most common source for obtaining travel data for

developing travel demand models. The mail back survey technique is most commonly used

for household surveys.

Crevo et al. (1995) presented techniques, results and costs of a statewide mail–out or mail–

back survey for Vermont, USA. Postal codes were used to generate the sample instead of a

telephone directory. The response rate of 8 to 10 percent was obtained due to a lack of

advance contact and complexity of the survey.

Table 2.3 An overview of household travel surveys

Source Name of travel

survey

Location of

survey

Type of survey

instrument

Sample

size*

Response

rate

SEQTS

(2008)

South East

Queensland

Travel Survey

(SEQTS)

South East

Queensland,

Australia

Hand delivered and

hand collected survey

accompanied by

telephone and postal

reminders

5,671

households

52.7%

PARTS

(2010)

Perth and

Regions Travel

Survey

(PARTS)

Perth, Australia Personal delivery and

pick up methodology

10,947

households

–

TDC (2010) Household

Travel Survey

(HTS)

Sydney Greater

Metropolitan

Area, Australia

Face to face

interviews

14,409

households

66.0%

Sharp and

Murakami

(2005)

National

Household

Travel Survey

(NHTS)

United States of

America

Two stage telephone

interviews

26,000

households

41.0%

MOT

(2006)

New Zealand

Household

Travel Survey

(NZHTS)

New Zealand Personal interviews 5,367

households

74.9%

DOT (2005) National

Household

Travel Survey

(NHTS)

South Africa Face to face

interviews

45,346

households

86.6%

Anderson et

al. (2010)

National travel

Survey (NTS)

United

Kingdom

Computer-Assisted

Personal Interviewing

(CAPI)

8,384

households

62.0%

SIKA

(2007)

RES Sweden Telephone interviews

supported by journal

entries

41,225

persons

67.6%

Note: The time period for which the travel surveys are conducted varies.

Literature review


Various nationwide household travel surveys were conducted by different agencies or

government institutions to study travel behaviour of people living at various places. Table 2.3

shows an overview of the state wide household travel surveys conducted for Australia and

some other national household travel surveys.

2.3.3.2 Employee travel survey

Generally, the journey to work and other related trips are not usually modelled by collecting

the employee data separately due to the large expenses involved. So acquiring quality

employment data is a major problem in transport modelling. The employment data is

particularly important in trip generation and trip distribution steps of travel demand

modelling (Souleyrette et al., 2001).

Zakaria (1986) described the development and findings of a mail back questionnaire survey

sampled for 236,000 employees for Center City Philadelphia. The survey included questions

on trip characteristics, usage of highway and transit modes, socioeconomic characteristics

and place of residence and work for assessing travel behaviour of employees.

2.3.3.3 Students’ travel survey

McMillan (2007) examined the influence of urban form and non – urban form factors on

children’s travel mode to school. She conducted surveys to identify key variables influencing

the decision about a child’s trip to school. The survey was conducted for elementary schools

(grade 3 – 5), surveys were distributed to the children to bring home to their caregiver for

completion and the completed survey was collected back at the school via child.

No specific survey has been found which is designed to study the travel characteristics of

university or high school students.

2.3.3.4 Shoppers’ survey

The relationship between land use and shopping travel behaviour, for before and after the

introduction of fringe shopping malls in the Prague metropolitan area, Czech Republic was

examined by Newmark et al. (2004). The shoppers were surveyed by doing a personal

interview survey at four malls and trip frequency, shopping activity duration, and mode

shares were compared against age groups, gender, income, car ownership and household size.

With the provision of new fringe shopping centres, patrons made fewer, longer trips and tend

to shift travel mode from pedestrian to vehicle, particularly the private vehicle.



2.3.4 Types of survey instruments

The previous section (Section 2.3.2) implies the extensive data requirements for travel

demand assessment purposes. Hence data collection is an inevitable part of the evaluation

process with the exception of a few cases where relevant data has already been collected for

some other purpose. Designing and conducting surveys is a critical aspect of this step.

Stopher & Greaves (2007) gave information about the various survey methods and

improvements or advancements in the survey methods from 1970’s to the present, listing the

changes in the data needs and survey methods. The standard method of conducting most

household travel surveys was found to be a diary, however the latest was GPS assisted trip

data collection.

Observational surveys, household self completion surveys, telephone surveys, intercept

surveys, household personal interview surveys, group surveys, in depth interviews, GPS

surveys, postal surveys and internet based household travel diary surveys were listed as

survey methods used for conducting a travel survey (Richardson et al., 1995). Characteristics

of four popular survey methods are discussed briefly based on the literature; GPS survey is

quoted as an emerging technique.

2.3.4.1 Home or personal interviews

Home or personal interviews produce more reliable and clearer data sets as the interviewer is

present at the time of obtaining responses and can help to better explain the survey questions.

However these surveys are time consuming and hence most expensive (Crevo et al., 1995;

Sharp and Murakami, 2005).

2.3.4.2 Telephone surveys

Telephone surveys are efficient and require fewer household contacts to obtain estimated

responses. Hence, it has been the most commonly used method in the United States in recent

years (Sharp and Murakami, 2005). Generally a random sample generation technique is used

for sample generation, but this technique does not guarantee a geographically representative

sample.

A computer assisted telephone interview (CATI) survey is sometimes used to increase the

response rate compared to postal surveys (Crevo et al., 1995; Stopher and Greaves, 2007).

One disadvantage of telephone surveys is that it is known that a significant numbers of trips

are omitted or are under reported at the time of telephone interview.

Literature review


2.3.4.3 Mail back surveys

Mail back surveys are mostly preferred due to their low cost and low administrative burden;

in addition mail back surveys can ensure representative samples by mailing to a greater

number of households (Crevo et al., 1995). Generally, postcard reminders and telephone calls

are used to increase the response rates. Pre–addressed, pre–paid envelopes are enclosed in the

survey envelopes so that respondents are not required to bear the expenses for returning back

the survey.

A self administered mail back survey was noted as a viable, cost–effective approach to gather

detailed information about household and personal travel information (Poorman and Stacey

1984; Crevo et al., 1995).

2.3.4.4 Internet based surveys

The internet based survey is a new cost–effective platform, becoming popular because of its

opportunity to deploy the important branching and data control features of the computer–

assisted telephone interview (CATI) instrument and the graphical features of the computer–

assisted self–interview (CASI) instrument.

For the surveyor, use of e–mail provides a convenient way to communicate with the

respondents and for the respondents this gives flexibility to complete the questionnaire at any

suitable time. When tested with other instruments, namely with conventional telephone and

mail administration technique for a household survey, the internet instrument had more trips

reported and lower non response rates (Alder et al., 2002).

In continuation to the earlier work, similar results were observed for the internet based

surveys by Abdel–Aty (2003) while performing an O–D travel survey for a toll plaza in

central Florida. An ordered Probit model was developed to explore the effects of individual

characteristics on the number of blank answers given. The individuals answered a specific

number of questions and the differences between mail and internet response were studied by

using a log–linear model.

Dillman (2006) listed criteria and principles of designing respondent friendly web

questionnaires. The major sources of errors were stated as coverage error, sampling error,

measurement error and non response error. Hence, this type of survey is suitable in the area

with web access and with a community having high computer usage.



2.3.4.5 GPS surveys

GPS surveys have been proposed as a supplement or replacement for diaries, panel surveys,

continuous measurement surveys and data fusion. The advantage is that the respondents are

not required to fill any questionnaire survey forms; instead they have to carry a GPS with

them for each trip undertaken (Stopher & Greaves, 2007). But these surveys include a higher

cost for procurement of a GPS unit for each member of the household.

Mixed approaches are being commonly used to strike a balance between survey costs and

data–quality issues, also most often response and coverage.

2.3.5 Design of survey instruments

2.3.5.1 Steps in design

The design of a survey instrument involves methodological and design considerations, as

listed in Sharp and Murakami (2005). The steps involved in designing a survey instrument

are (Smith, 1979):

Decision of the purpose of the survey

Decision of the variables to be measured to fulfil the survey questions

Decision of suitable survey method to measure the variables adequately

Determination of coefficients of variation of the variables in the question

Decision of level of accuracy and confidence limit

Computation of sample size

The steps for conducting a travel survey generally include sample generation, questionnaire

delivery, reminder call or postcard, data collection and data entry (Goldenberg, 1998).

2.3.5.2 Selection of survey instrument

The method of data collection is largely dictated by the population coverage and sample

frame; other determinants include survey costs, response rates and data quality issues. The

method selection can also be influenced by the complexity and the length of survey and

timeliness needs (Sharp and Murakami, 2005).

Goldenberg (1998) examined methodological options and interactions between five options;

24–h versus 48–h recording period, shorter versus longer series of questions about each

activity recorded, three types of incentives, booklet versus log format of diaries and telephone

versus mail back survey. Chi–square tests were conducted to consider the effects of

methodological options on response rates, and the effects of different pre–test design factors

Literature review


on completion rate were assessed using a logistic regression model. The 24–h diary and

telephone surveys were found to have greater completion rates.

2.3.5.3 Design of questionnaire form

Design of questionnaire requires careful consideration of the data needs, the questions

included should give the variables needed for planning or modelling. Generally the questions

about demographics and socio–economic characteristics are asked at the end of the

questionnaire (Zakaria, 1986). An introductory letter is always useful in offering a good

understanding or overview of the survey. The questions should be objective in nature so that

they are easy to answer. Keeping the layout simple and giving the appropriate choices

(options) reduces confusion and helps to increase the responses. Pre–tests or pilot studies are

performed in most cases to assess the suitability of the method and discover the quality of

questions based on the responses received.

2.3.5.4 Pilot surveys

Pilot testing is one of the most important components of the survey procedure and also can be

one of the most neglected. Pilot surveys or pre–tests should be performed to guide adequacy

of the sampling frame, variability of parameters within the survey population, non–response

rates, method of data collection, question wording, layout of questionnaire, adequacy of the

questionnaire in general, efficiency of interviewer and administrator training, data entry,

editing and analysis procedure, cost and duration of survey and efficiency of survey

organisation. The size of a pilot survey is related to the size of the main survey; as a rule of

thumb 5 to 10 percent of budget is expected to be spent on pilot surveys (Richardson et al.,

1995).

2.3.5.5 Sample selection

Generally, use of area telephone directories was found to be prominent in the literature for

sample selection. The number of samples varies mainly according to the population size and

size of the area surveyed. Generally, a sample size of 2 to 5 percent of the population is used

in the literature.

Smith (1979) illustrated a procedure to estimate the required sample size from the local data

for conducting a home–interview origin–destination (O–D) survey based on the sample sizes

required to calibrate the travel demand models rather than the sample sizes required to

duplicate travel patterns. It was also noted that travel demand models can be developed from

a survey of less than 1,000 (900 – 1,200) households. The formula for computing sample size

is as given in Equation 2.1.



𝑛 = 𝐶2𝑍2 𝐸2 Equation 2.1

where,

𝐶 = coefficient of variation

𝐸 = accuracy level expressed as a proportion rather than a percentage

𝑛 = number of samples and

𝑍 = normal variates corresponding to the confidence limit selected

In 1986, Zakaria used a cluster sampling procedure for distribution of survey questionnaire to

employees after considering practical, administrative and cost constraints. The sample size

was determined on the basis of specified levels of sampling error and confidence interval in

the survey results, using a relationship as shown in Equation 2.2. The sampling error was

high if the number of responses to a question was small.

ℎ = 𝑧2 𝑛 𝑝 ∙ 𝑞 1 2 Equation 2.2

where,

ℎ = specified (tolerable) sampling error

𝑧 = confidence interval or the multiple of standard errors corresponding to

the specified probability of obtaining the specified precision

𝑛 = sample size

𝑝 = probability that the population possesses certain characteristics

𝑞 = 1 − 𝑝

A new parcel based sampling strategy, called spatial sampling, was used to draw respondents

from a spatially delineated frame (Lee and Moudon, 2006).

2.3.6 Response rate

The response rate for a travel survey was found to vary depending upon the survey method,

follow up procedure and incentives offered. The response rate also fluctuated depending upon

the age, sex, occupation and income (Alder et al., 2002; Newmark et al., 2004; Abdel–Aty,

2003). The response rates were high for the surveys which offer some kind of incentives to

the respondents. Repeated reminders and resending of the questionnaire were found to

increase the response rates.

Literature review


Korimilli et al. (1998) estimated a linear regression model to predict response rates with a

sample of 35 transport surveys, using survey length, number of reminders, number of survey

stages, number of questions, geographic location of survey, method of survey administration,

type of survey instrument and presence of incentives as the design parameters.

2.3.7 Data analysis

Before analysing the data, the details collected by various survey methods for travel surveys

needs to be entered and organised in a format that is suitable for data analysis. Due to

advances in the computer technology, to assess the large amount of data collected, the data

processing is performed by computer using spreadsheet and database programs (such as

Microsoft Excel, Microsoft Access, Louts 1–2–3, dBase, FoxBASE/FoxPro). Zakaria (1986)

analysed the responses using a FORTRAN program with Contingency checks. Two types of

analysis are mainly made on the data; explanatory analysis and statistical analysis. The details

of these methods of data analysis are discussed in Section 2.4.

2.3.8 Summary of data collection

The data collection involves multiple aspects starting from selection of appropriate

methodology for conducting surveys through to obtaining a good response rate and

organising the collected data for analysis. A summary of different survey tools is given in

Table 2.4. The data collection method differs depending on the type of respondents. Personal

interviews were found to be the most efficient but most expensive tool for conducting the

data collection exercise. The variables required for analysis dominated the selection of survey

instrument. In addition, the pilot surveys or pre–tests were quoted as critical for improving

the quality of questions and response to the questions by a respondent.

The selection of sample was generally made randomly and should be such that it should

represent the population characteristics. Follow up mails, telephone calls and incentives were

highlighted as techniques to increase the response rates. The data collected then will be used

to perform statistical analysis or travel demand modelling. The techniques for data analysis

are explained in the following section.



Table 2.4 Summary of different survey instruments

Author Survey

instrument Description Inherent bias

Crevo et al.

(1995); Sharp and

Murakami (2005)

Home or

personal

interviews

Time consuming and most

expensive technique.

Give good quality of data and

achieve higher response rates

Biased responses as the

respondent is being

observed by interviewer

Sharp and

Murakami (2005);

Crevo et al.

(1995); Stopher

and Greaves

(2007)

Telephone

surveys

Popular method in US, obtain

better response rates than mail back

surveys.

Significant shortfalls in trip

reporting, ethical barriers in

obtaining the telephone numbers,

selected sample does not ensure

population representation

Excludes people with no

telephone facility

Poorman and

Stacey (1984);

Crevo et al. (1995)

Mail back

surveys

Traditional method, higher non

response rates and underreporting

of trips than telephone surveys, can

ensure representative sample, non

response bias can not be evaluated

Excludes unstructured /

choice based questions,

restricts the format of

questionnaire design

Exclude young children

and illiterate population

Alder et al. (2002);

Abdel–Aty (2003);

Dillman (2006)

Internet

based surveys

Cost effective method, requires less

resources and good internet access

for respondents, can track

continuity and completeness of

reported trips, observes lower

response rates

Excludes non–computer

users, this will have an

age and underprivileged

bias

Stopher and

Greaves (2007)

GPS surveys Precise data can be collected for

multiple days.

Higher cost for procurement of

GPS unit and can face problems of

signal loss at some places.

Limits responses to

technically sound users,

geographical area bias

2.4 Travel data analysis

2.4.1 Background

TOD enables clustering of activities with a wide range of choices offered by a combination of

mixed land use, pedestrian friendly environment, good quality public transport service,

increased density and affordable housing. Looking from the transport point of view, the

presence of all these things at a place makes the travel behaviour different, so needs to be

studied or analysed specifically. Previous studies have tried to establish a link between these

variables and travel behaviour of people but fail to suggest a suitable methodology for overall

travel demand assessment and in turn the travel behaviour of people living in and using a

TOD. The previous studies are explained below for their methodology used and a summary

of all studies is presented in Section 2.4.4.

Literature review


2.4.2 Comparative and statistical analysis

Handy (1996) reviewed the alternative approaches to explore the link between urban form

and travel behaviour and listed strengths and weaknesses of each basic approach. The studies

were classified into five categories; namely simulation studies, aggregate analyses,

disaggregate analyses, choice models and activity based analyses. She found that most of the

studies fell into the first three classes.

2.4.2.1 Parking requirements

Transport and land use planners set parking requirements for TODs (both maximum and

minimum) to encourage transit use and avoid excess parking supply. Higgins (1993)

presented a method for setting parking requirements for office, commercial and industrial

developments in vicinity to transit stations and stops. The method was illustrated using 1991

employee transport survey data for the city of San Diego, California. Employee densities for

various land uses were used with other variables like mode share, number of visitors or

shoppers per employee and proportion of walk–in shoppers, their mode of travel, vehicle

occupancy and volume of shoppers in normal versus holiday periods. The maximum and

minimum parking demand was derived on the basis of employee mode shares. The parking

demand was observed to be sensitive to employee density.

Developers of TODs claims reduced parking requirements and transportation impact fees as a

result of reduced level of automobile usage at such places. To test this claim, Steiner (1998)

compared the parking requirements of six prototypical traditional shopping districts in the

Oakland–Berkeley subarea of the San Francisco Bay Area, California based on the trips

generated in each shopping centre. It was found that, the claim of reduced parking

requirements can not be supported wholeheartedly, if the Saturday peak loads are considered.

Deakin et al. (2004) presented findings of a study conducted in downtown Berkeley,

California, during autumn 2002 which focused on land use, parking supply and use, mode

choices, and housing and jobs development. The multiple roles of parking management and

efficacy of TOD in smaller cities were also highlighted. Four special studies were conducted

to understand specific transport issues which include surveys for workers, shoppers, residents

and analysis of an on–street parking occupancy and turnover. A separate survey of new

housing in downtown was carried out. Intercept method was used for conducting the workers

and shoppers surveys while 2000 census data were collected for the residents survey along

with supplementary interviews. Parking occupancy and turnover for each parking space were

checked out hourly between 9 am and 5 pm on weekdays in the study area.



2.4.2.2 Car ownership

Hess and Ong (2002) quantified the effect of traditional neighbourhood land use patterns on

the automobile ownership using Portland, Oregon as a model. The probability of automobile

ownership was calculated by using Equation 2.3. An ordered logit regression was used to

model the decision to own zero, one, two, or three or more cars as dependant variable.

𝐴𝑢𝑡𝑜𝑚𝑜𝑏𝑖𝑙𝑒𝑜𝑤𝑛𝑒𝑟𝑠ℎ𝑖𝑝𝑖 = 𝑓 𝐻𝑖 ,𝑁𝑖 , 𝐿𝑖 Equation 2.3

where,

𝐻𝑖 = vector of household characteristics

𝑁𝑖 = vector of neighbourhood characteristics

𝐿𝑖 = vector of urban design characteristics

An international comparative analysis of relationships between car ownership, daily travel

and urban form using travel diary data from US and Great Britain (GB) was conducted by

Giuliano and Dargay (2006). Metropolitan size and population density were used as measures

of land use. A structural model including car ownership was estimated along with a reduced

form of model without car ownership. The total distance travelled by individual travel in

miles for the entire day by all modes was given by Equation 2.4. The results showed that

distance travelled was inversely related to residential density more strongly for non–work

travel than for work travel and for GB than US. Metropolitan size was found to have no

consistent influence on travel. The car ownership at the household level was represented by a

discrete variable and modelled using ordered probit specification.

𝑌𝐷 = 𝑓 𝑋𝐷 ,𝑇𝐷 , 𝐿𝐷 ,𝑢 Equation 2.4

where,

𝑌𝐷 = daily travel distance

𝑋𝐷 = attributes of the individual

𝑇𝐷 = transportation resources available to the individual

𝐿𝐷 = attributes of the residential location

𝑢 = unobserved factors

Literature review


2.4.2.3 Transit use

Messenger and Ewing (1996) described the variables affecting transit ridership in the

metropolitan Dade Country, Florida by using 1990 data. Five different variables were used to

model bus mode shares; socioeconomic variables, land – use variables, street network

variables, transit service variables and other (interaction) variables. Initially, the analysis of

bus mode share by place of residence was started with the estimation of a single equation

using stepwise regression analysis (ordinary least squares). Then a system of three equations

were used to model the bus mode share and the full – information maximum likelihood

(FIML) method was used to estimate the system of equations simultaneously. The bus mode

share by place of residence was found primarily to be dependent on automobile ownership

and secondarily on jobs–housing balance and service frequency, while the bus mode share by

place of work was dependant on the cost of parking, transit access to downtown, and overall

density through a web of interrelationships. The overall density had large indirect effect on

bus mode shares through car ownership and parking charges.

Lund (2006) reported the results of a survey of households (605 people) who moved to TODs

within past 5 years in the San Francisco Bay Area, Los Angeles or San Diego and studied the

factors that lead these households in TODs to move to a TOD and its implications on transit

use. Binary logistic regression analysis was used to predict the probability that a survey

respondent cited a particular factor as one of their household’s top three reasons for choosing

to live in a TOD.

The influence of built environment based on TOD on the level and temporal distribution of

metro ridership was examined using 46 metro stations in Taipei City, Taiwan, China. Two

regression models (one for weekday and another for weekend) were calibrated by using the

ordinary least–squares (OLS) method. The empirical results showed that daily ridership was

positively affected by the floor space area of the station areas, negatively affected by the

percentage of four–way intersections, and insignificantly affected by mixed land use. The

ridership dispersion in time was positively influenced by sidewalk length, negatively affected

by retail and service floor–space area, and insignificantly influenced by density (Lin and

Shin, 2008).

2.4.2.4 Travel mode

Binomial logit regression probability models were used to examine the likelihood of a child

walking or bicycling to school versus travel by private vehicle or neighbourhood carpool

(McMillan, 2007). The results of the analysis supported the hypothesis that urban form is



important, but not the sole factor that influences a caregiver’s decision about a child’s trip to

school. Other factors may be equally important such as neighbourhood safety, traffic safety,

household transportation options, caregiver attitudes, social or cultural norms and socio–

demographics.

The relationships between five urban form variables and walking in specific demographic

subgroups were assessed using stratified logistic models and controlling for participant

demographics (Kerr et al., 2007). The travel data for two day period from 3161 youths

between 5 and 18 years of age in Atlanta, US was considered for analysis. All five urban

form and recreation measures were related to walking among people from European ancestry,

but only land use mix and access to recreation spaces were significantly related to walking in

people not of European ancestry. There were more significant urban form physical activity

associations in high–income than in low–income households. More urban form variables

were related to walking in households with three or more cars than in households with no

cars. Living in mixed use–areas and having access to recreational space were related to youth

walking for transport in 11 of 13 population subgroups studied.

Cao et al. (2009) explored the relationship between the residential environment and non work

travel frequencies by auto, transit, and walk or bicycle modes controlling for residential self

selection by using the seemingly unrelated regression equations (SURE) model. The model

showed that neighbourhood characteristics were associated with individuals’ travel decisions,

especially non–motorized travel frequency. The mixed land uses tended to discourage auto

travel and facilitate the use of transit and non–motorized modes; the availability of transit

service and walking or biking infrastructures were important predictors for transit and non–

motorized travel; and walking or biking behaviour was also affected by the aesthetic quality

and social context of the built environment.

2.4.2.5 Travel behaviour

Khattak and Rodriguez (2005) examined differences in travel behaviour in a matched pair

neighbourhoods (one conventional and one neo–traditional) in Chapel Hill and Carrboro,

North Carolina. A detailed behavioural survey of 453 households and two–stage regression

models suggested that single–family households in the neo–traditional development made a

similar number of total trips, but significantly fewer automobile trips and fewer external trips,

and they travelled fewer miles, than households in the conventional neighbourhood, even

after controlling for demographic characteristics of the households and for resident self–

selection.

Literature review


A structural modelling approach was used to explore the relationships among changes in the

built environment, changes in auto ownership, and changes in travel behaviour (Cao et al.,

2007). The residential self selection was addressed in several ways. The study found that the

residential self–selection has significant direct and indirect impacts on travel behaviour. The

changes in the built environment have a statistically significant association with changes in

travel behaviour, controlling for current attitudes and changes in socio–demographics, and

taking multiple interactions into account. The influence on the built environment was found

not only to be statistically significant but also practically important.

2.4.2.6 Modelling vehicle availability

Generally, the emphasis of analysis is mainly on whether the residents walk or use transit or

how many residents drive their car. But instead of looking into this aspect we should study

whether the residents have any other competent option to their existing travel choices

(Handy, 1996). In order to undertake this type of study vehicle availability needs to be

considered and quality of service needs to be assessed subsequently.

An improved model for forecasting vehicle availability was described by incorporating transit

accessibility and land use indicators along with usual demographic variables using a two–

stage approach (Allen and Perincherry, 1996). A look up table was used to identify an initial

estimate of the proportion of the number of vehicles on the basis of the household’s size,

number of workers and income quartile in the first step. An incremental logit model as shown

in Equation 2.5 was applied to initial proportions to consider the effect of transit accessibility

and land use form. Equation 2.6 was used to calculate the utility function. The model was

validated using 1990 PUMS records for the Washington region. It was found that good transit

service and high development density were associated with lower vehicle ownership.

𝑝𝑓𝑣 =𝑝𝑖𝑣 × 𝑒𝑈𝑣

𝑝𝑖𝑣 × 𝑒𝑈𝑣 3+0

Equation 2.5

𝑈𝑣 = 𝑎𝑣 + 𝑏𝑣 × 𝑇𝑟𝑛𝐴𝑐𝑐 + 𝑐𝑣 ×𝑊𝑎𝑙𝑘𝐴𝑐𝑐 + 𝑑𝑣 × 𝐸𝑚𝑝𝑙𝐼𝑛𝑡 Equation 2.6

where,

𝑈𝑣 = utility of accessibility in the vehicle availability decision for vehicle

group 𝑣 (dimensionless)

𝑇𝑟𝑛𝐴𝑐𝑐 = number of employees within 𝑋 min by peak transit service

𝑊𝑎𝑙𝑘𝐴𝑐𝑐 = number of employees within 𝑌 min walk time



𝐸𝑚𝑝𝑙𝐼𝑛𝑡 = proportion of the total region’s employment within 𝑍 mi

𝑝𝑓𝑣 = final percentage of households with 𝑣 vehicles available

𝑝𝑖𝑣 = initial percentage of households with 𝑣 vehicles available (based on

size, income and workers)

𝑎, 𝑏, 𝑐,𝑑 = coefficients that vary by vehicle availability group 𝑣

2.4.2.7 Land use

Travel Demand

Cervero (1991) studied the relationship between land use and various indicators of travel

demand for 83 office buildings at six different suburban activity centres across the United

States. Stepwise regression analyses was performed to identify the influence of project size,

density, land–use mixing, and parking facilities on three measures of transportation demand;

trip generation rates, work–trip mode splits, and automobile occupancy levels and the

strongest measures of travel demand in terms of land use factors were identified. It was

observed that the single–tenancy and mixed–use buildings were associated with low vehicle–

trip generation rates.


Cervero and Gorham (1995) compared commuting characteristics of transit–oriented and

auto–oriented suburban neighbourhoods in the San Francisco Bay Area and in Southern

California. Regression models were prepared to elaborate the relationship between

neighbourhood type, transit mode shares and generation rates. This study indicated the

importance of location of the TOD to make transit work. This research suggested that,

specifically, neighbourhood design affected the degree to which people drive alone to work

or the degree to which they walk and bicycle. But the effect of neighbourhood type on transit

commuting was less clear.

Household travel

The relative influence of socioeconomic and land use factors on households’ travel behaviour

was investigated based on the number of household daily trips and VKT (Sun et al., 1998).

The travel data from 1994 Portland Activity – Based Travel Survey and land use information

from (Portland, Oregon) Metro’s GIS data resource centre was used. The Analysis of

variance (ANOVA) process was used to investigate the significance of selected variable and

multi-linear regression analysis was conducted to study the complex relationship between

Literature review


various attributes of household travel and land use data, followed by sensitivity analysis for

analysing the elasticity of each variable.

Internal Capture

Ewing et al. (2001) studied 20 mixed–use communities in South Florida to determine the

effect of land use on internal capture rate of trips, using data from the Southeast Florida

Travel 2000 Survey. The internal capture rate was calculated by dividing the number of trip

ends for trips internal to the community with the total number of trip ends produced or

attracted by the community. The different land use measures included; size measure, density

measure, entropy measure, balance measure and accessibility measure. However, community

size and one accessibility measure were the only variables included in the final model,

concluding that the internal capture rates increased with community size and decreased with

accessibility to other regional trip attractions, while density and land use mix did not have

independent predictive powers. When tested for effect of retirement population, the

proportion of retirees did not prove significant after controlling for size and regional

accessibility.

Travel impacts

McCormack et al. (2001) empirically explored the transportation impacts of mixed land use

neighbourhoods using a two day travel diary data collected over three neighbourhoods of

greater Seattle, Washington area by comparing household location and commercial

establishment, trip stops, transit, pedestrian trips, number of trips, travel time, travel speed

and travel and socioeconomic characteristics. This data set was then compared with

countrywide identical household travel data. ANOVA technique was used to demonstrate the

variations in travel measures with household or socioeconomic categories.

Travel patterns

Coevering and Schwanen (2006) evaluated the impact of urban form on travel patterns by

considering the role of individual travellers and space–time context of cities in Europe,

Canada and the USA. The 1990 data was augmented with information on housing,

development history and socio–demographic situation and studied how these factors,

alongside land use and transport infrastructure are related to travel patterns. The analysis was

done using Ordinary Least Square Regression modelling in SPSS software.



Individual travel behaviour (walking trips)

The influence of population density, relative share of commercial and service land uses, and

relative share of vacant land on individual’s propensity to make home–based, nonwork, non–

school (HB NWNS) walking trips was analysed for assessing the influence of land use on

travel behaviour in Santiago, Chile (Zegras, 2004). The number of trips by each travel mode

was chosen by an individual to maximise the function given in Equation 2.7 built on

microeconomic characteristics of travel behaviour, subject to constraint shown in Equation

2.8. Further, the number of HB NWNS trips was estimated using an ordered probit estimation

method.

𝑀𝑎𝑥𝑖𝑚𝑖𝑠𝑒,𝑈 𝑎,𝑤, 𝑏, 𝑥 Equation 2.7

subject to, 𝑦 = 𝑥 + 𝑎𝑝𝑎 + 𝑤𝑝𝑤 + 𝑏𝑝𝑏 Equation 2.8

where,

𝑈 = a utility function of benefits using time for each purpose

𝑎 = vector of number of auto trips for each purpose

𝑤 = vector of number of walk trips for each purpose

𝑏 = vector of number of bus trips for each purpose

𝑥 = composite variable of time spent on other activities

𝑝𝑖 = representative vector of time for each trip type in each mode 𝑖

𝑦 = total available time

2.4.2.8 Density

Mixed land–use and density

Tong and Wong (1997) described the land use and transport characteristics of a high density,

mixed land–use, linear urban development located along the northern shore of Hong Kong

Island using results of a home interview survey. Major comparisons were done for car

ownership and mode choice, trip rate and trip time, and self–containment. The degree of self

containment was measured by the percentage of trips made by residents which have both trip

ends within the study area. The development was shown to have four advantages; economy in

land utilization, less roads, commercially viable public transport, and high accessibility for

residents, in spite of a low private car ownership rate.

Literature review


Allowable development densities

TOD planning generally emphasizes development efficiency, but ignores two aspects of

sustainability: environment quality and social equity. Hence, to assist TOD planners in

reviewing the development density regulations around the subway station areas, a multi –

objective programming model was developed (Lin and Gau, 2006). Based on the concepts of

sustainability, three objectives were defined as maximising subway system ridership,

enhancing living environmental quality and maintaining social equity in land development.

The ratios of floor space to site space (RFS) for different land uses were considered as

decision variables and limitations on land use density, land use combinations and level of

facility services were considered as constraints. The model was applied to review the

development density regulations in the Chunghsiao – Fuhsing station area in Taipei and

sensitivity analysis was carried out.

2.4.2.9 Modelling accessibility

Mode accessibility plays a strong role in whether or not a trip is internalised in the originating

Transportation Analysis Zone (TAZ). Higher transit and walking times between origin and

destination reduce the likelihood of keeping a trip within a zone (Greenwald, 2006).

Waddell and Nourzad (2002) tested the effects of neighbourhood and regional accessibility

on residential location, controlling for housing and neighbourhood characteristics, and using

a spatially disaggregate model. The access measure for each location was given by Equation

2.9.

𝐴𝑎𝑖 = 𝐷𝑗 𝑒𝐿𝑎𝑖𝑗

𝐽

𝑗

Equation 2.9

where,

𝐷𝑗 = quantity of activity in location 𝑗

𝐿𝑎𝑖𝑗 = composite utility, or logsum, for vehicle ownership category 𝑎, from

location 𝑖 to 𝑗, scaled to a maximum value of 0 for the highest utility

interchange

2.4.2.10 Modelling other aspects of TOD

Sadownik and Jaccard (2001) used a spreadsheet model to evaluate aggregate energy–related

emissions in the year 2015, resulting from two alternative scenarios of urban growth

throughout China. The model focused on how energy demand, residential energy technology



penetration and transportation mode choices were affected by factors of density and mix of

use in neighbourhood development.

Cervero (2006) presented examples of post–processing and direct (off–line) modelling, for

rail and transit–oriented land use proposals for greater Charlotte, the San Francisco Bay area

exurbs and south St. Louis County in California. These alternative approaches to traditional

four step modelling approach were used for scenario testing, which revealed that,

concentrating development near rail stations produced an appreciable ridership bonus.

The effects of residential relocation to Shanghai’s suburbs on job accessibility and

commuting focusing on the influences of proximity to metro rail services and neighbourhood

environments on commute behaviour and mode choice were examined by using data from

recent movers to three suburban neighbourhoods. The modelling was done using path

diagram shown in Figure 2.1. It was found that moving near a suburban rail station

significantly moderated the travel–consumption impacts of relocation, especially from the

central city to the outskirts. Notably, households that relocated in a neighbourhood served by

Shanghai’s metrorail system and lived within 1km of a metrorail station had substantially

higher access to jobs (and most likely other destinations as well) following the move than

similar households in otherwise comparable non–rail settings. Living near a suburban

metrorail station was also associated with commute–mode changes from Non–Motorised

Transport (NMT) and bus transit to rail commuting. The enhanced accessibility associated

with living in a rail–served community also correlated with reductions in the time spent

getting to and from work, controlling for other factors (Cervero and Day, 2008).

Source: Cervero and Day (2008)

Figure 2.1 Path diagram of factors influencing changes in job accessibility and commuting

behaviour

Δ Location Δ Job

Accessibility

Δ Commute

Mode

C

o

n

t

r

o

l

s

Δ Commute

Time

Literature review


2.4.3 Travel demand modelling

2.4.3.1 Four step modelling

The most common method of regional travel demand modelling is four step modelling

(Greenwald, 2006). A traditional four step model with interaction between trip generation,

trip distribution, modal split, and trip assignment is shown in Figure 2.2. The travel demand

analysis is usually done for various trip purposes. The trip purposes used for modelling the

SEQ region in Brisbane Strategic Transport Model (BSTM) (SKM, 2006) are as listed below:

Home Based Work – Blue collar (HBWB)

Home Based Work – White collar (HBWW)

Home Based Education – primary and secondary only (HBE)

Home Based Education – Tertiary only (HBET)

Home Based Shopping & personal business (HBS)

Home–based Other – including HBRec (HBO)

Work Based Work (WBW)

Other Non–Home Based – excluding WBW (ONHB)

Commercial Vehicles Medium (CVM)

Commercial Vehicles Heavy (CVH)

Source: Ortuzar and Willunsen (1994)

Figure 2.2 The classic four–stage model

Zone networks Base year data Future planning data

Database

Base year Future

Trip generation

Assignment

Evaluation

Distribution

Modal split

Output

Iter

atio

ns



The traditional four step modelling approach has been proved to be a versatile modelling

approach for modelling areas or sites with conventional methods of development. A

traditional four step model assumes homogeneous land use, does not consider short trips; the

intra zonal trips are neglected and the trips by walk and bicycle are also not considered. Apart

from this, the land use and neighbourhood features (area characteristics) are not considered in

the modelling, making the model unable to internally predict the travel demand under

changed development conditions (specifically land use and neighbourhood characteristics).

These constraints of a traditional four step model make it unsuitable for modelling the travel

demand of special developments like TODs having mixed land use, high density

development located near transit station. Hence, the model should be updated to make it

suitable for analysing TODs. Some studies have attempted to address this issue; details of

them are provided in the following sub sections.

Demand models

Rossi et al. (1993) revised the travel demand models to include the variables representing

atypical land use patterns such as residential and employment density, heterogeneity and

quality of pedestrian environment to test the alternative land use patterns as a part of a project

“Making the Land Use, Transportation and Air Quality Connection” (LUTRAQ) for

Portland, Oregon.

The main four revised components of the model were auto ownership, destination choice

model, pre–mode choice model and mode choice model. The pre–mode choice model

estimated the percentage of trips using the walk or bicycle modes for each origin–destination

pair while the mode choice model determined the number of trips using auto mode and

transit. The auto ownership model included a variable measuring the quality of pedestrian

environment, called “pedestrian environment factor”, (PEF) based on ease of street crossing,

sidewalk continuity, local street characteristics and topography. The destination choice model

was formulated as a logit model.

Similar to the above work, the Auckland Strategic Planning model (ASP) was designed to

investigate strategic futures for Auckland, New Zealand over a 30 year planning horizon

(Winder, 1994). The model represented interaction between land use policies, transport

policies, infrastructure investment, and development controls and their impact upon urban

form and the transport system. A location model, a transport model, a regional demographic

Literature review


model, a model of regional employment growth and an evaluation module were composed in

the ASP model. The transport model was implemented in EMME/2.

Miller et al. (1999) described the analytical tools for integrated land use – transportation

planning by describing available integrated models, characteristics of an “ideal” integrated

model and steps that a planning organisation should take in order to support and expand such

modelling capability, to address the transportation – air quality relationship. Six levels of

modelling capability were described with a checklist of input and analytical requirements for

each level. The integrated land use – transportation models combined the travel demand

forecasting functions with the land use forecasting functions. The idealized integrated land

use – transportation system is as shown in Figure 2.3.

Source: Miller et al. (1999)

Figure 2.3 Idealised integrated urban modelling system

In another attempt to model the emerging requirements of land use and transport modelling,

Waddell (2002) modelled the metropolitan areas by developing a discrete model system

called “UrbanSim”. UrbanSim was designed specifically to address the policy analysis

requirements of metropolitan growth management, with particular emphasis on land use and

transportation interactions. The model was applied to Eugene – Springfield, Oregon region

and was validated using data from 1980 to 1994.

Demographics

Regional economics

Government policies

Transport system

Land use

Location choice

Auto ownership

Flows, times, etc. External impacts

Activity / Travel and

goods management



Rodier et al. (2002) evaluated the effect of land use, transit and auto pricing policies in the

Sacramento, California (US), region by applying two urban models; the advanced travel

demand model, SACMET96 and the integrated land use and transport model, the Sacramento

MEPLAN model. Different scenarios were described for the light rail network to take into

account the effect of different policy measures. The land use, travel results and emissions of

both the models were compared to get some insights into heuristic policy strategies and to

determine strengths and weakness of each model.

The land use, transit policies and the addition of modest auto pricing policies were found to

reduce VKT and emissions by about 5 to 7 percent and by about 4 to 6 percent respectively in

a 20 year time horizon. The development taxes and land subsidy policies was found as not

sufficient to generate effective transit oriented land uses without strict growth controls

elsewhere in the region. Further, it was concluded that the parking pricing should not be

imposed in the areas served by light rail lines and in areas with increased densities with

promoted land subsidy policies.

The relationship between urban form, as shaped through transit – oriented urban design, and

transport demand was investigated by using regional travel demand forecasting model

technique (Dock and Swenson, 2003). The impact of urban transit oriented design on

suburban land use and trip patterns for Minneapolis – St. Paul Metropolitan region in the

USA was studied. The existing forecasting model was used with three enhancements for

addressing; shorter–distance trips and the effects of greater proximity to transit at the

transport analysis zone (TAZ) level, changes in travel within TOD which was caused by

changes in urban form, land use density and mix of uses, and interaction between a TOD with

adjacent development.

The traditional techniques of windowing and focusing zones were used to divide the regional

model into smaller pieces and to add more details of roads and transit routes. Transport

analysis zones were subdivided using a layered approach. Finely grained socioeconomic data

were given as basic inputs for trip generation. The trip distribution matrices were generated

using off line estimation techniques to facilitate the elasticity based adjustments. Intrazonal

and interzonal adjustment factors were applied and the model was validated without

considering the expansion of highways and transit networks. The model was applied to two

TOD scenarios (modifications to highways network and modification to transit network) and

the results were compared with conventional land use and TOD patterns. The main drawback

Literature review


of the study was the basic of off–line estimation techniques on extrapolated data relationships

from other studies.

Trip generation

Steiner (1998) studied the trip generation rates of six prototypical traditional shopping

districts in the Oakland–Berkeley subarea of the San Francisco Bay Area using the pedestrian

count, retail gross leasable area, automobile share and daily hours of operation. Peak hour

and average daily trip generation rates were calculated for weekday and weekend for

shopping centre as a whole and compared with the ITE trip generation rates. A customer

intercept survey was used along with the data on overall activity levels to estimate the

number of person trips and trip ends in the shopping area. It was observed that the trip

generation rates for shopping centres with food shops was almost double the average ITE, the

trip vehicle rates were also found to exceed the ITE average hourly rate.

Arrington and Sloop (2009) derived evidence on trip generation and parking from 17

residential TOD projects in four metropolitan areas located in the USA. The 24–hour TOD

vehicle trip rates were compared with ITE trip rates. Over a typical weekday period, the

surveyed housing projects averaged 44 percent fewer vehicle trips than that estimated by the

ITE report. The research also concluded that TOD households were twice as likely to not own

a car and own roughly half as many cars as comparable households not living in TODs.

Trip distribution

The trips can be classified mainly as interzonal trips and intrazonal trips. An interzonal trip

has different trip origin and trip destination zones, while an intrazonal trip has the same trip

origin and trip destination zone. In strategic modelling, very often, the intrazonal or short

trips are not considered or underestimated and the mode choice models informed by trip

distribution assign private mode to the majority of intrazonal trips (Cervero, 2006). Hence,

most of the studies found were focused on the study of interzonal trips.

The relationship between land use, destination choice and travel mode choice for intrazonal

trips was addressed using data from a 1994 Household Activity and Travel Diary Survey

(Greenwald, 2006). The travel mode choice and decision to internalize trips were measured

by using multinomial logit and binary logistic models. The explanatory variables influencing

trip making were grouped into four categories; namely standard demographic traits, costs of

trip making behaviour, schedule limitation and availability of alternative mode choices and



land use measures. A modelling package EMME/2 was used to calculate door to door

network distances.

Mode choice

The influence of presence of retail activities on the commuting choices of residents was

explored using data from 1985 American Housing Survey. The effects of land–use

environments on mode choice were modelled using binomial discrete choice models of the

form shown in Equation 2.10. The neighbourhood densities showed a stronger influence than

mixed land–uses on all commuting mode choices except walking and cycling. For non–

motorised commuting, the presence or absence of shops was a better predictor than

residential densities (Cervero, 1996).

𝑝𝑟𝑜𝑏 𝑌𝑖 = 𝑘 =1

1 + 𝑒𝑥𝑝 − (𝑈𝑖|𝑌𝑖 = 𝑘) Equation 2.10

where,

𝑈𝑖 = 𝑏0 + 𝑏1𝑋1𝑖 + 𝑏2𝑋2𝑖 +⋯… . .…+ 𝑏𝑞𝑋𝑞𝑖

𝑌𝑖 = commuting mode 𝑘 for employed resident 𝑖

𝑋 = independent utility variable, 1…… . . 𝑞

In an another study, Cervero and Radisch (1996) studied the effects of new urbanism design

principles on non – work and commuting travel by comparing modal split between two

distinctly different neighbourhoods in the San Francisco Bay Area; Rockridge and Lafayette.

