Dissertation

Short-Stay Car Parking Choice

Behaviour A Case Study of Cardiff City Centre

Chao Qi

Student Number: 1302064

MSc Transport and Planning

Supervisor: Dr. Dimitris Potoglou

1

Abstract:

Studying individuals’ parking choice behaviour can significantly contribute to the

parking policy making for urban areas. Based on a parking-user survey conducted in

the main short-stay parking places around the Cardiff city centre, this thesis provides

a thorough analysis of individuals’ parking behaviours. Through descriptive statistics

of respondents’ personal and travelling characteristics, an understanding of parking

users’ basic profiles has been gained. From investigating parking users’ perceptions

of the parking service, potential issues with regard to parking pricing, availability,

safety and information clarity have been identified. Relative suggestions have also

been proposed for improving the parking service in Cardiff city centre.

Through the application of chi-square tests, underlying relations across parking users’

profiles have been revealed. Compared with the working population, individuals who

travel to Cardiff city centre for shopping or leisure tend to bring companions with

them and park less frequently. Meanwhile, female parking users are more likely to

travel for shopping or leisure purposes, while male parking users tend to come for

work reasons.

As the core of the thesis, discrete choice models have been developed to acquire

parking users’ sensitivities to parking charge and parking availability. It is found that

£1 increase in parking fare or one-minute increase in searching time can generally

decrease the log odds of continuing to park by 1.492 and 0.226 separately. In

addition, ‘taste variation’ across various parking user groups has also been obtained.

In terms of increases in parking charge, individuals aged 25-44 tend to be more

sensitive, while people who travel with companions show a lower sensitivity.

Compared with working people, those who travel for non-work reasons are found to

be more sensitive to the increase in searching time. The above finding will provide

useful references to the parking policy improvement within the Cardiff context.

(Word Count: 19830)

2

Acknowledgements

Thank you to my supervisor Dr. Dimitris Potoglou for his professional guidance and

support throughout the dissertation. I have appreciated his good teaching in the MSc

Transport and Planning which has inspired me both in term of academic knowledge

and of career planning.

Thank you to the Principal Transport Planner Miriam Highgate and all the other

transport experts at Cardiff Council for throughout providing useful information and

suggestions for the research.

Thank you to the Public Affairs & Research Coordinator Carrie Drage and all the

other professionals at the British Parking Association for providing professional

parking knowledge for this thesis. In particular, I greatly appreciate the £1500 John

Heasman Bursary provided by the British Parking Association, which has enabled the

research to conduct a large scale high-quality parking-user survey.

Thank you to transport expert Richard Carr for sharing his valuable experience in the

parking-user survey design. Richard’s precious advice has helped the survey acquire

the most relevant and targeted data sets.

Thank you to my classmates who helped me to conduct the parking-user survey in

the Cardiff city centre. Finally, many thanks to all the respondents who have spent

their precious time supporting this research.

3

Table of Contents

Abstract: .................................................................................................................................... 1

Acknowledgements ................................................................................................................... 2

Table of Contents ...................................................................................................................... 3

Lists of Figures and Tables ......................................................................................................... 6

1. Introduction ....................................................................................................................... 9

1.1 Background of the study .............................................................................................. 9

1.2 Motivation of the study ............................................................................................. 13

1.3 Structure of the thesis ............................................................................................... 15

2. Literature Review ................................................................................................................ 16

2.1 Influential features in parking choice behaviour ....................................................... 16

2.2 Parking pricing ........................................................................................................... 20

2.3 Parking availability ..................................................................................................... 23

2.4 Parking policy and public attitudes ........................................................................... 25

2.5 Models applied to analyse parking choice behaviour ............................................... 27

2.6 Improvement of discrete choice modelling .............................................................. 31

2.7 Conclusion of literature review ................................................................................. 34

3. Methodology ....................................................................................................................... 37

3.1 Overview of the Methodology .................................................................................. 37

3.2 Conceptualisation of the study.................................................................................. 38

3.3 Design of the questionnaire ...................................................................................... 39

3.3.1 Profile of parking users ................................................................................... 39

3.3.2 Issues in parking service from users’ prospective .......................................... 41

4

3.3.3 Driving forces behind parking choice behaviour and sensitivities to parking

features ................................................................................................................... 42

3.4 Pilot survey ................................................................................................................ 45

3.4.1 Findings from the pilot survey ........................................................................ 45

3.5 Implication of the main survey .................................................................................. 49

3.6 Data analysis methods ............................................................................................... 50

3.6.1 Chi-square test ............................................................................................... 50

3.6.2 Logistic regression .......................................................................................... 50

3.7 Conclusion of the Methodology ................................................................................ 54

4. Data Analysis ....................................................................................................................... 55

4.1 Descriptive and frequencies statistics of parking users’ profiles .............................. 55

4.1.1 Gender and age group .................................................................................... 55

4.1.2 Originations .................................................................................................... 56

4.1.3 Travel purpose to Cardiff city centre .............................................................. 58

4.1.4 Travel group size ............................................................................................. 58

4.1.5 Parking duration ............................................................................................. 59

4.1.6 Parking frequency ........................................................................................... 60

4.1.7 Reasons for parking location choice ............................................................... 61

4.1.8 Searching time for parking spaces .................................................................. 62

4.1.9 Distance to destinations ................................................................................. 63

4.1.10 Park Mark ..................................................................................................... 63

4.1.11 Conclusion of parking users’ profiles............................................................ 64

4.2 Parking users’ perceptions to parking service ........................................................... 66

4.2.1 Rating of parking charge ................................................................................. 67

5

4.2.2 Rating of parking availability .......................................................................... 68

4.2.3 Ratings of information clarity and payment options...................................... 69

4.2.4 Ratings of personal safety and vehicle safety ................................................ 70

4.2.5 Conclusion of parking users’ perceptions and relative suggestions............... 71

4.3 Relations across parking profiles ............................................................................... 75

4.3.1 Travel purpose and travel group size ............................................................. 75

4.3.2 Travel purpose and parking frequency ........................................................... 76

4.3.3 Travel purpose and age .................................................................................. 78

4.3.4 Travel purpose and gender ............................................................................ 78

4.3.5 Travel purpose and parking duration ............................................................. 79

4.3.6 Conclusions of relations in parking users’ profiles ......................................... 80

4.4 Logistic Regression .................................................................................................... 81

4.4.1 Choosing frequencies of alternatives ............................................................. 81

4.4.2 Independent samples t-test ........................................................................... 83

4.4.3 Modelling parking users’ general sensitivity to parking features .................. 84

4.4.4. Modelling parking users’ taste variations to parking features ...................... 86

4.4.5 Modelling sensitivities of parking users with various characteristics ............ 91

4.5 Conclusion of the data analysis ................................................................................. 94

5. Conclusions and Recommendations ................................................................................... 96

References: ............................................................................................................................ 102

Appendices ............................................................................................................................ 109

6

Lists of Figures and Tables

List of Figures:

Figure 1.1 Study area: the Cardiff city centre

Figure 1.2 Survey location: short-stay parking at Cardiff City Hall (CF10 3ND)

Figure 1.3 Survey location: short-stay parking in St. Andrews Crescent (CF10 3DB)

Figure 2.1 The process of parking choice

Figure 3.1 Main survey location: Cardiff City Hall CF10 3ND

Figure 3.2 Main survey location: St. Andrews Crescent CF10 3DB

Figure 4.1 Percentages of parking users' age groups

Figure 4.2 Percentages of parking users’ origination localities

Figure 4.3 Percentages of parking users' origination natures

Figure 4.4 Percentages of parking users' travel purposes

Figure 4.5 Percentages of travel group size (adults)

Figure 4.6 Percentages of travel group size (children)

Figure 4.7 Distribution of short-stay parking users' parking duration

Figure 4.8 Percentages of individuals' parking frequency

Figure 4.9 Percentages of reasons for specific parking choices

Figure 4.10 Percentages of searching time for parking spaces

Figure 4.11 Percentages of walking time to destinations

Figure 4.12 Percentages of ratings for parking charge

7

Figure 4.13 Percentages of ratings for parking availability

Figure 4.14 Percentages of ratings for information clarity on payment machines

Figure 4.15 Percentages of ratings for payment options

Figure 4.16 Percentages of ratings for personal safety

Figure 4.17 Percentage of ratings for vehicle safety

Figure 4.18 Ratings of parking features

Figure 4.19 Choosing frequencies of alternatives

Figure 4.20 Percentages of parking users' choices if they choose not to park at the

current location

List of Tables:

Table 1.1 On street short-stay parking tariffs across city centres in the UK

Table 2.1 Studies on influential factors of car parking choice behaviour

Table 2.2 Studies modelling parking choice behaviour

Table 3.1 Questions and options related to parking users’ profiles

Table 3.2 Q10: Parking users’ prospective to the parking service in Cardiff city centre

Table 3.3 Attributes and levels for the stated preference questions

Table 3.4 Full factorial design result (with blocks)

Table 3.5 Attributes and alternatives for discrete choice questions

Table 4.1 Descriptive statistics of parking users’ basic profiles

Table 4.2 Summarisation of parking users’ ratings to parking features

Table 4.3 Work_or_Not * Companion Crosstabulation

Table 4.4 Work_or_Not * Parking Frequency Crosstabulation

8

Table 4.5 Work_or_Not * Agegroup Crosstabulation

Table 4.6 Work_or_Not * Gender Crosstabulation

Table 4.7 Work_or_Not * Parking_duration Crosstabulation

Table 4.8 Independent samples t-test result

Table 4.9 Binary regression result for general parking users

Table 4.10 Binary regression result for parking users with different genders

Table 4.11 Binary regression result for parking users with different travel purposes

Table 4.12 Binary regression result for parking users belonging to different age

groups

Table 4.13 Binary regression result for parking users with different travel group sizes.

Table 4.14 Summarisation of parking user groups’ different sensitivities to parking

features

Table 4.15 Binary regression result for parking users with various characteristics

(step 1)

Table 4.16 Binary regression result for parking users with various characteristics

(step 2)

9

1. Introduction

Based on a parking-user survey conducted in Cardiff city centre, this study provides

an analysis of individuals’ parking choice behaviour. Through descriptive statistics,

the study has identified the basic profiles of parking users in Cardiff city centre.

Meanwhile, their perceptions to parking features have also been analysed to

discover the potential issues in the parking service. Using chi-square test, deeper

relations in parking users’ profiles have been explored. It has been found that

generally parking users with different genders and travel group sizes tend to differ in

travel purposes. In addition, people with different travel purposes have shown

variations in parking frequencies. As the core of the thesis, logistic regression

models have been developed to test parking users’ sensitivities to parking pricing

and parking availability in the context of Cardiff city centre. Through modelling the

stated-preference data, it is found that increases in parking pricing and searching

time for parking spaces have definitely adverse impacts on the possibility of

continuing to park. A £1 increase in parking fare or a one-minute increase in

searching time will decrease the general log odds of continuing to park by 1.492 and

0.226 separately. Meanwhile, through adding parking users’ personal characteristics

into the modelling, the study has shown the varied sensitivities across different

parking user groups. In the context of Cardiff city centre, parking users aged 25-44

are more sensitive to the increase in parking charge, whereas travellers with

companions tend to be less sensitive to the changes in parking pricing. Furthermore,

parking users who travel for shopping or leisure purposes have shown a higher

sensitivity to the searching time for parking spaces. Based on the findings from the

analysis, several suggestions for improving the parking policy in Cardiff city centre

are made in the concluding section.

1.1 Background of the study

Parking is a necessary component of motor vehicle travels. Travellers usually attach

great importance to the conditions of parking (Clinch and Kelly 2003). Variations in

10

parking features can definitely influence individuals’ choices of parking location or

travel modes (Thompson and Richardson 1996). Understanding the impacts of

parking features can help transport planners to devise more efficient parking policies.

Parking policy is an essential tool for controlling the travel demand in urban areas.

Through parking pricing and parking supply management, such policy can discourage

the private vehicle usage and parking in urban areas (Feeney 1988). Thus, many

issues such as congestion and air pollution in urban areas can be alleviated.

Cardiff city centre, the study area of this thesis, is the central business district (CBD)

of Cardiff, Wales. It is the sixth most successful shopping hub in the UK, with a

shopper footfall of 55 million during 2008-2009(Bolter 2009). For visitors in motor

vehicles, the Cardiff Council provides various facilities to accommodate their parking

demands, such as park-and-ride, off-street car parks and on-street parking (Cardiff

Council, 2014a). Facing the challenge of large footfall, effective parking policies are

essential to the normal operation of Cardiff city centre. Although parking

enforcement to crack down on illegal parking has been applied in Cardiff city centre

(Guardian 2014), without the proper intervention of parking policy tools, congestions

and relative externalities will still jeopardise the sustainability of the city centre

development. As a part of the research, parking experts at the Cardiff Council has

been approached for their current perspectives on the parking policies and

suggestions for this specific thesis.

During the visit to the Cardiff Council, those parking experts communicated that the

Council is currently focusing on improving the on street short-stay parking policy in

Cardiff city centre. In Cardiff, there are three main on-street parking areas: City

Centre, Butetown and Cardiff Bay (Cardiff Council 2014b). In Cardiff city centre, the

long-stay on street parking is mainly provided for working people. People who travel

for shopping or leisure will usually choose the short-stay parking. The Council is

concerned more about improving parking policies for people who come to Cardiff

11

city centre for various purposes than for those who come only because of work. Thus,

the Council has recommended that this certain thesis should study the parking

choice behaviour of short-stay parking users in the context of Cardiff city centre.

Figure 1.1 Study area: the Cardiff city centre

Source: http://en.wikipedia.org/wiki/File:Cardiff_UK_location_map.svg

https://maps.google.co.uk/maps/ms?ie=UTF8&oe=UTF8&msa=0&msid=117253141803189732484.00047fcaf6

8f6f0a33950&dg=feature

The study is also supported by the British Parking Association (BPA). BPA has

provided a £1500 John Heasman Bursary for this thesis as the research fund (BPA

2014a). With the help of this bursary, the study is able to hire assistants to help

conduct a larger scale survey. Inspired by the Cardiff Council and the BPA, a

face-to-face parking user survey was conducted in July, 2014. This survey comprised

a total of 233 respondents from two on street short-stay parking places in Cardiff city

centre: Cardiff City Hall (Figure 1.2) and St. Andrews Crescent (Figure 1.3). A

satisfying data set related to individuals’ parking behaviour has been collected and

this will contribute to both this thesis and the future studies.

12

Figure 1.2 Survey location: short-stay parking at Cardiff City Hall (CF10 3ND)

Source: The Author

Figure 1.3 Survey location: short-stay parking in St. Andrews Crescent (CF10 3DB)

Source: The Author

13

Meanwhile, according to the previous work by Cardiff Council, the charge of on

street short-stay parking in Cardiff is relatively modest compared with city centres

across the UK (Table 1.1). In this thesis, discrete choice models will be applied to

acquire parking users’ sensitivities to the parking charge increase under this

currently modest tariff. In addition, individuals’ sensitivities to the parking availability

will also be obtained through the modelling.

Table 1.1 On street short-stay parking tariffs across city centres in the UK

Parking Time

City Centre

1h 2h 4h 5h

Westminster £3.25 £6.50

Hounslow £1.25 £2.60 £9.50

Birmingham(inner) £1.50 £3.75

Leicester £1.58 £2.15

Nottingham £2.00 £2.00 £6.25

Bristol(max stay 2 or 6

hours)

N/A £3.50 £7.00 N/A

Cardiff £1.70 £2.80 £3.60

Source: Cardiff Council

1.2 Motivation of the study

Studying parking behaviour is fundamental to the design of targeted parking policies

for the sustainable development of urban areas. However, it seems that very little

specific parking research has been done in the context of Cardiff. The Cardiff City

Centre Users Survey has collected some parking data on individuals’ usual parking

types and frequencies of encountering parking problems (Cardiff Council 2007).

However, a lot more is required in order to thoroughly analyse people’s parking

behaviour.

14

The first step in studying parking behaviour is acquiring parking users’ basic profiles

which relate to their personal and travel characteristics. The collected information

can help the research understand ‘who the parking users are in the background of

Cardiff city centre’. Meanwhile, parking users’ satisfaction levels to the parking

service can directly influence their willingness to obey the applied parking policies

and rules (Jones 1990). It is important for the research to obtain parking users’

perspectives on the parking service of Cardiff. Through this information, underlying

issues related to several parking features such as parking charge, availability and

parking safety, etc. can be identified in the context of Cardiff city centre. Thus,

relative improvement suggestions can be made for the efficient implementing of

parking policy tools.

As a study which aims to contribute to the parking policy making at Cardiff Council,

the thesis has decided to obtain parking users’ sensitivities to the two most

important parking policy tools: parking charge and parking spaces supply. To achieve

this aim, stated-preference data should be acquired from the parking-user survey

and relative discrete choice models need to be developed. Several studies in terms

of discrete parking choice modelling have been conducted in different areas such as

CBD of Edmonton, CBD of Oregon and CBD of Sydney (Hunt and Teply 1992; Hess

2001; Hensher and King 2001). However, in the context of Cardiff city centre, the

study in this domain is still blank. Meanwhile, the results of these studies may not be

suitable for the parking policy making for the Cardiff locality. This is because

sensitivities to parking features will vary across different regions (Hess and Polak

2004). Parking users in Cardiff will have their own specific sensitivities under the

certain local conditions. Hence, this study will be innovative and meaningful in filling

the gap in parking choice behaviour modelling for Cardiff. Moreover, parking users’

characteristics can also influence their sensitivities against the changes in parking

features. With the help of data on parking users’ profiles, the thesis will try to

achieve this taste variation across different parking user groups. The main findings of

15

the study will provide important references for the parking policy revision and

improvement in the context of Cardiff city centre. Three main research objectives

can be summarised as the follows:

What is the profile of short-stay parking users in Cardiff city centre?

What are the current parking issues in Cardiff city centre from the users'

perspective?

What are the drivers for people’s parking choice behaviour and degrees of

people’s sensitivities to different parking features?

1.3 Structure of the thesis

Chapter 2 will provide a critical literature review of previous studies. A wide range of

research in terms of influential factors on parking choice behaviour, parking charging,

parking availability and discrete choice models will be reviewed. The study has

identified the gaps in these studies and will try to fill them in this thesis. The Chapter

3 will illustrate the methodology applied for this specific research including the

conceptualisation of the research objectives, the design process of the parking-user

survey, the implication of the main survey and rationales behind the adopted

discrete choice models. As the core of the thesis, Chapter 4 provides a thorough data

analysis in order to study the parking choice behaviour in the context of the Cardiff

city centre. Descriptive statistics has been applied to obtain parking users’ profiles

and their perceptions to parking service. Chi-square tests have been conducted to

explore underlying relations across individuals’ travel characteristics. Additionally,

discrete choice models have been developed to achieve individuals’ sensitivities to

the changes in parking charge and availability. Taste variations across different

parking user groups are also obtained in this chapter. Finally, the Chapter 5 will

summarise the main findings of the study and will give specific suggestions for the

parking policy improvement in Cardiff city centre. The limitations of this research

and relative recommendations for future studies will also be demonstrated in the

final chapter.

16

2. Literature Review

2.1 Influential features in parking choice behaviour

It is necessary to understand what features can influence motorists’ parking choice

behaviour for urban planners to design more targeted and efficient parking policies.

Parking policy acts an essential part in contemporary travel demand controlling

measures for urban transport planning. An effective parking policy can help reduce

the private car usage in city centres and encourage people to use sustainable modes

such as public transport and cycling. Therefore, it is meaningful in terms of

alleviating traffic congestion, air pollution and improving residence quality for city

centres (Feeney 1988; Hess and Polak 2004).

In practice, parking choice is a complex process and is influenced by many factors. It

has been asserted by Thompson and Richardson (1996) that parking choice

behaviour could be considered as a search-process in which parking users make a

series of relative decisions based on their own experience and the conditions

provided by a specific parking place. When a motorist is deciding whether or not to

park at the usual parking place, he/she will firstly evaluate the conditions of the car

park. In this period, many factors can influence a parking user’s decision. These

include travel time to parking places, availability of parking spaces, ease of entering

or exiting the parking lot, walking time to destination, the distance to pay machines,

pricing. Meanwhile, a motorist’s characteristics such as gender, age and travelling

group, etc. will also have impacts on parking choices (Young 1986; Waerden et al.

2003). Dissatisfied parking users will try to find another parking place and restart the

above evaluation process (Thompson and Richardson 1996) or choose to use an

alternative mode. Other users who are satisfied with the overall condition of the

parking space after trade-off will choose to park there. Meanwhile, this process can

is illustrated by the Figure 2.1.

17

Figure 2.1 The process of parking choice

Source: Thompson and Richardson 1996.

There are also other studies which have been conducted in terms of various

determinants on parking choice behaviour. Feeney (1988) has reviewed several

mode choice models and argues that compared with internal cost such as fuel cost,

parking charges and other out-vehicle costs are more valued by parking users.

Another study conducted by Young et al. (1991) finds that the importance of time

from parking space to destination is weighed approximately 2.5 times more than in

vehicle time which illustrates that motorists commonly prefer to park at locations

closer to their destinations. Similarly, parking charge is also proved more

predominant than other generalised trip costs. A study to examine individuals’

sensitivities to parking cost and walk time from parking place to destination is

conducted in central Toronto by Miller (1993). The result shows that these two

features are determinants for both parking location choice and transport mode

choices.

Yes

Yes

No

Search for parking space

Examine car park

Drive to next car park

Evaluate car park

Determine route to next

car park

Wait

Accept

car park

Car park

available

Park

No

18

The above studies have identified several factors that can significantly influence

parking users’ behaviours. However, they fail to compare individuals’ different levels

of sensitivities to various parking features. Tsamboulas (2001) asserts that pricing is

the most influential determinant in parking choice behaviour. Besides parking fare,

walking time to destinations could also influence parking users’ location choices.

However, this research has not included the impact of cruising time on parking

choice behaviour. It is argued that limited research has been conducted on studying

the influences of parking features other than pricing, such as walking distance and

cruising time for parking spaces (Lambe 1996; Tsamboulas 2001). The study by Golias

(2002) has filled this gap to some extent. Although parking price has the most

important impact on peoples’ parking choice behaviour, searching time for an

available parking space and distance from parking place to destination could also

significantly influence parking users’ location choices (Golias et al. 2002).

However, with regard to an efficient parking policy, it is infeasible to take all

influential factors mentioned above into account. Some features relating to parking

users’ micro-behaviour such as distance to pay machine and distance to end

destination vary largely among different individuals and are meaningless as a

summary for designing parking policies for the public. Major features whose

variations are in general initially considered by parking users and can lead to changes

in parking choice behaviour are the favorites of parking policy designers. It has been

pointed out (Feeney 1988) that parking policies mainly contain two aspects: the first

is changing the structure or levels of parking fare and the other is controlling the

supply of parking spaces. Meanwhile, the supply control not only concerns the

physical supply of parking locations, but also includes access restrictions for certain

time periods or specific parking users. Changes in parking users’ behaviour towards

an applied parking policy might be altering parking locations, parking time, travelling

modes or abandoning the trip (Feeney 1988). Influential factors in parking choice

behaviour concluded from above studies are summarised in the table 2.1.

19

Table 2.1 Studies on influential factors of car parking choice behaviour

Studies Influential factors on parking choice behaviour

Young (1986) Travel time to parking place

Walking time from parking space to destination

Availability of parking spaces

Available shade

Feeney (1988) Parking charge and other out-vehicle costs

Supply of available parking spaces

Young et al. (1991) Egress time /Walking time to destination

Parking charge

Parking supply restraints/ Duration restrictions

Miller (1993) Parking cost

Walking time after parking

Tsamboulas (2001) Parking pricing (most influential)

Walking distance to destination

Golias et al (2002) Parking charge(most important)

Searching time for parking places

Walking time to destination

Parking duration

Waerden et al (2003) Distance to the ticket machine

Ease of ingress/egress

Distance to destination

Personal characteristics (minor)

Moreover, Hess and Polak (2004) have proved the existence of the ‘taste variation’

among parking users in different regions. This variation can be explained as different

weights valued by motorists on various parking features based on different local

20

conditions. For example, there are differences which exist in sensitivities to

searching time for parking spaces between parking users in two UK cities,

Birmingham and Sutton Coldfield (motorists in Birmingham showed higher sensitivity

to search time). Therefore, it is important to have research to work out the specific

‘taste of parking’ of local parking users. Meanwhile, the basic characteristics of

parking users should be acquired to gain a deeper understanding of their parking

choice behaviours (Hess and Polak 2004).

2.2 Parking pricing

The popularity of non-motorised transport modes or sustainable modes with high

capacity such as bus or tram can help reduce the externalities associated with

congestion in urban areas (Johansson-Stenman 1999). However, Bonsall (2000)

argues that common reliance on private cars has caused dilemma for urban

transport policy makers. It has also been proved that improvement in public

transport service is less effective with regard to reducing private car usage (Bonsall

2000). Relative research was conducted in five UK cities by Dasgupta et al. in 1994.

They have found that a 50 percent reduction in public transport fares would reduce

car usage by only 1-2 percent, while doubling the parking charge could significantly

transfer 20 percent of total car usage in urban areas to public transport or other

modes. It has also been proven (Calthrop et al. 2000) that pricing measures are the

most efficient solution for urban transport demand management (TDM) since pricing

can encourage people to reduce the usage of private cars and move to other

sustainable modes which can ease the traffic burden on roads. Meanwhile, among

different pricing measures, parking pricing is regarded as one of the most powerful

components besides road and congestion pricing (Calthrop et al. 2000). Clinch and

Kelly (2003) also argue that the cost of parking is generally the most important

component of the expenditure generated by an urban car journey (even larger than

oil consumption). Therefore, parking charge could be an extremely powerful tool in

changing motorists’ decisions for mode choice or parking location choice (Clinch and

21

Kelly 2003). Moreover, compared with other pricing measures such as road pricing,

parking charging is more easily implemented since it has been recognised generally

and charging facilities are simple and inexpensive. Meanwhile, people tend to be

more averse to road charging or congestion charging and the establishment of

relative charging facilities is much more complex. Therefore, with regard to peoples’

acceptance and to implement costing, parking pricing has its unique advantages

compared with other pricing measures (Arnott et al. 1990; Clinch and Kelly 2003).

Another relative study of the impacts of park-pricing on parking choice and mode

choice has been conducted by Clinch and Kelly (2003) in Dublin. Under the condition

that there is a lack of alternative public transport modes available at that time, most

people choose to change the parking locations when parking fares increase. The

result shows that pricing is a viable policy tool for managing peoples’ parking choice

behaviour (Clinch and Kelly 2003).

The necessity of implementing parking charge has also been argued by Shoup (2011).

It is estimated that 99% of parking places are free in the U.S., which has caused great

problems caused by excess parking behaviours such as oil consumption, air pollution

and congestion. Charging for parking is suggested by Shoup as a way to reducing the

travel demand generated by private cars. Meanwhile, to ease the resistance to the

introduction of charging, Shoup suggested the concept of ‘parking benefit districts’

which would invest the revenues from parking charging in transport improvement

such as the construction of bike paths or the enhancement of public transit services.

This could make residents accept and even welcome the park-pricing policy (Shoup

2011).

However, Calthrop (2002) argues that few sophisticated studies have been done in

terms of on-street parking pricing solutions. Arnott and Rowse (1998) suggest that

the best pricing amount for every parker should be the marginal external cost added

to other parking users. In other words, a parker who has taken up an empty

22

on-street parking space has to pay for the additional cost of increased parking spot

searching time for other motorists. Meanwhile, other studies suggest that the pricing

policies should take into account more externalities generated by on-street parking

such as road congestion, air pollution, disadvantages to other transport modes, etc.

Thus, more external costs will be covered by revenues from parking fees (Calthrop

2002; Anderson and Palma 2004). Although these pricing methods make sense

theoretically, the implementation might be difficult since there is no specific scale

and method to calculate this kind of cost.

Moreover, Kelly and Clinch (2006) point out that although altering the parking price

could bring aggregate parking behaviour change as transport planners would hope

for a specific area, the change might cause the extinction of certain travel types

(business or shopping, for instance). This is because travellers for different purposes

should have different sensitivities to the pricing change. At a certain tariff increase

rate (the threshold), the relatively more sensitive groups would cancel their journeys

which could lead to homogeneous trip purpose in a specific urban area (Tsamboulas

2001; Lam et al.2006). Meanwhile, it is of great importance for transport planners to

keep the diversity of trip purposes in city centres. Kelly and Clinch (2006) have

interviewed approximately 1000 on-street parking users in Dublin and have found

that significantly different sensitivities exist between business and non-business

users to parking price. The parking choice behaviour of non-business travellers is

more likely to be influenced by price increase, while motorists travelling for business

purposes present relatively lower sensitivity. Kelly and Clinch suggest for planners

that different pricing schemes should be considered and tested to maintain the

diversity of parking users to city centres. Based on Kelly and Clinch’s findings, it

would be useful to categorise parking users via different standards (gender, travel

purpose, parking frequency, etc.) and find out their sensitivities separately in order

to identify whether significant differences exist among different groups.

23

2.3 Parking availability

Button and Verhoef (1998) argue that the potential advantage of parking policies

derives from the fact that on-street parking is closely related to the road capacity.

Therefore, an efficient parking behaviour management should be beneficial in

reducing congestion. Meanwhile, another important parking feature: searching time,

which can reflect levels of ease of finding an available parking space, should also be

carefully considered by planners. A long searching time for parking space can slow

down the whole operation system of a car park and be a significant contributor to

road congestion (Button and Verhoef 1998). The vacancy rate of a parking place is

partly associated with the parking price, because cheap fares would cause

overcrowding in car parks, especially in busy periods. Laurier (2005) discusses the

issue that motorists in the stage of on-street parking searching usually experience

adverse emotions. Firstly, a driver has to slow down to look for a vacant spot, which

might cause a collision with other vehicles following behind. Secondly, the

distraction could also bring safety risks. Moreover, the driver would tend to be

anxious with the searching time extending, especially when frequently honked by

other vehicles (Laurier 2005). Therefore, parking users will definitely be unwilling to

continue to park at the current location if they find it extremely difficult to find an

available parking space (Button and Verhoef 1998).

Arnott and Rowse (1998) have developed a parking model to test the adverse

impacts generated by searching time for a vacant parking spot. They find that the

‘searching process’ could increase traffic volumes and lower the speed of traffic flow

which would generate externalities such as costs of delayed time. Therefore, besides

parking pricing, planners should also appropriately adjust the provision of parking

spaces to simplify the searching process to attract motorists to park (Arnott and

Rowse 1998). From another perspective, reducing the number of available parking

spots could also be an efficient way to reduce the private car usage in city centres.

However, the premise of applying the limitation still has to guarantee the simplicity

24

in finding an available parking spot for users who choose to continue to park at the

current location.

Shoup (2006) concludes that the general cruising time for an available on-street

parking space in congested urban areas is between 3.5 and 14 minutes on average,

and as many as 74 percent of motorists cannot find parking spaces immediately

during busy periods (Shoup 2006). It is argued that, even though finding a parking

space would be much easier in off-street private car parks, people still tended to

search for a curb space because of the cheaper fares. This finding shows that

motorists will not only consider one single feature when they are making decisions

on whether to park at the current location. In this case, they are likely to take both

price and availability into account and show a stronger sensitivity to price. Shoup has

expressed the concern that inexpensive on-street parking prices would cause

overabundant parkers to choose curb parking, which could largely extend the

cruising time and cause other issues such as congestion and air pollution. It is

suggested that further research should be conducted to help city councils to find the

equilibrium point between on-street parking pricing and parking space supply to

manage the quantity of curb parking users (Shoup 2006). Anderson and Palma (2004)

even regard the search for parking as a major congestion source in urban areas. One

study by them, focusing on short stay parking users (mainly shoppers), has shown

the importance of regular parking pricing in reducing the general cruising time. The

long cruising would not only lead to endogenous congestion of parking users’

vehicles, but also adversely influence other vehicles running on roads which might

cause exogenous congestion to the whole transport system (Anderson and Palma

2004). Therefore, only applying parking space restraint policy without the assistance

of a pricing method to reduce the demand might cause more serious issues and

externality costs to city centres generated by increased cruising time. An efficient

parking policy should combine pricing and parking space supply harmoniously.

25

Meanwhile, from the above literature, it can be illustrated that although parking

pricing and controlling the supply of parking spaces (change level of availability of

parking spaces) are two different methods in urban parking management, they are

closely connected to each other. First of all, as determinants to peoples’ parking

choice behaviour, they both aim to encourage motorists to park at dispersal

locations or change to other transport modes to ease congestion and pollution

issues in city centres. Secondly, increasing pricing rate in a specific location will

definitely cause a reduction trend in the number of parking users which can improve

the availability. Therefore, in terms of on-street parking management, many studies

have considered both pricing and availability aspects because of this strong

connection. In other words, researchers and planners are seeking a balance between

pricing and availability in order to design the optimum parking policies.

2.4 Parking policy and public attitudes

Parking policy concerns the management of parking infrastructure and measures to

control travel demand. As introduced above, two policy tools are mainly

implemented by transport planners: pricing and parking space supply management

(Valleley 1997, p. 105; Feeney 1988). Applying parking policy is a relatively

convenient and inexpensive way to manage car usage in urban areas. Meanwhile, it

can bring substantial revenue for a government. Valleley (1997, p. 139) has proposed

three key principles of successfully implementing a parking policy: (1) the

government should guide and support the parking planning of local transport

authorities to make it coordinate with the integral urban development project. (2)

formulating, operating and enforcing a parking policy unitedly in ‘parking working

groups’ which consist of different transport departments to increase efficiency. (3)

producing a parking plan integrated with urban policy objectives to avoid conflicts

between transport and urban scopes. Though these are useful suggestions, they are

limited to the operation and management of a policy and they neglect other

important aspects: how to increase the efficiency of a parking policy and control its

26

disadvantages. As discussed above, appropriately balancing the connection between

pricing and parking space supply can help manage parking behaviour efficiently.

Meanwhile, the level of public compliance can also determine the effect of parking

policies. Peoples’ perspectives of parking service in terms of charging, information

clarity and safety, for instance, can directly influence their willingness to park at a

specific location and obey the relative rules. Some illegal parking behaviour may be

encouraged under unsatisfying parking conditions such as irrational fare levels and

poor availability of parking spaces. Jones (1990) argues that ambiguous signs and

interpretation on parking facilities could have an adverse impact on parking users’

experiences. Therefore, a study is needed to acquire the levels of satisfaction of

parking uses with regard to various parking features. The relative findings may help

to reveal issues existing in parking service which can adversely influence the

effectiveness of parking policy.

However, the parking policies, in particular the restraint policies, have noticeable

flaws such as the risk of jeopardizing the vitality of city centres, which has caused the

concern with many authorities and retailers. Still and Simmonds (1999) have

examined the impacts of parking restraint policy on the economic prosperity of city

centres in a UK context. They find that shoppers are generally more sensitive to

parking conditions (tariff and availability) than people who travel for work or

business and can freely change shopping locations if they are unsatisfied. Therefore,

many local authorities and retailers think that parking provision in urban centres

should be improved in order to attract more shoppers from competing towns (Still

and Simmonds 1999). However, the finding of another study by Mingardo and

Meerkerk (2012) has rejected the common belief of retailers that higher parking

charges can cause declining profits in city-centre shopping. In contrast, increasing

parking capacity is shown to be positive to the turnover of a regional urban centre

with specific catchment area for car users. Though divergence exists with regard to

the effects of parking charges, improving the availability of parking places close to

27

city centres is recognised as helpful to their prosperity through attracting more

shoppers. To overcome parking policy’s potential risk of city centre prosperity, the

perspectives on parking services and sensitivities to parking features of parking users,

especially shoppers, should be obtained and analysed to assist parking policy

making.

2.5 Models applied to analyse parking choice behaviour

Discrete choice models have been used widely to analyse individuals’ choices among

finite alternatives under different attributes across many areas in transport research.

Discrete choice models can help predict how individuals’ travel behaviour will

change under changes in related attributes. For example, Train (1977) demonstrates

the feasibility to forecast mode choice (bus, car and rail) to work under different

attributes such as family income, cost and walking time, etc. With regard to parking

behaviour study, discrete choice models such as the multinomial logit and nested

logit model have been generally adopted by researchers (Hess and Polak, 2004).

A multinomial logit (MNL) model is developed by Teknomo and Hokao (1997) to

study parking users’ location choice in the Central Business District (CBD) of

Surabaya. They model three choice options: ‘on-road parking, off-street at surface

parking, off-street on multistory parking’. The result demonstrates that availability of

parking spaces, walking time to destination, pricing and comfort etc. can influence

motorists’ choices of parking locations. A similar model has also been used by Spiess

(1996), who has modelled the parking lots choice behaviour with a focus on

park-and-ride users. Hess (2001) conducts another MNL analysis to investigate the

influences of free parking on parking demand and mode choice. Respondents in

Portland’s CBD were offered three options: drive alone, carpool and transit. It is

found that under free parking, as many as 62 percent people would drive alone.

However, when a $6 daily pricing was applied, 21 percent of previous drive-alone

commuters would alter to carpool or transit.

28

Hunt and Teply (1992) conducted a nested logit (NL) model for parking location

choices using data collected in the CBD of Edmonton, Canada. The alternatives of the

model are hierarchically structured so that an alternative can be composed of other

subset alternatives. They design three choice alternatives for respondents: off-street

parking, on-street parking and employer-managed parking. Meanwhile, the off and

on-street parking alternatives consist of several individual locations as subsets.

Distance to work place, parking cost and waiting time, etc. are considered as

attributes which could influence parking users’ choices. The result successfully

proves that the above factors could influence parking choice behaviour, whereas the

research also has limitation in treating all parking users homogenously. Therefore,

Hunt and Teply advise that this model should be conducted separately for different

parking user groups for future research. The NL model is also adopted by Bradley

(1993) et al to examine the impacts of parking policy on mode choice and parking

type choice behaviours. Hensher and King (2001) develop another NL model based

on data from a stated preference in Sydney’s CBD to study the influences of tariff

schedule and operation-hour (availability) on parking choices. The design of

alternatives is highly hierarchical. For instance, the alternative ‘continue to drive and

park’ is classified into drive and park elsewhere in the CBD，at the fringe and beyond

the fringe of the CBD. They assert that changes in parking behaviour in Sydney’s CBD

can be mainly attributed to pricing policy (97%), while the contribution of supply

management only accounts for 3% (Hensher and King 2001).