A dummy variable was used to represent the trips from the two fundamentally different built

environments. Two separate travel surveys were carried out for work and non work trips on

randomly selected households. The data for up to three non work trips per day were

considered instead of a whole travel diary. A Binomial logit model was developed to predict

the probability of using a non car mode for non work trips as a function of the type of

neighbourhood of respondents as well as other control variables. Another Binomial logit

model was developed for work trips to predict the probability of commuting by a non single

occupant vehicle (non–SOV). The results of the logit model were used to simulate mode

choice based on neighbourhood origin and number of vehicles per household.

Rajamani et al. (2003) investigated the impact of urban form on nonwork trip mode choice by

using the travel data from 1995 Portland Metropolitan Activity Survey conducted by Portland

Metro, which collected travel information from members of a sample of households over a

Literature review


two weekday period. A multinomial logit mode choice model was developed consisting of

drive–alone, shared–ride, transit, walk and bike as choice set. A geographic information

system (GIS) based method was used to develop urban form measures at the neighbourhood

level of each household. The effect of household and individual socio–demographics, level–

of–service of travel modes, and urban form measures on nonwork travel mode choice was

determined using a discrete choice methodology.

Kim et al. (2007) analysed the factors that influence mode choice for trips between home and

light rail stations using on–board passenger survey data describing St. Louis MetroLink riders

in the United States. A multinomial logit model (MNL) was used to model discrete mode

choices of drive and park, pick–up or drop off, bus and walk (Equation 2.11). The

coefficients in the model were estimated using the method of maximum likelihood.

It was found that crime at the station had an impact of mode choice in particular for female

transit riders. Further private vehicle availability, bus availability and convenience was

associated with choice of drive and park and bus mode respectively. The airline (straight–

line) distance between transit riders’ home and the station is considered but the actual

walking distance might be more, which was a drawback.

𝑃𝑛𝑖 =𝑒𝛽𝑖𝑥𝑛𝑖

𝑒𝛽𝑖′ 𝑥𝑛𝑖 ′𝐼𝑖 ′=1

Equation 2.11

where,

𝛽𝑖 = estimable mode specific constants

𝑥𝑛𝑖 = observable characteristics of the modes, trip makers and the

surroundings

Trip assignment

Miller and Ibrahim (1998) correlated energy consumption through private automobile usage

in terms of total number of vehicle kilometres travelled (VKT) by using EMME/2 road

network assignment procedure with particular focus to 24–hour home–based–work (HBW)

trips in greater Toronto area, Canada. The VKT for a given origin–destination (O–D) pair

was calculated by multiplying the equilibrium O–D travel distances over the road network by

the O–D flows for a given trip purpose. Trip geocodes were used to measure the average

straight–line distances for intrazonal trips. A linear regression of the form shown in Equation

2.12 was used to calculate the variation of HBW VKT with density and distance from CBD.



𝑉𝐾𝑇 = 𝑎 + 𝑏 × 𝐷𝐶𝐵𝐷 Equation 2.12

where,

𝑉𝐾𝑇 = average daily HBW VKT per worker produced by the residential

zone

𝐷𝐶𝐵𝐷 = straight line distance (km) from the zone to the Toronto CBD

2.4.3.2 Agent based simulation

Zhang and Levinson (2004) developed an agent–based travel demand model considering

interactions from three types of agents in the transportation system; node, arc, traveller. It

was found that simple rules of agent behaviours solved the complicated problems such as trip

distribution and trip assignment.

Agent–based models have a unique feature of explicitly modelling the goal, knowledge,

searching behaviour, and learning ability of related agents. Although agent–based modelling

technique provides flexible travel forecasting network that facilitates the prediction of

macroscopic travel patterns from microscopic agent behaviours, the studies on individual

travel behaviours are mandatory to be done. In order to assess the individual behaviour at the

microscopic level the corresponding attributes at the macroscopic level need to be measured.

The effect of land use regulations on travel behaviour was examined using agent-based

modelling. A simulation model for a hypothetical urban area loosely based on the Chicago,

Illinois, metropolitan area was used to study the impact of six land use regulation scenarios

on transit use and urban form. The results from the simulations showed that although the land

use regulations that were designed to increase the density near the transit station or in and

near the urban core were able to achieve the intended land use patterns, they did not increase

the transit mode share for the region in a significant manner. More detailed examination of

the output revealed that as long as the rules for mode choice, the distribution of employment,

and the transit network remained unchanged, land use regulations that affected residential

locations produced limited effects on transit use (Lu et al., 2008).

2.4.3.3 GIS based modelling

Modelling Accessibility

Grengs (2004) demonstrated a method to measure changes in transit accessibility on one

neighbourhood from Buffalo and another from Rochester, New York by developing a gravity

model using Geographic Information Systems (GIS). This method separated the combined

Literature review


spatial effect of shifts in land use patterns and transit service. The results were obtained in the

form of a dimensionless neighbourhood accessibility index (NAI) and were compared at two

points in time, 1990 and 1997. NAI was based on proportional coverage, frequency factor,

travel time and employment within walking distance of transit stop (Equation 2.13). A

variable was used to represent the share of accessibility change attributable to transit. The

accessibility was found to vary due to different causes, in Buffalo case it improved because of

changes in transit service while in Rochester study, accessibility improved because of

changes in land use. The analysis was mainly for transit dependant poor people who live in

inner–city neighbourhoods.

𝑁𝐴𝐼𝑘 = 𝐸𝑗

𝑛

𝑗=1

𝑇𝑘𝑗 −𝛽

Equation 2.13

where,

𝑁𝐴𝐼𝑘 = index indicating accessibility between census tract 𝑘 and the set of 𝑗

destination transit stops

𝐸𝑗 = number of jobs at transit stop 𝑗

𝑇𝑘𝑗 = travel time by transit between census tract 𝑘 and transit stop 𝑗

𝛽 = constant to represent distance decay

Modelling walkability

Understanding the opportunities for pedestrian movement is a key component in

understanding and evaluating TOD projects. In order to address this TOD – pedestrian link,

Schlossberg and Brown (2004) analysed twelve geographic information system (GIS) based

walkability measures to visualise and quantify the pedestrian environments at each site,

across eleven TOD sites in Portland, Oregon. The street networks were classified into

pedestrian – friendly and pedestrian – hostile categories. This refined street data were used to

identify the quantity of different street types, densities of good intersections and dead ends,

and the catchment areas to which pedestrians are likely able to reach. The comparative

analysis was extended to two specific sites (one positive and one negative example of

walkability) to demonstrate the usability of analysis technique. Access, connectivity, and

pedestrian choice (which can be derived using various elements of street network) were noted

as key elements in understanding the pedestrian environments.



The neighbourhood walkability was quantified by using destinations, distance, density and

route as core constructs (Lee and Moudon, 2006). The four step analytical process comprised

of initial variable screening, variable grouping, variable prioritization and statistical

modelling. The environmental variables were measured using GIS. The likelihood of walking

moderately and sufficiently for health purpose relative to not walking was measured by two

multinomial logit models, using airline and network measures.

2.4.3.4 Activity based travel analysis

The activity based approach reveals the travel patterns in the context of a structure of

activities, of the individual or household, with a framework emphasising the importance of

time and space constraints (Kitamura, 1988).

Zhang (2005) presented an activity–based time–use analysis of the relationship between

urban form and nonwork travel, using data from 1991 Boston Activity–Travel Survey. The

role of spatial accessibility was tested as a composite measure of urban form in explaining

individuals’ nonwork activity participation, travel time and travel frequencies. All activities

taking place inside home were grouped into one category, outside home work related

activities were differentiated from nonwork activities and nonwork activities were divided

into six types as school, shopping, social, personal, pick up or drop off and other. Each of

these six types of activities was given two components; activity and travel associated with it.

A gravity model – based measure of spatial accessibility was adopted to characterise the

opportunities and constraints to individual’s activity participation in the urban form. The

analysis was performed in two parts; part one estimated a set of multinomial logit models of

activity participation in which activity duration and travel times were combined for each type

of nonwork activity and part two separated activity duration from travel and analysed travel

duration and activity – travel frequencies. The results showed varying effects of modifying

spatial accessibility on nonwork activity participation among different activity categories.

Activity based travel analysis technique is not further explored because the data collection

involves “in–depth” interviews, which are uneconomical and limits the sample size. The

interviews are mainly option–specific thus lacks spontaneity. Further, these models have

limited applications to a wide range of planning and policy problems (Kitamura, 1988).

2.4.3.5 Tour based modelling

The relative association between travel time, costs, and land use patterns where people live

and work and its impact on modal choice and trip chaining patterns was studied for Central

Literature review


Puget Sound (Seattle, Washington) region (Frank et al., 2008). A tour–based modelling

framework and highly detailed land use and travel data was used to understand how

community design influences travel choices. The findings suggested that Puget Sound area

residents made travel choice decisions based on several factors, the most important being

time. A number of land use variables were also found to be statistically significant for all tour

types modelled. The degree to which participants chain trips together into tours was also

highly correlated to the land use characteristics where residents live and work.

Lee et al. (2009) examined the effects of proximal residential density, roadway accessibility,

land use mix, and transit accessibility on individual tour–based travel behaviour across three

trip categories in detail. Cross–sectional activities were obtained from household activity

travel survey data from the Atlanta Metropolitan Region. Time durations allocated to

weekdays and weekends were also compared. The censoring and endogeneity between

activity categories and within individuals were captured using multiple equations Tobit

models. The analysis and modelling revealed that land–use characteristics such as net

residential density and the number of commercial parcels within a kilometre of a residence

were associated with differences in weekday and weekend time–use allocations. Household

type and structure were significant predictors across the three activity categories, but not for

overall travel times. The tour characteristics such as time–of–day and primary travel mode of

the tours also affected traveller’s out–of–home activity–tour time–use patterns.

2.4.4 Summary of travel data analysis

The studies reviewed on TOD evaluation can be categorised into two types; statistical

analysis and travel demand modelling. A summary table is given for studies on comparative

and statistical analysis and studies on travel demand analysis in Table 2.5 and Table 2.6

respectively.

The key observations from studies on travel demand analysis are noted below.

The statistical analysis studies illustrated the effect of car ownership, land use, density

and neighbourhood design impacts on travel characteristics of people. Regression

analysis and ANOVA techniques were found to be the most commonly used

techniques to analyse the effect of different variables on travel.

Analysis for mode shares was found to be the topic of interest for many researchers

followed by the calculation of trip generation rates. The issue of parking requirements

was also addressed by comparing the situation for different TODs.



Table 2.5 Summary of studies on comparative and statistical analysis

Parameter Author/s Technique / method

Parking requirements

Higgins (1993) Setting minimum and maximum parking

requirements

Steiner (1998)

Deakin et al. (2004) Study of land use, parking supply and use, mode

choices, and housing and jobs development

Car ownership

Hess and Ong (2002) Ordered logit regression

Giuliano and Dargay

(2006)

Comparative analysis

Structured model for total distance travelled

Transit use

Messenger and Ewing

(1996)

Stepwise regression analysis

Full – information maximum likelihood

Lund (2006) Binary logistic regression analysis

Lin and Shin (2008) Regression models calibrated using Ordinary

Least squares method

Travel mode

McMillan (2007) Binomial logit regression probability models

Cao et al. (2009) Seemingly unrelated regression equations

(SURE) model

Walking Kerr et al. (2007) Stratified logistic models

Travel behaviour

Khattak and Rodriguez

(2005)

Determination of differences in matched pair

neighbourhoods

Two–stage regression models

Cao et al. (2007) Structural modelling approach

Modelling vehicle

availability

Allen and Perincherry,

(1996)

A two–stage approach; look up table and

incremental logit model

Travel demand Cervero (1991) Stepwise regression analyses


Cervero and Gorham

(1995)

Comparison of commuting characteristics of

transit–oriented and auto–oriented suburban

neighbourhoods

Regression models

Household travel Sun et al. (1998) The Analysis of variance (ANOVA), multilinear

regression analysis, sensitivity analysis

Internal Capture Ewing et al. (2001) Regression analysis

Travel impacts McCormack et al.

(2001)

Comparison

ANOVA technique to study variation

Travel patterns Coevering and

Schwanen (2006)

Ordinary Least Square Regression modelling

Individual travel

behaviour (walking

trips)

Zegras (2004) Maximisation technique

Ordered probit estimation

Mixed land–use and

density

Tong and Wong (1997) Comparative analysis

Allowable

development densities

Lin and Gau (2006) Maximisation and optimisation technique

sensitivity analysis

Accessibility Waddell and Nourzad

(2002)

Energy–related

emissions

Sadownik and Jaccard

(2001)

Spreadsheet model to evaluate aggregate energy–

related emissions

Scenario testing Cervero (2006) Post–processing and direct (off–line) modelling

Relocation impacts Cervero and Day (2008) Statistical modelling

Literature review


Table 2.6 Summary of studies on travel demand modelling

Modelling approach Author/s Method / Approach

Four step modelling

Rossi et al. (1993); Winder

(1994); Miller et al. (1999);

Waddell (2002); Rodier et

al. (2002); Dock and

Swenson (2003)

Travel demand models for scenario testing

Four step modelling: Trip

generation

Steiner (1998); Arrington

and Sloop (2009)

ITE comparison


distribution (Internalisation

of trips)

Greenwald (2006) Multinomial logit and binary logistic

models

Four step modelling: Mode

choice

Cervero (1996) Binomial discrete choice models

Cervero and Radisch (1996) Binomial logit model

Rajamani et al. (2003) A geographic information system (GIS)

based and discrete choice method

Kim et al. (2007) A multinomial logit model (MNL),

Maximum likelihood method


assignment

Miller and Ibrahim (1998) EMME/2 and linear regression equation

Agent based simulation Lu et al. (2008) A simulation model for a hypothetical

urban area

GIS based modelling

Grengs (2004) Gravity model to measure changes in

transit accessibility

Schlossberg and Brown

(2004)

GIS based walkability measures to

quantify the pedestrian environments

Lee and Moudon (2006) Quantification of neighbourhood

walkability, multinomial logit models

Activity based travel

analysis

Zhang (2005) Analysis of relationship between urban

form and nonwork travel

Kitamura (1988) A gravity model – based measure of

spatial accessibility

Tour based modelling

Frank et al. (2008) Investigation of impact of relative

associations between travel time, costs,

and land use patterns where people live

and work on modal choice and trip

chaining patterns

Lee et al. (2009) Effects of proximal residential density,

roadway accessibility, land use mix, and

transit accessibility on individual tour–

based travel behaviour

The two approaches, agent based simulation and activity based modelling, were

alternatives to four step modelling but they are not considered in detail because of the

limitations associated with them.

GIS modelling was performed to study the transit accessibility and walkability

indicators of TOD. This method is suitable to assess the accessibility and walkability

because it considers the geography of the area.



The four step modelling procedure was found by researchers to be more suitable for

analysing the travel demand at TODs. But it was found that this technique requires

modifications in order to represent the trips made for various trip purposes.

The travel demand models developed for TOD evaluation consider land use variables

in addition to the conventional variables. The models were mostly developed to test

the hypothetical development alternatives. These models also included walk mode

and short trips.

Very little research has tried to address the details at any individual level in the four

step modelling process. Intrazonal trips were found to be important for modelling trip

distribution while use of discrete mode choice models (binomial and multinomial

logit models) found suitable for mode choice analysis. A software package EMME/2

was found to be the most commonly used software for four step modelling.

2.5 Summary of literature review

2.5.1 Strengths of the literature

Strong evidence supporting the benefits and in turn success of TODs was found from

a wide range of studies that were mainly conducted in the USA, whose cities have

similar urban development characteristics to Australian cities.

The papers dealing with the statistical analysis of various TOD variables shows the

contribution of these variables towards assessing the travel behaviour of residents at

TODs.

These analyses confirm the need of consideration of these aspects into travel demand

modelling in addition to the traditional demographic and socioeconomic variables.

The studies showing the comparison of the travel characteristics of people living in

TOD and outside TOD confirms the difference in travel behaviour of people living in

these two forms of communities.

Strong evidence was found claiming the increased mode share of walking trips and

reduced VKT for non work travel in TODs for residents. The car ownership level was

also found to decrease with an increase in density. These findings are in support of

TODs.

The four step modelling approach is found as a viable approach for studying travel

demand at various levels.

Literature review


2.5.2 Gaps in the literature

The theoretical background explains the relationship between travel and urban design. The

past studies provide an insight in to the relationship between TOD and travel behaviour of

people. Following are some of the gaps found while reviewing the literature.

The review of literature demonstrates many studies explaining the effects of various

indicators of TOD on travel characteristics of residents. However, more detailed study

of various trip purposes and trip characteristics of all users using these types of

developments should be carried out.

Another shortcoming is that the data used for developing the models is not directly

applicable or not solely collected for model development purposes, making the quality

of the results limited. In most of the cases, the developments emerged as TODs were

considered; whereas the travel data from a planned TOD needs to be considered for

improving the quality of the results.

Very few studies considered the special design characteristics of TODs when

preparing travel demand models.

The specifications of traffic generation rates for mixed land use developments and for

walking and bicycle trips were not given separately. This demands more research in

the area of traffic generation for TODs. This will also help in studying the overall

traffic impact of the TOD.

It was observed that TODs need to be assessed for the amount of self containment

with a proportion of internal trips with special consideration to trip length and mode

split. The number of trips staying within the TOD needs to be determined and

compared with respect to the number of trips going outside of the TOD for various

trip purposes.

Many studies considered car ownership and mode choice of people but failed to

analyse transit use with respect to the quality of service available, demanding more

research in vehicle availability and transit availability.

The various TODs have mainly been concentrated on the aspect of residential land

uses and some have dealt with the commercial, retail, official land uses, however

studies investigating the educational and recreational aspects of TODs are missing.

Most of the studies considered rail based TODs while not much evidence was found

for bus or Bus Rapid Transit (BRT) based TODs.



None of the studies reviewed have focused on ways of quantifying the transport

related impacts of TODs from an Australian perspective to aid in planning and design

of TODs in this country.

2.6 Recommendations flowing from literature review

TODs are special developments which are designed to increase quality of life and reduce the

adverse environmental impacts. As the outcomes of TOD differ depending upon location,

size, type of land uses, density, accessibility, quality of public transport system and

pedestrian environment available; the travel characteristics of the people are also diverse

depending upon the type of community and its travel habits. Hence, each TOD needs to be

evaluated individually by giving consideration to the various design aspects of TOD and

travel habits of people living there. In order to do this, the literature review has identified the

following areas of research which warrant further investigations:

There is a need to develop a generalised (comprehensive) methodology for assessing

the travel demand of TODs. This study aims to develop this and also plans to test it

using a case study TOD.

For assessing travel demand the number of trips generated should be estimated

accurately. Previous studies (Steiner, 1998; Lin and Gau, 2006) have used ITE trip

generation rates for this purpose, but these trip rates are not necessarily directly

applicable as these assume homogeneous land uses. This research investigates the trip

generation rates for a TOD for various modes of transport.

The traditional travel demand models neglect the intrazonal trips. But in the case of

TOD, these require consideration as TOD has clustering of activities in a small place.

The trip lengths and mode choice for short trips need to be considered. The trip

distribution and mode choice steps should take into account these aspects.

The data for travel demand modelling should be collected for various categories of

users performing various activities at a TOD.

Comparison of trip characteristics for various trip purposes needs to be made for

checking the performance of a TOD as a whole system rather than considering

residents’ data only to confirm the claims of TOD.

The travel demand model should be developed considering the TOD design, location,

size, type of land uses, densities, accessibility, quality of public transport system and

pedestrian environment available. A multivariate analysis should be performed to

Literature review


investigate the relative importance of these variables and their effect on travel

demand.

The traditional four step modelling approach can not be applied directly for TOD

evaluation. Specifically, the procedure needs some advances or improvements in

order to make it suitable for travel demand analysis of TODs.

The trips should be modelled considering private car availability and transit

availability (quality of service), as the decision for mode choice may be driven by

these two factors.

The need for analysing TOD in a regional context should be noted. In order to assess

TOD at a higher level, detailed analysis of TOD at the zonal level should be

undertaken.

Activity based and tour based modelling techniques are emerging fields for travel

demand modelling which should be explored further for assessing their suitability.

Mixed results were obtained for the benefits of TODs for developing countries. Also,

stronger evidence was observed for studies in USA than in UK.

Travel at Australian TODs should be studied in order to gain understanding and

suitability of TODs from Australian perspective.

2.7 Chapter close

This chapter presented a review of literature, the strengths of the literature and the gaps in the

knowledge. These gaps provide direction for this research. The next chapter presents a

methodology for evaluating TODs, which will help to reduce the gaps in knowledge related

to TOD evaluation. The studies presented in the data requirements section offer base

information for the data collection process presented later in Chapter 5.




Chapter 3

Methodology for evaluating transport impacts of transit oriented

developments

3.1 Introduction

To achieve the objectives of this research and assess the performance of Transit Oriented

Developments (TODs) from transport point of view, this chapter proposes an appropriate new

methodology for evaluating TODs. Importantly, this methodology discloses the key

characteristics of a TOD within its first stage. It is these characteristics that are used to state a

clear and concise definition of a TOD to be used in this thesis from this chapter onwards.

The proposed methodology will be further investigated using an appropriate case study TOD.

The observations made while undertaking TOD investigations will be useful for

implementing this methodology more generally for assessing various TODs. The following

section presents the measures used in evaluating transport impacts followed by a detailed

description of each step involved in the proposed methodology. Later a brief summary,

applications and chapter close are presented.

3.2 Measures for transport evaluation of TODs

The transport evaluation of TOD consists of determination of traffic generation and the

details of travel at a TOD. The various measures of transport performance of TODs have

been considered carefully based on literature review and are shown in Figure 3.1.

The performance measures include traffic generation, household characteristics and travel

characteristics. The performance of these measures needs to be determined for various groups

of TOD users to get a complete picture of travel at a TOD. Typically, travel characteristics of

all user groups should be assessed and the household characteristics and trip characteristics

for TOD residents should be evaluated. The new methodology is proposed in this research to

evaluate the transport impacts through these measures, which is presented in the following

section.



Figure 3.1 Measures for transport evaluation of TODs

3.3 Proposed new research methodology

This research proposes a new methodology to evaluate TODs from a transport perspective.

The conventional procedure does not take into account the effects of atypical development

characteristics of TODs for example, presence of mixed land uses and characteristics of

various user groups using these land uses etc. To consider these effects, this methodology

incorporates some additional steps to the existing procedure. The methodology for TOD

evaluation involves a stepwise approach. The first step of TOD evaluation is assessing a

development before considering it as a TOD and the last step is determination of

transportation impacts through the measures specified in previous section. Figure 3.2

represents the flow diagram explaining the major steps in TOD evaluation. The five step

approach to TOD evaluation include; pre–TOD assessment, traffic and travel data collection,

determination of traffic and travel impacts and obtaining outcomes. Following sections

explain the process involved in each step with detailed flow diagrams.

Transport impacts

assessment measures

Travel

characteristics

Household

characteristics

Household size

Number of

vehicles/household

Mode share

Trip lengths

Trip characteristics

Number of

bicycles/household

Number of

bedrooms/household

Traffic

generation

Methodology for evaluating transport impacts of TODs


Figure 3.2 Proposed new methodology for evaluating transport impacts of TODs

3.3.1 Step I: Pre–TOD assessment

The first step of pre–TOD assessment is selection of a development and confirming its

function as a TOD. The steps involved in pre–TOD assessment are shown in Figure 3.3. The

shaded box in the flow diagram indicates new or additional step for pre–TOD assessment.

Often a mixed use development or any developments around a transit station can be

presumed to have certain TOD like transport characteristics without substantiation. Although

having some similar characteristics, these developments may not really operate as TODs. To

ensure the function of development as a true TOD, the land use mix, parking provision for

cars, and infrastructure for sustainable modes of transport needs to be appropriate.

The land use mix at a TOD is important for integrating various activity nodes by providing

greater accessibility. These diverse land uses produce and attract various trip types.

Noticeably, at a TOD presence of trip attracting land uses is in greater proportion than the trip

producing land uses. Hence, TODs normally have higher trip attraction rate than trip

production rate for all modes of transport. To ensure the use of sustainable modes of transport

for these trips, the provision for walking, cycling and public transport is essential. The use of

public transport, walking, and cycling is widely regarded in the transport professional

community as being more sustainable in comparison to the use of other motorised modes of

transport, specifically the car.

Step II: Traffic and travel data

Step IV: Travel impacts

Step I: Pre-TOD assessment

Step III: Traffic impacts

Step V: Outcomes



Figure 3.3 Step 1: Pre–TOD assessment

Awareness of parking infrastructure is also vital because the provision of large numbers of

parking spaces may attract, or produce, more car trips rendering the development less

pedestrian friendly, which defeats the purpose of a TOD. On the contrary, the provision of

fewer parking spaces may harm the development by distracting the attention of commercial

or retail development users. The optimum supply of parking infrastructure supports the

economic activities at a TOD and confines the avoidable use of car for transport, supporting

the more sustainable transport modes. The thresholds for parking supply can be obtained

from the standard guidelines followed by local government. Many local governments are now

stipulating thresholds for TODs or TOD like developments.

To evaluate a study development for TOD specification, data for three characteristics, land

use mix, parking provision for cars, and infrastructure for sustainable modes of transport,

should be gathered and evaluated to ensure correct anticipated provisions of a TOD rather

than assuming so and undertaking further steps of evaluation. If the selected development

does not exhibit appropriate TOD characteristics then another development form should be

selected and assessed for its suitability. Table 3.1 lists the measures used for assessing the

Yes

Selection of the development

Assess suitability of the development

Obtain land use mix and transport details

Is

suitable?

No

Proceed to step II

Represents new/additional steps





Step V: Outcomes



suitability of a TOD, and in doing so, importantly disclosing the key characteristics of a

TOD.

Table 3.1 Measures for assessing suitability of a TOD

TOD attribute Measure of assessment

Land use mix Presence of diverse land uses, self containment

Road infrastructure Layout of road network, its quality and connectivity to

major transport corridors

Infrastructure for walking Provision of sidewalks, quality of sidewalks and secured

pedestrian crossings

Infrastructure for cycling Dedicated bicycle tracks or bicycle lanes, its connectivity to

bicycle network and secured parking for bicycles

Location of nearest public

transit node

Major public transport node accessible within easily

walkable distance (400m)

Quality of public transport Level of service for transit availability and comfort and

convenience

Parking supply Optimum parking supply

Based on the measures listed in Table 3.1, the following definition of a TOD is used from this

point onwards in this thesis:

A TOD is defined as a higher density mixed land use development located within easy

walkable distance of a major public transport node having excellent facilities for walking and

cycling without compromising infrastructure for private motorised modes of transport

particularly car. Basically, a TOD provides competent mobility choices for sustainable

modes of transport such as walk, cycle and public transport to promote its use and

discourage use of private motorised modes of transport.

3.3.2 Step II: Traffic and travel data collection

After selecting the study TOD, next step is gathering of traffic and travel data. This data is

essential to overview transport at a TOD, to study the trip rates for various modes of transport

and to obtain travel characteristics of various user groups. The steps for traffic and travel data

collection are given in Figure 3.4. The shaded boxes in the flow diagram indicate new or

additional steps for data collection.

First the existing data availability needs to be checked before proceeding to data collection. If

the data is available then the data should be checked for its suitability. Classified cordon

counts for all access points of the development should be available and the travel data for all

user groups at TOD should be available to proceed to further steps. If the data is not available

then it should be collected by a two–step process; cordon counts and travel surveys.



Figure 3.4 Step II: Traffic and travel data

Before starting the field studies, the observation of the traffic and user groups at a TOD

should be made to gain an understanding of the development. The cordon counts for 24 hour

or for the specified period of the day should be conducted at all the access locations of the

development for both directions (inbound and outbound). The counts should be conducted for

all modes of transport as study of traffic generation for sustainable modes of transport as well

as lesser sustainable modes of transport is important. This data will provide input to Step III.

To collect TOD users’ travel data, the travel surveys should be conducted for all TOD users,

including residents as well as visitors. The user groups can be obtained from the land use

used by a user. For example, the employee user group utilising commercial land use/s. For


Travel surveys for all TOD

user groups

Identification of TOD user groups

Input for step IV

Travel data

available

No

Yes

Input for step III

Multimodal cordon counts for TOD

and commercial / retail land use

Observation of traffic and

determination of access points

Cordon data

available

Yes

No

Cordon data TOD users’ travel data





Step V: Outcomes



each user group, a suitable survey instrument should be selected as different users tend to

exhibit different characteristics that affect the response rate. The survey should be

implemented using a specifically designed questionnaire form. The selection of the

methodology and questions for the travel survey is based on the composition of the users,

such as age and occupation and the availability of the resources. Generally, TOD residents’

travel diaries for a typical working day and TOD visitors’ travel data for specific trip

purposes should be obtained along with the demographic characteristics. Before conducting

any full scale surveys, the survey process should be tested on a sample population to gain an

insight into respondents’ opinion about survey design. For a TOD, preferably the travel

surveys should be distributed to the whole population rather than undertaking sampling. The

travel data obtained through travel surveys provide input to Step IV.

3.3.3 Step III: Determination of traffic impacts

After obtaining traffic and travel data, the next step is Step III: Traffic impacts determination.

The steps involved in determining the traffic impacts are given in Figure 3.5. The shaded box

in the flow diagram indicates new or additional step for traffic impact determination. The first

task of analysing cordon counts is to determine the total trips generated at the TOD by

excluding the through traffic movements. Later the cordon data should be analysed to obtain

the trip rates by mode of transport and their characteristics. Further, the traffic generation

rates for complete development obtained from cordon counts should be compared with the

totals derived using the standard trip rates specified, for example, American context (ITE,

2008) and Australian context (RTA, 2002). The traffic generation rates should be also

compared with similar sized non – TOD development to determine any actual variation in

traffic generation. The results of these comparisons provide input for determining the

outcomes in Step V.



Figure 3.5 Step III: Traffic impacts

3.3.4 Step IV: Determination of travel impacts

The travel impacts of a TOD are determined from analysis of travel data gathered from travel

surveys. The responses from travel surveys need to be compiled before proceeding to

analysis. The process of determination of travel impacts is shown in Figure 3.6. The shaded

box in the diagram indicates new or additional step for determination of travel impacts. Once

the compiled data set is obtained, the preliminary data analysis should be performed to obtain

various characteristics of the TOD users. These characteristics include personal, household,

and travel characteristics, and perceptions, issues, suggestions and comments about the

development. The personal characteristics include age group, occupation, employment status,

gender, driving licence availability, etc and the household characteristics contain household

size, vehicle ownership, bicycle ownership, type of dwelling and number of bedrooms in the

household. The travel characteristics can be explained with mode shares, trip lengths, parking

details such as parking location and parking fee and public transport details such as travel

time, number of routes used and access and egress walking times or distances. In addition the

perception and issues about public transport and about the development can be obtained.

Trip generation for all modes of transport

Traffic characteristics

Comparison with standard trip rates

and/or with non – TOD development

Input for Step V

Step II: Cordon data






Step V: Outcomes



These results help to gain a better understanding about the travel at the TOD under

consideration.

Figure 3.6 Step IV: Travel impacts

Further, the travel and demographic characteristics should be compared with that of non –

TOD or conventional developments having some similar characteristics, such as size of the

development, distance from CBD or household characteristics. A regional comparison can

also be conducted by comparing travel data of TOD users with respective travel data for

entire region. The data for non – TOD or conventional developments can be obtained from

various data sources such as household travel surveys, census database etc. The preliminary

analysis of travel data provides a base for the travel demand analysis. The travel mode is a






Step V: Outcomes

Travel demand

analysis

Comparison with non –

TOD developments

Demographic

characteristics

Travel

characteristics

Perceptions

and issues

Characteristics of all user groups at TOD

Input to Step V

Step II: TOD users’ travel data

Data compilation and analysis



prominent aspect of each trip hence essentially this aspect needs to be investigated. In

addition, trip distribution or internalisation can be studied depending on data availability. The

outcomes from step IV provide the travel impacts which are used to draw outcomes in Step

V.

3.3.5 Step V: Outcomes

The outcomes obtained from various steps of analysis are displayed in Figure 3.7. The shaded

boxes in the flow diagram indicate new or additional outcomes obtained from the proposed

methodology. The first step of TOD evaluation provides insight into selection process of a

development for TOD evaluation. The second step of data collection provides findings useful

for collecting traffic and travel data at TODs; these can be useful in planning and designing

of future data collection processes at TODs. The comparison of total traffic generation

indicates the performance of TOD with respect to the traffic at the conventional

developments (Step III) and the results from the comparative analysis reveal how the TOD

under consideration is performing in comparison to other conventional developments and at a

regional level in terms of transport and demographics (Step IV: travel impacts). The findings

for data collection, traffic and travel impacts are the outcomes obtained through TOD

evaluation. The positive differences in traffic generation as compared to standard trip rates

and in travel characteristics as compared to non – TOD developments for a TOD indicate

sustainable transport and encourage further investigation, and on the contrary the negative

differences calls for more stringent evaluations and reviewing of sustainable transport claims

of TODs made by transport planners. Further, the results from trip rates and travel demand

analysis provide guidance for planning future TODs.



Figure 3.7 Step V: Outcomes

3.4 Summary

In evaluating a TOD for transport impacts, the starting point was selection of a case study

development, by observing the mix of land uses and sustainable transport infrastructure and

assessing its suitability, followed by collection of traffic and travel data for all the user groups

of the respective development from the classified multimodal cordon counts and travel

surveys. The characteristics obtained from the analysis of data helped to explain the travel

behaviour at a TOD. This also points towards the performance of a development which helps

in providing the evidence for TOD’s travel behaviour. The additional or new steps proposed

in this methodology aid in addressing the following gaps in the literature noted in Chapter 2.

A detailed study of various trip purposes and trip characteristics of all users was

missing. The additional steps suggested in this method provide a complete picture of

travel at TODs by studying all user groups at TOD under consideration.






Step V: Outcomes

Step II

Guidelines for

conducting travel

surveys

Traffic

impacts

Step III

Trip rates for

various modes

Step IV


for various user groups

Travel

impacts

Travel mode

assessment

Trip distribution/

internalisation



Some data quality and reliability issues were observed for previous studies as mostly

the data was not collected solely for TOD evaluation, the pre–TOD assessment step of

this methodology ensures selection of a development with all essential characteristics

of a TOD and the data collection step ensures the quality and reliability of the results.

This methodology assesses the overall traffic generation and traffic impacts of TODs

by conducting traffic counts for various modes of transport and also provides various

trip rates useful for planning TODs.

Quality of transit service is an important aspect of TOD hence required more

investigation while travel demand analysis. This methodology considers quality of

transit service while selection TOD as well as while analysing the travel demand.

Previously most of the studies considered only residential land use at a TOD for

evaluation, this methodology proposes study of transport aspects of various land uses

located within a TOD including residential land uses, commercial, retail, official land

uses and educational land uses giving complete picture if travel at TODs.

3.5 Application

This chapter provides an overview of the methodology for evaluation of transport impacts of

a TOD. Although this methodology is broadly applicable for assessing various TODs, some

case study specific modifications may be required.

3.6 Chapter close

The procedures followed in this research forms a basis for all chapters in this thesis. Each

step or combination of two or three steps suggested in this methodology forms a chapter in

the thesis. The next chapter outlines the details of Step I: Pre–TOD assessment which

includes gathering of case study development information and evaluating whether it can be

considered as a TOD for further analysis.


Chapter 4

Selection of case study transit oriented development

4.1 Introduction to case study selection

A case study Transit Oriented Development (TOD) which has good transit service is essential

to conduct research on inhabitants’ travel behaviour and achieve the objectives of this

research. To ensure selection of a suitable case study TOD, the characteristics of the

development and public transit availability are used as indicators. The purpose of this chapter

is to introduce the case study TOD and demonstrate the selection process followed in this

research to test its suitability.

The first section gives an overview of the location and various land uses at the case study

TOD site. The following section continues with the description of the transport facilities. The

criterion used for assessing the suitability of the case study TOD site and the analysis for the

same are listed in the next section. Finally the interpretation of the results and a brief

summary of the case study TOD are provided.

A candidate case study TOD, the Kelvin Grove Urban Village (KGUV) in Brisbane,

Australia, was selected for investigation. KGUV was tested for its suitability as a case study

TOD by evaluating quality of service (QoS) indicators for public transport service in order to

better appreciate whether it has the appropriate transit availability to support its function as a

true TOD.

4.2 Description of KGUV

Kelvin Grove Urban Village (KGUV), designed as a sustainable and mixed use development,

is situated in the inner suburb of Kelvin Grove, approximately 2km northwest of Brisbane’s

Central Business District (CBD). The greater Brisbane metropolitan area had a population of

approximately 1.9 million at the time of writing. KGUV has been developed as a joint

venture between the Queensland Department of Housing and Queensland University of

Technology (QUT) based on the ecological sustainable development (ESD) principles.

KGUV spans over 16.57 Ha of land area and to the best of the authors’ knowledge is the first

of its kind of development in Australia. It is surrounded by inner city suburbs, Spring Hill,

Herston, Red Hill, Newmarket and Wilston. The mixed use development consists of

educational, residential, commercial, recreational, retail and office land uses. Young, single



students and workers are expected to comprise the majority of the population. KGUV has a

university campus and a state government high school located on its northern boundary. This

site is expected to be fully developed by late 2009.

KGUV is an educational based mixed use development comprised of four distinct precincts;

the village centre, QUT Kelvin Grove campus extension, QUT’s Institute of Health and

Biomedical Innovation (IHBI), and QUT’s Creative Industries Precinct. The building height

varies between 4 and 7 stories. The details of mixed uses are as shown in Table 4.1.

Table 4.1 Mixed land uses at KGUV

Land use Size Description

Residential 35,668sqm Includes affordable accommodation, managed

accommodation for seniors, student accommodation,

investor apartments, townhouses (900 residential units

including 200 affordable housing units)

Education 14,770sqm QUT campus extension, Queensland Academy for Creative

Industries (QACI)

Retail Not Available Village centre; street level shops, catering for extended hour

demand

Commercial 3,878sqm Creative industries precinct, Institute of Health and

Biomedical Innovation (IHBI), health services and standard

commercial office facilities and tenancies with opportunity

to implement innovative structures

Lifestyle 6,897sqm La Boite theatre, Victoria Park golf course, network of

parks

Mixed 11,995sqm

26,014sqm

Mixed use area for residential, commercial and retail places

Mixed use area for education and commercial places

4.3 Transport facilities at KGUV

The main aspect of a TOD is the transport facilities, both public as well as private. KGUV is

well connected to arterial roads and has an internal street network forming a grid pattern, with

parks and open spaces. KGUV has sidewalks on both sides of all streets and through the

parks, and cycle lanes on road to encourage and support walking and cycling. BCC (2000)

stipulates car parking rates of minimum 3 spaces per 50m2 GFA at ground floor level and 1

space per 30m2 GFA above ground level for centre activities, 1 space for two staff and 1

space per 10 students at tertiary institutions and 1 space per 30m2 GFA for offices. In KGUV,

the number of car spaces is restricted to 1 space per 30m2 GFA for all non residential

development. These restricted parking facilities are provided at a TOD to discourage drivers

from driving their cars and to promote the use of sustainable modes of transport including

walking, cycling and use of public transport.

Selection of case study TOD


Figure 4.1 shows an overview of KGUV, the public transport corridors and its proximate

transit stops (development has occurred on the site since the photograph was taken). In this

figure, the area bounded by red polygon represents KGUV, the dotted yellow lines show the

public transport corridors and the blue symbols indicate location of bus stops or busway

(BRT) stations catering for KGUV users. The major public transport corridors for KGUV run

along east and west flank with an intercampus shuttle bus service running through its centre.

The three transit corridors catering for residents and visitors at KGUV are:

The Inner Northern Busway located on the eastern flank of the QUT campus.

Kelvin Grove Road located on the western flank of KGUV. Many of its services leave

the Busway at Normanby immediately to the south.

The QUT intercampus shuttle service operating along Musk Avenue within QUT KG

campus/KGUV and runs express to QUT Gardens Point (City) campus 3.5km to the

south.

Figure 4.1 Aerial overview of Kelvin Grove Urban Village (KGUV)

KGUV is close to two major busway (BRT) stations (ST1 and ST8), four express bus stops

(ST3, ST4, ST5, and ST7) and two local bus stops (ST2 and ST6); the locations of bus stops



can be seen in Figure 4.1. KGUV is served by 16 bus services including nine express and

very high frequency services. Table 4.2 identifies the stops and describes the bus routes

observing them. (The details of public transport services were obtained from the TransLink

Transit Authority’s public transport information website, www.translink.com.au.)

Table 4.2 Buses observing transit stops in KGUV

Stop Location Routes Destination

ST1 QUT KG Busway

Station on Inner

Northern Busway

330e, 333b, 340e, 376e,

392e, 393l, 680l

Inbound to CBD and

outbound to northern suburbs

ST2 Victoria Park Road 361a, 364a

ST3 Start of Musk Avenue 391e QUT Gardens point

(intercampus shuttle service)

ST4 Middle of Musk

Avenue (Near IHBI)

391e, QUT Gardens point

(intercampus shuttle service)

ST5 Kelvin Grove Road at

Blamey street

344cp, 345b, 351r, 357e,

359e, 390a

Inbound to CBD


School street

390a, 364a


Prospect terrace

344cp, 345b, 351r, 357e,

359e, 390a

Outbound to northern suburbs

ST8 Normanby Busway

Station on Inner

Northern Busway

330e, 333b, 340e, 344cp,

345b, 351r, 357e, 359e,

376e, 390a, 393l

Inbound to CBD and

outbound to northern suburbs

Note: The suffix to the bus numbers indicates type of bus service. Where e = express service, b = buz

service, l = link service, a = all stop service, cp = city precinct service, and r = rocket service

There are six main types of services operating on these corridors;

Buz (routes 333, 345), which are very high frequency buses with limited stops (10

minutes – 15 minutes, early morning till late night).

Express buses (routes 330, 340, 357, 359, 376), which are high frequency buses with

limited stops (10 minutes – 15 minutes).

City precincts (route 344), which stop at limited stops and run only at peak time.

Rocket buses (route 351), which stop only at few specific stops and run only at peak

times serving commuter markets.

All-stops buses (routes 361, 364), which stop at all bus stops on the specified route

across the day.

Links (route 393, 680), which are linked to other modes of transport such as train and

/ or ferry at transfer stations (early morning till evening).

http://www.translink.com.au/



Special purpose QUT intercampus shuttle service (routes 391, 392), which are only

available to QUT students and staff members (morning till night).

Generally TOD developments are planned around a major transit node. As the distance from

the node increases the density often decreases. However, KGUV does not have a major

transit node at its centre; rather, transit services are available along the corridors mentioned

above. Although the three main bus corridors suggest good public transport coverage, the

Quality of Service (QoS) needs to be assessed further. The analysis for the availability

criteria of the QoS framework for KGUV is presented in the following section.

4.4 Suitability of case study TOD

4.4.1 Background

The suitability of the case study TOD will be analysed using the Quality of Service (QoS)

framework described in the Transit Capacity and Quality of Service Manual (TCQSM) (TRB,

2003). A QoS framework is a tool for assessing effectiveness or usefulness of transit systems.

The QoS for a transit facility is the measure of the performance of the system within a

particular area from the passenger’s point of view. The performance measures used for

evaluating the QoS can be qualitative or quantitative in nature. The QoS framework given in

the TCQSM (TRB, 2003) is shown in Table 4.3. The analysis for QoS for a transit facility is

based on the type of service provided for the transit system; either fixed-route service or

demand-responsive service. The QoS framework is divided into “Availability” and “Comfort

and Convenience” measures. A primary measure of QoS is assigned to each of these

attributes, at three levels of scale; for individual transit stops, for route segments / corridors,

and for the whole system. The QoS is determined by calculating the level of service (LOS)

separately for each parameter. The LOS is graded on a scale from A to F, with “A” being best

and “F” being the worst result. The various performance measures used are listed in Table

4.3.