Hess and Polak (2004) have successfully developed a more advanced model: mixed

multinomial logit (MMNL) in their study. It has made up for the shortage of advanced

choice model applications for previous parking choice behaviour studies. MMNL has

its unique advantages over other discrete choice models. It can help acquire the

random taste variations of different individuals. In other words, MMNL is able to

capture the heterogeneities in parking users’ sensitivities to different parking

features under various backgrounds in terms of income, gender and travel purpose,

29

etc. (Train 2003; Hess and Polak 2004). Thus the result modelled by MMNL is closer

to the practical situation than other modes which assume that all the motorists with

different characteristics share a homogenous ‘taste’ for parking features. Hess and

Polak find that the heterogeneity in motorists’ tastes can lead to significantly

different attitudes with regard to parking features such as tariff, search and egress

time. Parking users’ profiles, such as travel purposes, are proved to be important

factors which affect parking choice behaviour.

Table 2.2 Studies modelling parking choice behaviour

Study Study

Area Alternatives Attributes

Adopted

Model

Hunt and

Teply (1992)

CBD of

Edmonton,

Canada

a. Employer

arranged parking

b. On street parking

with subset

alternatives

c. Off street parking

with subset

alternatives

a. Money cost;

b. Distance to

destination;

c. Trip position

(home-work or

work-home);

d. Nature of

parking

surface;

e. Searching and

waiting time

Nested

Logit Model

30

Teknomo

and Hokao

(1997)

CBD of

Surabaya,

Indonesia

a. On street parking

b. Off street surface

parking

c. Off street

multistory parking

a. Availability;

b. Travel

purpose;

c. Search time;

walking time;

d. Pricing;

e. Safety;

f. Comfortability

Multinomial

Logit Model

Hess (2001) CBD of

Oregon,

US.

a. Drive alone

b. Carpool

c. Transit

a. Cost of parking

(free and not

free);

b. Transit Travel

Time

Multinomial

Logit Model

Hensher and

King (2001)

CBD of

Sydney,

Australia

a. Parking close to

the CBD

b. Parking elsewhere

in the CBD

c. Parking at the

fringe of the CBD

a. Operation

hours;

b. tariff schedule;

c. Walking time

to destination

Nested

Logit Model

Hess and

Polak (2004)

West

Midlands

region. UK

a. Free-on-street

b. Charged-on-street

c. Charged-off-street

d. Multi-story car

parking

e. Illegal parking

a. Searching time

for parking

spaces

b. Ingress/egress

time

(hierarchical

parking behaviors

across different

travel purposes

and

characteristics)

Mixed

Multinomial

Logit Model

31

Although MMNL can help capture taste variations across different parking user

groups, it is complex and has higher requirements for data computation. Other

discrete choice models such as binary logit, multinomial logit and nested logit are

obviously used more frequently because they are simpler and easier to understand.

Meanwhile, the relative inaccuracy of these models is possible to be controlled to

some extent. In other words, some functions of MMNL such as ‘random taste

variation’ could also be achieved through appropriately conducting the data

collection and estimation. Parking users’ profiles such as gender, age, travel purpose

and parking duration can be obtained from background questions in the survey.

During the analysis, respondents’ choices to the discrete choice questions are able to

be classified and grouped according to different profiles and modelled to obtain

specific sensitivities to parking features of a certain age group, travel purpose group.

Thus, the ‘random taste variation’ could also be achieved by other models instead of

MMNL.

2.6 Improvement of discrete choice modelling

The data for discrete choice models generally belong to stated preference (SP), since

respondents need to make their choices from provided alternatives based on their

perspectives to different attributes. Indeed, adopting the SP data can be regarded as

an advantage of choice modelling. Respondents’ preferences can be shown through

their trade-off among hypothetical but plausible situations. Besides SP data, other

parametres underlying background questions (travel purpose, age and gender, for

instance) or revealed preference (RP) questions could act as supplementary to the

discrete choice models, in order to achieve more accurate results with taste

variations.

Many studies have demonstrated the effects of unobserved factors, e.g. respondents’

characteristics on discrete choice modelling. A model that treats individuals’ taste

homogeneously will usually generate less practical results. For example, Hensher

(2001) talks about this kind of limitation of generally used standard models such as

32

multinomial logit (MNL) with regard to estimating the value of travel time savings

(VTTS). In Hensher’s study, it is shown that individuals’ heterogeneity and other

unobserved variances can affect the result of VTTS estimation. Greene et al. (2006)

develop a discrete choice model which integrates the heterogeneity variance of

unobserved effects so as to estimate travellers’ VTTS. The model has contained

individuals’ specific characteristics within the data to achieve taste variances. Greene

et al. argue that this method can create a better model fit to obtain different

sensitivities of various travel behaviours with regard to VTTS. Srinivasan and

Mahmassani (2003) develop another model to study route switching dynamics. The

result shows observed influential attributes on route-switching such as timeliness,

level of service and real time information. Many unobserved factors with regard to

peoples’ different travel behaviours and experiences are also proved to have great

influences. Horsky et al. (2006) develop a model to study consumer choice behaviour

and has demonstrated the importance of combining observed preference data from

stated preference data with the unobserved data related to individuals’ specific

preferences. They argue that the prediction accuracy of the model has been

improved through taking preference variations across households into account. The

result can show the heterogeneity across different respondents.

Compared with hypothetical SP questions, RP questions have the advantage of

revealing respondents’ actual behaviours under practical situations. In the transport

domain, combining SP and RP data appropriately can usually make a survey more

efficient (Ben-Akiva and Morikawa 1990). Hensher and Bradley (1992) assert that the

joint utilization of RP and SP data could complement each other’s weaknesses and

help obtain a better understanding of choice behaviours. On the one hand, SP data

can increase the predictive ability of a RP based model. On the other hand, the

explanatory power of a SP based discrete choice modelling can be improved through

connecting to RP results, which can help towards designing of proper attributes and

33

levels for SP questions and obtaining variations among respondents’ choices

(Hensher and Bradley 1992).

This kind of dual data strategy has been applied in many studies. Adamowicz et al

(1997) combine RP and SP data on recreational location choices to model peoples’

perceptions of environmental quality measures. It is asserted that the RP-SP mixed

model is superior to single SP models and has largely improved the modelling

performance in terms of environmental valuation. Jovicic (1998) conducts a

hierarchical multinomial logit model through the union of SP and RP data for

forecasting the influential factors in good transport mode choice among rail, sea and

lorry. The result successfully proves transport cost and time are more influential than

other features such as safety, delay and frequency, etc. Brownstone et al. (1999) use

mixed stated and revealed preferences data to model people’ choices for alternative

fuel vehicles. The single analysis of SP data shows individuals’ general preferences

among petrol, electric and natural gas vehicles, etc. under various attributes. In

contrast, after mixing the RP data, the study finds great heterogeneity existing in

households’ preferences for alternative-fuel vehicles. Therefore, Brownstone et al.

argue that pure SP models might provide implausible results while the mixed use of

SP and RP might make the forecast more accurate.

From the above studies, in order to obtain more accurate and practical results, the

discrete model should take into account data from background or RP questions to

achieve the heterogeneity among different individuals. With regard to parking choice

behaviour, pricing and availability are well known as most influential factors.

Meanwhile, peoples’ specific characteristics such as gender and age can also affect

parking behaviour, although their influences cannot be captured directly. Obtaining

taste variations across different motorists can help us better understand parking

choice behaviour (Hess and Polak 2004).

34

2.7 Conclusion of literature review

Studies focusing on influential factors in parking choice behaviour are reviewed

initially in this chapter. It is found that various factors with regard to the conditions

of parking places and parking users’ characteristics can have greater or lesser

impacts on parking choices. Among them, three main factors have been identified

from the review: pricing, ease of finding a parking space and walking time from

parking lot to end destination. Meanwhile, in terms of parking policies, transport

planners usually consider from two aspects: parking pricing and parking spaces

supply management. The study is intended to investigate the parking choice

behaviours of short-stay (the long-stay is mainly for working people) on street

parking users in Cardiff city centre to make contributions on the policy level for

Cardiff Council and the British Parking Association (BPA). Therefore, the research will

also mainly focus on the influences of parking charge and parking availability (ease of

finding a parking space). Meanwhile, the information of parking users’ basic profiles

(gender, parking duration and travel purpose, for instance) will also be studied to

identify the reasons for the ‘taste variation’ of parking users to different parking

features in Cardiff city centre.

Based on a series of studies on parking pricing, it can be argued that parking charge

is an efficient method for urban transport demand management. Compared to other

pricing methods such as congestion pricing, parking charge has its unique

advantages in public acceptance and operation expenditure. Parking charge is a

significant component in urban parking policy. Free parking without charging will

cause serious congestions and relative externalities in urban areas. Meanwhile,

many studies have stated the pricing amount should be equal to the cost of

externalities generated by parking behaviour. However, this method might be

practically unfeasible, since this cost is hard to be monetarily quantified. The

research will try to solve this question from another angle: motorists’ sensitivities.

Besides pricing, parking users also attach importance to searching time for an

35

available parking space. Reducing parking spaces supply is able to encourage people

to use sustainable transport modes or change parking location. However, as many

studies have argued, a single implementing of the supply restraint could cause

extended cruising time for on-street parking spaces, which would cause serious

transport externalities for urban areas. Thus, an optimum parking policy should

involve both pricing and supply management. This research seeks to quantify parking

users’ sensitivities with regard to pricing and availability through discrete choice

modelling. The findings will help to identify the mutual relation between them and

provide suggestions for transport planners on using these two features to create an

efficient parking policy for Cardiff.

Moreover, the public attitudes to parking policy also have an impact on the

effectiveness of parking policies. It is important for this research to obtain parking

users’ levels of satisfaction to various features of parking services (parking charge,

safety and clarity of guidance information, etc.) in Cardiff city centre. This can help to

identify the underlying parking issues in Cardiff city centre and to make relative

improvements to make people more satisfied with and willing to support the parking

management policies.

In terms of modelling, the study has reviewed various examples of using discrete

choice models to forecast peoples’ parking choice behaviour. Standard models such

as multinomial logit and nested logit models are applied widely. However, a more

complex mixed multinomial logit model (MMNL) has been shown to achieve more

practical results, since it can obtain the random taste variations of different parking

user groups. The research will try to use basic models to achieve the similar function

as MMNL through jointly utilising data from both stated preference questions and

background questions. Respondents will be classified into different groups by

profiles (travel purpose, gender and age for instance) acquired from background

questions. The sensitivities to parking features of different groups will be estimated

36

separately through modelling to acquire random taste variation. The existence of

heterogeneity variance has been demonstrated by many relative studies. It has been

widely proved that this kind of mixed data usage in discrete choice modelling can

enrich models’ forecast function and achieve practical taste variations across

respondents.

The study will conduct a parking-user survey at main short-stay on street parking

places around Cardiff city centre. The questionnaire for the main survey will contain

both background and stated preference questions. Parking users’ basic

characteristics and their perceptions to parking services in Cardiff city centre will be

summarised at the data analysis stage. A discrete choice model will be developed

using data from stated preference questions to obtain parking users’ general

sensitivities to parking charge and parking availability in Cardiff. Meanwhile, taste

variations of different parking user groups will also be shown with the help of data

from background questions. This study can help to obtain a thorough understanding

of individuals’ parking choice behaviour in Cardiff city centre which will fill the gap

that few previous studies have been conducted for this specific area. In particular,

applying logistic regression analysis to discover influential factors on parking choice

behaviour will be innovative to this specific region. The relative findings will be

helpful for the parking policy making within Cardiff context.

37

3. Methodology

3.1 Overview of the Methodology

The research has adopted a quantitative method which is able to analyse data via

mathematical, numerical or statistical techniques to investigate a social

phenomenon. The quantitative research is able to explain and conclude empirical

observations using mathematical expressions and can discover deep connections

between different factors (Given 2008). The research will initially raise specific

questions and then collect enough numerical data from sampled participants. Finally,

it will try to analyse the collected data to answer the research questions in an

unbiased and objective manner via statistical models.

The study has envisaged both background and stated preference questions in the

questionnaire. The background questions are designed to acquire the basic

characteristics of short-stay on street parking users in Cardiff. Meanwhile, the stated

preference (SP) section has provided discrete choice questions which propose finite

alternatives for respondents to choose under different hypothetical situations (Train

and Winston 2007). In this research, the analysis of SP data can help find attributes

affecting short-stay parking users’ choice behaviour as well as people’s sensitivities

to different parking features (price, convenience).The primary designed

questionnaire has been examined by a pilot experiment prior to the main survey.

Based on the findings from the pilot survey, the research has modified the primary

questionnaire and determined the survey period and locations of the main survey.

The software used to analyse the data of this research is IBM SPSS MODELER (SPSS).

SPSS is a popular program for statistical analysis in social science (Levesque 2007).

The data analysis will be conducted through SPSS using cross tabulation, chi-square

test and logistic regression, etc. The main statistical outcomes will be demonstrated

in the data analysis chapter. This Methodology chapter will first of all introduce the

38

conceptualisation process of the research objectives. Then, the survey design

process will be elaborately illustrated in Chapter 3.3 and 3.4. The implication of the

main survey will be demonstrated in Chapter 3.5. Finally, statistical methods applied

to data analysis will be introduced in Chapter3.6.

3.2 Conceptualisation of the study

Following the literature review, the research initially focused on Cardiff Council as a

source of enquiry into the context of parking in Cardiff city centre. Cardiff council has

provided helpful suggestions for this research. According to their response, on street

short-stay parking around Cardiff city centre is an area worthy to be studied. As

long-stay parking is mainly used by working persons, the council is more concerned

about the parking policy making for short-stay parking users coming to Cardiff city

centre for various reasons, e.g. shopping or leisure. Thus, in terms of this specific

study for short-stay parking users, three research questions are proposed.

What is the profile of short-stay parking users in Cardiff city centre?

What are the current parking issues in Cardiff city centre from the users'

perspective?

What are the drivers for people’s parking choice behaviour and degrees of

people’s sensitivities to different parking features?

Richard Carr, an expert in transport planning, also provides valuable suggestions for

this study. He suggests that the hypothetical choice questions should be more simply

designed since respondents may not want to spend time on understanding difficult

questions with complex attributes and alternatives. Meanwhile, a pilot survey should

be conducted to examine whether the questionnaire is designed appropriately.

During the survey design phase, Cardiff Council and the British Parking Association

(BPA) have also provided invaluable advice for the research. The details will be

illustrated in the following survey design section.

39

3.3 Design of the questionnaire

The questionnaire is designed to help collect data related to the three main research

questions mentioned above. It contains three main sections: (1) Questions for

parking users’ basic profiles (Q1-Q9 and Q11-Q13). (2) A question to obtain parking

users’ perspectives on the parking service in Cardiff city centre (Q10) (3). Discrete

choice questions to observe individuals’ choices under the variations of parking price

and availability (B1-B4). The following paragraphs will introduce the design

procedures and rationales of these three sections separately.

3.3.1 Profile of parking users

The basic profile of short stay parking users in Cardiff city centre includes people’s

travel purpose, intended parking duration, parking frequency, originations, distance

to destinations, number of adults/children they are travelling with, time taken to

find the parking space, age and gender. The relevant variables with corresponding

options to acquire parking users’ profiles are listed in the table 3.1. Meanwhile, all

questions and options are coded with specific numbers to make it convenient for

data entry and analysis in SPSS afterwards.

Table 3.1 Questions and options related to parking users’ profiles

Variables Options

Q1.Locality of origination Locality1a(more specific than Cardiff)

Postcode1b:___________________

Q2. Nature of origination 1. Home

2. Work

3. Other

Q3 Travel purpose to Cardiff city centre 1. Shopping

2. Work/Business

3. Leisure

4.Other_________

40

Q4. Intended parking duration Please write the parking duration

__________

Q5. Reason for parking at the specific

location

1.It is the only one I know

2.Close to destination

3. Easy to find a parking space

4. Reasonable parking price

5.It is safe to park here

6. Other___________

Q6. Time spent to find an available

parking space

1. Immediately upon arrival

2. Or record time (minutes)

___________

Q7. Parking frequency 1. Every weekday

2. 2-3 times a week

3. Once a week

4. 2-3 times a month

5. Every fortnight

6. Once a month

7. This is the first time I visit Cardiff city

centre.

Q8. Distance (walking time) to

destination

Please write here:________________

Q9. Size of travel group Adults9a(note number) _____________

Children9b(note number)

_____________

Q11. Gender 1. Male

2. Female

Q12. Age group 1.17-24 2. 25-34 3. 35-44

4. 45-55 5. 55-65 6. Over 65

41

3.3.2 Issues in parking service from users’ prospective

To investigate issues underlying the parking service around Cardiff city centre, a

question concerning people’s levels of satisfaction of several parking features has

been designed. The parking features include parking charge, ease of finding a parking

space, clarity of information on pay and display machines (information on pricing,

length of stay, etc.), range of payment options, personal safety and vehicle safety.

The rating method is adopted and respondents will be asked to score about their

levels of satisfaction for every aspect listed above from 1 to 5 (5 being very satisfied

and 1 being very dissatisfied). The following data analysis will summarise the score

for each aspect and will find out which aspect has the lowest/highest average value.

It is obvious that potential issues are likely to exist in the generally lower-rated

aspects. Therefore, the research has recorded every respondent’s reasons for why

low ratings are given to specific options in the main survey. Through summarising

these reasons, potential issues underlying the parking service in Cardiff city centre,

as well as relative improvement solutions, will be revealed.

Table 3.2 Q10: Parking users’ prospective to parking service in Cardiff city centre

Aspects Rate Reasons for low rating

Parking charge

Ease of finding a parking space

Clarity of information on pay and display

machines e.g. pricing, length of stay, etc.

Range of payment options

Personal safety

Vehicle safety

42

3.3.3 Driving forces behind parking choice behaviour and sensitivities to

parking features

In this section, discrete parking choice questions are designed to acquire stated

preference data. Based on people’s choices under various combinations of parking

charge (increase in parking charge) and availability (time to find a parking space),

logistic models can be developed to investigate how these two parking features

motorists’ parking choice behaviour. Meanwhile, developing the model for different

parking user groups classified by travel purpose, gender, and age, etc. can help to

show their different sensitivities (taste variations) to changes in parking conditions.

Attributes and levels:

From the literature review chapter, it is concluded that an efficient parking policy

should focus on managing two parking attributes: parking charge and parking supply.

Cardiff Council has provided suggestions for the levels of attribute ‘parking pricing’.

Instead of providing a specific parking tariff to respondents, it is suggested that an

‘increase in parking price’ should be used to simplify the situation that different

parking lots have different charging schemes. There are six levels with regard to

parking price increase: £0.50, £1.00, £1.50, £2.00, £2.50 and £3.00. With regard to

parking availability, the research envisages four levels for ‘time spent to find a

parking space’: immediately (0 minute), 2 minutes, 4 minutes and 6 minutes. The

rationality and scientific nature of designed attributes and levels have been tested in

the pilot survey prior to the main survey. The result suggests that all attributes and

levels are appropriate and can reflect the practical situation.

Table 3.3 Attributes and levels for the stated preference questions

Attributes Levels

Increase in parking price £0.50, £1.00, £1.50, £2.00, £2.50, £3.00

Time to find a parking space immediately, 2 minutes, 4 minutes ,6 minutes

43

Full factorial design and block:

The full factorial design is a statistics method which can create all possible

combinations of factors (attributes) and levels. It is required that the design must

consider at least two factors which consist of several discrete levels separately (Box

et al. 2005). In this case, the research has designed two attributes with discrete

levels for stated-preference questions. Therefore, it is feasible to use full factorial

design to obtain all possible combinations of attributes and levels.

There are in total 24 combinations in this discrete choice model design. It is

necessary to include every combination into stated-preference questions which

means there should be 24 hypothetical questions in the questionnaire. It is

unfeasible to include all the 24 questions into a single questionnaire, since it would

make the survey lengthy and respondents would be unwilling to participate.

Therefore, ‘Blocking’ should be applied in this case. In statistics, ‘Blocking’ means

arranging experimental combinations into groups/blocks to shorten a questionnaire

and simplify the research (Gates 1995). This study has separated the 24 situations

into 6 groups (blocks) and created 6 versions of questionnaires (BLCK_1 to BLCK_6).

Each version has only four discrete choice questions in the hypothetical section,

which has largely simplified the questionnaire. Meanwhile, four alternatives are

provided for each choice question: ‘Continue to park here’, ‘Park elsewhere’, ‘Travel

by other mode’ and ‘Not make the trip’. The following is one example of designed

stated preference questions in BLCK_1 version questionnaire. All combinations

acquired by full factorial design are listed in Table 3.4.

‘If the cost to park today increased by £1.00 and you could find parking space

immediately, what would be your choice?’

口 1 Continue to park here.口 2 Park elsewhere.口 3 Travel by other mode:

_________. 口 4 Not make the trip.

44

Table 3.4 Full factorial design result (with blocks)

Block Number Increase in parking charge

(£)

Time to find a parking

space (minutes)

1 1.00 0

2.00 2

1.50 6

2.50 2

2 2.00 6

1.50 2

1.50 4

1.00 2

3 2.00 4

3.00 2

0.50 0

2.50 4

4 3.00 4

0.50 4

1.50 0

3.00 6

5 2.00 0

1.00 4

0.50 6

2.50 6

6 2.50 0

3.00 0

0.50 2

1.00 6

45

3.4 Pilot survey

The pilot survey is an experiment that uses a relatively small-scale sample to test the

questionnaire designed for the main survey. It is prior to the main survey and aims at

discovering the potential issues underlying the survey design. Conducting a pilot

survey is helpful to revise the type, format or process of the main research to make it

more efficient (Sincero 2012).

To test the efficiency of the survey designed for short-stay parking users in Cardiff,

this study comprises a pilot survey in several parking places around Cardiff city

centre. There are two main objectives of the pilot survey. The first is to investigate

and determine the appropriate parking places for the main survey. The second is to

discover flaws in the designed survey which enables the research team to improve

the questionnaire and methods for the main survey. The primary questionnaire

(BLCK_1) used in the pilot survey is in Appendix I.

The pilot was conducted on 7th and 9th July 2014 at four main council-managed short

stay parking places near Cardiff city centre: St. Andrews Crescent, North Road,

Cardiff City Hall (City Hall Rd and Museum Ave) and Sophia Gardens (Cathedral Road).

There were two surveyors conducted the pilot from 9:00am to 12:00am each day.

3.4.1 Findings from the pilot survey

1) People generally responded positively to the questionnaire and sometimes

gave more information than what the questionnaire asked. Occasionally,

people refused to participate but this only happened when someone was in a

hurry for work or a meeting.

2) Respondents were able to understand questions well, including the

hypothetical questions in Sections B which we considered might cause

misunderstanding initially.

3) With regard to Question 9 (‘How many adults/children are travelling with you

today?’), it is thought that this question could cause misunderstanding,

46

resulting in many alone travellers still answering ‘1’. Therefore, it is decided

that this question should be altered to ‘Including you, how many

adults/children are travelling with you today?’ in the main survey.

4) Question 10 should add another column to record the reasons why people give

low rating to a specific aspect. This would help to identify current issues in the

parking service in Cardiff city centre and make relative solutions.

5) The best locations for the main survey should be at City Hall and St. Andrews

Crescent (maps listed in Figures 3.1 and 3.2), since these two parking places

were always busy during the pilot survey and plenty of respondents were

obtained. North Road did not seem an appropriate place, since few cars parked

there. The Sophia Gardens location was mixed with long-stay and short-stay

parking, which would bring difficulty to a survey for short-stay parking users.

Meanwhile, most people chose to park there for access to work in the nearby

offices, thus preventing the research from obtaining respondents coming for

shopping or leisure. Therefore, Sophia Gardens was also omitted from the main

survey locations.

6) In terms of survey time, it was found that all parking spaces were relatively

quiet during 8:30am to 9:30am and only respondents coming for work were

obtained in this period. The busiest period was 9:30am-11:00am and

respondents travelling for various purposes (work/business, shopping and

leisure) were obtained. After 11:30am, fewer cars parked since most parking

spaces were filled up. Meanwhile, in the afternoon, people tended to come to

the parking spaces to leave Cardiff city centre instead of arriving. Therefore, it

was decided that the main survey would not be conducted at this period. Based

on these findings, the main survey was determined to be conducted at

9:00am-12:00am (mainly focusing on 9:30am-11:30am to obtain respondents

coming for various reasons).

47

7) The British Parking Association suggested adding a simple Yes/No question in

the questionnaire: Have you heard of ‘Park Mark’? Managed by the BPA, Park

Mark is a safer parking scheme which aims to reduce crime activities in parking

facilities. Parking operators who take measures to prevent criminal behaviours

have the opportunity to be awarded ‘Park Mark®’ (BPA 2014b). Through this

question, the study could help show the popularity of this safety scheme

among general short-stay parking users.

8) During the pilot, each research personnel managed to obtain approximately 10

respondents each day. So it was predicted that the main survey could achieved

200-240 respondents with the help of four hired surveyors conducting the

research on six weekdays.

9) A prior analysis was conducted for the stated preference data from pilot. The

aim was to test whether attributes and levels were designed rationally. The test

successfully proved the hypothesis that as parking price and searching time

increase, people tend not to park at the current location. The binary regression

result also demonstrated a rational trend in general parking behaviour change

against variations in parking charge and availability (

Meanings of the coefficients can be seen in Chapter

3.6.2).

48

Figure 3.1 Main survey location: Cardiff City Hall CF10 3ND

https://www.google.co.uk/maps/place/Cardiff+Register+Office/@51.4851523,-3.1789675,193m/data=!3m1!1

e3!4m2!3m1!1s0x0:0x230a8c559bb66e0

Figure 3.2 Main survey location: St. Andrews Crescent CF10 3DB

https://www.google.co.uk/maps/place/St+Andrew's+Crescent,+Cardiff+CF10+3DB/@51.4851303,-3.17

41938,109m/data=!3m1!1e3!4m2!3m1!1s0x486e1cb9a12cc25f:0xf84e462d50a83f51

49

3.5 Implication of the main survey

In terms of data collection, the main survey obtained respondents at two

council-managed short-stay parking places near Cardiff city centre: Cardiff City Hall

and St. Andrews Crescent. The survey lasted six weekdays in July 2014. Simple

random sample method was adopted. It was guaranteed that each respondent had

been chosen entirely by chance during the whole process. Each parking user in the

survey areas had the same possibility of being chosen or not which ensured the

analyse result unbiased (Yates et al. 2008). The main survey obtained a total of 233

respondents.

With regard to survey methods, the study determined to conduct a face-to-face

survey. Compared with other methods, face-to-face is the most suitable one for this

research. This is because parking is short-time behaviour. Only by reaching people

who have just finished parking can the surveyors obtain fresh first-hand data

(destination, reason to park here and satisfaction with the service, for instance).

Meanwhile, a face-to-face survey offers the chance to provide more complex

questions such as hypothetical discrete choice questions to help obtain deeper

discoveries, since surveyors will have the opportunity to explain these relative

complex questions to respondents. This kind of data would be difficult to achieve by

other survey modes such as telephone or e-mail. During the main survey, the

interviewers arrived at the specific short-stay parking locations at arranged time

periods to access respondents. Surveyors read questions and recorded answers

provided by parking users who consented to participate. However, the main

disadvantage of face-to-face survey is money-consuming, since on-the-spot

interviewers need to be hired and paid (Doyle 2003).

The research received support from the British Parking Association (BPA) with a

funding of £1500, which enabled the research to hire four assist surveyors and

obtain a satisfying sample size of 233 respondents in total.

50

3.6 Data analysis methods

In the analysis stage, the data with regard to parking users’ personal and travel

behaviour characteristics will be studied through frequencies, percentages, sum,

mean value and other basic statistical methods to outline a generalised profile of

parking users in Cardiff city centre. Meanwhile, chi-square test and discrete choice

modelling (binary logistic regression) will also be developed to obtain a deeper

understanding of peoples’ parking choice behaviour in the context of Cardiff city

centre.

3.6.1 Chi-square test

The chi-square test belongs to statistical hypothesis test methods to identify

whether the ‘opposite of a hypothesis’ (null hypothesis) is true (Greenwood and

Nikulin 1996). This test can be conducted by SPSS. The significance coefficient

(p-value) stands for the possibility that a null hypothesis is true. If the p-value is less

than 0.05 or 0.1, it can be asserted that the observed result is highly unlikely to

belong to the situation under null hypothesis (Stigler 2008). In other words, the

hypothesis will be correct and there is a significant association between two tested

variables. Therefore, the p-value can be the criterion for judging whether or not a

hypothesis is right.

The study will use chi-square test to discover deeper relations underlying peoples’

parking behaviour, e.g. to examine whether significant parking duration difference

exists across different travel purposes or age groups. A number of hypotheses

concerning this kind of relations will be envisaged and tested through chi-square test

in SPSS. The results will be helpful in forming a better understanding of peoples’

parking behaviour in Cardiff city centre.

3.6.2 Logistic regression

In this research, four alternatives: ‘Continue to park here’, ‘Park elsewhere’, ‘Travel

by other modes’ and ‘Not make the trip’ are provided for respondents. Interviewees

51

will make their choices based on their perceptions to combinations of two attributes:

‘increase in parking cost’ and ‘time to find a parking space’.

Table 3.5 Attributes and Alternatives for Discrete Choice Questions

Subject Attributes Alternatives

Parking Choice

Behaviour in Cardiff

city centre

a. Increase in parking fare

b. Time to find a parking

space

a. Continue to park here

b. Park elsewhere

c. Travel by other modes

d. Not make the trip

Discrete choice models can obtain individuals’ choice preferences among finite

alternatives under different combinations of considered attributes. As an essential

part of discrete choice models, logistic regression is a statistical model for predicting

the probabilities of people making certain choices (Bishop 2006). A discrete choice

model will be developed to analyse the stated preference data and acquire parking

users’ sensitivities to parking charge and availability. The modelling will be

developed based on several rational hypotheses related to the main findings of

previous studies.

For instance, previous studies have found that parking charge and availability can

influence motorists’ parking choice decision (Feeney 1988; Hunt and Teply 1992;

Golias et al.2002). Therefore, one hypothesis could be: as the parking charge and

searching time for available parking spaces increase in a parking space around

Cardiff city centre, individuals tend not to continue to park at this location. Moreover,

Hess and Polak (2004) find that travel purpose and individuals’ characteristics can

also influence parking choice behaviour. Thus, hypotheses related to travel purpose,

age and gender, etc. will also be proposed and tested. This study will try to discover

various influential factors on parking choice behaviour in Cardiff city centre.

52

During the modelling, this study will combine the alternatives ‘Park elsewhere’,

‘Travel by other modes’ and ‘Not make the trip’ into ‘Not continue to park here’

because the survey is not intended to acquire data to identify the drivers underlying

the choices among these three options. Therefore, the model will have two response

variables: ‘park here’ and ‘not park here’, which mean the logit regression will be

binary. The form of a binary logit equation is:

[ ( )] [ ( )

( )]

P = the possibility that a specific case happens

α= the constant of the equation

β = the coefficient of the predictor variables.

X = predictor variables

The meaning of the equation can be interpreted that for one unit change in X, the

log odd of P will increase/decrease by the absolute value of β. With regard to this

study, the ‘P’ represents for the probability of choosing to continue to park at the

current location, thus it can be expressed as Ppark. Meanwhile, the possibility of the

opposite choice: ‘not continue to park here’ can be represented by Pnot_park. Two

attributes: ‘increase in parking price’ and ‘time to find a parking space’ are envisaged

in this case and can be represented by and separately.

Therefore, the specific binary logistic equation for this research should be:

[

] [

]

Ppark = the possibility of ‘continue to park here’

Pnot_park = the possibility of ‘not continue to park here’ (Ppark + Pnot_park = 1)

53

α= the constant of the equation

β1= the coefficient of variable ‘increase in parking price’

X1= variable ‘increase in parking price’

β2= the coefficient of variable ‘time to find a parking space’

X2= variable ‘time to find a parking space’

In terms of odds, the equation can also be rewritten as:

( )

( )

( )⁄

( )⁄

The modelling result through SPSS will identify the value and of α, β1 and β2 as well

as the corresponding significance coefficients. Thus, parking users’ sensitivities to the

attributes pricing and availability can be obtained. Meanwhile, as mentioned above,

to acquire the taste variations across individuals, the model will consider the

influence of peoples’ characteristics on parking choice. For example, another dummy

‘Female’ (1 for Female and 0 for Male) can be added to the equation to observe

whether different sensitivities exist between male and female parking users. To

simplify the expression, ‘Cost’ in the following equation represents the variable

‘increase in parking charge’ and ‘Time’ will stand for the variable ‘time spent finding

a parking space’.

[

]

In this case, ‘Male’ parking users are treated as the reference variable. If the resulted

p-value of or is less than 0.05, it can be proved that there are differences

54

existing between male and female parking users in terms of sensitivities to parking

charge or parking availability (a minus represents a higher sensitivity, while a

plus represents a lower sensitivity). In contrast, insignificant results (p-value > 0.05)

will mean no sensitivity difference exists between different genders.

3.7 Conclusion of the Methodology

This chapter has introduced the methodology used for this study. The

conceptualisation of study objectives is firstly introduced. Then, the survey design

process is illustrated. With regard to the core part of the questionnaire: ‘stated

preference questions’, this chapter has demonstrated the design process of

attributes, levels and alternatives in detail. A full factorial design has also been

conducted to create six versions of questionnaires. Meanwhile, the rationales of

statistics tools, in particular the logistic regression model applied for data analysis,

are also introduced. Based on the findings from the pilot survey and testing results,

the primary questionnaire for pilot experiment has been improved for the main

survey. The finalised main survey questionnaire is listed in the Appendix II (BLCK_1).

In the next chapter, data analysis will be conducted to study peoples’ parking choice

behaviour in the context of Cardiff city centre.

55

4. Data Analysis

This chapter mainly contains four sections. The first section will use descriptive

statistics to introduce the basic profiles of parking users in Cardiff city centre.

Through the analysis, a basic understanding of the respondents’ parking behaviour

can be formed. The second section will analyse respondents’ perceptions to the

parking service in terms of several parking features. The result will help discover the

underlying issues in the parking service of Cardiff city centre. The third section will

explore deeper relations across parking users’ characteristics. Based on hypotheses

testing through chi-square test, several underlying relations in terms of parking users’

profiles can be revealed. The last section will develop discrete choice models to

analyse parking users’ sensitivities to parking charge and parking availability.

Meanwhile, different sensitivities (taste variations) across various parking user

groups are also obtained. The relative findings will be helpful to the parking policy

making in Cardiff city centre.

4.1 Descriptive and frequencies statistics of parking users’ profiles

First of all, parking users’ basic profiles will be analysed with the help of descriptive

statistics. Based on the relative findings, a basic understanding of travellers’ parking

behaviour in Cardiff city centre can be obtained. Meanwhile, the result is also useful

to the development of the following hypotheses testing and logistic regression

modelling.

4.1.1 Gender and age group

The main survey has obtained a total of 233 respondents. Among these parking users,

106 are males (45.5%) and 127 are females (54.5%). In terms of parking users’ age

groups, it is found that individuals aged from 25 to 55 account for the largest part of

respondents (22.7% for 25-34, 29.2% for 35-44 and 20.2% for 45-55). Meanwhile,

younger motorists aged 17-24 take up 12.4% and those aged 55-65 take up 13.3%.

Parking users aged above 65 only account for 2.1% of all respondents (Figure 4.1).

56

4.1.2 Originations

With regard to the localities of parking users’ originations, it is found that among the

233 respondents, most of them are local (184) or from surrounding cities or towns

such as Newport (23), Swansea (7) and Bristol (7). Occasionally, the research heard

from respondents coming from relatively distant places such as Kilmarnock or Essex,

but the frequencies are usually only one in these cases. Thus, the study will code

them together as ‘Others’. Figure 4.2 describes the composition of parking users’

localities.

12.4%

22.7%

29.2%

20.2%

13.3%

2.1%

Figure 4.1. Percentages of parking users' age groups

17-24

25-34

35-44

45-55

55-65

Over 65

57

The study has also acquired information on the nature of parking users’ originations

(home, work or other place). It is found that 87.6% of the parking users come from

home, while 12.4% travel from work or other places (Figure 4.3).

78.97%

9.87%

3.00% 3.00%

5.15%

Figure 4.2 Percentages of parking users' origination localities

Cardiff

Newport

Bristol

Swansea

Others

87.6%

9.4%

3.0%

Figure 4.3 Percentages of parking users' origination natures

Home

Work

Other

58

4.1.3 Travel purpose to Cardiff city centre

In terms of respondents’ main reasons for travelling to Cardiff city centre, 40.3% of

parking users come for shopping, 28.3% come for work or business and 15.0% travel

for leisure (Figure 4.4). Meanwhile, people also travel to Cardiff city centre for

various other reasons. For example, there were 23 respondents who had come to

attend graduation ceremonies at Cardiff University during the survey. The other

reasons include appointments, medical treatment and court appearances, etc.

4.1.4 Travel group size

The survey has provided a question to obtain data on the sizes of parking users’

travelling groups. 58.4% of parking users travel to Cardiff city centre without other

adult companions while the other 41.6% of respondents travel with one or more

adult passengers (Figure 4.5). Meanwhile, most respondents (79.8%) do not travel

with children. The percentages of parking users who travel with 1 or 2 children are

12.9% and 5.6% separately. Only 1.7% of respondents travel with 3 or more children

(Figure 4.6).

40.3%

28.3%

15.0%

16.3%

Figure 4.4 Percentages of parking users' travel purposes

Shopping

Work/Business

Leisure

Other

59

4.1.5 Parking duration

Since the survey is aimed at short-stay parking users, the obtained data on parking

durations are all equal or less than 5 hours (the maximum duration for short-stay

parking in Cardiff city centre). The average parking duration of all parking users is

approximately 3.10 hours. Meanwhile, 74 out of 233(31.8%) users choose to park for

the maximum 5 hours and 101 persons (43.3%) choose to park under 2 hours. Only

58 respondents (24.9%) intend to park for 3-4 hours (Figure 4.7).

58.4% 30.5%

7.7% 3.4%

Figure 4.5 Percentages of travel group size(adults)

1

2

3

4

79.8%

12.9%

5.6% 1.7%

Figure 4.6 Percentages of travel group size(children)

0

1

2

3 or more

60

4.1.6 Parking frequency

In terms of parking frequency, 27.5% of travellers park at the current location at

least once a week. While, the other 72.5% of respondents park less frequently

including 38 (16.3%) visitors stated that ‘this is the first time parking at this specific

parking place’. Parking users whose parking frequency is once a month account for

the largest 37.8% across all the parking frequency categories (Figure 4.8).

39

62

40

18

74

0

10

20

30

40

50

60

70

80

1 or less 2 3 4 5

Frequencies

Hours

Figure 4.7 Distribution of short-stay parking users' parking duration

61

4.1.7 Reasons for parking location choice

To identity parking features which are considered by individuals, the survey has

sought respondents’ reasons for choosing to park at the specific parking location.