Table 4.3 Quality of service framework: Fixed – route (TRB, 2003, Exhibit 3–1)

Service measures

Stop Corridor

(Route segment)

System

Availability Frequency Hours of service Service coverage

Comfort and

convenience

Passenger load Reliability Transit – Auto travel time



4.4.2 Transit availability

“Availability” estimates how accessible the service is to the passengers, whenever required,

without considerable waiting time and walking distance (TRB, 2003). The transit availability

is measured from supply side without considering the public transport demand.

For transit stops, availability is measured by service frequency of the transit service. The

service frequency is calculated by considering various destinations from a particular transit

stop. Several routes serving the same destination and available from a particular transit stop

can be combined for analysis. According to the TCQSM (TRB, 2003), buses serving the

same destination and arriving at a stop within three minutes of each other are counted as one

bus for calculation of service frequency. Bus services to different destinations from a

particular transit stop should not be combined for calculation of service frequency. The LOS

is determined using values of average headway or average frequency.

For route segments or corridors, availability is measured by service span (hours of service)

for a transit service between two locations during the entire day. This is to check whether

service is provided when it is needed. Service span is calculated by route rather than by trip.

When the service is provided at least once in an hour then an extra hour is added to the

calculated day service span. This additional one hour takes into account the last hour of the

service provided.

For the system, availability is measured by the system area served, which is the percentage of

the inhabited area within walking distance of a transit stop. This analysis can be performed in

two ways; either using a Geographical Information System (GIS) or a manual method. The

GIS method involves drawing circular buffers of 400m (for a bus stop) and 800m (for a

premium node such as a busway or railway station) around the transit stop or station. These

buffers of 400m and 800m represent the normal walking distance for the passengers using the

transit service; corresponding to 5 minutes and 10 minutes walk respectively. In the manual

method, the actual transit service radius is calculated by using a street connectivity factor,

grade factor, population factor and a pedestrian crossing factor. A physical (actual) map is

observed for determination of values of these factors. Alternatively, actual walk buffers may

be determined.

All three measures are useful for evaluating the availability of transit service to the potential

passengers within a case study area such as KGUV. Evaluation using these measures has

been performed to determine the LOS for transit availability at KGUV.



4.4.3 Comfort and convenience

“Comfort and convenience” is the standard or quality of service provided; including comfort

onboard, reliability for the service, and comparative journey time (TRB, 2003).

For transit stops, comfort and convenience is measured by passenger loads onboard,

determined by calculating the load factor and / or standing passenger area during the study

period/s. These require knowledge of the characteristics of the buses observing the stop and

the number of passengers onboard.

For route segments or corridors, comfort and convenience is measured by reliability; either

on-time performance, or headway adherence. On time performance measures the percentage

of buses maintaining their time within 5 minutes of their schedule along the route. Buses

arriving early are not counted as on time. Headway adherence is used to determine the

reliability of transit services operating at headways of ten minutes or less, and is based on the

coefficient of variation of headways.

For the system, comfort and convenience is measured by the travel time difference between a

door to door trip made by car and the same trip made by using transit. For car, the travel time

includes in vehicle time, parking time, walking time from parking to destination and for

transit, the travel time includes the time taken by the user to reach bus stop, waiting time,

time on the bus, walking time to the destination and the transfer time if applicable. For a

system a basket of trips may be used during a given day or across a number of days.

All three measures are useful for evaluating the comfort and convenience of transit service to

the potential passengers within a case study area such as KGUV. But the evaluation using

these measures is more data intensive and requires skilled personnel; hence for time and

expense reasons, the comfort and convenience analysis was not performed for assessing

suitability of KGUV. However, it is reported that Brisbane buses services overall offered

better performance in terms of on time running and comfort for its users (TransLink, 2010).

4.5 Transit availability for KGUV

4.5.1 Analysis background

A TOD is a compact area with increased population density and employment opportunities

and some people may live and work in the TOD, however others will live there and work

offsite, while others still will live off site and visit the TOD for work, shopping, education or

recreational purpose. This is expected to be the case for KGUV.



A readily available transit system will enhance transit mode share for trips between the site

and off site locations. So the analysis for determining QoS for a development should consider

it both as an important destination and origin for trips. Thus, all analyses have been

performed for both directions.

As discussed earlier, KGUV has three fixed route transit corridors; the analysis has been

performed separately for each corridor. The buses passing by a transit stop but not observing

it were not considered in the analysis. Analyses have been performed for four distinct time

periods corresponding to the following ticket validity periods set by the transit planning and

delivery agency, TransLink:

morning peak between 7am and 9am,

off – peak day time between 9am and 3pm,

evening peak between 3pm and 7pm, and

off – peak evening time between 7pm and 10pm.

The extended peak hour for evening peak is considered to be four hours rather than normal

two hours. Services operating during the early morning before 7am and late evening after

10pm were not considered in the analysis. The schedules for weekend days, public holidays

and NightLink services (operating only on Friday and Saturday nights) have not been

included.

The local bus stop at School Street on Kelvin Grove Road was not considered separately in

the analysis, because all the buses observing these stops also observe the bus stop on Kelvin

Grove Road at Blamey Street. The bus stop at Victoria Park Road was also not considered

separately for the same reason. This stop serves only route 364 operating from Herston to the

CBD in the evening and on weekends. No direct bus route was found to and from the north-

east side of KGUV so no destination was considered in that direction.

The analysis for transit frequency and hours of service was performed considering KGUV as

a destination for various origins located to the north and south of KGUV, and vice versa.

Figure 4.2 shows the location of KGUV with respect to the Brisbane CBD and various

northern suburbs. For this analysis, Kelvin Grove Road and QUT KG Busway Station were

considered separately. The Normanby Busway Station was not considered separately because

this would have given an optimistic picture of the LOS for transit frequency and hours of

service. All the buses served Normanby Busway Station and then split their routes to Kelvin

Grove Road or to QUT KG Busway Station. In order to avoid double counting, Normanby



Busway Station was not considered in analysis. On the contrary, the Normanby Busway

Station was considered for calculation of transit supportive area (TSA) to determine the

proportion of area served by Normanby Busway Station. In this case, even if the TSA overlap

is observed it will not be double counted.

Source: www.whereis.com

Figure 4.2 Regional map showing various offsite attractions considered in analysis

4.5.2 Availability – Transit stops

Table 4.4 defines the fixed route service frequency LOS ranges according to the TCQSM

(TRB, 2003). UK traffic identifies 12 – 15 minutes time interval for a random arrival for a

passenger (Balcombe et al., 2004) while US guidelines denote 10 – 14 minutes time interval

for a random arrival (TRB, 2003). This difference might be because of difference in attitudes

between public transport users. For cultural reasons Australian users are more likely to

behave as users from the United States of America; hence the guidelines provided by TRB

(2003) are used in this study.



Table 4.4 Fixed – route service frequency LOS (TRB, 2003, Exhibit 3 – 12)

LOS Average headway

(min)

veh / h Comments

A < 10 > 6 Passengers do not need schedule

B 10 – 14 5 – 6 Frequent service, passengers consult schedules

C 15 – 20 3 – 4 Maximum desirable time to wait if bus / train missed

D 21 – 30 2 Service unattractive to choice riders

E 31 – 60 1 Service available during the hour

F > 60 < 1 Service unattractive to all riders

For those living at KGUV various destinations were analysed. The popular destinations for

residents such as Brisbane CBD and Cultural Centre (both to the south of the study area), and

QUT Gardens Point (City) campus were considered. It is noted that busway stations in the

CBD and Cultural Centre are the key interchange hubs for regional transport, both bus and

rail. For reference, a major destination along each corridor to the north of KGUV was

considered; Carseldine, Aspley, Chermside and Everton Park. The locations of the various

destinations are shown in Figure 4.2. The timetables, obtained from the TransLink website

(www.translink.com.au), for all bus routes serving the same destinations from the same

transit stop were compiled to calculate the service frequencies. Average frequency was

calculated by dividing the total number of buses in the time period by the number of hours.

Average headway was calculated by taking the inverse of average frequency and then

converting to minutes. When combined, certain buses arriving at the transit stop within three

minutes of each other were counted as one service. LOS results obtained and the total

numbers of bus services noted during the analysis period of 7.00am to 10.00pm are shown in

Table 4.5 and Table 4.6. These are not the daily totals.

Table 4.5 LOS for various trip destinations originating from KGUV (Kelvin Grove Road bus

stops)

Time Period Gardens

Point

CBD

(1)

Cultural

Centre (1)

Aspley

(1)

Everton

Park (1)

Carseldine

7am to 9am D/C A A C F C

9am to 3pm C A B/C C E C

3pm to 7pm C A C B C E

7pm to 10pm E B C B/C No service E

Total No of

Services 48 145 67 68 23 34

Note: (1) – Indicates results for Kelvin Grove Road bus stops



Table 4.6 LOS for various trip destinations originating from KGUV (QUT KG Busway Station)

Time Period CBD (2) Cultural Centre (2) Chermside (2)

7am to 9am A A A

9am to 3pm A A A

3pm to 7pm A A A

7pm to 10pm A A B/C

Total No of services 177 121 127

Note: (2) – Indicates results for QUT KG Busway Station

The same procedure was then repeated for those visiting KGUV by considering the locations

mentioned above as origin of trips and KGUV as a destination. The results obtained and the

total numbers of bus services noted during the analysis period of 7.00am to 10.00pm are

listed in Table 4.7 and Table 4.8.

Table 4.7 LOS for various trip origins where destination is KGUV (Kelvin Grove Road bus

stops)

Time Period Gardens

Point

CBD

(1)

Cultural

Centre (1)

Aspley (1) Everton

Park (1)

Carseldine

7am to 9am C A C A A E

9am to 3pm C A C C E C

3pm to 7pm C A B C E D

7pm to 10pm E B/C C B/C No service E

Total No of

services 48 145 67 66 23 33

Note: (1) – Indicates results for Kelvin Grove Road bus stops

Table 4.8 LOS for various trip origins where destination is KGUV (QUT KG Busway Station)

Time Period CBD (2) Cultural Centre (2) Chermside (2)

7am to 9am A A A

9am to 3pm A A A

3pm to 7pm A A A

7pm to 10pm A A A

Total No of services 171 119 127

Note: (2) – Indicates results for QUT KG Busway Station

The following points are observed from the LOS results from the point of view of residents of

KGUV who are living in KGUV and going outside KGUV for various purposes:

Transit is a very good option for KGUV residents commuting to the CBD and

Cultural Centre (and connecting to/from other transit services at these locations) and

to Chermside.



Transit seems to be a fair to poor option for KGUV residents commuting to the other

outlying suburbs Aspley, Everton Park and Carseldine in the morning and home in the

evening.

Transit remains a poor option at night for KGUV residents except to or from the

CBD, Cultural Centre, and Chermside, where it is good.

The following points are observed from the LOS results from the point of view of visitors to

KGUV for various purposes:

Transit is a very good option for visitors from the CBD and Cultural Centre (and

those connecting to/from other transit services at these locations).

Transit is a good option for visitors from Aspley in the peak periods and fair in the off

peak period.

Transit is a fair to good option for visitors from Everton Park in the peak periods and

poor in the off peak period.

Transit is a very good option for visitors from or to Chermside throughout the day.

Transit is a poor option for visitors from Carseldine except during the outbound

morning peak period and daytime off peak period when it is fair.

Transit is a poor option for the visitors going back to the outlying suburbs at night,

except Aspley and Chermside.

Transit offers a fair option for students and staff members using the 391 QUT

intercampus shuttle service during the day but a poor option in the evening.

4.5.3 Availability – Route segments / corridors

Table 4.9 defines the fixed route hours of service LOS ranges according to the TCQSM

(TRB, 2003).

Table 4.9 Fixed – route hours of service LOS (TRB, 2003, Exhibit 3 – 13)

LOS Hours of service Comments

A 19 – 24 Night or “owl” service provided

B 17 – 18 Late evening service provided

C 14 – 16 Early evening service provided

D 12 – 13 Daytime service provided

E 4 – 11 Peak hour service only or limited midday service

F 0 – 3 Very limited or no service

For calculation of hours of service with respect to route segments the same destinations were

considered as above. The hours of service were calculated for a round trip; originating from a

bus stop and terminating at the same bus stop. For example, if we consider Kelvin Grove



Road as the origin and the CBD as the destination, then the round trip from Kelvin Grove

Road – CBD – Kelvin Grove Road was considered. The hours of service were calculated by

subtracting the time of last service departing from CBD to Kelvin Grove Road from the time

of departure of first service from Kelvin Grove Road. Although suggested by the TCQSM

(TRB, 2003), an extra hour has not been added to calculate the hours of service because it

was considered to give an overly optimistic picture of this LOS measure. The results for

hours of service for various round trips with the numbers of bus routes are listed in Table

4.10.

Table 4.10 Hours of service LOS for different corridors with bus route numbers

Route Segment (Round trip) Bus routes LOS

QUT Kelvin Grove Campus – QUT Gardens Point

Campus – QUT Kelvin Grove Campus 391

D

QUT Gardens Point Campus – QUT Kelvin Grove

Campus – QUT Gardens Point Campus C

Kelvin Grove Road – CBD – Kelvin Grove Road 344, 345, 351,

357, 359, 390

B

CBD – Kelvin Grove Road – CBD B

QUT KG Busway station – CBD – QUT KG Busway

Station 330, 333, 340,

376, 393, 680

A

CBD – QUT KG Busway Station – CBD B

Kelvin Grove Road – Cultural Centre – Kelvin Grove

Road 345 B

Cultural Centre – Kelvin Grove Road – Cultural Centre C

QUT KG Busway Station – Cultural Centre – QUT KG

Busway Station 330, 333, 340

A

Cultural Centre – QUT KG Busway Station – Cultural

Centre B

Kelvin Grove Road – Aspley – Kelvin Grove Road 345

C

Aspley – Kelvin Grove Road – Aspley B

QUT KG Busway Station – Chermside – QUT KG

Busway Station 330, 333, 340,

680

B

Chermside – QUT KG Busway Station – Chermside A

Kelvin Grove Road – Everton park – Kelvin Grove Road 351, 357, 359

E

Everton park - Kelvin Grove Road – Everton park E

QUT KG Busway Station – Carseldine – QUT KG

Busway Station 340, 392 C

Carseldine – QUT KG Busway Station – Carseldine B

The following points are observed from the LOS results for hours of service for the round trip

undertaken by residents of KGUV:

Transit is a good option for KGUV residents commuting to the CBD and Cultural

Centre throughout the day (and night and commuting to or from other services at

these locations).



Transit is a good to fair option for KGUV residents commuting to outlying suburbs,

except for Everton Park with limited evening service.

Transit is a limited option for KGUV students and staff members commuting to GP

Campus with service limited to the daytime and early evening.

The following points are observed from the LOS results for hours of service for the round trip

undertaken by visitors to KGUV:

Transit is a good option for visitors coming to KGUV from the CBD and Cultural

Centre across the day (and night and commuting to or from other services at these

locations).

Transit is a fair to good option for visitors coming to KGUV from outlying suburbs

throughout the day, except from Everton Park with no late evening service provided.

Transit is a limited option for students and staff members visiting to KGUV with

service limited to the daytime and early evening.

4.5.4 Availability – System

Table 4.11 defines the fixed route service coverage LOS ranges according to the TCQSM

(TRB, 2003).

Table 4.11 Fixed – route service coverage LOS (TRB, 2003, Exhibit 3 – 14)

LOS % TSA covered Comments

A 90.0 – 100.0% Virtually all major origins & destinations served

B 80.0 – 89.9% Most major origins & destinations served

C 70.0 – 79.9% About ¾ of higher – density areas served

D 60.0 – 69.9% About two – thirds of higher – density areas served

E 50.0 – 50.9% At least ½ of the higher – density areas served

F < 50% Less than ½ of higher – density areas served

Transit supportive area (TSA) is the portion of the area being analysed having a household

density of at least 7.5 units per gross hectare or an employment density of at least ten jobs per

gross hectare. All 16.57 Ha area of KGUV is TSA. The system availability was calculated

using the MapInfo Professional 8.5. The availability for system was analysed for the

following distinct service coverage areas:

for bus stops on Kelvin Grove Road (separately for stops in both directions) and

Normanby Busway Station,

the QUT KG Busway Station, and

the area covered by the QUT intercampus shuttle service (Figure 4.1).



The image of KGUV was obtained from the Google Earth. A GPS survey was carried out to

determine the exact latitude and longitude of bus stops (accurate position of bus stops) at

KGUV and data for four distinct points at each end was collected for the image registration in

MapInfo 8.5. A GPS, Garmin 76S was used for data collection. The data for inbound and

outbound bus stops was noted separately. Buffers were drawn around the bus stops with

specific radius. The radius of the buffers was calculated using Equation 4.1 (TRB, 2003).

𝑟 = 𝑟0𝑓𝑠𝑐𝑓𝑔𝑓𝑝𝑜𝑝 𝑓𝑝𝑥 Equation 4.1

where,

𝑟 = radius of the buffer

𝑟0 = the ideal transit stop service radius

𝑓𝑠𝑐 = the street connectivity factor

𝑓𝑔 = the grade factor

𝑓𝑝𝑜𝑝 = the population factor

𝑓𝑝𝑥 = the pedestrian crossing factor

The value of 𝑟0 is 400m for a bus stop and 800m for a busway station or rail station. The

value of 𝑓𝑠𝑐 the street connectivity factor was universally equal to 1.0 due to the grid street

layout. The 𝑓𝑝𝑜𝑝 was universally 1.0 as most of the residents are young singles and 75 % of

them are ages under 45 (The Hornery Institute and Hassell, 2004). The 𝑓𝑝𝑥 was universally

equal to 1.0 as all pedestrian crossing delays were less than 30 seconds. The grade factor (𝑓𝑔)

was calculated by taking the average grade for various walking distances placed at extreme

ends of KGUV to the related bus stops. Then after image registration and establishing exact

location of bus stops, the circular buffers of the radius obtained by calculation were drawn for

all cases by considering the respective bus stops. The buffers for the bus stops are shown in

the Figure 4.3 to Figure 4.6. Table 4.12 lists the radius of buffer, percentage of TSA covered

and LOS for system availability.



Table 4.12 LOS for different bus stops

Transit service Buffer

radius (m)

% TSA on

KGUV covered

LOS

QUT intercampus shuttle service 400 100 A

QUT KG Busway Station 760 100 A

Kelvin Grove Road Stop at Blamey Street

and Normanby Busway Station

400 or 760 96 A

Kelvin Grove Road Stop at Prospect

Terrace and Normanby Busway Station

400 or 760 95 A

The following points are observed from the LOS results of service coverage area from the

point of view of both residents of, and visitors to, KGUV:

QUT KG Busway Station provides excellent coverage to the whole study area,

considering that it provides a premium service that people are prepared to walk

further to access, as reflected by the 760m radial catchment. This caters for

passengers to or from the northern and northeast suburbs including Chermside and

Carseldine, the CBD and Cultural Centre.

The two QUT 391 intercampus shuttle service bus stops located on Musk Avenue

provide excellent coverage to the study area, which is to be expected given that they

are on the main street spine of KGUV. These cater for passengers to or from QUT’s

Gardens Point (City) campus.

The bus stops on Kelvin Grove Road inbound and outbound to the CBD and

Normanby Busway Station provide excellent coverage to the study area. These cater

for passengers to or from the northern and northwest suburbs including Aspley and

Everton Park, the CBD and Cultural Centre.



Figure 4.3 Buffers for bus stops for QUT route 391 intercampus shuttle service (ST3 & ST4)

Figure 4.4 Buffer for QUT KG Busway Station (ST1)



Figure 4.5 Buffers for bus stop at Kelvin Grove Road at Blamey Street (ST5) and Normanby

Busway Station (ST8) (inbound to CBD)

Figure 4.6 Buffers for bus stop at Kelvin Grove Road at Prospect Terrace (ST7) and Normanby

Busway Station (ST8) (outbound from CBD)



4.6 Interpretation of results

The following observations can be drawn from the Transit QoS Availability analysis

undertaken for KGUV:

Transit availability to the Brisbane CBD and Cultural Centre (CBD is to the south of

KGUV) is very good in terms of frequency and hours of service for KGUV residents

and visitors.

Transit is a fair option for students and staff members using QUT intercampus shuttle

service in terms frequency in both directions however hours of services are limited to

the daytime and early evening only.

Transit remains a good to fair option for KGUV residents travelling to certain

outlying suburbs in terms of frequency and hours of service,

Transit is a good option for visitors coming from certain outlying suburbs but for

others it offers fair to poor transit availability and hours of service.

All the bus stops, and both busway stations, offer very good transit service coverage

for KGUV residents and visitors.

The overall observation of results shows that KGUV has good transit availability and

therefore as a TOD represents a worthy case study. The QoS determination for transit

availability was effective and useful for this desktop study, however, comfort and

convenience analysis would require a substantial field data base, which reduces the method’s

effectiveness. This aspect is an area proposed for future research.

4.7 Summary

The composition of mixed land uses at KGUV suggests that the site has appropriately placed

land uses for self containment and reduced parking facilities to restrict car use. And the

analysis for QoS for transit availability indicates that KGUV has overall good public transit

availability to or from various origins and destinations. Due to these qualifying characteristics

KGUV was selected as the case study TOD site for further research.

4.8 Chapter close

This chapter provided details about the case study TOD site and the criterion for assessing its

suitability which completes the first step of TOD evaluation, pre–TOD assessment. Next

chapter provides the details of data collection process undertaken to collect the transport data



for the case study TOD, KGUV. The details of KGUV will also be useful for the traffic

generation calculations displayed in Chapter 6.


Chapter 5

Data collection

5.1 Introduction to data collection

The main objective of this chapter is to document the procedure followed for gathering traffic

and travel data for Kelvin Grove Urban Village (KGUV) and list the findings flowing from it.

As an initial step, the availability of existing datasets was checked. No dataset was found

explaining traffic and travel at KGUV. KGUV was a newly planned development hence the

data was required to be collected by conducting surveys. Before collecting the data, an

observational study was conducted to gain an understanding of the traffic and users at

KGUV. The next section presents an overview of the users using various land uses at KGUV

followed by the details of cordon counts conducted for gathering traffic data. Later the

methodology used for conducting the travel surveys is explained with the help of procedural

details along with sample sizes and response rates. In the last section, the lessons learned

from the data collection process are reported, followed by a brief summary and chapter close.

5.2 User groups at KGUV

In case of a Transit Oriented Development (TOD), the diverse mix of land uses provides

space for various categories of people to interact in a relatively small area. Various people

interacting at KGUV include residents, students, shoppers, employees and recreational users.

Mainly, the people residing in the TOD are termed as „residents‟ and people using the TOD

but residing outside the TOD boundary are termed as „visitors‟. For the purpose of this

research, the user groups were denoted based on the user characteristics and land uses they

use. In general, more than one user group was assigned for one land use. For example, two

user groups were specified for the users of the educational land use; namely students and

employees. The users of residential land use are termed as Residential Land Use (RLU) users

and Non Residential Land Use (NRLU) users. The following subsections note the overview

of land use and the users using the respective land uses.

5.2.1 Employees

The commercial land use at KGUV is comprised of a centrally located shopping centre, and

office land use comprising education related and private sector employment. The shopping

centre has retail outlets serving the KGUV users. The office land use has a separate block for



private companies and combined research facilities for the university (QUT). The office

spaces for private companies were under development so were not considered in this

research. The persons working at commercial and education land use were termed as

employees, who comprised the “employee” group for data collection.

The employees at KGUV were divided into two groups depending on the type of employment

they were engaged in; professional employees working at the university campus extension

and retail shop employees working at the centrally located shopping centre at KGUV. The

research students working at the research facility of the university were included in the

professional employee group because these two groups possessed similar characteristics such

as type of work, access to facilities and services, etc.

5.2.2 Shoppers

KGUV has a centrally located small local service function shopping centre. The shopping

centre has speciality shops and retail outlets. The shopping centre serves the KGUV and area

located within close proximity of KGUV. Anyone who shops at this shopping centre is

termed as a shopper at KGUV and these users formed the “shoppers” group for the purpose

of data collection. This group mainly consisted of employees, students at KGUV and some

people residing within close vicinity of KGUV.

5.2.3 Students

KGUV has a major share of education land use attracting a significant proportion of students.

It has a university campus extension and a high school for specialised education, which is a

separate entity from Kelvin Grove State College, which is located just on the north boundary

of the study area. The students undertaking studies at these facilities formed the “students”

group for analysis.

Similar to the employee user group, the students at KGUV were also divided into two groups

depending on the type of education; school students and university students. The students

studying at the high school formed school students group and the students studying at the

university formed the university students group.

5.2.4 Residents

The residential land use at KGUV is comprised of affordable apartments (public housing),

student accommodation, apartments and townhouses. Majority of the apartments are one

bedroom or two bedroom apartments. Most of the affordable housing apartments are

Data collection


occupied by single parent or young couples and the apartments are rented or owned by young

students or small sized families. The student accommodation was occupied by full time

students studying at universities located nearby KGUV. The persons living in these

apartments formed the “residents” group.

These diverse residential developments at KGUV provided living opportunities for people

representing various household characteristics. The residents at KGUV comprised of students

studying at the university, young couples and senior citizens possessing different personal

characteristics. The Student Residents (SR), living in student accommodation, and Non

Student Residents (NSR), living in other apartments, were the two groups of residents.

5.2.5 Recreational users

The recreational land use at KGUV consists of a theatre, an activity centre, parks and open

spaces. The various KGUV users, residents at KGUV and visitors use these facilities. These

users are termed as “recreational” users.

Figure 5.1 gives an overview of the various land uses and user groups at KGUV, which are

considered for further investigation.

Figure 5.1 Overview of land uses and user groups at KGUV

TOD Land uses Users User groups

Kelvin Grove

Urban Village

Residential

Commercial

or retail

Education

Students

University students

Non student residents

Employees

Shoppers

Residents

Professional employees

Retail shop employees

Shoppers

School students

Student residents



5.3 Cordon counts

Cordon counts were conducted to collect the traffic data for KGUV and determine the traffic

characteristics of this stated TOD. The locations for the cordon counts were selected from the

observational study conducted for KGUV. The access points were primarily considered as the

locations of cordon counts. In total 18 access points were targeted to gather the traffic data

for KGUV. Eight locations were designated for conducting counts at 18 access points. Thus

the recorder or observer standing at one location could collect traffic data for more than one

access point. Figure 5.2 shows the locations of the cordon counts for KGUV.

The traffic counts were not conducted for 24 hours due to a lack of resources; instead three

distinct time periods, morning peak (7:45am to 9:30am or 7:15am to 9:30am), midday peak

(11:30am to 1:30pm), and evening peak (3pm to 6pm) were chosen for traffic data collection.

From the many observations done at various times of the day, it was noticed that much of the

travel activity occurred during this time period. The midday peak was intentionally

considered to determine the details of activities during the typical lunch hour period. The

counts were conducted for a typical weekday during the school term. The cordon count for

AM peak period was undertaken on Wednesday morning and all other counts (midday peak

and PM peak) were undertaken on Thursday.

The bi–directional counts were undertaken for all modes of transport; car, buses, pedestrians,

motorcycles and bicycles, to capture vehicle and pedestrian movements inbound and

outbound of KGUV. The car occupancy was noted along with the number plates. The string

of first four characters of the number plate of each vehicle was noted. This data was later

cross matched to determine the amount of through traffic. The data was recorded in 15

minutes intervals for all three time periods. A sample of the form used for recording the

cordon data is included in Appendix B.

In addition to the above mentioned cordon count, a separate count was done for the centrally

located shopping centre. The shopping centre was considered separately because it was the

highest trip attractor, especially for the intra-zonal trips due to the retail mix. Hence it was

important to gain an understanding about the pedestrians as well as vehicular activities for the

shopping centre, as these could be different from the conventional shopping centre. The

shopping centre counts were conducted at two locations for four access points. The number

of cars, pedestrians, motorcycles and bicycles were noted. The car occupancy was noted but

the data for number plates were not collected. The bi–directional counts were conducted for

Data collection


the same three time periods which were used for cordon counts were used for collecting the

traffic data for the shopping centre.

Represents the locations of cordon counts for whole of KGUV

Represents the locations of counts for Village Centre

Figure 5.2 Locations of cordon counts at KGUV

5.4 Travel surveys

Revealed preference travel surveys were carried out in phases for all groups of KGUV users

(except recreational users) mentioned in Section 5.2; namely one for shoppers, one for

employees (professional employees and retail shop employees), one for students (school

students and university students) and one for residents (student residents and non student

residents). It is acknowledged that there can be some overlap of users using multiple

developments and responding to the same survey. No method was used to cross check

responses from the same person for multiple questionnaires as this might have confused the

respondents. An overview of the general methodology is outlined in the following section

followed, by critical issues and observations made while conducting each survey in

subsequent sections.



5.4.1 General methodology for conducting travel surveys

The step by step procedure followed in conducting the travel surveys for each user group is

demonstrated in Figure 5.3 with the help of a flowchart. The first step for travel data

collection involves selection of variables about which the information needs to be collected

from the travel survey. The variables were selected from existing literature and practical

conditions of the study area. The selection of survey instrument and the design of survey

questions were governed by these chosen variables. Sample size and the required target

response rate were the main factors determining the type of survey instrument to be used.

Selection of an acceptable response rate was crucial as it controlled the decision of whether to

stop or continue the main surveys. After this step, the questionnaire survey form was

designed in such a way that all questions were easy to understand by a lay person. Use of

technical terms was avoided while designing the questionnaire form to simplify the questions.

The survey questions and survey technique were tested for suitability, before conducting the

main travel surveys, by conducting pilot studies on some respondents. Responses to the

questions implied suitability of the survey technique and layout of the questions. Some

changes to the survey instrument or the survey questions were required at this stage in order

to improve the response rate and quality of responses for main surveys. The descriptions of

the changes made are noted in Section 5.4.5.

After updating the questionnaire from the pilot study, the main survey was conducted for

people selected randomly in the user group under study. Once the required response rate was

achieved, the surveys were discontinued and all collected responses were compiled and

combined together for preliminary data analysis. In subsequent stages of the research this

travel data can be used for travel demand modelling. This process was repeated for all user

groups at KGUV.

Data collection


Figure 5.3 Steps involved in data collection process for a user group travel survey

No

No

Yes

Yes

Yes

No

Update

questions?

Main surveys

Organize responses and

prepare database

Preliminary data analysis

Collect more survey data

Update questions

Update survey

method?

Is response rate

acceptable?

Variables

selection

Literature

review

Investigation of various

survey instruments

Selection of appropriate

survey instrument

Design questionnaire form

Pilot study

Decision of

sample size and

response rate

Review responses



5.4.2 Selection of survey instrument

The selection of survey instrument depends on the characteristics of respondents. After

reviewing the available survey techniques it was found that, due to the diverse mix of people,

a single methodology was not suitable for all users at KGUV. Hence a combination of

various survey techniques, such as mail back survey, internet based survey, personal

interview and intercept survey, was used to collect data related to KGUV travel. An

appropriate survey technique for each group of user was selected considering the data

collection time and cost involved in conducting the surveys. A computer assisted personal

interview (CAPI) survey was selected for collecting the travel data for shoppers. CAPI

surveys were favoured because survey techniques involving computer assisted methods, new

technologies and face–to–face techniques shown elsewhere to provide a higher response rates

(Korimilli et al., 1998). Shoppers were also given an option of taking the survey forms with

them with a reply paid envelope.

Different survey methods were adopted for the two types of employee groups. An internet

based survey was chosen for the professional employees because it is less time consuming,

more cost effective, and flexible for respondents. All of the professional employees also had

work email addresses and good access to internet at their work places. By contrast, the

employees at the shopping centre were surveyed using a CAPI survey. This method was

chosen as these employees did not necessarily have access to internet at their workplace.

Their working hours were variable; hence it was necessary to book times for meetings and to

contact them personally.

The two student groups at KGUV were also surveyed using separate survey instruments. An

internet based survey technique was used for the university students. This was done for

similar reasons that applied to professional employees. A mail back survey technique was

chosen for the school students. This method was selected particularly because it was

important to obtain the responses with written consent from parents of students who were

mostly under 18 years of age (adulthood).

Similar to other user groups, two different methods were used to survey the residents. In case

of residents the methods used for other user groups were not appropriate as it was not

possible to obtain either contact email addresses of the residents or any personal contact

details like telephone number, names of persons. These details were not available due to

Privacy law. A modified mail based method was used to gather data related to NSR‟s travel.

The respondents were given a choice of mail back, internet based surveys, telephone and

Data collection


personal interviews. An introductory letter was posted to NSRs requesting the preferred

method and the surveys were conducted based on the method chosen by the respondents.

Similar to NSRs, the SRs were given an option of mail back survey or internet based survey.

A notice was placed on the notice boards of the student accommodation with a web link for

the survey for the SRs at KGUV asking for response. If mail back survey was chosen then the

questionnaire survey was posted to the respondent.

To acknowledge the participants for their time and input some incentives were given to

certain user groups. The details of the same are noted in Section 5.4.4.

5.4.3 Design of questionnaire form

A separate questionnaire form was designed for the users of the non residential land use and

the users of the residential land use. Ethical clearance was obtained from the Research Ethics

Unit of the QUT to ensure the appropriate design of questionnaire forms. The following

sections explain details of each questionnaire form.

5.4.3.1 Questionnaire for non residential land use users

A separate questionnaire survey was designed for each category of user. The users of non

residential land use were asked about their travel details for the specific trip to KGUV by

their most usual way, along with personal and household details. Most of the questions were

multiple choice and stated in a layperson‟s language. While specifying possible responses,

broad ranges were given instead of asking overly specific questions. For example, the age

group of a respondent was asked instead of exact age. The respondents were given five

choices of age groups to choose from (0–18years, 18years–30years, 30years–45years,

45years–65years and 65years and above). The school students‟ survey was designed as a

short survey keeping in mind the age of the respondents. Some questions were excluded from

the general questionnaire form. A copy of the final questionnaire form can be obtained from

Muley et al. (2008). The questions in the questionnaire form asked information on the

following aspects:

Mode of travel to work, mode choices (if available), mode specific questions (like

parking fees, parking place, boarding and alighting stop location, public transport

route number, transfer location, walking time to and from the stop) and reasons for

choosing the selected mode of travel

Perceptions of existing public transport at KGUV and any improvements the

respondent considers are required to improve existing public transport



Perceptions of KGUV

Personal and household information (age, gender, occupation, vehicle ownership,

number of driving license holders, size of household, etc.)

Figure 5.4 Questionnaire details for non residential land use users

Yes

Introduction to travel survey

Usual mode of travel to KGUV

Private

vehicle Bicycle

Public

bus Train Ferry Taxi

Walk

only Other

Parking details Public transport details Specify

Other option

available? Specify options

Perception about public transport at KGUV

Activity purpose

(if not travelling home)

Perception about KGUV

Comments

Personal and household information

No

Frequency of travel to KGUV

Data collection


The flow of the survey questions is shown in Figure 5.4. The answer to the first question

asking the usual mode of travel to KGUV was compulsory. The respondent who did not

answer this question was directed towards the last page. Skip logic was used to direct

respondents to direct them to provide their travel details based on their mode of transport.

All the questionnaire forms were designed using the popular American web based survey

design site called www.surveymonkey.com. It was estimated that each respondent would take

approximately 10 to 15 minutes to complete a survey. Although a separate questionnaire

survey was designed for each group of visitors, all questionnaire forms essentially collected

the same travel data.

5.4.3.2 Questionnaire for residents

A questionnaire form was designed for both resident groups. The residents were asked to

record the travel details for all trip purposes for the whole day. This was necessary as the

residents were the only trip producers in KGUV. Although different survey techniques were

offered to the residents, the design of questionnaire for all the survey techniques was similar

in order to obtain the same data set for all the respondents. The same online survey tool was

used to design this questionnaire survey. A working Wednesday was chosen as the travel day

for the residents.

The questionnaire was composed of two forms; household information form and travel diary.

A copy of questionnaire form can be obtained from Muley et al. (2009). The household

information form collected details about the household characteristics such as type of

household, number of bedrooms in the household, household size, vehicle and bicycle

ownership, number of valid driving licence holders and street name.

A travel diary was designed to collect travel data related to residents‟ travel and personal

information. The travel diary was divided into three main sections. The first section asked

questions related to personal information like age, gender, occupation, employment status and

driving licence availability. The second section included questions related to each trip made

on the travel day. The travel details for each trip were similar to those listed for non resident

users‟ travel data. An option for filling maximum 10 trips was given to the residents, in case a

resident made more than 10 trips, a separate travel diary was provided upon request for

additional trips. The details of each trip included origin and destination of trip, start time and

end time of trip, mode specific questions and purpose of making the trip. The third and last

section asked questions related to the perceptions about public transport at KGUV,

http://www.surveymonkey.com/



perceptions about KGUV and any special comments. An overview of the survey form design

for the users of residential land use is given in Figure 5.5. It was estimated that each

respondent would take no more than 20 minutes to complete the questionnaire form.

Figure 5.5 Questionnaire details for residential land use users

Frequency of activities at KGUV

Perception about public transport KGUV

Comments

Perception about KGUV

Travelled more?

No

Yes

Mode choice/s (if available)

Trip purpose

Mode of travel for the trip

Private

vehicle Bicycle

Public

bus Train Ferry Taxi

Walk

only Other

Parking details Public transport details Specify

Time and location of origin and destination of the trip

Introduction to travel survey

Personal information

Household information

Travelled on

assigned travel day?

Frequency

of travel

No

Yes

Data collection


5.4.4 Reminder letters and incentives

All user groups were posted a reminder letter after two weeks of initial contact. If the number

of responses was less than the targeted sample size then a second reminder letter was sent

after four weeks of first reminder letter. The reminder letters emphasized the importance of

residents‟ response and participation, confidentiality and future benefits.

In previous research (Murakami & Watterson, 1992 and Tooley, 1996) it was found that use

of pre–incentives helped to achieve a better response rates, hence the participants of the

survey were given some small gifts to appreciate their participation and time. The

nonmonetary incentives were given to the respondents instead of monetary incentives. All the

respondents of CAPI surveys (shoppers and retail shop employees) were given a pocket pal

guide, providing information about the bus timetables obtained from the region‟s TransLink

Transit Authority as a reward for participating in the survey.

The NSR were posted a free coffee voucher of value $3.50 as a reward for participation in

advance with the questionnaire removing the ambiguity associated with the lucky draw prizes

in order to improve response rates. On the other hand, the participants of the SR survey were

given a book shop voucher of $5 for their contribution. No incentive was given to the

respondents of the internet based surveys (professional employees & university students) and

respondents of school students‟ survey because these was no individual contact made with

these users and also their personal contact details was not known to deliver the incentives due

to anonymity of the collected responses.

5.4.5 Process of conducting surveys

As stated in Section 5.4.2 the various KGUV users were surveyed using different survey

techniques, as the users had different characteristics. All the user groups were given a

covering participant information sheet that provided some information about the research,

contacts of research team, and the Research Ethics Committee at QUT for reporting any

issues or concerns about the project or questionnaire form. The procedure for conducting the

travel surveys for each user group was distinct; the procedural details for each user group are

listed in following subsections.

5.4.5.1 Shoppers’ survey

Initially, a CAPI survey was selected for shoppers. Pilot surveys were conducted to test the

selected methodology. A full questionnaire survey (with similar questions as noted in

employee survey (Muley et al., 2008), which took 10 to 15 minutes to complete, was offered



to shoppers, but it was found that they were always in a rush. Only two shoppers out 15

agreed to participate in the survey, but after talking to them it was found that these two

shoppers visited KGUV for the first time and they were not frequent visitors. Some people

were reluctant to stop and hear about the survey. Hence the size of questionnaire was required

to be reduced significantly for this group of subjects.

Only five questions were retained for the shoppers‟ survey, seeking information about their

shopping trip. The answers were noted down in a tabular format (Appendix B), so as to

reassure the shoppers that the survey would not take more than 1 or 2 minutes. Completing

the responses in the table made the survey process faster and easier for interviewers as well.

A good response rate of 72.7 percent was subsequently gained when this revised

methodology was tested. Considering the good response rate and time required to conduct the

survey, this methodology was used for the final survey.

5.4.5.2 Professional employee survey

To test the internet based survey technique for professional employees, a pilot survey was

sent to 30 professional employees selected at random. The employees were working at the

educational land use and the survey was sent to their work email addresses. A response rate

of 30 percent was obtained and the comments about the layout and design of the survey

questionnaire indicated that there was no need for any changes. Hence the same methodology

was adopted for conducting the main travel surveys. The participant information sheet was

attached to the email and displayed as a first page of the questionnaire form.

5.4.5.3 Retail shop employee survey

Initially, the questionnaire form that was used for professional employee survey was used for

conducting the main travel survey for employees at the retail shops. Before conducting the

main travel survey, the survey methodology was tested on only two employees and the

questionnaire form was shown to the owners or managers of all the shops to obtain

permission to conduct the surveys. These two surveys took around 10 to 15 minutes to

complete as expected. More employees were not surveyed for the pilot study as this is a small

scale shopping centre with about 125 employees. It was then decided to proceed directly with

the main survey.

When the main surveys commenced and the survey was conducted for all the employees, it

was observed that completion of survey form for some respondents took almost 30 to 45

minutes as the respondents were serving customers while the interview was in progress. After

Data collection


considering the time constraints on the respondents, some questions were removed as they

took more time to answer, these being the rating scale questions related to the perceptions

about public transport and KGUV and the reasons for choosing main mode of travel. The

survey methodology was also modified to “pen and paper” based interviews instead of using

“CAPI surveys”. This eliminated the time required to set up the laptop and gave flexibility to

respondents (some respondents preferred to fill the survey on their own). This accelerated the

data collection process and it took approximately 5 to 7 minutes to complete a questionnaire

survey.

5.4.5.4 School students’ survey

A mail back survey was used for school students‟ survey. Pilot surveys were not carried out

for school students due to the limited number of students. The questionnaire form used for

professional employees was used as a base for preparing the questionnaire form for school

students. The professional employee survey questionnaire was shortened keeping in mind

school students‟ age group. The questions related to the reasons for choosing the usual mode

of transport were omitted from the base questionnaire. Some questions related to personal

information like age, occupation, industry, and work postcode were also removed from base

questionnaire. A copy of the questionnaire can be found at (Muley et al., 2010). The survey

method and questionnaire form was approved by the School Principal before delivering it to

the students. The survey forms and parent consent sheets were handed over to the school

students at the school and the responses were collected via reply paid envelope provided with

the questionnaire form. The students were required to complete the questionnaire form and

return it using the reply paid envelopes provided with the signed parent consent form. The

responses without parent consent form were not considered for analysis. It was estimated that

a school student would take approximately 10 minutes to complete the survey.

5.4.5.5 University students’ survey

The same survey used for professional employees was utilised for university students. Hence,

a separate pilot survey was not conducted for the university students. The web link for an

internet based survey was placed at a university webpage, which the students were accessing

frequently. This web link appeared on each student‟s access page for two weeks. This method

was chosen because it was the best way to contact students without obtaining their personal

contact details. Similar to the professional employees, the participant information sheet was

displayed as the questionnaire cover page.



5.4.5.6 Non student residents’ survey

At the first stage, the NSRs were asked for their consent to participate and the method of

conducting the survey was described by way of letter box drop. In this letter, the residents

were given a choice of mail back survey, internet based survey, telephone interview or

personal interview. The residents were asked to register their interest with the number of

people living in the household by contacting any of the team members personally, through

email or telephone. After registration, the surveys were scheduled as per the chosen survey

technique and the place and time were decided as per the respondents‟ convenience. To test

this method, a pilot survey was undertaken for 42 households living in one affordable

housing block. No responses were obtained after posting the introductory letter. To modify

the process, discussions were arranged with the mangers of the residential apartments and the

method was modified to mail back questionnaire technique. The residents were also given a

choice of internet based surveys by providing a web link to enter into the survey.