According to the result, most individuals (68.7%) choose to park at the specific

parking spaces because of the short distance to their destinations. The reasonable

parking price (15.5%) and ease of finding a parking space (9.9%) are the second and

third most important reasons for respondents’ certain parking choices. Parking

safety (2.1%) seems to be a feature neglected by parking users in Cardiff city centre.

Meanwhile, 3% of parking users claimed that the current location was the only

parking space they know.

3.9%

13.7%

9.9%

18.5% 37.8%

16.3%

Figure 4.8 Percentages of individuals' parking frequency

Every weekday

2-3 times a week

Once a week

2-3 times a month

Once a month

First time

62

4.1.8 Searching time for parking spaces

In terms of parking users’ parking experience, the study has obtained the data about

time taken to find a parking space. The average time respondents spend on parking

spaces searching is 1.48 minutes which illustrates that the current parking availability

in Cardiff city centre is relatively satisfying. A total of 146 (62.7%) parking users

stated that they found their parking space immediately upon arrival, while 71(30.5%)

respondents found a parking space in 5 minutes. Only 16 (6.9%) motorists had to

spend more than 5 minutes searching for an available parking spot (Figure 4.10).

68.7%

9.9%

15.5%

2.1% 3.0% 0.9%

Figure 4.9 Percentages of reasons for specific parking choices

Close to destination

Easy to find a parking space

Reasonable parking price

It is safe to park here

Only car park I know

Other

62.7%

30.5%

6.9%

Figure 4.10 Percentages of searching time for parking spaces

Immediately

1-5 minutes

6-20 minutes

63

4.1.9 Distance to destinations

As mentioned above, as many as 68.7% parking users choose to park at the specific

place because of short distance to their destinations. This finding is also supported

by the data collected from Q8 (‘How long will it take from this car park to your trip

destination by walk?’). The average walking time of all the 233 respondents is only

4.75 minutes. During the survey, 202 (86.7%) respondents stated that it would take

them less than 5 minutes to walk to their trip destinations, while 25(10.7%) would

spend 6-10 minutes to reach destinations. Only 6 (2.6%) persons claimed that the

walking time would be above 10 minutes (Figure 4.11).

4.1.10 Park Mark

The question ‘Have you heard of Park Mark?’ is proposed by the British Parking

Association. Park Mark – The Safer Parking Scheme is an initiative of the Association

of Chief Police Officers. It is aimed at reducing crime and the fear of crime in parking

facilities. A parking facility which has applied measures to prevent crime activities

and met police requirements will be rewarded the safer parking status: Park Mark®.

However, from Figure 4.9, it can be seen that on street short-stay parking users tend

to neglect the safety condition of the parking places. Maybe this can explain why

only 13(5.6%) parking users said they have heard of Park Mark during the survey.

The Park Mark seems not to be popular among short-stay parking users in Cardiff

86.7%

10.7%

2.6%

Figure 4.11 Percentages of walking time to destinations

1-5 minutes

6-10 minutes

Above 10 minutes

64

city centre. There is a possibility that people who use off-street private parking

facilities such as the NCP would pay more attention to the Park Mark. Criminal

activities usually do not happen at places exposed to the public such as on-street

parking facilities. People tend not to worry about their personal and vehicle safeties

when they choose to park at on-street facilities for a short period. This will be

supported by the finding in Chapter 4.2.

4.1.11 Conclusion of parking users’ profiles

From the above paragraphs, a basic understanding of short-stay parking users’

profiles can be achieved in the context of Cardiff city centre. Through descriptive and

frequencies statistics, the profiles of parking users’ travelling characteristics,

including travel purposes, parking reasons and parking duration, etc. as well as

personal characteristics such as gender and age, can be generally formed. The

relative results can help solve the first research objective: What is the profile of

short-stay parking users in Cardiff city centre. The following Table 4.1 summarises

the relative statistics results of parking users’ profiles (the sequence follows the main

survey questionnaire).

Table 4.1 Descriptive statistics of parking users’ basic profiles

Profiles Categories Frequency Percentage

Q1.Localities of

originations

Cardiff

Newport

Swansea

Bristol

Other

184

23

7

7

12

78.97

9.87

3.00

3.00

5.15

Q2.Natures of

originations

Home

Work

Other

204

22

7

87.6

9.4

3.0

65

Q3. Travel Purpose Shopping

Work/Business

Leisure

Other

94

66

35

38

40.3

28.3

15.0

16.3

Q4. Parking duration

(Mean: 3.10 hours)

1 hour or less

2 hours

3 hours

4 hours

5 hours

39

62

40

18

74

16.7

26.6

17.2

7.7

31.8

Q5. Reasons for

parking

Only car park I know;

Close to destination;

Easy to find a parking

space;

Reasonable parking

price;

Safe to park here;

Other

7

160

23

36

5

2

3.0

68.7

9.9

15.5

2.1

0.9

Q6. Searching time

for a parking space

(Mean: 1.48mins)

Immediately

1-5 minutes

6-20 minutes

146

71

16

62.7

30.5

6.9

Q7 Parking

Frequency

Every weekday;

2-3 times a week;

Once a week;

2-3 times a month;

Once a month;

First time parking;

9

32

23

43

88

38

3.9

13.7

9.9

18.5

37.8

16.3

66

Q8. Distance to

destination(Mean:

4.75mins by walk)

1-5 minutes

6-10 minutes

Above 10 minutes

202

25

6

86.7

10.7

2.6

Q9. Travel Group

Size (Adult)

1

2

3

4

136

71

18

8

58.4

30.5

7.7

3.4

Q9. Travel Group

Size (Children)

0

1

2

3 or more

186

30

13

4

79.8

12.9

5.6

1.7

Q11. Park Mark Yes

No

13

220

5.6

94.4

Q12. Gender Male

Female

106

127

45.5

54.5

Q13. Age Group 17-24

25-34

35-44

45-55

55-65

Over 65

29

53

68

47

31

5

12.4

22.7

29.2

20.2

13.3

2.1

4.2 Parking users’ perceptions to parking service

Parking users’ levels of satisfactions regarding the local parking service is a very

important reference to the effectiveness of the parking management. Any

unsatisfying condition related to parking services can adversely influence parking

users’ perceptions and their willingness to obey the parking policies. Thus, it is

significant for this study to acquire the data on individuals’ satisfaction levels to

67

various parking features. The findings can help to reveal the underlying issues and to

propose relatively pointed solutions to improve parking services of Cardiff city centre.

In this case, the survey has recorded each respondent’s rating (from 1 to 5, 1

represents ‘very dissatisfied’ and 5 represents ‘very satisfied’) with regard to their

satisfaction levels to various parking features: parking charge, ease of finding a

parking space, clarity of information on pay machines, range of payment options,

person safety and vehicle safety. Meanwhile, the survey has also recorded parking

users’ reasons for giving low rating to specific features. This can help the research

better perceive the potential issues in parking services.

4.2.1 Rating of parking charge

The average score of parking charge is 3.55. 53.2% of parking users are generally

satisfied with the parking charge in Cardiff city centre and have scored 4(32.2%) or 5

(21.0%). Meanwhile, 15.0% of respondents seem dissatisfied with the parking charge,

including 4.7% users who only rate 1 for parking charge (Figure 4.12). In almost all

cases, parking users prefer lower parking fares. During the survey, some respondents

even stated that the parking should be free. However, extremely low parking charge

or free parking will cause congestion issues and relative externalities to urban area.

A rational parking tariff is of great significance to control the parking demand and

ease congestions (Shoup 2011). Meanwhile, based on the mean score of 3.55, it can

be asserted that the parking pricing for short-stay parking in Cardiff city centre is

relative moderate. Most people (85%) showed understanding for the parking pricing

during the survey.

68

4.2.2 Rating of parking availability

The mean rating of ‘ease of finding a parking space’ is 3.88. The relative higher score

corresponds with the finding that most parking users’ can find a parking space in five

minutes in Cardiff city centre (Chapter 4.1.8). 67.4% of respondents are satisfied

with the parking availability and 21.9% give the average score of 3. 10.7% of parking

users assert that it is sometimes hard to find an available parking space (Figure 4.13).

In spite of the small percentage, respondents’ reasons for the low rating of parking

availability are noticeable. In their opinion, the short-stay parking places will become

quite crowded and inaccessible at around 10.30am. This has also been proved by our

observation during the survey. Some respondents who found a parking space easily

during the survey also stated that things would be different if they arrived later.

Therefore, there is a potential issue on how to accommodate the parking demand of

parking users who reach Cardiff city centre at relative late time periods.

4.7%

10.3%

31.8%

32.2%

21.0%

Figure 4.12 Percentages of ratings for parking charge

1

2

3

4

5

69

4.2.3 Ratings of information clarity and payment options

Two features, clarity of information and payment options, are both related to the

condition of pay machines. The average ratings of them are very close (3.84 for

information clarity and 3.86 for payment options). Meanwhile, the percentage

distributions from scores 1 to 5 are also similar (Figures 4.14 and 4.15). It can be

seen from the results that parking users are generally satisfied with the conditions of

pay machines. However, there are also issues existing with regard to pay machines.

During the survey, 21 respondents complained that the payment guidance on the

machine is confusing and possibly be difficult for first-time users. Meanwhile,

although the machines have the device to support payment by card, about 20

parking users argued that the machines sometimes do not work for cards in practice.

Maintenance of payment machines should be more regular to guarantee card

payment. Moreover, 9 parking users complained that there was no change given for

cash payments, which adversely influenced their parking experience.

2.1%

8.6%

21.9%

33.5%

33.9%

Figure 4.13 Percentages of ratings for parking availability

1

2

3

4

5

70

4.2.4 Ratings of personal safety and vehicle safety

Among all the parking features, safety gets the highest average rating (4.40 for

personal safety and 4.28 for vehicle safety). This result proves that the overall safety

condition of on-street parking around Cardiff city centre is satisfying, whereas it can

be seen that the mean score of personal safety is a bit higher than the score of

vehicle safety. Compared with personal safety, parking users tend to be more

concerned with the vehicle safety. 91% parking users are satisfied with personal

safety in the parking place, whereas the satisfying percentage of vehicle safety is 3.4%

3.9%

9.4%

18.9%

34.3%

33.5%

Figure 4.14 Percentages of ratings for information clarity on payment machines

1

2

3

4

5

3.0%

9.9%

17.2%

38.2%

31.8%

Figure 4.15 Percentages of ratings for payment options

1

2

3

4

5

71

lower (Figures 4.16 and 4.17).Several respondents said that rubs between vehicles

could happen because of the small size of parking spaces. Moreover, 2 parking users

concerned that the absence of CCTV could cause a risk to the safety of their vehicles.

4.2.5 Conclusion of parking users’ perceptions and relative suggestions

The average rating across all features is 3.97 which means parking users are

generally satisfied with the parking service in Cardiff city centre. Among the 6

parking features included in the survey, personal safety has the highest rating of 4.40,

and the personal safety is the second most satisfying parking feature with a rating of

4.28. With regard to the conditions of pay machines, the average score is

0.4%

8.6%

41.2%

49.8%

Figure 4.16 Percentages of ratings for personal safety

2

3

4

5

0.90% 11.60%

46.40%

41.20%

Figure 4.17 Percentage of ratings for vehicle safety

2

3

4

5

72

approximately 3.85. Parking availability is slightly more satisfying with a rating of

3.88, whereas the parking charge gets the lowest score of 3.55.The variations in

ratings for different parking features are shown by Figure 4.18.

According to respondents’ reasons for low rating, several underlying issues can be

seen. About 15% of parking users complain that the tariff of short-stay parking in

Cardiff is expensive. 10.7% of respondents are dissatisfied with the parking

availability. They demonstrate that it is sometimes hard to find a parking space

around Cardiff city centre, especially when they arrive at late time periods. Around

13% of parking users have scored under 3 for the conditions of payment machines.

The most common reason for low rating is that the payment guidance is lengthy and

confusing. According to the findings in 4.1.6, 37.8% of parking uses visit Cardiff once

a month and 16.3% of respondents have stated that it is their first time of parking at

the specific parking place. Thus, the complex guidance will cause inconvenience to

these 54.1% of infrequent travellers who are not familiar with the parking pay

machines in Cardiff city centre. Meanwhile, during the survey, around 20

respondents complained that the payment could not support card payment and no

change was given if they paid by cash. With regard to safety, parking users are

overall not concerned with their personal safety, whereas several respondents worry

3.55

3.88

3.84

3.86

4.4

4.28

3 3.5 4 4.5

Parking charge

Parking availability

Information clarity

Range of payment options

Personal safety

Vehicle safety

Figure 4.18 Ratings of parking features

73

about the vehicle safety since the parking space is too small to ensure that no rubs

happen between vehicles.

Based on the above findings, several suggestions can be made. First of all, a more

customer-friendly parking pay machine would greatly improve users’ parking

experience. The guidance information on payment machines should be simplified. It

will largely shorten the time spent on paying for parking. Meanwhile, more regular

maintenance of pay machines should be conducted to assure the varieties of

payment options. Card payment should be consistently supported since it is

convenient compared to cash payment and many respondents’ prefer using their

cards. Moreover, it would be better if machines offered change during cash payment

for parking users who have not carried enough small change.

With regard to vehicle safety, according to the observations from the survey and

respondents’ reflections, there is indeed a risk of collision between vehicles during

the parking process, especially during busy periods. The small size of parking space

can make parking difficult when the space on the two sides has already been taken

up. One possible solution might be properly increasing the width of parking spaces

to leave more room for parking users. However, it can adversely affect the parking

availability since less available parking space will be provided in that location.

Parking charge and parking availability are special features which usually constitute a

dilemma for parking planners. On the one hand, people from their own perspectives

will always prefer lower charging and a larger supply of parking places. On the other

hand, the parking pricing and supply must be controlled to a certain degree to

encourage a percentage of parking users to use other transport modes. As the most

important tools to manage parking demand in urban areas, they cannot be changed

by totally following parking users’ perceptions. Otherwise, the sharply increasing

parking demand will cause congestion, air pollution and safety issues to city centres

(Anderson and Palma 2004; Shoup 2011). Thus, parking polices related to the

74

variations in parking charge and parking supply should be very carefully thought out.

The findings in this section have only provided the perceptual information with

regard to individuals’ satisfactions. In Chapter 4.4, through logistic regression, the

study will preciously capture how peoples’ parking choice behaviour will change

against variations in these two parking features.

Table 4.2 Summarisation of parking users’ ratings to parking features

Parking features Rating Frequency Percentage

Parking charge 1

2

3

4

5

11

24

74

75

49

4.7

10.3

31.8

32.2

21.0

Ease of finding a parking space 1

2

3

4

5

5

20

51

78

79

2.1

8.6

21.9

33.5

33.9

Clarity of information on pay

machines

1

2

3

4

5

9

22

44

80

78

3.9

9.4

18.9

34.3

33.5

Range of payment options 1

2

3

4

5

7

23

40

89

74

3.0

9.9

17.2

38.2

31.8

75

Personal safety 1

2

3

4

5

0

1

20

96

116

0

.4

8.6

41.2

49.8

Vehicle safety 1

2

3

4

5

0

2

27

108

96

0

.9

11.6

46.4

41.2

4.3 Relations across parking profiles

In this section, the study will try to discover deeper relations underlying parking

users’ profiles. Virtually, instead of being independent, many profiles of parking

users are mutually connected to each other. These connections can be discovered

through statistical methods. Relevant hypotheses will be proposed and then tested

using cross tabulation and chi-square test in SPSS. The collected data will be

appropriately recoded for the analysis. The findings will help form a better

understanding of short-stay parking users’ behaviour in Cardiff city centre.

4.3.1 Travel purpose and travel group size

People travel with different purposes might differ in travel group sizes. It is

hypothesised that individuals who visit Cardiff city centre for non-work reasons are

more likely to travel with companions compared with people who travel for work

purpose. First of all, data on ‘travelling purpose’ (Q3) should be recoded into a new

variable ‘Work_or_Not’ in SPSS for conducting the chi-square test. The options

‘shopping’, ’leisure’ and ‘other’ are combined and coded as 0 which stands for

‘Non-work’ and the option ‘work/business’ is recoded to ‘1’ for ‘Work’. Meanwhile,

76

the data from Q9 in the questionnaire should also be processed. Firstly, ‘Compute

variables’ function in SPSS should be used to add the number of companion children

(variable Q9_CH in SPSS) and number of companion adults (variable Q9_AD in SPSS)

together for each respondent into a new variable ‘Travelgroup’ (Travelgroup =

Q9_CH + Q9_AD). Then, ‘Travelgroup’ will be recoded into a new variable

‘Companion’. Variables in ‘Travelgroup’ that equal 1 will be recoded into 0 which

represents travel alone. The others will be recoded into 1 which means travel with

companions. The test result is shown in the Table 4.3.

Table 4.3 Work_or_Not * Companion Crosstabulation

Companion Total

Travel alone Travel with

companions

Work_or_Not

Non-work Count 51 113 164

Expected Count 71.8 92.2 164.0

Work Count 48 14 62

Expected Count 27.2 34.8 62.0

Total Count 99 127 226

Expected Count 99.0 127.0 226.0

Significant Association. Fisher’s Exact Test. p=0.000.

From the result, significant association can be found between ‘work purpose’ and

‘whether travel with companions’ (p-value equals 0.000). The proposed hypothesis is

correct. It is obvious that parking users who visit Cardiff city centre for work purpose

tend to travel alone, while people travelling for non-work reasons such as shopping

and leisure are more likely to bring companions with them.

4.3.2 Travel purpose and parking frequency

People travelling for different purposes might have significant difference in parking

frequencies. According to common sense, a hypothesis could be proposed that

people driving to Cardiff city centre for work purpose have higher parking

frequencies than individuals coming for other reasons such as shopping or leisure.

The test result for this hypothesis is shown in Table 4.4

77

Table 4.4 Work_or_Not * Parking Frequency Crosstabulation

Parking Frequency

Every

weekday

2-3 times

a week

Once a

week

2-3 times a

month

Work_or_Not Non-work Count 1 15 15 19

Expected

Count

6.5 22.9 16.5 18.6

Work Count 8 17 8 7

Expected

Count

2.5 9.1 6.5 7.4

Total Count 9 32 23 26

Expected

Count

9.0 32.0 23.0 26.0

Significant Association. Chi-Square = 34.100 (df= 6) p=0.000

With regard to this hypothesis, the significance coefficient (p-value) is 0.000 which

indicates that difference in parking frequency exists between individuals who visit

Cardiff city centre for work and non-work purposes. Thus, the hypothesis is right.

From the ‘Expected Count’, it can be referred that parking users for work purposes

tend to park more frequently than the other users for non-work purposes such as

shopping and leisure.

Parking Frequency Total

Every

fortnight

Once a

month

First time to

Cardiff city

centre

Work_or_Not Non-work Count 15 71 31 167

Expected

Count

12.2 63.1 27.2 167.0

Work Count 2 17 7 66

Expected

Count

4.8 24.9 10.8 66.0

Total Count 17 88 38 233

Expected

Count

17.0 88.0 38.0 233.0

78

4.3.3 Travel purpose and age

People belonging to different age groups might tend to differ for travel reasons. It is

hypothesised that relatively younger visitors tend to visit Cardiff city centre for work

purpose while the older ones are more likely to travel for shopping or leisure. Data

on respondents’ age from Q13 should be coded into a new variable ‘Age_group’ for

the test. Respondents aged from 17-34 are coded as 1 in SPSS. Meanwhile, people

aged from 35-55 are coded as 2 and the others aged over 55 are coded to 3. The test

result is listed in the following Table 4.5.

Table 4.5 Work_or_Not * Agegroup Crosstabulation

Age group Total

17-34 35-55 Over 55

Work_or_Not

Non-work Count 62 77 28 167

Expected Count 58.8 82.4 25.8 167.0

Work Count 20 38 8 66

Expected Count 23.2 32.6 10.2 66.0

Total Count 82 115 36 233

Expected Count 82.0 115.0 36.0 233.0

Insignificant association. Chi-square test = 2.547 (df =2). P= 0.280

From the test result, there is no significant correlation found between parking users’

age and travel purpose. The resulted p-value equals 0.280 which is much larger than

the threshold value (0.05) for significant association. Hence, the above hypothesis is

null. People of different age groups are visiting Cardiff city centre for both work and

non-work reasons. No travel-purpose difference exists across different age groups.

4.3.4 Travel purpose and gender

Another interesting hypothesis is proposed that males and females might have

general differences with regard to travel purposes. The association between the

variables ‘Gender’ (Q12) and ‘Work_or_Not’ is cross-tabulated to examine this

hypothesis here.

79

Table 4.6 Work_or_Not * Gender Crosstabulation

Gender Total

Male Female

Work_or_Not

Non-work Count 67 100 167

Expected Count 76.0 91.0 167.0

Work Count 39 27 66

Expected Count 30.0 36.0 66.0

Total Count 106 127 233

Expected Count 106.0 127.0 233.0

Significant Association. Fisher’s Exact Test. p=0.013.

The p-value of the test is 0.013 (less than the threshold value 0.05) which means that

significant differences exist in males’ and females’ travel purposes. Based on the

comparison of ‘Count’ and ‘Expected Count’ value, it can be argued that female

parking users are more likely to visit Cardiff city centre for non-work reasons

(shopping or leisure), while male parking users tend to travel for work or business

reasons.

4.3.5 Travel purpose and parking duration

The last hypothesis is designed to examine whether parking users with different

travel purposes differ significantly in parking durations. The data from Q4 on

respondents’ intended parking duration are coded as the numeric variable ’Park_

duration’ in SPSS with five categories: 1 hour or less, 2 hours, 3 hours, 4 hours and 5

hours. The variables ‘Work_or_Not’ and ‘Parking _duration’ are cross tabulated to

test this hypothesis. The result is shown by Table 4.7

80

Table 4.7 Work_or_Not * Parking_duration Crosstabulation

Parking_duratoin Total

1 or less 2 3 4 5

Work_or_Not

Non-work

Count 26 47 29 13 52 167

Expected

Count 28.0 44.4 28.7 12.9 53.0 167.0

Work

Count 13 15 11 5 22 66

Expected

Count 11.0 17.6 11.3 5.1 21.0 66.0

Total

Count 39 62 40 18 74 233

Expected

Count 39.0 62.0 40.0 18.0 74.0 233.0

Insignificant association. Chi-square test = 1.091 (df =4). P= 0.896

The p-value is very high (0.896) in this case, which illustrates that there is definitely

no association between parking durations and travel purposes of parking users in the

context of Cardiff city centre. The hypothesis is null and no difference in parking

durations exists between individuals who visit Cardiff for work or non-work

purposes.

4.3.6 Conclusions of relations in parking users’ profiles

Through proposing hypothesis and the relative chi-square tests, several noticeable

findings in terms of interactions across parking users’ travelling characteristics are

obtained.

First of all, people who visit Cardiff city centre for work reasons are more likely to

travel alone, whereas individuals coming for shopping or leisure purposes tend to

bring companions with them (p-value equals 0.000). Meanwhile, people travelling

for work generally park more frequently than visitors for non-work reasons (p-value

equals 0.000). Finally, female parking users are more likely to travel for shopping or

leisure in Cardiff city centre, while male parking users tend to travel for working or

business reasons (p-value equals 0.013).

81

In addition, from null hypotheses, it can be argued that people in different age

groups generally do not have differences in travel purposes to Cardiff city centre

(p-value equals 0.280). Meanwhile, the parking durations of short-stay parking users

do not tend to vary against variations in travel reasons (p-value equals 0.896).

4.4 Logistic Regression

From previous studies, it can be concluded that individuals’ parking choice behaviour

is influenced by various factors such as parking charge, parking availability, distance

to destination, safety and parking users’ characteristics, etc. (Hunt and Teply 1992;

Teknomo and Hokao 1997; Hensher and King 2001;Hess and Polak 2004). In this

section, the study will develop logistic regression models to obtain parking users’

sensitivities to the most policy-related parking features: parking pricing and

availability (time to find a parking space) in the context of Cardiff city centre. The

core data used are from the discrete parking choices questions of the questionnaire

(B1 to B4). Meanwhile, other data related to parking users’ basic profiles will also be

used to test the ‘taste variations’ to parking features across various travel groups.

4.4.1 Choosing frequencies of alternatives

The main survey has obtained 233 completed questionnaires and each questionnaire

has four discrete choice questions. Thus, the study has a total of 932 sets of

hypothetical parking choice data. Meanwhile, the study provides four alternatives

for respondents: (1) Continue to park here，(2) Park elsewhere, (3) Travel by other

modes and (4) Not make the trip. The following Figure 4.19 describes the

percentages of respondents’ choosing frequencies for these four alternatives.

82

It is obvious that the alternatives ‘continue to park here’ and ‘park elsewhere’ have

been chosen most frequently with the percentages of 45.5% and 43.6% separately.

However, the other two alternatives take a much smaller account. Based on the fact

that Alternative 3 and 4 are rarely chosen (9.8% and 1.2% separately) by the

respondents, it is reasonable to combine them with Alternative 2 to form the new

category: ‘Not continue to park here’. Therefore, the options provided can be

classified into two categories: ‘Continue to park here’ (Alternative 1) and ‘Not

continue to park here’ (Alternatives 2, 3 and 4). Meanwhile, descriptive statistics can

be conducted for respondents who have chosen ‘Not continue to park here’. It is

found that most parking users (79.9%) will search for another parking place if they

are not satisfied with the parking charge and availability of the current place. 17.9%

of the respondents will transfer to other transport modes such as bus, train and

cycling. Only a few, 2.2%, will not make the trip (Figure 4.20).

45.50%

43.60%

9.80%

1.20%

Figure 4.19 Choosing frequencies of alternatives

1.Continue to park here

2. Park elsewhere

3.Travel by other modes

4. Not make the trip

83

Moreover, the survey is not intended and designed to acquire the data to identify

the factors which can influence parking users’ trade-off among Alternatives 2,3 and 4

when they choose not to continue to park at the current location. Hence, in terms of

the logistic regression, this combination is more practical and can assure the

accuracy of the results. In SPSS, people’s choices will be coded into the variable

‘Park_or_not’ where 1 represents the choice of ‘continue to park here’ (Alternative 1)

and 0 stands for choosing ‘not continue to park here’(Alternatives 2, 3 and 4).

4.4.2 Independent samples t-test

A t-test should be conducted prior to the regression to examine the validity of the

collected data. As with the increase in parking charge and decline in parking

availability, the probability of a parking user choosing to continue to park will

decrease (Feeney 1988; Young et al.1991). Therefore, compared with people who

choose ‘continue to park here’, respondents who choose ‘not continue to park here’

should have higher mean values in terms of the two attributes: ‘Increase in parking

price’ and ‘Time to find a parking space’. Otherwise, the validity of the collected data

will be dubious. Therefore, the t-test is necessary to be conducted to identify

whether there are significant differences in the mean values of these two groups.

79.9%

17.9%

2.2%

Figure 4.20 Percentages of parking users' choices if they choose not to park at the current location

Park elsewhere

Travel by other modes

Not make the trip

84

The dependent variables for the t-test are ‘increase in parking price’ (variable

‘INCR_COST’ in SPSS) and ‘time to find the parking space’ (variable ‘TIME’ in SPSS).

And the independent variables are people choices between ‘park’ and ‘not park’.

Table 4.8 Independent samples t-test result

Park_or_not N Mean Std. Deviation Std. Error Mean

INCR_COST Continue to park here 424 1.283 .7572 .0368

Not continue to park here 508 2.137 .7144 .0317

TIME Continue to park here 424 2.55 2.167 .105

Not continue to park here 508 3.38 2.209 .098

INCR_COST: Significant difference. Sig.(2-tailed): 0.000. Mean difference: -0.8538

TIME: Significant difference. Sig.(2-tailed): 0.000. Mean difference: -0.830

It can be seen from the t-test result that significant differences exist in the mean

values of INCR_COST and TIME (both Sigs=0.000) between two the choice groups.

The mean value of ‘INCR_COST’ of people who choose ‘continue to park’ is about

£0.85 lower than people who choose ‘not continue to park’. Similarly, the mean

value of ‘TIME’ of the former is also 0.83 minutes lower than the later. Hence, this

result in the context of Cardiff city centre corresponds to the findings of previous

studies and the collected data is proved to be valid.

4.4.3 Modelling parking users’ general sensitivity to parking features

Based on the previous studies, with regard to the context of Cardiff city centre, it can

be hypothesised that increases in parking pricing (INCR_COST) and time to find

parking place (TIME) can adversely influence individuals’ probability to continue to

park at the current location. Binary logistic regression will be developed to test this

hypothesis and obtain parking users’ general sensitivities to these two parking

features.

85

The regression equation is:

Y: the log odds of choosing ‘continue to park here’ (compared with the odds of ‘not

continue to park’). Y= [

]

α: the constant of the equation

β1= the coefficient of variable INCR_COST

INCR_COST= Increase in parking charge

β2= the coefficient of variable TIME

TIME= Time to find a parking space

The regression result is shown by Table 4.9 (Full regression result is listed in

Appendix III)

Table 4.9 Binary regression result for general parking users

B S.E. Wald df Sig. Exp(B)

INCR_COST -1.492 .106 196.302 1 .000 .225

TIME -.226 .036 39.159 1 .000 .797

Constant 3.034 .239 161.198 1 .000 20.789

The significant coefficients of INCR_COST and TIME are both 0.000 (less than 0.05),

which has illustrated that increase in parking charge and searching time have an

important influence on peoples’ parking choice behaviour.

From the modelling result, the regression equation should be:

If searching time is controlled at a certain degree, one unit (£1) increase in parking

charge will decrease the log odds of continuing to park by 1.492 (the probability will

be times as much as the previous probability). Similarly, 1 minute

86

increase in searching time will decrease the log odds by 0.226 (the possibility will be

0.797 times as much as the previous one).

4.4.4. Modelling parking users’ taste variations to parking features

The above model has obtained parking users’ general sensitivities to changes in

parking features. Whereas, different user groups usually tend to have variations in

sensitivities (taste variation). Hess and Polak (2004) have applied a MMNL model to

acquire the taste variations across parking users. But this thesis will try to use

another method to obtain this taste variation in the context of Cardiff city centre.

The study assumed that parking users with different personal characteristics (gender

and age) and trip characteristics (travel purpose and travel group size) might have

different sensitivities to the changes in parking pricing and availability. Binary logistic

models will be adopted to test the assumptions.

Gender:

The regression equation in this case should be:

: Variable INCR_COST times the dummy value ‘Female’

: Variable TIME times the dummy value ‘Female’

Compared with the equation to model the general sensitivity, two new variables are

added into the new equation: and . ‘Female’

is a dummy variable in the equation. If a respondent’s gender is female, the variable

‘Female’ will equal 1.Otherwise it will be 0 (male as the reference). If the resulted

coefficient or is significant (sig<0.05), then it can be argued that compared to

the males, female parking users have different sensitivities to parking features.

The binary regression result is shown in Table 4.10.

87

Table 4.10 Binary regression result for parking users with different genders

B S.E. Wald df Sig. Exp(B)

INCR_COST -1.473 .123 144.562 1 .000 .229

TIME -.203 .047 18.382 1 .000 .816

INCR_COST Female -.034 .115 .088 1 .766 .966

TIME Female -.043 .058 .564 1 .453 .958

Constant 3.036 .239 161.356 1 .000 20.814

The significances of ‘INCR_COST∙Female’ and ‘TIME∙Female’ are 0.766 and 0.453

separately. They are both much larger than the threshold value 0.05. Thus, there is

no sensitivity difference found between male and female parking users.

Travel Purpose:

The regression equation to test taste variations across travel purposes should be:

‘Work’ is the dummy variable in this equation. It equals 1 if a parking user travels for

work reasons and equals 0 if an individual comes for shopping or leisure (references).

To be noticed, respondents come for reasons such as a ceremony or an examination

are also coded to ‘Work’ since they also have to reach a specific location at a certain

time like other working people. The test result is shown by Table 4.11.

Table 4.11 Binary regression result for parking users with different travel

purposes

B S.E. Wald df Sig. Exp(B)

INCR_COST -1.530 .121 161.048 1 .000 .217

TIME -.285 .045 39.288 1 .000 .752

INCR_COST∙Work .042 .118 .127 1 .722 1.043

TIME∙Work .121 .059 4.247 1 .039 1.129

Constant 3.074 .241 162.725 1 .000 21.624

88

From the result, no significant difference (Sig of ‘INCR_COST∙Work’ equals 0.722) is

found between people who travel for work and non-work reasons in terms of

sensitivity to parking charges. However, people coming for work purposes are less

sensitive to the changes in searching time compared to people coming for non-work

reasons ( equals 0.121, sig =0.039). Under a unit change in searching time for

parking places, the log odds of continuing to park of working people is 0.121 larger

than that of non-working people (the possibility is 1.129 times as much as that of

non-work people). This finding is different to the common notion that parking users

with work purposes might be more sensitive to the searching time for parking spaces.

This can be explained as being that people travel for work reasons are less free in the

choice of parking places since they usually have to work at a certain location.

Meanwhile, since working people usually arrive relatively earlier than other parking

users, in practice the searching process would be not difficult for them. Therefore,

they tend to be less sensitive to searching time for parking spaces.

The resulted equation should be:

Age group:

Respondents’ age are classified into three categories for developing the model:

17-24 (reference), 25-45 (Variable ‘Age25_44’) and over 45 (Variable ‘Ageover45’ ).

The regression equation should be:

over45 + over45

In this case, ‘Age25_44’ and ‘Ageover45’ are dummy variables and respondents who

aged from 17- 24 are the reference. The regression result is shown in Table 4.12.

89

Table 4.12 Binary regression result for parking users belonging to different age

groups

B S.E. Wald df Sig. Exp(B)

INCR_COST -1.212 .172 49.616 1 .000 .298

TIME -.276 .077 12.716 1 .000 .759

INCR_COST∙Over45 -.144 .175 .680 1 .409 .866

TIME∙Over45 .020 .088 .050 1 .824 1.020

TIME∙Age25_44 .084 .085 .978 1 .323 1.088

INCR_COST∙Age25_44 -.489 .171 8.131 1 .004 .613

Constant 3.059 .241 160.858 1 .000 21.309

It is obvious that parking users aged from 25-44 are more sensitive to changes in

parking price (sig=0.004). Compared with the individuals aged from 17-24, their

general log odds of continuing to park here will be 0.489 lower (the possibility will be

0.613 times as much as that of people aged 17-24) against a £1 increase in parking

charge. Meanwhile, no sensitivity difference has been found with regard to parking

users aged over 45 (Sigs>0.05). The equation in this case is:

Travel group:

Finally, whether the size of travel group can influence the parking probability will be

tested. Based on the survey data about travel group size, a new dummy variable

‘Companion’ is created in SPSS. In ‘Companion’, 1 stands for people who travel with

companions while 0 stands for people who travel alone (reference). The form of the

regression equation should be:

90

Table 4.13 Binary regression result for parking users with different travel group

sizes.

B S.E. Wald df Sig. Exp(B)

INCR_COST -1.648 .132 156.405 1 .000 .192

TIME -.201 .048 17.138 1 .000 .818

INCR_COST∙Companion .251 .119 4.436 1 .035 1.286

TIME∙Companion -.047 .058 .650 1 .420 .954

Constant 3.057 .241 161.436 1 .000 21.268

According to the result, parking users who travel with companions are less sensitive

to the increase in parking price (sig=0.035). Against one unit (£1) increase in parking

charge, their log odds of continuing to park at the current location is 0.251 larger

than that of lone travellers (the possibility is 1.286 times as much as that of lone

travellers). Meanwhile, different travel sizes tend not to differ in the sensitivity to

searching time for parking spaces (sig=0.420). Thus, the finalised equation should be:

The study has by now developed logistic models for different parking user groups.

The main findings are listed in the following table.

Table 4.14 Summarisation of parking user groups’ different sensitivities to parking

features

Note: [+] means less sensitivity to a parking feature; [-] means more sensitivity to a parking

feature; [N/A] means no sensitivity difference exists for a specific variable.

Classification

standards of parking

user groups

Variables(Categories) sensitivity

to parking

charge

sensitivity to parking

availability

(searching time)

Gender Female (Dummy)

Male (Reference)

[N/A]

[N/A]

[N/A]

[N/A]

91

Travel purpose Work (Dummy)

Non-work (Reference)

[N/A]

[N/A]

[+]

[-]

Age group 25-44(Dummy)

Over 45(Dummy)

17-24(Reference)

[-]

[N/A]

[+]

[N/A]

[N/A]

[N/A]

Travel Group Size Companion(Dummy)

Travel

alone(Reference)

[+]

[-]

[N/A]

[N/A]

4.4.5 Modelling sensitivities of parking users with various characteristics

The above chapter has classified parking users into different categories through

various standards: age, gender, travel purpose and travel group size. Varied

sensitivities to parking price and availability have been identified across these

categories. However, there is a limitation of the above binary logistic models in that

each of them only takes one characteristic into account for the prediction. In almost

all cases, the profile of a parking user is the combination of various characteristics.

The above models cannot predict the parking possibility against the mutual impact

of these characteristics. For example, they cannot answer the question that what is

the parking possibility of a 30-year-old male parking user who travels alone to Cardiff

city centre for work purpose against the changes in parking features? The following

model will consider all the possible influential factors on sensitivities to parking

features simultaneously and try to predict the parking possibility of parking users

with various characteristics. Thus, the regression equation should be:

92

The following Tables 4.14 and 4.15 show the result of the variables in the equation.

The full regression result can be seen in Appendix IV.

Table 4.15 Binary regression result for parking users with various characteristics

(step 1)

B S.E. Wald df Sig. Exp(B)

INCR_COST -1.404 .226 38.461 1 .000 .246

TIME -.333 .104 10.263 1 .001 .717

INCR_COST∙Female -.096 .123 .619 1 .432 .908

TIME∙Female -.018 .061 .083 1 .774 .983

INCR_COST∙Work .066 .125 .281 1 .596 1.068

TIME∙Work .124 .062 3.992 1 .046 1.132

INCR_COST∙Age25_44 -.459 .175 6.857 1 .009 .632

TIME∙Age25_44 .105 .088 1.414 1 .234 1.110

INCR_COST∙Over45 -.131 .178 .539 1 .463 .877

TIME∙Over45 .019 .091 .045 1 .832 1.019

INCR_COST∙Companion .274 .126 4.700 1 .030 1.315

TIMECompanion -.006 .062 .010 1 .921 .994

Constant 3.132 .245 163.249 1 .000 22.925

93

Variables with insignificant coefficients (sig>0.05) do not tend to influence the

probability of parking and should be removed from the model. Meanwhile,

significant variables will be taken into account for Step 2 of the modelling. The

remained variables for Step 2 are: ‘INCR_COST’, ‘TIME’, ‘Time∙Work’,

‘INCR_COST∙Age25_44’ and ‘INCR_COST∙Companion’.