Each household was mailed two travel diaries, a household information form, a free coffee

coupon and an introductory letter with instructions. The residents were requested to ask for

more travel diaries in case the number of people in the household was more than two, as one

travel diary was required for filling travel details of one person. The responses were collected

through a reply paid envelop provided with the questionnaire form. A response rate of 16

percent was obtained for the pilot survey undertaken for 48 households, so this method was

retained for the main surveys. The actual time required for filling the survey was not known.

The responses to the questions indicated that no change was required in layout of the

questions and language of the questions.

5.4.5.7 Student residents’ survey

The questionnaire form used for non student residents‟ survey was used for student residents‟

survey. Before conducting the main survey, the methodology for the SRs‟ survey was tested

on three students residing in the student accommodation. On average 15 minutes were

required to complete one questionnaire form. For the main survey, a notice was posted on the

notice boards of the student accommodation to gather interest in the survey and study.

However, only two responses were obtained for the internet based survey. In order to obtain

more responses, the methodology was modified to “intercept surveys”. Although the

questionnaire form was the same, this method obtained good response rates for the pilot

surveys because of personal contacts. Typically the intercept surveys were carried out during

late afternoon or early evening time periods. These time periods were chosen because the

Data collection


students returned home from university or came out for the evening activity and it was

noticed that they had some time to stop and answer the questions. If the students were in

hurry and were interested in participating in the survey then they were given the

questionnaire form, book voucher and a reply paid envelope. It was observed that the students

took 10 to 15 minutes of time to complete the questionnaire.

Table 5.1 presents summary of the travel surveys used for collecting travel data for various

groups of KGUV users. It should be noted that the details of the final surveys are listed.

Table 5.1 Summary of travel surveys

User group Survey

period

Survey

instrument

Type of travel

data

Survey

completion time

Professional

employees

March &

April 2008

Internet based Usual trip to work 10 to 15 minutes

Retail shop

employees

March &

April 2008

Pen & paper

based interviews

Usual trip to work 5 to 7 minutes

University

students

August 2008 Internet based Usual trip to

university

10 to 15 minutes

School students October 2008 Mail back surveys Usual trip to

school

10 minutes

Shoppers May 2008 Intercept surveys Specific trip on

survey day

1 to 2 minutes

Student

residents

March 2009 Intercept surveys Travel diary for a

weekday

10 to 15 minutes

Non student

residents

August &

September

2008

Mail back surveys Travel diary for a

weekday

20 minutes

5.4.6 Sample size and response rates

Determination of sample size is an important step before conducting any kind of survey.

Sample size is selected in such a way to represent the average household characteristics of the

study area. In the case of a TOD, the size is usually small; hence the travel surveys should be

given to all people using the case study TOD. This offers the sample size as 100 percent of

the population size. So there was no requirement to obtain the sample size using a specified

method. This decision sought to minimise coverage error and sampling error.

For KGUV, only people interacting within the KGUV boundary were targeted for data

collection. The sample size was kept as 100 percent of the population size to reduce coverage

and sampling error. For the shoppers‟ survey, best efforts were made to survey the entire

population; however some error of about 5 percent is estimated.



The number of responses and in turn the response rate both depend on the sample size.

Reasonable numbers of responses were obtained from the main travel surveys after

modifying the methodologies. Reminder letters were used only in the cases of non student

residents and professional employees. The reminder letters obtained additional 14 and 17

responses for non student residents and professional employees respectively. In general, a

minimum 30 responses or a response rate of 10 percent (whichever is more) was desired for

the travel survey of each user group.

The response rate for each survey was determined by dividing the number of complete

(usable) responses received by the total number of surveys sent (which is equivalent to the

size of population). For the shoppers‟ survey, instead the population size, the number of

KGUV shoppers approached for participation is taken as the number of survey sent. The

employees of educational land use showed a response rate of about 10 percent while the

response rate for employees of the retail shopping centre was approximately 31.2 percent.

The response rate for shoppers‟ survey was about 67.6 percent. Responses obtained from the

school students and university students were 20 percent and 15 percent respectively. While

for the student residents observed response rate was 11 percent and for non student residents

it was 8 percent. Although the response rate for NSR‟s survey was lower, the data collection

was stopped because of the limitations on the process of conducting the survey. For all other

surveys, the numbers of responses obtained were considered sufficient for travel

characteristics determination; hence the surveys were ceased at this point.

The mail back survey for residents showed the lowest response rate of 8 percent. This may be

due to large number of international students and young adults living in single family

households. Over-surveying of the residents can also be an important factor in lower response

rates as the residents had reportedly filled nine different surveys, not related to this research,

over a six months time period. It was also found that incentives did not help much in

obtaining a better response rate for the residents. The internet based surveys obtained

relatively lower response rates, around 10 to 15 percent compared to personal interviews (30

to 60 percent). Personal interviews, in our circumstances, are regarded as the best approach to

achieve a high response rate. A summary of survey instruments adopted and the final

response rates obtained after the surveys is given in Table 5.2.

Data collection


Table 5.2 Sample sizes and response rates

User

group

TOD user

group

Survey

instrument

Sample

size

Response

rate

Reason

Shoppers Shoppers Intercept

surveys 117 68%

Responses collected

in a tabular format

Employees

Professional

employees

Internet based 125 10%

Lack of direct contact

with respondents

Retail shop

employees

Personal

interviews 39 31%

Personal contact and

shorter questionnaire

Students

School

students

Mail back

surveys 28 20%

Use of shorter

questionnaire

University

students

Internet based 89 15%

No direct contact

with respondents

Residents

Student

residents

Intercept or

personal

interviews

51 11% Over surveying of

respondents by other

studies Non student

residents

Mail back

surveys 34 8%

Note: Sample size indicated number of usable responses

5.4.7 Sample bias

Generally, bias in the sample composition is inherited from the survey sample. Sampling bias

is caused due to unrepresentative sample, measurement error, sampling error and survey bias

(StatTrek, 2009). For all the surveys, the sample sizes were kept as close as possible to 100

percent of the population. Further, most suitable survey instruments were used for each user

group. In this case, there was no sampling technique applied, the survey was conducted under

the best possible environment for respondents, and the sample size was highest for maximum

coverage. Overall, it is believed that these survey responses contain minimal bias, although

any bias in the residents‟ data could be reduced if a larger sample size could be obtained.

However, the discussion above demonstrates the difficulty of achieving this.

5.5 Lessons learned

The observations and experiences from conducting the travel surveys at a stated TOD are

summarised below:

Multiple choice questions were quicker for respondents to answer, compared to

questions related to perceptions, which are usually designed on a rating scale.

The shoppers‟ survey had to be designed as a quick response survey in order to

maximise responses. A good response rate is likely to be obtained for a questionnaire

of one A4 page sheet or questions listed in the form of a table.



A pen and paper based survey was preferred over a computer based interview (CAPI)

for employees at retail shops. A higher response rate was obtained for personal

interview surveys compared to mail back survey forms.

An internet based survey was suitable for professional employees and university

students, as in both cases these groups had good internet access. This method offers

flexibility to these respondents, which improves response rates.

Similar to the observation made by Shih and Fan (2008), the newest internet based

survey technique obtained the lowest response rate for professional employees when

compared to other survey techniques, which include personal contact. The response

rate for internet based surveys was found to be dependant mainly on the age, and the

profession of the respondents.

Personal contact was the best way to approach to the respondents, although this

approach is time consuming and somewhat expensive.

The questionnaire for the school students needed to be short and simple with more

multiple choice questions, which are easy to answer. The mail back survey technique

is suitable for school students.

Respondents prefer a short survey over a lengthy survey, as was seen for the retail

shop employee survey.

For a TOD, ideally the entire population should be surveyed. If this is not possible for

a specific user group then maximum possible number of TOD users should be

contacted to achieve a sample size as close to 100 percent as practicable.

Although the mail back survey technique is most commonly used for a residents‟

survey, intercept surveys can yield better response rates and good quality data.

5.6 Summary

KGUV incorporates four prominent user groups; employees, shoppers, students and

residents. Due to various activities of these users, KGUV experiences traffic movements. The

cordon counts gathered traffic data for all modes of transport for KGUV. The diverse mix of

users required different survey instruments; hence various survey techniques were adopted to

collect travel data for all user groups. Variations were required to achieve better response

rates. An internet based survey was used for professional employees and university students,

personal interviews for retail shop employees, an intercept survey for shoppers and SRs and a

Data collection


mail back survey was used for NSRs and school students. The responses obtained from the

travel surveys yield information about transport at a TOD.

5.7 Chapter close

This chapter presented the methodology used for obtaining transport data, which completes

Step II of the TOD evaluation. The findings of this chapter provide the base for Step V of the

TOD evaluation. The subsequent chapters use traffic and travel data collected using

procedures described in this chapter for analysing TODs from a transport perspective. The

next chapter presents analysis and results for the traffic data analysis in terms of the traffic

generation. Chapters 7, 9 and 10 present the analysis and results obtained from travel survey

data analysis.




Chapter 6

Traffic generation at Kelvin Grove Urban Village

6.1 Introduction

A limited number of studies have been undertaken to study the traffic impacts of Transit

Oriented Developments (TODs) or mixed use developments. This chapter provides a detailed

overview of the traffic generated at the case study TOD, Kelvin Grove Urban Village

(KGUV), based on the cordon count data collected for KGUV and its centrally located

shopping centre known as Village Centre (VC). The traffic generation is compared with the

combined standard trip rates provided for homogeneous land uses because the data for similar

sized non – TOD developments or guidelines for trip rates for mixed land uses was not

available.

The first section of this chapter provides the methodological details of the cordon data

analysis. The following section presents an overview of the traffic at KGUV with the help of

car occupancy, hourly volumes and directional distribution for various modes of transport.

The comparison of peak period traffic with the trip rates from ITE (ITE, 2008) and Australian

sources (RTA, 2002) is undertaken in the following section. The final section presents a

summary of the chapter with a chapter close.

6.2 Analysis of cordon data

The details of the data collection were provided in Section 5.3. The analysis of cordon data

involved a stepwise approach, which is explained with the help of a flowchart in Figure 6.1.

The first step of analysis was to determine the amount of through traffic and eliminate it from

the recorded traffic volume. This task was performed only for the KGUV cordon counts and

not for the VC counts; all other tasks were similar for both the analysis. The possible

directions of the through traffic movements are shown in Figure 6.2. The through traffic

consists of the traffic that uses the road network at KGUV but does not stay in KGUV. Those

cars travelling in each direction were hence removed from recorded data for each time period.

There were six pairs of access cordon points, which were used by through cars. The through

traffic was only determined for cars. Through movements for all other modes of transport

were not determined because their data were not recorded with any identification number, as

in case of number plates for cars, to identify the through movements. It should also be noted



that KGUV has only one bus service running through it; the frequency of this bus service was

known hence number of buses were not counted during cordon counts. All other public

transport services were running on the edge of KGUV, hence these services did not use the

road network at KGUV hence these buses were not counted during cordon counts. However,

the frequency of buses can be used as a measure to determine the public transport traffic.

Figure 6.1 Process of cordon data analysis

The traffic data obtained after removal of through traffic was used for further analysis. The

calculations were undertaken for all modes of transport; cars, motorcycles bicycles, and

pedestrians. The person movements were also determined by calculating the total number of

persons travelling in or out of KGUV using various modes of transport excluding buses.

Initially, the total vehicle traffic as well as person traffic was determined for each 15 minute

time interval, which was then added to obtain the total traffic for each analysis period. The

directional distribution for each mode of transport was calculated by dividing the total traffic

in a particular direction by the total traffic in both the directions of transport (inbound and

outbound to KGUV). Later, the car person occupancy was obtained by dividing total number

of people in the car by total number of cars in both directions.

The hourly volumes for each mode of transport were determined by adding the traffic

volumes for four consecutive time intervals during the specified time period. The hourly

volumes observing the highest person movements in the respective time periods were termed

as the peak hour in each time period. The vehicle or person movements occurring during this

time period represented the maximum hourly volumes for each mode of transport. Using a

Cordon count data

Eliminate through traffic

Obtain total traffic volumes

Determine traffic flow attributes

Compare with ITE (USA) trip rates

Compare with RTA (Australian) trip rates

Determine differences in traffic volume

Traffic generation at KGUV


similar rationale, the minimum traffic volumes and the corresponding time period were

determined for all three time periods.

Source: Department of Housing (2008)

Figure 6.2 Direction of through traffic at KGUV

After determination of the maximum hourly traffic volumes, the respective values for AM

and PM peak period were compared with that of peak hourly volumes derived from the

average rates and regression equations provided in ITE (2008) for the United States of



America and the maximum of peak period traffic from the Australian sources (RTA, 2002).

This task pointed out the differences in the traffic volumes between the actual counts and the

standard practice. If the actual counts indicate less traffic than the standard guidelines then

the case study TOD exemplifies the claim of reduced traffic at TODs and the opposite trend

indicates that the TOD principles are not sufficient in reducing the traffic generation at this

particular development. On the other hand, if the amounts of observed and calculated traffic

are similar then the case study TOD does not display any change in traffic due to its stated

qualities. The following section presents the results of the analysis for the shopping centre

counts as well as cordon counts conducted for KGUV.

6.3 Conditions of the survey period

As stated before, KGUV includes educational land uses on its periphery. Hence, it was

necessary to conduct the cordon counts when the university and schools were open.

Considering the school and university semester timetables, the counts were undertaken during

September 2008 on a typical weekday. A typical weekday was chosen as a weekday during

the school term, falling in the middle of the week. Monday and Friday are not considered as

typical weekdays because members of the community tend to take leave on these days more

so than midweek days. More specifically, the Village Centre counts were undertaken on

Thursday for three time periods. For the whole KGUV, the AM peak period counts were

undertaken on Wednesday morning and midday peak and PM peak period counts were

undertaken on Thursday. The cordon counts were not conducted for 24 hours; instead typical

peak periods were identified by observation of traffic at KGUV.

6.4 Total traffic at KGUV

This section presents the overview of the traffic at KGUV for all modes of transport and its

attributes. First the details for the VC counts are presented, followed by the results from the

cordon counts for whole KGUV.

6.4.1 Traffic at the Village Centre

A detailed analysis was conducted for the counts undertaken for the VC. The details of the

time periods are listed in Table 6.1. The maximum hourly person movements were observed

between 8:00am to 9:00am, 11:45am to 12:45pm, and 16:45pm to 17:45pm.



Table 6.1 Time periods for analysis for Village Centre

Details AM peak Midday peak PM peak

Time period 7:15 to 9:30 11:30 to 13:30 15:00 to 18:00

Peak hour 8:00 to 9:00 11:45 to 12:45 16:45 to 17:45

Lowest volume hour 7:15 to 8:15 12:30 to 13:30 15:00 to 16:00

The traffic for VC consists of the traffic generated by the mixed uses, which include

residential apartments, commercial and retail shops. The share of the residential traffic is

significant in the case of the vehicular traffic (specifically car) because there is only one

access point for the car park entry, for the shoppers as well as residents. Pedestrians have

separate access points for entering the residential apartments, and shoppers have other access

points. So there numbers are not included in the count. The results for the total volume,

minimum, and maximum hourly volumes are displayed in Table 6.2.

Table 6.2 Traffic generated at Village Centre by mode on study day

Mode of

transport Traffic volume AM peak Midday peak PM peak

Cars

Total volume 245 246 538

Max hourly volume 117 129 211

Min hourly volume 108 134 140

Motorcycles

Total volume 4 2 6



Bicycles

Total volume 5 0 28



Walk only

(Pedestrians)




Total persons




It can be observed that the VC had very minimal bicycle as well as motorcycle traffic for all

three time periods. The AM peak observed the least pedestrian movements while the midday

peak showed the highest total person movements as well as pedestrian movements. This

indicated that the users of non residential land uses visited the shopping centre during the

midday peak period for brief activities. The car traffic was less in the AM peak and more in

the PM peak, indicating that people used the car for home based or work based shopping trips

not originating from KGUV.

The PM peak experienced the maximum hourly volume for cars (211 cars/hour) and bicycles

(18 bicycles/hour) while the AM peak observed the maximum hourly volume for motorcycles



(6 motorcycles/hour). The midday peak showed the maximum hourly volumes for pedestrians

(799 pedestrians per hour) and subsequently total person movements (955 persons/hour).

Table 6.3 and Table 6.4 present the outcomes for the car occupancy and directional

distribution calculations for cars and other modes of transport respectively. The car

occupancy was almost same for all three time periods; the AM peak had highest car

occupancy rate while the midday peak had highest proportion of single occupant cars at the

VC. The directional distribution for car also indicates close to a 50/50 distribution of traffic

during all three periods, highlighting an association with short stay visits.

Table 6.3 Car occupancy and directional distribution for Village Centre

Time period AM peak Midday peak PM Peak

Car person occupancy 1.24 1.20 1.21

Cars In (%) 53.47 48.78 51.49

Cars Out (%) 46.53 51.22 48.51

The pedestrians entering the VC were predominant in the AM peak and midday peak while a

50/50 distribution was observed for the PM peak. The midday peak was dominated by the

lunch hour activities undertaken by various users of KGUV or facilities located on the edge

of KGUV. The motorcycles also had an even directional distribution except during the PM

peak, where the motorcycles coming in were in large number compared to the other

directional movement, although numbers are too small for this observation to be statistically

significant. The bicycles observed 50/50 directional distribution for PM peak and more

bicycles were going out of VC as compared to the bicycles coming into the VC during AM

peak.

Table 6.4 Directional distribution for Village Centre

Time period Pedestrians Bicycles Motorcycles*

% In % Out % In % Out % In % Out

AM Peak 54.0 46.0 40.0 60.0 50.0 50.0

Midday peak 57.1 42.9 NA NA 50.0 50.0

PM peak 49.6 50.4 50.0 50.0 66.7 33.3

* Note: Small sample sizes

6.4.2 Traffic at whole of KGUV

The analysis of cordon data was performed for three time periods; a specific 60 minute peak

hour and lowest traffic hour were noted for each time period (Table 6.5). The analysis

includes the traffic observed at all cordons to KGUV. The traffic volumes for the whole

KGUV were determined instead of individual land uses. The various land uses at a TOD



interact with each other; hence the traffic generated by the whole TOD should be assessed

instead of traffic generated by individual land uses for studying its traffic generation.

Table 6.5 Time periods for analysis of whole KGUV traffic

Details AM peak Midday peak PM peak

Time period 7:45 to 9:30 11:30 to 1:30 3:00 to 6:00

Peak hour 8:00 to 9:00 12:00 to 1:00 4:15 to 5:15

Lowest volume hour 7:45 to 8:45 12:30 to 1:30 3:00 to 4:00

During the survey KGUV experienced a considerable amount of through traffic on its street

network. Not eliminating through traffic would have provided a false impression of traffic

generated by KGUV. Table 6.6 gives the proportion of through cars for each time period. The

highest proportion of through traffic was observed for midday peak period. The through

traffic was mainly accessing the university campus located on the northern boundary of the

KGUV.

Table 6.6 Through car traffic at KGUV by time of day

Time period Observed volumes Without through traffic % Through traffic

AM peak 2182 1784 18.2

Midday peak 2217 1712 22.8

PM peak 3552 2773 21.9

Table 6.7 Total traffic generated at KGUV by mode on study day

Mode of

transport Traffic volume AM peak Midday peak PM peak

Cars




Motorcycles




Bicycles




Walk only

(Pedestrians)




Total persons




The cordon data analysis determined traffic characteristics of various modes of transport. The

details of hourly traffic are explained in Table 6.7. The total person volumes for whole

KGUV had the major share by pedestrian mode followed by car mode. A comparison of the



total volumes indicates that KGUV had very little bicycle and motorcycle traffic throughout

the three time periods on the survey day. For cars, the maximum hourly volume occurred

during the AM peak, closely followed by the PM peak hour, while the minimum peak hourly

volume occurred during the midday peak period. The PM peak experienced the minimum as

well as maximum hourly flow of pedestrians; it also had the highest hourly volume in

comparison to its counterparts.

The person occupancy for cars shows maximum occupancy later in the day and PM peak and

slightly lower occupancy for the AM peak period (Table 6.8), which is likely reflection of a

greater proportion of commute trips during this early time of day. The car person occupancy

values are very similar to those for the shopping centre. A balanced distribution was observed

for cars except during the midday and evening peaks, while for the AM peak period the

proportion of cars entering KGUV was higher than the cars travelling out of KGUV again

likely reflection of commute car traffic entering in a more concentrated fashion in the

morning.

Table 6.8 Car occupancy and directional distribution at KGUV

Time period AM peak Midday peak PM Peak

Car person occupancy 1.24 1.28 1.28

Cars % In 56.28 50.00 48.25

Cars % Out 43.72 50.00 51.75

The directional distribution for pedestrians, bicycles and motorcycles is shown in Table 6.9.

The highest proportion of inbound movements was observed in the AM peak for all these

modes, which is again likely reflection of a marked inbound commute during this morning

peak period. For pedestrians slightly more entered than exited during the midday peak, likely

the reflection of an outbound lag. Pedestrian activity was relatively evenly split during the

PM peak. Bicycle entry was also predominant during the AM peak, while exiting was

predominant during the PM peak, but interestingly also during the midday peak. Similar to

pedestrians, slightly more motorcycles entered during midday peak and exited during PM

peak possibly due to similar reason.

Table 6.9 Directional distribution at KGUV

Time period Pedestrians Bicycles Motorcycles

% In % Out % In % Out % In % Out

AM Peak 61.1 38.9 60.3 39.7 63.9 36.1

Midday peak 55.7 44.3 35.3 64.7 52.7 47.3

PM peak 49.0 51.0 36.2 63.8 43.8 56.2



6.5 Comparison of peak hourly traffic with published rates

This comparison was undertaken for car trip ends only, as guidelines for other modes of

transport are not available. The observed maximum AM peak and PM peak hourly rates were

compared with the ITE trip generation rates for the United States (ITE, 2008). The Maximum

peak hourly rate amongst all three time periods was compared with the RTA traffic

generation rates for Australia (RTA, 2002).

For comparison of the VC traffic counts alone, the traffic generated by residential apartments

and the shopping centre were considered because they have a common access point for

vehicular traffic, hence it was not possible to distinguish the traffic individually for the

distinct land use. This statement is not applicable for the pedestrian traffic because the

residents had separate access points if they were on foot.

The total trip rates for KGUV estimated using the ITE and RTA rates respectively were

determined by calculating the car trip generation for all individual land uses at KGUV and

adding them together. These totals were compared with that obtained from the cordon count

analysis. The detailed calculations for ITE and RTA comparisons and their outcomes are

presented in Appendix C.

6.5.1 ITE comparison

6.5.1.1 ITE trip rates

The trip generation guidelines (ITE, 2008) provided trip rates for various land uses using an

independent variable. The trip rates were based on mainly the studies conducted in various

parts of the United States. The following points provide some information about ITE trip

rates.

Generally, the sites were surveyed between the 1960s and the 2000s. Some of the data

collected was few decades ago which might represent different traffic generation

characteristics than the current characteristics.

The regression equations are provided when three criteria are met; results for

minimum 4 studies were available, R2 value is greater than 0.5 and number of trips

increase as the value of independent variable increases.

The number of studies conducted for each land use type also varied greatly,

specifically for the land uses used in this case from 4 to 173 (Appendix C). This large

variation in number of studies raises data quality issues when the trip rates were



combined to obtain whole traffic generation in case of TODs. Further, the land uses

which contained fewer studies required caution while using the trip rates.

The sites were surveyed using various data collection techniques, which affects the

quality of the data collected.

No information was provided about the time of year during which the various studies

were conducted. So these trip rates do not account for seasonal variations if any.

The trip rates are based on one independent variable. The trip rates do not consider the

variation in other independent variables or multiple independent variables, which

could affect the vehicular traffic generation.

Although some shortcomings or data quality issues were found for the ITE trip rates; these

rates were used for comparison because they are widely used as guidelines in various parts of

the world. The following section presents the comparison for Village Centre and the whole of

KGUV.

6.5.1.2 Comparison of shopping centre traffic volumes

Table 6.10 shows the comparison of AM peak and PM peak traffic volumes with the ITE trip

generation rates obtained with the help of average trip rate and regression equation provided

in ITE, 2008. For determination of the peak period traffic, the trip rates for the peak period of

the generator were considered because the traffic volumes from peak periods of the Village

Centre were used for comparison.

Table 6.10 ITE comparison for traffic at the Village Centre

Description AM peak PM peak

Actual count 117 211

Volume based on ITE (average trip rate) 525 624

Volume based on ITE (Regression equation) 530 613

Difference with respect to ITE average trip rates -78% -66%

Difference with respect to ITE regression equation -78% -66%

Note: All volumes are in veh/h

From the percentage differences, it can be concluded that the VC generated in excess of 66

percent less traffic than estimated by ITE guidelines. Overall, the shopping centre shows

reduced vehicular traffic for the centrally located shopping centre (VC) at KGUV based on

ITE guidelines. One reason for that was many shoppers accessed the VC on foot as an

intrazonal trip for KGUV, or the surrounding land uses including QUT Kelvin Grove Campus

and the State School.



6.5.1.3 Comparison of total traffic volumes at KGUV

The comparison of the maximum hourly traffic for the AM and PM peak hours at KGUV

respectively is presented with the ITE trip rates in Table 6.11. For calculation of trips

generated under ITE guidelines, the trip rates for the peak hour of the adjacent street were

used. Where the data was not available the rates for the peak hour of the generator were used.

Commonly, the peak hour of the generator was not used because the peak hour traffic of

KGUV as a whole was used for comparison and all land uses at KGUV might not have the

same peak hour for traffic generation as the whole KGUV, this implied use of peak hour of

adjacent traffic was more appropriate. In addition, the regression equations for the AM and

PM peak hour for the high school were not provided so the value obtained by the average

rates were used for determination of total traffic generated by regression equation. A similar

case was observed for the AM peak of the adjacent street traffic for the shopping centre. A

specific category for the student accommodation was not found so the trip rates were

calculated based on the guidelines for similar land use of Senior Adult Housing–Attached.

Table 6.11 ITE comparison for traffic at KGUV

Description AM peak PM peak

Actual count 1078 1051

Volume based on ITE (average trip rate) 1462 1813

Volume based on ITE (Regression equation) 1513 2296

Difference with respect to ITE average trip rates -26% -42%

Difference with respect to ITE regression equation -29% -54%


The ITE trip rate comparison clearly showed that actual traffic generated at the KGUV is less

than that of the ITE guidelines. Reductions of 26 percent and 42 percent were observed when

compared with the average trip rates for the AM and PM peak respectively. The same

reduction increased to 29 to 54 percent when the regression equations were used.

6.5.2 Australian sources (RTA) comparison

6.5.2.1 RTA trip rates

Similar to ITE (2008), the RTA (2002) presents trip rates for various land uses. The trip rates

were derived from Road and Traffic Authority’s (RTA) Land Use Traffic Generation – Data

and Analysis reports. These rates are based on the studies conducted by RTA, New South

Wales (NSW), Australia. These trip rates are widely used in Australia by various government

authorities. Hence, to compare the traffic generation at KGUV with local guidelines RTA

(2002) trip rates were used despite of the following observations.



Generally, RTA (2002) provides average trip rates. In cases where it provides

regression equations, the equations use one or more independent variables to

determine trip rate.

The land uses covered in RTA (2002) were not as extensive as ITE (2008).

Site by site variations from the average values were not considered while determining

the trip rates.

6.5.2.2 Comparison of shopping centre traffic volumes

The peak hour traffic generation rate provided by RTA (2002) varied depending on size of

the shopping centre and the day of the week. The peak hour generation rate for Thursday was

considered for comparison as the VC counts were undertaken on Thursday. The comparison

of PM peak traffic showed that the RTA guidelines overestimated the vehicular (specifically

car) traffic generation by almost 63 percent. The PM peak traffic was considered for

comparison because RTA (2002) provided guidelines for peak period traffic and in the case

of the Village Centre the maximum car traffic was observed in the PM peak when compared

with AM peak and midday peak traffic volumes. This is in line with the general findings

where peak weekday traffic is observed for Thursday evening (RTA, 2002).

Table 6.12 RTA comparison for traffic at the Village Centre

Description PM peak

Actual count 211

Maximum Volume based on RTA guidelines 571

Difference with respect to RTA regression equation -63%


6.5.2.3 Comparison of total traffic volumes at KGUV

The RTA guidelines provided various rates for traffic generation by different land uses. The

maximum rate was used for each land use. The most similar available rate for student

accommodation was considered as that of Aged or Disabled persons. The maximum value

corresponded to the resident funded developments. The RTA (2002) did not provide any trip

rates for the educational land use. Hence, the rate provided by ITE (2008) was used.

Table 6.13 RTA comparison for traffic at the KGUV

Description AM peak

Actual count 1078

Volume based on RTA guidelines 1857

Difference with respect to regression equation -42%




The comparison of the peak hour counts and the RTA trip rates indicated that similar to the

ITE rates, the RTA rates also overestimated traffic generation (by 42 percent) when

compared to the actual measured traffic at KGUV.


The maximum person movement was seen for the PM peak for KGUV as a whole and

midday peak in case of the shopping centre. The Village Centre as well as KGUV as a whole

had significant pedestrian activity, with a higher mode share, more than any other mode of

transport. This indicates that this TOD’s users performed more walking trips and accessed the

development by walking; which accords with Steiner (2008). Similar observations were made

for residents at a traditional neighbourhood (Handy, 1996).

The comparison for the shopping centre showed reduced traffic generation by more than 63

percent when compared with the ITE (2008) and RTA (2002) guidelines for determining the

trip rates. When KGUV as a whole was considered, a reduction of about 42 percent was

observed for peak period traffic (RTA comparison) and on average a reduction of about 27 to

48 percent was observed for the AM and PM peak hours respectively (ITE comparison).

These finding are in line with the results found by Arrington and Sloop (2009) who observed

TOD housing producing 50 percent fewer trips than the conventional development. These

reductions support the presumption that a TOD reduced traffic generation due to its atypical

development characteristics.

The sensitivity analysis of actual traffic counts showed that an increase of 72.3 percent in

actual traffic count matched the traffic obtained by RTA peak period trip rates. While an

increment of 35.6 percent to 40.4 percent was required to match the AM peak traffic volume

by ITE average trip rates and by regression equations respectively. Similarly, an increase of

72.5 percent and 118.5 percent was required to equal the PM peak traffic counts by ITE

average trip rates and by regression equations respectively.

The findings from this analysis also emphasize the need for a special land use category for

TODs, which may need to address the mix, proximity and therefore interaction in

accessibility between land uses. The specifications for traffic generation at TODs should not

only address car traffic but also pedestrian and bicycle traffic as provision of infrastructure

for these facilities including parking facilities depend on the traffic generation. Some research



on mixture of land uses at TODs should be conducted to determine the level of proper land

use mix and proportion of self containment of trips.

6.7 Summary

In summary, an analysis of cordon data indicated that the Village Centre of KGUV as well as

KGUV as a whole, exhibit a decrease in traffic generation in excess of 27 to 48 percent and

42 percent respectively when compared to the ITE (2008) and RTA (2002) rates provided for

homogeneous land uses. This shows that the particular investigated TOD generated less

traffic, likely because of mixed use developments and provision of public transport service. It

is recognised that similar studies at a range of TODs in Australia will be required to

investigate this claim in depth and use those results as a practice or norm while designing

new TODs.

The use of ITE (2008) showed that the manual does not have specific trip rates for student

accommodation. In the case of the RTA manual, no guidelines were found for determination

of trip generation by education land uses as well as student accommodation. TODs should be

assessed as a special land use category and land use mix at a TOD should be assessed for self

containment. Further, no specifications for other means of transport (pedestrians, bicycle, and

motorcycle) were given in either. These aspects ought to be investigated through further

research.

6.8 Chapter close

This chapter provided a detailed overview of traffic characteristics at the TOD, KGUV. This

completes Step III of TOD evaluation which deals with traffic impact determination and

provides input to Step V. The findings from this chapter provide some of the basis for the

conclusions and recommendations in Chapter 11. This chapter contributes to the knowledge

in the area of trip generation at a TOD, which was highlighted as a gap in the literature

review (Chapter 2).


Chapter 7

Characteristics of Kelvin Grove Urban Village users

7.1 Introduction

A Transit Oriented Development (TOD) can be better explained in terms of its travel demand

by identifying characteristics of various trips undertaken by its different users. This can be

explained by analysing demographic and travel characteristics. This chapter presents the

analysis of demographic and travel characteristics of each group of TOD users to better

explain Kelvin Grove Urban Village (KGUV) in terms of its transport activity.

The next section presents details of the analysis performed on responses obtained from the

travel surveys of KGUV users to determine the desired characteristics. The later sections

describe the demographic as well as travel characteristics of each group of KGUV user. First

the characteristics of the shopping trips undertaken by shoppers are explained, followed by

the characteristics of work trips undertaken by employees, education trips undertaken by

students, and trips undertaken by the residents. An interpretation of the characteristics is then

given by comparing characteristics of various groups of KGUV users. A brief summary of

the chapter and chapter close are then offered.

7.2 Determination of users’ characteristics

All responses collected from various KGUV users were entered into the online survey tool

used for travel data collection. These responses were downloaded from the website in

Microsoft Excel format and then the responses in the Microsoft Excel sheet were rearranged

in a convenient format for analysis. The formatted sheets were then analysed using Microsoft

Excel‟s functions to obtain the characteristics of KGUV users. Responses from the travel

surveys for each category of KGUV user were compiled separately. The responses from the

pilot studies were combined with responses obtained from the full scale surveys before

conducting the analysis. This study only conducts preliminary data analysis to determine

users‟ characteristics. A detailed cross tabulation analysis would provide better information

about KGUV users‟ characteristics but due to limited number of datasets, this was not

performed. However, the detailed cross tabulation analysis should be undertaken for a case

study site with larger dataset. The analysis of the travel survey data mainly provided two



types of characteristics; demographic and travel characteristics. A brief description of the

preliminary analysis undertaken is provided in the following subsections.

7.2.1 Demographic characteristics

The demographic characteristics describe the structure or profile of users based on the

personal and household characteristics. The personal characteristics of KGUV users are

explained with the help of gender, age distribution, employment status, frequency of visiting

KGUV, share of valid driver‟s licence holders, and average working hours at KGUV. In

addition to the personal characteristics, the household characteristics are described using

household size (adults and children), number of vehicles in the household and number of

valid driver‟s license holders in the household. These characteristics were directly obtained

from the responses of the travel surveys.

7.2.2 Travel characteristics

To gain an overview of the travel at KGUV, the travel characteristics of various user groups

were determined. The travel characteristics of KGUV users are explained with the help of

mode shares, trip lengths by mode of transport, parking details such as location and parking

fee, and public transport characteristics such as number of legs of the journey, transfer

locations, and access and egress times. The proportions of choice riders and captive riders are

also explained.

The mode shares and mode choices, parking details and public transport characteristics were

obtained directly from the responses. All three modes of public transport, namely public bus,

train and ferry were combined and denoted as a public transport mode. The share of

sustainable modes of transport was also determined by combining the mode shares for public

transport, bicycle and walk only as these modes are termed as more sustainable means of

transport than their other counterparts, specifically car. The trip length for each trip was

calculated with the help of the “home suburbs” specified using “Google Maps”

(www.maps.google.com.au). Trip length was the actual road distance travelled by car from

home suburb centroid to centrally located Village Centre for the employee, students and

shoppers‟ trip. The trip lengths for residents‟ trips were calculated considering the Village

Centre as an origin and the trip end suburb centroid as the destination. The trip length details

were derived from these calculated values.

The following sections provide details of demographic and travel characteristics of each user

group.

http://www.maps.google.com.au/

Characteristics of KGUV users


7.3 Shoppers’ shopping trips at KGUV

The shopping centre (also known as Village Centre) has some street level family owned

businesses; this gives the feeling of a village atmosphere and made the shopping centre more

attractive to shoppers. A data set for 117 respondents to the quick interview shoppers‟ survey

was used for analysis. The responses from the shoppers‟ pilot survey were not included in the

analysis as they represented weekend data.


The demographic characteristics of the shoppers are explained with the help of employment

status and age group distribution. The frequency of shoppers‟ shopping trips to the Village

Centre was also analysed to explain the rate of the recurrence of shopping trips. The

proportion of shoppers belonging to each category of employment status is shown in Table

7.1. It can be seen that more than 60 percent of respondents were students, and others were

residents or visitors coming from the suburbs located in close vicinity to KGUV. The

students of the educational land use on KGUV, and from the school and university campuses

located on the boundary of KGUV constituted a significant proportion of the shoppers.

Table 7.1 Employment status of shoppers at KGUV

Description Proportion of shoppers

Employed 23.9

Student 63.2

Homemaker 3.4

Retired 3.4

Not available 6.0

Note: All values are presented in percentage

Figure 7.1 illustrates the distribution of age groups for shoppers at KGUV. It can be seen that

73 percent of respondents were young adults (between 18 and 45 years) and very few

respondents were above 65 years of age. The proportion of students seemed to have a great

influence on the distribution of age groups.

The distribution of frequency of shopping trips for shoppers at KGUV is shown in Figure 7.2.

A zero frequency of shopping trip includes the respondents who visited KGUV for the first or

second time. A frequency of 2.5 indicates that on an average the shopper visited the shopping

centre two to three times a week. More than 50 percent of respondents visited the shopping

centre once to thrice in a week. Remaining respondents visited the shopping centre more than

three times in a week. This indicates that students use the shopping centre for their

convenience (day to day) shopping. It was also observed that the shoppers who were



residents of KGUV made 3.5 trips a week; likely due to close proximity of the shopping

centre. These trips contributed to internal trips which are mostly made by walking.

Figure 7.1 Distribution of shoppers’ age groups in years

Figure 7.2 Frequency of shopping trips per week

0 - 18

14%

18 - 30

53%

30 - 45

20%

45 - 65

11%

> 65

2%

0

5

10

15

20

0 0.5 1 2 2.5 3 3.5 4 5 6 7

Per

cen

tage o

f sh

op

per

s (%

)

Frequency of shoppers' shopping trips per week




7.3.2.1 Mode shares

Mode share is an important variable to be considered when assessing travel characteristics.

The classified distribution of mode shares for shoppers at KGUV is shown in Figure 7.3. The

public transport trips include trips made by public bus and train; no ferry trips were observed

as there is no ferry terminal nearby KGUV. Only 27 percent of shopping trips were

undertaken by car, while very few shoppers used bicycle and motorcycle. It may be

postulated that the mixed land uses promoted walking trips (43 percent). The mode share

distribution indicates that more than 70 percent of the shoppers travelled by one of the more

sustainable modes of transport, including public transport, walking and cycling. This is a

good indication of success of this TOD in terms of sustainable transport mode share.

Figure 7.3 Mode shares for shoppers’ at KGUV

7.3.2.2 Trip lengths

The trip lengths for each shopping trip were calculated and the average trip lengths are

tabulated in Table 7.2. The overall average trip length for shoppers at KGUV was calculated

to be 7.6km. The minimum trip length was theoretically 0km for internal trips and maximum

trip length was 94.5km. The internal trips originated at KGUV and terminated at the Village

Centre so a 0km trip length was assigned for these trips. About 17 percent of trips were

internal trips and 83 percent trips were external or generated from the non residential land

Car

27%

Public

transport

24%

Walk only

43%

Bicycle

4%

Motorcycle

2%



uses. From the trip length values it was observed that most trips were from suburbs located in

close proximity to KGUV.

Table 7.2 Trip lengths by mode of transport for shoppers at KGUV

Mode of transport Trip length (km)

Minimum Average Maximum

Car 0.9 9.4 69.4

Public transport 0.9 15.7 94.5

Walk only 0.0 1.0 5.1

Bicycle 0.0 2.0 4.1

Motorcycle 1.9 3.7 5.5

Combined (Overall) 0.0 6.9 94.5

7.3.2.3 Time of day

The shopping activity was distributed all over the day. Shoppers visited the shopping centre

mainly between 7.30am to 9am, 11.30am to 1pm and 3.30pm to 6pm. The shopping activity

was very much dependant on the working hours of students and employees at KGUV.

7.4 Employees’ work trips at KGUV

KGUV has a university campus extension and a research centre which contains employment

for professionals (professional employees), and also has employees working at the centrally

located shopping centre (retail shop employees). A data set of 125 responses obtained from

the internet based survey was used for analysing professional employees and their travel. The

personal interviews carried out for retail shop employees collected travel data of 39

employees. This dataset was used to determine the characteristics of the retail shop

employees. These two datasets incorporated responses from the pilot surveys.


The demographic characteristics of employees are explained with the help of personal

characteristics such as gender, employment status, proportion of licence holders, and

frequency of visiting KGUV and the household characteristics are explained with the help of

household size, vehicle ownership, and number of valid driver‟s licence holders in the

household.

7.4.1.1 Personal characteristics

Table 7.3 lists the personal characteristics of KGUV employees. The proportion of female

respondents was higher than that of male respondents for both employee groups. The retail

shops provided an employment opportunity to students studying at KGUV or at the adjacent



university. These work trips undertaken by students were most likely combined with their

educational trip, hence can be counted as intra-zonal trips. The proportion of driver‟s licence

holders was high for both categories of employees, but particularly so for the professional

employees.

Table 7.3 Personal characteristics of employees’ at KGUV

Characteristic Professional employees Retail shop employees

Male 37.5 38.5

Female 62.5 61.5

Valid driver‟s licence holders 92.4 84.6

Employed full time 62.3 53.8

Employed part time 11.5 7.7

Employed part time & student full time* 10.7 25.6

Self employed – 12.8

Research students full time* 14.8 –

Employed full time & student part time 0.8 –

*Research students are treated as employees


The distribution of age groups (in Figure 7.4) shows that the retail shop employees have more

young workers. The difference in age groups can be related to the type of work as the retail

shop employees perform mostly hospitality or customer service and professional employees

do research and education oriented work.

Figure 7.4 Distribution of age group for employees’ at KGUV

0

10

20

30

40

50

60

70

0-18 18-30 30-45 45-65

Per

cen

tag

e o

f em

plo

yees

(%

)

Age group (years)





Figure 7.5 compares the frequency of employees‟ work trips per week. Almost 31 percent of

retail shop employees worked on weekends while professional employees worked typically

five days per week.

Figure 7.5 Frequency of work trips at KGUV

7.4.1.2 Household characteristics

The variation in the household size for the professional employee households and the retail

shop employee households is presented in Figure 7.6. The size of a retail shop employee

household varied from one to eight with an average size of 3.4 persons in the household. The

average household size for the professional employee‟s household was 2.6 (lesser than its

counterpart) with a minimum of one and maximum of six. As the household size increased

the proportion of professional employee households decreased. With an increase in the size of

household, the proportion of retail shop employee households did not vary greatly.

Table 7.4 shows an overview of the vehicle and bicycle ownership along with the number of

valid driver‟s licence holders in an employee‟s household. Although the retail shop

employees had higher proportion of households with no car; a higher proportion of them

possessed more cars per household, hence they had higher number of valid driver‟s licence

holders in their households. The majority of households in both employee groups did not own

0

10

20

30

40

50

60

1 or 2 3 4 5 > 5

Per

cen

tag

e o

f em

plo

yees

(%

)

Frequency of work trip per week





a motorcycle. The households of professional employees had higher bicycle ownership than

the retail shop employees.

Figure 7.6 Household size distribution at employees’ households

Table 7.4 Vehicle ownership and licence availability at employees’ households

Parameter Quantity Professional employees Retail shop employees

Cars

0 4.2 20.5

1 55.9 20.5

2 23.7 38.5

> = 3 16.1 20.5

Bicycles

0 32.4 51.3

1 22.5 20.5

2 17.6 17.9

>= 3 27.5 10.3

Motorcycles

0 86.9 94.9

1 10.7 2.6

2 2.4 2.6

Valid driver‟s licence

holders

0 1.7 5.4

1 22.9 5.4

2 53.4 48.6

>=3 22.0 40.5


0

5

10

15

20

25

30

35

40

45

1 2 3 4 5 >= 6

Per

cen

tag

e o

f em

plo

yee

ho

use

ho

lds

(%)

Household size






7.4.2.1 Mode shares

The mode share distribution for the professional employees at KGUV is presented in Figure

7.7. The professional employees travelled by many modes of transport with 50 percent by

car, about one third by public transport, and almost 20 percent by walking or cycling. The

professional employees travelled typically during peak hours when public transport provision

is most frequent, which might be a reason for its higher public transport mode share. Cycling

was also a high mode share for this group, which is attributed to good quality access and trip-

end facilities (such as bikeways, showers, cycle lockers, etc). The greater bicycle ownership

also contributed to a high mode share. Most of the professional employees had another choice

for making their work trips; about less than a third professional employees were captive

riders. Overall, about 50 percent of professional employees travelled by the more sustainable

modes of transport.

Figure 7.7 Mode shares for professional employees

Figure 7.8 shows the distributions of mode shares for work travel for employees working at

the retail shops at KGUV. In contrast to the professional employees, the retail shop

employees travelled by only three modes of transport; car, public transport and walk only,

with car comprising 60 percent of mode share. This higher mode share for car travel is

attributed to odd (late night or early morning) working hours and less frequent public

transport or no public transport during off peak times (see Muley et al. (2007)) and

Car

50%

Public

transport

28%

Walk only

10%

Bicycle

10%

Other

2%



availability of free parking spaces. This infers that a lower proportion of retail shop

employees (about 40 percent) travelled by sustainable modes as compared to the professional

employees. Almost 50 percent of the retail shop employees had another choice for their work

travel while others did not have any other choice than their chosen mode.

Figure 7.8 Mode shares for retail shop employees


The maximum, minimum and average trip lengths for employees at KGUV are listed in Table

7.5 and Table 7.6. Though the average trip length for both types of employees is similar, the

maximum trip length varies almost two fold. The longer trip lengths for professional

employees are attributed to the presence of specialised educational facilities (including a

university extension comprising a research centre).

Table 7.5 Trip lengths by mode of transport for professional employees at KGUV

Mode of transport Trip lengths (km)


Car 0.9 14.0 44.3


Walk only 0.9 3.0 6.5

Bicycle 2.9 5.8 8.7

Other 3.8 5.2 6.5

Combined (overall) 0.9 12.7 93.8

Car

59%Public

transport

20%

Walk only

21%



Table 7.6 Trip lengths by mode of transport for retail shop employees at KGUV



Car 0.9 15.8 53.1


Walk only 0.0 0.9 2.6


7.4.2.3 Time of day

The working hours for professional employees were mostly between 8am and 6pm. These

employees sometimes worked in the evenings depending upon the class times. The average of

working hours for professional employees was 8.4 hours. For the retails shop employees, the

working hours varied greatly from early in the morning (5am) to late in the evening

(9.30pm). The average of working hours was 8.2 hours.

7.4.2.4 Parking characteristics

The majority of the professional employees who drove parked their car at their workplace.

The rest of the employees first searched for a free parking space on street. If not available

then they parked at paid parking. Around 50 percent of car drivers paid less than $10 per day

as a parking fee and others parked for free.

Mostly, the retail shop employees who drove parked their car at their workplace and they had

free parking available. The availability of the free parking was a strong factor in choosing car

as a mode of travel.

7.4.2.5 Public transport characteristics

Table 7.7 provides the public transport trip details for employees at KGUV. None of the

employees performed three legged journey by public transport. The majority of employees

who took public transport had two legs to their trip (77 percent and 75 percent). Roma street

public bus / train station and the Cultural Centre Busway Station were the most popular

transfer locations. Mostly, the employees walked for 5 to 10 minutes to reach the bus stop.

The employees walked for about 10 minutes to reach to their workplace. These distances are

standard walking distances for premium services (TRB, 2003 and Queensland Transport,

1999). The employees who walked more than 15 minutes to reach KGUV accessed the site

from the Roma Street train station located 1.9km from KGUV, on the western edge of the

Brisbane Central Business District.



Table 7.7 Public transport trip details for employee trips at KGUV

Description Specification Professional employees Retail shop employees

No of legs 1 22.9 25.0

2 77.1 75.0

Access time

< 5 minutes 28.6 25.0

5 – 10 minutes 45.7 62.5

10 – 15 minutes 8.6 12.5

> 15 minutes 17.1 0.0

Egress time

< 5 minutes 35.3 62.5

5 – 10 minutes 32.4 37.5

10 – 15 minutes 11.8 0.0

> 15 minutes 7.0 0.0


7.5 Students’ education trips at KGUV

KGUV has a university campus extension and a specialised government sponsored high

school for students. The data collected from both the users (school students and university

students) was used for analysis. The university students‟ dataset contained 89 responses and

the school students‟ dataset had 28 responses, which were used to explain the characteristics

of students at KGUV. The characteristics of both the groups were determined separately and

then compared with the other group. There were no pilot surveys conducted hence no

responses available from the pilot surveys in case of students surveys.


The students‟ demography was described using the same parameters used for the explaining

the demographic characteristics of employees in previous section (Section 7.4.1).


Table 7.8 indicates the gender distribution, employment status distribution and the proportion

of valid driver‟s licence holders for school students and the university students. It can be seen

that for both student groups the proportion of female student respondents is almost double

and triple for school students and university students respectively. Only few university

students were undertaking their studies as part time but others were full time students. More

than 60 percent of students had some casual or part time employment in addition to their full

time university commitments. Most of the school students were at school full time and few

students undertook part time employment. As the school students were below 18 years of age,

they did not possess an open driver‟s licence. But a significant proportion of university

students possessed a valid driver‟s licence.



Table 7.8 Personal characteristics of students at KGUV

Characteristic School students University students

Male 32.1 21.3

Female 67.9 78.7

Full time students 85.7 30.7

Full time students & employed part time 14.3 63.6

Part time students NA 5.7

Valid driver‟s licence holders NA 78.7


Figure 7.9 compares the distribution of age groups for the students at KGUV. All the school

students were in the age group of 0 to 18 years. Almost 85 percent of the university students

were in the age group of 18 years to 30 years. This high proportion of young students is

related to the type of university precinct (Creative Industries).

Figure 7.9 Distribution of age group for students at KGUV

The comparison of frequency of education trip (Figure 7.10) indicates that the school students

visited school more consistently, that is five times a week. While the frequency of trip varies

for university students, a major proportion of students visit university three or four times a

week. This may be due to the diverse arrangement of university educational commitments.

0

20

40

60

80

100

0-18 18-30 30-45 45-65 >65

Per

cen

tag

e o

f st

ud

ents

(%

)

Age group (years)

School students

University students



Figure 7.10 Frequency of education trips at KGUV


The household size variation for school and university students‟ households is shown in the

Figure 7.11. The average household size for school students was four persons with a

minimum of two and maximum of six persons in a household. Similarly, for university

students, the average household size was 3.6 persons but with a minimum of one and

maximum of six persons in the household. Two university students were living in a shared

accommodation, which had a household size greater than six. These responses were omitted

before plotting the graph as these were classed as outliers.

Table 7.9 lists the vehicle ownership and the proportion of the driver‟s licence holders in each

category for school students and university students‟ households. A significant proportion of

households possessed two or more than two cars in both the cases. More than 25 percent and

35 percent of school students‟ and university students‟ households respectively did not own

any bicycle. The 43.5 percent of school students‟ households possessed three or more than

three bicycles; this might be because these households use bicycles for recreational purposes

but not for commute purposes. Only small proportion of students‟ households owned a

motorcycle. In case of school students, the proportion of two driver‟s licence holders was

highest as compared to other groups. For the university students, the proportion of three or

more than three driver‟s licence holders was highest.

0

20

40

60

80

100

1 or 2 3 4 5 > 5

Per

cen

tag

e o

f st

ud

ents

(%

)

Frequency of education trip per week

School students

University students



Figure 7.11 Household size distribution at students’ households

Table 7.9 Vehicle ownership and licence availability at students’ households

Parameter Quantity School students University students

Cars

0 0.0 9.2

1 11.1 25.3

2 51.9 33.3

> = 3 37.0 32.2

Bicycles

0 26.1 36.8

1 13.0 32.9

2 17.4 15.8

>= 3 43.5 14.5

Motorcycles 0 87.0 93.2

1 13.0 6.8


holders

0 – 2.2

1 10.7 10.1

2 57.1 25.8

>=3 32.1 61.8



The students‟ education travel is explained with the help of mode share, and trip lengths. The

average working hours, and time of day of travel are also noted in the following subsections.

7.5.2.1 Mode shares

The mode shares for school students are presented in Figure 7.12. The school students

travelled by only two modes of transport; public transport and car. Almost 86 percent of

students used public transport for their trip to school and remaining 14 percent students were

0

5

10

15

20

25

30

35

40

1 2 3 4 5 6

Per

cen

tage

of

stu

den

ts h

ou

seh

old

s

(%)

Household size

School students

University Students



dropped at school by car. The share of public transport represented the high share of use of

sustainable modes of transport. This student group did not have valid driver‟s licences hence

by and large had no access to car as a driver. Due to this, this group of KGUV users can be

termed as „captive riders‟. The responses to the mode choice questions indicated that 69.2

percent of respondents were choice riders and 30.8 percent of them were captive riders who

did not have any other option to travel to KGUV. There were no walk only and bicycle trips

for school students; these two modes were not considered any further as an alternative option

for journey to school. (It is noted that QACI is an elite school, to which students will travel

from across Brisbane.)

Figure 7.12 Mode shares for school students

Figure 7.13 shows the mode shares for university students. Similar to retail shop employees,

the university students travelled by only three modes of transport; walk only, car or public

transport. Similar to school students, the university students exhibited more use of the more

sustainable modes of transport (84 percent). Similar to school students, no bicycle trip was

recorded for university students. Almost two-thirds of students travelled to university by

public transport and only 16 percent arrived by private car. The higher public transport mode

share may be attributed to the student having no driver‟s licence or to the cost involved in

using a car. In the case of university students, 53.9 percent students had another option for

travelling to KGUV and 46.1 percent students did not have any other option to perform their

education trip.

Car

14%

Public

transport

86%



Figure 7.13 Mode shares for university students


Table 7.10 and Table 7.11 tabulate the details of trip lengths for school students and

university students‟ education trip at KGUV respectively. The minimum trip length for both

groups of students was the same (0.9km). However, the maximum trip lengths varied greatly,

with for university students nearly twice of that for school students. In spite of having a lower

value of maximum trip length, school students showed slightly higher value of average trip

length compared to the university students.

Table 7.10 Trip lengths by mode of transport for school students at KGUV



Car 0.9 14.0 38.4



Table 7.11 Trip lengths by mode of transport for university students at KGUV



Car 3.6 16.7 42.6


Walk only 0.9 1.6 3.9


Car

16%

Public

transport

75%

Walk only

9%



7.5.2.3 Time of day

The school students attended the school typically between 8am and 4pm. The average stay of

the students at their school was 7.3 hours. Some university students were engaged in the

evening classes hence they stayed until late. On an average, a student was at university for

7.8 hours.

7.5.2.4 Parking characteristics

The school students at the university were dropped off at the school, hence their vehicle was

not parked, and hence they did not pay any parking fee. The university students who travelled

by car parked their vehicle on street or at the university parking facility. Most of them paid a

parking fee of less than ten dollars a day if they could not find a free car parking space.

7.5.2.5 Public transport characteristics

The most popular public transport transfer locations for school students were Roma Street

public bus / train station or Cultural Centre Busway Station, while the King George Square

Busway Station and Roma Street public bus / train station were the main transfer stations for

university students. Some students had suburban train stations as their transfer locations.

Table 7.12 indicates the distribution of access and egress time and number of legs for the

journey. The university students observed highest egress time of more than 15 minutes when

they accessed KGUV through the Roma Street station. Although few three leg journeys were

observed; again majority was two legs for public transport trips suggesting interchange

locations and their facilities are important.

Table 7.12 Public transport trip details for student trips at KGUV

Description Specification School students University students

No of legs

1 12.5 9.7

2 75.0 79.0

3 12.5 11.3

Access time

< 5 minutes 29.2 41.3

5 – 10 minutes 41.7 33.3

10 – 15 minutes 25.0 15.9

> 15 minutes 4.2 9.5

Egress time

< 5 minutes 20.8 53.7

5 – 10 minutes 70.8 13.0

10 – 15 minutes 8.3 14.8

> 15 minutes – 18.5




7.6 Residents’ trips at KGUV

KGUV has two types of residents; non student residents, who lived in the privately owned or

rented apartments, and student residents who lived in the managed student accommodation. It

should be noted that there were some university students renting the apartments in the Village

Centre. A dataset of 34 responses for non student residents‟ and 51 responses for student

residents‟ survey was used for analysis. These responses included the responses from the

pilot surveys as well as from the different phases of travel surveys.


To assess the residents‟ demographic characteristics, the same parameters were used as per

previous user groups. In addition to those, the household characteristics are described by the

type of dwelling and the number of bedrooms in that dwelling.


Table 7.13 lists the demographic characteristics of the residents at KGUV. The proportion of

female respondents was a little higher than male respondents in the case of student residents,

while for non student residents a balanced distribution was obtained. The majority of

residents in both user groups were employed full time or studying full time. Only six percent

of the non student residents were unemployed, and these were mostly retired persons. A

significant proportion of non student residents (88.2 percent) possessed a valid driver‟s

licence. The variation in the distribution of the age groups is shown in Figure 7.14. Around

90 percent of student residents were young adults (age group 18 years to 30 years), which is

obvious. The non student residents contained some older residents when compared with

student residents.

Table 7.13 Personal characteristics of residents’ at KGUV

Characteristic Non student residents Student residents

Male 50.0 43.1

Female 50.0 56.9

Valid driver‟s licence holders 88.2 68.6

Employed full time 38.2 –

Students full time & employed part time 8.8 23.5

Employed full time & student part time 2.9 –

Employed full time & student full time 2.9 –

Students full time 35.3 76.5

Employed part time 5.9 –

Retired / unemployed 5.9 –




Figure 7.14 Distribution of age group for residents’ at KGUV


The apartments at KGUV were typically one or two bedroom while the students were living

in shared units having 1 to 6 bedrooms. Figure 7.15 shows the variation in the number of

bedrooms in a household. The variation in the household size is displayed in Figure 7.16.

There was a drop in one person households compared to the number of bedrooms. This was

largely because of a couple or family of two persons living in a one bedroom apartment.

Figure 7.15 Distribution of number of bedrooms in residents’ household

0

10

20

30

40

50

60

70

80

90

0-18 18-30 30-45 45-65

Per

cen

tag

e o

f re

sid

ents

(%

)

Age group (years)


Student residents

0

20

40

60

80

1 2 3 > 3

Per

cen

tag

e o

f re

sid

ent

hou

seh

old

s (%

)

Number of bedrooms in residents' household


Student residents



Figure 7.16 Distribution of household size at residents’ household

Table 7.14 Vehicle ownership and licence availability at residents’ households

Parameter Quantity Non student residents Student residents

Cars

0 20.6 41.2

1 55.9 47.1

2 20.6 7.8

> = 3 2.9 3.9

Bicycles

0 82.4 88.2

1 14.7 9.8

2 2.9 2.0

Motorcycles 0 100.0 98.0

1 – 2.0


holders

0 5.9 7.0

1 52.9 14.0

2 38.2 34.9

3 – 20.9

> 3 2.9 23.3


Vehicle ownership is an important household characteristic describing private vehicle

dependency of residents. The distribution of vehicle ownership for KGUV residents is given

in Table 7.14. KGUV residents had lower car ownership compared to a high driver‟s licence

possession. Around one fifth of households did not have a car. A large proportion of KGUV

households did not possess either a bicycle or a motorcycle, indicating very low vehicle

0

5

10

15

20

25

30

35

40

45

50

1 2 3 > 3

Per

cen

tag

e o

f re

sid

ents

ho

use

ho

lds

(%)

Household size


Student residents



ownership, which can result in low private vehicle usage. During the interviews, some of the

respondents pointed out that they did not have a car or bicycle because they did not need it.

Most of the student residents left their car at home. They believed that attractions were

sufficiently close to each other that they did not require a vehicle for transport. This is argued

by many transport professionals to be one of the biggest advantages of TODs.


The travel diaries for a complete day were collected from the residents‟ travel surveys. The

complete trip details were determined for the trip making characteristics. Later, the first trip

of the day was chosen for evaluating the travel characteristics as this trip was originating

from KGUV and was of most interest for the purpose of this study. The parking

characteristics and public transport characteristics are not listed for residents travel; instead

an insight into the internalisation of trips is provided.

In the case of non student residents only one person did not travel on the assigned travel day,

while in case of student residents three students did not travel on the assigned travel day. A

retired person who did not travel on the assigned day mostly travelled on the pension day and

sometimes for visiting a doctor. The students did not travel because they did not have any

academic engagements on that day. A set of 33 and 48 travel diaries were analysed for non

student residents and student residents respectively.

7.6.2.1 Trip characteristics

Table 7.15 lists the minimum, average and maximum number of trips made by the residents

at KGUV. The minimum number of trips was 0 as the respondents did not perform any trip

on the assigned travel day. A non student resident made more trips than a student resident,

partly because of the various activities required to perform by a household (pick up and drop

off formed a major share of this). On a typical weekday, the residents mostly travelled for

work or education during the day and in addition to a return trip home for shopping or

recreation during the evening. The evening shopping trip was mostly on foot.

Table 7.15 Number of trips for residents at KGUV

Description Number of trips per person


Non student residents 0.0 2.9 6.0

Student residents 0.0 2.4 4.0



7.6.2.2 Mode shares

Mode share was assigned based on the first trip of the day. The travel mode share for

residents living in apartments at KGUV is shown in Figure 7.17. Similar to the university

students and retail shop employees, the residents at KGUV typically travelled by either car,

public transport or walk only. No train, ferry or bicycle trips were reported; the reason for no

train or ferry trips was that KGUV does not have a train station or a ferry terminal within

easily walkable distance. The mode shares indicated that 63 percent of the residents travelled

by the more sustainable modes of transport. More than 50 percent (54.5 percent) of the

residents had another mode choice for performing this trip.

Figure 7.17 Mode shares for non student residents at KGUV

Figure 7.18 shows a pie chart showing the distribution of mode shares for the students living

in student accommodation for their first trip of the day. It can be noted that these residents

travelled by only three modes of transport; car, walk only and public transport (specifically

public bus as no train or ferry trips were reported) similar to non student residents. No

resident used a bicycle or taxi for arriving at their desired destination. The reason for there

being no bicycle trips is postulated to be due to the limited bicycle connections available to

more remote areas, heavy traffic around the area, and the hilly terrain of the area.

Public bus was the most preferred mode with a share of 48 percent. Most of the residents

(More than 90 percent) used the modes of transport labelled as sustainable (which include

Car

37%

Public

transport

36%

Walk only

27%



walk only and public transport) and only eight percent residents‟ used car as their mode of

transport. When asked about mode choices, around 77 percent of residents did not perceive

that they had any choice other than their chosen mode of transport; we therefore consider

them to be captive users.

Figure 7.18 Mode shares for student residents at KGUV


The minimum, average and maximum trip lengths for residents at KGUV were calculated for

the first trip listed in the travel diary. The details are listed in Table 7.16 and Table 7.17 for

non student residents and student residents respectively. Theoretically, the minimum trip

length was zero. This is a trip undertaken by a student to the university or a shopping trip

undertaken by walking. For non student residents, the maximum trip length was observed as

94.4km for a non student resident by car and the overall average trip length was 6.1km.

Similarly, the maximum trip length for student residents was observed by car as 40.7km and

overall average trip length was 3.1km. The overall average trip lengths indicate that the trips

originating from KGUV were distributed to a relatively smaller area specifically for student

residents. The residents used car for accessing destinations located away from KGUV. This

also points out that the mixed uses can help in containing the trips over a smaller area.

Car

8%

Public transport

48%

Walk only

44%



Table 7.16 Trip lengths by mode of transport for non student residents at KGUV



Car 0.9 12.8 94.4


Walk only 0.0 0.9 4.8


Table 7.17 Trip lengths by mode of transport for student residents at KGUV



Car 2.2 15.8 40.7


Walk only 0.0 0.2 1.6


7.6.2.4 Internalisation of trips

The amount of internal trips is an important factor of a TOD travel. In order to gain an

understanding about the various activities performed by the residents at the KGUV, the

different activities were noted with their frequency. Table 7.18 represents the percentage of

activities and their frequency for non student residents. Table 7.19 lists the details for the

student residents‟ trips. Shopping at the Village Centre was the most popular activity for the

residents at KGUV. Other activities undertaken by residents were visiting cafes and farmers

market on Saturday. Also some students used the newly opened activity centre. Further, the

more than 60 percent of student residents at KGUV never made an education trip to KGUV

because they were studying at the campuses located outside KGUV specifically the QUT

campus located on the edge of KGUV or in the Brisbane CBD.

Table 7.18 Activity frequency for non student residents at KGUV

Activity Shopping Visiting parks Education trip Travel to work

Once a week 15.6 16.7 3.4 6.9

Twice a week 15.6 3.3 10.3 –

Thrice a week 28.1 – – 6.9

Four times a week 12.5 – 6.9 –

Five times a week 9.4 3.3 13.8 17.2

More than five

times a week

15.6 6.7 3.4 –

Never – 30.0 31.0 20.7

Once a fortnight – 6.7 3.4 –

Once a month – 26.7 – –

Not Applicable 3.1 6.7 27.6 48.3




Table 7.19 Activity frequency for student residents at KGUV

Activity Shopping Visiting parks Education trip Travel to work

Once a week 14.0 14.0 2.0 2.0

Twice a week 22.0 2.0 4.0 –

Thrice a week 30.0 2.0 6.0 –

Four times a week 8.0 – 2.0 –

Five times a week 10.0 – 6.0 10.0

More than five

times a week

14.0 4.0 4.0 4.0

Never 2.0 48.0 62.0 48.0

Once a fortnight – 12.0 – 10.0

Once a month – 16.0 14.0 10.0

Not Applicable – 2.0 – 16.0


7.7 Comparison of KGUV users’ characteristics

The characteristics of KGUV users can be further explored by comparing the characteristics

of each user group. For this purpose, the characteristics of the main user groups were

determined by combining the characteristics of the subgroups in that user group. For

example, the characteristics of professional employees and retail shop employees were

combined to obtain the characteristics of the employee user group. Table 7.20 and Table 7.21

represent a comparison of personal and household characteristics respectively. A comparison

of travel characteristics is shown in Table 7.22.

7.7.1 Comparison of demographic characteristics

7.7.1.1 Comparison of personal characteristics

The comparison of the personal characteristics of KGUV users (Table 7.20) indicates most of

the KGUV users were between 18 to 30 years of age. The proportion of female respondents

was more for users of Non Residential Land Use (NRLU) while the Residential Land Use

(RLU) exhibited balanced distribution. Most of the employees in all user groups were

employed full time or were full time students. Amongst all user groups, employees at KGUV

visited the development more frequently than any other NRLU user group. Further, the

highest proportion of employees possessed a valid driver‟s licence, followed by residents and

students; with students having higher proportion of users who do not possess a valid driver‟s

licence.



Table 7.20 Comparison of personal characteristics of KGUV users’

Characteristic Description Shoppers Employees Students Residents

Age group

0 to 18 years 14.0 0.6 30.8 7.1

18 to 30 years 53.0 45.3 64.1 74.1

30 to 45 years 20.0 37.1 2.6 11.8

45 to 65 years 11.0 17.0 1.7 7.1

65 years and above 2.0 0.0 0.9 0.0

Gender Female NA 62.3 76.1 54.1

Male NA 37.7 23.9 45.9

Employment

status

Employed full time

or Student full time

87.1*

71.4 47.4 78.0

Self employed 3.1 NA 0.0

Employed part time

and student full time 14.3 48.3 18.3

Employed part time /

student part time 10.6 4.3 0.0

Employed full time

and student part time 0.6 0.0 1.2

Homemaker / retired 6.8 0.0 NA 2.4

Other 6.0 0.0 0.0 1.2

Frequency of

trip (per week)

Rarely 6.0 NA NA NA

Once or twice 38.5 7.9 6.8 NA

Two to three times 4.3 – – NA

Three times 17.9 7.3 30.8 NA

Three to four times 0.9 – – NA

Four times 12.8 14.6 20.5 NA

Five times 9.4 51.2 33.3 NA

More than five times 10.3 18.9 8.5 NA

% valid driver‟s licence holders NA 90.5 59.8 76.5

Note: All values are given in percentage

*indicates combined value for employees and students

7.7.1.2 Comparison of household characteristics

The comparison of household characteristics was made for employees, students and residents

at KGUV. Shoppers were not considered in this comparison as shopper‟s household data was

not available (Table 7.21). The household size comparison shows that the student households

had higher household size than the employee and residents households. Most of the KGUV

employees‟ and residents‟ households owned one car but the students‟ households possessed

more cars (two or more). All three user households had higher proportion of households with

no bicycles and motorcycles. The student households possessed higher proportion of driver‟s

licence holders which supports the higher car ownership.



Table 7.21 Comparison of household characteristics of KGUV users’

Characteristic Description Employees Students Residents

Household size

1 11.5 1.8 24.3

2 38.9 13.2 41.9

3 24.8 33.3 18.9

4 14.0 25.4 1.4

5 7.6 18.4 13.5

> = 6 3.2 7.9 0.0

Car ownership

0 8.3 7.0 32.9

1 47.1 21.9 50.6

2 27.4 37.7 12.9

> = 3 17.2 33.3 3.5

Bicycle ownership

0 37.6 34.3 85.9

1 22.0 28.3 11.8

2 17.7 16.2 2.4

> = 3 22.7 21.2 0.0

Motorcycle

ownership

0 89.4 91.8 98.8

1 8.1 8.2 1.2

2 2.4 0.0 0.0

Driver‟s licence

holders

0 2.6 1.7 6.5

1 18.7 10.3 31.2

2 52.3 33.3 36.4

> = 3 26.5 54.7 26.0

Note: Values are given in percentage and trip lengths are in km

Indicates minimum values for each user group

Indicates maximum values for each user group

7.7.2 Comparison of travel characteristics

The comparison of travel characteristics of the KGUV users (Table 7.22) shows that

employees at KGUV have highest car mode share while students have the lowest car mode

share. This might be due to high access to car for travelling to KGUV for employees at

KGUV. Although the KGUV students have lower maximum trip length, they had highest

average trip length by when travelled by car. Comparing the public transport trip details, the

students used more public transport for undertaking education trips while least proportion of

shoppers used public transport for their shopping trips. The residents of KGUV have lowest

value of maximum trip length and average trip length when travelled by public transport.

Similar to car trips, KGUV students had maximum average trip length for travel by public

transport. When walk only trips were compared shoppers showed highest mode share because

of internalisation and students indicated lowest walk trip mode share might be because of

higher trip lengths. The residents exhibited lowest average trip length for walk trips while

employees indicated the highest average trip length. When the bicycle mode share was



compared two groups of KGUV users; shoppers and employees used bicycle with employees

showing maximum values of bicycle mode share and shoppers having the minimum values.

Table 7.22 Comparison of travel characteristics of KGUV users’

Mode Characteristic Shoppers Employees Students Residents

Car

Mode share 27.4 52.4 15.4 19.8

Minimum tip length 0.9 0.9 0.9 0.9

Maximum trip length 69.4 53.1 42.6 94.4

Average trip length 9.4 14.6 16.0 13.6

Public transport

Mode share 23.9 26.2 77.8 43.2

Minimum trip length 0.9 1.3 0.9 2.1



Walk only

Mode share 42.7 12.8 6.8 37

Minimum trip length 0 0 0.9 0



Bicycle

Mode share 4.3 7.3 0 0

Minimum trip length 0 2.9 – –

Maximum trip length 4.1 8.7 – –

Average trip length 2.0 5.8 – –

Other

Mode share 1.7 1.2 0 0

Minimum trip length 1.9 3.8 – –

Maximum trip length 5.5 6.5 – –

Average trip length 3.7 5.2 – –

Note: Mode shares are given in percentage and trip lengths are in km

Indicates minimum values for all user groups

Indicates maximum values for all user groups

7.8 Transport issues related to TOD from users’ perspective

The respondents were asked to rate the public transport and highlight any transport related

issues. A detailed overview of the results of perception analysis is given in Appendix D.

Some key issues about the transport facilities at KGUV for all respondents were noted as

below.

Respondents placed a strong emphasis on the frequency and reliability of public

transport service to afford a mode shift from car to public transport. The travel time

difference and the absence of a direct public transport link1 were also pointed out as

the main reasons for using personalised modes of transport. This indicated that for a

TOD to be successful from a transport point of view, a good quality direct public

1 Brisbane has a hub and spoke public transport network with most services intersecting at the CBD. One often

needs to change service, particularly for public buses, to access a destination on the other side of the CBD. The

imposition of a seat change has been reported to make public transport less attractive.



transport service is required from various destinations, not only from the CBD. Some

respondents preferred train over public bus because of its on time performance or

reliability.

As this area lies near the CBD, some respondents suggested having a “loop service”

(with minimum or no cost) running at a 15 minute interval from the CBD to KGUV

which also connects the nearby Roma street railway station to make KGUV more

attractive. (It is noted that, after this survey, TransLink implemented a high frequency

bus service, Route 66, along the busway system between Kelvin Grove busway

station to the east of KGUV, through the CBD via Roma Street railway station and on

to the inner southeast suburb of Woolloongabba. A service called 933 between Roma

Street train / public bus station and QUT KG Busway Station now augments this in

the peak.)

A professional employee also suggested having strictly enforced parking restrictions

on the local streets in KGUV with increased parking cost at the work place and

incentives for employees who travel by sustainable modes of transport from the

organisations to promote the more sustainable modes of transport.

The responses also indicated that on time performance and actual arrival of the public

bus at scheduled time or some information which would provide them an idea about

the public bus arrival were highlighted as important factors. This was linked to the

overcrowding and passenger discontent, as some passengers were unable to board the

first available public bus due to high demand. This issue was prominent in peak times

when visitors travelled to KGUV and also back home after finishing their activity at

KGUV. Access to the real time mobile information and linked information with the

nearby major train stations or busway stations was suggested as a solution for this.

A perceived lack of public transport in outlying suburbs was also a concern and

reason for using personalised modes of transport.

Some respondents perceived the cost of the public transport as expensive. Most of the

students were quite happy with the cost as they travelled on a concession fare which is

half the price of a full fare.

Few respondents suggested having route maps at the bus stop and signage on the

public buses to guide users. There are route maps in the city area but these may need

to be extended to the outer suburbs or the places of public interest as this may aid

users to provide more information about the bus routes.



Some respondents also highlighted that they have infrequent or inadequate services

during off peak times and due to work commitments it was not always possible to

travel during peak times. The frequency of services during off peak hours to some

areas demanded more attention. This indicates need for public transport service to or

from various destinations for complete day not only in peak periods.

The accessibility to the Roma Street station and Normanby Busway Station was stated

as a major concern. The respondents were ready to walk for about 2km provided they

have good infrastructure for walking as this will also yield them health benefits. This

implies that connections to Normanby Busway and Roma Street station could be

enhanced.

The respondents demanded better signage and better sun / rain protection at public

bus stops. Public bus stops should not only look aesthetically good but also need to

perform better in terms of form and function. Decent shelter and accurate timetables

at the public transport stops were the amenities requested by a user.

The respondents requested bicycle lanes and protected bicycle parking facilities at the

KGUV in addition to their workplace. Wider footpaths were suggested for

accommodating bicycles to avoid collision. Demand for bicycle facilities indicated

that a good bicycle path is needed not only in KGUV but also from home to the

workplace. Riding a bicycle on Brisbane‟s roads was described as “scary”.

A pedestrian friendly sequencing of signals (or scramble crossing) and pedestrian

priority at intersections were suggested for improving the pedestrian facilities at

KGUV.

The presence and quality of bicycle ways in Brisbane was stated by a professional

employee as generally poor. It was stated that presence of a strong network of bicycle

ways throughout inner city suburbs and the CBD may encourage more people to

consider cycling to / from work (which maintains fitness and is sustainable).

7.9 Interpretation of TOD users’ characteristics

A comparison of various user characteristics was undertaken to determine the similarities or

differences. The proportion of female respondents was high for employees and students while

for residents a balanced distribution was obtained. The household size for student households

was more than the employee and residents‟ households. When the vehicle ownership was

compared for the subgroups it was found that the retails shop employees‟ and school



students‟ households had higher vehicle ownership as compared to their counterparts. Very

few households in each user group possessed a motorcycle. The comparison of bicycle

ownership showed that the households of non TOD resident user groups (employees‟ and

students‟ households) owned more bicycles than the KGUV resident households.

The mode share comparison suggested that a significant proportion of each user group

travelled by sustainable modes of transport; the lowest proportion was for employees (46

percent) and highest was for residents (80 percent). The comparison for average trip lengths

for various modes of transport showed that the KGUV residents used car for covering longer

trip lengths while the other user groups (students, employees and shoppers) preferred public

transport for longer trips as opposed to use of a car.

7.10 Summary

The preliminary analysis of travel data provided the various user characteristics as the

outcome. The mode share plots for KGUV users demonstrated encouraging mode share

values for the more sustainable modes of transport. Fewer cycling trips were reported by

KGUV users, possibly due to relatively treacherous cycling conditions on certain Brisbane

arterial roads and hilly topography around KGUV. Despite some steep terrain, walk trips

were considerable due to the complementary land uses placed together and attractive walk

paths constructed at KGUV. The share of public transport was significant, which is an

encouraging result. The comparison of overall average trip length for all user groups

indicated that of all users the KGUV residents had the lowest average trip lengths. The

students travelled farther to access the specialised education facilities.

The household characteristics of residents of KGUV indicated that KGUV residents had

lower car ownership compared to high valid driver‟s licence possession. The KGUV

residents also reported low bicycle and motorcycle ownership. This might be due to the

proximity of mixed land uses and availability of good public transport service. These

characteristics of KGUV users need to be investigated further by performing comparative

analysis and travel demand analysis for evaluating travel for the users of KGUV.

7.11 Chapter close

This chapter provided an insight into demographic as well as travel characteristics of various

KGUV users, which partly completes Step IV of TOD evaluation. These characteristics will

be used as a reference for further data analysis. Although the characteristics for each



subgroup of KGUV were explained separately in this chapter, considering the number of

responses and diversity of the user groups, further analysis was conducted for the combined

datasets. The combined trips undertaken by shoppers, employees and students were used for

further analysis. The next chapter presents the comparative analysis of KGUV users‟

characteristics with regional and similarly located suburban characteristics. The personal and

travel characteristics obtained in this chapter will also be used for the travel demand analysis

of KGUV users in Chapter 9 and Chapter 10 respectively.


Chapter 8

Comparative analysis of Kelvin Grove Urban Village’s users’

characteristics

8.1 Introduction

To assess the travel impacts of Kelvin Grove Urban Village (KGUV) with respect to other

non Transit Oriented Developments (TOD), the various characteristics need to be compared

with non TOD (or conventional) developments to determine the variations. This chapter

explains the differences in characteristics of KGUV users with respect to the Brisbane

regional and similarly located suburban users’ characteristics. The following section

describes details of these comparisons and later sections present the comparison of

characteristics of each group of KGUV users with their counterparts. The comparison of

travel characteristics for all KGUV user groups is demonstrated by comparing the mode

shares and trip lengths with regional and suburban values and a comparison of KGUV

residents’ household characteristics, with regional and suburban household characteristics is

also presented. Finally, an interpretation of the results from the comparative study, summary

and chapter close are provided.

8.2 Basis for comparison

The comparative analysis was mainly conducted using the travel characteristics of KGUV

users; namely mode shares and average trip lengths. In addition, the residents’ household

characteristics were compared. Basically, this comparison needs data from non TOD

developments. In this case, data from a specific non TOD development with similar size was

not available; hence the data availability for various suburbs located at similar distance from

the CBD was checked but due to insufficient datasets, the comparisons were undertaken by

grouping data from inner city suburbs. Firstly, a regional comparison was made to determine

the variation of KGUV travel with respect to Brisbane as a whole (known as the Brisbane

Statistical Division (BSD)). Secondly, suburban comparisons were conducted. A group of

suburbs located within similarly close proximity to Brisbane CBD as KGUV were considered

for comparison. The suburban comparison consisted of two groups of suburbs, Brisbane Inner

North Suburbs (BINS) and Brisbane Inner South Suburbs (BISS). BINS and BISS are typical

cases of suburbs located within close vicinity of an Australian capital city CBD. Note that



BISS travellers often need to cross the Brisbane River, which has a limited number of

crossings or bridges and is a natural barrier.

The travel data for various groups of KGUV users was obtained from preliminary analysis

(Chapter 7). Specifically four major trip types were considered, namely shopping trips (by

shoppers), work trips (by employees), education trips (by students), and residents’ trips. The

travel data for BSD and its inner suburbs (BINS and BISS) was obtained from the South East

Queensland Travel Surveys (SEQTS) conducted during 2006–08 (SEQTS, 2009).

Specifically, the SEQTS survey for Brisbane statistical division undertaken in 2006, known

as BSEQTS06, was used for analysis in this chapter. The BSEQTS06 collected travel data

and household data using a self–completion questionnaire which was hand-delivered to, and

hand-collected from, the survey households (SEQTS, 2008).

The BSEQTS06 obtained 4178 travel diaries from 1564 households and this data was

analysed using Microsoft Access®. The travel characteristics for shopping trips, work trips

and education trips were determined by considering the trips terminating at these suburbs,

while the characteristics for residents’ were obtained by considering their location of the

household. The mode shares were obtained by considering the “LINKMODE” assigned to

each trip. LINKMODE was the priority mode assigned for each trip after considering distinct

modes for each leg of the journey. The average trip lengths were determined by considering

the “QT_CUMDIST” for each trip. QT_CUMDIST was the total network distance travelled

for each trip noted in km. In case of BISS and BSD work trips, few interstate trips having

very large trip lengths were observed, so to avoid a false impression, these values were

termed as outliers and were removed from analysis. The absolute values were used instead of

the weighted values from SEQTS.

The following sections present comparison for each trip type at KGUV, BSD, BINS and

BISS as explained before. It should be noted that the public transport mode share and average

trip length comprised of trips undertaken by public bus, train, ferry and school bus.

8.3 Comparison of shoppers’ shopping trips

8.3.1 Mode share comparison

When comparing users’ shopping trip mode shares to KGUV with those of BSD, BINS, and

BISS, Table 8.1 shows that the car mode share for shoppers visiting KGUV was one third of

BSD and BISS, and one half of BINS. Significantly higher public transport and walking

Comparative analysis of KGUV’s users’ characteristics


mode shares for shopping trips were observed at KGUV compared to BSD, BINS, and BISS.

A bicycle mode share of 4.3 percent was observed for KGUV shopping trips compared to 0.6

percent for BSD, zero for BINS and 1.7 percent for BISS. Similar to BISS shoppers, none of

the KGUV shopping trips was by taxi as opposed to 0.2 percent for BSD and 0.4 percent for

BINS.

Overall, shoppers visiting KGUV used more sustainable modes of transport (70.9 percent)

compared to BSD shoppers (14.7 percent), BINS shoppers (42.2 percent) and BISS shoppers

(11.1 percent). KGUV has a small scale shopping centre, which caters for the daily needs of

its inhabitants and apparently some of the population of the immediate surrounding area. The

higher walking and cycling mode shares may be attributable to the various mixed land uses

placed together, which in theory induces more intrazonal trips, which are mostly undertaken

by walking.

Table 8.1 Mode share comparison for shopping trips

Mode of

transport KGUV

Brisbane

statistical division

Brisbane inner

north suburbs

Brisbane inner

south suburbs

Public transport 23.9 4.7 12.5 3.4

Walk only 42.7 9.4 29.7 6.0

Bicycle 4.3 0.6 0.0 1.7

Subtotal labelled

as “sustainable” 70.9 14.7 42.2 11.1

Car 27.4 84.2 54.7 88.0

Taxi 0.0 0.2 0.4 0.0

Other 1.7 0.9 2.6 0.9

Note: Mode share values are in percentage

8.3.2 Trip length comparison

Table 8.2 lists comparison of average trip lengths by modes of transport for KGUV, BSD,

BINS and BISS shoppers. The overall average trip length was KGUV and BSD (6.9km and

7.0km) shoppers were same, this was slightly higher than average trip length for BINS

shoppers (6.4km) and considerably higher than BISS shoppers (4.8km). The KGUV shoppers

exhibited higher average trip length by car when compared to its counterparts. While the

average trip length by public transport and bicycle did not show much variation. Noticeably,

KGUV shoppers on an average walked longer (1km) than BSD (0.9km), BINS (0.6km), and

BISS (0.5km) shoppers.



Table 8.2 Comparison of average trip lengths for shopping trips

Mode of

transport KGUV

Brisbane


Brisbane inner

north suburbs

Brisbane inner

south suburbs


Walk only 1.0 0.9 0.6 0.5

Bicycle 2.0 2.6 – 1.9

Car 9.4 7.3 7.2 4.6

Taxi – 5.7 11.7 –

Other 0.0 4.2 5.2 6.0

Overall 6.9 7.0 6.4 4.8

Note: Average trip lengths are in km

8.4 Comparison of employees’ work trips


The mode shares for work trips by employees at KGUV were compared with those of BSD

and the inner suburbs of Brisbane; BINS and BISS (Table 8.3). The comparison shows a

considerably smaller car mode share for KGUV employees, 29 percent and 22.8 percent less

for KGUV employees compared to BSD and BISS employees respectively. Their mode share

is a more modest 5.2 percent less when compared to BINS employees. When public transport

trips are compared, there is not much difference for BINS and KGUV employees; however

these two values are both considerably higher than the BSD and BISS public transport mode

share of 10.2 percent and 13.6 percent respectively. A similar trend exists for walk only mode

shares. The employees of BINS and KGUV likely exhibited higher mode shares than BSD

for public transport because of good quality services to the CBD and surrounding suburbs,

particularly during peak times. A high mode share for bicycle for KGUV employees may be

attributable to good quality access and end-of-trip facilities (such as showers and cycle

lockers). The higher car mode share in case of BISS as compared to BINS may be attributed

to the strict parking restrictions imposed for inner city and northern suburbs.

Overall, KGUV employees demonstrated more sustainable travel choices (46.3 percent)

compared to BSD employees (17.6 percent) and BISS employees (23.4 percent) and a little

improvement in use of sustainable modes of transport than BINS employees (41.6 percent)

for their work trips.



Table 8.3 Mode share comparison for work trips

Mode of

transport KGUV

Brisbane


Brisbane inner

north suburbs

Brisbane inner

south suburbs


Walk only 12.8 6.2 14.5 8.0

Bicycle 7.3 1.2 1.5 1.8

Subtotal labelled

as “sustainable” 46.3 17.6 41.6 23.4

Car 52.4 81.4 57.6 75.2

Taxi 0.0 0.3 0.6 0.9

Other 1.2 0.6 0.2 0.4



The comparison of average trip lengths for KGUV employees work trips (Table 8.4) indicate

that overall KGUV had lower average trip length (12.3km) than BSD (14.7km), BINS

(20.7km), and BISS (18.0km) employees. The average trip length for car trips was same for

KGUV and BSD employees which was about 40 percent lesser than average trip lengths for

BINS and BISS employees. Similarly, KGUV employees have almost 40 percent, 50 percent,

and 35 percent less average trip length when travelled by public transport compared with

BSD, BINS, and BISS employees respectively. Despite lower average trip length by public

transport and car, KGUV employees walked longer to reach their workplace than BSD, BINS

and BISS employees. Similarly, KGUV employees had higher average trip length for work

trips by bicycle when compared with BSD and BINS employee trips, but lesser when

compared with BISS employee bicycle trips.