Table 4.16 Binary regression result for parking users with various characteristics

(step 2)

B S.E. Wald df Sig. Exp(B)

INCR_COST -1.555 .132 139.152 1 .000 .211

TIME -.297 .043 48.074 1 .000 .743

TIME∙Work .145 .045 10.562 1 .001 1.156

INCR_COST∙Age25_44 -.248 .088 8.001 1 .005 .780

INCR_COST∙Companion .239 .091 6.948 1 .008 1.270

Constant 3.132 .245 163.895 1 .000 22.915

Based on the result, the finalised equation should be:

The interpretation of the equation can be: As the increase in parking charge and

searching time for a parking space, parking users will be less likely to continue to

park at the current location. Meanwhile, people who travel to Cardiff city centre for

work purposes are less sensitive to the increase in searching time than the others

who come for shopping or leisure. Parking users aged from 25-44 are more sensitive

to the increase in parking price. Finally, people who travel with companions tend to

be less sensitive to the parking charge. Theoretically, this model can be applied to

94

obtain the sensitivities to parking features for every individual with unique travel

characteristics. For example, the equation to predict the parking possibility of a 30-

year-old male parking user who travels alone to Cardiff city centre for work purpose

should be:

= 3.132-1.803 INCR_COST - 0.152 TIME

A one-unit (£1) increase in parking charge will decrease his log odds of continuing to

park by 1.803. Meanwhile, a one-minute increase in searching time will cause the log

odds to decline by 0.152.

4.5 Conclusion of the data analysis

Through the analysis of the survey data, the study has obtained important findings.

First of all, parking users’ profiles in the Cardiff context is acquired. It can help

transport planners better understand the question of who the parking users in

Cardiff city centre.

With regard to perceptions to parking service, parking users are generally satisfied

with the parking service provided by Cardiff city centre (the average rating is 3.97/5).

However, several issues have been identified from their responses. For example, the

guidance information on the pay machine is confusing and can cause inconvenience

to unfamiliar parking users. 15% of the respondents think the short-stay parking

charge in Cardiff city centre is still too expensive to them. Meanwhile, the

overcrowded parking places in the late-time period have also caused parking

difficulties to about 10% of parking users.

The study has also identified several important relations underlying parking users’

profiles. It is found that parking users who travel for work purposes tend to drive

alone and park more frequently compared with the others who come for non-work

95

reasons. Meanwhile, female parking users are more likely travel to Cardiff city centre

for shopping or leisure than male parking users.

Finally, the study develops binary regression models to test parking users’

sensitivities to the changes in parking charge and availability. For general parking

users, a £1 increase in parking charge will cause the log odds of continuing to park to

decline by 1.492. A one-minute increase in searching time will decrease the log odds

by 0.226. Meanwhile, the study has also obtained the taste variations across

different parking user groups. It has found that characteristics such as age, travel

purpose and travel group size can lead to significant differences in parking users’

sensitivities to parking features.

96

5. Conclusions and Recommendations

Based on a parking user survey in the main short-stay parking places around Cardiff

city centre, the thesis has provided a thorough analysis of individuals’ parking choice

behaviour. Through developing statistical tools, various important discoveries have

been achieved. The findings of this thesis will directly contribute to the parking

policy improvement in Cardiff.

In Chapter 2, a critical literature review has been demonstrated. According to the

previous studies, individuals’ parking choice behaviour is influenced by many factors.

Parking pricing, parking availability, walking distance to destination, ease of

ingress/egress and even personal characteristics can affect motorists’ parking

decisions (Young 1986; Miller 1993; Waerden et al. 2003). Across the various factors,

parking charge and parking availability are identified as the features which are most

related to parking policy. Transport planners can use parking pricing and parking

supply management to achieve travel demand control for urban areas (Feeney 1988;

Calthrop et al. 2000). As a study that is aimed at contributing to parking policy

improvement, the thesis decides mainly to explore the influences of these two

parking features on people’s parking behaviour. The necessity of applying parking

pricing to ease congestion and relative externalities in city centres has been

demonstrated by previous studies (Clinch and Kelly 2003; Shoup 2011). However, the

pricing scale is found to be hard to determine based on the studies so far.

Meanwhile, another important parking policy tool, parking availability management

is shown to cause more serious congestion issues to urban areas if applied alone. An

efficient parking policy should implement these two policy tools simultaneously

(Anderson and Palma 2004; Shoup 2006). The thesis has tried to seek the balance

between applying parking charge and parking supply management from a new angle:

parking users’ sensitivity. Through acquiring people’s sensitivities to parking features,

variations in parking possibilities against unit changes in parking pricing or parking

availability can be predicted. Thus, the scales of parking pricing and parking spaces

97

supply control can be determined, based on transport planners’ anticipations.

Meanwhile, discrete choice models should be developed to obtain parking users’

sensitivities. Several previous studies have successfully modelled individuals’ parking

choice behaviour in the context of different regions in the world. However, no

specific modelling has been developed to study the parking choice behaviour in the

Cardiff background up to now. This thesis has decided to fulfil this gap. Across the

various types of discrete choice models, the Mixed Multinomial Logit (MMNL) model

can provide more accurate predictions since it can obtain the taste variations across

parking users’ characteristics (Hess and Polak 2004). The thesis has tried to

innovatively achieve the similar function of MMNL model using standard binary

logistic models. Compared with MMNL, the developed models in this study are more

simplified. Taste variations across parking users are captured with the help of the

data set from background questions in the survey. The innovation of mixing

background data into stated-preference based modelling has largely improved the

prediction accuracy of standard discrete choice models and can provide references

for future studies.

In Chapter 3, the conceptualisation process of the study is introduced. Through the

literature review and the suggestions from the parking experts at Cardiff Council and

the British Parking Association, the objectives of the study have been identified

explicitly. The targeted questionnaire for the survey has been designed based on the

determined research objectives. It mainly contains three sections: (1) Background

questions in terms of parking users’ personal and travel characteristics. (2) Rating

questions to acquire respondents’ levels of satisfaction with the parking service in

Cardiff city centre. (3) Stated-preference discrete choice questions to observe

individuals’ parking choices against various combinations of parking features. A prior

pilot survey has also been conducted to test the rationality of the questionnaire

design and determine the survey time and locations for the main survey. Using

98

random sample method, the main survey of this research has successfully obtained

233 respondents in the main short-stay parking places around Cardiff city centre.

In the core data analysis chapter, descriptive statistics are initially applied to acquire

the basic profiles of parking users. The parking users in Cardiff city centre are mostly

from the locality (78.97%) and mainly travel to the city centre from home (87.6%)

rather than work or other places (12.4%). Meanwhile, 58.4% of the parking users

travel without adult companions and 79.8% travel without child companions. In

terms of travel purpose, 55.3% of parking users travel to Cardiff city centre for

shopping or leisure; 28.3% travel for work reasons and 16.3% come for other specific

reasons such as a graduation ceremony or an appointment. Short-stay parking users’

average parking duration is 3.10 hours.27.5% of them park at the specific short-stay

parking location at least once a week, whereas the other 72.5% tend to park less

frequently. On average, parking users have to spend 1.48 minutes searching time for

available parking spaces and the mean distance from the car park to their trip

destinations is 4.75 minutes on foot. The main reason of respondents for choosing

the specific parking place is ‘close to destination’ (68.7%), while the most overlooked

parking feature is parking safety (2.1%). Meanwhile, during the survey, only 5.6%

respondents have stated that they have heard of the ‘Park Mark’ of the BPA which

illustrates that the popularity of this safety scheme in short-stay parking spaces is

not satisfying in the Cardiff city centre background. Details of the statistics of parking

users’ profiles are contained in Chapter 4.1.

Parking users are generally satisfied with the parking service provided in Cardiff with

an average rating of 3.97/5 across all parking features. However, several underlying

issues have been found during the analysis. 15% of respondents complain that the

current parking tariff is still expensive. Meanwhile, about 10.7% of parking users are

not satisfied with the parking availability in Cardiff city centre. Finding a parking

space will be quite hard for motorists who arrive at late time periods. In addition,

99

many respondents have complained that the guidance information on parking pay

machines is confusing and inconvenient. Thus, the thesis recommends that this

information be revised and made into a more straightforward version. Although the

pay machines have the device to support card payment, it is observed during the

survey that the function does not always work in practice. Thus, it is also suggested

that more frequent maintaining of the machines be conducted to ensure the ranges

of payment options. Moreover, although more than 90% of parking users are

satisfied with the parking safety, some still show concern that the parking spaces are

too small to avoid rubs between vehicles. Solving the above issues can directly help

to improve the parking service and encourage individuals’ compliance to the parking

policy in Cardiff city centre. The analysis details of parking users’ perceptions can be

seen in Chapter 4.2.

Chi-square tests have also been conducted to explore the underlying relations across

parking users’ profiles. It is found that, in the context of Cardiff city centre, people

who travel for work purposes tend to park more frequently compared with people

who come for shopping or leisure (p-value=0.000). Meanwhile, parking users for

non-work reasons are more likely to bring travel companions with them, while the

others who come for work purposes tend to travel alone (p-value=0.000). Also,

differences in travel purposes have been identified between different genders.

Female parking users are more likely to travel for shopping or leisure, whereas male

parking users are more possibly coming for work reasons (p-value =0.013). Details of

the chi-square test results have been illustrated in Chapter 4.3.

The validity of the collected stated-preference data has first of all been confirmed

through the independent sample t-test. Discrete models have subsequently been

developed to simulate the impacts of parking features on individuals’ parking choice

behaviour. With regard to parking users’ general sensitivities to parking charges and

parking availability, the resulted regression equation is: ‘

100

’. A £1 increase in parking charge will decrease the log

odds of continuing to park at the current place by 1.492 (the parking possibility will

be times as much as the possibility before a pricing increase).

Meanwhile, a one-minute increase in searching time for parking spaces will decrease

the log odds of parking by 0.226 (the possibility will be 0.797 times as much as the

possibility before a searching time increase). Thus, the result has quantified parking

users’ reflections against variations in parking features. It can provide an important

reference to the parking policy making. For instance, if planners aim to reduce the

parking demand by about 25 percent in Cardiff city centre, they can increase the

general parking charge by 20 pence or reduce the parking availability to increase the

general searching time by 1.3 minutes (the possibility will decline by 25.8%).

Moreover, the combined changes in both the two parking features can be applied to

achieve the same goal and might be more effective because of the joint utilisation of

parking policy tools. The balance between parking pricing and supply management

can be found and examined in practice. The thesis has also modelled the taste

variations across different parking users groups. It has found that people travelling

for shopping or leisure purposes are virtually more sensitive to the increase in

searching time. This result has supported another finding under the UK context: Still

and Simmonds (1999) argue that shoppers are more sensitive to the adverse parking

conditions because they are freer in parking location choice compared to working

people. Meanwhile, parking users aged from 25-44 are more sensitive to the

changes in parking charge than individuals belonging to other age groups.

Additionally, people who travel with companions tend to be less sensitive to the

increase in parking pricing (Details in modelling can be seen in the Chapter 4.4).

Based on these sensitivity variations across parking users, more targeted parking

policies can be made if planners want to change the parking behaviour of a specific

parking user group. Finally, a model has been developed to predict individuals’

parking possibilities against the combined effect of personal and travel

101

characteristics. The resulted regression equation is:

In almost all cases, in statistics, every individual can be described as the combination

of various characteristics. In this case, every individual has their unique combination

of travel purposes, age and travel group size. Therefore, this result has expanded the

model’s function to predicting the parking possibility for each individual. The model

can contribute to the design of more detailed and pointed parking policy making

within the Cardiff city centre context.

However, there are also several limitations in this thesis. First of all, this parking

choice behaviour study is conducted under the scope of council-managed on-street

short-stay parking. Other parking types in the Cardiff background such as off-street

private car parks, multistory parking and park-and-ride are not included in this

research. Future studies are recommended to analyse parking users’ behaviour in

the context of other parking types. Thus, a more comprehensive understanding of

parking choice behaviour under various parking types can be obtained. Secondly, in

this thesis, the influences of the two most policy related parking features (parking

pricing and availability) have been explored. However, other detailed factors such as

safety, comfortability and ease of ingress/egress, etc. can also influence individuals’

parking choice decisions. Hence, it is suggested that future studies should take more

influential factors into account in their analyses. Finally, in terms of individuals who

choose not to continue to park, this study has not acquired enough information to

identify the factors which can impact on their trade-off among parking elsewhere,

travelling by other modes and not making the trip. Hence, further studies are

recommended to obtain the relative data and fill this gap. Multinomial logit models

and nested logit models are suggested if future studies can acquire more thorough

data sets which can help classify individuals’ parking choices into more categories.

102

References:

Adamowicz, W. et al. 1997. Perceptions versus objective measures of environmental quality

in combined revealed and stated preference models of environmental valuation. Journal of

environmental economics and management 32, pp. 65-84.

Anderson, S.P. and A. de Palma. 2002. The economics of parking. Journal of Urban

Economics 55, pp. 1-20.

Arnott, R.et al.1990. A temporal and spatial equilibrium analysis of commuter parking.

Journal of Public Economics, vol. 45, pp. 301-335.

Arnott, R. and Rowse, J. 1998. Modeling Parking. Journal of Urban Economics, vol. 45, pp.

97-124.

Ben-Akiva, M. and Morikawa, T. 1990. Estimation of switching models from revealed

preferences and stated intentions, Transport Research A 24, pp. 485-495.

Bishop, C.M. 2006. Pattern recognition and machine learning. Springer.

Bolter, A. 2009. Capital investment pushes Cardiff up retail rankings. Available at:

http://www.walesonline.co.uk/business/business-news/capital-investment-pushes-cardiff-u

p-2072100 [Accessed 15 July 2014]

Box, G. E. et al. 2005. Statistics for Experimenters: Design, Innovation, and Discovery (2nd

ed.). Wiley. ISBN 0-471-71813-0.

Bradley, M.et al. 1993. A joint model of mode/parking type choice with supply constrained

application, Proceedings of the 21st Annual Summer PTRC Meeting on European Transport,

Highways and Planning, pp. 61-73.

103

British Parking Associating. 2014a. Cardiff University student awarded funding from the John

Heasman Bursary. Available at:

http://www.britishparking.co.uk/News/cardiff-university-student-awarded-funding-from-th

e-john-heasman-bursary [Accessed 10 July 2014]

British Parking Association (BPA). 2014b. Park Mark-The Safer Parking Scheme. Available at:

http://www.britishparking.co.uk/Park-Mark---The-Safer-Parking-Scheme. [Accessed 17 July

2014]

Brownstone, D. et al. 1999. Joint mixed logit models of stated and revealed preferences for

alternative-fuel vehicles. Transportation Research Part B 34, pp. 315-338.

Bonsall, P. 2000. Legislating for modal shift: background to the UK’s new transport act.

Transport Policy 7, pp. 179–184.

Button, K. J. and Verhoef, E. T. 1998. Road Pricing, Traffic Congestion and the Environment.

Cheltenham&Northhampton: Edward Elgar.

Calthrop, E. et al. 2000. Parking policies and road pricing. Urban Studies 37, pp. 63–76.

Calthrop, E. 2002. Evaluating on-street parking policy. Energy, Transport & Environment.

Available at: http://www.econ.kuleuven.be/ete/downloads/ete-wp02-03.pdf [Accessed: 1

July 2014]

Cardiff Council. 2007. Cardiff City Centre Users Survey 2006. Cardiff Research Centre.

Cardiff Council. 2014a. Parking in Cardiff. Available at:

https://www.cardiff.gov.uk/ENG/resident/Parking-roads-and-travel/Parking-in-Cardiff/Pages

/default.aspx [Accessed 15 July 2014]

Cardiff Council.2014b. On street parking. Available at:

https://www.cardiff.gov.uk/ENG/resident/Parking-roads-and-travel/Parking-in-Cardiff/On-st

reet-parking/Pages/On-street-parking.aspx [Accessed 15 July 2014]

104

Clinch, J.P. and Kelly, J.A. 2003. Testing the sensitivity of parking behaviour and modal choice

to the price of on-street parking. Environmental Studies Research Series (ESRS) Working

Paper 03/03. Department of Environmental Studies, University College Dublin.

Dasguapa, M. et al. 1994. Impact of transport policies in five cities. Transport Research

Laboratory, Project Report 107.

Doyle, J.K. 2003. Face to Face Surveys. Department of Social Science and Policy Studies.

Worcester Polytechnic Institute. Available at:

https://www.wpi.edu/Images/CMS/SSPS/Doyle_-_Face-to-Face_Surveys.pdf [Accessed 17

July 2014]

Feeney.B.P.1988. A review of the impact of parking policy measures on travel demand.

Transportation Planning and Technology, vol. 13, pp. 229.

Fisher, R.A. and Yates, F. 1974. Statistical Tables for Biological Agricultural and Medical

Research, 6th ed. Oliver & Boyd, Ltd., Edinburgh.

Gates, C. E.1995. What really is experimental error in block designs? The American

Statistician 49 (4): 362–363.doi:10.2307/2684574. JSTOR 2684574.

Given, Lisa M. 2008. The Sage encyclopedia of qualitative research methods. Los Angeles,

Calif: Sage Publications.

Golias, J. et al. 2002. Off-street parking choice sensitivity. Transportation Planning and

Technol, vol. 25, pp. 333–348.

Greene, W. H. 2006. Accounting for heterogeneity in the variance of unobserved effects in

mixed logit models. Transportation Research Part B 40, pp. 75–92.

Greenwood, P.E. and Nikulin, M.S. 1996. A guide to chi-squared testing. Wiley, New

York. ISBN 0-471-55779-X

105

Guardian Cardiff. 2014. Cardiff Council takes on illegal parking. Available at:

http://www.theguardian.com/cardiff/2010/jul/05/cardiff-council-illegal-parking-fines-enforc

ement [Accessed 14 July 2014]

Hensher, D. A. and Bradley, M.1992. Using stated response choice data to enrich revealed

preference discrete choice models. Marketing Letters. pp. 139-151.

Hensher,D.A. and King, J. 2001, Parking demand and responsiveness to supply, pricing and

location in the Sydney central business district, Transportation Research 35A(3), pp.

177-196.

Hensher, D.A. 2001. The sensitivity of the valuation of travel time savings to the specification

of unobserved effects. Transport Research Part E 37, pp. 129-142.

Hess, D.B. 2001. The effect of free parking on commuter mode choice: evidence from travel

diary data, Transportation Research Record 1753, pp. 35 – 42.

Hess, S. and W. Polak, J. 2004. An analysis of parking behaviour using discrete choice models

calibrated on SP datasets. Centre for Transport Studies - Imperial College London. Available

at:http://www.researchgate.net/publication/23731358_An_analysis_of_parking_behaviour

_using_discrete_choice_models_calibrated_on_SP_datasets/file/79e4150d086ad64a90.pdf

[Accessed 30 June 2014]

Horsky, D. et al. 2006. Observed and unobserved preference heterogeneity in brand-choice

models. Marketing Science, vol. 25, pp. 322-335.

Hunt, J.D. and Teply, S. 1992. A nested logit model of parking location choice. Transport

Research, vol. 27B, No. 4, pp. 253-265.