Table 8.4 Comparison of average trip lengths for work trips

Mode of

transport KGUV

Brisbane


Brisbane inner

north suburbs

Brisbane inner

south suburbs


Walk only 2.2 0.7 0.6 0.7

Bicycle 5.8 5.2 4.6 6.7

Car 14.6 14.6 22.4 19.3

Taxi – 6.4 3.8 3.1

Other – 13.3 18.3 21.5

Overall 12.3 14.7 20.7 18.0




8.5 Comparison of students’ education trips


When comparing user students’ education trip mode shares to KGUV with those of BSD,

BINS and BISS, Table 8.5 shows that only 15.4 percent of KGUV students used a car for

their education trip compared to 58.4 percent for BSD, 40 percent for BINS, and 54.6 for

BISS. These equate to shares 52.9 percent, 28.3 percent and 43.5 percent higher shares for

public transport than those of BSD, BINS and BISS respectively. This may be attributed to

the high quality of public transport facilities for access to KGUV.

Although the public transport mode share is particularly high, KGUV students had a lower

walking mode share for education trips compared to its counterparts, perhaps due to the hilly

terrain and highly trafficked roadways, such as Kelvin Grove Arterial Road on the western

flank. Further, and perhaps for similar reasons, no bicycle mode shares for education trips to

KGUV were reported, in this case consistent with BINS. Although dedicated bicycle facilities

are located nearby, including a segregated facility running alongside the M3 motorway to the

south east, students may be deterred from cycling due to highly trafficked crossings such as

Kelvin Grove Arterial Road, as well as missing links. Long travel distances may also be

playing a part in these results. The bicycle mode share for BSD students indicate that the

students studying in education institutions located in outer suburbs used bicycle for their trip

to school.

Table 8.5 Mode share comparison for education trips

Mode of

transport KGUV

Brisbane statistical

division

Brisbane inner

north suburbs

Brisbane inner

south suburbs


Walk only 6.8 13.8 9.5 11.1

Bicycle 0.0 2.9 0.0 0.0

Subtotal labelled

as “sustainable” 84.6 41.6 59.0 45.4

Car 15.4 58.4 40 54.6

Taxi 0.0 0.1 0.0 0.0

Other 0.0 0.0 1.1 0.0


When mode shares for all sustainable modes of transport were combined and compared,

KGUV user students showed a significant difference in mode shares for their education trips

at 84.6 percent, compared to 41.6 percent, 59 percent and 45.4 percent for BSD, BINS, and

BISS respectively.




When comparing the average trip lengths by mode of transport for education trips at KGUV,

BSD, BINS and BISS students (Table 8.6) shows that KGUV students have a higher overall

average trip length (17.2km) than BSD (7.0km), BINS (14.7km), and BISS (7.0km) students.

This was almost 2.5 times higher than BSD and BISS students and 17 percent more than

BINS students. The average trip length for car trips was more than two fold and three fold for

BINS, and BSD and BISS students respectively. Similar to shopping and work trips, students

at KGUV walked longer distances than students of its counterparts. The students at KGUV

travelled on an average 19km by public transport, which was more than BSD (13.3km) and

BISS (12.4km) student trips and less than BINS (23.8km) education trips.

Table 8.6 Comparison of average trip lengths for education trips

Mode of

transport KGUV

Brisbane


Brisbane inner

north suburbs

Brisbane inner

south suburbs


Walk only 1.6 1.1 0.9 0.9

Bicycle – 2.9 – –

Car 16.0 5.9 7.1 4.9

Taxi – 18.1 – –

Other – – 5.0 –

Overall 17.2 7.0 14.7 7.0


8.6 Comparison of residents’ trips

8.6.1 Household characteristics

Table 8.7 represents a comparison of KGUV residents’ household characteristics with the

residents of BSD, BINS and BISS. It should be noted that characteristics of KGUV student

residents were not included in the analysis as it is argued that the characteristic of students’

accommodation are different than of typical conventional households due to difference in

demographic characteristics. So in order to obtain a fair comparison, the data for this user

group was not included. The comparison of household size suggests that KGUV residents

have slightly lower household size than residents of BINS and BISS and considerably smaller

household size than BSD households. This is likely due to the higher number of single and

double bedroom apartments at KGUV.

KGUV has a lesser average number of bedrooms per household, indicating these apartments

are highly attractive to small family households; typically to young adults and families with

no children. However, a TOD should also cater for large families as they tend to drive more



for children’s activities, and living in such an environment may help to reduce the number of

vehicle trips by a household considerably.

Table 8.7 Comparison of residents’ household characteristics

Household

characteristics KGUV

Brisbane

statistical

division

Brisbane inner

north suburbs

Brisbane inner

south suburbs

Average household size 2.0 3.4 2.3 2.2

Average of number of

bedrooms 1.4 3.5 2.8 2.5

Average motor vehicles

per household 1.1 2.0 1.5 1.3

Average bicycles per

household 0.2 1.8 1.2 1.0

Note: These are Non Student Residents’ (NSRs) characteristics only

Transport professionals often postulate that TOD residents have low vehicle ownership; this

is true in case of KGUV, with 1.1 vehicles per household in comparison to 2.0, 1.5, and 1.3

vehicles per household for BSD, BINS and BISS respectively. This finding was similar to the

findings of previous studies (Deakin et al., 2004 and Giuliano and Dargay, 2006). When

asked about vehicle ownership while conducting the surveys; the respondents indicated that

they do not own a car because they do not need one. One respondent said, “Everything is so

close here. I can access everything by walk and I like it very much”. The low vehicle

ownership requires less parking infrastructure, and as such this aspect of TOD needs to be

studied separately for appropriate parking arrangements, as does making these developments

more pedestrian friendly.

Transport professionals also often postulate that TOD residents have higher walking and

cycling trip rates, which means that they ought to have higher bicycle ownership. However,

KGUV residents exhibit an opposite trend, having only 0.2 bicycles per household in

comparison to 1.8, 1.2 and 1.0 bicycles per household by BSD, BINS and BISS households.

This may attributable to the reasons cited above.

8.6.2 Trip characteristics

When the average trip characteristics for KGUV residents and BSD, BINS and BISS

residents were compared (Table 8.8), it was found that on an average KGUV residents

undertook fewer trips on the given travel day (2.6 trips/person) compared to BSD (3.1

trips/person), BINS (3.6 trips/person) and BISS (3.5 trips/person) residents. The minimum

trips were the same (zero) as there were few respondents in each category who did not travel

on the assigned travel day. KGUV residents made a quarter and one third fewer trips when



compared with BSD and BISS, and BINS residents respectively. Previous research suggested

that TOD residents make fewer car trips and more walk trips. Car trips are stated to be

replaced by walk trips, but making same number of trips as for residents living in

conventional development (Sun et al., 1998). However KGUV residents made less number of

trips exhibiting an opposite trend.

Table 8.8 Comparison of residents’ average trips per person

Description KGUV Brisbane


Brisbane inner

north suburbs

Brisbane inner

south suburbs

Minimum trips by

a person 0 0 0 0

Maximum trips by

a person 6 23 18 23

Average trips per

person 2.6 3.1 3.6 3.5


The mode shares for residents’ first trip of the day were compared to each other. Table 8.9

lists the details of mode shares for KGUV, BSD, BINS and BISS residents. Only 20 percent

of KGUV residents used the car, compared to around 70 percent for other inner suburban

residents. KGUV residents used public transport at around twice and four fold the rate

compared to other inner suburban residents and BSD residents respectively. Similarly,

KGUV residents walked at around four fold the rate compared to other inner suburban

residents. A small proportion of the BSD, BINS and BISS residents used bicycle for their

travel while none of the KGUV resident used bicycle. Noteworthy, taxi was not a popular

mode for the first trip of the day in any group of Brisbane or inner suburban residents.

Table 8.9 Mode share comparison for residents’ first trip of the day

Mode of

transport KGUV

Brisbane


Brisbane inner

north suburbs

Brisbane inner

south suburbs


Walk only 37.0 7.4 7.9 8.8

Bicycle 0.0 1.5 0.9 2.4

Subtotal labelled

as “sustainable” 80.2 20.6 26.4 30.4

Car 19.8 78.5 71.8 68.2

Taxi 0.0 0.2 0.9 0.4

Other 0.0 0.7 0.9 0.8


Overall, 80.2 percent of KGUV residents used sustainable modes of transport for making

their first trip of the day. On the other hand, only 20.6 percent, 26.4 percent and 30.4 percent



residents of BSD, BINS and BISS used such modes for making their first trip of the day

respectively. The relatively high mode share for KGUV residents was due to the fact that

many residents were living in KGUV and working / studying not too far from their place of

residence. Further, they have high quality public transport facilities close by and services to /

from various destinations. This outcome was in line with the previous findings from past

studies (Cervero and Gorham, 1995; Cervero, 1996; and Hess and Ong, 2002). This

demonstrates that KGUV has a greater tendency towards using modes of transport labelled as

sustainable; supporting the hypothesis that TOD development will help to reduce transport

congestion in urban areas over traditional development of the same size.


When the average trip lengths for KGUV, BSD, BINS and BISS residents’ first trip were

compared (Table 8.10), it was found that overall KGUV residents travelled within close

vicinity as compared to BSD, BINS and BISS residents. The average trip length by public

transport for KGUV residents was one sixth and one half of the corresponding trip lengths for

BSD, and BINS and BISS residents respectively. Similarly, the residents at KGUV had very

short walking trips (0.4km) compared to BSD (1.2km), BINS (1.5km), and BISS (1.0km)

residents. The residents of BINS had lower bicycle average trip length than BSD and BISS

residents possibly due to the hilly conditions noted before. The average trip lengths by car

suggests a similar value for KGUV and BSD residents, this was lower than BINS residents’

and higher than BISS residents’ average car trip lengths. The higher trip length for car trips

was in contradiction to the claim of TOD residents’ shorter car trips made by (Sun et al.,

1998 and Steiner, 1998). It should be noted that local factors will specifically impact the trip

lengths.

Table 8.10 Comparison of average trip lengths for residents’ first trip of the day

Mode of

transport KGUV

Brisbane


Brisbane inner

north suburbs

Brisbane inner

south suburbs


Walk only 0.4 1.2 1.5 1.0

Bicycle – 4.6 2.2 5.5

Car 13.6 12.3 16.1 7.7

Taxi – 13.5 8.2 5.8

Other – 9.2 2.9 2.9

Overall 4.4 12.8 13.0 7.1





The results of a comparative analysis indicated that TODs are transport efficient in practice

when compared with the other non TOD areas. On an average, KGUV residents made fewer

trips than BSD, BISS and BINS residents. Due to presence of mixed land uses the KGUV

residents possessed fewer motor vehicles (1.1) per household in comparison to its other

counterparts (BSD at 2.0, BINS at 1.5 and BISS at 1.3). This reduces the dependence on

motor vehicle and ultimately restrains its usage.

The comparison shows that all user groups of KGUV used more sustainable modes of

transport compared to those of BSD, BINS and BISS (Figure 8.1). This outcome provides

evidence to support the presumption that TOD is a more sustainable form of development

than traditional development, with respect to travel modes of users. This high proportion of

sustainable mode usage resulted in more environment friendly practices supporting

transportation claims of TODs and making them more transportation efficient. A good public

transport connection to or from various destinations at KGUV attracted higher public

transport mode shares and subsequently reduced car usage. Clustering of activities was

correlated with more walking trips; specifically for shopping and recreational trips.

Figure 8.1 Comparison of sustainable transport mode share

In general, the mode shares of BISS and BSD were very similar when the trips by Non

Residential Land Use (NRLU) users were considered while mode shares for BINS and

0

10

20

30

40

50

60

70

80

90

Shopping trips Work trips Education trips Residents’ first

trip

Su

sta

inab

le t

ran

sport

mod

e sh

are

(%

)

Trip type

KGUV

BSD

BINS

BISS



KGUV were similar with KGUV showing more sustainable travel choices. This improvement

in the mode shares can be attributed to the atypical TOD development characteristics mainly

the quality of public transport service. BINS has better connections to CBD than BISS or

BSD overall.

The comparison of overall average trip length for KGUV, BSD, BINS and BISS users

(Figure 8.2) indicates that KGUV shoppers and students have greater average trip lengths and

KGUV employees and residents have lower average trip lengths as compared to its

counterparts in BSD, BINS and BISS. The higher trip length can be attributed to the

specialised education facilities at KGUV and the shopping trips attracted because of these

specialised facilities. The reduction in average trip length for residents shows that the TOD

residents travel fewer kilometres, this was similar to the finding determined by McCormack

et al. (2001).

Figure 8.2 Comparison of overall average trip length

When the average trip lengths by car and public transport were compared for all trips

undertaken by NRLU users, it was found that for all user groups at KGUV as well as BSD,

BINS and BISS, the average trip lengths by public transport were higher than average trip

lengths by car indicating that public transport users travelled longer distances. When a similar

comparison was made for residents, those of KGUV and BINS used car for travelling farther

0

5

10

15

20

25

Shopping trips Work trips Education trips Residents’ first trip

Over

all

aver

age

trip

len

gth

(k

m)

Trip type

KGUV

BSD

BINS

BISS



and used public transport for accessing destinations located closer to their residence. This

may be because of availability of good public transport service to the destinations located

closer to their homes and strict parking conditions in those areas. On the contrary, the

residents of BSD and BISS exhibited an opposite trend than KGUV and BINS residents. This

might be possibly due to the similar factors as of KGUV and BINS residents but exhibiting

an opposite trend.

8.8 Summary

The comparison of residents’ household characteristics showed that KGUV residents have

low vehicle and bicycle ownership as compared to BSD, BINS and BISS residents. The

outcomes of comparative analysis indicate that KGUV users use more sustainable modes of

transport than BSD, BINS and BISS users. The overall average trip length for shopping and

education trips at KGUV shows higher overall trip lengths than its counterparts. On the other

hand, shorter trips were observed for KGUV residents and employees when compared with

their respective counterparts. These results provide a means of comparing transport

performance of KGUV with respect to non TOD conventional developments, indicating the

travel impacts of one kind of TOD. The outcomes from this comparative analysis should,

however, be applied with caution while planning future TODs, as each TOD has its own

location, geographic, demographic, socio–economic and built form characteristics. No two

TODs are exactly alike. The finding that sustainable travel choices are made at this site,

however, supports the notion of development of future TODs from KGUV users’ perspective.

Although this study supports TOD planning, travel data for more case studies at various

scales and characteristics should be examined. In the Australian context, this task will

become easier as more TODs are developed and are able to be studied by transport

researchers using standardised techniques.

8.9 Chapter close

This chapter provided a comparison of travel characteristics of all TOD users, and household

characteristics of residents, with corresponding regional and suburban characteristics. The

results of this chapter provided the travel impacts of KGUV; these outcomes will provide

some information influencing the conclusions in Chapter 11. This partly completes the fourth

step of TOD evaluation. To study this outcome in detail and make use of it for further

planning purposes, it is also important to assess the travel demand at KGUV, which is next



sub step in TOD evaluation. Chapter 9 and Chapter 10 undertake this task and explore the

travel at KGUV by conducting the travel demand analysis.


Chapter 9

Kelvin Grove Urban Village users’ travel demand analysis

9.1 Introduction

After assessing the travel demand of a Transit Oriented Development (TOD) with respect to

other non TOD areas, the next step is to carry out travel demand analysis for detailed travel

demand investigation. The issue of mode choice is probably the single most important

element in transport planning and policy making. It affects the general efficiency of urban

transport (Ortuzar and Willumsen, 1994). Hence, an investigation into the travel modes was

conducted for all user groups at Kelvin Grove Urban Village (KGUV) to determine the

influence of various characteristics on the travel modes of TOD users. For mode choice

analysis, two types of models can be developed, models based on aggregate demand and

models based on disaggregate demand. For analysing the travel modes of KGUV users,

disaggregate demand was used as this will enable more realistic models to be developed

(Ortuzar and Willumsen, 1990). This chapter presents the analysis and results for the same.

The following section provides the background and details used for analysing KGUV users‟

travel modes. Later the equations for travel mode determination are provided for individual

TOD user groups; namely for shoppers‟ shopping trips, employees‟ work trips, students‟

education trips and residents‟ first trip of the day. An interpretation of the analysis is provided

in the next section. At the end, a summary and chapter close are presented.

9.2 Analysis background

The investigation into travel modes was made using a logistic regression technique. Logistic

regression is a widely used technique for performing statistical analysis of survey data. This

consists of fitting a linear logistic model to an observed data set in order to measure the

relationship between the outcome variable and one or more explanatory variables. Logistic

regression was used as it is more flexible than other techniques. Also, it does not assume

distributions of the predictor variables, hence the variables do not have to be normally

distributed, linearly related or of equal variance within each group. The predictors in the

logistic regression analysis can be a mix of continuous, discrete or dichotomous variables.

Unlike multiple regression analysis, logistic regression cannot produce negative predicted

probabilities (Tabachnick and Fidell, 2007). Further, logistic regression can be used to



determine the effect of size of the independent variables on the dependent variable; to rank

the relative importance of independents; to assess interaction effects; and to understand the

impact of covariate control variables. The impact of predictor variables is usually explained

in terms of odds ratios.

Binary logistic regression was used to predict the probability of a KGUV user using

sustainable modes of transport. Firstly KGUV users‟ travel modes were divided into two

categories; sustainable transport modes (walking, cycling and public transport) and the

private car. Separate logistic regression analysis was carried out for each trip purpose to

determine the probability of a KGUV user choosing a sustainable transport mode. The

general form of logistic regression is shown in Equation 9.1 and Equation 9.2.

Logistic regression function,

𝑝(𝑦) =𝑒𝑧

1 + 𝑒𝑧 Equation 9.1

The linear regression equation,

𝑍 = 𝑎0 + 𝑎1 × 𝑥1 + 𝑎2 × 𝑥2 + 𝑎3 × 𝑥3 + ⋯ Equation 9.2

where,

𝑝(𝑦) = Probability of predicting y, in this case use of a sustainable

transport mode, using independent variables via Z

𝑎0 = Regression constant

𝑎1,𝑎2,… = Regression coefficients

𝑥1, 𝑥2,… = Independent variables

𝑍 = Linear function of independent variables

The independent variables were obtained from the travel surveys. The travel survey mainly

gathered information on respondents‟ trip characteristics, household characteristics, personal

characteristics and perception about KGUV and transport at KGUV. The perception based

variables were discarded from travel demand analysis and analysed separately (Appendix D)

as they are mainly used for qualitative analysis, whereas travel demand analysis is

quantitative analysis. Amongst the household characteristics, vehicle availability is the most

important and governing variable. Initially to represent vehicle availability for an individual

trip, the variable named car availability was used. To represent the trip characteristics trip

KGUV users’ travel demand analysis


length by car, LOS for public transport and travel time saving of car over public transport

were used, which incorporate the characteristics of public as well as private modes of

transport.

From the observations, it was noted that, in this case, personal characteristics were the

governing factors in determining the travel mode. Hence all variables related to personal

characteristics were considered in the analysis. This inclusion of personal variables also fills a

knowledge gap, as previously no study was found demonstrating the influence of personal

characteristics on choice of travel mode at the disaggregate level.

Based on the above arguments, the probability of choosing sustainable modes of transport

was calculated as a function of personal characteristics, such as age group, occupation,

employment status, type of student, frequency of shopping trip and gender, and transit

characteristics such as (Quality or) Level of Service (LOS), trip length, and travel time saving

(TTS). The LOS for frequency of public transport service was included in the analysis

because it was rated as the most important parameter for public transport service by KGUV

users (Appendix D). Each variable was coded using the coding system specified in Table 9.1.

Actual values of the continuous variables including trip length and travel time saving were

used rather than coded values.

The personal characteristics of the users and the trip length values were obtained from the

travel survey analysis illustrated in Chapter 7. The travel time saving was the time saving if a

user travelled by public transport. It was calculated as the difference between travel time by

car and the travel time by public transport. The car travel time was the time required to travel

the specified trip length by car (Given by Google Maps, www.maps.google.com.au). The

public transport travel time was calculated using the journey planner on the TransLink

Transit Authority‟s public transport information website www.translink.com.au. The origin

location was the home suburb and for convenience the destination was given as the QUT

Kelvin Grove Busway Station for trips by visitor groups and vice versa for the residents‟

trips. The option of fastest travel time for a weekday was considered in travel time

calculations. An additional travel time of 10 minutes and 15 minutes was added to the

estimated travel time for car and public transport respectively to take into account the access

times. In the TransLink journey planner, the options of arrive before start time and leave

after start time were selected for visitors and residents of KGUV respectively. The travel

time saving was calculated by subtracting public transport travel time from car travel time.





The travel time saving for the cases in which no public transport option was available was

coded as missing values.

Table 9.1 Details of coding system for travel mode investigation

Variable Coding

Mode of transport Sustainable mode of transport = 1

Private car mode = 0

Age Group (AG)

0 to 18 years = 0

18 years to 30 years = 1



65 years and above = 4

Occupation (OCCU)

Full time worker = 0

Student = 1

Part time worker = 2

Part time student = 3

Homemaker / retired = 4

Employment Status (ES)

Employed full time = 0

Employed part time & student full time = 1

Self employed = 2

Employed part time = 3

Student part time & employed part time = 4

Unemployed / retired = 5

Gender (GDR) Female = 0

Male = 1

Type of Student (TS)

Full time student = 0

Full time student & Casual employment = 1

Student part time = 2

Licence Availability (LA) Licence available = 1

Licence not available = 0

Car Availability (CA) Car available = 1

Car not available = 0

Level of Service (LOS)

LOS A = 6

LOS B = 5

LOS C = 4

LOS D = 3

LOS E = 2

LOS F = 1

No service = 0

The LOS for public transport frequency of service was determined using the procedure stated

in the TCQSM (TRB, 2003). The LOS was determined for each journey undertaken by a

KGUV user by public transport. It was observed before that the KGUV users had multiple

legs to their journeys. The LOS for each leg was determined and the minimum LOS was

assigned for the whole journey. The public transport frequency calculations were made for



the AM peak period of 7am to 9am as most of the trips were undertaken during this time

period.

The final regression model was obtained iteratively for trip undertaken by each user group.

Iterations were terminated when the change in parameter estimates was less than 0.001. To

test the goodness – of – fit of models, the Omnibus test for model coefficients was performed

to obtain the model Chi-square (2) from the log–likelihood technique. The statistically

significant difference between the full model and constant only model should be at a level of

at least p < 0.05 (Tabachnick and Fidell, 2007).

The effect of multicollinearity of independent variables was assessed from the value of the

coefficient and standard error. If a variable had high degree of multicollinearity then the

regression coefficients become unstable and the standard errors for the coefficients become

wildly inflated (UCLA, 2010). Any such variable with a high degree of multicollinearity was

removed from analysis. The importance of the individual variables was assessed using the

Wald statistics and the significance (p) value. Wald statistics should be significantly different

from zero to assume that the predictor is making a significant contribution in predicting the

outcome (Field, 2009). On the contrary, the significance value closer to the zero is desired for

a variable to be noted as significant in predicting the outcome. The variable with the value of

least significance is the most significant in predicting the outcome (Hair et al., 2006 and

Warner, 2008).

The outcomes from the logistic regression are explained with the help of odds ratios. Odds

are defined as the ratio of the probability of an event occurring to the probability of the event

not occurring (Hair et al., 2006). The odds ratio is defined as the change in odds of one of the

categories of outcome when the value of a predictor variable increases by one unit

(Tabachnick and Fidell, 2007).

The statistical software package “Statistical Package for Social Scientists (SPSS)” was used

for conducting the logistic regression analysis (Kinnear and Gray, 2009). The initial

regression analysis was undertaken considering all variables. The variable “Car Availability”

was found to have high degree of multicollinearity hence was removed from final analysis.

The values of regression coefficient and standard error can be observed from Muley et al.

(2009). The following section presents the final binary Logistic regression models developed

for shoppers, employees, students and residents at KGUV.



9.3 Analysis for shoppers’ shopping trips

The analysis of travel modes used for shopping trips by shoppers at KGUV was conducted

using LOS, trip length, travel time saving, frequency of shopping trip, age group and

occupation as independent variables, with 117 cases. Equation 9.3 shows the linear regression

equation and Table 9.2 lists the detailed results of logistic regression analysis. The final

model was statistically significant at a significance of 0.001 with 2 of 30.930 predicting 79.8

percent cases correctly. The values of regression coefficient and standard error indicated no

variable with high degree of multicollinearity. The comparison of Wald statistic and

significance values identified occupation as the least significant variable and age group as the

most significant variable in determining the travel mode of a shopper.

𝑍 = 1.168 + 0.155 × 𝐿𝑂𝑆 + 0.042 × 𝑇𝐿 + 0.069 × 𝑇𝑇𝑆 + 0.353 × 𝐹𝑅𝑄

− 1.122 × 𝐴𝐺 − 0.104 × 𝑂𝐶𝐶𝑈 Equation 9.3

The odds ratio for frequency of shopping trips (1.423) showed the highest odds of increased

use of a sustainable mode of transport with a unit increase in frequency of shopping trip. This

is justified because many employees and students visit the shopping centre often by walking

as this is their intrazonal trip within a short walking distance. In contrast, the odds ratio for

age group indicated that a unit increase in age group reduces the odds of using of a

sustainable mode of transport by 0.326. A unit increase in the LOS score, trip length and

travel time saving increase the odds of choosing walking, cycling or public transport by

1.167, 1.043 and 1.072 respectively. This isn‟t a fair comparison for trip length and travel

time because they exercise much greater variability i.e. across broader scales.

Table 9.2 Travel mode analysis for shoppers’ shopping trips

Variable Coefficient

(B)

Standard

error

Wald

statistic

Significance

(p)

Odds ratio

Exp(B)

Level of service (LOS) 0.155 0.308 0.253 0.615 1.167 Trip length (TL) 0.042 0.030 1.926 0.185 1.043 Travel time saving (TTS) 0.069 0.034 4.169 0.046 1.072 Frequency of shopping

trip (FRQ)

0.353 0.154 5.258 0.023 1.423

Age group (AG) -1.122 0.318 12.432 0.000 0.326 Occupation (OCCU) -0.104 0.258 0.162 0.688 0.902 Constant 1.168 1.826 0.410 0.522 3.217

No of cases = 117

% cases correctly predicted = 79.8 (criterion, if estimated probability > 0.500, the predicted mode is

sustainable mode of transport)

2model = 30.930, degree of freedom = 6, significance = 0.001

2critical = 22.458



Similar to age group, the unit increase in occupation reduces the odds of a using sustainable

mode of transport by 0.902. The odds ratio for occupation also indicates better odds of

choosing a sustainable mode for students and full time employees. The retired persons and

homemakers have a greater tendency to drive to the shopping centre.

Figure 9.1 plots the changes in the probability of selecting sustainable mode with the travel

time saving. The probabilities were calculated for a range of travel time saving (50 minutes to

-50 minutes) for a typical shopper who is full time employed (18 to 30 years of age) and has a

LOS of five with trip length of 7km visiting shopping centre three times a week. As the car

travel time becomes more than the public transport travel time the probability of choosing

public transport, walk or cycle increases. This is obvious because the shoppers aim at travel

time savings. For zero travel time saving the probability is 0.89 which is quite impressive.

Figure 9.1 Sensitivity of typical shopper’s sustainable travel mode probability with travel time

saving

9.4 Analysis for employees’ work trips

The logistic regression for work trips by employees at KGUV was undertaken for a combined

data set of retail shop employees and professional employees with 164 cases using LOS, trip

length, travel time saving, age group, employment status and gender as independent

variables. Unlike shoppers shopping trips analysis, the frequency of work trips was not

considered because the employment status indirectly considered the frequency of travel. An

additional variable, gender of employees, was used for this analysis. The travel mode was

0

0.2

0.4

0.6

0.8

1

-60 -40 -20 0 20 40 60

Pro

bab

ilit

y

Travel time saving (minutes)



used as a dependent variable. The results of Wald statistic and significance values indicated

that age group was statistically the most significant variable governing the travel mode; while

LOS was statistically the least significant. The values of regression coefficient and standard

error indicated no variable with high degree of multicollinearity. The model predicted 69.1

percent of cases correctly. The 2 model was 26.391 at a significance level of 0.001. The

details of the analysis are given in Table 9.3. Equation 9.4 notes the linear regression

equation obtained from the logistic regression analysis.

𝑍 = 1.842 + 0.012 × 𝐿𝑂𝑆 − 0.018 × 𝑇𝐿 + 0.014 × 𝑇𝑇𝑆 − 0.759

× 𝐴𝐺 − 0.602 × 𝐸𝑆 + 0.375 × 𝐺𝐷𝑅 Equation 9.4

Table 9.3 Travel mode analysis for employees’ work trips


(B)

Standard

error

Wald

statistic

Significance

(p)

Odds ratio

Exp(B)

Level of service (LOS) 0.012 0.183 0.004 0.948 1.012 Trip length (TL) -0.018 0.017 1.103 0.294 0.982 Travel time saving (TTS) 0.014 0.017 0.659 0.417 1.014 Age group (AG) -0.759 0.256 8.766 0.003 0.468 Employment status (ES) -0.602 0.214 7.952 0.005 0.547 Gender (GDR) 0.375 0.379 0.976 0.323 1.455 Constant 1.842 1.218 2.286 0.131 6.309

No of cases = 164



2model = 26.391, degrees of freedom = 6, significance = 0.001

2critical = 22.458

The details of analysis indicate that a male employee at KGUV has 1.455 times higher odds

of choosing a sustainable mode of transport than their female counterpart. It should be noted

that the proportion of female employees was higher than male employees. The higher odds

might be attributable to convenience and personal security strengths associated with the car.

The odds of choosing sustainable mode of transport reduce by 0.468 times as the age group

of a KGUV employee increases. This variation is shown in Figure 9.2. The model was

evaluated for a typical full time male employee who has LOS of five for public transport and

has a trip length of 12km with 20 minutes higher travel time by public transport. The highest

probability of choosing a sustainable transport mode was 0.85 for an employee 18 years or

younger. This probability was decreased to 0.22 for an employee 65 years or older. This may

be attributable to better availability of parking spaces for more senior employees. Overall, the

odds of choosing sustainable modes of transport increased with an increase in LOS, travel



time saving, and gender, while the odds decreased with an increase in trip length, age group

and employment status. Generally, the odds ratios followed the trend because public transport

use increases with an increase in public transport availability, and the use of public transport,

walking and cycling reduces as trip length increases. For an employee, use of sustainable

transport modes depends upon frequency of travel and parking space availability.

Figure 9.2 Sensitivity of typical employee’s sustainable travel mode probability with age group

9.5 Analysis for students’ education trips

The travel mode analysis for education trips for students at KGUV was undertaken using

school students‟ and university students‟ data (117 cases). A variable termed „licence

availability (LA)‟ was introduced to better distinguish between two groups of students. In

addition, transit characteristics such as LOS and travel time saving, and personal

characteristics such as LA, age group, gender and type of student were utilised in the logistic

regression analysis. Trip length was not included in the model because inclusion of trip

length provided a 2

model of 12.321 at a degree of freedom 7 at a significance level of 0.090.

This significance was greater than 0.05 hence the model did not fit as well statistically. So to

obtain a statistically fit model the least significant variable, trip length, was removed and

logistic regression analysis was again conducted.

The results of the analysis revealed that type of student was statistically the most significant

variable, while LOS was statistically the least significant variable. This might be because the

y = -0.1625x + 1.0388

R² = 0.9954

0

0.2

0.4

0.6

0.8

1

0 - 18 18 - 30 30 - 45 45 - 65 65 & above

Pro

ba

bil

ity

Age group (years)



students were mostly captive riders. The values of regression coefficient and standard error

indicated no variable with high degree of multicollinearity. The analysis predicted 87.2

percent cases correctly. The model has a significance value of 0.055 which is very close to

the desired level of significance (p < 0.05) with 2 of 12.311. Although the significance and

2

model values were at par of desired values, the model was used due to data constraints. The

data for education trips was biased with a high proportion of public transport trips. Equation

9.5 shows the linear regression equation for logistic regression and Table 9.4 explains the

details of analysis.

𝑍 = 3.263 − 0.054 × 𝐿𝑂𝑆 + 0.015 × 𝑇𝑇𝑆 − 1.934 × 𝐿𝐴 − 0.446 × 𝐴𝐺

+ 0.753 × 𝐺𝐷𝑅 + 1.362 × 𝑇𝑆 Equation 9.5

Table 9.4 Travel mode analysis for students’ education trips


(B)

Standard

error

Wald

statistic

Significance

(p)

Odds ratio

Exp(B)

Level of service (LOS) -0.054 0.255 0.045 0.832 0.947 Travel time saving (TTS) 0.015 0.032 0.218 0.641 1.015 Licence availability (LA) -1.934 0.903 4.586 0.036 0.145 Age group (AG) -0.446 0.652 0.468 0.494 0.640 Gender (GDR) 0.753 0.826 0.830 0.362 2.123 Type of student (TS) 1.362 0.630 4.677 0.033 3.904 Constant 3.263 1.715 3.620 0.057 26.131

No of cases = 117




2critical = 12.396

The odds ratios from the analysis indicated that the odds of choosing a sustainable mode of

transport highly depend on the type of student; the odds of a student choosing public

transport are 3.904 times higher for a unit change in type of student. This means under similar

conditions a full time student having casual employment had 3.904 times higher odds of

choosing sustainable modes of transport in comparison to a full time student. The LA had

0.145 odds of reducing public transport or walk only travel mode choice. The travel time

saving increases the odds of using a sustainable mode by 1.015 with a unit increase in travel

time saving. Similar to employees, and perhaps for similar reasons, male students had higher

odds (2.123) of choosing a sustainable transport mode. The LOS, LA and age group reduce

the odds of using a sustainable transportation mode. The reason for LOS results exhibiting an



opposite trend to the general trend may be that a large proportion of students used public

transport and this provided less knowledge about car trips made by a student to the university.

The regression equation was evaluated for a typical female full time student who has a valid

driver‟s licence, has a LOS of 4, and needs an extra 20 minutes if travelling by public

transport instead of car. Figure 9.3 represents the sensitivity of age group for probability of a

student choosing a sustainable mode of transport for that student. The probability shows a

reduction (0.69 – 0.27) as age group increases.

Figure 9.3 Sensitivity of typical student’s sustainable travel mode probability with age group

9.6 Analysis for residents’ first trip of the day

A binary logistic regression was performed on KGUV residents‟ data to predict the

probability of using sustainable modes of transport with six predictor variables; age group,

employment status, gender, LOS, trip length and travel time saving with 81 cases. Table 9.5

reveals the results of analysis and Equation 9.6 presents the regression equation. The values

of regression coefficient and standard error indicated no variable with high degree of

multicollinearity. The model predicted almost 89 percent of cases correctly. The 2 was

41.558 with a significance of 0.001. The values for Wald statistic and significance indicated

age group as the most significant variable in determining the travel mode while employment

status and LOS as the least influential variable. The residents had a good quality of service to

y = -0.1055x + 0.7986

R² = 0.9995

0

0.2

0.4

0.6

0.8

1

0 - 18 18 - 30 30 - 45 45 - 65 65 & above

Pro

bab

ilit

y

Age group (years)



several destinations (See Chapter 4) hence the availability of public transport was not a

concern for this group of users.

𝑍 = 4.590 + 0.498 × 𝐿𝑂𝑆 − 0.432 × 𝑇𝐿 + 0.115 × 𝑇𝑇𝑆 − 1.578 × 𝐴𝐺

− 0.461 × 𝐸𝑆 − 2.166 × 𝐺𝐷𝑅 Equation 9.6

Table 9.5 Travel mode analysis for residents’ first trip of the day


(B)

Standard

error

Wald

statistic

Significance

(p)

Odds ratio

Exp(B)

Level of service (LOS) 0.498 0.527 0.893 0.345 1.646

Trip length (TL) -0.432 0.225 3.674 0.059 0.649

Travel time saving (TTS) 0.115 0.066 3.071 0.085 1.122

Age group (AG) -1.578 0.646 5.973 0.019 0.206

Employment status (ES) -0.461 0.492 0.881 0.348 0.630

Gender (GDR) -2.166 1.207 3.222 0.079 0.115

Constant 4.590 3.598 1.628 0.202 98.532

No of cases = 81




2critical = 22.458

Figure 9.4 Sensitivity of typical resident’s sustainable travel mode probability with travel time

saving

The odds ratios indicate that the LOS variable has highest odds of increasing the use of

sustainable modes of transport (1.646). Unlike other users, male residents tend to use car

0

0.2

0.4

0.6

0.8

1

-60 -40 -20 0 20 40 60

Pro

bab

ilit

y

Travel time saving (minutes)



more than their female counterparts. The odds of a male resident using sustainable modes of

transport were less by 0.115 than a female resident. Generally, the odds of a resident using a

sustainable mode of transport increased with unit increase in LOS, and travel time saving. On

the contrary, the odds decreased with unit increase in trip length, age group, employment

status and gender. As the residents travelled longer distances the probability of choosing

public transport decreased and as the travel time saving increased the probability of using

public transport.

Figure 9.4 shows the variation of probability of a resident using sustainable modes of

transport with the variation in the travel time saving. The plot was drawn for a female full

time employee resident having an age group of 18 to 30 years with LOS of five, and a trip

length of 5km. The model indicates that the elder residents used car, likely due to the

availability of car and parking space at the destination. The full time employees and students

used more sustainable modes of transport as they were working in the CBD (where they have

good public transport service) or their university was within walkable distance of KGUV.


Table 9.6 presents the values of Wald statistic, and in parentheses significance, of different

independent variables for various user groups at KGUV. The comparison indicates that LOS

is least significant for travel mode determination, while age group is most significant except

for students‟ education trips. The trip length was significant only for residents‟ trips and

travel time saving was significant only for the shoppers‟ trips and residents‟ trips. The

employment status was significant in case of students and employee trips.

Table 9.6 Significance of variables for travel mode determination for KGUV users

Variable Shoppers trips Employee trips Students trips Residents

first trip

LOS 0.253 (0.615) 0.004 (0.948) 0.045 (0.832) 0.893 (0.345)

Trip length 1.926 (0.165) 1.103 (0.294) NA 3.674 (0.059)

Travel time saving 4.169 (0.041) 0.659 (0.417) 0.218 (0.641) 3.071 (0.085)

Age group 12.432 (0.000) 8.766 (0.003) 0.468 (0.494) 5.973 (0.019)

Employment status/

Occupation/Type of

student

0.162 (0.688) 7.952 (0.005) 4.677 (0.033) 0.881 (0.348)

Gender NA 0.976 (0.323) 0.830 (0.362) 3.222 (0.079)

Frequency of trip 5.258 (0.022) NA NA NA

Licence availability NA NA 4.586 (0.036) NA

Note: NA is Not Applicable



When the odds of different user groups were compared (Table 9.7), it was observed that all

TOD users except shoppers showed lesser odds of choosing public transport for greater trip

lengths. The most distinct was the residents‟ first trip. For all user groups, the increase in

travel time saving marginally increased the odds of using sustainable modes of transport, as is

to be expected. The increase in LOS increased the odds of using public transport and walk for

all users except for student trips where the odds ratio was subtly less than unity. The increase

in odds with LOS was most distinct for the residents‟ first trip.

The older KGUV users showed higher odds of using car compared to younger KGUV users.

This trend was consistent for all user groups. Considering employment status as an indication

of frequency of travel, the most frequent travellers in each group exhibited higher probability

of using sustainable modes compared to less frequent travellers, except for students‟

education trips. This is in line with the trend because the students having part time

employment with full time studies tend to use cars more to manage time and commitments.

The female residents showed higher tendency of using sustainable mode of transport as

opposed to the visitor groups, namely employees and students, where males did so. The

higher use of car by female is justified by previous research (Cervero and Radisch, 1996) but

the female residents at KGUV present an opposite trend.

Table 9.7 Comparison of odds of KGUV users’ for choosing a sustainable mode of transport

Variable Shoppers

trips

Employee

trips

Students

trips

Residents

first trip

LOS 1.167 1.043 0.947 1.646

Trip length 1.043 0.983 NA 0.649

Travel time saving 1.072 1.015 1.015 1.122

Age group 0.326 0.437 0.640 0.206

Employment status /

Occupation / Type of student 0.902 0.632 3.904 0.630

Gender NA 1.582 2.123 0.115

Frequency of trip 1.423 NA NA NA

Licence availability NA NA 0.145 NA

Note: NA is Not Applicable

9.8 Summary

The binary logistic models revealed that personal and transit characteristics have an impact

on the decision of mode selection. Overall, age group was the most significant variable for

determining the sustainable travel mode in case of shoppers, employee and residents‟ trips

while type of student and driver‟s licence availability was the most significant in case of



determining mode for students trips. The inclusion of LOS in the travel mode analysis

illustrates the effect of LOS on the travel mode selection. This adds to the existing state of

knowledge as this was noted as the least explored area in Section 2.5.2.

The logistic regression analysis presented in this chapter is useful to explain mode shares and

also in predicting mode shift. For example, application of the equations presented include

predicting mode shift to public transport, assessing the effect of sprawl on travel mode

(increased trip lengths), and determining the effect of aging population on the travel mode. It

is also important to note that the results from the statistical analysis may be applicable in

mode choice estimation in the strategic four step modelling process for certain TOD types.

9.9 Chapter close

This chapter presented the equations to assess the influence of personal and transit

characteristics on travel modes of a KGUV user. The following chapter (Chapter 10) derives

the models for travel modes of travel at TODs based on the equations presented in this

chapter. The outcomes presented in this chapter also contribute to Chapter 11.




Chapter 10

Models for travel modes of transit oriented development users

10.1 Introduction

In Chapter 9, we have seen the effect of personal and transit characteristics on selection of

travel modes of Kelvin Grove Urban Village (KGUV) users and the variation caused in the

probability of choosing sustainable modes of transport. These equations need to be simplified

for application into planning of Transit Oriented Developments (TODs) as it is not always

possible to predict all the variables for a newly planned development. This chapter presents

models that in the absence of site specific data or local data may be used to forecast the

probability of choosing a sustainable mode of transport for TOD users, which are derived

from the equations presented in Chapter 9. First a brief background for model development is

provided, followed by the models for each user group’s travel; namely shoppers’ shopping

trips, employees’ work trips, students’ education trips and residents’ first trips of the day.

Later a brief discussion about the models and their application is presented with a brief

summary and chapter close.

10.2 Analysis background for model development

The models predicting the probability of choosing sustainable mode were developed by

simplifying the equations produced in the previous chapter. In order to simplify a model,

variables which were not evidently significant in determining the sustainable travel mode

were removed from the equation, and the final model was obtained using the significant

variables. The Wald statistic was used to determine the significance of each variable (Hair et

al., 2006).

The Wald statistic is analogous to the t–statistic in linear regression. A higher value of Wald

statistic (significantly different than zero) denotes greater significance and a value closer to

zero represents least significance. The Wald statistic is the value of regression coefficient

divided by its associated standard error (Field, 2009). Mathematically,

𝑊𝑎𝑙𝑑 =𝐵

𝑆𝐸𝐵 Equation 10.1



where,

𝑊𝑎𝑙𝑑 = Wald statistic for a variable

𝐵 = Regression coefficient of the variable

𝑆𝐸𝐵 = Standard error of the variable

The Wald statistic follows a 2 distribution. The value of each Wald statistic is compared

with a 2 distribution with a degree of freedom of one (Bewick et al., 2005). If the

2 value of

the distribution is less than the observed value (Wald statistic) then the variable is significant

in predicting the travel mode of a TOD user. Only significant variables were retained for

inclusion in the final model development. The significance level indicates the percentage of

confidence level, that is, a significance of 0.05 indicates 5 percent confidence level. The 2

values of significance greater than 0.25 are not available (Tabachnick and Fidell, 2007),

hence the variables having significance in their Wald statistics in excess of 0.25 were denoted

as the least significant variables in predicting the travel mode.

After removing the least significant variables, logistic regression analysis was performed and

the model performance was noted. The final model was tested by assessing the proportion of

cases predicted correctly and the significance of the 2 obtained from the Omnibus test for

model coefficients. The significance of the Wald statistic of the variables in the final model

was also noted. The linear equation obtained from the final model (Equation 10.3) provided a

measure of individual’s preference for travel by sustainable modes of transport, which was

further used in determining the probabilities of using the sustainable modes of transport p(y)

using Equation 10.2. The details of model development for each user groups are explained in

following sections.