Johansson-Stenman, O. 1999. Regulating road transport externalities: pricing versus

command and control. In: Sterner, T. (Ed.), The Market and the Environment. Edward Elgar,

Cheltenham.

106

Jones, P. 1990. Traffic Quotes: Public perceptions of traffic regulations in urban area. HMSO,

London.

Jovicic, G. 1998. Application of models based on stated and revealed preference data for

forecasting danish international freight transport. Available at:

http://www.trafikdage.dk/td/papers/papers98/trafikmodeller/g_jovicic/g_jovicic.pdf

[Accessed 30 June 2014]

Kelly, J.A. and Clinch, P.J. 2006.Influence of varied parking tariffs on parking occupancy levels

by trip purpose, Transport Policy 13,pp. 487–495.

Lam, W.H.K. et al. 2006. Modelling time-dependent travel choice problems in road networks

with multiple user classes and multiple parking facilities. Transport Research Part B 40. pp.

368-395.

Lambe,T.A.1996. Driver choice of parking in the city. Socio-economic Planning Science 30 (3),

pp. 207-219.

Laurier, E. 2005. Searching for a parking space. Intellectica, pp. 2-3, 41-42, 101-116.

Levesque, R. 2007.SPSS Programming and Data Management: A Guide for SPSS and SAS

Users. Fourth Edition, SPSS Inc., Chicago Ill. ISBN 1-56827-390-8

Mingardo, G. and Meerker, V. J. 2012. Is parking supply related to turnover of shopping

areas? The case of the Netherlands. Journal of Retailing and Consumer Services 19, pp. 195–

201.

Miller, E.J. 1993. Central area mode choice and parking demand. Paper presented to the

Transportation Research Board Annual Meeting, Washington D.C. January 1993.

Shoup, D.C. 2006. Cruising for parking. Transport Policy 13, pp. 479–486.

Shoup, D. C. 2011. The high cost of free parking. APA Planners Press; Updated edition.

107

Sincero, M.S. 2012. Pilot Survey. Explorable. Available at:https://explorable.com/pilot-survey

[accessed 16 July 2014].

Spiess, H. 1996. A logit parking choice model with explicit capacities. Working Paper,

EMME/2 Support Center, Aegerten, Switzerland.

Srinivasan, K.K. and Mahmassani, H.S. 2003. Analyzing heterogeneity and unobserved

structural effects in route-switching behavior under ATIS: a dynamic kernel logit formulation.

Transport Research Part B 37, pp. 793-814.

Stigler, S. 2008. Fisher and the 5% level. Chance 21 (4): 12. doi:10.1007/s00144-008-0033-3.

Still, B. and Simmonds, D. 2000. Parking restraint policy and urban vitality. Transport Reviews,

vol. 20, no. 3, pp. 291-316.

Teknomo, K.and Hokao, K. 1997. Parking Behaviour in Central Business District: A Case Study

of Surabaya,Indonesia. EASTS Journal 2.

Thomson, R. G. and Richardson, A. J. 1998. A parking search model. Transportation Research

A, Vol. 32, pp.159-170.

Train, K. 1978. A validation test of a disaggregate mode choice model. Transportation

Research, Vol. 12, pp. 167-174.

Train, K. 2003. Discrete choice methods with simulation. Available at:

http://eml.berkeley.edu/books/choice2nd/Ch06_p134-150.pdf [Accessed 2 July 2014]

Train, K. and Winston, C. 2007. Vehicle Choice Behavior and the Declining Market Share of

US Automakers. International Economic Review, vol. 48, no. 4, pp. 1469-1496.

Tsamboulas,D.A. 2001. Parking fare thresholds: a policy tool. Transport Policy 8, pp. 115-124.

Valleley,M. 1997. Parking Perspectives: a source book for the development of parking policy.

Landor Publishing.

108

Waerden, V.D.P. et al. 2003. Travelers micro-behavior at parking lots: a model of parking

choice behavior. Available at: http://www.ltrc.lsu.edu/TRB_82/TRB2003-000415.pdf

[Accessed 29 June 2014]

Yates, D.S. et al. 2008. The Practice of Statistics, 3rd Ed. Freeman. ISBN 978-0-7167-7309-2.

Young, W. 1986. PARKSIM/1: a network model for parking facility design. Traffic Engineering

and Control 27, pp. 606-613.

Young,W. et al. 1991. A Review of Urban Car Parking Models. Transport Reviews, vol. 11(1),

pp 63-84.

109

Appendices

Appendix I: Questionnaire for the pilot survey (BLCK_1)

Cardiff City Centre Parking-User Survey

Hello my name is {your name} and is a student at Cardiff University.

As part of a research project, we are conducting a survey of parking users. The

survey will only take 5-7 minutes of your time. Would it possible to ask you a few

questions about parking? You can refuse to answer any of the questions that

follow. Your answers are anonymous, and only aggregate statistics will be

published. You can refuse to answers any questions you do not wish to and you can

quit at time during this survey. Thank you.

Interview start time {record start time}

Section A: Parking information

Q1. Where are you travelling from? {ask for name of place and postcode}

Locality1a(more specific than Cardiff)

______________________________

Postcode1b: _________________

Q2. Would that be home, work or other place?

1 Home 2 Work 3 Other_________________

Q3. What is the main reason for travelling to Cardiff City centre today?

1 Shopping 2 Work/Business

3 Leisure 4 Other

Q4. How long are you planning to park here?{record the time and minutes/hours}

_________________

Q5. What is the reason for choosing to you park here? {mark the most appropriate to the

answer}

1 It is the only one I

know

2 Close to destination 3 Easy to find a parking space

4 Reasonable parking

price

5 It is safe to park here 6 Other (please

specify)____________

Q6. How long did it take to find a parking space? {Check either immediately or stated time}

1 Immediately upon arrival OR record time (minutes) ________________________

110

Q7. How often do you park here?

1 Every weekday 2 2-3 times a week 3 Once a week 4 2-3 times a

month

5 Every fortnight 6 Once a month 7 This is the first time I visit Cardiff City

centre

Q8. Approximately how far (in minutes) is your end destination from this car park?

Please write here: ______________________________

Q9. How many adults/children are travelling with you today?

Adults10a (note number) ______________ Children10b (note number) ________________

Q10. Please rate from 1 to 5 how satisfied you are with the following aspects of

parking? (5 being very ‘satisfied’ and 1 being ‘very dissatisfied’ )

Aspects Rate

Parking charge

Ease of finding a parking space

Clarity of information on pay and display machines e.g. pricing, length of stay, etc.

Range of payment of options

Personal safety

Vehicle safety

Section B: Hypothetical parking choices

B1. If the cost of the fare for you to park today increased by £1.00 and you could find parking

space immediately, then what would be your choice?

1 Continue to

park here

2 Park elsewhere 3 Travel by other mode:

______________________

4 Not make the

trip

B2. If the cost of the fare for you to park today increased by £2.00 and the time to find a park

space was 2 minutes, then what would be your choice?

1 Continue to

park here

2 Park elsewhere 3 Travel by other mode:

______________________

4 Not make the

trip

B3. If the cost of the fare for you to park today increased by £1.50 and the time to find a park

space was 6 minutes, then what would be your choice?

1 Continue to

park here

2 Park elsewhere 3 Travel by other mode:

______________________

4 Not make the

trip

111

B4. If the cost of the fare for you to park today increased by £2.50 and the time to find a park

space was 2 minutes, then what would be your choice?

1 Continue to

park here

2 Park elsewhere 3 Travel by other mode:

______________________

4 Not make the

trip

Section C: Parking user information

Q11. {surveyor records this without asking} 0 Male 1 Female

Q12. What is your age? You only need to indicate in which age group you belong to.

Would you say?

117-24 225-34 335-44 445-55 555-65 6Over 65 7Refused

Thank you very much for your time!

It is very much appreciated. My department is the School of Planning and Geography in Cardiff

University. Should you have any enquire about this study, please contact my supervisor Dr

Dimitris Potoglou on tel. 02920876088; Email: [email protected]

Interview End time {record end time}

112

Appendix II: Questionnaire for the main survey (BLCK_1)

Cardiff City Centre Parking-User Survey

Hello my name is {your name} and is a student at Cardiff University.

As part of a research study, we are conducting a survey of parking users. The survey

is about 5 minutes of your time. Would it possible to ask you a few questions about

parking? Your answers are anonymous. You can refuse to answers any questions

you do not wish to and you can quit at time during this survey. Thank you.

Interview start time {record start time}

Section A: Parking information

Q1. Where are you travelling from? {ask for name of place and postcode}

Locality1a(more specific than Cardiff)

_____________________________

Postcode1b:_____________________

Q2. Would that be home, work or other place?

1 Home 2 Work 3 Other_________________

Q3. What is the main reason for travelling to Cardiff City centre today?

1 Shopping 2 Work/Business

3 Leisure 4 Other

Q4. How long are you planning to park here? {record the time in minutes/hours}

_________________

Q5. What is the reason for choosing to park here? {mark the most appropriate to the

answer}

1 It is the only car park

I know

2 Close to destination 3 Easy to find a parking space

4 Reasonable parking

price

5 It is safe to park here 6 Other (please

specify)____________

Q6. How long did it take to find a parking space? {Check either immediately or stated time}

1 Immediately upon arrival OR record time (minutes) ________________________

113

Q7. How often do you park here?

1 Every weekday 2 2-3 times a week 3 Once a week 4 2-3 times a

month

5 Every fortnight 6 Once a month 7 This is the first time I visit Cardiff City

centre

Q8. Approximately how far (in minutes) is your end destination from this car park?

Please write here: ______________________________

Q9. Including yourself, how many adults/children are travelling with you today?

Adults10a (note number) ______________ Children10b (note number) ________________

Q10. Please rate from 1 to 5 how satisfied you are with the following aspects of

parking? (5 being very ‘satisfied’ and 1 being ‘very dissatisfied’)

Aspects Rate Reason for low rating

Parking charge

Ease of finding a parking space

Clarity of information on pay and display machines e.g. pricing,

length of stay, etc.

Range of payment of options

Personal safety

Vehicle safety

Q11. Have you heard of 'Park Mark?' 0 Yes 1 No

Section B: Hypothetical parking choices

B1. If the cost to park today increased by £1.00 and you could find parking space immediately,

then what would be your choice?

1 Continue to

park here

2 Park elsewhere 3 Travel by other mode:

______________________

4 Not make the

trip

B2. If the cost to park today increased by £2.00 and the time to find a park space was 2 minutes,

then what would be your choice?

1 Continue to

park here

2 Park elsewhere 3 Travel by other mode:

______________________

4 Not make the

trip

114

B3. If the cost to park today increased by £1.50 and the time to find a park space was 6 minutes,

then what would be your choice?

1 Continue to

park here

2 Park elsewhere 3 Travel by other mode:

______________________

4 Not make the

trip

B4. If the cost to park today increased by £2.50 and the time to find a park space was 2 minutes,

then what would be your choice?

1 Continue to

park here

2 Park elsewhere 3 Travel by other mode:

______________________

4 Not make the

trip

Section C: Parking user information

Q12. {surveyor records this without asking} 0 Male 1 Female

Q13. What is your age? You only need to indicate in which age group you belong to.

Would you say?

117-24 225-34 335-44 445-55 555-65 6 Over 65 7Refused

Thank you very much for your time!

It is very much appreciated. My department is the School of Planning and Geography in Cardiff

University. Should you have any enquire about this study, please contact my supervisor Dr

Dimitris Potoglou on tel. 02920876088; Email: [email protected]

Interview End time {record end time}

115

Appendix III: Logistic regression result of parking users’ general sensitivities to

parking charge and parking availability.

Case Processing Summary

Unweighted Casesa N Percent

Selected Cases Included in Analysis 932 100.0

Missing Cases 0 .0

Total 932 100.0

Unselected Cases 0 .0

Total 932 100.0

a. If weight is in effect, see classification table for the total number of cases.

Dependent Variable Encoding

Original Value Internal Value

Not continue to park here 0

Continue to park here 1

Block 0: Beginning Block

Classification Tablea,b

Observed Predicted

Park_or_not Percentage

Correct Not continue

to park here

Continue to

park here

Step

0

Park_or_not Not continue to park

here

508 0 100.0

Continue to park here 424 0 .0

Overall Percentage 54.5

a. Constant is included in the model.

b. The cut value is .500

Variables in the Equation

B S.E. Wald df Sig. Exp(B)

Step 0 Constant -.181 .066 7.550 1 .006 .835

116

Variables not in the Equation

Score df Sig.

Step 0 Variables INCR_COST 234.435 1 .000

TIME 32.118 1 .000

Overall Statistics 264.948 2 .000

Block 1: Method = Enter

Omnibus Tests of Model Coefficients

Chi-square df Sig.

Step 1 Step 296.848 2 .000

Block 296.848 2 .000

Model 296.848 2 .000

Model Summary

Step -2 Log likelihood Cox & Snell R Square Nagelkerke R Square

1 987.597a .273 .365

a. Estimation terminated at iteration number 5 because parameter estimates changed by less

than .001.

Classification Tablea

Observed Predicted

Park_or_not Percentage

Correct Not continue

to park here

Continue to

park here

Step

1

Park_or_not Not continue to park

here

389 119 76.6

Continue to park here 118 306 72.2

Overall Percentage 74.6

a. The cut value is .500

Variables in the Equation

B S.E. Wald df Sig. Exp(B)

Step 1a INCR_COST -1.492 .106 196.302 1 .000 .225

TIME -.226 .036 39.159 1 .000 .797

Constant 3.034 .239 161.198 1 .000 20.789

a. Variable(s) entered on step 1: INCR_COST, TIME.

117

Appendix IV: Logistic regression result of parking users’ sensitivities under the

combinations of various characteristics.

Step 1:

Case Processing Summary

Unweighted Casesa N Percent

Selected Cases Included in Analysis 932 100.0

Missing Cases 0 .0

Total 932 100.0

Unselected Cases 0 .0

Total 932 100.0

a. If weight is in effect, see classification table for the total number of cases.

Dependent Variable Encoding

Original Value Internal Value

Not continue to park here 0

Continue to park here 1

Block 0: Beginning Block

Classification Tablea,b

Observed Predicted

Park_or_not Percentage

Correct Not continue

to park here

Continue to

park here

Step

0

Park_or_not Not continue to park

here

508 0 100.0

Continue to park here 424 0 .0

Overall Percentage 54.5

a. Constant is included in the model.

b. The cut value is .500

Variables in the Equation

B S.E. Wald df Sig. Exp(B)

Step 0 Constant -.181 .066 7.550 1 .006 .835

118

Variables not in the Equation

Score df Sig.

Step 0 Variables INCR_COST 234.435 1 .000

TIME 32.118 1 .000

INCR_COSTFemale 54.462 1 .000

TIMEFemale 17.694 1 .000

INCR_COSTWork 14.524 1 .000

TIMEWork .238 1 .626

INCR_COSTAge25_44 78.964 1 .000

TIMEAge25_44 16.667 1 .000

INCR_COSTOver45 12.614 1 .000

TIMEOver45 3.750 1 .053

INCR_COSTCompanion 29.472 1 .000

TIMECompanion 7.825 1 .005

Overall Statistics 286.180 12 .000

Block 1: Method = Enter

Omnibus Tests of Model Coefficients

Chi-square df Sig.

Step 1 Step 328.128 12 .000

Block 328.128 12 .000

Model 328.128 12 .000

Model Summary

Step -2 Log likelihood Cox & Snell R Square Nagelkerke R Square

1 956.317a .297 .397

a. Estimation terminated at iteration number 5 because parameter estimates changed by less

than .001.

119

Classification Tablea

Observed Predicted

Park_or_not Percentage

Correct Not continue

to park here

Continue to

park here

Step

1

Park_or_not Not continue to park

here

399 109 78.5

Continue to park here 117 307 72.4

Overall Percentage 75.8

a. The cut value is .500

Variables in the Equation

B S.E. Wald df Sig. Exp(B)

Step

1a

INCR_COST -1.404 .226 38.461 1 .000 .246

TIME -.333 .104 10.263 1 .001 .717

INCR_COSTFemale -.096 .123 .619 1 .432 .908

TIMEFemale -.018 .061 .083 1 .774 .983

INCR_COSTWork .066 .125 .281 1 .596 1.068

TIMEWork .124 .062 3.992 1 .046 1.132

INCR_COSTAge25_44 -.459 .175 6.857 1 .009 .632

TIMEAge25_44 .105 .088 1.414 1 .234 1.110

INCR_COSTOver45 -.131 .178 .539 1 .463 .877

TIMEOver45 .019 .091 .045 1 .832 1.019

INCR_COSTCompanion .274 .126 4.700 1 .030 1.315

TIMECompanion -.006 .062 .010 1 .921 .994

Constant 3.132 .245 163.249 1 .000 22.925

a. Variable(s) entered on step 1: INCR_COST, TIME, INCR_COSTFemale, TIMEFemale,

INCR_COSTWork, TIMEWork, INCR_COSTAge25_44, TIMEAge25_44, INCR_COSTOver45,

TIMEOver45, INCR_COSTCompanion, TIMECompanion.

120

Step 2:

Case Processing Summary

Unweighted Casesa N Percent

Selected Cases Included in Analysis 932 100.0

Missing Cases 0 .0

Total 932 100.0

Unselected Cases 0 .0

Total 932 100.0

a. If weight is in effect, see classification table for the total number of cases.

Dependent Variable Encoding

Original Value Internal Value

Not continue to park here 0

Continue to park here 1

Block 0: Beginning Block

Classification Tablea,b

Observed Predicted

Park_or_not Percentage

Correct Not continue

to park here

Continue to

park here

Step

0

Park_or_not Not continue to park

here

508 0 100.0

Continue to park here 424 0 .0

Overall Percentage 54.5

a. Constant is included in the model.

b. The cut value is .500

Variables in the Equation

B S.E. Wald df Sig. Exp(B)

Step 0 Constant -.181 .066 7.550 1 .006 .835

121

Variables not in the Equation

Score df Sig.

Step 0 Variables INCR_COST 234.435 1 .000

TIME 32.118 1 .000

TIMEWork .238 1 .626

INCR_COSTAge25_44 78.964 1 .000

INCR_COSTCompanion 29.472 1 .000

Overall Statistics 282.449 5 .000

Block 1: Method = Enter

Omnibus Tests of Model Coefficients

Chi-square df Sig.

Step 1 Step 322.990 5 .000

Block 322.990 5 .000

Model 322.990 5 .000

Model Summary

Step -2 Log likelihood Cox & Snell R Square Nagelkerke R Square

1 961.455a .293 .392

a. Estimation terminated at iteration number 5 because parameter estimates changed by less

than .001.

Classification Tablea

Observed Predicted

Park_or_not Percentage

Correct Not continue

to park here

Continue to

park here

Step

1

Park_or_not Not continue to park

here

393 115 77.4

Continue to park here 117 307 72.4

Overall Percentage 75.1

a. The cut value is .500

122

Variables in the Equation

B S.E. Wald df Sig. Exp(B)

Step

1a

INCR_COST -1.555 .132 139.152 1 .000 .211

TIME -.297 .043 48.074 1 .000 .743

TIMEWork .145 .045 10.562 1 .001 1.156

INCR_COSTAge25_44 -.248 .088 8.001 1 .005 .780

INCR_COSTCompanion .239 .091 6.948 1 .008 1.270

Constant 3.132 .245 163.895 1 .000 22.915

a. Variable(s) entered on step 1: INCR_COST, TIME, TIMEWork, INCR_COSTAge25_44,

INCR_COSTCompanion.