𝑝 𝑦 =𝑒𝑍

1 + 𝑒𝑍 Equation 10.2

𝑍 = 𝑎0 + 𝑎1 × 𝑏1 + 𝑎2 × 𝑏2 + 𝑎3 × 𝑏3 + ⋯ Equation 10.3

where,


𝑎0 = Regression constant

𝑎1, 𝑎2, 𝑎3 … = Regression coefficients

Models for travel modes of TOD users


𝑥1, 𝑥2, 𝑥3 … = Independent variables

𝑝 𝑦 = Probability of predicting y, in this case use of a sustainable

transport mode, using independent variables via Z

Similar to Chapter 9, the effect of each variable in the final model is described using odds

ratios. The odds ratios are exponentiated coefficients which will not have negative values. An

odds ratio above 1.0 reflects a positive relationship and a value less than 1.0 reflects a

negative relationship (Hair et al., 2006).

10.3 Model for shopping trips

Table 10.1 presents the details of original travel mode analysis and the χ2

critical for each

variable. The original equation considered Level (Quality) of Service (LOS), trip length,

travel time saving, frequency of shopping trip, age group and occupation as independent

variables for predicting probability of choosing sustainable mode of transport. The Wald

statistic indicated that LOS variable and occupation are the least significant variables

(Significance > 0.25) hence these variables were removed from the final model. Further,

comparison of Wald statistic implies that trip length, travel time saving, frequency of

shopping trips and age group are significant in predicting the probability of choosing

sustainable travel mode. Hence, the final regression model was developed using four

independent variables. The model form is given in Equation 10.4 and the details of the model

are given in Table 10.2.

Table 10.1 Original model for shoppers’ shopping trips


(B)

Standard

error

Wald

statistic

Significance

(p)

2critical

Level of service (LOS) 0.155 0.308 0.253 0.615 NA

Trip length (TL) 0.042 0.030 1.926 0.185 1.922

Travel time saving (TTS) 0.069 0.034 4.169 0.046 4.030

Frequency of shopping

trip (FRQ)

0.353 0.154 5.258 0.023 5.238

Age group (AG) -1.122 0.318 12.432 0.001 10.828

Occupation (OCCU) -0.104 0.258 0.162 0.688 NA

Constant 1.168 1.826 0.410 0.522 –

No of cases = 117



2 model = 30.930, degree of freedom = 6, significance = 0.001

2critical = 22.458



The final form of the model predicted 76.9 percent cases correctly and was statistically

significant at 0.001 with 2 of 30.380. Although the percentage of cases correctly predicted

reduced slightly the final model form was acceptable because the χ2

model was considerably

higher than its critical value.

𝑍 = 1.964 + 0.040 × TL + 0.075 × 𝑇𝑇𝑆 + 0.311 × 𝐹𝑅𝑄 − 1.043 × 𝐴𝐺 Equation 10.4

The odds of the final model for shoppers’ shopping trips indicates that a unit increase in trip

length, travel time saving and frequency of shopping trips increased the odds of choosing

sustainable modes of transport by 1.041, 1.078 and 1.365 respectively. While the odds of

choosing sustainable modes of transport decreases by 0.352 with a unit increase in age group.

These odds exhibit a similar trend to the original model, indicating removal of a variable

changes the magnitude of odds but does not affect the inclination of variables.

Table 10.2 Revised model for shoppers’ shopping trips


(B)

Standard

error

Wald

statistic

Significance

(p)

Odds ratio

Exp(B)

Trip length (TL) 0.040 0.024 2.755 0.097 1.041

Travel time saving

(TTS)

0.075 0.029 6.654 0.010 1.078

Frequency of

shopping trip (FRQ)

0.311 0.143 4.740 0.029 1.365

Age group (AG) -1.043 0.284 13.489 0.001 0.352

Constant 1.964 0.570 11.867 0.001 7.130

No of cases = 117




2critical = 16.266

10.4 Model for work trips

The analysis for work trips for employees was undertaken similar to the previously explained

case. The sustainable travel mode equation containing LOS, trip length, travel time saving,

age group, employment status and gender as independent variables was used for analysis. The

details of the original analysis are listed in Table 10.3 with χ2

critical values for each variable.

The significance of the Wald statistic for LOS, trip length, travel time saving and gender

indicate that these variables are not significant in predicting the probability of choosing a

sustainable travel mode hence should be removed from the final model. Further, the

comparison of Wald statistic with χ2

critical shows that age group and employment status are

important for predicting the probability of choosing sustainable travel mode so these



variables were considered in final analysis and logistic regression was performed. The results

of logistic regression are listed in Table 10.4 and the regression equation is presented as

Equation 10.5. The final mode model form contained only two variables and predicted 64.2

percent of cases correctly. The 2

model was 21.043 at a significance level of 0.001. Although a

drop in the percent of cases predicted correctly was observed and the model was acceptable

because 2

model was greater than 2

critical.

Table 10.3 Original model for employees’ work trips


(B)

Standard

error

Wald

statistic

Significance

(p)

2critical

Level of service (LOS) 0.012 0.183 0.004 0.948 NA Trip length (TL) -0.018 0.017 1.103 0.294 NA Travel time saving

(TTS)

0.014 0.017 0.659 0.417 NA

Age group (AG) -0.759 0.256 8.766 0.003 0.468 Employment status

(ES)

-0.602 0.214 7.952 0.005 0.547

Gender (GDR) 0.375 0.379 0.976 0.323 NA Constant 1.842 1.218 2.286 0.131 –

No of cases = 164



2model = 26.391, degrees of freedom = 6, significance = 0.001

2critical = 22.458

The odds ratios imply that a unit increase in the age group of an employee decreases the use

of sustainable modes of transport by 0.459. A full time employee exhibited higher odds of

using sustainable modes of transport than a part time employee. In accordance with the

shoppers’ trends in odds ratio, the trend for odds ratios were similar in case of the original

and final model, although the magnitude varied slightly.

Table 10.4 Revised model for employees’ work trips


(B)

Standard

error

Wald

statistic

Significance

(p)

Odds ratio

Exp(B)

Age group (AG) -0.779 0.236 10.926 0.001 0.459

Employment status

(ES)

-0.545 0.201 7.343 0.007 0.580

Constant 1.452 0.440 10.889 0.001 4.272

No of cases = 164



2 model = 21.043, degrees of freedom = 2, significance = 0.001

2critical = 13.816



𝑍 = 1.452 − 0.779 × 𝐴𝐺 − 0.545 × ES Equation 10.5

10.5 Model for education trips

The details of original logistic regression for student’s education trips using LOS, travel time

saving, licence availability, age group, gender and type of student is shown in Table 10.5.

The significance of the Wald statistic indicates that LOS, travel time saving, age group and

gender are not significant in predicting the probability of choosing a sustainable travel mode.

Comparison of the Wald statistic with χ2critical shows that licence availability and type of

student are the only two variables predicting the probability of choosing a sustainable mode.

Hence, the final model contained these two variables for predicting the probability of

choosing a sustainable travel mode. The details of logistic regression of these two variables

are shown in Table 10.6 and the final model form is given by Equation 10.6.

Table 10.5 Original model for students’ education trips


(B)

Standard

error

Wald

statistic

Significance

(p)

2critical

Level of service

(LOS)

-0.054 0.255 0.045 0.832 NA

Travel time saving

(TTS)

0.015 0.032 0.218 0.641 NA

Licence availability

(LA)

-1.934 0.903 4.586 0.036 4.502

Age group (AG) -0.446 0.652 0.468 0.494 NA

Gender (GDR) 0.753 0.826 0.830 0.362 NA

Type of student (TS) 1.362 0.630 4.677 0.033 4.644

Constant 3.263 1.715 3.620 0.057 –

No of cases = 117




2critical = 12.3967

The final model form for student’s education trips was significant with significance level of

0.014 and χ2

model of 8.484. The model predicted 85.3 percent cases correctly, the percent of

cases predicted reduced compared to the original model. In this case, although the χ2

model and

χ2

critical are very close, the final model was accepted because the significance level was

reduced from original value increasing the model fit and data constraints noted earlier.

The odds ratios indicate that a unit increase in type of student increases the odds of a student

using sustainable modes of transport by 2.958. If a student possesses a valid driver’s licence



then odds of the student using public transport or walking to school or university decreases

by 0.170 as compared to a student who does not have valid driver’s licence.

𝑍 = 2.414 − 1.775 × LA + 1.085 × TS Equation 10.6

Table 10.6 Revised model for students’ education trips


(B)

Standard

error

Wald

statistic

Significance

(p)

Odds ratio

Exp(B)

Licence

availability (LA)

-1.775 0.724 6.009 0.014 0.170

Type of student

(TS)

1.085 0.558 3.784 0.052 2.958

Constant 2.414 0.605 15.903 0.001 11.179

No of cases = 117




2critical = 8.721

10.6 Model for residents’ first trip of the day

The model for the travel mode of the first trip of the residents’ mode was determined using

details shown in Table 10.7. The initial model contained LOS, trip length, travel time saving,

age group, employment status and gender as independent variables. The Wald statistic

significance of LOS and employment status was greater than 0.25 hence these variables were

removed from the final analysis. In addition, the comparison of Wald statistic of each

variable with the corresponding 2

critical indicates that trip length, travel time saving, age

group and gender were significant variables. Hence, these four variables were retained for

further analysis. The final model form of the linear regression equation is given in Equation

10.7 and the statistical details are presented in Table 10.8.

The final model predicted 90.1 percent cases correctly with a χ2 of 39.894 at a significance of

0.001. The percent cases predicted correctly slightly increased from the original model.

Although χ2

model reduced slightly, when compared with the final degree of freedom at a given

significance, it was higher than the critical value. So the final model form was acceptable.



Table 10.7 Original model for residents’ first trip of the day


(B)

Standard

error

Wald

statistic

Significance

(p)

2critical

Level of service (LOS) 0.498 0.527 0.893 0.345 NA

Trip length (TL) -0.432 0.225 3.674 0.059 3.636

Travel time saving

(TTS)

0.115 0.066 3.071 0.085 3.045

Age group (AG) -1.578 0.646 5.973 0.019 5.667

Employment status

(ES)

-0.461 0.492 0.881 0.348 NA

Gender (GDR) -2.166 1.207 3.222 0.079 3.182

Constant 4.590 3.598 1.628 0.202 –

No of cases = 81



2

model = 41.558, degree of freedom = 6, significance = 0.001

2critical = 22.458

The odds ratios for final model indicates that the odds of a resident choosing sustainable

modes of transport increases with a unit increase in travel time saving (1.146) and decreases

with a unit increase in trip length and age group by 0.662 and 0.215 respectively. The odds of

0.154 for gender show that female residents have more tendency of using sustainable modes

of transport than their male counterparts. These trends were similar to that observed in the

original model.

𝑍 = 6.999 − 0.413 × 𝑇𝐿 + 0.136 × 𝑇𝑇𝑆 − 1.538 × 𝐴𝐺 − 1.871 × 𝐺𝐷𝑅 Equation 10.7

Table 10.8 Revised model for residents’ first trip of the day


(B)

Standard

error

Wald

statistic

Significance

(p)

Odds ratio

Exp(B)

Trip length (TL) -0.413 0.204 4.096 0.043 0.662

Travel time

saving (TTS)

0.136 0.064 4.478 0.034 1.146

Age group (AG) -1.538 0.628 6.007 0.014 0.215

Gender (GDR) -1.871 1.114 2.820 0.093 0.154

Constant 6.999 1.916 13.345 0.001 1095.764

No of cases = 81




2critical = 18.467



10.7 Interpretation

The models for travel modes derived by removing the least significant variables indicate that

the χ2

model reduces from each original model, but the reduction in the degrees of freedom

increases the model fit. Similarly, the percentage of cases predicted correctly observes a

small reduction. This small reduction in percentage of cases predicted correctly indicated the

piecemeal influence of removal of variables on the overall model. The odds ratios also

exhibit a similar trend with a slight variation in the odds. The significance value for model fit

remained constant in the case of all models, except for education trips, where the significance

was decreased significantly from the original model increasing the model fit.

The final models indicate that the travel mode of shoppers and residents trips was dependant

on personal as well as travel characteristics, while the travel mode of employees and students

was determined from their personal characteristics only. The LOS variable was not retained

for any user group, but the effect if transit service was taken into account through travel time

saving for residents and shoppers; travel time savings was noted as an important variable

when asked about perceptions (Appendix D). Although employees and students noted travel

time saving as a prominent variable, statistically it was least significant, hence was removed

from the final model. The models developed based on KGUV users’ travel data can be used

for planning future TODs, which can further help in planning the infrastructure for

sustainable modes of transport. A brief overview of the procedure for model application is

noted in the following section.

10.8 Model application

As stated earlier, in the absence of site specific or local data, the models proposed in this

research may be used to forecast travel and planning of a future TOD. However, these models

first need to be generalised by considering travel data from various forms of TODs and

general models for all trips undertaken at TODs need to be developed. It should be noted that

the general equation is obtained from the raw travel data of various TOD sites. Before model

development; the variables in the raw data should be normalised, i.e. they should be

transformed to a normal distribution with unit standard deviation and a zero mean. This step

is necessary as the characteristics of each case study TOD varies (for example, scale of

TODs), which is not explained by the variables under consideration. From the normalised

data, the generalised (global) equation for each trip type is obtained of the form given in

Equation 10.8. This equation provides basis for estimating mode split.



𝑍 = 𝑏0 + 𝑏1𝑥1 + 𝑏2𝑥2 Equation 10.8

where,


𝑏0 = Global regression constant

𝑏1,𝑏2 = Global regression coefficients

𝑥1, 𝑥2 = Independent variables

This following section illustrates a methodology which may be adopted in applying the

general model for a specific trip type for TOD planning. The similar procedure should be

adopted for all trip types. However, it must be noted that this methodology has not been

tested through this research; such testing is a proposed topic for future research. The

methodology for application of a general equation given, as noted by Bruton (1985), consists

of five steps as explained below:

1) Assume the mean (µ1, µ2,...) and standard deviation (σ1, σ2,...) for each variable for the

development under consideration by using the values from past travel data and

applying the experience and considering the proposed characteristics of the users.

2) Derive the appropriate equation coefficients from the global regression constants

applicable for the subject development. The new regression coefficients, applicable

for subject development, can be derived by dividing global regression coefficients by

the standard deviation of the respective variable. The new regression coefficients can

be derived from the equation below:

𝑏1′ =

𝑏1

𝜎1, 𝑏2

′ =𝑏2

𝜎2, … Equation 10.9

3) Determine the value of 𝑍

The value of linear regression function (𝑍) can be calculated by inputting the new

regression coefficients into Equation 10.8. The new form of the equation is given by

Equation 10.10. It should be noted that the global regression constant is dropped as it

has no relevance for the TOD under consideration. The calibration of the equation is

achieved using a scaling factor while determining the probability.

𝑍 = 𝑏1′ μ1 + 𝑏2

′ μ2 + ⋯ Equation 10.10



4) Apply 𝑍Z and scaling factor (𝑥) to determine the probability of choosing sustainable

travel mode

The value of scaling factor, ranging from 0 to 1, is obtained considering the variables

under consideration and from raw data or proposed modal split. The scaling factor can

be derived from Equation 10.12. The probability of a TOD user using sustainable

modes of transport can therefore be obtained using Equation 10.11.

𝑝(𝑦) =𝑒𝑥+𝑍

1 + 𝑒𝑥+𝑍 Equation 10.11

𝑥 = − 𝑏𝑖′𝜇𝑖

𝑖=𝑛

𝑖=0

+ 𝑙𝑜𝑔𝑒

𝑎2

𝑎1 Equation 10.12

where,

𝑏𝑖′ = Coefficients for the variables under consideration

𝜇i = Mean of the variable under consideration

𝑛 = Number of variables under consideration

𝑎2

𝑎1

= Ratio of more sustainable mode users to less sustainable mode

users

5) Determine the mode split using the projected population for the subject development

The mode split for the subject development can be determined by multiplying

probability obtained by applying Equation 10.11 with the population of the respective

user group undertaking the respective trip type.

10.9 Summary

The models developed in this chapter allow the probability of TOD users choosing a

sustainable mode of transport. In the absence of site specific or local data, these model forms

may be used in forecasting travel patterns of new greenfield or brownfield TODs. It must be

stressed that the models presented in this study are based on data from only one development,

so these models need to be refined by observing travel data from other TODs, as each TOD

has its own characteristics due to location, geography, mix of land uses and socio – economic

characteristics.



10.10 Chapter close

This chapter exhibited the models for travel modes a TOD; these models complete the fourth

step of TOD evaluation. The input from this step completes the fifth step of determining the

outcomes. The developments in this chapter provide base for conclusions of this research

which are presented in the following chapter (Chapter 11).


Chapter 11

Conclusions and recommendations

11.1 Introduction

This research was aimed at developing a universal methodology for evaluating the transport

impacts of Transit Oriented Developments (TODs). This methodology was elaborated upon,

using a case study TOD. This chapter presents the conclusions and recommendations from

this research. Firstly, the main conclusions are provided followed by reflections from case

study TOD and applications of this research. The contribution to knowledge, limitations of

this research and areas of further research are noted in the following sections. Finally, the

chapter close is provided, which also provides closure to this thesis.

11.2 Conclusions from this research

Based on the research conducted to achieve the objectives, this thesis maintains that TODs

are special property developments; hence from a transport perspective needs to be evaluated

differently from conventional developments. These developments should be assessed as a

complete system considering characteristics of residential as well as non residential land uses

and their users. Simply evaluating the transport at the residential land use of a TOD, for

assessing its performance, can lead to only a partial impression of transport at TODs.

The residential land use at TODs should be designed to cater for small as well as larger

family households. Further, instead of strictly developing TODs around a major public

transport node, the public transport nodes might be able to be located on the edge / boundary

of the TOD while still obtaining transport efficiency. The underlying principles for public

transport design should be that the TODs should not only have good quality of public

transport service to and from key centres such as Central Business Districts (CBD) but also to

and from various surrounding key origins and destinations to serve the transport demand of

residents as well as visitors of TODs. This is important because the travel mode is governed

by personal characteristics as well as on available public transport service, rather than the

development characteristics only.

To illustrate the above mentioned aspects, the key conclusions from this research are noted

below, by linking them with the aims and objectives of this research, which were set out in

Chapter 1. This is explained with the work reported in respective chapters to achieve these



objectives. The work reported in some chapters helped to achieve multiple objectives, so

these chapters may appear more than once in following sections.

11.2.1 Build up an understanding of the concept of TOD and various aspects related to

it with a detailed knowledge on TOD evaluation

Chapter 2: An extensive review of the literature was carried out to study the existing state of

knowledge about TOD and studies conducted on TOD evaluation from a transport point of

view. The review suggested that in the main, TODs were evaluated based on data from

residential land use only. But this is not completely valid because, due to presence of mixed

land uses, TODs attract higher trips by different users. Hence a study of data for users of non

residential land uses was noted as the least explored area, and proposed for action in this

research.

11.2.2 Develop a methodology for evaluating the transport impacts of TODs

Chapter 3: This chapter presented the detailed methodology for TOD evaluation. The

proposed five step methodology was comprised of some new or additional steps for TOD

evolution in addition to the standard practice of the literature. The main steps were pre–TOD

assessment, traffic and travel data collection, obtaining traffic impacts, determining the travel

impacts and drawing the outcomes.

The following sections explain key features for implementing this methodology for

evaluating a TOD.

Chapter 4: The pre–TOD assessment of the development for assessing its suitability as a

TOD should be made by studying the development characteristics and evaluating the quality

of service (QoS) for public transport service to and from various origins and destinations. The

framework provided by TRB (2003) can be used to determine the QoS of public transport

service, while universal frameworks for assessing walking and cycling infrastructure, as well

as the mix of land uses, remains to be developed.

Chapter 5: A cordon count should be conducted at all access points to obtain the classified

traffic counts. The travel data collection is an extensive and expensive process. The detailed

procedure for travel data for each TOD user group was explained in this chapter. Different

survey instruments should be employed to gather travel data for the various groups of TOD

users, depending on their demographic profile or characteristics. The questionnaire form

should be designed as a short form using a simple language so that it is easily understood by a

layperson; the questions can be simplified by providing multiple choices for answer. Pilot



surveys are an integral part of travel data collection, as these test and refine the survey

process. Incentives for those surveyed helped to increase the response rates, but to a limited

extent. Importantly, the sample size for a TOD should ideally be 100 percent of its

population, making it a census. However, this becomes more challenging over time.

Chapter 6: For determination of traffic impacts, a comparison of published trip rates is more

appropriate over the comparison of the trip rates obtained by household travel surveys. The

data collected for determining published trip rates is mainly collected for impact assessment

purposes, while the data from the travel surveys is mainly collected for strategic modelling

purposes. Further, the purpose of data collection governs the details and the extent of data

collected. The trip rates from travel surveys are on the sample basis while the published trip

rates are based on the population data. Hence this research proposes use of published rates for

determination of traffic impacts.

The traffic impacts of a TOD can be obtained from analysis of the cordon data. It is not

reasonable to expect that a road network at a TOD in an open environment will only be

carrying the traffic generated by the users of development. Hence, it is important to eliminate

through traffic from observed traffic so as to determine the exact traffic generated by the

development. The peak hour/s for a TOD should be identified based on the maximum person

movements rather than maximum vehicle movements. This is because TODs tend to attract

large numbers of pedestrian, bicycle and public transport trips. The details such as directional

distribution, and car occupancy for all modes of transport; cars, motorcycles, pedestrians,

bicycles, and public transport should be determined to gain a full appreciation of traffic

movement at the TOD. The total traffic generated, considering this as a conventional

development, can be determined from combining the trip rates specified for each land use by

the standard guidelines (such as ITE, 2008; RTA, 2002). Admittedly, data for some land uses

can be hard to come by. Trip rates should be compared to the observed maximum trip rate to

determine differences, and hence the traffic differences of the TOD.

It was found that only car traffic generation can be compared with standard guidelines to

determine traffic impacts. No trip rates for pedestrian or bicycle trip generation are specified

in standard literature. A study of this aspect is necessary for a TOD, as this governs the

provision of infrastructure for these modes. Further, TODs should be assessed as a special

land use category and land use mix and proportion of self containment of trips at a TOD

should be assessed.



Chapter 7: The requirement for determination of travel impacts is made by obtaining

datasets for comparison; specifically for TOD users and non TOD users. The dataset for TOD

users can be obtained from analysis of the travel data. The compiled dataset for each group of

TOD users should be analysed to obtain the demographic and travel characteristics. The

demographic characteristics sought to include personal characteristics such as age group,

gender, frequency of visiting TOD, employment status, driver’s licence availability and

household characteristics such as household size, number of bedrooms in the household,

vehicle and bicycle ownership, number of valid driver’s licence holders. The travel

characteristics include mode shares, trip lengths, parking details such as parking location and

parking fee, and public transport trip details such as access and egress times, number of

transfers and transfer locations. Additionally, the trip making characteristics such as number

of trips per person and internalisation of travel activities should be obtained for TOD

residents.

Chapter 8: The dataset for non TOD users can be obtained from secondary sources or by

conducting a separate travel survey. The decision to conduct a separate travel survey depends

on the time and resources available. Typically, the dataset for non TOD users can be obtained

from a development which has similar size and distance from the CBD as does the TOD

development. A regional comparison should also be made to determine any variation.

After obtaining the datasets for comparison, the travel characteristics of users of Non

Residential Land Uses (NRLU) should be compared and the travel as well as household

characteristics of the users of Residential Land Uses (RLU) should be compared. The travel

characteristics can include (but not limited to) mode share, trip length, access distances,

parking characteristics and trip making characteristics (for residents). The household

characteristics can include (but not limited to) motor vehicle ownership, bicycle ownership

and household size.

Chapter 9 and Chapter 10: Travel demand analysis for investigating travel at a TOD should

be primarily undertaken for assessing the travel modes of TOD users. This can further be

extended for studying the distribution of trips. Further, a detailed analysis should be

conducted to investigate the travel modes and distribution of trips, to develop models for

travel modes and trip distribution.

Chapters 5, 6, 8, 9, 10: The findings from the data collection provide guidelines for

conducting data collection at a TOD. The results from the investigation carried out for



determining traffic impacts indicate variation in traffic generation as well as trip rates for

various modes such as cars, pedestrian, bicycles, public transport and motorcycles. The

comparative analysis of users’ characteristics provide an appreciation of the variation in

travel characteristics of TOD users and household characteristics of TOD residents, which

provide detail on the travel impacts of a TOD. Further, the investigation into travel modes

provides more information about travel demand at TODs. These outputs complete the

analysis for obtaining the final outcomes.

11.2.3 Demonstrate the methodology by implementing it on an Australian case study

TOD

For implementing the methodology developed in this study, a fully planned development,

Kelvin Grove Urban Village (KGUV), located 3km northwest of Brisbane’s CBD, in

Australia, was used as a case study TOD. The selected development was assessed for its

suitability as a TOD by studying the development characteristics and evaluating the QoS for

transit availability for various origins and destinations. The development was reported to

have a mix of land uses, good quality of public transport service, walking and cycling

infrastructure. This was the first step of TOD evaluation. (Chapter 4)

The traffic data was collected by conducting the classified cordon counts for KGUV as a

whole, and of the centrally located shopping centre. The travel data was collected by

conducting travel surveys. The professional employees and university students who had good

internet access were surveyed using internet based surveys. Retail shop employees and

shoppers who may not have internet access or office email were surveyed using the personal

interview technique. For these two groups, initially CAPI surveys were planned but the

survey instrument was modified to pen and paper form for convenience and improved

uptake. The non student residents and school students were surveyed using a mail back

survey technique and the student residents were surveyed using intercept surveys. This

process completed the second step of TOD evaluation being traffic and travel data collection.

(Chapter 5)

The traffic impacts of KGUV were determined by following the procedure mentioned in the

third step of TOD evaluation. The maximum trip rate for car was compared with the total

traffic volumes derived from guidelines specified in ITE (2008) and RTA (2002). A similar

procedure was repeated for the cordon counts conducted for centrally located shopping

centre. (Chapter 6)



The travel data from shoppers, employees, students and residents at KGUV was collected to

compile datasets for comparison of KGUV (which is a TOD in this case). The travel impacts

were determined by comparing the characteristics of KGUV users with those of respective

user groups from regional (Brisbane Statistical Division, BSD) and suburban (Brisbane Inner

North Suburbs, BINS and Brisbane Inner South Suburbs, BISS) users’ characteristics.

(Chapter 7 and Chapter 8)

Travel demand analysis for investigating the travel modes of TOD users was undertaken, as

travel mode is a prominent aspect of a trip. The effect of personal characteristics such as

employment status, type of student, gender, age group, frequency of trip and travel

characteristics such as trip length, LOS and travel time savings were evaluated using the

logistic regression technique. Separate equations were developed for each TOD user group

predicting the probability of choosing sustainable modes of transport. Further, the

investigation into the travel modes of TOD users was undertaken and models were developed

for determining the travel modes of different trips at KGUV. (Chapter 9 and Chapter 10)

The findings from analysis (Chapter 6, 8, 9, and 10) provided transport impacts and

completed the evaluation of KGUV.

11.2.4 Determine trip rates for various modes of transport for a TOD and assess the

travel demand of TOD users

Chapter 6: Analysis of cordon data provided the peak periods for person movements. The

trips undertaken during these time periods can be used as trip rates. It was observed that on

given survey day the maximum vehicle trips were observed for PM peak for shopping centre

(211 veh/h) and 1078 veh/h for AM peak when KGUV was considered as a whole. The

highest pedestrian movement was 799 ped/h for midday peak and 1375 ped/h for PM peak

for the shopping centre and the entire KGUV respectively.

Chapter 9: The travel demand for each group was assessed by conducting binary logistic

regression analysis for travel modes. The travel modes were divided into two parts;

sustainable modes (walk only, bicycle, and public transport) and less sustainable modes (car,

motorcycle, and taxi). The results indicated that overall age group was the most significant

variable for determining the sustainable mode choice in case of shoppers, employee and

residents’ trips, while type of student and driver’s licence availability were the most

significant in case of determining mode for students trips.



Chapter 10: Travel demand assessment was continued for determining the models for travel

mode. The models were developed for trips by all user groups at KGUV.

11.2.5 Evaluate the transport impacts of TODs from an Australian perspective by

comparing the results with characteristics of conventional development

The results obtained from an Australian case study TOD, KGUV, helped to determine the

transport impacts of TODs from an Australian perspective.

Chapter 6: The impacts on the traffic generation at the TOD were determined from the

differences between total traffic generation at KGUV and ITE (2008) and RTA (2002) trip

rates. The comparison for the shopping centre showed reduced traffic generation by more

than 63 percent when compared with the ITE (2008) and RTA (2002) guidelines. When

KGUV as a whole was considered, a reduction of about 42 percent was observed for peak

period traffic (RTA, 2002 comparison) and a reduction of about 27 to 48 percent was

observed for the AM and PM peak hours respectively (ITE, 2008 comparison).

Chapter 8: The travel impacts of the TOD were obtained by comparing the characteristics of

KGUV users with Brisbane Statistical Division (BSD), Brisbane Inner North Suburbs

(BINS), and Brisbane Inner South Suburbs (BISS) users’ characteristics. The outcomes of

comparative analysis indicated that the KGUV users used more sustainable modes of

transport than BSD, BINS and BISS users. While the overall average trip length for shopping

and education trips at KGUV shows higher trip lengths than its counterparts. Shorter trips

were observed for KGUV residents and employees. The KGUV residents exhibited a trend of

undertaking fewer trips and possessing fewer motor vehicles than BSD, BINS and BISS

residents showing reduced dependence on the car. The residents also had smaller household

size and a smaller number of bicycles per household than their counterparts.

Although KGUV was not fully developed when the data collection was undertaken, the

results will be still applicable in the future. The further development of KGUV will

incorporate additional users from office and residential land use and these users are more

likely to possess similar travel characteristics to those of existing KGUV users due to

availability of infrastructure for more sustainable modes of transport and the development’s

proximity to the CBD.



11.3 Reflections from case study TOD

The application of the proposed methodology to an Australian case study provided some

insights into TOD implementation. It was observed that some aspects of case study TOD

were more successful than others. The key indications observed from case study TOD which

can be useful for TOD planners are as noted below:

For successful implementation of TOD, good quality of public transport service is

required from key origins and destinations. Provision of competent public transport

service leads to increased public transport usage (as observed from increased mode

shares).

The topography of the surrounding area governed the bicycle ownership and in turn

the bicycle usage for TOD residents. Further good connections to various origins and

destinations across Brisbane (for example bicycle tracks, dedicated bicycle facilities)

are desired for promoting bicycle usage for all TOD users. Trip end facilities have

also shown considerable influence on the bicycle mode share in case of users of non

residential land uses.

The public transport node may be located at the centre or on the edge of development

but should be easily accessible to encourage use of public transport. Good

accessibility is important rather than its location.

The provision of specialised facilities at case study TOD exhibited higher trip lengths

for the concerned TOD users as compared to the users of conventional land uses.

The comparison of trip characteristics indicated that the case study TOD residents

made fewer trips and the trips were contained in a nearby area of TOD. TOD residents

used car for making longer trips.

The provision of limited parking facilities restricted the car trips which have

ultimately increased use of more sustainable modes of transport.

It is necessary to provide some form of shelter at the stops and for sidewalks

considering the local climatic conditions.

Inclusion of education land use and provision of supporting infrastructure generated

more local walking trips. On the negative side the area was very quiet during

university holidays making it less attractive.

The careful placement of various land uses such as parks and open spaces,

recreational facilities made the TOD environment more vibrant. Presence of centrally

located shopping centre kept trips intrazonal undertaken by walk. The scale of



shopping centre is vital in keeping shopping and conveyance trips intrazonal and

attracting interzonal trips.

The various land uses should be carefully mixed together to ensure proper functioning

of each land use especially care should be exercised while locating residential land

use. Careful mixing of various land uses also reduce the total vehicular traffic

generation.

The TOD should have houses/apartments with more number of bedrooms per

household. Restricting the development to one or two bedroom apartments attracts

young couples, single family households or families with no children but deters large

family households who tend to drive more due to children’s activities.

Adjoining areas may also benefit from the TOD’s atypical development

characteristics. So careful planning of land uses should be done to serve this

additional demand due to adjoining land uses.

The affordability is a vital consideration when deciding the pricing for the facilities.

Excessive pricing discourage an average person from using the facilities at a TOD.

11.4 Applications of this research methodology

The research methodology developed in this study is applicable to any TOD irrespective of

its size and location. The application of this methodology for various TOD assessments can

lead to a framework for TOD evaluation. Further, the outcomes obtained from TOD

evaluation can be used for designing new TODs and also can be used to assess performance

of existing TODs.

To apply this methodology to another TOD some site specific modifications may be required.

The following points provide some possible alterations.

In the case of a TOD served by more than one mode of public transport, a step for

determination of the combined quality of service for public transport needs to be

added. This should be an additional step for pre–TOD assessment.

An additional step of sample size determination may be required for travel data

collection in case of large scale TODs or where time and resources are limited.

If the public transport station is at the centre of a TOD then the public transport

demand also needs to be determined for the calculation of trip rates. The public

transport demand for a TOD should be determined based on the type of station. If the



public transport station is a major interchange or the station is also serving to an

adjacent development then an additional step is required in Step III of TOD

evaluation to determine the accurate demand generated by the TOD.

Generally, for a TOD the centrally placed land use is the highest trip generator which

attracts trips from various land uses. A TOD which has other land use, such as an

activity centre, as the major trip attractor then the traffic generation of that particular

land use should be determined, in addition to the whole TOD traffic generation.

11.5 Contribution to knowledge

The main contribution from this research is the development of a comprehensive

methodology for TOD evaluation and an insight into the travel characteristics of TODs

considering both residential and non residential land uses. The key features achieved while

studying these two aspects are as noted below:

This research proposes a comprehensive methodology for evaluating the transport

impacts of TODs which considers various user groups of TODs with the total traffic

generation considering the TOD as a whole.

The previous studies considered in the main only residents’ travel data, and traffic

generated by TOD residents. This study focuses on the users of non residential land

use, which has been a neglected focus, and improves understanding about users of

these land uses. Thus a complete picture of travel at TODs may be provided. The

outcomes of this research will help planners while designing various land uses at

TODs.

This research provides trip rates for various modes of transport and assesses the traffic

impacts of TOD as a whole. These will help transport engineers and planners, and

policy makers to set appropriate targets of traffic reduction at TODs.

This research used an Australian case study for demonstrating the research

methodology, which provides insight into TOD evaluation from an Australian

perspective. This fills a major gap in knowledge as only a limited amount of previous

study explaining transport impacts of TODs from Australian perspective were found.

The Australian perspective will provide input to stakeholders for setting up more

TODs and develop activity centres.



This study collects the travel data and perceptions of all user groups at a TOD, which

can be used by transport engineers and planners for further planning of TODs and

extending the research in identified areas.

The travel mode analysis was conducted to study how personal and transit

characteristics affects the sustainable mode choice at a disaggregate level. These

findings can be used for assessing the mode shifts for various groups of TOD users.

These results will aid planners while planning transport for TODs.

The results from this research provide evidence to support claims in the literature of

the travel efficiency of TODs by comparing South East Queensland Travel Survey

(SEQTS) dataset with case study data set. The outcome will help local government to

support TOD implementation locally.

Review of the research suggested that most focus was on the rail based TODs; very

little evidences were found for bus or Bus Rapid Transit (BRT) based TODs. The case

study TOD considered a bus or BRT based TOD and assessed its performance.

11.6 Limitations of this research

Although the scope of this research was limited to evaluating the transport impacts of TODs,

some limitations to the methodology applied for assessing KGUV were noted. These

limitations are listed below.

One major limitation of this study is it does not separate the intrazonal and inter zonal

trips. This means the travel surveys do not consider the multiple activities undertaken

by the visitors within KGUV, which give rise to intrazonal trips.

Flowing from the previous point, there is a possibility of double counting of trips, for

example, the trip by a resident is separately noted as a shopping trip when the

shoppers’ survey was conducted, but it may also be considered as a trip in the

residents’ surveys. Correlation was not established between the different surveys.

This methodology was demonstrated through one TOD only; the outcomes may be

used as indicators for planning future TODs, but needs refinement through more

studies for them to be used in general application.

The traffic counts conducted for KGUV were for a limited number of hours of the

day. Full cordon surveys for 24 hour period were not conducted for determining the

total traffic generation. Hence, this study provided trip rates for the peak periods only.



The models for determining travel mode of TODs include the complete dataset. The

mode choice models were not developed by removing the responses obtained from

captive users.

11.7 Areas of further research

While conducting this research for evaluating TODs from a transport pint of view, several

research areas were observed. These areas are listed below, which provide directions for

further future research. Points 1 to 10 denote areas of future research related to application of

generalised methodology developed in this research and points 11 to 17 present areas of

future research identified for gaining more understanding of travel at TODs and are not

covered in this research.

1. The methodology proposed in this research should be refined by applying it to various

TODs and a framework for evaluating TODs should be developed based on the

experiences of more TODs to reach more generalised conclusions.

2. This research provides an insight into travel behaviour of TOD residents and visitors.

To test the methodology proposed in this research and verify the TOD performance,

various TODs should be assessed. The performances for TODs of different scale

which are located at various distances from CBDs need to be considered.

3. In addition to the comparison of traffic generation with the standard trip rates, the

traffic generation at a TOD should also be compared with a similar sized non TOD

development to determine actual variation in traffic flow.

4. The ITE (2008) and RTA (2002) trip generation guidelines do not have trip rates for

student accommodation, and in addition the RTA guidelines do not have any

specifications for education land uses. So these areas need to be explored in future for

a better understanding under Australian conditions.

5. The various land uses at TODs do not act as stand alone. Hence, to obtain accurate

traffic generation, standard guidelines for total traffic reduction due to a TOD need to

be set. Every TOD is different; hence the standards for total traffic reduction will be

more appropriate than obtaining trip rates for individual land uses. The parking

requirements need to be studied separately, as this will be affected due to reduced

traffic generation.

6. The quality of service for comfort and convenience should be included in the

selection criteria for a TOD, along with criteria for walking and cycling infrastructure.



7. The travel of TOD users can be compared with the households having similar

demographic characteristics. Further comparison of Journey to Work (JTW) data by

suburbs can be conducted to study the work travel at a TOD.

8. A suitable method for assessing statistical significance of the comparative analysis

needs to be developed to strengthen the outcomes.

9. More residents’ data should be collected and explored by considering various trip

types.

10. The personal characteristics are not considered as measure for transport evaluation

although they found to have influence on the travel demand. Hence for future TOD

evaluation, the personal characteristics should be included as the measures of

transport evaluation.

11. This research does not investigate the characteristics of intrazonal trips and

multipurpose trips undertaken by various TOD user groups. So these aspects need to

be studied in detail for understanding intrazonal trip making patterns at TODs. This

will lead to study distribution of trips at a TOD.

12. To gain better understanding study of the travel at a TOD in detail, cross tabulation

analysis of the travel data for each user group at a TOD should be undertaken.

13. Before and after studies should be conducted for assessing the transport impacts of

TOD development.

14. This study considers personal and transit characteristics, but does not consider the

effect of neighbourhood characteristics. The effect of neighbourhood characteristics

on travel of TOD users should be studied in future research while assessing TODs.

15. Appropriate travel demand models for TODs including mode specific mode choice

models for travel at TODs should be developed.

16. The effect of location of TOD in terms of its distance from CBD and other key centres

should be assessed.

17. Various TODs should be studied to determine the amount of self containment such

that guidelines for placing appropriate land uses together may be developed.

11.8 Chapter close

This chapter documented the main conclusions of the study along with applications of this

research and future areas for investigation. This completes documentation of all aspects of

this research.




References

Abdel-Aty, M. A. (2003). Hybrid distribution and response techniques for an origin-

destination travel survey. Institute of Transportation Engineers, ITE Journal, 73(2), 22–

27.

ABS (2007). 9314.0 – Sales of new motor vehicles, Australia, Sep 2007, Australian

Government, Australian Bureau of Statistics.

Aditjandra, P.T., Mulley, C. and Nelson, J.D. (2009). Neighbourhood design impact on travel

behaviour: A comparison of US and UK experience. Journal of the Massachusetts

Institute of Technology (MIT), Department of Urban Studies and Planning, Planning for

Sustainable Transportation: an International Perspective, 9, 28 – 52.

AGO (1998). The national greenhouse strategy: Strategic framework for advancing

Australia's greenhouse response. Canberra ACT: Australian Greenhouse Office.

AGO (2006). 2006 Tracking to the Kyoto target: Australia’s greenhouse emissions trends

1990 to 2008–2012 and 2020. Canberra ACT: Australian Greenhouse Office,

Department of the Environment and Heritage.

AGO (2007). National greenhouse gas inventory: analysis of trends and greenhouse

indicators 1990 to 2005, Australian Greenhouse office, Australian Government,

Department of the Environment and Water resources.

Alder, T., Rimmer L., and Carpenter, D. (2002). Use of internet-based household travel diary

survey instrument. Transportation Research Record, 1804, 134–143.

Allen, W. G., and Perincherry, V. (1996). Two-stage vehicle availability model.

Transportation Research Record, 1556, 16–21.

Anderson, Tracy, Rofique, Josi, Pickering, Kevin, and Tipping, Sarah. (2010). National

Travel Survey 2009: Technical Report, Prepared for the Department for Transport,

United Kingdom, June 2010. Accessed on 19th

September 2010 at

http://www.dft.gov.uk/pgr/statistics/datatablespublications/nts/technical/nts2009technical

.pdf.

http://www.dft.gov.uk/pgr/statistics/datatablespublications/nts/technical/nts2009technical.pdf

http://www.dft.gov.uk/pgr/statistics/datatablespublications/nts/technical/nts2009technical.pdf



Arrington, G. B., and Cervero, R. (2008). TCRP Report 128: Effects of TOD on housing,

parking, and travel. Washington, D.C.: Transportation Research Board.

Arrington, G. B., and Sloop, K. I. (2009). New transit cooperative research program research

confirms transit–oriented developments produce fewer auto trips. Institute of

Transportation Engineers, ITE Journal, 79(6), 26–29.

Badoe, D. A., and Miller, E. J. (2000). Transportation–land–use interaction: Empirical

findings in North America, and their implications for modelling. Transportation

Research Part D, 5(4), 235–263.

Bajracharya, B., and Khan, S. (2005). Planning for sustainable transit oriented development:

the case of south east Queensland. The Queensland University of Technology Research

Week International Conference. Brisbane, Australia, Queensland University of

Technology, Australia.

Balcombe. R., Mackett, R., Paulley, N., Preston, J., Shires, J., Titheridge, H., Wardman, M.,

and White, P. (2004). TRL Report 593 The demand for public transport: a practical

guide. Transport Research Laboratory (TRL) Limited. Accessed on 10th

October 2010 at

www.demandforpublictransport.co.uk.

BCC (2000) Brisbane city plan 2000. Brisbane city council, Brisbane (Qld) council

http://www.brisbane.qld.gov.au/masterview/masterplan/enquirer/publishR.aspx?page

Accessed on 26 May 2008.

Bewick, V., Cheek L., and Ball, J. (2005). Statistics review 14: Logistic regression. Critical

Care, 9(1), 112–118.

Bruton, M. J. (1985). Introduction to transportation planning. Tiptree, Essex, Great Britain:

Anchor Brendon Ltd.

BTA. (2007). Bicycle transportation alliance. Accessed on 7 November 2007.

http://www.multiline.com.au/%7Ebta/goodidea/trends.shtml.

BTRE (2007). Estimating urban traffic and congestion cost trends for Australian cities,

working paper no. 71, Canberra, Australia: Australian Government, Bureau of Transport

and Regional Economics, Department of Transport and Regional Services.

http://www.demandforpublictransport.co.uk/

http://www.brisbane.qld.gov.au/masterview/masterplan/enquirer/publishR.aspx?page

http://www.multiline.com.au/~bta/goodidea/trends.shtml

References


Burke, M., and Brown, A. L. (2005). Rating the transport sustainability of transit oriented

developments: will developments achieve objectives? Transit Oriented Development –

Making it happen. Perth, Western Australia, 5–8 July 2005.

Calthrope, P. (1993). The next American metropolis: Ecology, community, and the American

dream. New York: Princeton Architectural Press.

Cao, X., Mokhtarian P. L., and Handy, S. L. (2007). Do changes in neighborhood

characteristics lead to changes in travel behavior? A structural equations modeling

approach. Transportation, 34(5), 535–556.

Cao, X., Mokhtarian P. L., and Handy, S. L. (2009). The relationship between the built

environment and nonwork travel: A case study of Northern California. Transportation

Research Part A, 43(5), 548–559.

Cervero, R. (1991). Land uses and travel at suburban activity centres. Transportation

Quarterly, 45(4), 479–491.

Cervero, R. (1996). Mixed land-uses and commuting: Evidence from the American housing

survey. Transportation Research Part A, 30(5), 361–377.

Cervero, R. (2006). Alternative approaches to modelling the travel demand impacts of smart

growth. Journal of the American Planning Association, 72(3), 285–295.

Cervero, R., and Day, J. (2008). Suburbanization and transit-oriented development in China.

Transport Policy, 15(5), 315–323.

Cervero, R., and Gorham, R. (1995). Commuting in transit versus automobile

neighbourhoods. Journal of the American Planning Association, 61(2), 210–225.

Cervero, R., and Radisch, C. (1996). Travel choices in pedestrian versus automobile oriented

neighborhoods. Transport Policy, 3(3), 127–141.

Chatman, D. G. (2003). How density and mixed uses at the workplace affect personal

commercial travel and commute mode choice. Transportation Research Record, 1831,

193–201.



Coevering, P. v. d., and Schwanen, T. (2006). Re-evaluating the impact of urban form on

travel patterns in Europe and North-America. Transport Policy, 13(3), 229–239.

Crevo, C. C., Niedowski, R. S., and Scott, D. J. (1995). Design and conduct of a statewide

household travel survey in Vermont. Transportation Research Record, 1477, 26–30.

Currie, G. (2005). Strengths and weakness of bus in relation to transit oriented development.

Transit Oriented Development–Making it Happen. Perth, Western Australia, 5–8 July

2005.

Deakin, E., Bechtel, A., Crabbe, A., Archer, M., Cairns, S., Kluter, A., Leung, K., and Li, J.

(2004). Parking management and downtown land development in Berkeley, California.


Department of Housing (2008). KGUV Development Opportunities. Queensland Government

and Queensland University of Technology, Accessed on 16th

July 2008 at

http://www.public-housing.qld.gov.au/partnerships/pdf/dev_opps.pdf.

Dill, J. (2008). Transit use at transit-oriented developments in Portland, Oregon, area.


Dillman, D. A. (2006). Mail and Internet Surveys: The Tailored Design Method, 2007

Update, John Wiley & Sons, Inc., Hoboken.

Dittmar, H., and Ohland, G. (2004). The new transit town: best practices in transit oriented

development. Washington, DC: Island Press.

Dock, F. C., and Swenson, C. J. (2003). Transit–oriented urban design impacts on suburban

land use and transportation planning. Transportation Research Record, 1831, 184–192.

DOT (2005). The First South African National Household Travel Survey 2003. Department

of Transport, Republic of South Africa. Accessed on 19th

September 2010 at

http://transport.dot.gov.za/projects/nts/techreport/index.html.

Ewing, R., Dumbaugh, E., and Brown, M. (2001). Internalizing travel by mixing land uses.


Field, A. (2009). Discovering statistics using SPSS. London: Sage Publications limited.

http://www.public-housing.qld.gov.au/partnerships/pdf/dev_opps.pdf

http://transport.dot.gov.za/projects/nts/techreport/index.html

References


Frank, L., Bradley, M., Kavage, S., Chapman, J. and Lawton, T. K. (2008). Urban form,

travel time, and cost relationships with tour complexity and mode choice.

Transportation, 35(1), 37–54.

Fujii, S., and Gӓrling T. (2003). Application of attitude theory for improved predictive

accuracy of stated preference methods in travel demand analysis. Transportation

Research Part A, 37(4), 389–402.

Gatzlaff, D., Sirmans, S., Carver, D. (1999). Marketing transit oriented development–Phase

III: Final Report, Primary factors in marketing TOD's as successful real estate ventures.

Tallahassee, Florida, The Marketing Institute: 21.

Gilbert, D., and Ginn, S. (2001). Transit oriented sustainable developments, Built

Environment Research Unit, Building Division, Queensland Department of Public

Works.

Giuliano, G., and Dargay, J. (2006). Car ownership, travel and land use: A comparison of the

US and Great Britain. Transportation Research Part A, 40(2), 106–124.

Goldenberg, L. (1998). Choosing a household survey method: Results of Dallas–Fort worth

pretest. Transportation Research Record, 1625, 86–94.

Government of South Australia (2010). The 30 – year plan for Greater Adelaide – A volume

of the South Australian planning strategy. Department of Planning and Local

Government. http://www.dplg.sa.gov.au/plan4adelaide/index.cfm Accessed on 15th

September 2010.

Gray, M., and Bunker, J. M. (2005). Kelvin Grove Urban Village regional accessibility by

active transport. Technical Report JB1205, School of Urban Development, Queensland

University of Technology.

Greenwald, M. J. (2006). The relationship between land use and intrazonal trip making

behaviors: Evidence and implications. Transportation Research Part D, 11(6), 432–446.

Grengs, J. (2004). Measuring change in small – scale transit accessibility with geographic

information systems. Transportation Research Record, 1887, 10–17.

http://www.dplg.sa.gov.au/plan4adelaide/index.cfm



Hair, J. F., Black, W. C., Babin, B. J., Anderson, R. E., and Tatham, R. L. (2006).

Multivariate data analysis. New Jersey: Pearson Education Inc.

Handy, S. (1996). Methodologies for exploring the link between urban form and travel

behavior. Transportation Research Part D, 1(2), 151–165.

Hess, D. B., and Ong, P. M. (2002). Traditional neighborhoods and automobile ownership.


Higgins, T. J. (1993). Parking requirements for transit–oriented developments.


ITE (2008). Trip Generation 8th

Edition. Institute of Transportation Engineers, Washington,

D.C., Volume 1 to 3.

James, B. (2005). Evolution of the importance of TOD in South East Queensland. Transit

Oriented Development – Making it happen. Perth, Western Australia, 5–8 July 2005.

Kerr, J., Frank, L., Sallis, J. F., and Chapman, J. (2007). Urban form correlates of pedestrian

travel in youth: Differences by gender, race–ethnicity and household attributes.

Transportation Research Part D, 12(3), 177–182.

Khattak, A. J., and Rodriguez, D. (2005). Travel behavior in neo-traditional neighborhood

developments: A case study in USA. Transportation Research Part A, 39(6), 481–500.

Kim, S., Ulfarsson, G. F., and Hennessy, J. T. (2007). Analysis of light rail rider travel

behavior: Impacts of individual, built environment, and crime characteristics on transit

access. Transportation Research Part A, 41(6), 511–522.

Kinnear, P. R., and Gray, C. D. (2009). SPSS 16 made simple. Hove, New York: Psychology

Press.

Kitamura, R. (1988). An evaluation of activity-based travel analysis. Transportation, 15(1-2),

9–34.

Korimilli, M. S., Pendyala, R. M., and Murakami, E. (1998). Metaanalysis of travel survey

methods. Transportation Research Record, 1625, 72–78.

References


Lee, C., and Moudon, A. V. (2006). The 3Ds + R: Quantifying land use and urban form

correlates of walking. Transportation Research Part D, 11(3), 204–215.

Lee, Y., Washington, S., and Frank, L. D. (2009). Examination of relationships between

urban form, household activities, and time allocation in the Atlanta Metropolitan Region.

Transportation Research Part A, 43(4), 360–373.

Lin, J. J., and Gau, C. C. (2006). A TOD planning model to review the regulation of

allowable development densities around subway stations. Land Use Policy, 23(3), 353–

360.

Lin, J.–J., and Shin, T.–Y. (2008). Does transit-oriented development affect metro ridership?

Evidence from Taipei, Taiwan. Transportation Research Record, 2063, 149–158.

Lu, Y., Kawamura, K., and Zellner, M. L. (2008). Exploring the influence of urban form on

work travel behavior with agent-based modeling. Transportation Research Record, 2082,

132–140.

Lund, H. (2006). Reasons for living in a transit-oriented development, and associated transit

use. Journal of the American Planning Association, 72(3), 357–366.

McCormack, E., Rutherford, G. S., and Wilkinson, M. G. (2001). Travel impacts of mixed

land use neighborhoods in Seattle, Washington. Transportation Research Record, 1780,

25–32.

McMillan, T. E. (2007). The relative influence of urban form on a child’s travel mode to

school. Transportation Research Part A, 41(1), 69–79.

Mepham, D. (2005). Transit oriented development – Options for Brisbane. Transit Oriented

Development – Making it Happen. Perth, Western Australia, 5–8 July 2005.

Messenger, T., and Ewing, R. (1996). Transit-oriented development in the Sun Belt.


Miller, E. J., and Ibrahim, A. (1998). Urban form and vehicular travel: Some empirical

findings. Transportation Research Record, 1617, 18–27.



Miller, E. J., Kriger, D. S., and Hunt, J. D. (1999). TCRP Report 48: Integrated urban models

for simulation of transit and land use policies: Guidelines for implementation and use.

Transportation Research Board, National Research Council, Washington, D.C.: 40.

MOT (2006). Travel survey results 2006. Land Transport and Safety Authority, Ministry of

Transport, New Zealand. Accessed on 19th September 2010 at

http://www.transport.govt.nz/research/Documents/travel-survey-results-2006.pdf.

Muley, D. S., Bunker, J. M. and Ferreira, L. (2007). Evaluating transit Quality of Service for

Transit Oriented Development (TOD). 30th Australasian Transport Research Forum.

Melbourne, Australia, 25–27 September 2007.

Muley, D. S., Bunker, J. M. and Ferreira, L. (2008). Conducting visitor travel survey for a

TOD – case study from South East Queensland. 31st Australasian Transport Research

Forum. Gold Coast, Australia, 2–3 October 2008, 223–238.

Muley, D. S., Bunker, J. M. and Ferreira, L. (2009). Assessing the Impact of Transit and

Personal Characteristics on Mode Choice of TOD Users. 12th TRB National

Transportation Planning Applications Conference, Houston, Texas, 17–21 May 2009.

Muley, D. S., Bunker, J. M. and Ferreira, L. (2009). Investigation into Travel Modes of TOD

Users: Impacts of Personal and Transit Characteristics. International Journal of ITS

Research. Special Issue: Sustainable Transportation Systems, 7(1), 3–13.

Muley, D. S., Bunker, J. M. and Ferreira, L. (2009). Transit Oriented Developments: Results

from a Travel Survey. The Second Infrastructure Theme Postgraduate Conference.

Brisbane, Australia: Queensland University of Technology, 26th March 2009, 273–284.

Muley, D. S., Bunker, J. M. and Ferreira, L. (2009). User characteristics of transit oriented

developments: the case of Kelvin Grove Urban Village. In: Yigitcanlar T., Sustainable

Urban and Infrastructure Development: Management, Engineering and Design.

Muley, D. S., Bunker, J. M., Ferreira, L. (2008). Investigation into Travel Modes for Transit

Oriented Development: Employees' Travel Survey of Kelvin Grove Urban Village.

http://eprints.qut.edu.au/14319/1/14319.pdf.

http://www.transport.govt.nz/research/Documents/travel-survey-results-2006.pdf

http://eprints.qut.edu.au/14319/1/14319.pdf

References


Muley, D. S., Bunker, J. M., Ferreira, L. (2009). Investigation into Travel of Transit Oriented

Development Residents’ of Kelvin Grove Urban Village.

http://eprints.qut.edu.au/28781/1/c28781.pdf.

Muley, D. S., Bunker, J. M., Ferreira, L. (2010). Kelvin Grove Urban Village: Travel survey

for school students. http://eprints.qut.edu.au/37691/1/School_students_survey.pdf.

Murakami, E., and Watterson, W. T. (1992). The Puget Sound Transportation Panel after two

waves. Transportation, 19(2), 141–158.

Newman, P. (2005). Transit Oriented Development: An Australian overview. Transit

Oriented Development – Making it happen. Perth, Western Australia, 5–8 July 2005.

Newmark, G. L., Plaut, P. O., and Garb, Y. (2004). Shopping travel behaviours in an era of

rapid economic transition: Evidence from newly built malls in Prague, Czech Republic.


NIPC (2001). Building, a regional framework: Transit – oriented development. Northeastern

Illinois Planning Commission.

OESR (2007). Petrol: Average retail price (a) by type, Brisbane, 1997 to 2007, Queensland

Government, Office of Economic and Statistical Research.

Ortuzar, J. d. D., and Willumsen, L. G. (1994). Modelling transport. Chichester; New York:

Wiley.

Parker, T., McKeever, M., Arrington, GB and Heimer, J. S. (2002). Statewide transit-

oriented development study: Factors for success in California, Final Report by Business,

transportation and housing agency, California department of transportation.

PARTS (2010). Perth and Regions Travel Survey: A Survey of Day-to-Day Travel. Data

Analysis Australia, Government of Western Australia. Accessed on 26th

September 2010

at http://www.daa.com.au/case-studies/parts/.

Poorman, J. P., and Stacey, D. J. (1984). Self-administered mailback household travel survey.


http://eprints.qut.edu.au/28781/1/c28781.pdf

http://eprints.qut.edu.au/37691/1/School_students_survey.pdf

http://www.daa.com.au/case-studies/parts/



Queensland Government (2005). South East Queensland regional plan 2005-2026,

Queensland Government, Office of Urban Management, Department of Local

Government, Planning, Sport and Recreation.

Queensland Transport (1999). Shaping Up 2nd edition. Shaping urban communities to

support public transport, cycling and walking in Queensland. Brisbane: Queensland

Transport.

Rajamani, J., Bhat, C. R., Handy, S., Khaap, G., and Song, Y. (2003). Assessing impact of

urban form measures on nonwork trip mode choice after controlling for demographic and

level-of-service effects. Transportation Research Record, 1831, 158–165.

Richardson, A. J., Ampt, E. S., and Meyburg, A. H. (1995). Survey methods for transport

planning. Melbourne, Australia: Eucalyptus Press.

Rodier, C. J., Johnston, R. A., and Abraham, J. E. (2002). Heuristic policy analysis of

regional land use, transit, and travel pricing scenarios using two urban models.

Transportation Research Part D, 7(4), 243–254.

Rossi, T. F., Lawton, K., and Kim, K. H. (1993). Revision of travel demand models to enable

analysis of a typical land use pattern. 4th National Conference on Transportation

Planning Methods Applications-Volume II. A Compendium of Papers, JerryM. Faris.

RTA (2002). Guide to traffic generating developments. NSW: Roads and Traffic Authority.

2.2.

Rural and Regional Affairs and Transport References Committee. (2006). Australia’s future

oil supply and alternative transport fuels. Interim Report ISBN 0 642 71706 0.

Russell, R. (2006). Transit’s potential to improve the urban situation in Perth. Transport

Engineering in Australia, 10(2), 87–110.

Sadownik, B., and Jaccard, M. (2001). Sustainable energy and urban form in China: The

relevance of community energy management. Energy Policy, 29(1), 55–65.

Schlossberg, M., and Brown, N. (2004). Comparing transit–oriented development sites by

walkability indicators. Transportation Research Record, 1887, 34–42.

References


SEQTS (2008). South East Queensland Travel Survey 2006 – 2008: Survey procedures and

Documentation – Brisbane Survey 2006. Queensland Department of Main Roads,

Queensland Government, December 2008.


Documentation – Gold Coast Survey 2008. Queensland Department of Main Roads,

Queensland Government, December 2008.


Documentation – Sunshine Coast Survey 2007. Queensland Department of Main Roads,

Queensland Government, November 2008.

SEQTS (2009). South East Queensland Travel Survey 2006 – 2008 Confidentialised Unit

Record Data Tables User's Guide - Brisbane, Gold Coast and Sunshine Coast Surveys.

Queensland Transport, Queensland Government, Version 2.0, February 2009.

Sharp, J., and Murakami, E. (2005). Travel surveys: Methodological and technology–related

considerations. Journal of Transportation and Statistics, 8(3), 1–15.

Shih, T.-H., and Fan, X. (2008). Comparing response rates from web and mail surveys: A

meta-analysis Field Methods, 20(3), 249–271. Accessed on 8th August 2008 at

http://fmx.sagepub.com/cgi/content/abstract/20/3/249.

SIKA (2007). RES 2005 – 2006 The National Travel Survey. Swedish Institute for Transport

and Communications Analysis, Sweden. Accessed on 19th September 2010 at

http://www.sika-institute.se/Doclib/2007/SikaStatistik/ss_2007_19_eng.pdf.

SKM (2006). Brisbane Strategic Transport Model – 2005 Update Brisbane City Council.

Brisbane: 1–59.

Smith, M. E. (1979). Design of small-sample home-interview travel surveys. Transportation

Research Record, 701, 29–35.

Souleyrette, R. R., Plazak, D. J., Strauss, T. R., and Andrle, S. J. (2001). Application of state

employment data to transportation planning. Transportation Research Record, 1768, 26–

35.

http://fmx.sagepub.com/cgi/content/abstract/20/3/249

http://www.sika-institute.se/Doclib/2007/SikaStatistik/ss_2007_19_eng.pdf



StatTrek (2009). AP Statistics Tutorial: Bias in Survey Sampling. Available at:

http://stattrek.com/AP-Statistics-2/Survey-Sampling-Bias.aspx?Tutorial=AP. Accessed

on 15 November 2009.

Steiner, R. L. (1998). Trip generation and parking requirements in traditional shopping

districts. Transportation Research Record, 1617, 28–37.

Stopher, P. R., and Greaves, S. P. (2007). Household travel surveys: Where are we going?

Transportation Research Part A, 41(5), 367–381.

Sun, X., Wilmot, C. G., and Kasturi, T. (1998). Household travel, household characteristics,

and land use: An empirical study from the 1994 Portland activity–based travel survey.


Tabachnick, B. G., and Fidell, L. S. (2007). Using multivariate statistics. USA: Pearson

Education, Inc.

TDC (2010). 2008/09 Household travel survey: Summary report, 2010 Release. Transport

Data Centre, NSW Government. Accessed on 3rd

October 2010 at

http://www.transport.nsw.gov.au/sites/default/file/tdc/documents/R2010-01-2008-09-

HTS-Summary-Report.pdf.

The Hornery Institute and Hassell (2004) Kelvin Grove Urban Village, Integrated master

plan, August 2004.

Tong, C. O., and Wong, S. C. (1997). The advantages of a high density, mixed land use,

linear urban development. Transportation, 24(3), 295–307.

Tooley, M. S. (1996). Incentives and rates of return for travel surveys. Transportation


TransLink (2010). TransLink transit authority annual report 2009–10. TransLink Transit

Authority, Queensland Government. Accessed on 11th

October 2010 at

http://events.translink.com.au/resources/about/2010-annualreport.pdf.

TRB (2003). TCRP Report 100 Transit Capacity and Quality of Service Manual 2nd Edition.

Washington D. C.: Kittelson & Associates, Inc, KFH Group, Inc, Parsons Brinckerhoff

Quade & Douglass, Inc and Hunter – Zaworski, K., Transportation Research Board.

http://stattrek.com/AP-Statistics-2/Survey-Sampling-Bias.aspx?Tutorial=AP

http://www.transport.nsw.gov.au/sites/default/file/tdc/documents/R2010-01-2008-09-HTS-Summary-Report.pdf

http://www.transport.nsw.gov.au/sites/default/file/tdc/documents/R2010-01-2008-09-HTS-Summary-Report.pdf

http://events.translink.com.au/resources/about/2010-annualreport.pdf

References


UCLA (2010). Regression with SPSS. UCLA: Academic Technology Services, Statistical

Consulting Group. Accessed on 16th

October 2010 at

http://www.ats.ucla.edu/stat/spss/webbooks/reg/chapter2/spssreg2.htm.

Waddell, P. (2002). UrbanSim: Modelling urban development for land use, transportation and

environmental planning. Journal of the American Planning Association, 68(3), 297–314.

Waddell, P., and Nourzad, F. (2002). Incorporating nonmotorised mode and neighborhood

accessibility in an integrated land use and transportation model system. Transportation


Warner, R. M. (2008). Applied statistics: From bivariate through multivariate techniques.

United States of America: Sage Publications, Inc.

Winder, P. D. (1994). Integrating planning for transport, land use and the environment: The

Auckland strategic planning model. Proceedings of the 17th ARRB Conference, Part 7,

ARRB Transport Research Ltd.

Zakaria, T. (1986). Employee transportation survey for centre city Philadelphia.


Zegras, P. C. (2004). Influence of land use on travel behavior in Santiago, Chile.


Zhang, L., and Levinson, D. (2004). Agent-based approach to travel demand modelling:

Exploratory analysis. Transportation Research Record, 1898, 28–36.

Zhang, M. (2005). Exploring the relationship between urban form and nonwork travel

through time use analysis. Landscape and Urban Planning, 73(2-3), 244–261.

www.maps.google.com.au

www.translink.com.au

www.whereis.com

http://www.ats.ucla.edu/stat/spss/webbooks/reg/chapter2/spssreg2.htm



http://www.whereis.com/




Appendix A

List of publications

Refereed journal paper

Muley, D. S., Bunker, J. M. and Ferreira, L. (2009). Investigation into Travel Modes of TOD

Users: Impacts of Personal and Transit Characteristics. International Journal of ITS

Research. Special Issue: Sustainable Transportation Systems, 7(1), 3–13.

Refereed conference papers

Muley, D. S., Bunker, J. M. and Ferreira, L. (2009). Transit Oriented Developments: Results

from a Travel Survey. The Second Infrastructure Theme Postgraduate Conference. Brisbane,

Australia: Queensland University of Technology, 26th March 2009, 273–284.

Muley, D. S., Bunker, J. M. and Ferreira, L. (2008). Conducting visitor travel survey for a

TOD – case study from South East Queensland. 31st Australasian Transport Research

Forum. Gold Coast, Australia, 2 – 3 October 2008, 223–238.

Muley, D. S., Bunker, J. M. and Ferreira, L. (2007). Evaluating transit Quality of Service for

Transit Oriented Development (TOD). 30th Australasian Transport Research Forum.

Melbourne, Australia, 25–27 September 2007.

Book chapter

Muley, D. S., Bunker, J. M. and Ferreira, L. (2009). User characteristics of transit oriented

developments: the case of Kelvin Grove Urban Village. In: Yigitcanlar T., Sustainable Urban

and Infrastructure Development: Management, Engineering and Design.

Conference presentation

Muley, D. S., Bunker, J. M. and Ferreira, L. (2009). Assessing the Impact of Transit and

Personal Characteristics on Mode Choice of TOD Users. 12th TRB National Transportation

Planning Applications Conference, Houston, Texas, 17–21 May 2009.


Deepti Muley 234


Appendix B

B.1 Classified traffic count data sheet

Date:

Time period: In 15 minutes interval

Name of recorder:

Approach

Vehicles coming in study area Vehicles going out of study area

1

occupant

cars

2

occupant

cars

3

occupant

cars

Pedestrians Bicycles Motor

cycles

1

occupant

cars

2

occupant

cars

3

occupant

cars

Pedestrians Bicycles Motor

cycles


B.2 Shoppers survey sheet

Date:

Time period:

Name of recorder:

Classification of age groups:

1 0 to 18 years

2 18 to 30 years

3 30 to 45 years

4 45 to 65 years

5 65 years and above

Mode of

transport

Car occupancy /

Bus route

Home

postcode

Frequency

(per week)

Age

group Occupation


Appendix C

Table C.1 AM peak ITE comparison for Kelvin Grove Urban Village

Land use ITE

code

Number

of studies ITE equation

Traffic based on ITE

guidelines Actual

counts

% difference with respect to

ITE trip rates

Average

rate

Regression

equation Average rate

Regression

equation

Residential

220

231

252

78

7

8

𝑇 = 0.49 𝑋 + 3.73

𝐿𝑛 𝑇 = 0.90𝐿𝑛 𝑋 − 0.07

𝑇 = 0.19 𝑋 − 13.86

253 278

Retail 850 5 NA 146 146*

Education

760

550

550

530

28

6

4

68

𝐿𝑛 𝑇 = 0.82𝐿𝑛 𝑋 + 0.33

𝑇 = 0.21 𝑋 − 69.14

𝐿𝑛 𝑇 = 0.64𝐿𝑛 𝑋 + 2.08 NA

895 875

Office 710 163 𝐿𝑛 𝑇 = 0.86𝐿𝑛 𝑋 + 0.24 168 214

Total 1462 1513 1078 26 29

Note: *indicates regression equation not available hence value obtained using average rate is used.


Table C.2 PM peak ITE comparison for Kelvin Grove Urban Village

Land use ITE

code

Number

of studies ITE equation

Traffic based on ITE

guidelines Actual

counts

% difference with respect to

ITE trip rates

Average

rate

Regression

equation

Average

rate

Regression

equation

Residential

220

231

252

90

6

8

𝑇 = 0.55 𝑋 + 17.65

𝐿𝑛 𝑇 = 0.89𝐿𝑛 𝑋 − 0.07

𝑇 = 0.24 𝑋 − 16.45

306 390

Retail 850 40 𝐿𝑛 𝑇 = 0.61𝐿𝑛 𝑋 + 3.95 426 498

Education

760

550

550

530

29

8

4

40

𝐿𝑛 𝑇 = 0.81𝐿𝑛 𝑋 + 0.40

𝑇 = 0.19 𝑋 + 118.58

𝐿𝑛 𝑇 = 0.652𝐿𝑛 𝑋 + 3.12 NA

920 1158

Office 710 173 𝐿𝑛 𝑇 = 0.37𝐿𝑛 𝑋 + 60.08 161 250

Total 1813 2296 1051 42 54

Table C.3 Peak period RTA comparison for Kelvin Grove Urban Village

Land use Max rate Traffic based on RTA guidelines Actual counts % difference with respect to RTA trip rates

Residential 0.5

0.2

280

Retail 12.3 464

Education* 980

Office 33.33 133

Total 1857 1078 42

Note: * indicates use of ITE (2008) values


Appendix D

Perceptions of Kelvin Grove Urban Village’s users

D.1 Introduction

Chapter 7 outlined the demographic and travel characteristics of Kelvin Grove Urban Village

(KGUV) users. To further explore the opinions of the KGUV users, this appendix documents

the perceptions of the various groups at KGUV about the public transport at KGUV and

KGUV as a TOD. The first section explains the perception of KGUV users about the existing

public transport service at KGUV and lists the suggestions and improvements. Later the

ratings of KGUV users on various aspects of KGUV are explained with the comments from

the respondents. In the following section, other development issues are mentioned with a

brief summary.

D.2 Determination of KGUV users’ perceptions

All the user groups surveyed were asked to specify their opinion or perception about the

public transport service and KGUV. The responses obtained from these questions were

analysed separately to gain the overall perception of KGUV from each user group. Same

parameters were provided and asked to rate on a five point rating scale. The lowest ranking

was assigned a value 1 and the highest tanking was assigned a value of 5 for the purpose of

analysis. Two outcomes were determined; the average of the ratings for each user group and

the average score for each parameter. The details of the outcomes are presented in the

following sections.

D.3 Perception about existing public transport

All the respondents were asked to rate their perception about the current public transport

service on a rating scale of one to five with one being very poor and five being excellent. An

option of not applicable was also provided if a respondent thought that he or she should not

answer the question or the parameter was not seen as appropriate. The current public

transport service at KGUV was rated based on the frequency of service, on time performance,

overall quality of service, waiting time, accessibility to stop, safety for passengers at stops,

public transport signage and cost of public transport. These indicators provided the perceived



quality of public transport service as opposed to the physical measures determined by the set

of guidelines.

To get an idea about the expected improvements in the public transport service, few KGUV

users were asked to specify the importance of the improvements required for the public

transport service. A rating scale was designated for this purpose with one indicating not

important and five representing extremely important or essential. The improvements were

ranked based on the minimum waiting time, more amenities at the station, good quality of

service, cost, walking time to and from the stop and frequency of service. The following

subsections report the responses of KGUV users for the perception about existing service and

the importance of the improvements.

D.3.1 Ratings of employees on existing public transport service

The professional employees were asked to rate their perception about current public transport

service as well as the importance of the improvements to it. These questions were not asked

to the retail shop employees due to the circumstantial constraints. Table D.1 presents the

results for the professional employee analysis. The individual average rating for the public

transport system ranged from 1 to 4.75 with an average of 3.3 out of 5. The highest rating

was obtained for safety and lowest rating was waiting time and on time performance.

Table D.1 Rating of current public transport by professional employees at KGUV

Description Very

poor Poor Average Good Excellent

Average

rating

Frequency of service 5.7 16.0 41.5 30.2 6.6 3.2

On time performance 9.5 20.0 42.9 22.9 4.8 2.9

Overall quality of service 2.8 16.0 43.4 35.8 1.9 3.2

Waiting time 7.6 22.9 40.0 27.6 1.9 2.9

Accessibility to stop 5.7 8.5 26.4 44.3 15.1 3.5

Safety for passengers at stop 2.8 5.7 26.4 51.9 13.2 3.7

Public transport signage 2.8 9.3 28.0 47.7 12.1 3.6

Cost of public transport 6.7 9.6 38.5 38.5 6.7 3.3

Note: The numbers represent the percentages for each ranking

D.3.1.1 Ratings of employees for improvements in public transport service

To improve the existing public transport the professional employees were asked to rate the

importance of various parameter. An overview of the responses is given in Table D.2. The

professional employees rated frequency of service as the most important factor which can

improve the public transport system. The improvements in the amenities at the station were

marked as the least priority parameter.

Perceptions of KGUV’s users


Table D.2 Improvements rating by professional employees

Description Not

important

Less

important

Medium

importance

Highly

important

Extremely

important

Average

rating

Minimum

waiting time

0.0 2.7 13.5 49.5 34.2 4.2

More

amenities at

station

17.3 40.4 26.0 13.5 2.9 2.4

Good quality

service

0.9 4.6 20.4 54.6 19.4 3.9

Should be

cheaper

2.9 19.0 29.5 29.5 19.0 3.4

Less walking

time to &

from the stop

7.5 28.0 27.1 21.5 15.9 3.1

Frequent

service

0.9 2.7 14.3 42.0 40.2 4.2


D.3.2 Ratings of students on existing public transport service

Table D.3 presents the proportion of the ratings for each rank for various variables. The

results were determined by combining the responses for each question from school students

as well as university students. The average rating for the existing public transport was 3.5 out

of five. The individual minimum average rating was two and the maximum rating was 4.9 out

of five. Similar to the professional employees, the safety was rated highest and waiting time

and on time performance was rated lowest. This was because of the excessive delays

observed while waiting for the bus during the peak periods, as the buses did not turn up on

the scheduled time and no information was provided to the users about the scheduled arrival.

Table D.3 Rating of current public transport by students at KGUV

Description Very poor Poor Average Good Excellent Average

rating




Waiting time 3.6 23.2 35.7 35.7 1.8 3.1








D.3.2.1 Ratings of students for improvements in public transport service

The university student students were asked to rate the improvements based on their views

while the school students were not asked this question because the questionnaire was

simplified for them and this question was removed from the survey forms. Table D.4 shows

the variation in the university students’ perception. The improvements ratings observed

similar trend in results as the professional employees. This indicated that although different

age and profile for these two user groups the priority parameters remain same.

Table D.4 Improvements rating by university students

Description Not

important

Less

important

Medium

importance

Highly

important

Extremely

important

Average

rating

Minimum

waiting time

1.1 1.1 21.3 46.1 30.3 4.0

More

amenities at

station

8.0 33 38.6 12.5 8.0 2.8

Good quality

service

0.0 12.5 30.7 35.2 21.6 3.7

Should be

cheaper

3.5 7.1 31.8 30.6 27.1 3.7

Less walking

time to &

from the stop

8.3 23.8 33.3 22.6 11.9 3.1

Frequent

service

0.0 0.0 11.2 41.6 47.2 4.4


D.3.3 Ratings of residents on existing public transport service

The results from the analysis for residents’ responses on current public transport system as

noted in Table D.5. The residents did not show much variation in the ratings of KGUV.

Accessibility to stop secured highest rating in addition to the safety and on time performance

and waiting time was biggest concern for the residents as well. These ratings indicate that the

visitors and residents at KGUV have similar perceptions about the public transport system.



Table D.5 Rating of current public transport by residents at KGUV

Description Very poor Poor Average Good Excellent Average

rating




Waiting time 3.8 10.3 30.8 43.6 11.5 3.5






D.3.4 Opinion about public transport system

These were the observations emphasized collectively by the respondents’.

Respondents placed a strong emphasis on the frequency and reliability of public

transport service to afford a mode shift from car to public transport. The travel time

difference and the absence of a direct public transport link1 were also pointed out as

the main reasons for using personalised modes of transport. This indicates that for a

TOD to be successful from a transport point of view, a good quality direct public

transport service is required from various destinations, not only from the CBD. Some

respondents preferred train over bus because of its on time performance or reliability.

As this area lies near the CBD, some respondents suggested having a “loop service”

(with minimum or no cost) running at a 15 minute interval from the CBD to KGUV

which also connects the nearby Roma street railway station to make KGUV more

attractive. (It is noted that, after this survey, TransLink implemented a high frequency

bus service, Route 66, along the busway system between Kelvin Grove busway

station to the east of KGUV, through the CBD via Roma Street railway station and on

to the inner southeast suburb of Woolloongabba. A peak hour service Route 933 was

also started to facilitate the KGUV users travel to and from city.)

A professional employee also suggested having strictly enforced parking restrictions

on the local streets in KGUV with increased parking cost at the work place and

incentives for employees who travel by sustainable modes of transport from the

organisations to promote the more sustainable modes of transport.

1 Brisbane has a hub and spoke public transport network with most services intersecting at the CBD. One often

needs to change service, particularly for buses, to access a destination on the other side of the CBD. The

imposition of a seat change has been reported to make public transport less attractive.



Shoppers suggested that the shopping centre should be an “all the time and one stop

shop” and have more retail outlets aimed at young customers.

The responses also indicated that on time performance and actual arrival of the bus at

scheduled time or some information which would provide them an idea about the bus

arrival were highlighted as important factors. This was linked to the over crowding

and passenger unhappiness as some passengers were unable to get on the bus. This

issue was prominent in peak times when visitors travelled to KGUV and also back

home after finishing their activity at KGUV. Access to the real time mobile

information was suggested as a solution for this.

The lack of public transport in outlying suburbs was also a parameter for concern and

reason for using personalised modes of transport. The employees analysed the cost

and benefits of a mode to determine the travel mode for them.

Brisbane has a public transport system with spoke pattern; all the buses, trains and

ferries are oriented to or from the CBD. This often requires a transfer from one

service to another. This adds the trip length by public transport and consequently the

travel time. This additional travel time was a prominent reason to distract users from

using public transport.

Although the campus shuttle service was very well received by the students and staff

members, there were some issues regarding the overcrowding and waiting time at KG.

An increase in the frequency was suggested for better service. A consistent service

throughout the year was desired rather than only during university times.

Some respondents perceived the cost of the public transport as expensive. Mostly, the

students were quite happy with the cost as they travelled on the concession ticket

which is half the price of a full fare.

Few respondents suggested having route maps at the bus stop and to show via sign on

the buses to guide users. There are route maps in the city area but these needs to be

extended to the outer suburbs or the places of public interest this will aid users if they

are unsure.

Some respondents also highlighted that they have infrequent / inadequate services

during off peak times and due to work commitments it was not always possible to

travel during peak times. The frequent service should be extended to off peak hours as

well in addition to the peak hours to cover everyone’s needs.



The accessibility to Roma Street and Normanby busway was a concern for all KGUV

users. The respondents were ready to walk for about 2km to access their places of

interest provided they had good infrastructure for walking as this will yield them

health benefits as well.

The respondents having young children were required to access various destinations

as they were involved in multiple activities which made use of existing public

transport unsuitable.

The respondents demanded better signage and better sun / rain protection at the bus

stop. So the bus stops should not only look aesthetically good but also need to

perform better in terms of its function. The strong sun was blinding the vision; it made

the passenger to miss the bus. So shelter is important. Decent shelter and accurate

timetables were amenities requested by a user. Additional amenities like more seating

and lighting at bus stops was requested.

KGUV has a bus stop located up the hill on Kevin Grove road when travelling

outbound from the city, so the pedestrians were required to walk back to access the

KGUV. As this stop did not have close proximity to KGUV the safety for respondents

in the evenings was a concern. This indicates that the public transport stops or staions

can be located at the centre of a TOD or on the edge of a TOD but not very away from

the TOD boundary.

One suggestion to improve public transport was, remove busses completely from the

equation and give the roads over to cars and private mini–van services such as in

developing countries and use trains for longer distances.

The public transport at KGUV was seen as much better than any other parts of

Brisbane. Having the extra 66 bus is extremely useful and has made it a lot easier to

get a seat on a bus. The connection from city was seen as very good.

The respondents demanded a better management of bus fleet as there were many 'Not

in service buses' that passed through KGUV Busway station. On a particular day one

respondent counted 11 “not in service" or "sorry bus full" busses in evening peak.

This wait time for the buses was the frustrating part for public transport users. There

were many students waiting for the buses during peak hour, the bus frequency should

be supplied to cater for the demand so as to reduce long waiting times and crowding

at the station platform.



D.4 Perception about KGUV

The users were asked to assess KGUV based on the various design parameters such as social

life, facilities or quality of facilities, environment or air quality, aesthetics of urban village,

quality of sidewalks, presence of sidewalks, transport infrastructure, parks and open spaces,

commercial activities, pedestrian friendly environment, parking facilities and overall rating of

KGUV. A rating scale of one to five was provided with one representing very poor and five

representing excellent. A not applicable option was provided similar to other rating scale

questions. The following subsections describe the perceptions of KGUV users.

D.4.1 Perception of employees

Table D.6 shows the perceptions of professional employees about KGUV. The KGUV

obtained an average individual rating of 3.4 with minimum of one and maximum of 4.8. The

lowest average rating was given to the parking facilities and highest was given to the footpath

availability. A professional employee described as KGUV has no soul.

Table D.6 Perception of employees about KGUV

Description Very


Average

rating

Social life 6.8 13.6 44.7 31.1 3.9 3.1

Facilities/quality of

facilities

1.7 5.8 27.5 49.2 15.8 3.7

Environment/Air quality 1.7 9.4 29.1 47.9 12.0 3.6

Aesthetics of urban village 5.1 6.8 33.1 39.8 15.3 3.5

Quality of sidewalks 6.7 3.4 22.7 45.4 21.8 3.7

Presence of sidewalks 5.0 4.2 22.5 46.7 21.7 3.8

Transport infrastructure 8.1 17.1 34.2 34.2 6.3 3.1

Parks and open spaces 2.5 12.7 29.7 42.4 12.7 3.5

Commercial activities 0.9 10.4 37.4 44.3 7.0 3.5

Pedestrian friendly

environment

7.5 13.3 22.5 44.2 12.5 3.4

Parking facilities 35.8 30.2 23.6 9.4 0.9 2.1

Overall rating of KGUV 3.4 7.7 29.9 47.9 11.1 3.6


D.4.2 Perception of students

Similar to the previous analysis, the perception analysis was also performed for a combined

dataset for school students and university students. Table D.7 outlines the details of the

rankings for students at KGUV. The individual average rating of 3.7 was obtained for the

design of KGUV. The KGUV was rated a minimum of 2.6 and a maximum of 4.8 by

students. Except for parking very few responses rated other parameters as very poor or poor.



Table D.7 Perception of students about KGUV

Description Very


Average

rating

Social life 0.0 8.0 39.3 42.0 10.7 3.6

Facilities/quality of facilities 0.0 0.0 23.9 54.9 21.2 4.0








Pedestrian friendly

environment 0.9 1.8 25.4 49.1 22.8 3.9




D.4.3 Perception of residents

All the residents’ responses were combined and analysed together to obtain their perspective

about KGUV. Table D.8 displays the percentage of responses for each ranking on the rating

scale. Similar to other user groups, the residents also gave lowest rating for the parking

facilities. All other aspects scored good scores for their design and provision. Overall, the

perception did not vary much by the user groups as usually seen in the travel characteristics.

One resident quoted, “Quality of life at Kelvin Grove Urban village is very excellent, shop,

dental clinic, gym, and parks are available, there is easy access to public transport. Now we

have market once a week”. The residents appreciated clustering of activities.



Table D.8 Perception of residents about KGUV

Description Very


Average

rating

Social life 6.3 10 27.5 46.3 10.0 3.5

Facilities/quality of facilities 1.2 4.9 21.0 54.3 18.5 3.9








Pedestrian friendly

environment

0.0 6.2 28.4 50.6 14.8 3.8




D.4.4 Comments about KGUV

The main concern in the development of this area was parking. More parking was

required & parking attendants need to enforce no parking more stringently.

Diverting the buses on Kelvin Grove road to pass through KGUV was suggested an

option to expose the KGUV to visitors by a retail shop employee as this might attract

more business opportunities for the retails shops.

Most of the KGUV users were happy with the standards of the KGUV; some

professional employees specially appreciated it. The growing popularity of the area

was a plus point. Provision of more places for social activities was suggested by

students.

To promote sport activities, a field for students was suggested for a casual sport.

One suggestion was giving right of way to pedestrians within the village and

minimum waiting time at the traffic lights for pedestrians to encourage walking.

A taxi rank was requested by employees as it was necessary for work purposes.

A professional employee suggested very high frequency public transport service

(similar to Hong Kong) to support KGUV. This suggestion should be considered by

keeping in mind the size respective transport system.

Some respondents were concerned about mixing land uses specially with the

education environment. A careful planning should be made before clustering of land

uses to maintain the performance of each land use.



Some respondents were having difficulties with the continual construction as it was

making the pedestrian movement difficult. It should be noted that the survey was

undertaken when the KGUV was underdevelopment. This problem will be resolved

after complete development of KGUV. However, this indicated the need for proper

planning of construction activities to minimize the interruption to a general user.

The residents appreciated the community feeling, centrally located shopping centre

and the proximity of bus stops. However, some students were concerned about the

pricing of the KGUV.

D.5 Other issues

The other development and transport issues were as noted below:

A suggestion was made to give pedestrians priority at intersections in KGUV to

reduce the walking time and make walking more attractive. Strong emphasis was

placed on having more pedestrian crossings and marked crosswalks.

Demand for bike facilities indicated that a good bike path is needed not only in

KGUV but also from home to the work place, as riding a bike on Brisbane’s roads

was described as “scary”.

One suggestion was to shift the residential area further away from commercial and

retail area.

The shoppers suggested that the shopping centre should be a “one stop shop”.

Currently, the Village Centre has a supermarket, food outlets, a pharmacy, a liquor

shop, a bar and coffee shops. He suggested to include a hair dresser or beauty salon,

deli and dress shop in the shopping centre. One more shopper also suggested having a

bigger scale shopping centre to offer more choices indicating the scope for expansion.

Another shopper found the shopping centre as convenient and accessible within a

short walk for people in the nearby suburbs.

The respondents were interested in having more greenery around the village to make

the development more attractive and provide shade during summer.

To obtain a better connection from the Roma Street station, a more environmental

friendly solution such as a free bike hire system was proposed where bikes can be

made available for a gold coin deposit to ride from KG to Roma Street and vice verse.

Some student residents quoted KGUV having good facilities but bit overpriced. The

issue of affordability should be considered especially for residents of KGUV.



D.6 Summary

The outcomes from this analysis indicated that the frequency of service was the most

important factor in deciding the mode choice. Hence, the effect of LOS for transit availability

was considered for investigating the travel modes of TOD users. The perceptions and

opinions will further aid practitioners while designing further TODs. KGUV observed good

rating except for its parking facilities. The inadequate parking facility was issue for all

KGUV users. However, these parking restrictions have encouraged the use of sustainable

modes of transport supporting that the reduced parking facilities at TODs contributes to the

sustainable mode share.